EXCUTIVE SUMMARY
1. Two non-spatial models have been developed for RCD in rabbits: a model of intermediate complexity, with simplified rabbit population dynamics, now submitted and accepted for publication in an international journal; and a model with detailed rabbit population dynamics, completed but not fully analysed or published. The former includes a submodel for a simple epidemic (ie without the rabbit population dynamics).
2. The model of intermediate complexity led to increased understanding about the epidemiology of disease. Specifically:
a) the final outcome of an epidemic (proportion surviving and the proportion of these immune) largely depends on the basic reproductive rate of the disease;
b) the speed of the epidemic depends both on the disease reproductive rate and the decay rate of free-living virus in the environment (the higher the reproductive rate and the decay rate, the more rapid the epidemic - the Gum Creek epidemic required that virus has a half-life of 1 day or less); and
c) persistence of disease depends on the decay rate of free-living virus but in the opposite way - persistence requires a half-life of at least 14 days. Hence the apparent irreconcilability of rapid epidemics and disease persistence, discussed in Barlow (1995, 1996).
3. Since this work was done, two mechanisms have been found which allow both the intense initial RCD epidemics observed, and local persistence of the disease as a classical biological control. Persistent, classical biological control, therefore, now appears to be more likely than suggested in the earlier reports.
4. One of these mechanisms is a novel model involving multiple disease transmission pathways. Direct rabbit-to- rabbit transmission gives rapid epidemics, while transmission via free-living virus in the environment or in vectors, with a virus half-life in the order of 2 weeks, gives disease persistence between epidemics.
5. The second mechanism arises from new data, applied to the original model with indirect transmission (i.e. through free-living virus). The latest serology results suggest that most, if not all, rabbits surviving an epidemic are recovered and immune, as opposed to initial indications that about 40% were still susceptible. This allows the possibility of a higher disease transmission rate and basic reproductive rate. This in turn allows a rapid and intense initial epidemic, which would otherwise be precluded by the long (2 week) free-living virus half-life necessary for disease persistence.
6. Even if RCD does not persist locally, global persistence may be possible through a 'hide-and-seek' effect, whereby the disease dies out in one place but is naturally re-introduced through dispersal or vectors once the population has recovered. Preliminary work with a spatial cellular automaton model for RCD suggested that the disease would not persist in this way, at least if it were spread through rabbit dispersal. Essentially, the disease acted too quickly relative to the recovery rate of the rabbits: by the time a local population had recovered, RCD was sufficiently far away that the chances of re-infection were negligible. Such a result may change if vectors are as important as they now appear to be, since the distances covered would be significantly greater. On the other hand, this would offset or negate the spatial 'hide-and-seek' effect, which depends on maintenance of spatial heterogeneity in disease prevalence. Extremely mobile vectors would tend to homogenise the environment in terms of the disease reservoir, and make the spatial model equivalent to the homogenous-mixing, non-spatial model described here, but applied to a geographically larger rabbit population with gross epidemic behaviour the average of many smaller, partially synchronised epidemics. Applied in this way, such a model would clearly need to be both fitted and validated at this larger spatial scale.
7. The new models giving RCD persistence and intense initial epidemics, predicted subsequent yearly epidemics in summer or autumn and sustained suppression of Otago rabbit populations which depended on the disease parameters but was frequently in the range of 75 95%.
8. In the above situation, repeat epidemics involved extremely low prevalences of the disease (< 5% rabbits infected at any one time), in spite of the fact that the disease was giving up to 95% control of the rabbits. This is because the disease tends to drive the rabbits down to a level close to the threshold for its own persistence. The practical implication is that low counts of dead rabbits in a post-RCD introduction situation convey no information about the effectiveness of the disease.
9. Allowing for seasonal variation in disease transmission, to represent the effect of vectors, gave little change in the pattern of epidemics or degree of rabbit suppression. The only effect was to shift the timing of the yearly epidemic slightly closer to the time of peak transmission. Thus the seasonal pattern of RCD appears to be determined at least as much by the intrinsic dynamics of the rabbit/RCD system as by seasonality in vector activity. These dynamics primarily involve the recovery rate of the susceptible rabbit population to the threshold density for an epidemic, and need to be taken into account when seeking correlations between RCD epidemics and climate. In the model, disease released at the time of minimum transmission (winter) was apparently as effective as disease released at any other time.
10. If RCD persists as a classical biological control but gives insufficient suppression of rabbits on its own, the question arises as to whether supplementary control is desirable. The model suggests that such control may have adverse consequences that are not intuitively obvious. In particular, the additional control may reduce rabbit densities to below the threshold for RCD persistence and the disease may die out. Clearly a number of options can be explored, such as a smaller supplementary kill, or re-inoculation of RCD once densities have built up to RCD- sustaining but non-damaging levels. Such specific evaluations will be most profitably carried out once more is known about the disease and the models better parameterised.
11. One of the new mechanisms identified for RCD persistence (high disease reproductive rate and few susceptibles surviving an epidemic) also allows its persistence in low-density lowland populations, though with the degree of suppression reduced from 95% to around 50%. This is true even for the simplest model with the disease contact rate linearly related to rabbit density, but additionally, and as stressed in Barlow (1996), it is possible that other disease scenarios which give disease persistence in high-density populations (eg mixed transmission) may also give persistence in low-density populations if the relationship between contact rate (number of potentially infectious contacts made per infectious rabbit per unit time) and rabbit density is non-linear (ie contacts do not decline proportionally as density declines).
12. If RCD needs to be reintroduced to replace or augment existing disease, its action is likely to be density- dependent: greater percent kills would be expected at higher rabbit densities than at low ones, compared with 1080 poisoning in which the percent kill should be independent of rabbit density. This is because of secondary cycling of disease once it is introduced - the natural epidemic resulting from rabbit-to-rabbit transmission. However, the initial knockdown from baiting with RCD, and the mortality resulting from rabbits infected directly from the bait, will be density-independent like any other control. If secondary cycling of disease occurs (and this can be tested by marking baits with rhodamine and counting dead rabbits with and without dye), then it follows that the best way of integrating RCD and conventional poisoning is to bait with RCD first then follow up with a poisoning operation once the initial RCD epidemic wanes. However, this should not be done if there is a prospect of the RCD persisting naturally on-site, because reducing the rabbit density too much can jeopardise survival of the disease (as above).
13. Monitoring of rabbit numbers (e.g. spotlight counts) in New Zealand should be carried out at least monthly during times when RCD epidemics are likely, if the data are to be useful in understanding RCD epidemiology. The relatively long intervals between rabbit counts in many of the Australian intensive monitoring sites, and the necessary absence of control populations, makes it difficult to determine the contribution of RCD to observed rabbit declines, or the intensity of such declines in relation to density.
Contact for Enquiries
Manager, Strategic Science Team
MAF Biosecurity New Zealand
PO Box 2526
Wellington
NEW ZEALAND
Phone: +64 4 894 0115
Fax: +64 4 894 0731
Contact this person
