Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Climatic and edaphic controls over soil δ15N in temperate grassland of northern China: A PLS-PATH analysis

  • Xianzhao Liu ,

    Roles Data curation, Funding acquisition, Investigation, Methodology, Writing – original draft

    1100095@hnust.edu.cn

    Affiliation School of Earth Science and space information Engineering, Hunan University of Science and Technology, Xiangtan, Hunan, China

  • Zhengying Luo,

    Roles Data curation, Software

    Affiliation School of Earth Science and space information Engineering, Hunan University of Science and Technology, Xiangtan, Hunan, China

  • Tianhao Wang ,

    Contributed equally to this work with: Tianhao Wang, Qing Su

    Roles Data curation, Investigation

    Affiliation School of Earth Science and space information Engineering, Hunan University of Science and Technology, Xiangtan, Hunan, China

  • Qing Su

    Contributed equally to this work with: Tianhao Wang, Qing Su

    Roles Formal analysis, Investigation, Writing – original draft

    Affiliation School of Life and Health Science, Hunan University of Science and Technology, Xiangtan, Hunan, China

Abstract

Identifying the impact path of climate and soil factors on soil δ15N is very crucial for better understanding the N turnover in soils and the integrated information about ecosystem N cycling. Many studies have showed that climate and soil variables influence the change of soil δ15N. However, most of the existing studies focused on the overall impact of factor on soil δ15N, without distinguishing between the direct and indirect effect. Although scholars have studied the relationships among temperature, precipitation, soil N, soil pH, and soil δ15N rather than estimating all the causal relationships simultaneously. To answer the above-mentioned questions, a regional-scale soil collection was conducted across a temperate grassland in northern China. Meanwhile, a PLS-PATH analysis was utilized to evaluate the direct and indirect effects of various factors on soil δ15N and to explore the causal relationships among variables. The results showed that along the transect, mean annual precipitation (MAP) and mean annual temperature (MAT) directly and significantly reduced soil δ15N, and indirectly affected soil δ15N through their effects on soil pH, soil clay, soil N and soil C/N. Soil C/N ratio has a significant direct impact on soil δ15N with a negative correlation. Soil clay, soil N content, and soil pH have a total positive effect on soil δ15N, but the total positive impact of soil pH is very weak because it has a negative indirect impact on soil δ15N by affecting soil clay, soil N and soil C/N ratio. The total influence is, in order, MAP > MAT > soil C/N > soil clay > soil N > soil pH (in absolute value). The above results will provide valuable information about ecosystem N cycle in temperate grassland of northern China.

Introduction

With the rapid development of isotope measurement technology, the analysis of soil nitrogen isotope (expressed as soil δ15N) has become an effective tool to explore the nitrogen (N) dynamics and soil development. Soil δ15N can potentially provide valuable information about N cycle in soil ecosystem, because it is mainly based on overall isotopic fractionation during microbial degradation and transformation or the preferential decomposition of the substrates depleted in 15N [18]. In general, compared with the substrate with lower decomposition, the older soil with higher microbial treatment is enriched in 15N [9,10]. Since 14N is preferentially lost from the ecosystem, resulting in the enrichment of soil 15N in the ecosystem with high N cycle openness, soil δ15N can be used as a valid indicator of ecosystem N cycle openness [11]. Generally, the N cycle in arid areas is more open than that in humid areas because soil δ15N value is often negatively correlated with precipitation on different spatial scales. Additionally, without considering the inputs of anthropogenic N and the influence of plant residues, the variation in soil δ15N in natural ecosystems is largely controlled by climate and soil factors during microbial decomposition and nitrogen turnover [1214]. Therefore, the signature of 15N in soil integrates a variety of information about the ecosystem N cycling [15,16]. Unfortunately, the influential mechanism that drive the relationships between climatic and edaphic variables and soil δ15N values in terrestrial ecosystem has not yet been fully understood, which affects the interpretation of global or regional patterns of N-cycling.

To understand the driving factors controlling soil δ15N along natural environment gradients, numerous scholars have conducted extensive research at regional and global scales [13,1722]. Most results have revealed that climatic and edaphic factors, including mean annual temperature (MAT), mean annual precipitation (MAP), soil pH, soil C and N content and soil texture, can affect soil δ15N values by affecting N transformation and release [13,23,24]. For instance, on a global scale, climatic variables can control soil δ15N, with values increasing in response to decreasing MAP and increasing MAT, which will enhance the process leading to N loss of but discriminate against 15N loss [12,21,25]. Recent studies indicated that climate factors control about 50% of the variation of soil δ15N across temperate grassland in northern China. It was found that soil δ15N decreased with the increase of MAP and MAT [26]. Further studies exhibited that drought can nonlinearly shape soil δ15N in arid and semi-arid grassland [11,22]. Therefore, on a certain spatial scale, soil δ15N values across climate gradient can mirror the relationship between N losses and turnover [23]. Besides climatic factors, soil factors have been demonstrated to affect the pattern of soil δ15N [12]. For example, soil δ15N values are negatively related to soil organic carbon and positively associated with soil N contents [10,13,27,28]. Soil pH may affect soil N availability via its influences on microbial nitrification, denitrification and anaerobic ammonium oxidation in terrestrial ecosystems, thus controlling the δ15N values in soils [9,24,25,29]. Similarly, soil texture can also impact soil N turnover by its indirect impact on other soil physicochemical properties (e.g., soil water, oxygen concentration and soil N content), thus driving changes in isotope values Similarly, soil texture can also indirectly affect other soil physical and chemical properties (such as soil moisture, oxygen concentration and soil nitrogen content) to affect soil nitrogen turnover, so as to promote the change of isotopic value [30]. On the whole, these studies regarding the pattern of δ15N values have greatly increased our understanding of ecosystem N-cycling.

Nonetheless, we still know little about how co-varying climatic and edaphic variables independently affect soil δ15N values along environmental gradients due to the following two reasons. First, because nitrogen cycle and the factors affecting soil δ15N values are very complex, the climate and soil variables determining the patterns of soil δ15N probably involve direct and indirect influence paths on a regional scale. For instance, low δ15N values usually occur in ecosystems with high precipitation and high carbon contents, but with low temperature and low N concentrations [15,25]. In other words, all the influences among the climate and soil factors will form a network, and a path analysis can disclose how these factors interact to affect soil δ15N values. However, most of the existing studies focused on the overall impact of single factor on soil nitrogen isotope without distinguishing the direct and indirect effects or failed to estimate the causal relationships among the variables. Second, most of the existing studies primarily used bivariate linear regression or multiple regressions to determine how climatic and edaphic variables influence soil δ15N values at different spatial scales. Such methods cannot evaluate the interplay of environmental factors on soil δ15N, and also fail to determine the extent to which variables explain the δ15N values owing to the covariation of environmental factors [21,31]. This may also be one of the reasons for the relatively low reliability of explaining ecosystem nitrogen cycle through δ15N signal in the existing studies.

Temperate grasslands in northern China account for approximately 12% of global grassland area and provide crucial ecosystem services. In particular, grassland soils have great potential to regulate biogeochemical cycle under the background of global climate change [32]. Therefore, it is vital to investigate the impact mechanisms of environmental factors on soil δ15N along temperate grassland for elucidating soil N-cycling on regional scale [20,33]. In this study, we use a path analysis based on partial least squares regression (PLS-PATH) to assess the direct and indirect impact of environment factors on soil δ15N values in the temperate grasslands in northern China and causal relationships among the variables. The research strives to answer the following three questions. First, what are the direct and latent indirect influences of the climate and soil variables on variance of soil δ15N value? Second, how do the causal path relationships between driving factors? Third, what is the order of all variables’ influences on soil δ15N with respect to their influential intensity?

Materials and methods

Study transect and field sampling

The study was conducted along a 1200 km transect across the temperate grasslands of northern China. The east-west transect covered longitudes from 112°53′ to 121°98′E and latitudes from 42°12′ to 43°98′N, with elevations between 158 and 1406m. The transect was characterized predominantly by a temperate continental monsoon climate, with a MAT gradient from 1.71 to 7.10°CC and a MAP gradient from 154 to 446 mm, respectively.

Along the transect from east to west, the main vegetation types are temperate meadow grassland, typical temperate grassland and temperate desert [34]. The soils related to these three grassland types were chernozem, chestnut soil and aeolian sand soil, respectively, which were the same substrate age [35]. The soil pH values of the 0~20cm layer varied between 6.2 and 8.5 [26]. With such climatic and edaphic gradients, this transect provides an ideal place for examining the direct and indirect effects of environmental factors on soil δ15N and the interrelationships between driving factors.

Our studies were approved by the Forestry and Grassland Bureau of Inner Mongolia Autonomous Region and by the Institute of Grassland Research of CAAS. During the period from July to August in 2017, a total of 40 undisturbed sites were chosen to collect soil samples along this transect. At every site, five 1 m ×1 m plots within an area of 10 m × 10 m were established for soil sampling. Within each plot, five soil cores with a diameter of 2.5 cm were randomly taken at 0 to 20cm depth. A total of 25 soil cores were obtained from each site and fully mixed as a composite sample. After removing fine roots and other coarse materials in soils, each composite sample was sieved by a 2.0 mm screen and then divided into two parts: one was stored in an incubator (at 4°C) for determining the physicochemical properties of the soil (e.g., soil C and N contents, soil clay and pH), and the other was air-dried and ground to uniformity with a ball mill (NM200, Retsch, Hann, Germany) for measurement of soil δ15N value. Additionally, the altitude, longitude and latitude of sampling sites were determined by Global Positioning System. The MAT and MAP of each sampling site were obtained from Data Center for Resource and Environment Science in China (https://www.resdc.cn/Default.aspx), China Meteorological Data Service Centre (http://data.cma.cn), and from local weather stations. Further detailed information about the sampling sites was summarized in Table 1.

thumbnail
Table 1. The general information of the sampled sites in the study area.

MAT and MAP in Table 1 are the abbreviations of mean annual temperature and mean annual precipitation, respectively (the same below). These climate data were the averages of observation data collected during a 30 year period (1987–2017). TMS, TTS and TD represent temperate meadow steppe, typical temperate steppe and temperate desert, respectively. The dominant soil types were from the “1:1,000,000 Soil Map of China in 2007”.

https://doi.org/10.1371/journal.pone.0265795.t001

Measurements of soil δ15N and soil properties

The δ15N value of soil sample was measured by a Finnigan DeltaPlusXP (Thermo scientific, Waltham, Massachusetts, USA) coupled with an automatic elemental analyzer (Flash EA1112, Thermo Finnigan). The isotope result of soil N was calculated with the following equation: δ15N (‰) = (Rsample/Rstandard -1)×1000, where Rsample and Rstandard represent the ratios of 15N/14N in the sample standard. The standard of δ15N is atmospheric N2. The standard deviation of above isotopic repeated measurements was ±0.15‰.

Soil pH value was determined using a pH meter (HI-9125, Hanna Instruments Inc, Woonsocket, RI) with a dry soil-water ratio of 1:2.5. Total organic carbon (TOC) and nitrogen (TN) concentrations of soil samples were measured by a TOC/TN analyzer (Multi N/C 3100CLD, Jena, Germany). Soil C/N ratio was expressed by the quotient of TOC and TN content. Another subsample was separated into clay (<2μm) by ultrasonic energy method. The result of particle size analysis was expressed as a percentage of the weight of oven-dried soil. The values of all the above observed variables, which includes the MAP and MAT of sampling locations, soil N, soil pH, soil C: N, soil clay and soil δ15N, were listed in S1 Appendix.

PLS-PATH analyses

Before PLS-PATH analyses, all data were tested for normality and standardized to eliminate the influence caused by different variable units. The potential relationship between various variables (MAP, MAT, soil N, soil C/N, pH and soil clay) and soil δ15N was analyzed by bivariate correlations and linear regressions. The normality test results of the variables showed that the absolute values of critical ratios were greater than 0.05, indicating that the kurtosis (or skewness) coefficients of all variables were significant (Table 2). This also meant that it was suitable for using path analysis in our study. Moreover, the correlation analysis results showed that there was obvious collinearity among some independent variables (Table 3), which means that the indirect impact is also the driver of the soil δ15N in ecosystem. To eliminate collinearity issues among the variables and address how climatic and edaphic variables interactively affect soil δ15N, a PLS-PATH analysis was used to explore the direct and indirect influence of each environmental variable on soil δ15N values and prospective causal relationships among variables at the regional scale. PLS-PATH analysis is considered as a special form of PLS-SEM [35]. Because of the lax requirements of data distribution and small sample size, it has become one of the most popular approaches for estimating complex path relationships. In the PLS-PATH analyses, the direct influence of variables on soil δ15N as well as indirect and total impacts of variables on soil δ15N can be expressed by direct, indirect and total path coefficients, respectively. At the same time, the causal relationship between variables can be shown by vectors in the path diagram. On the whole, PLS-PATH analysis is more comprehensive and accurate compared with simple regression analysis. The basic idea of PLS-PATH analysis is as follows: first of all, assume that the explained variable y has several explanatory variables xi (i = 1, 2, 3,…, n), and the relationship between each explanatory variable xi and y is linear. Second, PLS regression is used to establish the optimal regression equation between explanatory variables and explained variable. Thereupon, the path coefficient of the explanatory variable xi to the explained variable y in the PLS regression equation can be decomposed into direct path coefficient (denoted as ) and indirect path coefficient (namely, the action coefficient of explanatory variable xi to explained variable y through explanatory variable xj, denoted as ). The calculation formulas of direct action coefficient, indirect action coefficient and total action coefficient of independent variable on dependent variable are as follows: (1) (2) (3)

thumbnail
Table 2. Statistical description and normality test of all the variables in the study transect.

The critical ratio is the skewness (or kurtosis) of each variable divided by its standard error.

https://doi.org/10.1371/journal.pone.0265795.t002

thumbnail
Table 3. Pearson’s correlation coefficients for climatic and edaphic variables in explaining soil δ15N.

https://doi.org/10.1371/journal.pone.0265795.t003

Where represent the partial correlation coefficient of xi and y; denotes the simple correlation coefficient of xi and xj; and Sy are the standard deviations of the explanatory variable xi and the explained variable y, respectively. Basic statistics of the data was conducted with SPSS 18.0, and path coefficients were estimated via the maximum likelihood method using SmartPLS V3.2.8 software [36].

Results

Path coefficients and decomposition results of climatic and edaphic variables on soil δ15N values

A path diagram that displays the variable relationships was established based on the PLS-PATH analysis. Fig 1 showed that, along the transect in temperate grassland of northern China, climatic and edaphic variables explained 64.8% of the total variance in soil δ15N values. MAP and MAT together with soil C/N directly determined soil δ15N values, and the three had a significant negative effect on soil δ15N values. On the contrary, soil clay, soil pH and soil N content had a positive effect. PLS-PATH analyses further indicated that among all independent variables, MAP, MAT and soil pH could significantly and directly alter other climate and/ or soil variables (soil N, soil clay and soil C/N) (Fig 1). As can be seen from the decomposition results of the total action coefficient of each variable (Table 4), MAP, MAT and soil C/N ratio had a total negative effect (direct and indirect) on soil δ15N values, and the standardized total effect of these variables on soil δ15N values was -0.677, -0.229 and -0.209 correspondingly; while soil N, soil clay and soil pH had positive total effects on soil δ15N values, and the standardized total effect of them was 0.114, 0.198 and 0.072, respectively. This showed that soil δ15N displayed a large increase with soil N, soil clay and pH across temperate grassland in northern China. Additionally, their total effects (expressed in absolute value, the same below) were, in proper order, MAP > MAT > soil C/N > soil clay > soil N > soil pH. Meanwhile, the MAP, MAT and soil C/N had a direct negative influence on soil δ15N, and their standardized direct effects were -0.337, -0.219 and -0.209, respectively. The sequence of direct influence of all factors on soil δ15N values was MAP > MAT > soil C/N > soil pH >soil clay > soil N, and the indirect influence of all factors on soil δ15N values was, in order, MAP >soil pH > soil clay > soil N > MAT> soil C/N.

thumbnail
Fig 1. Diagram of path analysis.

The PLS-PATH analysis was used to explore the direct and indirect effects of each variable on soil δ15N value, as well as the causal relationship between variables. The number adjacent to arrow is standardized path coefficients, showing the influence size of the relationship; Continuous and dotted black arrows denote positive and negative relationship, and green lines indicate no significant relationship. *, ** and *** represent significant differences at the levels of 0.05, 0.01 and 0.001, respectively.

https://doi.org/10.1371/journal.pone.0265795.g001

thumbnail
Table 4. The decomposition results of the path analysis.

Through PLS-PATH analysis, the total impact of each variable on soil δ15N is divided into direct and indirect effects. The negative sign indicates that the variable makes the soil δ15N decrease.

https://doi.org/10.1371/journal.pone.0265795.t004

Influence mechanism of climatic and edaphic variables on soil δ15N values

Fig 2 described the direct and indirect effects of various factors on soil δ15N. As could be seen from Fig 2A, soil δ15N was very sensitive to both direct and indirect impacts of MAP. When MAP increased by 1 standard deviation, the soil δ15N directly decreased by 0.337 standard deviation. Meanwhile, the indirect effect of MAP on soil δ15N through MAT, soil clay, soil N, soil pH and soil C: N was -0.340 (Table 4). As shown Fig 2B, the contribution of MAT to soil δ15N mainly came from its direct impact, and the direct impact of MAT on soil δ15N was -0.219. In addition, MAT had a negative indirect effect on soil δ15N through soil N and C:/N ratio; whereas exhibited a positive indirect effect on soil δ15N through soil pH and soil clay. Similarly, following Fig 2C, the contribution of soil clay to soil δ15N mainly was from its direct effect. The total indirect impact of clay content on soil δ15N was fairly small (0.063). Among them, the indirect influence of soil clay on soil δ15N through soil N was 0.013, while the indirect influence through soil C/N ratio was 0.050. The direct contribution of soil pH increasing soil δ15N was 0.208, and the indirect impacts of soil pH through soil C/N, clay content and soil N were -0.054, -0.053 and -0.029 respectively (Fig 2D), amounting to -0.136 (Table 4).

δ

thumbnail
Fig 2. Influence path of climate and soil factors on soil δ15N.

(A) The paths of MAP on soil δ15N. (B) The paths of MAT on soil δ15N. (C) The paths of soil clay on soil δ15N. (D) The paths of soil pH on soil δ15N. (E) The path of soil C: N on soil δ15N, yet no indirect effect of soil C: N ratio on soil δ15N was found. (F) The paths of soil N on soil δ15N.

https://doi.org/10.1371/journal.pone.0265795.g002

The soil C/N ratio had no indirect effect on soil δ15N (Table 4). The increase of soil C/N directly leaded to significant decrease of soil δ15N. When the soil C/N ratio increased by 1 standard deviation, the soil δ15N directly decreased by 0.209 (Fig 2E). As shown in Fig 2F, soil N had a positive direct impact on soil δ15N with influence strength of 0.091. However, the indirect impact of soil N on soil δ15N through soil C/N was quit small (0.023).

Discussions

Many scholars have studied how climate and soil factors influence on soil δ15N. The study used the PLS-PATH method to estimate the direct and indirect impact of various factors on soil δ15N and their causal relationships. Some of our results confirmed previous conclusions. On the whole, climatic factors had stronger effects than those of edaphic factors on soil δ15N values (Table 4). The total impact strength of climate factors on soil δ15N was 0.90 (in absolute value); whereas the total impact intensity of edaphic factors on soil δ15N was 0.574 (in absolute value). This indicated that the soil δ15N value was more susceptible to changes of climatic factors than to those of edaphic factors. The following was an analysis of how various climatic and edaphic factors control soil δ15N.

Climatic controls on soil δ15N

Many scholars generally think that the change of soil δ15N is mainly determined by the isotopic values of input N and output N. Since the δ15N of input N is close to zero and usually lower than that of soil, the N output process may leave more 15N in soil compared with the input [8,11]. In our study, soil δ15N decreased significantly with increasing MAP along the transect (Fig 3A), which is in accordance with the results obtained by other scholars [13,15,37,38]. The strong negative effect of MAP on soil δ15N in temperate grassland of northern China may be caused by the following two reasons. First, the openness of nitrogen cycle is determined by precipitation and decreases with precipitation. Thus, the more open N-cycle in arid environment compared to that in wet environment would likely lead to disproportionate loss of inorganic N through NH3 volatilization (the N loss caused by leaching and denitrification is generally very small due to the low soil moisture content in our study area) during nitrification process, which causes a greater 15N-enrichment in soil organic N and an increase of soil δ15N [39,40]. Second, precipitation is one of the most critical determinants of N cycle in arid and semi-arid areas. The N utilizing efficiency by plants may increase with MAP increasing when soil water content is unsaturated and remains aerobic, which results in more 15N-depleted in the soil pool. Meanwhile, as precipitation increase, the incomplete decomposition of plant residue may bring more enriched 14N organic matter to soil N pool, thus causing soil δ15N values to be lower [18]. Besides that, MAP can indirectly affect soil δ15N through its effects on other factors (Fig 2A). For example, when the MAP increases by 1 standard deviation, the soil N will decrease by 0.598 standard deviation, which leads to a reduction of soil δ15N owing to the positive correlation between soil δ15N and soil N (Fig 3B). For another instance, precipitation usually affects the leaching of alkaline cations in soil, so as to affect the soil pH value. Since soil pH value is negatively correlated with MAP across the transect (Fig 3C), the relatively high soil pH in the dry sites is easy to accelerate the volatilization of NH3, resulting in positive soil δ15N values [15,41]. But, remarkably, the indirect effect strength of MAP on soil δ15N values is different because of different relative importance of driving factors.

thumbnail
Fig 3. Changes of soil δ15N with various factors and the relationships between some factors.

The linear regression method was used to analyze how the soil δ15N changes with various environmental factors (MAP, MAT, soil N, soil C/N, pH, soil clay and longitude) and the relationships between some factors. (A) Change of soil δ15N with the MAP. (B) Change of soil δ15N with the soil N. (C) Change of soil pH with the MAP. (D) Change of soil δ15N with the MAT. (E) Change of the MAP with the longitude. (F) Change of the MAT with the longitude. (G) Change of soil δ15N with the MAT at the western part of the transect. (H) Change of soil δ15N with the soil clay. (I) Change of soil δ15N with the soil pH. (J) Change of soil δ15N with the soil C:N ratio.

https://doi.org/10.1371/journal.pone.0265795.g003

Most studies show that soil δ15N values usually increase with MAT [15,26,27,37]. However, the relation between soil δ15N and MAT in temperate grassland of northern China was clearly inconsistent with those obtained from other regions and globally. We argued that the negative soil δ15N-MAT relationship (Fig 3D) may be owing to these two reasons. First, such a pattern was possibly related to the narrow temperature range (MAT: 1.71 to 7.10°CC) along the study transect. As is known to all, the shift in N cycling from organic or ammonium-dominated status to nitrate-dominated status is usually considered as the potential control of soil δ15N [42]. Accordingly, compared to the wide temperature range obtained from global and other regional scales, the narrow range of MAT across our study area was insufficient to cause a large impact on the N-cycling process and, thus, on affecting soil δ15N values [18]. Second, the negative correlation of soil δ15N with temperature may also be due to simultaneous effect of rain and heat. In our study area, approximately 60% to 80% of annual precipitation occurs in summer season [34]. Although increasing temperature can accelerate the rates of soil organic N mineralization and NH3 volatilization, resulting in accelerated 14N loss and positive soil δ15N values, due to the synchronization of rain and heat at the eastern part of the transect (Fig 3E and 3F), increasing precipitation can contribute more negative soil δ15N values in this region, thus masking the positive correlation between soil δ15N and MAT to a certain extent [18]. This result was supported by the indirect effect of MAP on soil δ15N through MAT (Fig 2A). To examine whether there is a positive relation between soil δ15N and MAT, we extracted 19 sites data at the western part of the transect, and found that soil δ15N of these sites increased significantly with the increase of MAT (Fig 3G). Furthermore, PATH analyses showed that MAT indirectly influenced soil δ15N through edaphic factors (Fig 2B). For instance, the standardized indirect influence of MAT on soil δ15N through soil N is -0.039, demonstrating that temperature also has an indirect negative effect on soil δ15N. This result confirms the view that the soil N retention rate is lower in high-temperature environments than in low-temperature regions [24], and also reflects the apparent temperature-dependence of soil N cycle processes [4].

Edaphic controls on soil δ15N

Besides climate, our PATH analyses also showed that edaphic factors play important roles in controlling soil δ15N values. In the current study, soil δ15N values exhibited a significant increase with soil clay across temperate grasslands in northern China (Fig 3H). The positive correlation between soil δ15N and clay may be directly driven by a larger proportion of discriminating gaseous N (N2O and NO) losses from soil with high clay concentrations. As a result, the large amount of gaseous N loss may bring about higher δ15N values in fine-textured soils because they have a great potential impact on 15N/14N fractionation [17,30]. Despite all this, the increase of soil δ15N with clay content is also inseparable from the indirect effect of soil clay (Fig 2C). That is because soil clay can influence soil N concentration and soil C: N ratio through its fixing more 15N-enriched soil organic matter [43]. Due to the greater decomposition of organic matter in soil with high clay concentration, the soils in warm and /or dry environments may contain soil organic matter with high δ15N because mineral-associated organic matter contains a greater proportion of N.

In our study area, soil δ15N exhibits a significant upward trend with soil pH values along the transect (Fig 3I), suggesting that soil pH has an important impact on ecosystem N cycle. The increase of soil δ15N with soil pH value might be caused by the increase of NH3 volatilization under alkaline conditions [26]. Some experimental evidences show that soil pH can influence internal N cycle by affecting microbial nitrification and denitrification, and also control gaseous N loss via NH3 volatilization because at high pH, the ammonium-ammonia equilibrium is inclined to the gaseous form, which enriches soil N with 15N [28,30]. Moreover, soil pH indirectly influences soil δ15N through influencing soil C: N ratio, soil clay, and soil N (Fig 2D). The total indirect effect strength of soil pH on soil δ15N is -0.136, indicating that soil pH is also a main indirect driver of nitrogen isotope fractionation.

In our regional scale, soil δ15N decreased significantly with the increase of soil C/N ratio along the transect (Fig 3J), and this total effect of soil C/N was all from its direct effect (Fig 2E). There are reports that soil C: N ratio can greatly influence the operation of N cycle process at regional scale [27,44]. Therefore, the reduction of soil δ15N with soil C: N ratio may be due to its impacts on soil gaseous N losses. It was reported that soil N2O emissions exhibited an exponential decline as soil C/N ratio increased [45]. The significant negative correlation of soil δ15N and C: N ratio indicates that a lower C: N ratio will enhance soil gaseous losses in the process of soil mineralization, nitrification and denitrification, and thus cause larger soil 15N enrichment. And it also means that the loss of gaseous N can regulate soil δ15N in temperate grassland of northern China.

Besides, some scholars found that soil δ15N increased or decreased as soil N contents increased [26,27], but we observed a significant positive correlation between soil δ15N and soil N across the transect (Fig 3B), which was consistent with the results previously reported by Shan et al. [44] on the Chinese Loess Plateau. The positive relationship between soil δ15N and soil N might result from the increase in the fractionation of 15N with the increase of soil N reserves. Normally, the 15N abundance of N input is close to zero and lowers than soil 15N. Therefore, N outputs and the fractionation of microbial activities in soils may have great influence on soil δ15N [20,46,47]. In this study, with the reduction of MAP and MAT from east to west along the transect (Fig 3E and 3F), warm and dry sites may have led to increase in microbial activity relative to cold and wet sites [35]. In the case of high soil N content, isotope depleted 14N is preferentially lost from the soil through NH3 volatilization and microbial denitrification, resulting in the enrichment of 15N in soil pool and subsequent increases in soil δ15N. Furthermore, soil N indirectly affected soil δ15N by affecting soil C/N, but the influence intensity is small (Fig 2F).

Conclusions

In short, a regional-scale soil samples were collected along an environmental gradient in temperate grassland of northern China. Four conclusions emanate from PLS-PATH analysis conducted herein, which will provide valuable information for the response of ecosystem N cycle to climatic and edaphic changes.

(1) Soil N, soil pH, and soil clay have positive total effect on soil δ15N, whereas MAP, MAT, and soil C/N ratio have negative total effect on soil δ15N. The total influence is, in order, MAP > MAT > soil C: N ratio > soil clay > soil N > soil pH (in absolute value). Among them, soil pH, soil clay, and soil N have a positive direct effect on soil δ15N. In addition, the indirect impact of all factors on soil δ15N is, in order, MAP > soil pH > soil clay > soil N > MAT > soil C/N. (2) Across the transect, MAP exerted the first-order controls on soil δ15N with negative correlation, and directly or indirectly influenced soil δ15N value through affecting MAT, soil clay, soil pH, soil N and C/N ratio. Soil δ15N values significantly decreased as MAT increased. The negative soil δ15N-MAT relationship, which is in contrast with previous research, may be due to the narrow temperature range and simultaneous effect of rain and heat in the study area. (3) Soil C: N ratio imposed a significant negative direct effect on soil δ15N. This result indicates that in temperate grassland, soil C/N ratio can regulate soil δ15N values by gaseous N loss processes. Soil clay, soil N content, and soil pH have a total positive effect on soil δ15N. However, soil pH has a significant negative indirect influence on soil δ15N through its effects on soil clay, soil N and C: N ratio. Consequently, the total positive impact of soil pH on soil δ15N is weak. (4) These results suggest that climatic and edaphic factors are to some extent coupled in controlling soil 15N abundance. It should be noted that the current research did not consider plant species and microbial activities as the factors affecting soil δ15N.

Supporting information

S1 Appendix. Geospatial, soil and climatic data.

All observed data are available, which includes the longitude and latitude of sampling locations, mean annual precipitation (MAP), mean annual temperature (MAT), altitude, soil N, soil pH, soil C:N, soil clay and soil δ15N.

https://doi.org/10.1371/journal.pone.0265795.s001

(PDF)

Acknowledgments

We are very grateful to the stations along the way for providing convenience and relevant meteorological data. We also thank the Forestry and Grassland Bureau of Inner Mongolia Autonomous Region and the Institute of Grassland Research of CAAS for their help to the project.

References

  1. 1. Ann-Kathrin S, Michael Z, Björn B, Sándor G, Ulrich H, Slobodan BM, et al.The late Quaternary loess record of Tokaj, Hungary: Reconstructing palaeoenvironment, vegetation and climate using stable C and N isotopes and biomarkers. Quaternary International. 2011; 240 (1–2): 52–61. https://doi.org/10.1016/j.quaint.2010.10.009.
  2. 2. Ariz I, Cruz C, Neves T, Irigoyen JJ, Garcia-Olaverri C, Nogués S, et al. Leaf δ15N as a physiological indicator of the responsiveness of N2-fixing alfalfa plants to elevated CO2, temperature and low water availability. Frontiers in Plant Science. 2015; 6: 1–10. pmid:26322051.
  3. 3. Gruber N, Galloway JN. An earth-system perspective of the global nitrogen cycle. Nature. 2008; 451(7176): 293–296. pmid:18202647.
  4. 4. Hobbie EA, Jumpponen A, Trappe J. Foliar and fungal 15N:14N ratios reflect development of mycorrhizae and nitrogen supply during primary succession: testing analytical models. Oecologia. 2009; 146: 258–268. pmid:16096847.
  5. 5. Houlton BZ, Bai E. Imprint of denitrifying bacteria on the global terrestrial biosphere. Proceedings of the National Academy of Sciences of the United States of America. 2009; 106(51): 21713–21716. pmid:19995974.
  6. 6. Houlton BZ, Sigman DM, Schuur EA, Hedin LO. A climate-driven switch in plant nitrogen acquisition within tropical forest communities. Proceedings of the National Academy of Sciences of the United States of America. 2007; 104(21): 8902–8906. pmid:17502607.
  7. 7. LeBauer DS, Treseder KK. Nitrogen limitation of net primary productivity in terrestrial ecosystems is globally distributed. Ecology. 2008; 89(2): 371–379. pmid:18409427.
  8. 8. Zhou L, Song MH, Wang SQ, Fan JW, Liu JY, Zhong HP, et al. Patterns of soil 15N and total N and their relationships with environmental factors on the Qinghai-Tibetan Plateau. Pedosphere. 2014; 24(2): 232–242. https://doi.org/10.1016/S1002-0160(14)60009-6.
  9. 9. Mayor JR, Wright SJ, Schuur EA, Brooks ME, Turner BL. Stable nitrogen isotope patterns of trees and soils altered by long-term nitrogen and phosphorus addition to a lowland tropical rainforest. Biogeochemistry. 2014; 119: 293–306. https://doi.org/10.1007/s10533-014-9966-1.
  10. 10. Menge D, Baisden WT, Richardson SJ, Peltzer DA, Barbour MM. Declining foliar and litter δ15N diverge from soil, epiphyte and input δ15N along a 120000 yr temperate rain-forest chronosequence. NewPhytologist. 2011; 190(4): 941–952. pmid:21272006.
  11. 11. Wang C, Wang X, Liu D, Wu H, Lü X, Fang Y, et al. Aridity threshold in controlling ecosystem nitrogen cycling in arid and semi-arid grasslands. Nature Communications. 2014; 5: 4799. pmid:25185641.
  12. 12. Bai E, Houlton BZ. Coupled isotopic and process-based modeling of gaseous nitrogen losses from tropical rain forests. Global Biogeochemical Cycles. 2009; 23(2): 269–277. https://doi.org/10.1029/2008GB003361.
  13. 13. Craine JM, Elmore AJ, Aidar MPM, Bustamante M, Dawson TE, Hobbie EA, et al. Global patterns of foliar nitrogen isotopes and their relationships with climate, mycorrhizal fungi, foliar nutrient concentrations, and nitrogen availability. New Phytologist. 2009; 183(4): 980–992. pmid:19563444.
  14. 14. Ma JY, Sun W, Liu XN, Chen FH. Variation in the stable carbon and nitrogen isotope composition of plants and soil along a precipitation gradient in northern china. PLoS ONE. 2012; 7(12): e51894. pmid:23272186.
  15. 15. Craine JM, Brookshire ENJ, Cramer MD, Hasselquist NJ, Koba K, Marin-Spiotta E, et al. Ecological interpretations of nitrogen isotope ratios of terrestrial plants and soils. Plant Soil. 2015a. 396(1–2): 1–26. https://doi.org/10.1007/s11104-015-2542-1.
  16. 16. Makarov I. The nitrogen isotope composition in soils and plants: Its use in environmental studies (a review). Eurasian Soil Science. 2009; 42: 1335–1347. https://doi.org/10.1134/s1064229309120035.
  17. 17. Robinson D. δ15N as an integrator of the nitrogen cycle. Trends in Ecology and Evolution. 2001; 16(3): 153–162. https://doi.org./10.11179580.
  18. 18. Liu WG, Wang Z. Nitrogen isotopic composition of plant-soil in the Loess Plateau and its responding to environmental change. Chinese Science Bulletin. 2009; 54(2): 272–279. https://doi.org/10.1007/s11434-008-0442-y.
  19. 19. Díaz FP, Frugone M, Rodrigo AG, Latorre C. Nitrogen cycling in an extreme hyperarid environment inferred from δ15N analyses of plants, soils and herbivore diet. Scientific Reports. 2016; 6: 22226. pmid:26956399.
  20. 20. Wu Y, Wang B, Chen D. Regional-scale patterns of δ13C and δ15N associated with multiple ecosystem functions along an aridity gradient in grassland ecosystems. Plant and Soil. 2018; 432: 107–118. https://doi.org/10.1007/s11104-018-3781-8.
  21. 21. Zhao X, Xu XL, Wang F, Greenberg I, Liu M, Che RX, et al. Climatic, edaphic and biotic controls over soil δ13C and δ15N in temperate grasslands. Forests. 2020; 11: 433. https://doi.org/10.3390/f11040433.
  22. 22. Luo ZK, Viscarra-Rossel RA, Qian T. Similar importance of edaphic and climatic factors for controlling soil organic carbon stocks of the world. Biogeosciences. 2021; 18: 2063–2073. https://doi.org/10.5194/bg-18-2063-2021.
  23. 23. Aranibar JN, Anderson IC, Epstein HE, Feral C, Swap RJ, Ramontsho J, et al. Nitrogen isotope composition of soils, C3 and C4 plants along land use gradients in southern Africa. Journal of Arid Environments. 2008; 72(4): 326–337. https://doi.org/10.1016/j.jaridenv.2007.06.007.
  24. 24. Nel JA, Craine JM, Cramer MD. Correspondence between δ13C and δ15N in soils suggests coordinated fractionation processes for soil C and N. Plant and Soil. 2018; 423(1/2): 257–271. https://doi.org/10.1007/s.11104-017-3500-x.
  25. 25. Ladd B, Pepper DA, Bonser SP, Laffan SW. Carbon (δ13C) and nitrogen (δ15N) stable isotope composition in plant and soil in Southern Patagonia’s native forests. Global Change Biology. 2012; 18(1): 311–321. https://doi.org./10.1111/j.1365-2486.2011.02494.x.
  26. 26. Cheng WX, Chen QS, Xu YQ, Han XG, Li LH. Climate and ecosystem 15N natural abundance along a transect of Inner Mongolian grasslands: Contrasting regional patterns and global patterns. Global Biogeochemical Cycles. 2009; 23(2): GB2005. https://doi.org/10.1029/2008GB003315.
  27. 27. Craine JM, Elmore AJ, Wang L, Augusto L, Baisden WT, Brookshire E, et al. Convergence of soil nitrogen isotope ratios across global climate gradients. Scientific Reports. 2015b; 5: 8280. pmid:25655192.
  28. 28. Yang YH, Ji CJ, Robinson D, Zhu B, Fang HJ, Shen HH, et al. Vegetation and soil 15N natural abundance in alpine grasslands on the Tibetan Plateau: patterns and implications. Ecosystems. 2013; 16: 1013–1024. https://doi.org/10.1007/s10021-013-9664-1.
  29. 29. Tang ZX, Sun XL, Luo ZK, He NP, Sun OJX. Effects of temperature, soil substrate, and microbial community on carbon mineralization across three climatically contrasting forest sites. Ecology and Evolution. 2018; 8(2): 879–891. pmid:29375762.
  30. 30. Chapin FS III, Matson PA, Vitousek PM. Principles of terrestrial ecosystem ecology. New York: Springer; 2011.
  31. 31. Liu X Z, Zhang Y, Li ZG, Feng T, Su Q, Song Y. Carbon isotopes of C3 herbs correlate with temperature on removing the influence of precipitation across a temperature transect in the agro-pastoral ecotone of northern China. Ecology and Evolution. 2017; 7(24): 10582–10591. pmid:29299240.
  32. 32. Conant RT, Cerri CEP, Osborne BB, Paustian K. Grassland management impacts on soil carbon stocks: A new synthesis. Ecological Applications. 2017; 27(2): 662–668. pmid:27875004.
  33. 33. Reynolds JF, Smith D, Lambin EF, Turner B, Mortimore M, Batterbury SP, et al. Global desertification: building a science for dryland development. Science. 2007; 316(5826): 847–851. pmid:17495163.
  34. 34. Yang LM, Han M, Zhou GS, Li JD. The changes in water-use efficiency and stoma density of Leymus chinensis along Northeast China Transect. Acta Ecological Sinica. 2007; 27(1): 16–23. https://doi.org/10.1016/S1872-2032(07)60006-7.
  35. 35. Hair JF, Howard MC, Nitzl C. Assessing measurement model quality in PLS-SEM using confirmatory composite analysis. Journal of Business Research. 2020; 109: 101–110. https://doi.org/10.1016/j.jbusres.2019.11.069.
  36. 36. Sarstedt M, Cheah JH. Partial least squares structural equation modeling using SmartPLS: A software review. Journal of Marketing Analytics. 2019; 7(3): 196–202. https://doi.org./10.1057/s41270-019-00058-3.
  37. 37. Amundson R, Austin AT, Schuur E, Yoo K, Matzek V, Kendal C, et al. Global patterns of the isotope composition of soil and plant nitrogen. Global Biogeochemical Cycles. 2003; 17(1): 1031–1035. https://doi.org/10.1029/2002GB001903.
  38. 38. Liu XZ, Wang GA. Measurements of nitrogen isotope composition of plants and surface soils along the altitudinal transect of the eastern slope of Mount Gongga in southwest China. Rapid Communications in Mass Spectrometry. 2010; 24(20): 3063–3071. pmid:20872640.
  39. 39. Frank DA, Evans RD, Tracy BF. The role of ammonia volatilization in controlling the natural 15N abundance of a grazed grass-land. Biogeochemistry. 2004; 68: 169–178. https://doi.org/10.1023/B:BIOG.0000025736.19381.91.
  40. 40. Chang GG, McCormick PV, Newman S, Elliott EM. Isotopic indicators of environmental change in a subtropical wetland. Ecological Indicators. 2009; 9: 825–836. https://doi.org/10.1016/j.ecolind.2008.09.015.
  41. 41. Murphy BP, Bowman DMJS. The carbon and nitrogen isotope composition of Australian grasses in relation to climate. Functional Ecology. 2009; 23(6): 1040–1049. https://doi.org/10.1111/j.1365-2435.2009.01576.x.
  42. 42. Sun FF, Kuang YW, Wen DZ, Hou EQ. Long-term tree growth rate, water use efficiency, and tree ring nitrogen isotope composition of Pinusmassoniana L. in response to global climate change and local nitrogen deposition in Southern China. Journal of Soils & Sediments. 2010; 10: 1453–1465. http://doi.org/10.1007/s11368-010-0249-8.
  43. 43. Sollins P, Kramer MG, Swanston C, Lajtha K, Filley T, Aufdenkampe AK, et al. Sequential density fractionation across soils of contrasting mineralogy: evidence for both microbial-and mineral-controlled soil organic matter stabilization. Biogeochemistry. 2009; 96: 209–231. https://doi.org./10.1007/s10533-009-9359-z.
  44. 44. Shan Y, Huang M, Suo L, Zhao X, Wu L. Composition and variation of soil δ15N stable isotope in natural ecosystems. Catena. 2019; 183: 104236. https://doi.org./10.1016/j.catena.2019.104236.
  45. 45. Stevenson BA, Parfitt RL, Schipper LA, Baisden WT, Mudge P. Relationship between soil δ15N, C/N and N losses across land uses in New Zealand. Agriculture Ecosystems & Environment. 2010; 139: 736–741. https://doi.org./10.1016/j.agee.2010.10.020.
  46. 46. Ren XY, Zeng GM, Tang L, Wang JJ, Wan J, Feng HP, et al. Effect of exogenous carbonaceous materials on the bioavailability of organic pollutants and their ecological risks. Soil Biology & Biochemistry. 2018; 116: 70–81. https://doi.org/10.1016/j.soilbio.2017.09.027.
  47. 47. Luo W, James JE, Lü XT, Wang ZW, Bai E, Yan CF, et al. Plant nutrients do not covary with soil nutrients under changing climatic conditions. Global Biogeochemical Cycles. 2015; 29(8): 1298–1308. https://doi.org/10.1002/2015GB005089.