Results

Biome-level FBC, BBC, and F:B Ratio

There are large variations in biome-level FBC, BBC, and F:B ratio (Table 1; P < 0.001 for FBC, BBC, and F:B ratio among biomes). Deserts exhibited the lowest FBC of 16.9 (95% range: 14.4~19.9) mg kg-1 and BBC of 6.8 (6.1~7.7) mg kg-1, while tundra habitats displayed the highest FBC of 3683.6 (1678.5~8083.9) mg kg-1 and BBC of 428.4 (237.0~774.3) mg kg-1. Boreal forests had significantly higher FBC than tropical/subtropical forests and temperate forests (1234.0 mg kg-1 for boreal forests vs. 258.4 mg kg-1 for temperate forests and 451.4 mg kg-1 for tropical/subtropical forests). Boreal forest and tropical/subtropical forests have significantly higher BBC than temperate forests (226.4 mg kg-1 for boreal forest, 210.9 mg kg-1 for tropical/subtropical forest vs. 53.0 mg kg-1 for temperate forest), with no significant differences in BBC were found between boreal forests and tropical/subtropical forests (Table 1 ). Pastures had significantly higher FBC and BBC than grasslands (632.2 mg kg-1 soil vs. 215.2 mg kg-1 soil for FBC and 270.7 mg kg-1 soil vs. 62.7 mg kg-1 soil for BBC); While we did not find differences in FBC across unvegetated ground, cropland, shrub, savanna, and natural wetlands; BBC was significantly higher in wetlands than in unvegetated ground (Table 1 ).
The F:B ratio varied less across biomes, with the lowest values in savannas and greatest values in tundra habitats (1.8 for savanna vs. 8.6 for tundra). We also found significantly higher F:B ratio in boreal forests and temperate forests than that in tropical/subtropical forests (5.0 for boreal forest, 4.9 for temperate forest vs. 2.2 for tropical/subtropical forest). No significant differences in F:B ratio were found across natural wetlands, unvegetated grounds, deserts, and shrubs (Table 1 ).

Biogeography of FBC, BBC, and F:B ratio

Both FBC and BBC exhibited inverse unimodal relationships with latitude, with lowest values at mid-latitudes (Fig. S2a-b ; P< 0.0001 for both FBC and BBC along latitude), whereas the F:B ratio was positively correlated with latitude (Fig. S2c ;P < 0.0001). Of climatic predictors, MAT showed an inverse unimodal relationship with FBC, with the lowest at 14-15°C (Fig. S3a ; P < 0.0001). Conversely, BBC showed no significant correlation with MAT (Fig. S3b ; P = 0.19). The F:B ratio showed a significantly negative linear relationship with MAT (Fig. S3c ; P < 0.0001).
Both FBC and BBC showed unimodal relationships with MAP, with peak FBC and BBC at approximately 2100-mm y-1 and 3000-mm y-1, respectively. While F:B ratio linearly decreased with MAP (Fig. S3d-f ; P FBC < 0.0001, P BBC < 0.0001,P F:B ratio < 0.0001). FBC increased in a non-linear manner with SM, while BBC linearly increased with SM (Fig. S4a-b ; P FBC < 0.0001,P BBC < 0.0001). Both FBC and BBC linearly increased with ST (Fig. S4d-e;P FBC < 0.0001,P BBC < 0.0001). F:B ratio increased with SM (Fig. S4c ; P < 0.0001) but decreased with ST (Fig. 4f ; P < 0.0001).
Vegetation controls on microbial biomass C differed in fungi and bacteria. While BBC significantly increased with Croot(Fig. S5a-b ; P FBC = 0.2,P BBC = 0.00035), no significant correlation between FBC and Croot occurred. The F:B ratio exhibited a unimodal correlation with Croot, with the peak F:B ratio associated with the Croot of 6.9 kg m-2 (Fig. S5c ; P < 0.0001). Both FBC and BBC linearly increased with NPP, while F:B ratio linearly decreased with NPP (Fig. S5d-f ; P FBC = 0.011, P BBC < 0.0001,P F:B ratio< 0.0001).
Microbial biomass was well correlated with edaphic factors. Both FBC and BBC exhibited inverse unimodal relationships with SOC, with minimum FBC and BBC at SOC of 142.1 and 222.7 g kg-1, respectively (Fig. S6a and b ; P FBC < 0.0001, P BBC = 0.0017), while F:B ratio linearly increased with SOC (Fig. S6c ; P < 0.0001). Both FBC and BBC linearly increased with TN, while F:B ratio exhibited unimodal relationship with TN, with the maximum F:B ratio at TN of 25.4 g kg-1 (Fig. S6d-f ;P FBC < 0.0001,P BBC = 0.011, P F:B ratio< 0.0001). Both FBC and BBC showed unimodal relationships with SOC:TN (C:N) ratio, with the maximum FBC and BBC at C:N ratio of 20.1 and 17.7, respectively (Fig. S6g-h ; P FBC< 0.0001, P BBC < 0.0001), while F:B ratio showed inverse unimodal relationship with C:N ratio, with minimum F:B ratio at C:N ratio of 7.1 (Fig. S6i ;P F:B ratio < 0.0001). In addition, both FBC and BBC showed inverse unimodal relationships with soil bulk density, with minimum FBC and BBC at bulk density of 1.5 and 1.4 g cm-3, respectively, while F:B ratio linearly decreased with bulk density (Fig. S6j-l ; P FBC< 0.0001, P BBC = 0.00035,P F:B ratio < 0.0001). Furthermore, we found that FBC, BBC, and F:B ratio all showed inverse unimodal relationships with soil pH, with minimum FBC, BBC, and F:B ratio at soil pH of 7.5, 7.4 and 6.3, respectively (Fig. S6m-o ;P FBC < 0.0001,P BBC < 0.0001,P F:B ratio < 0.0001). We also found the highest FBC and BBC in clayey s, but the highest F:B ratio in sandy soil (Fig. 6p-r; P FBC < 0.0001,P BBC < 0.0001,P F:B ratio < 0.0001).

Quantitative Assessment of Controls on Microbial Biogeography

We constructed generalized linear models to disentangle the effects of climate (MAP and MAT), plant (NPP and Croot), soil microclimate (SM and ST), and edaphic properties (SOC, TN, soil pH, clay, sand, and bulk density), on the variation in FBC, BBC, and F:B ratio. The variance inflation factor (VIF) test revealed no multicollinearity among variables. Environmental factors in total explained a large proportion of variation in microbial biomass (81.9% for FBC, 84.8% for BBC, and 71.2% for F:B ratio) (Fig. 2 ). Notably, the edaphic properties were the most important drivers in FBC and BBC, with 66.4% and 70.4% of the variations in FBC and BBC being explained by edaphic properties and the interaction with other factors, respectively (Fig. 2a-b ). Complex interactions between the groups of variables explained 23.7% of the variation in FBC (Fig 2a ). In contrast, variation in BBC was explained primarily by the interactions between edaphic properties and climate (13.9%), multiple interactions (11.91%), and edaphic properties alone (10.22%). Climate was the most important predictor of F:B ratio. Climate alone explained 11.6%, and climate interactions with other variables explained 35.5% of the variation in the F:B ratio (Fig. 2c ).

Global Carbon Storage in Fungal and Bacterial Biomass

Based on our findings of environmental controls on FBC and BBC at the biome and global scales, we further developed an empirical model for F:B ratio considering the higher proportion of missing data in FBC (14.8%) and BBC (16.3%) relative to F:B ratio (1.9%) (Materials and Methods; Table S2 ). Combined with a global microbial biomass C dataset reported by Xu et al. (2013), we further produced global maps of BBC and FBC in topsoil (Fig. 3 ). The global FBC and BBC are estimated to be 12.56 (6.64~16.42) Pg C, and 4.34 (0.47~10.26) Pg C in BBC for 0-30 cm topsoil. Taking the global estimates of SOC (684~724 Pg C in 0-30 cm), approximately 1.8% and 0.6% of SOC is stored in soil fungi and bacteria, respectively. The highest FBC density occurs in northern high-latitude regions while lowest values are characteristic of low-latitude regions (Fig. 3b ). On the contrary, the highest BBC was found in high-latitude and equatorial regions, and the lowest in mid-latitude regions (Fig. 3c) .
At biome-level, boreal forests stored large FBC (3.60 Pg C) and tropical/subtropical forests had the largest BBC storage (0.85 Pg C), while shrub contributed least to both FBC and BBC (0.39 Pg C for FBC and 0.14 Pg C for BBC) (Table 2 ). Although boreal forests do not occupy the Earth’s largest surface area (11.82 million km2), the high FBC density contributes to its prominent FBC storage. The large microbial C storage in pasture was primarily due to its large area (27.0 million km2). Tropical/subtropical forests have relatively high BBC density, along with the second largest area (16.44 million km2), tropical/subtropical forests thus stored the largest BBC across the globe. The smallest FBC and BBC storage in shrub was primarily due to its small area (8.11 million km2) and the low FBC and BBC densities (48.06 g C m-2 for FBC and 17.31 g C m-2 for BBC). The small FBC and BBC storage in desert primarily resulted from their low FBC and BBC densities (Fig S5 ), while the small FBC and BBC storage in tundra and natural wetland may be due to the small area (5.75 million km2 for tundra and 6.91 million km2 for natural wetlands). Eventually tundra has high densities of FBC and BBC (226.96 g C m-2 for FBC and 32.65 g C m-2 for BBC).