Livestock as sentinels for TB in possums
Farm/forest margin habitat, Karamea – Caroline Thomson
Substantial research effort over several decades has gone into improving surveillance techniques for assessing the status of TB (caused by Mycobacterium bovis) in New Zealand wildlife.
Driven by an effective collaborative feedback loop between management questions and scientific inquiry, important advances have been made in diagnostics and in understanding the behavioural ecology of host species and multi-species epidemiological dynamics (see Anderson et al. 2015 New Zealand Veterinary Journal 63). Diagnostics within an animal has progressed at three levels: gross lesion identification, culture techniques of mycobacteria, and genetic identification M. bovis strains. Possums emerged as a maintenance host, while other wildlife species were identified as spillover hosts, which do not maintain the disease but serve as useful sentinels to indicate its persistence. Research findings on movement behaviour and disease transmission rates have informed predictive modelling of TB persistence in specified wildlife control operations. All of these advances led to the increasingly frequent situation in which no TB is detected in wildlife surveys, and has created the need for a predictive tool to quantify the level of confidence in the absence of TB given no detections (Proof of Freedom Utility; Nugent 2015 in Kararehe Kino 25). The Utility now plays a key role in TBfree New Zealand’s effort to eradicate TB from the entire country by progressively declaring vector control zones (VCZ) free of the disease. Given the very large area to survey, the high cost of possum and wildlife-sentinel surveys, and severe budget limitations, the emerging need is for a low-cost and wide-spread surveillance system.
One solution to this problem is to use livestock-TB-surveillance data to make inferences on the probability of disease absence in sympatric or adjacent-living possum populations. The disease is transmitted in both directions between possums and livestock; therefore, the absence of TB in livestock tells us something about the disease status in wildlife. While TB testing of livestock is not inexpensive per se, it is conducted as part of routine herd-health management and will be done in the future. Dean Anderson has developed a user-friendly tool for TBfree New Zealand disease managers to predict the probability of disease freedom in wildlife given negative livestock surveillance data. To detect TB in an infected possum population using livestock data, the following sequence of events must occur: 1) livestock must encounter infected possums and become infected themselves, 2) infected livestock are tested for TB, and 3) the test is positive. To estimate the probability of these events, the model incorporates the spatial layout of possum habitat and farms, and the number of animals tested.
To demonstrate the model, Dean applied the Livestock as Sentinels Tool to a 10 000-ha VCZ with livestock testing data from 37 farms from 2000 to 2013. Not all farms were surveyed in any given year. If possum control had occurred in the VCZ, population density would be expected to be low, and if infection persisted, the proportion of the area with TB possums would be low. In this low-disease-prevalence situation it would be difficult to find one of the few remaining infected animals, which results in low surveillance sensitivity and low probability of eradication (black line in Fig. 1a, b). If no control had occurred, possum density and TB prevalence would be high, and the surveillance would have a high sensitivity and would likely detect TB. Given that no TB was detected, the predicted probability of eradication using livestock data alone would exceed 95% (blue line, Fig. 1a). Importantly, in this situation, wildlife control and surveillance, and the associated costs of both could be completely avoided.
In high TB-risk areas, (i.e. where TB had been previously found), possum control and surveillance would be advisable. The Livestock as Sentinels Tool can be used to supplement wildlife surveillance data to decrease the time to successfully declare eradication or reduce the amount of money spent on wildlife surveillance. In the trial VCZ, wildlife surveillance data were collected from 2006–2009, and by itself never achieved the target 95% probability of eradication (dashed line, Fig. 1b). However, the combined analysis of livestock and wildlife data exceeded the target probability (red line, Fig. 1b).
Wildlife disease management is expensive and logistically complicated. The Livestock as Sentinels Tool takes advantage of existing data to either supplement or replace direct surveillance of wildlife disease status. The only cost is the minimal time required to acquire, process and analyse the data. The potential benefits are substantial reduction in surveillance costs and reduced time to achieve a target probability of disease eradication in wildlife. While the tool was described in the context of TB in New Zealand, it could be applied to any wildlife disease that affects regularly tested livestock.
The work was funded by TBfree New Zealand.
