Increasing SOC sequestration and closing N cycle during post-agricultural restoration in karst region, Southwest China

Purpose Post-agricultural restoration affects soil organic carbon (SOC) sequestration and ecosystem nitrogen (N) cycle. However, the control mechanism of SOC sequestration and alteration of ecosystem N status following post-agricultural restoration are not well understood in karst regions. Methods Croplands, abandoned croplands, and native vegetation forests were selected to represent three stages following post-agricultural restoration using a space for time substitution approach in a karst critical zone in Guizhou province, Southwest China. The variations of soil aggregate associated SOC and relationships between soil Ca and SOC were analyzed to identify SOC sequestration potential. Foliar δ 15 N composition and soil to plant 15 N enrichment factor (EF = δ 15 N litter − δ 15 N soil ) were analyzed to determine ecosystem N status. and the N and were to identify SOC sequestration potential and ecosystem N status following post-agricultural restoration in the During post-agricultural restoration, macro-aggregate proportions and the SOC concentrations in bulk soils signicantly increased. Soil Ca concentrations non-linearly increased with increasing SOC concentrations. Soil macro-aggregates and Ca 2+ played important roles in enhancing SOC storage and stabilization, which was critical to SOC sequestration. Foliar δ 15 N values and EF values were more negative following post-agricultural restoration, which was directly associated with the plant uptake of 15 N-depleted inorganic N produced from SON mineralization and nitrication. During post-agricultural restoration, increased plant biomass and decreased SON mineralization led to the more sucient absorption of the produced inorganic N, less NO 3− loss, and a more closed N cycle. These results underscore the increasing SOC sequestration potential and the closing of N cycles following post-agricultural restoration in the karst region.


Introduction
The soil organic carbon (SOC) pool (1550 Gt) is approximately double the size of the atmospheric C pool (760 Gt) and triple the size of the pant C pool (560 Gt) (Lal 2004). The dynamics of the SOC pool likely cause signi cant in uences on atmospheric CO 2 concentration and global climate change ; Han 2020, 2021). Nitrogen (N) is an essential bio-element for vegetation growth, which generally restricts the primary productivity in many terrestrial ecosystems at middle and high latitudes (Vitousek and Howarth 1991;Davidson et al. 2007; Song et al. 2021), and C storage in plants and soils (Luo et al. 2004;Xia et al. 2020). Post-agricultural restoration, as one type of land use change, has been widely recognized to affect SOC dynamics ( Poeplau et al. (2011) found that SOC storage decreased with the conversion from natural lands to croplands in several decades, but it could be recovered during post-agricultural restoration. In many studies, SOC accumulation during post-agricultural restoration is closely linked to the increased litterfall and root input with increasing vegetation biomass (Ghafoor et al. 2017; Scott et al. 2021). Generally, SOC storage is primarily determined by the balance between organic debris input and microbial decomposition loss (Jobbagy and Jackson 2000; Yang et al. 2016;Juhos et al. 2021). However, only the accumulation of SOC pool which can undergo long-term stabilization is meaningful for global climate because short residence time (several days to several months) of liable SOC pool determines the negligible in uence for the balance between soil C pool and atmospheric C pool.
SOC sequestration mainly depends on the accumulation of the stable SOC which is not easily mineralized in the soil environment (Fan et al. 2020). Thus, identifying SOC stabilization mechanism is key to understanding the SOC sequestration potential following post-agricultural restoration (Li et al. 2012). The three mechanisms of SOC stabilization have been widely proposed (Sollins et al. 1996 (2) chemical stabilization through association with clay particles; and (3) physical stabilization via the protection of soil aggregates. Generally, the primary SOC stabilization approach differs between sites due to the discrepancies in land use management types and soil physicochemical properties (Fan et al. 2020). Thus, it is necessary to con rm the main SOC stabilization approach following post-agricultural restoration at a speci c site.
In terrestrial ecosystems, SOC accumulation signi cantly affects soil N processes because both of them are existing mainly as organic complexes. The 15 N stable isotope ratio (δ 15 N) has been widely considered to evaluate N processes and sources, including N fertilizer application (Choi et al. 2017), symbiotic N uptake by mycorrhiza (Hobbie and Ouimette 2009;Taylor et al. 2019), N 2 xation of N 2 -xing plants (Hogberg 1997); atmospheric N deposition Liu et al. 2006), plant N uptake and microbial N assimilation (Fowler et (Six et al. 1998(Six et al. , 2000(Six et al. , 2002. According to these ndings, we hypothesized that soil aggregates and Ca are the major factors determining SOC sequestration during post-agricultural restoration in the karst region (H1). Additionally,  suggested that post-agricultural restoration signi cantly enhanced soil N availability based on the increased gross N mineralization rate. The increment of gross N mineralization is mainly attributed to the increased gross microbial biomass carbon C and liable SON. However, the proportion of SON which can easily mineralize decreases during post-agricultural restoration, due to increased SON stabilization via OM-Ca 2+ -mineral complexes and soil aggregates. Moreover, vegetation restoration enhances plant N uptake ). Thus, we hypothesized ecosystem N cycles are towards closed during postagricultural restoration in the karst region (H2). The research objectives were to: (1) determine the control mechanism of SOC sequestration during post-agricultural restoration by analyzing the relationships of SOC with soil aggregates and Ca; (2) determine the alteration of ecosystem N cycle during postagricultural restoration by analyzing plant and soil δ 15 N composition and EF value; and (3) establish a conceptual pattern of SOC sequestration and ecosystem N cycle during post-agricultural restoration in the karst ecosystem. SOC sequestration and ecosystem N cycle are closely linked to greenhouse gas release, ecosystem primary productivity, soil fertility (Robinson 2001;Lal 2004). The GGP program is no longer to only seek a better social and ecological environment for residents, possibly it is of great importance to global climate change, ecosystem development, and soil health based on our study objects.

Study area
The study area is located in the Chenqi catchment (26°15.  Table S1 and photographs of the three stages are shown in Fig. 1.

Soil and leaf sampling
In June 2016, a total of 18 soil sampling sites from croplands (CL, n = 8), abandoned croplands (AL, n = 5), and native vegetation lands (NV, n = 5) were selected. Information about topography and cropping history at each soil site is exhibited in Table 1. The sampling sites at the same stage of post-agricultural restoration spaced 100 m apart. A 1 m × 1 m square was set up at each sampling site, and three duplicate subsites were selected from the corners of the square. A 0.3 m× 0.3 m plot with 0.5 m depth at each subsite was dug to collect soil samples. Soil samples were collected from the layer at the 0 ~ 10, 10 20, and 20 ~ 30 cm depth, orderly. The three duplicate soil samples at the same depth were mixed to be one sample. ithout cultivation and fertilization > 100 y, shurb-arbor land at present The dominant vegetation species under the three land use types were identi ed in the eld. Leaf samples from at least ve plants with same species were selected at the same stage of post-agricultural restoration. The mature leaves of the dominant vegetation species were collected at the high, middle, and low tree height. The leaves from each tree height and same species were mixed to be one sample. Litter samples were collected within a 1 m × 1 m square, in which the center of the square corresponds to one of the soil sampling sites.

Sample analysis
After washing the dust on the leaf surface with pure water for 3 times, the leaf and litter samples were dried at − 40℃ in a freezer dryer, then ground into powder by an attritor ). Soil samples were air-dried at room temperature (25℃) for at least 20 days after removing big gravel and roots. One part of the dried soil samples was conserved as the bulk soils after passing through a 2 mm sifter. Other parts of the dried soil samples, which were not crushed, were used to separate different-sized aggregates physically by the modi ed wet sieving method (Six et al. 1998). Concretely, the dried soil samples were slowly wetted via capillary water absorption, then naturally disintegrated into a series of different-sized aggregates. In pure water, these different-sized aggregates were passed through a 2000 µm, 250 µm, and 53 µm sifter in that order. The aggregates over 2000 µm in diameter were crushed by a tweezer, to make all aggregates passing through the 2000 µm sifter. Macro-aggregates (250 ~ 2000 µm) and microaggregates (53 ~ 250 µm) were collected after passing through the 250 µm and 53 µm sifters, respectively. The silt + clay sized fractions (< 53 µm) were extracted from the residual mixed liquid by centrifugation. The moist soil aggregates were dried in an oven at 55°C until constant weight, then weighed to calculate the mass-proportions of the different-sized aggregates.
The samples of bulk soils and different-sized aggregates were ground into powder by an agate mortar and then passed through a 149 µm sifter. The pulverized samples (< 149 µm) were treated using 0. repeatedly until the neutrality of the supernatant liquid, then were dried and ground into powder (< 149 µm). The SOC and SON concentrations in the pulverized samples were measured using a multi-elemental analyzer (Elementar, Vario TOC, Germany) in the Laboratory of Sur cial Environment Hydrogeochemistry, China University of Geosciences (Beijing). The precision was greater than ± 0.01% for C and better than ± 0.02% for N. Actual SOC and SON concentrations in the original bulk soils or aggregates can be calibrated based on the loss of carbonates and inorganic N, according to the method by . In brief, the measured values are multiplied by the ratio of sample mass after treating to before treating to obtain the actual SON and SOC concentrations in different-sized aggregates and bulk soils. The foliar C and N concentrations were also analyzed using the multi-elemental analyzer.
The N stable isotope ratio ( 15 N/ 14 N) of SON in the treated bulk soils and aggregates were determined by an isotope mass spectrometer (Thermo, MAT-253, USA) in the Central Laboratory for Physical and Chemical Analysis, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences. The measurements are expressed in standard δ notation (‰) to indicate the differences between the 15 N/ 14 N ratio of the samples and accepted standard materials (atmospheric N 2 ), where: Reference material GBW04494 (δ 15 N Air : 0.24‰ ± 0.13‰) was monitored by repeatedly measuring to determine the precision (greater than ± 0.1‰). The δ 15 N values of leaf and litter samples were also analyzed using the isotope mass spectrometer.

Statistical analysis
One-way ANOVA with the least signi cant difference (LSD) test was conducted to identify the signi cant differences in the SOC and SON concentrations of bulk soils, the proportions of different-sized aggregates, the aggregate-SOC and SON concentrations in bulk soils, soil Ca concentrations, soil Ca/Al ratios, the δ 15 N values of SON in bulk soils and different-sized aggregates, and the EF values at the three stages following post-agricultural restoration, at the signi cance level of P < 0.05. Linear and non-linear regression analyses were used to determine the variations of foliar N concentrations, δ 15 N values, and C/N ratios following post-agricultural restoration. Based on the linear and non-linear regression analyses, the equations of best-t lines were determined, and the coe cients of r and P-value showed the tting degree and validity, respectively. Pearson correlation and Spearman's rank correlation analyses were performed to determine the linear relationships and rank order relationships of SOC and SON concentrations in bulk soils and different-sized aggregates with soil Ca concentrations and soil Ca/Al ratios, respectively. Statistical analyses were performed by the SPSS 18.0 software (SPSS Inc., Chicago, IL, USA). All gures were done by the SigmaPlot 12.5 software (Systat Software GmbH, Erkrath, Germany) and Adobe Illustrator CS2 software (Adobe Systems Inc., California, USA).

Variations of foliar N concentrations, δ 15 N values, and C/N ratios
Regression analyses were used to determine the variations of foliar N concentrations, δ 15 N values, and C/N ratios following post-agricultural restoration (Fig. 2). Although their variations did not strictly correlate with the restoration time, signi cantly decreasing foliar N concentrations and δ 15 N values and increasing foliar C/N ratios following post-agricultural restoration were still observed. Additionally, foliar N concentrations of C 4 vegetation were lower than those of C 3 vegetation (Fig. 2a), while foliar C concentrations between them were not signi cantly different (Table S1). Thus, foliar C/N ratios of C 4 vegetation were higher than those of C 3 vegetation (Fig. 2c). However, foliar δ 15 N values between C 3 vegetation and C 4 vegetation were not signi cantly different (Fig. 2b).

SOC and SON concentrations and aggregate distribution
The SOC and SON concentrations of native vegetation lands were signi cantly higher than those of croplands and abandoned croplands at the 0 ~ 10 and 10 ~ 20 cm depth, while there were no signi cant differences between them at the 20 ~ 30 cm depth ( Fig. 3a and 3b). These results indicate that SOC and SON concentrations increased following post-agricultural restoration but only in the surface soils (0 ~ 20 cm depth). Macro-aggregates accounted for the largest proportion (> 60%) of total aggregates in the calcareous soils (Fig. 3c). In the soils at the 0 ~ 10 cm depth, macro-aggregate proportions in croplands were signi cantly lower than those in abandoned croplands and native vegetation lands, while the proportions of micro-aggregates and silt + clay sized fractions were signi cantly higher. Silt + clay sized fraction proportions at the 10 ~ 30 cm depth and macro-aggregate proportions at the 20 ~ 30 cm depth were not signi cantly different between the three land use types. These results indicated that macroaggregates increased, while micro-aggregates and silt + clay sized fractions decreased following postagricultural restoration. Furthermore, these variations were affected by soil depth. Additionally, compared to the croplands, macro-aggregate proportions in the 3 ~ 8 years abandoned croplands signi cantly increased, while the SOC and SON concentrations did not. The results indicated that the response of soil aggregates to post-agricultural restoration was more rapid than the response of SOC and SON.

Aggregate-SOC and SON concentrations in bulk soils
In the soil layers of the 0 ~ 10 and 10 ~ 20 cm depth, macro-aggregate-SOC and SON concentrations in bulk soils under native vegetation lands were signi cantly higher than those under croplands (Fig. 4). These concentrations under abandoned cropland were intermediary and did not show signi cant differences with those under cropland (lowest) and native vegetation land (highest). However, these concentrations under the three land use types were not signi cantly different at the 20 ~ 30 cm depth.
Additionally, the SOC and SON concentrations in micro-aggregates, silt + clay sized fractions, and bulk soils were not signi cantly different between the three land use types at the all soil depths. These results indicated that the macro-aggregates played the most important role in increasing SOC and SON storage during post-agricultural restoration.

Soil Ca concentrations and Ca/Al ratios
Soil Ca concentrations under native vegetation lands were signi cantly higher than those under croplands and abandoned croplands at the 0 ~ 10 and 10 ~ 20 cm depth; while there were no signi cant differences at the 20 ~ 30 cm depth (Fig. 5a). However, Ca belongs to an easily migrated element in the natural soil environment. The lower or higher Ca concentrations at different soil sites with the same bedrock can be caused by, in addition to the depletion or enrichment of Ca with land use change, the different degrees of chemical weathering (Li and Han 2021). Al is di cult to migrate during chemical weathering, thus soil Ca/Al ratios can eliminate the effects of chemical weathering on soil Ca concentrations and indicate the degree of soil Ca depletion or enrichment only caused by land use change (Balls et al. 1997). In the present study, soil Ca/Al ratios under native vegetation land were also signi cantly higher than those under croplands and abandoned croplands at the 0 ~ 10 and 10 ~ 20 cm depth (Fig. 5b). The result indicated that soil Ca in native vegetation land was indeed enriched compared to that in croplands and abandoned croplands at the 0 ~ 20 cm depth.

Soil δ 15 N of SON in bulk soils and aggregates and soil to plant 15 N EF values
In the bulk soils and different-sized aggregates at the 0 ~ 10 cm depth, the δ 15 N values of SON under croplands were signi cantly higher than those under abandoned croplands and native vegetation lands; while there were no signi cant differences at the 10 ~ 20 and 20 ~ 30 cm depth (Fig. 6). These results indicated that the SON of bulk soils and aggregates gradually enriched 15 N following post-agricultural restoration. Additionally, in the soils at all depths under the three land use types, the δ 15 N values of aggregate-associated SON followed the order: micro-aggregates < macro-aggregates < silt + clay sized fractions. Although these differences of δ 15 N values between different-sized aggregates were not statistically signi cant, the same trends that occurred in all soil samples indicated that the discrepancies of δ 15 N values in different-sized aggregates is relevant.

Increasing SOC sequestration during post-agricultural restoration
In the present study, SOC concentrations increased during post-agricultural restoration in surface soils  (Fig. 3c). These results indicated that soil aggregation gradually intensi es with the cease of agricultural disturbances. The increased soil aggregates will provide more physical protection for SOC (Oades and Waters 1991), which is conducive to enhancing SOC accumulation and stabilization in the soil environment (Six and Paustian 2014).
Furthermore, only macro-aggregate associated SOC concentration in bulk soils signi cantly increased during post-agricultural restoration (Fig. 4a). This result indicates that macro-aggregates mainly affect SOC storage after cropland abandonment, relative to micro-aggregates or silt + clay sized fractions.
Generally, macro-aggregates are more sensitive to land use management than other small-sized aggregates (Franzluebbers and Arshad 1997), which is determined by the physical and chemical stability of different-sized aggregates (Six et al. 2004). It suggests that tillage activities are more likely to destroy macro-aggregates rather than small-sized aggregates, and macro-aggregates can be more rapidly restored after stopping cultivation. Moreover, the SOC within macro-aggregates accounts for 60% ~ 80% of the total SOC in bulk soils ). Thus, the variations of macro-aggregates are closely associated with SOC dynamics during post-agricultural restoration.
The calcareous soils developed from limestones (mainly CaCO 3 ) contain abundant Ca 2+ , which can easily absorb clay minerals and organic matter to form stable OM-Ca 2+ -mineral complexes . As a form of physical protection, the organic-inorganic complexes can enhance SOC stabilization to resist microbial decomposition (Schmidt et al. 2011). Thus, soil Ca concentrations generally show a positive effect on SOC sequestration. Soil Ca in native vegetation lands were enriched compared to that in croplands and abandoned croplands (Fig. 5). This result is likely attributed to the strong loss of Ca via leaching along with nitrate and harvesting under long-term cultivation (Guo et al. 2010), while soil Ca is lost only slightly under the native vegetation. As the main source of soil Ca, rock weathering should continue for several hundred years at least, which can cause a signi cant increase in soil Ca concentration. Theoretically, soil Ca concentration decreases slowly during several decades of post-agricultural restoration, while soil Ca accumulation is almost impossible. However, soil Ca accumulated in the limestone region after a relatively short-term post-agricultural restoration (Fig. 5). On the one hand, the rapid limestone weathering rate can replenish depleted soil Ca pool after cropland abandonment ). On the other hand, the formation of OM-Ca 2+ -mineral complexes in the SOC-rich soils also reduces Ca 2+ leaching with increasing SOM. The rapid replenishment of soil Ca is closely associated with increased SOC sequestration following post-agricultural restoration in the karst catchment. Li et al.
(2017) reported that soil Ca played an important role in improving SOC and N storage after agricultural abandonment. Pearson correlation coe cients showed that there were no linear relationships between soil Ca concentration and SOC concentration of bulk soils and different sized aggregates (Table 2).
However, a signi cant non-relationship (P < 0.01) between them was present at all soil depths according to Spearman's rank correlation coe cients. These results veri ed the close relationships between soil Ca replenishment and SOC sequestration in the karst region, and underscored a non-linear increase between them.  Similar to SOC, SON was protected by soil aggregates and OM-Ca 2+ -mineral complexes due to the extensive homology (SOM) between them, as shown in Fig. 4b. In addition to increased SON concentration (Fig. 3b), these protection mechanisms also enhance SON stabilization following postagricultural restoration (Six et al. 1998(Six et al. , 2000(Six et al. , 2002. The increased SON stabilization likely restricts N mineralization and subsequent nitri cation to produce available N, resulting in decreasing soil N availability. There were two pieces of evidence to verify the decreased soil N availability during postagricultural restoration. Firstly, foliar N concentrations decreased and C/N ratios increased after cropland abandonment ( Fig. 2a and 2b), which could affect litter N concentrations and C/N ratios. Generally, the decomposition of litter with a low N concentration and high C/N ratio will be restrained at the initial stage of litter decomposition (Xia et al. 2020). Thus, the N mineralization rate of litter is also slower under native vegetation lands than that under croplands. Secondly, the δ 15 N values of aggregate-associated SON followed the order: micro-aggregates < macro-aggregates < silt + clay sized fractions (Fig. 6) (Six et al. 2000). The SON within micro-aggregates is more stable compared to within macro-aggregates due to the stronger stabilization of micro-aggregates and the protection provided by macro-aggregates (Beare et al. 1994). During post-agricultural restoration, decreased macro-aggregate turnover with the cease of cultivation activities is conducive for the formation and stabilization of micro-aggregates inside (Six et al. 2000). Accordingly, fresh SON within micro-aggregates can be physically protected as well.
Thus, the N mineralization rate of fresh SON is also slower under native vegetation than that under cropland, due to increases in soil aggregation.
During post-agricultural restoration, increased plant biomass enhances available N uptake, while decreases in soil N availability slow down the supply of available N via SON mineralization. After cropland abandonment, the slow SON mineralization and faster plant N uptake lead to less inorganic N loss, i.e., tending towards closed N cycles. The more negative foliar 15 N-abundance (Fig. 2b) and soil to plant 15 N EF values (Fig. 7) indicates the alteration from leaky to closed N cycles following postagricultural restoration (Xiao et al. 2018

A conceptual model of post-agricultural restoration in the karst ecosystem
In the present study, we propose a conceptual model about the increasing SOC sequestration and closing N cycles following post-agricultural restoration in the karst ecosystem (Fig. 8). However, considering the differences in soil type, climate, vegetation restoration approach (natural restoration without disturbance, forest plantation, grass plantation, and so on), the main mechanism of SOC sequestration and soil N transformation processes and magnitude differ between the GGP programs at different region. For future research, the conceptual model should be adjusted according to the local situation.   The concentrations of aggregate-SOC (a) and SON (b) in the bulk soils at the three stages following postagricultural restoration. Error bar is standard error (SE) of mean. Different lowercase letters indicate signi cant differences in aggregate-SOC and SON concentrations in the same depth of soils between the three stages following post-agricultural restoration, based on the one-way ANOVA with LSD test at the level of P < 0.05. CL, cropland; AL, abandoned cropland; NV, native vegetation land A conceptual model about the increasing SOC sequestration and closing N cycle following postagricultural restoration in the karst ecosystem. The green arrows indicate the positive processes; the red arrows indicate the negative processes.