Depth- and Time-resolved Estimates of Soil Organic Carbon Change under Agroforestry: A Global Meta-analysis

Agroforestry (AF) is widely promoted as a pathway to rebuild soil organic carbon (SOC) while enhancing ecosystem resilience. Global estimates, however, remain heterogeneous due to variation in baselines, system archetypes, climate, and measurement depth. We synthesize two complementary datasets—the literature-derived database of Beillouin et al. and paired field observations from Cardinael et al.—to generate depth‑ and time‑resolved estimates of SOC change following transitions into AF. Observations were harmonized to 0-30 cm and 0-100 cm. Effects are expressed as log response ratios (lnRR = ln [SOC_AF/SOC_ref]), percent change = [exp(lnRR) − 1]×100, and absolute difference ΔSOC = SOC_AF − SOC_ref (Mg C ha⁻¹). For Dataset 1 (D1), unweighted summaries were used because study‑level variances are inconsistently reported; D1 subsets with variances were analyzed using random‑effects meta‑analysis. Age trajectories for Dataset 2 (D2) were modeled with cubic B‑splines (df = 4; knots at empirical quantiles) and mixed‑effects regressions with study/site random intercepts to avoid pseudo‑replication. Across the assembled evidence, AF indicates a mean SOC increase of 19% [95% CI: 14-24%] at 0–30 cm and +11 Mg C ha⁻¹ [95% CI: 7-15] at 0-100 cm, with larger effects on prior cropland and in humid tropical multistrata systems. Detectable gains typically emerge after ~7-10 years and extend below 30 cm in mature systems. These results estimate durable SOC recovery when AF is implemented on degraded croplands, while emphasizing context dependence and remaining uncertainty in subsoil horizons and temperate regions.