Bayesian estimation of multiple clade competition from fossil data

Détails

Ressource 1Télécharger: ffar3010.pdf (359.15 [Ko])
Etat: Public
Version: Final published version
ID Serval
serval:BIB_9717D8A0D667
Type
Article: article d'un périodique ou d'un magazine.
Collection
Publications
Titre
Bayesian estimation of multiple clade competition from fossil data
Périodique
Evolutionary Ecology Research
Auteur(s)
Silvestro D., Pires M.M., Quental T.B., Salamin N.
ISSN
1937-3791
ISSN-L
1522-0613
Statut éditorial
Publié
Date de publication
2017
Peer-reviewed
Oui
Volume
18
Pages
41-59
Langue
anglais
Résumé
Background: The diversification dynamics of clades is governed by speciation and extinction processes and is likely affected by multiple biotic, abiotic, and stochastic factors. Using quantitative methods to analyse fossil occurrence data, one may infer rates of speciation and extinction in a Bayesian framework. Moreover, Silvestro et al. (2015a) recently developed a Multiple Clade Diversity Dependence birth-death model (MCDD) to determine whether diversification dynamics can be explained by positive or negative effects of interactions within or between co-existing clades. However, the power and accuracy of this model and its general applicability have yet to be thoroughly investigated.
Aims: Explore the properties of the existing MCDD implementation, which is based on Bayesian variable selection. Introduce an alternative parameterization based on the Horseshoe prior and show the properties of this approach for Bayesian shrinkage in complex models. Test the ability of the model to correctly identify within and between diversification interference under a suite of different diversification scenarios.
Methods: Use simulations to assess and compare the power and accuracy of the two algorithms.
Results: Diversity dependence within and between clades can be inferred with confidence in a wide range of scenarios using the MCDD model. The two implementations provide comparable results, but the new Horseshoe prior estimator appears to be more reliable, albeit slightly more conservative. The MCDD model is a powerful framework to analyse the putative effects of ecological interactions on macroevolutionary dynamics using fossil data and provides a sound statistical basis for future method developments.
Mots-clé
Bayesian shrinkage, diversity dependence, fossils, macroevolution
Web of science
Création de la notice
03/04/2017 14:53
Dernière modification de la notice
20/08/2019 15:59
Données d'usage