Earlier this year, Ecological Applications published one of QAEcologist Darren Southwell‘s PhD thesis chapters as an article. In it, Darren builds a model of a threatened species following mainland-island metapopulation dynamics and investigates the relative merits of adding new habitat patches, extending the area of existing patches and leaving the metapopulation alone.
It’s clear pretty early on that the best approach for population persistence depends on the colonisation rate. If the species colonises new habitat patches well, then creating new patches is worthwhile because we can trust individuals to find their way over and establish a new subpopulation. If the species disperses to new patches rarely, we may be better off expanding existing habitat patches to secure these local populations.
But what if we don’t know the species’ colonisation rate? Darren goes on to build an optimisation that allows for uncertainty in this important parameter. Even better, it factors in the potential learning opportunities that pop up when we monitor colonisation into empty habitat patches. This is adaptive management in its most quantitative form.
Darren’s use of stochastic dynamic programming and beta-binomial updating make this a great methodological companion piece to Mick McCarthy’s article on adaptive vegetation management, my PhD research on harvest management and Tracy Rout’s Honours research on threatened species translocation. It’s an elegant approach that prevents the range of learning possibilities from spiralling beyond computational bounds. (Nevertheless, the types and sizes of problems that can be addressed are quite limited.) As a co-author, I most enjoyed diving into this technical detail with Darren.
For those less excited about recursive equations, there’s a nice case study based on a Bay Checkerspot butterfly metapopulation.
Southwell D.M., Hauser C.E. & McCarthy M.A. (2016) Learning about colonization when managing metapopulations under an adaptive management framework. Ecological Applications 26: 279-294. doi