Digital Mapping of Vegetative Great Groups to Inform Management Strategies

Abstract

Ecological site descriptions have become a prominent way of describing plant communities across rangelands. Disturbance response groups (DRGs) stratify landscapes by grouping ecological sites on the basis of their responses to natural or anthropogenic disturbances. DRGs allow managers to organize, scale, and evaluate information collected on the ground, thus creating expectations of how sites with similar characteristics will respond to disturbance and management. While the importance and utility of these concepts are well understood, the location and spatial extent of DRGs are not. Uncertainty of DRG location and extent make it challenging to evaluate trends or degradation risks of a given area and difficult to define and organize adaptive management concerns and opportunities on a landscape scale. DRGs are organized by major land resource areas (MLRAs), which can make real-life applications across MLRA boundaries for natural phenomena (e.g., wildfire boundaries) repetitive for specific management objectives. Vegetative great groups have been used to overcome this challenge while retaining the state-and-transition model importance of ecological sites. Presented here is a gridded process for vegetative great group mapping across MLRA boundaries, as well as an assessment of the ecological implications of the information gained about the plant communities through the mapping efforts. The scale and output are designed to fit the Landsat library grid and its derived information. Computer machine learning was used to generate spatial maps of vegetative great groups that were compared with Natural Resources Conservation Services soil survey maps, which are currently used by public land management agencies. Machine learning enhanced accuracy by 14% versus conventional soil mapping, providing a more accurate way to conceptualize and manage plant communities at the landscape scale. Further, predictor variables used in machine learning can supplement our knowledge of ecological process information on sites and aid land managers in understanding the various plant community responses to disturbance.