Modeling elk distribution in Bandelier National Monument, New Mexico

Elk (Cervus elaphus nelsonii) are large ungulates that can have significant impacts on ecosystems. Bandelier National Monument, New Mexico, is part of the winter range for the large herd of elk that summer on the nearby Valles Caldera National Preserve. The proportion of the herd that winters on Bandelier appears to have increased after the 1977 La Mesa fire burned through the area, and created favorable forage conditions at elevations that commonly have snow conditions conducive to wintering elk. Potential damage to archeological resources and disruption of ecosystem function from use by elk is of primary concern to area resource managers.

Winter aerial surveys have shown that elk use of the area is not uniform and that some areas are used much more heavily than are others. Our objectives were to assess wintering elk population density, identify seasonal distribution, habitat associations, and movement patterns to be used as inputs for the development of a spatially explicit computer model to predict the amount and potential duration of winter elk use on Bandelier. Elk (n = 114) were captured and radio-collared in and around Bandelier, and monitored from January 2000-February 2002 by fixed-wing aircraft, with most intensive data collection in December, January, and February. We estimated winter population numbers and density distribution by mark-resight analysis using a combination of fixed-wing aircraft telemetry surveys and visual helicopter surveys. We found a correlation between snow depth in the area and the number of elk wintering on Bandelier.

We quantified habitat variables in a 300m^ grid cell overlay of the study area in a GIS environment, and assigned a Desirability Index ranking system to cells as an index to the likelihood of winter elk use. We developed a computer program that assigns elk to cells based on decision criteria supported by literature review and observed elk behavior. Initial conditions were assessed by static habitat parameters (elevation, aspect, slope, cover type, etc.) and then modified by the current values of the dynamic parameters (snow depth, amount of standing crop, number of elk present, etc.).

We found good correlation between DI score and mean number of telemetered locations per cell in all 3 winters studied. We corroborated model function through correlation of model output elk-days use to the density of elk recorded by aerial telemetry flights, concurrent helicopter visual surveys, and historic helicopter visual surveys. Model output was well correlated to telemetry data. Correlation of model output to helicopter survey data was reduced by poor sightability of elk from the helicopter, the limited number of survey flights conducted, and limitations of snow depth estimation techniques, but remained consistent under widely disparate winter conditions. Model output was most sensitive to variations in critical snow depth threshold at which elk are excluded from cells.

While model performance declined under low snow conditions, the model was flexible and dynamic and approximated the process intended. It produced results which could be corroborated quantitatively or qualitatively, can be easily analyzed and displayed, reflects current biological understanding of elk and their interaction with their environment, and can be readily modified as the knowledge base increases.