Cool beans. Unfortunately it doesn’t seem to be saying much :P. None of the parameters are wildly big or small.
But looking back at your data, have you considered doing some sort of zero-inflated model looking thing? Check page 197, 13.7. Zero-Inflated and Hurdle Models in the 2.17.0 manual. It looks more like the situation is that than a mixture of two normals. Maybe that’d work better.
Back to “I wasn’t worried, since the divergences are quite well distributed in space” – it’s kinda hard to get much out of where the divergences are when they’re all spread around unfortunately. Any divergences are still bad news bears. @martinmodrak has a neat example here Getting the location + gradients of divergences (not of iteration starting points) - #5 by martinmodrak, but that’s down the line.