A field site was established at Beltsville, MD, in 1986 to assess the effect of conventional and no-till cultural practices on the movement of pesticides into shallow groundwater. Groundwater samples taken from unconflned (<1.5 m deep) and confined (<3 m deep) monitoring wells in 1986-1988 were analyzed for atrazine [6-chloro-A'-ethyl-A'-(l-methylethyl)-l,3,5-triazine-2,4-diamine], deethylatrazine [6-chloro-./V-( 1 -methylethyl)-!,3,5-triazine-2,4-diamine], alach-. lor [2-chloro-A'-(2,6-diethylphenyl)-A'-(methoxymethyl)acetamide], cyanazine [2-([4-chloro-6-(ethylamino)-l,3,5-triazine-2-yl]amino)-2methylpropanenitrile], and carbofuran (2,3-dihydro-2,2-dimethyl-7benzofuranyl methylcarbamate). Atrazine was found in groundwater all year, while cyanazine, alachlor, and carbofuran were present only for a short period (<3 mo) after pesticide application. Fairly constant background levels of <0.5 /tg L ' atrazine were found under fields treated before 1986, while levels under continuously treated fields were <2.0 /ig L" 1 for 22 of 25 samplings. Pesticide residues in unconfined groundwater were usually higher (ca. 2 to 4X) than in confined groundwater. Rainfall timing relative to pesticide application was critically important to pesticide leaching. A prolonged rain immediately after the 1988 application resulted in peak atrazine and cyanazine levels of ca. 200 fig L~' in unconfined and ca. 30 to 40 Mg L ' in confined groundwater, which resulted in short-term levels ca. 2 to 50 X greater under no-till than conventional till plots. Results of this study suggest that preferential transport occurred.

The percentages of technical aldrin, chlordane, endrin, heptachlor, Dilan, isodrin, BHC, and toxaphene remaining in Congaree sandy loam soil after 14, years were 40, 40, 41, 16, 23, 15, 10, and 45, respectively; those of purified aldrin and technical dieldrin after 15 years were 28 and 31, respectively; and the percentage of technical DDT in three soils after 17 years was 39. Treatments and maintenance of the soils were such that leaching, volatilization, photodecomposition, mechanical removal, and probably biological decomposition were at a minimum. These values may approach an upper limit of persistence of insecticides in soil.

Two or more pesticides together may produce a growth response in plants that is not predictable by their individual or independent toxicities. This unpredicted (dependent) response results from an interaction, a concept that usually is not easily interpreted. Dependent responses are further complicated by the fact that they can be either synergistic or antagonistic. Several methods exist for identifying and measuring phytotoxic interactions. Nearly all methods have certain shortcomings, however. Additive and multiplicative models (mathematical expressions) are the two basic approaches to determining pesticide interactions. The two-parameter, isobole, and calculus methods axe additive; whereas, Colby and regression estimate are multiplicative models. Regression estimate analysis considers deviations due to experimental errors, and a statistical significance can be attached to the interaction magnitude, thereby overcoming the deficiencies of the Colby method, but both methods seem to be limited to response data in which the combined pesticide concentration is the sum of the individual pesticide concentrations. The two-parameter method seems to be limited to response data in which the combined concentration is equal to the individual pesticide concentration and to response data in which a pesticide concentration necessary to produce a 50% of control value is interpolated rather than extrapolated. The calculus method is a mathematical expression of the growth response, and interaction is measured by derivation of the equation obtained. The calculus method is difficult to interpret and has a major weakness because it depends upon the multiple regression equation of the observed data. The regression estimate method is recommended as a reasonable approach to interpretation of interaction type data, with a SAS language computer program available from the author.

No abstract