Pattern Analysis of Fish Communities Upstream/ Downstream of Pulp and Paper Mill Discharges on Four U.S. Receiving Waters

摘要

This study compares the spatial and temporal patterns of fish communities seen along longitudinal gradients in four rivers that include pulp and paper mill effluent point source discharges. This project is part of the larger Long-Term Receiving Water Study (LTRWS), a 10 to 20 year program designed to address questions of community gradients along receiving waters, being conducted on four North American rivers. The four receiving waters represent a diversity of geographic, mill process, and river conditions. Codorus Creek in south central Pennsylvania is an effluent-dominated wadeable stream with a bleached kraft mill; the Leaf River in south central Mississippi is a large river with a bleached kraft mill; the McKenzie River in central western Oregon is a large river with an unbleached kraft mill; and the Willamette River in western Oregon is a large river with a bleached kraft/deink operation. Collection methodologies varied between rivers due to river size and site accessibility; the smallest receiving water, Codorus Creek, had only backpack electrofishing and the three larger rivers had nearshore backpack as well as boat electrofishing. For the McKenzie and the Willamette Rivers, only the boat electrofishing data were analyzed. This report presents multivariate analysis of fish community patterns for each river for multiple years and seasonal pattern analysis for three of the rivers (Codorus, McKenzie and Willamette). River to river comparisons of the assemblage patterns is made on the assumption that sample collection methods, while varying between rivers, were all designed to allow for capture of a representative fish assemblage for each river. Analysis for patterns was done using multivariate statistics; evenness and diversity indices were also calculated for the fish communities for each site on each sampling date for all the rivers and then analyzed for significant differences. Codorus Creek had an upstream to downstream gradient in fish communities with each site significantly different from the rest. The fish community pattern, which was driven by 13 fish, was correlated at a low level with the water variable pattern, including some variables considered to be effluent markers. The fish pattern was not significantly different between seasons or between years. The Leaf River demonstrated no significant pattern between sites; however, there was a significant difference between years. The McKenzie River had a significant but weak difference between the upstream sites and the two sites immediately downstream of the mill. This pattern was driven primarily by salmonids and largescale suckers and was correlated with water quality variables not exclusively associated with mill effluent. The largest river, the Willamette, had no fish community gradient, with sites not significantly different from each other. The evenness measurement, Pielou's J', and the diversity measurements, Shannon H' and Simpson's 1-λ, showed no consistent pattern of significant differences for any of the rivers. This study demonstrates that multivariate analysis of fish assemblage data can identify community patterns and the organisms and water quality parameters driving the patterns. In addition, this analysis has demonstrated a pattern of relative magnitude of response in receiving waters from the small effluent dominated receiving water, with a significant pattern response, to the large non-dominated receiving waters with little or no detectable changes in fish community patterns.