Taxonomic richness is one of the most important measures of biological diversity in ecological studies, including those with stream macroinvertebrates. However, it is impractical to measure the true richness of any site directly by sampling. Our objective was to evaluate the effect of sampling effort on estimates of macroinvertebrate family and Ephemeroptera, Plecoptera, and Trichoptera (EPT) genera richness at two scales: basin and stream site. In addition, we tried to determine which environmental factors at the site scale most influenced the amount of sampling effort needed. We sampled 39 sites in the Cerrado biome (neotropical savanna). In each site, we obtained 11 equidistant samples of the benthic assemblage and multiple physical habitat measurements. The observed basin-scale richness achieved a consistent estimation from Chao 1, Jack 1, and Jack 2 richness estimators. However, at the site scale, there was a constant increase in the observed number of taxa with increased number of samples. Models that best explained the slope of site-scale sampling curves (representing the necessity of greater sampling effort) included metrics that describe habitat heterogeneity, habitat structure, anthropogenic disturbance, and water quality, for both macroinvertebrate family and EPT genera richness. Our results demonstrate the importance of considering basin- and site-scale sampling effort in ecological surveys and that taxa accumulation curves and richness estimators are good tools for assessing sampling efficiency. The physical habitat explained a significant amount of the sampling effort needed. Therefore, future studies should explore the possible implications of physical habitat characteristics when developing sampling objectives, study designs, and calculating the needed sampling effort.
Keywords: Biodiversity; Physical habitat; Richness estimators; Sampling efficiency.