Labile organic carbon pools and enzyme activities of Pinus massoniana plantation soil as affected by understory vegetation removal and thinning

The effects of forest management on carbon (C) sequestration are poorly understood, particularly in the Three Gorges Reservoir area. We aimed to identify the effects of forest management on C sequestration in Pinus massoniana plantations. An intact control forest (CK), a site undergoing regular shrub cutting with the simultaneous removal of residues (SC), a site under low-intensity thinning (LIT), and a site under high-intensity thinning (HIT) were compared for soil labile organic carbon (LOC), related enzyme activities, and soil characteristics. Soil organic carbon (SOC) significantly decreased in the HIT treatment as compared with that in the CK treatment. Soil EOC, DOC, MBC contents in treated plots were higher than those in the CK treatment; particularly, the HIT treatment significantly increased those values in 0–10 cm layer. Thinning resulted in a decrease in cellulase and amylase activities, but an increase in invertase activity. In addition, the SOC content was significantly correlated with four enzymes activities and LOC components, which suggested that the soil LOC components and enzymes activities were sensitive to the changes of SOC. Our results suggest that high-intensity thinning treatment in Pinus massoniana plantation could significantly decrease the SOC content and lead to an increase of LOC components.

The forest ecosystem accounts for approximately 73% of the terrestrial soil carbon (C) pool, which is an important part of the global terrestrial ecosystem 1 . Thus, the forest ecosystem plays a crucial role in the C cycle 2 . A major focus of forest management is to promote the increment of C pool [3][4][5] . In soil, according to their mean residence times, soil organic carbon (SOC) can be divided into recalcitrant and labile components. Management practices have little effect on recalcitrant components because of their longer turnover time in soil 6 , while soil labile organic carbon (LOC) fractions are more responsive to changes in forest management strategies. Although soil LOC fractions make up a relatively small part of SOC [7][8][9] , they can serve as indicators of minor changes in SOC.
Responses of soil LOC components are often indicated by easily oxidised organic carbon (EOC), dissolved organic carbon (DOC), and microbial biomass carbon (MBC) 6,10 . Previous studies have shown that EOC, DOC, and MBC contents affect C sequestration capacity of soil and the emission of greenhouse gases 11 thus indicating that they are important sources of C that are released from the soil to the atmosphere and aid in the decomposition of recalcitrant C 12 . Thus, soil LOC fractions in forests are imported for maintaining balance in the soil C pool under different forest management strategies. Moreover, the activities of enzymes related to the soil C cycle (e.g., invertase, amylase, and cellulase) participate in the SOC decomposition and indicate the status of the available C resources. Therefore, these enzyme activities can contribute to our understanding of the variations in SOC in response to forest management 13,14 .
SCIEntIFIC RepoRtS | (2018)8:573 | DOI: 10.1038/s41598-017-18812-x Thinning is a common strategy of forest management used in plantations forests, and it is a tool for controlling the species composition 15 . Thinning commonly decreases SOC by reducing litter inputs into soil and possibly accelerating decomposition rates owing to change in microclimate 3,12 and their change degree of SOC among different soil layers were varied 12 . Understory vegetation can modulate SOC by affecting the soil characteristics and changing the organic inputs and the leaching of dissolved organic matter 16 . Different forest management strategies can change soil temperature and moisture, and the composition of aboveground vegetation, thereby influencing C and nutrient cycles in forests [17][18][19] . For example, whole-tree harvesting has the most significant effect on the soil C pool, because it causes direct damage to vegetation via SOC release, increased soil temperature, and accelerated erosion 20,21 . Accumulating evidence shows that for most tree species, the effect of forest thinning on SOC dynamics is complex, and that the intensity and method of forest management affect the degree to which microclimate and residual vegetation composition are affected, thereby affecting SOC sequestration 3,22 . However, studies on forest C storage under different forest management strategies have produced contradictory conclusions 23,24 . It has been reported that decreasing tree density in stands decreases total forest C stores 25,26 , whereas others have found that this action can maintain or increase live tree C due to the increased growth rate of trees grown at lower densities 27 , especially in the case of long-term responses to thinning 28 . Short-term studies have revealed that thinning consistently decreases aboveground C 26,29 , indicating that low densities of small trees do not fully offset the loss of C 30 . Thus, it is still unclear whether forest management strategies are compatible with the purpose of increasing forest C storage for climate change mitigation. In addition, exploring the effects and mechanisms affecting forest management strategies on SOC sequestration is a key research area in both forestry and C cycling science, while relatively less attention was paid to the effect of short-term forest management activities on soil LOC fractions 31 .
Plantations are a key component of global forest resources, and play an important role in sustainable forest management [32][33][34] . In China, pure plantations forests constitute approximately 80% of all plantations 16 . Pinus massoniana is the main tree species used for afforestation in South China. It plays a major role in providing forest resources and ecological services 35 , and it covers the largest area (67.83 × 10 4 ha) in the Three Gorges Reservoir area located in subtropical China 36 . However, due to anthropogenic effects and the complex terrain in this region, the soil C content is relatively low 37 . Thus, in recent years, recreating forest structures and optimizing the use of soil by forest management activities were a major focus of the sustainable management of Pinus massoniana aimed at maintaining ecosystem sustainability and soil productivity 12 .
In this study, the objective was to assess the effect of forest management in term of SOC, LOC fractions, and activities of related enzyme at three mineral soil layers (0-10 cm, 10-20 cm, 20-30 cm). The aim of the project was to investigate (1) how different forest management treatments (0%, 15% and 70% stem thinning and understory vegetation removal) influence the contents of SOC, soil LOC fractions (i.e. DOC, MBC, and EOC), and related enzyme activities (i.e. cellulase, amylase, invertase, and catalase) and (2) whether relationships exist among soil LOC fractions, enzyme activities and other soil characteristics (soil pH and fertility characteristics). Our hypotheses are that 1) thinning and understory vegetation removal treatment will decrease the SOC, enzyme activities and increase LOC pools, and 2) the soil LOC fractions are linked to enzyme activities and partial soil characteristics.

Results
Soil chemical properties. Soil total nitrogen (TN), total phosphorous (TP), available potassium (AK), NH 4 + -N, and NO 3 − -N contents decreased with increasing soil depth while soil pH increased with increasing depth, although significant difference were only observed in the soil TN content and pH among the soil layers (p < 0.05). Besides this, in the 0-10 cm soil layer, TN, total potassium (TK), AK and NO 3 − -N contents were significantly lower in the three treated plots than those in the control plots (p < 0.05). In the 10-20 cm soil layer, AK and NO 3 − -N contents in the three treated plots were significantly lower than those in the control plots (p < 0.05). In addition, TN content in the SC treatment plots was greater than those in the thinning (LIT, HIT) plots in 0-10 cm and 10-20 cm soil layer (p < 0.05). In the 20-30 cm soil layer, NH 4 + -N and NO 3 − -N contents in three treated plots were significantly lower than that in the control plots (p < 0.05) ( Table 1). LOC fraction. SOC, EOC, DOC, and MBC decreased with increasing soil depth, and significant difference were only observed in the SOC and DOC contents among the three soil layers (p < 0.05). The EOC content in the 0-10 cm layer was significantly higher than that in the other soil layers (p < 0.05). In the 0-10 cm soil layer, SOC content under the LIT and HIT treatment were significantly lower than those in the CK and SC plots, but in 0-30 cm soil layer, significant difference between the HIT and CK treatment was only observed (p < 0.05). The contents of DOC under the SC and HIT treatment were significantly higher than those in CK and LIT treatment in 0-10 cm soil layer (p < 0.05), and the corresponding values in the 10-20 cm layer were significantly higher than those in the CK (p < 0.05). The content of EOC in the 0-10 cm layer was significantly higher in the HIT treatment than that in other three treated plots (p < 0.05). In addition, the content of MBC in the three treated plots were higher than those in the CK, and, in particular, there was only a significant difference between LIT and CK treatment observed in three soil layers (p < 0.05) (Fig. 1).
Soil enzyme activity. Soil enzyme activity decreased with increasing soil depth from the overall, which was consistent with the trend of vertical change in LOC content (Fig. 2). Soil cellulase enzyme activities in the 0-10 cm soil layer were higher than those in the other two soil layers (p < 0.05), and the activities of cellulase and invertase in the 10-20 cm soil layer were significantly higher than those in the 20-30 cm layer (p < 0.05).
Comparison with CK revealed that, the LIT and HIT treatment resulted in lower levels of cellulase, amylase, and catalase activity, whereas invertase activity increased. Cellulase activity in the 0-10 soil layer in the CK plots was significantly higher than that in the SC and thinning plots (p < 0.05). Amylase activity in the 0-10 and SCIEntIFIC RepoRtS | (2018)8:573 | DOI:10.1038/s41598-017-18812-x 10-20 cm soil layers in the CK plots were significantly higher than those in the LIT and HIT plots (p < 0.05). The differences of soil invertase activity between the LIT, HIT, and CK plot the in 10-20 cm and 20-30 cm layers were significant (p < 0.05).

Relationships between soil LOC fractions and soil enzyme activities. SOC content was signifi-
cantly positively correlated with the content of DOC, MBC, and EOC (r = 0.898, p < 0.01; r = 0.704, p < 0.01; r = 0.466, p < 0.01, respectively), as well as with the activities of cellulase, amylase, catalase, and invertase (r = 0.930, p < 0.01; r = 0.311, p < 0.05; r = 0.725, p < 0.01; r = 0.570, p < 0.05, respectively) ( Table 2). The content of DOC, MBC, and EOC were significantly positively correlated with cellulase, catalase, and invertase activity and  Table 1. Soil chemical properties at three soil depths in the four forest management treatments (mean value ± standard error; n = 3). Significant differences among different soil layers subjected to the same treatments are identified with A, B, and C (p < 0.05). Significant differences among different treatments of the same soil layer are identified with a, b, c, and d (p < 0.05), based on the analysis of variance.

Discussion
The content of soil TN, TP, AK, NH 4 + -N and NO 3 − -N decreased with increasing soil depth, and pH exhibited the opposite trend, which is consistent with previous findings [38][39][40] . According to Zhang et al. 39 , SOC and TN of a chestnut forest decreased after shrub cutting. In this study, contents of TN, AK, and NO 3 − -N in the 0-10, 10-20, and 20-30 cm soil layers were reduced in the LIT and HIT treatment. This might be due to a reduction in nutrient elements such as N, P, and calcium being returned to the soil through the litter, during to the thinning proces 41 . Moreover, we did not find any significant effect of treatment on soil pH and TP of all the soil layers, which were consistent with the findings of many studies [42][43][44] . This might be due to the positive and negative effects of the treatments on soil nutrients 42 . It is also possible that the effects of treatments on these properties will be manifested in the long term but not observed in the short period of this study 45 .
In this study, SOC content of soil (0-30 cm) in the CK, SC, LIT, and HIT treatment were 53.01, 53.84, 53.31, and 48.70 g kg −1 , respectively, suggesting that HIT treatment reduced the SOC content remarkably, which was partially consistent with the hypothesis. Similar results were found by Chen et al. 12 and Achat et al. 46 . The decrease in SOC content was mainly caused by the fact that, after the thinning, substrate inputs to the soil were reduced 3 . The microclimate induced an increase in the rate of SOC decomposition following decreased canopy closure and reduced SOC content 47 . In addition, the partial removal of tree canopies caused acceleration SOC leaching losses 12 . Moreover, we found that the response of SOC to thinning and understory vegetation removal was significantly different among three different soil layers because of the SOC in deeper soil layer with a longer residence time is less sensitive to disturbances 48 . These results may be due to the different compositions of plant species in different treatment plots wherein their roots reach different soil depths to control C sequestration and decomposition 49 .
SOC, DOC, MBC, and EOC content decreased with increasing soil depth, which is consistent with previous findings 8,50-52 . This may be related to the spatial distribution of root residual input and decomposition 52,53 .
Although little is known about the composition of soil active C, the method has shown that it is more sensitive to soil management strategies than SOC is, and more closely related to soil biological properties 7,54-56 . According to Chatterjee et al. 57 and Diochon et al. 58 , forest thinning causes in the return of a large amount of residual organic matter to the surface, accompanied by changes in light conditions, which can lead to significant changes in LOC mineralisation. In our work, the contents of soil EOC, MBC, and DOC in thinning treatments were higher than those in the CK plots was consistent with our hypothesis; particularly, the HIT treatment significantly increased these contents in the 0-10 cm layer and the effect was more marked in the upper than in the lower soil layers. The result was supported by the findings from a study on coniferous forests in Wyoming, USA 57 and those of Chen et al. 12 . However, high-intensity thinning promoted non-tree vegetation owing to the sparse forest canopy. Subsequently, their fine roots would increase the LOC input by root exudates. In addition, high-intensity thinning accelerated the decomposition of litter and residue, and the accumulation of LOC fractions, consequently the potential SOC mineralisation rate increased 12,59 .
Soil enzymes participate in almost every transformation process of litter decomposition and play a central role in maintaining forest soil fertility by releasing plant available mineral nutrients from complex organic resources 60,61 . Soil invertase, cellulase and catalase activities decreased with increasing soil depth, in accordance with the results of Xiao et al. 52 and Chen et al. 12 . A high content of organic matter in surface soil is beneficial to the growth of microorganisms with active metabolic processes, which in turn leads to the accumulation of soil enzymes in the surface layer. In this study, the overall cellulase and amylase activities decreased after thinning, which is supported by Chen et al. 12 . Similar results were observed in the New Jersey Pine Barrens, where cellulase and phenol oxidase activities significantly decreased after one year of thinning 62 . The results were partially consistent with hypothesis of this study. These variations in enzyme activities can be due to reductions in root activity and changes in microbial composition. Furthermore, the extent of utilisation of the C and N sources for soil enzymes differs, with varying effects on soil enzyme activity under different management strategies. Li et al. 63 reported that invertase can break down some carbohydrate polymers to release the nutrients from organic compounds through its role in the first phases of degradation of organic compounds. During this phase, molecular size is reduced and smaller organic structures are produced, which facilitates microbial enzyme activities. We found that the thinning treatments had greater invertase activity than that in the CK treatment. A plausible explanation is that invertase activity is positively correlated with soil pH, TN, and TP, but it is negatively correlated with TK and NO 3 − -N. We found that the SOC content was significantly correlated with the four enzymes activities and soil LOC components, which was observed in earlier studies 63,64 . The result suggested that the soil LOC components and these enzymes activities were sensitive to the variations of SOC which was consistent with the hypothesis. Significant correlations were found among the LOC fractions and invertase, catalase, and cellulase activities in the soil, which is consistent with findings of Paz-Ferreior 65 . In addition, significant correlations were found between each component of SOC and the soil TN content in the soil, consistent with findings of Geng et al. 62 . These correlations might have arisen because the N content in the soil organic matter affects the rate of soil organic matter decomposition and consumption by microorganisms. Nitrogen-rich organic matter is easily and rapidly decomposed, transferred, and converted by microorganisms, thus increasing the SOC content in the soil. As soil enzymes directly participate in the utilisation of soil nutrients, they indirectly reflect the dynamic state of the conversion of soil nutrients. Taken together, a similar conclusion to that of Ma et al. 66 can be drawn in that enhancing soil nutrient content is the key factor for increasing the accumulation of LOC fractions.

Conclusions
This study demonstrates the distribution of chemical properties of soil, LOC fractions, and enzyme activities, and provides insight into their relationships in Pinus massoniana plantations under different forest management  approaches. Our results showed that the content of soil LOC fractions, TN, TP, AK, NH 4 + -N, NO 3 − -N and enzyme activities decreased with increasing soil depth in all the treatments. High-intensity thinning reduced the SOC content remarkably. Soil EOC, DOC, and MBC contents were higher in thinning treatments than those in the control; especially high-intensity thinning treatment significantly increased those contents in 0-10 cm layer. Simultaneously, thinning resulted in a decrease in cellulase and amylase activities, but an increase in invertase activity. The variations of SOC, EOC, DOC, MBC, cellulose, amylase, and invertase were more marked in the top soil layer than in the deeper soil layers. The correlation analysis showed that the soil LOC components and enzymes activities were sensitive to the variations in SOC. Our results indicate that high-intensity thinning treatments in Pinus massoniana plantation decreased the SOC content and might lead to an increase of LOC components, in which the effect was more marked in the 0-10 cm soil layer than in the deeper layers.

Materials and Methods
Study site. The experimental plots were located in the Jiulingtou Forest in Zigui County, Hubei Province, China, where Pinus massoniana were aerially seeded in the 1970s (Fig. 3). The soil type of the plots was haplic luvisol 67 , and the annual mean temperature was approximately 16.9 °C with annual rainfall varying between 1000 and 1250 mm, occurring primarily from April to September 68 . Twelve 20 m × 20 m plots within Pinus massoniana monoculture plantations, laid out in an orthogonal design and separated from each other by at least 2 m, were defined in September 2013. Each plot contained four treatments, randomly assigned per row. Treatments consisted of (1) intact forest (control, CK), (2) all shrubs harvested and residual materials removed (shrub cutting, SC), (3) low-intensity thinning (LIT) with 15% of the basal area of the trees removed, and (4) high-intensity thinning (HIT) with 70% of the basal area of the trees removed. The study area was relatively steep with a northwest-facing slope of 34°. The experiment was conducted in mid-October 2013 and included a control intact forest, and thinning and shrub cutting treatments using chain saws. Only the harvested trunks were removed and the residue produced by the harvest was not cleared from the plots. Except for plots subjected to the SC treatment, the understory shrub layer was dominated by Lespedeza bicolor Turcz., Pyracantha fortuneana (Maxim.) Li, and Litsea pungens Hemsl., with an average diameter at breast height of 5.00 cm and average height of 5.60 m. Herbs in the study area were mainly Woodwardia japonica L.f., Carex tristachya Thunb., Aster ageratoides Turcz., and Parathelypteris nipponica (Franchet and Savatier) Ching. Soil analysis. The twelve plots (4 treatments × 3 replications per treatment) were divided into 24 subplots (10 m × 20 m). Six randomly-placed replicate soil samples were collected from each subplot in June 2016 at depths of 0-10, 10-20, and 20-30 cm yielding 432 (24 subplots × 6 replicates per subplot × 3 soil layers) samples, which were collected using a soil sampler 30 cm in length. These samples were placed in plastic bags and stored in a portable cooler for transport to the laboratory. The soil samples were divided into two subsamples of equal volume. One was passed through a 2-mm sieve to remove impurities of soil and stored at 4 °C before testing. This subsample was used to determine NH 4 + -N, NO 3 − -N, DOC, MBC, EOC, and enzyme activities (cellulase, amylase, invertase, and catalase). The other subsample was air-dried and sieved before use for the analysis of SOC and other soil properties (TN, TP, TK, AK, AP, and pH). Sample analyses. Soil chemical analysis. SOC content was measured using dichromate oxidation 69 . Soil TN was determined using the Kjeldahl method 70 . NH 4 + -N and NO 3 − -N concentrations were determined using a flow injection analyser, TP, TK, AK, and AP were measured using inductively coupled plasma mass spectrometry (ICP-MS) analysis (IRIS Intrepid II XSP system; Thermo Electric Co., USA). Soil pH was determined from a soil water (1:5 w/v) suspension, prepared by shaking 30 min, using a conductivity meter.
Total DOC content was measured using dichromate oxidation titration 71 , and EOC content was analysed using 0.333 mol L −1 KMnO 4 oxidation 72 . MBC content was measured using chloroform fumigation extraction 73 .
Soil enzyme activity analysis. Soil amylase activity was measured using 2 g of fresh soil incubated for 24 h at 37 °C according to Ebregt's method 74 . Soil invertase activity was measured as at 30 °C and pH 4.65 in Na-acetate buffer according to Gianfreda's method 75 . Soil cellulase activities were detected by an incubation according to Sharma's method 76 , and soil catalase activity was determined at pH 7.0, following the monitoring of the decomposition of H 2 O 2 at 240 nm with an extinction coefficient of 43.6 M −1 cm −1 according to Roggenkamp and Sahm 77 . Data analyses. All data were analysed using SPSS 22.0 for Windows (SPSS Inc., Chicago, IL, USA). One-way analyses of variance (ANOVA) and comparisons among means were made using the least significant difference (LSD) test, with p < 0.05 regarded as significant. Pearson's correlation coefficients of soil LOC fractions (EOC, DOC, and MBC) with enzyme activities (invertase, cellulase, catalase, and amylase) and other soil characteristics were estimated, with a significance level of p < 0.05.