Soil organic nitrogen variation shaped by diverse agroecosystems in a typical karst area: evidence from isotopic geochemistry

Study area description and soil sampling

The study area (27°35′–28°21′N and 108°18′–108°48′E, Fig. 1) is situated in Muhuang Town, the largest town in Yinjiang County, which is a karst area in northeast Guizhou Province. Yinjiang County possesses 13 towns with a 1,968.1 km² area, and the average population intensity is 145 persons per square kilometer. Yinjiang County is controlled by a subtropical monsoon climate with a combination of precipitation and heat. The number of rainfall days in 2016 was 147, with the majority occurring in summer, and the annual relative humidity was 78% in Yinjiang County (Tongren Bureau of Statistics, 2017). The extent of soil erosion in southwestern China is shown in Fig. 1A. Most soils in Yinjiang County slowly evolved from carbonate rocks and sandstone, which are characterized by weak soil aggregate stability and strong leaching (Zhang et al., 2021). Furthermore, the abundant mountains and frequent high-intensity precipitation in the study area also easily induce soil loss. The statistical yearbook of Tongren City in 2017 reported that the actual arable land covered 18,588 hectares and crop production reached 144,800 tons, while the stock of goats reached 66,100 at the end of 2016 in Yinjiang County (Tongren Bureau of Statistics, 2017).

A large amount of cropland has been abandoned to protect the ecological environment under the policy of grain for green in the last 20 years. The afforestation area reached 9,171.3 hectares in 2016 (Tongren Bureau of Statistics, 2017). However, the total area of arable land has limited variation, and the number of grazing goats has increased in recent years because agriculture is considered the primary industry in Yinjiang County. The wide distribution of abandoned cropland and forest land provides an appropriate opportunity to analyze the disturbance of soil properties by agricultural activities. A total of 47 samples were collected from three soil profiles under different land uses in September 2016 (Fig. 1). Of which 20 soils were obtained in the profile under secondary forest land (SF profile), 16 soils were obtained in the profile under abandoned cropland for 3 years (AC profile), and 11 soils were obtained in the profile under grazing shrubland for 5 years (GS profile). This study represents multiple stages of abandoned cropland evolution using three land uses (abandoned cropland–shrubland–forest land). Long-term changes can also be employed as an alternative by substituting the spatial variation for reference with temporal changes within a given region (Xiao et al., 2018). The detailed description of sampling methods was reported by Han & Xu (2022), more than three profiles were dug within a distance of 1 m at each site. All results shown in this study are averages of samples from duplicate profiles at the same depth. An interval of 5 cm was applied to the sampled surface soil, and an interval of 10 cm was applied to the deeper soil. Based on the depth from the surface to the bedrock, the thicknesses of the SF and GS profiles were 160 and 70 cm, respectively. According to the soil sampling standards in China, the sampling depth in croplands was defined as 100 cm. Therefore, a sampling depth of 130 cm was selected to ensure data reliability in the AC profile.

Determination of soil composition

All chemical analyses were based on dried soil samples and excluded plant residues and stones. The soil texture was determined using a laser particle size analyzer, and a soil particle diameter less than 2 μm was assigned to clay, ranging from 2 to 53 μm in the silt, and from 53 to 250 μm in the sand (Soil Survey Staff, 2014). Soil pH was measured by a pH meter in the suspension of soil (sample diameter < 2 mm): water = 2 : 5 with a precision of ± 0.05 (Han et al., 2020). The remaining samples were sieved at 75 μm (approximately 200 mesh) for composition measurements. Finer soils were digested with 2 mol/L KCl for 1 d to remove inorganic N and dried at 55 °C after repeated washing to neutral pH with pure water. SON and SOC contents were measured using treated soil samples with an elemental analyzer (Vario TOC cube; Elementar Analysensysteme GmbH, Langenselbold, Germany) with a precision of ±0.02%. The SOC contents and δ¹³C_SOC values have been reported by Han, Zhang & Xu (2023a). The nitrogen isotopes were determined using a stable isotope mass spectrometer (Thermo, MAT-253, USA) with a precision better than 0.2‰. All values of δ¹⁵N_SON (¹⁵N_SON/¹⁴N_SON) were normalized to the atmospheric N₂ standard in ‰ unit:

$${\delta }^{15}{\mathrm{N}}_{\mathrm{S}\mathrm{O}\mathrm{N}}=\left[\left({\delta }^{15}{\mathrm{N}}_{\mathrm{M}}-{\delta }^{15}{\mathrm{N}}_{\mathrm{A}}\right)/{\delta }^{15}{\mathrm{N}}_{\mathrm{A}}\right]\times 1000$$where M means measured samples, and A means atmospheric N₂ standard. All measurements were performed at the Institute of Geographic Sciences and Natural Resources Research, CAS, with duplicate determination to ensure accuracy.

The map of the study area was graphed using ArcMap (version 10.8; Esri ArcGIS Desktop; Esri, Redlands, CA, USA). All data analyses of SON content, δ¹⁵N_SON values, and related soil properties were performed using SPSS (version 25.0; IBM SPSS Statistics, Chicago, IL, USA), and all figures were analyzed using Origin (version 2017; OriginLab, Northampton, MA, USA).

Soil properties

The data of the soil physical parameters are given in Table S1, and the soil particle distributions in the three profiles are plotted in Fig. 2. Detailed data on soil pH variation and soil particle distributions have been reported by Han, Zhang & Xu (2023b). The AC profile had an acidic soil environment, and medium to slightly alkaline soil environments were observed in the SF and GS profiles. The three soil profiles in Yinjiang County were characterized by lower clay and sand contents, according to the United States Department of Agriculture (Soil Survey Staff, 2014), and the soils were silt and silt loam soil types. Stronger soil coarsening in abandoned croplands showed higher desertification threats than in other lands. A larger proportion of sand fraction was observed in the AC profile, and soil particle distributions were more variable in the AC profile.

Geochemical characteristics

Soil organic carbon (SOC) has been suggested to be an essential factor affecting SON dynamics by regulating N release from plants (Han et al., 2020). The variations in SON content, δ¹⁵N_SON values, and the ratios between soil organic carbon (SOC) content and SON content (SOC/SON, all values in this study are mass ratios) under the three land uses in Yinjiang County are listed in Table S2, and the ranges and coefficient of variations (CV) are presented in Table 1. The SON contents ranged from 0.06% to 0.23%, 0.08% to 0.22%, and 0.12% to 0.63% in the SF, AC, and GS profiles, respectively. The values of δ¹⁵N_SON ranged from 4.35‰ to 7.59‰ in the SF profile, 3.79‰ to 7.23‰ in the AC profile, 1.87‰ to 7.08‰ in the GS profile. The varied range of SON contents in the three soil profiles (0.06% to 0.63%) were similar to those in the Puding catchment (0.05% to 0.77%) and Libo County (0.08% to 0.75%) with karst structure, and lower than that in the Jiulongjiang Basin (0.16% to 0.91%), while the δ¹⁵N_SON compositions in the study area (1.11‰ to 9.46‰) were narrower than those in the Puding catchment (1.87‰ to 7.59‰) and Libo County (1.20‰ to 10.0‰), but obviously higher than that in Jiulongjiang Basin (0.8‰ to 5.7‰) (Han et al., 2020; Li, Han & Tang, 2014; Liu & Han, 2022). The ratios of [SOC/SON] ranged from 5.21 to 18.13 in the SF profile, 4.81 to 8.15 in the AC profile, and 6.11 to 17.18 in the GS profile. The CV values showed vertical variations in SON contents, and its isotopic fractionations were limited in the AC profile and animated in the GS profile.

The vertical distributions showed that the SON contents, δ¹⁵N_SON values, and ratios of [SOC/SON] fluctuated greatly within the 0 to 30 cm soils of the three profiles (Fig. 3). SON content was in the sequence of GS profile (mean: 0.31%) > SF profile (mean: 0.09%) ≈ AC profile (mean: 0.10%) and tended to a similar value in the soils below 30 cm. In contrast, the δ¹⁵N_SON values were at the highest level in the AC profile and the lowest level in the GS profile in the 0 to 30 cm soil layer at the same depth. The values of δ¹⁵N_SON increased in the 0 to 30 cm soils of the three profiles. Notably, the δ¹⁵N_SON values consistently increased in soils below 30 cm in the GS profile, whereas those values hardly changed in the SF and AC profiles. The variations in SOC/SON ratios decreased as the soil depth increased in the SL (range 9.65) and GS profiles (range 8.82), especially in the soil above 30 cm. The values of SOC/SON remained fairly constant in the AC profile (range 3.35) and were almost at the lowest level among the three profiles.

Spatial patterns of SON contents under diverse disturbed mode

Land evolution and utilization possibly more affect the variation in SON contents, rather than the climate and geological background because of their analog within the small catchment. The spatial patterns of SON in the three profiles were presented in Fig. 3. Covered vegetation can influence the contents of SON and the bioaccumulation of soil microbes by regulating litter inputs (Ambrosino et al., 2021; Yan et al., 2020). Furthermore, the SON contents are strongly controlled by microbiotas activities by N mineralization, ammonium, nitrification immobilization, and nitrate immobilization (Buzin et al., 2019; Li et al., 2018). The microbial biomass is generally at the highest level in forests and at the lowest level in agricultural land (Li et al., 2018). However, a lower SON content was observed in the AC profile. It may be closely linked to the application of manure and nitrogenous fertilizers, which are used to promote crop growth in agricultural activities. Synthetic N fertilizer is generally considered the largest contributor to the N level of cultivated soil, which is also accompanied by high loss rates of N (such as NO₃^- leaching and NO₂ emission) (Choi et al., 2017). However, SON will continue to decompose and be assimilated by plants to become lost in abandoned croplands with the cessation of the N fertilizer supply (Liu et al., 2010). In addition to the utilization of inorganic N by plants, soil organism is also regarded as forceful factor in regulating SON dynamics. The distribution of soil microorganisms plays a forceful role in shaping the SON pool size by facilitating the conversion of inorganic to organic N and serves as a crucial repository (Buzin et al., 2019). The biomass accumulation of microbiotas tends to be higher in shrubs than in forests or areas without plants, particularly in the surface soil (Yan et al., 2020). Microbiotas may cause the highest level of SON contents in the surface soil of the GS profile.

The SOC/SON ratios in the three profiles are shown in Fig. 3. The variation in the SOC/SON ratios in the cultivated land (<8.5) was significantly lower than that in the other lands, while the SON content in the cultivated land was similar to that in the forest land. It possibly has resulted from lower levels of SOC contents (0.51% to 1.77%) in the AC profile (Han, Zhang & Xu, 2023a). Following the abandonment of cultivation, the cessation of fertilizer supplies disrupted soil components to lose SOC and could not revert to the original SOC levels within a short period. The SON contents in the GS profile were much higher than that in the SF profile, whereas the SOC/SON ratios showed similar values, indicating the existence of additional SOC and SON in the GS profile. Neither forested nor abandoned land had additional organic matter input in recent years, while goat-grazed shrubland land may sustainably obtain soil organic matter (SOM) from goat excreta, which may cause the obviously higher SON content in the surface soils of the GS profile. The SOC/SON ratios presented obvious decrease trends in the 0 to 30 cm soil layers in both the SL and GS profiles. Generally, SOM is enriched in surface soil and decreases with increasing depth, which is regarded as a forceful factor in regulating the dynamics of SON (Liu, Han & Zhang, 2020). Higher ratios of SOC/SON were less suitable for SOM decomposition than lower ratios of SOC/SON because of restraining by microorganisms (Lan, Hu & Fu, 2020). The higher SOC/SON in the SF and GS profiles may have resulted from additional SOC inputs by plant litter and excreta. The accumulation of SOM in deeper soils mainly combines with Ca²⁺ to form stable complexes, especially in calcareous soils (Li et al., 2018). The lower SOC/SON ratios with limited fluctuation in the deeper soil of the three profiles may indicate a stable form of SOM, which was difficult to decompose.

Environmental factors on SON contents and isotopic composition

Soil texture can also regulate SON and δ¹⁵N_SON isotopic fractionation in soil. Soil texture can directly influence SON retention and storage, which is easily altered by land uses conversion (Lan, 2021; Xia, Rufty & Shi, 2020). Additionally, increasing proportions of silt and clay can also control the δ¹⁵N_SON composition by increasing the abundance of microbes (such as fungi and filamentous bacteria) (Hobbie & Ouimette, 2009; Xia, Rufty & Shi, 2020). In this study, the δ¹⁵N_SON values were stable in the deep soils of all profiles, which may inherit from the characteristics of regional bedrock and have been less disturbed of biological activities. However, the soil components are easily transferred during pedogenesis. The relationships among SON content, δ¹⁵N_SON value, Ca content, soil pH, and soil particle proportion are illustrated in Fig. 4. Both soil pH and Ca content showed no correlation with SON content and δ¹⁵N_SON value in the SF profile, suggesting that the controlling factor of SON dynamics could be attributed to others. Soil pH was negatively correlated with SON content in the GS profile but positively correlated with δ¹⁵N_SON values in the AC and GS profiles. Ca content showed positive correlations with SON content and a negative correlation with δ¹⁵N_SON values in both the AC and GS profiles. The complexions between organic matter and minerals are strongly affected by soil Ca, and polyvalent cations (the primary cations are Ca and Mg in neutral to alkaline environments) can promote SOM stability (Li et al., 2017). The Ca content in the study area is at a higher level in Chinese soil because of the karst structure (China Environmental Monitoring Station (CEMS), 1990). The complexation of the OM-Ca²⁺-minerals is considered a vital mechanism to reserve SOM in karst areas in southwest China due to the enrichment of Ca (Li et al., 2017; Liu & Han, 2021b). Therefore, Ca may play a vital role in SON distribution by increasing the stability of the remaining SON in the study area. The fraction of SON fixed by minerals was possibly enriched with lighter N isotopes in the AC and GS profiles.

A positive correlation was found between the SON content and sand proportion in the SF profile, and negative correlations were found between the clay proportion and SON content in the AC and GS profiles. The N requirement for abundant biomass and rapid turnover of microbiotas in the shrubland may be responsible for accelerating SON decomposition (Stoner et al., 2021; Xia, Rufty & Shi, 2020). In the sand fraction, the enzyme-mediated biochemical mineralization of SON is stronger and the SON content decreases, whereas clay supplies more capacity for SON by the stabilization of organic-mineral compounds (Xia, Rufty & Shi, 2020). The results showed that all soil profiles were silt loamy in texture, and the sand proportions were at the highest level in the AC profile (4.00% to 9.38%, sand proportions in the other profiles ranged from 0.52% to 2.66%), which may improve the loss of SON. Generally, soil properties can indirectly affect δ¹⁵N_SON fractionation by controlling the stability and diversity of microbial biomass and products (Khalil, Hossain & Schmidhalter, 2005). Only the sand fraction showed a correlation with the SON content and δ¹⁵N_SON value, whereas the proportion of sand was less than 1.5% in the SF profile. It can be assumed that the influence of the soil texture on δ¹⁵N_SON is limited. The δ¹⁵N_SON values were negatively correlated with the clay and silt fractions in the AC and GS profiles, which is consistent with the silt loamy soil. Therefore, ¹⁵N-enriched N may prefer to remain in clay and silt particles.

SOC and δ¹³C_SOC were employed to further discuss the factors controlling the transformation of SOM (Fig. 5). Except for the AC profile, the SOC/SON ratios showed a positive linear correlation with the SON contents in the other profiles. It can be suggested that SOM decomposition strongly influenced the distribution of SON in the SF and GS profiles. The lower SOC and SON contents in the AC profile may have resulted from elimination by leaching and crop uptake. Negative correlations between SON contents and δ¹⁵N_SON values were also found in the three profiles (Fig. 5B), which suggests that the transformation of SON affected δ¹⁵N_SON fractionation. Generally, the decomposition rate of SOM increases when the mass ratio of SOC/SON is less than 25, and at a higher level when the ratio is less than 15 (Brust, 2019). The SOC/SON mass ratios in almost all soils were less than 10, indicating a high rate of SOM decomposition. The ¹⁵N-enriched N prefers to accumulate in microorganisms and nitrifying bacteria, and returns to the soil in the form of residual SOM by dead microbiota (Craine et al., 2015). In contrast, ¹⁵N-depleted N is produced in the soil by SON mineralization and nitrification, and is subsequently lost or absorbed by plants (Craine et al., 2015; Li et al., 2018). Therefore, the SOC/SON values were negative for δ¹⁵N_SON in most soils of the three profiles (Fig. 5C). The correlation between SOC/SON and δ¹⁵N_SON values was slightly positive in the SF profile below 20 cm, with a narrow variation in δ¹⁵N_SON values (1.14‰), which possibly indicated the limited influence of SOM decomposition.

Soil δ¹⁵N_SON fractionation affected by agricultural activities

As an active area for plant–soil–microbiota interaction, the surface soil layer is a container, in which a large amount of material transformation and energy exchange occurs (Daly et al., 2021). Fluctuations in δ¹⁵N_SON values were also concentrated in the soils over 30 cm in the three profiles (Fig. 3). In addition to direct inheritance from direct SON contributors (e.g., fertilizer and plant litter), the δ¹⁵N_SON composition also changes during the transformation of various N compounds and biological processes (Han et al., 2020; Loss et al., 2017). Generally, plants prefer to assimilate ¹⁵N-depleted N after SON nitrification, whereas heavier δ¹⁵N_SON is enriched in microbiota (Dijkstra et al., 2006; Hobbie & Högberg, 2012). Therefore, the ¹⁴N will be removed, and the ¹⁵N will be accumulated in deeper soil. Due to the decomposition of SOM, the δ¹⁵N_SON and δ¹³C_SOC values generally increase downward in natural land without the transition of C₃ to C₄ covered vegetation and exogenous disturbance (Han et al., 2020). The vertical δ¹⁵N_SON composition in the three profiles was possibly controlled by the plants. Analyses of δ¹⁵N_SON and δ¹³C_SOC were employed to further analyze nutrient cycling in plant–soil systems (Fig. 5D). C₃ plants widely covered by vegetation in the study area may have a great influence on the isotopic composition of soils. The δ¹⁵N_SON values of leguminous plants (mean δ¹⁵N_SON: 1‰) and non-leguminous plants (mean δ¹⁵N_SON: 9‰) have a large difference (Deniro, 1987; Hoefs, 2009). Obviously, the distribution of δ¹⁵N_SON values trended toward the end-member of non-leguminous plants in the GS profile while the δ¹⁵N_SON composition of soils in the SF and AC profiles may be strongly influenced by leguminous plants. Plant uptake and utilization may be directly affected by changes in the soil environment caused by agricultural disturbances. Prolonged fertilizer input during tillage easily causes a decrease in δ¹⁵N_SON values in the soil due to the relatively lower value of δ¹⁵N_SON (Bateman & Kelly, 2007). The long-term cultivation of the AC profile location was accompanied by abundant inputs of synthetic N fertilizer; thus, the δ¹⁵N_SON values in the AC profile were relatively lower. In contrast, the decomposition of organisms may induce sustainable accumulation of ¹⁵N-enriched materials, which can be improved during the tillage process (the largest reach within 5–7‰) (Hoefs, 2009). Therefore, the δ¹⁵N_SON values in the deeper soil were higher than those in the shallow layers of cropland.

However, the δ¹⁵N_SON values were lower in the surface soils of the GS profile than in those of the others. Létolle (2012) reported that δ¹⁵N values in soils mostly varied from 5.1‰ to 12.3‰. However, the values of δ¹⁵N_SON in the topsoil were lower than 5.1‰, particularly in the GS profile (δ¹⁵N_SON < 5.1‰ in soils at 0–25 cm depth). Previous studies have demonstrated that a higher grazing intensity would lead to less N residence time in the soil, and further decrease the δ¹⁵N_SON value (Golluscio et al., 2009). Additional SOM inputs may promote ammonification, and further lead to an increased ratio of SOC/SON and soil pH (Khalil, Hossain & Schmidhalter, 2005). Similar to the unreturnable crops in the cropland, returnable plant biomass on the grazing shrubland was reduced. The N ingested by animals is absorbed and transformed, and lighter N is generally excreted preferentially, while heavier N is enriched within the tissues (Wittmer et al., 2011). Generally, the N-isotopic composition will be heavier in animals than in ingested materials, especially in elderly individuals (Wittmer et al., 2011). Goat grazing may be an important factor causing ¹⁵N depletion in the SON of the GS profile. Moreover, the rate of SON mineralization in grazed land is lower than that in non-grazed land (Golluscio et al., 2009). Therefore, δ¹⁵N_SON values were lower in the GS profile than in the other profiles.

Erosion risk of soil nutrient loss in karst area

Long-term application of organic fertilizer regulates soil SON and SOC levels, while adversely affecting the stability of SON (Xu et al., 2023). Unreasonable agricultural structures possibly cause soil acidification, which also accelerates the loss of soil nutrients by inhibiting the transformation of SON to available N in plants (Brust, 2019). Soil acidification also promotes NO₃^– leaching and the loss of exchangeable cations (e.g., Ca and Mg) (Čakmak et al., 2014). Moreover, it also increases the solubility of potentially toxic heavy metals in the soil and ultimately threatens human health (Čakmak et al., 2014). Different land uses can alter SON dynamics by changing the soil biogeochemistry under artificial disturbance (Fig. 6). The pH values ranged from 4.8 to 5.2 in the AC profile, showing an obvious acidification trend. The Fe, Mn, and Pb contents in the abandoned cropland were also higher than those in shrubland, while both were lower than those in the forest land in Yinjiang County (Han & Xu, 2022; Han, Zhang & Xu, 2023b). It confirms that cultivation may cause damage to soil quality and cannot be recovered within a short time after the cessation of agricultural activities. As shown above, sustainable cultivation results in relatively large soil disturbances, and self-recovery after stopping tillage makes it difficult to restore soil quality in the short term. In contrast, the SON content of shrublands possessed a higher level to sustain soil N supplementation.

The karst area in Guizhou Province is characterized by abundant precipitation and predominantly mountainous areas that are sensitive to environmental changes. A high soil moisture content may promote the conversion of SON to inorganic N, resulting in the loss of SON (Li et al., 2022). Additionally, a large slope further intensifies the loss of soil nutrients by soil erosion. Furthermore, the additional N may also stimulate the emission of N₂O, and be further unfavorable to the eco-environment (Harris et al., 2022). However, slope farming is widely distributed in the Guizhou Province. Agriculture has become the primary economic industry in many parts of the Guizhou Province. Crop–livestock production systems have been demonstrated to improve the utilization of organic fertilizer (i.e., animal excreta) rather than chemical fertilizer and soil stability, further forming regional nutrient recycling (Jalilpour, Chavoshi & Jalalian, 2022; Xu et al., 2023). Combining crop–livestock production with agriculture may reduce the degradation of fragile lands compared with the simple type. It is important to focus on changes in soil quality and soil nutrient loss, and to optimize agroecosystem functions in the karst region of southwest China.