Niche differentiation increases along a continental temperature gradient and is linked to coexistence between Actinobacteria and Proteobacteria in paddy soils

Understanding how temperature relates to microbial competition and coexistence in managed paddy-soil bacterial communities is essential for predicting climate-change impacts on soil biodiversity, yet the underlying mechanisms remain poorly resolved. Here, we collected 429 soil samples from 39 paddy plots in 13 rice-producing regions along a north–south climatic transect in eastern China, with mean annual temperature (MAT) ranging from 1.5 to 23.8 °C. We integrated high-throughput 16S rRNA gene amplicon sequencing, GeoChip 5.0 functional gene profiling, co-occurrence network analysis, and modern coexistence theory (MCT) to characterize bacterial co-occurrence patterns and the genetic potential for antagonistic metabolite biosynthesis at the community level, and to derive MCT-informed, network-derived niche-difference (ND) and fitness-difference (FD) metrics for the focal Actinobacteria–Proteobacteria pair. Bacterial co-occurrence network complexity, indicated by connectance and clustering coefficient, was positively associated with MAT, while the diversity and richness of functional genes associated with antagonistic metabolite biosynthesis concurrently declined. Applying this proxy-based MCT framework to Actinobacteria and Proteobacteria, which together accounted for approximately 48% of bacterial abundance across plots, we found that 27 of 39 plots (69.2%) were classified within the MCT-predicted coexistence region, and the network-derived FD metric indicated a relative fitness advantage for Actinobacteria in 36 of 39 plots (92.3%). Spatial generalized least-squares analysis further supported a positive association between MAT and the network-derived niche-difference metric after accounting for spatial dependence and soil variation. Collectively, these results indicate that, for the focal Actinobacteria–Proteobacteria pair, higher MAT was associated with greater network-derived niche differentiation rather than fitness equalization, alongside reduced community-level genetic potential for antagonistic metabolite biosynthesis. Within this space-for-time design, these observational patterns generate testable hypotheses about how climatic variation relates to microbial coexistence in paddy soils.