A framework for coupling explanation and prediction in hydroecological modellingq

Causal explanation and empirical prediction are usually addressed separately when modelling ecological systems. This paper proposes a novel, integrated modelling approach which couples explanatory modelling for causal understanding and input variable selection with a machine learning approach for empirical prediction. Exemplar datasets from the field of freshwater ecology are used to develop and evaluate the modelling framework, based on 267 stream and river monitoring stations across England, UK. These data describe spatial patterns in two benthic macroinvertebrate community indices that are hypothesised to be driven by meso-scale physical and chemical habitat conditions. Explanatory models were developed using Structural Equation Modelling. These models accounted for 64% to 70% of the among site variation in macroinvertebrate community indices that reflect the sensitivity of taxa to organic pollution and to flow conditions within streams and rivers. Predictive models based on Extremely Randomised Trees demonstrated only moderate performance, with R2 based on 10-fold cross validation reaching 50% to 61% for the same macroinvertebrate community indices. These results emphasise the distinction between explanatory and predictive power in environmental modelling. Limited predictive power reflects the complexity of ecosystems at community levels of organisation and the lack of empirical data with which to describe all significant causal relationships within these systems. However, by coupling explanation and prediction, our proposed modelling framework provides opportunities to enhance feedback between causal theory, empirical data and prediction within environmental systems.

SURRIDGE Ben W.;  BIZZI Simone;  CASTELLETTI Andrea; 

2014-11-26

ELSEVIER SCI LTD

JRC85349

1364-8152,   

http://www.sciencedirect.com/science/article/pii/S1364815214000632,    https://publications.jrc.ec.europa.eu/repository/handle/JRC85349,   

10.1016/j.envsoft.2014.02.012,