Nature: Arid and semi-arid ecosystems cover ~40% of Earth’s terrestrial surface1, but we know little about how climate change will affect these widespread landscapes. Like many drylands, the Colorado Plateau in southwestern United States is predicted to experience elevated temperatures and alterations to the timing and amount of annual precipitation2, 3, 4. We used a factorial warming and supplemental rainfall experiment on the Colorado Plateau to show that altered precipitation resulted in pronounced mortality of the widespread moss Syntrichia caninervis.
Increased frequency of 1.2 mm summer rainfall events reduced moss cover from ~25% of total surface cover to <2% after only one growing season, whereas increased temperature had no effect. Laboratory measurements identified a physiological mechanism behind the mortality: small precipitation events caused a negative moss carbon balance, whereas larger events maintained net carbon uptake. Multiple metrics of nitrogen cycling were notably different with moss mortality and had significant implications for soil fertility. Mosses are important members in many dryland ecosystems and the community changes observed here reveal how subtle modifications to climate can affect ecosystem structure and function on unexpectedly short timescales. Moreover, mortality resulted from increased precipitation through smaller, more frequent events, underscoring the importance of precipitation event size and timing, and highlighting our inadequate understanding of relationships between climate and ecosystem function in drylands.
Ecosystem responses to climatic change have been documented globally and discerning the mechanisms behind and consequences of these responses is a central theme in contemporary ecological research. However, our grasp of how dryland ecosystem structure and function will respond to changing climate remains poor5. Dryland ecosystems comprise a substantial proportion of total land cover and constitute a significant component of global biogeochemical cycles1, 6, yet owing to strong limitations by water and nutrients7, 8, undisturbed drylands are typically thought to maintain relatively low annual rates of ecosystem processes—such as plant photosynthesis5 (but see refs 9, 10)—and to harbour biological communities that change composition on relatively slow timescales11.
Predictions of dryland organism response to increased temperature and altered precipitation remain under debate. Some research suggests dryland plants and soil food webs will be relatively resilient to changes in climate, as they are well adapted to extreme environments7, 12. In contrast, other studies suggest many dryland organisms are already living at the edge of their ecological tolerance and may respond significantly and nonlinearly to even subtle climatic changes7, 13, 14, 15, 16, 17, 18. Furthermore, studies of vascular plants suggest that changes to the timing and size of individual rainfall events may be as or more important for ecosystem function than changes to the absolute amount of annual rainfall13, 14, 15, 16, 17, 18. Of particular interest is the fate of biological soil crusts (biocrusts), a diverse soil community comprising mosses, lichens and cyanobacteria6 that represent a major component of dryland ecosystems. How biocrusts will respond to climate change remains almost wholly unknown; nevertheless, these organisms play crucial roles in dryland function6: fixing C and N (ref. 19), regulating hydrology, affecting seedling establishment and stabilizing soils6.
We conducted a field experiment to assess the response of biocrust communities to altered climate and a laboratory experiment to explore the underlying mechanisms behind the observed effects. For five years, we applied full-factorial field warming (three years at 2 °C above ambient and a subsequent two years at 4 °C above ambient) and watering treatments (quadrupling the long-term average frequency of ≤1.2 mm summer rainfall events) in twenty 5 m2 plots on the Colorado Plateau (Supplementary Table S1 and Figs S1–S3). We monitored biological and biogeochemical soil responses to treatments and found that, whereas 2–4 °C warming had no significant effect on biocrust communities, increasing the frequency of small summer watering events led to pronounced mortality of the dominant moss (S. caninervis; comprising >25% of total plot cover). Not only was this die-off event sizeable—reducing cover to nearly zero in watered plots—it was rapid: mortality was observed after a single season of treatment and the moss did not recover (Fig. 1a). We also conducted a laboratory experiment, collecting moss samples adjacent to the field experiment and subjecting them to 1.25 and 5 mm watering events. We found the 5 mm watering additions resulted in net C fixation and an overall positive C balance (Fig. 2). In contrast, the 1.25 mm events resulted in more respiratory loss than photosynthetic gain and thus a net negative C balance.
