Boat-electrofishing transect location and flow levels: influence on riverine fish monitoring in non-wadeable habitats.

Fisheries monitoring can be improved by studying the influence of gear selectivity, sampling design, and habitat conditions. We used boat-electrofishing data to investigate how sample unit placement (shoreline and channel transects) and sampling conditions (low and high flow years) affect detection of fishes in a highly regulated Ontario (Canada) river system. Species detection histories associated with a spatially replicated sampling design was fit to a Bayesian hierarchical site occupancy model for 14 fishes. Habitat (transect location) had a significant effect on detection probabilities (p) for all species, with shoreline sampling more likely to detect species than channel. Sampling year had a significant effect on detectability of six species. The relative influence of habitat and sampling year varied among species. Detection probabilities based on combined shoreline and channel transect data across both years ranged from 0.09 to 0.48 and were positively correlated to species abundance. High detection probabilities and precise occupancy estimates were associated with sunfish and bass abundant in shoreline habitats. Small-bodied species closely associated with the riverbed or exhibiting schooling behavior tended to be poorly detected. Power to detect future changes in species distribution is expected to differ based on habitat sampled, with greater power associated with shoreline boat-electrofishing. Detecting small to moderate changes is not likely for most species. The addition of channel transects did not improve the effectiveness of shoreline sampling. Improvements to the current design could be realized by reallocating sampling effort from channel to shoreline and testing different gear in deep (> 3 m) habitats.