A novel data filtering method resolves the controversy in the phylogeny of the Chondrichthyes

Abstract

Phylogenomics, which uses genome-scale data for phylogenetic inference, has clarified many controversial nodes in the tree of life. Such extensive data improve tree resolution and better reflects organismal history compared to analyses based on single or a few genetic loci. However, some relationships within the tree of life remain unresolved, as increased data can yield high node support without ensuring accuracy due to systematic errors. For example, the order-level relationships among chondrichthyans are still contentious despite the use of phylogenomic data. To address systematic errors, complex models have been developed, and filtering for less erroneous data shows great promise. Current metric-based filtering methods rank loci based on overall tree statistics, but problematic signals are often local; and topology-based data filtering approaches struggle with circular assumptions. In this study, we introduced two novel metric-based data filtering methods based on the ratio of local branch length or GC content between problematic clades. We applied these methods to a dataset of 4,452 single-copy exons extracted from 98 chondrichthyan species. The results using all loci showed that the Hexanchiformes was positioned at the root of Elasmobranchii, pulling other squalomorphs to the basal position and rendering Squalomorphii paraphyletic. Contrastingly, filtering for loci with more even branch length, the branch ratio method (absRatioLen) strongly supported the monophyly of all superorders of the chondrichthyans as well as their higher classification grouping, such as Selachii and Batoidea. By concentrating on problematic nodes, our assumption-free filtering methods demonstrate significant potential in resolving contentious relationships in the tree of life.