Elemental migration and enrichment are important processes influencing tea plant growth and the assembly of rhizosphere bacterial communities within the rock–soil–plant continuum. This study explores how soil parent materials (granite, quartz schist, and sericite schist) are potentially associated with these processes and their observed associations with the elemental composition of tea leaves. Exploratory statistical analyses revealed distinct, lithology-specific biogeochemical patterns that serve as a foundation for hypothesis generation. In granite soils, chlorite correlated with the mobility of Cr, Pb, Cu, Ni, Mg, and Na, coinciding with shifts in the relative abundances of Verrucomicrobia, Armatimonadetes, and Chloroflexi. In quartz schist, kaolinite exhibited notable correlations with the dynamics of Pb, Cr, Ni, Zn, and As, which were statistically linked to Planctomycetes, Proteobacteria, and Acidobacteria. Complex mineral–microbe interactions were observed in sericite schist soils, where clay minerals (e.g., chlorite, illite) were closely associated with the migration of multiple elements (Pb, K, Ca, Cd, As, Al, Fe, Zn), paralleling structural variations in communities of Actinobacteria, Planctomycetes, Chloroflexi, and Proteobacteria. Potassium (K), calcium (Ca), and manganese (Mn) showed bioaccumulation tendencies in tea leaves across all lithologies, with an enrichment capacity order of Ca > K > Mn > Mg > Na > Al. Exploratory Classification and Regression Tree (CART) analysis suggested that the migration of K, Ca, Cu, Zn, and Hg corresponded most closely with their soil concentrations. Manganese (Mn) exhibited a mineral-associated trend, with kaolinite content as a potential correlate, while cadmium (Cd) migration was statistically linked to the relative abundance of Armatimonadetes. These findings highlight potential candidate relationships between mineralogy, microbes, and elemental mobility rather than confirming causal mechanisms, emphasizing the need for further validation in larger or experimental datasets.
Li et al. (Sat,) studied this question.