A provisional fish index of biotic integrity for assessing Ouachita Mountains streams in Arkansas, U.S.A.

Multimetric indices are often used to monitor aquatic-resource conditions. We used existing fish-collection data from streams to develop an Index of Biotic Integrity (IBI), which is a multimetric index, for the Ouachita Mountains ecoregion in Arkansas, U.S.A. Each fish-collection site was categorized as reference or non-reference. We examined 62 candidate IBI metrics, and selected 12 non-redundant metrics that differentiated best between reference and non-reference sites. The selected metrics were: Percent (of individuals) as Black Bass; Percent as Benthic Feeders; Percent as Centrarchids; Percent as Cyprinids; Percent as Ictalurids; Percent as Mineral, Site-Prep Spawners; Percent as Mineral, Site-Prep, Parental-Care Spawners; Percent as Simple, Mineral Substrate Spawners; Percent as Miscellaneous, Site-Prep, Parental-Care Spawners; Total Number of Centrarchid Species; Total Number of Arkansas Department of Environmental Quality (ADEQ) Ouachita Mountains Indicator Species; and Total Number of ADEQ Ouachita Mountains Key Species. We standardized each metric to score from 0 to 10 by using linear equations and threshold limits. Some selected IBI metrics had their scoring criteria adjusted to account for watershed size (i.e., stream size). We standardized the IBI to score from 0 to 100. In addition, we determined that our Percent as Black Bass and Percent as Benthic Feeders metrics contributed most to IBI scores in reference conditions, but their contributions decreased with decreasing stream conditions. Reproductive metrics contributed most in degraded stream conditions. Furthermore, we identified relations between IBI metrics and water-quality and land-use variables; some relations were counterintuitive. Unexpected relations may be random observations explained by limited ranges of land-use and water-quality variables. When select water-quality and land-use variables were included in a principal component analysis, a composite Land Use Intensity variable explained most of the model variance. Although the IBI has not been independently validated, the PCA, as well as other superficial analyses, indicated that the IBI should be able to differentiate stream conditions.