Trajectory of the Yellowstone Grizzly Bear Population under Alternative Survival Rates

The grizzly bear population inhabiting the GYE is of national and international interest. Although this population has increased in size and extent in recent years (Eberhardt et al. 1994, Eberhardt 1995, Boyce et al. 2001, Schwartz et al. 2002), isolation from other grizzly bear populations and continuing human development along its geographic margins justify continued concern about its future. Since the adoption of the federal recovery plan for grizzly bears in the United States (U.S. Fish and Wildlife Service 1993), mortality of grizzlies in the GYE has been monitored and a standard for acceptable mortality limit established. One important component of the limits of acceptable mortality is an estimate of the maximum human-caused mortality sustainable by a grizzly bear population (Harris 1986). This level was generated for a generic bear population, but recent information specific to the GYE population now allows for improvements to this estimate. Here, we use data from the period 1983–2002 (Haroldson et al. 2006, Schwartz et al. 2006a,c) as the basis for deterministic calculations and short-term stochastic projections of the GYE grizzly bear population under a range of survival rates for independent females (i.e., those no longer under the care of their mothers) that might apply in the future. Our approach to stochastic simulations was to produce a series of basic projections using parsimonious interpretations of data from Schwartz et al. (2006a,c) and Haroldson et al. (2006). We faced a number of different ways to project populations and interpret results, and we considered them as alternatives explored through sensitivity analyses. In generating trajectories, we wished to estimate not only the expected (or most likely) outcome but also the probability of decline (because declines are possible even when expected k . 1). Thus, we emphasized appropriate treatment of yearly variability in vital rates. Although analyses by Schwartz et al. (2006a,c) and Haroldson et al. (2006) identified strongly supported environmental covariates, these failed to explain the full range of yearly variation in vital rates. A mechanistic model that simulated these environmental factors directly (and linked vital rates to them) would have yielded less yearly variation than was observed during 1983–2002. We therefore integrated all factors contributing to yearly variation (both identified and unknown) via our estimates of the true process variance (yearly variation of the population only, excluding sampling variation). Because our objective was to understand survival rates that minimized the risk that k would decline below 1.0, we focused on females. However, male mortality rates are relevant to more general conservation concerns, so we also examined the behavior of simulated populations under alternative male survival schedules. We claim no ability to predict future reproductive or survival rates as environmental or management factors change. We can, however, use our knowledge of patterns in vital rates from 1983 to 2002 to understand population trajectories associated with a range of plausible future vital rates.