Bayesian Ecosystem and Natural Capital Models to Understand the Effect of Offshore Renewables in the Marine System

Modelling Spatial and Temporal Species Dynamics in Response to Changes in Climate and Productivity in the North Sea

Neda Trifonova, University of Aberdeen

There is about to be an abrupt step-change in the use of our coastal seas, specifically by the addition of large-scale offshore renewable developments to combat climate change. New modelling frameworks are required to aid site determination and purpose and to estimate the effects of different uses on our natural resources. Being able to forecast the ecological benefits and trade-offs that will occur with the operation of offshore renewable devices and future climate change is vital for the sustainable management of all uses of our marine ecosystems. Understanding how usage of spatial habitat of marine species may change with climate change and offshore renewable devices is challenging but of key importance and essential for the sustainable management of their populations. In this study, computational Bayesian ecosystem models will be applied to provide indications of how ecosystems are likely to change. Bayesian networks are models that graphically and probabilistically represent dependency relationships among variables. They can integrate physical and biological variables presented at different scales. In addition, Bayesian networks integrate the uncertainty associated with species dynamics due to the action of multiple driving factors. In the initial stages of this study, optimization techniques (“search-score”) have been applied to find the data-driven interactions among a set of physical and biological variables within three spatial areas of the North Sea: northern, southern North Sea and west of Scotland. These data-driven dependencies were found over different temporal and seasonal windows. Specifically, a hill-climb technique with random restart, was used to find the level of confidence of each relationship throughout space and time. In this way, we can identify significant relationships that shape species dynamics in both space and time. Using the identified relationships, ultimately Bayesian ecosystem models will be built to investigate the impact on the species dynamics, using outcomes of climate change on the physical habitats and productivity. Model scenarios will be investigated to examine the likely outcomes of alternative management and climate scenarios, and for evaluating trade-offs and benefits to aid site determination and purpose in the context of large-scale offshore renewable devices.


Trifonova, N., Kenny, A., Maxwell, D., Duplisea, D., Fernandes, J. and Tucker, A. (2015).Spatio-temporal Bayesian network models with latent variables for revealing trophicdynamics and functional networks in fisheries ecology. Ecological Informatics, 30, 142-158.

Sadykova, D., Scott, B., Dominicis, M., Wakelin, S., Sadykov, A., Wolf, J. (2017). Bayesianjoint models with INLA exploring marine mobile predator-prey and competitor space habitatoverlap. Ecology and Evolution, 7 (14), 5215-5226.

Scott, B., Sadykova, D., Sadykov, A., Dominicis, M., Wakelin, S., Wolf, J. (2017). Local vsglobal: How to assess the ecological cost and benefit of large-scale renewables? OCEANS2017, 1-2.Page