The correlation between bacterial communities and physicochemical characteristics
PCoA showed that the PC 1 axis and PC 2 axis explained 70.19% and 15.96% the difference of bacterial communities in the samples, respectively (Fig. 3A). The bacterial communities in the S1, S2, S3, and S4 were clearly separated from those in the S5, S6, S7 and S8 at PC 1 axis. ANOSIM analysis also supported the PCoA result, suggesting that the bacterial communities in the S1, S2, S3 and S4 were significant (p <0.05) different from those in the S5, S6, S7 and S8 (Fig. 3B). The same result was obtained from Bray Curtis analysis, which showed that the distance of bacterial communities in the S1, S2, S3 and S4 sites were wider than those in the S5, S6, S7 and S8 sites.
RDA analysis showed that the sand physiochemical properties explained 80.25% (RDA1 explained 70.78% and RDA2 explained 9.47%) the variances of bacterial community at the phylum level (Fig. 4). The sand pH (RDA1 = 96.5%, r2= 0.38, p = 0.01) and P (RDA1 = 99.8 %, r2 = 0.38, p =0.019) significantly positively correlated with RDA1, and altitude (RDA1 = 100%, r2 = 0.50, p = 0.001) and Ca2+ (RDA1 = 97.2 %, r2 = 0.44,p =0.006) significantly negatively correlated with RDA1, indicating that these factors were vital to explain the variations of bacterial community structure. At the OTU level, CCA analysis showed that the similar correlation between the factors and the bacterial communities, namely, pH (r2 = 0.49, p =0.002) and P (r2 = 0.37, p =0.035) significantly positively correlated with CCA1, and altitude (r2 = 0.55, p = 0.001) and Ca2+ (r2= 0.48, p =0.003) significantly negatively correlated with CCA1, suggesting that these factors had a similar important effects on bacterial communities.
VPA analysis showed that both altitude and pH explained 42.7% of the variations bacterial community. Ca2+, altitude and pH explained 50.3% of the variations of the bacterial community. TOC, altitude, and pH explained 46.4% of the variations. EC, altitude, and pH explained 40.4% of the variations (Fig. S2).
The Spearman correlation analysis was evaluated the correlation between different phyla and each of the sand physiochemical properties (Fig. 5). The sand physiochemical properties were divided into two groups. Group 1 included EC, altitude and Ca2+. Group 2 included Cl-, WC, TOC, Cl, TC, Mg2+, TN, Na+, V, Zn, Zr, Y, Ti, Mn, pH, and P. Besides, the bacterial phyla were grouped into three clusters. Cluster 1 was significantly positively correlated with altitude and Ca2+ but negatively correlated with V, Zn, Zr, Y, Ti, Mn, pH, and P. Cluster 2 was significantly positively correlated with Ca2+, TOC, Cl, Mg2+, Na+, V, Zr, Y, Ti, Mn, and P, but negatively correlated with Cl-. Cluster 3 was significantly negatively correlated with EC, altitude, and Ca2+, but positively correlated with TC, TN, V, Zn, Zr, Y, Ti, Mn, pH, and P.