Prey specialization and morphological conformation of wolves associated with woodland caribou and moose

Morphological analysis of wolves associated with woodland caribou in late succession boreal coniferous forests north of the commercial cut line and those associated with moose in early succession boreal deciduous forests south of the commercial cut line were studied in Ontario. So-called “moose-wolves” could readily be distinguished from “caribouwolves” in both genders using a few morphological measurements. Wolves associated with woodland caribou were significantly smaller in most measurements, and increased in size within seven years post-harvest as moose totally replaced caribou in the ecosystem. Whether this change in wolf morphology is related to micro-evolutionary change, the migration of larger “moose-wolves” into the area, or both, remains unclear.


Introduction
Historically, woodland caribou (Rangifer tarandus caribou) dominated most of the Boreal Forest Biome throughout Canada (Bergerud, 1974) and few moose (Alces alces) existed in the northeast of Ontario prior to the mid-1800s (Krefting, 1974).Woodland caribou originally extended as far south as Manitoulin Island and north to approximately 46° latitude (Cumming and Beange, 1993).However, since the 1880s this subspecies has disappeared in the southern part of the range.The present range distribution is north of 50° latitude (Cumming & Beange, 1987;Cumming & Beange, 1993), which approximates the northern limit of commercial timber harvest between Cochrane and Kenora (Racey et al., 1991).
Moose have benefited from early succession forests created by timber harvest (Krefting, 1974) and it has been hypothesized that wolves increase pack sizes (Bergerud, 1974;Messier, 1985;Seip, 1992;Rettie & Messier, 1998) and body mass (Hillis, 1990;Mallory & Hillis, 1995) due to increased prey size.This shift in the predator-prey system is hypothesized to result in the extirpation of local woodland caribou populations due to greater wolf numbers and wolf body mass and the loss of prime habitat (Hillis, 1990;Seip, 1992;Mallory & Hillis, 1995).Bergerud (1990) hypothesized that rareness was an anti-predator strategy of caribou, and Lariviere et al. (2000) and Cumming (1975) suggested the clumping strategy of caribou would fail when moose were present in high densities.Evidence indicates that disturbance has a major negative impact on caribou (Schaefer & Pruitt, 1991;Cumming, 1992;Antoniak, 1993;Chubbs et al., 1993;Cumming & Beange, 1993;Antoniak & Cumming, 1996;Hillis et al., 1998;Hillis & Mallory, 2004).Woodland caribou are now listed as vulnerable by COSEWIC because of the historical recession of their range associated primarily with commercial logging (Brown et al., 2003).No declines in caribou have been recorded in undisturbed Ontario populations.The primary goals of this study were: (1) to assess the status of woodland caribou in northwestern Ontario, (2) to understand habitat utilization by this subspecies, and (3) to understand the impact of human-related land use activities on this species and its predators.
Moose are the largest member of the Cervidae and are found only in boreal forest dominated by spruce, fir, and pine, where disturbance is a major factor shaping the vegetative communities (Odum, 1983;Telfer, 1984;Karns, 1998).In late winter, moose are commonly found on south-facing slopes at higher elevations, in association with protective cover; while in summer they are found near water with abundant aquatic plants (Karns, 1998).Moose numbers were few in the area north of Lake Superior until the early twentieth century; due to the predominance of old growth coniferous forest, and have only recently become common in response to logging and other disturbance (Peterson, 1953(Peterson, , 1955;;DeVos, 1958;Karns, 1998).Since the middle of the twentieth century, moose populations have shown positive growth across the continent (Bergerud, 1981;Crete, 1987;Thompson & Euler, 1987;Karns, 1998), which is believed to be due to a reduction in predators, reduced deer populations in the north due to the reversion of farmland to forest, larger clearcuts, and increased legal protection (Aldous & Krefting, 1946;Karns et al., 1974;Peek et al., 1976;Hicks, 1986;Boer, 1992;Alexander, 1993;Bontaites & Guftason, 1993;Morris & Elowe, 1993;Karns, 1998;Peek, 1998).
It is hypothesized that the decline in woodland caribou (Rangifer tarandus caribou) in Ontario is due in part to predation.Since moose predation exposes wolves to greater risk of injury or death (Weaver et al., 1992;Mallory et al., 1994), natural selection would favor larger wolves in habitats dominated by moose and smaller wolves in habitats dominated by caribou.As wolves select for the most common ungulate in the system (Lamothe & Parker, 1989), if moose dominate, the increased numbers of wolves and selection for larger wolves would subject the smaller, less common caribou to increased predation risk and population decline into a "predator pit" or local extirpation (Mallory et al., in press).Presumably, if the smaller prey is more abundant, natural selection would result in smaller-sized wolves (less food) and the larger, less common prey would present a more dangerous prey choice and tend to be avoided by the predator, resulting in sustained or increasing populations of the larger prey species.

Material and methods
Wolf carcasses were obtained from trappers across northwestern Ontario between 1997 and 1998 and wolf scats were collected from moose and caribou ranges.Necropsy procedures for wolf carcasses were followed as outlined by Roffe et al. (1996) and measurements were taken as described by Hillis (1990) and Mulders (1997).Wolf skulls were fleshed and dried prior to cleaning by dermestid beetles and the suture between the basiosphenoid and basioccipital bones was checked for ossification to identify adult animals.Only skulls of adults were used in the statistical analysis (Kolenosky & Stanfield, 1975;Skeel & Carbyn, 1977;Mulders, 1997).A total of 45 cranial, mandibular, and dental measurements to the nearest 0.01 mm were taken using digital calipers (Fig. 1; Appendix 1).
Total body weight with skin removed was measured to the nearest 0.5 kg using a Salter Scale (up to 100 kg) and external body measurements were made to the nearest mm with a fiberglass measuring tape (Fig. 2).A Sartorius Electronic Balance (up to 4 kg) was used to weigh internal organs to the nearest 0.1 g.Differences in morphology between wolves (Canis lupus) from caribou range and those from moose ranges were analyzed.Adipose tissue depth measurements were made to the nearest mm with calipers and kidney fat was estimated as described by Harder & Kirkpatrick (1996).
Wolf stomach and scat contents were washed through a 0.21-mm sieve and hairs were selected randomly and rinsed in methyl salicylate.Imprints of each hair were made in clear nail polish for prey species identification (Adorjan & Kolenosky, 1969).Scats were stored in plastic Whirlpak bags and autoclaved to kill endoparasites before handling.Each sample was assigned either a single prey category or-when there were multiple prey items in the stomach or scat-the most abundant item was used (Floyd et al., 1978;Weaver, 1993).Frequencies (%) were calculated within each prey category to allow comparisons among the 3 different habitat types: (1) no disturbance, (2) 0 to 7 years post-harvest, and (3) more than 7 years post-harvest.
Statistical analyses were used to identify differences between items in wolf diets, skull morphology, and body and organ morphology for males and females.All multivariate and univariate statistical analyses were done using a SPSSx program, and Discriminant Analysis was used to identify differences in skull and body measurements that best distinguished the wolves from caribou and moose ranges, while Student's t-tests, Mann-Whitney U, and Kruskal-Wallis one-way Analysis of Variances were performed on adipose parameters (Zar, 1974).Corresponding descriptions are listed in Table 1.

Wolf diet
Stomach and scat contents were analyzed from undisturbed areas (stomachs -n = 94; scats -n = 16) and from recently logged woodland caribou range 0 to 7 years post-harvest (stomachs n = 89; scats -n = 12).Only wolf stomach contents (n = 110) were analyzed from logged moose ranges greater than 7 years post-harvest.The percentage of caribou in wolf diets declined significantly, from 21% in areas with no disturbance, to 6% in areas where logging had occurred during the previous 0 to 7 years (c 2 (1) = 7.259; P<0.01) and to 0% in areas logged earlier than 7 years (Fig. 3).The percentage of moose in the diet of wolves was 37% in areas of no disturbance and 33% in areas logged approximately 0 to 7 years earlier and 49% on ranges where logging had occurred earlier than 7 years.The presence of white-tailed deer (Odocoileus virginianus) in wolf diets also increased significantly from 0% on ranges with no disturbance to 6% on ranges with logging approximately 0 to 7 years earlier and 9% on ranges where logging had occurred more than 7 years earlier (c 2 (2) = 8.400; P<0.025).In contrast, the percentage of beaver (Castor canadensis) in wolf diets declined from 26% on ranges with no disturbance to 22% in ranges with logging approximately 0 to 7 years earlier and to 12% on ranges where logging had occurred more than 7 years earlier.Significant changes in prey classified as "Other" were 0% on ranges with no disturbance, 28% on ranges with logging approximately 0 to 7 years earlier, and 14% on ranges where logging had occurred more than 7 years earlier (c 2 (1) = 4.024; P<0.05); prey diversity was highest in ranges 0 to 7 years post-harvest and lowest in ranges with no disturbance (Fig. 3).

Skull morphology
Male wolves A comparison of the means of adult male wolf skull measurements using a Student's t-test and a nonparametric Mann-Whitney U test is presented in Table 1.Canonical Discriminant Analysis was used to find which of 45 skull measurements best discriminate between moose-and caribou-wolves.Sixtynine (64%) of the 107 wolf skulls were excluded from the Discriminant Analyses, as they were juveniles or damaged.Twenty-three skulls of adult male wolves were used, 7 from areas within the undisturbed range of woodland caribou and 16 from ranges more than 7 years post-harvest, where moose predominated and caribou were absent.Male moose-wolves had significantly (P<0.05)larger values on numerous skull measures including: Mandible Width, Length of Premolar 4 , Zygomatic Breadth, Cheek T W, and Width Across the 1 st Molars.The maximum Zygomatic Breadth, which is associated with bite force (Wiersma, 2001), differed significantly between adult male moose-and caribouwolves from the two ranges (F = 8.3; P=0.009).Zygomatic Breadth entered the discriminant function first and explained 28% of the variance, and the Occipital Condyle Width entered the function next and accounted for an additional 19% of the variance.Only the first five measurements were allowed to enter the Discriminant Analysis to ensure that multiple measurements that vary similarly did not influence the results.Once the 5 measurements had entered the function, only 23% of the variance in all skull measurements remained to be explained, thus 77% of the variance was accounted for by the measurements listed in Table 2.A Wilks' Lambda value of 0.23 indicated that there was a highly significant difference between the two groups of adult male wolves using the 2 value from Table 2 (2(5) = 27.15;P<0.001).The discriminant function effectively assigned 100% of the male wolves that were used to generate the function to the range from which they were collected.When the function was tested on adult male wolves not used to generate the function, it correctly classified 95.7% to their respective ranges.

Female wolves
One-way Analysis of Variance and Canonical Discriminant Analysis were used to find which of 45 skull measurements best discriminated between adult female moose-and caribou-wolves.Measurements were made on 19 skulls: 8 wolves were from the undisturbed woodland caribou range and 11 from the moose-dominated and caribou-absent range, where logging had occurred more than 7 years earlier.
Adult female moose-wolves had significantly larger values (p<0.05) in Length of Premolar 4 and Width Between the 2 nd Premolars (Table 3).A One-way Analysis of Variance indicated that adult female caribou-wolves had greater Palatal Width across the 2 nd Premolars (mean = 35.1 mm) than adult female moose-wolves (mean = 33.5 mm, F = 5.9; P=0.03).Wilks' Lambda indicated that 26% of the variance between the 2 groups was accounted for by the Palatal Width measurement and the five measurements accounted for 85% of the variance.The Wilks' Lambda value (0.15) for the entire function and its 2 value indicated that the two groups separated well using the function.When the Discriminant Analysis was used to calculate discriminant scores (D) for each skull, the function correctly assigned 100% of the adult female wolves used to generate the function to their respective ranges.When the function was tested on wolves not used to generate the function, it correctly assigned 94.7% of them to their respective ranges, with only one of the 19 wolves being erroneously classified as a caribouwolf (Table 4).

Body morphology Male wolves
Table 5 compares the morphological parameters of adult male wolves collected from undisturbed caribou ranges and ranges where logging had occurred more than 7 years earlier and where moose dominated and caribou were absent.Levene's test for equality of variances was used to select the appropriate t-test.Two-tailed Student's t-test significance was confirmed by non-parametric Mann-Whitney U tests and the results of these analyses indicated that adult male moose-wolves were larger than adult male caribou-wolves.Adult male moose-wolves had higher Body Mass (31.6 kg) compared to adult male caribou-wolves (24.7 kg).Male moose-wolves also had greater mean Neck Girth (397.1 mm) than male caribou-wolves (344.3 mm).In addition, adult male moose-wolves had greater Heart, Lung, Liver, and Spleen Weights (Table 5).

Male wolves
Twenty male adult wolves were used for a Discriminant Analysis with 8 body measurements and 5 organ weights (Table 6).One-way Analysis of Variance indicated that adult male caribou-wolves had smaller Neck Girths (mean = 335.00mm) than moose-wolves (mean = 395.70mm, n = 20, F = 25.700;P<0.001).Neck Girth entered the discriminant function on the first step and accounted for 59% of the variability between wolves from the two areas.Neck Girth and Kidney Weight were the only two variables to enter the function and accounted for 65% of the total variance.Wilks' Lambda value for the function was 0.35 and its 2 value in Table 6 indicated that there was a significant difference (P<0.001) between the two groups of wolves using the body measurements listed.When scores were calculated for each male wolf used to generate the discriminant function, it correctly assigned 83% of the male wolves to their respective areas.When applied to wolves not used to generate the function, 88% of them were correctly classified (Table 6).

Female wolves
Comparisons of the mean organ masses and external body measurements of adult female wolves from undisturbed woodland caribou ranges and ranges where logging had occurred more than 7 years earlier were compared (Table 7).Levene's test for equality of variances was used to select the appropriate t-test.Two-tailed Student's t-test significance was confirmed by non-parametric Mann-Whitney U tests.The results indicated that no significant differences existed between the female moose-wolves and female caribou-wolves (P>0.05).

Discriminant analysis-Female wolves
Seventeen adult female wolves were used for a Discriminant Analysis using 8 body measurements and 5 organ weights (Table 8).One-way Analysis of Variance indicated no differences in the means of body measurements between adult female moose-wolves and adult female caribou-wolves.The Heart Weight (F = 1.900;P = 0.185) entered the discriminant function on the first step and accounted for 11% of the variability between adult female wolves from the two ranges.Kidney Weight and Neck Girth were the only other variables that entered the function and accounted for 50% of the total variance.Wilks' Lambda value for the function was 0.50 and its 2 value in Table 8 indicated that there was a significant difference (P = 0.025) between the two groups of adult female wolves using the body measurements listed.The function correctly classified 86% of the wolves used to generate the function.When applied to a sample of 22 wolves not used to create the function, it correctly classified 73% (Table 8).

Condition indices
The condition of moose-and caribou-wolves determined by adipose measurements is presented in Tables 9 and 10.Levene's test for equality of variances was used to select the appropriate t-test and two-tailed Student's t-test significance was confirmed by nonparametric Mann-Whitney U tests.No significant differences were found in adipose tissue of moose-and caribou-wolves; however, adipose indices were greater in moose-wolves for all variables measured.

Male wolves
A comparison of condition indices of adult male wolves collected from undisturbed woodland caribou ranges and disturbed moose-dominated ranges with no caribou was undertaken (Table 9).
Female wolves A comparison of condition indices of adult female wolves collected from undisturbed woodland caribou ranges and disturbed moose ranges with no caribou was undertaken and no significant differences were found (Table 10).

Wolf injuries
Old skeletal injuries distinguished by calcification (Phillips, 1984) were evident in 4 of the 106 wolves.Three of the 4 injured wolves were males from moose range.Male injures included a crushed anterior end of the nasal bones, a shattered humerus and scapula, a cracked and deformed premaxilla, and a cracked back of the skull where the occipital and sagittal crests meet.The single adult female caribou-wolf had a cracked sagittal crest.

Wolf diet
Wolf diet changed from essentially a 3 prey system (moose, caribou, beaver) in undisturbed boreal forest to a 9+ prey system in ranges 0 to 7 years post-harvest and finally to a moosespecialized system in ranges more than 7 years post-harvest.However, as moose were available within 10 to 60 km from the core caribou wintering area in this study, their presence in the stomachs and scats of wolves in the "no disturbance" area was expected.Lariviere et al. (2000) found the highest wolf densities in areas where wolves had access to white-tailed deer and moose and hypothesized that multipleprey-based ecosystems sustained higher densities of wolves than systems with single prey bases.Empty stomachs were least frequent (6%) in logged habitats 0-7 years post-harvest compared to the other habitats (15%-16%  1980-1985, 1985-1990, and 1990-1995, respec-tively.These increases were not uniform and occurred in spite of moose declines in late successional areas, such as Wabakimi Provincial Park and the area north of Lac Seul that lacked substantial timber harvest. Moose are the primary prey of wolves throughout the southern half of Quebec, with a direct exponential relationship between wolf and moose densities, particularly when moose exceeded 0.3 animals/km 2 (Lariviere et al., 2000).Below 0.3 moose/km 2 , wolf density remained relatively constant at slightly less than 1 animal/km 2 .Analysis of moose densities in Ontario between 1975and 1995(McKenney et al., 1998) indicated that initial scattered patches of higher moose density had expanded to cover a vast area across northern Ontario.The boundary separating areas of moose density above and below the provincial average of 0.209/km 2 closely approximates the southern limit of caribou distribution and the northern limit of timber harvest.Caribou continue to exist along the northern limit of timber harvest where moose densities are below 0.209.Zones of lower than average moose density south of the caribou line are separated from caribou ranges by bands of high moose density.Using the regression line of Lariviere et al. (2000), the band of high moose density would represent an ecological barrier of elevated wolf density through which caribou would be unable to re-colonize their former range.
A more varied prey base implies that wolves benefit from a more constant food supply (Cumming, 1975).Prey items such as snowshoe hare, small mammals, fish, birds, and vegetation increased from 0 in undisturbed ranges to 28% in areas 0 to 7 years post-harvest to 14% in ranges more than 7 years postharvest.Increases in diversity during the transitional stage (0 to 7 years post-harvest) can be explained by the presence of species associated with both early and late successional habitat.Forbes and Theberge (1996) found that beaver acted as a buffer prey species when ungulate densities were low.They noted that the inspection of beaver houses was a common activity among Algonquin Provincial Park wolves.In this study, winter observations of wolf tracks in a caribou wintering area indicated that wolves checked a previously opened beaver house.Tracks indicated that wolves had watched the opened house, suggesting that beaver might continue to use a house previously opened by wolves.Beaver were found in 26% of the scat/stomach samples from undisturbed ranges and in 12% of the samples from ranges more than 7 years post-harvest.
While moose were available in remote undisturbed ranges, their densities were low.Dale et al. (1995) found that caribou densities below 0.2/km 2 were associated with wolves switching to preying on moose.Messier (1985) found that wolves changed territory boundaries, increased extraterritorial excursions to white-tailed deer areas, and became more susceptible to mortality when moose density was below 0.2/km 2 .In this study-during 8 days of observations including 228 km of transects by snowmobile, snowshoe, and aircraft in the caribou wintering area-the only moose sign observed were tracks and pellet groups found in poplar stands >10 km south of the core caribou wintering area.The absence of moose was also noted while making daily observations of caribou totaling 110 individuals.These observations suggest that, historically, woodland caribou ranges had few to no moose present and that wolf density was low (Bergerud, 1990), as the migratory and clumping behaviour of woodland caribou would leave vast tracts of land seasonally devoid of ungulate prey (Cumming, 1975).
The diet analysis indicated that as sampling moved to areas with increased disturbance, moose increased in importance in the diet, wolves had fewer empty stomachs, and prey diversity increased.It was also concluded that increases in moose density associated with disturbance resulted in the increased use of moose by wolves (Messier, 1995).In addition, it was concluded that increased disturbance subjected woodland caribou to increased predation by wolves, contributing to their extirpation from much of their former range.As timber harvest moved northward, there has been a general increase in moose density (Schwartz & Franzmann, 1991;McKenney et al., 1998) coupled with a decline in woodland caribou (Bergerud, 1974;Darby et al., 1989;Mallory & Hillis, 1996;McKenney et al., 1998).The decline in caribou and the increase in moose (McKenney et al., 1998) as one moves south was also reflected in the diet of wolves (Fig. 3).Brousseau (1978), using pellet counts, documented a sharp decline in woodland caribou from 0.86/km 2 to 0.00/km 2 within 5 years following the commencement of timber harvest in the Cliff Lake area, northwestern Ontario.Aerial surveys also indicated a 75% decline in the number of caribou using the area during the 11 years of study.Similar to Brousseau (1978) and Kohira & Rexstad (1997), the wolf diet in this study supported the hypothesis that commercial timber harvest changed the predator-prey relationship.

Skull morphology
Adult male moose-wolves had significantly larger values than adult male caribou-wolves in a number of skull measures including: Mandible Width, Length of Premolar 4 , Zygomatic Breadth, Cheek T W, and Width across the 1 st Molars.The Canonical Discriminant Analysis correctly classified 95.7% of male wolves to their respective ranges using these 5 skull measures.Adult female moose-wolves had larger values than adult female caribou-wolves in two skull measures: Length of Premolar 4 and Width between the 2 nd Premolars.The Canonical Discriminant Analysis correctly assigned 94.7% of the adult female wolves to their respective ranges with only one of 19 being erroneously classified.The fact that male wolves were more different than female wolves is significant.Similar results were reported by Hillis (1990) for arctic wolves along the west coast of Hudson Bay, Nunavut.In addition, the fact that adult male moosewolves had significantly larger skull parameter values that are associated with prey capture, supports the conclusion that these differences are directly related to capture of larger, more dangerous prey.Increased width across the Zygomatic Breadth without a correlated increase in the width of the brain case provides space for larger masseter muscles.Thomason (1991) and Wiersma (2001) demonstrated that bite force increased with greater Zygomatic Breadth and Mandible Width.Larger masseters would increase the bite force and should be selected for in predators preying on larger and more dangerous prey (Mallory et al., 1994).The high classification success of the discriminant functions indicated that there are significant morphological differences in male skull parameters between moose-and caribou-wolves.Mallory et al. (1994) and Hillis & Mallory (1996) hypothesized that during hunting, adult male wolves usually make first contact with large ungulate prey and are most prominent in the capture and killing of risky prey.This behavior would strongly influence the natural selection for male wolf body size and skull morphology in relation to prey specialization, gender, and sexual dimorphism (Mallory & Hillis, 1995;Hillis & Mallory, 1996) and also influence differences between related species, such as the coyote (C.latrans) (Mallory & Edwards, 1996).

Body morphology and condition indices
As diet changed from caribou to moose, the body size of male wolves increased significantly.Male moosewolves had significantly greater mean neck girth (397.1 mm) than caribou-wolves (344.3 mm) and significantly larger hearts, lungs, livers, and spleen masses.In addition, male moose-wolves had significantly greater body mass (31.6 kg) than male caribouwolves (24.7 kg), while adult female wolves had few significant differences in body or organ weights.Schmitz & Kolenosky (1985) and Schmitz and Lavigne (1987) found that over a period of 25 years  there was an increase in mean body weight and length of coyotes at the same time as there was a decrease in the mean body length of wolves in southeastern Ontario.Prior to European settlement, wolves in southeastern Ontario were typically predators of large ungulates like moose (Kolenosky & Stanfield, 1975) and Schmitz & Lavigne (1987) concluded that changes in wolves were correlated with an increase in the use of smaller, more abundant white-tailed deer as primary prey.Increases in body size were observed in coyotes that recently arrived in the Maritimes (wolves had been extirpated since the 1800s) and it has been hypothesized that the size increase was due to specialization on the larger prey such as white-tailed deer (Crete & Desrosiers, 1995).Hillis & Mallory (1996) found that the greatest sexual dimorphism in wolves (n = 425) from Nunavut was in body and visceral organ masses, particularly of the heart, lungs, liver, and spleen.They attributed these differences to the division of labor between the sexes, with males being highly specialized for capturing and killing large ungulate prey.In this study, we found that male moose-wolves were larger than male caribou-wolves for the same set of organ measurements and concluded that the differences were due to specialization for capturing larger prey.Thicker necks in moose-wolves would assist in pulling prey down and the increase in bite force due to wider zygomatic arches (Thomason, 1991;Wiersma, 2001) and larger masseter muscles would complement increased neck girth.Larger hearts and lungs would increase the cardiovascular fitness in moose-wolves, and the larger spleen-a blood reservoir-would release more erythrocytes in response to drops in blood oxygen or blood pressure (Walker and Liem, 1994).
The fact that condition indices did not differ between moose-and caribou-wolves suggests that the body size and pack size of each wolf type adjusted to the nutritional input available from their primary prey.Although no significant differences occurred in condition indices, moose-wolves consistently had higher levels of adipose tissue.Hall (1989), working on arctic fox (Alopex lagopus), found larger litter sizes when adipose reserves increased due to an abundance of small mammals.The greater amount of adipose reserves also resulted in higher ovulation rates and we expect a similar relationship in wolves, which would affect wolf pack size.
The results from this study support the conclusion that the greater adult skull and body size of moosewolves relative to caribou-wolves was related to specialization on larger prey.The data also support the conclusion that wolves adapted to larger prey would be more efficient at capturing smaller ungulate prey and impact these populations negatively.Gates & Larter (1996) observed that after wood bison (Bison bison athabascae) were introduced to the MacKenzie Wood Bison Reserve, moose and caribou populations in that region declined, which likely reflects the same shift in the predator-prey system as described in this study on moose and woodland caribou in Ontario.

Fig. 3 .
Fig. 3. Percentage of stomach and scat contents of Canis lupus in three different habitats (no logging-caribou range; logging-caribou range; logged-moose range) from northwestern Ontario.

Table 1 .
Skull morphology of adult male wolves (Canis lupus) collected from undisturbed woodland caribou ranges and disturbed moose ranges, where logging had occurred more than 7 years earlier.Levene's test for equality of variances was used to select the appropriate t-test (equal variances assumed or not).Two-tailed Student's t-test significance was confirmed by non-parametric Mann-Whitney U tests (C = caribou range, M = moose range, Wt = weight w[g] (unless otherwise indicated), Lth = length [mm], P< 0.05*, P<0.01**, P<0.001***).

Table 2 .
Canonical Discriminant Analysis of skull measurements of adult male moose-and caribou-wolves (n = 23).

Table 3 .
Morphological measurements of adult female wolves (Canis lupus) collected from undisturbed woodland caribou ranges and disturbed moose ranges, where logging had occurred more than 7 years earlier.Levene's test for equality of variances was used to select the appropriate t-test (equal variances assumed or not).Two-tailed Student's t-test significance was confirmed by non-parametric Mann-Whitney U tests.(C = caribou range, M = moose range, Wt = weight [g], Lth = length [mm]), P< 0.05*, P<0.01**, P<0.001***).

Table 4 .
Canonical Discriminant Analysis of skull measures of adult female moose-and caribou-wolves (n = 19).

Table 5 .
Morphological parameters of adult male wolves (Canis lupus) collected from undisturbed woodland caribou ranges and ranges where logging had occurred more than 7 years earlier where moose dominated and caribou were absent.Levene's test for equality of variances was used to select the appropriate t-test (equal variances assumed or not).Two-tailed Student's t-test significance was confirmed by non-parametric Mann-Whitney U tests (C = caribou range; M = moose range; Wt = weight [g] (unless otherwise indicated); Lth = length [mm]), P< 0.05*, P<0.01**, P<0.001***).

Table 6 .
Canonical Discriminant Analysis of body measures of adult male moose-and caribou-wolves.

Table 7 .
Morphological measurements of adult female wolves (Canis lupus) collected from undisturbed woodland caribou ranges and moose ranges.Levene's test for equality of variances was used to select the appropriate t-test (equal variances assumed or not).Two-tailed Student's t-test significance was confirmed by non-parametric Mann-

Table 8 .
Canonical Discriminant Analysis of body measures of adult female moose-and caribou-wolves (n = 17).

Table 9 .
Condition indices of adult male wolves (Canis lupus) collected from undisturbed woodland caribou ranges (caribou range) and disturbed moose ranges.Levene's test for equality of variances was used to select the appropriate t-test (equal variances assumed or not).Two-tailed Student's t-test significance was confirmed by nonparametric Mann-Whitney U tests (C = caribou range; M = moose range, P< 0.05*, P<0.01**, P<0.001***).

Table 10 .
Condition indices of adult female wolves (Canis lupus) collected from undisturbed woodland caribou ranges and disturbed moose ranges.Levene's test for equality of variances was used to select the appropriate t-test (equal variances assumed or not).Two-tailed Student's t-test significance was confirmed by non-parametric Mann-Whitney U tests (C = caribou range; M = moose range, P< 0.05*, P<0.01**, P<0.001***).