Impact of Ghost Introgression on Coalescent-Based Species Tree Inference and Estimation of Divergence Time

The species studied in any evolutionary investigation generally constitute a small proportion of all the species currently existing or that have gone extinct. It is therefore likely that introgression, which is widespread across the tree of life, involves ghosts, that is, unsampled, unknown, or extinct lineages. However, the impact of ghost introgression on estimations of species trees has rarely been studied and is poorly understood. Here, we use mathematical analysis and simulations to examine the robustness of species tree methods based on the multispecies coalescent model to introgression from a ghost or extant lineage. We found that many results originally obtained for introgression between extant species can easily be extended to ghost introgression, such as the strongly interactive effects of incomplete lineage sorting (ILS) and introgression on the occurrence of anomalous gene trees. The relative performance of the summary species tree method (ASTRAL) and the full-likelihood method (*BEAST) varies under different introgression scenarios, with the former being more robust to gene flow between nonsister species, whereas the latter performing better under certain conditions of ghost introgression. When an outgroup ghost (defined as a lineage that diverged before the most basal species under investigation) acts as the donor of the introgressed genes, the time of root divergence among the investigated species generally was overestimated, whereas ingroup introgression, as commonly perceived, can only lead to underestimation. In many cases of ingroup introgression that may or may not involve ghost lineages, the stronger the ILS, the higher the accuracy achieved in estimating the time of root divergence, although the topology of the species tree is more prone to be biased by the effect of introgression. [Anomalous gene trees; divergence time; ghost introgression; multispecies coalescent; simulation; species tree.]

Automated summary

This study examines the impact of ghost introgression on species tree estimations and found that many results obtained for introgression between extant species can be extended to ghost introgression. The performance of summary species tree method (ASTRAL) and full-likelihood method (*BEAST) varies under different introgression scenarios. When an outgroup ghost acts as the donor of introgressed genes, the time of root divergence is generally overestimated, while ingroup introgression leads to underestimation. The accuracy of root divergence estimation is higher with stronger incomplete lineage sorting (ILS), although the topology of the species tree is more prone to be biased by introgression.