A New Method for Quantifying Uncertainty for Landscape Genetics Studies
In the past few years there have been some very nice advances in methods for estimating relationships between landcover and connectivity using landsdape genetics (Shirk et al. 2010, Dudaniec et al. 2013, Peterman et al. 2014). These methods are potentially powerful for conservation because they allow spatial predictions of genetic connectivity in relation to landcover to be made. Hence, we should be able to predict the effects of landscape change and conservation interventions on genetic connectivity using these methods. But, there may be considerable uncertainties associated with the parameter estimates that link landcover and connectivity (Graves et al. 2013). To date it has been impractical to estimate uncertainties associated with these parameter estimates in large landscapes because of computational limitations. Yet, estimates of uncertainty are crucial for using these models for conservation decisions.
In a recent paper of ours (Dudaniec et al. 2016) we develop a new method for estimating uncertainty in parameters that link landcover to genetic connectivity. We use a constrained grid search method for parameter esitmation and an information-theoretic appoach to quantify uncertainty in parameters. We apply this approach successfully to three marsupial species in Queensland, Australia and simulations demonstrate that the approach captures actual parameter uncertainty well in the vast majority of cases. Although we haven’t yet applied this to any conservation problems, it makes an important contribution to being able to capture uncertainty in landscape genetic models and hence improving the quality of information for conservation purposes.
Graphical representation of uncertainty. Bar heights represent support for different parameter values.