Despite the importance of agroforestry parkland systems for ecosystem and livelihood benefits, evidence on determinants of carbon storage in parklands remains scarce. Here, we assessed the direct and indirect influence of human management (selective harvesting of trees), abiotic factors (climate, topography, and soil) and multiple attributes of species diversity (taxonomic, functional, and structural) on aboveground carbon (AGC) stocks in 51 parklands in drylands of Benin. We used linear mixed-effects regressions and structural equation modeling to test the relative effects of these predictors on AGC stocks. We found that structural diversity (tree size diversity, HDBH) had the strongest (effect size β = 0.59, R2 = 54%) relationship with AGC stocks, followed by community-weighted mean of maximum height (CWMMAXH). Taxonomic diversity had no significant direct relationship with AGC stocks but influenced the latter indirectly through its negative effect on CWMMAXH, reflecting the impact of species selection by farmers. Elevation and soil total organic carbon content positively influenced AGC stocks both directly and indirectly via HDBH. No significant association was found between AGC stocks and tree harvesting factor. Our results suggest the mass ratio, niche complementarity and environmental favorability as underlying mechanisms of AGC storage in the parklands. Our findings also highlight the potential role of human-driven filtering of local species pool in regulating the effect of biodiversity on AGC storage in the parklands. We conclude that the promotion of AGC stocks in parklands is dependent on protecting tree regeneration in addition to enhancing tree size diversity and managing tall-stature trees.
© 2023. The Author(s).