Well the thing with the planets with just Bio sites is that they’re all landables without volcanism.Just FYI, from what we've been told, not only are there no POIs, but non-landable planets don't currently have a real surface. They only get a generated texture - terrain generation is not run at all.
The aspect of this that really puzzles me is that FSS results for planets with only bio sites can apparently be generated quite quickly. It seems like those would also depend on terrain generation for placement. I wonder if, for bio sites, there is a short circuit whereby their existence and approximate position is determined while the terrain generation is still at a coarse level. That would explain why the sites sometimes jump around on final approach - perhaps the exact placement is only determined when the fine terrain mesh is generated.
I’d always wondered similarly if the issue was just with geo sites, but from what Stephen said it’s both geo and bio, which means it’s the volcanism part where the issue is. That actually makes a lot of sense as simming up a planet with active geology would presumably be quite a bit more involved than simming up an inactive planet.
Also, some bio sites definitely have a location dependency on specific surface features. Ejecta craters in the case I know of, but that’s quite a large scale feature. However, similar and more fine scale dependencies may well be there as well.