Evaluation of subsampling efforts in estimating community indices and community structures

Even though subsampling can minimize the research efforts needed to deal with numerous specimens, the accuracy of bioassessment and community analysis can be affected according to subsampled individuals. In this study, the effects of subsampling on community indices, the parameters of the species abundance distributions (SADs) and ordination based on communities were evaluated using benthic macroinvertebrates from three different stream pollution levels: less, intermediately, and heavily polluted. The values of the community indices and the parameters of the SADs were clearly different according to the differences in the pollution levels, regardless of subsampling efforts. Detrended correspondence analysis was applied to characterize similarities between the subsamples with different individuals based on community indices and the parameters of the SADs. A self-organizing map was used to characterize the community similarities and properties of the subsamples, differentiating the subsamples from their original datasets. Finally, our results revealed that the effects of subsampling were reflected differently in different indices, parameters, and ordinations of communities at different pollution levels, suggesting that subsample size should be determined differently at different pollution levels depending on the objectives of the study. However, if only one subsample size is to be used for simplicity and to reduce the cost of aquatic ecosystem health monitoring and assessment, then we recommend at least 400 individuals as the subsample size. At least 800 individuals should be subsampled for biodiversity studies.