Among‐Pass, Interregional, and Single‐ versus Multiple‐Season Comparisons of Detection Probabilities of Stream Fishes

Abstract

AbstractUnequal detection of stream fishes must be accounted for when estimating assemblage composition. Detection probabilities (p) may differ among electrofishing passes, regions, and methods of estimation. We used data sets collected from (1) the middle New River basin, Virginia, using three‐pass electrofishing; (2) the upper Wabash River system of northern Indiana, using the U.S. Geological Survey (USGS) National Water Quality Assessment (NAWQA) protocol; and (3) the NAWQA database, collected throughout the United States. We tested for among‐pass differences in p using the New River data set. To test for interregional differences, we compared average detection probability (pavg) between the New and Wabash river data sets. We compared single‐ and multiple‐season estimates using the New River and NAWQA data sets. We found no differences among pass‐specific p for 97% of the species in the New River data set. No significant differences in pavg were found between the New and Wabash river systems for 73% of the species shared by the two data sets. Single‐season estimates of pavg did not differ from multiple‐season estimates of p for 86% of the species found in both data sets. Two of the three species for which single‐ and multiple‐season estimates of p differed also had interregional differences. Differences in pass‐specific detection probabilities may be attributable to decreased abundance or behavioral changes in subsequent passes. Interregional differences may be attributable to dissimilarities in abundance, habitat, or sampling protocols between the two systems. Similar factors may have affected single‐ versus multiple‐season comparisons. Our results show that p does not significantly differ among electrofishing passes in three‐pass samples for most species and that pavg from single‐season data is comparable to p from multiple‐season data sets. More research is needed to determine whether differences in p across regions are common. Our approach provides a relatively quick and reliable method for estimating the detection probabilities of stream fishes and will benefit managers who seek to attain accurate estimates of stream fish assemblage composition while making the best use of their time and resources.