Population estimation from single-pass electrofishing in small coldwater streams in Michigan

Objective The Michigan Department of Natural Resources uses standardized electrofishing and capture–recapture methods to estimate the population of salmonids in small coldwater streams. These methods require 2 d of effort per site, which limits the number of sites that can be sampled per year. In this paper, I explore how estimates of capture probability from sites with full capture–recapture data can be used to allow for population estimates at sites with only a single pass. Methods I used simulation modeling to evaluate the performance of four different approaches using Chapman estimates of population size from capture–recapture sampling to calibrate catches at single-pass sites. These approaches include computing the mean capture probability across all calibration sites, two regression approaches, and a ratio estimator approach. I also compared results from calibrated single-pass estimates to Chapman estimates in 26 streams. Results Performance of these methods depends on the population size, the capture probability, and the process variation in capture probability among sites. Each of these approaches has unique attributes such that no single method is generally preferable. In situations where capture probability is low (i.e., 0.2), single-pass estimates perform better than independent estimates using a Chapman estimator. At population sizes and process variation typical of data from Michigan coldwater streams, the Chapman estimator performs better than single-pass estimates but comes at double the cost of sampling effort. Conclusions Estimates from a single pass generally parallel trends from the Chapman estimator, indicating that single-pass sampling may be used to evaluate population trends in most cases at a substantially reduced cost.