Rangelands are a major type of land found on all continents. Though they comprise around 70% of the world's land area, knowledge of rangelands is limited and immature. Rangelands supply humans with food and fiber at very low energy costs compared to cultivated lands. They are inherently heterogeneous, highly variable in time and space. Rangeland management needs to consider the impacts of long-term vegetation transition. It needs a conceptual framework defining potential vegetation communities, describing the management induced transition of one vegetation community to another, and documenting the expected benefits provided by the various potential vegetation communities. The most widely used conceptual unit in the rangeland discipline is the "ecological site". Ecological sites can be an effective unit that should respond to management consistently and can help managers understand the site's potential to meet human needs. A state and transition model (STM) brings ecological sites and their potential vegetative states together to build a conceptual framework showing the major causes of transitions between states of an ecological site and thus helping make adaptive management decisions. Within the STM there is a need for an indicator of ecosystem health. Ecosystem services can be important to evaluate alternative states. Ecosystem services do not pass through a market for valuation, though often the cost would be very high if, through mismanagement, the ecosystem is no longer capable of providing those services. Vegetation communities are constantly facing reversible or irreversible transitions triggered by natural events and/or management actions. The framework generated in this study is significant in using remote sensing to generate state and transition models for a large area and in using ecosystem services to evaluate natural and/or management induced transitions as described in the STM. This dissertation addresses the improvement of public rangelands management in the West. It applies geospatial technologies to map ecological sites and states on those sites, characterizes transitions between states and selects a desired state to manage towards based on a systematic assessment of the value of flows of environmental services. The results from this study are an evaluation of improved draft ecological site maps for a larger area using remote sensing images, a simplified state-and-transition model adapted to remote sensing capabilities to study transitions due to climatic events and management practices, and a constrained optimization model that incorporates ecosystem services and the simplified STM to evaluate management costs and conservation benefits. The study showed that brush treatment is the most effective management practice to cause state transitions. The highest increase in the high cover state was by 24%. Areas under grazing and drought show slow transitions from brush to grass and also after prescribed fire vegetation take at least two years to recover.

Nest site selection and nesting success of plains sharp-tailed grouse were examined on the Grand River National Grassland in South Dakota during the nesting season from 2009-2012. We used conditional logistic regression to assess vegetation production, ecological site description, and landscape position on nest site selection. Two competing models regarding nest site selection: top model consisted of non-native forbs and native cool-season grasses, second best model included all grass and forb. Nine ESDs were used for nesting; loamy and clayey ecological sites most frequently used and produced the highest standing crop. Most frequent observed nest site State were Annual/Pioneer Perennial and Introduced and Invaded Grass. Top model for nest daily survival rates included litter, second-best model included ESD; second-best model showed negative effect for nests initiated in thin claypan, limy backslope, and sandy ecological sites. Based on daily survival estimate and 23-day incubation period, nests were 59% successful.

Erosion and weed dominance often limit the recovery of burned piñon-juniper woodlands. Soil water repellency (SWR) is one factor that may contribute to this by increasing overland flow and impeding seedling establishment. In spite of these effects, the extent of SWR within piñon-juniper woodlands is unknown. In this study, the extent, severity and thickness of SWR were sampled across 41 1,000 m2 plots within three 2009 Utah wildfires. Predictive models of SWR were built from ecological site characteristic data collected at each site. Across the study, SWR was found at 37% of the points sampled. SWR extent was strongly related to piñon-juniper canopy cover (r2 = 0.60) and was found to be significantly higher in tree/shrub mound zones (71%) as compared to interspaces (16%). Endorsed predictive models of SWR extent and severity had R2 adj values of 0.63 and 0.61; both models included piñon-juniper canopy cover and relative humidity the month before the fire as coefficient terms. These results suggest that as piñon-juniper canopy cover increases due to infilling processes in the coming years, post-fire SWR extent and severity will increase. As the effects of a changing climate in the Intermountain West link additively with infilling processes to increase the frequency and intensity of wildfires, the net effect will be stronger SWR over a greater spatial extent. To cope with these changes, land managers can apply the predictive models developed in this study to prioritize fuel control and post-fire restoration treatments with respect to SWR.

Monitoring and identifying the state of rangelands on a landscape scale can be a time consuming process. In this thesis, remote sensing imagery has been used to show how the process of classifying different ecological sites and states can be done on a per pixel basis for a large landscape. Twenty-seven years' worth of remotely sensed imagery was collected, atmospherically corrected, and radiometrically normalized. Several vegetation indices were extracted from the imagery along with derivatives from a digital elevation model. Dominant vegetation components from five major ecological sites in Rich County, Utah, were chosen for study. The vegetation components were Aspen, Douglas-fir, Utah juniper, mountain big sagebrush, and Wyoming big sagebrush. Training sites were extracted from within map units with a majority of one of the five ecological sites. A Random Forests decision tree model was developed using an attribute table populated with spectral biophysical variables derived from the training sites. The overall out-of-bag accuracy for the Random Forests model was 97.2%. The model was then applied to the predictor spectral and biophysical variables to spatially map the five major vegetation components for all of Rich County. Each vegetation class had greater than 90% accuracies except for Utah juniper at 81%. This process is further explained in chapter 2. As a follow-on effort, we attempted to classify vegetation ecological states within a single ecological site (Wyoming big sagebrush). This was done using field data collected by previous studies as training data for all five ecological states documented for our chosen ecological site. A Maximum Likelihood classifier was applied to four years of Landsat 5 Thematic Mapper imagery to map each ecological state to pixels coincident to the map units correlated to the Wyoming big sagebrush ecological site. We used the Mahalanobis distance metric as an indicator of pixel membership to the Wyoming big sagebrush ecological site. Overall classification accuracy for the different ecological states was 64.7% for pixels with low Mahalanobis distance and less than 25% for higher distances.

This study tested the performance of classification, regression, and ordination techniques to evaluate the spatiotemporal dynamics of threats to big sagebrush ecological sites. The research was focused on invasion by annual exotic grasses and encroachment by woodlands. We sought to identify those areas that have had a persistent coverage of cheatgrass (Bromus tectorum) in big sagebrush ecological sites. We took advantage of the contrast in greenness between multi-temporal (within one year) remotely sensed vegetation indices captured in the spring and summer to find a distinct phenological signature that allowed mapping cheatgrass. We utilized support vector machines (SVM) to classify three temporal scenarios for which field data sets were available. SVM performed very well with accuracies of 70% (producer's) and 95% (user's) for the class of interest (presence of cheatgrass). This was the focus of chapter 2. In chapter 3 we report the development of vegetation continuous fields (VCF) for three years of interest 1996, 2001, and 2007 in order to detect active woodland encroachment. We prepared VCF for shrubs, trees, herbaceous vegetation, and bare ground using a suite of remotely sensed spectral reflectance, vegetation indices, and transformations. We compared the performance of multivariate regression trees (MRT) and random forests (RF) to develop the VCF multi-temporal series. RF outperformed MRT in both accuracy and ability to appropriately map the continuum of percent cover across large landscapes. We estimate that 17,570 hectares of big sagebrush lands showed encroachment by woodlands. Our goal in chapter 4 was to develop a similarity index for large rangeland landscapes. Trend assessments field sites and a long-term annual series (1984 - 2008) of remotely sensed imagery were used in conjunction with multidimensional scaling (MDS) to measure ecological distance to undesired states such as invasion by exotic annuals and encroachment by woodlands. In this chapter our units of analysis were soil-mapping units, which were predominantly composed of one ecological site (>60%). Our MDS results show that different ecological sites can be identified in the reduced MDS statistical space. The observed transitions and trajectories of mountain, Wyoming, and basin big sagebrush sites correlated well with the ecological expectation in semiarid lands. We anticipate that managers can use our protocols to update ecological site descriptions and state and transition models from a remotely sensed perspective.

Over the past 100–150 years Texas rangelands have dramatically changed from native open savannahs to dense woodlands. On the Edwards plateau, a major management concern is the increasing encroachment of Ashe juniper (Juniperus ashei). Preceding an anticipated brush management program, I investigated the presence, co-occurrence, and habitat relationships of 7 breeding bird species in the Leon River Watershed in central Texas, USA: black-capped vireo (Vireo atricapillus), golden-cheeked warbler (Dendroica chrysoparia), northern bobwhite (Colinus virginianus), white-eyed vireo (Vireo griseus), Bell’s vireo (Vireo bellii), painted bunting (Passerina ciris), and brown-headed cowbird (Molothrus ater). Vegetation characteristics were compared between sites occupied by each species and unoccupied sites using univariate analysis. Models for predicting species site occupancy were developed (using logistic regression) based on habitat characteristics correlated with the presence of each species. Two species of special concern, the endangered black-capped vireo and golden-cheeked warbler occupied 5.6% of sites and 13.8% of sites respectively, while the brood parasite brown-headed cowbird was the most widespread, occupying 86.8% of sites. Species co-occurrence patterns revealed significant associations between the golden-cheeked warbler and each of 5 other species. For most species, variables included in habitat models could be explained by knowledge of species known habitat associations. For example, the black-capped vireo was positively associated with increasing low-growing (<1.5 m) hardwood cover and with Low Stony Hill ecological sites. The golden-cheeked warbler was positively associated with increasing density of larger juniper trees, increasing variability in vertical vegetation structure, and decreasing midstory canopy of deciduous nonoaks (e.g., cedar elm [Ulmus crasifolia]). It also preferred Low Stony Hill and Steep Adobe ecological sites. Site occupancy seemed to be driven by variables that describe overall vegetation structure. In particular, cover of low-growing non-juniper vegetation and juniper tree density appeared to be important in determining site occupancy for several species. Although the models constructed were not very robust, resource managers can still benefit from such models because they provide a preliminary examination of important controlling variables. Managing rangelands to maintain or restore a mosaic of juniper patches and open shrublands are likely to help meet the habitat requirements of these bird communities.