Albedo feedbacks to future climate via climate change impacts on dryland biocrusts

Drylands represent the planet’s largest terrestrial biome and evidence suggests these landscapes have large potential for creating feedbacks to future climate. Recent studies also indicate that dryland ecosystems are responding markedly to climate change. Biological soil crusts (biocrusts) ‒ soil surface communities of lichens, mosses, and/or cyanobacteria ‒ comprise up to 70% of dryland cover and help govern fundamental ecosystem functions, including soil stabilization and carbon uptake. Drylands are expected to experience significant changes in temperature and precipitation regimes, and such alterations may impact biocrust communities by promoting rapid mortality of foundational species. In turn, biocrust community shifts affect land surface cover and roughness—changes that can dramatically alter albedo. We tested this hypothesis in a full-factorial warming (+4 °C above ambient) and altered precipitation (increased frequency of 1.2 mm monsoon-type watering events) experiment on the Colorado Plateau, USA. We quantified changes in shortwave albedo via multi-angle, solar-reflectance measurements. Warming and watering treatments each led to large increases in albedo (>30%). This increase was driven by biophysical factors related to treatment effects on cyanobacteria cover and soil surface roughness following treatment-induced moss and lichen mortality. A rise in dryland surface albedo may represent a previously unidentified feedback to future climate.

. Percent change from control for soil surface roughness and moisture. Percent change was calculated from the treatment averages for both soil roughness and moisture. For soil roughness, a negative value indicates a smoother soil surface from the control. Negative soil moisture indicates a dryer soil surface compared to the control. The soil roughness percentage values are derived from a soil roughness index calculated following Saleh 1993. The soil moisture percentages were calculated from direct measurements of volumetric water content (see Methods Figure S1: Boxplot of at surface irradiance in drylands and non-drylands in the USA Global Horizonal Irradiance (W m -2 ) data compiled from 10 different Department of Energy NREL (National Renewable Energy Laboratory) locations (n=5 for dylands; n=5 for nondrylands) in 2012. The monthly average irradiance in 2012 was averaged for a yearly average irradiance value. Drylands significantly differed in surface irradiance from non-dryland locations (P < 0.0001). Significant differences between dryland and non-dryland locations were tested with a one-way ANOVA using the statistical program R.  (Fig. 3, S4).

Calculation used in estimation of radiative forcing values
The following references are specific only to the Supplementary Information radiative forcing calculation above and not also cited in the manuscript: The following three figures (S2-S4) include the outlier in the dataset. These figures display the large variance in the watering treatment ( Figure S2 and S4) and can be used to compare with the figures in the manuscript, which do not include the outlier. Figure S3 has the same R 2 and Pvalue independent of outlier removal.

Figure S2 | Boxplot of albedo separated by treatment with outlier in Water treatment
Significant differences were found between the control albedo and the warm (P < 0.01) and water + warm (P < 0.05) treatments, but did not significantly differ from the water treatment (P = 0.09) with the outlier included. The large variance, due to possible measurement error, in the water treatment led to the conclusion to remove the outlier and use the Kruskal-Wallis analysis of variance.

Figure S3 | Linear models relating albedo to the proportional cover of cyanobacteria, soil surface roughness and soil moisture
The (a) proportional cover of cyanobacteria within biocrust communities of experimental plots was collected in point-intercept frames in autumn 2014, and calculated as the ratio of points intercepting cyanobacteria relative to total biotic cover (sum total of cyanobacteria, moss, and lichen points). Soil surface roughness (b) was measured in spring 2014 to calculate a roughness index to characterize the soil surface roughness upslope and across slope within each plot. Soil moisture (c) at a depth of 5 cm was measured as the hourly average of values recorded every five minutes during the same time as the albedo measurements. R 2 and P-values are from simple linear regression. Climate treatments are denoted by symbol colors (green = control, red = warming, blue = watering, purple = warming + watering). The outlier in the water treatment is included in this analysis (n = 5 compared to Figure 2 where n = 4).

Figure S4 | Estimated global mean radiative forcing from anticipated changes in biocrust communities and IPCC AR5 from 1750 to 2011.
Global radiative forcing values resulting from changes in biocrust cover were calculated here using all data (i.e., the data shown in Figure 3 plus the noted outlier for the watered treatment) in equation (1) described in the methods. Uncertainties for treatment radiative forcing are represented by 95% confidence intervals (black bars). Effective radiative forcing (ERF) values were used for total anthropogenic, aerosol interactions, and well-mixed greenhouse gasses. All other IPCC derived values are of radiative forcing (RF). Uncertainties for the IPCC AR5 RF and ERF values are represented by 5 to 95% confidence intervals.

Figure S5 | Diagram of spectro-goniometer collection angles (zenith and azimuth)
Spectra were collected at several, repeated zenith (a) and azimuth (b) angles to more accurately estimate albedo. Central zenith angles along the spectro-goniometer arc were chosen to limit the amount of interference from plot instrumentation and non-soil surfaces (e.g., vascular plants). Specific azimuth angles (or planes) were used to capture the maximum amount of scattering reflected light. Figure S6: Linear regression models relating albedo to surface cover of key biocrust organism groups The proportional cover of dominant biocrust community groups was collected via point-intercept frames in autumn 2014, and calculates the ratio of points intercepting cyanobacteria (a), moss (b), or lichen (c) to total biotic cover (sum total of cyanobacteria, moss, and lichen points). Albedo was significantly related to all three biocrust community groups, with cyanobacterial ground-surface cover explaining the largest amount of variation in albedo.