Accuracy of Removal Electrofishing Estimates of Trout Abundance in Streams

Removal electrofishing is frequently used to estimate fish distribution and abundance in streams because it is simple and requires only one visit to a study site. However, because the removal method often overestimates capture efficiency and therefore underestimates fish abundance, some biologists have questioned its use in favor of less biased methods. In southern Idaho streams in the summers of 2006 and 2007, trout were marked and released in blocknetted reaches with backpack electrofishers using pulsed DC, and four-pass removals were conducted the following day. Removal electrofishing underestimated the abundance of trout ≥10 cm by 17, 22, and 25% for four-, three-, and two-pass removals, respectively, whereas for trout <10 cm, equivalent underestimates were 27, 27, and 37%. Removal estimates were biased in part because capture efficiency progressively decreased for fish ≥10 cm, from 58% in pass one to 37, 30, and 18% in passes two, three, and four, respectively; a similar decline was not as evident for fish <10 cm. Increases in channel complexity, in the form of boulder substrate, water depth, instream wood, and stream shading, resulted in higher bias in removal estimates. Linear regression models incorporating these and other variables explained between 47 and 65% of the variation in this bias. Visiting new sites in the summer of 2009 with a new field crew produced nearly identical amounts of removal estimate bias, but in some cases, predictive models did not accurately predict the bias we measured at individual sites. Our results suggest that multiple pass removal sampling in typical Rocky Mountain streams can produce population estimates that are minimally biased and therefore probably adequate for most basic fish population monitoring, especially if electrofisher settings and crew training balance the need to minimize injury with effective fish sampling.