State-space analysis of power to detect regional brook trout population trends over time

Abstract

Threats to aquatic biodiversity are expressed at broad spatial scales, but identifying regional trends in abundance is challenging owing to variable sampling designs and temporal and spatial variation in abundance. We compiled a regional data set of brook trout (Salvelinus fontinalis) counts across their southern range representing 326 sites from eight states between 1982 and 2014 and conducted a statistical power analysis using Bayesian state-space models to evaluate the ability to detect temporal trends by characterizing posterior distributions with three approaches. A combination of monitoring periods, number of sites and electrofishing passes, decline magnitude, and different revisit patterns were tested. Power increased with monitoring periods and decline magnitude. Trends in adults were better detected than young-of-the-year fish, which showed greater interannual variation in abundance. The addition of weather covariates to account for the temporal variation increased power only slightly. Single- and three-pass electrofishing methods were similar in power. Finally, power was higher for sampling designs with more frequent revisits over the duration of the monitoring program. Our results provide guidance for broad-scale monitoring designs for temporal trend detection.