Fast and accurate detection of evolutionary shifts in Ornstein–Uhlenbeck models

Abstract

1. The detection of evolutionary shifts in trait evolution from extant taxa is motivated by the study of convergent evolution, or to correlate shifts in traits with habitat changes or with changes in other phenotypes. 2. We propose here a phylogenetic lasso method to study trait evolution from comparative data and detect past changes in the expected mean trait values. We use the Ornstein-Uhlenbeck process, which can model a changing adaptive landscape over time and over lineages. 3. Our method is very fast, running in minutes for hundreds of species, and can handle multiple traits. We also propose a phylogenetic Bayesian information criterion that accounts for the phylogenetic correlation between species, as well as for the complexity of estimating an unknown number of shifts at unknown locations in the phylogeny. This criterion does not suffer model overfitting and has high precision, so it offers a conservative alternative to other information criteria. 4. Our re-analysis of Anolis lizard data suggests a more conservative scenario of morphological adaptation and convergence than previously proposed. Software is available on GitHub.

Detecting Polygenic Adaptation in Admixture Graphs

“…Our use of the Normal model of drift as an approximation to the Wright-Fisher diffusion is closely analogous to the use of Brownian motion models in some phylogenetic methods (Venditti et al 2011;Eastman et al 2011;Revell et al 2012;Rabosky et al 2013;Jhwueng and O'Meara 2015). It may also be worth exploring the relationship between Ornstein-Uhlenbeck models for phenotypic evolution on phylogenies (Uyeda and Harmon 2014;Khabbazian et al 2016) and the aforementioned hypothetical extension of our method to include stabilizing selection, as the two processes are closely related (Lande 1976;Simons et al 2017). Tables Graph Trait P P Table 1 Trait-associated variants with Bonferroni-corrected significant evidence of being under polygenic adaptation in the 1000 Genomes data, using the Q X statistic: P < 0.05/n where n is the number of GWAS tested, assuming a χ 2 distribution.…”

Widespread ecomorphological convergence in multiple fish families spanning the marine–freshwater interface

“…Such contrasts in criterion results are not entirely unexpected: AIC C -based '1ou tends to be liberal (finding not only many true shifts but also some false positives), whereas '1ou coupled with pBIC tends to exhibit high precision (any detected shifts are typically true) but a low recall rate (i.e. false negatives) [43]. '1ou analyses (using AIC) focusing on specific PC axes identified higher levels of convergence on PC1 (11 total shifts with nine collapses onto multiple convergent optima) aligning closely with dietary habits.…”

“…To identify potential convergence onto phenotypic optima without a priori designation of selective regimes we used the '1ou package [43]. A model-based approach to detecting convergence, '1ou employs a regularized least absolute shrinkage and selection operator (LASSO) method to determine the optimal number of selective regimes in a phylogeny, and can accommodate single or multiple traits.…”

BBMV: an R package for the estimation of macroevolutionary landscapes

“…Comparing model parameters among clades is usually done for BM or OU models by singling out one or several clades from a larger phylogeny (O'Meara et al 2006, Uyeda and Harmon 2014, Khabbazian et al 2016. This can be used for two main purposes: 1) to increase statistical power in the estimation of the macroevolutionary landscape by combining data from multiple clades or 2) to test whether parameters of the model statistically differ between clades.…”

“…The package thus complements existing packages like 'geiger' (Pennell et al 2014), 'l1ou' (Khabbazian et al 2016), 'bayou' (Uyeda and Harmon 2014), 'OUwie' (Beaulieu et al 2012) or 'phytools' (Revell 2012), which all implement other models of trait evolution based on BM and OU. BBMV can accommodate interesting evolutionary scenarios like evolution between bounds, directional trends, disruptive selection, or evolution towards two distinct peaks, thus offering new perspectives for the analysis of trait evolution along phylogenies.…”