I'm not sure where the problem is. The output seems to show there might be something different about the range of values in the data. Try 3dUnifize first or scale separately with 3dcalc. Another way is to align to a template with something like the method used in the macaque alignment script by transforming the template brain back to the native space (See macaque_align.csh). Also consider you may not need to skullstrip at all depending on what you want to accomplish.