Estimating Sustainable Harvest in Wolverine Populations Using Logistic Regression

Abstract: Population viability analysis (PVA) is a common tool to evaluate population vulnerability. However, most techniques require reliable estimates of underlying population parameters, which are often difficult to obtain and PVA are, therefore, best used in a qualitative context. Logistic regression is a powerful alternative to traditional PVA methods but has received surprisingly limited attention. Logistic regression fits regression equations to binary output from PVA models at a specific point in time to predict probability of a binary response over a range of parameter values. We used logistic regression on output from stochastic population models to evaluate the relative importance of demographic parameters for wolverine (Gulo gulo) populations and to estimate sustainable harvest in a wolverine population in Alaska. Our analysis indicated that adult survival is the most important demographic parameter to reliably estimate in wolverine populations because it had a greater effect on population persistence than did both fecundity and subadult survival. In accordance with this, harvest rate had a greater effect on population persistence than did any of the other harvest- and migration-related variables we tested. Furthermore, a high proportion of harvested females strengthened the effect of harvest. Hypothetical wolverine populations suffered high probabilities of both extinction and population decline over a range of realistic population sizes and harvest regimes. We suggest that harvested wolverine populations must be regarded as sink populations and that source populations in combination with sufficient dispersal corridors must be secured for any wolverine harvest to be sustainable.