Hierarchical multi-population viability analysis.

Abstract

Population viability analysis (PVA) uses concepts from theoretical ecology to provide a powerful tool for quantitative estimates of population dynamics and extinction risks. However, conventional statistical PVA requires long-term data from every population of interest, whereas many species of concern exist in multiple isolated populations that are only monitored occasionally. We present a hierarchical multi-population viability analysis model that increases inference power from sparse data by sharing information among populations to assess extinction risks while accounting for incomplete detection and sampling biases with explicit observation and sampling sub-models. We present a case study in which we customized this model for historical population monitoring data (1985-2015) from federally threatened Lahontan cutthroat trout populations in the Great Basin, USA. Data were counts of fish captured during backpack electrofishing surveys from locations associated with 155 isolated populations. Some surveys (25%) included multi-pass removal sampling, which provided valuable information about capture efficiency. GIS and remote sensing were used to estimate August stream temperatures, peak flows, and riparian vegetation condition in each population each year. Field data were used to derive an annual index of nonnative trout densities. Results indicated that population growth rates were higher in colder streams and that nonnative trout reduced carrying capacities of native trout. Extinction risks increased with more environmental stochasticity and were also related to population extent, water temperatures, and nonnative densities. We developed a graphical user interface to interact with the fitted model results and to simulate future habitat scenarios and management actions to assess their influence on extinction risks in each population. Hierarchical multi-population viability analysis bridges the gap between site-level field observations and population-level processes, making effective use of existing datasets to support management decisions with robust estimates of population dynamics, extinction risks, and uncertainties.