The SAP noted this was a good start and that the Agency should refine its objectives and then develop a proposal, with clear goals and budgets, for further review. Given the uncertainty with many aspects of the design, the SAP also suggested that before proceeding with a National survey, a pilot survey on a local regional scale should be carried out to test estimation techniques, stratification issues, individual CWS sampling issues and the validity of the many assumptions used in building the study design. Also, it was noted that not all the information desired had to be collected across the entire survey. Some subsets of the samples in the study could address various details to reduce design and cost problems.
Other general comments are summarized by topic. These comments are followed by responses to the specific charge to the Panel.
Study Design and Funding
The SAP agreed with the Agency that the general goals of monitoring need to be two-fold: 1) obtain accurate estimates of exposure for as many pesticides as economic constraints will allow, and 2) collect the information so as to maximize use of the data to construct predictive models. One of the objectives of the monitoring study is to try to collect enough information on several pesticides in a single effort so that continuous monitoring is not required to estimate future exposure. The only way this objective can be met is if these data can be used to construct predictive models which could subsequently be applied to predict exposure in the future as land use patterns change, pesticide use patterns change, new pesticides are introduced, and precipitation changes.
However, the level of funding specified for this effort--7-10 million dollars--seems inadequate to meet the stated goals. At this level of funding, exposure estimates for only a handful of pesticides could be made with the precision needed. More importantly, this level of coverage would not be adequate to produce a predictive model that would estimate exposure reliable enough to negate the need for additional monitoring over time. The proposal by the American Crop Protection Association (ACPA) calls for a single-year study, which will not capture significant changes in weather or year-to-year variability in land management practices and pesticide use by agricultural producers. The ACPA statistical design, on which the cost estimate of 7-10 million dollars was initially based, seems flawed in the degree to which it assumes CWSs (and the watersheds providing the water for CWSs) to be homogenous. An appropriate statistical design must allow for the variability from year to year and the variability among CWSs relative to changes in pesticide use and pesticide fate and transport within each watershed. To collect sufficient monitoring data, in a one-shot sampling program that will be adequate for generating reliable prediction models so that annual monitoring would not be needed, would require a minimum investment of something on the order of 50 million dollars spread over a three year period. Even then, there would be a need to continue monitoring at some level to verify that