Within-Community-Sampling Power Analysis to Detect Richness Change

Eden W. Tekwa, Jake Lawlor, Matt Lemay, Ryan R.E. Stanley, Nick Jeffery, Emily Rubidge, Jennifer Sunday

Reliable biodiversity monitoring requires understanding how sampling effort influences detectability of meaningful changes in species richness. Increased sampling within independent units has been shown to reduce measurement error, while sampled richness estimates are often subject to bias. However, robust methods for quantifying the relationship between sampling effort and the power to detect biodiversity change remain limited. Here we present a simulation-based power analysis framework that explicitly links sampling effort within communities to the ability to detect richness change between them. The approach uses empirical pilot data to simulate sampling and applies a dimensionless coverage metric to translate simulated effort into real sample sizes within communities. This produces direct relationships among power, effect size, and coverage, enabling estimation of the sampling required to achieve a specified probability of detecting richness change in the correct direction. We also provide a variant designed for before–after monitoring scenarios in which pilot data are available from a single community and the post-impact state is unknown. We demonstrate the application of this framework using a field-based environmental DNA biodiversity dataset. Additionally, we show that sample size recommendations rapidly converge when we apply the initial recommendation and reanalyze the next survey. The method is implemented in the R package BioDivPoweR, enabling broad application for designing efficient biodiversity monitoring programs and evaluating emerging sampling technologies.