Comparison of Fish Communities at Random and Nonrandom Locations in a Sand‐Bed River

Abstract

AbstractSurveys based on nonrandom site selection, or convenience samples, are often a necessary part of large‐scale monitoring programs to help minimize costs. The reliability of convenience samples to inform managers about distributions or population status of imperiled species is questionable, however, because the samples may not be representative of the whole population. We compared fish community data from 20 nonrandom, long‐term monitoring sites for Rio Grande Silvery Minnow Hybognathus amarus with those from 20 randomly chosen samples collected during two surveys (one in summer, one in autumn) in the Rio Grande, New Mexico. We compared the species richness, community composition, and the catch per unit effort (CPUE). Fish species compositions, which were similar between both sets of summer and autumn surveys, were nearly identical in the autumn surveys. Similarly, we found consistent Rio Grande Silvery Minnow CPUE between surveys; summer random surveys estimated 0.32 fish/100 m2 sampled, whereas summer nonrandom surveys estimated 0.37 fish/100 m2 sampled. In autumn, both surveys showed a marked decline in Rio Grande Silvery Minnow; random surveys found 0.08 fish/100 m2 sampled (95% confidence interval 0.04–0.18), whereas the nonrandom surveys failed to collect any Rio Grande Silvery Minnow. Both surveys showed a reduction in species richness between summer and autumn with a corresponding increase in dominance by Red Shiner Cyprinella lutrensis and a decline in Rio Grande Silvery Minnow. We failed to find any meaningful differences in either fish community or Rio Grande Silvery Minnow CPUE between random and nonrandom sites, suggesting that the long‐term, nonrandom locations currently used to monitor the Rio Grande Silvery Minnow population are representative of the middle Rio Grande. We believe our results are applicable to many monitoring programs in systems with a homogeneous distribution of mesohabitats; nonetheless, we recommend that managers assess potential bias in monitoring programs based on convenience samples.