Quantile-based sensitivity and risk thresholds reveal asymmetric responses of tropical forest productivity to individual and compound hot–dry extremes

Abstract

Gross primary productivity (GPP) regulates terrestrial carbon uptake and is strongly constrained by temperature and water availability. While increasing evidence shows that heat and drought can substantially suppress productivity, most studies rely on mean-based analyses and univariate frameworks, obscuring heterogeneous and compound impacts. Here we investigated how temperature and soil moisture anomalies influence the entire distribution of GPP in tropical deciduous and evergreen forests in Peninsular India. Using satellite-derived GPP and a quantile regression framework, we quantified productivity responses across multiple quantiles, thereby examining heterogeneous sensitivities beyond the mean. Further, we identified probabilistic thresholds of productivity loss under individual temperature and moisture stressors and evaluated how the joint occurrence of hot–dry events amplifies the risk of extreme GPP reductions. Results revealed differential responses across forest types, where seasonally dry deciduous forests exhibited responses that are highly heterogeneous across quantiles and strongly seasonally modulated compared to evergreen forests, highlighting strong plant functional type (PFT) contrasts among ecosystems. We found region- and PFT-specific thresholds, where seasonally dry deciduous forests exhibited higher probabilities of productivity loss under relatively small temperature and soil moisture anomalies, while evergreen forests showed greater resistance and required larger anomalies to trigger comparable declines. In addition, compound hot–dry conditions substantially increased the probability of extreme productivity losses compared to individual stressors alone. These results highlight the need to resolve distributional sensitivities and risks under individual and compound hot–dry conditions, to improve understanding of ecosystem responses to climate stressors and our capacity to anticipate vegetation responses in a warming and increasingly variable climate.