Wednesday, 26 September 2012

Decision tools for reintroduction but who actually uses them?

My post today was going to be a summary of a paper by Adam Schapaugh and Andrew Tyre of the University of Nebraska-Lincoln on Markov decision processes (MDPs). These enable conservation decision-making by dictating what action should be taken based on the state a system is in, and incorporates a reward for having taken the action. Schapaugh and Tyre have used as one of their examples, an hypothetical reintroduction to demonstrate this. MDPs require the user to describe the state variables, set what actions are associated with all the combinations of state variables, and construct a reward system that means the actions can be optimised to create the best outcome. In the paper this means that a set of state variables (e.g. source population size) affect which actions (e.g. capture and release) are undertaken to produce a target population of a given size through reintroduction.

Schapaugh and Tyre present a way of selecting the most relevant variables to include in order to make the best informed choices using MDPs. Their algorithm chooses the actions that maximises rewards where the rewards are set by the user to match their targets (e.g. creating a population of the target species). The actions which reap largest rewards are selected first; the actions which provide negligible or no reward are removed from the process. There is obvious application to species recovery programmes as a decision tool for when and how to act and this paper seems to present an approach for streamlining this potentially complex process.

I wish I was in a position to comment on the utility of their approach and critique it in a way that would be useful to conservation translocations. However, I am new to MDPs and instead would like to pose the question: who uses MDPs in the real world? This then leads me to ask: how do they (the practitioners) make the jump from algorithms in a paper to policy and implementation? And finally, how long does it take for advances such as the one presented in this paper to be absorbed into practice?

Feel free to comment - the questions above are not rhetorical and are a genuine attempt to understand the use of MDPs in relation to translocations.

Schapaugh, A. W., & Tyre, A. J. (2012). A simple method for dealing with large state spaces. Methods in Ecology and Evolution. doi:10.1111/j.2041-210X.2012.00242.x

No comments:

Post a Comment