Creating predictive models of community structure and change in marine systems by combining sparse data and expert opinion.

This source preferred by Roger Herbert and Rick Stafford

Authors: Stafford, R., Williams, R.L. and Herbert, R.J.H.

Start date: 11 May 2014

Effective ecological predictions require large amounts of data for model parameterisation. Furthermore, although uncertainty of predictions is reported more frequently now than historically, understanding of uncertainty is not evident in many management or public engagement approaches. Bayesian belief networks (with some limitations, addressed in this study) create an intuitive method to model changes in ecological communities and present intuitive measures of certainty in their outputs. They are able to incorporate sparse or uncertain data, or data from multiple sources. This study explores the modification and use of these networks to predict changes in marine systems, it also explores how expert opinion can be combined with limited ecological data to create effective predictions. Models are based on rocky shore communities; however, the underlying concepts are applicable across a range of marine systems. Expert evaluation of interaction strengths (from rocky shore ecologists from different areas of the world to the study conducted, with a minimum of postgraduate expertise), combined with data collected from shores in the UK, created robust models predicting major changes in community structure following disturbance, which were verified by experimental manipulations. As such, the models provide a robust method to develop end-user friendly predictive models with limited available data.

The data on this page was last updated at 04:40 on June 25, 2017.