Abstract
We compared results of the hydroacoustic and netting methods of estimating guided and unguided fish passage and evaluated fish-guidance efficiency (FGE) of an extended submersible bar screen at John Day Dam on the Columbia River. Hydroacoustic counts of guided fish were significantly correlated with concurrent gatewell catches (r2 = 0.73; N = 39), as were hydroacoustic counts of unguided fish with fyke-net catches (r2 = 0.71; N = 39). However, hydroacoustic sampling significantly underestimated both guided and unguided fish passage relative to netting estimates. We could not explain the underestimates by modeling hydroacoustic detectability, and the distribution of fish passage across the intake width was not skewed away from transducer sampling volumes. Hydroacoustics provided relatively unbiased estimates of fish guidance efficiency (guided/(guided + unguided)) because of compensating errors in the numerator and denominator. The best correlation between net and hydroacoustic estimates of efficiency (r2 = 0.85; N = 40) had a slope of 0.91 when the intercept was set to zero. Precision of hydroacoustic estimates increased 50% and the r2 of the correlation line increased 19% when hydroacoustic sampling duration was extended from the typical netting duration of 1–2 h to 4 h. Further increases in hydroacoustic sampling duration from 5 to 9 h provided no significant improvement in correlations. Strong correlations between estimates of FGE derived from netting and hydroacoustic sampling are reassuring and useful because both methods have advantages that can be exploited to improve overall sampling effectiveness at a hydropower project. The derivation of a universally applicable relation between hydroacoustic and physical-capture estimates of fish passage is not possible given many potential deployment-dependent biases in estimates.