论文摘要：

1. Arid and semi‐arid grassland ecosystems cover about 15% of the global land surface and provide vital soil carbon (C) and nitrogen (N) sequestration. Although half of the soil C and N is stored in deep soils (below 30 cm), no regional‐scale study of microbial properties and their functions through the soil profile has been conducted in these drylands.
2. To explore the distribution and determinants of microbial properties and C and N mineralization rates through soil profile along aridity gradient at a regional scale, we investigated these variables for four soil layers (0–20, 20–40, 40–60 and 60–100 cm) in 132 plots on the Mongolia Plateau.
3. Soil microbial properties (biomass and bacteria:fungi ratio) and C and N mineralization rates decreased with increasing soil depth and aridity at the regional scale. Aridity‐induced declines in soil microbial properties mainly resulted from the negative effects of aridity on ANPP/root biomass and soil organic C (SOC) in the surface soil layers (0–20 and 20–40 cm) but from the direct and indirect (via SOC and soil C/N) negative effects of aridity in the deep soil layers (40–60 and 60–100 cm).
4. Aridity‐induced declines in soil C mineralization rates mainly resulted from the negative indirect effect of aridity on SOC and microbial properties in each soil layer, with weaker effects of SOC and stronger effects of soil microbes in the deep soil layers. Aridity‐induced declines in soil N mineralization rates mainly resulted from the negative indirect effect of aridity on SOC in the three soil layers above 60 cm and mainly resulted from the negative direct effect of aridity in the 60–100 cm soil layer.
5. Aridity via direct or indirect effects strongly determined the patterns of soil microbial properties and C and N mineralization throughout soil profiles on the Mongolian Plateau. These findings suggest that the increases in aridity are likely to induce changes in soil micro‐organisms and their associated functions across soil depths of semi‐arid grasslands, and future models should consider the dynamic interactions between substrates and microbial properties across soil depths in global drylands.