An Integrated GIS Model of Bovine Tuberculosis in a Population of African Buffalo (Syncerus caffer): Exploring Management Options in Kruger National Park, South AfricaEPA Grant Number: FP916382
Title: An Integrated GIS Model of Bovine Tuberculosis in a Population of African Buffalo (Syncerus caffer): Exploring Management Options in Kruger National Park, South Africa
Investigators: Ryan, Sadie J.
Institution: University of California - Berkeley
EPA Project Officer: Jones, Brandon
Project Period: January 1, 2004 through December 31, 2004
Project Amount: $104,772
RFA: STAR Graduate Fellowships (2004) RFA Text | Recipients Lists
Research Category: Fellowship - Terrestrial Ecology and Ecosystems , Academic Fellowships , Ecological Indicators/Assessment/Restoration
The objective of this project is to create a geographic information system (GIS) to project model simulations of management strategies for a bovine tuberculosis (BTB) epidemic in African buffalo (Syncerus caffer) in Kruger National Park, South Africa. This GIS will elucidate the spatial characteristics of the epidemic through a spatially explicit individual-based model (IBM) comprising three main modeling components: demography, age-structured disease transmission, and a habitat-dependent herd movement model. Management strategies, such as culling or vaccination, will be modeled as modifications of these components and the results demonstrated in a GIS.
Ongoing ecological analyses of tracking data, long-term demographic data, and remotely sensed vegetation data will yield parameter estimates for the overarching model of the system. A spatially explicit model of the system comprising 19,000 cells, each representing 1 km2 of Kruger National Park will be set up in the C+ language. The structure will be based on an IBM, with each herd on the landscape representing an individual. I will use cellular automata theory to describe local movements of herds within a small neighborhood of cells, in which the movement will be decomposed to a vector and a speed. The rules of movement will be determined by the ‘attractiveness’ of neighboring cells, based on the habitat quality derived from the results of ongoing habitat selection analyses, distance to water, geological substrate, and occupancy by another herd. The individual herds will experience demographic events (births, deaths, and aging) and infection (mass action transmission) through a straightforward stage-based model and a proportion of immigration and emigration as a result of fission/fusion and individual dispersal events. To compliment and enhance the utility of extant spatial data on buffalo distributions from census datasets, digitized routes, topological hydrology maps, prior landscape classifications, and vegetation assessments, I am using remotely sensed data, from the National Oceanic and Atmospheric Administration’s Advanced Very High Resolution Radiometer (AVHRR) to quantify vegetation and moisture availability, using the Normalized Differential Vegetation Index (NDVI) algorithm at a 1 km resolution to explore the interplay of dynamics between buffalo and their mixed savanna ecosystem. I anticipate that the model time step should be on the order of available habitat data—the 10-day increment of the NDVI data. To simulate a 10-year projection, this will mean approximately 360 time steps. Whereas this seems computationally intensive, each simulation only requires loops across each herd, not each cell, and a single habitat update. Management strategies for BTB in this population will be modeled as modifications of model components. For example, a vaccination strategy might target a specific age group or specific herds in the population. This would modify the infection model component, and the result of its effect would be compared to a no-vaccine model, and the distribution of infection among herds can be demonstrated visually in the GIS. Similarly, a fence could prevent infected herds from moving into areas of no infection and could be modeled as habitat modification. Cells in the model at the border of the fence would have no attraction, or simply prevent movement. Directed culling of individuals or herds would modify both the spatial distribution of herds and the demographic component of the model. Management strategy results then will be reprojected in the GIS and presented to managers.