Regional frameworks and candidate metrics for assessing biotic integrity in mid-Atlantic highland streams

Regionally stratified biological criteria are being used increasingly to as sess stream quality. We used multivariate analysis of variance and canonica l analysis to examine the utility of two regional frameworks (basins and ec oregions) and 14 candidate metrics of local fish assemblages for assessing the biotic integrity of streams in the mid-Atlantic highlands (montane area s from Pennsylvania to Virginia). In particular, we determined (1) how metr ics varied naturally among basins and ecoregions and (2) which metrics vari ed most consistently with site quality. We also examined the ability of pre liminary multimetric indices (MMIs) to distinguish site quality. Metrics va ried meaningfully among both basins and ecoregions, but most metrics differ ed more among basins. The basin effect was especially strong for taxonomic metrics (e.g., number of species [TOTSP]), which reflected the influence of zoogeography on fish community composition. Few metrics differed strongly among both basins and ecoregions. Collectively, metrics distinguished among high, medium-, and low-quality sites within most regions, but the discrimi native ability of individual metrics differed by region. The number of dart er or sculpin species (DOSSP) was the only metric related to site quality b oth in most basins and most ecoregions. Metric differences among site-quali ty classes were more consistent with a priori expectations within basins th an within ecoregions. In each of five regions, we built an MMI from the mos t discriminative metrics. Only DOSSP and the proportional abundance of tole rants were included in all five MMIs. All MMIs included taxonomic and repro ductive (e.g., proportional abundance of simple lithophils, excluding toler ants) metrics, but not all included trophic metrics (e.g., proportional abu ndance of invertivores). Multimetric indices distinguished between high- an d low-quality sites in each region tested, but they usually did not do so t o a greater degree than did taxonomic metrics alone. Among the metrics incl uded in MMIs, TOTSP was most consistently related to site quality. Our find ings indicate that both basins and ecoregions provide useful frameworks for regionalizing biotic assessments based on fishes and that metric utility m ay vary considerably among regions even when regions are environmentally si milar. To enhance MMI performance, we encourage an increased reliance on re gion-specific empirical relations in the development of metrics and scoring criteria.