Successional retrogression of biocrusts: a common response to perturbation
Eight years of reduced precipitation input led to lower late successional biocrust cover and higher lightly-pigmented cyanobacterial cover, which likely expanded into areas previously occupied by mosses (Figure 1). The moss constituent of the biocrust, dominated bySyntrichia caninervis in this region, has previously shown pronounced and rapid mortality due to carbon starvation in response to an increased frequency of small summer watering events relative to a lesser frequency of larger hydration events (Coe et al . 2012; Reed et al . 2012). Because mosses, like all biocrust organisms, lack roots and do not actively regulate water loss (e.g., do not have stomates), their activity and carbon gain are directly tied to the amount of time that soil surfaces are wet (Mishler & Oliver 2009). Furthermore, upon rehydrating, they initially lose more CO2 to respiration than they gain from photosynthesis (Coe et al . 2012). If the biocrusts dry again before equilibrium is reached beyond the compensation point for C fixation, then they experience a net C loss over that hydration event. Repeated short-duration hydration (e.g., small amounts of water added during summer monsoon precipitation) can nearly eliminate mosses in approximately one year (Reed et al . 2012). Our results indicated that a similar trajectory may unfold more gradually if the amount of precipitation is decreased across all rain events—with the 35% year-round reductions in this experiment effectively shifting the storm size distribution towards a greater frequency of smaller events. Less precipitation reaching the surface is likely to shorten hydration periods on average, offering less C gain per hydration event, and thus potentially more hydration events that result in net C-losses. Syntrichia caninervis did appear to be substantially more resistant to an overall reduction in precipitation than previous evidence on sudden simultaneous shifts in hydration frequency and magnitude. Relative to mosses, lightly-pigmented cyanobacteria (usually Microcoleus spp .; Garcia-Pichelet al . 2013) appear much more resistant and able to expand in response to both kinds of precipitation change (Ferrenberg et al . 2015). Contrary to our hypothesis lichen cover did not experience the same negative impact as mosses with precipitation reduction despite being a late-successional species.
One of the more common successional sequences observed in biocrusts across dryland ecosystems is early colonization of lightly-pigmented pigmented, filamentous cyanobacteria, followed by darker pigmented cyanobacteria, followed by lichens and/or mosses (Weber et al . 2016). This sequence is not universal but has been widely documented in our study area and many other locations. Thus, our observations are consistent with a significant slowing of succession or even with a reversal of succession. Ferrenberg et al ., (2015) demonstrated that similar reversals are a common response to climate change and recurrent pulse perturbations, including stressful shifts in watering frequency, experimental warming, and physical trampling. We add long-term experimental precipitation reduction to the perturbations that can induce this response in biocrust successional maturity.
A successional retrogression of the biocrust state points toward cascading effects on multiple dryland ecosystem processes and functions. In comparison to the functions provided by earlier successional biocrust, reduced abundance of late successional biocrust implies a decrease in carbon and nitrogen fixation (Housman et al . 2006), soil moisture (Eldridge et al . 2020), soil stability (Ficket al . 2019), and increases in soil redistribution, runoff, surface albedo, soil microtopography, and alterations to land-surface energy (Rutherford et al . 2017; Xiao & Bowker 2020). Moreover, the elevated bare soil (13.1% higher in droughted plots) implies that not only are the droughted biocrusts less functional, but precipitation reduction may cause an overall lower cover of biocrusts across a wide range of dryland types, consistent with global models predicting an overall decline in biocrust coverage as the climate changes (RodrĂ­guez-Caballero et al . 2018).