Multiple outgroups can cause random rooting in phylogenomics.

The diagram on the left shows our strategy for examining outgroup impact on ingroup topology using several recent phylogenomic datasets. Basically, the approach entails examining multiple three taxon statements concurrent with the use of outgroups that are farther and farther away from the ingroup. We find a strong correlation of outgroup distance from ingroup and inaccurate phylogenetic inference. The figure on the right shows results of ASTRAL analysis for decision branch length and normalized quartet score. As distance of the outgroup to ingroups gets larger and larger, both measures of phylogenetic accuracy get worse and worse. [Display omitted] Outgroup selection has been a major challenge since the rise of phylogenetics, and it has remained so in the phylogenomic era. Our goal here is to use large phylogenomic animal datasets to examine the impact of outgroup selection on the final topology. The results of our analyses further solidify the fact that distant outgroups can cause random rooting, and that this holds for concatenated and coalescent-based methods. The results also indicate that the standard practice of using multiple outgroups often causes random rooting. Most researchers go out of their way to get multiple outgroups, as this has been standard practice for decades. Based on our findings, this practice should stop. Instead, our results suggest that a single (most closely) related relative should be selected as the outgroup, unless all outgroups are roughly equally closely related to the ingroup. [ABSTRACT FROM AUTHOR]