The more than 1 million acres comprising the Uintah and Ouray Indian Reservation in northeastern Utah, USA, have not been widely studied, and access to non-tribal members is highly restricted. We assembled a unique dataset of the vegetation in this region by sampling vegetation on 300,000 acres of previously unsurveyed Reservation lands and combining these data with stocking rate surveys conducted by the Bureau of Indian Affairs (BIA) from 2010-2015. We then associated these vegetation data with biophysical covariates including site soil characteristics and precipitation, vapor pressure deficit, and NDVI measured in the sampling year. We built a random forest model using this database to predict annual aboveground vegetation and forage productivity throughout the Reservation and surrounding lands from 1984-2017.
Results/Conclusions
This model improves the inventory of rangelands in the region by providing fine-scale predictions of forage availability and total productivity that are sensitive to site-specific characteristics. These predictions preserve spatiotemporal variability typically lost when calculating stocking rates through conventional methods. Forage varies drastically through space and time, from 5 pounds per acre to over 500 pounds per acre throughout the study area. Recognizing forage variability is valuable for examining site stability in response to disturbances like drought and can improve management decisions by revealing how forage fluctuates in different areas through time. Forage availability in drought years is up to 30% lower than forage in typical years, revealing the potential for overgrazing or insufficient forage availability in drought years. Forage predictions may also be useful to monitor trends in vegetation production over time, and effects of management decisions.