Assessment and prediction of electroshocked-induced injury in North American fishes

Abstract

Electrofishing has served as an efficient method for scientific sampling of freshwater fishes since the mid-1900s, but it has become apparent since the 1990's that electroshock can cause fish injury. Electroshock-induced fish injury (damage to hard or soft tissues), which is primarily manifested as vertebral fracture or hemorrhage (broken blood vessels) along the backbone, can be a critical determinant of fish survival. The ability to predict factors influencing fish injury rate (the proportion of. injured fish in a sample) would be very useful to biologists. To test the null hypothesis of no effect of electrical waveform (W), voltage gradient (E), and fish size (S) on injury rate, I conducted controlled electroshock experiments on chinook salmon Oncorhynchus tshawytscha, rainbow trout O. mykiss, channel catfish Ictalurus punctatus, largemouth bass Micropterus salmoides, bluegill Lepomis macrochirus, and hybrid striped bass Morone saxatalis x M. chrysops. Data collected included electrical stimulus, fish behavioral response (R), length (L) and weight (W), and injury status (present/absent). Vertebral injury was determined using radiography, and hemorrhage by bilateral filleting. My model selection criteria, which was based on Akaike's Information Criterion (AIC), indicated that risks for both types of injury in chinook salmon and channel catfish were best represented by the (W, E, S) model, the (W, S) model for both types of injury in rainbow trout, the (W, E) model for hemorrhage and the (W, E, S) model for vertebral injury in largemouth bass, the (W) model for both injury types in hybrid striped bass, and, that risk for injury in bluegill injury was best described by the null model (no effect of W, E, S). A mechanistic model relating electrical stimulus, the force of contraction, and the resistance to contraction to electroshock-induced injury, using (R) as a surrogate for electrical stimulus, (L) as a surrogate for force of contraction, and vertebral count (V) as a surrogate of resistance to injury, was explored. Application of the mechanistic model (R, L, V) to the pooled data set demonstrated a strong predictive relationship. This model offers guidance for the reduction and prevention of electroshock-induced injury for all species in all situations.