Sample Size and Sensitivity in the Detection of Community Impact

Power is a measure of the sensitivity of a statistical test. Considerations of power are, or should be, a primary concern in designing monitoring programs. Increasing the number of samples increases power, but at an obvious increase in the cost of collecting and processing those samples. Here we consider the effects of community structure (specifically species diversity and density) on the power of simple two-sample tests to detect changes in several indices of community structure. Our results show that tests based on community-level diversity indices have higher power than those based on the abundance of single indicator species, especially if those species are rare. Different indices respond differently to changes in the overall species pool and the density of the community. Our approach could be used to derive quantitative estimates of sample size necessary to obtain given levels of power using only very coarse resolution biological data.