Most Genomic Loci Misrepresent the Phylogeny of an Avian Radiation Because of Ancient Gene Flow

Abstract

Phylogenetic trees based on genome-wide sequence data may not always represent the true evolutionary history for a variety of reasons. One process that can lead to incorrect reconstruction of species phylogenies is gene flow, especially if interspecific gene flow has affected large parts of the genome. We investigated phylogenetic relationships within a clade comprising eight species of passerine birds (Phylloscopidae, Phylloscopus, leaf warblers) using one de novo genome assembly and 78 resequenced genomes. On the basis of hypothesis-exclusion trials based on D-statistics, phylogenetic network analysis and demographic inference analysis, we identified ancient gene flow affecting large parts of the genome between one species and the ancestral lineage of a sister species pair. This ancient gene flow consistently caused erroneous reconstruction of the phylogeny when using large amounts of genome-wide sequence data. In contrast, the true relationships were captured when smaller parts of the genome were analyzed, showing that the “winner-takes-all democratic majority tree” is not necessarily the true species tree. Under this condition, smaller amounts of data may sometimes avoid the effects of gene flow due to stochastic sampling, as hidden reticulation histories are more likely to emerge from the use of larger datasets, especially whole-genome datasets. In addition, we also found that genomic regions affected by ancient gene flow generally exhibited higher genomic differentiation but a lower recombination rate and nucleotide diversity. Our study highlights the importance of considering reticulation in phylogenetic reconstructions in the genomic era.

Assembling an illustrated family-level tree of life for exploration in mobile devices

“…Third, phylogenies are prioritized based on the number of characters employed for their reconstruction, thus favoring the recent phylogenomic analyses over traditional single-locus or multi-locus approaches. Although it has been shown that a higher amount of characters does not always translate into a more accurate species tree (e.g., Zhang et al 2021) and are still prone to different sources of systematic error (Philippe et al 2017; Reddy et al 2017), we believe that the advantages of larger datasets outweight their drawbacks in most cases, especially considering that recent phylogenomic studies are thoroughly analyzing their data with scrutiny to account for potential misleading issues (e.g., Arcila et al 2017; Betacur-R et al 2019; Redmong & McLysaght 2021; Zhang et al 2021). Higher taxon sampling is used as a tiebreaker in instances of equal number of characters, due to its positive impact in breaking long branches and improving model selection (Nabhan & Sarkar 2012).…”

Evolutionary relationships among the snakelike pygopodid lizards: a review of phylogenetic studies of an enigmatic Australian adaptive radiation

“…It was also surprising that the concatenated UCE loci yielded a tree containing likely spurious sister species pairings of D. butleri / D. tincta and D. haroldi / D. borea . But given that the members of this five-species group (except D. grayii ) in addition to D. borea and D. tincta have overlapping or proximal contemporary geographic ranges ( Storr, Smith & Johnstone, 1990 ), the possibility exists that ancient hybridization involving these lineages is clouding their true phylogenetic relationships in molecular studies (e.g., Lásková, Landová & Frynta, 2015 ; Zhang et al, 2021 ).…”

“…However, as the recent study by Zhang et al (2021) showed, extensive ancient gene flow between non-sister species can confound species tree estimates based on genome-wide datasets but perhaps not negatively impact tree estimates inferred from smaller numbers of genomic loci. This finding raises the possibility that the D. borea / D. tincta sister species pair in the MSC-UCE tree may actually be an artifact of hybridization, a plausible scenario given that both of these species are sympatric with each other ( Storr, Smith & Johnstone, 1990 ).…”

Reticulate evolution in a neutral model: speciation, extinctions, and hybridizations

“…Phylogenies that do not take into consideration processes that lead to reticulation, like hybridization, can lead to the construction of incompatible trees that are not verisimilar representations of the real tree (Zhang et al, 2021). Detecting hybridization, however, can be challenging.…”

“…This allows us to compute true and complete phylogenies, from its root at T = 0 to the leafs at present time, containing all events that ever occured in that population. This offers a theoretical methodology that avoids the common loss of evolutionary information caused by gene flow (Zhang et al, 2021).…”