Livingstone et al. New Zealand Veterinary Journal http://dx.doi.org/*** S1 Development of the New Zealand strategy for local eradication of tuberculosis from wildlife and livestock PG Livingstone* 1, N Hancox*, G Nugent, G Mackereth and SA Hutchings* *OSPRI New Zealand, PO Box 3412, Wellington, 6140, New Zealand Landcare Research Ltd, Lincoln, New Zealand Environmental Science & Research, Upper Hutt, New Zealand Author for correspondence. Email: Paul.Livingstone@tbfree.org.nz Supplementary Information 1: Approach and assumptions used to predict outcomes under the No Control option. Modelling approach A spreadsheet deterministic model was developed in 2006 by AHB (Livingstone et al. 2008) to predict the rate of VRA expansion and infection consequences for cattle and deer should TB testing and TB-related possum control cease after 2010. It assumed that stopping possum control would allow possum densities to increase, with a resulting increase in the number of tuberculous possums within VRA. In the absence of control measures, tuberculous possums would also spread to adjacent parts of VFA, thus effectively expanding the VRA. The predicted rate of VRA expansion and implications for herds are identified in the key assumptions below. The modelled annual expansion of VRA was mapped and geospatially combined with mapped locations of dairy, beef and deer herds. This enabled the number and type of herds potentially exposed to tuberculous possums each year to be predicted for each annulus of tuberculous possum expansion into the VFA. Based on increasing exposure to tuberculous possums over time, the model then predicted the number of new infected herds, and the number of possum-related infected cattle and deer associated with those herds. The number of infected herds within the expanded VRA was calculated based on the vector-related herd incidence rate for the particular year after commencement of the model run. The number of possum-related infected TB cattle and deer was calculated based on the annual animal incidence rate set for that year. To this was added the number of cattle and deer that became infected from contact with infectious herd mates. This enabled annual calculation of the total number of infected cattle and deer herds, and total numbers of individual TB cattle and deer. 1 The content of this supplementary information has not been edited. All risk and liability rest with the authors.
Livingstone et al. New Zealand Veterinary Journal http://dx.doi.org/*** S2 Key assumptions No nationally organised possum control and herd testing programme, no records of animals and herds tested or their TB status, no records of TB wildlife presence, and no legally enforceable powers to manage TB testing, infected herds or TB in wildlife Tuberculous possums spread from the existing (2007) VRA boundaries at an initial rate of 0.4km/year, increasing to a maximum rate of 4km/year in year 10 (Livingstone et al. 2015) which then stabilises. The initial low rate of spread was designed to take account of the very low possum densities expected in most VRA boundary areas at July 2010. Over the following 10 years with no formal control, possum densities were postulated to increase, leading to an increased rate of spread. The rate of spread is conservative, as it is likely that the expanding tuberculous possum fronts would encounter moderate to high possum densities in previously uncontrolled VFA well within the ten year period, which would be expected to amplify disease expansion. The initial tuberculous possum-related annual herd incidence rate for exposed herds is set initially at 0.15% (compared with actual rates of 0.04% and 0.6% in VFA and VRA respectively in 2007). The annual herd incidence rate within VRA is assumed to increase over a 20 year period to 6% per year, this being the maximum annual herd incidence rate observed in VRA during 1993 to 1996 when there was limited possum control, but herds were under annual TB test. The herd annual incidence rate then remained at 6% Each herd contained the average number of dairy, beef and deer animals per herd for the respective herd types in VRA in 2006 The initial number of tuberculous possum-related infected cattle and deer in newly infected herds was set initially to 1 and increases over 10 years to an annual incidence rate of 6%, this being equivalent to 60% of the mean TB incidence rate observed in cattle herds that were under an intense programme of TB testing and no possum control in the Buller South area in 1970 and 1971 (Livingstone et al. 2015). Herd TB testing would be voluntary and paid for directly by farmers, with no legal requirement to slaughter TB reactors, no compensation payments for TB reactors and no quarantine conditions imposed on infected herds Three percent of infected herds will clear infection per year as a result of TB testing, but would be liable to become re-infected following contact with tuberculous possums
Livingstone et al. New Zealand Veterinary Journal http://dx.doi.org/*** S3 TB animals not removed from herds become infectious to herd mates within two years. The daily rate of infection is based on the equation N x β, where N = the number of susceptible animals in the herd, and β is the transmission coefficient = 2.79 x 10-5 (Barlow et al. 1997). A maximum of 20% of susceptible animals can become infected in an infected herd per year Fourteen percent of animals in each herd would be culled each year and replaced by TB free animals.
Livingstone et al. New Zealand Veterinary Journal http://dx.doi.org/*** S4 Supplementary Information 2. Parameters for the Modified No Control Modelling Option Modelling of the No Control option predicted extremely high livestock infection rates which did not reflect the reality of how New Zealand farmers and industry would respond to the lack of a formal TB control programme. Therefore the assumptions of the No Control option were modified to include some possum control and TB testing in order to provide what was considered a more realistic baseline (the Modified No Control option). The Modified No Control option used the same modelling approach described in Appendix 1 for modelling the expansion of VRA and determining the number of possum-related herd and animal infections, but a number of the key assumptions were modified and new assumptions added as follows: All herds in which TB was found would be tested annually at the farmers expense That $20m/year ($8m from government and $12m from farmers and local government) would be spent on formal TB-related possum control (37% of 2008 vector control expenditure) Possum control funding would be insufficient to achieve eradication of TB from possum populations or stop the spread of tuberculous possums into the VFA Possum control funding would be sufficient to reduce the rate of spread of tuberculous possums to half the annual rate used in the No Control option. Once the maximum annual dispersal rate of 4km/year is reached, it is then maintained Tuberculous possum-related annual herd incidence rate begins at 0.15% and slowly increases to 6% (incidence rate observed during the period 1993/94 1995/96) over a 30 year period (two-thirds the annual rate of increase used in the No Control option) and remains at this level Possum control reduces the tuberculous possum-related infectious contact rate. As a consequence the number of M. bovis-infected livestock starts at 1 per infected herd and slowly increases such that after 10 years, it reaches 3% (approximately 25% of the TB incidence rate observed in the Buller South area of the West Coast under intensive testing and no possum control in 1970 and 1971) (Livingstone et al. 2015) and then remains at this level Initially 30% of infected herds clear infection through TB testing, the rate reducing to 14% after 13 years. The reduction is related to the inability to maintain low possum numbers on infected properties, such that increasingly they remain infected Four percent of animals will be sold annually from herds in VFA and VRA TB could spread between herds through movement of infected animals.
Livingstone et al. New Zealand Veterinary Journal http://dx.doi.org/*** S5 Supplementary Information 3. Parameters for the Ad Hoc Control modelling Option The Modified No Control option was considered by some stakeholders to still be too conservative in its assessment of the TB control effort that would be implemented in the absence of a formal regulated programme. In response, a further scenario the Ad Hoc option - was developed, which included a lesser amount of formal possum control funding, but an increase in the level of farmer funded TB testing. The Ad Hoc option used the modelling approach described in Appendix 1 for modelling the expansion of VRA and determining the number of possum-related herd and animal infections. Assumptions for the Ad Hoc option are the same as the Modified No Control except where modified as follows: All infected herds to be tested annually until the herd tests clear All dairy herds to be tested annually if located in areas where tuberculous possums had been identified or suspected (areas would be poorly defined given the lack of a formal reporting and recording scheme), or triennially Initially only 25% of beef and deer herds located in areas where tuberculous possums had been identified or suspected would be TB tested, increasing to 90% within 15 years In areas where tuberculous possums had not been identified or suspected, initially only 5% of beef and deer herds would be TB tested, increasing to 40% within 15 years Due to the increasing proportion of TB cattle and deer moving between herds there would be a corresponding increase in pre-movement TB testing of cattle and deer, such that within 15 years, pre-movement testing would be required for 90% of animals moving from areas where tuberculous possums had been identified or suspected, and for 25% of animals moving in other areas That $12m/year (22% of 2008 vector control expenditure) would be spent on TB-related possum control Possum control funding would be insufficient to achieve eradication of TB from possum populations or stop the spread of tuberculous possums into the VFA Possum control reduces the tuberculous possum-related infectious contact rate. As a consequence the number of M. bovis-infected livestock starts at 1 per infected herd and slowly increases such that after 20 years, it reaches 4.5% per year (approximately 33% of the TB incidence rate observed in the Buller South area of the West Coast under intensive testing and no possum control in 1970 and 1971) (Livingstone et al. 2015) and remains at this level.
Livingstone et al. New Zealand Veterinary Journal http://dx.doi.org/*** S6 Supplementary Information 4. Scenario modelling 2008. Scenario modelling can be used to explore a wider range of scenario variants than can be possible with the (effectively) single variant forecasting models used previously to evaluate TB management strategies. The interactive model employed in this exercise was designed to link scenarios for expenditure on possum control (with input variability related to control frequency, type and the extent or depth of control into forests) to the long-run outcomes in terms of average possum density and TB levels in possums. These in turn were linked to the levels of infection in livestock. The interactive model simplistically assumed that each scenario was applied indefinitely (except for eradication scenarios), and predicted the long-run steady state equilibrium outcomes under those assumptions. Although not explicitly spatial, the interacting model simulated each VRA and its outer buffer and inner core separately, and then summed the outcomes to produce annual predictions of national control costs, numbers of infected herds, prevalence of herd infection, together with the number of infected cattle and deer for the life of each option simulated. Prompted by funder requests for evaluation of more detailed and complex scenarios than could easily be addressed by the forecasting models above, a deterministic spreadsheet model was developed in 2008 (Nugent et al. 2008). This was used to compare a number of alternative scenarios that spanned a wide range of expenditure levels and strategic approaches (Livingstone et al. 2008). The interactive model was developed to provide a steady-state whole-of-vra view of TB in possums under various levels of control and the consequential impacts on the number of infected herds and TB cattle and deer within the VRA. Under the interactive model, expenditure on possum control was related to the control effort required for each of the different control scenarios. Based on average possum densities, frequency of control and assessed likely results, the distribution of surviving tuberculous possums was simulated using a negative binomial distribution. Thus the model was able to evaluate relationships between expenditure on possum control (variability related to control frequency, type (ground/aerial) and extent of control into forests), the number of tuberculous possums and number of possum-related infections in cattle and deer herds and animals in the VRA. The model was able to treat each VRA and its outer buffer and inner core individually. Each of these areas was further categorised as being subject to ground or aerial control. The total area (in ha) was available for each category. The model outputs provided annual estimates of control costs, number of infected herds, and prevalence of herd infection, together with the number
Livingstone et al. New Zealand Veterinary Journal http://dx.doi.org/*** S7 of infected cattle and deer for the life of the proposed TB programme, for the selected options modelled. The interactive model incorporated the following components: An input component, in which a range of possum-control strategies were specified in terms of the type (aerial or ground), frequency (annual vs biennial vs triennial), cost ($/ha per operation), location (VRA and location within VRA), and scheduling (when control was first implemented) A possum density component in which the density of possums for particular areas was predicted based on the type and intensity of control applied, and the elapsed time since most recent application of control A TB-possum component, where the prevalence of TB in the possum population was predicted given current density and previous control A TB-cattle component in which the number of infected herds was predicted given the number of infected possum in the area. The interactive model assumed that: New Zealand was categorised into five major VRA (Central North Island, Southern North Island, Northern South Island, West Coast-Tasman and Southern South Island), forecast to be present in 2010 Each VRA was comprised of an outer buffer and an inner core Tuberculous possums were contained by VRA boundaries through maintaining low possum densities in the buffers or by use of major geographical barriers to their movement TB could not persist in the possum population below a minimal threshold of 5% RTCI (assumed to equate to 1 possum per ha) The maximum TB prevalence of 1% occurred once the possum population density reached a RTCI of 25%, or 5 possums per ha A tuberculous possum survived for one year Tuberculous possums migrated from forest to farmland based on outputs derived from the spatial possum model (Ramsey and Efford 2005; Barron et al. 2015) Tuberculous possums were distributed proportionately to the area (in ha) with and without cattle and deer, enabling the mean number of tuberculous possums per farm to be estimated The frequency distribution of tuberculous possums per farm was simulated assuming a negative binomial frequency distribution which achieved a high level of clustering to emulate spread of infection between possums
Livingstone et al. New Zealand Veterinary Journal http://dx.doi.org/*** S8 The likelihood of at least one animal in a herd becoming infected per year was calculated from the number of infected possums per farm using the equation (1-(1-p) N TB-possums ) where p was the probability of any one possum transmitting TB to livestock assumed to be 0.2 Based on this calculation, the number of new possum-related transmissions of infection per year was estimated There was a lag of 1.5 years in detecting infection in livestock and in clearing infection once detected The period prevalence of TB in livestock was calculated from the number of herd breakdowns adjusted upward to allow for the lag in detecting and clearing infection from livestock The number of infected herds identified in the VRA was used to calculate the number of infected herds in the VFA based on the ratio of infected herds in VRA : infected herds in VFA in 2008/09, which was 3:1. Thus 25% of New Zealand s infected herds will be found in VFA representing the spread of infection through infected cattle and deer movements, largely from VRA.
Livingstone et al. New Zealand Veterinary Journal http://dx.doi.org/*** S9 References cited in Supplementary Information Barlow ND, Kean JM, Hickling G, Livingstone PG, Robson AB. A simulation model for the spread of bovine tuberculosis within New Zealand cattle herds. Preventive Veterinary Medicine 32: 57 75, 1997 Barron MC, Tompkins DM, Ramsey DSL, Bosson MAJ. The role of multiple wildlife hosts in the persistence and spread of bovine tuberculosis in New Zealand. New Zealand Veterinary Journal, doi: 10.1080/00480169.2014.968229, 2015. *Livingstone P, Bosson M, Knowles G, Crews K, Braaksma N, Nugent G, Barron M, Mackereth G. Response options analysis: Designing containment buffers for the geographic containment of TB to existing vector risk areas. http://www.tbfree.org.nz/portals/0/2014augresearchpapers/designing%20containment%20buffers%20 for%20the%20geographic%20containment%20of%20tb%20to%20existing%20vector%20risk%20area s.%20animal%20health%20board%20(wellington)%20internal%20report.%2015p,%202008.pdf (accessed June 2 2014). Review of the Bovine TB National Pest Management Strategy: A response options analysis report. Animal Health Board, Wellington, New Zealand, 2008 Livingstone PG, Nugent G, de Lisle GW, Hancox N. Toward eradication: The effect of Mycobacterium bovis infection in wildlife on the evolution and future direction of bovine tuberculosis management in New Zealand. New Zealand Veterinary Journal doi:10.1080/00480169.2014.971082, 2015. *Nugent G, Barron M, Livingstone P, Braaksma N, Mackereth G. Response options analysis: options for modelling herd prevalence and cost of TB strategy options. http://www.tbfree.org.nz/portals/0/2014augresearchpapers/bovine%20tb%20national%20pest%20m anagement%20strategy%20response%20options%20analysis%20response%20options%20modelling. %20A%20report%20prepared%20for%20the%20Animal%20Health%20Board,%20Wellington,%20Pp %209,%202008%20.pdf (accessed May 30 2014). Review of the Bovine TB National Pest Management Strategy: A response options analysis report. Animal Health Board, Wellington, New Zealand, 2008 *Ramsey DSL, Efford M. Eliminating Tb results from a spatially explicit, stochastic model. http://www.tbfree.org.nz/portals/0/2014augresearchpapers/ramsey%20dsl,%20efford%20m.%20eli minating%20tb%20-%20results%20from%20a%20spatially%20explicit,%20stochastic%20model.pdf (accessed May 2 2014). Animal Health Board project no. R-10619. Landcare Research contract report LC0405/118 to the Animal Health Board, Wellington, New Zealand, 2005.