Introduction
Study Area
Methods
Results
Discussion
Management Implications

Introduction

Wolf numbers in Minnesota, Wisconsin, and Michigan have exceeded the criteria for recovery and removal from the federal list of endangered species in those states (Michigan Department of Natural Resources 1997, Wisconsin Department of Natural Resources 1998, Berg and Benson 1999). After delisting, each state will regain management responsibility with temporary federal oversight. One biological and political issue important to the design of a sound wolf management plan is the question to what extent wolves affect deer hunting, for deer are the primary prey of wolves in all 3 states. Minnesota alone hosts some 400,000 deer hunters (Fuller 1990).

Many aspects of wolf-deer interactions have been studied (Stenlund 1955; Pimlot et al. 1969; Mech and Frenzel 1971; Kolenosky 1972; Potvin et al. 1988; Mech et al. 1991). However, the resulting information is only partly relevant to the question of wolf effects on deer hunting. The latter subject has had little scientific attention. In the Superior National Forest (SNF) of northeastern Minnesota, Stenlund (1955) concluded that wolves reduced deer browsing pressure, and thus in some ways, benefited the herd while at other times wolves competed with hunters. In part of this region where deer habitat was poorest, wolves and severe winters extirpated the wintering deer herd during 1968-1974 and reduced deer numbers in the surrounding area (Mech and Karns 1977). This area included 3,000 km² of wilderness largely inaccessible during the hunting season. Because deer declined throughout northern Minnesota, the Minnesota Department of Natural Resources closed the deer hunting season for 1971 and implemented more restrictive antlerless deer hunting regulations in 1972 and 1973, and restricted the take to bucks only since 1974 (Mech and Karns 1977).

A few attempts have been made to numerically examine the interactions among wolves, deer, and hunters. For the area described above, a simple model using wolf and deer numbers predicted the deer demise (Mech and Karns 1977). A more complex model, utilizing data on hunter harvest and winter severity, indicated for the region around the void, that "without wolf predation the deer herd would have declined very little by 1976 but that with the known wolf densities the deer population would drop to less than 0.4 deer/km²" (Mech and Karns 1977:21). The actual density dropped to 0.3-0.7 deer/km² (Nelson and Mech 1986a), with wolves killing 20% of the legal bucks in the area and hunters taking 30% (Nelson and Mech 1986b).

For an area 130 km west of ours, Fuller (1989) modified a model by Keith (1983) to examine interactions among wolves, human hunters, and deer and showed graphically the minimal hypothetical effect of wolves on human harvesting of deer. The most direct study of wolf competition with human hunting was conducted in Quebec where wolf numbers were experimentally reduced by 40-71%; the authors concluded that "the harvest of bucks was not affected" (Potvin et al. 1992:1595).

These studies yielded certain insights into the effects of wolf predation on deer harvesting by humans. However, they also had limitations. The deer decline on SNF in the early 1970's was extreme, and the exact role of poor habitat and several severe winters was unknown. The Keith-Fuller models were hypothetical and based on assumptions that might not be valid. For example, wolf predation and hunting mortality were considered completely additive to other mortality factors. Furthermore, in Fuller's (1989,1990) study area, only 10% of the deer mortality was due to wolves, whereas some 77% was due to humans, so wolf predation was relatively light. In the Quebec wolf-removal experiment, wolves had repopulated the removal areas within 8 months, greatly confounding that study (Potvin et al. 1992).

Thus, additional information is needed on the question of wolf competition with human harvests of deer. One approach to the subject is to examine data on wolf numbers and deer harvests in an area where wolves and humans have both killed deer over a long enough period to include a good representation of weather conditions and their effects on deer numbers. We use such data from the central SNF of northeastern Minnesota from 1975 to 1997 to test the extent to which wolves might influence deer hunting there.

Study Area

Our wolf census area encompasses some 2,060 km² immediately east of Ely in the east-central SNF (48°N, 92°W) of Minnesota (Fig. 1). The topography varies from large stretches of swamps to rocky ridges, with elevations ranging from 325 to 700 m above sea level. Winter temperatures < -35°C are not unusual, and snow depths (usually from about mid-Nov through mid-Apr) generally range from 50 to 75 cm on the level. Temperatures in summer rarely exceed +35°C.

We used 2 sources of data for our analyses: direct aerial counts of individual packs in the wolf-census area, and buck-harvest statistics for parts of the census area and the area immediately to the west. We conducted the wolf census by aerially tracking and counting radio collared wolf packs from December through March each winter and aerially counting tracks of any noncollared packs in the census area (Mech 1977, 1986). For each pack, we considered the highest number of wolves (or tracks in nonradioed packs) seen as being the pack size for that winter. The total population for the census area was the sum of all of the packs living there. This approach does not include assessment of numbers of lone wolves. However, the fact that we used the highest figure for each pack greatly minimizes any inaccuracy caused by lone wolves because most lone wolves are individuals that recently dispersed from packs (Fritts and Mech 1981, Messier 1985, Fuller 1989).

To maintain the same size census area each year while individual wolf packs shifted their use of the area somewhat, we only counted the number of wolves proportionate to the percent of the census area that a given pack used that winter, based on radio tracking data. We subtracted the number of wolves killing primarily moose from our total wolf census to derive the number of wolves dependent on deer (Mech 1986; L. D. Mech, U.S. Geological Survey, unpublished data).

Information on buck harvest was obtained from the mandatory registration of bucks with the Minnesota Department of Natural Resources by hunters in various "kill blocks" in and adjacent to our wolf-census area (Lenarz 1997 and M.S. Lenarz, Minnesota Department of Natural Resources, personal communication). These kill blocks included an area east of Ely, an area around Isabella, and an area south of Ely (Fig. 1). There was a good relationship between the trends of the harvests in the latter 2 areas (r² = 0.60, P <0.001), but not between the harvest in the first area and in either of the other two. We did not use numbers of antlerless deer harvested because those numbers fluctuated with the number of permits granted.

No measure of hunting effort was available for our study area to test whether variable hunting pressure obscured effects of wolves. Nevertheless, we hypothesized that if wolves had a strong negative effect on number of bucks harvested, we should find an inverse relationship between wolf numbers in the deer-killing packs one winter and the buck harvest the following fall. Thus, we used simple linear regression to compare harvest statistics to wolf numbers (Statistix 4.1 1994).

So as not to overlook possible relationships that might support our hypothesis, we deliberately ran regressions on whatever combinations of our 2 variables we thought logical. This approach would assure that if we did not find significant relationships, that negative finding would tend to indicate either that wolves were having little effect or that variable hunting effort might be masking any wolf effect. We analyzed data from each pack and from our entire census area against harvest statistics in 2 kill blocks partly in the wolf-census area and harvest data from the zones immediately west of our wolf-census area (Fig. 1). We assumed that annual changes in estimates of wolf density represented changes in the surrounding area as well. We also examined relationships between buck harvest in each kill block and wolf populations in and near each of those blocks.

With our largest data set, we also examined the individual annual changes in wolf and deer numbers (Table 1) and examined plots for any lag between wolf numbers and buck harvest that might confound regression analyses (Fig. 2). Annual numbers of deer hunters fluctuated widely in our study area before 1988 and after 1995, but remained reasonably constant from 1988 to 1995 (M. S. Lenarz, Minnesota Department of Natural Resources, personal communication). Therefore, we conducted separate analyses for 1988 to 1995.

Results

From 1975 through 1997, we counted 4-6 wolf packs that were dependent on deer in our census area, and their numbers ranged from 2 to 14 per pack each winter (Mech 1986, and L. D. Mech, U.S. Geological Survey, unpublished data). Annual buck harvest varied from 28 to 97 for the Isabella kill block, and 5 to 111 for the Ely kill block (Table 1). Total buck harvest for the zones west of our wolf census area varied from 200 to 930 each year (Table 1).

We found no significant relationship between any of our individual wolf pack sizes and either the Isabella or Ely buck harvest over the entire 23-year period, even though some of the wolf packs inhabited those kill blocks. The total number of wolves from all packs showed a marginally significant (P = 0.08) inverse relationship (r² = 0.14) with the Isabella buck harvest (Table 2).

Upon inspecting the scatter plots of the regressions, we noticed an apparent outlier in one of the plots. Although we knew of no reason to remove the outlier from the analysis, we did so arbitrarily to see how much this maneuver would force the data to fit our hypothesis. The result was an increase to an r² of 0.22 for the total of the wolf packs on a deer economy versus the Isabella buck kill (Table 2).

Plotting annual total wolf numbers against Isabella buck harvest from 1975 to 1997 showed no lag effect (Fig. 2). In fact, from 1975 through 1984, the wolf population tracked the decreasing deer harvest but continued downward through 1991 after deer harvest increased. Wolf numbers then increased again.

Annual decreases or increases of >5 each of wolves and deer were inversely related in only 9 (41%) of the 22 years. Inverse relationships occurred in 7 other years but in those years the increase or decrease for one species was