model to better understand results from various animal studies that showed signs of carcinogenicity from MTBE exposure. Key to this understanding, said Borghoff, is linking MTBE exposure to the dose of MTBE and metabolites that actually reach the target tissue and cause some effect. After contamination was found in 11 drinking water sources in California, state officials in February started requiring all public water systems to test for MTBE. The worst contamination to date resulted from underground gas tank or pipeline leaks in Santa Monica, where concentrations reached 94 ppb. "We have shut down the majority of the city's
wells," said Craig Perkins, director of the city's Environmental and Public Works Department. Perkins said the city believes MTBE has worked its way into both shallow and deep groundwater. Cleanup of Santa Monica's drinking water is at a standstill because the state doesn't know how to filter out MTBE contamination, said Perkins. Researchers and water authorities have found MTBE to be highly soluble and resistant to most filtration systems. Oregon Graduate Institute researcher Clinton Church said aeration seems to be the best way to get MTBE out of water. However it is much harder to remove the compound from the environ-
ment, said Church, because it vaporizes when removed from water and is extremely persistent. Church said researchers have found that MTBE biodegradation takes from one to three years. Another researcher at the Oregon Graduate Institute, James Pankow, has shown that the atmosphere in major cities where MTBE is being used may have become a nonpoint source for the transport of MTBE into shallow groundwater. MTBE in urban atmospheres partitions into rain in easily quantifiable measures according to Pankow's study. He will present his results at the American Chemical Society's meeting this month. VINCENT LECLAIR
EPA issues new guidance to improve states' water quality reports This month, EPA will issue new guidance that it hopes will lead to a more complete and accurate assessment of the nation's aquatic health. Under the Clean Water Act, section 305(b), states are required to submit a biennial report to EPA assessing the condition of their waters. The reports have been criticized for poor data quality resulting from nonrepresentative sampling methods, limited state monitoring resources and inconsistent use of biological indicators. According to Barry Burgan, EPA's national 305(b) coordinator, the reports cover less than onefifth of the nation's waters and data quality varies widely from state to state. Following the recommendations of an advisory group convened last year, the new 305(b) guidance advises states to submit reports on a fiveyear cycle, adopt a probabilistic sample design, and report on biological indicators. State water quality officials Steve Butkus of the Washington Department of Ecology and Chris Yoder with the Ohio EPA applauded the guidance as a way to reduce bias in assessments. But they questioned how many states would adopt the voluntary guidance. The switch to a five-year reporting cycle is being made, according to Burgan, because states were spending too much time preparing reports and not enough time monitoring water quality.
Beginning this August, EPA will ask states to report annually on 20% of their waters, so that after five years all bodies of water in each state will have been covered. Butkus called the switch from census sampling to probabilistic sampling "one of the most positive changes in the new guidance." Census sampling design requested that states directly
The agency advocates biological indicators, probabilistic sampling on 305(b) reports. sample all water bodies, but states have not had the resources to undertake such extensive sampling. Butkus reports that Washington State, for instance, was able to report on only 4% of its streams in 1996. He noted that census sampling led to targeted sampling of a few large water bodies, resulting in a national 305(b) report that was largely incomplete and heavily biased toward a few sampling stations. Probabilistic sampling takes data from a random sample of a state's watersheds and extrapolates the results to all watersheds in the state. According to Burgan, probabilistic sample design will enable EPA for the first time to track trends in water quality.
Butkus cautioned that states will need assistance because "probabilistic sampling design requires GIS [geographic information systems] skills for designing sample surveys and assessing data." Dave Chestnut, watershed coordinator with the South Carolina Department of Health and Environmental Control, questioned whether this approach would be economically feasible. He claimed that in his state the effort spent on sampling would have to nearly double because census stations that target problem areas could not be abandoned The move to report on biological indicators shows that "EPA is starting to recognize that there is a gradient of data quality among the states' 305(b) reports," concluded Yoder. He explained that although some states may base their water quality assessments only on chemical data from a few stations, others may have hundreds of stations sampling a full range of physical and biological data. Therefore Yoder said results among states are not easily compared Despite improvements to the 305(b) reporting process, critics are not entirely satisfied with the changes. Yoder noted that although EPA "has done everything to promote biological monitoring, short of requiring it," further incentives will be needed to get states to adopt it. JANET PELLEY
VOL.31, NO. 4, 1997/ENVIRONMENTAL SCIENCE & TECHNOLOGY / NEWS • 1 7 7 A