A growing number of studies have demonstrated that biodiversity is a strong and positive predictor of ecosystem temporal stability by simultaneously affecting multiple underlying mechanisms of stability i.e. dominance, asynchrony, and averaging effects. However, to date, no study has disentangled the relative role of these key mechanisms of stability in biodiversity experiments. We created a species richness gradient by mimicking a loss of rare species and assessed the role of species richness on community stability and, more importantly, quantified the relative role of three stabilizing mechanisms i.e. dominance (stabilization due to stable dominants compared to the rest of the species in the community), asynchrony (stabilization due to temporal asynchrony between species), and averaging effects (pure effect of diversity) on community stability across a species richness gradient. We found that extreme species loss negatively impacted community stability, but just three species were enough to stabilize biomass production to a level similar to highly diverse communities. However, the similar stability of communities resulted from differing contributions from each stability mechanism, depending on the community diversity. Since less abundant species were more temporally variable, species loss stabilized the populations of the remaining species. The loss of rare and subordinate species reduced the dominance and averaging effects, but increased the asynchrony effect. Hence, the asynchrony effect played a major role in the stability of species poor communities, while the averaging effect drove most of the stability of species rich communities. Overall, dominance played only a minor role, accounting for 5-15% of the stabilization, while asynchrony and averaging effects were dominating forces contributing to ~ 85-95% of the total stabilization.

Synthesis. This study highlights the importance of biodiversity and roles of dominant and rare species for long-term community stability and, for the first time, disentangles relative roles of dominance effect, asynchrony, and averaging effect on community stability in a real-world biodiversity experiment.