Developing Statistical Learning Methods for a Charismatic Tropical Ecology System

We invite applications for an exciting PhD opportunity to develop and apply cutting-edge statistical and machine learning approaches to better understand the complex dynamics of ant-following bird systems in African ecosystems. These unique systems offer an incredible chance to explore interactions between species, including the behaviour of birds that follow army ants (Dorylus sp.) to feed on insects flushed out by the ants’ movements.

Dorylus driver ants are keystone species in African rainforests – they profoundly alter ecosystems and provide a bounty of prey for birds, chimpanzees, pangolins and myriad arthropods. African rainforests are quickly being cut, and fragmented by roads, but we don’t at all understand how much space these keystone ants (and their dependent birds) need to sustain their populations, nor do we understand if roads create barriers to their movement. Ongoing empirical research in Equatorial guinea has been mapping Dorylus colony movements and attaching tiny GPS units to ant-following birds, as well as performing transects across different road types to understand barriers to movement. Given the empirical data we now have, a great opportunity exists to combine empirical data with cutting edge modelling techniques to determine how these keystone animals–and their dependent species–move about in disturbed landscapes. With a fully fitted model that realistically describes how these animals move, we will be in a strong position to design protected areas systems and ecological restoration projects that conserve fully intact ecosystems. While ant-following bird systems have been a focus of some ecological research, this project aims to push boundaries by applying innovative statistical models and machine learning techniques to investigate species interactions, movement patterns, and population dynamics. This project will take an interdisciplinary approach, combining statistics, ecology, and computer science.

The precise direction of the PhD will be agreed by the student with the supervisory team. 

Some possible components include:

·     Agent-Based Models: Develop and refine agent-based models (ABMs) to simulate bird and ant behaviour, exploring the drivers of movement, foraging efficiency, and interspecies interactions.

·     Mark-Recapture Analysis: Implement rigorous statistical techniques, including advanced mark-recapture models, to assess population dynamics, survival rates, and movement patterns of ant-following birds across various African landscapes.

·     Machine Learning for Visual Learning: Employ computer vision and machine learning algorithms to automate and improve the accuracy of identifying bird species and their behaviour from field data, such as camera traps, and describing ant raids using video footage.

Methodological Focus:

The project will emphasize rigorous statistical modelling, integrating classical approaches with contemporary machine learning methods. This PhD will contribute to advancing ecological research by refining tools for complex systems. This research aims to build on existing knowledge, applying modern quantitative methods integrated with innovative fieldwork from our collaborators to enhance our understanding of these systems.

Skills Required:

Strong background in statistics, mathematics, or quantitative ecology, in particular, Bayesian statistics would be beneficial.

Experience or interest in machine learning and/or computer vision techniques.

Programming skills (e.g., R, Python, or Julia) for data analysis and model development.

Knowledge or strong interest in ecology, particularly bird, arthropod or animal space use, is advantageous but not essential.

What We Offer:

This PhD offers the opportunity to join a vibrant research community at the School of Mathematics and Statistics, University of St Andrews. The successful candidate would also be a member of the world-leading Centre for Research into Ecological and Environmental Modelling (CREEM). The project is the result of a brand new collaboration between the modelling team at St Andrews, Dr Fergus J Chadwick and Dr Ben Swallow, and fieldwork team at Cibio/Biopolis in Portugal, led by Dr Luke L. Powell. Training in advanced statistical methods and machine learning will be provided, along with support for visiting the field team, publication, and conference attendance.