Over the past 15 years the endangered eastern timber wolf (Canis lupus lycaon) has been slowly recolonizing northern Wisconsin and, more recently, upper Michigan, largely by dispersing from Minnesota (where it is listed as threatened). We have used geographic information systems (GISs) and spatial radiocollar data on recolonizing wolves in northern Wisconsin to assess the importance of factors in defining favorable wolf habitat. We built a multiple logistic regression model applied to the northern Great Lakes states to estimate the amount and spatial distribution of favorable wolf habitat at the regional landscape scale. Our results suggest that areas with high probability of favorable habitat are more extensive than previously estimated in the northern Great Lake States. Several variables were significant in comparing new pack areas in Wisconsin to nonpack areas, including land ownership class, land cover type, road density, human population, and spatial landscape indices such as fractal dimension (land cover patch boundary complexity), land cover type contagion, landscape diversity, and landscape dominance. Road density and fractal dimension were the most important predictor variables in the logistic regression models. The results indicate that public forest land and private industrial forest land are both important in managing for a broad‐ranging animal such as the wolf. Our data portray favorable habitat that is highly fragmented along development corridors in northern Wisconsin, which may be responsible for the slow growth of the wolf population. Upper Michigan, which is just beginning to be colonized by wolves, has very large, contiguous areas of likely habitat approaching the importance of those in northeastern Minnesota. If continuing development or wolf control restrict dispersing wolves from moving from Minnesota to Wisconsin, and Wisconsin habitat becomes more marginal through further fragmentation, Michigan has the potential to maintain a significant wolf population independent of Minnesota and serve as a source population for Wisconsin. However, a simple island/corridor model of wolf habitat in Wisconsin does not seem to apply. Wolves apparently move throughout the landscape, across many unfavorable areas, but establishment success is restricted to higher quality habitat. Source‐sink dynamics may be operating here, and they suggest that reduction of the Minnesota population in the near term may affect recovery in Wisconsin and Michigan. Our analysis is an example of use of long‐term monitoring data and large‐scale cross‐boundary regional analysis that must be done to solve complex spatial questions in resource management and conservation.

Many carnivore populations escaped extinction during the twentieth century as a result of legal protections, habitat restoration, and changes in public attitudes. However, encounters between carnivores, livestock, and humans are increasing in some areas, raising concerns about the costs of carnivore conservation. We present a method to predict sites of human-carnivore conflicts regionally, using as an example the mixed forest-agriculture landscapes of Wisconsin and Minnesota (U.S.A.). We used a matched-pair analysis of 17 landscape variables in a geographic information system to discriminate affected areas from unaffected areas at two spatial scales (townships and farms). Wolves (Canis lupus) selectively preyed on livestock in townships with high proportions of pasture and high densities of deer (Odocoileus virginianus) combined with low proportions of crop lands, coniferous forest, herbaceous wetlands, and open water. These variables plus road density and farm size also appeared to predict risk for individual farms when we considered Minnesota alone. In Wisconsin only, farm size, crop lands, and road density were associated with the risk of wolf attack on livestock. At the level of townships, we generated two state-wide maps to predict the extent and location of future predation on livestock. Our approach can be applied wherever spatial data are available on sites of conflict between wildlife and humans. Predicción de Conflicto Humano-Carnívoro: un Modelo Espacial Basado en 25 Años de Datos de Depredación de Ganado por Lobos Resumen: Muchas poblaciones de carnívoros lograron evitar la extinción durante el siglo veinte debido a protecciones legales, restauración de hábitat y cambios en las actitudes del público. Sin embargo, los encuentros entre carnívoros, ganado y humanos están incrementando en algunasáreas, lo cual es causa de preocupación en cuanto a los costos de la conservación de carnívoros. Presentamos un método para predecir los sitios de conflictos humanos -carnívoro a nivel regional, utilizando como ejemplo los paisajes mixtos de bosques-agricultura de Wisconsin y Minnesota (E. U. A.). Utilizamos un análisis apareado de 17 variables del paisaje en un sistema § §Current address: Living Landscapes Program, Wildlife Conservation Society, Treves et al. Predicting Human-Carnivore Conflict 115de información geográfica para discriminaráreas afectadas deáreas no afectadas a dos escalas espaciales (municipios y establecimientos). Los lobos (Canis lupus) depredaron selectivamente el ganado en municipios con proporciones altas de pasto y altas densidades de venado (Odocoileus virginianus) combinadas con proporciones bajas de terrenos agrícolas bosques de coníferas, humedales herbáceos y cuerpos de agua abiertos. Estas variables, junto con la densidad de caminos y el tamaño del establecimiento, permitieron además predecir el riesgo para establecimientos individuales cuando analizamos solamente el estado de Minnesota. En Wisconsin, solamente el tamaño del establecimiento, los terrenos agrícolas y la densidad de cam...

Managers of recovering wolf (Canis lupus) populations require knowledge regarding the potential impacts caused by the loss of territorial, breeding wolves when devising plans that aim to balance population goals with human concerns. Although ecologists have studied wolves extensively, we lack an understanding of this phenomenon as published records are sparse. Therefore, we pooled data (n = 134 cases) on 148 territorial breeding wolves (75 M and 73 F) from our research and published accounts to assess the impacts of breeder loss on wolf pup survival, reproduction, and territorial social groups. In 58 of 71 cases (84%), ≥1 pup survived, and the number or sex of remaining breeders (including multiple breeders) did not influence pup survival. Pups survived more frequently in groups of ≥6 wolves (90%) compared with smaller groups (68%). Auxiliary nonbreeders benefited pup survival, with pups surviving in 92% of cases where auxiliaries were present and 64% where they were absent. Logistic regression analysis indicated that the number of adult‐sized wolves remaining after breeder loss, along with pup age, had the greatest influence on pup survival. Territorial wolves reproduced the following season in 47% of cases, and a greater proportion reproduced where one breeder had to be replaced (56%) versus cases where both breeders had to be replaced (9%). Group size was greater for wolves that reproduced the following season compared with those that did not reproduce. Large recolonizing (>75 wolves) and saturated wolf populations had similar times to breeder replacement and next reproduction, which was about half that for small recolonizing (≤75 wolves) populations. We found inverse relationships between recolonizing population size and time to breeder replacement (r= —0.37) and time to next reproduction (r= —0.36). Time to breeder replacement correlated strongly with time to next reproduction (r=0.97). Wolf social groups dissolved and abandoned their territories subsequent to breeder loss in 38% of cases. Where groups dissolved, wolves reestablished territories in 53% of cases, and neighboring wolves usurped territories in an additional 21% of cases. Fewer groups dissolved where breeders remained (26%) versus cases where breeders were absent (85%). Group size after breeder loss was smaller where groups dissolved versus cases where groups did not dissolve. To minimize negative impacts, we recommend that managers of recolonizing wolf populations limit lethal control to solitary individuals or territorial pairs where possible, because selective removal of pack members can be difficult. When reproductive packs are to be managed, we recommend that managers only remove wolves from reproductive packs when pups are ≥6 months old and packs contain ≥6 members (including ≥3 ad‐sized wolves). Ideally, such packs should be close to neighboring packs and occur within larger (≥75 wolves) recolonizing populations.

Environmental hazards are distributed in nonrandom patterns; therefore, many biologists work to predict future hazard locations from the locations of past incidents. Predictive spatial models, or risk maps, promise early warning and targeted prevention of nonnative species invasion, disease spread, or wildlife damage. The prevention of hazards safeguards both humans and native biodiversity, especially in the case of conflicts with top predators. Top predators play essential ecological roles and maintain biodiversity, but they can also threaten human life and livelihood, which leads people to eradicate predator populations. In the present article, we present a risk map for gray wolf (Canis lupus) attacks on livestock in Wisconsin between 1999 and 2006 that correctly identified risk in 88% of subsequent attack sites from 2007 to 2009. More-open habitats farther from any forest and closer to wolf pack ranges were the riskiest for livestock. Prediction promotes prevention. We recommend that the next generation of risk mappers employ several criteria for model selection, validate model predictions against data not used in model construction before publication, and integrate predictors from organismal biology alongside human and environmental predictors.

Recovery of populations of wolves (Canis lupus) and other large, wideranging carnivores challenges conservation biologists and resource managers because these species are not highly habitat specific, move long distances, and require large home ranges to establish populations successfully. Often, it will be necessary to maintain viable populations of these species within mixed-use landscapes; even the largest parks and reserves are inadequate in area. Spatially delineating suitable habitat for large carnivores within mixed, managed landscapes is beneficial to assessing recovery potentials and managing animals to minimize human conflicts.Here, we test a predictive spatial model of gray wolf habitat suitability. The model is based on logistic regression analysis of regional landscape variables in the upper Midwest, United States, using radiotelemetry data collected on recolonizing wolves in northern Wisconsin since 1979. The model was originally derived from wolf packs radio-collared from 1979 to 1992 and a small test data set of seven packs. The model provided a 0.5 probability cut level that best classified the landscape into favorable (road density Ͻ 0.45 km/km 2 ) and unfavorable habitat (road density Ͼ 0.45 km/km 2 ) and was used to map favorable habitat with the northern Great Lake states of Wisconsin, Minnesota, and Michigan. Our purpose here is to provide a better validation test of the model predictions based on data from new packs colonizing northern Wisconsin from 1993 to 1997. In this test, the model correctly classified 18 of 23 newly established packs into favorable areas. We used compositional analysis to assess use of the original habitat probability classes by wolves in relation to habitat class availability. The overall rank of habitat preference classes (P, the percentage favorability from the original model), based on the new packs, was probability class 2 (P ϭ 75-94%) Ͼ 3 (P ϭ 50-74%) Ͼ 1 (P ϭ 95-100%) Ͼ 4 (P ϭ 25-49%) Ͼ 5 (P ϭ 10-24%) Ͼ 6 (P ϭ 0-9%). As more of the landscape becomes occupied by wolves, classes of lower probability than the 95% class, but above the favorability cut level, are slightly more favored. The 95% class is least abundant on the landscape and is usually associated with larger areas of classes 2 and 3. Wolves may continue to occupy areas of slightly lower habitat probability if adequate population source areas are present to offset the greater mortality in these lower quality areas. The model remains quite robust at predicting areas most likely to be occupied by wolves colonizing new areas based on generally available road network data. The model has also been applied to estimate the amount and spatial configuration of potential habitat in the northeastern United States.