Moving forward: insights and applications of moving‐habitat models for climate change ecology

Harsch, M. A., Phillips, A., Zhou, Y., Leung, M. R., Rinnan, D. S., & Kot, M. (2017). Moving forward: insights and applications of moving‐habitat models for climate change ecology. Journal of Ecology.

The threats to biodiversity are varied (e.g. habitat loss, climate change), and the responses of populations and species to these threats are difficult to predict. There are multiple possible responses, including range shifts, changes in phenology, acclimation or adaptation, and extinction; how each species reacts to a given change is determined by a combination of biotic interactions, genetic differences, and abiotic conditions (e.g. variation in warming). These many interacting components mean that predicting population responses to global change is difficult. In this paper, Harsch et al. (2017) review current approaches and outline a new generalizable framework for prediction of population responses to shifting habitats.

Currently, four main types of models are used to assess and predict species’ vulnerability to global change. Species distribution models relate current distributions to environmental variables, and in doing so consider habitat static and usually ignore biological interactions, which makes prediction using these models difficult. Reaction-diffusion equations focus on dispersal, which incorporates movement of organisms, and can easily incorporate field data. However, they assume Gaussian dispersal kernels, which ignores the potential long-distance dispersal that may be important under habitat changes. Individual-based models simulate individuals over various environments and interactions, making them flexible, but they are computationally intensive and sometimes difficult to implement. The authors propose use of integrodifference equations, which are simple models that track population densities over continuous space in discrete time.

This model, the “moving-habitat model,” tracks population densities over a continuous space (a line segment) in discrete time. In each time step, this homogeneous habitat moves by a fixed amount, which is referred to as the velocity of climate change. To parameterize the model, one must: (1) estimate the dispersal kernel (usually using a known or accepted distribution for a given species), (2) estimate a growth function, (3) estimate habitat size (either from environmental metrics or patch size), and (4) estimate the velocity of climate change. In this paper, the authors fit the velocity of climate change a free parameter, allowing them to explore how a population will respond to different rates of climate change. In doing so, they show that there is a critical rate of climate change at which a population can no longer track its habitat and goes extinct. This rate depends on both the habitat size and the dispersal kernel, where species with larger habitats and greater dispersal distances can tolerate faster rates of climate change.

Harsch et al. also show two extensions of their basic model. First, they model habitat in two dimensions, where instead of a line, habitat is a rectangle that shifts a certain amount in a given direction during each time step. In doing so, they show that the optimal habitat shape depends on the dispersal kernel shape. Specifically, long corridors are better for populations with platykurtic dispersal kernels (which have a large number of individuals dispersing short distances and very thin, narrow tails), whereas a species with a leptokurtic dispersal kernel would be better served by a wide corridor. However, as the rate of climate change increases, the opposite direction becomes important for each species. Second, they incorporate life history structure into this model, showing (intuitively) that dispersal becomes more important than longevity with increasing rates of climate change. Finally, the authors outline four priorities for future research using this framework: Allee effects, asymmetric shift rates between upper and lower range margins, biotic interactions (i.e, competition and facilitation), and infectious disease dynamics.