Taxonomic and structural diversity indices predict soil carbon storage better than functional diversity indices along a dieback intensity gradient in semi-arid oak forests

Key message The use of structural diversity indices can be efficient tools to increase understanding of changes in soil carbon storage as a key ecosystem function along the oak dieback intensity. Investigating the relationships between biodiversity and ecosystem functions within the mass‐ratio and niche complementary hypotheses is still a challenging issue in terrestrial ecology. These relationships have not been studied along a dieback gradient of disturbance in semi-arid forest ecosystems. To fill this gap, we investigated the relationships between a main ecosystem function -soil carbon storage- and various diversity indices along a dieback intensity gradient (no, low, moderate or high dieback intensity) in protected or intensively managed mixed oak forests in western Iran. We used different diversity approaches and calculated the functional divergence index (FD var ), community weighted mean trait values; CWM, taxonomic diversity indices (richness, Shannon–Wiener diversity and evenness) and structural diversity indices (MI: mingling index, HD: height differentiation, DD: diameter differentiation). Soil carbon storage was significantly influenced by the type of management, the intensity of dieback and their interactions and was higher in the protected areas (95.90 ± 4.62 ton ha −1 ) than in the intensively managed areas (76.52 ± 2.04 ton ha −1 ). It showed a humped-shaped pattern along the dieback intensity gradient in the protected areas, as it peaked at the low dieback intensity (122.47 ± 10.12 ton ha-1), indicating that soil function was maximized at a low disturbance level. Soil carbon storage was positively and significantly correlated with all structural and taxonomic diversity indices, except for evenness, but negatively with most functional composition indices (CWM of LNC: leaf nitrogen concentration, H: mean of woody species height and LDMC: leaf dry-matter content). It was best predicted by a structural index (Mingling index: R 2 = 0.214) followed by a taxonomic index (species richness, R 2 = 0.173) and a CWM index (CWM LDMC, R 2 = 0.158). Our results emphasize the role played by the diversity indices to predict ecosystem functions in contrasted management conditions and along a dieback gradient. They also provide evidence to support both the mass‐ratio and niche complementary hypotheses.