A composite disturbance index for monitoring Sudanian dryland forests: structural and compositional indicators diverge under anthropogenic pressure

We developed and tested a field-based Composite Disturbance Index (CDI) for monitoring anthropogenic pressure in Sudan dryland forest, integrating stump density, partial extraction, grazing, agriculture, and fire into a single normalized score, and evaluated its performance against structural and compositional indicators of forest condition in Abu-Gadaf Natural Reserved Forest (AGNRF), Sudan. Using field plots combined with remote sensing covariates, we tested the CDI and its component metrics against tree diversity and forest structure using partial redundancy analysis, boosted regression trees, and AIC-based model comparison, keeping the aggregated index and its components in separate models to avoid structural collinearity. Anthropogenic predictors accounted for an adjusted 15.0% of variation in the multivariate diversity matrix ( p = 0.001), while the pure environmental fraction was indistinguishable from zero ( p = 0.939). Stump density was the strongest single correlate of compositional indicators, accounting for 60.7% of relative influence in boosted regression trees, whereas the composite index was the better predictor of size-weighted structural metrics. Critically, structural and compositional indicators diverged in their sensitivity to disturbance: basal area ( p = 0.002), stem density ( p = 0.0004) and aboveground biomass ( p = 0.007) all differed significantly among disturbance classes, while Shannon diversity ( p = 0.152) and species richness ( p = 0.147) did not, and aboveground biomass was predicted almost entirely by basal area regardless of disturbance intensity. No unimodal disturbance–diversity relationship was supported for any indicator. This divergence means that monitoring frameworks built on a single indicator class risk an incomplete read of forest condition: structural indicators central to carbon-focused reporting (e.g. UNFCCC/REDD+) responded sharply to the same disturbance gradient that biodiversity-focused indices (e.g. IPBES) did not detect. We recommend basal area and stump density as complementary, cost-effective indicators for dryland forest monitoring, and argue that safeguard frameworks pairing structural and compositional metrics are needed to avoid systematically overlooking degradation in reserves like AGNRF.