Factors associated with the transmission dynamics of bovine tuberculosis in New Zealand : a dissertation presented in partial fullfilment [sic] of the requirements for the degree of Doctor of Philosophy at Massey University, Institute of Veterinary, Animal and Biomedical Sciences, Massey University, Palmerston North, Turitea, New Zealand

Abstract

This thesis presents the results of a series of studies on the epidemiology of TB in brushtail possum and domestic cattle populations in New Zealand. The first set of studies provides an analysis of the results of routine TB testing carried out in the Featherston area from July 1980 to June 2004. The median annual incidence rate of TB reduced from 4.7 cases per 1000 cattle-years at risk for the period 1986 to 1991 to 1.8 cases per 1000 cattle-years at risk for the period 1992 to 2003, coincident with the use of poisoning to control possums in the surrounding forest park (a major possum habitat area). We identified clusters of cattle TB cases adjacent to the forest park and found no evidence of spatio-temporal interaction of TB risk among farms. Our findings support the hypothesis that possums living in the forest park are a source of bovine TB in this area and that farm-to-farm transmission was not an important mechanism of infection spread. A mixed-effects Poisson regression model was developed to investigate the influence of farm-level covariates on the number of cattle confirmed with TB. The model showed that, despite intensification of possum control activities, proximity to forest parks remained a significant predictor of the number of confirmed TB cases per farm per year. Our analyses identified a significant, 3-fold increase in TB risk in dairy cattle relative to beef conditional on the size of local possum habitat, and confirmed the positive influence of cattle population size and the presence of previous infection status as a determinant of the number of confirmed TB cases per farm per year. The second set of studies investigates details of capture events recorded during a longitudinal, capturemark- recapture study of possums in a 22-hectare study site near Castlepoint, from April 1989 to August 1994. Social network analyses were used to identify contact patterns and to estimate the influence of contact on R0 for bovine TB. The average number of contacts per possum ranged from 20 to 26 per year. We estimated that TB would spread if an average of between 1.94 and 1.97 infective contacts occured per year per infected possum. We evaluated the effect of sex, habitat and contact behaviour of 26 postmortem confirmed TB cases in possums with those of 104 matched controls. Unit increases in the number of infected contacts increased the odds of TB infection by 2.61 (95% CI 1.29 – 5.29, P <0.01). Our results show that individual contact behaviour is a determinant of the presence of TB foci within this population and challenge the hypothesis that contact with many individuals increases the probability of infection. A model to predict spatial variation in possum abundance was developed using a Geographic Information System. Details of possum capture events were obtained from 157 10-trap lines distributed within 42 randomly located transects at Molesworth Station. Two GIS-based models were developed to predict the number of possums caught per line using Poisson regression techniques. The first model used remotely sensed environmental data; the second used a combination of remotely sensed and fine-scale data. Both models provided adequate predictive ability with Pearson correlation coefficients greater than 60%. We conclude that the prediction maps produced from this model provide a useful decision support tool for possum control managers. These results have implications for the management of TB in this area of New Zealand, providing the information that will allow effective control activities to be applied at significantly lower cost.