Is it Possible to Model the Probability of where African Elephants, Loxodonta africana, are Likely to Raid Farms?

Abstract

The African elephant is the largest terrestrial mammal on the planet. With elephants being so large they have huge dietary requirements (not only volume but also quality) and huge ranges to meet this demand. After elephant numbers plummeted due to poaching, Kenya banned all forms of hunting in 1977. As both elephant and human populations and territories have increased, the negative interactions between them have also increased – known as Human-Elephant Conflict (HEC). HEC is mainly in the form of crop raiding and when they do raid they can put a farmer in financial and sustenance crisis. This project compares the outputs generated by Generalised Linear Models (GLMs), MaxLike, and MaxENT to assess where could be vulnerable to future elephant raids of farms during the different seasons. This project finds that GLM’s tend to create an ‘overgeneralised’ output and that MaxENT creates an under-representation. MaxLike is deemed unsuitable for this type of modelling. There is huge scope for the continuation of this kind of work especially to assess the accuracy of a model where GLM’s and MaxENT are merged as well as using agent based modelling to better assess where elephants are likely to raid farms.