Effects of Grassland Conversion in the Chinese Chernozem Region on Soil Carbon, Nitrogen, and Phosphorus

: Converting natural vegetation to other types of land utilization is presently the most common land use change around the world. Conversion of grassland plays an important role in estimating the regional nutrients budget and assessing ecological effects. However, few studies have estimated the impacts of grassland conversion on soil nutrients under different soil pedogenic horizons. This study selected three sites, which were covered by grassland, shelterbelt, and cropland. The study evaluated the effects of grassland conversion and soil pedogenic horizon (to a depth of 100 cm) on the soil organic carbon (SOC), soil total nitrogen (STN), and soil total phosphorus (STP) concentrations and stocks in the Chinese Chernozem region. The results revealed that signiﬁcant ( p < 0.05) differences were seen after grassland conversion for concentrations and stocks of SOC, STN, and STP. The transformation from grassland to shelterbelt and cropland plantations caused soil carbon and soil nitrogen losses but caused soil phosphorus accumulations. Moreover, conversion of grassland made SOC, STN, and STP all drop below the B k -horizon. Changes in the SOC and STP on an area basis were the greatest after conversion of grassland to cropland, for concentrations of − 16% and 26% and for stocks of − 15% and 32%, respectively. Land use change and soil pedogenic horizon primarily inﬂuenced the distribution patterns of nutrients concentrations and stocks. However, grassland conversion effects on nutrients were mainly at surface horizons. Soil properties, such as calcium carbonate (CaCO 3 ) and soil texture, affected the nutrients from the B k -horizon to the C-horizon. This study indicates that land use management policies should protect natural grasslands to minimize losses of SOC, STN, and STP.


Introduction
Grasslands, which occupy approximately 20% of the terrestrial land surface and store 34% of the total terrestrial carbon, have stored about 89% of the sequestration carbon in the soil [1,2]. The conversion of grassland to other land uses is vital from the perspective of soil fertility and sustainable management and for their effects on atmospheric CO 2 concentration and global warming. Such conversion has many negative influences on soil organic carbon (SOC), soil total nitrogen (STN), and soil total phosphorus (STP) circulation by altering soil C, N, and P accumulation and turnover [3][4][5][6]. Nonetheless, the change patterns of SOC, STN, and STP are not fully understood, and knowledge of the driving factors remains unclear.
The effects of grassland conversion on SOC, STN, and STP have received more and more attention all over the world because the massive emission of the oxides of C and N has worsened the earth's climate. Conversion from grassland to shelterbelt and cropland resulted in great soil SOC and STN loss [7,8]. However, there is a lack of comprehensive understanding of the changes of SOC, STN, and STP after conversion of grassland to other land use types. Knowing the distribution of SOC, STN, and STP concentrations and stocks more than 70% of the precipitation occurring during June, July, and August. The soil is identified as Chernozem by Chinese Soil Taxonomy, or as Ustoll by US Soil Taxonomy. The main native vegetation is meadow steppe, which initially covered the whole study area. There is an obvious accumulation of calcium carbonate under the humus horizon. Moreover, the humus horizon in the area is thin. At present, a considerable part of the natural vegetation in the area of Chernozem have been successively reclaimed for farmland, resulting in fertility degradation and acidification of soil. Figure 1. Location of the study area and sampling sites for different land-use types in the Chinese Chernozem region. Lowercase letters: a represents grassland, b represents shelterbelt, and c represents cropland. Numbers after lowercase letters indicate the replicate sites for each land use. Site 1 was located in Guyang Village, site 2 was located in Dafu Village, and site 3 was located in Wanshun Village.

Soil Sampling and Measurements
Samples were collected following soil pedogenic horizons for a depth of 100 cm in late October 2017. Three sampling sites were covered by native grassland (meadow steppe), shelterbelt (Populus simonii Carr.) as farmland shelterbelt, and cropland ( Figure 1). The shelterbelt and cropland are adjacent. The dominant plant species in the grassland was Leymus chinensis (Trin.) Tzvel. The crop species primarily planted in the cropland was maize (Zea mays L.). The cropland was converted from grassland in the 1960s, and the shelterbelt was converted in the 1980s. The grassland was only grazed in the autumn and winter, with hay cut in the autumn. At each sampling site center, we established three representative plots (1 × 1 m for grassland and cropland, 5 × 5 m for shelterbelt) with a spacing of 15-30 m. Soil samples were gathered at five soil pedogenic horizon from top to bottom, such as A-horizon (humus horizon), AB-horizon (transition horizon), Bk-horizon (eluvial horizon, that is caliche), BC-horizon (transition horizon), and C-horizon (parent material horizon), by Soil Taxonomy according to soil color, lime reaction, lime pseudomycelium, plant root distribution, and clay for each plot using a soil shovel. Under three land use types, the average depth of each soil pedogenic horizon was 0-30, 30-55, 55-76, 76-90, and 90-100 cm. Therefore, a total of 135 composite soil samples (3 sites × 9 plots × Figure 1. Location of the study area and sampling sites for different land-use types in the Chinese Chernozem region. Lowercase letters: a represents grassland, b represents shelterbelt, and c represents cropland. Numbers after lowercase letters indicate the replicate sites for each land use. Site 1 was located in Guyang Village, site 2 was located in Dafu Village, and site 3 was located in Wanshun Village.

Soil Sampling and Measurements
Samples were collected following soil pedogenic horizons for a depth of 100 cm in late October 2017. Three sampling sites were covered by native grassland (meadow steppe), shelterbelt (Populus simonii Carr.) as farmland shelterbelt, and cropland ( Figure 1). The shelterbelt and cropland are adjacent. The dominant plant species in the grassland was Leymus chinensis (Trin.) Tzvel. The crop species primarily planted in the cropland was maize (Zea mays L.). The cropland was converted from grassland in the 1960s, and the shelterbelt was converted in the 1980s. The grassland was only grazed in the autumn and winter, with hay cut in the autumn. At each sampling site center, we established three representative plots (1 × 1 m for grassland and cropland, 5 × 5 m for shelterbelt) with a spacing of 15-30 m. Soil samples were gathered at five soil pedogenic horizon from top to bottom, such as A-horizon (humus horizon), AB-horizon (transition horizon), B k -horizon (eluvial horizon, that is caliche), BC-horizon (transition horizon), and C-horizon (parent material horizon), by Soil Taxonomy according to soil color, lime reaction, lime pseudomycelium, plant root distribution, and clay for each plot using a soil shovel. Under three land use types, the average depth of each soil pedogenic horizon was 0-30, 30-55, 55-76, 76-90, and 90-100 cm. Therefore, a total of 135 composite soil samples (3 sites × 9 plots × 5 soil horizons) were collected. Soil samples did not exceed 100 cm depth, because the parent material horizon was generally above 100 cm depth, as well as for comparison analysis between different land uses.
Samples were allowed to air dry and were ground through a 2 mm sieve to measure soil texture. Samples that were sifted through a 0.25 mm mesh were used to determine the Sustainability 2021, 13, 2554 4 of 13 SOC, STN, and calcium carbonate (CaCO 3 ). Samples that were sifted through a 0.15 mm mesh were used to determine the STP concentration. Undisturbed samples were also excavated to measure the soil bulk density (BD) in the center of each horizon by using a metal cutting ring (diameter: 5 cm, length: 5 cm). The SOC concentration was measured by the dichromate oxidation method [13]. The STN concentration was measured by the Kjeldahi method [13]. The STP concentration was measured by acid-soluble molybdenum antimony colorimetric method after digesting with HClO 4 -H 2 SO 4 at high temperatures [13]. The CaCO 3 content was determined by using the acid-alkali neutralization titrimetric method [24]. The soil texture was determined according to the pipette method [24]. The soil bulk density was measured by oven-drying each soil sample at 105 • C for 24 h. The soil physicochemical characteristics of the three land uses are shown in Table 1.

Calculations and Statistical Analyses
The SOC, STN, and STP stocks were used to show the SOC, STN, and STP weight of the unit area of a certain soil horizon, which were calculated by the following equation [25]: where Stock (kg m -2 ) is the storage of SOC, STN, and STP; Con i is the SOC, STN, and STP concentration of the ith horizon (g kg -1 ); BD i is the soil bulk density of the ith horizon (g cm -3 ); ST i is the proportion of coarse fragments (>2 mm) of the ith horizon (ST can be considered to be zero, for no visible coarse fragments); ∆d i is the thickness of the ith horizon (cm); and n is the number of soil pedogenic horizons. The factor 0.01 is the unit conversion coefficient. Statistical analyses were carried out with SPSS 17.0 statistics software. One-way ANOVA and the least significant difference (LSD) test were used to determine differences between treatment means when the land use change and/or soil pedogenic horizon effects were significant. Two-way ANOVA was used to analyze the interaction between soil pedogenic horizons and land uses on the soil nutrients concentrations and stocks. We investigated the relationships among SOC, STN, STP, and influencing variables using the Pearson correlation coefficients analysis. Then, in order to explore the contribution of different soil properties to the variation of SOC, STN, and STP concentrations, redundancy analysis (RDA) was performed with CANOCO 5.0. RDA is a linear constrained ordination method [26], which conforms to the explanatory variable gradient of <3.0 SD (Standard Deviation) in the study case. The factors, log (1 + x) transformed, centralized, and standardized in CANOCO 5.0, were identified to show the most influential gradients by forward selection. Adding explanatory variables until further adding of variables did not significantly (p < 0.05) improve the explanatory ability of the model. This was assessed with Monte Carlo permutations test.

Variations in SOC, STN, and STP Concentrations
After grassland conversion, SOC concentration demonstrated significant differences (p < 0.01) among the whole profile (except AB-horizon and B k -horizon), especially in the A-horizon (p < 0.001) ( Figure 2). Higher SOC concentration occurred in grassland than shelterbelt and cropland among the whole profile, with the exception of A-horizon ( Figure 2a). Compared with grassland, SOC concentration decreased by 17-43% in shelterbelt site (except A-horizon) and decreased by 2-35% in cropland site for different soil horizons, with an average loss of 0.73 g kg -1 and 1.24 g kg -1 , respectively.
Statistical analyses were carried out with SPSS 17.0 statistics software. One-way ANOVA and the least significant difference (LSD) test were used to determine differences between treatment means when the land use change and/or soil pedogenic horizon effects were significant. Two-way ANOVA was used to analyze the interaction between soil pedogenic horizons and land uses on the soil nutrients concentrations and stocks. We investigated the relationships among SOC, STN, STP, and influencing variables using the Pearson correlation coefficients analysis. Then, in order to explore the contribution of different soil properties to the variation of SOC, STN, and STP concentrations, redundancy analysis (RDA) was performed with CANOCO 5.0. RDA is a linear constrained ordination method [26], which conforms to the explanatory variable gradient of <3.0 SD (Standard Deviation) in the study case. The factors, log (1 + x) transformed, centralized, and standardized in CANOCO 5.0, were identified to show the most influential gradients by forward selection. Adding explanatory variables until further adding of variables did not significantly (p < 0.05) improve the explanatory ability of the model. This was assessed with Monte Carlo permutations test.

Variations in SOC, STN, and STP Concentrations
After grassland conversion, SOC concentration demonstrated significant differences (p < 0.01) among the whole profile (except AB-horizon and Bk-horizon), especially in the A-horizon (p < 0.001) ( Figure 2). Higher SOC concentration occurred in grassland than shelterbelt and cropland among the whole profile, with the exception of A-horizon ( Figure  2a). Compared with grassland, SOC concentration decreased by 17-43% in shelterbelt site (except A-horizon) and decreased by 2-35% in cropland site for different soil horizons, with an average loss of 0.73 g kg -1 and 1.24 g kg -1 , respectively. Horizon distribution of (a) soil organic carbon (SOC) concentration, (b) soil total nitrogen (STN) concentration, and (c) soil total phosphorus (STP) concentration in grassland, shelterbelt, and cropland. Error bars represent the standard error of the mean. Within each horizon, different lowercase letters indicate significant differences in the SOC, STN, and STP concentrations among the three land uses (p < 0.05). On the vertical axis, A represents A-horizon, AB represents AB-horizon, Bk represents B k -horizon, BC represents BC-horizon, and C represents C-horizon.
The profile distribution of the STN concentration was similar to that of SOC after grassland conversion. The profile distribution of the STN concentration exhibited much less fluctuation than the SOC concentration ( Figure 2b). Higher STN concentration occurred in grassland than shelterbelt among the whole profile (except A-horizon), but that in the soils of cropland showed a slight change more than the soils of grassland in the AB-horizon and B k -horizon. Compared with grassland, STN concentration decreased by 2-18% in the shelterbelt site (except A-horizon) and decreased by 3-21% in cropland site (except Sustainability 2021, 13, 2554 6 of 13 AB-horizon and B k -horizon) for different soil horizons, with an average loss of 0.05 g kg -1 and 0.03 g kg -1 , respectively.
The mean STP concentration exhibited slight differences from B k -horizon to C-horizon responses to the conversion of grassland ( Figure 2). STP concentration of grassland was lower than shelterbelt and cropland above the B k -horizon, but higher than shelterbelt and cropland below the B k -horizon ( Figure 2c). Moreover, STP concentration of grassland had the least changes among the whole profile with respect to cropland and shelterbelt. STP concentration increased by an average of 6% in the shelterbelt site and increased by an average of 22% in the cropland site compared with the grassland site for different soil horizons.

Changes in SOC, STN, and STP Stocks
The study found that land use changes significantly affected SOC, STN, and STP stocks ( Figure 3). The total SOC stocks in the grassland were notably different (p < 0.001) from, and more than, those of the shelterbelt and cropland. The distribution of SOC stocks was highly significantly (p < 0.001) greater in the grassland than in the shelterbelt and cropland from the B k -horizon to the C-horizon (Figure 3a). The total STN stocks in the grassland were higher than those of the shelterbelt and cropland, but not significantly different from the B k -horizon and the C-horizon (Figure 3b). The STP stocks of grassland were significantly lower (p < 0.05) than the shelterbelt (except AB-horizon) and cropland above B k -horizon, but were significantly higher (p < 0.05) than shelterbelt and cropland below B k -horizon (Figure 3c).
lowercase letters indicate significant differences in the SOC, STN, and STP concentrations among the three land uses (p < 0.05). On the vertical axis, A represents A-horizon, AB represents AB-horizon, Bk represents Bk-horizon, BC represents BC-horizon, and C represents C-horizon.
The profile distribution of the STN concentration was similar to that of SOC after grassland conversion. The profile distribution of the STN concentration exhibited much less fluctuation than the SOC concentration (Figure 2b). Higher STN concentration occurred in grassland than shelterbelt among the whole profile (except A-horizon), but that in the soils of cropland showed a slight change more than the soils of grassland in the ABhorizon and Bk-horizon. Compared with grassland, STN concentration decreased by 2-18% in the shelterbelt site (except A-horizon) and decreased by 3-21% in cropland site (except AB-horizon and Bk-horizon) for different soil horizons, with an average loss of 0.05 g kg -1 and 0.03 g kg -1 , respectively.
The mean STP concentration exhibited slight differences from Bk-horizon to C-horizon responses to the conversion of grassland (Figure 2). STP concentration of grassland was lower than shelterbelt and cropland above the Bk-horizon, but higher than shelterbelt and cropland below the Bk-horizon (Figure 2c). Moreover, STP concentration of grassland had the least changes among the whole profile with respect to cropland and shelterbelt. STP concentration increased by an average of 6% in the shelterbelt site and increased by an average of 22% in the cropland site compared with the grassland site for different soil horizons.

Changes in SOC, STN, and STP Stocks
The study found that land use changes significantly affected SOC, STN, and STP stocks ( Figure 3). The total SOC stocks in the grassland were notably different (p < 0.001) from, and more than, those of the shelterbelt and cropland. The distribution of SOC stocks was highly significantly (p < 0.001) greater in the grassland than in the shelterbelt and cropland from the Bk-horizon to the C-horizon (Figure 3a). The total STN stocks in the grassland were higher than those of the shelterbelt and cropland, but not significantly different from the Bkhorizon and the C-horizon (Figure 3b). The STP stocks of grassland were significantly lower (p < 0.05) than the shelterbelt (except AB-horizon) and cropland above Bk-horizon, but were significantly higher (p < 0.05) than shelterbelt and cropland below Bk-horizon (Figure 3c).  Figure 3. The amount of (a) SOC, (b) STN, and (c) STP stocks in each horizon of the soil profile under different land use types. Different lowercase letters indicate significant differences in the SOC, STN, and STP stocks among different land use types in the same soil horizon (p < 0.05). The error bars are the standard errors. On the vertical axis, A represents A-horizon, AB represents AB-horizon, Bk represents B k -horizon, BC represents BC-horizon, and C represents C-horizon.
Compared to the grassland, the ∆SOC, ∆STN, and ∆STP stocks in cropland and shelterbelt express the carbon sequestration capability of land use changes. Positive values of ∆SOC, ∆STN, and ∆STP stocks illustrate carbon, nitrogen, and phosphorus accumulations; however, negative values illustrate carbon, nitrogen, and phosphorus losses. The ∆SOC stocks indicated carbon generally lost in the whole profile after conversion of grassland, with the exception of the A-horizon in shelterbelt and AB-horizon in cropland (Figure 4a). The highest SOC loss was detected in the cropland (Figure 4a), especially in the A- horizon. ∆STN stocks indicated that nitrogen was lost through the whole profile, with the exception of the A-horizon in the shelterbelt and AB-horizon in the cropland (Figure 4b). In contrast to the ∆SOC and ∆STN stocks, positive ∆STP stocks were detected after the grassland conversion to cropland and shelterbelt in the A-horizon and AB-horizon; however, negative ∆STP stocks were found under B k -horizon (Figure 4c). The mean ∆SOC, ∆STN, and ∆STP stocks of each horizon in the shelterbelt site and the cropland site were −0.26 and −0.33 kg m -2 , −0.02 and −0.01 kg m -2 , 0.01 and 0.03 kg m -2 , respectively. On the horizontal axis, A represents A-horizon, AB represents AB-horizon, Bk represents B k -horizon, BC represents BC-horizon, and C represents C-horizon.

Factors Affecting SOC, STN, and STP
Significant influences of land use change and soil horizon on the concentrations and stocks of SOC, STN, and STP (p < 0.05) can be seen in Table 2. However, land use change had lower effects than soil horizon on nutrients. For example, land use change had weaker influences on STN concentrations and stocks with respect to soil horizon. Some interactions between land use change and soil horizon had a highly significant effect on nutrient elements (p < 0.05), whereas effects on the concentrations and stocks of STN weakened (p < 0.05). Pearson correlation analysis indicated that SOC, STN, and STP concentrations were mainly correlated with CaCO 3 , pH, and soil texture (Table 3). CaCO 3 content significantly affected soil nutrients, which usually occurred from the B k -horizon to the C-horizon. pH value significantly affected SOC, and STN was usually under the B k -horizon. Soil texture significantly influenced SOC, STN, and STP concentrations in the B k -horizon. Therefore, CaCO 3 , pH, and soil texture mainly controlled the spatial patterns of nutrients in the deep horizons. However, land use changes significantly affected the soil nutrients in the surface horizons, mainly above the B k -horizon ( Figure 2). Note: * Significant at 0.05 levels (bilateral); ** Significant at 0.01 levels (bilateral).
The RDA was also conducted to study the relative contributions of the dominating soil factors to SOC, STN, and STP in each soil horizon ( Figure 5). On average, land use, CaCO 3 , pH, and clay + silt explained 17.1%, 15.4%, 17.6%, and 29.1% of the variations in the SOC concentrations; 11.4%, 22.4%, 18.4%, and 23.4% of the variations in the STN concentrations; and 18.0%, 21.1%, 14.7%, and 27.0% of the variations in the STP concentrations, respectively. The relative contributions of land use were greater than other factors in the A-horizons (except to STP). Nevertheless, as soil horizons down, the impacts of land use generally reduced, contributing the least to SOC, STN, and STP concentrations in the B k -horizon. The effects of pH did not control soil nutrients steadily. CaCO 3 and clay + silt contents have begun to play a significant role from B k -horizon to C-horizon. Therefore, land use changes controlled the distributions of SOC, STN, and STP concentrations in the upper horizons, whereas CaCO 3 and soil texture principally influenced the soil nutrients in the deep horizons. On the vertical axis, A represents A-horizon, AB represents AB-horizon, Bk represents Bk-horizon, BC represents BC-horizon, and C represents C-horizon.

Changes of Nutrients along Soil Pedogenic Horizons After Grassland Conversion
Land use changes, soil pedogenic horizon, and their interactions significantly affected concentrations and stocks of SOC, STN, and STP (Table 2).
Grassland conversion has been displayed to diverge significantly in some soil characteristics, particularly in the allocation of C, N, and P along soil pedogenic horizons [3,14,27]. Generally, the study discovered that grassland conversion depleted SOC, while STN pools increased STP pools. Meanwhile, land use changes played a significant role on SOC, STN, and STP, mainly in the A-horizon, which was consistent with previous research results [5]. The soil pedogenic horizon provides a vital role by influencing nutrient

Changes of Nutrients along Soil Pedogenic Horizons After Grassland Conversion
Land use changes, soil pedogenic horizon, and their interactions significantly affected concentrations and stocks of SOC, STN, and STP (Table 2).
Grassland conversion has been displayed to diverge significantly in some soil characteristics, particularly in the allocation of C, N, and P along soil pedogenic horizons [3,14,27]. Generally, the study discovered that grassland conversion depleted SOC, while STN pools increased STP pools. Meanwhile, land use changes played a significant role on SOC, STN, and STP, mainly in the A-horizon, which was consistent with previous research results [5]. The soil pedogenic horizon provides a vital role by influencing nutrient profile distributions. The topsoil nutrients are mainly extracted from litter, plant roots, and root exudates; however, the subsoil nutrients are mainly extracted from root shedding and leaching of nutrients [28]. Therefore, the effects of land use change on SOC, STN, and STP were significant in surface soils. However, the effect of grassland conversion should not be ignored below the B k -horizon ( Figure 5).
Intriguingly, grassland conversion led to significant differences in SOC concentrations and stocks (p < 0.05) throughout the whole profile (Figures 2 and 3). STN concentrations and stocks exhibited much less variation than SOC concentrations and stocks. STP concentration, however, showed slight differences after grassland transition to shelterbelt and cropland under the B k -horizon. P is a sedimentary element, mainly from lengthy time periods of rock weathering. Nevertheless, the soil material in the study area is mainly Quaternary loess deposits, the source of soil P is relatively fixed, and the degree of weathering in the whole profile difference is not large. Thus, the spatial distribution of soil P storage between soil pedogenic horizons is relatively stable [29]. The highest STP stocks were found in the cropland, which can be primarily attributed to the input of external phosphate fertilizer.
The grasslands are mainly composed of poaceae and legume grasses with high aboveground biomass and high hay yield [30]. It suggested that grassland vegetation as a dominant genus had the highest C and N storage in the study area. The SOC and STN stocks in the grassland were 11.30 and 1.18 kg m -2 , respectively. The SOC stocks were just slightly less than the overall grassland SOC stocks trends in China (13.16 kg m -2 ) [25], and the STN stocks were also slightly less than the overall grassland STN stocks trends in China (1.25 kg m -2 ) [31]. Meanwhile, the SOC and STN stocks in the non-degraded grassland were 8.51 and 0.96 kg m -2 in the whole profile (0-20 cm) of the Qinghai-Tibetan Plateau [32]. Due to the spatial heterogeneity of soil, the SOC and STN stocks in our studied grassland site were unlike those in other areas.

Effects of Grassland Conversion and Soil Pedogenic Horizon on Changes of Nutrient Stocks
Increasing populations inevitably gave rise to a wide range of land use conversions. Moreover, land use changes had various influences on soil carbon and soil nitrogen sequestration [33]. In particular, land use transitions to or from grass will promote significant shifts in soil C and N storage [34]. The results illustrated this point (Figure 4). In the study area, the cropland exhibited a large loss in SOC and STN stocks, the ratio of which reached 15% and 4%, compared with grassland. Due to lack of nutrient replenishment in the shelterbelt, SOC and STN stocks were lost at about 12% and 9% respectively, with respect to grassland. Generally, the root system, root exudates, and dissolved organic matter are considered the primary sources of C and N in the lower soil horizons [35]. Grasslands usually have a well-developed deep root system, and root biomass reductions were slower from upper horizons to deeper horizons than shelterbelt and cropland, thus causing slower reductions in SOC and STN storage in the lower soil horizons than in the shelterbelt and cropland. Thus, the grassland may be a better choice for soil carbon and nitrogen sequestration in the study sites of semiarid areas with precipitation ranging from 370 mm to 470 mm.
The results indicated that grassland conversion depleted cropland the most in terms of SOC and STN sources in the A-horizon, causing 23% and 16% losses, respectively, but contributed to cropland the most in terms of STP storage in the A-horizon with respect to other soil horizons. This finding conformed to the results of a few studies [23,36]. The decreases in SOC and STN stocks in the A-horizon after the transformation from grassland to cropland were potentially due to agricultural activities. Tillage could give rise to soil erosion and disruption in soil aggregates, which might wash away the soil organic matter [37]. In addition, decreases in the SOC, STN, and STP storage after conversion from grassland to cropland were also discovered from B k -horizon to C-horizon (except that STP stocks increased in the B k -horizon) (Figure 4), indicating that C, N, and P in the deep soil horizons could also be exhausted by agricultural practices.
Grassland conversion led to 12% SOC accumulations in the A-horizon of the shelterbelt, whereas it resulted in massive losses in other horizons. This corresponded with the study showing that a significantly higher SOC concentration in shelterbelt than ingrassland was evident only in the topsoil [38]. Therefore, grassland conversion to shelterbelt led SOC stocks loss of 1.31 kg m -2 in sum. Additionally, differences in STN losses between shelterbelt and cropland were small, and STPs of two sites above B k -horizon were all accumulated for the two adjacent sites [39]. To sequester C and N, the shelterbelt is not suitable for all circumstances, so more grassland than shelterbelt should be established in sandy land and low hills [40].

Variation of Nutrients with Soil Properties
Land use changes contributed to SOC, STN, and STP mainly at the surface horizon; however, the effects of this could not be ignored in the deep horizons. Grassland conversion changed the basic physical and chemical properties of soil, which were associated with changes in SOC, STN, and STP. Some research enables us to better understand the relationship between soil physical and chemical characteristics and SOC, STN, and STP sequestrations after conversion of grassland [5,9,32].
Chernozem has a calcic horizon (B k -horizon) under the humus horizon and with typical morphological characteristics, including aggregations of CaCO 3 . The results showed that CaCO 3 content of the soil had significant effects on the SOC, STN, and STP concentrations from the B k -horizon to the C-horizon (Table 3, Figure 5). The large aggregations of CaCO 3 in the B k -horizon, to some extent, can protect SOC from rapid mineralization. CaCO 3 can increase the physical protection of organic carbon in soil aggregates and provide chemical protection of organic carbon through calcium ion bridge effect, which reduces the sensitivity of organic carbon to be decomposed by microorganisms [41][42][43]. The mean CaCO 3 content of grassland was the highest among the three land uses, so grassland can be expected to protect more SOC within Ca 2+ aggregates, forming stable SOC. However, the mechanisms of how CaCO 3 affects STN and STP are not presently well understood.
In the study, it was discovered that SOC, STN, and STP were mainly positively correlated with pH, suggesting that grassland conversion aggravated the SOC, STN, and STP losses by changing the pH of soil. The content of CaCO 3 increases soil pH value, maintaining the capacity of SOC absorption [44].
Grassland was dominated by sandy soil texture with raised clay contents along the soil pedogenic horizon in our study area, which was associated with rapid grassland establishment and high productivity. Generally, clay content in soil was conducive to enhancing the soil fertility and plants, as well as growth nutrient absorption capacity of plants. Approximately 55-57% of the organic C and 62-68% of the organic N were stored in the clay composition [45]. The variation of clay in shelterbelt and cropland reached the maximum in the B k -horizon, compared with grassland. Soil texture markedly influenced SOC and STN concentrations in the B k -horizon, mainly by reducing microbial decomposition or increasing water holding capacity [46]. However, the increase in CaCO 3 appears to induce the proportion of particles <0.25 mm to decrease [47,48]. In accordance with this, the mean CaCO 3 content was the highest in grassland, which had the lowest clay among the three land uses. However, the decrease in clay in grassland without the decrease in SOC and STN suggested that it may have resulted from high aboveground biomass and the strong root system of grassland.

Conclusions
Variations in soil nutrients concentrations and stocks along the soil pedogenic horizons after conversion of grassland in the Chinese Chernozem region were revealed in this study. The results showed that changes in SOC, STN, and STP mainly depended on soil pedogenic horizons, land use changes, and soil properties. Grassland conversion notably impacted the SOC, STN, and STP concentrations and stocks along the soil pedogenic horizons (p < 0.05). Conversion of grassland resulted in losses of soil carbon and nitrogen but accumulations of phosphorus. It had different edaphic factors controlling the accumulation of SOC, STN, and STP in different soil pedogenic horizons with grassland conversion. The upper soil pedogenic horizons were found to be principally influenced by land use changes, while the soil below the B k -horizon was controlled by both land uses and soil properties. CaCO 3 and soil texture affected the nutrients concentrations in and below the B k -horizon, possibly for the aggregations of CaCO 3 in the B k -horizon. Based on the research, grassland conversion should be paid more attention in semiarid regions.