Why may repeatability of highly flexible traits say little about their evolutionary potential?