Widespread leaf phenology change detected by the leaf-age-dependent leaf area index (Lad-LAI) in pantropical evergreen broadleaf forests

Abstract Leaf phenology in tropical evergreen broadleaf forests (EBFs) often exhibits weak seasonality, making it challenging to detect the start of season (SOS) and end of season (EOS) from satellite-based vegetation indices particularly challenging. Leaf age structure describes the seasonal variability in canopy leaves of different age cohorts (young, mature, old) and for characterizing leaf phenology in tropical EBFs. Here, we built a methodological framework to produce a 0.25° leaf phenology dataset for tropical EBFs across 30°S–30°N using a leaf-age-dependent leaf area index product (Lad-LAI) for 2001–2023. SOS and EOS from the new dataset showed strong agreement with PhenoCam-derived metrics and good correlations with EVI-, NIRv-, and kNDVI-derived metrics. Across continents, the Amazon exhibited relatively later SOS dates, mainly during DOY 210–330, and earlier EOS dates, mainly during DOY 60–210. Conversely, the Congo and tropical Asia were characterized by relatively smaller SOS and larger EOS. Notably, SOS increases while EOS decreases progressively from northern to southern latitudes, showing contrasting latitudinal patterns. From 2001 to 2023, tropical EBFs experienced a widespread advance in SOS, which was most pronounced in the Congo region, with an average rate of −1.39 days yr−1. In contrast, EBFs mostly experienced a delay in EOS, which was strongest in tropical Asia, with a mean delay rate of 2.12 days yr−1. Consequently, the growing season (EOS minus SOS) increased significantly across all three continents. Overall, our leaf-age-dependent phenology dataset provides fundamental data for leaf phenology monitoring, carbon-cycle modeling, and climate-change impact predictions in tropical EBFs.