Introduction

Globally, soil is the largest carbon (C) pool in terrestrial ecosystems1. Approximately 1,500 ± 230 Gt of soil organic carbon (SOC) is stored in the first meter of soil, nearly twice as much as atmospheric carbon (828 Gt as CO2)2 and almost 44% is stored in forests3. A better understanding of soil C and N balance in forest floor and mineral soil can help guide the implementation of mitigation policies to reduce the emission of greenhouse gases by sources or enhance their removal from the atmosphere by “sinks” which refers to forests, vegetation or soils that can reabsorb CO23,4. In contrast to the clear inventory-based assessments of aboveground C on both regional5 and global scales3, the C and N pools in forest floor and mineral soils remain uncertain because of the higher variations among global forest ecosystems6,7.

Natural environmental factors and human activities generate abrupt, large scale, irreversible changes and thus alters forest composition8,9, consequently resulting in the changes of soil C and N cycles10,11,12. The impacts of converting natural forest (NF) to plantation forest (PF) on soil C and N cycles have been well examined at the stand scales, relating their variances in above- and below-ground C input through litter-fall, root exudation, and their inherently different management practices13,14,15. However, the influence of the wider regional or global scale conversion of NF to PF on changes of soil C and N cycles is not yet well understood. This is perceived as a key bottleneck in improving the prediction and feedback of soil C and N mitigation related to land use change (LUC). Previous studies have suggested that the impact of converting natural forests to plantations significantly impacts the quantity and quality of C and N input between natural forests and plantations12,14,15. Finzi et al.10 for example, reported that changes in the species composition control forest floor and surface soil C and N dynamics while converting natural evergreen broadleaf forests to moso bamboo plantations significantly decreased the concentrations of C within the 0–40-cm.

The natural abundance of soil 13C and 15N (expressed as °/oo) has uniquely been used to estimate soil C turnover rate16,17,18,19,20. The 13C and 15N abundance in soils is a dynamic function of the rate and isotopic composition of inputs and outputs, and the internal C, N transformations that occur in a soil system. It can also be used as a gross indicator of environmental processes that impact soil C storage in forest ecosystems and therefore provide integrated insights into soil C and N cycles21,22. Litter inputs generally lowers soil δ13C and δ15N values while higher decomposition leads to increased δ13C and δ15N values21,23. Moreover, it has been well evaluated that for sites with stable vegetation and low human disturbance, SOC turnover rates do not significantly change according to vertical soil-profile SOC contents and δ13C20,22,24. The values of soil δ15N vary with some soil N cycle processes, e.g., N deposition, soil N nitrification and denitrification, and are used as a tool to examine soil N availability19,25,26. In addition, N cycle processes in stand-scale forests altered by human disturbance and climatic factors27,28. However, details pertaining changing soil δ13C and δ15N with NF converted to intensively managed plantations along climatic gradients of MAT and MAP at a regional scale is still not well understood.

China, with a land area of 9.63 million km2 encompasses a climatic gradient, from cold temperate to tropical climate zones, and therefore has diverse land uses (LU) ranging from deciduous forests in the north to evergreen forests in the south7,29. China can be viewed as a unique “laboratory”, with its complex interactions between the varied climatic zones and intensive human activities, thus providing an excellent opportunity to examine simultaneous climate and human impacts on the forest soil C and N pools30. To date, several studies have been carried out to evaluate soil C pools across China’s forests using statistical or biogeochemical models7,31,32. However, there are some inconsistencies in soil C and N pool estimations such as the increasing of plant cover through reforestation and afforestation programs, the lack of data from repeated inventories, the variety of methods used to assess the carbon balance of China and the variation of climate conditions within those studies which vary from one period to another. Among others, these are either because of the insufficient observations or inconsistent measurement methods among the studies32. Recently, Tang et al.33 conducted a field survey involving 14,371 field plots to evaluate the current C stocks in mineral soil of China’s terrestrial ecosystems in order to examine their biogeographical patterns and potential climatic drivers. However, the aforementioned studies evaluated soil C pools only in the NF. The increasing needs for timber and other economic forest products in China, means that large areas of NF has been converted to intensively managed plantations12,34,35 but there is still significant uncertainty concerning changes in forest floor and mineral soil C and N stocks and turnover processes potentially induced by forest conversion along the 4200 km transect from northern China to southern China.

The objectives of this study are, in a range of climate zones in eastern China to: (i) determine the patterns of soil C and N contents (g kg−1) and δ13C and δ15N values for 1 meter vertical depth in NF and PF in forest floor and mineral soil, (ii) quantify changes in soil C and N contents and turnover rates as induced by LUC, and (iii) determine the effect of climate factors on the patterns of soil C and N contents, and natural abundance of soil 13C and 15N after LUC. We hypothesize that: (i) C, N, δ13C and δ15N contents in forest floor would significantly differ between NF and PF, and vary significantly among the study sites in the climate gradient, (ii) site specific content and turnover times of soil C and N would be increased by LUC, but (iii) overall climate controlled variations in soil C, N, δ13C and δ15N content between site would not be significantly affected.

Results

Carbon and nitrogen contents, and C/N ratios in forest floor and mineral soils

Forest floor C, N contents and C/N ratio were significantly different among the study sites (P < 0.001, for all except for N content P < 0.01) and their values were considerably affected following conversion of natural forest (NF) to plantation forest (PF) (Table 1). Furthermore, forest floor C mean content was lower in NF with 399.8 ± 86.1 g kg−1 across six sites as compared to PF with 438.6 ± 70.9 g kg−1 while an opposite trend was observed in forest floor N content with 13.6 ± 1.8 g kg−1 in NF and 10.3 ± 2.6 g kg−1 in PF. We observed an increase of forest floor C varying from 2–17% among sites following the order: DH > JF > MH > QY > HT > XY. Moreover, a general increase of C/N ratio and decrease of soil N in forest floor were observed except at XY site (Table 2).

Table 1 The two-way ANOVA results for all soil variables in forest floor.
Full size table
Table 2 Mean values of C and N content (g kg−1) and C/N ratios in forest floor and mineral soil layers in natural forests and plantations at six sites across the eastern China.
Full size table

Soil C ranged from 2.98 g kg−1 to 49.55 g kg−1 whereas soil N was in the range of 0.23 g kg−1 to 4.23 g kg−1 in mineral soil across the sampling sites (Table 2). There was a significant difference (P < 0.01) in soil C, N content and C/N ratio among sites (Table 3). The statistical analysis showed that soil C and C/N ratios were significantly (P < 0.05) altered following the conversion from NF to PF but soil N did not change significantly (P = 0.214). In addition, we observed a gain of the mean values of soil C content over 1 m by +6% (HT), +39% (XY) and a loss varied between −38% to −12% at DH > MH > JF > QY (Fig. 1a). Soil C/N ratio increased at XY, HT and JF sites (Table 2). On the other hand, although an increase of soil N contents after NF conversion to PF was observed at XY, there were no obvious differences among sites (Fig. 1b). Our results also indicated a decrease in soil C, N values and C/N ratio with soil depth. The statistical analyses affirmed this trend and revealed that soil depth significantly (P < 0.01) alters C, N values and C/N ratio. However, the conversion from NF to PF strongly influenced C and C/N ratio (Table 3). In addition, a significant and positive correlation was found between C and N (r = 0.879, P < 0.01) (Table 4).

Table 3 The two-way ANOVA results for all soil variables over 1 m depth layer.
Full size table
Figure 1
figure 1

Median of (a) soil C content (g kg−1), (b) soil N content (g kg−1), (c) soil δ13C (‰) and (d) soil δ15N (‰) at 0–100 cm soil depth across different forest type and among sites. NF: Natural Forest; PF: Plantation Forest. A segment inside the rectangle shows the median. The boundary of the box indicates the distribution of soil samples (25th and 75th percentile). Error bars denote the maximum and minimum (90th and 10th percentiles).

Full size image
Table 4 Pearson’s coefficients correlation between soil variables, land use and depth.
Full size table

The natural abundance of 13C and 15N in forest floor and mineral soils

Forest floor δ13C ranged between −31.58‰ and −27.73‰ in all sites sampled while forest floor δ15N ranged between −3.65‰ to 0.20‰ in NF, and between −4.66‰ to 3.66‰ in plantations. In general, forest floor δ13C values were significantly lower in NF than PF in XY and HT (Fig. 2) while the change in forest floor δ15N was not obvious except for DH and JF (Fig. 3). There were significant (p < 0.001) differences in forest floor δ13C and δ15N values among sites. In addition, LUC significantly increased forest floor δ13C values (Table 1) but not δ15N (P = 0.155).

Figure 2
figure 2

The values of δ13C in forest floor and mineral soil layers in natural forests and plantations at six sites across the eastern China. The error bars indicate standard deviation (n = 3). MH (Mohe), QY (Qingyuan), XY (Xinyang), HT (Huitong), DH (Dinghushan), JF (Jianfengling). *, ** and *** indicate a significant level at p < 0.05, p < 0.01 and p < 0.001, respectively.

Full size image
Figure 3
figure 3

The values of δ15N in forest floor litter and soil layers in natural forests and plantations at six sites across the eastern China. The error bars indicate standard deviation (n = 3). MH (Mohe), QY (Qingyuan), XY (Xinyang), HT (Huitong), DH (Dinghushan), JF (Jianfengling). *, ** and *** indicate a significant level at p < 0.05, p < 0.01 and p < 0.001, respectively.

Full size image

Soil δ13C and δ15N values significantly (P < 0.001) varied among sites in mineral soil after the conversion from NF to PF. In addition, soil depth, LU and site × soil depth interaction were significantly altered after the conversion from NF to PF (Table 3). Soil δ13C and δ15N were enriched with depth after the conversion from NF to PF (Figs 2, 3), especially at the northern (MH and QY) and southern sites (DH and JF) indicating the high decomposition rate of soil organic matter (SOM). However, there were no differences in soil δ13C between NF and PF at QY and HT. Furthermore, soil δ13C was significantly different between NF and PF among 0–10 cm, 10–20 cm, 20–40 cm, 40–60 cm and 80–100 cm depths at DH and JF. Soil δ15N was enriched along soil profiles with the mean values of 2.98‰ in 0–10 cm, 4.29‰ in 10–20 cm and 5.35‰ in 20–40 cm but it decreased between 40–100 cm depth layers.

Relationships of climatic factors to C and N contents, C/N ratios and δ13C and δ15N

The mean annual temperature (MAT) ranged from −4.14 °C at MH to 21.08 °C at DH, and mean annual precipitation (MAP) increased from 436 mm to 2499 mm from northern to southern China (Table 5). We observed a linear relationship between soil C/N ratio and MAT (Fig. 4a), and a quadratic relationship between soil C/N ratio and MAP in NF (Fig. 4b). Soil δ13C were related to MAT and MAP in both NF and PF (Fig. 4c,d), with a linear relationship of soil δ13C and MAT in PF (r = 0.22, P = 0.004; Fig. 4c), and a quadratic relationship of soil δ13C and MAT in NF (r = 0.29, P = 0.004; Fig. 4c), and there were quadratic relationships of soil δ13C and MAP in both NF and PF (Fig. 4d).

Table 5 Location and characteristics of forest stands at six sites across the eastern China.
Full size table
Figure 4
figure 4

Regression relationships between (a) soil C/N ratios and mean annual temperature (MAT), (b) soil C/N ratios and mean annual precipitation (MAP), (c) soil δ13C and MAT, and (d) soil δ13C and MAP across forest type at 0–100 cm depth layer. The solid lines indicate regression relationships in natural forest stands, and the dotted lines indicate regression relationships in plantation stands. NF (Natural Forest), PF (Plantation Forest).

Full size image

Discussion

Effects of forest conversion on carbon and nitrogen contents, and C/N ratios in forest floor and mineral soils

The present study provides a quantitative overview of C and N contents, and natural abundances of δ13C and δ15N in forest floor layer and mineral soil layers over 1-meter depth under NF and converted plantations across eastern China. Firstly, we found high forest floor N, but low C and C/N ratios in forest floor layer under NF as compared to PF. This point to the better quality of leaf litter materials in NF stands and large release potential of N during decomposition of the litter. This result was consistent with that found by Chen et al.35 who suggested the lower C/N ratios in forest floor litter in NF compared to hoop pine (Araucaria cunninghamii) plantations in Australia. The increase in forest floor C and C/N values and decrease in forest floor N may indicate that forest management practices provide favorable conditions for decomposing microorganisms thus alter their values as previously reported. In China, forest management practices have been directed towards timber production and sustained yield of wood supply, a practice which has significantly altered the balance between heterotrophic litter decomposition and litter inputs, thus impacting on the C content in the forest floor. This trend is consistent with the results of Vesterdal et al.36 who reported that the accumulation of nutrients in the forest floor is altered with increasing thinning intensity. There were significant differences in C, N and C/N within the study sites because of the effects of the forest conversion, including tree species that may affect the decomposition rate and the turnover of these elements in the soil37,38. Tree species differ in their carbon sequestration potential, LUC by changing species composition, tree density and forest structure altering their sequestration potential. Similar results have been reported by Vesterdal et al.39 who observed that forest floor C and N contents and C/N ratio were strongly affected by trees species. Besides, Fonseca et al.37 reported that forest floor C and N under the coniferous species had a large quantity of organic materials poorly decomposed, while a high rate of transformation of forest floor and incorporation in mineral soil have been observed in broad-leaved species. In our study, the forest floor litter was mainly composed of deciduous or evergreen broad-leaf litter in NF, while forest floor in PF is almost purely needle litter. Broad leaves are generally thought to produce mull forest floors that are richer in nutrients and promote rapid decomposition40. In addition, the higher forest floor N observed in NF at MH and QY is probably a result of the low temperature in North part of China which negatively affects the decomposition process and consequently leads a larger accumulation of organic matter. Thus, this study revealed that forest floor C and N content is strongly affected by human disturbances, consequently by LUC.

In mineral soil, C values were significantly altered after the conversion from NF to PF among site probably because of the difference of soil type, vegetation and trees species. The vegetation cover influences the storage of its elements in that it reduces the arrival of solar radiation directly to the soil. Yet, litter decomposition rates are controlled by the temperature and moisture which directly affect soil microbial activity. These findings are consistent with those of Jobbágy & Jackson41 who reported that the variation of soil C with depth in the profile varies strongly with vegetation type. Ramesh et al.42 also noted that quality and quantity of different soil organic carbon pools change with time depending on the rate of photosynthetic C addition and their losses through decomposition. Our study therefore suggested that LUC alters the carbon-holding capacity of soil in short carbon retention capacity of soil. Furthermore, the changes of soil C and N contents after NF converted to PF were not obvious. We observed a gain in soil C values at XY and HT sites and loss at DH, MH, JF, QY sites after the conversion from NF to PF suggesting that LUC from NF to PF influences C inputs. Similarly, Lewis et al.43 found that the effects of change from NF converted to introduced Pinus sp. plantations were highly site-specific and might have a positive, negative, or no influence on the variation of soil C values. Smith et al.34 also demonstrated that conversion from natural Amazonian forest to plantations altered soil organic C with an increase in surface under Euxylophora paraensis Hub. plantation and decreases under Pinus caribaea var. honduensis Barrett and Golfari. The result could be related to different quantity and quality of C input through root exudation, litter inputs and different management practices between NF and PF44. Although biological N fixation is the primary source of nitrogen input45, soil N values did not change significantly within the study sites after the conversion from NF to PF in the present study indicating that LUC did not alter significantly the balance between N input and loss.

In general, soil C and N where stored in the 0–20 cm depth segment of the overall profile, i.e. 67 and 57%, respectively. This is consistent with the results found by Batjes et al.46 who reported 50% the amount of OC located in the upper 30 cm of the soil organic carbon in the layer 0–100 cm. This result suggests a greater soil C sequestration within 0 to 20 cm soil depth. Furthermore, we found that soil C and N content decreased with soil depth may be due to the vertical distribution of roots. Indeed, Jobbágy et al.41, found that root distributions affect the vertical placement of C in the soil, and above- and below-ground allocation affects the relative amount of C that eventually falls to the soil surface from shoots. In addition, it well known that N status is a crucial factor driving forest soil C dynamics and high N availability can promote a greater soil C sequestration13,47. This trend is in line with our study which reported a significant and positive correlation between C and N probably through the effects on organic matter decomposition45.

Effects of forest conversion on natural abundance of 13C and 15N in forest floor and mineral soils

Stable isotopic abundances of δ13C and δ15N have been used as powerful index to evaluate the long-term alterations of C and N cycles16,17,19,20,22. In the present study forest floor δ13C were enriched after LUC in subtropical and tropical regions while δ15N depleted. The depleted natural abundance of 13C and 15N in litterfall could account for low values of forest floor δ13C and δ15N in NF because of the generally higher above-ground litterfall input in NF compared to PF7,48. Moreover, we found that soil δ13C were strongly enriched at XY and HT sites after LUC. These findings might also be related to the less above-ground litterfall and below-ground roots inputs in these sites. Hertel et al.49 confirmed this trend by reporting that the conversion of tropical forest into plantations decrease C flux with fine root mortality to soil organic C pool. Moreover, soil organic C and N accumulation and stability have been suggested to be strongly influenced by litter quality, resulting in more stable organic C accumulated in NF soils with high quality of litter substrates50. On the other hand, the intensive management practices in PF (e.g., clear-cutting and slash burning, site preparation and pruning) accompanied by higher temperature and precipitation conditions in tropical and subtropical (at the DH and JF sites) regions could enrich soil 13C and 15N in plantation because of more soil C and N loss27,44. However, soil organic C accumulation and stability were also influenced by soil matrix structure51, and were related to the saturation of SOC, climatic zones and ecosystem types52. It was difficult to explain clearly the effect of LUC on soil δ15N variation within the study sites because of the complex interactions between abiotic and biotic factors.

A number of studies observed an increasing tendency of δ13C and δ15N with soil depth20,21,23,53. In our study, we observed an enrichment of soil δ13C following the increase of depth layer by 3.09‰ in surface mineral soil (0–10 cm) after LUC probably because deeper soil layer has a greater humification of organic matter decomposition or largely to the increasing residence time of the organic C in the soil54. This result was in line with Brunn et al.20 who found the mean 13C enrichment between δ13C values of the Oi horizon and δ13C values of soil organic matter (SOM) in 10 cm soil depth was 3.4 + 0.2‰ under temperate beech forests. Several processes might cause enrichment of δ13C with increasing soil depth e.g., the depletion of 13C in the atmosphere due to combustion of fossil fuel, and the considerable presence of litter and roots with depleted 13C in the upper soil. Accoe et al.55 observed that the average rate of change in soil δ13C is directly related to organic matter decomposition rates in different parts of the soil profile. Thus, the extent of change in 13C-abundance with increasing soil depth may indicate the quality or stability of SOM under continuous C3 vegetation55. Furthermore, isotopic fractionation during microbial metabolism of SOM and the downward cycling of hydrophilic 13C enriched decomposing products with dissolved organic carbon fluxes possibly contributing to the establishment of the vertical δ13C depth trends54,56. The pattern of soil δ13C observed in PF of DH showed an increase from 0 to 40 cm and then a decrease until 100 cm depth layer reflecting the more complex processes of microbial degradation and mixing of soil C of different ages57. On the other hand, the low soil δ15N values observed at the surface soil across the study sites particularly at XY could be due to the continuous addition of plant residues with extremely low δ15N values58.

Effects of climatic factors on carbon, nitrogen and their isotopic abundances

Climatic variables have strong impacts on soil organic C and N contents59,60, and natural abundance of soil 13C20,22 and 15N61,62. There were linear and non-linear regressions of soil C/N ratios and δ13C with MAT and MAP in both NF and plantations, which indicated that SOC turnover rates were largely determined by MAT and MAP. Our result was consistent with Wang et al.22 who examined large-scale controls of climate over patterns of SOC turnover across terrestrial biomes worldwide using a meta-analysis of soil δ13C in previous literatures and demonstrated that SOC turnover was substantially accelerated with increasing MAT. In addition, the distribution of soil δ13C was slightly correlated with MAP after LUC probably due to the replacement of vegetation types which determines changes in the relative vertical distribution of soil C along rainfall gradients41 after LUC. This result is in line with those found by Burke et al.63 who reported that precipitation clearly has a direct role regionally and globally in the amount of soil C stored. Thus, our findings revealed that MAP is a key factor controlling soil C accumulation and decomposition after the conversion from NF to PF. According to Jia et al.64, vegetation type and soil type rather than MAT explained the variability in soil δ13C along the 400 mm isohyet in China.

Previous studies suggested that climate is the primary factor controlling soil δ15N61. Contrary to these results, we did not observe clear relationship between climate factors and δ15N in the present study. This could be due to the climate change which may change soil δ15N values through altered precipitation patterns, elevated temperatures, and more frequent and extreme weather events. In addition, climate change can affect temperature-sensitive biogeochemical processes, including N mineralization, nitrification and denitrification, soil respiration, litter decomposition, as well as root dynamics and plant productivity65, consequently, altering the rate of δ15N accumulation.

Materials and Methods

Study sites and soil sampling

In August 2017, soil samples were collected from one natural forest (NF) and one plantation forest (PF) stand at six sites including Mohe (MH), Qingyuan (QY), Xinyang (XY), Huitong (HT), Dinghushan (DH), Jianfengling (JF) across eastern China forest (Fig. 5, Table 5). To minimize the impact of local terrain on vegetation or trees, the topography of all selected plots was uniform and flat. In each one of these, the forest floor litter was collected using three 10 cm × 10 cm wooden frames in each stand. Sampling was done carefully in order to avoid contamination with the mineral material. Forest floors were very thin in most stands and were dried at 65 °C for 72 h and roots were sorted out and weighed to determine dry mass. Mineral soils were sampled in the same points where forest floors had been removed. For soil sampling, three pits were dug at 1-meter depth, and mineral soil samples were collected at 0–10, 10–20, 20–40, 40–60, 60–80 and 80–100 cm along soil profiles. A total of 252 samples were collected from 6 sites × 2 land uses × 3 pits × 7 layers. Mean annual temperature (MAT) and mean annual precipitation (MAP) data at six sites were recorded from the adjacent climate observing stations during 1960–2014.

Figure 5
figure 5

The location of forest stands at six sites across eastern China. MH (Mohe), QY (Qingyuan), XY (Xinyang), HT (Huitong), DH (Dinghushan), JF (Jianfengling).

Full size image

Soil C and N contents, and δ13C and δ15N analysis

Forest floor litter samples were oven-dried at 65 °C for 72 h, and mineral soil samples were air-dried at room temperature for 2 weeks before passing through a 2 mm sieve to remove coarse stones, plant residues, and roots. Then, the forest floor and soil samples were ground to fine powders using a ball mill (JXFSTPRP-64, Jingxin Co., Ltd, China) for measurements of soil C and N contents, and δ13C and δ15N.

Soil C and N pools and stable isotope 13C and 15N composition were measured using an isotope ratio mass spectrometer (IRMS) (IsoPrime 100, Isoprime Ltd., UK), connected to a CN elemental analyzer (Vario MICRO cube, Elementar, Germany). Carbon and nitrogen stable isotope abundances were calculated as δ13C and δ15N (‰) using the following formula:

$${\delta }^{13}{\rm{C}}\,{\rm{or}}\,{\delta }^{15}{\rm{N}}(\textperthousand )=({{\rm{R}}}_{{\rm{sample}}}/{{\rm{R}}}_{{\rm{standard}}}\,-\,1)\,\ast \,1000,$$

where Rsample is the13C:12C or 15N:14N ratio in the samples and Rstandard is the 13C:12C or 15N:14N ratio in the standard. The Vienna Pee Dee Belemnite (VPDB) and atmospheric N2 (δ15N = 0.0‰) were used as the standard, respectively. The precision of isotopic composition was checked using internal standards i.e. acetanilide, L-histidine, D-glutamic and glycine28. In general, the analytical precision for δ13C and δ15N was better than 0.2‰.

Statistical analysis

The results are the average of the replicates determined from three subsamples of the same site. Statistically significant differences were determined by P < 0.05 unless otherwise stated. The sample differences of δ15N, δ13C values, C and N concentrations, the effect of LU and depth on C, N, δ13C, and δ15N in whole soil and the comparison of C and N among LU types and depths were tested with analysis of multivariance (ANOVA). Linear regression and nonlinear regression analysis were used to test the relationships of C/N, soil δ13C with climatic factors (MAT, MAP) across forest type. The samples from each depth were used as dependent measurements in the regression. All statistical analyses were performed with the SPSS version 20.0 (Systat Statistical Software Package for Windows)66.