Landscape epidemiology of hantavirus in the Atlantic Forest of Paraguay

Abstract

Hantaviruses are zoonotic, RNA viruses that are harbored by muroid rodents of the families Muridae and Cricetidae. While the virus is endemic, and mostly non-symptomatic in its rodent reservoirs, when humans contact the virus it can result in serious disease. My purpose in this dissertation is to investigate the effect that landscape patterns and land cover condition can have on pathogen prevalence in a hantavirus reservoir species (Akodon montensis) within the Atlantic Forest region of Eastern Paraguay and to investigate ways to analyze those patterns using remotely sensed data. The first component to this research is to test potential improvements to image classifications on land use/land cover classifications useful for the study of small mammal communities. An object-based classification produced the best results with seven classes: Forest, Wet Cerrado, Dry Cerrado, Latifundia, Minifundia, Dry Pasture, and Wet Pasture. The classified imagery was then used to assess landscape effects on the presence of hantaviral antibodies (a 'marker' for exposure to the virus) in populations of A. montensis. In the overall landscape, proximity of similar habitat patches was related to seroprevalence in Akodon. When considering only the forest class, high amount of forest, high number of forest patches, and high diversity in forest patch sizes were all associated with seroprevalence. Next, was an analysis of ways to distinguish understory density variables through the use of satellite imagery. Horizontal and vertical density in the understory has been associated with the presence of hantavirus in A. montensis. Vertical and horizontal density measurements were correlated with NDVI and the Fourth band in the Tasseled Cap transformation. Finally, I consider the relationship between small mammal community diversity and seroprevalence, and their association with NDVI. Diverse small mammal communities are associated with low hantavirus seroprevalence. Low diversity metrics and high hantavirus seroprevalence were associated with high mean NDVI values. Many aspects of landscape patterns are important to hantavirus seroprevalence in small mammal communities in Eastern Paraguay. Several of the landscape patterns important to hantavirus seroprevalence can be studied using satellite-derived data.