The topology of this tree, which – confirming our 2008 plastid tree – shows several deep incongruences with the nuclear-based species trees (ITS-based, Grimm et al. 2006, or phylogenomic: Li et al. 2019), devalidates your follow-up paper published in Genomics a year later, where you base all analyses on the combined nuclear(435kbp)-plastid(150kpb) data set.

Given the here shown results and the contrasting topology of your follow-up combined tree – being nuclear-constrained, most deep incongruences have been masked – wouldn't it have been necessary to provide an updated, tip-richer complete plastome tree, and use that tree as basis for the explicit biographic analysis?

Note that in all known (and well-documented) cases for nuclear-plastid incongruence, very common in extra-tropical tree genera (e.g. Quercus, Nothofagus s.str., Fagaceae at family level), one can observe that plastid differentiation is strongly controlled by geography (past and recent). Thus, while nuclear data will give a better species tree, plastid data may better capture ancient geographic patterns, including those predating the formation and stabilisation of modern-day infra-generic lineages. On the other hand, the combination of two incongruent genealogies will lead to branching artefacts.

Specifically, the nuclear (-dominated, when combined) trees place the species-rich East Asian clades (our 2006 'platanoid cluster') as the earliest diverging lineage, while this tree (and any other plastid-data based tree) recognises trans-continental pairs/few-species groups as the earliest diverging (plastid) lineages. The latter pretty fitting to the early fossil record of the genus (High-Arctic, E.Siberia/Kamchatka to Spitsbergen, A. arcticum p.p.)

Or have you reached the conclusion that this (and other plastid-data based) topology – Renner et al. 2008, Yu et al. 2022 – is just wrong and can be discarded?