Space-time clustering of cowpox virus infection in wild rodent populations

1. The spatial and temporal scale at which infectious animals pose a transmission risk to others may be inferred by a study of space-time interaction in the clustering of cases of infection. 2. We analyse the distribution of bank voles and wood mice infectious with cowpox virus, using 7 years of monthly live-trapping data from a woodland in north-west England. 3. A published method, K-function analysis, is adapted to measure space-time interaction among infectious animals trapped on a grid system, where the populations at risk may vary in density. 4. We hypothesized that the risk of cowpox transmission from an infectious animal would be contained within a temporal scale of one infectious period and a spatial scale of one home-range diameter. 5. Cowpox cases among wood mice showed significantly greater spatial clustering than the population at risk at the smallest spatial scale, corresponding to animals caught in the same location. At larger scales, and for bank voles at all scales, cowpox cases showed no spatial clustering beyond that expected from an association with host distribution. 6. Cowpox cases among wood mice showed strong evidence of temporal clustering beyond that of the host population, indicating short-lived outbreaks. No such effect was apparent among bank voles. 7. Significant space-time interaction among cowpox cases was detected at the hypothesized spatial and temporal scales in both host species. This suggests that the risk of transmission is concentrated locally. 8. Results are discussed in terms of transmission, including consideration of the appropriateness of distance and nearest-neighbour based clustering tests to infections transmitted by local density and frequency dependent mechanisms.