Your mistake is in thinking the null hypothesis is that a species-wide, complex morphological feature is non-adaptive.