Cheers! There are a few models that try to account for this bias, here is one.

I agree WRT the species/groups with very very few records, and just adding a footnote to say so.

WRT attribution for data etc, it's something I'd like to do, but haven't learnt how, yet..

Cheers!