Proto-Austronesian has a limited set of reconstructed vowel-initial roots, all of which are kin terms, body parts, or other readily possessed nouns.
The complete list of vowel-initial Proto-AN roots reconstructed by Blust is as follows: These are all the kinds of words expected in inalienable-possession systems.
[1] Blevins suggests that inalienable possession was lost from Proto-Austronesian, presumably after epenthetic *q- was added to vowel-initial words.
[2] Blevins suggests that in all three cases, the Proto-Austronesian *w- reflects the Proto-Ongan possessive prefix *gw- 'his, her', which remained as a fossil in some daughter languages.
Van Driem (2011) considers Blevins' evidence as "not compelling", although he leaves the possibility open that some resemblances could be the result of contact/borrowing, a position also held by Hoogervorst (2012).
Blust concludes that the grammatical comparison does not hold up, and also cites non-linguistic (such as cultural, archaeological, and biological) evidence against Blevins' hypothesis.