Distinct associations of rhizoid-sphere and endophytic bacteria with dominant moss traits along an elevational gradient on Segrila Mountain

Microbial communities associated with alpine mosses remain poorly characterized in southeastern Tibet, particularly the distinction between bacteria at the rhizoid-soil interface and those inside moss tissues. We used 16S rRNA gene amplicon sequencing to characterize 36 rhizoid-sphere and 12 endophytic samples from three dominant mosses ( Actinothuidium hookeri , Ptilium crista-castrensis , and Dicranum fuscescens ) across four elevations (3,700–4,300 m) on Segrila Mountain. We combined community analysis, SPIEC-EASI networks, random forests, and path modeling to examine how habitat, elevation, soil properties, and bacterial community attributes were related to moss traits. The endophytic network showed higher clustering (0.665) and modularity (0.84) and a larger proportion of positive partial correlations (94.71%) than rhizoid-sphere’s (0.305, 0.79, and 83.48%). Elevation uniquely explained 25.9% of the variation in rhizoid-sphere bacterial composition ( p = 0.0001), compared with 7.9% for moss species ( p = 0.0093), while neither factor significantly explained endophytic composition (both p > 0.05). Soil properties explained rhizoid-sphere community variation (adjusted R 2 = 0.199, p = 0.001), whereas the endophytic model was nonsignificant (adjusted R 2 = 0.029, p = 0.239). Random-forest models most strongly predicted moss tissue K (LOOCV R 2 = 0.58) and N (LOOCV R 2 = 0.48). In path modeling, rhizoid-sphere Shannon diversity was positively associated with morphology ( β = 0.500, p = 0.0005) and negatively associated with nutrient ( β = −0.263, p = 0.0036). These findings highlight the rhizoid-sphere as an interface linking elevation, soil conditions, bacterial community variation, and moss traits.