4 Discussion
Our results declared that both N addition and drought reduced the complexity of interaction network via reducing CFG, ID and IS (Figure 2). The decreasing CFG and ID of arthropod networks suggested that some species have been lost (Eklof & Ebenman, 2006; Tylianakis et al., 2010) and the decreasing IS suggested that the interdependence among functional groups was weakened (Rzanny & Voigt, 2012). Notably, N addition reinforced partial relationships between some herbivores and carnivores, which broke the balance of the network and reduced its complexity, while drought eliminated some relationships among functional groups by filtering out some plant species that were related to arthropods. Compared with the ever-lasting effect of N addition and drought on arthropod communities in the peak of growing season, delayed snowmelt had a direct effect on arthropods only in the early growing season (Figure.S1); although delayed snowmelt had no direct effect on arthropod community in the peak growing season, it had significant interactive effects with N addition and/or drought on partial arthropod functional groups (Table.1&2).
Specifically, the effect of plant N could be cascaded up to higher trophic levels in N addition plots (Fig.1c&2), and foliar N was widely believed to have profound influence on arthropod’s interaction both within and among trophic levels (Ilić et al., 2021). Changes in foliar N affect network complexity by interacting with foraging behavior of herbivores, and such effect could also be cascaded up (Bukovinszky et al., 2008). In our experiment, the relative abundance of plant suckers, detritivores and vegetation hunters was increased by N addition (Fig.1c). Generally, abundance of nitrophilous consumers increased due to elevated N content of primary producers when they were relieved from N restriction (Haddad et al., 2000; Elser et al., 2007; Wimp et al., 2010). For instance, aphids or leafhoppers (plant suckers) fed by piercing the phloem of their food plant and thus they are sensitive to changes in plant quality (Muller et al., 1999); In particular, a significant increase in relative abundance of plant suckers and detritivores resulted in reinforcement of partial relationships (e.g., plant suckers vs. web spiders, detritivores vs. parasitoids) under N addition treatment; however, plant chewers (e.g. grasshopper) were insensitive to N addition (Ritchie, 2000) (Table.3). The changes of different functional groups could be attributed to the specific responses of their N-related traits to N addition. Thus, N addition could weaken the net effect of other interactions thereby reducing ecosystem network complexity by strengthening trophic chains in single pathway (Fig.1&2).
Drought and its interaction with N addition both decreased the network complexity as indicated by declined connectance, ID and IS (Fig.2). Our results showed such decrease was realized not only through increasing foliar N, but also via reducing plant species richness (Fig.1b&2). Generally, more diverse producer communities could support more diverse primary consumers, which could be cascaded up to secondary consumers (Scherber et al., 2010; Rzanny & Voigt, 2012). Indeed, our results highlighted the negative effect of reduced plant richness on arthropod network features under drought conditions (Fig.1&2). Many network properties had been changed due to species loss and especially its resulted loss of relationship among arthropods within the networks under drought conditions. For example, omnivores (e.g. ants and beetles) had been transformed into the core of networks under drought conditions because they are better adapted to dry environments with low plant diversity. Such dry habitats are less attractive to most other arthropods, and are thus difficult to maintain stable interactions among arthropods and between arthropods and plants. More importantly, omnivores were found to be linked with many more other functional groups under experimental drought (Fig.1b&1d), suggesting that omnivores may play a critical role in maintaining the arthropod networks (Fig.1b&1d). Although some species could still adapt to the dry habitats, the change of microhabitat structure significantly impact arthropod functional groups and could not offer a suitable environment for predation (Fig.3). For instance, experimental drought decreased the relative abundance of web spiders and their links with other functional groups due to decreased vegetation height and uniformity (Figure.1). Weaving webs requires appropriate two-dimensional or complex three-dimensional environment (Gibson et al., 1992), but drought-induced high structural heterogeneity and low vegetation height are disadvantageous for predation of web spiders, leading to the disappearance of links of web spiders with plant suckers and parasitoids. Similar results had been obtained by different ecosystem studies measuring structure of food webs both qualitatively and quantitatively (Ledger et al., 2012; Woodward et al., 2012; Rosenblatt & Smith-Ramesh, 2017; Amundrud & Srivastava, 2019). However, several researches demonstrated that plant biomass is the key driver of arthropod community under multiple global changes in grassland, but the plant species richness of their study sites was much lower than ours (Wimp et al., 2010; Prather et al., 2020). Although drought strongly impacted plant biomass in the present study (Table.1), the spatial structure (βbal and βbal) of plant communities did not change significantly (Fig.4). Therefore, compared to the variation of plant biomass in grassland, the existence of plant species under climate changes is more important and thus appears the key factor for the arthropod ecological network stability and complexity (Brodersen et al., 2018).
Delayed snowmelt enhanced abundance of arthropod community in June, but showed little effect during the rest of the growing season (Fig.S1), despite that, delayed snowmelt showed significant interactive effects with drought or N addition on part of functional groups (Table.2&3; Fig.S1). For example, the interaction of delayed snowmelt with drought or N addition significantly influenced parts of carnivorous functional groups, including web spiders and parasitoids, which suggests that the global change events occurring in different seasons may interact (Table.2&3). It is conceivable that the effects of multiple global change factors on plants may be amplified and cascaded up to partial functional groups at higher trophic levels, ultimately leading to the mismatch between trophic levels (Post & Forchhammer, 2008; Rosenblatt & Smith-Ramesh, 2017), which will undoubtedly affect ecosystem stability (Baert et al., 2018; De Laendera et al., 2016; De Laender, 2018). The responses of carnivores as a whole to the interaction of snowmelt delay with drought or with N addition were modest, but significant changes did occur to its finer functional groups (including decreasing ground hunters and parasitoids). Especially, the diversity of parasitoids and web spiders differed in the responses to the interaction of snowmelt delay with drought (Table.2&3). It means that the responses of arthropods to multiple global change factors might be dependent on their specific traits (Voigt et al., 2003). Even now the reason for the responses of these functional groups remains unclear, but our study suggests that it is necessary to examine more complete taxa or functional groups in future researches, with the purpose of capturing the panorama of the responses of entire ecosystems. Meanwhile, long-term and slow global changes such as snowmelt can cause subtle biases in ecosystems that need to be taken into account in global change studies (Kroel-Dulay et al., 2015).