Pesticide analysis quality assurance
Chapter 9, PLAP - Monitoring results May 1999 - June 2002
Scientifically valid methods of analysis are essential for the integrity of the present monitoring programme. The field monitoring work has therefore been supported by intensive quality assurance entailing continuous evaluation of the analyses employed. Pesticide analysis quality assurance (QA) data for the period July 2001-June 2002 are presented below, whereas the QA data for the preceding monitoring periods are given in Kjær et al. (2001) and Kjær et al. (2002).
9.1 Materials and methods
The pesticide analyses were carried out at two commercial laboratories selected on the basis of a competitive tender. In order to assure the quality of the analyses, the call for tenders included requirements as to the laboratory's quality assurance (QA) system comprising both an internal and an external control procedure. In addition to specific quality control under the PLAP, each of the laboratories takes part in the proficiency test scheme employed by the Danish Environmental Protection Agency when approving laboratories for the Danish Aquatic Monitoring and Assessment Programme (NOVA-2003).
9.1.1 Internal QA
With each batch of samples the laboratories analysed one or two control samples prepared at each laboratory as a part of their standard method of analysis.
9.1.2 External QA
Every third month, two control samples were analysed at the laboratories along with the various water samples from the 6 test sites. Two stock solutions of different concentrations were prepared from 2 standard mixtures in ampoules prepared by Promochem, Germany (Table 24). Fresh ampoules were used for each set of low and high standard solutions. 150 µl or 350 µl of the pesticide mixture was pipetted into a preparation glass containing 10 ml of ultrapure water. The glass was closed and shaken thoroughly and shipped to the staff collecting the samples. The staff finished the preparation of control samples in the field by quantitatively transferring the standard solution to a 3-l measuring flask. The standard solution was diluted and adjusted to the mark with groundwater from an upstream well. After thorough mixing, the control sample was transferred to a sample bottle and transported to the laboratories together with the regular samples. The standard solutions were prepared 2 days before a sampling day. The pesticide concentration in the solution is indicated in Table 24. Blank samples consisting of HPLC water were also included in the external QA procedure every month. All samples included in the control were labelled with coded reference numbers so that the laboratories were unaware of which samples were controls and blanks.
Table 24. Pesticide concentrations in the spike solution and in the high-level and low-level control samples.
9.2 Results and discussion
9.2.1 Internal QA
The internal QA data have been analysed to obtain an impression of the day-to-day variation and within-day variation. The statistical analysis encompasses all duplicate pesticide analyses, single analyses being excluded. One-way analysis of variance was used to separate day-to-day variation from within-day variation. The results are presented in Table 25.
With 38 out of 47 of the pesticides, day-to-day variation accounted for most of the uncertainty. Thus when st exceeded 10, this was due to a high day-to-day variation. F>Fcritical indicates that the day-to-day variation is significantly higher than the within-day variation (95% confidence interval). F
The overall standard deviation (st) of the various pesticide analyses lie within the range 0.003-0.029 µg/l. Reproducibility of the degradation products was generally found to be poorer than that of the mother compounds, although within the same range. Standard deviation for mother compounds and degradation products was in the range 0.003-0.025 µg/l and 0.004-0.029 µg/l, respectively. With seven of the compounds, reproducibility was relatively poor (st ³0.02).
Table 25. One-way analysis of variance of pesticide analyses. a = 0.05.
9.2.2 External QA
Table 26 provides an overview of the recovery of all spiked samples based on 1-3 observations. Recovery of the spiked samples is generally good (>70%). Exceptions are desmedipham and phenmedipham, glyphosate, flamprop free acid, fenpropimorph and pirimicarb, for which recovery was low at some of the field sites.
Table 26. Average recovery (%) at low/high concentration level indicated for each site. Recovery refers to the ratio of the observed and nominal concentrations.
The low recovery reported for desmedipham and phenmedipham is most likely due to stability problems. Degradation products were thus detected in the spiked samples - see Appendix 11, where the concentration of mother compounds and degradation products in the spiked samples is indicated on the control cards. Stability problems were also observed with these particular compounds by Kjær et al. (2002) in an analysis of the stability of a large number of compounds. Total recovery, including also the degradation products, thus provides a more realistic picture of the recovery. As acidity and other water quality parameters differ between sites, matrix effects may explain some of the differences in recovery between sites.
Table 27 provides an overview of the number of times each compound was detected at each site during the 2001/-2002 monitoring period together with the maximum concentration. Nineteen pesticides and twelve degradation products were detected in samples from the experimental fields, and QA data connected to these findings are of special interest.
Table 27. Number of times each compound was detected at each site during the 2001/2002 monitoring period together with the maximum concentration (µg/l) in parentheses.
Recovery of pesticides in external QA samples was found to be acceptable for the great majority of pesticides detected in field samples (recovery >=70%) Exceptions are desmedi pham and pirimicarb at Silstrup, fenpropimorph, glyphosate (low concentration) and pirimicarb at Estrup and glyphosate at Slaeggerup. The internal QA data generally showed good recovery. With some compounds, however, a few of the results are either high (>130%) or low (<70%). This is also reflected in the table of standard deviations (Table 25), typically resulting in st values above 0.02 µg/l. For all compounds that have been detected in more than one sample, the external and internal QA results are shown in Appendix 11.
No pesticides were detected in blank samples, thus indicating that no contamination of the samples occurred in the laboratory. Samples found to contain pesticides and their degradation products are thus regarded as true positive findings. All the pesticides in the spiked samples were detected in all samples.
9.3 Summary The overall quality of the pesticide analysis was considered satisfactory. The QA system showed that:
- Reproducibility of the pesticide analyses was good, standard deviation being in the range 0.003-0.025 µg/l.
- Reproducibility of the degradation products was a little poorer than that of the mother compounds, being in the range 0.004-0.029µg/l.
- Recovery was generally good (>70%) in external QA samples, exceptions being fenpropimorph, desmedipham, glyphosate, phenmedipham and pirimicarb at single sites.
- Variations in recovery of the same compound in spiked samples from all field sites in dicate uncertainties in analysis caused by differences in matrix composition.
- No contamination of samples occurred during collection, storage and analysis.