Persistently highest risk areas for hantavirus pulmonary syndrome: Potential sites for refugia

Interannual variation in the number of cases of human disease caused by hantaviruses in North America has been hypothesized to reflect environmental changes that influence rodent reservoir populations. This hypothesis postulates that when cases are rare reservoir populations are geographically restricted in patches of suitable habitat. Identifying these sites, which is needed to test the hypothesis, has proven to be a challenge. Satellite imagery of the U.S. Southwest has shown associations among the likelihood of human hantaviral disease and increases in the rodent populations, as well as increased prevalence of Sin Nombre virus (SNV) in rodent populations. In this study we characterize local areas that had environmental signatures that persisted as predicted highest risk sites for human disease through much of the 1990s. These areas represent a small percentage (0.3%) of the region. Exploratory analyses indicate that these areas were not randomly distributed, but were associated with certain landscape characteristics. Characteristics of elevation, slope, aspect, and land cover were associated with persistent high risk. Using multivariate Poisson regression to control for confounding effects, sites with deciduous- or mixed-forest land cover on moderate to steep slopes (>5°) above 2130 m elevation were associated with increasing numbers of years at highest risk. These are candidate locations for refugia. Sites associated with cleared ground or shrubland were less often associated with high risk compared to reference conditions. The seasonal patterns of vegetation growth in persistently high-risk areas were compared to matched locations using MODIS (moderate resolution imaging spectroradiometer) NDVI (normalized difference vegetation index) during a time of a severe drought in the region from 2002 to 2004. Despite the drought and regardless of land cover, the NDVI in persistently highest risk areas had an early onset, with significantly higher levels of green vegetation that lasted longer than at comparable sites. These observations identify locations that can be monitored for the abundance of P. maniculatus and presence of SNV. If these sites are refugia, we predict they will be occupied by infected deer mice when other monitored sites are unoccupied.