Sampling effects on the effectiveness of ecological indicators in detecting fishery-induced community changes

Effective ecological indicators (EI) should reflect changes to ecosystem status in a timely manner to guide fisheries management; however, the robustness of EIs in the face of sampling uncertainty is not well understood and sampling errors may result in delayed or even unhelpful actions for management. In this study, we use a size-spectrum model to evaluate the effectiveness of EIs in detecting fishery-induced ecosystem changes given various levels of sampling uncertainty. We demonstrate that there is a time-lag that exists between changes in fishing pressure and EI response. The selectivity of survey gears can strongly determine the level of EI responses within certain size ranges. EIs may lose statistical power once sampling errors exceed a certain level, implying that several decades of monitoring data may be needed to be sure of detecting even a large change. Multivariate methods can strengthen the statistical powers of EIs, but only when the level of sampling noises is low. This study suggests the need for considering the impact of sampling uncertainty on the use of EIs in fisheries management.