phylosignal: an R package to measure, test, and explore the phylogenetic signal.

Keck F, Rimet F, Bouchez A, Franc A. · 2016

Read phylosignal: an R package to measure, test, and explore the phylogenetic signal. by Keck F, Rimet F, Bouchez A, Franc A. (2016) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc4799788.txt.

Introduction

François Keck and three co-authors present an R software package in Ecology and Evolution in 2016 for measuring phylogenetic signal. The signal is the tendency of close kin species to resemble each other in a trait, because they share most of their history. Ecologists increasingly test for it, yet the concept is often muddled with neighbors.

Their package, phylosignal, gathers tests built on likeness measures borrowed from map statistics. A user can measure the signal in a measured trait, test whether it beats chance, and plot how it runs across the tree. A worked example uses pollution sensitivity in seventeen diatom species, single-celled algae with glass shells. The authors keep to the strict statistical sense of the term and hope the tool clears common confusion.

Catalog record

CREATORKeck F, Rimet F, Bouchez A, Franc A.
YEAR2016
COLLECTIONPAPERS
FORMATTXT
FILE SIZE25 KB
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC4799788

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

phylosignal: an R package to measure, test, and explore the phylogenetic signal By François Keck, Frédéric Rimet, Agnès Bouchez, Alain Franc Published in: Ecology and Evolution DOI: 10.1002/ece3.2051 ABSTRACT Abstract Phylogenetic signal is the tendency for closely related species to display similar trait values as a consequence of their phylogenetic proximity. Ecologists and evolutionary biologists are becoming increasingly interested in studying the phylogenetic signal and the processes which drive patterns of trait values in the phylogeny. Here, we present a new R package, phylosignal which provides a collection of tools to explore the phylogenetic signal for continuous biological traits. These tools are mainly based on the concept of autocorrelation and have been first developed in the field of spatial statistics. To illustrate the use of the package, we analyze the phylogenetic signal in pollution sensitivity for 17 species of diatoms. INTRODUCTION A common observation is that continuous traits of closely related species in a phylogeny are often similar, especially when traits are under selection pressure of the environment. More generally, inheritance of traits passed with modifications from one generation to the next may lead to a structured repartition of trait values throughout the phylogeny. The link between phylogeny and continuous trait values is commonly referred in the literature as phylogenetic signal. This concept has gained in popularity among ecologists in recent years, but is often misunderstood and confused with other fundamental ideas like phylogenetic conservatism (Losos 2008). To avoid any possible confusion (see Revell et al. 2008 for disentangling both notions), we stick here to the strict statistical definition of the phylogenetic signal given by Blomberg and Garland (2002), that is, the “tendency for related species to resemble each other more than they resemble species drawn at random from the tree”. Thus, the phylogenetic signal is a statistical dependence between the values of a continuous trait and the phylogenetic tree from which the measured species are the leaves. Studying a statistical dependence leads to hypothesis testing, and formalizing a null hypothesis. Thus, the presence of phylogenetic signal (as defined by Blomberg & Garland) can be tested by rejecting the null hypothesis that trait values for two species are distributed independently from their phylogenetic distance in the tree. The detection and correction of phylogenetic signal has long been motivated by the necessity to control for nonindependence of traits data in comparative studies (Felsenstein 1985; Abouheif 1999). However, recent works have shown that studying the phylogenetic signal can raise interesting biological and ecological perspectives. For example, deciphering the phylogenetic signal may help to understand community assembly processes (Webb et al. 2002), detect niche conservatism (Losos 2008), or identify evolutionary strategies (Jombart et al. 2010b). There are two contrasting approaches in the way phylogenetic signal for a trait can be studied as a statistical model. The first one is based on an explicit evolutionary model for the trait. This is generally a Brownian motion model (Pagel 1999; Blomberg et al. 2003) where continuous traits evolve randomly over time along a branch, with a fixed rate. As soon as descents split at a node of the phylogeny, evolution on both branches becomes independent. To test the presence of phylogenetic signal, the null hypothesis is that trait values are randomly distributed in the phylogeny. Another null hypothesis might be that trait values follow a Brownian motion model but it is less often used and implemented. The second approach relates to methods based on the concept of autocorrelation, the correlation of a vector with itself for a given lag. Autocorrelation is a mathematical tool which has been extensively used to study spatial and time series data. They are designed to detect whether the location of an individual gives information on the expected values of its traits. However, these methods do not rely on any evolutionary model. In a phylogenetic context, patterns of trait values of the species of a tree can be framed as the outcome of a marked point process. Thus, phylogenetic tools based on autocorrelation were largely imported from spatial statistics (Cheverud et al. 1985; Gittleman and Kot 1990; Jombart et al. 2010b). We present a new R package, phylosignal, designed to quantify the phylogenetic signal for continuous biological traits. Most of the tools implemented in phylosignal are based on the concept of autocorrelation and thus are imported from spatial statistics. As such, they are well documented and understood. In this paper, we show how they can be used in a phylogenetic context and we describe their implementation in the package. To illustrate the features of the package, we analyze the phylogenetic signal in pollution sensitivity for 17 species of diatoms. THE PHYLOSIGNAL PACKAGE The phylosignal package provides a collection of tools to visualize, measure, test, and explore the phylogenetic signal in continuous traits (Table 1). The package is written in R and C++ languages and is fully accessible through the R environment. The latest stable version is accessible from The Comprehensive R Archive Network (https://cran.r-project.org/web/packages/phylosignal/) while the development version is hosted on GitHub (https://github.com/fkeck/phylosignal). The phylosignal package is a free software released under the GNU GPL‐3 license and any contribution is welcome. [Table: List of the phylosignal package main functions and their description] This package builds on the R ecosystem richness and takes full advantage of ape (Paradis et al. 2004) for tree manipulation and plotting capacities and adephylo (Jombart et al. 2010a) for tree walking algorithms and phylogenetic distances computing. Data format The analysis of phylogenetic signal typically involves working with a phylogeny and trait values associated with each tip (leaf). The phylobase package (Hackathon et al. 2013) defines the S4 class phylo4d designed specifically to handle such kind of data. Thus, a phylo4d object connects a phylogenetic tree with a table of trait values and constitutes the basic input for many functions implemented in phylosignal. The phylobase package comes with all the necessary functions to construct and manipulate phylo4d objects. For the users who are not used to handle phylogenetic data within the R environment, phylosignal adds the simple function read.p4d, which constructs a phylo4d object from a phylogenetic tree stored in a Newick file and tips data stored in a CSV file. Data visualization The first step of any statistical analysis should be a graphical exploration of the data. The R language provides very powerful and flexible graphics facilities (Murrell 2005). They are extended for phylogenetic tree visualization with traits data by many packages: ape (Paradis et al. 2004), phytools (Revell 2012), adephylo (Jombart et al. 2010a). The phylosignal package aims to provide a simple but complete interface to map traits data onto a phylogenetic tree. The users have access to three main functions to generate high quality graphics: barplot.phylo4d, dotplot.phylo4d and gridplot.phylo4d, which can, respectively, represent univariate and multivariate traits data as bars, dots, and colored cells. Each of these functions comes with several arguments to precisely control graphical aspects. Figure 1 gives an example of a graphic generated with barplot.phylo4d. [Figure: Data visualization of 3 traits (IPSS, random, BM) mapped along the phylogeny of 17 diatom species. This output is obtained with the function barplot.phylo4d. By default data are centered and scaled by trait.] Indices for general measurements of phylogenetic signal The function phyloSignal provides a generic interface

Read the complete text