Early Warning Signals of Biodiversity Collapse

Abstract

Ecological regime shifts -- abrupt, self-reinforcing transitions from one stable ecosystem state to another, often ecologically degraded, state -- represent among the most dangerous consequences of biodiversity loss, because they can be triggered by gradual changes in driver intensity that produce catastrophic ecological change with little apparent warning before the tipping point is crossed. The theory of critical transitions predicts that approaching tipping points produce characteristic statistical signatures -- critical slowing down (CSD) -- in the temporal dynamics of ecosystem state variables: as a system approaches its tipping point, its recovery from small perturbations slows progressively (increasing autocorrelation; increasing variance; increasing skewness), providing early warning signals (EWS) that precede the actual transition. This study provides the most comprehensive empirical validation of biodiversity EWS yet conducted, testing CSD indicators in 247 long-term biodiversity time series from 84 ecosystems across 18 biome types that experienced documented regime shifts (comparison sites that did not shift used as controls). Lag-1 autocorrelation (AR1) and coefficient of variation (CV) of species richness time series both increased significantly before 74.7% of documented regime shifts (AR1 mean +0.38; CV mean +47.4% in the decade before transition vs. control sites; both p < 0.001). Lead time of EWS before the actual transition averaged 8.4 +- 3.4 years -- sufficient for conservation management intervention if monitoring systems detect the signal. Functional diversity variance (FDiv CV) outperformed species richness CV as an EWS (84.7% vs. 74.7% detection rate; p = 0.024), with a longer mean lead time (12.4 years vs. 8.4 years) -- confirming that functional diversity time series provide earlier and more sensitive early warning than species richness alone.