Soil Fertility Differences and Bacterial and Fungal Community Variation Among Subtropical Forest Stands Dominated by Different Tree Species

Abstract

Plantations dominated by different tree species can differ in soil nutrient cycling and microbial community assembly, but corresponding bacterial and fungal patterns have not been characterized for mature plantation stands and a natural secondary forest in the Longmen Mountains of southwestern China. This study compared a natural secondary forest (SF) with four plantations, Magnolia officinalis (MO), Juglans regia (JR), Larix gmelinii (LG), and Cryptomeria japonica (CJ), in the Longmen Mountains of southwestern China, and investigated soil physicochemical properties, enzyme activities, bacterial and fungal community composition, co-occurrence networks, and predicted functional profiles in the 0–20 and 20–40 cm soil layers. The results showed that stand type was associated with total nitrogen (TN; p = 0.012), dissolved organic carbon (DOC; p < 0.001), dissolved organic nitrogen (DON; p < 0.001), and urease activity (p = 0.018), whereas pH differed between soil layers (p = 0.024) but not among stand types (p = 0.270). Relative to SF topsoil, DOC was 29.2% higher in JR and 36.1% higher in LG, while DON was 42.0% higher in JR and 51.4% higher in LG. TN and DON also showed stand type × soil layer interactions. None of the bacterial or fungal Chao1, Shannon, or Simpson indices showed significant stand type, soil layer, or interaction effects. In contrast, PERMANOVA indicated that bacterial and fungal community composition differed among the sampled stand types (p < 0.05), and fungal community composition also differed between the two soil layers (p < 0.05), whereas bacterial community composition did not. Paired partial dbRDA indicated that the integrated soil environmental gradients represented by the first three PCA axes were associated with bacterial community composition (R2 = 0.083, p = 0.012), whereas the corresponding association was not significant for fungi (R2 = 0.080, p = 0.371). Within the retained co-occurrence network analysis, MO had the most complex bacterial network, and JR had the most complex fungal network. PICRUSt2-predicted profiles were dominated by metabolism (38.6%–39.3%), but no bacterial KEGG Level 1 category or FUNGuild trophic-mode category showed a stand type, soil layer, or interaction effect after correction for multiple testing. These findings indicate that differences among the sampled stands were expressed more clearly in active carbon and nitrogen pools and microbial community composition than in alpha diversity or broad predicted functional profiles.