Belowground carbon pools and dynamics in China’s warm temperate and sub-tropical deciduous forests

We report the first estimates of pools and dynamics of microbes, roots, plant litter and soil organic carbon (SOC) in three dominant types of China’s vast deciduous forest area:Betula platyphylla, Quercus liaotungensis, andQuercus alienavar. acuteserrata. Organic matter degradation rates overshadowed litter inputs as the main determinant of the soil carbon stocks. Across the three forests, rates of litter decomposition were also indicative for turnover rates of SOC. Litter and SOC decay was faster in the subtropical than in the warm-temperate forests. Among the latter, SOC turnover was highest in the forest producing the higher-quality litter. Microbial biomass was, as expected, correlated with SOC content. Microbial activity, in contrast, was highest at the sub-tropical forest, despite the lower SOC availability, lower fraction of labile SOC, and lower soil microbial biomass. These results may contribute to increased understanding of controls over belowground carbon cycling in deciduous forests.


Introduction
With the significant increase in atmospheric greenhouse gas concentrations and the potential for global climate change, studies of the terrestrial carbon cycle have gained attention over the last 20 years (Houghton et al., 2001;Callesen et al., 2003).Soil carbon is an important terrestrial carbon reservoir and plays a key, yet poorly understood role in the global carbon cycle and its feedback to climate change (Post et al., Correspondence to: C. W. Xiao (cwxiao@ibcas.ac.cn) 1982; Davidson and Janssens, 2006).Therefore, the study of soil organic carbon (SOC) dynamics is critically important to our ability to understand the global carbon cycle and its response to future global change (Davidson et al., 2000).
In this study, we compared pools and dynamics of fine roots, soil carbon pools, and soil microbes among Asia white birch, East-Liaoning oak, and Sharptooth oak forests.We applied density fractionation (e.g.Elliott and Paustian, 1996;Zimmermann et al., 2007) to soil samples of the different forests to address the pool sizes and kinetics of fast-and slow cycling organic matter pools.The overall objective of this study was to examine soil carbon quantity and quality in these important forest types.The specific objectives of this study were: (1) to determine the total SOC pool and its components in the three forest types, and (2) to determine the rates of carbon cycling through the litter and SOC pools.
Published by Copernicus Publications on behalf of the European Geosciences Union.C. W. Xiao et al.: Belowground carbon pools and dynamics in deciduous forests 2 Materials and methods

Site description
Three deciduous broad-leaved forest types were assessed in this study: (1) Asia white birch forest; (2) East-Liaoning oak forest; and (3) Sharptooth oak forest.All three forests were growing on clay-poor soils (less than 5%), with pH values varying between 6.5 (Sharptooth oak) and 6.9 (Asia white birch and East-Liaoning oak).Similar clay contents and pH values across the three sites are favourable for comparisons of SOC pools, although we acknowledge that different clay mineralogies may exhibit different SOC stabilization potentials (Six et al., 2000).The study sites of Asia white birch and East-Liaoning oak were located in the Donglingshan Mountains, Beijing (39 • 48 -40 • 00 N, 115 • 24 -115 • 36 E).These two sites were situated in the warm temperate climate zone, and the two sites were characterized by a warm temperate, semi-wet monsoon climate.Long-term mean annual precipitation in this area was 612 mm and mean annual air temperature was 4.8 • C. The soil in both stands was classified as a Eutric cambisol (FAO-WRB, 1998), with a depth of about 60 cm.The Sharptooth oak forest was located in the Shennongjia Mountains, Hubei Province (31 • 15 -31 • 57 N, 109 • -110 • 58 E).This forest was located in the sub-tropical zone, and was characterized by a sub-tropical monsoon climate, with mean annual precipitation of 1514 mm and mean annual air temperature of 10.6 • C. The sandy-loam soil in this forest is classified as a Haplic cambisol (FAO-WRB, 1998) with a depth of some 100 cm.
A 0.25 ha plot of 60-year-old Asia white birch forest (39 • 57 01 N, 115 • 25 07 E, elevation 1380 m a.s.l.) was selected for this study.The inclination of the site was 32 • .The Asia white birch forest was dominated by Asia white birch, admixed with associated species (Betula utilis and Populus alba), and an abundance of shrubs including Sorbus pohuashanensis, Lonicera japonica, Prunus armeniaca, Corylus mandshurica, Acer mono, Abelia biflora, Leptodermis oblonga, Spiraea sargentiana, Macrocarpium officinalis.Tree density at the plot was 1234 trees ha −1 , with a mean diameter at breast height (DBH) of 13.2 cm and a mean tree height of 8.5 m.
We also selected a 0.25 ha plot of 60-year-old East-Liaoning oak forest (39 • 57 04 N, 115 • 25 04 E, elevation 1200 m a.s.l.), with an inclination of 28 • .This East-Liaoning oak forest was dominated by East-Liaoning oak, and admixed with B. utilis as associated tree species and some shrubs (S. sargentiana, A. mono, Lespedeza bicolor, L. japonica, C. mandshurica, and Deutzia scabra).Tree density was 1262 stems ha −1 , with a mean DBH of 12.2 cm and a mean tree height of 6.8 m.
Primary forests of Asia white birch and East-Liaoning oak have been intensely disturbed by human activities and disappeared completely.The contemporary Asia white birch and East-Liaoning oak forests are secondary and are currently protected and naturally regenerating (Chen, 1997).Sharptooth oak forests were much less disturbed by human activities, and although our study site was not a primary forest, it has been less intensively managed/disturbed than the other two study sites.For all three species, leaves tend to appear by the end of April, and most of the litterfall occurs between early September and end of October.

Soil analyses
For the determination of bulk density, five soil cores were taken at different depths (0-5, 5-15, 15-30, 30-45, 45-55 cm) in all plots in May 2006.A special coring device for the determination of bulk density (volume = 100.0ml) was applied.In July 2006, surface organic horizon mass was quantified with a metal cylinder inserted down to the mineral soil (n=5).Five soil columns were collected in each plot for the determination of light fraction organic carbon (LF-OC), heavy fraction organic carbon (HF-OC) and total SOC.These samples were randomly taken by coring with a sharp-edged metal cylinder with an inner diameter of 3 cm and a length of 10 cm.Samples were separated according to depth (0-5, 5-15, 15-30, 30-45, 45-55 cm) and the fresh samples were passed through a 2-mm sieve and manually cleaned of any visible plant tissues.Density fractionation of SOC physically separates soil into low-and high-density fractions, referred to as LF-OC and HF-OC.LF-OC is commonly referred to as a plant-derived and less stable fraction with high C concentration (Golchin et al., 1994).HF-OC is assumed to be a more stable and high-density organo-mineral fraction, having lower C concentrations (Golchin et al., 1995).The LF-OC was determined using the density fractionation method (Sollins et al., 1984).Air-dried soils were passed through a 2 mm mesh sieve and 5.0 g (dry weight equivalent) of airdried soils was transferred to a tube and dispersed in 20 mL of NaI solution adjusted to a density of 1.7 gmL −1 .The suspension in the tube was shaken thoroughly for 15 min, and after standing overnight, separated light and heavy fractions.The light fractions at the surface of the density liquid were aspirated, and trapped onto a membrane filter paper (Whatman, Grade 1:11µm), rinsed with deionized water, and then oven-dried at 50 • C and weighed.Total SOC and LF-OC were determined with the dichromate oxidation method (Lovell et al., 1995).Briefly, 0.2 g of ground soil was digested with 5 ml of 2 M K 2 Cr 2 O 7 and 5 ml of concentrated H 2 SO 4 at 170 • C for 10 min, followed by titration of the digests with 2 M standardized FeSO 4 .The HF-OC was determined by subtracting LF-OC from the total SOC.

Soil microbial biomass carbon and activity
Five soil core samples per plot were randomly collected for determination of soil microbial biomass carbon (SMB-C) and soil microbial activity (SMA) in May, July and September 2006.Samples were taken from the 0-15 cm mineral soil layer using a sharp-edged metal cylinder with an inner diameter of 10 cm and a length of 15 cm.Each sample was labeled, and then stored at 2 • C in a cooler for transport to the laboratory.In the laboratory, the fresh samples were passed through a 2-mm sieve and manually cleaned of any visible plant tissues.
Soil microbial biomass carbon (SMB-C) was measured using the chloroform fumigation-extraction method (Vance et al., 1987).Twenty grams (dry weight equivalent) of fumigated and non-fumigated soil samples were extracted with 0.5 M K2SO4.Extracts were filtered through 0.45-µm filters and frozen at −20 • C before analysis of extractable carbon by dichromate digestion as described by Lovell et al. (1995).SMB-C was calculated as the difference in extractable carbon of fumigated and non-fumigated soil samples.To correct for incomplete extractability, a conversion factor (Kec) of 0.38 was used to obtain SMB-C (Vance et al., 1987).
Soil microbial activity (SMA), i.e. soil microbial respiration, was estimated by determining CO 2 evolution over a 2week incubation period.First, 20.0 g (dry weight equivalent) of soil was brought to 60% of the water holding capacity and incubated at 25 • C for 2 weeks.Respired CO 2 was captured in 5.0 ml of 0.5 M NaOH suspended inside a Mason jar, and the NaOH solution was subsequently titrated to determine the amount of CO 2 evolved (Hu and van Bruggen, 1997).

Forest floor mass and litterfall
The three forests exhibited a moder type of litter layer (Müller, 1889).Forest floor mass was measured on five randomly located, 0.5 m×0.5 m subplots from each plot in May, June, July, August, September, and October 2006 and in April 2007.Forest floor mass was sorted into coarse woody debris and surface organic matter.Litterfall was trapped and collected in May, June, July, August, September, and October 2006 and in April 2007 using five randomly located 0.45 m×0.35 m rectangular baskets, and sorted into woodyand non-woody fractions.Dry litter mass was determined after oven-drying at 75 • C for 2-3 days.

Fine root biomass, production and turnover rate
Fine root (<2 mm) biomass was determined in each plot by core sampling (Roberts, 1976) to a depth of 55 cm in May, July and September 2006.At each sampling date, 10 sample columns were randomly excavated using a sharp-edged metal cylinder with an inner diameter of 10 cm and a length of 20 cm.Samples from different depths (0-5, 5-15, 15-30, 30-45, 45-55 cm) were separated and labeled.Fine roots were manually removed from the soil samples and washed.Live and dead root fragments were subsequently separated by visual inspection.The xylem of dead roots looks darker and deteriorated, the degree of cohesion between the cortex and the periderm decreases, and root tips become brittle and less resilient (Janssens et al., 2002).Dry biomass was determined after oven-drying at 75 • C for 2-3 days.
Fine root (<2 mm) production during the growing season was estimated with a modified in-growth core technique (Lund et al., 1970).The 10 holes created by the root biomass in each plot were refilled early May 2006 with native soil obtained from the root biomass experiment and their boundaries were marked with sticks.The in-growth cores were harvested at the end of October 2006.Soil samples from different depths (0-5, 5-15, 15-30, 30-45, 45-55 cm) for each in-growth core were labeled, and fine root biomass was subsequently estimated using exactly the same procedures as described above.Total fine root production was estimated as the sum of live and dead roots present in the in-growth core in end October 2006.
Fine root turnover rate is defined here as the rate of the total amount of fine root produced in the growing season over the mean standing biomass of fine roots (Aber et al., 1985).Mean fine root biomass was estimated as the average of live root biomass on May, July and September 2006.

Decomposition of leaf litter, fine roots, LF-OC, HF-OC and SOC
Decomposition rates of leaf litter and fine root, LF-OC, HF-OC and SOC were determined using the nylon bag (or litterbag) method (Wen, 1984;Lin et al., 1992Lin et al., , 2005;;Arunachalam et al., 1996).Recently fallen leaves, fine root and soil from the 0-10 cm mineral soil layer were collected from the forests.The fresh soil samples were first passed through a 2-mm sieve.Each nylon bag had a dimension of 10 cm×15 cm and a mesh of 1 mm for leaf litter and fine roots, and of 48 µm for soil.40 nylon bags for Asia white birch and East-Liaoning oak forests and 35 nylon bags for Sharptooth oak forest containing 3 g of air-dried leaves and fine roots, and 100 g of air-dried soil were placed in nylon bags and the edges heat-sealed, respectively.In the Asia white birch and East-Liaoning oak forests these nylon bags were inserted on 2 May and collected after 0, 30, 59, 91, 123, 151, 179 and 365  121, 174, 365 days.Five nylon bags were collected at each sampling date.Mass of leaf litter and fine roots in each nylon bag was determined after oven-drying at 75 • C for 2-3 days.LF-OC, HF-OC and total SOC of soil in each litterbag were determined using the density fractionation method described above.
At the onset of the decomposition experiments, we also determined total C and N, lignin and soluble phenol of leaf and fine root material, and total N of LF-OC, HF-OC and of total SOC.Total C was determined by the standard method of wet-combustion, and total N by semi-micro Kjeldahl method (Bao, 1999).Lignin was determined with the thioglycolic acid method (Dean, 1997).Soluble phenol concentrations were analyzed using a combination of methanol extraction and the Folin-Ciocalteau assay (Waterman and Mole, 1994).Initial chemical characteristics for the substrates used in the decomposition studies are shown in Table 1.

Statistical analysis
Data management and statistical analyses were performed using SPSS software (SPSS, Chicago, IL).The decay constant (K) and the average rate of litter loss were determined by fitting the following exponential function: X t = X 0 e −kt (Olson, 1963).One-way ANOVA was used to test for significant differences of initial chemical content of leaf litter, fine root, and LF, HF and total soil, soil bulk density, surface organic, coarse woody debris, LF-OC, HF-OC and total SOC, fine root biomass, production and turnover rate.Repeated Measures Analysis of Variance was used to detect the significant differences of seasonal variation of forest floor mass, SMB-C, SMA and fine root biomass.Multiple comparisons were also performed to permit separation of effect means using a least significant difference test at a significance level of P <0.05.

Soil carbon pools
The seasonal evolutions of the forest floors in the three investigated forests were relatively similar (Fig. 1), exhibiting a continuous slow decrease until October, when annual leaf litterfall commenced.Nonetheless, total forest floor mass was significantly (P <0.05) higher in the temperate Asia white birch and East-Liaoning oak than in sub-tropical Sharptooth oak forest (Table 2); a difference that was mainly related to differences in the non-woody fraction of the surface organic horizon (Fig. 2).There was no significant difference in coarse woody debris among the three forests (Fig. 2).In the Asia white birch and East-Liaoning oak forest, coarse woody  debris comprised about 30% of the forest floor mass, whereas in the less disturbed Sharptooth oak forest coarse woody debris accounted for 36% of the forest floor mass.In contrast to the non-woody fraction of the forest floor, seasonal fluctuation of woody debris was very little (data not shown).
In the various soil layers down to 55 cm, LF-OC, HF-OC and total SOC differed significantly among the studied forests (P <0.05, Fig. 2;  oak forests.Although LF-OC and HF-OC differed among these forests in parallel to total SOC, the relative differences in LF-OC were more pronounced than those in HF-OC (the proportion of labile components was lower where total SOC was lowest; Table 2).In absolute terms, labile C amounted to 3.2 g C m −2 in Asia white birch, 2.4 g C m −2 in East-Liaoning oak and 2.1 g C m −2 in Sharptooth oak.In relative terms, the difference in the labile C pools thus amounts to 50%.The chemical composition of the various SOC pools also differed among forest types (Table 1).If C:N ratio, lignin:N ratio, or soluble phenolics concentrations are used as proxies determining the chemical recalcitrance to microbial decay (Swift et al., 1979;Melillo et al., 1982), then the East-Lioning oak forest produced the most recalcitrant organic matter and the Asia white birch the least recalcitrant.

Microbial biomass and activity
In accordance with the SOC availability, mean SMB-C of the two temperate forests was significantly higher than that of the sub-tropical Sharptooth oak forest (P <0.05, Fig. 3a).In contrast, SMA exhibited exactly the opposite trend, and this throughout the entire growing season (P <0.05, Fig. 3b).In all three forests, both SMB-C and SMA were significantly higher in July than in May and September 2006 (P <0.05, Fig. 3a and b).
The mass loss patterns of decomposing leaf litter, fine roots, LF-OC, HF-OC and total SOC are shown in Fig. 4. Fine roots decomposed fastest (42-58% mass loss per year; Table 3), followed by leaf litter, LF-OC, Total OC, and last HF-OC that decomposed with an annual mass loss of 4.1-5.5%.Across all litter and SOC types, the decay constant and mass loss rates decreased from Sharptooth oak, Asia white birch to East-Liaoning oak.Differences in decomposition rates were, however, significant only for leaf litter mass, fine root mass, LF-OC and total SOC, and not for HF-OC (Table 3, Fig. 4).

Fine root biomass and production
As with microbial biomass, integrated fine root biomass in the soil layers down to 55 cm was significantly higher in July than in May and September (P <0.05, Fig. 5) in all three forests.Statistically significant differences in fine root biomass among the forests occurred in the 0-35 cm soil layers in May, and in the 5-25 cm soil layers in July and September (P <0.05, Fig. 5).
Integrated over all depths and averaged over the growing season, mean fine root biomass was 0.33±0.02,0.30±0.02and 0.26±0.01kg C m −2 in Asia white birch, East-Liaoning oak and Sharptooth oak forests, respectively.However, it should be kept in mind that in the first two forests, almost the entire soil profile was sampled, whereas in the Sharptooth oak forest, where soil depth was around one meter, total fine root biomass was underestimated by sampling only to a depth of 55 cm.Fine root production in the 0-55 cm soil layer decreased from Sharptooth oak, Asia white birch to East-Liaoning oak forests, and significant differences were observed in the 0-25 cm soil layers (Fig. 6).Integrated over all depths, fine root production was lowest in East-Liaoning oak forests (Table 2).

Soil carbon inputs and their residence times
The seasonal patterns of litterfall were very similar among the different forests, with the majority of the annual litter production in September and October.Seasonal fluctuations in branch litterfall were, however, very little.Annual aboveground litter inputs were significantly (P <0.05) higher in Asia white birch and Sharptooth oak forests than in the East-Liaoning oak forest (Table 2).In agreement with above- ground litterfall, below-ground litter production, estimated as being equal to fine root production (under the assumption of interannual steady state in fine root biomass), was also higher in Asia white birch and Sharptooth oak than in East-Liaoning oak (Table 2).The quality of leaf litter for decomposition decreased in the same order as the quantity of litter inputs.For every measured proxy for the quality of leaf litter for decomposition (soluble phenolics, C:N ratio, lignin:N ratio, lignin content), East-Liaoning oak exhibited the lowest quality litter and Asia white birch the highest (Tables 1 and 3).In contrast to leaf litter, however, fine root quality differed only very slightly among the three tree species.
The residence time of the surface litter inputs in the forest floor (calculated as the ratio of the maximum carbon content in October over the carbon loss between October and September), is much shorter in the sub-tropical Sharptooth oak forest (3.6 years) than in the Asia white birch forest (4.3 years).The East-Liaoning oak forest (5.4 years) has the longest residence time in the forest floor.This pattern is also obtained when the forest floor residence time is calculated from the ratio of forest floor mass over leaf litter inputs.According to this computation, the Sharptooth oak forest floor exhibited a mean residence time of less than three years, whereas carbon resides for more than four years in the forest floor of the East-Liaoning oak forest.
When considering the total unprotected SOC (surface litter+LF-SOC), the residence times calculated as the ratio of the carbon stock over the total litter inputs (above+root litter inputs) follow the same pattern as those in the surface layer.According to these calculations, labile carbon resides in the litter and LF-OC for slightly more than 4 years in the Sharptooth oak forest, up to 6.5 years in the East-Liaoning oak forest, with Asia white birch as an intermediate (5.8 years).(n=5).Different letters within a soil layer are significantly different (P <0.05) according to the least significant difference test.Absence of letters implies that no significant differences were detected.

Decomposition of various SOC types
Decomposition of litter and SOC is an important process contributing to carbon and nutrient cycling (Vogt et al., 1991;Christensen, 2001;Cornelissen et al., 2007) and is mediated primarily by climate and organic matter quality (Harmon et al., 1990;Sinsabaugh et al., 2002;Fioretto et al., 2007).We assessed decomposition rates by means of litter bags.Litter bag results are prone to errors (both overestimations and underestimations are possible; Swift et al., 1979).One of these errors is the loss or dissolved or particulate through the mazes in the mesh bags.This may be especially relevant in our study, because the Sharptooth oak forest was exposed to much higher precipitation than the other two forests.The reader should bear this potential pitfall in mind, because if important, this would have influenced the calculated turnover rates reported in the results section.
In all three forests, the quality of the SOC for microbial decay decreased in the sequence fine roots, leaf litter, LF-OC and HF-OC, as indicated both by decay constants as by the chemical analyses.Moreover, independent of the type of SOC (from fresh litter to HF-SOC), Sharptooth oak always exhibited the fastest decay rates and East-Liaoning oak the slowest.Thus, it appears that the decomposability of the deposited litter is a proxy for the decomposability of the SOC derived from it.It can also not be excluded that the priming mechanism (i.e. a stimulation of decomposition of more recalcitrant material by the addition of labile substrates; Kuzyakov et al., 2000) was more pronounced where the deposited litter was most labile (Subke et al., 2004).Although priming could explain why the order of decay con- stants across sites is sustained with depth or with substrates of varying recalcitrance, it can not explain why also the chemical proxies varied accordingly.Hence, priming may contribute, but cannot be the sole process explaining why SOC decomposition is highest where litter decomposition is highest.
In agricultural systems, turnover time of LF-OC typically varies from a few months to a few years, while in natural ecosystems turnover times of LF-OC may amount to decades and centuries, depending on local conditions (Post and Kwon, 2000;Leifeld et al., 2009;Schulze et al., 2009).In contrast, HF-OC is stabilized through mineral surface interactions and micro-aggregation (Torn et al., 1997;Kögel-Knabner et al., 2008) and its turnover time is on the order of decades to millennia (Leifeld et al., 2009;Schulze et al., 2009).In agreement with observations in other studies (Sollins et al., 1996;Swanston et al., 2002), the decay constants of LF-OC were considerably higher than those of HF-OC in all three studied forests.This difference in decay constants is likely mainly attributable to the stabilization by the mineral surfaces, but was nonetheless also related to the differences in C:N ratio and lignin:N ratio, two proxies for decomposability (Melillo et al., 1982).The observed C:N ratio's of the LF-OC were 31-38, and thus within but in the lower end of the range (24-86) observed in forest soils (Strickland and Sollins, 1987;Swanston and Myrold, 1997).The C:N ratio's of  were much lower than those of LF-OC, consistent with other studies (Whalen et al., 2000;Tan et al., 2007).Although the C/N ratio was correlated with decomposition rate constants within all of our organic matter types, it cannot explain the differences among substrates (the more recalcitrant substrates have lower C/N).It is clear that the lower decomposition rates of HF-OC relative to LF-OC are due to the stronger stabilization mechanisms, and that the higher N contents are related mainly to the longer residence times and associated degrees of humification.

What determines the differences in SOC stocks?
The observed total soil organic carbon stocks (14-17 kg m −2 ) were within, but at the low end of the range reported for forest ecosystems (8-48 kg m −2 ; Dixon et al., 1994).The relatively small pools compared to other forests is, however, probably due to the shallower soils in our study compared to the review by Dixon et al. (1994) and to the low clay contents and thus low potential for SOC stabilization (Torn et al., 1997;Kögel-Knabner et al., 2008).Total soil carbon stocks were lowest in Sharptooth oak forest, highest in Asia white birch, and intermediate in East-Liaoning oak forests.However, it should be kept in mind that in the two temperate forests, the entire soil profile was sampled (bedrock at about 55 cm), whereas in the sub-tropical Sharptooth oak forest, where soil depth was around one meter, total SOC was underestimated by sampling to the same depth as in the other forests.
In soils with similar clay contents (as the three study sites included in this study) and assuming similar dominant clay minerals at the sites, one could argue that the stabilization potential of SOC is similar.Under these conditions, SOC contents are determined by the balance of soil carbon inputs and carbon losses.Would it then be possible to explain the observed differences in the SOC stocks with those in the litter inputs and decomposition rates?One prerequisite is for sure that carbon losses other than decomposition do not export substantial amounts of C. Given the steep slopes at the sites, it is likely that part of the deposited litter is transported to the valley bottoms during heavy rains, although we did not observe substantial transport of litter, we can not exclude that this potential loss of surface litter may have confounded our turnover estimates.Nonetheless, all three sites exhibited similar slopes and thus likely similar degrees of erosion and litter transfers, implying that the site comparison is still relevant.
The sub-tropical Sharptooth oak and temperate Asia white birch forest exhibited very similar amounts of litter input, both above-and below-ground.Given the higher quality for decomposition in the Asia white birch forest, as indicated by the chemical composition and decay constants of the litter and LF-OC, one would expect lower SOC stocks in the Asia white birch forest.However, the opposite is observed.Both in the forest floor as in the LF-OC, the residence time in the Asia white birch forest is 50% higher than in the Sharptooth oak forest and hence, the SOC stocks are 22% higher in the Asia white birch forest (despite the similar litter inputs).Under the assumption that differences in SOC stabi-lization mechanisms are negligible, this result indicates that the negative effects of the poorer SOC quality on its decay in the Sharptooth oak forest are overshadowed by the positive effects of the more favorable to decomposition sub-tropical climatic conditions.Does this imply that the differences in chemical composition are unimportant in these deciduous forests?When comparing the two temperate forests, it becomes clear that chemical quality does play a critical role.Although microclimate may have differed, climatic differences are probably minor and thus do not confound the observations.Total litter inputs are 33% higher in the Asia white birch -than in the East-Liaoning oak forest, yet the total SOC content is only 11% higher.This discrepancy between site differences in litter inputs and carbon stocks is attributable solely to processes occurring in the surface organic layer.Belowground, the 25% difference in root litter inputs is reflected in the LF-OC pool, which exhibits a very similar relative difference between both forest types (+29%).In contrast, aboveground litter fall is 40% higher in the Asia white birch than in the East-Liaoning oak forest, whereas the forest floor does not significantly differ in carbon content between both forest types (statistically insignificant difference of 7%).Because leaf litter is of much higher quality in the Asia white birch forest and leaf litter decomposition proceeds much faster, the higher above-ground carbon inputs are almost completely offset by the higher carbon losses due to the difference in above-ground litter quality.
Thus, litter quality is a very important determinant of SOC cycling in these temperate forests, but among our study sites less so than the effect of climate.Nonetheless, we would like to re-iterate that this reasoning only holds if erosion and litter export by heavy rains are comparable in the two temperate sites (which are likely given the same climate, slopes, and soil types).

Soil microbial biomass and activity
Soil microbial biomass represents an important labile pool of nutrients and carbon (Henrot and Robertson, 1994).Changes in the size of the microbial biomass pool may indicate changes in the substrate availability that are otherwise not easily detectable.In this study, SMB-C and SMA were higher in July than in May and September in all three forests, reflecting that substrate availability must have varied considerably during the growing season.Similar findings were also observed in other studies (Wardle, 1998;Michelsen et al., 2004;Shishido et al., 2008).
Soil microbial biomass of the sub-tropical Sharptooth oak forest was lower than that of East-Liaoning oak and Asia white birch forests.This pattern reflected well that of the labile SOC pools, confirming that substrate availability might be an important control over the size of the SMB-C pool (Wardle, 1992).www.biogeosciences.net/7/275/2010/Biogeosciences, 7, 275-287, 2010 In contrast to SMB-C, SMA was higher in the Sharptooth oak forest than in the Asia white birch and East-Liaoning oak forests.This pattern was very surprising given that SMA was determined in the lab under similar climatic conditions, and that the Sharptooth oak soil contained the least available SOC, which also was less degradable than the SOC in the Asia white birch soil.Based on the observed SOC quantity and quality, we would have expected the highest SMA in the Asia white birch, and lower ones in the Sharptooth oak forest because of the lower SOC availability.We can only speculate why the SMA observations contrasted our expectations.One potential explanation could be that the environmental conditions in the lab during the SMA experiments resembled the climatic conditions in the sub-tropical Sharptooth oak forest much better (both in terms of temperature and soil moisture) than in the two temperate-zone forests.Soil microbial population may adapt either physiologically or structurally to temperature (e.g.Bradford et al., 2008) and perhaps the warm and moist conditions during the lab incubations were more optimal for the microbial populations in the subtropical soil, and supra-optimal for those in the temperate soils.

Fine root dynamics
Our estimates of mean fine root biomass (between 510 and 660 g dry matter m −2 ) were well within the range reported by Vogt et al. (1996) for fine root biomass in temperate broadleaved forests (243-999 g dry matter m −2 ).Fine root growth at our study sites (270 to 350 g dry matter m −2 year −1 ) was much higher relative to that of a temperate oak forest in a more moderate, maritime climate (Quercus robur; fine root productivity of 180 g dry matter m −2 year −1 ; Curiel Yuste et al., 2005).Luyssaert et al. (2007) reported a fine root productivity average across 52 temperate deciduous forests of 440 g dry matter m −2 year −1 (recalculated from carbon units assuming that dry matter contains 50% carbon).Although our sites are all below this global average, they are all well within the 25-75% percentile range (204-460 g dry matter m −2 year −1 ) reported in the review by Luyssaert and co-workers (2007).
Fine root biomass and -production both decreased with soil depth.Similar findings have been observed in many studies (Vogt et al., 1981;Olsthoorn, 1991;Xiao et al., 2003;Konôpka et al., 2006) and have been attributed to declining nutrient availability and changing physical conditions with depth.Also fine root turnover declined with depth (data not shown).Averaged over all depths, turnover of roots <2 mm at our sites (0.45 to 0.67 year −1 ) was lower than the global mean turnover rate (0.8 year −1 ) for forest fine roots <2 mm reported in the review by Gill and Jackson (2000), but nonetheless well within the range for broadleaf forests in similar climates (0.2-1.4 year −1 ).
It should be noted that estimates of fine root productivity are notoriously difficult and that different techniques can produce largely different estimates, even within the same site (Gill and Jackson, 2000;see Milchunas, 2009 for an intercomparison study of most applied techniques).As stated by Milchunas (2009), the ingrowth core technique applied here may produce either over-or underestimations, depending on specific methodological details.It is therefore extremely difficult to quantify the accuracy of our root production -(and thus also root turnover) estimates.Nonetheless, because we applied the same methodology at all three sites, it is fair to assume that the relative differences can provide valuable information.The faster root turnover in the Asia white birch forest relative to the East-Liaoning oak forest, for instance, thus provides important information.Rates of root turnover are influenced by climate (Vogt et al., 1986;Hendrick and Pregitzer, 1993;Pregitzer et al., 2000) and nutrient availability (Crick and Grime, 1987;Schoettle and Fahey, 1994;Janssens et al., 2002), amongst others.Root functioning is optimal when the resource capture efficiency (uptake of nutrients or water per unit C cost) is maximized (Eissenstat and Van Rees, 1994).Rapid fine root turnover constitutes a large energy and nutrient cost for the plant, while long lifespans result in reduced rates and lower efficiency of resource uptake (Schoettle and Fahey, 1994).Rapid root turnover may be advantageous in nutrient-rich environments, where resource capture efficiency is likely to be maximized by reducing root longevity, thus simultaneously increasing nutrient uptake capacity and reducing root maintenance costs (Crick and Grime, 1987).In contrast, increased longevity would be more favourable under highly competitive, nutrient-poor conditions because nutrient losses through root mortality will need to be avoided.
The difference in root turnover in our two temperate forests can not be climate driven and might thus be related to faster decomposition rates and thus nutrient cycling.The higher fine root turnover rate observed in the sub-tropical Sharptooth oak forest might be due merely to the warmer and wetter conditions, that favour root production (Gill and Jackson, 2000).However, as discussed above, the more favourable climate also accelerates decomposition and thus nutrient availability to roots (Wardle, 1992;Zaman et al., 1999;Gill and Jackson, 2000;Tu et al., 2003;Xiao et al., 2007).Hence, it is impossible to state from this limited number of forest sites whether climate or nutrient cycling is the dominant control over root turnover.
In conclusion, our results show that there are obvious differences in pool size and decomposition rates of litter and SOC, SMB-C and SMA, and fine root biomass, production and turnover rate among Asia white birch, East-Liaoning oak and Sharptooth oak forests.These results provide basic information in estimating the effectiveness of belowground carbon dynamics and sequestration in the three forests. 34

Table 1 .
Initial chemical content of leaf litter, fine root litter, light fraction organic carbon (LF-OC), heavy fraction organic carbon (HF-OC) and total soil organic carbon (SOC) in the 0-10 cm mineral soil layer used in the decomposition experiments in Asia white birch and East-Liaoning oak forests of warm temperate zone and Sharptooth oak forest of sub-topical zone.Values represent mean ± standard error (n=5).Different letters in each row are significantly different (P <0.05) according to the least significant difference test.

Table 2 .
Soil organic carbon inputs and pools in Asia white birch and East-Liaoning oak forests of warm temperate zone and Sharptooth oak forest of sub-topical zone.Values represent mean ± standard error (inputs in g C m −2 yr −1 ; pools in kg C m −2 ; n=5).Different letters in each row are significantly different at P <0.05.Labile carbon was assumed to include non-woody surface litter and light-fraction SOC.Recalcitrant carbon was estimated as woody debris plus heavy-fraction SOC.
2 Fig. 1.Seasonal changes of forest floor mass in Asia white birch and East-Liaoning oak forests of warm temperate zone and Sharptooth oak forest of sub-topical zone from May 2006 to April 2007.Vertical bars indicate standard errors of means (n=5).

Table 2 )
. In accordance with the carbon stores in the surface horizon, we observed the largest SOC pool in Asia white birch and the lowest in Sharptooth www.biogeosciences.net/7/275/2010/Biogeosciences, 7, 275-287, 2010

Table 3 .
Mean annual decay/mineralization constants (k) of leaf litter and fine root, and light fraction organic carbon (LF-OC), heavy fraction organic carbon (HF-OC) and total soil organic carbon (SOC) in Asia white birch and East-Liaoning oak forests of warm temperate zone and Sharptooth oak forest of sub-topical zone.Values represent mean ± standard error derived from eight-time sampling with five replicates for Asia white birch and East-Liaoning oak forests, and seven-time sampling with five replicates for Sharptooth oak forest.Mass loss per year was estimated as: 100× (initial mass -final mass)/initial mass.Different letters in each column are significantly different (P <0.05) according to the least significant difference test.R 2 , multiple coefficient of determination and significance level of the exponential model ( * P <0.05; * * P <0.01) are also given.