The idea of ancestral area reconstruction is to reconstruct ancestral areas. For this, each tip taxon has to be coded for its modern area (ideally not more than one, because all AARs tend to end up with ambiguous reconstructions the deeper the node).

You coded each species for the entire area of the genus, in your data mostly represented exclusively by (sometime geographically very restricted, see Appendix 4b to the cited Grimm & Renner 2013) Chinese species. And why code both Corylus the same if C. wangii only occurs at the Weixi Xian mountain (Yunnan: 1) but C. avellana across western Eurasia (2 according your coding)? Likewise, B. cordifolia (E Canada, NE U.S) should be just 3 in contrast to the circum-arctic B. nana (the only actual 123).

Generalising coding is generally a bad idea and particular problematic in Betulaceae because Ostrya spp. are not part of an exclusive clade (when using plastid data): the North American (earliest diverging) Ostrya (not included in your sample) are sister to the East Asian Ostrya + Carpinus (the latter is predominantly Eurasian with a single species each in North America and Central America) Alnus and Betula have relative complex biogeographic patterns.

AAR should include at least data on one representative per occupied area, otherwise it's completely useless.