Abstract
Transportation networks threaten global forests, but prior assessments have been regional or limited to single metrics (e.g., forest cover). Here, we present a global analysis of multidimensional road effects on forests, using high-resolution remote sensing data and a Grid-wise Environmental Matching for Background Reference (GEM-BR) strategy. We detect 18.6% lower forest cover, 2.7 m shorter canopy height, 52.2 gC m(-2) yr(-1) reduced net primary productivity, and 23.0 patches per km(2) higher fragmentation within 1 km of roads compared to reference areas. Impacts extend up to 5 km with a clear distance decay effect, totaling 4.26 million km(2) of forest loss-equivalent to 10.7% of the 2020 global forest extent. The Global South (tropics accounting for 54.8%) faces severe, worsening degradation (2000-2020), while the Global North shows milder impacts, with partial recovery. Critically, 89% of grid cells exhibit conflicting long-term trends across metrics, highlighting the inadequacy of cover-only assessments. We further find that road-linked degradation is tightly coupled with local human activity, and that global protected areas have insufficient capacity to curb ongoing degradation. Differences in impacts among regions suggest that road-linked forest degradation is tied to governance choices-urging integrated transport-forest planning to balance development and conservation.