PREDICTIVE GIS MODELING OF RARE PLANT OCCURRENCE

The differential success of plant species across heterogeneous landscapes is of potential value in predicting the locations of previously unidentified populations of plants. Species in the mustard family (Brassicacea) are known for rarity, occurring locally in patches and often isolated. Ecological genetic studies of the wild upland mustard species Boechera stricta in the northern Black Hills of South Dakota indicate that local range may be constrained because allocation to chemical defenses inhibits tolerance to the stressful environment beyond patch boundaries. Thus, B. stricta is particularly sensitive to environmental factors that determine range and represents an ideal system to develop efficient GIS models for locating additional populations.

Based on potential environmental factors that are correlated with boundaries of known B. stricta patches, we hypothesize that elevation, soil type, surrounding vegetation, and slope gradient are the most important environmental controlling factors on the success of B. stricta. A weighted overlay analysis of these four factors was used to build a predictive model of B. stricta occurrence. Known B. stricta localities in the northern Black Hills exist at elevations between 1648 m and 1770 m. Data obtained from the USDA soils layer and from soil analysis from selected sites indicate that soils in which B. stricta occurs are largely silt loam to sandy loam, which were in turn given the highest rank in the model. Black Hills National Forest vegetation data, along with field observations, show that B. stricta commonly grows near aspen and paper birch and less frequently near ponderosa pine. These vegetation types were ranked accordingly in the model. Slope gradient of known B. stricta localities ranges from 4% to 25%, and therefore slopes outside of this range were excluded as potential B. stricta habitat.

Based on these four environmental factors, 374 km² were identified as optimal B. stricta habitat from within the total study area of 1347 km². Accessibility to field sites is a consideration, so a 0.8 km buffer was placed on roads and hiking trails, which further reduced the search area to 327 km². In the next phase of research, these areas will be inspected for the presence of B. stricta. Additional iterations of the model will be completed as new field data are gathered.