Minnesotans For Sustainability©
Sustainable Society: A society that balances the environment, other life forms, and human interactions over an indefinite time period.
Patterns of Prey Selection by Wolves
L. David Mech, Thomas J. Meier,
Wolves tend to prey on young and older animals and those in poor condition (Murie
1944, Mech 1966a, 1970, Pimlott et al. 1969, Mech and Frenzel 1971a, Mech and
Karns 1977, Haber 1977, Peterson 1977, Fritts and Mech 1981, Peterson et al.
1984, Ballard et al. 1987). However, most studies documenting these patterns
were conducted where there was 1 primary prey species. In addition, several
studies involved prey that were also subject to human harvest, and many relied
on aging by tooth wear, which is less precise than counting cementum annuli
(Wolfe 1969, for moose). We examine patterns of selection by wolves on three
species of prey from a system in which neither wolves nor prey are harvested,
and the ages of prey were based on cementum annuli.
This study was conducted from 1986 through early 1992 in Denali National Park and Preserve (Denali), Alaska (63° N, 151° W). Denali includes about 9,200 km2 of "old" park (formerly Mount McKinley National Park), where wolves are legally protected, and an apron of 15,200 km2 of new park and preserve where wolves can be taken under various restrictions. (During the study, four wolves were known to have been killed in this area). Denali is home to about 2,000 moose (Meier et al. 1991), 3,000-4,000 caribou, including those adjacent to the park that are within range of park wolf packs (Adams et al. 1989a, Shults and Adams 1990), and approximately 2,000 Dall sheep. Little information exists to evaluate trends in the sheep and moose populations during the study, although moose estimates for both 1986 and 1991 were about the same (Meier 1987, Meier et al. 1991). Caribou numbers increased an average of 8% annually through 1990, then declined by 18% in 1991 (Adams et al. 1989a, 1993).
The wolf population increased an average of 27%/year from about four to eight wolves /1,000 km2 in late winter 1987 to 1990 (Meier et al. 1993). Denali wolves prey on all 3 ungulate species as well as on various minor prey species (Murie 1944,Mech et al. 1991a).
Snowfall during the study ranged from the second lowest on record in 1985-86 to one of the highest in 1990-91 (Table 1).
We aerially located radio-tagged wolves to find their kills, and examined the remains from the ground. Wolves were captured primarily by darting from a helicopter and were fitted with radio collars. Collared wolves were located an average of three times per month with antennae-equipped PA-18 Supercub and Cessna 185 aircraft. Not all packs were located during each flight, and some packs were located more frequently than others. Wolves were located more often in late winter and spring than during late autumn and early winter; summer data were especially sparse.
Kill sites were examined from the ground to determine the species, cause of death, age, sex, and condition of prey. The position of the carcass, presence of blood, signs of a struggle, and manner of feeding provided clues as to whether wolves killed or merely scavenged a particular carcass. Sex was determined by the presence of antlers or antler pedicels (moose), antler size (caribou), horn shape and size (sheep), mandible, metatarsus, or metacarpus length (caribou and sheep), or pelvis shape (all species). Teeth, usually incisors, were collected for age determination by sectioning and counting cementum annuli (Matson's Laboratory, Milltown, Montana). Age of younger animals was determined by the pattern of tooth eruption. Sheep were aged by annular rings on the horns and by tooth eruption and tooth sectioning. Marrow (usually femur marrow) was collected, dried, and weighed to determine percent fat content (Neiland 1970). The reported fat percentage is probably higher than actual because some specimens were not retrieved from kills until long after the animal had died.
All bones found at kill sites were examined for abnormalities, and certain bones were collected, if available, for subsequent cleaning and examination.
Because of the disparity in sizes of the wolf's prey, intermittent sampling of wolf locations would result in a bias toward locating the largest prey, which provide more food and occupy wolves longer. Therefore, we used two approaches to analyzing the kill data for determining the proportions of moose, caribou, and sheep killed. First we examined the observed data directly. Second, we assumed that the time wolves spent on a given age, sex, and species of prey was directly proportional to the weight of that class of prey, so the relative proportions of kill classes were adjusted accordingly. The observed number of each age, sex, and species of prey was multiplied by the reciprocal of the assumed weight of each prey class to arrive at an adjusted proportion of each kill class. For example, assuming bull moose weighed four times as much as caribou cows, we multiplied the number of caribou cows by one and bull moose by one quarter. Sample sizes of prey remains data vary for each type of data because not all types of data could be collected from every kill or prey carcass found. We performed two analyses of the monthly kill proportions because of strong bias during summer against finding calf kills and disproportionate sampling of certain packs. One year-round analysis excluded calves, and one analysis for October through April included calves.
To determine if any class of prey was killed disproportionately in any given
month, we used the Chi2 test to compare monthly
proportions of each prey class against the mean of the monthly proportions of
the year-round sample. We performed this test for adults only and adult and calf
samples. Significance for a given month was assumed if P0.05
for that month. For comparing age structures, we used the Kolmogorov-Smirnov
test (Hollander and Wolfe 1973).
One hundred and seven wolves were captured and radio-tagged in 25 packs distributed throughout Denali during 1986 to 1991. Some packs included up to four wolves wearing active collars concurrently because radio-collars usually lasted for several years. During each year, five to 19 packs included radio-tagged members.
Remains of 294 moose, 225 caribou, and 63 sheep were found by tracking wolves. Of these, 245 moose, 221 caribou, and 60 sheep were considered wolf kills or probable wolf kills (hereafter pooled as "kills"). This is our basic sample from which various subsamples were examined for different analyses. Some 167 moose, 165 caribou, and 49 sheep were examined from the ground to determine species, age, sex, condition, abnormalities, and cause of death.
Our tallies of kills found with collared packs probably were not a representative sample of the kills made by these packs for several reasons: 1) flying efforts were not distributed evenly over the year; 2) wolves spend more time at the kills of larger animals, and 3) larger kills such as moose are more visible and identifiable from the air.
Overall, moose represented 47% of the kills found, caribou 42%, and sheep 11%
(Table 2). Because of the biases (previously discussed), this sample probably
exaggerates the relative numbers of moose taken and minimizes the proportion of
sheep. However, the proportions may more accurately represent the relative
biomass of the three prey consumed. About 15% of all moose eaten by wolves
(about 40% of the bulls and 9% of cows and calves) were thought to have been
scavenged. They were not included in the kill sample. Haber (1977) also found
that Denali wolves scavenged considerably on bull moose.
The composition of wolf kills varied by both year and month. Moose formed the largest percentage of the sample during the first 4 years of the study, but caribou predominated in 1989-90 and 1990-91 (Table 2, Fig. 1). As caribou kills increased, both moose and sheep kills decreased. This trend held true both for the observed proportions of kills in our sample and for the proportions derived after adjustment for weight.
On a monthly basis, the total sample of kills suggested that wolves tended to
kill moose calves year around, caribou calves in May, June, March, and April
(1990-1992 only), caribou bulls year around, but primarily in July through
November, bull moose year around, but primarily in November and December,
caribou cows primarily February through June, cow moose October through May, and
sheep primarily September through April and especially December (Table 3, Fig.
2). Some of these trends may have resulted from sampling error or bias. Chi
square analysis indicated that significantly more moose were taken November
through February, more caribou cows and calves in March, and more sheep in
December. Although on a monthly basis the preponderance of caribou bulls taken
August to October was not significantly different from its proportion of prey in
other months (Table 3), bull caribou formed the single most predominant prey
type for any single period of the year (Fig. 2).
Denali wolves killed primarily calves and old moose and caribou of both sexes (Figs. 3 and 4), except during multiple caribou kills (see below). Although our sample of Dall Sheep was small, apparently individuals aged 5 years were taken disproportionately (Fig. 5). These results parallel those of Murie (1944), Burles and Hoefs (1984), Sumanik (1987), and Hayes et al. (1991) for sheep; Mech (1966a), Peterson (1977), Haber (1977), Fuller and Keith (1980), Peterson et al. (1984), Ballard et al. (1987), Bjorge and Gunson (1989), Hayes et al. (1991), and Gasaway et al. (1992) for moose; and Parker and Luttich (1986) for caribou. Our oldest kills of each species were estimated at 20 years old, 22 years old, and 13 years old for moose, caribou, and sheep, respectively.
There was no significant difference between the sex ratio of wolf-killed moose and the sex ratio of the herd (Meier et al. 1991), but wolves killed significantly more male caribou (119:100; = 13.7).
This differs from the even sex ratio found by Parker and Luttich (1986) for wolf-killed caribou in Labrador. Samples of the males of each species killed by wolves were younger than those of the females (Figs. 3-5), although only the male moose and sheep were significantly younger.
Ballard et al. (1987) found no significant difference between the sex ratio of wolf-killed caribou and that of caribou surveyed by air in south-central Alaska.
During late winter and spring 1990 and 1991, record amounts of snow fell in
the study area (Table
1), and several multiple wolf kills of caribou were found. For both males
and females, the age distributions of caribou killed during these bouts of mass
predation (Fig. 4) were significantly younger than those killed individually and
at other times of the year (P <.01). Contrary to reports from other
areas (Kelsall 1957, Eide and Ballard 1982, Miller et al. 1985, 1988b), Denali
wolves returned repeatedly to feed on the frozen carcasses of the multiple kills.
The most striking change in the composition of wolf kills during our study was the major increase in the proportion and composition of caribou killed beginning the second consecutive winter (1989-1990) with above-average snowfall (Table 1, Fig. 1). The proportion of adult moose we found killed in that winter greatly decreased (Fig. 1). Although winter 1988-89 was the first year of above-average snowfall during the study, the increased proportion of wolf-killed caribou cows was not found until 1989-90 and 1990-91 (Table 2), both of which also had above-average snowfall (Table 1). This may be evidence for a cumulative snow effect (Mech et al. 1987) on caribou condition. The proportion of caribou in observed wolf kills each year, including those we could not sex or age, increased with the cumulative snowfall) for that year (R2 = 0.70; P = 0.02). When a sub-sample of caribou kills (Table 2) that could be sexed and aged was examined relative to snowfall, it was the proportion of caribou cows in the wolf kills that varied most with snowfall (R2 = 0.54; P = 0.06; bulls, R2 = 0.33; P = 0.17). There was no substantial change in proportion of time spent locating various wolf packs, or in the geographic area of coverage, that might account for the switch to caribou during the study.
The occurrence of caribou calves in the sample of winter wolf kills was related to snowfall (Table 1) during the winter they were in utero. Caribou calves were killed by wolves during winters that followed winters of above-average snowfall, whereas after winters of below-average snowfall, no wolf-killed caribou calves were found the next winter (Table 2). This parallels the results of neonate caribou survival in the same herd. Following two winters of below-average snowfall, mortality of calves 30 days old averaged 39%, whereas following three winters of above-average snowfall, the average was 67% (Adams et al. 1993).
No corresponding relationship was found between winter snowfall and proportion of moose calves killed by wolves. Where moose are the primary prey of wolves, calf vulnerability to wolf predation is related to snowfall (Peterson 1977). In Denali, increased moose calf vulnerability based on snowfall may have been masked by the increased kill of caribou.
The mean marrow-fat content of moose, caribou and sheep killed by wolves was
low during each of the winters (October-April) of the study. Means for moose
were lowest in 1990-91 (Table 4), when snow was deepest (Table
1). Marrow fat was less in moose and caribou calves than in adults, less in
male caribou during October through May (no January data) than in June through
September, and less in bull moose than in cows (Table 5). Percent marrow fat in
29 caribou killed in multiples during winters of deep snow averaged 66±4%
compared with 47±5% for 44 caribou killed individually during all winters. The
distribution of marrow fat for each species contained individuals with
<20%-90% (Fig. 6).
More than a third of the wolf-killed moose examined showed mandibular necrosis, and a third or more had arthritis in their lumbosacral or coxofemoral joints (Table 6). Similarly, jaw necrosis and arthritis was found in wolf-killed moose on Isle Royale (Peterson 1977), on Kenai Peninsula, Alaska (Peterson et al. 1984), and earlier in Denali (Haber 1977). Arthritis afflicted male moose as young as five years of age, whereas no sign of it was found in cows 14 years old (Table 6). This contrasts with reports that arthritis afflicted moose cows as young as eight years on Isle Royale (Peterson 1977) and six years of age on Kenai Peninsula (Peterson et al. 1984). On Isle Royale, as well as in our study, the incidence of arthritis was higher in male moose (Peterson 1977).
More wolf-killed sheep than moose showed necrosis (Table 6). Murie (1944) also reported a high rate of necrosis in Denali sheep. We found no arthritis in our small sheep sample, except for a severely arthritic mandibular ramus on a 10-year-old ram.
The incidence of mandibular necrosis and arthritis in wolf-killed caribou was
lower than in sheep or moose (Table 6), but the incidence of necrosis in our
sample was higher than the incidences reported by Doerr and Dietrich (1979) and
several authors whose work they summarized. The incidence was higher in male
caribou than in females, but not significantly so (P = 0.14). Doerr and
Dietrich (1979) found significantly more males with necrosis than females.
The Denali wolf-prey system is, in all important respects, a natural system. Thus, our findings should represent a view of the way wolf-prey systems have functioned in the past throughout at least the northern range of the wolf. Differences with studies elsewhere could be related primarily to the fact that in most other study areas, either the wolves, the prey, or both, have been harvested or otherwise modified by humans.
In the Denali system, numbers of moose, caribou, sheep, and wolves have been interacting and fluctuating naturally for many years. During our study, caribou increased (Adams et al. 1993). This finding may represent one of the exceptions implied by Seip (1991:51), when he concluded that, "Caribou generally appear unable to survive in areas where there is extensive overlap with wolves and alternate prey species."
We do not know the trend in populations of moose and sheep. No drastic decline was not discernible in either, and both species remained about as widely distributed as they have been for decades (Haber 1977). Wolves consumed their prey as completely as possible, leaving only hair, rumen contents, and bones, most of them chewed. Thus, there was no indication, such as incomplete consumption of kills (Pimlott et al. 1969, Mech and Frenzel 1971a, Peterson 1977), that conditions were especially extreme for prolonged periods during most of the study.
Under these circumstances, the Denali wolves found enough prey not only to survive and reproduce, but to double in numbers (Meier et al. 1993) during a period of widely variable snowfall. They showed a high degree of selectivity in their predation patterns over different dimensions in their relationships with prey. The commonality of each dimension, however, was prey vulnerability.
Vulnerability took the form of youth, old age, poor condition, and hindrance by snow, and it varied by species, sex, time of year, and snow depth. Calves were especially important prey during summer when they are weakest, bulls before, during and after their autumn rut when they are most vulnerable, and cows in late winter when snow depth, negative energy-balance, and the drain of pregnancy reduce nutritional state. Probably the poor nutritional condition of caribou bulls during and after the rut (Table 5) explains why wolves took proportionately more bulls than cows. Although cows and calves become more vulnerable primarily during or after winters of above-average snowfall, bulls must rut every year. Rutting ungulates are generally in poor nutritional condition due to fighting and chasing females rather than feeding (Bergerud 1973, Geist 1974, Clutton-Brock et al. 1982). The reason for caribou bull vulnerability before the rut is currently under investigation (Adams et al. in preparation).
Our sample of sheep kills was small and biased. Nevertheless, it is clear that wolves kill sheep year around, apparently more during late fall, possibly a result of the vulnerability of rutting males. Murie (1944) and Haber (1977) also found that sheep were important to wolves in Denali.
Skeletal abnormalities and other possibly-debilitating factors were found among prey remains. The incidence of these conditions in the general prey population is unknown, and to what degree they contribute to prey vulnerability is open to conjecture. Jaw necrosis can result in abnormal occlusion and tooth loss, and was common in adult moose and sheep taken by wolves. Arthritis of the lumbosacral joint (between the sacrum and sixth lumbar vertebra) appears to be related to age in moose, with severe arthritis common in animals 15 years. Few skeletal abnormalities have been found among the remains of caribou eaten by wolves.
The mean ages of female prey animals taken were older than those of males. Because arthritis did not afflict cow moose until much older than bulls, this suggests that the arthritis may have helped predispose older individuals to predation.
Although Denali wolves were able to survive and increase during periods of below-average snowfall, above-average snowfall in Denali helped predispose prey to wolf predation. Deep snow had a direct effect on reducing prey condition and mobility and thus increasing predation by wolves (Mech and Frenzel 1971a, Mech and Karns 1977, Peterson 1977, Haber 1977, Nelson and Mech 1986b). We also found evidence of an indirect effect of snow depth on caribou calves that had been in utero and thus were predisposed to wolf predation during the next summer (Adams et al. 1993) and winter (Table 2), similar to findings in other wolf-prey systems (Mech and Karns 1977, Peterson 1977, Mech et al. 1987, 1991b). The most important common denominator in predisposing Denali prey to wolves was probably nutritional condition indicated by the low marrow fat content in all of the prey species, ages, and sexes of our wolf-kill sample (Table 4). Considering that some individuals must have been predisposed by physical frailties not apparent in the bones, which were usually all that could be examined for most kills (Mech 1970), and that such animals would not necessarily show low marrow fat (Mech and DelGuidice 1985), the low average percent fat we found is striking. This is especially true given that our values are probably artificially high (see Methods).
In caribou, femur fat <70% indicates that the animal's body weight has declined about to its limit (Dauphine 1971), with total body fat <5% (Huot and Goudreault 1985). Adult caribou cows in good condition possess 11-14% body fat (Dauphine 1971, Huot and Goudreault 1985), and bulls 31% (Dauphine 1971), with marrow fat 70%. Starvation has been documented in adult moose at a mean marrow fat level of 52% (S.E.= 15.3) (Ballard et al. 1987).
However, several workers believe that marrow fat must reach much lower levels before indicating that an animal is near death. Stephenson and Johnson (1972), Franzmann and Arneson (1976), Peterson et al. (1984) and Hayes et al. (1991) used 20% marrow fat in adults and 10% in calves as indicators of starvation. Although there is value in being conservative, using such low levels ignores starvation physiology and risks reaching erroneous conclusions.
Ungulate marrow fat 70-87%, depending on species, is a direct indicator of total body fat, but by the time the marrow fat is as low as this threshold, the greater majority of body fat has already been lost (Dauphine 1971, Huot and Goodreault 1985, Watkins et al. 1991, Holand 1992). As ungulates lose fat stores, they also lose protein, or muscle mass (Leibholz 1970, Paquay et al. 1972, Hovell et al. 1987, Torbit et al. 1985, DelGuidice et al. 1990). In adult white-tailed deer (Odocoileus virginianus), for example, the R2 between weight loss and protein (muscle) loss was 0.91 (DelGuidice et al. 1990). At maximal work loads, such as when running from wolves, it is muscle glycogen that forms the major source of fuel (Froberg et al. 1971, Hultman and Nilsson 1971). Furthermore, blood glucose which also is important to a running animal, in starved individuals falls to about a third of its level in fed animals, and insulin which fosters glucose use, drops to one-tenth (Smith et al. 1983:542).
Thus, marrow-fat percentage should be viewed not so much in terms of a fat indicator, but as an indicator of fat, muscle, and energy depletion, and any level below the threshold indicates an animal in marginal condition. While it certainly is true that some individuals do not actually die until their marrow fat is almost depleted, loss of vigor and vitality is a matter of degree rather than an all-or-none phenomenon. Additional stressors such as fighting, plowing through snow, or being chased by wolves probably would raise the marrow-fat threshold at which individuals in marginal condition would perish. This relationship could explain Ballard's et al. (1987) starved moose with a mean of 52% marrow fat.
Given the above considerations, we believe that most of our wolf-killed moose and caribou were in poor condition. Because such individuals would have lost considerable muscle mass as well as fat, these animals would have had little energy left to withstand chases by wolves.
The marrow fat content of Denali wolf-killed prey was consistently low despite relatively low snow depths in some years. There seemed to be no relationship between percentage marrow fat in our wolf kills and the snowfall, except that marrow fat of our moose kills was lowest during the winter of deepest snow (Table 1, Table 4).
The preponderance of low marrow fat despite low snow depth during three of
the seven years of the study indicates that the unharvested Denali prey herds
must include a certain proportion of individuals that are unable to secure
sufficient food, even under average environmental conditions. Our data indicate
that such individuals are among the oldest, youngest, and the bulls during the
rut. This situation probably is typical of natural ungulate populations
unharvested by humans. Wolves could depend on such vulnerable members of natural
prey populations, along with the young which are generally more vulnerable, to
sustain their own numbers during most years. When snowfall, or other weather
factors become extreme and increase prey vulnerability, wolf populations can
increase (Mech 1977b, Peterson 1977) to make use of the sudden increase in
resources, such as the caribou in our study.
This study was funded primarily by the U.S. National Park Service, Natural
Resources Preservation Program. Denali National Park, the U.S. Fish and Wildlife
Service, and the U.S.D.A. North Central Forest Experiment Station also
contributed to the project.
[MFS note: For the extensive list of materials cited, please see the original
Please send mail to
firstname.lastname@example.org with questions or comments about this web site. Minnesotans For Sustainability
(MFS) is not
affiliated with any government body, private, or corporate entity.
Copyright © 2002 Minnesotans For Sustainability