More diverse and structurally complex forests are less tightly coupled to ENSO forcing

Anomalies in the El Niño–Southern Oscillation (ENSO) generate extreme hydroclimatic events that shape and alter forest characteristics, yet their spatially explicit and time-lagged effects across different tropical forest types remain poorly quantified. We developed a 25-year, pixel-wise time-series modeling framework to assess how ENSO-related oceanic indicators influence vegetation dynamics along the Marañón Valley in northern Peru, a region with the highest ENSO-driven precipitation amplitudes globally. For each 1-km pixel, we fitted an independent model relating monthly mean Enhanced Vegetation Index (EVI) to four oceanic indicators (Niño 1+2 SST, Trans-Niño Index (TNI), and Multivariate ENSO Index (MEI)) at lags of 0–6-months. We then compare how model predictive power (R 2 ) and significant estimates from each variable change across forest types and forest characteristics using field data from the Peruvian National Forest Inventory. The modelling approach showed a higher R 2 in dry forests than in Amazonian forests and locally high values near major rivers. The TNI and Niño 1+2 SST exerted the strongest effects, with peak responses at 2-month and 6-month lags. In the dry forest tree species diversity and forest basal area correlated positively with mean and temporal variability of EVI but negatively with models R 2 , indicating that more diverse and structurally complex forests are less tightly coupled to ENSO forcing. These demonstrate that specific oceanic indicators, particularly TNI and Niño 1+2 can underpin operational early-warning tools for ecosystem monitoring along the dry–to-rainforest transition.