Assessing ecosystem health in order to advise policy makers and direct conservation efforts requires us to understand historical, current and future biodiversity changes and their causes. The biodiversity simulator, EcoSISTEM [1, 2], that the supervisory team have developed with their collaborators, can mathematically model hundreds of thousands of species responding to reconstructions of the global climate at a local, national or continental scale and presents an opportunity to understand how biodiversity responds both to ongoing climate and other environmental changes as well as directed mitigation efforts. The simulator is, however, limited in how it can learn from recorded biodiversity data, with the complex mathematical model making traditional statistical inference extremely difficult.
This interdisciplinary project will combine statistics with mathematical biology to develop efficient statistical methods that can learn the parameters of the mathematical models from real biodiversity data. The complexity of EcoSISTEM prevents the use of traditional statistical techniques, so instead the focus is on developing computational techniques which bypass the need for a formal statistical model. The project will use Approximate Bayesian Computation, e.g. [3,4], to run steadily more relevant simulations to provide a distribution of relevant parameters representative of the ecosystems being modelled. This will be combined with the mathematical knowledge of the model dynamics to provide an early stopping system [5] which allow the simulations to focus on the relevant parameter combinations. The project will then look at how statistical data techniques can be used to link missing, biased, and incomplete data to the complex dynamics represented within the simulator.
Project Benefits and Opportunities
Conferences – The project has funding to attend projects across ecology, mathematics, and statistics.
Collaborative Visits – Funding is provided to make multiple trips to visit collaborators at Biomathematics and Statistics Scotland and the Natural History Museum, London.
Centre for Mathematics Applied to the Life Sciences – Opportunity to engage with mathematical life science modellers across Glasgow.
Analytics for Digital Earth – Be part of a new and growing theme within the school with workshop and networking events.
Computational Biology Community – Join a cross-school interdisciplinary community with regular seminars, a yearly conference, and leadership opportunities.
Boyd Orr Centre for Population and Ecosystem Health – Join the University of Glasgow-hosted multi-disciplinary centre with a particular focus on integrating mathematical modelling with empirical data.
Additional Training
APTS – Attend a residential training programme for statistical Ph.D. students.
SMSTC – Option to take relevant mathematical courses.
Data Analytics – Option to take courses from our online Data Analytics MSc.
Application
Please send your CV and a short description of why you would like to do the project through the enquiry system below confirming that you are eligible for UK funding.