Assessing road-proximity patterns of biomass and modeled species richness in tropical dry forests

Tropical dry forests provide essential ecosystem services, including carbon storage and biodiversity maintenance, while road networks increasingly fragment these systems and are associated with complex edge-related and landscape-scale gradients. Despite their ecological relevance, most studies rely on simplified edge–interior contrasts or uniform road classifications, overlooking non-linear, distance-dependent responses and forest-type variation. This study evaluated associations between road proximity and aboveground biomass and modeled species richness across semi-evergreen, semi-deciduous, and deciduous forest, accounting for environmental gradients associated with landscape heterogeneity. A combined framework of covariate-balanced matching and generalized propensity score–based continuous-treatment modeling was applied across multiple road categories and distance thresholds up to 2 km. Environmental covariates, including topography, land cover, climate, and proximity to other infrastructure, were used to improve comparability between road-proximity and interior sites. Results showed lower aboveground biomass and modeled species richness near roads and gradual attenuation toward interior forests, with non-linear distance–response trajectories reflecting broader landscape-scale variability. Semi-evergreen forests exhibited the highest baseline values, while aboveground biomass and modeled species richness showed differences in response magnitude and variability across forest types. These patterns indicate that associations between road proximity, forest structure, and diversity are non-monotonic, scale-dependent, and modulated by forest type. By revealing the spatial structure of associations between road proximity, forest carbon, and biodiversity, this framework provides a transferable approach to support conservation planning, carbon accounting, and landscape management in fragmented tropical dry forests.