Cost Effective Methods of Monitoring Pesticide Pollution in Water Systems
Report No. 1120/1/04

March 2004


South Africa is the largest market for pesticides in sub-Saharan Africa. Despite this, relatively little is known about the environmental and health impacts of widespread pesticide use, particularly in rural agricultural areas. Previous research supported by the Water Research Commission (WRC Report No. 795/1/00) found consistent low level pesticide contamination of rural water sources in three Western Cape agricultural areas and highlighted the need for monitoring of water sources for pesticides. In response to this finding, a 3-year follow-up study between 2000 and 2002 was undertaken to explore the potential development of cost-effective methods for monitoring for pesticide pollution. The research was conducted collaboratively by the Occupational and Environmental Health Research Unit at the University of Cape Town and the Departments of Analytical Chemistry and Chemical Engineering at the Peninsula Technikon (PENTECH) - the same group that completed the baseline contamination study from 1997 to 1999.

The study was motivated by the fact that in South Africa, infrastructure is poorly developed to monitor and control pesticides reaching water sources, particularly in rural farming areas. Methods for the detection of pesticides in water are currently constrained by a number of practical limitations: 1) relatively few laboratories undertake pesticide analysis and few provide details of rigorous quality assurance; 2) high expense of testing; 3) barriers posed by commercialisation of testing services, and 4) practical difficulties relating to collection and transport of samples. All these factors result in poor availability of testing.

Moreover the human resource and technical, support available in rural communities to implement any testing programmes for pesticides is largely unknown.

The project therefore addressed some of these serious gaps in the environmental management of pesticides. At the same time, the project intended to contribute both to improving quality of rural life, while at the same time promoting environmental justice and empowerment of rural communities. It is these vulnerable groups, rural communities and farm workers, who are most at risk from undetected and uncontrolled pesticide exposures in water.


The aim of the project was to evaluate cost-effective methods for monitoring pesticide pollution in rural water.

The objectives were:

  1. To evaluate the utility of solid phase micro extraction (SPME) fibres and a programmable sampler as tools to achieve integrated sampling of water sources being tested for pesticide pollution.
  2. To evaluate the utility of enzyme immunoassay kits for the quantitative and semi quantitative identification of pesticides in water .
  3. To develop a set of procedures /guidelines for use by rural communities and/or rural local authorities for monitoring water sources for pesticide pollution (based on the results of objectives 1 and 2, in conjunction with findings of the WRC project No. 795/1/00).
  4. To estimate the costs associated with the procedures evaluated under objectives 1 and 2, and compare these costs to conventional methods for monitoring.
  5. To formulate the procedures evaluated under objectives 1 and 2 in the form of a manual.
  6. To evaluate the capacity existing in rural communities and local authorities to implement monitoring of water sources for pesticides, and to make recommendations for strengthening such capacity.

The project therefore comprised a set of discrete sub-studies, which included in-situ laboratory development (SPME utility), analysis of field samples (SPME utlility, immunoassay testing), field surveys (human resource and community capacity) and health economic analysis (costing). The study area and sample sites were those that had been extensively studied by the research team in the previous investigation. The Hex River district, an intensive grape-farming area with high pesticide usage was therefore chosen as the study area for this project. Samples were taken from 2 points along the course of the Hex River which represented a sequential flow downstream, two water reservoirs situated alongside vineyards, and a drain in the vineyards that collected water from soil surface run-off. Similarly, the two pesticides examined, endosulfan and chlorpyrifos, were those previously found to be most prevalent in the region.


Solid-phase microextraction (SPME) utilizes a short, thin, solid rod of fused silica that is coated with an absorbent polymer. The coated SPME fibre is attached to a metal rod, and both are protected by a metal sheath that covers the fibre when not in use. Solid phase microextraction (SPME) was introduced to eliminate the problems associated with Solid Phase Extraction (SPE). SPME sampling involves the extraction of an analyte into the polymer coating, an equilibrium process based on the partitioning of the analyte between the sample phase and the polymeric coating. By using SPME rather than SPE, solvents are completely eliminated, blanks are greatly reduced and extraction times may also be greatly reduced. It is not only a quicker alternative, but potentially a more cost-effective one as well.

The experimental work confirmed that vigorous stirring (agitation) of the sample is essential in order to achieve reproducible results. A series of experiments using absorption times of 5 to 25 minutes, in increments of 5 minutes, showed that the improvement in reproducibility beyond an absorption time of 20 minutes for the two pesticides under consideration, was neglibile. An absorption time of 25 minutes was used in the majority of subsequent analyses as this conveniently coincides with Gas Chromatograph (GC) cycle times, and is similar to absorption times reported in the literature.

The time required to achieve equilibrium (maximum) absorption was explored by increasing the absorption time from 20 minutes to 140 minutes, at intervals of 20 minutes. Maximum absorption, for the analytes concerned and under the experimental conditions, occurred for an absorption time between 60 and 80 minutes.

The reproducibility of the SPME/GC sampling and analytical method was assessed by comparing the differences between replicate pairs. Under the optimum agitation conditions (N = 625rpm) and the absorption time (25 minutes) established for the analytes under consideration, the SPME method yielded excellent reproducibility. The relative standard deviation of the differences between 17 replicate standard analyses was about 1% for 3 of the analytes, and less than 10% for endosulfan II, for standards with concentrations in the range 0.1 to 5g/L. For the 3 field samples analysed, the relative differences between replicates ranged between 1 and 10% for the 4 analytes under consideration, for analyte concentrations in the range 0.1 to 6.0 g/L.

Method Detection Limits were estimated based on a ratio of 3x the Standard Deviation of a set of 7 determinations, similar to methods recommended by the US Environmental Protection Agency (USEPA). Limits of 0.01g/L to 0.02 g/L were established for the analytes of interest.

The SPME method is quicker and simpler to use than traditional Solid-Phase Extraction. The fibres are reusable for up to 100 analyses. When compared to other techniques of analysis, such as Solid Phase Extraction (SPE) and Enzyme Linked Immunoassays (ELISA), the SPME method rendered more reliable results on a consistent basis without being time consuming or labour intensive. Both reproducibility and detection limits are superior to the SPE method.


An accurate assessment of the water quality in a particular region currently requires a relatively frequent (and costly) sampling regime because it relies on multiple grab samples to characterise contamination over a particular period. However, the costs of attempting to assess water quality at a number of sites through frequent grab sampling are prohibitive. In practice, grab samples are taken at intervals varying from twice per week to weekly or at greater time intervals, and average values are inferred from what are actually relatively infrequent grab samples. A sampling method that yields a Time Weighted Average (TWA) concentration over a comparatively extended period (24 hours or longer), using a single sample, would give a more accurate picture of prevailing contaminant levels.

An unsteady state theoretical analysis of the rate of mass transfer of the analyte (pesticide) from the sample, assumed to be flowing perpendicular to the face of the fibre, to the fibre is presented. The theoretical analysis develops previous steady state analyses, used to establish the basis for measuring TWA concentration of volatile or semi-volatile compounds in air, and uses the mathematical results of analogous unsteady state heat transfer problems. The theoretical analysis presented in this report showed that the SPME system could be adapted to obtain a TWA sample of pesticides in water, at concentrations of 1 g/L. The system could be operated with mass loadings of 30-40% of the equilibrium values, with a corresponding increase in method sensitivity (detection limit) provided that the linearity of the mass loading rate through the sampling period was preserved by operating the system at a sufficiently low Mass Transfer Biot number for the sample period under consideration. The theoretical analysis also demonstrated the impossibility of satisfying both linearity and sensitivity criteria with a single fibre located in a particular configuration if the analytes of interest have a wide range of distribution constants (Kfs values) - in this case the ratio of the highest to lowest Kfs values, based on literature data, is approximately 50:1. A multi-fibre system is required for these cases. Temperature dependence of Kfs values may be significant, therefore in-field temperature control of the fibre should be considered for maximum reproducibility.

In-situ laboratory experiments were conducted to estimate: 1) the diffusivities in water of the two analytes under investigation 2) the distribution constant values (Kfs) 3) analyte retention on the fibre 4) linearity of mass absorption with respect to time and concentration, and 5) the effect of exposure concentration profiles on detections. Comparison was then made between the modelled TWA based on SME analysis and average concentration of 24 separate hourly grab samples collected by an autosampler.

The experimental results confirmed that in principle and in a practical environmental setting, the SPME device could be adapted to obtain a TWA sample of the pesticides of interest, over a sampling period of up to 3 hours. The laboratory-based experiment showed that a similar (agreement within 6%) TWA sample was obtained for two markedly different exposure patterns with the same theoretical TWA concentration, a key requirement for the success of the method.

The field test over 24 hours showed that a TWA sample could be obtained with minimal effort. However, the TWA sample result significantly underestimated the average concentrations of 24-hourly samples. However, the fibre failed to retain, for the time period required, 100% of the analyte absorbed, with losses of over 50% over a 2-hour period. Based on limited experimental data, this loss factor was assumed to be constant, and the calculated TWA result was adjusted accordingly. The validity of this calculation procedure has yet to be confirmed experimentally. The loss of analyte clearly degrades the sensitivity (and possibly the reproducibility as well) of the method. The losses over 24 hours may be as high as 90%. The loss of analyte from the fibre may be overcome by in-fibre derivitisation, or chemically fixing the absorbed analyte, but this work was beyond the scope of the present project.

At the end of the study, insufficient experimental results were obtained to provide estimates of the accuracy and sensitivity of the SPME-based TWA sampling method. However, the prototype field sampling system that was used is essentially simple. It consists of a custom-made fibre holder and a small sampling pump. Further research would be needed to overcome the problem of desorption of the analyte during the extended sampling period.


Enzyme-linked Immunoassays are tests based on the reaction of selective antibodies attached to solid supports with sensitive enzymes and analytes. They are capable of detecting low levels of pollutants with high specificity. For example, detection limits for chlorpyrifos vary from 0.1g/L to 3.0g/L and for endosulfan (as mixed isomers) from 0.08g/L to 1g/L. Immunoassay methods vary depending on the compound or compounds they are capable of detecting. The assays can be class-specific (they are able to screen for a certain class of compound), or compound-specific (they have a high specificity for the compounds of interest).

Two variations are commonly in use, one using an immobilizing antibody and the other an immobilizing conjugate on a solid-phase support such as a 96-well microtiter plate. The coating conjugate on the microtiter plate must differ from the immunogen to prohibit binding from antibodies that recognize the carrier protein. ELISAs for endosulfan and chlorpyrifos are in the form of plate kits. ELISA methods hold the promise of decreased sample processing, high compound specificity and significant increase in the number of samples that can be analyzed compared to traditional analytical techniques.

To evaluate the ELISA methods for measurement of endosulfan and chlorpyrifos, samples were tested in the laboratory to estimate the calibration curve, and on field samples to compare ELISA methods to SPME (1 set of tests on samples from 9 field sites taken on the same day). In addition, a set of two triplicate analyses for endosulfan at 1ppb and at 0.1ppb respectively, were conducted to assess replicability at different concentrations.

Consistent detections by ELISA were made in the field samples. Contamination was comparable to that found in the previous study (WRC Report No. 795/1/ 2000). When comparing the ELISA method to the Solid Phase Extraction (SPE) method, qualitative agreement (present/absent) was high (8/9) for chlorpyrifos, but much lower (5/9) for endosulfan. One third of endosulfan results were false positives. Quantitative agreement between the two methods was weak. Triplicate analyses suggested good reproducibility for endosulfan estimation at both concentrations tested (range of difference = 15% at 1.0 ppb, and 7% at 0.1ppb).

From the tests performed, it appeared that the ELISA method could be useful as a semi-quantitative tool. Experience in using the ELISA for analysis indicated that the reagents were safe to use but were not as long-lasting, effectively limiting the value of the method unless there are high volumes of tests required. The ELISA method is therefore likely to be a useful screening tool for multiple samples as the method is not as time consuming as traditional methods of analysis.


To establish the utility of different monitoring methods for pesticide pollution, comparisons were made of the costs of the three analytical methods, viz.1) ELISA, 2) SPE and 3) SPME; and of the costs of the two integrated methods of sampling 1) 24-hour TWA SPME and 2) 24-hour auto sampling SPME. Comparisons in all cases were made in terms of the average cost of analyses per sample.

Standard econometric methods were used for the costing, which included estimation of direct costs (including capital costs such as equipment and recurrent costs such as labour) and indirect costs.

Of the three analytical methods, ELISA, SPME and SPE, the cost appraisal found SPME to have the lowest cost per sample (R253.86), followed by SPE (R329.86), while ELISA had the highest cost per sample, appreciably higher than SPME. For all three methods, the annualised capital cost was low and formed a low percentage (5-9%) of the total annual cost. Recurrent costs therefore formed the bulk of the costs of all three methods, with transport costs high for all three methods, and personnel costs for SPE and SPME being substantially higher than ELISA. The cost of materials were high for ELISA (R 240 per sample).

Sensitivity analysis showed that if the full cost of capital items (as totally dedicated to the analyses of the 384 annual samples) were included then the cost per sample for all methods increases substantially. ELISA would then rise to R450.66, for SPE to R429.86 and SPME to R502.01. In order to obtain the low cost per sample quoted above, 786 samples per annum are required for ELISA and 3200 samples per annum for SPE and SPME.

Despite an a priori expectation that the ELISA would be a useful method if there was sufficient economy of scale, costing analysis has shown that there is little to save with high volumes of tests since the highest costs for ELISA are not personnel but reagents. Moreover, given that SPME and SPE produce quantitative data with good detection limits (SPME 0.01-0.02 g/L), and that the ELISA measurements are semi-quantitative, there appears to be little advantage to promoting the wider use of ELISA's at this stage.

The costs associated with using the autosampler compared to TWA for obtaining an integrated estimate for 24 hours were almost twice (R1707) that of the TWA derived from the SPME fibre (R865), largely because of the capital costs of the autosampler and time required to analyse each sample. The use of the autosampler in the field combined with laboratory SPME and GC analysis yields a quantitavely superior result compared to the use of the SPME fibre for in-field TWA sampling. However, it is evident that neither of these two methods are presently feasible for routine use in monitoring due to the logistics and costs associated with the collection and analyses of samples. Further investigation would be needed to establish their utility in improving the characterisation of rural water contamination before such high costs could be considered, or further refinements need to be developed to reduce costs.


In order to implement monitoring of pesticides in water, it is important to evaluate the capacity of rural local authorities in the Western Cape and other public sector responsible for water management.

The survey was intended to characterise the knowledge, perceptions and practices of rural target groups in relation to pesticide pollution of water, and to identify needs for training, technical and logistic support. A cross-sectional survey of farm workers, environmental health officers (EHOs), local authority management, rural NGOs and Government department staff was planned to meet these objectives. Questionnaire development, and field interviews of local authority officials and farm workers was undertaken by environmental health students from PENTECH supervised by 2 senior researchers from PENTECH and one from UCT. The survey was conducted during September and October 2001. For logistic reasons, local authorities and farms were surveyed separately, and different questionnaires were applied.

Final participants in the study included 8 EHOs. These participants were drawn from 7 rural farming districts in the Western Cape, which cover 21 towns with a total population of 503 000 From the management of water quality perspective, it was reported that in all 7 districts surveyed, water sources on farms are tested by EHOs. While the number of EHO's per district varies according to the population in the area, it is clear that in some areas there is a shortage of personnel. Only 3 (37%) of respondents felt that there were enough persons conducting water monitoring in their area. All respondents indicated that the costs of monitoring are borne by the local municipality, and only one respondent indicated that private farm owners may cover the cost if they request testing. Feedback of results of pesticide analysis was not common (only one respondent), while feedback to the farmer after request was relatively common (3/7 respondents).

The 63 farm personnel who participated in the study, included 6 sprayers (10%), 35 managers (57%) and 20 other workers (33%). They resided on 16 farms in 3 districts. Farm personnel reported their main water source for domestic use as being boreholes (25%), dams or rivers (44%), mountain springs (16%), rain water (19%) or reticulated water via municipal systems (19%). Unprotected sources for domestic water use was cited as canal water (13%), river or dam water (44%), and irrigation water (6%),). Farm personnel from 8 (50%) out of 16 farms indicated that water was tested on their farms. Four (50%) reported that water was tested for chemicals, 4 (50%) mentioned bacteria, 1 (13%) mentioned pH and 4 (50%) mentioned colour. However, the veracity of these reports could not be confirmed, and it is extremely unlikely that pesticides are tested on 50% of farms in the region. This is borne out by responses from farm personnel when asked who carried out testing, they indicated mainly non-statutory bodies or individuals (6 of the 8 farms). Regularity of testing varied widely from once daily (n=1) to once yearly (n=2).

Although water monitoring was mentioned by respondents, a number of problems emerged. General monitoring appeared to take place relatively inconsistently in farming areas, and no monitoring for pesticides was reported. Although EHO's stated that they had a good knowledge of the health hazards of pesticides, there was a shortage of staff in many areas and there were not enough laboratories in the Western Cape to conduct pesticide water analysis. Lack of feedback to owners on results of analyses also appeared to be an area in need of intervention.


A guideline document for use by rural communities and/or rural local authorities which sets out relevant aspects for conducting pesticide water monitoring is important in the context of this study which aims to develop screening methods for the pesticide monitoring of water. However, it was probably somewhat ambitious to try to produce both new technologies (methods for analysis, and instruments for integrating exposure in the field) as well as to recommend procedures based on these new technologies. For this reason, the project has been limited to recommendations on how a guideline document may be developed. It is based on the results of analyses of screening methods developed in Chapters 2-4, and the costs associated with these methods detailed in Chapter 5. The guidelines also reflect on the findings discussed in Chapter 6, in conjunction with the WRC project No. K5/795/00, of the capacity within rural communities to conduct water monitoring

The proposed guideline document would take into account factors which would optimise its utility in addition to content. Factors optimising the utility include physical structure and comprehensibility which should be undertaken with the assistance of language and media consultants. The content areas of the proposed guideline document should start by setting out a brief background on pesticide water pollution, with reference to WRC project No. 795/1/00 for additional information The guideline document will include sections covering the following:

  1. purpose of the document
  2. identification of target goups and potential applications
  3. knowledge required before sampling for pesticides can take place
  4. identification of areas and sites
  5. identification of pesticides to be monitored.
  6. procedures to follow and equipment required
  7. methods available for analysis
  8. selection of laboratories and analysis of samples
  9. interpretation of results, and
  10. management of water polluted by pesticides.


This project sought to establish new methods of pesticide analysis for monitoring of rural water sources, and to use these data to strengthen capacity for rural communities to monitor their own water sources. These objectives were only partly fulfilled in the course of the project. Clearly what has emerged is the utility of SPME fibres as a method of analysis for organic compounds, specifically two pesticides, chlorpyrifos and endosulfan, in water. The method produces cost-effective analyses of high sensitivity and repeatability.

Extending the use of the SPME fibres to produce a field device which would give an integrated estimate of water pollution has only been partly successful in this project. A theoretical model of the rate of absorption of an analyte (pesticide) into the fibre coating, in a particular physical configuration, has been developed and tested using a prototype field device. Data produced in this project, though, has been too limited to draw any conclusions as to the validity or usefulness of the approach. Further research is recommended to overcome the problem of analyte retention during the extended sampling period, and to assess the importance of temperature control of sampling conditions in the field.

As for ELISA analysis, it is unlikely, based on the preliminary work conducted in this project, that ELISA analysis could become a useful adjunct in pesticide monitoring. Costs are high, much higher than anticipated in the literature, and accuracy weak. The use of ELISA even as a routine screening method could not be supported by the present associated costs.

Rural capacity to monitor pesticides remains limited, although awareness, as demonstrated in this and previous studies, is high. Costs are the chief barrier, and future work to reduce such costs will be the rate-limiting step for any widespread adoption of monitoring. However, the value of developing tools that give rural communities, local authorities, farmers and farm workers the capacity to initiate, coordinate, understand and interpret monitoring for pesticides in water, is highly desirable.


There is a need to develop pesticide monitoring methods with a view to the implementation of future full-scale monitoring. In particular, further development of the validity of the calculation adjusting for the loss factor, in-fibre derivation, and/or chemically fixing the absorbed analyte which was beyond the scope of the present project, should be explored in future research to establish integrated methods of monitoring pesticides.

Development of in-house plate kits which could substantially reduce costs associated with ELISA could be explored.

The preamble guidelines for monitoring pesticides in water developed in this study, should be further developed.

The capacity of State laboratories to measure pesticide water pollution at low levels (0.01 mg/L) using either SPE or SPME methods needs to be enhanced, especially if routine monitoring is implemented.

The problem of lack of feedback to farmers of the water results should be addressed.

It is important for farm residents to be informed about pesticide pollution of water and about the possible adverse health effects of pesticides. This could best be done by the farmer, farming co-op or supplier.