The three factors, extent of use area, environmental fate properties, and risk, should indeed be the primary factors in the process for selecting chemicals in a national drinking water survey design. These are factors that have been reviewed and discussed for years as key elements for such designs. None of the individual selection criteria provides a logical or completely satisfactory set of target pesticides. This monitoring study has multiple objectives, each requiring a different emphasis on pesticide selection factors. Therefore, the combination of criteria should be used, although not necessarily equally weighted.
Extent of use area is appropriate for a national survey of this scope that will focus on multiple pesticides. A national focus will also help insure that a range of climatic-hydrogeologic conditions is encompassed for model development. Generally, high risk pesticides should be included. The Agency stated that the list of pesticides requiring exposure estimates have been reduced using screening models (pesticides that are not expected to occur in water) and using extant monitoring data. Candidates high on the list included pesticides where the "risk cup" was partially (or nearly) full based only on exposure through food. This is the appropriate priority for meeting the goals of the Food Quality Protection Act, but it cannot be expected to provide enough variability in fate and transport characteristics to construct prediction models that are robust enough to make estimates of pesticide exposure for pesticides not included in the survey or for new pesticides. The SAP raised the question if non-human, ecological risk should also be considered, particularly if raw water occurrence was also a target for the survey, in addition to finished drinking water.
For evaluation of model performance, one needs to make sure that a reasonable range of pesticide fate properties are represented (as well as a range of environmental conditions). Additional pesticides might need to be added because of their fate and transport characteristics. Without a "representative" set of pesticides relative to fate and transport characteristics, reliable predictive models cannot be estimated, and additional monitoring might be required to make exposure estimates for the remaining pesticides. Therefore, the multiple objectives of this study require some flexibility in the selection procedures.
Also, another practical (and economic) factor that should be considered is what pesticides and transformation products might be covered by multi-analyte methods. Since cost containment may be a limiting factor in this monitoring study, it is important to maximize the information that can be obtained from efficient chemical analyses. However, this should not be the primary factor in selecting the target pesticide. Also, pesticide degradation/transformation products have been mentioned to be included. The purpose of including them should be carefully spelled out.
In addition, results from previous monitoring studies (USGS-NAWQA, etc.) should be consulted to identify "problem" chemicals that need to be included. Inclusion of chemicals that pose low probabilities of detection might be avoided, but this must be done with care. Advancing model development and testing the actual site selection protocol will be a critical component. A range of high to low use, a range of hydrogeologic, and fate and transport conditions must be