Biocrusts are compositionally resistant to climate perturbation
We found a general compositional resistance indicated by the lack of substantial shifts in biocrust species composition in response to the imposed 8 year drought. Our analysis based on the NMDS ordination suggested only weak evidence of compositional shift; the lack of a stronger trend could be due to the heterogeneity of the communities, the variance in successional state among communities, and the often strong random component of the community responses to perturbations (Schmitz 1997). Part of the heterogeneity of the sites involves the long history of disturbances of the Colorado Plateau, like grazing by domestic livestock and wildfire, that are associated with a slow recovery rate due to the low productivity of the system (Schwinning et al . 2008). This may have resulted in a landscape where our experimental sites had different initial biocrust states and communities did not start at their maximum potential, having a realistic mosaic of communities of the region. Nonetheless, the weak direction of the compositional shifts correlated with the aridity gradient suggests that drought could slowly drive biocrusts to a species composition characteristic of more arid environments, underlying the common successional retrogression response of community maturity previously shown. This could occur both through alterations to the pace of recovery following disturbance and through unique successional trajectories caused by an altered climate. These minor shifts in biocrust composition, here measured as dissimilarity indices, were partitioned into two components: species gains and species losses. The biocrust species gains and losses indices were on average equivalent, showing that neither of the two processes was dominant in the compositional changes, regardless of the indication given by successional retrogression in which a consistent species loss would be expected as late successional cover decreases. However, we consider that perhaps a longer experimental time period than our eight-year drought could be necessary to shift composition, as biocrust species reach environmental conditions outside their tolerance range.
We expected that diversity would increase biocrust resistance to drought based on our second hypothesis. However, our models showed that the dissimilarity between control and droughted communities was not driven by the control diversity or by site aridity. Although diversity is rarely the primary driver of stability (Worm & Duffy 2003), finding no predictive power of diversity on the compositional resistance was unexpected, and leads us to infer that the degree of the compositional resistance may be driven by other community attributes rather than taxonomic diversity.
Additionally, we found a weak but positive relationship between overall dissimilarity and control plot diversity, suggesting that more diverse biocrusts are slightly less resistant due to greater shifts in composition as they gain and lose species (Figure 3). These results do not support our initial diversity-stability hypothesis; however, resistance, as a component of stability, refers to the difference in a system property from its unperturbed condition to a perturbed one. If under undisturbed conditions a property increases with diversity, the post-disturbance relationship is likely to have a less positive slope. Because species-poor systems already have low values of such properties under undisturbed conditions, they may be less affected by disturbance than species-rich systems (Pfisterer & Schmid 2002). Past studies have documented a negative relationship between diversity and resistance in systems with diversity-dependent productivity (Pfisterer & Schmid 2002; Vogel et al . 2012). This rationale explains the stronger negative impact of drought on richer biocrust communities, where the change in composition accounts for changes in relative abundance or cover (a productivity proxy), in a system that has shown richness to enhance productivity and functionality (Maestre et al . 2012; Bowkeret al . 2017).
While previous studies have shown aridity to drive biocrust composition and functionality (Bowker & Belnap 2008; Delgado-Baquerizo et al . 2016), which could create variations in the strength of biotic interactions (Bowker et al . 2010; Maestre et al . 2010), we found no support here for the effect of aridity in biocrust compositional resistance. This lack of an aridity effect could also reflect a strong adaptation of the biocrust community to ambient aridity. Similarly, previous work at these same study sites found a surprising lack of interaction between drought and soil attributes in determining plant response (Hoover et al . 2015), which led to the suggestion that vegetation communities are adapted to their local soil conditions such that drought responses are similar across contrasting soil types. However, it is also possible that the range of our environmental gradient, our sample size, varied land use history, or the experimental duration, was not broad enough to capture the interaction of aridity and drought on biocrust composition resistance, or that other environmental factors not considered could have a stronger effect on the impacts of global change in our system.