Sensitivity analysis of Repast computational ecology models with R/Repast

Area of Science:

• Computational Ecology
• Ecological Modeling
• Systems Ecology

Background:

• Individual-based modeling (IBM) is crucial for understanding complex ecological dynamics, capturing individual variability and nonlinearities in ecosystems.
• IBM's bottom-up approach offers insights into local mechanisms generating observed global ecological patterns.
• Rigorous analysis of model results, including sensitivity analysis, is essential for reliable conclusions in in-silico studies.

Purpose of the Study:

• To present R/Repast, a GNU R package designed for running and analyzing Repast Simphony models.
• To demonstrate the application of global sensitivity analysis within the R/Repast framework.
• To provide guidance on interpreting the results of sensitivity analyses for ecological models.

Main Methods:

• Development and presentation of the R/Repast package for integrating Repast Simphony with the R statistical environment.
• Implementation of global sensitivity analysis techniques for in-silico ecological experiments.
• Illustrative examples showcasing the practical application and interpretation of sensitivity analysis results.

Main Results:

• The R/Repast package facilitates the execution and analysis of individual-based models within the R ecosystem.
• Worked examples demonstrate effective methods for performing and interpreting global sensitivity analysis on ecological simulations.
• The study underscores the importance of sensitivity analysis for validating and understanding ecological model outputs.

Conclusions:

• R/Repast provides a valuable tool for computational ecologists to conduct robust in-silico experiments.
• The integration of sensitivity analysis enhances the reliability and interpretability of individual-based model results.
• This approach aids in understanding the impact of input uncertainty on ecological model predictions.