A Bayesian sensitivity analysis applied to an Agent-based model of bird population response to landscape change

  • Authors:

  • Hazel R. Parry;Christopher J. Topping;Marc C. Kennedy;Nigel D. Boatman;Alistair W. A. Murray

  • Affiliations:

  • Food and Environment Research Agency, Sand Hutton, York YO41 1LZ, United Kingdom;Department of Bioscience, Aarhus University, Grenåvej 14, 8410 Rønde, Denmark;Food and Environment Research Agency, Sand Hutton, York YO41 1LZ, United Kingdom;Food and Environment Research Agency, Sand Hutton, York YO41 1LZ, United Kingdom;Food and Environment Research Agency, Sand Hutton, York YO41 1LZ, United Kingdom

  • Venue:
  • Environmental Modelling & Software
  • Year:
  • 2013

Quantified Score

Hi-index 0.00

Visualization

Abstract

Agricultural land management has important impacts on land use and vegetation that can rapidly induce ecosystem change. Birds are often used as indicators of such impacts of landscape change on ecosystems. However, predicting the response of birds to changes in their environment is an ongoing challenge. Agent-based models (ABMs) have the potential to provide useful insights but have not been widely used in such studies to date. This paper illustrates the use of agent-based modelling for policy decision-making, using the case study of the impacts of the removal of set-aside land on Skylark populations in Denmark. In order to address the importance of critical interpretation of ABMs, we introduce a novel methodology with which to analyze the sensitivity of an ABM, Bayesian Analysis of Computer Code Outputs (BACCO). BACCO constructs an emulator of the model in order to provide a rapid and thorough sensitivity analysis. This allows us to identify input parameters in the model that require more rigorous parameterization, as some parameters are highly sensitive and are found to produce spurious results when varied even a small amount.