Tree suitable habitats and colonization potential dynamics at range boundaries under climate change - an integrated modelling perspective

Background/Question/Methods

Changes in climate forecast for this century is likely to cause shifts in suitable habitats for various tree species including the creation of some novel assemblages of community types. These habitats will be differentially colonized by the species, based on landscape heterogeneity and species interactions. Our integrated multi-stage approach consists of using decision tree ensembles to predict current and future suitable habitats of 134 tree species in the eastern US, with climate, soil and elevation variables as predictors (DISTRIB). The colonization potential of suitable habitats is estimated by using a spatially explicit cellular model called SHIFT. It uses a fat-tailed dispersal kernel mimicking historical migration rates, thereby enabling source regions to differentially colonize suitable fragmented sink habitats, revealing dispersal constraints beyond the current species boundary. Combining DISTRIB and SHIFT enables us to assess colonization potential under future climates and explore suitable migration pathways. Biological and disturbance factors not considered by DISTRIB and SHIFT can modify final outcomes, and are hard to model. These are separately evaluated using a weighted decision support system (ModFacs) that examines 9 biological and 12 disturbance factors to distill DISTRIB+SHIFT results in order to facilitate communication of model uncertainty and likely outcomes.

Results/Conclusions

Several groups of tree species were examined using our integrated multi-stage approach to spatially evaluate the nature of habitat variability as well as colonization potential near range margins. Predominantly climate-driven species (eg. oaks) can extend their suitable habitats northward under warming climate. Using our integrated approach, we  show a) how the carbon emission pathway humanity adopts can affect future species habitat availability,  b) that the likelihood of oaks colonizing the large northern habitats available to them in the future, is low due to dispersal constraints c) what mitigating factors can operate for vulnerable northern species, like aspen-birch and spruce-fir which are predicted to lose suitable habitats in the eastern US, and d) how DISTRIB+SHIFT can be used to evaluate suitable patches and corridors to facilitate migration of vulnerable species. Our macro-scale, integrated modelling approach is one among many needed to uncover the multi-faceted and dynamic effects of the rapidly increasing anthropogenic impacts on tree habitats.