Regional Assessment of Land Use Impacts on Stream Channel Habitat in the Middle Columbia River Basin

Many human land uses and land cover modifications (e.g., logging, grazing, roads) tend to increase erosion, leading to an increase in fine sediment supplied to streams that can degrade aquatic habitat for benthic organisms. This study evaluated potential human impacts on streambed substrate in the John Day River basin (a tributary to the Columbia River) by relating stream habitat survey data from a randomized survey design (n=72) to human disturbance within the riparian zone and upstream catchment at local to whole-catchment scales inferred from field observations and remote sensing data. We assessed potential excess fine sediments using a bed stability index formulated as the log of the ratio of observed geometric mean particle diameter to the predicted critical diameter at bankfull flow. This bed stability index was significantly positively related to watershed area and estimated mean annual precipitation (p<.001) in the study area. Accounting for these natural landscape controls, our data showed no significant relationship between the bed stability index and any of the riparian or watershed human disturbance measures we tested for streams draining catchments underlain by resistant rocks. For streams with >10% of basin area underlain by more erodible bedrock, however, bed stability was a significantly negatively correlated (p<.05) with an index of local riparian human disturbance and with the frequency of upstream road-stream crossings. In general, we found only weak associations between remotely sensed land cover metrics and bed stability, possibly due to the relatively poor quality of available remotely sensed data. We examined this issue using 22 sites for which more intensive data were available. For these sites, the strength of the inverse relationship between bed stability and road-stream crossing frequency increased from very weak (R2=.09) for metrics based on the U.S. Census Bureau's TIGER/Line files to significant and much stronger (R2=0.38, p=.002) for metrics based on TIGER data updated using recent U.S. Forest Service fire road maps, and was stronger yet for the 10 sites with catchments having >10% erodible bedrock (R2=0.66).