The principal aim of this project was to project changes in fuels, fire dynamics, and associated responses of vegetation and breeding birds that might inform selection and prioritization of management actions in the Great Basin. Our original six objectives were to model percent cover of cheatgrass (Bromus tectorum) across the Great Basin and project changes in cheatgrass cover as a function of precipitation; increase the accuracy of models of percent cover of sagebrush and herbaceous vegetation; use field data to train and validate models of vegetation cover and to relate cheatgrass cover and biomass; model probability of fire across the Great Basin as a function of percent cover of cheatgrass and precipitation; project changes in habitat quality and occupancy for sensitive-status birds given changes in vegetation; and test whether ecological thresholds (values of an environmental pressure at which ecosystem state changes abruptly and nonlinearly) occur in projected future vegetation cover, habitat quality, and occupancy.

We established study sites in four biogeographically distinct regions of the Great Basin. We collected field data on cheatgrass cover and biomass and other aspects of vegetation composition and structure, captured ultra-high resolution aerial images, and collected point-count data on breeding birds and their habitat.

The current body of public data does not allow reliable estimation of percent cover of cheatgrass as a continuous variable. Across the Great Basin, categorical percent cover of cheatgrass increased as primary productivity early in the growing season, which was correlated positively with precipitation, increased. Both the presence and proportion of years of grazing increased the probability of presence and prevalence of cheatgrass. Our results do not support the use of livestock grazing to suppress cheatgrass, and especially not in unburned areas.

Remote estimation of cheatgrass cover on the basis of visible spectral bands remains quite difficult unless the phenological stage of cheatgrass contrasts sharply with that of surrounding vegetation and cheatgrass cover exceeds about 20%. Ilastic, software with a random forest algorithm that uses spectral and texture features, proved to be a generally accurate classifier of land cover within sagebrush shrubsteppe on the basis of ultra-high resolution images.

Relations between percent cover and biomass (which is strongly related to fire likelihood) of cheatgrass were moderate to strong. We are clarifying whether the strength of the relation varies among regions or wet and dry years, or on the basis of sampling methods. From 1992–2012, about 52% of fires in the Intermountain West were human-ignited (75% in areas with cheatgrass), and that human ignitions more than doubled the length of fire season. The frequency of fires increased substantially in areas in which observed cheatgrass cover was > 1%.

Environmental associations with the abundances of individual species of breeding birds often were similar among regions, but apparent random or directional movement of individual birds during the breeding season complicates model fit. It appears that some species of birds are moving directionally along elevational gradients both within seasons and among years. The abundances of many riparian-obligate and facultative riparian species appeared to decline sharply beyond a given threshold of within-canyon riparian area.