Vegetation structure of Caragana tibetica communities in inner Mongolia in relation to environmental factors

Abstract

Caragana tibetica communities represent a key vegetation type across the grassland–desert transition zone in central and western Inner Mongolia. Their community structure and ecosystem stability play vital roles in maintaining regional ecosystem functions. In this study, we investigated Caragana tibetica communities in Inner Mongolia. Community classification was conducted on the basis of plant importance values, and differences in vegetation characteristics and species diversity among the classified groups were analyzed. Redundancy analysis (RDA) was further applied to explore the relationships between community traits and environmental factors processed via principal component analysis. The main findings are as follows: (1) A total of 94 plant species belonging to 62 genera and 31 families were recorded. Poaceae, Fabaceae and Asteraceae were the dominant families. The flora comprised 22 geographical elements, dominated by Gobi-Mongolian and East Palaearctic components, with diverse floristic compositions. (2) Combined with dominant species traits, the two-way indicator species analysis (TWINSPAN) quantitative classification and standard nomenclature principles, the C. tibetica communities were classified into four associations: C. tibetica–Pennisetum flaccidum (PF), C. tibetica–Stipa breviflora + Convolvulus ammannii (SCO), C. tibetica–S. breviflora–Cleistogenes spp (SCL), and C. tibetica–Allium spp.–Stipa krylovii (OS). (3) Significant differences in community characteristics were detected among the four associations (p < 0.05). PF exhibited higher shrub coverage and height, while the highest shrub evenness was associated with SCO. Herb coverage in SCL and OS was significantly lower than that in SCO, and the herb coverage in OS was markedly lower than that in SCO and SCL. (4) Shrub traits were regulated mainly by meteorological factors, whereas herb traits were affected primarily by soil properties. The divergence between PF and OS was driven by hydrothermal conditions: PF occurred in areas with favourable hydrothermal regimes, while OS was distributed under inferior hydrothermal conditions. In contrast, SCO and SCL were predominantly influenced by soil factors; SCO was distributed in habitats with relatively high soil nutrient levels, and SCL was distributed in nutrient-poor soils. This study provides scientific reference and basic data for the conservation of C. tibetica communities, restoration of degraded vegetation, ecological management and sustainable development in desert regions.

1 Introduction

China encompasses a vast area of deserts, and the desert ecosystems widely distributed across the Inner Mongolia Plateau play a significant role in shaping the ecological security of northern China and providing windbreak and sand fixation functions. Through long-term natural selection, desert vegetation has developed unique adaptive characteristics and community structures (Fan et al., 2025). Shrub plants, which are highly tolerant to drought and barrenness, are often the constructive and dominant species in desert plant communities and play an irreplaceable role in maintaining the stability of desert ecosystems (Li et al., 2009). Caragana tibetica, also named spiny Caragana, is a constructive species of specific plant communities. These communities are distributed mainly in the narrow transitional belt between the desert steppe and the steppified desert in Inner Mongolia, China, with scattered distributions extending to the Qinghai‒Tibet Plateau (Liu, Y et al., 2024; Zhao et al., 2025). Caragana tibetica is an important sand-fixing plant and a pioneer species for ecological restoration. The community characteristics of C. tibetica, including coverage, height, species composition and diversity, are affected by multiple factors, such as precipitation, temperature, soil texture, topography and wind erosion. The coupling relationship between the community structure and environmental factors has become a key research direction in desert ecology.

There are obvious research gaps in existing studies on C. tibetica. Most studies have focused on its growth habits, community characteristics (Zhang et al., 2011; Qianqian et al., 2019; He et al., 2024; Liu L. E. et al., 2024; Han M et al., 2024; Liang et al., 2025) and association classification (Inner Mongolia and Ningxia Scientific Expedition Group, 1985), while research on the correlations between community characteristics and environmental factors is inadequate. Some scholars have investigated C. tibetica when exploring the relationships between vegetation patterns and environmental factors in the arid zone on the western slope of Helan Mountain (Zheng et al., 2013). Nevertheless, in-depth studies on community structure, species diversity and associated environmental factors at the association level are still scarce, which restricts a comprehensive understanding of the assembly and maintenance mechanisms of such desert communities.

Although Zhang Kun et al. have classified associations, their research is limited mainly to the taxonomic description of vegetation (Zhang K et al., 2025), and exploration of the environmental driving mechanisms of community structure differentiation at the association scale is lacking. In addition, research on shrub communities in arid areas has made remarkable progress in recent years. Studies have revealed that aridity, soil organic matter and total phosphorus are the dominant factors driving shrub–herb interactions in the desert steppes of Inner Mongolia (Hou et al., 2024). Another study conducted by Xu et al. on the western Loess Plateau indicated that precipitation is the major environmental factor causing differences in diversity and community assembly patterns among various shrub communities (Xu et al., 2024). However, none of these studies took C. tibetica communities as research objects or carried out analyses at the association level. Moreover, research on the spatial pattern of C. tibetica shrubs has proven that habitat heterogeneity profoundly influences the spatial distribution of shrub populations; however, it has failed to connect spatial distribution patterns with the environmental interpretation of community structure and diversity (Liu Y et al., 2024).

Substantial differences in climate, soil properties and community structure exist among different associations. Therefore, the use of mean community values alone cannot reflect the spatial heterogeneity of regional vegetation. The classification of associations can further clarify the driving effects of hydrothermal conditions, soil properties and species adaptability on community structure and prevent habitat differences from being concealed by generalized analysis. This study provides a scientific basis for the conservation and restoration of C. tibetica communities as well as for ecological research on vegetation in arid regions.

In view of the above research gaps, this study proposes the following two hypotheses: (1) There are significant differences in climate, soil properties and community structure among different associations, and (2) the classification of associations can effectively reveal the driving effects of hydrothermal conditions, soil characteristics and species adaptability on community structure and prevent the concealment of habitat differences through generalized analysis.

To verify these hypotheses, this study considered C. tibetica communities in Inner Mongolia as the research object. Quantitative classification methods were applied to classify associations, and the species composition, diversity and community structure characteristics were systematically analyzed. By examining environmental factors, including climate, soil and topography, this study aims to clarify the influences of dominant environmental factors on the communities. These findings can provide a theoretical basis for maintaining the ecological balance of C. tibetica communities and protecting biodiversity and offer scientific support for the conservation, restoration and sustainable management of regional desert ecosystems.

2 Materials and methods

2.1 Overview of the study area

In Inner Mongolia, C. tibetica occurs mainly in the central and western region, including Wulate Zhongqi of Bayannur City; Etuoke Qi, Etuoke Qianqi, and Hangjin Qi of Ordos City; the Dalhan Maomingan Joint Banner (hereinafter referred to as the Damao Banner) of Baotou City; and the Alxa Left Banner of the Alxa League. These areas are located between 105.716°E and 110.362°E and between 38.139°N and 42.219°N. The landform types include the Yinshan Mountains (northern foot), Ordos Plateau, and Alxa Plateau. The study area has a typical continental arid climate with four distinct seasons, including cold winters and hot summers, and notable temperature differences between day and night. The annual average temperatures range from 3.3 °C to 7.2 °C, and the annual average precipitation amounts range from 162.5 to 261.8 mm, with precipitation concentrated from July to September. Numerous sandstorms occur in spring, and the dominant wind direction is from the northwest (Liu, Y et al., 2025; Maulana, et al., 2023).

2.2 Plot establishment

This study was conducted mainly during the vigorous growing season from 2023 to 2025. The sampling sites were arranged along spatial gradients based on the natural distribution of C. tibetica communities in the Inner Mongolia Autonomous Region, covering Wulate Middle Banner of Bayannur City, Hangjin Banner, Otog Banner and Otog Front Banner of Ordos City, Alxa Left Banner of Alxa League, and Darhan Muminggan United Banner of Baotou City, to guarantee the spatial representativeness of the samples.

The plots were selected on the premise that C. tibetica served as the constructive species with a well-preserved community structure of native vegetation. Areas with intense anthropogenic disturbances, such as roads and grazing lands, were excluded. Typical landforms, including flat land and gentle slopes, were considered, and various hydrothermal and soil habitat types across the study area were fully covered to ensure that the sampling points could objectively reflect the status of local natural communities. A total of 53 plots with a size of 300 m × 300 m were randomly established and labelled with the prefix A plus serial numbers. Small quadrats were randomly arranged within each plot for field investigation, and basic information, including longitude, latitude, altitude and administrative location, was recorded. Restricted by the natural distribution of plant populations in the field, the density of sampling points varied across different regions. This study focused only on the characteristics of existing communities and did not analyse their long-term dynamic succession processes.

2.3 Vegetation field survey

Vegetation surveys were conducted by combining vegetation type records along transects and plot investigations in typical habitats. A total of 11 sampling sites were established in Bayannur City, 40 in Ordos City, and 1 site each in Baotou City and Alxa League of Inner Mongolia (Figure 1).

FIGURE 1

In accordance with the field survey standards for desert shrub communities and methods adopted in relevant studies (Ministry of Ecology and Environment, 2021), three 20 m × 20 m shrub quadrats and three 1 m × 1 m herb quadrats were randomly established at each sampling site. In total, 139 shrub quadrats and 139 herb quadrats were surveyed. Repeated sampling reduced the interference of spatial heterogeneity and improved the representativeness of the field data.

Basic information, including longitude, latitude, altitude, landforms and vegetation type, was recorded for each quadrat. Within each 20 m × 20 m quadrat, we documented the species name, height and crown width of all the shrubs. For each 1 m × 1 m quadrat in the herb layer, the species name, height, abundance and visually estimated coverage (individual and community coverage) of all vascular plants were recorded (HJ 1170—2021, 2021).

2.4 Environmental factor measurement

Regional surface meteorological elements, such as temperature and precipitation, were obtained from the National Earth System Science Data Center (http://www.geodata.cn) (average values for 2000–2020). Other climate data were derived from CHELSA (Swiss Federal Institute for Forest, Snow and Landscape Research WSL). ArcGIS 10.6 (Environmental Systems Research Institute, Inc.) was used to extract the climatic characteristics of the communities. Ten climatic variables were ultimately acquired, including the annual mean temperature (bio1), mean diurnal range (bio2), maximum temperature of the warmest month (bio5), minimum temperature of the coldest month (bio6), annual temperature range (bio7), mean temperature of the warmest quarter (bio10), mean temperature of the coldest quarter (bio11), annual precipitation (bio12), precipitation of the wettest month (bio13) and precipitation of the wettest quarter (bio16).

2.5 Sample collection

Soil samples were collected beneath the surveyed shrub patches, with sampling points no more than 15 cm away from individual plants. One sampling point was randomly selected within each shrub quadrat, and soil cores were collected from the 0–40 cm soil layer using a soil auger.

2.6 Index measurement

The collected soil samples were transported to the laboratory, air-dried and sieved through a 2 mm mesh prior to the determination of soil physicochemical properties. Soil particle size was measured using a particle size analyzer, and soil particles were classified into three fractions: sand (2–0.005 mm), silt (0.05–0.005 mm) and clay (<0.005 mm). Soil organic carbon was determined by the potassium dichromate volumetric method. Total phosphorus was measured via acid digestion combined with molybdenum-antimony differential spectrophotometry. Total nitrogen was analyzed using the perchloric acid-sulfuric acid digestion method. Total potassium was determined by acid digestion and flame photometry. Soil pH was measured with the potentiometric method (Sparks et al., 1996).

2.7 Data processing

The importance values of the shrubs and herbs in the Caragana tibetica communities were calculated on the basis of the species coverage, height and individual number acquired from the quadrat surveys Four α diversity indices, namely, the Shannon–Wiener index, Pielou evenness index, Simpson dominance index and Patrick richness index, were subsequently computed (Shannon and Weaver, 1998; Wilsey and Potvin, 2000; Simpson., 1949; Wu and Ding, 2020). These indices, including their ecological implications, calculation formulas, analytical software and corresponding references, are summarized in Appendix Table 1.

2.8 Association grouping classification

All the Chinese and Latin names of the plants in this study were verified and confirmed according to the methods of Flora of China (The Editorial Committee of Flora of China, 1986). Two-way indicator species analysis (TWINSPAN) was applied to classify the community associations (Roleček, et al., 2009).

2.9 Data analysis

Data processing, statistical analysis and chart generation were completed using Excel and Origin 2024. Preliminary data processing and diversity index calculations were both performed in Excel. The classification of floristic geographical elements was conducted with reference to the Classification of Vascular Plants in Inner Mongolia and Their Floristic Ecogeographical Distribution by Zhao (2012).

TWINSPAN cluster analysis was carried out in R 4.4.3 to classify the plots into different association groups. To mitigate the interference of multicollinearity among climatic factors on the stability and explanatory power of the redundancy analysis (RDA) results, the variance inflation factor (VIF) of each climatic factor was calculated using the vif. cca function in the vegan package. The VIF quantifies the degree to which a variable can be linearly explained by other variables in the model: the higher the VIF value is, the stronger the multicollinearity.

In accordance with the general standards in ecology and statistics, a VIF > 10 was set as the threshold for severe multicollinearity. When the VIF of a climatic factor exceeds 10, more than 90% of its variation can be explained by other factors, with limited independent information, leading to unstable regression coefficients and distorted significance test results (Hair et al., 1998). Therefore, climatic factors with a VIF > 10 were eliminated stepwise until the VIF values of all the remaining variables were less than 10. The final retained climatic factors with low multicollinearity were used for subsequent ordination analysis.

A total of 15 environmental variables were ultimately retained for analysis, including the mean diurnal range (bio2), annual temperature range (bio7), mean temperature of the warmest quarter (bio10), and precipitation of the wettest month (bio13), as well as longitude, latitude, altitude, soil organic carbon (SOC) and total nitrogen (TN).

Vegetation grouping (PF, SCO, SCL, OS) served as a four-level fixed factor in this study. One-way ANOVA was separately applied to density, coverage, height and alpha diversity to identify group disparities. Significant overall differences (α = 0.05) were followed by Tukey’s HSD pairwise comparisons. Origin 2024(Origin Lab Corporation, Northampton, MA, United States) software and R ((TWINSPAN, VIF calculation, PCA, DCA, RDA) were conducted in R 4.4.3 (R Core Team, Vienna, Austria)) were used for all statistical computations and visualization.Normality tests and homogeneity of variance tests were conducted in advance to verify the prerequisites for parametric tests before further statistical analysis. Principal component analysis (PCA) was applied to reduce the dimensionality of the 15 environmental factors. Detrended correspondence analysis (DCA) was used to calculate the gradient length of the vegetation community data. The gradient length of the first axis was 3.6188, ranging from 3 to 4, indicating that the response of species to environmental factors was between linear and nonlinear patterns. Considering the research objectives and data characteristics, redundancy analysis (RDA) was selected to explore the coupled relationships between vegetation and environmental factors. This method performs well in quantifying the degree to which environmental factors explain community variation, and its results are easy to interpret and compare. Canonical correspondence analysis (CCA) is another alternative, yet it has stricter requirements for the nonlinear response of data. Although the present data fell within an intermediate gradient range, the overall trend of species distribution was gentle and lacked obvious unimodal characteristics. Accordingly, RDA was more suitable for revealing the linear coupling between communities and environmental factors and facilitating interpretation of the results. Hence, RDA was employed to analyse the relationships among the vegetation characteristics, biodiversity and environmental factors of the C. tibetica communities.

3 Results and analysis

3.1 Composition of Caragana tibetica communities

3.1.1 Species composition of Caragana tibetica communities

A total of 94 plant species belonging to 31 families and 62 genera were recorded in the 53 plots. Among them, Poaceae constituted the largest family (11 genera and 17 species), which included mainly perennial herbaceous plants, such as Cleistogenes songorica, Cleistogenes squarrosa, Stipa tenuissima var. klemenzii, Stipa breviflora, Stipa caucasica subsp. glareosa, Agropyron cristatum, Agropyron desertorum, and Enneapogon desvauxii, as well as annual herbaceous plants, such as Tragus mongolorum, Setaria viridis, Aristida adscensionis, and Eragrostis pilosa. The second-largest family was Leguminosae (5 genera and 14 species), whose members included mainly C. tibetica, Caragana stenophylla, and Caragana microphylla, and the remaining semishrubs included mostly Lespedeza potaninii, Lespedeza davurica, and Oxytropis aciphylla. The third-largest family was Asteraceae (7 genera and 13 species), with shrubs and semishrubs primarily including Artemisia ordosica and Ajania achilleoides, as well as annual and biennial herbaceous plants, such as Artemisia annua, Artemisia scoparia, and Artemisia frigida.

3.1.2 Floristic characteristics of the Caragana tibetica communities

The 94 species were divided into 22 floristic geographical elements, among which Gobi–Mongolian and East Asiatic Holarctic species were the most common (16 species each), mainly including C. songorica, O. aciphylla, A. achilleoides, Echinops gmelinii, Corispermum mongolicum, Leymus chinensis, Gueldenstaedtia stenophylla, and Astragalus scaberrimus. Moreover, there were 14 East Asian species, such as T. mongolorum, S. caucasica subsp. glareosa, and C. stenophylla, and 8 Tethyan species, including Neotrinia splendens, Pennisetum flaccidum, Bassia prostrata, and Peganum nigellastrum. Finally, there were 6 Holarctic species, including E. desvauxii, A. annua, Euphorbia esula, and Lappula myosotis. Numerous species with a single floristic geographical element, such as Loess Plateau species, Kazakhstan–Mongolian species, and Loess Plateau–eastern Mongolian Plateau species, were observed (Figure 2).

FIGURE 2

3.2 Classification of Caragana tibetica association groups

As shown in Figure 3, the communities were divided into four association groups via TWINSPAN. The first group was the Caragana tibetica–Pennisetum flaccidum association group (abbreviated as the PF group), which was dominated by the perennial rhizomatous grass Pennisetum flaccidum in the herb layer. The average height of the community was 16.7 cm. Common species in the shrub layer included Oxytropis aciphylla and Caragana stenophylla, while common species in the herb layer included Saussurea japonica, Cleistogenes songorica, and Allium tenuissimum, covering plots such as A25, A27, and A28. The second group was the Caragana tibetica–Stipa breviflora + Convolvulus ammannii association group (abbreviated as the SCO group), with the tufted grass Stipa breviflora as the dominant species and Convolvulus ammannii as the subdominant species in the herb layer. The average community height of this group was 9.26 cm. Common species in the shrub layer included Caragana stenophylla and Oxytropis aciphylla, while common species in the herb layer included Allium tenuissimum, Cleistogenes squarrosa, Allium mongolicum, Polygala wattersii, Peganum harmala, and Potentilla bifurca, covering plots such as A26, A39, and A45. The third group was the Caragana tibetica–Stipa breviflora–Cleistogenes serotina association group (abbreviated as the SCL group), with the tufted grass Stipa breviflora as the dominant species and Cleistogenes species as the subdominant species in the herb layer, and the average community height was 10.68 cm. In addition, common species in the shrub layer included Convolvulus tragacanthoides and Caragana stenophylla, while those in the herb layer included Artemisia frigida, Cleistogenes songorica, Allium polyrhizum, and Convolvulus ammannii, covering plots such as A14, A19, and A30. The fourth group was the Caragana tibetica–Allium spp.–Stipa tianschanica var. klemenzii association group (abbreviated as the OS Group), which was dominated by Allium spp. in the herb layer and Stipa tianschanica var. klemenzii as the subdominant species, with an average community height of 9.58 cm. In addition, common species in the shrub layer included Artemisia xerophytica, Ajania fruticulosa, Krascheninnikovia ceratoides, Caragana brachypoda, Sarcozygium xanthoxylon Bunge, and Atraphaxis pungens, while the herb layer included Stipa glareosa, Peganum nigellastrum, Agropyron desertorum, and Cleistogenes squarrosa, covering plots such as A1, A2, and A8.

FIGURE 3

3.3 Differences in community characteristics among the different association groups of Caragana tibetica

With respect to shrubs, herb coverage in the PF group was significantly greater than that in the OS group, and herb height in the PF group was significantly greater than that in the SCO group (p < 0.05). Among the herbaceous plants, there were significant differences in coverage and abundance but no significant difference in height. Notably, herb coverage in the SCO and SCL groups was significantly greater than that in the OS group (p < 0.05), and herb abundance in the SCO group was significantly greater than that in the SCL and OS groups (p < 0.05). With respect to the biodiversity of the Caragana tibetica community, only the Pielou evenness index of the shrubs significantly differed among the groups, with a significantly greater shrub Pielou evenness index in the SCL group than in the SCO group (p < 0.05). The biodiversity of the herb layer did not significantly differ among the groups (p > 0.05).

3.4 Relationships between community characteristics and environmental factors influencing Caragana tibetica

The PCA results revealed that the cumulative variance contribution rate of the first five principal components reached 74.43%. Among them, the first principal component mainly represented soil nutrient indicators (soil organic carbon (SOC), total nitrogen (TN), soil silt (SIL), and soil sand (SAN)). Moreover, the second principal component represented primarily meteorological factors (annual temperature range (bio7) and precipitation in the wettest month (bio13)), and the third principal component represented the longitude (EE) and latitude (NN). The fourth and fifth principal components largely represented altitude (ALT), total phosphorus (TP), and the mean temperature of the warmest month (bio10) (Table 1).

Redundancy analysis (RDA) revealed clear spatial differentiation among the four associations of Caragana tibetica communities in the ordination space. The sampling points of the OS group were distributed mainly along the positive directions of the RDA1 and RDA2 axes, and its community characteristics were strongly correlated with the environmental gradients of PC1 and PC2. The SCO group occurred along the negative RDA1 axis and the middle section of the RDA2 axis and was closely associated with PC3, PC4 and PC5. The PF group was concentrated in the negative zones of both the RDA1 and the RDA2 axes. The SCL group was located near the centre of the ordination diagram, which was affected by multiple environmental gradients.

4 Discussion

4.1 Differences in the community characteristics of Caragana tibetica among the different association groups

A total of 94 plant species belonging to 62 genera and 31 families were recorded in this survey. Among them, Poaceae, Fabaceae, and Asteraceae were the dominant families and constituted the most species-rich groups in Inner Mongolia. The top five dominant families within the community were Poaceae, Fabaceae, Asteraceae, Chenopodiaceae, and Liliaceae, which collectively accounted for 64.89% of all the species, which is close to the proportion (67.7%) reported for the desert area north of the Lang Mountains (Yong and Zhao, 1979). Most of the remaining families contained 1 to 3 species, and 19 families included only one species. The high proportion of monotypic families corresponds to the typical floristic characteristics of steppe and desert regions in Central Asia (Barbolini et al., 2020; Zhang et al., 2023; Cao et al., 2026). The floristic geographic elements of Caragana tibetica communities included mainly Gobi–Mongolian elements (17.02%), East Palaearctic elements (17.02%), and East Asian elements (14.89%), followed by Palaeo-Mediterranean and Holarctic elements. Overall, the community exhibited transitional characteristics, reflecting an overlap between Central Asian desert flora, eastern steppe flora, and East Asian flora (Zhang et al., 2023; Cao et al., 2026; Ma et al., 2019). Temperate floristic elements predominated in the study area, which is consistent with the floristic characteristics of the Hexi Corridor (Dongxue et al., 2026).

However, the Gobi-Mongolian, East Palaearctic and East Asian distribution types and their variants accounted for the greatest proportion, followed by the Tethyan and Pan-Arctic elements. This floristic pattern differs from that of the Hexi Corridor but is more closely related to the flora of the Alxa Plateau (Zhang, K.,2025). This finding further demonstrates the transitional nature of local vegetation. From the perspective of desert flora, the study area lies at the ecotone between desert and grassland. It harbours typical Gobi Desert species and steppe‒desert taxa, resulting in relatively diverse floristic elements. These findings further indicate that climate plays a dominant role in shaping regional vegetation. Shrubs and subshrubs constitute the primary life forms in this region. As a xerophytic shrub, C. tibetica acts as a constructive species, while most perennial herbs exist as companion species, collectively forming a xerophytic community dominated by shrubs and subshrubs.

The association group is a key unit in vegetation classification and aggregates plant associations with similar synusia structures or dominant layers (Jennings et al., 2009; Lang et al., 2021). Classification of association groups helps characterize C. tibetica communities at a relatively fine taxonomic scale. Our results revealed significant differences among the four association groups in terms of shrub coverage, shrub height, Pielou evenness, and herb coverage and abundance (Figure 4). Such divergences in community traits are attributed mainly to variations in species composition and environmental conditions. For instance, compared with the OS and SCO groups, the PF group had distinctly greater shrub coverage and height, indicating its greater potential for wind prevention and sand fixation (Cao et al., 2024).

FIGURE 4

Moreover, compared with the SCO and OS groups, the SCL group had a significantly greater Pielou evenness index. This suggests that the shrubs in SCL were distributed more evenly, with weaker dominance of constructive species, less intense interspecific competition, and a more stable community structure (Wang et al., 2024). From the perspective of community assembly theory, the relatively mild habitat stress in SCL weakened the filtering effect of the environment, facilitating the coexistence of multiple species. Species realize the complementary utilization of nutrients and space via niche differentiation, which effectively alleviates interspecific competition (Kusumoto et al., 2021). In contrast, the SCO and OS groups were constrained by stronger environmental stress. Dominant species occupied most of the available resources, leading to lower species evenness. This fully reflects that community differentiation is jointly driven by environmental filtering and interspecific interactions.

In terms of herbaceous plants, herb coverage and abundance were significantly greater in SCO than in OS, reflecting the highest degree of steppe tendency. Herbaceous species contributed greatly to the assembly of the SCO community. Overall, community patterns in arid regions are shaped primarily by environmental filtering and niche differentiation (Chase and Myers, 2011). The SCO group is well adapted to the habitats of steppe‒desert ecotones, where environmental filtering is relatively weak. Herbaceous species achieve complementary coexistence through niche differentiation in space and nutrient use, which facilitates their colonization and community expansion. In contrast, the OS group shows typical desert characteristics under intense environmental filtering. Herbs exhibit poor adaptability here, resulting in sparse and structurally simple herb layers. Additionally, the dense herb cover in the SCO group accelerates soil nutrient cycling via litter input and root turnover and increases nutrient accumulation in the topsoil. This creates positive feedback between vegetation development and soil fertility improvement, further maintaining the dominance of the herb layers (Daniels and Nemenman, 2015). Previous studies have demonstrated that herb cover and diversity serve as key drivers for the restoration of desertified lands (Du and Tu, 2020). Excluding external factors, such as soil moisture and human disturbance, this study further reveals that the high herb abundance and coverage in the SCO group result from the combined effects of community structural differentiation, nutrient cycling regulation and trade-offs in interspecific competition. These features not only reflect the steppe tendency of the community but also enhance community stability and resistance to disturbance by optimizing resource utilization and sustaining soil nutrient cycling. These findings are consistent with the classical theory that herb functional groups maintain ecosystem stability in arid regions (Fernando et al., 2009).

4.2 Relationships between the community characteristics of Caragana tibetica and environmental factors among the different association groups

Precipitation, temperature, and soil nutrients are core factors determining the distribution and growth of plant communities. Fluctuations in environmental factors, such as plant height, coverage, and the richness of associated herbaceous species, directly affect the vegetation characteristics of C. tibetica communities. In this study, the ecological characteristics of the shrubs in the C. tibetica communities were affected mainly by meteorological factors, whereas those of the herbaceous plants were controlled mainly by soil factors (Figure 5).

FIGURE 5

Yang et al. (2008) reported that precipitation and minimum temperature govern the population density of shrubs in the Alxa Desert. As large-scale environmental filters, climatic factors regulate water availability and plant phenology, thereby shaping shrub community patterns. In contrast, studies from the Loess Plateau have indicated that herb growth is constrained primarily by soil conditions, which influence herb development via nutrient cycling and substrate filtering (Ma, 2005; Liu, 2012). Such discrepancies reflect a rule for community assembly in arid regions: community dynamics are dominated by climate-induced water limitation at the regional scale, while soil resource distribution plays a leading role at the local scale. This conclusion is in line with previous findings (Chase and Myers, 2011; Fernando et al., 2009).

Further analysis of the relationships between the different association groups and environmental factors revealed that the difference between the PF and OS groups was explained mainly by PC2 (meteorological factors). Geographically, the PF group occurred largely in the Hanggin Banner and the Otog Front Banner on the Ordos Plateau, whereas the OS group was concentrated in the Urad Middle Banner within the northern piedmont area of the Yin Mountains. The two groups differed considerably in terms of the mean precipitation and annual average temperature. The mean annual precipitation in the OS group was 183.6 mm, and the mean annual temperature was 6.8 °C, whereas the figures in the PF group were 248 mm and 8.2 °C, respectively. Therefore, compared with the PF group, the OS group had poorer hydrothermal conditions (Figure 5). Precipitation and mean annual temperature affect vegetation coverage and plant height directly by regulating water supply and plant metabolism (Mao et al., 2020; Ullah et al., 2025). The PF group occurs in areas with superior hydrothermal conditions, which alleviate water limitation and accelerate soil nutrient cycling. Adequate resources consequently enable these shrubs to achieve significantly higher coverage and height than those of the OS group. This conforms to the general rules of community assembly in arid regions, where hydrothermal regimes modulate plant growth and community structure through resource redistribution, which is consistent with the findings of previous related studies (M J C, A J M, 2011).

Moreover, herb coverage was the lowest in the OS group and was significantly lower than that in the SCL and SCO groups (p < 0.05), although no significant difference was found between the OS and PF groups (p > 0.05). These findings further indicated that hydrothermal availability notably limits the growth and distribution of herbaceous plants (Liu L et al., 2024). Relatively unfavourable hydrothermal conditions may also partly explain the lower shrub evenness in the OS group than in the other groups (Wang et al., 2009). In contrast, the relatively high Pielou evenness of the shrubs in the SCL and PF groups (although the difference between the PF and OS groups was not significant) was associated with more favourable hydrothermal conditions, reflected in the lower PC2 values. Shrubs can alter their physiological and functional traits in response to stressful environments, leading to more uniform species distributions and greater structural stability (Guo Y et al., 2025). Under limited resource availability, intensified competitive exclusion reduces the coverage, abundance, and evenness of both shrubs and herbs in the OS group, resulting in uneven species allocation and a relatively simple community structure.

With respect to the SCL and SCO groups, PC1 (soil nutrients) was the dominant influencing factor. The SCL group occurred mainly in areas with low soil nutrient levels, whereas the SCO group occurred in regions with high soil nutrient levels (Figure 5). Differences in soil nutrient levels primarily governed herb performance. Specifically, herb abundance was significantly greater in the SCO group than in the SCL group. Gao Sheng reported that herb coverage was positively correlated with soil nutrient indices (Gao et al., 2012), which is consistent with our observation that higher soil nutrient levels corresponded to greater herb coverage in the SCO group.

5 Conclusion

References

• 1

  AsigbaaseM.SjogerstenS.LomaxB. H.DawoeE.SchererL. (2019). Tree diversity and its ecological importance value in organic and conventional cocoa agroforests in Ghana. PLOS ONE14 (1), e0210557. 10.1371/journal.pone.0210557

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 2

  BarboliniN.WoutersenA.Dupont-NivetG.SilvestroD.TardifD.CosterP. M. C.et al (2020). Cenozoic evolution of the steppe-desert biome in central Asia. Sci. Adv.6, eabb8227. 10.1126/sciadv.abb8227

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 3

  CaoY.XuH.LiY.SuH. (2024). Vegetation growth and physiological adaptation of pioneer plants on Mobile sand dunes. Sustainability16 (20), 8771. 10.3390/su16208771

  • CrossRef
  • Google Scholar

• 4

  CaoY.MoY.JiaC.MiaoB.ShiK.LiY.et al (2026). Plant endemism and biodiversity conservation in the alashan–ordos dryland of Inner Mongolia. Diversity18, 128. 10.3390/d18020128

  • CrossRef
  • Google Scholar

• 5

  DanielsB. C.NemenmanI. (2015). Automated adaptive inference of phenomenological dynamical models. Nat. Communications68133. 10.1038/ncomms9133

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 6

  De WaalC.AndersonB.EllisA. G.BartomeusI. (2015). Relative density and dispersion pattern of two southern African asteraceae affect fecundity through heterospecific interference and mate availability, not pollinator visitation rate. J. Ecol.103 (2), 513–525. 10.1111/1365-2745.12358

  • CrossRef
  • Google Scholar

• 7

  DongxueL.YifanX. U.GuodongD.MinghanY. (2026). Characteristics of flora in the desert-steppe ecotone at the northeastern margin of ulanbuhe Deset[J/OL]. Sci. Soil Water Conservation. 1–19.

  • Google Scholar

• 8

  DuY. L.TuE. G. (2020). Vladimirov Dmitrii.Effects of the microhabitat formed by introduced Hippophae rhamnoides on plant community and soil in desert land of Northwest Sichuan Province. Chin. Agric. Sci. Bull.36 (31), 71–76. 10.11924/j.issn.1000-6850.casb2020-0146

  • CrossRef
  • Google Scholar

• 9

  FanZ.XieT.ShanL.WangH.MaJ.YueY.et al (2025). Soil-driven coupling of plant community functional traits and diversity in desert–oasis transition zone. Plants14 (13), 1997. 10.3390/plants14131997

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 10

  FernandoT. M.MatthewA. B.MaríaD. P.Belén HinojosaM.MartínezI.García-PalaciosP.et al (2009). Shrub encroachment can reverse desertification in semi-arid mediterranean grasslands. Ecol. Letters12 (9), 930–941. 10.1111/j.1461-0248.2009.01352.x

  • CrossRef
  • Google Scholar

• 11

  GaoS.WangL.XueJ. H.WuY. B.RongY. (2012). The relationship between coverage of herbaceous vegetation communities and soil nutrients in the karst area of Guizhou province. J. Nanjing For. Univ. Sci. Ed.36 (1), 5. 10.3969/j.jssn.1000-2006.2012.01.016

  • CrossRef
  • Google Scholar

• 12

  GuoY.QinH.HeM.HanG. (2025). A comparative evaluation of rehabilitation approaches for ecological recovery in arid limestone mine sites. J. Environ. Manag.2025, 373 123876–123876. 10.1016/J.JENVMAN.2024.123876

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 13

  HairJ. F.AndersonR. E.TathamR. L.BlackW. C. (1998). Multivariate Data Analysis. 5th ed.Upper Saddle River, NJ: Prentice Hall.

  • Google Scholar

• 14

  HanM.LiangY.GaoY.YangW.GuoY. (2024). Relationship between root system-soil C:N:P and soil microbial diversity at different evolutionary stages of Caragana tibetica scrub in arid desert grassland, Northern China. Front. Plant Sci.15, 1423536. 10.3389/fpls.2024.1423536

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 15

  HeM.HanY.GaoY.HanM.DuanL. (2024). Decoding the metabolomic responses of Caragana tibetica to livestock grazing in fragile ecosystems. Front. Plant Sci.15, 1339424. 10.3389/fpls.2024.1339424

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 16

  HJ 1170—2021 (2021). Technical Specification for National Ecological Status Investigation and Assessment – Field Observation of Desert Ecosystem. Beijing: Ministry of Ecology and Environment.

  • Google Scholar

• 17

  HouH.ZhangY.ZhouJ.GuoY.LiuH.LiY.et al (2024). Aridity and soil properties drive the shrub–herb interactions along drought gradient in desert grassland in Inner Mongolia. Agronomy14 (11), 2588. 10.3390/AGRONOMY14112588

  • CrossRef
  • Google Scholar

• 18

  Inner Mongolia and Ningxia Scientific Expedition Group (1985). Chinese Academic of Sciences.Vegetation of Inner Mongolia. Science Press(Beijing, China: In Chinese with English abstract).

  • Google Scholar

• 19

  JenningsM. D.Faber-LangendoenD.LoucksO. L.PeetR. K.RobertsD. (2009). Standards for associations and alliances of the U.S. National Vegetation Classification. Ecol. Monogr.79, 173–199. 10.1890/07-1804.1

  • CrossRef
  • Google Scholar

• 20

  KusumotoB.KubotaY.BaselgaA.Gómez‐RodríguezC.MatthewsT. J.MurphyD. J.et al (2021). Community dissimilarity of angiosperm trees reveals deep‐time diversification across tropical and temperate forests. J. Veg. Sci.32 (2), e13017. 10.1111/JVS.13017

  • CrossRef
  • Google Scholar

• 21

  LangX. D.LiuW. D.LiuJ.SuJ. R. (2021). A discussion on the improvement of Chinese vegetation classification system and nomenclature. Bull. Botanical Res.41 (05), 641–659. 10.7525/j.issn.1673-5102.2021.05.001

  • CrossRef
  • Google Scholar

• 22

  LarjavaaraM. (2014). The world's tallest trees grow in thermally similar climates. New Phytol.202 (2), 344–349. 10.1111/nph.12656

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 23

  LiY.CuiJ.ZhangT.OkuroT.DrakeS. (2009). Effectiveness of sand-fixing measures on desert land restoration in Kerqin Sandy Land, northern China. Ecol. Engineering J. Ecotechnology35 (1), 118–127. 10.1016/j.ecoleng.2008.09.013

  • CrossRef
  • Google Scholar

• 24

  LiangY.HeM.JiaR.TianQ.ChenJ.LiM. (2025). Effects of simulated nibbling intensity on growth and physiological characteristics of Caragana tibetica. Front. Plant Sci.16, 1625476. 10.3389/fpls.2025.1625476

  • CrossRef
  • Google Scholar

• 25

  LiuB. J. (2012). The Characteristics of Community Succession and Interaction with the Soil Environment in Abandoned Croplands. Xi’an, Shaanxi: Xi’an University of Science and Technology.

  • Google Scholar

• 26

  LiuY.DongL.WangJ.LiJ.YiL.LiH.et al (2024). Spatial heterogeneity affects the spatial distribution patterns of Caragana tibetica scrubs. Forests15 (12), 2072. 10.3390/f15122072

  • CrossRef
  • Google Scholar

• 27

  LiuL. E.HongYanLiJiaXinLiXinJuanLiHuNaSunJ.et al (2024). Chloroplast genomes of Caragana tibetica and Caragana turkestanica: structures and comparative analysis. BMC Plant Biol.24 (1), 254. 10.1186/S12870-024-04979-9

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 28

  LiuL.GouX.WangX.YangM.QieL.PangG.et al (2024). Relationship between extreme climate and vegetation in arid and semi-arid mountains in China: a case study of the Qilian Mountains. Agric. For. Meteorology348, 109938. 10.1016/J.AGRFORMET.2024.109938

  • CrossRef
  • Google Scholar

• 29

  LiuY.SunS.YangX.WangX.LiuK.DongH. (2025). Estimating biomass carbon stocks of inner Mongolia grasslands using multi-source data. Remote Sens.17 (1), 29. 10.3390/rs17010029

  • CrossRef
  • Google Scholar

• 30

  ChaseJ. M.MyersJ. A. (2011). Disentangling the importance of ecological niches from stochastic processes across scales. Philosophical Transactions R. Soc. Lond. Ser. B, Biol. Sciences366 (1576), 2351–2363. 10.1098/rstb.2011.0063

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 31

  MaX. H. A. (2005). Study on the Relationship Between Vegetation Restoration Succession and Soil Factors in Abandoned Cropland of the Loess Hilly and Gully Region.

  • Google Scholar

• 32

  MaZ.BussmannR. W.HeH.CuiN.WangQ.XuZ.et al (2019). Traditional utilization and management of wild allium plants in Inner Mongolia. Ethnobot. Res. Appl.18, 1–14. 10.32859/era.18.16.1-14

  • CrossRef
  • Google Scholar

• 33

  MaoL.SwensonN. G.SuiX.ZhangJ.ChenS.LiJ.et al (2020). The geographic and climatic distribution of plant height diversity for 19,000 angiosperms in China. Biodivers. Conserv.29, 487–502. 10.1007/s10531-019-01895-5

  • CrossRef
  • Google Scholar

• 34

  MaulanaR.Helms-LorenzM.KlassenR. (2023). Effective Teaching Around the World: Theoretical, Empirical, Methodological and Practical Insights (Springer). 10.1007/978-3-031-31678-4

  • CrossRef
  • Google Scholar

• 35

  PoorterH.JagodzinskiA. M.Ruiz-PeinadoR.KuyahS.LuoY.OleksynJ.et al (2015). How does biomass distribution change with size and differ among species? An analysis for 1200 plant species from five continents. New Phytol.208 (3), 736–749. 10.1111/nph.13571

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 36

  QianqianG.JiuyanY.FengshiL. I.HuanY.YajieZ. (2019). Adaptative characteristics of leaf epidermis micromorphology of caragana sp. in different climates and environments in inner Mongolia plateau. Chin. J. Appl. Environ. Biol.25 (2), 281–290. 10.19675/j.cnki.1006-687x.2018.06010

  • CrossRef
  • Google Scholar

• 37

  RolečekJ.TichýL.ZelenýD.ChytrýM. (2009). Modified TWINSPAN classification in which the hierarchy respects cluster heterogeneity. J. Veg. Sci.20 (4), 596–602. 10.1111/j.1654-1103.2009.01062.x

  • CrossRef
  • Google Scholar

• 38

  ShannonC. E.WeaverW. (1998). “The mathematical theory of communication,”. IL, USA: University of Illinois Press.

  • Google Scholar

• 39

  SimpsonE. H. (1949). Measurement of diversity. Nature163, 688. 10.1038/163688a0

  • CrossRef
  • Google Scholar

• 40

  SparksD. L.PageA. L.HelmkeP. A.LoeppertR. H.SoltanpourP. N.TabatabaiM. A.et al (1996). [SSSA Book Series] Methods of Soil Analysis (Part 3 Chemical Methods). 10.2136/sssabookser5.3

  • CrossRef
  • Google Scholar

• 41

  StampfliA.ZeiterM. (2004). Plant regeneration directs changes in grassland composition after extreme drought: a 13-year study in southern Switzerland, 92(4), 568–576. 10.1111/j.0022-0477.2004.00900.x

  • CrossRef
  • Google Scholar

• 42

  The Editorial Committee of Flora of China (1986). “Chinese academy of sciences,” in Flora Republicae Popularis Sinicae Volume 73, Part 1. Beijing: Science Press. (In Chinese with English abstract).

  • Google Scholar

• 43

  UllahH.WangX.AliS.KongD.WangX.YangS.et al (2025). Significant spatiotemporal shifts of land cover over the Mongolian Plateau through the past three decades. Sci. Total Environment999, 180364. 10.1016/J.SCITOTENV.2025.180364

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 44

  WangT. C.LongJ. R.WangL. Q.JingZ. C.ShiJ. J. (2009). Changes in plant diversity, biomass and soil C, in alpine meadows at different degradation stages in the headwater region of three rivers, China. Land Degrad. and Dev.20 (2), 187–198. 10.1002/ldr.879

  • CrossRef
  • Google Scholar

• 45

  WangY.TangC.WangR.SuY.LiF. (2024). “The relationship between the community landscape characteristics and soil properties in the Liangucheng national nature reserve in minqin, Gansu,” in 2024 the 8th International Conference on Energy and Environmental Science (ICEES 2024). ICEES 2024. Editor LiuY. (Cham: Environmental Science and Engineering. Springer). 10.1007/978-3-031-63901-2_6

  • CrossRef
  • Google Scholar

• 46

  WilseyB. J.PotvinC. (2000). Biodiversity and ecosystem functioning: importance of species evenness in an old field. Ecology81 (4), 887–892. 10.1890/0012-9658(2000)081[0887:BAEFIO]2.0.CO;2

  • CrossRef
  • Google Scholar

• 47

  WuH.DingJ. (2020). Abiotic and Biotic Determinants of Plant Diversity in Aquatic Communities Invaded by Water Hyacinth [Eichhorniacrassipes (Mart.) Solms]. Front. Plant Sci.11, 1306. 10.3389/fpls.2020.01306

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 48

  XuJ.DangH.HuD.ZhangP.LiuX. (2024). Patterns of diversity and community assembly and their environmental explanation across different types of shrublands in the Western Loess Plateau. Forests15 (2), 222. 10.3390/F15020222

  • CrossRef
  • Google Scholar

• 49

  YangZ. L.BaY.ChenJ.QinM. L.DuW. X.XuX. P. (2008). Relevance analysis of population densities of five desert shrub species to water-thermal factors. Chin. J. Agrometeorology29 (2), 177–180.

  • Google Scholar

• 50

  YongS. P.ZhaoY. Z. (1979). Basic characteristics of the flora in the typical desert Region of Northern Langshan. J. Inn. Mong. Univ. Sci. Ed. (02), 91–113. (In Chinese with English abstract).

  • Google Scholar

• 51

  ZhangP.YangJ.ZhaoL.BaoS.SongB. (2011). Effect of Caragana tibetica nebkhas on sand entrapment and fertile islands in steppe–desert ecotones on the Inner Mongolia Plateau, China. Plant Soil347 (1-2), 79–90. 10.1007/s11104-011-0813-z

  • CrossRef
  • Google Scholar

• 52

  ZhangW. H.LiL.DijkstraF.ZhuB.BaiW.TianQ. (2023). Plant-soil interactions in grasslands of the Mongolian plateau under global change. Plant and Soil491 (1/2), 1–7. 10.1007/s11104-023-06291-1

  • CrossRef
  • Google Scholar

• 53

  ZhaoY. Z. (2012). Classification and Its Floristic Ecological Geographic Distribution of Vascular Plants in Inner Mongolia. Hohhot: Inner Mongolia University Press(In Chinese with English abstract).

  • Google Scholar

• 54

  ZhaoW.LiuY.LiY.ZouC.ShimizuH. (2025). Unveiling the impact of climatic factors on the distribution patterns of Caragana spp. in china’s three northern regions. Plants14 (15), 2368. 10.3390/PLANTS14152368

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 55

  ZhengJ.ChenY.WuG. (2013). Association of vegetation patterns and environmental factors on the arid Western slopes of the helan Mountains, China. Mt. Res. Dev.33 (3), 323–331. 10.1659/mrd-journal-d-12-00088.1

  • CrossRef
  • Google Scholar