Identification of Microbial Indicator Species Synchronously Responding to Arsenic Contamination and Nutrient Variation via Co-Occurrence Network

Traditional arsenic (As) risk assessments rely predominantly on total arsenic (TAs) concentrations and overlook nutrient-driven masking effects on As biotoxicity. Microbial biomarkers hold promise for As pollution diagnosis, yet existing indicators only respond to As stress and cannot disentangle nutrient confounding effects. This study adopted co-occurrence network analysis to screen new microbial indicators and compared their association patterns with those of conventional α-/β-diversity regarding synchronous responses to As contamination and nutrient variation across 10 sampling sites in ecologically restored Yangzong Lake. Results showed pronounced spatial heterogeneity: TAs accumulated in western riparian soils (up to 123.60 mg·kg−1) and northern water (up to 32.86 μg·L−1), whereas soil organic matter (SOM) and nutrients concentrated in southwest and partial western zones. Notably, bioavailable, easily exchangeable As (Asexc) was geochemically decoupled from SOM in western soils, despite co-occurring high TAs and SOM. Bacterial α-diversity showed no significant correlation with As, while separate CCA axes reflected either As or nutrient gradients. Co-occurrence networks identified candidate indicator taxa, i.e., Actinomyces sp. in surface water, positively correlated with TAs (r = 0.714) but negatively with TP (r = −0.844), as well as Luteitalea sp. and RB41 sp. in riparian soil, positively correlated with Asexc (r = 0.802~0.941) but negatively with SOM (r = −0.694~−0.627). These taxa outperformed conventional α-/β-diversity indices, offering promising candidates for joint assessment of As contamination and nutrient status during ecological restoration.