Effects of Grazing Intensity on Plant-Soil C: N: P Stoichiometry with Precipitation Changes During the Growing Season in Desert Grassland

Background and aims High-intensity grazing in the Mongolian grassland has led to the general deterioration of biodiversity and ecosystem functioning. Although abundant evidence shows that grazing affects the structure and function of grassland ecosystems, research on the impact of precipitation, especially under drought and overgrazing. Methods We examined the effects of heavy grazing, moderate grazing and no grazing on plant communities; plant and soil C, N and P contents; and plant and soil C:N:P stoichiometry in the desert grassland in different years with different amounts of precipitation. Results There was no signicant difference in the species diversity between the grazing and no grazing treatment, while the no grazing treatment was signicantly higher than the heavy grazing treatment. Compared with the amounts in the no grazing and moderate grazing treatments, the N and P contents of the plants in the heavy grazing treatment were the highest, and the N content of the soil increased. There was a positive correlation between precipitation and the N and P contents of plants and the C and N contents of the soil at 0-10 cm and 10-20 cm. Conclusions Our study suggest that a large amount of precipitation of plant growth will drive changes in the community species diversity. Grazing promoted the ow of N between plants and the soil, especially under heavy grazing. Under grazing stress, plants maintain the potential of compensatory growth, and precipitation in the peak season of plant growth induces rapid growth, suggesting that precipitation is an important factor driving grazing ecosystems.


Introduction
Grassland ecosystems play an important role in regulating the climate and maintaining global biodiversity (Habel et al., 2013;Binder et al., 2018). Grazing is the most common land use in grassland ecosystems (Allred et al., 2011). In recent decades, under the continuous in uence of human factors, large grassland areas on the Mongolian Plateau have been degraded to varying degrees (Wang et al., 2017). However, in desert grasslands where the average annual precipitation is less than 200 mm and long-term overgrazing occurs, plant communities remain stable and maintain a relatively high biomass (Wang et al., 2014a). Therefore, understanding the ecological process utilized by vegetation and soil in desert steppes under grazing stress is very important for the protection and management of plant species and for understanding community feedback and maintenance mechanisms.
As herbivores are responsible for rangeland degradation, it is obviously not tenable to explain plant response processes with livestock feeding. The existing literature suggests that grazing not only directly affects the community structure, primary productivity, biodiversity and ecosystem stability of grasslands ( (Sterner et al., 2002;Elser et al., 2000;Yan et al., 2016). Through the study of the C, N and P contents in plant leaves and C, N and P stoichiometry, we can clearly understand the nutrients in the soil and the nutrient absorption and assimilation abilities in plants because there is usually a reciprocal nutritional relationship between plants and soil.
In addition, studies have shown that when large herbivores are present, precipitation is an important factor affecting community characteristics, especially in arid and rainless desert grasslands (Wang et al., 2014a;Bai et al., 2004).
Water availability is the primary factor affecting the community structure and composition (Bai et al., 2012). Bai et al. (2004) suggested that precipitation from January to July was the main factor driving changes in grassland communities. However, the amount of precipitation in the desert steppe is far less than the amount of evaporation. If the precipitation period is not synchronous with the plant growth period or the time interval between the two periods is too long, precipitation cannot signi cantly affect the community (Westoby, 1998). Westoby (1998) assumed that changes in the community structure required special rainfall events, such as rare heavy rain. We hypothesize that the precipitation during the peak plant growth period will not signi cantly affect the characteristics of the community but will alter the essential nutrients in plants (e.g., C, N and P) and even in the soil.
Previous studies have focused on the changes of grass underground vegetation (such as the composition, structure, diversity and biomass) and soil characteristics (such as the water content and carbon storage) (Zhang et al., 2011;Jing et al., 2014). However, this study analysed the characteristics of plant communities and C: N: P stoichiometry in a desert steppe under different grazing treatments and precipitation distributions. Speci cally, we addressed the following questions: 1) How do different grazing treatments affect the coverage, biomass, litter biomass and species richness of a plant community? 2) How do plant and soil C, N and P contents and stoichiometry respond to grazing treatments? 2 Materials And Methods

Study area
The experimental area is located at the long-term grazing experiment site of the Stipa brevi ora desert steppe of the Inner Mongolia Plateau. The geographical location of the study area is E 112°47′16.9″ and N 42°16′26.2″, and the elevation is 1100-1150 m. The average temperature is 4.3 ℃, the highest temperature is 38.7 ℃, the lowest temperature is -38.8 ℃, the winter is long and cold, the spring is dry and windy and the summer is hot. The frost-free period lasts approximately 130 days, with annual precipitation amounts ranging from 170 to 190 mm on average.
Years with more than 30% of the average precipitation are classi ed as wet years, and years with less than 30% of the average precipitation are classi ed as dry years. In this study, 2016 is a normal precipitation year, 2017 is a dry year and 2018 is a wet year (Table S1).

Experimental design
The grazing experiment was started in 1999. Grazing experiments began in May and ended in October. After the grazing experiments were completed, an enclosure was constructed to prevent interference from large herbivores.
Three grazing treatments were established in the grazing experiment, and each treatment was repeated three times. The area of each plot was 2.60 ha. No grazing, moderate grazing and heavy grazing were accomplished with 0, 5 and 8 sheep, respectively. The stocking rates for the three treatments were 0 sheep ha −1 a −1 , 1.92 sheep ha −1 a −1 and 3.08 sheep ha −1 a −1 , respectively. The weight, sex and health status of the sheep used in the experiment were similar (Liu et al., 2019).

Data collection
Field data were obtained from July 20 to August 10, which represented the period with the highest annual plant growth peaks in 2016, 2017 and 2018. In each treatment, ve 1 × 1 m quadrats were established randomly to observe the plant community coverage and species number, and then the aboveground biomass and litter biomass were clipped at the ground level. All plant materials in each sample were dried in an oven at 65 ℃ for 48 h and weighed. The average total dry weight of each species, the number of species used to estimate the aboveground biomass and the species richness were calculated. Leaf samples of all species were collected in the same quadrat for the C, N and P analysis (Bai et al., 2012).
Soil samples were obtained as composite samples from soil cores with a diameter of ve 5 cm at depths of 0 to 10, 10 to 20, 20 to 30 and 30 to 40 cm. The ve cores from each quadrat were mixed in situ as one composite sample, and rocks and visible plant materials were removed by hand. The soil was air-dried and ground to pass through a 1 mm sieve for the C, N and P analysis.
The total C and total N in the soil and plant samples were analysed with an Elementar Vario MACRO CUBE. The total P was determined by the H 2 SO 4 -HClO 4 fusion method. Based on the content, the stoichiometries of the plant and soil samples, i.e., the ratios of C:N, C:P and N:P, were calculated.
Meteorological data were obtained from a micro weather station (GroWeather® software version 1.2; Davis Instruments Corporation, Hayward, CA, USA) located at the experimental site.

Data analysis
Analysis of variance (ANOVA) was used to examine the effects of grazing on the community coverage, biomass, litter biomass, species richness, plant C, N and P contents, plant C:N:P stoichiometry, soil C, N and P contents and soil C:N:P stoichiometry at different depths in 2016, 2017 and 2018. The grazing treatment, year and all interactions were used as xed factors. The ANOVA was followed by Duncan's multiple range tests to compare grazing effects (Bai et al.,

2012).
To analyse the driving factors affecting the community biomass and species richness, the partial least squares (PLS) method was used to construct regression equations. A regression equation for the community biomass, grazing treatment, precipitation during the peak period of plant growth and the C, N and P contents and C:N:P stoichiometry of plants and soils at different depths was constructed. A regression equation for the species richness, grazing treatment, precipitation during the peak period of plant growth and the C, N and P contents and C:N:P stoichiometry of plants and soils at different depths was also constructed. The variable importance of projection (VIP) was determined according to the regression equations. The greater the VIP value of the independent variable was, the stronger the in uence of the independent variable on the community biomass and species richness was. If the independent VIP was > 1, the independent variable was considered an important index; if the independent VIP was < 0.5, the independent variable was considered a non-important index (Li et al., 2015).
To understand the relationship between precipitation during the peak period of plant growth and the C, N and P contents and C:N:P stoichiometry of plants and soils at different depths, grazing was used as a xed factor for a partial correlation analysis. The above statistical analyses were carried out in SAS 9.1 (SAS Institute Cary NC USA, 2003).

Coverage, biomass and species richness
Precipitation during the peak period of plant growth signi cantly affected the plant community coverage, biomass, litter biomass and species richness (Table 1; P<0.05), and grazing affected the plant community biomass and litter biomass. The community biomass and litter biomass were the highest in the non-grazing treatment, followed by those in the moderate grazing treatment and then the heavy grazing treatment (Fig. 1D-I   For the C:N:P stoichiometry, the C:N ratio was affected by the grazing treatment and the amount of precipitation during the peak period of plant growth. The no grazing treatment had the highest ratio, followed by the moderate grazing treatment and then the heavy grazing treatment. 3.3 C, N and P contents and C:N:P stoichiometry of soil Table 3 Means and analysis of variance on the effects, their probability of grazing treatment in desert community of Inner Mongolia on soil C, N, P content over 3 years. Means having a different lower case letter within a subset of a column are different (P > 0.05).  Table 4 Means and analysis of variance on the effects, their probability of grazing treatment in desert community of Inner Mongolia on soil stoichiometry over 3 years. Means having a different lower case letter within a subset of a column are different (P > 0.05). 3.4 Relationships among biomass, species richness and precipitation during the peak period of plant growth and the C, N and P contents and C:N:P stoichiometry The VIP values showed that grazing and precipitation were important indicators of the community biomass and species richness (VIP > 1). Among them, the plant N content, the C:N ratio and the P content in the soil at 0-20 cm were important factors affecting the species richness, while the soil C:N ratio and the N content in the soil at 0-10 cm were important factors affecting the community biomass ( Fig. 2; VIP > 1). Partial correlation analysis showed that precipitation during the peak period of plant growth was positively correlated with the plant N content and the soil C and N contents at 0-10 cm and 10-20 cm and negatively correlated with the plant C content, the C:N ratio and the soil P content (Fig. 3).

Effects of grazing on community characteristics
Grazing is a continuous and highly complex disturbance that can change the characteristics of grassland plant communities, such as the biomass and species diversity. In this study, community biomass and litter decreased as the grazing intensity increased, which was consistent with the results of Sun et al. (2013). The coverage of plant communities uctuated greatly by year, especially in 2018 (wet year). The coverage under the heavy grazing treatment was 156% higher than that under the no grazing treatment, but the biomass under the heavy grazing treatment was 45% lower than that under the no grazing treatment. This result seems to be contradictory and may be due to the following reasons. 1) Precipitation can explain 55-86% of the variation in a plant community (Bai et al., 2012). During the analysis of the precipitation data, a large amount of continuous precipitation before the sampling period was noted. Continuous precipitation can lead to a surge in plant growth with relatively little interspeci c competition under heavy grazing conditions, which is referred to as the "opportunism" strategy. During the process of observing the community coverage in the eld, the coverage of these "opportunistic" plants was also measured, and the results showed an increased coverage in the heavily grazed communities. The individual moisture contents of these "opportunistic" plants increased; however, the biomass that accumulated in the community under heavy grazing was still lower than that under no grazing. Westoby (1998) assumed that community succession requires special rainfall events, such as rare heavy rains, to drive community structure changes. This theory explains our research results very well because there was no signi cant difference in the community structure between the grazing and the no grazing treatments in 2016 and 2017, but the no grazing treatment had a signi cantly different structure than the heavy grazing treatment in 2018.
Grazing makes the grassland appear to be low competition, increases the plant sites for plant seeds, improves the reproduction and distribution ranges of grazing-tolerant species and maintains species richness and livestock feeding in a delicate temporary steady state (Liu et al., 2017). In addition, the desert steppe is dry and rainless, the community structure is simple and the number of species is small. Thus, there was no signi cant difference in species richness among the grazing treatments. However, due to the continuous large amount of precipitation during the peak plant growth period in 2018, the "opportunistic" plants under the grazing treatments grew in large quantities and destroyed the relative balance between the other grassland species and livestock. The competitive disadvantage of the other grassland species coupled with the high-intensity feeding of livestock resulted in a decline in species richness under the grazing treatments, especially the heavy grazing treatment, in 2018.

Effects of grazing on plant and soil C, N and P pools
Previous studies have suggested that for the studied species, nutrients are less affected by the environment and are relatively stable. Grazing changes the species composition, thereby altering the community stoichiometry because species have different nutrient components (Ritchie et al., 1998;Bardgett and Wardle, 2003). This theory supports our research to a certain extent; for example, in 2016 and 2017, the plant C contents were the highest under the no grazing treatment, followed by the moderate grazing and heavy grazing treatments, while there were no signi cant differences in C contents in 2018. In addition, these results may be due to the characteristics of desert grassland plants and their adaptation to grazing disturbances. The most prominent climatic characteristics in arid areas are the scarcity of precipitation and the large inter-annual and inter-seasonal variations. In drought years, the leaf area and leaf width showed a decreasing trend. In contrast, the leaf area and leaf length showed a preferential growth characteristic in wet years ( In addition, the results showed that plants under the heavy grazing treatment had higher N and P contents in plant leaves than those under the no grazing and moderate grazing treatments. We propose the following potential mechanisms to explain these phenomena. Plants induce livestock to feed on plant tissues, organs or populations with low P and N contents (Sun et al., 2014). Sheep feeding not only stimulates the growth of pasture plants but also promotes the redistribution of N and P to young organs, thus increasing the N and P contents in living plants under grazing treatments, leading to the maximum average N and P contents under heavy grazing conditions. This phenomenon implies that plants manipulate the nutrient cycle by altering the diet of livestock to a certain extent. In the desert steppe, a large amount of precipitation will signi cantly increase the water content of plant leaves. The daily evaporation of the desert steppe is very high, which means that large amounts of precipitation will increase the evaporation capacity of plants per unit time (the transpiration tension is strengthened). Higher transpiration rates will promote the synthesis of more N-rich transporters in roots for transporting nutrients to compensate for the plant leaves. To support the rapid growth of plants, ribosomes must rapidly synthesize protein, which means that plants must allocate more P to rRNA, which improves growth rates and increases the N and P contents in plants.
By analysing the soil N content, it was found that the soil N content under the heavy grazing treatment was higher than that under the no grazing treatment, and this is consistent with Bai et al. (2012). This result indicates that N is in a circulating state in heavy grazing ecosystems. This may be because grazing increases C-rich root exudates that stimulate microbial activity and transformation, ultimately leading to an increased availability of soil nutrients to plants ( Bardgett et al., 1998;Hamilton and Frank, 2001 shown that the C:N ratio in soil can reveal the relationship between the transformation of carbon and nitrogen during soil biological decomposition. When the soil C:N ratio was between 5.6 and 11.3, the soil microbial biomass began to increase, and soil N mineralization increased signi cantly. When the C:N ratio was between 15.3 and 20.6, the soil microbial biomass increased rapidly, organic matter decomposition weakened and mineralized N was released (Gundersen et al., 1998). In this study, compared with the no grazing treatment, the soil C:N ratios under the heavy grazing and moderate grazing treatments were in the range of 5.6-11.3, which further veri ed that grazing accelerated the ow and transformation of N in the soil.
Based on the analysis of the C, N and P contents in plants, we propose that plants under grazing conditions, especially heavy grazing, may adapt to a severe living environment in the following ways. Under long-term grazing stress, plants can induce livestock to eat speci c plant tissues and organs or plant species with low N and P contents. To continue to grow and reproduce, the surviving plants must remain in a relatively stable maintenance state (because the rapid growth of plants requires the synthesis of a large amount of proteins and RNA, which requires them to have high N and P contents). Partial correlation analysis revealed that precipitation was positively correlated with plant N and P contents, suggesting that precipitation was the "trigger" that disrupted the equilibrium. In the soil, accelerated N cycling could meet the N needs for plant growth, and precipitation was positively correlated with the C and N contents in the 0-20 cm soil layer, suggesting that precipitation is an important factor driving the changes in grazing ecosystems. Our results further validate the hypothesis that grazing and precipitation are important indicators of community biomass and species richness.

Conclusions
The results indicated that in the desert steppe, abundant precipitation during the peak period of plant growth could induce changes in the community species composition. Grazing promoted the ow of N between plants and soil, especially under heavy grazing. Under grazing stress, plants maintain the potential of compensatory growth with high N and P contents, and precipitation in the peak season of plant growth induces rapid growth, suggesting that precipitation is an important factor driving grazing ecosystems. Figure 1 Effects of grazing on the community coverage, biomass, litter biomass and species richness (error bars are denoted by SE). Bars with the same letter were not signi cantly different in Duncan's multiple range tests reported from one-way ANOVA, and NS indicates non-signi cant (P>0.05).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. TableS1.docx