Bringing new knowledge to an old problem: Building a biotic index from lotic macroinvertebrate traits

Global environmental change necessitates a new approach to lotic bioassessment. Although biotic indices are widely accepted by regulatory authorities they are labor intensive to develop and biogeographically restricted in application. While trait-based assessment circumvents these problems, difficulties associated with non-discrete responses to stressors, less robust metric formulation (e.g. % abundance) and complications of the trait syndrome limit its potential uptake. We assumed that trait combinations, within the syndrome, would effectively represent synergistic, antagonistic and null interactions with respect to stress sensitivity. Applying this principle of positive and negative complementarity we constructed indicator scores for macroinvertebrate families for the complex stressor of organic pollution based on potentially relevant traits: respiration, habit, trophic class, flow preference, thermal preference, size and activity/mobility. We compared trait-derived sensitivity scores to empirically-derived scores from three versions of a biotic index: the BMWP, a statistically-based revision of the BMWP (s_BMWP) and the BMWP modified for the Iberian Peninsula (IBMWP). Trait-derived sensitivity scores were significantly correlated with empirical sensitivity scores: IBMWP>s_BMWP>BMWP. Despite the overall good correspondence, trait-derived scores tended to be consistently underestimated for Heteroptera, Odonata and Ephemeroptera in comparison with BMWP scores. Conversely sensitivity scores for Hirudinea and Gastropoda tended to be overestimated. While revision of the composite trait algorithm could, in some cases, improve these disparities they also indicate the importance of evolution and phylogeny in determining organism sensitivity to river pollution. Despite the potentially contentious issues of trait selection and trait weighting, the concept of trait complementarity is arguably more scientifically justifiable than regarding traits as isolated entities. It may offer the most pragmatic means to elaborate biotic indices where empirical methods are problematic. The subsequent ability to deconstruct sensitivity scores to verify their appropriateness would greatly facilitating moves towards harmonized assessment.