Date of Graduation
2015
Document Type
Dissertation
Degree Type
PhD
College
Davis College of Agriculture, Natural Resources and Design
Department
Forest Resource Management
Committee Chair
James A Thompson
Committee Co-Chair
Louis McDonald
Committee Member
Brenden McNeil
Committee Member
James Rentch
Abstract
Soils form the dynamic interface of many processes key to the function of terrestrial ecosystems. Many soil properties both influence and are influenced by activity of flora and fauna. Interactions between soils, biota, and climate determine the potential ecosystem services that a given unique ecological site (ES) can support, and how resilient a site is to various pressures and disturbances. Soil data are needed to fully understand how these factors interact, but because this data is difficult to obtain, existing soil maps are sometimes not detailed enough to fully explore relationships. Environmental raster GIS data layers were used to increase the detail of maps by representing soil forming factors and associated ecological pedomemory legacies important to understanding ecological potential. This dissertation presents methods and tools to help create these new soil maps at appropriate resolution and theme for field scale assessment of ecological sites that enable land managers to plan and implement appropriate management decisions.;USDA-NRCS soil surveys were disaggregated to higher resolution maps using a semi-automated expert training routine to implement a random forest classification model. This transformed soil map polygons of variable thematic and spatial resolution (soil map unit concepts) to a consistent 30-meter raster grid of unified theme (soil taxa). Disaggregated maps (DM) showed highly variable accuracy (25--75% overall validation accuracy) that mirrored that of the original soil surveys evaluated in Arizona (AZ) and West Virginia (WV). However, disaggregated maps expressed the soil data at a much more detailed spatial scale with a more interpretable legend. The WV surveys exhibited much lower accuracy than the AZ survey evaluated. This lower accuracy in WV is likely due to the forested setting and highly dissected landscape, two factors that create more intrinsic soil variability that is harder to explain with spatial covariates.;Ecological site descriptions (ESD) document soil-ecosystem groups that produce unique amounts and types of biological constituents and respond similarly to disturbance and environmental variation. ESD are linked to soil map unit components in USDA-NRCS soil surveys and are used as the basis for land management planning on rangelands and forestlands. The component level connection makes DM a good way to spatialize ESD because both are spatially represented at the same thematic level, whereas conventional soil maps have polygons that often have multiple components linked to a delineation.;However, in the evaluation of mapping ESD via DM, the DM turned out not to document the key difference in spodic soil properties that distinguished the important ecotone between northern hardwood and alpine red spruce conifer ESDs in Pocahontas and Randolph counties, WV. So, to adjust, spodic soil properties were mapped directly using digital soil mapping approaches. A strong spatial model of spodic soil morphology presence was developed from a random forest probability model and showed correspondence to red spruce and hemlock occurrences in local historic land deed witness trees from records between 1752 and 1899. From this result, areas with spodic soil properties were assumed to be associated with historic red spruce communities, although 68% of those areas in the WV study area are currently under hardwood cover. This would seem to indicate that hardwoods have encroached on the historic extent of spruce, which is consistent with other recent studies. O-horizon thickness was also observed to be one cm thicker for every 10% greater importance value of red spruce or hemlock versus that of hardwood species at field sites. From these observations, it was calculated conservatively that at least 3.74-6.62 Tg of C have likely been lost from red spruce influenced ecological sites in WV due to historic disturbance related conversions of forest to hardwood composition. These results highlight the value of working within a soil-ecological factorial framework (e.g. an ESD) to contextualize land management options and potential derived services or negative consequences of each available action.
Recommended Citation
Nauman, Travis, "Spatializing the Soil-Ecological Factorial: Data Driven Integrated Land Management Tools" (2015). Graduate Theses, Dissertations, and Problem Reports. 6299.
https://researchrepository.wvu.edu/etd/6299