Development and evaluation of sampling protocols for at-risk fishes in wadeable warmwater streams

Abstract

Natural resource managers make decisions based on the analyses of fish sample data; hence it is imperative to use unbiased data with low variance. The ability to capture fishes (capture efficiency) varies among methods and site- and species-characteristics. Failure to account for capture efficiency can bias sample data. High variance also influences data quality and can obfuscate important relationships and population trends. I evaluated two commonly-used fish sampling protocols in 31 streams the Upper Coosa Basin, Georgia, modeled fish capture efficiency, and evaluated the sources of sample variance. Capture efficiency was low and varied among species, between methods, and was related to stream habitat characteristics. I also estimated fish movement out of unblocked sample units during sampling and calculated unconditional capture efficiency. Failure to account for incomplete capture of fish introduces bias and variance in sampling data and may lead to poor inference, particularly for at-risk fishes.