Plant functional traits determined the latitudinal 1 variations in soil microbial functions : evidence from a 2 forest transect in China 3 4

Abstract. Plant functional traits have increasingly been studied as determinants of ecosystem properties, especially for soil biogeochemical processes. While the relationships between biological community structures and ecological functions are a central issue in ecological theory, these relationships remain poorly understood at the large scale. We selected nine forests along the North–South Transect of Eastern China (NSTEC) to determine how plant functional traits influence the latitudinal pattern of soil microbial functions, and how soil microbial communities and functions are linked at the regional scale. We found that there was considerable variation in the profiles of different substrate use along the NSTEC. Soil microorganisms from temperate forests mainly metabolized high-energy substrates, while those from subtropical forests used all the substrates equally. The soil silt content and plant functional traits together shaped the biogeographical pattern of the soil microbial substrate use. Soil organic matter decomposition rates were significantly higher in temperate forests than in subtropical and tropical forests, which was consistent with the pattern of soil microbial biomass carbon concentrations. Soil organic matter decomposition rates were also significantly and negatively related to soil dissolved organic carbon concentrations, and carboxylic acid, polymer, and miscellaneous substrates. The soil microbial community structures and functions were significantly correlated along the NSTEC. Soil carbohydrate and polymer substrate use were mainly related to soil G+ bacterial and actinomycetes biomass, while the use of amine and miscellaneous substrates were related to soil G− bacteria, fungal biomass, and the F/B ratio. The contributions of different groups of microbial biomass to the production of soil enzyme activities differed. The relationship between soil microbial community structure and functions supported that there was functional dissimilarity.



Introduction
The catabolic diversity of soil microbial communities is a useful indicator of how microbial functions adapt to environmental stress and can be used to test fundamental questions about soil biological resistance and resilience (Jagadamma et al., 2014;Swallow and Quideau, 2015).We need robust information about functional diversity to understand the role of microbial communities in different environments (Preston-Mafham et al., 2002).Biological community structure and function are intimately linked in ecological processes, and their relationships are a central issue in ecological theory (Talbot et al., 2014).Therefore, a major goal in ecological research is to identify and understand the mechanisms and relationships that control the structure and function of microbial community at large spatial scales.
Numerous studies have documented how environmental and anthropogenic perturbations impact on the structure, diversity (Zhou et al., 2016), and enzyme activities (Peng and Wang, 2016;Xu et al., 2017) of soil microbial communities (Tu et al., 2016), and have reported forests in the same climatic zone develop similar microbial communities.Other researchers have examined spatial patterns in soil microbial function at different scales.For example, in their study of Changbai Mountain, China, Tian et al. (2015) found that the soil microbial metabolic activity was moderately spatially dependent, and that the functional diversity was much more spatially dependent.Other researchers have reported differences in soil microbial activities among forest types, with high local variation and significant separation along regional climate gradients (Brockett et al., 2012;Cao et al., 2016).Soil microbes from different climatic zones have different affinities for carbon substrates.For example, microorganisms from boreal pine forest soils used carboxylic acids more efficiently, but decomposed amino acids much less efficiently, than microorganisms from temperate forest soils (Klimek et al., 2016).Despite this, because of limitations in analytical methods, questions still remain about how soil microbial functions vary at the regional scale.
The functional diversity of soil microbial communities is regulated by physico-chemical soil properties (Gartzia-Bengoetxea et al., 2016), climate (Cao et al., 2016), and the composition of plant cover (Sherman and Steinberger, 2012).For example, the geographic patterns of soil microbial activity reflect the climate, soil pH, and total phosphorus concentrations over large spatial scales (Cao Biogeosciences Discuss., https://doi.org/10.5194/bg-2018-499Manuscript under review for journal Biogeosciences Discussion started: 24 January 2019 c Author(s) 2019.CC BY 4.0 License. et al., 2016).Research has shown that substrate-induced respiration rates were higher in soil microbial communities that developed under beech and holm oak forests than under oak and pine forests (Gartzia-Bengoetxea et al., 2016).Plant functional traits have increasingly been studied as determinants of ecosystem properties, especially for soil biogeochemical processes (De Vries et al., 2012;Pei et al., 2016).Soil bacteria phospholipid fatty-acids (PLFAs) were found to be positively correlated with the community-weighted means (CWM) of plant functional traits (leaf nitrogen (N) concentration) (De Vries et al., 2012).The plant leaf dry matter content and the leaf carbon (C) to nitrogen (N) ratio both influence the multivariate soil microbial community structure, and these factors positively promote the abundances of specific microbial functional groups (Pei et al., 2016).
Limited soil resources, particularly in tropical forests, mean that soil microorganisms may be more reliant on plants than soil for C and nutrients via rhizosphere exudation or litter production, which varies among plant species (Russell et al., 2007;Raich et al., 2014;Waring et al., 2015).While soil functional diversity has been used as an indicator of microbial metabolic potential, there have been few studies of the integrated effects of climate, vegetation, and soil substrate availability on largescale soil microbial functional diversity.
Although the functional characteristics of soil microorganisms are at least as important as their patterns of community structure in biogeochemical studies, the links between microbial community structure and microbial functions are poorly understood.There are two current hypotheses about how microbes determine ecosystem process rates.In functional redundancy, different microbes perform the same function and so changes in the microbial community structure do not necessarily lead to a change in soil function (Balser and Firestone, 2005;Strickland et al., 2009).For example, Banerjee et al. (2016) showed that the abundance of different bacterial and fungal groups changed by up to 300-fold under straw-and nutrient-amended treatments but that the decomposition rate remained similar, indicating possible functional redundancy.The functional redundancy hypothesis has recently been challenged by a counter-hypothesis, referred to as functional dissimilarity, which suggests that diversity brings stability, and that every species plays a unique role in ecosystem function (Fierer et al., 2007;Waldrop and Firestone, 2006).Soil microbial community composition therefore, combined with environmental variables, may ultimately determine ecosystem process rates.Waldrop and Firestone (2006) showed that G + bacteria were mainly responsible for the Biogeosciences Discuss., https://doi.org/10.5194/bg-2018-499Manuscript under review for journal Biogeosciences Discussion started: 24 January 2019 c Author(s) 2019.CC BY 4.0 License.decomposition of pine needles and soil organic matter, but G − bacteria were mainly responsible for the decomposition of starch and xylose, which are easy to break down.Philippot et al. (2013), when studying the diversity of denitrifiers, showed that the loss of microbial diversity could result in decreases of between 4-and 5-fold in denitrification activity.In the Mediterranean, losses in the mass of decomposing leaf litter from shrub species accelerated as detritivore assemblages became more functionally dissimilar (Coulis et al., 2015).These studies suggest that the importance of functional redundancy in soil microbial communities has been overstated, so studies of the relationships between soil microbial communities and their functions in natural ecosystems are urgently needed.
The North-South Transect of Eastern China (NSTEC) extends from a cold temperate coniferous forest in the north to a tropical rainforest in the south, and includes almost all the forest types found in the Northern Hemisphere (Zhang and Yang, 1995) (Fig. 1 and Table 1).This transect, therefore, provides the optimal environment for investigating geographical patterns in microbial communities and their responses to environmental changes at the large scale.In this study, we examined spatial patterns in soil labile C concentrations, soil organic matter (SOM) decomposition rates, and metabolic activity and functional diversity of microbes in nine forest biomes along the NSTEC.We assessed how abiotic factors, such as climate, soil physical and chemical properties, and biotic factors, in the form of community-weighted means (CWM) of plant functional traits, contributed to soil functional diversity at the regional scale.We also examined the links between soil microbial community structure (PLFAs) and function (SOM decomposition rate, enzyme activities, and microbial substrate use).We tested the following three hypotheses in this study, that (1) the profiles of soil microbial substrate use would vary along a latitudinal gradient, (2) biogeographical patterns of soil microbial substrate use would be constrained by climate and plant functional traits, and (3) the relationships between soil microbial community and functions would demonstrate functional dissimilarity.
Soil samples were collected from four random plots in July and August 2013.The information of the sampling process are available in Xu et al. (2017).Briefly, we established four sampling plots measured 30 × 40 m and collected soil samples from a depth of between 0 and 10 cm at between 30 and 50 points in each plot along an S-shape.On return to the laboratory, the fresh soil samples were immediately sieved through a 2-mm mesh and subdivided into three subsamples.One subsample was stored briefly at 4°C until analysis for soil enzyme activities and soil pH.Another was stored briefly at −20°C until analysis for PLFAs and Eco-Biolog.The third was air-dried, sieved through a 0.25 mm mesh, and analyzed for soil nutrients.

Soil analysis
Soil pH was measured at a soil-to-water ratio of 1:2.5.Soil organic carbon (SOC) and total N (TN) concentrations were determined by dry combustion of ground samples (100-mesh) in a C/N analyzer (Elementar, Vario Max CN, Germany).Total phosphorus (TP) was determined with a flow injection auto-analyzer following digestion with H2SO4-HClO4 (Huang et al., 2011).After extraction with distilled water at a soil: distilled water ratio of 1:5, dissolved organic carbon (DOC) concentrations were determined by Liqui TOC II (Elementar, Liqui TOC II, Germany) (Jones and Willett, 2006).
Soil microbial biomass carbon (MBC) was measured using the chloroform fumigation and direct extraction technique (Vance et al., 1987).A conversion factor of 2.64 was used to convert extracted C to biomass C. The silt fractions (<53 μm) of the samples were separated by wet-sieving and then were freeze-dried in the laboratory, as described by Six et al. (2000).The soil properties are shown in Table 2.We followed the method described by Bå å th et al. (2003) for PLFA analysis and PLFAs are expressed in units of nmol g -1 .The four enzymatic activities of β-glucosidase (BG), Nacetylglucosaminidase (NAG), acid phosphatase (AP), and leucine aminopeptidase (LAP) responsible for soil C, N, and phosphorous cycling, were measured following the procedure outlined Biogeosciences Discuss., https://doi.org/10.5194/bg-2018-499Manuscript under review for journal Biogeosciences Discussion started: 24 January 2019 c Author(s) 2019.CC BY 4.0 License.
in Saiya-Cork et al. (2002) and are expressed in units of nmol h -1 g -1 .Information about PLFA and enzyme activities are presented in Table S1.

Vegetation data
As described by Xu et al. (2018), we collected litter and sun-exposed and mature leaves (leaf blades for trees) from between five and ten individuals of each plant species at each site and determined their TN and TC concentrations.We calculated the specific leaf area (SLA, the one-sided area of a fresh leaf divided by its oven-dried mass, m 2 kg -1 ), leaf dry matter content (LDMC, the oven-dried mass of a leaf divided by its water-saturated fresh mass, mg g -1 ), leaf C concentrations (leaf C, g kg - 1 ), and leaf N concentrations (leaf N, g kg -1 ) for ten fully expanded leaves of each sampled individual.
We also calculated the community-weighted means (CWM), as described by Garnier et al. (2004).
The diversity of the tree species and plant functional traits are summarized in Table S2.

Microbial substrate use
Microbial functional diversities were determined using a Biolog EcoPlate TM (Biolog Inc., Hayward, California, USA) as described by Garland and Mills (1991).Briefly, approximately 10 g of fresh soil was suspended in 100 ml saline solution (0.85% NaCl) and shaken on an orbital shaker for 30 min at 190 rpm.A 150 μl aliquot of supernatant from 1:1 000 dilutions of each soil sample was added to each well.The plates were incubated at 25°C, and the absorbance at 590 nm was measured using a microplate reader (GENios Pro TM , Tecan Trading AG, Mä nnedorf, Switzerland) every 24 h up to 240 h (0, 24, 48, 72, 96, 20, 144, 168, 192, 216, and 240 h).
The Richness (R), Shannon-Weiner diversity index (H′), Shannon evenness index (E), and Simpson dominance index (D) were calculated from the absorption values after EcoPlate TM incubation for 96 h (Gomez et al., 2006).Additionally, the 31 C sources were divided into six groups, namely carbohydrates, carboxylic acids, amines, amino acids, polymers, and miscellaneous, as suggested by Zak et al. (1994).The average absorbance of all C sources within each group was computed as the intensity of the single substrate use.The soil microbial metabolic intensities (S) were estimated by the area underneath AWCD vs. t, and were obtained by integrating the equation against time t (Guckert et al., 1996) where vi was average optical density of the ith incubation time.

SOM decomposition rate
Four replicates from each sampling site with a 60% water-holding capacity were incubated at 20°C .
In brief, 40 g of each fresh soil sample were put into a 150-ml incubation bottle, and the samples were then adjusted so that their moisture content corresponded to a water-holding capacity of 60%.
During the 4-week incubation period, the soil respiration rates were measured on days 1, 7, 14, 21, and 28 using an automatic system.The SOM decomposition rates were calculated as described in the study of Xu et al. (2015).

Statistical analysis
One-way analysis of variance (ANOVA) followed by a post hoc Tukey HSD test were used to test the significance of the differences among the soil properties, C use, functional diversity, and SOM decomposition rates in the different forest ecosystems.We tested the relationships between labile C, soil microbial community structure, microbial function, and the SOM decomposition rates with the Pearson correlation test.Differences were considered significant when P<0.05, with marginal significance set at P<0.01.All ANOVA and regression analyses were performed using SPSS 19.0 for Windows.Data are reported as the mean ± SE.
We used redundancy analysis (RDA) to examine the relationship between the environmental variables and soil microbial substrate use.The environmental variables were the same as those described in Xu et al. (2018), including climate, soil properties, litter properties, and plant functional traits.Before RDA, we conducted forward selection of the environmental variables that were significantly correlated with variations in the microbial substrate use profile using stepwise regression and the Monte Carlo Permutation Test.We used CANOCO software 4.5 (Ter Braak and Smilauer 2002) for the RDA and stepwise regression.The environmental properties, which were significantly correlated with the microbial substrate use in the RDA, were stressed in the plots.
3.1 Patterns in the microbial substrate use, soil labile carbon, and SOM decomposition rates Of the forests along the NSTEC, the C metabolic intensity of soil microbes was lowest in HZ and LS; the C metabolic intensity of soil microbes differed significantly between JF and the other forests (Fig. 2), which indicates that the color development was significantly higher in the tropical forest soils than in the subtropical and temperate forest soils and is consistent with the variations in the AWCD (Fig. S1).The average values of R, H′, and D were significantly different among the nine forest soils and were highest in JF, SN, and CB (Table 3).
Across the nine forests, soil microorganisms used the six substrate groups in the same order; the carboxylic acid substrate was used most, followed by amino acids, carbohydrates, polymers, amines, and miscellaneous substrates (Fig. 3).Microorganisms in the boreal and temperate forests mainly metabolized carbohydrates, amino acids, and carboxylic acids, while those from the subtropical and tropical forests used the substrates in equal proportions.
Overall, soil MBC concentrations in the boreal and temperate forests were three to eight times higher than those of the subtropical and tropical forests.In contrast, the average DOC concentration in the tropical and subtropical forest soils ranged from 311 to 458 mg kg −1 , which was significantly higher than the average concentration in the temperate and boreal forest soils, where the average concentrations ranged from 204 to 284 mg kg −1 (Table 2).The average SOM decomposition rates in the subtropical forests ranged from 0.64 to 2.42 μg C g −1 d −1 , and were significantly lower than the rates in the temperate forests, which ranged from 3.43 to 4.61 μg C g −1 d −1 (Table S3).

Effect of environmental properties on soil microbial substrate use
Redundancy analysis showed that the variations in soil microbial substrate use were strongly and positively correlated with the CWM values of LDMC, leaf N, and leaf C, and strongly and negatively correlated with the soil silt content and SMC (Fig. 4).The RDA2 of soil microbial substrate use was strongly positively correlated with TN and SOC, but negatively correlated with mean annual precipitation (MAP) (Fig. 5).RDA1 mainly represented the plant functional traits, soil texture, and micro-meteorological conditions, while RDA2 represented climate and soil nutrients.Overall, the soil silt content and the CWM values of plant functional traits were the main predictors of the  Miscellaneous substrate use was positively related with fungal biomass and G + /G − bacterial biomass (Fig. 5).
The abundance of G − bacteria was positively associated first with the specific activities of BG, whereas actinomycetes and G + bacteria were positively associated with BG and LAP.Soil fungi were negatively associated with BG (Fig. 5).

Relationships between SOM decomposition rate, PLFAs, enzyme activity, and microbial metabolic activities
The SOM decomposition rates were significantly and positively related to soil MBC concentrations but significantly and negatively related to soil DOC concentrations (Fig. 6a and b).Except for amino acid and amine substrates, the SOM decomposition rates were significantly and positively related to microbial metabolic activities (AWCD) and carbohydrate substrate use (Fig. 6c and d) and negatively related to carboxylic acid, polymer, and miscellaneous substrate use (Fig. 6e, g, and i).
The SOM decomposition rates were significantly and positively correlated with total PLFAs (r=0.456,P=0.005), bacteria (r=0.3836,P=0.021), actinomycetes (r=0.500,P=0.002), and G − bacteria PLFAs (r=0.520,P=0.001) (Fig. 7a, b, d, and f) but were negatively correlated with fungal PLFAs (r=−0.370,P=0.026), F/B (r=−0.513,P=0.001), and the G + /G − (r=−0.496,P=0.002) (Fig. 7c,    g, and h).Except for LAP activity, soil enzyme activities were significantly and positively correlated with the SOM decomposition rates (P<0.01)(Fig. 7i, j, and l).Soil organic matter is one of the most important C pools in terrestrial ecosystems.The concentrations of soil DOC in the temperate forests were lower than those in subtropical forests but soil MBC concentrations were higher in temperate forests than in subtropical forests.This reflects the results of previous regional and global studies (Tian et al., 2010;Xu et al., 2013), and shows that the production/consumption ratio of soil DOC was lower, but that microbial C immobilization was higher, in the high latitude forests (Fang et al., 2014).Soil DOC, as a labile SOM fraction with a rapid turnover, is one of the primary energy sources for microorganisms.The higher temperatures and precipitation in subtropical and tropical forests lead to higher turnover rates (Fang et al., 2014), so soil DOC concentrations were highest in subtropical, and MBC concentrations were lowest, in tropical forests.However, in temperate forests, more C is assimilated into microbial biomass, so that less C is lost through chemical and physical processes (Liu et al., 2010).Also, because the decomposition ability of different microbe groups varies, the differences in the soil microbial communities in different forest ecosystems may also be responsible for the spatial variations in the soil DOC and MBC concentrations along the NSTEC (Hagedorn et al., 2008).
Heterotrophic soil respiration is sustained by the decomposition of SOM.The SOM decomposition rates along the NSTEC were greater in temperate forests than in subtropical forests, which was consistent with the variations in the soil MBC and SOC concentrations.These results indicate that, as found in other studies, large scale SOM decomposition rates are driven by the amounts of substrate available (Yu et al., 2010).Changes in the availability of C in SOM may affect the microbial resource strategies, which may in turn influence the SOM decomposition rate.

Latitudinal variation in microbial substrate use
The AWCD reflects the sole C source use ability of the soil microbial community (Garland and Mills, 1991).Of the six groups of C substrates, microbial communities in the temperate forests mainly used carbohydrates, carboxylic acids, and amino acids, which suggests that microorganisms in temperate forests probably use high-energy substrates that degrade easily (Kunito et al., 2009).The latitudinal pattern of soil microbial C substrate use was mainly related to the soil silt contents and the CWMs of LDMC, leaf C, and leaf N concentrations, indicating that the quality of nutrients from plant inputs had a major influence on microbial carbon use efficiency.Plant species with high SLA, high N Biogeosciences Discuss., https://doi.org/10.5194/bg-2018-499Manuscript under review for journal Biogeosciences Discussion started: 24 January 2019 c Author(s) 2019.CC BY 4.0 License.concentrations in leaves, and low LDMC can result in bacterial-dominated soil microbial communities in grasslands (Orwin et al., 2010).Looking beyond individual traits, related tree species may cultivate microbial communities with similar preference for carbon sources through the coevolution of plants and microbes (Liu et al., 2012;Buscot, 2015).
As hypothesized, the soil microbial community composition was explained by the CWMs of plant traits at the regional scale.Carbon substrate use was negatively correlated with leaf N concentrations (Table S2).Bacterially dominated soil microbial communities develop from leaf litter comprised of N-rich leaves from fast growing species (De Vries et al., 2012), while leaves with low N concentrations will promote fungal domination (Orwin et al., 2010;De Vries et al., 2012).In line with this, fungal biomass decreased, and bacterial biomass increased, as the CWM leaf N content increased, and is associated with fast-growing, N-exploitative plants (Xu et al., 2018).Leaf N concentrations are considered as indicators of plant growth and resource uptake (Wright et al., 2004).
The results from this study show that, along the NSTEC, high leaf N restrained microbial C substrate use and was a good indicator of the competition between plants for soil N (Pei et al., 2016).Soil microbes and nearby plants may have been competing for N in the soil.
We also found that the C substrate use was negatively correlated with the leaf C concentrations (Table S2).High latitude plants may have higher leaf C concentrations than plants at lower latitudes so that they can balance the osmotic pressure of cells and resist freezing (Millard et al., 2007;Hoch and Körner, 2012).The increased C was most likely in the form of an increase in non-structural C, including starch, low molecular weight sugars, and storage lipids that are easy to break down.Plant functional traits play an important role in shaping soil microbial communities (Pei et al., 2016), so soil microorganisms from the temperate forests mainly metabolized high-energy substrates (carbohydrates, carboxylic acids, and amino acids).
The LDMC is the ratio of the leaf dry weight to the fresh weight and has been used as a proxy for the ratio of structural compounds to assimilatory tissue (mesophyll and epidermis, Van Arendonk and Poorter, 1994).High values of LDMC indicate large amounts of vascular tissue, cellulose, insoluble sugars, and leaf lignin that are difficult to decompose (Poorter and Bergkotte, 1992); C substrates such as carbohydrates, carboxylic acid, and amino acid are, however, easy to decompose (Myers et al., 2001).In line with this, the use of carbohydrate, carboxylic acid, and amino acid Biogeosciences Discuss., https://doi.org/10.5194/bg-2018-499Manuscript under review for journal Biogeosciences Discussion started: 24 January 2019 c Author(s) 2019.CC BY 4.0 License.
substrates was negatively related to the CWMs of the LDMC (Table S2).Pei et al. (2016) reported that the LDMC was an important driver of multivariate soil microbial community structure and G − bacterial abundance.
Soil texture regulates soil biological processes and so affects the soil microbial community structure (Sessitsch et al., 2001).In the present study, microbial C substrate use was significantly and positively related to the soil silt content.Soil types and textures varied along the NSTEC.Soil texture influences how microbes use organic matter, and has a strong influence on soil moisture, nutrient availability, and retention (Veen and Kuikman, 1990).Fine-textured soils with a higher silt content are known to be more favorable for bacterial growth than soils with a lower silt content because of their greater water-holding capacity and nutrient availability, and because they are better protected from bacterial grazers (Carson et al., 2010).We found that the microbial C substrate use was higher in LS, CB, SN, and JL than in the other forests, reflecting their fine-grained soils and high silt contents, which ranged from 60% to 80%.

Links between soil microbial community structure and function
The quality and changes in the amounts of SOM are influenced by the biomass, vegetation coverage, root distribution, microbial specie (Raich and Schlesinger, 1992).The SOM decomposition rates were higher in temperate forests than in tropical forests and may reflect the higher soil microbial biomass (Wang et al., 2016).In line with this, SOM decomposition rates were positively related with soil MBC concentrations and different groups of PLFAs.The inverse relationships between SOM decomposition rates and DOC, and between SOM decomposition rates and the use of some individual C substrates along the NSTEC, indicate a shift in the soil C turnover from open to closed with increases in the soil labile C concentrations.Further, soil nutrients have a strong influence on the spatial patterns of soil microbial communities.Thus, soil DOC and MBC do not influence SOM decomposition rates directly, but indirectly by regulating microbial properties (Boberg et al., 2014;Wei et al., 2014).Because different communities of microbes have different SOM use efficiencies (Balser and Wixon, 2009;Lipson et al., 2009;Monson et al., 2006), changes in the microbial community structure may influence the decomposition rates of organic matter (Lipson et al., 2009;Keiblinger et al., 2010).
Shifts in microbial community composition may influence enzyme production (DeForest et al., 2012;Waldrop et al., 2000;Brockett et al., 2012).Different microbial groups require different amounts of nutrients to construct biomass, or have enzymes that differ in their affinity for nutrients.
For example, fungi tend to have higher C/N or C/P ratios while heterotrophic bacteria typically have lower C/N or C/P ratios (Sterner and Elser, 2002).We found that the relative abundances of the G + bacteria and actinomycetes communities were associated with the specific activities of hydrolytic enzymes involved in C and N acquisition (BG and LAP), whereas the relative abundance of the G − bacteria was correlated with soil NAG activities involved in chitin degradation.Waldrop et al. (2000) found that phosphatase activity was significantly correlated with the abundance of various bacterial PLFAs.Soil BG was mainly responsible for cellulose degradation and was involved in breaking down complex organic compounds (cellobiose) into small molecule substrates (glucose) in favor of acquiring C through microbial community growth.Other studies have found that G + bacteria were positively correlated with the cellobiohydrolase that was responsible for degrading complex compounds (Waldrop et al., 2000).Fungi are commonly considered as producers of oxidative enzymes.Therefore, the influence of fungal biomass on variations in enzyme activities was minimal (Kivlin and Treseder, 2014).The linkages between enzyme activity and community composition may provide some insight into the microbial mechanisms that drive the decomposition of macromolecular C compounds.
The soil microbial community structure and functions were significantly correlated along the NSTEC.Soil carbohydrate and polymer substrate use were mainly related to soil G + bacterial and actinomycic biomass, but amines and miscellaneous substrates were mainly related to soil G − bacterial, fungal biomass, and the F/B ratio.Soil bacteria mainly decomposed simple carbohydrates, organic acids, and amino acids, whereas soil fungi mainly decomposed recalcitrant compounds (Myers et al., 2001;Treonis et al., 2004).Shifts in the microbial community composition may influence enzyme production if microbial groups need nutrients at lower concentrations to construct biomass, or have enzymes that differ in their affinity for nutrients.In agreement with our study, numerous other researchers have reported significant correlations between PLFA profiles and enzyme activities (DeForest et al., 2012;Brockett et al., 2012;Riah-Anglet et al., 2015).Soil BG and AP activities were positively related with bacterial and actinomycic biomass and negatively related with Biogeosciences Discuss., https://doi.org/10.5194/bg-2018-499Manuscript under review for journal Biogeosciences Discussion started: 24 January 2019 c Author(s) 2019.CC BY 4.0 License.fungal biomass.Soil NAG activities were weakly and positively related with fungal biomass in the present study, and may have been mainly produced by fungal populations (Valášková et al., 2007).
These results suggest that that overall ecosystem functioning may suffer if soil microbial groups are lost, which confirms the functional dissimilarity hypothesis.However, to gain an improved understanding of the mechanisms that drive these relationships, we need to carry out further studies with different experimental techniques.

Conclusions
In this study we examined the patterns in labile C concentrations, SOM decomposition rates, microbial substrate use, and functional diversity and identified a combination of abiotic and biotic factors that influenced soil microbial functional diversity at the regional scale.The MBC concentration and SOM decomposition rates were significantly lower, and the soil DOC concentrations and microbial metabolic activities were higher, in the subtropical and tropical forests than in the temperate forests.For the first time, we showed that, along with the soil silt content, CWM plant traits explained variations in soil microbial C substrate use at the regional scale.Soil microbial community structure and function were strongly related, which suggest that the loss of soil microbial groups may have consequences for overall ecosystem functioning, which confirms the functional dissimilarity hypothesis.Different lowercase letters indicate significant differences among forests in the same climate zone.The abbreviations of the sampling sites are given in Table 1.

Supporting information
Table S1 Average values of forest soil enzyme activities and different PLFA groups along the NSTEC.

Table S2 Plant diversity and community weighted means of plant functional traits
Table S3 Soil organic matter (SOM) decomposition rates during the28 days of incubation time (Mean±SE) (μg C g −1 d −1 ) Fig. S1 Variations in the average well color development (AWCD) values during a 240-h incubation for the nine forests.The abbreviations of the sampling sites are the same as those in Table 1

3. 3
Relationships between soil microbial substrate use, enzyme activities, and PLFAs Microbial carbohydrate use was positively related with bacterial biomass and actinomycic biomass (Fig.5).Microbial polymer use was negatively related with bacterial biomass and actinomycic biomass.Microbial amines use was negatively related with G − bacterial and fungal biomass.
of soil labile C and SOM decomposition rates to variations in forest type Biogeosciences Discuss., https://doi.org/10.5194/bg-2018-499Manuscript under review for journal Biogeosciences Discussion started: 24 January 2019 c Author(s) 2019.CC BY 4.0 License.

Fig. 2 .
Fig.2.Variations in soil microbial substrate use during a 240-h incubation for the nine forests.Different colors represent different forest types: Yellow, coniferous forest; Dark yellow, coniferous broad-leaved mixed forest; Purple, deciduous broad-leaved forest; Olive, subtropical evergreen broad-leaved forest; Orange, Tropical monsoon forest.

Fig. 4 .
Fig.4.Redundancy analysis (RDA) ordination biplot of soil microbial carbon sources use efficiency and environmental properties.The representatives of different colors were showed in Figure 2. The dotted lines and solid lines represent the environmental variables and lipid signatures and carbon sources.The abbreviations of the variables in this figure are as follows: MAP, mean annual precipitation.The vegetation data: LDMC, leaf dry matter weight;Leaf C, leaf carbon content; Leaf N, leaf nitrogen content; SLA, specific leaf area.Soil properties included SMC, soil moisture content; Silt, soil silt content; TN, soil total nitrogen; SOC, soil organic carbon.The abbreviations of the sampling sites were given inTable 1

Figure 5 .Figure 6 .)
Figure 5.The heatmap of the Pearson's correlation coefficients between the use of individual substrates and microbial PLFAs and soil enzyme activities.Note: The abbreviations of the variables:

Figure 7 .
Figure 7. Relationships between soil carbon mineralization rates (μg C g −1 d −1) and different groups of soil microbial PLFAs (a-h) and enzyme activities (i-l).

Table 1 .
The main characteristics of the sampling sites along the North South Transect of East China Biogeosciences Discuss., https://doi.org/10.5194/bg-2018-499Manuscript under review for journal Biogeosciences Discussion started: 24 January 2019 c Author(s) 2019.CC BY 4.0 License. 620

Table 3 .
Functional diversity of soil microbial communities in forest ecosystems along the NSTEC Indices were calculated based on the optical density values after incubation for 96 h.Data are expressed as 625 means±standard errors.Different lowercase letters indicate significant differences among forests.The abbreviations626of the sampling sites are the same as those used in Table1.
24 633 Figure 2. Variations in soil microbial substrate use during a 240-h incubation for the nine forests.Different colors 634 represent different forest types: Yellow, coniferous forest; Dark yellow, coniferous broad-leaved mixed forest; 635 Purple, deciduous broad-leaved forest; Olive, subtropical evergreen broad-leaved forest; Orange, Tropical monsoon 636 forest.Different lowercase letters indicate significant differences among forests in the same climate zone.The 637 abbreviations of the sampling sites are given in Table 1.