Effects of Heavy Grazing on Interspecific Relationships at Different Spatial Scales in Desert Steppe of China

Abstract

This study investigates the effects of grazing intensity and spatial scale on the important values, interspecific relationships, and community stability of desert steppe plant communities in Siziwang Banner, Inner Mongolia, China. Using vegetation data collected at three spatial scales (50 m × 50 m, 25 m × 25 m, and 2.5 m × 2.5 m) under two grazing conditions (no grazing and heavy grazing), we employed ecological statistics, including variance ratio analysis, ${\chi }^2$ tests, and the Jaccard index, to analyze species interactions and community structure. The results indicated that the important values of species vary with both spatial scale and grazing intensity; for example, Stipa breviflori and Chenopodium aristatum exhibited significantly higher important values in heavily grazed areas. Larger spatial scales enhanced the dominance of Cleistogenes songorica and Chenopodium aristatum, while smaller scales favored Stipa breviflori and Caragana stenophylla. Furthermore, interspecific associations were stronger in heavy grazing conditions. The community demonstrated consistent instability; however, no grazing areas were more stable than heavily grazed ones. These findings highlight that species importance, interspecific relationships, and community stability are closely linked to grazing intensity and spatial scale, emphasizing the critical role of sustainable grazing management in maintaining the long-term stability and resilience of desert steppe ecosystems. By emphasizing the need for targeted and sustainable management strategies, this study aims to contribute to the restoration and preservation of these vital ecosystems.

1. Introduction

The study of interspecific relationships is a cornerstone of community ecology and is essential for exploring species distribution, clustering, and their responses to environmental changes [1,2,3]. Interspecific relationships refer to the interactions between different populations in a plant community, which comprehensively reflect the mutual relationships and influences among populations. These determine the composition, structure, functions, and dynamics of a community [2]. The analysis of interspecific relationships can reveal the competitive exclusion effect between species. In general, positive correlations reflect similarities in resource use between species. In contrast, negative correlations suggest that one or both parties are disadvantaged and may encounter interspecific competition and interference [4], which are crucial for understanding plant community formation and evolution [5].

In recent years, many studies have focused on the effects of interspecies relationships on ecosystems and their responses to environmental changes. For example, Bronstein pointed out in his research that symbiosis between plants plays an important role in the evolutionary process and is significant for plant community construction and ecosystem stability [6]. In addition, the study of Kraft emphasized the impact of interspecies competition on ecosystem structure and function [7]. The importance of resource utilization, space occupation, and other factors in plant community dynamics was revealed too. Therefore, the study of interspecific relationships is meaningful for us to understand the relationships between species interactions and provide a scientific basis for maintaining the biodiversity and stability of ecosystems [8].

When studying the influencing factors of interspecific relationships, it is necessary to consider habitat characteristics, resource utilization, competition or symbiosis, climate change, and other factors [9,10,11]. Most past studies have explored the factors affecting interspecific relationships considering these aspects, but there are few studies on how grazing intensity and spatial scale affect interspecific relationships.

Grassland covers about 40% of the Earth’s land surface and is one of its important ecosystems, mainly used for livestock grazing [12]. Grazing activities are an important control factor for biodiversity and composition in grasslands [13]. The impact of grazing on ecosystems is complex and varied, relying on various factors such as herbivore species, plant species, climate, and vegetation cover types. In appropriate grazing systems, grazing can promote plant diversity and create heterogeneous microenvironments. However, in some cases, especially in areas with short grazing histories or where grazing activities have been newly introduced, it may lead to vegetation depletion, homogenization, and systemic imbalance. Additionally, changes in grazing systems or the abrupt cessation of grazing can both pose threats to ecosystems [14]. Grazing activities have long been recognized as posing significant threats to biodiversity, particularly to certain species or taxonomic groups. Within the framework established by the International Union for Conservation of Nature—Conservation Measures Partnership (IUCN-CMP) for classifying direct threats to biodiversity, grazing is categorized under “2. Agriculture and Aquaculture” [15]. Grazing intensity has a significant effect on the composition and dynamics of plant communities and can change the structure type and diversity of grassland plant communities [16]. In addition, changes in grazing intensity can lead to changes in major plant functional groups, thus affecting forage yield, nutritional value, and plant persistence [17]. The response of plant composition will change according to different grazing intensities [18].

Increasing grazing intensity may change the grassland vegetation structure and affect the relationships between species. High-intensity grazing may reduce vegetation and increase herbaceous litter, thereby altering habitat conditions and affecting the survival and reproduction of other plants and animals [19]. In grassland ecosystems, there is competition among plant populations, and improper grazing activities can intensify this competition, resulting in the increased dominance of some species and the exclusion of others [20]. On the other hand, grazing activities may also affect the symbiotic relationships between plants and soil microbes, which, in turn, affects nutrient cycling and the health of soil ecosystems. Therefore, understanding the relationship between grazing intensity and interspecific associations is significant for the conservation and sustainability of grassland ecosystems [21].

The spatial scale of species is a key concept in ecological research. It is used to describe the range of distribution, spatial structures of biological populations in geographical space, and their interactions with the environment [22]. In ecology, studying spatial scale changes in species is essential for understanding biodiversity distribution, ecosystem function, and stability [23]. Species spatial scale involves not only the distribution range of a single species, but also the spatial interactions between species and the organizational structures and functional characteristics of biomes at different spatial scales [24].

There is a close interaction between spatial scale and interspecific association. Spatial scale refers to the range and scale of species distribution and interaction in an ecosystem, while interspecific correlation describes the spatial interaction and dependence between different species. These two concepts play important roles in ecology and are intertwined on multiple levels. At smaller spatial scales, interspecific associations are more often reflected as direct interactions, such as competition, predation, and symbiosis. These interactions shape the distribution patterns and community structures of species in local environments [22]. For example, competitive relationships between plant populations may determine their abundance and distribution within a local area. In addition, interspecific association also includes symbiosis, such as mutualism between plants and rhizosphere microorganisms, which is of great significance for plant growth and survival. However, at larger spatial scales, interspecific associations are often more indirect and complex, and are influenced by factors such as geographical environment and topographic structure. The composition and structure of biological communities may be constrained by geographical factors such as topography, climate, and soil characteristics, which, in turn, affect the distribution range and diversity of species [23]. For example, geographical barriers like mountains and rivers can restrict species migration and dispersal, leading to the formation and differentiation of local communities.

In ecology, the stability and dynamics of interspecific interactions are also influenced by spatial scale. At smaller scales, local environmental changes have a greater impact on species interactions, leading to increased dynamism in interspecific associations. Conversely, at larger scales, long-term climate change and human activities can exert broader and more persistent effects on species distribution and interactions, thereby influencing the stability of interspecific relationships. Therefore, comprehending the relationship between spatial scale and interspecific associations is crucial for elucidating ecosystem structure and functioning, predicting species responses and adaptability, as well as formulating effective strategies for ecological conservation and management [24].

Compared with other grasslands, desert grassland is the most susceptible to environmental pressures because it is located in a transitional landscape between typical grasslands and deserts, with lower precipitation and marginalized vegetation coverage. Due to harsh environmental factors and overgrazing, desert grassland may undergo desertification. In this process, soil and vegetation degradation occurs, altering the structure and function of ecosystems, as well as the productivity of grasslands [25,26]. Additionally, the loss of vegetation accelerates wind erosion rates, leading to dust storms and other environmental issues [27].

The area of desert grassland in Inner Mongolia is approximately 112,000 square kilometers, accounting for 10.7% of Inner Mongolia’s grassland area [28]. In recent decades, human activities and climate change have contributed to the deterioration of desert grasslands, with the former being considered as the main cause of this trend [29]. According to Li, reporting from 1982 to 1997 [29], the degraded pasture area in Inner Mongolia expanded at an annual rate of 1.9%, increasing by 173,619 square kilometers over fifteen years. Therefore, we selected vegetation in the Inner Mongolian desert grassland as our research object. Through studying the interspecific relationships among species, we can better understand their roles and impacts within the desert grassland ecosystem, which will contribute towards protecting and restoring the ecological balance. In addition, the rational utilization of interactions among different plant species can enhance vegetation coverage, improve soil quality, and slow down the process of desertification.

This study explored changes in plants’ importance value, overall associations, interspecific associations, and stability under different grazing intensities and spatial scales in the Inner Mongolia desert steppe. The purpose was to study the effects of different spatial scales and grazing intensities on interspecific relationships. The changes in species importance and community stability under different grazing intensities and spatial scales were analyzed to investigate the variation in interspecific competition intensity with spatial scale. These studies will provide a scientific basis for effective environmental management and the health and stability of ecosystems.

2. Materials and Methods

2.1. Study Site Description

This experiment was performed in the semi-arid grassland at Siziwang Banner in Inner Mongolia, China (41°47′17″ N, 111°53′46″ E, 1450 m; Figure 1) [30]. Animal husbandry is the most important industry in this area. The grazing intensity had been heavy over a long period before the establishment of our study site (25 ha). The health and stability of the ecosystem are seriously affected by heavy grazing. At present, the grassland has been seriously degraded, and the incomes of herders are limited accordingly. The fluctuating climate makes it worse. The mean annual precipitation in the study area is about 210 mm, falling mainly from May to September. The annual evaporation is about 2300 mm. In most years, annual evaporation is 8-10 times the amount of precipitation.. The average annual temperature in the region is 3.6 °C. The soil of this region is classified as Kastanozem (FAO Working Group), with a sandy loam texture. The studied desert steppe ecosystem is composed of about 27 planted species and can be classified into the following four plant function groups: perennial grasses, shrubs and semi-shrubs, perennial forbs, and annuals/biennial grasses. Annual/biennial forbs and some perennial forbs are highly palatable in the sheep diet [31]. The three dominant species are the perennial grasses Stipa breviflora and Cleistogenes songorica and semi-shrub Artemisia frigida.

Experimental Setup

In June 2023, a grazing experiment was established with two treatments, which covered about 8.4 ha. Two levels of grazing intensity treatments were implemented, no grazing (control check, CK) and heavy grazing (i.e., 12 sheep in one plot, stocking rate was 0.45 sheep ha⁻¹ month⁻¹) (HG). CK1 and HG1 in Figure 1 were selected as test cells. The annual grazing occurred during the vegetation period, from June to the end of November, and from 6 am to 6 pm every day. Adult two-year-old wethers were used for grazing.

In terms of spatial scale sampling, in the peak season of plant growth in mid-to-late August 2023, representative plots with a plot size of 50 m × 50 m were selected in the two experimental areas with levels of grazing, and a vertex of the square plot was specified as the coordinate origin O. The 50 m × 50 m plot was divided into 25 10 m × 10 m grids on average, with a total of (5 + 1) × (5 + 1) = 36 grid vertices. A 0.25 m² quadrate was placed in the center of each vertex, and the plant species, height, coverage, and density of each specie in the quadrate were measured and recorded as the data of the 50 m × 50 m spatial scale. Sample plots of 25 m × 25 m (grid part in Figure 2) were selected according to the O point and divided into 25 grids of 5 m × 5 m. The method of obtaining data was the same as that for 50 m × 50 m. A plot of 2.5 m × 2.5 m (the brown part in Figure 2) was selected according to the O point, the coordinates of each plant were located in the plot, and the height, coverage, and density data of the plant were measured as vegetation data on three spatial scales of 2.5 m × 2.5 m.

In summary, vegetation at three spatial scales of 50 m × 50 m, 25 m × 25 m, and 2.5 m × 2.5 m were measured, respectively, under two stocking rate levels. This can be used to analyze the variation in interspecific relationships with different spatial scales.

2.2. Data Analysis

2.2.1. The Importance Value

The importance values (IVs) can reflect the role and status of a species in a community [32]. The larger the IV of a species, the greater the dominance of the species in the community. The calculation formulas are as follows:

$$IV={\displaystyle \frac{d_i/{\displaystyle \sum _{i=1}^S{d}_i}+h_i/{\displaystyle \sum _{i=1}^S{h}_i}+c_i/{\displaystyle \sum _{i=1}^S{c}_i}}{3}}$$

(1)

where $d_i$ is the number of individuals of the ith species, $h_i$ is the total height of the ith species,  is the total cover of the ith species, and $S$ is the total number of species.

2.2.2. The Overall Association

The overall association of the communities was measured by the variance ratio (VR) test [33]. The W statistic was calculated to test whether the overall association was significant. The calculation formulas are as follows:

$$S_{\mathrm{T}}^2=1/N{\displaystyle \sum _{j=1}^N(}{T}_j-t{)}^2$$

(2)

$${\delta }_{\mathrm{T}}^2={\displaystyle \sum _{i=1}^S{P}_i}(1-P_i)$$

(3)

$$VR=S_{\mathrm{T}}^2/{\delta }_{\mathrm{T}}^2$$

(4)

$$W=VR\times N$$

(5)

where $P_i=n_i/N$, $n_i$ is the number of quadrats in which the ith species appeared; $N$ is the total number of quadrats; $S$ is the total number of species; $T_j$ is the number of individuals appearing in the jth quadrat; and $t$ is the average number of species in the quadrats.

When the sample is independent and the null hypothesis is satisfied, the expectation of VR is 1. When VR = 1, there is no overall association of major species. When VR > 1, the overall association of major species is positive. When VR < 1, the overall association of major species is negative. W is used to test the significance of the VR deviation from 1, When ${\chi }_{0.95}^2(N)<W<{\chi }_{0.05}^2(N)$, the major species are not significantly correlated overall. Otherwise, the major species are significantly correlated overall.

2.2.3. Interspecific Association

Whether the species pairs appear in the quadrat is arranged in a 2 × 2 contingency table, in the shape of $\left[\begin{array}{cc}a& b\\ c& d\end{array}\right]$, where a is the quadrat number in which both species appear; b and c are the quadrat numbers in which only species A or only species B appear, respectively; and d is the quadrat number in which neither species appear. Based on a, b, c, and d, the following methods can calculate the degree of connection between species pairs.

Chi-square (${\chi }^2$) statistics were calculated by the Yates continuous correction method [34], and the formula is as follows:

$${\chi }^2={\displaystyle \frac{N{(\left|ad-bc\right|-0.5N)}^2}{(a+b)(c+d)(a+c\left)\right(b+d)}}$$

(6)

If ${\chi }^2<3.841$, it can be considered that the 2 species have no interspecific associations. If $3.841\hspace{0.33em}<{\chi }^2<6.635\hspace{0.33em}$, there are some associations between 2 species; if ${\chi }^2>6.635$, there is a significant association between 2 species. When $ad>bc$ there is positive association between species pairs, and when $ad<bc$, there is negative association between species.

The connection coefficient (AC) is calculated in the following three cases [35]:

$$\mathrm{When} ad\ge bc, AC={\displaystyle \frac{ad-bc}{(a+b)(b+d)}}$$

(7)

$$\mathrm{When} bc>ad\mathrm{and} d\ge a, AC={\displaystyle \frac{ad-bc}{(a+b)(a+c)}}$$

(8)

$$\mathrm{When} bc>ad\mathrm{and} d<a, AC={\displaystyle \frac{ad-bc}{(b+d)(d+c)}}$$

(9)

AC is in the interval $[-1,1]$, and the closer the value of AC is to 1, the stronger the positive association between species pairs is. The closer the value of AC is to $-1$, the stronger the negative correlation is. If AC = 0, the species pairs are completely independent.

The χ² test can accurately and objectively reflect the significance of interspecies association, but cannot distinguish the strength of the association, while the Jaccard index (JI) can express the strength. Therefore, the Jaccard index (JI) was used in this study to represent the probability and degree of association of species pairs [36], as follows:

$$JI={\displaystyle \frac{a}{a+b+c}}$$

The JI range is [0, 1]. The larger the value is, the stronger the positive correlation is; otherwise, the stronger the negative correlation is.

2.2.4. Community Stability Index

Using the Godron stability index to measure the stability of the community [37], the frequency of all plants in the community was arranged from highest to lowest, and the cumulative reciprocal percentage and cumulative relative frequency were calculated. The model of these two values was established. The coordinates of the intersection in the graph of this model and the equation y = 100 − x were calculated. The closer the coordinates are to the stable point (20, 80), the more stable the community is.

3. Results

3.1. The Important Value of Species

We calculated the important values of species (

Table 1. The importance values of species.

No.	Species	Important Value IV
		2.5 m × 2.5 m		25 m × 25 m		50 m × 50 m
		No Grazing	Heavy Grazing	No Grazing	Heavy Grazing	No Grazing	Heavy Grazing
1	Stipa breviflora	0.42	0.52	0.3	0.41	0.35	0.36
2	Cleistogenes songorica	0.21	0.1	0.11	0.15	0.18	0.18
3	Caragana stenophylla	0.16		0.04	0.00	0.04
4	Asparagus	0.01		0.00		0.00
5	Convolvulus ammannii	0.12	0.00	0.17	0.03	0.16	0.04
6	Lagochilus ilicifolium	0.03		0.01		0.01
7	Allium tenuissimum	0.01		0.03		0.02
8	Krascheninnikovia ceratoides	0.01		0.1		0.06
9	Kochia prista	0.00		0.05		0.02
10	Leymus chinensis			0.03		0.02	0.00
11	Artemisia frigida			0.06		0.05
12	Potentilla bifurca			0.01	0.00	0.01
13	Chenopodidm glaucum			0.02		0.02
14	Chenopodium aristatum	0.01	0.3	0.02	0.33	0.04	0.36
15	Cleistogenes squarrosa					0.00
16	Salsola collina	0.00	0.01	0.04		0.01	0.00
17	Parthenocissus tricuspidata		0.06		0.06		0.05
18	Stipa krylovii			0.01
19	Neopalasia spectate						0.00

). At the three spatial scales and two stocking rates, Stipa breviflora and Cleistogenes songorica were always the dominant species with the highest IV values. Convolvulus ammannii was dominant in no grazing areas, but it was not dominant in heavy grazing areas. The opposite was true for Chenopodium aristatum. Krascheninnikovia ceratoides was dominant only in 25 m × 25 m grazed areas.

3.2. The Overall Association of Desert Grasslands

At the three spatial scales (

Table 2. The overall association of desert grasslands at different spatial scales and different grazing intensities.

Spatial Scale	Habitat Type	Variance Ratio	Statistic W	(Chi0.95, Chi0.05)	Association
2.5 m × 2.5 m	No grazing	1.03	25.67	(14.61, 37.65)	Positive correlation, not significant
	Heavy grazing	1.24	30.91	(14.61, 37.65)	Positive correlation, not significant
25 m × 25 m	No grazing	0.65	23.41	(23.27, 51.00)	Negative correlation, not significant
	Heavy grazing	1.26	45.46	(23.27, 51.00)	Positive correlation, not significant
50 m × 50 m	No grazing	0.69	24.96	(23.27, 51.00)	Negative correlation, not significant
	Heavy grazing	1.67	59.83	(23.27, 51.00)	Positive correlation, significant

), the overall association (VR) in no grazing areas was always greater in heavy grazing areas than in no grazing areas. This implies an increase in positive correlation in heavily grazed areas. VR < 0, and the interspecific relationship was negative in the 25 m × 25 m and 50 m × 50 m in no grazing areas, while VR > 0 and the interspecific relationship was positive in other situations. The plant community in the desert steppe was generally insignificant, as the W statistic was in (Chi0.95, Chi0.05). Only 50 m × 50 m heavy grazing showed a significant positive correlation, and W was not in the interval.

3.3. ${\chi }^2$ Test and Comparison of Jaccard Index

The ${\chi }^2$ test results (

Table 3. ${\chi }^2$ test and comparison of Jaccard index at different spatial scales and different grazing intensities in desert steppe.

.			$$\mathbf{Chi}-\mathbf{Square} \mathbf{Test} {\mathit{\chi }}^2\mathbf{Test}$$				Jaccard Index
Spatial Scales	Habitat Type	Pair Species Number	Positive Association	No Association	Negative Association	Positive and Negative Association Ratio	JI ≥ 0.3	JI < 0.3	JI = 0
2.5 m × 2.5 m	No grazing	55	8 (14.55%)	31 (56.36%)	16 (29.09%)	0.5	9 (16.36%)	46 (83.64%)	17 (30.91%)
	Heavy grazing	15	5 (33.33%)	6 (40%)	4 (26.67%)	1.25	6 (40%)	9 (60%)	1 (6.67%)
25 m × 25 m	No grazing	105	1 (0.95%)	100 (95.24%)	4 (3.81%)	0.25	9 (8.57%)	96 (91.43%)	24 (22.83%)
	Heavy grazing	21	5 (23.81%)	9 (42.86%)	7 (33.33%)	0.71	12 (57.14%)	9 (42.86%)	2 (9.52%)
50 m × 50 m	No grazing	105	4 (3.81%)	91 (86.67%)	10 (9.52%)	0.4	11 (10.48%)	94 (89.52%)	35 (33.33%)
	Heavy grazing	21	6 (28.57%)	8 (38.1%)	7 (33.33%)	0.86	10 (47.62%)	11 (52.38%)	2 (9.52%)

() is the ratio of the number of species in the habitat to the total number of species.

) indicated that, at different spatial scales, the proportions of unrelated species pairs in comparison between no-grazing and heavy-grazing areas were as follows: within 2.5 m × 2.5 m areas, 56.36% and 40%; within 25 m × 25 m areas, 95.24% and 42.86%; and within 50 m × 50 m areas, 42.86% and 38.1%, respectively. Across these scales, the majority of populations showed no correlation. Among them, only the positive and negative correlation ratio of the 2.5 m × 2.5 m heavy grazing area was greater than 1, and the positive and negative correlation ratio of other spatial scales and grazing intensity was less than 1.

According to the results of the Jaccard index (Table 3), under different spatial scales and grazing intensities, the number of JI < 0.3 pairs exceeded 50%, except for 25 m × 25 m heavy grazing areas, among which, the 25 m × 25 m no grazing areas were JI < 0.3, accounting for 91.43% of the total species logarithm.

3.4. ${\chi }^2$ Test and Jaccard Index Semi-Matrix at Different Spatial Scales

3.4.1. ${\chi }^2$ Test and Jaccard Index of 2.5 m × 2.5 m Spatial Scales

The ${\chi }^2$ test could accurately reflect the significance level of the interspecific association and provide a quantitative index for evaluating the significance. Among 55 species pairs in the 2.5 m × 2.5 m no grazing area (Figure 3), 31 pairs showed no correlation, accounting for 56.36% of the total logarithm. There were 16 pairs of negative correlation, accounting for 29.09% of the total logarithm, among which, 14 pairs had an extremely significant negative correlation and 8 pairs had a positive correlation. Among the 15 species pairs in the 2.5 m × 2.5 m heavy pasturing area (Figure 4), 6 pairs showed no correlation, accounting for 40% of the total logarithm. There were four pairs of negative correlation, all of which had an extremely significant negative correlation, and five pairs had a positive correlation.

3.4.2. ${\chi }^2$ Test and Jaccard Index of 25 m × 25 m Spatial Scales

Among 105 species pairs with no grazing at 25 m × 25 m (Figure 5), 100 pairs showed no correlation, accounting for 95.24% of the total logarithm. Four pairs were significantly negatively correlated, and one pair with an extremely significant positive correlation was Stipa breviflora and Cleistogenes songorica. Among 21 species pairs in the 25 m × 25 m heavy pasturing area (Figure 6), nine pairs showed no correlation, accounting for 42.86% of the total logarithm. Seven pairs had a significant negative correlation, accounting for 33.33% of the total logarithm; five pairs had an extremely significant positive correlation.

3.4.3. ${\chi }^2$ Test and Jaccard Index of 50 m × 50 m Spatial Scales

Among 105 species pairs in the 50 m × 50 m no grazing area (Figure 7), 91 pairs showed no correlation, accounting for 86.67% of the total logarithm. Ten pairs were negatively correlated, accounting for 9.52% of the total logarithm. There were four pairs of positive correlation, of which two pairs had a significantly positive correlation and two pairs had an extremely significant positive correlation. Among the 21 species pairs in the 50 m × 50 m heavy grazing area, 8 pairs showed no correlation (Figure 8), accounting for 38.1% of the total logarithm. Eight pairs of negative correlation; six pairs were positively correlated, of which one pair showed a significant positive phase.

3.5. AC Value Semi-Matrix at Different Spatial Scales

3.5.1. AC Value of 2.5 m × 2.5 m Spatial Scales

The semi-matrix diagram of the association coefficient (AC) is shown. In the 2.5 m × 2.5 m no grazing area (Figure 9), 31 pairs had a significant negative correlation and 1 pair had a weak negative correlation. Three pairs were not associated and four pairs had a weak positive correlation. There was a significant positive correlation between 16 pairs. In the 2.5 m × 2.5 m heavy pasturing area (Figure 10), five pairs had a significant negative association, one pair had a weak negative association, one pair had no association, one pair had a weak positive association, and seven pairs had a significant positive association.

3.5.2. AC Value of 25 m × 25 m Spatial Scales

The semi-matrix of association coefficient (AC) shows that 26 pairs of 25 m × 25 m no-grazing areas had a significant negative correlation and 18 pairs had a weak negative correlation (Figure 11). In total, 28 pairs were not associated; 18 pairs had a weak positive correlation; and 14 pairs were significantly positively correlated. In the 25 m × 25 m heavy pastoral area, six pairs had a significant negative correlation (Figure 12) and four pairs had a weak negative correlation. One pair was not associated; four pairs had a weak positive correlation; and six pairs were significantly positively correlated.

3.5.3. AC Value of 50 m × 50 m Spatial Scales

The semi-matrix diagram of the correlation coefficient (AC) shows that, in the 50 m × 50 m no grazing area, 35 pairs had a significant negative correlation (Figure 13) and 16 pairs had a weak negative correlation. In total, 17 pairs were not associated and 21 pairs had a weak positive correlation. There was a significant positive correlation between 16 pairs. In the 50 m × 50 m heavy grazing area, there were six pairs with a significant negative correlation (Figure 14), two pairs with a weak negative correlation, four pairs with a weak positive correlation, and nine pairs with a significant positive correlation.

3.6. Plant Community Stability in Desert Steppe

According to the stable fitting curve of M.Godron (Figure 15 and Figure 16) and its fitting equation (

Table 4. Stability analysis of desert steppe at different spatial scales and grazing intensities (M.Godron method).

Spatial Scale	Habitat Type	Fitting Equation	Determination Coefficient	Intersection Coordinate	Euclidean Distance	Stability
2.5 m × 2.5 m	No grazing	$y=-1.6681+2.4619x-0.0148x^2$	0.9875	(34.44, 65.56)	20.42	unstable
	Heavy grazing	$y=-13.085+2.4413x-0.0131x^2$	0.9868	(38.51, 61.49)	26.17	unstable
25 m × 25 m	No grazing	$y=9.7604+2.0178x-0.0117x^2$	0.9941	(35.16, 64.84)	21.44	unstable
	Heavy grazing	$y=-5.5102+2.3565x-0.013x^2$	0.9985	(36.63, 63.37)	23.52	unstable
50 m × 50 m	No grazing	$y=12.272+2.0178x-0.0117x^2$	0.9858	(33.39, 66.61)	18.94	unstable
	Heavy grazing	$y=-7.2435+2.4472x-0.0138x^2$	0.9986	(36.42, 63.58)	23.22	unstable

), it can be seen that the intersection coordinates of the 2.5 m × 2.5 m no grazing area with the 2.5 m × 2.5 m heavy grazing area and the 25 m × 25 m no grazing area with the 25 m × 25 m heavy grazing area and the 50 m × 50 m heavy grazing area were (34.44, 65.56), (38.51, 61.49), (35.16, 64.84), (36.63, 63.37), and (36.42, 63.58). The five intersections were farthest from the stable point of the community. The intersection coordinates of the 50 m × 50 m no grazing area were (33.39, 66.61), closest to the community stability point (20.00, 80.00), but it was still in an unstable state.

4. Discussion

4.1. Changes in Species Importance Value Under Different Spatial Scales and Grazing Intensity

The role and intensity of each species in the ecosystem are different. The importance value index (IV) helps to evaluate an individual species’ importance in the study area. Species with higher IV values will be more stable in terms of adaptability and growth [38]. The results showed that the IVs of Stipa brevifloris and Chenopodium aristatum under the same spatial scale were significantly higher in heavy grazing areas than in no grazing areas. This indicates that the importance and dominance of these two species increased significantly in heavy grazing areas. Additionally, Parthenocissus tricuspidate was not observed in no grazing areas, but appeared in heavy grazing areas. In contrast, other species such as Caragana stenophylla and Convolvulus ammannii had higher IVs in no grazing areas than in heavy grazing areas. Some species were not observed in heavy grazing areas at all. The main reason for this may have been the selective feeding behavior of livestock and their different responses to grazing disturbances. The cuticle thickness of Stipa breviflorum increased significantly in heavy grazing areas. The thickening of the cuticle is a common defense response of grassland plants against grazing animals. This is for reducing its palatability to achieve grazing tolerance [39]. At the same time, livestock trampling causes fragmentation of the clusters of Stipa brevifloris. That means that, with an increase in grazing intensity, large clusters of Stipa brevifloris separate into several small clusters, which increases the population density and its dominant position in the community. On the contrary, domestic animals have an obvious feeding preference for other species, such as Caragana, which leaves less time for the plants to rest and compensate. It is not conducive to their reproduction and is more conducive to the growth and reproduction of species with a poor palatability, such as Chenopodium aristatum and Parthenocissus tricuspidate. According to the above results, it can be concluded that grazing changes the community composition and the species’ relative dominance/importance. This is consistent with Yan’s conclusion [40].

Under the same grazing intensity, the IVs of Stipa breviflora and Caragana stenophylla decreased with an increasing spatial scale. Conversely, the IVs of Cleistogenes songorica and Chenopodium aristatum increased. This may have been due to the differences in species adaptability, abundance, and distribution, with certain species showing greater advantages at smaller scales and lower importance at larger scales [41]. At smaller scales, some species may exhibit a higher abundance and clustering because they are better adapted to local conditions. At larger scales, these species may become more dispersed, leading to a decrease in their importance [42]. However, the reasons for the effects of spatial scale on species’ importance values still require further observational studies.

4.2. Changes in Interspecific Connectivity of Community Species Under Different Spatial Scales and Grazing Intensities

Interspecific associations can reflect the dynamics of a community and the interactions between species. Positive correlations indicate that species have similar demands for environmental resources, exhibiting a strong complementarity, so the species have more efficient resource utilization [5]. In contrast, a negative correlation indicates that species, due to significant differences in their biological characteristics, have adaptations to environmental heterogeneity, which makes them more likely to occupy space and compete for resources, and, therefore, exhibit exclusion [43]. We used the variance ratio (VR), Chi-square (${ \chi }^2$) test, and Jaccard index (JI) to reveal the changes in the interspecific associations in desert steppe ecosystems at different spatial scales and grazing intensities. It was found that VR < 1 only in the 25 m × 25 m and 50 m × 50 m areas with no grazing, which showed a negative correlation. Conversely, the VR > 1 in the 25 m × 25 m and 50 m × 50 m areas with heavy grazing, which indicated a positive correlation. In addition, at the same spatial scale, with an increase in grazing intensity, the VR also increased, which means species were more closely related to each other. This is consistent with the results of Liu’s study [44]. The results of the ${\chi }^2$ test and JI showed that, at the same spatial scale, the proportion of positively associated species pairs increased with an increase in grazing intensity. At the same time, the proportion of unassociated species decreased. This indicated that grazing interference had some effects on the interspecific associations. The proportion of JI ≥ 0.3 also increased with an increase in grazing intensity, which meant a stronger correlation between species. This may have been because, without a disturbance to grazing, resources are the main limiting factor for the growth and reproduction of species, and species have fierce competition for resources. However, with a disturbance to grazing, the main limiting factor becomes the grazing behavior of animals, which controls the growth and reproduction of species. To maintain their position and function in the community, species form cooperation to fend off grazing disturbances. Therefore, the community showed a positive correlation with heavy grazing as a whole.

Spatial scale can influence interspecific interactions too. At smaller spatial scales, interactions between species may be more intense and significant, and competition between species becomes more intense. However, at larger spatial scales, interactions between species may become more complementary and competition may weaken [45]. In no grazing areas, the results showed that the VR decreased with an increase in the spatial scale. An increase in the spatial scale would lead the community to an unstable state of development, which unbalances the community structure and species composition, thereby increasing competition. These results contradict MacArthur’s conclusion. On the contrary, in heavy grazing areas, the positive correlation of VR was more significant with an increase in the spatial scale, indicating that an increase in spatial scale will also enhance the complementary between species, which is consistent with the conclusion of MacArthur. The results of the ${\chi }^2$ test showed that, in no grazing areas, the proportion of positively associated species pairs at a small scale was higher than that at the medium scale and large scale. But there was no significant difference in heavy grazing areas. The results of JI showed no significant difference in spatial scale, regardless of no grazing or with heavy grazing. Therefore, the main factor affecting interspecific association was grazing, and spatial scale had no significant effect on interspecific association.

4.3. Changes in Community Stability at Different Spatial Scales and Grazing Intensities

The stability of a community is a dimension of the stability of the ecosystem, which can be defined as the ability of the community to resist external disturbances and maintain its original structure and state after disturbances. It includes resistance and resilience [46]. Generally, the stability of a community is related to its self-repair, during which, it may also be affected by factors such as the variation in the competitions of intraspecific and interspecific, influence of the environment, and the interference of humans, etc. [47]. The stability assessment method of M. Godron is a systematic and comprehensive mathematical approach [48]. The method was used in this study and the results showed that the coordinates of the intersection points of the six different habitat types in the experimental area were all far from the community stability point (20.00, 80.00), which indicated that they were all in the stage of unstable succession. Moreover, under the same spatial scale, the no grazing area was more stable than the heavy grazing area. The reason for this is the stability of grassland communities in northern China is very sensitive to precipitation and temperature [47,49]. In particular, the desert steppe in Inner Mongolia is a fragile ecosystem in the interlace zone between desert and steppe, with low vegetation coverage, little species diversity, and extreme sensitivity to climate change and grazing disturbance.

5. Conclusions

Grazing intensity and spatial scale significantly impact interspecific connectivity, community stability, and species importance values in the Inner Mongolian desert steppe. The analysis of interspecific associations revealed that, with an increase in grazing intensity, the interspecific associations became stronger, and the ratio of positive and negative associations widened. Furthermore, the influence of grazing intensity on interspecific relationships was more pronounced than that of spatial scale. Consequently, grassland species selection should be tailored to grazing intensity, and the ratio of positive and negative species correlations should be adjusted reasonably to enhance community stability and biodiversity.

Our study on the effects of grazing intensity and spatial scale on community stability showed that six different habitat types in the experimental area were all distant from the stable point of the community, indicating their unstable state. Notably, the stability of the no grazing area was higher than that of the heavily grazed area. These findings suggest that reducing grazing intensity or implementing rotational grazing practices could help to stabilize community structures and promote ecosystem recovery.

Moreover, the study on species importance values found that, with an increasing grazing intensity, the importance of Stipa breviflora and Chenopodium aristatum increased, with Chenopodium aristatum exhibiting significantly higher importance values in heavily grazed areas compared to ungrazed areas. These shifts in dominant species not only reflect ecological changes, but also have practical implications for grassland management. For instance, managers can consider incorporating these species into restoration projects or using them as indicators of grazing pressure.

From a management perspective, our findings offer data support and theoretical guidance for understanding local ecosystem recovery and sustainable utilization. Specifically, managers and administrators can use our results to design targeted restoration projects, mitigation and compensation measures, and ordinary management actions focused on specific grazing regimes. For example, implementing rotational grazing or reducing the overall grazing intensity in heavily grazed areas can help to stabilize communities, enhance biodiversity, and support sustainable livestock production. Additionally, incorporating native species like Stipa breviflora and Chenopodium aristatum into restoration efforts can accelerate ecosystem recovery and improve resilience to future disturbances.

Overall, we plan to expand our research scope to surrounding areas by establishing zones with different grazing intensities to deeply analyze their impact on grassland ecosystems and assess the generalizability and regional differences of our findings. This expanded research aims to deepen our understanding of grassland dynamics and provide practical recommendations for sustainable grassland management.