Rphylopars: fast multivariate phylogenetic comparative methods for missing data and within-species variation

Goolsby, EW, Bruggeman, J and Ané, C 2016 Rphylopars: fast multivariate phylogenetic comparative methods for missing data and within-species variation. Methods in Ecology and Evolution, 8 (1). 22-27. https://doi.org/10.1111/2041-210X.12612

Goolsby_et_al-2017-Methods_in_Ecology_and_Evolution.pdf - Published Version
Available under License All Rights Reserved.

Download (235kB) | Preview
Official URL: https://doi.org/10.1111/2041-210X.12612


Over the past several years, phylogenetic comparative studies have increasingly approached trait evolution in a multivariate context, with a number of taxa that continues to rise dramatically. Recent methods for phylogenetic comparative studies have provided ways to incorporate measurement error and to address computational challenges. However, missing data remain a particularly common problem, in which data are unavailable for some but not all traits of interest for a given species (or individual), leaving researchers with the choice between omitting observations or utilizing imputation-based approaches. Here, we introduce an r implementation of PhyloPars, a tool for phylogenetic imputation of missing data and estimation of trait covariance across species (phylogenetic covariance) and within species (phenotypic covariance). Rphylopars provides expanded capabilities over the original PhyloPars interface including a fast linear-time algorithm, thus allowing for extremely large data sets (which were previously computationally infeasible) to be analysed in seconds or minutes rather than hours. In addition to providing fast and computationally efficient implementations, we introduce in Rphylopars methods to estimate macroevolutionary parameters under alternative evolutionary models (e.g. Early-Burst, multivariate Ornstein-Uhlenbeck). By providing fast and computationally efficient methods with flexible options for various phylogenetic comparative approaches, Rphylopars expands the possibilities for researchers to analyse large and complex data with missing observations, within-species variation and deviations from Brownian motion.

Item Type: Publication - Article
Additional Keywords: ast methods; linear-time algorithm; missing data; multivariate Ornstein-Uhlenbeck; phylogenetic comparative method; phylogenetic generalized least squares; phylogenetic imputation
Subjects: Biology
Computer Science
Divisions: Plymouth Marine Laboratory > Science Areas > Marine System Modelling
Depositing User: Jorn Bruggeman
Date made live: 28 Sep 2020 11:43
Last Modified: 13 Dec 2023 12:21
URI: https://plymsea.ac.uk/id/eprint/7588

Actions (login required)

View Item View Item