Evaluating the predictive capacity of coral reef resilience assessments.

Resilience assessments, which rely on ecological indicators to predict future ecosystem resilience across a management jurisdiction, are used to prioritize limited conservation resources toward "resilient" locations with the best chance of surviving climate change. While resilience assessments have been widely applied to inform management, particularly for coral reefs, their precision and accuracy have rarely been validated. We used a timeseries of coral reef 3D models to (1) conduct multiple resilience assessments of fixed sites over time, and (2) track the resistance and recovery of those same sites over a decade. This allowed us to compare resilience predictions across assessments (precision) and test whether resilience predictions aligned with observed patterns of resistance and recovery (accuracy). We found that resilience assessments are capable of generating consistent resilience predictions over time, but in our case those predictions were not correlated with observed resistance or recovery dynamics. We recommend modelling resistance and recovery as distinct processes, and emphasize the importance of validating predictions with long-term monitoring. We caution against using resilience assessments in isolation to drive conservation decision making. Instead, spatial prioritization should be driven by local community input, with management of priority sites informed by functional indicators of resistance and recovery.