Should I spread my risk or concentrate my efforts: Is triage of a subpopulation ever the best decision?

Threatened species often exist in a small number of isolated subpopulations. Given limitations on conservation spending, we must ask the question: should we put all our eggs in one basket and manage the best quality subpopulation or the subpopulation most likely to benefit from management, or should we hedge our bets and manage both subpopulations? A further complexity arises when we consider that most threatened species are cryptic and their presence in an area can be uncertain as a result of the imperfect nature of most detection methods. Managers of cryptic species thus face several dilemmas: if they are unsure whether a species is present in an area or has been extirpated, should they continue to manage the species in that area or instead invest some of their limited resources in surveying to determine if the species is still present (and viable)? How much negative evidence do they need in order to give up and make the decision to cease management? The ecology and conservation literature present little guidance on how to approach such problems, though some analogous problems have been tackled within a decision theory framework (Gerber et al. 2005; Regan et al. 2006; Wilson et al. 2006). Here we build on lessons from these studies and others investigating optimal conservation decision making (Possingham et al. 2001; Dorazio & Johnson 2003) to develop a coherent decision framework for allocating resources between two subpopulations of a threatened species where we are uncertain about the persistence of the species in our management areas. In this problem we must make a decision about how to allocate finite resources to three separate actions in each subpopulation; management, surveying and doing nothing. Management reduces a subpopulation's risk of extinction. Surveying, while not reducing extinction risk, improves our knowledge about whether the species is present, therefore avoiding costly unnecessary expenditure. Both management and surveying cost money and thus the decision to perform either of these actions in a subpopulation will alter the resources available and therefore the success of the action implemented in the other subpopulation. At any point in time managers will have a belief about whether a subpopulation is still extant. In this paper we assess how our optimal decisions change as a function of those beliefs and the time remaining in the management period. The goal of efficient conservation planning and management is to find a resource allocation strategy, or set of actions, that maximises the net expected long-term benefit. Here the optimal strategy involves a trade-off between the persistence of our subpopulations at the end of the management period, and the impact of our decisions on the probability of subpopulation extinction. We pose this problem as a Partially Observable Markov Decision Process (POMDP) and solve a multi time-step scenario using the incremental pruning algorithm (Cassandra et al. 1997). The POMDP algorithm finds an optimal resource allocation each year given the current belief about the state of the species (extant or extinct) in each subpopulation. This paper has two major aims; (i) to extend the framework proposed by Chades et al. (in review) to incorporate two subpopulations of a threatened species, addressing the issue of triage in conservation management, and (ii) to introduce more ecological complexity and realism to the problem by considering subpopulationws of differing habitat quality