Quantifying the global human-wildlife interface

Project overview

This project will develop a high-resolution mapping framework to analyse human-wildlife interactions within increasingly anthropogenic landscapes. As natural habitats are altered by human activities, from agriculture to urban development, traditional land-use models fail to capture the full complexity of these shared spaces.

Using machine learning, this project will classify remote-sensed land-use data into detailed “bioanthromes” categories that reflect both human and wildlife use. These bioanthromes will consider agricultural systems, habitat connectivity, and landscape dynamics, providing a fine-scale perspective on how human and animal interactions are structured in these transformed environments.

The student will then use regression models to understand how various mammal and bird species interact with these bioanthromes, to uncover patterns of habitat use within modified landscapes. Human mobility data will also be integrated to explore how people occupy and move through these spaces, enhancing predictions about ecological connectivity and human-wildlife interactions.

This research provides a globally relevant, data-driven approach for mapping critical human-wildlife interactions at fine scales. Its insights will support conservation, biodiversity monitoring, and land-use management, equipping stakeholders with tools to better understand and address the complexities of anthropogenic landscapes.

Project Specific Training

The student undertaking this PhD project will receive project-specific training in advanced data science techniques, with an emphasis on big data handling, artificial intelligence (AI), and geospatial statistics.

Training in big data and AI methodologies will be delivered primarily through one-to-one instruction from the supervisory team, focusing on the use of machine learning models, e.g. Convolutional Neural Networks (CNNs) and knowledge-guided AI. This training will enable the student to develop models for complex classification of remote-sensed land-use data and predict high-risk human-animal contact zones.

For geospatial statistics, the student will work closely with geospatial experts on the supervisory team, gaining hands-on experience in processing and analysing high-resolution spatial datasets. This training will include working with satellite and mobility data, applying Bayesian regression models, and developing spatial kernels for mapping human and wildlife behaviours.