Chemistry, physical habitat, and fish communities

Abstract

Water resource monitoring is often focused on exceptional sites, that is, sites of noticeably high or low quality. There may be both intentional and unintentional reasons or causes for the selection of sampling sites. One sound reason for this sampling bias is the high social and scientific priority that our society places on preservation of sites with the best conditions and remediation of those with the worst. The emphasis on impaired water monitoring was codified in the Clean Water Act requirement for 303(d) listings. More recently, relatively undisturbed “reference quality” sites have received attention for their role in setting expectations for biological criteria. The Clean Water Act antidegradation mandate applies, however, not only to exceptional sites but to all waters. Reporting on the overall condition of aquatic resources as mandated by provision 305(b) has been supported by maintenance of a network of long-term monitoring sites. Longterm monitoring of repeat sites is a good tool for detecting trends, but may not be the best estimate of overall status. More importantly, there is no guarantee that hand-picked sites are representative of overall resource quality. For example, easy-access sites may be of relatively low quality due to their proximity to human influence, whereas resource scientists’ “preferred” sites may be of unusually high quality. Unbiased sampling can be achieved either via exhaustive census or true random sampling. Random but “even” sampling of an unevenly distributed resource presents a statistical challenge. The probabilistic site sampling methodology associated with EPA’s EMAP (Environmental Monitoring and Assessment Program) project was developed as an answer to this challenge. Sites are selected from a representation of the known extent of the target population and weighted based on their proportion in the population; the sampling algorithm is spatially balanced but random. Results can be extrapolated to the target population with known confidence. In the case of this report, measurements are extrapolated to perennial wadeable stream miles in the state of Kansas. The randomly selected sites in this study were established as part of a larger project begun in 1994-95 to evaluate the health of fisheries in USEPA Region VII. This dataset also includes a number of putative reference sites, handpicked by regional resource scientists.