Oral Presentation Society for Freshwater Science 2026 Annual Meeting

Using remotely sensed data to quantify the influence of beaver dams on surface water in small prairie streams. (136094)

Matt Webster 1 , Colleen Piper 1 , Lisa Eby 1 , Blake Hossack 1 2 , Patrick Donnelly 1 3 , Thomas Riecke 1
  1. University of Montana; W.A. Franke College of Forestry and Conservation, Missoula, Montana, United States
  2. U.S. Geological Survey; Northern Rocky Mountain Science Center, Missoula, Montana, United States
  3. University of Montana; Numerical Terradynamic Simulation Group (NTSG), Missoula, Montana, United States

Water limits productivity in semi-arid and arid regions. Beaver-based restoration is increasingly being used to conserve and restore streams and riparian areas and to promote ecosystem resilience to drought. However, the influence of beaver dams on surface water has not been examined in small prairie streams. Watersheds in the northern Great Plains are characterized by erosive soils and sparse woody vegetation, often resulting in deeply incised channels with confined stream flow. This region is also known for variable and unpredictable summertime flows across intermittent and permanent streams. Although data from satellite imagery has dramatically increased our ability to monitor beaver dams and surface water change across landscapes, assessing the effects of beaver in prairie watersheds has been hindered by many factors, including stream size, flow intermittence, and deeply incised stream banks. To quantify the effects of beaver in prairie streams in eastern Montana, we created a random forest-based land cover classifier that uses fine scale, freely available satellite imagery to identify surface water across landscapes. We then coupled this surface water data with drainage area, topographic confinement and stream flow data to quantify the relative influence of beaver dams on surface water, relative to modeled stream flow. Validation results show that our random forest land cover classifier performs well, with an overall accuracy of 0.86 for all land cover types (water, vegetation, bare ground), and an average consumer’s accuracy of 0.95 for surface water. Despite considerable variation in the magnitude of effect among watersheds, topographic ponding capacity and stream flow were clear drivers of surface water area. Beaver dams increased surface water area within a limited window of stream flow conditions. Notably, this window tends to correspond with mid- to late summer stream flow levels, which has important implications for late season mesic productivity. Our research offers a framework by which to explore the hydrologic and topographic conditions that moderate beavers’ influence on surface water in prairie streams, which can help target areas for beaver-based restoration, and inform expectations for when dams may increase surface water availability.