Rabbit haemorrhagic disease in New Zealand: field test of a disease–host model

Abstract

An earlier published model, parameterised from qualitative data from Europe and detailed observations of the first simple epidemic in Australia, gave a good fit to longer-term (3 year) data from New Zealand, without any re-tuning of parameters. Changing some of the unknown disease parameters further improved the model's fit, but two problems remained. Firstly, predicted proportions seropositive are too high if rabbit densities are as low as observed, and if the proportions seropositive are correct then the predicted densities are too high. Secondly, observed rabbit densities do not show obvious peaks of recruitment, as predicted by the model with and without RHD. Possible reasons for these discrepancies are suggested, and initial trials with the model suggested that a novel transmission mechanism involving both direct (rabbit to rabbit) and indirect (via free-living virus) transmission may help explain both high suppression and low antibody levels. The main conclusions from the original model remain unaffected by its testing against new data, namely: rabbit populations are likely to be suppressed in the long term by about 75%; the pattern of epidemics is determined largely by intrinsic disease behaviour rather than seasonality, though the latter may tune this to some extent; there tend to be yearly epidemics; percentages of rabbits infected at any one time are low (around 5%) but this does not imply low impact; maternal antibodies have little effect on RHD dynamics; RHD may persist in low-density as well as high-density populations but give less suppression; and additional control may eradicate disease, at least temporarily.