From 2011-2018 USGS biologists recorded vegetation and biological soil crust (BSC) cover by species and tracked survival of tagged individual plants (388 in total) across 40 locations where paired experimental plots had been installed in 2010. Plant cover was visually estimated using four 75 x 100 cm survey frames. Each site contained a two plots measuring 1.5 by 2.0 meteres: a control plot and a plot covered by a shelter that excluded 35% of incoming precipitation. Plots were selected to represent shallow vs. deep soils, sandstone vs. shale parent material, and dominant plant species on the Colorado Plateau around Moab, Utah. We used an information theoretic approach using generalized linear models to determine the combination of factors that best predicted mortality. We included treatment, year, and species as fixed effects in our first order models to test for treatment effects on mortality while accounting for the influence of interannual-climate variability and species-level differences. Models also included individual plant ID nested within site as random effects to account for pseudo-replication across sites and tagged individuals. We continued with a second set of models by adding abiotic variables including elevation (m), soil depth (shallow or deep), and parent material as additional explanatory variables to the best-fit model.