Desert ecosystems are thought to be highly responsive to natural and anthropogenic environmental changes. The Great Basin of western North America, a "cold desert" of more than 425,000 km2 , currently receives < 250 mm annual precipitation. Approximately 75% of the region is managed by federal and state resource agencies, yet managers often lack even baseline data for ecological assessment and decision-making. Knowledge of the extent to which land cover and measures of biological diversity vary in space and time is essential for making accurate inferences and taking appropriate management action, especially when the consequences of those actions may be irreversible. We aim to understand ecological responses both to implementation of treatments like prescribed fire and in the absence of deterministic treatments.

With colleagues in the Great Basin Ecosystem Management Project and Great Basin Invasive Species and Remote Sensing Network, we are documenting decadal changes in land cover and the response of native animals to these changes. Since the mid 1990s, we have been developing an integrated geospatial database of the central Great Basin. The database includes environmental measurements derived from remote sensing and digital elevation models, as well as comprehensive data on the distribution of butterflies and birds.

We have developed a statistically rigorous framework for examining the generality of predictors of species richness (number of species) and species occurrence (presence or absence of individual species) using an iterative process of model building, validation, and refinement. The framework identifies predictors of species occurrence at resolutions of several km2 over extents of 100s to 1000s of km2, which correspond to the scale at which many land-use decisions must be made. Successful models can be linked with alternative land-cover scenarios to evaluate the potential effects of each scenario in terms of sustaining, increasing, or decreasing native biodiversity.

Riparian systems are a primary focus of our work. Although riparian areas cover less than two percent of the Great Basin, they support a disproportionately large proportion of the region's economic activity and native species. Existing data suggest that resources in riparian areas are being lost or degraded by a current episode of stream channel incision. We are using remote sensing and field measurement of vegetation to identify factors controlling the distribution of riparian meadows, to assess stream and meadow condition, and to prioritize restoration efforts. This will help us to achieve a better understanding of the structure and functioning of riparian ecosystems within the Great Basin and to develop guidelines for maintaining or restoring their integrity.

Analysis of remotely sensed data is facilitating calculation of rates of environmental change related to invasion of cheatgrass (Bromus tectorum) and expansion of pinyon-juniper woodlands. For example, cheatgrass germinates and senesces earlier in the year than most native perennial species. Because indices of greenness can be derived from satellite data, these differences in phenology can be observed with remote sensing. In addition, the frequent occurrence of fire in cheatgrass-dominated landscapes results in distinct spectral signatures. Woodland expansion is slow (50-100 years) relative to the repeat measurement times and inventories of remotely sensed data. Nevertheless, later stages in the process can be identified remotely as a steady interannual increase in green cover.

Our ongoing work suggests that integration of remote sensing, spatial analysis, and field data may be one of the most productive avenues for developing adaptive management strategies that will conserve both biodiversity and the ecological processes that sustain it.