Effects of sampling effort, assemblage similarity, and habitat heterogeneity on estimates of species richness and relative abundance of stream fishes

We estimated the sampling effort required to accurately estimate species richness and to detect changes in catch-per-unit-effort (CPUE) in four Great Plains, USA, streams. The number of sampled reaches (i.e., <1 km) required to estimate stream-segment (i.e., 20-28 km) species richness decreased with increased sampled reach length (i.e., 10, 20, 40, or 60 mean stream widths, MSW), whereas total sampling effort decreased with a greater number of shorter sampled reaches. Collecting all species in a stream required all sampled reaches (i.e., 10) of a length equal to 40 or 60 MSW. The number of stream reaches sampled with lengths equal to 40 MSW required to detect a 50% change in CPUE of common species (i.e., total abundance > 1% of total catch) with beta = 0.80 ranged from 7 to 630 (mean = 99) and decreased with longer sampled reaches. A greater number of sampled reaches were needed to detect 90% of species richness and 25% changes in CPUE when Jaccard's similarity of samples of stream fish assemblages and habitat heterogeneity was lower within streams. Our results suggest that homogeneous stream segments require more sampled reaches to characterize fish assemblages and monitor trends in fish abundance.