Institut Méditerranéen d’Océanologie
Accueil du site > Agenda > Thèses > Publication MIO : Clement Aldebert (MIO) , Daniel B. Stouffer - Community (...)

Publication MIO : Clement Aldebert (MIO) , Daniel B. Stouffer - Community dynamics and sensitivity to model structure : towards a probabilistic view of process-based model predictions

Version imprimable de cet article

in Journal of the Royal Society Interface

Received October 5, 2018.
Accepted November 5, 2018.

DOI : 10.1098/rsif.2018.0741

Context : This work emerged from a seminar given by Clement Aldebert (postdoc MIO) at the School of Biological Sciences at the University of Canterbury (Christchurch, New-Zealand) in February 2018. Clement Aldebert was visiting the group of Daniel Stouffer (Assoc. Prof at UC), a visit that ended up in a collaboration that already materialized with this paper.

Abstract :

Statistical inference and mechanistic, process-based modelling represent two philosophically different streams of research whose primary goal is to make predictions. Here, we merge elements from both approaches to keep the theoretical power of process-based models while also considering their predictive uncertainty using Bayesian statistics. In environmental and biological sciences, the predictive uncertainty of process-based models is usually reduced to parametric uncertainty. Here, we propose a practical approach to tackle the added issue of structural sensitivity, the sensitivity of predictions to the choice between quantitatively close and biologically plausible models. In contrast to earlier studies that presented alternative predictions based on alternative models, we propose a probabilistic view of these predictions that include the uncertainty in model construction and the parametric uncertainty of each model. As a proof of concept, we apply this approach to a predator–prey system described by the classical Rosenzweig–MacArthur model, and we observe that parametric sensitivity is regularly overcome by structural sensitivity. In addition to tackling theoretical questions about model sensitivity, the proposed approach can also be extended to make probabilistic predictions based on more complex models in an operational context. Both perspectives represent important steps towards providing better model predictions in biology, and beyond.

(Shorter) Media summary :

Statistical inference and process-based modelling represent philosophically different streams of research to make predictions. Here, we merge them to keep the theoretical power of process-based models while explicitly considering their predictive uncertainty. Predictive uncertainty is often equated with parametric uncertainty, which neglects the structural uncertainty arising from the choice between alternative —biologically plausible— mathematical models. Our study introduces a probabilistic approach considering both uncertainties. Surprisingly, we observe that classic parametric sensitivity is regularly overcome by structural sensitivity. Our approach allows theoretical investigations on model sensitivity and operational probabilistic predictions, two major steps toward better model predictions in biology and beyond.

Voir en ligne : http://rsif.royalsocietypublishing....