Assessing the effect of estimation error on population viability analysis: an example using the black-capped vireo

Abstract

Population viability analysis (PVA) is usually conducted assuming that values of factors that characterize both average conditions and natural variability (stochasticity) are error-free. However, those values are often estimates of true parameters and, therefore, have an associated estimation error. This error, also called uncertainty, arises from limitations of the methods used to estimate parameter values, such as sampling, measurement, and expert opinion error. Natural resource management decisions must be made in spite of incomplete information, and the impact of uncertainty when establishing specific management objectives must be assessed. We propose a strategy to reflect our current degree of knowledge about stochastic and average system conditions by accounting for error in parameter estimates of PVA models and assessing its effect on establishing endangered species conservation objectives. Using the computer simulation model vortex, we applied this strategy to the black-capped vireo ( Vireo atricapillus), an endangered neotropical migrant land bird species. We pursued two conservation goals in this study: probability of persistence and retention of genetic diversity of at least 95 and 90%, respectively, over the next 100 years. We evaluated two situations, one accounting for uncertainty, the other not. Achieving the conservation goals while explicitly accounting for estimate uncertainty required more demanding management objectives than that assuming no uncertainty. Under static carrying capacity, we had to increase fecundity 40% from 2.5 to 3.5 young fledged per female per year to meet the conservation goals. Explicitly incorporating estimation error into PVA allows us not only to assess the range of possible fates an endangered species may face, but also to establish measures necessary for meeting the conservation goals despite current levels of uncertainty.