As a matter of fact, the preface of BDAG said it uses NA27. Therefore they would reference alternative readings as they did when they specifically endorsed the NA27 reading and called the other reading alternate because of the lack of punctuation in the earliest manuscripts.
The principal New Testament textual base is Nestle-Aland(27) (s. N. at the beginning of List 1). Some of the words appearing in BDAG are not noted in the apparatus of that edition. Frequent account is taken of variant readings which are not recorded in the Nestle-Aland apparatus. Thus the word gene,qlioj appears Mk 6:21 in codex D, and is so noted in BAAR, but not in the Nestle-Aland apparatus. Although some words, such as avkaqa,rthj or na,rkh, would not ordinarily merit attention, either because they are editorial inventions which have wormed their way into respected editions without strict accounting (e.g., some Erasmian readings) or are the result of erroneous restoration of a fragmentary text, their appearance in learned discussions requires identification with cautionary notice in this lexicon. Since little advantage is to be gained by cluttering the lexicon with an unscientific collection of selected manuscript sigla, users of the lexicon are encouraged to find the sources for such variants in earlier editions of Nestle, in Tischendorf (s. Tdf. at the beginning of List 1), or in other critical editions. For related reasons I have not included in citation of text the brackets used in many passages in Nestle-Aland.