Predicting distribution changes of a mire ecosystem under future climates
Mire ecosystems are threatened by global climate change but have important roles in biodiversity conservation, carbon storage, landscape-scale hydrological function and in providing ecosystem services. We aimed to: (1) estimate change in areas environmentally suitable for mires under future climates; (2) evaluate the sensitivities of projected change to uncertainties in future climate and model structure; (3) evaluate the effect of global mitigation actions on distribution change; (4) identify potential climate refuges for future adaptation actions.
We developed and evaluated correlative bioclimatic models for an Australian mire ecosystem by: (1) selecting environmental predictors representing ecological processes that mediate ecosystem occurrence and dynamics; (2) using a high-performance modelling algorithm; (3) quantifying predictive performance by cross-validation; (4) cross-checking responses to predictor variables between different algorithms; (5) comparing the modelled responses with expected mechanistic responses; (6) evaluating extrapolation risks by quantifying the deviation between future and current environmental domains of the study area and by assessing the temporal constancy of correlations between variables; (7) using a geographically stratified cross-validation to verify spatial consistency of the model; and (8) quantifying the robustness of predictions of climate change impacts to uncertainty in both climate and ecological models.
All combinations of global circulation models and distribution model projected declines of at least 30% in both area and suitability of environments for the mire ecosystem and in projecting a contraction of range to the southwest. We identified a likely refuge in the south of the distribution and two less certain, emerging areas of suitable environment west and south of the current distribution.
We conclude that southern mire ecosystems are highly susceptible to climate change. Our approach will be useful for the prediction of climate impacts on other ecosystems for which there is enough knowledge to map distributions and develop plausible hypotheses about environmental factors that influence them.