Introduction
For decades we have witnessed the rapid decline of biodiversity across ecosystems due to the synergistic impacts of natural and human-caused disturbances (Stork 2010; Pimm et al . 2014). The combined effects of current and future disturbances represent a threat to the stability of ecosystems that can have significant and broad-scale impacts on multiple environmental services that are linked to and supported by biodiversity (Millennium Ecosystem Assessment 2005). The stability of ecosystems in response to disturbance has intrigued ecologists for decades (Odum 1950; Holling 1973; McCann 2000) and considerable research has been devoted to the relationship between diversity and ecosystem stability. The work of MacArthur (1955) and Elton (1958) pioneered the diversity-stability hypothesis, which states that diversity promotes stability. However, later theoretical and applied ecological studies found evidence of positive, negative, neutral, and dynamic effects of diversity on ecosystem stability (Ives & Carpenter 2007; Tilmanet al . 2012; Isbell et al . 2015; Pennekamp et al . 2018). These varied findings and lack of consensus correspond to the multifaceted nature of disturbances, species diversity, and stability, coupled with the limitations of what can be measured by ecologists (Donohue et al . 2016).
Ecological stability is a broad concept that can be described using three main components in response to a disturbance: temporal variability in a system property, persistence of functions and structure in the face of a change in external conditions (resistance), and rate of recovery from disturbance (resilience) (Pimm 1984; McCann 2000; Donohue et al . 2016). Empirical studies of biodiversity-stability have focused primarily on the temporal variability of one major measure of ecosystem function – plant productivity – in order to establish and measure the integrity of a community or ecosystem (Tilman 1996; Pfisterer & Schmid 2002; Vogel et al . 2012; Isbell et al . 2015; Lepš et al . 2018). But it is also important to assess the ecological resistance as a measure of stability, as pressures of climate change become more frequent, that can help inform policies governing land und use and biodiversity conservation (Donohue et al . 2016; Pennekampet al . 2018).
We focused our study on the Colorado Plateau, a dryland ecoregion of the Southwest United States, where increasing aridity and drought severity are among several expressions of climate change likely to impact these ecosystems and the human societies that depend on their services (Seageret al . 2007; Munson et al . 2011; Cook et al . 2015; Hoover et al . 2015). While responses of dryland ecosystems to the changing climate are still uncertain (Huang et al . 2017), global drylands could exhibit shifts in the distribution, structure, function, and composition of ecosystems; however, geographical and latitudinal variation suggest different outcomes for particular regions (Trenberthet al . 2014; Huang et al . 2017). The cold desert of the Colorado Plateau in southern Utah is one of several dryland regions suffering the combined effects of climate change, changes in land use and invasive species (Munson et al . 2011; Copeland et al . 2017).
One nearly ubiquitous feature of drylands is the presence of a soil surface biocrust community. In these systems, the soil beneath and between the canopy of widely spaced vascular plants is generally dominated by the soil surface community of cyanobacteria, algae, fungi, lichens, bryophytes, and soil microbes collectively known as a biological soil crust (biocrust). This community is a major ecosystem component involved in water balance maintenance, nitrogen fixation, soil stability, wind and water erosion resistance, and dynamic interactions with other biota of dryland ecosystems (Belnap & Gardner 1993; Belnap & Weber 2013; Maestre et al . 2016; Seitz et al . 2017; Havrilla et al . 2019; Eldridge et al . 2020).
Biocrusts often follow a pattern of successional sequence, in which the bare ground is colonized by lightly-pigmented, filamentous cyanobacteria that aggregate the soil particles, making soils suitable for subsequent colonization by nitrogen-fixing, darkly pigmented cyanobacteria, and then by mosses and/or lichens (Read et al . 2016; Weber et al . 2016). Based on the dominance of cyanobacteria versus mosses and/or lichens, biocrusts are often classified into early and late successional stages, respectively, which inform the properties of the community (Lanet al . 2012). Despite being well-adapted to harsh arid conditions, biocrusts and particularly mosses are sensitive to shifts in the timing and magnitude of precipitation and to rising temperatures that shorten the hydration periods in which biocrusts are physiologically active (Reed et al . 2012). Therefore, climate change is a disturbance that can potentially lead to a decrease of the late successional constituents, mosses and lichens, i.e., a reversal of succession (Ferrenberg et al . 2015).
We used a long-term precipitation reduction experiment in the field to investigate the effects of drought on biocrust community composition and successional maturity, and to test the diversity-stability hypothesis using resistance to climate disturbance as one component of ecosystem stability (Pimm 1984). In this context, we defined resistance as the ability of a community to maintain compositional and structural integrity following a disturbance (Sankaran & McNaughton 1999; Connell & Ghedini 2015). Biocrust composition was measured by species presence and abundance, while the successional maturity was considered as the abundances of the functional groups of the community. We hypothesized that: (1) Precipitation reduction would induce a successional reversal in biocrust composition, which would be indicated by a decline in moss and lichen cover and increase in cyanobacterial cover; (2) Higher initial diversity would confer greater compositional resistance, which would be indicated by less extreme changes of community composition to the imposed climate perturbation; and (3) The degree of compositional resistance observed in the biocrust community would be inversely related to the degree of the environmental stress experienced. Testing these hypotheses helped to illuminate the factors controlling community change under a climate change scenario and to move us closer to being able to predict the magnitude of changes in dryland communities.