Assessing stream biological integrity across gradients of flow permanence and hydrologic connectivity

Le résumé fourni par la source

Ideally, stream bioassessment indices would be applicable, effective, and directly comparable across the full range of fluvial landscapes that occur within and across regions. However, few bioassessment indices exist for use in nonperennial streams, and those that have been used in nonperennial streams often perform poorly. In this study, we assessed how well 16 macroinvertebrate-based indices performed across streams that differed markedly in flow permanence and hydrologic connectivity. We asked if we could achieve acceptable index performance across a large arid region by 1) accounting for the effects of naturally occurring differences in flow permanence, seasonality, flashiness, and hydrologic connectivity on macroinvertebrate assemblages; 2) incorporating biological traits that reflect differences among taxa in their resistance and resilience to drying; 3) including rare taxa that are often excluded from indices developed for perennial streams; and 4) basing assessments on a single, most-sensitive metric. We developed indices with data collected from 382 reference-quality and 91 anthropogenically degraded streams in the western United States that spanned a wide range of flow regimes and degrees of hydrologic connectivity. We used random forest models to account for natural variation in all indices. Candidate predictors in these models included 33 frequently used landscape-scale predictors and 20 more-recently available measures of flow regime and hydrologic connectivity. We assessed several measures of index performance across 3 gradients of flow permanence or hydrologic connectivity. Performance measures included the ability of indices to correctly infer the condition of reference or physically degraded sites identified a priori. Measures of flow permanence were inferred from the USA’s National Hydrography Dataset and from models that predicted long-term (>40 y) means of the annual number of zero-flow days expected at a site. Additionally, to validate index performance, we used an independent set of sites from California that were classified as perennial or nonperennial based on more-direct measures of flow permanence, including a combination of site visits, water-level loggers, and best professional judgement. Based on the validation data, indices correctly inferred condition of perennial sites (56–93%) more often than nonperennial sites (46–81%). Given that index performance varies with flow permanence, different index threshold values may be needed for regulatory purposes. Index performance was not improved by inclusion of refined measures of flow regime and hydrologic connectivity as predictors or inclusion of resistance/resilience biological traits in indices. In contrast to observed/expected (O/E) indices developed for perennial streams, O/E indices that included rare taxa outperformed those that excluded rare taxa. Simple, modeled richness indices performed just as well as, if not better than, O/E and modeled multimetric indices, which indicates they may be practical, intuitive measures of biological condition universally applicable to streams that vary in flow permanence and hydrologic connectivity.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.