Rapid assessment and extinction prediction using stochastic modeling of the endangered amargosa vole

Abstract

The Amargosa vole, Microtus californicus scirpensis, is an endangered microtine rodent obligately found in marshes near the Amargosa River, Mojave Desert in California. Very few data to inform modeling and adaptive management. If interventions are postponed until data are available, the vole could go extinct in the interim, making a more flexible modeling approach imperative. The voles face threats from environmental and demographic stochasticity, Allee effects, inbreeding, genetic drift, intense predation, and disease. The modeling approach used here is based on diffusion methods for time series of population size constrained by a carrying capacity, focusing on environmental stochasticity and the probability that the variance in population growth could allow the population to encounter the lower “absorbing” boundary and go extinct. We parameterized the model with Amargosa vole data that stand as Bayesian “priors” for carrying capacity, until more data can be obtained and allow us to refine a more accurate estimate. There are no multiple-year time series data or data for most demographic characteristics of the Amargosa vole, forcing us to look to California vole time series as a Bayesian prior. Our analysis indicated expected 20-24 years to extinction and 4-5% probability of extinction in one year due to environmental stochasticity: the real time could even be shorter if there is significant demographic stochasticity. Implementation of management based on best available modeling will be crucial to avert this risk. This modeling approach also has merit for other species in urgent need of management even in the face of early projects lacking mature data sets.