Characterizing and mitigating confounding by unreplicated evolutionary events in phylogenetic regression
Evolution Letters · 03.08.2026
DOI: 10.1093/evlett/qrag032 ↗Kurz gefasst
Fachartikel in der Zeitschrift Evolution Letters, online erschienen am 03.08.2026. Verfasst von Qifan Wang und 3 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 67 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Cornell University, National Institute of General Medical Sciences.
Abstract (Originalsprache)
When researchers want to test for a direct effect of one attribute on another using interspecific data, they need to account for the confounding influence of other attributes that have phylogenetic structure. Accordingly, a suite of approaches have been developed for this purpose, mostly notably the use of phylogenetic regression. However, it has been shown that such methods can be misled by evolutionary events with large effects. We suggest that one widely applicable solution to this problem can be found by borrowing both from the statistical genetics literature and by digging into the past of phylogenetic comparative methods. In conventional phylogenetic regression, the phylogenetic variance–covariance matrix is included as a random-effects term; here, we show that it can be advantageous to also include the leading eigenvectors of this matrix as fixed effects. Building on the results of Schraiber et al. (Schraiber, J. G., Edge, M. D., & Pennell, M. (2024). Unifying approaches from statistical genetics and phylogenetics for mapping phenotypes in structured populations. PLoS Biology, 22(10), e3002847), we use mathematical analysis and simulations to identify the scenarios where this hybrid strategy is effective as well as where it will not be. To aid researchers’ intuition, we develop a novel approach for visualizing the contributions of different branches of the phylogeny to the eigenvectors of the variance–covariance matrix. We then apply the mixed approach to investigate the coevolution of gene expression levels of different genes across the phylogeny of cichlid fishes; we show that including the leading eigenvectors likely reduces the false-positive rate. To facilitate the use of our visualization approach and of including eigenvectors in phylogenetic regression models, we present a new R package called $\tt {EIGER}$. We argue that the approaches we explore here can both help address the statistical problems that arise from large, historical events and, more generally, provide richer insights into the nature of phylogenetic confounding.
Förderung
- Cornell University
- National Institute of General Medical Sciences
Zitieren
APA
Wang, Q., Edge, M. D., Schraiber, J. G., & Pennell, M. (2026). Characterizing and mitigating confounding by unreplicated evolutionary events in phylogenetic regression. Evolution Letters. https://doi.org/10.1093/evlett/qrag032
Vancouver
Wang Q, Edge MD, Schraiber JG, Pennell M. Characterizing and mitigating confounding by unreplicated evolutionary events in phylogenetic regression. Evolution Letters. 2026. doi:10.1093/evlett/qrag032
BibTeX
@article{Wang2026,
author = {Wang, Qifan and Edge, Michael D and Schraiber, Joshua G and Pennell, Matt},
title = {Characterizing and mitigating confounding by unreplicated evolutionary events in phylogenetic regression},
journal = {Evolution Letters},
year = {2026},
doi = {10.1093/evlett/qrag032}
}
EUVIMED