Scholarship & Research
Permanent URI for this communityhttps://scholarworks.montana.edu/handle/1/1
Browse
2 results
Search Results
Item Generalizing and transferring a GIS-based species distribution model: from one hot spot to another(Montana State University - Bozeman, College of Agriculture, 2018) Garcia Neto, Narciso; Chairperson, Graduate Committee: Clayton B. MarlowSpecies distribution models (SDMs) are efficient simulations of the distribution of species across geographical space and help to understand the spatial patterns of biological diversity. However, they are not designed to provide a description of species habitats. Geographic information systems (GIS) combined with SDMs have been used to illustrate the distribution and infer the sustainability and capability of habitats, to explore ecological relationships, serve as selection of vegetation types, avoidance of habitat disturbed by humans, establishing factors like predation, and to identify landscapes favorable for establishment of a new population. Despite the large number of SDMs papers published within the last decade, the practical utility of these models in the conservation management field remain sparse. The main objective of this research was to develop techniques for habitat modelling based on presence/availability data depicted by illustrative habitat maps and to test the new model on different landscapes. Resource selection function was used to develop a new model for the Yellowstone bison herd from published habitat maps. The predictor variables within the new model were elevation, ruggedness, profile curvature, percent of tree cove, Horizontal and vertical distance to water. The new model was then transferred and tested with field data from the National Bison Range and Grand Teton bison herds. The top predictive model performed better for the Yellowstone and Grand Teton herds than for National Bison Range herd. The output of this research indicated that habitat maps could work as source of land use by wildlife through transference to new areas of interest especially when local use data is not available.Item Predicting dominant species on grasslands at the National Bison Range, Moiese, Montana(Montana State University - Bozeman, College of Agriculture, 2014) Garcia Neto, Narciso; Chairperson, Graduate Committee: Clayton B. MarlowUnder ecologically sustainable conditions, a landscape should retain representative climax vegetation. Thus, a method to predict the climax species component of a functioning vegetation community is an important tool for restoration projects. Based on descriptions of the Palouse Prairie grassland the National Bison Range managers selected bluebunch wheatgrass, Idaho fescue, and rough fescue as target species for management and restoration objectives. An indicator called Relative Effective Annual Precipitation (REAP) was created by Montana Natural Conservation Service (NRCS) to express the amount of water available to the plants, at a specific location, taking into account precipitation, slope and aspect, and soil properties. Using Geographic Information System (GIS) and REAP as the predictor variable, a map to predict the occurrence of species within grassland communities was developed to guide restoration and management efforts on the USFWS National Bison Range. REAP values were calculated for sample sites from three earlier rangeland assessments and related to actual field measures of the target species. Classes of REAP intervals were defined to bracket the range in value for each species. Classes were also created for target groups (bluebunch and fescue) sorted by genus. REAP values for sites dominated by bluebunch wheatgrass were significantly different from values for sites dominated by Idaho fescue and rough fescue (P < 0.0001). However, there were no statistical differences between REAP values for Idaho fescue and rough fescue (P=0.989).The mean probability of the REAP model to accurately predict the occurrence individual target species was 0.55 and for the target group was 0.64. NBR and should be dominated by grasses, but there were patches of conifer forest. The values of REAP related to the forest patches were compared against REAP values for grassland areas to learn if the model could differentiate between the two major cover classes. The REAP values for the forest patches were higher than values predicted for grasslands (P=0.0026). So, prediction of areas dominated by grasslands was different from forest sites. However, the discrimination between Idaho and rough fescue was not successful.