Resolving Recent Plant Radiations: Power and Robustness of Genotyping-by-Sequencing

Abstract

Disentangling species boundaries and phylogenetic relationships within recent evolutionary radiations is a challenge due to the poor morphological differentiation and low genetic divergence between species, frequently accompanied by phenotypic convergence, interspecific gene flow and incomplete lineage sorting. Here we employed a genotyping-by-sequencing (GBS) approach, in combination with morphometric analyses, to investigate a small western Mediterranean clade in the flowering plant genus Linaria that radiated in the Quaternary. After confirming the morphological and genetic distinctness of eight species, we evaluated the relative performances of concatenation and coalescent methods to resolve phylogenetic relationships. Specifically, we focused on assessing the robustness of both approaches to variations in the parameter used to estimate sequence homology (clustering threshold). Concatenation analyses suffered from strong systematic bias, as revealed by the high statistical support for multiple alternative topologies depending on clustering threshold values. By contrast, topologies produced by two coalescent-based methods (NJ$_{\mathrm{st}}$, SVDquartets) were robust to variations in the clustering threshold. Reticulate evolution may partly explain incongruences between NJ$_{\mathrm{st}}$, SVDquartets and concatenated trees. Integration of morphometric and coalescent-based phylogenetic results revealed (i) extensive morphological divergence associated with recent splits between geographically close or sympatric sister species and (ii) morphological convergence in geographically disjunct species. These patterns are particularly true for floral traits related to pollinator specialization, including nectar spur length, tube width and corolla color, suggesting pollinator-driven diversification. Given its relatively simple and inexpensive implementation, GBS is a promising technique for the phylogenetic and systematic study of recent radiations, but care must be taken to evaluate the robustness of results to variation of data assembly parameters.

Maintenance of genetic and morphological identity in two siblingSyrrhopodonspecies (Calymperaceae, Bryopsida) despite extensive introgression

“…The lower number of SNPs obtained in mosses may reflect a truly lower genetic diversity caused, among others, by high rates of clonality, but may also call for protocol improvements in these non‐model organisms. Nevertheless, the present results, together with those of Lewis et al (), suggest that SNP data are a promising tool for shallow systematics (Fernández‐Mazuecos et al, ; Ding et al, ) and fine‐scale genetic structure (Attard et al, ) in bryophytes when traditional markers such as low‐copy nuclear genes and chloroplast genes are not polymorphic.…”

Macroevolutionary dynamics of nectar spurs, a key evolutionary innovation

“…Supinae (Fig. ) may have been favoured by their specialized, predominantly self‐incompatible flowers with both an occluded personate corolla and a nectar spur, together with Mediterranean conditions and climate changes since the late Miocene (Fernández‐Mazuecos & Vargas, ; Blanco‐Pastor et al ., , ; Blanco‐Pastor & Vargas, ; Fernández‐Mazuecos et al ., 2013a, ). High diversification rates are also possible in the absence of nectar spurs, as shown by Antirrhinum , where geographic speciation under Mediterranean conditions and pollinator specialization by evolution of corolla length are proposed as main drivers (Vargas et al ., , , ; Wilson & Hudson, ).…”

“…Similarly, in Linaria subsect. Versicolores , spur length plays a role, in conjunction with tube width, in determining pollinator strategies in closely related species (Fernández‐Mazuecos et al ., 2013a, ).…”

Cause of gene tree discord? Distinguishing incomplete lineage sorting and lateral gene transfer in phylogenetics

“…Do these neL resolutions represent "Lhole-genome evidence"? Or reflect biases in data processing steps and tree reconstruction methods (e.g., Dell'Ampio et al, 2014;Fernández-Mazuecos et al, 2017)? Or are driven by strong signals in small number of genes or sites (Lhich could be outliers; e.g., Shen et al, 2017;Xi et al, 2014)?…”