Multiple rounds of ancient and recent hybridizations have occurred within the Aegilops-Triticum complex.

We agree with Sandve et al. (2015) that the nomenclature of Aegilops /Triticum lineages is complex. Indeed, there is a contradiction in how they themselves have defined ‘theDgenome lineage’ in their Letter (Sandve et al., 2015), compared to their earlier paper (Marcussen et al., 2014). Fig. 1 of Sandve et al. (2015) defines it as a clade comprising D + S* + M genome species; this definition is consistent with the clade we identified in our cpDNA phylogeny (Fig. 1 of Li et al., 2015). By contrast, the nuclear gene phylogeny of Marcussen et al. (2014) places M-genome species (e.g. Ae. comosa) within theB-genome lineage, not theD-genome lineage (see Fig. S6 of Marcussen et al., 2014). This contradiction between the chloroplast and nuclear phylogenies is consistent with our previous inference of a complex hybridization history for the D-genome lineage. It is not consistent with a single homoploid hybrid origin of the D + S* +M clade, as inferred by Sandve et al. (2015). Moreover, contrary to the assertion by Sandve et al. (2015) that we confounded the modern D genome and the D-genome lineage (as defined in Fig. 1 of Sandve et al., 2015), we explicitly drew this distinction in our analyses (e.g. Table 2 of Li et al., 2015; showing results both for extant D-genome species and for ‘the ancient D-genome lineage’). We also explicitly considered the single homoploid hybrid originmodel favored by Sandve et al. (2015) for the D-genome lineage (see Fig. 2a of Li et al., 2015); however, the conflicting chloroplast and nuclear phylogenies described earlier led to our inference of a more complex hybridization history.