The complexities of costing eradications: a reply to Donlan & Wilcox


M. de L. Brooke, Department of Zoology, University of Cambridge, Downing Street, Cambridge, CB2 3EJ, UK.

We thank Donlan & Wilcox (2007) for their interest in our study (Martins et al., 2006). We would like to begin our reply by re-iterating that the object of our work was not to provide an estimate of the cost of proposed programmes accurate to the nearest dollar – the error terms are too great for that to be realistic – but to provide an estimate sufficiently robust to feed into an assessment of the prioritized ranking of proposed eradication programmes. This ranking requires a system for assessing not only the costs but also the conservation benefit of projected programmes, a topic on which we are currently working.

That observation leads to our one profound disagreement with Donlan & Wilcox, who believe that ‘until such (detailed cost) data is available, prioritization of invasive eradications should proceed without input costs’. In contrast, we concur with Naidoo et al. (2006) in their eloquent championing of the case for integrating economic costs into conservation planning, even if, as here, the cost data remain coarse. In our particular case, we reported that the slope of log cost of eradication on log island area is significantly <1, in fact 0.77. If the benefit of planned programmes scales with an area at a slope greater than 0.77, then the greatest benefit/cost is achieved by targeting large islands, and if benefit scales with an area at a slope <0.77, then managers should preferentially target small islands. Without considering costs, the conservation community simply cannot make the crucial decision whether to concentrate its limited eradication resources on smaller or larger islands.

Turning to the particulars of the Donlan & Wilcox critique, we do not dispute that the costs of eradication include both fixed and variable costs. The former is of course why the cost of eradicating from an imaginary island without area is greater than zero. The latter are not trivial but again we emphasize that our aim was to step back from the particular to view matters from a wider perspective. Where variable costs result in an eradication costing an unexpectedly large amount, such as Anacapa, this does not undermine our costs model – which is simply an empirically derived GLM based on available data – but, rather, raises questions about whether the conservation gains justified the exceptional costs in the example they cite.

While, as Donlan & Wilcox point out, helicopter broadcasting of bait is likely to require higher application rates than a hand-distribution bait-station approach, our limited data on the two approaches do not indicate that one is more or less costly per unit area than the other. However, the practicalities dictate that aerial broadcasting of bait is the method of choice for islands larger than a few square kilometres.

If costs per unit area have decreased over time, it is likely to be due to efficiency gains rather than due to the passage of time per se. Strictly our analysis did not distinguish the two. Donlan & Wilcox suggest that the efficiency gains over time are not linear. And indeed they are not linear in our model because of the log-transformation of the cost variable. The curve of cost versus year shows a decelerating decline as is typical of efficiency savings generally.

Transferring those likely efficiency gains to new systems may be difficult but it is certainly preferable to ignoring the accumulation of eradication expertise altogether. In fact we would argue that transfer is facilitated because most major projects use consultant expertise, equipment and even bait from the world leaders in the technology, generally from New Zealand, Australia and the United States.

Donlan & Wilcox make a valid point about the statistical problems that result from the regrettably small sample sizes for ungulates and cats. A consequence of this was that we were not able to formally consider interaction terms in the model, such as an area × taxon interaction. Their Fig. 1 is highly suggestive of such an interaction but the issue will only be clarified by further data.

None of our comments in any way detract from our belief, shared with Donlan & Wilcox, that substantial refinement of our analysis is entirely desirable. Such refinement will depend on further cost data from eradication projects entering the public domain. It will then help ensure that the conservation community delivers to posterity as many islands as possible that are both of high conservation value and free of alien vertebrates.