Additional Files captions

Table S1 Coverage of plant species at the different desertification stages.
Table S2 Primers and thermal profiles used for real-time PCR quantification of the different phylogenetic and functional genes.
Table S3 ANOSIM and ADONIS analyses of microbial beta-diversities between desertification stages.
Table S4 ANOSIM and ADONIS analyses of microbial beta-diversities between superficial and deep layers.
Table S5 Results of least-squares linear regressions analysing ecosystem functions and community metrics. The multifunctionality index calculated with all possible combinations of one, two and three functions.
Figure S1 Vertical variation in C, N, and P functions during desertification. Significant differences between desertification sites were based on a one-way ANOVA followed by an LSD test. Linear least-squares regression relationships between multifunctionality and soil depth were estimated. The bold lines denote the least-squares linear regressions across soil depth, with their 95% confidence intervals (grey-shaded areas). PD: potential desertification; MD: moderate desertification; SD: severe desertification.
Figure S2 Variation in dominant phyla of soil microbiomes during desertification development.
(A): Bacteria; (b): fungi; (C): archaea. Significant differences between desertification stages were tested by the Kruskal-Wallis H test. PD: potential desertification; MD: moderate desertification; SD: severe desertification.
Figure S3 Variation in dominant phyla of soil microbiomes in superficial and deep soils during desertification. Significant differences between desertification sites were based on a one-way ANOVA followed by an LSD test. PD: potential desertification; MD: moderate desertification; SD: severe desertification.
Figure S4 Relationships between the dominant phyla of soil microbiomes and multifunctionality at all sites. The red lines denote the least-squares linear regressions with their 95% confidence intervals (grey-shaded areas).
Figure S5 Relationships between the dominant phyla of soil microbiomes and multifunctionality at different desertification stages. The red lines denote the least-squares linear regressions with their 95% confidence intervals (grey-shaded areas). S : slope. PD: potential desertification; MD: moderate desertification; SD: severe desertification.
Figure S6 Variation in microbial alpha-diversity in superficial and deep soils during desertification. Significant differences between desertification sites were based on a one-way ANOVA followed by an LSD test. PD: potential desertification; MD: moderate desertification; SD: severe desertification.
Figure S7 Direct and indirect effects of soil depth, pH, electrical conductivity (EC), bulk density (BD), microbial species index or biodiversity index on multifunctionality. Structural equation models are shown for the all data sets. Numbers adjacent to arrows are indicative of the effect-size (bootstrap P value) of the relationship. R2 denotes the proportion of variance explained. Standardized total effects (direct plus indirect effects) derived from the structural equation models depicted above. The abundances of standardized dominant phyla of soil bacteria (Chloroflexi and Acidobacteria), fungi (Ascomycota and Basidiomycota), and archaea (Thaumarchaeota and Euryarchaeota) were then averaged to obtain an overall microbial species index. These standardized diversity indexes (Shannon and phylogenetic diversity) of soil bacteria, fungi, and archaea were then averaged to obtain an overall biodiversity index.
Figure S8 Changes of plant litter and belowground biomass during desertification development. Significant differences between desertification sites were based on a one-way ANOVA followed by an LSD test. The plant was dug up and the aboveground litter were clipped and dried to obtain the litter biomass, and the roots were washed with tap water and dried at 70 °C for 48 h to obtain belowground biomass.