Modelling elephant behaviour using satellite data to estimate food availability.

Vicky Boult

In the time of a changing Africa, wildlife is continually being squeezed into smaller spaces with ultimate impacts on demographics and ranging behaviour. One species in which this is apparent is the African elephant (Loxodonta africana), which once had a continental range but is now increasingly fragmented and largely in decline. As a keystone species, it is important that scientists and wildlife managers are able to make accurate predictions about how future changes to ecosystems are likely to impact elephant populations.

Elephant demographics and ranging behaviour are largely governed by access to forage. Therefore, if managers have information regarding how food availability changes through space and time, they should be able to predict how elephant populations will respond. This project makes use of remotely-sensed information on vegetation to map food availability across time and space. The resulting demographic rates and ranging behaviours of elephants are predicted using an individual-based model (IBM) approach. The IBM incorporates a newly proposed energy budget model so that variation in food availability results in differential survival, growth and reproductive rates. Complex social systems also play a large role in governing elephant ranging behaviour. As such, it will be important to include rules to simulate the social aspects of an elephant’s decision making process.

The model is being developed and tested using data from the long-running Amboseli Elephant Research Project. It is hoped that the finished model will provide a useful tool for predicting the response of an elephant population to projected changes in its ecosystem. This will be of increasing importance as Africa’s human population grows and the human-wildlife interface expands.

Click here to see the time-evolving map of vegetation in the Amboseli Ecosystem and here to see my poster at the 2016 British Ecological Society conference in Liverpool.