1. Introduction

New roads are anticipated to expand by 25 million kilometers globally by 2050 (Laurance et al., 2014). This rapid development is expected to have profound impacts on global ecosystems. Studies have noted that road construction can lead to vegetation destruction, habitat fragmentation, and biodiversity loss (IPBES, 2019) and that rapid road development poses a significant obstacle to achieving biodiversity targets (CBD, 2020). Fortunately, road construction does not inevitably result in biodiversity loss. Although roads are commonly believed to facilitate the spread of exotic species while being detrimental to native species (Ansong and Pickering, 2013; Okimura et al., 2016), a majority of studies have shown that roads have a positive impact on native species (Lázaro-Lobo and Ervin, 2019). In addition, current research overemphasizes the impact of roads on exotic species and neglects their comprehensive impact on community diversity (Suárez-Esteban et al., 2016). Consequently, there remains considerable uncertainty regarding how roads affect the biodiversity of plant communities (Feng et al., 2012; Zhou et al., 2020; Li et al., 2022).

Ecosystems that may be particularly impacted by road disturbance include those at high latitudes or altitudes. There are several reasons these cold ecosystems may find it more challenging to recover from road construction damage than those in warmer regions (Wang et al., 2004; Bölter and Müller, 2016). Expanding road networks have been shown to provide additional avenues for species dispersal (Vorstenbosch et al., 2020), which may reduce local diversity and change species composition (Lembrechts et al., 2014). Furthermore, global warming mitigates the low-temperature limitations in these regions, potentially allowing for the colonization of exotic species. Importantly, high-altitude regions often harbor many endemic species and are priority areas for biological conservation (Myers et al., 2000). In these biodiversity hotspots, land-use pressures, including road construction, may reduce biodiversity integrity (Newbold et al., 2016). Consequently, researchers have recommended that these areas be protected as nature reserves, and that road disturbance be minimized (Laurance et al., 2014).

The Tibetan Plateau is Earth's widest and highest orogenic system (Ding et al., 2022) and a global hotspot for species endemism (Myers et al., 2000). Over recent decades, extensive road development has occurred on the Tibetan Plateau to meet the demands of economic development (Gao et al., 2019), raising concerns about the associated biodiversity risks. Nevertheless, how roads affect plant diversity on the Tibetan Plateau remains poorly understood, as research has primarily concentrated on biophysical conditions (Zhou et al., 2023), wild animal migration (Ru et al., 2022), soil bacterial community structure (Liu et al., 2021), and ecosystem services (Yang et al., 2023).

The ecosystems on the Tibetan Plateau most influenced by roads are alpine grasslands, including alpine steppes and meadows (Chen et al., 2003). Alpine grassland plant richness and evenness have been shown to be negatively affected by road disturbance, and roads were not found to differentially affect alpine meadows and alpine steppes (Guo et al., 2007). In contrast, some reports have argued that alpine meadows are more sensitive to road disturbances than are alpine steppes (Ran et al., 2019) and less resilient to risks (Chen et al., 2007). Thus, it remains unclear whether road disturbance uniformly affects plant diversity across different biomes and environmental gradients on the Tibetan Plateau.

This study used large-scale field surveys to comprehensively examine the impacts of road disturbance on the plant diversity of alpine grasslands on the Tibetan Plateau. Our specific objectives were to (a) evaluate the collective impact of roads on plant diversity and species composition at a regional level and (b) investigate how environmental factors affect the response of plant diversity to road disturbance. Assessing the risk of biodiversity loss in the alpine grasslands of the Tibetan Plateau due to long-term road operation is of great significance for biodiversity conservation in high-altitude regions worldwide.

2. Materials and methods 2.1. Study area

We investigated 25 sites on the Tibetan Plateau, including 10 in alpine meadows and 15 in alpine steppes (Fig. 1). The climate and vegetation of these sites vary considerably (Wang et al., 2022). Alpine meadows receive an annual precipitation of approximately 400 mm, with July temperatures below 9 ℃. Annual precipitation in alpine steppes ranges from 200 to 400 mm. Drought-tolerant graminoids dominate alpine steppes, whereas mesic and meso-xeric species prevail in alpine meadows (Wang et al., 2022).

2.2. Experimental design and field measurements

Vegetation surveys were conducted during the peak growing season from July through August 2020 to 2022. To ensure a comprehensive assessment of road impacts, we employed four criteria in selecting survey sites: (a) Areas where roads were built alongside rivers or mountains were excluded; in these regions, the impact of natural features may outweigh that of the roads, posing challenges for investigating the pure effects of roads on plant diversity in adjacent grasslands; (b) Grasslands with significant slopes were also excluded, as slopes can lead to water accumulation, which may confound the effects of roads; (c) To minimize interference from other human activities, we avoided sites near towns, villages, and those with visible signs of vehicles or construction equipment; (d) Sites with non-representative biome types were not considered; for instance, occasional alpine meadows within regions characterized by alpine steppes, altered due to their proximity to rivers, were excluded from this study. We ultimately selected 25 sites that met these criteria and collected data on slope, biome type, road class, road age, moisture, and temperature conditions (Appendix A).

Previous studies have indicated that in alpine grasslands of the Tibetan Plateau, road impacts on vegetation are typically restricted to 100 m (Wang, 2015; Bao et al., 2024). A recent study reported that plant species composition did not differ significantly between 100 m and 250 m from roads on the Tibetan Plateau (Tan et al., 2024). Roads also often do not affect abiotic factors beyond the 100-m range, such as soil temperature (Zhou et al., 2022), soil moisture (Feng et al., 2012), nitrogen deposition (Truscott et al., 2005), or heavy metal enrichment (Zhang et al., 2015). Therefore, we established a 100-m transect perpendicular to the direction of the road at each site. We set up plots (1 m × 1 m) at varying distances of 2, 10, 20, 30, 50, and 100 m from the base of the roadside slope, with the plots at 100 m serving as controls. At each sample site, vegetation surveys were conducted, plant species were recorded, and their height and cover were measured. With the help of experts, we identified all plant species within the sampling plots and further categorized them into four plant functional groups: grasses, sedges, legumes, and forbs (Wang et al., 2023b). Each plot (1 m × 1 m) was divided into 100 sub-plots (0.1 m × 0.1 m), yielding 121 points. The height of a given species was recorded as the average value of five random measurements (using a ruler), and the cover was recorded as the number of times it occurred at these points divided by 121.

2.3. Biodiversity indexes

Species richness (R) indicates the number of species in a grassland. Pielou's evenness (J) represents the uniformity of species distribution within the community, independent of richness. As a complement to Simpson's dominance index, Simpson's diversity index (D) negatively correlates with the dominance level of the primary species within the community. Shannon–Weiner's diversity index (H) is a composite measure for species number and evenness. They were calculated as (Lauren et al., 2018):

(1)

(2)

(3)

(4)

where S is the number of species present within the plot, and P_i is the importance value of the ith species, calculated as:

(5)

where H_i is the height, and C_i is the cover of the ith species. The importance value of a given plant functional group was defined as the sum of the importance values of all species within it.

The β-diversity indexes were used to quantify the dissimilarity between grasslands near the road and those away from the road (Koleff et al., 2003). Sørensen's dissimilarity index (β_Sør), one of the most common β-diversity indexes, was calculated as follows:

(6)

where a is the number of shared species, and b and c represent the number of species unique to the plots at the roadside and away from the road, respectively. Simpson's dissimilarity index (β_Sim) was used to quantify true species turnover without the influence of richness differences:

(7)

The difference between these two indexes was referred to as nestedness (β_Nes), i.e., dissimilarity due to differences in species richness (Baselga, 2010):

(8)

2.4. Statistical analyses

To make the data comparable at different sites, we quantified the impacts of the roads on plant diversity using response ratios (Hedges et al., 1999). A logarithmic transformation (lnRR) was used to improve the normality of the data distribution:

(9)

where X is the diversity index close to the road, and X_CK is the diversity index of the control group located 100 m away from the road.

To eliminate the interference of spatial heterogeneity across sample sites, we modeled road impacts with a linear mixed-effects model (Zuur et al., 2009), which assumes that the road's effects (intercepts) vary at different sites. For α-diversity, the model was set to be:

(10)

where (1│Site) represents the random intercept among sites. X_CK represents the α-diversity 100 m away from the road, and Distance represents the distance from the road's edge, both of which were Min-Max normalized (Jain et al., 2018) before entering the model. Biome, Class, Mois, and Temp are all categorical variables, representing biome type, road class, moisture condition, and temperature condition, respectively. The multiplication symbol "×" represents that the separate effects and interactions of the variables on either side of it were considered. We did not include the interaction between X_CK and Distance in the analysis because the interaction term would introduce strong multicollinearity into the models, resulting from both variables being continuous. Since β-diversity indicates the dissimilarity of grasslands near the road to grasslands 100 m from the road, the model for it has X as the dependent variable and excludes X_CK from the independent variables:

(11)

The full models, as shown in Equations (10) and (11), include all potential predictors. Nonetheless, a model with partial predictors may provide more significant explanatory power than the full model. Consequently, we tested all subsets of the predictors and employed the restricted maximum likelihood method to determine the optimal model (Gurka, 2006). Linear mixed-effects models were fitted, and model selection was performed using the R packages "lme4" (Bates, 2010) and "MuMIn" (Bartoń, 2013), respectively.

3. Results 3.1. Impact of roads on α-diversity

Alpine grasslands on the Tibetan Plateau experienced neutral impacts from roads on species richness, Shannon–Weiner's diversity, and Simpson's diversity but showed a significant increase in Pielou's evenness (Fig. 2a). The α-diversities of grasslands exhibited no evident variation with distance from roads (Fig. 2b–e). Except for species richness, higher α-diversities of grasslands 100 m from roads correlated with increased detrimental effects of roads on roadside grassland diversities (Table 1 and Fig. S1). A significant interaction between road distance and temperature conditions influenced road effects on Simpson's diversity (Table 1). In the plateau temperate zone, road-edge grasslands showed increased Simpson's diversity (p = 0.013, Fig. S2), with this effect diminishing as the distance from the road increased (p = 0.071, Fig. S2). Conversely, road distance in the plateau sub-frigid zone failed to explain the change in Simpson's diversity (Fig. S2).

3.2. Impact of roads on β-diversity

We used β-diversities to quantify differences in species composition between grasslands near and far away from roads. Simpson's dissimilarity and nestedness are components of Sørensen's dissimilarity, and they represent species turnover and species loss (or gain), respectively. Simpson's dissimilarity accounted for 75.7% (0.193 vs. 0.255) of Sørensen's dissimilarity (Fig. 3a), indicating species turnover dominated species variation near roads. Increasing distance from roads led to decreased Sørensen's and Simpson's dissimilarity (Fig. 3b, c and Table 2). Alpine steppes had lower Sørensen's and Simpson's dissimilarity than did alpine meadows (Table 2).

4. Discussion 4.1. Road disturbance has a neutral effect on plant richness but slightly increases evenness

We investigated how roads affect α-diversity in alpine grasslands on the Tibetan Plateau. We found that the effect of roads on species richness was neutral within 50 m, although roads slightly increased Pielou's evenness, leading to a non-significant increase in Shannon–Wiener's diversity. These findings are inconsistent with those of two previous studies that showed roads reduced plant richness and diversity in Tibetan Plateau grasslands (Guo et al., 2007; Tan et al., 2024). The inconsistency between our study and these may be the result of two different factors. First, previous studies were conducted on sections of the Qinghai-Tibet Highway, limiting their representativeness to those specific areas. Our results, which are based on more systematic and comprehensive sampling, may, therefore, more accurately reflect the overall impact of road disturbance on Tibetan Plateau alpine grasslands. Second, the effects of road disturbance on plant diversity may be heterogeneous. Roads may have positive or negative impacts at different sites (Fig. S1), which balance each other, leading to insignificant pooled results in this study. Comparable results to previous studies were observed at some of our sites (e.g., S06, S09, S11, S18, S23), suggesting that their results correspond to several samples in our study. These discrepancies highlight the complexity and scale-dependent nature of road impacts and emphasize the need for large-scale investigations on the Tibetan Plateau.

Roads modify the competitive advantage of grassland plants by influencing abiotic factors. Specifically, roads alter roadside temperatures and wind speeds, enriching road edges with water and nutrients (Suárez-Esteban et al., 2016; Lázaro-Lobo and Ervin, 2019). Dominant species well-adapted to the original environment often become less adjusted to the new environment created by road disturbance, whereas species adapted to these changes increase their competitive advantage (Hillebrand et al., 2008). Hence, road disturbance promotes more even species distributions across most regions (Feng et al., 2012). For example, Carex montis-everesti dominates in S01, with over 90% coverage. Roads allow runoff to collect along their edges, increasing soil moisture (Coffin, 2007). Improved moisture conditions significantly enhance the importance of moisture-loving species like Argentina anserina, promoting the evenness of grassland plants. In contrast, when species evenness surpasses specific thresholds (Fig. S3, Shannon–Wiener's diversity: 2.13; Pielou's evenness: 0.86; Simpson's diversity: 0.83), the dominance of species adapted to road disturbance results in reduced evenness and diversity (Wang et al., 2014). For example, in S23, species distribution was initially even (Pielou's evenness: 0.91). However, Carex parvula became dominant along the roadside, with dramatic declines observed in Asteraceae forbs (e.g., Youngia simulatrix, Saussurea arenaria, and Ajania khartensis). This shift resulted from improved hydrothermal conditions and increased resource availability at road edges (Lázaro-Lobo and Ervin, 2019), expanding the ecological niche occupied by sedges over forbs and aiding in the restoration of degraded grasslands (Luo et al., 2018).

Road impacts are argued to peak at road edges and diminish with distance (Cui et al., 2009; Zeng et al., 2011). However, in this study, the overall effect of roads on plant diversity at road edges was negligible due to the high spatial heterogeneity of road impacts, where positive and negative effects balanced each other. Given the minor effect of roads on plant diversity, its lack of attenuation with distance is not unexpected. In the plateau temperate zone, Simpson's diversity was higher at road edges, with impacts diminished farther away from roads. In contrast, similar patterns were absent in the plateau sub-frigid zone (Fig. S2). This result suggests that roads reduce the dominance of keystone species in roadside grasslands less in the plateau sub-frigid zone than in the plateau temperate zone. The plateau sub-frigid zone experiences a colder and harsher climate, where species are highly adapted to extreme environments (Sun et al., 2014). These dominant species pose a challenge for exotic species introduced by roads to colonize, highlighting the greater resilience of the plant diversity in the plateau sub-frigid zone to road disturbance (Hillebrand et al., 2008).

4.2. Road disturbance slightly altered plant species composition

Differences in species composition between roadside grasslands and those away from roads have been attributed to plant seed dispersal via roads (Von Der Lippe and Kowarik, 2007). Moreover, roads have been shown to alter grassland habitat (Gardiner et al., 2018), eliminating some pre-existing species that were unable to adapt to the new environment, whereas new species adapted to the changed conditions increased their competitive ability. Specifically, we found that the predominance of grasses increased, whereas that of sedges and legumes decreased on the roadsides (Fig. S4), which may be strongly related to nitrogen from vehicle emissions (Shen et al., 2024). Our finding that road disturbance did not greatly affect plant species composition is consistent with previous studies (e.g., Wang et al., 2023a). This stability is attributed to effective ecological conservation measures employed during road construction, which minimized disturbance to vegetation and facilitated its restoration (Cheng, 2009). More importantly, exotic species are poorly adapted to the harsh environment of the Tibetan Plateau, making it difficult for them to colonize this region (Pauchard et al., 2009, 2016). This finding emphasizes the stability of community structure and resistance to disturbance in Tibetan Plateau grassland plant diversity.

The dissimilarity was partitioned into true species turnover (species replacement) and nestedness (species increase or decrease), with the former dominating. Evidence suggests that the entry of new species or the exit of pre-existing species rarely occurred independently since most sites exhibited minimal changes in species richness (Fig. S5). Instead, species turnover was a much more common pattern (typical at S02, S03, S08, S16, S24, etc.), dominating the variation in species composition. Nestedness played an essential role in species dissimilarity only at individual sites (Fig. S5), where considerable changes in species number occurred (e.g., S4 and S17). Our results confirm previous reports that species dissimilarity between roadside and nearby habitats is mainly due to species turnover rather than nestedness (Vanneste et al., 2020; Dániel-Ferreira et al., 2023), emphasizing that limited species turnover does not lead to significant loss of species diversity.

Our analysis revealed a significant correlation between distance from the road and variations in species composition. Sørensen's and Simpson's dissimilarity decreased as distance from the road edge increased (see Fig. S6a, b), indicating that roads contribute to species variation (especially true spatial turnover). This finding aligns with previous research (Guo et al., 2007), and likely arises from the constrained dispersal and colonization ability of plant propagules facilitated by roads (Christen and Matlack, 2006). The negligible change in nestedness can be attributed to the lack of substantial fluctuation in plant richness across the distance gradient from the road (Fig. 2b).

In contrast to alpine meadows, alpine steppes exhibited remarkable species composition similarity between areas adjacent to and distant from roads (Fig. S6c, d), indicating superior resistance and stability of community composition in alpine steppes in response to road disturbance. The lower species turnover observed in alpine steppes can be primarily attributed to two factors. First, the harsh environment of the Tibetan Plateau favors native species with high resistance to external pressures; conversely, exotic species originating from regions with mild climates are poorly adapted to the environmental stresses of this region (Alpert et al., 2000). For example, previous research has demonstrated that alpine steppes, which are drier than alpine meadows (Wang et al., 2022), harbor species more resistant to adverse conditions (Zhang et al., 2019). Hence, in alpine steppes, exotic species face more severe competition challenges against native species (Pauchard et al., 2009, 2016). Second, the lower species diversity in alpine steppes (Fig. S7) implies that fewer non-adapted species are eliminated due to road disturbance.

4.3. Implications for ecological security at high altitudes

This study evaluated the impacts of road disturbance on the richness, evenness, and species composition of alpine grassland plants on the Tibetan Plateau through field investigations at 25 sites (Fig. 4). Our results indicate that roads have a weak positive effect on Pielou's evenness, suggesting enhanced adaptation and resilience to climate change and anthropogenic disturbance in alpine grasslands on the Tibetan Plateau (Cole et al., 2019; Bai et al., 2022). Conversely, species richness, Simpson's diversity, and Shannon–Wiener's diversity showed no significant response to roads, highlighting the robust resistance of alpine grassland plant diversity to road disturbance. Grassland community structure was found to be more stable in the colder plateau sub-frigid zones compared to the plateau temperate zones, with plant diversity exhibiting lower susceptibility to road disturbances.

Our analysis of the dissimilarity of species composition between roadside and off-road grasslands revealed a limited impact of roads on species composition, indicating stability in the species composition of alpine grasslands on the Tibetan Plateau. Nestedness (species loss or gain) contributed insignificantly to species alteration, primarily influenced by the spatial turnover of species (species replacement). Due to the more severe environment, alpine steppes exhibited fewer road-induced species changes than alpine meadows.

Our findings indicate that current road construction projects pose a minor risk to the biodiversity of Tibetan Plateau grasslands, attributable to vegetation protection and restoration measures implemented during road construction (Zhou et al., 2008) and the competitive advantage of resistant native species. Our conclusions can be cautiously generalized to high-altitude regions globally, which share similar topography, climate, and resistant native species. Notably, although cold climates at high altitudes have traditionally hindered the establishment of exotic species, these species are more likely to colonize such areas in a warming future. Despite the limited ecological risk of road disturbance, climate change may destabilize plant community structure, increasing the risk of invasive alien species (Pyšek et al., 2011; Petitpierre et al., 2016). This risk should be considered in future road development plans, which have significant implications for the ecological security of global high-altitude regions, as exemplified by the Tibetan Plateau.

4.4. Limitations

This study investigated the overall impacts of road disturbance on grassland plant diversity on the Tibetan Plateau through a large-scale field survey. However, several limitations in terms of methodology and data should be acknowledged.

Firstly, due to limited funding, our sampling was conducted across a broad environmental gradient at the expense of replication at individual sites. While gradient designs are superior for monitoring and quantifying complex ecosystem responses (Kreyling et al., 2018), the absence of replicate sampling introduces some uncertainty to our results. In future studies, trade-offs should be considered between replicate and gradient designs to improve the reliability of findings.

Secondly, our sampling sites included alpine steppes and meadows, encompassing several key eco-geographic regions (Fig. S8). We did not conduct field surveys in the forest-dominated southeastern Tibetan Plateau or in the northwestern Tibetan Plateau, where road disturbance is minimal, as these areas did not meet our sampling criteria. Despite the Qilian mountains in eastern Qinghai meeting our criteria, no surveys were conducted in this region. We emphasize the necessity for further research there to mitigate sampling bias, thereby advancing our understanding of grassland responses to road disturbances on the Tibetan Plateau.

Thirdly, this study did not consider road age, a significant factor influencing plant species composition (Pourrezaei et al., 2020). Plant diversity is reported to increase with road age and to plateau 20 years after road construction (Zeng et al., 2011). Because most sample sites in this study were on roads older than 20 years of age (Appendix A), road age introduces minimal uncertainty to our results. Moreover, we did not include traffic volumes in our analysis but substituted road class instead. Although road class is strongly correlated with traffic volume (Fitch and Vaidya, 2021), traffic volume, a continuous variable, would be more effective in predicting road impacts. Road types (fenced vs. non-fenced, or asphalt vs. dirt) may also affect plant diversity responses, but this was not examined in this study due to data accessibility constraints. Therefore, to improve our understanding of road impacts, we recommend that future studies incorporate predictors such as traffic volume and road type.

Finally, this study did not consider the effects of grazing, an important disturbance of grasslands on the Tibetan Plateau, which may confound road impacts. However, the effects of grazing on plants within 100 m of the road were similar to those of controls (Gelbard and Harrison, 2003; O'Farrell and Milton, 2006; Inger et al., 2011). Thus, by calculating the relative changes in plant diversity near the roads (lnRR = ln(X/X_CK)) compared to the control (Hedges et al., 1999), we eliminated potential confusion by grazing impacts. Note that this analysis does not account for potential interactions between road and grazing impacts, which introduces some uncertainty to our results. We recommend integrating the roles of roads and grazing disturbance in future studies to allow for a more comprehensive picture of human activities on grassland biodiversity on the Tibetan Plateau.

Acknowledgments

This work was funded by the Second Tibetan Plateau Scientific Expedition and Research Program (Grant No. 2019QZKK0403).

Data availability statement

The data supporting this study's findings are available from the corresponding author upon reasonable request.

CRediT authorship contribution statement

Ziwei Chen: Writing – original draft, Methodology, Investigation, Formal analysis, Conceptualization. Dongsheng Zhao: Writing – original draft, Supervision, Investigation, Conceptualization. Siqi Deng: Writing – original draft, Visualization, Investigation. Yu Zhu: Investigation. Ke Wang: Investigation. Shunsheng Wang: Investigation. Du Zheng: Supervision.

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.2024.09.006.