Critically evaluating the use of dark trait measurement in selection