Modelling bird migration

An inter-disciplinary approach to mechanistic modelling of migration patterns

This project arose at the Santa Fe Institute Complex Systems Summer School, in collaboration with Dr. Marius Somveille from the University of Oxford (now at Yale and the Max Planck Institute for Ornithology).

The work was recently published in Nature Scientific Reports. All simulation code from the project is available Here.

We set out to understand the factors influencing bird migration in the hope that we could ultimately predict pathways and distribution patterns using only environmental data.

To do this we created an "environmental potential" landscape directly from environmental data, and allowed birds to move as particles within this landscape. Using a combination of wind and chlorophyll density data, with only 2 unknown parameters, we were able to demonstrate a close fit between simulated pathways from breeding colonies in the Southern Ocean to known dispersal patterns for Black-Browed Albatrosses, with best-fit parameters conserved across breeding colony locations.

Shown above is the predicted dispersal pattern for 16 individuals from the breeding colony on South Georgia and the Sandwich Islands.