7942 - Laserfiche WebLink

Home Browse Search

Interlaboratory Cross-checks Results of statistical analyses performed on the two sets of cross-check data are presented in Table 3. Results for indi- vidual samples are tabulated with corresponding monitor- ing data in Appendix A. Compared with earlier years (Schmitt et al. 1981), inter- laboratory agreement was generally good. Lab A significantly overestimated the lipid content relative to that estimated by CNFRL (Table 3). The fact that the paired- comparisons ANOVA and the intercept, but not the slope, were significant suggests a consistent additive (rather than multiplicative) disparity. Significant differences between our laboratory and Lab A in the results of one or more statis- tical tests were demonstrated for most of the compounds measured in 1976 and 1977. Agreement in all comparisons was evident only for 7-BHC, for which residues were at or near 0.0 µg/g in all samples (Table 3). Discrepancies were especially noticeable for multi-component residues, such as PCB's, toxaphene, and the chlordane-heptachlor complex, and for residues that co-elute with components of these mixtures (e.g., p,p'-DDT homologs). Interlaboratory agreement for 1978-79 was better than that for 1976-77 (Table 3). The Great Lakes Fishery Laboratory cross-checked fewer compounds than did Lab A, but significant differences occurred for fewer com- pounds; agreement was close for lipid content and for resi- dues of total DDT and total PCB. As was true for Lab A, however, agreement was generally poor for multi- component residues (e.g., PCB mixtures, toxaphene, and chlordane). General Trends in Residue Concentrations Mean, minimum, and maximum residue concentrations and percent occurrence of the compounds measured were determined for each collection period (Table 4). These statistics are based on all samples collected. We computed ANOVA for 1976-79 for 102 stations and for 1974-79 for 78 stations (Table 5). Means for each of these collection periods and the significance of differences among the means are presented in Table 6. The general interpretation of ANOVA results for resi- due monitoring data was discussed by Schmitt et al. (1981). As these authors explained, the "species within" line of the ANOVA represents variation between species at different locations and time periods. Comparison of the "species within" source of variation, for corresponding wet-weight and lipid-weight residues allows one to determine whether adjusting for lipid content explains differences between species. A further indication of the variation due to lipid content can be found by examining r2, which is the pro- portion of the variation in the residue levels explained by sources other than "within species" (error mean-square in Table 5); if accounting for variation attributable to per- cent lipid is important, then ANOVA in which lipid-weight residues are used should yield higher r2 values than the same analyses in which wet-weight residues are used. As Table 5 illustrates, our data generally refute the hypothesis that differences in organochlorine residues be- tween species at a given site are related to their differing lipid levels. "Species within" F-values for lipid-weight resi- dues, although generally lower than those for correspond- ing wet-weight analyses (Table 5), were nevertheless sig- nificant for all compounds except those where the data set was dominated by zero or very low values (Aroclor 1248, a-BHC, HCB, heptachlor, and [when 1974 data were included] endrin). The r2 values showed little, if any, improvement for most compounds when lipid-weight rather than wet-weight residues were considered; in fact, r2 decreased as often as it increased (Table 5). These results support two conclusions previously reported by Schmitt et al. (1981): (a) lipid content alone does not adequately ex- plain differences in residue levels between species at a given location; and therefore (b) more than one species must be collected if mean residue levels at each location are to be accurately reflected by the monitoring effort. Figure 2 is a matrix of inter-compound correlations for the years 1976-77 (below the principal diagonal) and 1978-79 (above the principal diagonal). Not surprisingly, weak correlations were present during both monitoring years between several compounds and percent lipid (posi- tive correlations) and percent moisture (negative correla- tions). Also not surprising were the strong correlations among several compounds and their metabolites and among related components of multi-residue mixtures (DDT homologs, PCB's, and members of the chlordane-hepta- chlor complex). Residues of persistent compounds that were formerly used to protect the same crops (Eichers et al. 1978) also tended to be intercorrelated (e.g., toxaphene, endrin, dieldrin, and DDT, which were all used in cotton produc- tion; and the cyclodiene insecticides chlordane, aldrin [which is metabolized to dieldrin], and heptachlor, all of which were used to combat corn rootworm [Diabrotica undecimpunctata] and a variety of cotton pests). The weak correlations between several chlordane components and PCB mixtures probably represent the co-occurrence of high levels of these compounds in the Great Lakes and in many northeastern and midwestern rivers. Temporal and Geographic Trends in the Levels of Specific Compounds DDT and its Metabolites As in previous years (Schmitt et al. 1981), residues of DDT or its metabolites were present at every NPMP station during the period 1976-79 (Table 4). Maximum total DDT residues (wet weight, about 4 µg/g in both collection periods; lipid weight, 211 µg/g in 1976-77 and 104 µg/g in 1978-79) were still high, but were markedly lower than the maxima of 13-48 µg/g wet weight and 402-873 µg/g