Assessing the potential of RAD-sequencing to resolve phylogenetic relationships within species radiations: The fly genus Chiastocheta (Diptera: Anthomyiidae) as a case study

Abstract

Determining phylogenetic relationships among recently diverged species has long been a challenge in evolutionary biology. Cytoplasmic DNA markers, which have been widely used, notably in the context of molecular barcoding, have not always proved successful in resolving such phylogenies. However, with the advent of next-generation-sequencing technologies and associated techniques of reduced genome representation, phylogenies of closely related species have been resolved at a much higher detail in the last couple of years. Here we examine the potential and limitations of one of such techniques—Restriction-site Associated DNA (RAD) sequencing, a method that produces thousands of (mostly) anonymous nuclear markers, in disentangling the phylogeny of the fly genus Chiastocheta (Diptera: Anthomyiidae). In Europe, this genus encompasses seven species of seed predators, which have been widely studied in the context of their ecological and evolutionary interactions with the plant Trollius europaeus (Ranunculaceae). So far, phylogenetic analyses using mitochondrial markers failed to resolve monophyly of most of the species from this recently diversified genus, suggesting that their taxonomy may need a revision. However, relying on a single, non-recombining marker and ignoring potential incongruences between mitochondrial and nuclear loci may provide an incomplete account of the lineage history. In this study, we applied both classical Sanger sequencing of three mtDNA regions and RAD-sequencing, for reconstructing the phylogeny of the genus. Contrasting with results based on mitochondrial markers, RAD-sequencing analyses retrieved the monophyly of all seven species, in agreement with the morphological species assignment. We found robust nuclear-based species assignment of individual samples, and low levels of estimated contemporary gene flow among them. However, despite recovering species’ monophyly, interspecific relationships varied depending on the set of RAD loci considered, producing contradictory topologies. Moreover, coalescence-based phylogenetic analyses revealed low supports for most of the interspecific relationships. Our results indicate that despite the higher performance of RAD-sequencing in terms of species trees resolution compared to cytoplasmic markers, reconstructing inter-specific relationships among recently-diverged lineages may lie beyond the possibilities offered by large sets of RAD-sequencing markers in cases of strong gene tree incongruence

Are glacial refugia hotspots of speciation and cytonuclear discordances? Answers from the genomic phylogeography of Spanish common frogs

“…Hence, the maintenance of ghost mitochondrial lineages can create mirages of cryptic species (Hinojosa et al, ), but at the same time mitochondrial capture and replacement may conceal genuine evolutionary divergences (Dufresnes, Mazepa, et al, ). Because these two phenomena might be more common than previously assumed (Dufresnes, Mazepa, et al, ), we recommend that taxonomic revisions involving taxa from refugial areas should be based on genomic evidence to make decisions on species‐level divergence (Suchan et al, ).…”

Mitonuclear discordance in wolf spiders: Genomic evidence for species integrity and introgression

“…Species with high levels of introgression are supposed to show close overall relatedness and intermixing of species in recently introgressed genes (Eaton et al., ). Historical or ongoing introgression often revealed by discordant patterns between mitochondrial and nuclear phylogenies (Linnen & Farrell, ; Papakostas et al., ; Suchan et al., ). Unlike the mitochondrial phylogeny, the nDNA phylogeny showed a clear distinction between the target species and did not support the idea of introgression at the genomic level in either Alopecosa or Pardosa .…”

DiscoSnp-RAD: de novo detection of small variants for population genomics

“…Tests on real data were performed on ddRAD reads previously obtained for the phylogenetic study of seed parasitic pollinators from the genus Chiastocheta (Diptera: Anthomyiidae). The dataset corresponds to reads from 259 individuals sampled from 51 European populations generated by Lausanne University, Switzerland [19] (data soon available on zenodo.org). Quality scores of reads 2 (average base quality over reads and across samples of 33.98 sd 1.78) were lower than those of reads 1 (average base quality over reads and across samples of 35.15 sd 1.70), which led us to focus the analyses on reads 1.…”

“…This latter treatment of predictions is considered as the default process of DiscoSnp-RAD outputs. Then, classical filters were applied to follow as much as possible the filters used in the Suchan et al [19]: a minimum genotype coverage of 6, a minimal minor allele frequency of 0.02 and a minimum of 20 samples with a non missing genotype for each variant. These filters remove less informative variants or alleles specific to a small subset of samples.…”