Reconsidering functional redundancy in biodiversity research