Evaluating sampling sufficiency in fish assemblage surveys: A similarity-based approach

The number and identity of fish species occurring at a site at a particular time provide basic information for assessing biological integrity, inferring fish assemblage - environment relationships, and determining biodiversity patterns. Conclusions are often dependent on how sufficiently species richness and composition of fish assemblages are characterized by sampling. The proportion of total species richness obtained in a sample is an explicit measure of sampling sufficiency. However, because total species richness (TSR_tru) at a site is often unknown, sampling sufficiency cannot be determined directly. To overcome this difficulty, we developed a new approach, which is based on a relationship between the proportion of TSR_tru or %TSR_tru and the similarity among replicate samples (autosimilarity). With autosimilarity measured with the Jaccard coefficient (JC), a simple relationship was established: %TSR_tru = 100JC. Fourteen sites where TSR_tru was reached or approached during sampling were selected from four surveys to validate this relationship. We used the approach to estimate the sample sizes required for 90, 95, and 100% TSR_tru, indicating that widely differing sampling efforts among sites are needed to obtain the same proportion of the local species pool. The results strongly support the use of the new approach in evaluating sampling sufficiency in stream and river fish surveys.