Resource–Disturbance Trade-Offs Regulate Grassland Plant Diversity Across Experimental and Model Systems

Abstract

Disentangling the joint effects of resource availability and disturbance on plant diversity is fundamental to understanding community assembly. We developed a stochastic extension of the Lotka–Volterra model that explicitly incorporates resource facilitation and disturbance-induced mortality, both mediated by species-specific trait responses. Combining simulations with a long-term field experiment manipulating nitrogen addition and mowing, we show that mowing consistently increased species diversity, whereas nitrogen addition reduced it, with no significant interaction between the two factors. Notably, mowing increased evenness, suggesting that higher diversity can coincide with more even abundance distributions. Simulations reproduced these patterns and revealed a non-linear resource–disturbance relationship: diversity declined under high-resource, low-disturbance conditions but was maintained at intermediate disturbance and moderate-to-low resource levels. This pattern was further supported by shifts in evenness and dominance across environmental gradients. Our results demonstrate that plant diversity emerges from a balance between resource-driven competitive exclusion and disturbance-mediated coexistence, modulated by species-specific traits.

1. Introduction

The mechanisms that maintain species diversity have always been one of the central topics in ecological research [1]. In community ecology, the Lotka–Volterra competition model serves as a fundamental theoretical framework for describing interactions between species [2,3], characterizing the dynamics of multi-species populations and their coexistence conditions through a series of differential equations [4,5]. The classical Lotka–Volterra model assumes that population growth rates are linearly regulated by intraspecific and interspecific competition, and its equilibrium solutions reveal the theoretical basis for the competitive exclusion principle and niche differentiation [6]. A growing body of empirical research has tested the applicability of this framework in real ecological systems by parameterizing or validating Lotka–Volterra dynamics with field and experimental data. For example, grassland competition experiments have directly estimated species interaction coefficients and shown that empirically derived parameters can reproduce coexistence patterns, while also revealing strong temporal variability in competitive strength across seasons [7]. Extensions incorporating non-linear density dependence, such as crowding effects, have further been supported by empirical observations that high-density mortality enhances intraspecific regulation and promotes multispecies coexistence [8,9].

These studies illustrate that while the classical Lotka–Volterra model provides a foundational theoretical framework, empirical evidence increasingly supports the need to incorporate environmental variability, non-linear interactions, and indirect effects to better explain species coexistence in natural ecosystems [10,11,12]. At present, there is no dynamic model that simultaneously incorporates resource limitation factors and environmental disturbance factors to simulate the long-term diversity changes in actual grassland communities [7,9,13,14]. Based on the classical Lotka–Volterra model, this study constructs a stochastic population dynamics model incorporating both resource facilitation effects and environmental disturbance effects [15]. To investigate the role of the aforementioned mechanisms in shaping species diversity, this study designed a large-scale simulation experiment: 100 plots with randomly generated initial species compositions were created, each with no more than 35 species initially. The model was iterated for 5000 iterations.

The core innovation of the model lies in the introduction of two key parameters: the population growth term $ep·e^{{\eta }_i}$ driven by resource abundance, and the population reduction term $ek·e^{{\theta }_i}$ caused by environmental disturbance. Here, ep and ek represent the levels of resource availability and environmental stress intensity, respectively, while ${\eta }_i$ and ${\theta }_i$ are species-specific trait parameters designed to characterize different species differential responses to resource fluctuations and environmental perturbations. This setup allows the model to simultaneously simulate the combined effects of positive facilitation (such as nutrient enrichment and interspecific facilitation) and negative disturbance (such as extreme climate and anthropogenic disturbances) on population dynamics, thereby better capturing the complexity of real ecosystems [7,15].

Afterward, the diversity indices of each plot were calculated, including species richness, Shannon–Wiener diversity index, Berger–Parker dominance index, and Pielou’s evenness index [16,17,18].

In addition, this study utilized controlled experiments simulating long-term nitrogen addition and mowing to validate the patterns predicted by the model [12,19]. By analyzing the changes in biodiversity under combinations of different resource levels and disturbance intensities, it aims to reveal how the trade-off between resource facilitation and environmental disturbance influences the process of community assembly [20]. The research results are expected to provide a new theoretical perspective for understanding the mechanisms that maintain biodiversity in natural ecosystems, and to offer simulation-based guidance for practical applications such as grassland management and biodiversity conservation.

2. Materials and Methods

2.1. Overview of the Research Area

The Ningxia Yunwu Mountain National Grassland Nature Reserve is located in the northeastern part of Guyuan City, Ningxia Hui Autonomous Region, between 106°21′–106°27′ east longitude and 36°10′–36°17′ north latitude (Figure 1). It lies within the temperate semi-arid climatic zone, characterized by a dry climate with scarce and concentrated rainfall. The average annual temperature is 5 °C, and the average annual precipitation is 445 mm [21].

2.2. Nitrogen Addition and Mowing Experiments

Since 2013, a randomized block design has been used, with a total of 108 plots measuring 5 m × 10 m, divided into 6 blocks. Each block contains different gradients of nitrogen enrichment and mowing treatments. To reduce edge effects, a 2 m wide buffer zone was set between all plots. The nitrogen enrichment treatment used urea as the nitrogen source, with 6 gradients set as 0, 5, 10, 20, 40, and 80 g m⁻² yr⁻¹ of urea, equivalent to 0, 2.34, 4.67, 9.34, 18.68, and 37.35 g N m⁻² yr⁻¹, applied uniformly in late April every year. The mowing treatment was set at 3 gradients of 0, 1, and 2 cuts every year, removing the aboveground parts of the vegetation.

2.3. Sample Plot Survey

This study conducted plot surveys in August 2024 and 2025 by randomly selecting standardized 1 m × 1 m grassland plots. We recorded the species names of all plants appearing within the plots, visually estimated the coverage of each species within the plot, measured the maximum natural height of the plants using a tape measure, and counted the aboveground ramet density to represent abundance.

2.4. Model Simulation

This study is based on the discrete-time Lotka–Volterra competition model and constructs a stochastic population dynamics model, including resource facilitation and environmental disturbance effects, which is used to simulate multi-species population dynamics in plant communities [7]. The model expression is as follows:

$$N_{i,t+1}=r_i{N}_{i,t}\left(1+{\displaystyle \sum _{j=1}^S}\left({\alpha }_{ij}{N}_{j,t}\right)+\mathrm{e}\mathrm{p}·e^{{\eta }_i}-\mathrm{e}\mathrm{k}·e^{{\theta }_i}\right)+N_{i,t}$$

(1)

Here, $N_{i,t}$ denotes the population size of species i at time t (generations), where t = 0, 1, …, T, represents the number of discrete generations; $r_i$ is the intrinsic growth rate of species i, reflecting the potential growth ability of the species under conditions without competition and disturbance [22]; and ${\alpha }_{ij}$ is the competition coefficient of species j against species i. The summation term characterizes the total competitive pressure on species i from all species within the community (including itself), where S represents the total number of species in the community. The term $ep·e^{{\eta }_i}$ represents the promoting effect of abundant resources on population growth, where ep is the coefficient of global resource availability, and ${\eta }_i$ represents the functional trait of species’ strategy and is used to simulate the differential responses of different species to resource fluctuations; this term is always non-negative, and the promoting effect is ensured not to become negative by using an exponential form. The term $ek·e^{{\theta }_i}$ denotes the inhibitory effect of environmental disturbance on population reduction, where ek is the global environmental stress coefficient, and ${\theta }_i$ is also a species-specific functional trait that is used to simulate the differential responses of different species to the intensity of disturbance. This non-negative term ensures the non-negativity of the disturbance effect through an exponential form.

In the model, resource facilitation and environmental disturbance are linearly added to the competition term, jointly regulating the population growth rate. To investigate the effects of resource facilitation and environmental disturbance on species diversity, this study randomly generated 100 virtual plots. The initial number of species in each plot was uniformly randomly selected between 3 and 35, and the initial population size of each species was 1. The model iterated for 10 generations, with 5000 time-step iterations. Each time-step iteration simulates population changes over 1 week, and every 50 iterations represent 1 year; that is, one generation.

2.5. Species Diversity Index

This study uses the relative abundance p of populations to calculate species diversity indices: Simpson diversity index, Shannon–Wiener diversity index, Pielou’s evenness index [23], and Berger–Parker dominance index [24]. These four indices are calculated as follows:

$$L=1-{\sum }_{i=1}^S{p}_i^2$$

(2)

$$H=-{\displaystyle \sum _{i=1}^s}{p}_i·\mathit{ln}{p}_i$$

(3)

$$J={\displaystyle \frac{H}{\ln s}}={\displaystyle \frac{-{\displaystyle {\sum }_{i=1}^s}{p}_i·\ln p_i}{\ln s}}$$

(4)

$$D={\displaystyle \frac{max\left\{n_1\left.,\dots ,n_i,\dots ,n_S\right\}\right.}{{\displaystyle {\sum }_{i=1}^s}{n}_i}}$$

(5)

where L represents the Simpson diversity index, H represents the Shannon–Wiener index, J represents Pielou’s index, D represents the Berger–Parker dominance index, p_i denotes the importance value of the i-th species in the community, ni represents the number of individuals in the i-th population in the community, and S is the number of populations in the community.

2.6. Statistical Test

In this study, for Shannon–Wiener diversity, which meets the assumptions of normal distribution and homogeneity of variance, we used two-way analysis of variance (ANOVA) to test the effects of nitrogen addition, mowing, and their interaction. For Simpson diversity, Pielou’s evenness, and Berger–Parker dominance, which do not meet the assumption of normal distribution, we used the Scheirer–Ray–Hare rank test to analyze the effects of nitrogen addition, mowing, and their interaction. All models and analyses were performed using R version 4.4.3.

3. Results

3.1. Resource Availability and Disturbance Intensity Change Coexistence Patterns

3.1.1. Resource Availability and Disturbance Intensity Affect the Development Process of Community Diversity

Simulation results show that resource availability and disturbance intensity together determine the dynamics of community diversity, there is a potential interaction effect between resources and disturbance (Figure 2).

At low disturbance intensity, in communities with low resource availability, species diversity decreases slowly with the shortening of generational time, whereas in communities with high resource availability, diversity declines rapidly, and in stable generations, diversity is even lower. Under moderate disturbance, species diversity in low resource availability communities drops sharply, while in communities with medium and high resource availability, diversity, after an initial decline, stabilizes quickly and shows a potential upward trend. Under high disturbance, species diversity in low resource availability communities decreases sharply, whereas communities with high and medium resource availability maintain good diversity. Overall, under moderate disturbance intensity, in scenarios with medium to low resource availability, the average final diversity maintained by simulated communities is actually higher.

3.1.2. Resource Availability and Disturbance Intensity Affect the Final Evenness of the Community

Heatmap analysis revealed the response patterns of the Pielou’s evenness index to different environmental factors of the simulated community (Figure 3). The index range is relatively narrow (0.17–0.23), and the overall uniformity is low. Overall, under conditions of moderate resources and medium to high disturbance, Pielou’s evenness index of the simulated community achieved the maximum value (>0.23), and the species distribution of the community was more uniform. However, under conditions of low resources and high disturbance, Pielou’s evenness index of the simulated communities was significantly lower, showing an abnormal diversity distribution pattern (<0.18), and overall Pielou’s evenness was higher under the condition of high resources.

3.1.3. Resource Availability and Disturbance Intensity Affect the Dominance of Dominant Species

The interaction effects of disturbance intensity and resource abundance on the Berger–Parker index (representing the relative abundance of the most dominant species in the community) were analyzed using a heatmap (Figure 4). The results showed that the index values ranged from 0.27 to 0.36, indicating that within the set parameter range, the communities were always dominated by a few dominant species, but the degree of dominance varied significantly.

The overall trend indicated that the Berger–Parker index peaked under the combination of high resources and low disturbance (maximum 0.36), suggesting that under these conditions, the dominance of a single species was most pronounced. Specifically, in regions with higher resource abundance (Y-axis > 5.0), dominance exhibited a non-linear pattern of initially decreasing and then increasing with increasing disturbance intensity: for example, at a resource abundance of 7.5, the index decreased from 0.36 under low disturbance to 0.28 at medium disturbance (5.0) but rose again to 0.33 under high disturbance (7.5). In contrast, in resource-poor regions (Y-axis < 5.0), dominance was generally lower (mostly between 0.30 and 0.31) and responded weakly to disturbance intensity, showing mild changes.

3.2. Nitrogen Addition and Mowing Significantly Affect Species Diversity

The results of the variance analysis indicate that, in the 2024 plots, fertilization had a highly significant effect on the Shannon–Wiener diversity index (F_5,102 = 9.61, p = 0.002), and mowing also had a highly significant effect on the Shannon–Wiener diversity index (F_2,104 = 124.70, p < 0.001), but the interaction between fertilization and mowing was not significant (F_17,89 = 0.33, p = 0.57) (Figure 5a). In the 2025 plots, fertilization had a highly significant effect on the Shannon–Wiener diversity index (F_5,102 = 11.79, p < 0.001), mowing also had a highly significant effect on the Shannon–Wiener diversity index (F_2,104 = 51.64, p < 0.001), but again the interaction between fertilization and mowing was not significant (F_17,89 = 0.814, p = 0.37) (Figure 5b).

The results of the rank test indicate that, in the 2024 plots, fertilization had a highly significant effect on the Simpson diversity index (H = 8.497, p = 0.131), and mowing also had a highly significant effect on the Simpson diversity index (H = 25.184, p < 0.001), but the interaction between fertilization and mowing was not significant (H = 1.881, p = 0.997) (Figure 5c). In the 2025 plots, fertilization did not have a significant effect on the Simpson diversity index (H = 9.828, p = 0.080), while mowing had a highly significant effect on the Simpson diversity index (H = 27.361, p < 0.001), but the interaction between fertilization and mowing was not significant (H = 5.428, p = 0.861) (Figure 5d).

Overall, the Shannon–Wiener diversity index and Simpson diversity index increased with the frequency of mowing and decreased with the level of nitrogen fertilization. The interaction effect of mowing and nitrogen fertilization level on the species diversity indices was not significant, while mowing frequency was a more important factor affecting species diversity.

3.3. Nitrogen Addition and Mowing Significantly Affect Community Structure

3.3.1. Nitrogen Addition and Mowing Significantly Affect Community Evenness Index

The results of the rank test indicate that, in the 2024 plots, fertilization had no significant effect on the Pielou’s evenness index (H = 5.741, p = 0.332), while mowing had a highly significant effect on the Pielou’s evenness index (H = 44.376, p < 0.001), but the interaction between fertilization and mowing was not significant (H = 5.911, p = 0.823) (Figure 6a). In the 2025 plots, fertilization had no significant effect on the Pielou’s evenness index (H = 7.536, p = 0.184), while mowing had a highly significant effect on the Pielou’s evenness index (H = 9.887, p = 0.007), but the interaction between fertilization and mowing was not significant (H = 6.067, p = 0.81) (Figure 6b).

Overall, Pielou’s evenness index significantly decreased with increasing mowing frequency and showed an increasing trend with higher nitrogen fertilization levels. The interaction effect of mowing and nitrogen fertilization levels on Pielou’s evenness index was not significant, whereas mowing frequency was a more important factor affecting Pielou’s evenness index.

3.3.2. Nitrogen Addition and Mowing Significantly Affect Community Dominance Index

The results of the rank test indicate that, in the 2024 plots, fertilization had no significant effect on the Berger–Parker dominance index (H = 4.7, p = 0.453), while mowing had a highly significant effect on the Berger–Parker dominance index (H = 13.511, p = 0.002), but the interaction between fertilization and mowing was not significant (H = 9.841, p = 0.455) (Figure 6c). In the 2025 plots, fertilization had no significant effect on the Berger–Parker dominance index (H = 8.982, p = 0.11), while mowing had a significant effect on the Berger–Parker dominance index (H = 6.674, p = 0.035), but the interaction between fertilization and mowing was not significant (H = 8.611, p = 0.569) (Figure 6d).

Overall, the Berger–Parker dominance index tended to increase with mowing frequency and decrease with increasing nitrogen fertilization levels. The interaction effect between mowing and nitrogen fertilization had no significant impact on the Berger–Parker dominance index, and mowing frequency was a more important factor affecting the Berger–Parker dominance index.

4. Discussion

4.1. The Dominant Role of Mowing-Induced Disturbance and Its Determinative Effect on Community Diversity

This study indicates that mowing (disturbance) is the primary driver affecting community diversity, with explanatory power significantly higher than nitrogen addition (resource enhancement) [12,25]. In data from both years, mowing had highly significant effects on the Shannon–Wiener index, Simpson index, and Berger–Parker index, although interactions with fertilization were not significant, indicating that the disturbance effect has strong independence in this system [26]. Periodic mowing removes aboveground biomass, primarily weakening the light acquisition ability of tall dominant species, thereby reducing their monopoly over light resources while alleviating asymmetric competition pressure. This process can be understood as a dynamic adjustment of interspecific competition coefficients or the competition terms in the Lotka–Volterra model, allowing originally disadvantaged mid- and low-stature species a window for recovery and expansion, thereby enhancing overall species coexistence [27].

This study further shows that mowing not only increases species diversity but also enhances Pielou’s evenness index of the community, suggesting that disturbance does not simply ‘equalize’ the community, but may reshape abundance distribution patterns by promoting rapid responses of some opportunistic species [12,28]. This result indicates that although disturbances maintain species coexistence, they may also reinforce internal dynamic imbalances within the community, thereby achieving community equilibrium and shifts in community structure [29,30,31]. This phenomenon is significant in grassland ecosystems, indicating that traditional assessments of community stability based solely on species richness or the Shannon–Wiener index may underestimate the complex structural changes within the community [32,33].

4.2. The Effect of Intensified Competition from Resource Enrichment and Its Mechanism for Inhibiting Diversity

In contrast to the disturbance effect, nitrogen addition generally reduced community diversity, especially showing a significant decrease in the Shannon–Wiener index, indicating that resource enrichment primarily drives diversity decline by enhancing interspecific competition [34,35]. This result is consistent with the resource competition theory and the classical conclusion that “eutrophication leads to dominance monopoly”, where enrichment with excessive amounts of nutrients is commonly seen as one of the main causes of diversity loss in terrestrial plant communities—as nutrient addition shifts the species interactions from competition for nutrients to competition for light [36,37], meaning that increased resources amplify the growth advantage of dominant species and accelerate competitive exclusion [11,38,39].

It is worth noting that the effect of nitrogen addition on evenness was either insignificant or showed a slight upward trend in different years. This may reflect that increased resources partially alleviate resource limitations for some species, allowing moderately competitive species to persist and thereby locally enhance evenness. However, this ‘increase in evenness’ did not translate into an overall increase in diversity; instead, it was accompanied by a simplification of species richness and dominance structure. This result highlights that the impact of resource enrichment on communities is multidimensionally inconsistent, and its ecological consequences need to be evaluated comprehensively using multiple diversity metrics [40].

4.3. The Consistent Results from Experiments and Model Simulations Indicate: A Diversity Maintenance Mechanism Driven by Resource–Disturbance Trade-Offs

The traditional Lotka–Volterra model is often limited to the interspecific relationship parameters between predators and prey, and lacks research on the effects of resource input and environmental disturbances on community structure. Based on an extended Lotka–Volterra stochastic model, this study reveals how the trade-off between resource facilitation ($ep·e^{{\eta }_i}$) and disturbance inhibition ($ek·e^{{\theta }_i}$) jointly shapes community diversity patterns. Simulation results show that under moderate disturbance and medium-low resource conditions, communities can maintain higher stable diversity [31,41], whereas under high resource and low disturbance conditions, diversity declines rapidly. This pattern is highly consistent with the empirical rule observed in experiments that fertilization reduces diversity while mowing increases diversity [19]. Heatmap analysis reveals that Pielou’s evenness peaks under a combination of medium resources and medium disturbance, while the Berger–Parker dominance is highest under high resource and low disturbance conditions [42,43]. This result mechanistically supports the Intermediate Disturbance Hypothesis [44] but simultaneously extends its meaning, as optimal diversity depends not only on disturbance intensity but also on the joint regulation by resource availability and species trait strategies [45,46].

Therefore, the results from the models and experiments consistently indicate that the maintenance of diversity is not always driven by a single decisive factor [47,48], instead, the dynamic changes in diversity within a community are jointly influenced by the complex trade-offs of ‘resources–disturbance–plant traits’(The coefficients of variation of the diversity index, evenness index, and dominance index of the simulated communities are driven by the ‘resource-disturbance’ factor, as shown in Figures S1–S4). This framework provides a new theoretical perspective for understanding the formation of diversity patterns in natural ecosystems and offers a quantitative basis for grassland management practices, such as the rational combination of mowing and fertilization.

4.4. Limitations and Future Research Directions

Although the experimental results of this study are highly consistent with model predictions, and the multi-species model constructed based on the assumptions of non-linear resource limitations and non-linear environmental disturbances possesses certain innovativeness, several limitations remain. First, the experiments were conducted only on a single grassland type, and the extrapolation of the research conclusions needs to be further validated across multiple types of grassland ecosystems [49]. Second, the model primarily focuses on diversity patterns after the stable development of communities, and its characterization of long-term community succession and stable-state transitions remains insufficient [35,50]. Additionally, the current analysis has not yet deeply explored the relative contributions of specific species to maintaining diversity [6], nor does the model incorporate spatial heterogeneity or complex ecological interaction networks among species (such as mutualism and cascade effects) [8,51]. Future research should expand to grassland ecosystems in different climatic zones, extend observation periods, and integrate species functional traits with spatially explicit models to more comprehensively reveal the mechanisms by which resource–disturbance–trait trade-offs affect community diversity dynamics.

5. Conclusions

This study systematically reveals the differentiated driving mechanisms of community diversity under mowing (disturbance) and nitrogen addition (resource enrichment). Mowing significantly increases species diversity, richness, and evenness by suppressing competitive exclusion and its effect is independent of fertilization treatment, supporting the classic theory that disturbance promotes coexistence. In contrast, although nitrogen addition has slightly increased evenness in certain years, it overall intensified the dominance of dominant species, leading to a significant decline in the Shannon–Wiener index, validating the mainstream hypothesis that resource enrichment weakens diversity through competitive exclusion. Combined with an extended Lotka–Volterra model and heatmap analysis, it further shows that diversity is jointly shaped by the complex trade-offs among ‘resources–disturbance–traits’, reaching an optimum under moderate disturbance and low to moderate resource conditions. This framework not only deepens the understanding of community assembly mechanisms but also provides a theoretical basis for the adaptive management of grassland ecosystems.