Northern Prairie Wildlife Research Center
|Figure 1. Locations of Minnesota farms suffering chronic depredations by wolves during 1989-1998 and farms not suffering losses, chosen as a matching sample.|
We chose to study chronic farms rather than those experiencing only occasional loss because chronic farms are more likely to have some characteristic that predisposes them to depredations. Farms that experience only occasional losses are more apt to be affected by random events, such as presence of a dispersing wolf passing through the area (Fritts 1982).
To assess which farms suffered chronic losses, we created a database from WS records of all verified wolf depredation complaints from 1989 to 1998. We ranked farms according to number of calendar years when they suffered verified losses. We defined chronic farms as those where WS personnel had verified at least one wolf depredation in each of 3 or more years during the 10-year period (Fritts et al. 1992).
We deemed 51 farms (4 sheep, 4 turkey, and 43 cattle) in 15 counties as chronic during 1989-1998. Because of the low number of sheep and turkey farms and the difficulty of finding a match for them, we considered only cattle farms. We used 41 of 43 cattle farms in the analysis as we were unable to interview owners of one chronic farm and unable to find a match for another. Though all had wolf losses during at least 3 years in 10, the history of wolf depredations on these farms varied considerably. Individual farms experienced up to 18 episodes of depredation during the study and had depredations during 8 years of the 10-year period.
Around each chronic farm, we attempted to locate other farms raising the same type of livestock (beef cattle or dairy cattle) where wolf depredation had not occurred (matching farms). To randomize our matching sample, we chose a cardinal direction from the depredated farm by throw of a die and first searched for matching farms in that direction within 8 km of the chronic farm. The principal method of locating matching farms was driving in the random cardinal direction looking for livestock, pasture areas, and hay storage. WS personnel, county extension agents, cattlemen's associations, and other farmers also were questioned as to the locations of potential matching farms.
If we did not find a farm without claimed wolf losses within 8 km in the initial compass quadrant, we extended the search in other directions, working clockwise from the initial random direction. In some cases, we needed to go up to 15 km from the chronic farm to locate non-problem farms to survey. Several farms were usually surveyed near each chronic problem farm until a suitable matching farm was found.
We avoided using as matching farms those with verified wolf problems that did not reach the level of chronic farms. If an operator claimed to have suffered wolf depredation, even if no losses had been verified in the last 10 years, we rejected that farm as a matching (non-depredated) farm and chose other matching farms.
We visited each of the chronic and matching cattle farms one to 4 times between July 1998 and January 1999 to survey the owner or manager in person. When this could not be done, we conducted telephone interviews (n = 15). Interviews covered location and size of the livestock operation, history of livestock raising and depredation problems, farm size, number of cattle, number of years raising cattle, amount of pasture bordered by brush or forests, longest distance of livestock from house, pasture characteristics, calving locations, number of times stock were checked each week, presence of carcass dump, and carcass disposal methods.
Besides the 41 farmers at chronic farms, we interviewed 145 farmers at matching farms and chose 41 matches that fit the criteria stated above. We then summarized the answers to the survey questions that might provide insight into factors predisposing livestock to wolf depredations.
A factor identified as possibly being important in predisposing certain farms to wolf depredations was leaving livestock carcasses where scavengers could use them (Fritts 1982, Fritts et al. 1992). As part of a separate survey involving use of rendering plants for carcass disposal, we requestioned farmers in our sample of matching farms about their carcass disposal methods. We attempted to phone each matching farm during 15 April to 2 May 1999.
The group of chronic farms we surveyed was essentially an entire population rather than a random sample. Therefore, to determine significant differences between measures derived for chronic farms versus measures for our sample of matching farms, we used the following approaches. We considered any average measure of the chronic population to differ significantly from that of the matching sample if the average for the chronic farms fell outside the 95% confidence limits of the average of the matching sample. To compare distributions of characteristics between our 2 types of farms, we used the chi-square test.
We created a GIS coverage of chronic farms and another of matching farms. Using ArcView (Environmental Systems Research Institute, Redlands, California, USA) GIS software, we created zones with radii of 1.6 km and 4.8 km around farms to examine surrounding habitat. We dropped one chronic farm and its match from the analysis of the 4.8-km radius because the radius extended out of Minnesota, where we had no habitat data.
For our habitat analysis we used a coverage assembled by the Minnesota Department of Natural Resources, Grand Rapids, which mapped 8 cover types (urban and rural development, cultivated land, hay-pasture-grassland, brush, forest, water, bog-marsh-fen, and mining). The source data were collected between 1987 and 1996 and were originally captured in 30-m (13 counties) and 90-m (2 counties) cells and then converted into a feature data source.
We then used ArcView to estimate percentage of each habitat type for chronic and matching farms within the 1.6-km and 4.8-km radii. The data for each kind of farm were pooled to give a single set of percentages of habitat for each kind. We hypothesized that if farms with chronic losses were surrounded by some specific cover or land-use type that predisposed them to wolf depredations, then the pooled data should differ from those for the matching farms in proportions of habitat types.
Because of the possible importance of carcass disposal as a predisposing factor and because improper carcass disposal is illegal, we attempted to cross-check reporting about this subject. For chronic farms, we asked WS personnel about their personal knowledge of carcass disposal at these farms and compared their replies with those obtained from direct interviews.
WS personnel had no personal knowledge of conditions on matching farms, however. Thus, as a cross-check for those farms, we compared replies about carcass disposal at matching farms with replies to a similar question asked of the same farms during the special telephone survey about rendering plants.