DISCUSSION
Many terrestrial ecosystems are increasingly experiencing drought events. However, despite much research, a general understanding of mechanisms underlying ecosystem resistance to drought and recovery after drought remains elusive. This is at least, in part, due to the focus on the functional resistance and recovery in previous studies of ecological stability, which have paid relatively little attention to structural resistance and recovery. Elucidating how ecosystems respond to drought is becoming particularly pressing when these ecosystems are simultaneously undergoing chronic environmental changes (e.g., increased N deposition), which may have the potential to alter ecosystem responses to drought (Xu et al., 2014; Shi et al., 2018). Using a manipulative field experiment conducted in a temperate semiarid grassland, our study showed that N enrichment and mowing had strikingly different effects on grassland biomass and structural resistance/recovery. Specifically, nitrogen addition resulted in reduced biomass resistance and increased biomass recovery, whereas annual mowing resulted in reduced structural resistance and increased structural recovery. Importantly, our study demonstrated that dominant species resistance/recovery, rather than species diversity, largely modulated the effects of nitrogen enrichment and mowing on both functional and structural resistance/recovery of our study grassland.
An appreciable number of studies have shown that nitrogen enrichment often reduces grassland biomass temporal stability (Yang et al., 2012; Hautier et al., 2014; Hautier et al., 2015; Song & Yu, 2015; Xu et al., 2015b; Zhang et al., 2016; Zhang et al., 2019; Hautier et al., 2020; but see Grman et al., 2010). Much less, however, is known about how nitrogen enrichment influences grassland biomass resistance to disturbance (Xu et al., 2014; Shi et al., 2018) and recovery after disturbance (Kinugasa et al., 2012; Xu et al., 2014). Our study showed that nitrogen addition decreased plant community biomass resistance to drought, but increased biomass recovery after drought. As nitrogen addition resulted in increased plant community biomass, indicating nitrogen limitation in our study grassland, this opposite effect of nutrient addition on biomass resistance and recovery also means that more productive communities lost more biomass during drought but recovered more biomass after drought (Fig. S7). The strong negative relationship between biomass resistance and pre-draught community biomass (r 2 = 0.918,P < 0.001) lends strong support to the hypothesis that drought-induced plant biomass reduction is sensitive to pre-drought community biomass (the biomass-dependent resistance hypothesis; Wang et al., 2007). This result is also consistent with the idea that more fertile grasslands should be more responsive to climate change (Grime et al., 2000). This result arose presumably because of increasing water shortage associated with communities that attained greater biomass, and therefore, increased evapotranspiration, following N addition (Friedrich et al., 2012; Valliere et al., 2017; Shi et al., 2018). More mechanistically, we found that nitrogen enrichment reduced community biomass resistance mainly by exacerbating the biomass loss of L. chinensis during drought. L. chinensis is a perennial rhizomatous grass with relatively high foliar nitrogen content (Cui et al., 2010), and nitrogen enrichment tends to increase its abundance, stature and root biomass (Pan et al., 2005; Bai et al., 2009). As the most dominant species in our study grassland, L. chinensis plants are known to show reduced stomatal conductance and photosynthesis, and in turn, reduced biomass production when experiencing drought (Xu & Zhou, 2006), especially under nitrogen enrichment (Shi et al., 2018).L. chinensis plants are also known to allocate more of their biomass into their belowground components when experiencing drought (Xu & Zhou, 2006), particularly under nitrogen enrichment (Shi et al., 2018), presumably as an adaptive strategy to maximize water intake. This strategy may have allowed L. chinensis to quickly expand its clonal reproduction when more water becomes available, contributing to increased community biomass recovery in the nitrogen addition plots after drought.
Compared with nitrogen deposition, we know relatively little about how mowing influences grassland stability. The few studies that examined the effect of mowing on grassland biomass temporal stability have produced mixed results (Yang et al., 2012; Zhang et al., 2017). Even less is known about how mowing influences grassland resistance and recovery. The only study on this topic, to our knowledge, examined the response of experimental grasslands to experimentally induced drought under mowing of different frequency, reporting that grassland biomass resistance declined with increasing mowing frequency (Vogel et al., 2012). However, our study found that mowing promoted community biomass resistance, primarily via enhancing the biomass resistance of L. chinensis (Fig. 3a). The fact that mowing reduced community and dominant species biomass and made our study grassland less vulnerable to drought is again consistent with the biomass-dependent resistance hypothesis (Wang et al., 2007). The apparent discrepancy between the results of our study and Vogel et al. (2012) may be explained by the difference in mowing frequency between the two studies. Whereas mowing was implemented once per year in our experiment, it was implemented two or four times a year in the experiment of Vogel et al. (2012), with only the more frequent mowing treatment diminishing biomass accumulation and reducing community biomass resistance to drought in the latter study. Note that increased species asynchrony under mowing (manifested in the SEM as an indirect effect of mowing via L. chinensis biomass resistance) also contributed to community biomass resistance (Fig. 3a). Mechanistically, mowing reduced the abundance of dominant, drought-sensitive species (i.e., L. chinensis ), which offered greater opportunities for species that were less abundant but more tolerant of drought (e.g.,C. duriuscula, S. baicalensis , Thermoposis lanceolata ) to compensate for its biomass loss under drought, leading to greater asynchrony.
One of our most important results is that only mowing altered grassland structural stability, resulting in reduced grassland structural resistance and increased grassland structural recovery. Notably, these effects of mowing on community structural stability were largely mediated via dominant species structural stability, which was strongly associated with community structural stability (Fig. 3). Importantly, mowing had opposite effects on L. chinensis biomass stability and dominant species structural stability. Mowing reduced dominant structural resistance despite its positive effect on L. chinensisbiomass resistance, as three of the other four dominant species exhibited the trend of declining resistance to drought under mowing (Fig. S8a-d). On the other hand, mowing increased dominant structural recovery despite its negative effect on L. chinensis biomass recovery, as all the other four dominant species exhibited the trend of increased recovery from drought under mowing (all P < 0.001; Fig. S8e-h). These results correspond to the fact that L. chinensis , as the most dominant species, imposed negative effects on the other dominant species (Fig. S9) but suffered substantial biomass loss from mowing (Fig. S5), with ensuing consequences for the abundance of other species and the structural stability of our study communities. Previous work has shown that mowing reduces the dominance of L. chinensi s by hampering the formation of regeneration buds and the propagation of rhizomes (Yang et al., 1995). Our results are consistent with previous studies reporting that mowing tended to reduce the abundance of dominant species, resulting in increased light availability (Molina et al., 2021) and, in turn, increases in grassland plant diversity (Leps, 2014; Yang et al., 2019; Molina et al., 2021). Our study, however, focused on examining the effects of mowing on community resistance and recovery, demonstrating the importance of dominant species for driving the reassembly of communities experiencing drought.
Overall, our results point to the importance of dominant species for determining both biomass and structural stability of our study grassland experiencing drought. Species richness was not identified as a significant predictor of grassland biomass or structural stability (Fig. 3). We attributed these results to the predominant role of dominant species in contributing to plant biomass production and driving plant community assembly in our study grassland. In our experiment, the most dominant species, L. chinensis, accounted for almost half of grassland aboveground biomass. Its response to drought not only largely drove grassland biomass resistance and recovery, but also imposed substantial influences on the responses of other species (including other dominant species) to drought, with subsequent influences on community structural resistance and recovery. Our results are thus consistent with the findings of several previous studies that plant community functional resistance and recovery are modulated by the traits of dominant species (Macgillivray et al., 1995; DeClerck et al., 2006; Hoover et al., 2014). Together, these results provide strong support for the mass ratio hypothesis that properties of an ecosystem are largely determined by its dominant species, rather than species diversity (Grime, 1998). Moreover, our finding that the response of dominant species to disturbance also strongly influenced the reassembly of communities undergoing disturbance suggests that the mass ratio hypothesis may also be applied at the community level. We suggest that elucidating traits that regulate dominant species resistance to disturbance and recovery from disturbance, such as the pre-draught biomass as identified here, are particularly important for predicting both community and ecosystem responses to increasingly frequent disturbance events.
A recent meta-analysis (Hillebrand & Kunze, 2020) reported that functional and structural stability of ecological communities tended to be positively correlated, such that functional recovery of communities following disturbance is more complete with greater structural recovery. In line with this trend, we found that the recovery of the structure of our study grassland promoted its biomass recovery (Fig. 3b; Fig. S6b). This pattern emerged as the recovery in the biomass of other species complemented that of L. chinensis in contributing to community biomass recovery, particularly under mowing that inhibited L. chinensis biomass accumulation. The meta-analysis of Hillebrand and Kunze (2020) also found that for some communities, functional recovery was possible in the absence of structural recovery, a pattern attributable to the functional redundancy among species (Yachi & Loreau, 1999; Allison & Martiny, 2008). Likewise, we found that the biomass resistance of our study communities was decoupled from their structural resistance (Fig. 3a). This pattern arose as drought substantially altered grassland community structure, but L. chinensis remained a predominant contributor to community biomass and drought-tolerant species partially compensated for the loss of biomass of drought-intolerant species.