Uncertainty analyses for Ecological Network Analysis enable stronger inferences

Abstract

Uncertainty analyses show how variability inherent in model parameters affects model outcomes. While conducting uncertainty analyses is considered best practice, technical and conceptual challenges limit applications for network models. This work adapts Linear Inverse Modeling (LIM) techniques to conduct uncertainty analysis on ecosystem flow networks, which represent the movement of energy-matter through ecosystems. We present a new R function for the enaR package to perform the analysis and use two case studies of previously published networks to demonstrate the power of this approach. The first case study examines a system with available flow uncertainty data to show how LIM uncertainty analysis can support stronger statistical inference. The second case study examines a system without available uncertainty data to illustrate how these techniques can determine the relative strength of model conclusions, even without quantitative data. The tools presented here represent an important step in the maturation of Ecological Network Analysis.