Thanks for your thoughts on this.
Temp. was 197 (on the lower side for the used PLA), tree support branch angle at 40°. The other paramters wer the defaults from the tree support, flow parameters were unchanged, too. Print speed was set to 40mm/s.
I don't think that the gaps are caused by bridges because I watched the print head going straight between those points. I don't have a video of that part, only of a later time. In this image from that video (sorry, no better quality) you can see (somehow) the edges near the yellow marker line:
They seem to be inverted and not caused by bridges.
The main model didn't turn out so good, but that's because it contains overhangs which warp during printing. Unfortunately the warping leeds to pressure on one side of a supported part by the print head. The support from the tree structure is to weak with the current settings and so these parts break loose when the print head moves into the warped area (does this make sense?). It is this model, btw: https://www.thingiverse.com/thing:4750833, scaled to 90%.
I think, this model will work better with classical support and some additional support structures for part of the wings...