Disease, predation and demography: Assessing the impacts of bovine tuberculosis on African buffalo by monitoring at individual and population levels

1. Understanding the effects of disease is critical to determining appropriate management responses, but estimating those effects in wildlife species is challenging. We used bovine tuberculosis (BTB) in the African buffalo Syncerus caffer population of Kruger National Park, South Africa, as a case study to highlight the issues associated with estimating chronic disease effects in a long-lived host. 2. We used known and radiocollared buffalo, aerial census data, and a natural gradient in pathogen prevalence to investigate if: (i) at the individual level, BTB infection reduces reproduction; (ii) BTB infection increases vulnerability to predation; and (iii) at the population level, increased BTB prevalence causes reduced population growth. 3. There was only a marginal reduction in calving success associated with BTB infection, as indexed by the probability of sighting a known adult female with or without a calf (P = 0·065). 4. Since 1991, BTB prevalence increased from 27 to 45% in the southern region and from 4 to 28% in the central region of Kruger National Park. The prevalence in the northern regions was only 1·5% in 1998. Buffalo population growth rates, however, were neither statistically different among regions nor declining over time. 5. Lions Panthera leo did not appear to preferentially kill test-positive buffalo. The best (Akaike's Information Criterion corrected for small sample size) AICc model with BTB as a covariate [exp(β) = 0·49; 95% CI = (0·24-1·02)] suggested that the mortality hazard for positive individuals was no greater than for test-negative individuals. 6. Synthesis and applications. Test accuracy, time-varying disease status, and movement among populations are some of the issues that make the detection of chronic disease impacts challenging. For these reasons, the demographic impacts of bovine tuberculosis in the Kruger National Park remain undetectable despite 6 years of study on known individuals and 40 years of population counts. However, the rainfall and forage conditions during this study were relatively good and the impacts of many chronic diseases may be a non-linear function of environmental conditions such that they are only detectable in stressful periods.