Résumé
The Amazon Forest is undergoing rapid ecological shifts driven by intensifying drought, rising temperatures, and widespread anthropogenic disturbance. Yet the reorganization of vegetation functional strategies under climate stress remains poorly quantified at the biome scale. Here, we show that the temporal stability of canopy reflectance offers a sensitive remote proxy for sclerophylly-leaf toughness, a key indicator of conservative, drought-adaptive plant strategies. By integrating ground-based trait data (specific leaf area, SLA) from over 3000 trees across 448 plots in the Amazon-Cerrado savanna transition zone with highresolution remote sensing imagery, we demonstrate that lower SLA-a well-established proxy for conservative leaf strategies-is associated with reduced dry-season variability in the blue band spectral reflectance of vegetation. Extending the analysis across 130 plots in nine Amazonian countries using 40 years of harmonised remote sensing data, we find that dry-season reflectance variability has declined by ~34% (a drop of ~10 percentage points) since 1984, indicating a biome-wide shift toward greater drought tolerance. This trend is most pronounced in the southern and eastern Amazon and closely tracks rising climate stress, particularly increased temperature, evaporative demand, and water deficit. If these patterns persist, much of the southern and eastern Amazon could reach reflectance-stability levels comparable to transitional zones with the Cerrado savanna biome within the next three to four decades. Our results show signals of an early-stage forest functional transformation that could reduce forest productivity and carbon uptake, increase vulnerability to fire, and diminish biodiversity. These findings highlight regions where early signs of reduced forest resilience are emerging, underscoring the need for spatially targeted conservation.