I recently updated my Pama-Nyungan database to show some summary stats. Here are the subgroups with the most data points:
So far, there are 3607 reconstructions. Yolŋu again leads, with 807, and Karnic is second. However, about 1/3 of the reconstructions haven’t yet been labelled with a subgroup so it’s hard to give meaningful figures yet.
Eventually, I will be able to give stats about innovation and retention among languages and subgroups, but that’s a long way off yet.
There are 190,000-odd language headwords in the database at present, but that’s about to go up substantially.