Accurate Estimation of Salmonid Abundance in Small Streams using Nighttime Removal Electrofishing: an Evaluation using Marked Fish

Abstract

Estimation of stream fish abundance using removal electrofishing is common and allows sampling of fish populations during a single site visit. However, recent evaluations have demonstrated that removal estimators can substantially underestimate fish abundance, raising concerns about using this method. We evaluated removal estimates of trout (family Salmonidae) abundance using night electrofishing in 200–300-m reaches of 8 Rocky Mountain streams and analyzed the data using new methods in Program MARK to account for potential sources of bias. The removal estimates were validated using populations of previously captured and marked resident fish. Overall, removal estimates were accurate estimates of the number of marked fish in study reaches (mean bias, −2.4% [<1 fish/reach]), and removal and mark–recapture estimates of total fish abundance also differed by less than 1 fish/reach on average. In general, capture probabilities were high (average = 97% over three passes), and removal estimates had narrow 95% confidence intervals that included the known number of marked fish for all but 1 of 17 estimates for three trout species. Capture probability increased with fish length and was greater on the first than on subsequent electrofishing passes. Capture probability also varied among sites, but this variation was not accounted for by any measured habitat variable. Finally, fish that had been previously marked were slightly more likely to be captured during electrofishing, even though the recovery periods for most fish exceeded 24 h. Simulations confirmed our finding that accounting for the heterogeneity in capture probability owing to individual fish size reduced the bias in removal estimates. Our results demonstrate that three-pass removal electrofishing can provide accurate estimates of fish abundance when field methods are designed to maximize capture probability and when modern analytical tools are used to account for the heterogeneity in capture probability.

Received June 18, 2010; accepted January 27, 2011

ACKNOWLEDGMENTS

We thank N. Cathcart, B. Dritz, M. Groce, A. Hansen, J. Mazzone, A. Romanyshyn, E. Saunders, V. Sednek, W. Stacy, and L. Young for assistance in the field. We thank W. Clements, J. Dunham, B. Kondratieff, P. Meiman, J. Peterson, G. Temple, and M. Young for helpful comments on the manuscript. Financial support for this research was provided by the Natural Resources Conservation Service (administered by K. Boyer and E. Hackett), U.S. Forest Service (D. Winters), Bureau of Land Management (J. Thompson), and Wyoming Game and Fish Department (D. Zafft), and graduate scholarships from Trout Unlimited (West Denver, Cutthroat, and Rocky Mountain Flycasters Chapters), the Robert J. Behnke graduate scholarship and Charles Bonham Memorial Fund (both of Colorado State University), and the Eugene Maughan Scholarship (Western Division of the American Fisheries Society).

Present address: Department of Watershed Sciences, Utah State University, 5210 Old Main Hill, Logan, Utah 84322, USA.