1. Introduction

Beta diversity refers to the variation in species composition across different locations, directly linking local biodiversity (α diversity) with the wider regional species pool (γ diversity) (Tuomisto, 2010; Anderson et al., 2011; Socolar et al., 2016). However, unraveling the contributions of individual species and their combinations on beta diversity poses challenges. The two primary approaches to measuring beta diversity, which are either additive (Hα + Hβ = Hγ) or multiplicative (Hα × Hβ = Hγ) (Jost, 2007; Baselga, 2010; Chao et al., 2012), can lead to alpha-dependent beta diversity indices (Jost, 2007). Consequently, ecologists recommend partitioning beta diversity into distinct components to avoid misinterpretations (Jost, 2007; Ellison, 2010; Anderson et al., 2011). Partitioning also supports the development of informed conservation strategies within ecosystems.

Previous research has calculated total beta diversity within a community by using an ecological uniqueness approach (Legendre and De Cáceres, 2013). This framework allows for further decomposition of overall beta diversity into the local contribution to beta diversity and species contribution to beta diversity (Nakamura et al., 2020; Archidona-Yuste et al., 2020). Local contributions to beta diversity (LCBD) reflect the ecological uniqueness of each sampling points (Dubois et al., 2020; Yao et al., 2021). Although sites with high LCBD do not always exhibit hyperdiversity or host rare species (Vilmi et al., 2017), they display distinct characteristics in terms of conservation and alteration systems compared to most other sites (Heino and Grönroos, 2017; Hill et al., 2021). For conservation planning purposes, sites with high LCBD values often exhibit unique species assemblages or, alternatively, low species richness due to degradation (Hill et al., 2021). These sites are prime candidates for ecological restoration efforts (Vilmi et al., 2017; Dubois et al., 2020). Meanwhile, the SCBD indicates the relative importance of species to the beta diversity observed across different sites. The SCBD identifies species that significantly influence differences in biological communities or require special consideration in conservation efforts (Qiao et al., 2015; Santos et al., 2021).

The ecological uniqueness approach integrates beta diversity with analytical methods designed to test hypotheses about the mechanisms that generate and sustain beta diversity (Legendre and De Cáceres, 2013; Yang et al., 2021). LCBD may correlate with various community metrics (da Silva et al., 2020), environmental conditions (Wang et al., 2020), and disturbance activities (Leão et al., 2020; de Paiva et al., 2021). Conversely, SCBD has been associated with species characteristics such as abundance, niche position, and biological traits (da Silva et al., 2018; Santos et al., 2021). These ecological and biological attributes are likely interrelated, highlighting the connections between species and their environments. However, most studies that quantify ecological uniqueness using LCBD, SCBD, or both have to date predominantly focused on freshwater ecosystems and their resident organisms, including zooplankton in lakes (Brito et al., 2020; Loewen et al., 2020), fish in river systems (Arantes et al., 2018; Duarte et al., 2022), and insects in streams (López-Delgado et al., 2020). While previous research has examined the local contribution, it has largely been limited to forest ecosystems (Yao et al., 2021; He et al., 2022) and subtropical reservoir phytoplankton (de Moura et al., 2022). Consequently, there is a pressing need for studies that explore general patterns of ecological uniqueness and their determinants across diverse ecosystems and biota.

Grassland ecosystems, covering approximately 40% of the Earth's land surface, play a vital role in providing essential ecosystem services, including erosion control, pollinator health, and the maintenance of vegetation productivity and stability (Bai and Cotrufo, 2022; Strömberg and Staver, 2022). In contrast to forest ecosystems with dense canopy structures (He et al., 2022), grasslands feature open spaces and herbaceous vegetation (Fontana et al., 2020), leading to distinct ecological dynamics and conservation requirements. The distinction is crucial, as grasslands are becoming increasingly susceptible to land-cover changes caused by intensive agriculture, afforestation, and the impacts of seasonal precipitation, as well as invasive species following fires and droughts (Buisson et al., 2022). Despite their global significance, no specific studies have examined the contributions of individual species and species assemblages to plant beta diversity within grassland ecosystems.

This study applied the ecological uniqueness approach to examine both local contributions to beta diversity (LCBD) and species contributions to beta diversity (SCBD) across six major geographical ranges in alpine grasslands in the Qinghai-Tibet Plateau. First, we examined how LCBD in alpine grasslands reflects gains or losses of plant species. Second, we determined which community metrics and environmental gradients influence LCBD. Finally, we examined how niche breadth (i.e., narrow vs. broad) contributed to beta diversity.

2. Materials and methods 2.1. Study site description

The Qinghai-Tibet Plateau (QTP) hosts vast alpine grasslands, covering over 60% of its area, which are critical for biodiversity conservation and pastoralism in the Palaearctic region (Dong et al., 2023). However, climate warming and vegetation degradation over the past few decades have intensified the conflict between ecological conservation efforts and the livelihood concerns of local herders (Pech et al., 2007; Dong et al., 2023; Zhang et al., 2024). Therefore, investigating the patterns of the ecological uniqueness of plant communities in alpine grassland is crucial for maintaining regional plant diversity and for developing effective strategies for sustainable grazing management in the QTP.

This study collected and analyzed data from six alpine grassland sites located at the northeastern edge of the Qinghai-Tibetan Plateau (Fig. 1). The soil in the study area is classified as alpine grassland soil (Cambisols and Gelisols), with high organic carbon content (5–8%) and an average pH range of 6.5–7.6 (Shi et al., 2004). These sites are important biodiversity hotspots, supporting unique plants such as Carex capillifolia and Carex moorcroftii, and animals such as plateau pikas (Ochotona curzoniae), and play a key role in regional water conservation and carbon sequestration (Wang et al., 2020). Over the past 20 years, some alpine grasslands have been leased to local herders for seasonal rotational grazing. The six survey sites, with elevations ranging from 3200 to 3750 m a.s.l., received an average annual precipitation between 250 mm and 653 mm, and were located in a cold and humid continental plateau climate (Table 1). The dominant vegetation in the alpine grasslands includes sedge species such as Carex capillifolia, C. humilis, and C. pygmaea. Additionally, key associated species include C. moorcroftii, Poa crymophila, Elymus nutans, Stipa purpurea, and Saussurea pulchra.

2.2. Data collection

Sixty 35 m × 35 m plots were established to represent the local vegetation type. To minimize the impact of non-natural factors, we selected survey sites with minimal human interference, grazing, and land use intensity, and those located far from main roads to reduce local disturbances. Specifically, ten plots were selected along the route at each site, spaced across different elevation gradients to capture a range of elevational variation. To avoid spatial dependence among sampling sites, adjacent sites were separated by a minimum distance of 10 km in this study. Plant communities were surveyed during the peak of the 2023 vegetation season (July–August) to maximize the completeness of the plant diversity data. In each plot, nine quadrats were arranged diagonally, with a spacing of approximately 6 m between them. Each quadrat measured 1 m by 1 m. All species within quadrats were identified and recorded. The needlepoint method was employed to measure species cover in each quadrat. This involved vertically inserting a needle into the ground and recording the plant species or parts it touched. The process was repeated 100 times across grid or random points within the quadrat to ensure representativeness. Simultaneously, species abundances (total number of each plant species) were recorded. The height of species was measured using a tape measure. An individual was operationally defined as a plant with a physically separated base and no visible aboveground connection to adjacent units. A cluster referred to a group of ramets or shoots sharing a basal connection. To ensure consistency in the field, units were considered distinct if their bases were at least 5 cm apart. If the abundance of a species exceeded 10 individuals or clusters, 10 were randomly sampled for measurement; otherwise, all individuals were measured.

2.3. Explanatory variables 2.3.1. Environmental variables

To evaluate factors that may affect the distribution and diversity of plant species, we examined six environmental variables that reflect the structural and bioclimatic characteristics of the alpine grasslands: (1) Elevation; (2) MAT (Mean Annual Temperature); (3) SI index (The SI index is a metric that measures the degree of seasonal variation in precipitation) (Walsh and Lawler, 1981); (4) NDVI (Normalized Difference of Vegetation Index); (5) MAP (Mean Annual Precipitation); (6) ET (Evapotranspiration). Elevation data were obtained from field measurements using a GPS device. MAP data for each point were obtained from the Earth System Science Data Center. Monthly precipitation raster data with a 1 km resolution from 1901 to 2021 in China. MAT and ET data were obtained from the China Meteorological Data Network. NDVI data were collected from the Moderate Resolution Imaging Spectroradiometer (MODIS) aboard NASA's Terra satellites, which were updated every 16 d with a resolution of 250 m.

2.3.2. Plant community characteristics

We assessed plant communities based on five key variables: species richness, relative abundance, and three functional trait indices. Species richness was defined as the average number of species per quadrat within a plot. The relative abundance of the plant community was calculated as the proportion of total plant abundance in each plot (35 m × 35 m), derived from the average plant abundance across its 9 quadrats (1 m × 1 m), relative to the total plant abundance across all surveyed plots. We chose functional indices — functional richness, functional evenness, and functional divergence — which are commonly used in plant communities. In this study, the species metrics included four variables: the number of sampled plots occupied, which reflected the extent of species throughout the plots, ranging from species absent in a plot to a species that occurred in all plots; individual abundance; plant height and cover.

2.4. Statistical analysis 2.4.1. SCBD and LCBD calculation

All statistical analyses and result visualizations were conducted in R v.4.2.0 (R Core Team, 2022). LCBD and SCBD indices were calculated using the beta.div function from the "adespatial" package (Dray et al., 2018). A Hellinger transformation was applied to the species-by-site presence/absence matrix, after which we calculated the squared deviations of each transformed value from the species mean. The total sum of squares (SS_T) was obtained by summing the matrix elements containing these squared differences. To calculate total beta diversity, we divided the total sum of squares by the number of sites minus one, resulting in values ranging from 0 to 1. A maximum value of 1 indicates that plant compositions at all alpine steppe sites are entirely distinct.

2.4.2. Turnover and nestedness calculations

To determine whether turnover or nestedness better accounts for the variation in LCBD in alpine grassland communities, we utilized the "beta.div()" function from the "adespatial" package. This method, designed for presence-absence data, decomposes total beta diversity into dissimilarity, species replacement (turnover), and richness differences (nestedness). Turnover and nestedness highlight the uniqueness of each site in terms of species replacement or richness variation. The function generates three matrices along with global outputs: total beta diversity (BD total), total turnover, and total nestedness.

2.4.3. Functional metric calculation

This study derived three functional indices for plant communities by creating a matrix of essential functional traits for the species involved. Plant functional richness was calculated using species versus twelve species key function traits matrices (see in supplementary materials). Gower's distance was utilized to develop a species-by-species functional distance matrix from the scaled and centered trait data (Villéger et al., 2008), employing the function provided in the "ade4" package in R (Pavoine et al., 2009). This method accommodates various types of variables, including quantitative, nominal, fuzzy, ordinal, and circular data. From this functional distance matrix, we subsequently computed three functional diversity indices using the dbFD function from the "FD" package in R (Laliberté and Legendre, 2010).

2.4.4. Statistical modeling

First, we assessed LCBD differences among six sites using ANOVAs. Tukey's HSD tests for pairwise comparisons were conducted using the "emmeans()" function from the "emmeans" package (Lenth, 2020). Before conducting these analyses, we checked the assumptions of homogeneity of variance using Levene's Test from the "car" package and normality using the Shapiro–Wilk test (Fox et al., 2007). Non-normally distributed data undergoes log transformation to meet statistical assumptions and optimize model fit. Statistical tests were deemed significant at p < 0.05.

To capture the variability between different locations and control for the autocorrelation between sites, this study applied a Linear Mixed-Effects Model (LMM) to model LCBD. We incorporated site as a random effect in the model to account for the non-independence of samples within the same sites, ensuring that intra-location correlations do not bias the estimates. Initially, we linked LCBD to various community metrics, including species richness, community relative abundance, and functional indices. Following this, we explored the influence of environmental variables as predictors for LCBD.

Considering the potential nonlinear effects of certain environmental variables, as highlighted by several studies (Heino and Grönroos, 2017; Tan et al., 2019; Wang et al., 2020), squared terms of these variables were incorporated into the final LMM models. The variance inflation factor (VIF) was calculated using the "car" package to avoid the impact of high multicollinearity on model interpretation. Only variables with VIF values below 5 were retained in the model. Meanwhile, Moran's I test was applied to the model residuals using the "MuMIn" package to assess spatial autocorrelation in the variables (Bartoń, 2024). The final LMM was selected based on the lower AICc value after testing quadratic terms. LMM model construction was performed using the "lme4" package (Bates et al., 2015).

This study adopted Piecewise Structural Equation Modelling (Piecewise SEM) to delve into the direct and indirect influences on LCBD (Lefcheck, 2016). To expedite the modeling process, all measured factors were transformed into "composite variables" based on the coefficients derived from the LMM analyses. This approach provided both the "marginal" and "conditional" contributions of climate and plant community predictors in driving LCBD. The "piecewiseSEM" packages in R facilitated these analyses (Lefcheck et al., 2016), with the Chi-squared test employed for the assessment of the model goodness-of-fit. Through iterative refinement of the models, driven by the significance of the paths (p < 0.05) and the overall model fit, this study discerned the standardized effect sizes of the relationships, thereby identifying the direct and indirect drivers of LCBD. This study also employed a random forest (RF) approach to assess the importance of the observed variables (Breiman et al., 2001). We use the "rfPermute" package to assess the significance of each predictor for the response variable (Jiao et al., 2018).

Next, this study linked SCBD to the following species traits: the number of species occupying the community, individual abundance, average plant height and coverage. When SCBD values were correlated with species metrics such as height, coverage, and abundance, our objective was to ascertain whether species that significantly influence beta diversity possess either small or intermediate ecological niches. In addition, the squared term of species occupancy was included in the final model to capture any potential nonlinear relationships. Plant height, coverage, and abundance were averaged across all quadrats within each plot. This approach represents the overall species traits at each plot level and does not include a nested structure by sub-location. Therefore, we applied beta regression with a logit link function to model SCBD using the "betareg" function (Cribari-Neto and Zeileis, 2010). This approach was chosen because beta regression is well-suited for response variables constrained between 0 and 1, providing a more precise fit for SCBD (Cribari-Neto and Zeileis, 2010).

3. Results

Across six study sites, we identified a total of 72 herbaceous species classified into 22 families, representing the γ diversity of the surveyed area. The number of species (α diversity) at each site ranged from 31 to 66 (51.50 ± 12.43 species; mean ± SD) (Table S1). Most species were common to multiple sites, and only three were unique to one site each: Carex kokanica and Euphrasia pectinate at LQ, and Euphorbia helioscopia at HY. The relative abundance of plant communities varied significantly between sites (F = 27.21, p < 0.001). The highest values of relative abundance of plant communities were recorded at LQ, whereas the lowest was at GH (Table S2). The most abundant plant was Carex humilis, which was observed at all sites. Other plants were usually abundant once they appeared, such as Poa crymophila, Leontopodium nanum, and Saussurea pulchra (Table S3).

3.1. LCBD of alpine grasslands

Plant beta diversity of alpine grasslands across all study sites was quantified at 0.499. The average LCBD was 0.017, with a range from 0.011 to 0.029 (Fig. 2a). Ecological uniqueness in species composition was high at two sites (i.e., MY and QL). Comparative analysis of sites revealed marked compositional differences (F = 93.56, p < 0.001; Fig. 2b), with one site (MY) diverging from all others (Table S4).

Partitioning of the total plant beta diversity revealed that the percentage of alternative ingredients (turnover = 0.899; 96%) was significantly higher than the total richness difference (nestedness = 0.033; 4%), indicating that the difference between the LCBD was caused by species replacement. Examination of the six sites individually indicated that species differences at one site (LQ) were caused by the nested component of the species (70%), whereas differences between other sites (i.e., HY, QL, GC, and MY) were caused by a larger proportion of species replacement (> 85%). Our analysis indicated that at one site (GH) species differences were caused by richness and species replacement (turnover = 51.8%; nestedness = 48.2%).

3.2. Explaining the LCBD of alpine grasslands

The community metric models examined in this study showed that LCBD was negatively correlated with species richness, relative abundance and functional richness of plant communities (Tables 2 and S5; Fig. 3a–d).

The models containing environmental covariables showed that SI index and evapotranspiration were correlated with LCBD (Tables 2 and S5; Fig. 3e and f). In particular, LCBD had a minimum value as the SI index increased. Furthermore, our model showed that the LCBD was negatively correlated with evapotranspiration.

Both local and regional plant diversity are directly and indirectly influenced by climate (Fig. 4). Notably, the direct influence of climate on LCBD was stronger than that of plant community characteristics. Although the SI index was the best predictor of large-scale LCBD dynamics, the relative abundance of plant communities was critical at a local level.

3.3. Explaining the SCBD of alpine grasslands

SCBD values spanned from 0.001 to 0.022, with 41 out of 72 species contributing to beta diversity above the average (> 0.0139) (Fig. 5a). Potentilla multifida was the most important plant for beta diversity (SCBD value: 0.022) (Fig. 5b), whereas Thlaspi arvense contributed the least to beta diversity (SCBD value: 0.001) (Table S6).

Beta regression of the species metrics had a positive effect on the number of plants occupying the communities (Table 3 and Fig. 6a). In contrast, none of the tested species metrics (height, cover, and abundance) were significant predictors of SCBD (Fig. 6b–d).

4. Discussion

We found that the total beta diversity values across six alpine grasslands were relatively low (0.499, from a maximum of 1) (Fig. 2a), indicating that our study area had a uniform species composition and a relatively small number of exclusive species. This pattern is expected since all our study sites were situated in the same alpine grassland region of the northeastern QTP. As a result, they shared similar habitat features, species interactions, and ecological histories. In general, sedges (e.g., C. pygmaea and C. humilis) are widely distributed, with some species being replaced by closely related allopatric or parapatric species across the region. For instance, several closely related species (e.g., Saussurea hieracioides and S. pulchra; Gentiana aristata, and G. straminea) were documented at different sites.

Most differences in our plant communities were driven by species turnover, with nestedness playing a lesser role (Fig. 2b). It indicates that the LCBD at each location was primarily influenced by concurrent species additions or removals, attributable to environmental filtering, competition, or historical influences, rather than shifts in species abundance. Environmental variations between sites, such as local disturbances, can benefit specific species, which could explain this pattern. However, two of our study sites (i.e., LQ and GH) differed for distinct reasons (Fig. 2c). The beta-diversity at LQ was almost strictly the result of nestedness, whereas that of GH was the result of turnover and nestedness. This pattern can most likely be explained by the antagonism of species richness (Radinger et al., 2016). GH and LQ were distinct sites with GH displaying notably low species richness and LQ exhibiting high richness values. Despite these contrasts, both sites showed elevated overall relative abundance. Historical factors such as habitat fragmentation, study area size, and habitat quality might explain the varied responses of grassland plants at these locations. These factors may have led to the increased or decreased presence of certain species in LQ and GH, thus highlighting the disparities in species richness. Our findings suggest that LCBD is co-determined by local environmental and community characteristics. Specifically, we observed a negative correlation between plant LCBD and α diversity, with the lowest LCBD values and the highest species richness and relative abundance recorded at the LQ site. This indicates that communities with high local diversity tend to be compositionally similar to others, thereby contributing less to beta diversity. Comparable patterns — i.e., a negative association between LCBD and species richness — have been reported in mammalian and forest ecosystems (Tan et al., 2019; Santos et al., 2021). In our case, LCBD was primarily driven by variation in the occurrence of rare species. While common species showed little variation across sites, the inclusion of rare species resulted in a positive correlation between LCBD and species richness. Kong et al. (2017) found that in the South-East Asian flood-pulse system, LCBD was positively associated with taxonomic diversity at certain fish sampling sites, highlighting the context-dependent nature of this relationship. Thus, the high proportion of common species may explain the negative correlation with LCBD. Consequently, HY and GH showed comparable LCBD values, indicating that their community compositions are less variable in relation to one another.

Overall, the best local predictor of LCBD was the relative abundance of plant communities at the regional scale. Higher community relative abundance often corresponded to lower LCBD (Fig. 3b), primarily because higher abundance typically indicates more homogenous communities with dominant species, reducing site uniqueness in terms of species composition and ecological function. Our study revealed an inverse relationship between LCBD values and plant functional richness and evenness in alpine grasslands (Fig. 3c and d). This observation aligns with previous research that found elevated LCBD values are associated with lower species and functional richness in some forest mammals (da Silva et al., 2020). This supports the concept that LCBD, based on composition, is closely linked to shifts in certain grassland plant functional metrics. These findings imply that an increase in species richness may result in reduced LCBD of plant communities, though this does not necessarily indicate higher specialization or original functional diversity. Species-rich areas tended to include plants with more similar traits, reflecting convergence within the local functional space. As observed in alpine grasslands, this suggests that the LCBD can be used to forecast functional changes in vegetation.

The ecological uniqueness was strongly correlated with precipitation seasonality (Fig. 4). Precipitation seasonality plays a critical role in the persistence of living organisms (Walsh and Lawler, 1981; Dayton, 2008). Precipitation tends to drive biomass allocation in alpine grasslands through its effects on plant community traits and functions (Sun et al., 2023). Seasonal variations in precipitation conditions affect the availability of resources and, in turn, the presence or absence of plants in the environment at both temporal and spatial scales (Dayton, 2008; Van Dyke et al., 2022). Over time, natural selection favors alpine grassland plants, which are physiologically adaptable, allowing them to maximize seasonal, sporadic resources and cope with extreme environmental conditions. In addition, high plant richness in the region may depends on precipitation stability, whereas habitat homogenization promotes the coexistence of species with similar ecological requirements. This leads to a reduction in LCBD through competitive exclusion processes.NDVI is commonly used to represent grassland productivity gradients and habitat homogenization (Yang et al., 2008), whereas plant beta diversity and ecosystem multifunctionality in grassland ecosystems exhibit elevation dependence (Kraft et al., 2011; Wang et al., 2023). However, unexpectedly, in our study, elevation and NDVI did not drive the patterns of LCBD, despite similar patterns being observed in other biological groups (Tan et al., 2019; Yang et al., 2021). One possible explanation for this is that NDVI might not capture the specific habitat homogenization that influences LCBD. Additionally, as our elevation gradient spans only an average of 850 m, the influence of elevation on species distribution might be masked by multiple environmental variables. Even so, the results support the argument that LCBD values are higher in areas with lower evapotranspiration because limited water availability and cooler conditions create more heterogeneous environments. This heterogeneity promotes ecological niche differentiation and increases species turnover across sites, leading to higher LCBD. Additionally, resource constraints in low evapotranspiration areas favor specialized species adapted to unique conditions, further amplifying site-to-site variability in species composition. Our study reaffirms that both local and regional plant diversity are directly and indirectly influenced by climate, highlighting climate as a key factor in shaping unique species assemblages across broad geographic scales.

Our study provides a ranking of species contributions to the plant beta diversity in the surveyed area (Table S6). This fundamental list serves as essential information for plant conservation, reflecting the ecological significance of herbaceous plants at larger spatial scales. Our model supports the observation that SCBD is influenced by native occupancy, with species exhibiting common to moderate occupancy levels contributing the most to beta diversity (Fig. 6a). For instance, in contrast to dominant and rare plants in alpine grasslands, the species with the highest incidence-based SCBD — Potentilla multifida, Stipa purpurea, Veronica polita, Gentiana dahurica, and Leontopodium nanum — are relatively common and possess intermediate-sized niches. More specifically, we observed that Potentilla multifida and Stipa purpurea had notably high SCBD values. These species were well distributed across all sites, representing 41–55% of the sampled plots. Studies measuring SCBD in other groups have similarly revealed positive correlations with species occupancy (Heino and Grönroos, 2017; Szabó et al., 2019), indicating a strong influence on this metric. This variation may be attributed to the differing occupancy levels of these species among sites (Gaston et al., 2006). However, it is important to note that SCBD does not correlate with species metrics (Fig. 6), such as the individual abundance of alpine grassland plants, which contradicts our initial expectations. This may be because SCBD can be predicted by niche location, whereas individual biological traits do not fit this equation well (Heino and Grönroos, 2017). Nevertheless, these findings do not imply that such effects are nonexistent in all grassland plant communities; rather, their impact may be less pronounced compared to other communities. Further investigation is necessary to explore these effects, particularly considering that dispersal models and plant size can constrain species distribution.

5. Conclusion

Our standardized quadrat survey approach effectively assessed the ecological uniqueness of grassland communities, identifying plants that contribute to current beta diversity patterns. Monitoring grasslands using LCBD offers valuable insights into changes in community structure, enhancing our forecasting ability to preserve a complete regional pool of herbaceous plants. For instance, under limited conservation resources, managers can prioritize plant communities for protection based on LCBD (Dubois et al., 2020), even if sustainability requires focusing on the entire study area when species replacement dominates the community (Socolar et al., 2016). Additionally, LCBD can highlight environmental gradients experiencing significant turnover. Habitats with high LCBD values warrant protection as they pinpoint areas where species richness and key functional traits undergo significant changes (da Silva et al., 2020). We also highlight the importance of SCBD, which identifies individual plants as viable ecological indicators, allowing conservation efforts to target particularly unique or locally rare plant species. Specifically, plants with high SCBD values should be prioritized if our goal is to protect areas with distinct richness or nestedness (Socolar et al., 2016), as these plants are crucial in maintaining diverse species compositions. Within this framework for sustaining grassland plant diversity, we advocate for further investigation into disturbance mechanisms, such as grazing and population outbreaks of wild small herbivores (Dong et al., 2023; Zhang et al., 2024). and their impacts on ecological uniqueness. This is crucial for effective small herbivore management and sustainable grazing system establishment.

Acknowledgements

This work was supported by the National Key Research and Development Program of China (2023YFF1304302), and the Qaidam basin and Qilian Mountains germplasm resources collection project (Grant No. SJCZFY2022-1-6). The authors would like to thank Huan Yang, Yuan Yuan Duan, Hao Hao Qi, Ling Ling Wang, Cai Feng Liu, Tong Wu, Yi Mo Wang, Xue Ting Xu, from Lanzhou University for the contributions made to this study through their field assistance and laboratory analysis.

CRediT authorship contribution statement

Jie Li: Writing – review & editing, Writing – original draft, Visualization, Methodology, Investigation, Data curation, Conceptualization. Xiao Pan Pang: Writing – review & editing, Methodology. Zheng Gang Guo: Writing – review & editing, Methodology, Investigation, Funding acquisition, Conceptualization.

Data availability statement

The data generated in this study are available on Zenodo at https://doi.org/10.5281/zenodo.13918347.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.pld.2025.05.003.