Aim: Tropical forests account for a quarter of the global carbon storage and a third of the terrestrial productivity. Few studies have teased apart the relative importance of environmental factors and forest attributes for ecosystem functioning, especially for the tropics. This study aims to relate

1. Forests account for a substantial part of the terrestrial biomass storage and productivity. To better understand forest productivity, we need to disentangle the processes underlying net biomass change. 2. We tested how above-ground net biomass change and its underlying biomass dynamics (biomass recruitment, growth and mortality) can be explained by four alternative and contested hypotheses; the soil fertility, biomass ratio, niche complementarity and vegetation quantity hypotheses. 3. Above-ground biomass dynamics were evaluated over a 5-year period in 200 permanent sample plots in 8 tropical dry forests in Brazil, and related to soil fertility, community-weighted mean (CWM) traits that are important for carbon storage and sequestration (wood density, specific leaf area, maximum stem diameter and deciduousness), species richness, functional diversity and initial stand biomass. 4. Initial stand biomass was the best predictor of all three processes of biomass dynamics, providing strong support for the vegetation quantity hypothesis. In these dry forests, the dominance of conservative species, rather than of acquisitive species, is associated with high biomass growth and storage, probably because their low specific leaf area and high wood density allow them to keep on functioning during drought stress. 5. Paradoxically, high soil fertility (Ca) led to low biomass productivity, probably because of nutrient imbalance. 6. In contrast to what is shown for controlled experiments, we found no support for niche complementarity (in terms of functional diversity or species richness) for forest productivity. Biomass storage was favoured by low-rather than high trait diversity, as most of the biomass is concentrated in species with large stem diameter and high wood density. 7. Synthesis. Biomass dynamics are mainly shaped by vegetation quantity, and then by vegetation quality, in line with the mass ratio hypothesis. Dry forests show different trait-productivity relationships than wet forests, as stands with 'slow' trait values are 'fast' in terms of productivity. Diversity matters, but in a different way than expected; high trait diversity does not enhance productivity, but instead, does low trait diversity enhance carbon storage.

No abstract