Field sampling data have been linked to sampling locations in the GIS and we are working our UNR collaborators to determine what relationships exist between rodent population numbers, infection status, ecosystem parameters, and time. The results of year 1 sampling indicate the relative density of deer mice was lower for salt desert and bare areas, and while there are no indications of a continuous relationship between relative density and prevalence of infection, regions with lower rodent densities do have much lower probabilities of infection. These findings are similar to those of Mills et al. (1997) who document that deer mice numbers and infection rates were only restricted at very high altitudes or in desert conditions.
Field observations indicate that the mapped distributions of salt desert in the GIS encompassed a rather wide range of vegetative cover densities. Based on this observation we found that a significant improvement in differentiating relative densities of deer mice could be made by intersecting the mapped distribution of salt desert with the NDVI data on vegetation abundance. Results show promise that a combination of map and image data could serve to estimate the controls on rodent populations and hence disease transmission in the host species.
During year one we have also started the compilation of map and image datasets for the entire Great Basin study area. Maps of vegetation communities are available, and Landsat imagery has been acquired to provide exhaustive coverage. In addition, we have compiled maps of census tracts for the entire Great Basin into the GIS and have linked a number of socio-economic datasets to these maps. The census is expected to provide useful indicators of exposure risk such as rural/urban population numbers, employment categories, and dwelling characteristics.
Spatial analysis of field data on relative densities and infection rates in deer mouse populations was performed using semivariograms. The data indicated a clustering of population numbers and infection rates at two different scales. At a broad scale these variables were determined in part by the distribution of appropriate habitat (i.e. not salt desert or bare). At a finer scale, clustering was apparent over smaller distances in the areas of suitable habitat. By plotting squared differences in values over distance, semivariograms provide information on spatial autocorrelation in datasets. In this case this would indicate the area over which clusters of infection extended. This analysis indicates that spatial clustering of infection rates in the study area typically extended over a range of less than one kilometer.
A prototype program for the cellular automata model of disease spread and recurrence has been developed. The program is written in C, and works on grids representing landscape parameters which will eventually be derived from the GIS. Year 2 field sampling efforts will focus on developing the method for parametrizing rate of spread estimates. These activities will be required by the cellular automata modeling effort to provide useful information on the spatial and temporal dynamics of Hantavirus in the natural environment. Field sampling design is being based on the aforementioned semi-variogram analysis with a clustered approach designed to capture transient events over a distance of a kilometer.
Further information on this project may be found at the Desert Research Institute's Hantavirus Web Page.
Hanta Project Introduction
Last updated: 27 Mar 2000