Project Overview: Biodiversity loss is a critical global challenge, driven by climate change, land-use transformation, and human disturbance. Predicting how ecological communities respond to these pressures is a complex problem due to the dynamic and nonlinear interactions among species and their environments. This PhD project, hosted at University College London, aims to revolutionize biodiversity forecasting by integrating artificial intelligence (AI) and network science with traditional ecosystem modelling.

Research Focus: The project will develop AI-integrated network ecosystem models that learn, represent, and forecast biodiversity change. Biodiversity dynamics are conceptualized as a spatiotemporal network problem, where species, their interactions, and environmental drivers form a dynamic graph evolving under anthropogenic pressure. The student will utilize large-scale biodiversity sensor networks (camera-trap and acoustic arrays), Earth observation, and environmental data to construct multi-layer species interaction networks, capturing community structure and change over time.

Methodology: The research framework consists of three layers: (1) Data layer—harmonizing heterogeneous ecological data into a spatiotemporal event graph; (2) Interaction layer—applying graph neural networks (GNNs) and probabilistic modelling to infer dynamic species–species and species–environment links, regularized by ecological priors; (3) Process layer—integrating learned networks into mechanistic ecosystem models to simulate dispersal, resource use, and competition, producing calibrated forecasts of biodiversity change under various scenarios.

Case Studies & Impact: The project will leverage long-term datasets from ecosystems under strong human pressure, such as the Maasai Mara (Kenya) and the Terai Arc (Nepal). These sites provide rich data for quantifying network reconfiguration, predicting biodiversity outcomes under land-use or climate scenarios, and assessing conservation and ecosystem resilience implications. The research sits at the intersection of ecology, AI, and environmental data science, offering transformative advances in predictive biodiversity science and actionable conservation forecasts.

Training & Opportunities: The student will receive comprehensive training in applied AI, biodiversity, and transferable professional and research skills. The project includes a placement with an AI-INTERVENE partner (3–18 months) and opportunities to present at national and international conferences, positioning the student at the forefront of the discipline and enhancing future employment prospects.

Eligibility: Suitable candidates will have a degree in ecology, engineering, environmental, physical, or computer science, with experience in computing and mathematics. The interdisciplinary supervisory team will support the student in developing expertise in ecological theory, computational modelling, and geospatial data analysis.

Funding: The studentship is fully funded by the AI-INTERVENE NERC Doctoral Focal Award, subject to a competitive selection process.

Application Deadline: January 19, 2026.

References: Emergent Global Patterns of Ecosystem Structure and Function from a Mechanistic General Ecosystem Model; Heterogeneous graph neural networks for species distribution modelling.

How to Apply: Apply via the University College London or University of Reading portal as indicated in the project listing. Prepare a CV and cover letter detailing relevant experience. For further information, contact the supervisory team.