Estimating abundances of age-0 rainbow trout by mark-recapture in a medium-sized river

We developed and evaluated a sampling methodology to obtain mark-recapture data to estimate abundances of age-0 rainbow trout Oncorhynchus mykiss in 70 125-m-wide reaches of the Henrys Fork of the Snake River, Idaho. Sampling by electrofishing, was concentrated in sample areas that were 100 m long and extended from bank to bank: these areas were electrofished 3-5 times within periods of 3-17 d. Adjacent 50-m-long areas upstream and downstream were sampled to quantify movements out of the 100-m sampleareas. We evaluated assumptions-closed population and equal catchability-using the field data, and we used simulation to identify the most appropriate abundance estimator for sparse data. Both closed and open population abundance estimators were evaluated. Most trout (84%) were recaptured in the area where they were marked. but about 10% had moved downstream and about 61% were recaptured upstream, Multistrata model analyses confirmed that apparent mortality rates. and hence movement rates. were low. The Chao M, estimator, which assumes that capture probabilities vary with capture occasion. performed best for simulated closed populations bias was minimal and interval coverage was near or at the nominal level. This estimator was also robust to minor violations of the closure assumption, performance was better for larger closure violations when capture probabilities were smaller. Application of the Chao M, estimator to our field data resulted in a median capture probability of 0.036, a median capture efficiency of 16.7%, and a median recapture rate of 5.4%. Average abundance estimates in the sample areas provided indices of abundance and extrapolated estimates provided total abundance estimates for river sections 1-4 kin long, Small capture probabilities and large confidence intervals made it possible to detect only relatively large changes in abundance, but this level of discrimination was sufficient to satisfy management needs.