Reviews and syntheses : Soil resources and climate jointly drive variations in microbial biomass carbon and nitrogen in China ’ s forest ecosystems

Microbial metabolism plays a key role in regulating the biogeochemical cycle of forest ecosystems, but the mechanisms driving microbial growth are not well understood. Here, we synthesized 689 measurements on soil microbial biomass carbon (Cmic) and nitrogen (Nmic) and related parameters from 207 independent studies published up to November 2014 across China’s forest ecosystems. Our objectives were to (1) examine patterns in Cmic, Nmic, and microbial quotient (i.e., Cmic /Csoil and Nmic /Nsoil rates) by climate zones and management regimes for these forests; and (2) identify the factors driving the variability in the Cmic, Nmic, and microbial quotient. There was a large variability in Cmic (390.2 mg kg ), Nmic (60.1 mg kg ), Cmic : Nmic ratio (8.25), Cmic /Csoil rate (1.92 %), and Nmic /Nsoil rate (3.43 %) across China’s forests. The natural forests had significantly greater Cmic (514.1 mg kg −1 vs. 281.8 mg kg) and Nmic (82.6 mg kg −1 vs. 39.0 mg kg) than the planted forests, but had less Cmic : Nmic ratio (7.3 vs. 9.2) and Cmic /Csoil rate (1.7 % vs. 2.1 %). Soil resources and climate together explained 24.4–40.7 % of these variations. The Cmic : Nmic ratio declined slightly with Csoil : Nsoil ratio, and changed with latitude, mean annual temperature and precipitation, suggesting a plasticity of microbial carbonnitrogen stoichiometry. The Cmic /Csoil rate decreased with Csoil : Nsoil ratio, whereas the Nmic /Nsoil rate increased with Csoil : Nsoil ratio; the former was influenced more by soil resources than by climate, whereas the latter was influenced more by climate. These results suggest that soil microbial assimilation of carbon and nitrogen are jointly driven by soil resources and climate, but may be regulated by different mechanisms.


Introduction
Microbial metabolism is closely coupled with ecological processes such as decomposition, nutrient cycling, and rhizosphere symbiosis, and hence plays a key role in regulating the biogeochemical cycle of terrestrial ecosystems (Chapin et al., 2011).Microbial communities convert soil organic carbon (C soil ) and nitrogen (N soil ) into microbial biomass and release CO 2 to the atmosphere, mineralize soil organic matter (SOM) and nutrients, and thus control energy and material flows between trophic levels (Miltner et al., 2012;Sinsabaugh et al., 2013).Although soil microbial biomass carbon (C mic ) and nitrogen (N mic ) only account for 0.9-6.5 % of C soil and 1.7-8.1 % of N soil , respectively (Xu et al., 2013), microbial decomposition of SOM releases about half of the total soil CO 2 efflux (Wang and Yang, 2007).Microbes also provide supplementary nutrients for the plants particularly in nutrient-limited ecosystems (Singh et al., 1989) because of the dichotomous seasonality between microbial biomass and soil nutrient availability (Kaiser et al., 2011).Therefore, understanding soil microbial metabolism and its feedback to climate change is critical in carbon cycling studies (Davidson and Janssens, 2006;Singh et al., 2010) especially in forest ecosystems that contain more than 70 % of the global C soil storage (Jandl et al., 2007).
Published by Copernicus Publications on behalf of the European Geosciences Union.
Recently, Xu et al. (2013) synthesized the global microbial biomass data of terrestrial ecosystems, of which only about 24 % were of forest ecosystems mainly from North America and Europe.China has diverse geographic and environmental conditions, including a latitudinal gradient from boreal to tropical climate zones, a longitudinal gradient from arid deserts to humid forests, and an altitudinal ladder-like topography from the east to the west, which allow various forests to exist (Fang et al., 2012;Yang et al., 2014).Research on the forest soil microbes in China had lagged behind many regions in the world, but has been catching up rapidly since 2000.During the past 15 years, large amounts of data on microbial biomass have been produced but not synthesized yet.Synthetic analyses of these data will help to understand the patterns and environmental controls of soil microbial biomass.
Soil resources saliently influence microbial growth (He et al., 2014;Jagadamma et al., 2014;Kaiser et al., 2014), because C soil and N soil are the fundamental sources of energy and nutrients for microbial metabolism (Wardle, 1992;Xu et al., 2014;Spohn, 2015).Soil microbial quotient (i.e., C mic / C soil and N mic / N soil rates) is an important index of carbon assimilation of microorganisms (Xu et al., 2014), and also a sensitive indicator of soil biological responses to environmental changes (Sparling, 1992).To date, however, it is not conclusive how soil substrates control microbial metabolic processes.For example, Xu et al. (2014), using field data and modeling simulation, indicated that soil microbial quotient decreased exponentially with the C soil : N soil ratio at a biome-level.On the other hand, Spohn (2015), based on laboratory incubation data, reported that microbial metabolic quotient (i.e., respiration rate per unit microbial biomass) linearly increased with the litter C : N ratio but decreased with the litter nitrogen concentration.Such diverse relationships between soil resources and microbial metabolism may be associated with environmental conditions and research scales, because the translocation of soil resources involves a series of biological activities and physical diffusion that are constrained by soil temperature and moisture regimes especially in frozen or arid harsh environment or seasons (Jefferies et al., 2010;Edwards and Jefferies, 2013).
At a large scale, geographic climate is the primary driver of vegetation communities (Drenovsky et al., 2010).Vegetation controls C soil inputs, alters soil habitats by modifying rhizospheric activities and elemental stoichiometry, and eventually influences microbial composition and metabolism (Paterson et al., 2007;Bell et al., 2014).Additionally, forest disturbances change soil and vegetation characteristics, which in turn alter microbial biochemical processes (Holden and Treseder, 2013).Of the diverse influencing factors, it is not clear which are the major drivers of the variability in soil microbial biomass (Wardle, 1998;Xu et al., 2013).
In this study, we retrieved peer-reviewed papers on microbial biomass in China's forest ecosystems published up to November 2014.The database included 689 measurements from 207 independent studies (Table S1 in the Supplement) that covered all the climate zones (frigid highland, cool temperate, warm temperate, temperate desert, subtropical/tropical zones) and management regimes (natural vs. planted forests; Fig. 1).The objectives of this synthesis were to (1) examine patterns in C mic , N mic , and microbial quotient by climate zones and management regimes for China's forest ecosystems; and (2) identify the factors driving the variability in the C mic , N mic , and microbial quotient.We hypothesized that (1) C mic , N mic , and microbial quotient vary with climate zones and management regimes because of different quality and quantity of soil resources and environmental conditions involved; and (2) the variability in C mic and N mic are jointly driven by soil resources and climate (a proxy of soil temperature and moisture regimes).

Data collection
We collected the data on soil microbial biomass of China's forest ecosystems published up to November 2014.The literature retrieval was conducted through the China National Knowledge Infrastructure (CNKI, http://www.cnki.net;almost all studies indexed are conducted in China) and the Web of Science (http://apps.webofknowledge.com)online databases using "soil microbial", "forest" and "China" or "Chinese" as the key words.To enhance the cross-study comparability, the original papers were further screened by the following criteria: (1) the studies must contain corresponding C soil or SOM and N soil data.The SOM values were converted into C soil values using the Bemmelen index (0.58,Mann, 1986).(2) Following previous synthesized studies (e.g., Don et al., 2011), the data with anthropogenic disturbances or management activities in recent years (< 7 years) were excluded.(3) Only the measurements from the control or untreated plots were included if a manipulated experiment was employed.And (4) the data on C mic and N mic determined only with the chloroform fumigation-extraction method were included, but the conversion factors were ignored due to lacking detailed description in the original literature.To our knowledge, there were no papers published before 2000 reporting China's soil microbial biomass that were measured with the chloroform fumigation-extraction method.Eventually, 689 measurements from 207 peer-reviewed papers met the criteria above and were included in this synthesis (Table S1).The data set covered the forest regions across China, divergent in climate zones and management regimes (Fig. 1).
We also obtained corresponding data on the geographic locations (latitude, longitude), climate conditions (mean annual temperature (MAT), mean annual precipitation; MAP), soil properties (bulk density), and vegetation characteristics (forest type, management regime).Unlike previous syntheses (e.g., Tian et al., 2010) that averaged the climate data across the whole climate zones for many years, the MAT and MAP in this study were calculated from the original literature and weighted by the numbers of measurements.This algorithm may more realistically reflect the environmental conditions driving the microbial metabolism, because microbial communities are more directly responsive to the local habitat, rather than the large-scaled environment.All data collected in this study were numerical data from the tables, text, or extracted from the figures in the original papers with the Origin 7.0 digital plugin (Digitize; OriginLab Ltd., China).

Data standardization and analyses
Considering the effects of climate, parent materials and vegetation, we standardized the data set by implementing the following procedures.First, we categorized the data set into five climate zones, i.e., frigid highland (FH), cool temperate (CT), warm temperate (WT), temperate desert (TD), and subtropical/tropical (ST) zones based on the 1 : 1 000 000 landuse map of China (Wu, 1988).Since forests are rarely distributed in the TD zone, we excluded it in this synthesis (Fig. 1).Second, we further divided the forests within each climate zone into natural and planted forests by management regimes.Third, the soil data were all standardized to the top 20 cm depth in order to maximize the cross-study comparison.The rationales for doing this are: (1) C mic and N mic are concentrated on the top soil layers (0-20 cm; Cleveland and Liptzin, 2007); and (2) direct comparisons with the raw data are problematic, because the individual studies included in the current data set applied different soil sampling methods (e.g., by soil profile horizons or fixed-depth layers) and sampling depths.About 75 % of the studies were implicated in the top 20 cm of the soil (Fig. 1 Inset 2).Therefore, we conducted the following procedures for the topsoil standardization: (1) in the cases that the 0-20 cm topsoil was divided into more than one sublayer, we obtained the bulk density (BD) weighted means of C soil , N soil , C mic , and N mic across the 0-20 cm depth.The BD, if missing in the original papers, was derived from an empirical relationship between SOM and BD (i.e., BD = 0.29 + 1.2033 exp (−0.0775 × SOM); Wu et al., 2003;Yang et al., 2007).(2) In the cases that the maximum sampling depth was less than 20 cm, we estimated the C soil , N soil , C mic , and N mic across the 0-20 cm depth with the empirical equations (C soil : y = 1 − 0.878x, R 2 = 0.95, P < 0.001; N soil : y = 1 − 0.893x, R 2 = 0.95, P < 0.001; C mic : y = 1−0.889x,R 2 = 0.97, P < 0.001; N mic : y = 1−0.869x,R 2 = 0.97, P < 0.001, where x is the soil depth (cm), and y is the cumulative fraction of the dependent variable.See details in Fig. S1 in the Supplement).(3) In the cases that the Inserted figures are the overall comparisons between the natural (NF) and planted forests (PF).C soil , N soil , C mic , and N mic stand for soil organic carbon, soil total nitrogen, microbial biomass carbon, and microbial biomass nitrogen, respectively.FH, CT, WT, and ST stand for frigid highland zone, cool temperate zone, warm temperate zone, and subtropical/tropical zone, respectively.Different lowercase letters denote significant differences among the climate zones or between the management regimes (mean ±SE, α = 0.05).seasonal dynamics in microbial biomass were reported, we calculated arithmetic means of the data.
An ANOVA procedure was performed to test the differences in all variables among climate zones and between management regimes.A multiple regression procedure was employed to examine the effects of soil resources and climate on C mic , N mic , C mic : N mic ratio and microbial quotient.A full model (including the independent variables of both soil resources and climate and possible interactions) and a reduced model (only including the independent variables of soil re- sources and possible interactions) for each dependent variable were developed to examine partial contributions of soil resources or climate to the variability.A backward elimination procedure was conducted to remove insignificant terms (α = 0.05) in the models.The variables, if needed, were transformed with a 10-based logarithm to meet the assumptions of normality and homoscedasticity for the regressions.
To examine the effect of soil quality on C mic , we subjectively divided the original data set into two groups by the median of the C soil : N soil ratio, and defined as high-Table 1.Multiple regression models of microbial biomass carbon (C mic ), microbial biomass nitrogen (N mic ), C mic : N mic ratio, and microbial quotients (i.e., C mic / C soil and N mic / N soil rates) against soil resources and climate.C soil and N soil stand for soil organic carbon and soil total nitrogen, respectively.Y , X 1 , X 2 , X 3 , X 4 , X 5 in the models stand for the dependent variable, C soil , N soil , mean annual temperature, mean annual precipitation, and C soil : N soil ratio, respectively.Y , X 1 and X 2 are 10-based log transformed.N, R 2 , P stand for sample size, determination coefficient, and probability, respectively.All the terms in the regression models are significant at α = 0.05.
quality group (C soil : N soil ratio ≤ the median, n = 251) and low-quality group (C soil : N soil ratio > the median, n = 250).Then we applied the standardized major axis (SMA) procedure to compare the C mic -C soil or C mic -N soil relationship between the two groups.The SMA procedure was also used to compare the C mic : N mic -C soil : N soil relationship between this study and the one derived from the global data set (Cleveland and Liptzin, 2007).All statistical analyses were performed with SPSS 19.0 for Windows; and the map (Fig. 1) was generated with ArcGIS 10.

Concentrations of soil microbial biomass carbon and nitrogen
By synthesized 689 measurements on soil microbial biomass carbon (C mic ) and nitrogen (N mic ) from 207 independent studies published up to November 2014 across China's forest ecosystems, we found that the C mic in the topsoil (0-20 cm) varied by 142 folds (mean ±SD: 390.2 ± 312.9 g kg −1 ), the N mic varied by 289 folds (60.1 ± 57.4 g kg −1 ), C soil varied by 72 folds (26.2 ± 21.9 g kg −1 ), and the N soil varied by 62 folds (2.06 ± 1.65 g kg −1 ).The C mic and N mic differed significantly between management regimes and among climate zones (Fig. 2).The natural forests had significantly greater C mic (514.1 mg kg −1 vs. 281.8mg kg −1 ) and N mic (82.6 mg kg −1 vs. 39.0 mg kg −1 ) than the planted forests.In the natural forests, the C mic and N mic displayed similar patterns across the climate zones: ST > FH > CT > WT (Fig. 2e  and g).The ST zone had the greatest C mic and N mic with the greatest MAT and MAP (Fig. 1 Inset 1) in spite of having the least C soil and N soil among the climate zones (Fig. 2a and c).The C soil significantly (P < 0.05) decreased from cold humid FH zone to warm humid ST zone, the former was one fold greater than the latter (Fig. 2a).This pattern was largely op-posite to that of MAT (Fig. 1 Inset 1).There was no significant difference in N soil (P = 0.673) among the climate zones (2.83 ± 1.85 g kg −1 ; Fig. 2c).
The planted forests had significantly lower concentrations of C soil , N soil , C mic , and N mic than the natural forests (P < 0.05); the latter had 2.2, 2.0, 1.8, and 2.1 times as much as the former, respectively (Fig. 2 Inset).Unlike the natural forests, the planted forests exhibited relatively consistent patterns of C soil , N soil , C mic , and N mic across the climate zones: FH > CT > WT or ST (Fig. 2b, d, f, and h).
The C soil , N soil , and their interactions explained 36.6 % of the variability in C mic , and 28.7 % in N mic ; MAT and MAP explained additional 2.6 and 12.0 %, respectively (Table 1).Both C mic and N mic were increasing in a power function with C soil or N soil (Fig. 3).Further SMA analysis showed that the high-quality (i.e., low C soil : N soil ratio) and lowquality (i.e., high C soil : N soil ratio) soil groups had a common slope (P > 0.05) of the log(C soil ) -log(C mic ) or log(N soil ) -log(C mic ) relationship (Fig. 4).However, the high-quality group had a significantly greater intercept (P < 0.001) than the low-quality group for the log(C soil )-log(C mic ) relationship (Fig. 4a), while the former had a significantly less intercept (P < 0.001) than the latter for the log(N soil )-log(C mic ) relationship (Fig. 4b).

Stoichiometry of soil microbial biomass carbon and nitrogen
The C soil : N soil and C mic : N mic ratios varied by 15 folds (13.7 ± 6.63) and 41 folds (8.25 ± 6.20) across the country, respectively.Management regime and climate zone significantly affected the C mic : N mic ratio (Fig. 5).The C mic : N mic ratio in the natural forests was significantly less (P < 0.05) than that in the planted forests (7.3 vs. 9.2; Fig. 5d Inset), while the C soil : N soil ratio in the former was significantly greater (P < 0.05) than that in the latter (15.2 vs. 12.4; Fig. 5b Inset).The C mic : N mic ratio showed a similar pattern across C soil , N soil , C mic , and N mic stand for soil organic carbon, soil total nitrogen, microbial biomass carbon, and microbial biomass nitrogen, respectively.FH, CT, WT, and ST stand for frigid highland zone, cool temperate zone, warm temperate zone, and subtropical / tropical zone, respectively.Different lowercase letters denote significant differences among climate zones or between management regimes (mean ± SE, α = 0.05).
the climate zones for both natural and planted forests, with the greatest ratio occurring in the CT zone and the smallest ratio in the FH zone (Fig. 5c and d).
Pooling the data across the climate zones and management regimes, we found a significant power relationship, rather than a linear, between C soil and N soil or between C mic and N mic (P < 0.001; Fig. S2).The C soil and N soil together explained 10.8 % of the variability in the C mic : N mic ratio, and climate explained another 13.6 % (Table 1).There was a weak negative correlation between C mic : N mic ratio and C soil : N soil ratio (n = 335, P < 0.01, R 2 = 0.04; Fig. 6a), which was consistent with the relationship derived from the global data set (Cleveland and Liptzin, 2007).The slopes of the relationships were −0.13 [−0.21, −0.05] (mean [95 % confidence interval]) from this study and −0.12 [−0.23, −0.01] from the global data set, while the intercepts were 9. 73 [8.45, 11.06] and 9.34 [7.86, 10.82], respectively.The C mic : N mic ratio increased with latitude (Fig. 6b), decreased with MAT (Fig. 6c), and was correlated with MAP in a quadratic polynomial function (Fig. 6d).

Soil microbial quotient
The C mic / C soil rate varied by 91 folds (1.92 % ± 1.17 %) across China's forest ecosystems, while the N mic / N soil rate varied by 480 folds (3.43 % ± 2.49 %).They differed significantly between management regimes and among climate zones (Fig. 5).The C mic / C soil rate in the natural forests was significantly less (P < 0.05) than that in the planted forests (1.7 % vs. 2.1 %; Fig. 5f Inset), while the N mic / N soil rate did not differ (3.5 % vs. 3.4 %; P > 0.05; Fig. 5h Inset).The C mic / C soil and N mic / N soil rates showed different patterns across the climate zones between the natural and planted forests (Fig. 5e-h).In the natural forests, the FH zone had the least C mic / C soil rate but the greatest N mic / N soil rate among the climate zones; the ST zone had greater C mic / C soil and N mic / N soil rates (Fig. 5e and g).In the planted forests, the C mic / C soil and N mic / N soil rates showed similar patterns across the climate zones: WT > ST > FH > CT (Fig. 5f and h).
The C soil , N soil , and their combinations explained 36.7 % of the variability in the C mic / C soil rate, and 17.3 % in the N mic / N soil rate; MAT, MAP, and their interactions explained another 2.4 and 13.3 %, respectively (Table 1).The C mic / C soil rate decreased in a power function with the C soil : N soil ratio (Fig. 7a), but the N mic / N soil rate increased linearly with the C soil : N soil ratio (Fig. 7b).Both C mic / C soil and N mic / N soil rates increased with MAT (Fig. 7c and d).There was a quadratic polynomial function between the C mic / C soil rate and MAP (Fig. 7e), but there was no significant relationship (P > 0.05) between the N mic / N soil rate and MAP (Fig. 7f).

Variability in forest soil microbial biomass carbon and nitrogen
The C mic and N mic in China's forest ecosystems varied substantially with climate zones and management regimes (CV up to 80-96 %), consistent with previous studies (Cleveland  and Liptzin, 2007;Tian et al., 2010;Xu et al., 2013).Forest disturbances contain various regimes and histories, and thus increase the complexity of the microbial biomass pattern associated with climate (Holden and Treseder, 2013).In this synthesis, the natural forests in the subtropical/tropical zones with the greatest MAT and MAP (Fig. 1 Inset 1) had the greatest C mic and N mic among the climate zones (Fig. 2), while the planted counterparts had the least C mic and N mic (Fig. 2f and h).Similarly, the planted forests in both warm and cool temperate zones had lower C mic and N mic than those in the frigid highland zone.These different patterns between the natural and planted forests may be associated with, in addition to climate, more afforestation and younger plantations in regions other than the frigid highland zone in China (Fang et al., 2012;Yang et al., 2014).
In this study, soil resources and climate together explained about 40 % of the variability in microbial biomass, to which C soil and N soil contributed 71-93 % (Table 1).All power functions of microbial biomass against soil resources had exponents of less than one (Fig. 3), indicating that microbes grow rapidly with soil resource availability increasing when the C soil or N soil is low, but the acceleration decreases as the C soil or N soil increases.This implies a shift in constraints of microbial growth from resource availability to other abiotic and/or biotic factors, such as soil moisture, temperature, pH, etc.Furthermore, the high-quality (i.e., low C soil : N soil ratio) soils had a greater intercept of the log(C soil )-log(C mic ) relationship (Fig. 4a) but a less intercept of the log(N soil )log(C mic ) relationship (Fig. 4b) than the low-quality soils.This result indicates that the microbes grow faster in the soils with a lower C soil : N soil ratio at a specific C soil level or in the carbon-richer soils at a specific N soil .These findings illustrate that the relationships between soil resources and microbial biomass are affected by the availability and stoichiometry of C soil and N soil .

Stoichiometry of forest soil microbial biomass carbon and nitrogen
The C mic : N mic ratio in the topsoil for the natural forests in this study (7.3) is slightly greater than the global mean (6.9; Cleveland and Liptzin, 2007;Xu et al., 2013).It varied substantially less than C mic (41 folds vs. 142 folds) or N mic (289 folds; Fig. 5).The relatively little variability in the C mic : N mic ratio, together with its slight decline with the C soil : N soil ratio (Fig. 6a), illustrates that soil microbes alleviate soil resource stoichiometry (Kaiser et al., 2014), rather than "you are what you eat" (Sterner and Elser, 2002).Nevertheless, the C mic : N mic ratio decreased by 13 % as the C soil : N soil ratio increased (Fig. 6a), consistent with other terrestrial ecosystems studies (e.g., Cleveland and Liptzin, 2007;Fanin et al., 2013); And it also changed with latitude (Fig. 6b) and climate (Fig. 6c and d).Although the R 2 val-Z.H. Zhou and C. K. Wang: Microbial carbon and nitrogen in China's forests ues of these relationships were small, they reflect a general trend of the stoichiometric flexibility of the microbial communities (Sterner and Elser, 2002;Li et al., 2012;Fanin et al., 2013).Small adjustments of microbial biomass stoichiometry may have a significant effect on soil resource cycles (e.g., respiration and N, P mineralization, c.f., Mooshammer et al., 2014).Further multiple regression analysis showed that soil resources and climate together explained 24.4 % of the variability in the C mic : N mic ratio (Table 1), which might be useful for modeling the biogeochemical cycle of forest ecosystems.However, no conclusive notion has been reached on the underlying mechanisms of such an environmental plasticity of microbial stoichiometry.We speculate that the greater C mic : N mic ratio in cold temperate zone (Figs. 1 Inset, 5c,  and d), perhaps similar to that in the plant leaves (He et al., 2006;Asner et al., 2014), might reflect an increased carbon investment on structural material (e.g., cell wall) as a defensive strategy in harsh environments.Additionally, environmental changes may shift microbial community composition, and thus change the C mic : N mic ratio (Drenovsky et al., 2010;Tischer et al., 2014), because fungi have higher C : N ratio than bacteria (Keiblinger et al., 2010).Clearly, disclosing the underlying mechanisms of microbial metabolism is challenging but critically needed.

Forest soil microbial quotient
Soil microbial quotient is largely determined by microbial assimilation and respiration that are mainly affected by soil resources and environmental conditions (Manzoni et al., 2012).In this study, we found that C mic / C soil and N mic / N soil rates responded to soil resources differently or even oppositely.First, soil resources explained 36.7 % of the variability in the C mic / C soil but only 17.3 % in the N mic / N soil (Table 1), indicating that the C mic / C soil rate was more controlled by soil resources.Second, the C mic / C soil and N mic / N soil rates responded oppositely to the C soil : N soil ratio, i.e., decreasing in a power function (Fig. 7a) versus increasing linearly (Fig. 7b).Third, much stronger interactions between MAT and MAP exerted on the N mic / N soil rate than the C mic / C soil rate, because weak or no correlations between the N mic / N soil rate and MAT (Fig. 7d) or MAP (Fig. 7f) despite that climate factors explained 4.5 times more variability in the N mic / N soil rate than in the C mic / C soil rate (13.3 % vs. 2.4 %; Table 1).These results, together with Fig. 4 (See the discussion above), suggest that soil microbial assimilation of carbon and nitrogen respond differently to soil resources and climate, and might be regulated by different mechanisms.
Vegetation and disturbances modify soil resources, interact with climate, and eventually affect microbial growth and metabolism (Figs. 5 and 7).The rapid decrease in C mic / C soil rate with C soil : N soil ratio increasing (Fig. 7a) suggests that soil microbes produce less biomass and respire more carbon in N-limited soils (Xu et al., 2014).The underlying mechanisms, however, are still not clear.Spohn (2015) sum-marized three potential mechanisms, i.e., microbial nitrogen mining, overflow respiration and enzyme inhibition at high nitrogen concentration.Additionally, microbial communities, in responses to changes in C soil : N soil ratio, may evolve to a new balance between species-specific resource demand and resource availability in the micro-habitat via inter-specific competition and shifting of specific extracellular enzymes stoichiometry, and eventually alter the resource use efficiency at the community level (Tilman, 1982;Cherif and Loreau, 2007;Kaiser et al., 2014;Mooshammer et al., 2014;Torres et al., 2014).

Conclusions
By synthesizing 689 measurements from 207 independent studies across China's forest ecosystems, we find large variations in soil microbial biomass carbon and nitrogen and microbial quotient that are subjected to changes in soil resources, climate, and management regimes.We also provide evidence for stoichiometric flexibility of microbial communities.These results improve our understanding of soil microbial metabolic processes, ecological stoichiometry, and microbial responses to environmental change.
The Supplement related to this article is available online at doi:10.5194/bg-12-6751-2015-supplement.

Figure 1 .
Figure 1.A map of China showing the distribution of sampling sites and data summary for this synthesis.The climate zones are categorized following Wu (1988) into frigid highland (FH), cool temperate (CT), warm temperate (WT), subtropical/tropical (ST), and temperate desert (TD) zones.The TD is excluded in the synthesis, because forests are rarely distributed in the zone.Inset 1: mean annual temperature (MAT) and mean annual precipitation (MAP) by climate zones.Inset 2: frequency distribution of maximum depth of soil sampling.Inset 3: the number of measurements of soil organic carbon (C soil ), soil total nitrogen (N soil ), microbial biomass carbon (C mic ), and microbial biomass nitrogen (N mic ) by climate zones.

Figure 2 .
Figure 2. Comparisons of C soil (a, b), N soil (c, d), C mic (e, f), and N mic (g, h) among climate zones and between management regimes.Inserted figures are the overall comparisons between the natural (NF) and planted forests (PF).C soil , N soil , C mic , and N mic stand for soil organic carbon, soil total nitrogen, microbial biomass carbon, and microbial biomass nitrogen, respectively.FH, CT, WT, and ST stand for frigid highland zone, cool temperate zone, warm temperate zone, and subtropical/tropical zone, respectively.Different lowercase letters denote significant differences among the climate zones or between the management regimes (mean ±SE, α = 0.05).

Figure 3 .
Figure 3. Relationships between soil resources and microbial biomass.C soil , N soil , C mic , and N mic stand for soil organic carbon, soil total nitrogen, microbial biomass carbon, and microbial biomass nitrogen, respectively.The regression models are given.

Figure 4 .
Figure 4. Relationships between microbial biomass carbon (C mic ) and soil organic carbon (C soil , a), and soil total nitrogen (N soil , b) by two soil quality groups: high-quality group (C soil : N soil ratio ≤ the median of the whole data set, n = 251) vs. low-quality group (C soil : N soil ratio > the median, n = 250; See details in the Methods).Log stands for 10-based logarithm.The standardized major axis (SMA) models are given for each group.

Figure 5 .
Figure 5. Comparisons of C soil : N soil ratio (a, b), C mic : N mic ratio (c, d), C mic / C soil rate (e, f), and N mic / N soil rate (g, h) among climate zones and between management regimes.Inserted figures are overall comparisons between the natural (NF) and planted forests (PF).C soil , N soil , C mic , and N mic stand for soil organic carbon, soil total nitrogen, microbial biomass carbon, and microbial biomass nitrogen, respectively.FH, CT, WT, and ST stand for frigid highland zone, cool temperate zone, warm temperate zone, and subtropical / tropical zone, respectively.Different lowercase letters denote significant differences among climate zones or between management regimes (mean ± SE, α = 0.05).

Figure 6 .
Figure 6.Relationships between C mic : N mic and C soil : N soil ratios (China: from this study vs. Globe: from Cleveland & Liptzin, 2007; a), latitude (b), MAT (c), and MAP (d), respectively.C soil , N soil , C mic , N mic , MAT, and MAP stand for soil organic carbon, soil total nitrogen, microbial biomass carbon, microbial biomass nitrogen, mean annual temperature, and mean annual precipitation, respectively.The regression models are given.

Figure 7 .
Figure 7. Relationships between C mic / C soil or N mic / N soil rate and C soil : N soil ratio (a, b), MAT (c, d), and MAP (e, f), respectively.C soil , N soil , C mic , N mic , MAT, and MAP stand for soil organic carbon, soil total nitrogen, microbial biomass carbon, microbial biomass nitrogen, mean annual temperature, and mean annual precipitation, respectively.The regression models are given.NS stands for non-significant.