Improving forecasting and management of fish stocks and forest pests using artificial intelligence and machine learning

Supervisors: Prof Justin Travis, Dr Wei Pang

Project description:

Biodiversity provides a wide range of ecosystem services and disservices. Managing these biodiversity-provisioned services effectively is a major challenge, especially in an era of rapid environmental changes. Underpinning successful future management will be a new generation of ecological models that simulate complex systems with greater realism and that take advantage of the emergence of big environmental and ecological data that can be used for parameterising, fitting, validating and sequentially updating these models (Urban et al. 2016). However, we currently lack effective approaches for integrating these new models and the ‘big data’ in a robust management context. Machine learning approaches have begun to be applied, while artificially intelligent management systems are just beginning to be considered in the context of managing complex ecological systems (Lindkvist et al. 2017).  The aim of this project is to develop a framework for exploiting both artificial intelligence and machine learning to make more effective use of ecological software and big data in managing vital resources.

As case studies for developing and testing the approaches the project will use fish stocks (a major ecosystem services) and forest diseases and pests (key ecosystem disservices). Largely because fisheries and forests are major industries in the UK the data availability is unusually large so they both provide ideal systems in which to develop novel approaches for applying artificial intelligence and machine learning. The project will utilise RangeShifter (Bocedi et al. 2014), a state-of-the-art ecological modelling platform developed by the Travis group, and will include using supervised and unsupervised learning approaches as well as reinforcement learning which is especially applicable in the context of adaptive management.  However, the student will be free to explore using approaches from across the diverse range rapidly emerging in both AI and machine learning.

This project offers a fantastic opportunity for a student to gain training in an emerging field, leading to excellent future employment opportunities in academia and/or industry.  They will benefit from being embedded in two groups, one at the forefront of developing models for ecological forecasting and management and the other in using machine learning and artificial intelligence across a range of topics. One supervisor (Justin Travis) is heavily involved in Masters training in Ecology and the other (Wei Pang) co-ordinates a Masters’ programme in Artificial Intelligence. Furthermore, the student will be closely linked with Marine Scotland and Forest Research who, in both cases, already work with the RangeShifter software and have access to extensive data sets. The student will spend time at both Marine Scotland and Forest Research gaining experience of the management and policy arenas within which their work will have impact.


Urban MC, 20 authors, Travis JMJ. 2016. Improving the forecast for biodiversity under climate change. Science 353: aad8466. 

Lindkvist E, Ekeberg O, Norberg J. 2017. Strategies for sustainable management of renewable resources under environmental change. Proc. Roy. Soc. B 284: 20162762

Bocedi G, Palmer SCF, Pe’er G, Watts K, Matsinos Y, Travis JMJ. 2014. RangeShifter: a platform for modelling spatial eco-evolutionary dynamics and species’ responses to environmental changes. Methods in Ecology and Evolution, 5: 388-396.


For information about eligibility, how to apply, documentation, etc., please see here.

Please apply for this project via this link by the 4 July 2018.