Discussion
This study compared microbial community stability components from different successional stages of terrestrial ecosystems and offered direct empirical evidence for the role interactions play in governing RSRs.
We found a rough trend that forests had higher resilience, but lower resistance, than shrub, grassland, and bare soil (Fig. 1A), which was opposite to the trend displayed by plant communities where complex systems were likely to have higher resistance (Isbell et al.2015). We suggest that such differences arise from the reproductive rates, physiological resistance, and differences in richness between plants and microbes (Curtis 2006; Konopka 2006). Compared to microbes, plants have much lower reproductive rates and richness (Curtis 2006; Konopka 2006). When faced with short term high temperature stress, on the order of only several days, the dominant forest plant keystone population was likely to remain unchanged unless large scale regional death occurred, and thus appearing to have high resistance (Curtis 2006; Konopka 2006). However, microbes have extremely high richness as compared to plants in terrestrial ecosystems, and are able to reproduce in a timespan as short as hours or even minutes (Curtis 2006; Konopka 2006). These indicate that competitive functional groups with different temperature range tolerances could potentially replace the original dominant groups and thus led to lower resistance of the microbial community (Pinsky 2019). However, with keystone species dominated by organisms with high stress tolerance, such as the communities from bare soil (Fig. 1A) that are exposed to high day-night temperature variation, dry-wet alteration, and high ultra-violate, could exhibit high resistance because those keystone species were not likely to change (Remias et al. 2012; Harrison & LaForgia 2019).
We found a strong negative linear relationship between microbial community resistance and resilience, which indicated a trade-off between resistance and resilience of microbial communities in the ecosystems studied here (Fig. 1B and Fig. 1C). This is plausible and inevitable from both basic logic and an evolutionary perspective. The essence of resistance and resilience is the ability of altering the relative abundance of species as conditions change. Thus, a microbial community that is more readily prone to change simultaneously has less resistance and higher resilience, and vice versa (Miller & Chesson 2009; Griffiths & Philippot 2013). From an evolutionary perspective, communities need to coordinate the functions of different components to ensure the continuation of key ecological processes for survival of the community under variable environmental conditions. This can be realized by assigning key functions to a few stress tolerant functional species (such as the high resistance community in bare soil) (Craine et al. 2013), or to alternative functional groups composed of different stress tolerant members (like the low resistance community in forests) (Whitham et al. 2006; Walworth et al. 2020), that can be treated as the K-strategy and r-strategy of community evolution or succession. However, to simultaneously possess K- and r-strategy wastes energy and is an evolutionary dead-end (Whitham et al. 2006; Liet al. 2020; Walworth et al. 2020). Thus, the RSRs varied according to the stability components that we observed.
Most importantly, our results confirmed the role that interaction types play in governing RSRs. Richness offers a basic available species pool from which the community derived, while interactions offer the basic functional organizational patterns for species to form the community (Whitham et al. 2006; Montesinos-Navarro et al. 2017; Walworth et al. 2020). Our results showed that only under approximately balanced proportions of positive and negative interactions did richness increase resistance and decrease resilience, and that an exceedingly high proportion of positive interactions caused richness to decrease resistance and increase resilience (Fig. 3). An extremely high proportion of positive interactions is theoretically unfavorable for resistance because the extinction of one species would threaten the survival of other cooperative or mutualistic species that rely on it, even though they maybe insensitive to the stress (May & MacDonald 1978; Damore & Gore 2012). The higher the richness, the greater the possibility for species co-extinction through positive interaction. Thus, when under exceedingly high proportions of positive interactions, the higher richness and the lower the community resistance (May & MacDonald 1978; Damore & Gore 2012). However, negative interactions compensate for the effect of the positive interactions on RSRs. Negative interactions include predation and competition, with competition the main consideration in bacterial and fungal community networks (Deng & Zhou 2015). Competition indicates an overlap of function and niche among community members (Pianka 1981; Pinsky 2019). Groups which conduct the same functions are capable of replacing species that had been killed off by stress, and protect other members from further extinction due to the disappearance of their functional partner(s) (Montesinos-Navarro et al. 2017; Qian & Akçay 2020). Thus, with appropriate proportions of positive and negative interaction, species can coexist in a manner without total dependency. The higher the richness, the higher the functional redundancy and increased community resistance (Pianka 1981; Pinsky 2019).
Theoretical ecologists have considered interaction types as one factor that influence RSRs in models, even though there has been no empirical evidence from previous field and laboratory studies to support this idea. Models constructed based on different constraints and underlying theories generate significantly different results, while our empirical results were supported in part by recent modelling studies which indicated that the appropriate ratio of positive and negative interactions facilitate RSRs (Mougi & Kondoh 2012; Qian & Akçay 2020). The constraints to meet these demands were, firstly, the quantitatively comparable and balanced effects on community from both positive and negative interactions and, secondly, the decreasing interaction strength with increasing interactions (Kondoh & Mougi 2015). This indicated that interaction strength and distribution were also potential governors of RSRs, and were capable of influencing the effects exerted by the interaction types. However, whether or not such constraints commonly exist in natural ecosystems is still an open question and requires further investigation to test the effect of interaction strength on RSRs. There are also other models that suggest the mixture of appropriate positive and negative interactions of inter-trophic community cannot promote positive RSRs (Suweis et al. 2013). Considering that most microbial networks cannot be divided into different trophic groups according to recent methods, further research is needed to test whether our conclusions are applicable to macro communities or not. According to the theoretical predication, exceedingly high proportions of negative interactions are also an unbalanced state and could possibly contribute to a negative richness-resistance relationship, similar to an exceedingly positive unbalanced state (Mougi & Kondoh 2012; Qian & Akçay 2020). Because under an exceedingly negative interaction state, it is possible that a large number of competitors that use similar resources and conduct similar functions may become simultaneously extinct due to environmental stress (Pianka 1981; Pinsky 2019), and thus the higher the richness, the greater the number of species that could potentially become extinct, and the lower the resistance. Unfortunately, microbial communities in our sampling fields, like many terrestrial ecosystems (Shang et al. 2018), were dominated by positive interactions, and thus, our results were incapable of directly supporting the conjecture that under an overly negative unbalanced state, richness also reduces resistance. Therefore, we suggest to test the idea in other negative interaction dominated ecosystems, such as aquatic or artificial ecosystems.