1. Introduction

Aboveground vegetation and soil seed bank are two distinct but interdependent components of the plant community, and their interactions are crucial for community succession (Török et al., 2018; Angert et al., 2009; Ma et al., 2021). Grasslands, with rapid species turnover rates and abundant seed production, typically exhibit stronger compositional coupling between aboveground vegetation and soil seed bank than other ecosystems (Hopfensperger, 2007). Moreover, such coupling is sensitive to anthropogenic and natural disturbances, with implications for community succession and ecosystem multifunctionality (Kiss et al., 2018). Understanding changes in the compositional relationship between aboveground vegetation and soil seed bank provides essential insights into enhancing grassland resilience in a changing world.

The compositional coupling or decoupling between aboveground vegetation and the soil seed bank is driven by abundance gradient and balanced variation in abundance (Baselga, 2013; Zhao et al., 2023). Abundance gradient indicates that some individuals are absent from one community relative to the others, while balanced variation in abundance refers to the individuals of some species in one community are replaced by the same number of individuals of other species in another community across different habitats and environmental changes (Baselga, 2013). For example, disturbance from N deposition may reduce the aboveground plant diversity, but may not affect the soil seed bank (Zhang et al., 2019; Bahn et al., 2022; Chen et al., 2025). This can exacerbate compositional dissimilarity between aboveground vegetation and the soil seed bank by increasing the abundance gradient. In contrast, other disturbances (e.g., mowing) may increase species diversity or abundance in aboveground vegetation and contribute additional seeds to the soil seed bank, thereby decreasing the compositional dissimilarity by reducing balanced variation in abundance (Miao et al., 2020). Nonetheless, the relative contributions of these components and processes governing community assembly to the compositional coupling between aboveground vegetation and the soil seed bank remain poorly understood.

While increased N availability is known to threaten plant diversity and alter ecosystem multifunctionality in global grasslands (Vitousek et al., 1997; Stevens et al., 2004; Simkin et al., 2016), its effects on the soil seed bank remain variable among studies. N deposition may deplete soil seed banks by reducing seed production and viability (Klaus et al., 2018; Ma et al., 2025), or it may have neutral effects due to the strong buffering capacity of soil seed banks (Zhang et al., 2019; Hábenczyus et al., 2024). These differential responses of aboveground vegetation and the soil seed bank to N deposition may alter their compositional coupling. For example, Zhao et al. (2023) reported that N inputs induced compositional decoupling between aboveground vegetation and the soil seed bank in a natural temperate steppe. However, whether such impacts of N deposition on the compositional coupling between vegetation and the soil seed bank are mediated by grassland management practices remains an open question.

Mowing is one of the most widespread management strategies across global grasslands (Molina et al., 2021), which can mitigate the adverse effects of N deposition on aboveground plant diversity (Yang et al., 2019; Bahn et al., 2022). Moreover, mowing alters the spatial heterogeneity of soil resources and vegetation structure, two key drivers of seed bank composition. On one hand, mowing enhances seed diversity and abundance by increasing light availability and creating favorable microsites for seed input (Miao et al., 2020). On the other hand, it has negative impacts by removing reproductive structures before seed maturation (Hábenczyus et al., 2024). Consequently, the compositional coupling between aboveground vegetation and the soil seed bank might be largely context-dependent. While a few studies have examined how mowing mediates the effects of N deposition on the soil seed bank in grasslands (Klaus et al., 2018; Hábenczyus et al., 2024), their interactive effects on the compositional coupling between aboveground vegetation and the soil seed bank remain unknown.

The different responses of plant functional groups with distinct strategies would provide a mechanistic explanation for changes in the compositional coupling between the soil seed bank and aboveground vegetation following disturbances (Tognetti et al., 2021). Dominant graminoids benefit more from N deposition due to their stronger tolerance to N-induced acidification and stronger competitive ability (Yang et al., 2019), whereas subordinate graminoids and forbs are often suppressed. Both subordinate graminoids and forbs would be favored by mowing, as it could substantially reduce the dominance of high-stature species (Zhang et al., 2025). However, such land use-induced compositional alterations in the aboveground vegetation might not be mirrored in the soil seed bank due to the interspecific variations in reproductive strategies. While forbs always deposit large amounts of seeds into the soil (Ma et al., 2025), most graminoids are clonal plants and characterized mainly by asexual reproduction (Ott et al., 2019). Therefore, quantifying the shifts in the abundance of different plant life-forms in both aboveground vegetation and the soil seed bank is essential to improve our mechanistic understanding of compositional decoupling induced by global change drivers and grassland management strategies.

To understand the compositional coupling of aboveground vegetation and the soil seed bank in response to N inputs and mowing, we examined the compositional responses of the aboveground plant community and soil seed bank to factorial N addition (5 g N m⁻² yr⁻¹) and mowing in a temperate grassland after 10 years of treatments. Specifically, two hypotheses were tested. (1) N addition would reduce the compositional coupling between aboveground vegetation and the soil seed bank by increasing abundance gradient, given that N addition often exerts stronger effects on the aboveground community composition than the soil seed bank (DeMalach et al., 2021). (2) Mowing might counteract the adverse effects of N addition and enhance the compositional coupling between above- and below-ground communities by reducing balanced variation in abundance, as mowing suppresses dominant graminoids and promotes subordinate species, thereby facilitating the incorporation of seeds into the soil seed bank (Miao et al., 2020).

2. Materials and methods 2.1. Study sites

Our experiment was carried out in an Inner Mongolian grassland dominated by C3 species (50°10′ N, 119°22′ E). Livestock grazing has been excluded by fencing since 2013. This region has a mean annual temperature of −2.4 ℃ and a mean annual precipitation of 360 mm (1970–2021). Nearly 80% of the total precipitation is concentrated from May to August. The soil type is chernozem according to the World Soil Information System Database.

2.2. Experimental design

An experiment with factorial N addition and mowing was established in May 2015, following a completely random design. Each of the four treatments, including control, N addition (N), mowing (M), and combined N addition and mowing (N + M), was replicated six times. In total, there were 24 plots with each size of 6 m × 6 m, with the neighboring plots being separated by 1 m aisles. N addition was annually applied with urea at a rate of 5 g N m⁻² year⁻¹ in mid-May. The fertilizer for each plot was first dissolved in 2 L of water and then sprayed onto the soil surface. To eliminate the confounding effects of water input, the plots without N addition received an equivalent amount of water. Annual mowing was conducted by a lawn mower in late August, with a stubble height of 10 cm. The mowing time and stubble height are designed to mimic the hay-harvest practice of local people.

2.3. Data collection

On April 1 of 2024, ten years after the experiment began, subject to annual treatment, we collected soil samples from the top 20 cm layer using a 5 cm diameter soil corer. Five soil cores from each plot were collected and then pooled into a composite sample. Each composite sample was placed in a labeled cloth bag and passed through a 2 mm sieve to remove roots, residuals, and rocks. The sieved samples were then air-dried at room temperature and stored for 20 days in cloth bags to prevent the intrusion of tiny airborne seeds.

The species richness and abundance of the soil seed bank were identified following the seedling emergence method in a greenhouse located in Shenyang (41°31′ N, 123°24′ E). Each sample was laid in a 2 cm layer over 5 cm of sterile and fine sand in 35 × 40 × 12 cm plastic trays. To detect any airborne seed contamination, we prepared 10 control trays containing only sterile sand. The moisture of the samples was maintained by watering daily (or as needed). Emerging seedlings were checked twice per week after one month of incubation, and were removed upon identification. Those unidentified species were transplanted into separate trays for further observation. We terminated the germination experiment after 20 consecutive days without the emergence of new seedlings (Eskelinen et al., 2023). The germination experiment lasted for seven months, from the beginning of May to the end of November 2024.

We investigated the aboveground plant community composition in two 20 cm × 20 cm quadrats within each plot on August 1, 2024. The quadrats were set at least 1 m away from the edges of the plot to minimize edge effects. Given the predominance of perennial graminoids in the community, we quantified species abundance by counting shoots (Te et al., 2025). Based on morphological and physiological characteristics, all plant species were classified into graminoids or forbs.

2.4. Statistical analysis

We converted the recorded shoot numbers and soil seed numbers into the abundance of aboveground vegetation and the soil seed bank per square meter. We used generalized linear mixed-effects models (GLMMs) to assess the effects of N addition and mowing on species richness and abundance of aboveground vegetation and the soil seed bank at both community and functional group levels. To evaluate the variations among the four treatments, we conducted the one-way ANOVA using GLMM (Y ~ treatment) with Gaussian distributions for species richness and a negative binomial distribution for abundance. The choice of distribution was guided by comparing Akaike Information Criterion (AIC) values among Gaussian, Poisson, and negative binomial models. GLMM models were used with the 'glmmTMB' package.

The compositional coupling between aboveground vegetation and the soil seed bank was evaluated by the abundance-weighted Bray–Curtis dissimilarity index, which accounts for both the presence-absence and abundance of different species. To determine the ecological processes associated with compositional coupling based on species abundance, we partitioned dissimilarity into abundance gradient and balanced variation in abundance (Baselga, 2013). The effects of N addition and mowing on compositional dissimilarity, abundance gradient, and balanced variation in abundance were assessed by the GLMM models with Gaussian distributions. Dissimilarity index was calculated with the 'vegdist' function in the 'vegan' package. Abundance gradient and balanced variation in abundance were calculated using the function 'beta.pair.abund' in the package 'betapart' (Baselga, 2013). All the statistical analyses were conducted with R 4.2.0 (R Core Team, 2022).

3. Results

N addition reduced the species richness of the total community and forbs, while mowing significantly increased the species richness of graminoids in the aboveground vegetation (p < 0.01; Fig. 1a–c). Neither N addition nor mowing affected species richness of the total community, graminoids, and forbs in the soil seed bank (Fig. 1d–e).

Mowing increased the total community and graminoid abundance, while N addition reduced the forb abundance in the aboveground vegetation (p < 0.001; Fig. 2a–c). Mowing increased the forb abundance in the soil seed bank, but both N addition and mowing had no impact on seed abundance of the total community or graminoids (p < 0.001; Fig. 2d–f).

N addition increased the abundance of the dominant graminoid, Leymus chinensis (p < 0.01; Fig. 3a), and mowing increased the subordinate graminoids, Koeleria macrantha and Cleistogenes squarrosa, in the aboveground vegetation (p < 0.01; Fig. 3a). Mowing and mowing with N addition increased the abundance of the forb, Artemisia scoparia, in the soil seed bank (p < 0.05; Fig. 3b). The most abundant species in the soil seed bank shifted from the graminoid, Carex duriuscula, in ambient and N addition plots to the forb, A. scoparia, in mowing and mowing with N addition plots (Fig. 3b).

N addition and mowing significantly interacted to alter the compositional dissimilarity between aboveground vegetation and the soil seed bank (p < 0.001; Fig. 4a) as well as abundance gradient (p < 0.01; Fig. 4b). N addition increased the compositional dissimilarity between aboveground vegetation and the soil seed bank in unmown plots but not in mown plots (p < 0.001; Fig. 4a). Mowing increased the compositional dissimilarity in mown plots (p < 0.001; Fig. 4a). Abundance gradient contributed to the increased dissimilarity in responses to N addition in unmown plots (p < 0.001; Fig. 4b), while balanced variation in abundance accounted for the increased dissimilarity induced by mowing (p < 0.001; Fig. 4c).

4. Discussion

We found that the impacts of N addition on compositional coupling between the aboveground vegetation and the soil seed bank depended significantly on whether the grasslands were mown or not. N addition reduced the compositional coupling only in unmown grasslands. Mowing, as a deterministic driver in community assembly, increased the compositional decoupling between aboveground vegetation and the soil seed bank. Our findings highlight the importance of the land-use background when assessing the impacts of N input on vegetation regeneration dynamics in grasslands.

Supporting our first hypothesis, N addition significantly increased compositional dissimilarity between aboveground vegetation and the soil seed bank in unmown grasslands by increasing the abundance gradient. This result implies that N addition had different impacts on species composition in aboveground vegetation and the soil seed bank. The strong sensitivity of aboveground plant composition to N addition has been extensively documented (Stevens et al., 2004; Suding et al., 2005; Clark and Tilman, 2008). Consistent with these results, we found that N addition reduced the richness and abundance of forbs but promoted graminoid dominance in the aboveground vegetation. For example, N addition reduced forb richness by half (Fig. 1c) and increased the relative abundance of the dominant grass, Leymus chinensis, from 36% to 59% in the aboveground vegetation (Fig. 3a). Such shifts are understandable as graminoids can outcompete forbs under elevated N availability through asymmetric competition for light and nutrients (Hautier et al., 2009; Borer et al., 2014; Chen et al., 2025). In contrast, we observed no alteration in the species richness and abundance of the soil seed bank after a decade of N addition. A similar pattern has been found in an alpine meadow of the Tibetan Plateau (Ma et al., 2025). Thus, N addition reduced the richness and abundance of forbs in the aboveground vegetation but not in the soil seed bank, thereby increasing their abundance gradient and intensifying the compositional decoupling between above- and below-ground communities.

Inconsistent with our expectation, mowing increased the compositional dissimilarity between aboveground vegetation and soil seed bank, which was largely caused by the increases in balanced variation in abundance. Mowing increased the richness and abundance of graminoids in the aboveground vegetation, while enhancing forb abundance in the soil seed bank. Such mismatched responses in functional groups jointly contributed to the compositional decoupling at the community level. Mowing facilitated the growth of subordinate graminoids, Koeleria macrantha and Cleistogenes squarrosa, in the aboveground vegetation. However, there were no corresponding changes in graminoids in the soil seed bank (Fig. 3). This is likely because graminoids rely primarily on the belowground bud bank instead of the soil seed bank for rapid regeneration after mowing (Ott et al., 2019). By contrast, mowing increased the relative contribution of forbs to the soil seed bank. For example, the forb, Artemisia scoparia, became dominant in the soil seed bank, replacing the dominance status of the graminoid, Carex duriuscula (Fig. 3). The prevalence of forbs in the soil seed bank in mown grasslands is likely linked to the dynamics of seed dispersal and establishment. Mowing strongly suppresses the growth of tall-statured graminoids (e.g., Leymus chinensis) and litter accumulation, thereby alleviating physical barriers to seed entry into the soil (Klimešová et al., 2023; Wang et al., 2024). Moreover, forbs typically produce smaller and numerous seeds than graminoids, facilitating wind dispersal and allowing prolonged dormancy in the soil seed bank (Klimešová et al., 2023). Therefore, mowing amplified the dominance of graminoids in the aboveground vegetation but forbs in the soil seed bank, thereby widening the functional divergence between above- and below-ground assemblies and increasing balanced variation in abundance. These changes intensify the decoupling between aboveground vegetation and the soil seed bank composition, which could ultimately influence vegetation succession and ecosystem stability (Hopfensperger, 2007).

N addition did not alter the compositional coupling between aboveground vegetation and the soil seed bank in the mown grassland. While a recent study from the unmown grassland reported that N addition decoupled the compositional association between aboveground vegetation and the soil seed bank (Zhao et al., 2023), our findings highlight that the impacts of N-inputs on such above- and below-ground compositional coupling vary with management context. Therefore, the N-induced compositional decoupling between aboveground vegetation and the soil seed bank might have been overestimated by the results from unmanaged grasslands. In addition, we found that the underlying mechanisms governing the coupling of aboveground vegetation and the soil seed bank in response to N addition differed between unmown and mown grasslands. For instance, abundance gradient accounted for the decoupling in unmown grasslands, whereas balanced variation in abundance played a critical role in mown grasslands. Such shifts in intrinsic mechanisms from unmown to mown grasslands were driven by the effects of mowing (Fig. 4). These results are in line with the widely reported variations in the effects of N addition on aboveground community composition between mown and unmown grasslands (Hautier et al., 2009; Zhang et al., 2017). Mowing can reverse the N-induced species loss by increasing light availability and alleviating competitive exclusion from dominant graminoids (Borer et al., 2014; Yang et al., 2015). In our study, N addition reduced the aboveground plant diversity in unmown grasslands, whereas this N effect was absent in mown grasslands, indicating that mowing diminished the N-induced species differences between aboveground vegetation and soil seed bank to some extent (Fig. 1a). Given the fact that many grasslands are subjected to different utilization practices globally (Tälle et al., 2016; Chen et al., 2025), our results suggest that local management practices may exert strong influence on the compositional coupling between above- and below-ground communities. These results underscore the necessity of accounting for land-use management effects (i.e., mowing, grazing, and burning) when evaluating the impact of N deposition on grassland ecosystem dynamics.

Although our study provides crucial evidence for the context dependency of N addition impacts on the compositional coupling between aboveground vegetation and the soil seed bank, some limitations should be acknowledged. Given the strong species-specific differences in seed dormancy and germination strategies, the species composition identified based on the greenhouse seedling-emergence approach may underestimate the abundance and diversity of soil seed banks (Ma et al., 2021). In addition, although we conducted seed sampling in the most representative early spring, when seeds have already undergone natural filtering processes (e.g., extreme cold and predation), one-time sampling may not provide complete information because the dynamics of soil seed banks vary among seasons. Therefore, we suggest that researchers collect soil seed samples across multiple seasons and integrate different germination methods to further reveal the temporal dynamics of the compositional coupling between aboveground vegetation and the soil seed bank.

In conclusion, N addition decoupled the compositional association between the soil seed bank and aboveground vegetation by increasing abundance gradient in unmown grasslands. In mown grasslands, however, N addition did not alter the coupling between the soil seed bank and aboveground vegetation. Given the increasing global trend of grassland use for production with substantial annual biomass removal (Bengtsson et al., 2019; Bardgett et al., 2021), our findings highlight the importance of considering ecosystem management practices when assessing the impacts of global change drivers on community dynamics and succession of grasslands.

Acknowledgements

We appreciate the logistical support from the National Field Observation and Research Station of Shenyang Agro-ecosystems and the constructive comments and suggestions from three anonymous reviewers. This study was supported by the National Key Research and Development Program of China (2023YFD1301702), Natural Science Foundation of China (32171543 and 32422057), and the China Scholarship Council joint Ph.D. program. PT was supported by the National Research, Development and Innovation Office [KKP144068 and K137573] during the manuscript preparation.

Author contributions

Niwu Te, Wen-Tao Luo, and Xiao-Tao Lü conceptualized the study and wrote the drafts with Péter Török and Anke Jentsch. Niwu Te, Xiao-Ru Zhang, Xiao-Sa Liang, Yuan-Xiu Wu, and Xiao-Jing Zhang collected the data. Niwu Te analyzed the data. All authors contributed critically to the drafts and gave final approval for publication.

Data availability statement

Data be archived in Science DB (https://doi.org/10.57760/sciencedb.34112).

Statement of inclusion

Our study brings together authors from a number of different countries, including scientists based in the country where the study was carried out. Whenever relevant, literature published by scientists from the region was cited.

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.