The suggestion by AIMED is good.
But if you can't scan again that subject, please give a try to this (it isn't a perfect alignment method, but still can work with a good % of approximation): take one of the two sans you have, RMB (Right Mouse Button) on the scan's name in the list of scans, flip normals, align the flipped one to the first scan. When the raw alignment is done, RMB on the same scan as before and flip again normals. That should work enough because the wings have a similar flat areas that is so thin to be considered like the same one (this could work with leafs, and sheets, too). Then adjust manually the second scan looking at the borders. David doesn't have a "pivot" feature (I mean a feature that allow the user to choose a customizable point that work as scan's center, so you could refine your manual alignment via other software that have this feature).
Give it a try if you don't need the perfect relative positions betwenn the two scans. Otherwise you are forced to scan again the butterfly, following the previous suggestion.
In any case, don't forget to set high the Fusion resolution, at the end. If you will set low Fusion values, you could have holes in the wings' areas.