Integrated demographic modeling and estimation of the central Georgia, USA, black bear population

Abstract

The central Georgia population (CGP) of black bears is considered to inhabit mostly forested land in and around 186 km2, and potentially an area of 1,200 km2, associated with the Ocmulgee River drainage system, and likely a core area of contiguous forest in the Oaky Woods and Ocmulgee Wildlife Management Areas (WMAs). We document the density, survival and reproduction, as well as genetic structure, of the CGP under the sampling protocol, over the duration of the study from 2003 to 2008. We describe a joint model of population abundance with three data structures (DNA hair snares, camera traps, and radiotelemetry) that incorporates genetic error from replicate genetic samples and a calibration sample of known individuals. The hierarchical joint Bayesian model incorporates Markov-Chain Monte Carlo (MCMC) methods with Gibbs, Metropolis-Hastings, and reversible jump Metropolis-Hastings sampling algorithms of posterior distributions. Median posterior abundance estimates within the WMA land over five seasons from 2004 to 2006 were: 2004 summer 213 (95% BCI: 144-354), 2004 fall 106 (95% BCI: 72-179), 2005 summer 184 (95% BCI: 137-266), 2005 fall 131 (95% BCI: 91-207), 2006 summer 192 (95% BCI: 143-280). Adult annual survival estimates were 0.861 (95% CI: 0.746-0.976) for females and 0.845 (95% CI: 0.754-0.937) for males. Reproduction rates were simulated from bootstrap simulations using mean birth interval and average number of cubs per female litter. Reproduction rates from the CGP only and the CGP combined with eastern black bear populations were 0.845 (95% CI; 0.843-0.847) and 1.139 (95% CI: 1.137-1.141), respectively. Population viability analyses using demographic parameters from the CGP and eastern black bear populations suggest that population growth is decreasing. The joint Bayesian hierarchical model also suggests that population growth is decreasing, since the Bayesian credible intervals of λ, the finite rate of population increase, included values above and below one. The λ from abundance models overlapped confidence intervals with λ from the population viability analyses, which suggest that conclusions based on increased harvest and population status are consistent with different data sources. Additional effort for the CGP should be focused on estimates of cub and sub-adult survival and reproduction.