Dynamic interaction balance control of relative abundances buffers community instability

How ecological communities sustain high diversity and dynamic persistence despite complex species interactions remains a central question in biology. Theoretical models suggest that non-random network architectures promote stability under static, equilibrium conditions. However, tracking how species interactions dynamically adjust to regulate real, non-equilibrium communities has remained empirically elusive. Here we show that interaction networks dynamically reorganize to regulate species dominance and rarity and buffer instability. By applying empirical dynamic modelling to infer time‑resolved interaction networks in aquatic microbial communities, sampled three times per week across two year-long experiments, we introduce the interaction balance index—the log ratio of total positive to negative interaction strengths acting on each species at each sampling time. We find that subdominant species (relative abundance ≤ 0.1) frequently experience a net positive interaction balance, whereas dominant species tend to experience a net negative balance. The strength of this abundance-dependent regulation within a community, termed interaction balance control (IBC), buffers short-term instability, with stronger IBC dampening interspecific interaction-driven amplification of abundance fluctuations. These findings reveal IBC as an emergent mechanism linking dynamic network structure to community persistence, providing crucial empirical evidence for how complex ecosystems maintain dynamic stability in nature.