General enquiries on this form should be made to:Defra, Science Directorate, Management Support and Finance Team,Telephone No. 020 7238 1612E-mail: research.competitions@defra.gsi.gov.uk

SID 5 Research Project Final Report

SID 5 (2/05) Page 1 of 28

NoteIn line with the Freedom of Information Act 2000, Defra aims to place the results of its completed research projects in the public domain wherever possible. The SID 5 (Research Project Final Report) is designed to capture the information on the results and outputs of Defra-funded research in a format that is easily publishable through the Defra website. A SID 5 must be completed for all projects.

A SID 5A form must be completed where a project is paid on a monthly basis or against quarterly invoices. No SID 5A is required where payments are made at milestone points. When a SID 5A is required, no SID 5 form will be accepted without the accompanying SID 5A.

This form is in Word format and the boxes may be expanded or reduced, as appropriate.

ACCESS TO INFORMATIONThe information collected on this form will be stored electronically and may be sent to any part of Defra, or to individual researchers or organisations outside Defra for the purposes of reviewing the project. Defra may also disclose the information to any outside organisation acting as an agent authorised by Defra to process final research reports on its behalf. Defra intends to publish this form on its website, unless there are strong reasons not to, which fully comply with exemptions under the Environmental Information Regulations or the Freedom of Information Act 2000.Defra may be required to release information, including personal data and commercial information, on request under the Environmental Information Regulations or the Freedom of Information Act 2000. However, Defra will not permit any unwarranted breach of confidentiality or act in contravention of its obligations under the Data Protection Act 1998. Defra or its appointed agents may use the name, address or other details on your form to contact you in connection with occasional customer research aimed at improving the processes through which Defra works with its contractors.

Project identification

1. Defra Project code PS2203

2. Project title

Describing the behaviour of ionic compounds for pesticide registration

3. Contractororganisation(s)

Central Science LaboratorySand HuttonYorkYO41 1LZ          

54. Total Defra project costs £ 151,171

5. Project: start date................ 01 January 2004

end date................. 31 December 2007

SID 5 (2/05) Page 2 of 28

6. It is Defra’s intention to publish this form. Please confirm your agreement to do so...................................................................................YES NO (a) When preparing SID 5s contractors should bear in mind that Defra intends that they be made public. They

should be written in a clear and concise manner and represent a full account of the research project which someone not closely associated with the project can follow.Defra recognises that in a small minority of cases there may be information, such as intellectual property or commercially confidential data, used in or generated by the research project, which should not be disclosed. In these cases, such information should be detailed in a separate annex (not to be published) so that the SID 5 can be placed in the public domain. Where it is impossible to complete the Final Report without including references to any sensitive or confidential data, the information should be included and section (b) completed. NB: only in exceptional circumstances will Defra expect contractors to give a "No" answer.In all cases, reasons for withholding information must be fully in line with exemptions under the Environmental Information Regulations or the Freedom of Information Act 2000.

(b) If you have answered NO, please explain why the Final report should not be released into public domain

Executive Summary7. The executive summary must not exceed 2 sides in total of A4 and should be understandable to the

intelligent non-scientist. It should cover the main objectives, methods and findings of the research, together with any other significant events and options for new work.

It is estimated that ca. 25% of the existing pesticides reviewed by the European Union as part of the re-registration process are either weak acids or bases (EU, 2002). These include compounds that have been found in groundwater and surface waters such as 2,4-D, bentazone, glyphosate, MCPA and mecoprop. Also, a significant and increasing proportion of new actives proposed for registration are ionisable (including most sulfonylureas) and the formation of acidic metabolites is common in degradation of pesticides. Strength of sorption and, in some cases, rate of degradation of ionisable pesticides may be strongly influenced by soil properties, particularly pH. Environmental modelling is used to estimate concentrations of pesticides leaching to groundwater and reaching surface water via surface runoff and drainflow. Sorption and degradation parameters are amongst the most sensitive inputs for the models, but approaches to selecting appropriate parameters for ionisable pesticides are relatively simplistic and/or rely on limited datasets.

The aim of this research was to generate a significant database on the influence of soil properties on sorption and degradation of ionisable pesticides and thus to refine existing procedures for assessing the environmental fate of this important group of compounds. The specific objectives were: (1) to investigate the factors influencing the sorption and degradation of ionisable pesticides in soil; (2) to investigate the link between sorption and degradation for ionisable pesticides; and (3) to refine regulatory procedures for assessing the risk of ionisable pesticides impacting on surface waters and groundwater.

An initial literature review collated all available information on the sorption and degradation behaviour of ionisable pesticides in soil. Next a large dataset for sorption and degradation of ionisable pesticides in multiple soils was generated and subjected to statistical analysis to identify factors influencing the two processes and any correlation between them. Nine arable soils were collected to give a gradient in pH (pH in 1M KCl from 4.4 to 8.0) and to have a range in texture (clay content from 6 to 42% and organic carbon content (0.8 to 3.2%). Six acidic pesticides and four basic pesticides were selected in consultation with PSD to give a range of properties. Batch sorption was determined for each soil-pesticide combination using OECD methodology. Rate of degradation was determined at 20oC and –33 kPa soil water content, again using OECD methodology. Statistical analysis with Genstat and MobyDigs was used to identify parameters influencing sorption and degradation. As well as soil properties, the analyses considered basic pesticide properties (solubility, lipophilicity etc) and a range of molecular descriptors generated with the Dragon Web molecular modelling software.

Differences in sorption of all six acidic compounds in nine contrasting soils could be described using a combination of soil organic carbon content and Log D (the logarithm of the octanol-water partition coefficient corrected for the pH of a particular soil). Current regulatory procedures do not facilitate the

SID 5 (2/05) Page 3 of 28

identification of relationships between sorption and more than one soil property. Derivation of a relationship including both pH and organic carbon content requires measurement of sorption in 8-10 contrasting soils. This is a larger number than is generally measured for regulatory datasets. Introduction of a pesticide parameter related to the atomic van der Waal’s volume allowed both relative differences in sorption of the acidic compounds in different soils and the magnitude of that sorption to be predicted. The resulting equation was successfully validated on an independent dataset of seven pesticides and 36 soils provided by BRGM, but predictions were relatively poor for sorption data obtained from disparate literature sources.

The sorption of four basic pesticides was more complex than that of acids, partly because of the larger number of sorption mechanisms that may be implicated in the process. Soil pH, organic carbon content, clay content and cation exchange capacity have all been proposed as important influences on sorption of basic compounds. Results suggest that relationships between sorption and soil properties should continue to be determined on the basis of individual chemicals. An important element of the work was to determine the extent to which sorption and degradation processes are correlated. Significant correlations between sorption and degradation were only observed for three of the ten pesticides. In each case, the correlation was negative (i.e. faster degradation in soils with stronger sorption), thus giving the opposite relationship to that which had been expected. No generalisation on the extent of correlation was possible despite the extensive dataset generated.

Further experiments were undertaken to investigate specific aspects of sorption behaviour. Outcomes can be summarised as: (1) There was no evidence of competition for sorption sites either between pairs of basic pesticides with relatively strong sorption or between acidic or basic pesticides and phosphate in soil. (2) The ionic composition of the soil solution (or background electrolyte used in batch experiments) was shown to have a significant impact on the sorption of acidic pesticides. Measured sorption coefficients generally increased with ionic strength of the background electrolyte. As standard tests are carried out with 0.01M CaCl2 but the ionic strength of soil solution rarely exceeds 0.001M, this may influence the applicability of laboratory measurements of sorption to the field situation. Differences in the effect were used to infer a cation-bridging mechanism as being responsible for part of the sorption of 2,4-D. (3) A centrifugation technique allowed measurement of sorption at realistic soil moisture contents. Strength of sorption was generally smaller than the batch value when measured by centrifugation after one day of incubation, but very similar to the batch value when measured after seven days of incubation. The discrepancy between one-day measurement via centrifugation and the batch value was larger for soil-pesticide combinations showing stronger sorption. Time-dependent sorption was also assessed. The increase in sorption between one and seven days was not directly related to the level of sorption although it was more important in soils containing more organic carbon.

Finally, the study considered implications of the results for the way that environmental fate of ionisable pesticides is assessed within regulatory procedures. Current approaches to select input parameters for environmental modelling follow FOCUS guidelines. These are pragmatic, but necessarily have some constraints because of the small number of experimental measurements (minimum is sorption measured in three soil types). For borderline cases involved ionisable compounds (e.g. maximum predicted environmental concentration in groundwater >0.01 μg/L), it is recommended that sorption should be measured in sufficient soils to determine any relationship between sorption and soil properties. The extent to which standard exposure scenarios used in risk assessment are protective for transport of ionisable compounds to water in the UK was also assessed. The FOCUS groundwater scenarios are considered protective for compounds with Koc >100 mL/g, but appear less protective for more mobile compounds because the more intrinsically vulnerable scenarios for such compounds have pH ≤7.0. The highest pH of any of the drainage scenarios likely to be relevant to UK conditions is 7.2, whereas ca. 35% of drained cereal land in England and Wales has pH >7.2. The FOCUS runoff scenarios are heavily weighted towards high pH soils and appear protective. It is recommended that the degree of protection associated with exposure estimates from the FOCUS scenario tools should be considered further.

The primary requirement for future research is for fundamental investigations into the mechanisms of pesticide sorption to soil. Whilst such studies are challenging and time-consuming, a range of approaches are now available that could improve our understanding of the sorption process. The lack of universal relationships at an empirical level, particularly for basic compounds, suggests that further progress in predicting extent of sorption in different soils is unlikely without a better quantification of the mechanisms involved.

This research was undertaken in collaboration with a French-funded study at BRGM, France. The French study is behind schedule, but implications of the research for regulatory assessment procedures will be revisited once results from that study become available.

SID 5 (2/05) Page 4 of 28

Project Report to Defra8. As a guide this report should be no longer than 20 sides of A4. This report is to provide Defra with

details of the outputs of the research project for internal purposes; to meet the terms of the contract; and to allow Defra to publish details of the outputs to meet Environmental Information Regulation or Freedom of Information obligations. This short report to Defra does not preclude contractors from also seeking to publish a full, formal scientific report/paper in an appropriate scientific or other journal/publication. Indeed, Defra actively encourages such publications as part of the contract terms. The report to Defra should include: the scientific objectives as set out in the contract; the extent to which the objectives set out in the contract have been met; details of methods used and the results obtained, including statistical analysis (if appropriate); a discussion of the results and their reliability; the main implications of the findings; possible future work; and any action resulting from the research (e.g. IP, Knowledge Transfer).

1 INTRODUCTION

Current procedures for assessing the risk of a crop protection product impacting on the environment partly rely on the use of mathematical models to predict likely concentrations in different environmental compartments. Previous research funded by Defra (PL0532) has demonstrated that predictions for concentrations of pesticides in groundwater are particularly sensitive to parameters related to sorption and degradation (Dubus et al., 2000). The same is true for predictions of transport to surface water via drainflow.

In the registration context, sorption is traditionally characterised using the Koc concept, which is based on the assumption that sorption of the pesticide is proportional to the organic carbon content of a soil or a sediment. Although this concept was initially derived for neutral (non-ionic) compounds, it is often applied to both ionisable and non-ionisable compounds in current registration procedures. There is widespread evidence showing that soil organic matter may not be the main determinant of sorption for ionisable pesticides and that pH is one of the main factors controlling sorption. The characterisation of sorption of ionisable compounds using the Koc concept is therefore often inappropriate. This most frequently arises because the dataset from which to derive relationships between sorption and soil properties is often extremely limited (a minimum of measurements in three soil types is required) and because generic understanding of likely relationships for different chemical types is poorly developed. Proper estimation of sorption properties for a range of soils (for which measurements will not always be available) is problematic and this may weaken subsequent exposure and risk assessments.

It is estimated that ca. 25% of the existing active substances which are currently being reviewed by the European Union as part of the re-registration process are either weak acids or bases (EU, 2002). These include actives that have been found in groundwater and surface waters such as 2,4-D, bentazone, glyphosate, MCPA and mecoprop. Also, a significant and increasing proportion of new actives proposed for registration are ionisable (including most sulfonylureas) and the formation of acidic metabolites is common in degradation of pesticides.

1.1 Aim and objectivesThe aim of this research was to generate a significant database on the influence of soil properties on sorption and degradation of ionic pesticides and thus to refine existing procedures for assessing the environmental fate of this important group of compounds.

The specific objectives were:

1. To investigate the factors influencing the sorption and degradation of ionic pesticides in soil;

2. To investigate the link between sorption and degradation for ionic pesticides;

3. To refine regulatory procedures for assessing the risk of ionic pesticides impacting on surface waters and groundwater.

1.2 Collaborative linksThis project was established with an agreement that this Defra-funded study would be undertaken in parallel with a French project starting concurrently that was funded by SSM and undertaken by BRGM, Orléans, France. Specific tasks would be undertaken within each project with results combined at the end of the projects to determine the overall implications for regulatory assessment procedures. Subsequent to the start of the UK

SID 5 (2/05) Page 5 of 28

project, there was a delay of 12 months before the French project began. Much of the work to be undertaken in France remains incomplete at the end of January 2007. The results obtained so far at BRGM (i.e. sorption data for seven ionisable pesticides in 36 temperate soils) are integrated in the present report. However, the delay in the work programme for the French study means that it is not considered possible to include the results on leaching behaviour and detailed investigation on sorption behaviour in a combined final report as initially planned. With agreement from PSD, this report covers primarily aspects undertaken within the UK project. Implications for regulatory procedures have been determined based solely on data from work carried out for project PS2203. However, this element of the work will be revisited based on the combined datasets and outputs from France and the UK once work in France is completed.

2 LITERATURE REVIEWA great deal of work has been undertaken concerning the sorption of ionisable pesticides in soils in the past 15 years. An extensive review of the literature was undertaken to introduce the main issues concerning the behaviour of ionisable pesticides in soils. The review was published in Reviews of Environmental Contamination and Toxicology in 2006 (Kah and Brown, 2006). The main conclusions can be summarised as follows:

1. Many retention mechanisms in addition to hydrophobic partitioning have been postulated as contributing to the sorption of ionisable pesticides in soils. These include ionic exchange, charge transfer, ligand exchange and cation (or water) bridging. However, relatively little experimental evidence is available.

2. The sorption of ionisable compounds in soils is strongly influenced by pH and this effect depends on soil composition and the characteristics of the compound. This pH dependence derives mainly from the different proportions of ionic and neutral forms of the pesticide present at each pH level and from differences in their strength of sorption. The influence of varying pH on the charge at the surface of soil particles may also play a role in some cases. A decrease in sorption with increasing pH is often observed for acids as well as for bases (Figure 1, curve A). However, bell shaped curves (curve B), increases in sorption (Curve C) and pH-independent behaviours have also been reported. The decrease in sorption at more acidic pH (Curve B) is generally attributed to competition between the cationic form of a base and other cations present in solution. Curve C may occur for some weak bases that are mainly adsorbed as neutral molecules.

3. Soil organic matter generally promotes the sorption of ionisable pesticides in soils although a negative influence has occasionally been observed. Clay and Al, Fe (oxi)hydroxides can also play a significant role and might have to be considered in some situations (i.e. tropical and sub tropical soils).

4. So far, no modelling approach has been applied successfully to a range of ionisable pesticides to predict their sorption in soils. Further experimental data are required to select the descriptors and parameters that should be included.

5. Degradation of ionisable pesticides is influenced by soil pH in a particular way that relates to changes in sorption, changes in composition and activity of the microbial community, and to shifts in the balance between different degradation mechanisms.

6. Questions remain concerning the link between the processes of sorption and degradation. Although i t is well documented that sorption processes may affect biodegradation (mainly by modifying chemical bioavailability) factors other than sorption also influence degradation rates. It is not appropriate to assume de facto that there will be a link between sorption and degradation for a particular compound, and the relationship should continue to be assessed case by case through analysis of experimental data.

Figure 1. Three sorption behaviours have often been recorded for ionisable compounds as a function of soil pH. A pH-independent sorption can also be observed in some cases.

pH

Ads

orpt

ion

A

B

C

SID 5 (2/05) Page 6 of 28

3 IDENTIFICATION OF FACTORS INFLUENCING THE SORPTION AND DEGRADATION OF IONISABLE PESTICIDES

The first experimental phase of the study involved generating a large dataset for sorption and degradation of ionisable compounds in UK soils. The dataset was processed using multivariate analysis to identify those factors determining the extent of sorption and degradation in soil and to elucidate any relationships between sorption and degradation.

3.1 Materials and methods

3.1.1 Soils

Nine arable soils were selected to give a gradient in pH (pH in 1M KCl from 4.4 to 8.0) and to have a range in texture (clay content from 5.6 to 41.5%) and organic carbon content (0.76-3.24 g kg -1). Soils were sampled from the top 20 cm in several locations in southern England. They were sieved to 3 mm, air-dried and analysed by the Laboratoire d’analyses des sols (INRA Arras, France) (Table 1).

Table 1. Properties of the nine soils selected for the study

Soil texture pH Clay Silt Sand OC CaCO3 CEC-33 kPa water content

water % % % % % cmol+ kg-1g water

/100 g dry soil

1 silty clay loam 8.20 38.5 48.7 12.8 1.77 76.4 6.96 32.22 sandy clay loam 7.81 25.7 24.8 49.5 3.24 36.3 16.6 28.03 sandy clay loam 8.08 27.5 21.0 51.5 1.08 0.49 12.9 17.74 sandy clay loam 7.91 34.5 21.5 44.0 2.00 0.70 18.1 26.95 sandy clay loam 6.85 19.9 26.5 53.6 2.38 0.09 11.6 26.36 sandy 7.07 5.6 4.6 89.8 0.77 0.21 3.41 9.77 loam 6.89 23.6 35.7 40.7 1.68 0.09 10.3 25.58 clay 5.96 41.5 33.0 25.5 3.23 0.09 22.3 35.59 sandy loam 5.28 13.5 33.1 66.4 1.50 0.09 6.62 20.1

3.1.2 Pesticides

Ten ionisable pesticides were selected in consultation with PSD to give a range of properties and to investigate six acids and four bases (Table 2). Pestanal analytical grade standards of 2,4-D, dicamba, fluroxypyr, metsulfuron-methyl, metribuzin, pirimicarb, fenpropimorph and terbutryn were purchased from Sigma-Aldrich (Seelze, Germany), fluazifop-P was supplied by Syngenta (Bracknell, UK) and flupyrsulfuron-methyl was supplied by E.I. DuPont de Nemours (Wilmington, DE, USA). Radiolabelled 2,4-D was purchased from American Radiolabelled Chemicals Inc. (St. Louis, MO, USA), dicamba from Izotop Institute of Isotopes Co., Ltd. (Budapest, Hungary) and metsulfuron-methyl and flupyrsulfuron-methyl were supplied by E.I. DuPont de Nemours (Wilmington, DE, USA). All organic solvents were HPLC grade (Fisher Scientific, UK). Fluroxypyr is applied in the field as fluroxypyr-meptyl, which is rapidly hydrolysed to the parent acid (half-life in soil-water slurries= 2-5 h at pH 6-7 and 22-24º C; Tomlin, 1997), so fluroxypyr was directly applied to the soils in this study. Similarly, the first metabolite of fluazifop-P-butyl (fluazifop-P) was used because the DT50 of the former is less than 24 h in soils (Tomlin, 1997).

Based on pre-experiments (data not shown) it was assumed that no competition effects operate at low concentration. Pesticides were paired (fluroxypyr with fluazifop-P, metribuzin with pirimicarb and fenpropimorph with terbutryn) and studied together for all non-radiolabelled sorption and degradation experiments, effectively doubling the number of samples that could be processed. Samples were analysed by either high-pressure liquid chromatography (HPLC) or gas chromatography with mass spectrum detection (GC-MS). Full details of methods are published in Kah and Brown (2007a) and Kah et al. (2007).

A molecular modelling package (Dragon Web Version 3.0, Talete srl) was used to predict a large number of molecular properties for the ten pesticides; these included properties that might influence sorption and/or degradation behaviour such as potential for formation of hydrogen bonds, features of molecular orbitals, size:charge ratios etc. Estimates of degradation potential were included in the list of pesticide descriptors as well.

SID 5 (2/05) Page 7 of 28

Six estimates of aerobic biodegradability were determined using BIOWINTM (2000). Three DT50 values from the literature were included as well for each pesticide (Footprint database, 2006).

The neutral and ionic forms of ionisable compounds have different polarities. Since their ratio varies with pH, the lipophilicity of ionisable pesticides is pH dependent. Lipophilicity profiles (Log D) represent the shift in octanol/water partition coefficient with pH as a consequence of dissociation. For acidic compounds, a sigmoidal decrease of Log D is obtained with pH. The curve closely resembles the speciation curves, but is offset towards higher pH values. Log D can be calculated at a certain pH, provided that the lipophilicity of the neutral form (Log P) and the dissociation constant (pKa) are known:

The decrease in Log D with pH was measured for each acid using the pH metric method (GLpKa, Sirius Analytical Instruments Ltd.).

Table 2. Properties of the ten pesticides selected for study. Values are from the literature except for Log P and pKa which were measured for the six acids.

apK a:

dissociation constantbKoc: sorption coefficient in soils normalised by the organic carbon contentcDT50:half-life in soil, time required for 50% of the initial dose to be degradeddLog P: octanol-water partition coefficient of the neutral form; indicates the lipophilicity of the compound. Log P was measured for the six acids whereas data for bases are from the literature.

3.1.3 Batch technique to measure sorption

Results of the sorption experiment have been published and a more detailed description of the procedure can be found in Kah and Brown (2007a). Sorption coefficients (Kd, mL g -1) were determined with four replicates using a standard batch equilibrium method following the OECD guidelines (1997). All experiments were carried out at 2 mg kg-1 with the exception of the two sulfonylurea herbicides, which were studied at 1 mg kg-1. After a pre-equilibration period of 14 h, soil suspensions in 0.01M CaCl2 were spiked with a pesticide solution, and returned to shaking for 72 h. The samples were then centrifuged at 5000g for 10 min, and the supernatant was analysed to measure the concentration of pesticide remaining in solution after sorption. Mixtures of unlabelled and radiolabelled compounds were used for 2,4-D, dicamba, metsulfuron-methyl and flupyrsulfuron-methyl, and supernatants were analysed by liquid scintillation counting. Concentrations of fluroxypyr and fluazifop-P were analysed by HPLC and concentrations for basic compounds were analysed by GC-MS. Samples were maintained in the dark at 4oC throughout the procedure to minimise degradation.

3.1.4 Incubation to measure degradation

Results of the incubation experiments have been published and a more detailed description of the procedure can be found in Kah et al. (2007). Degradation of the pesticides was determined at native soil pH under standard incubation conditions following the OECD guidelines (2002). Samples of fresh soil were pre-incubated for eight days prior to application of pesticide (moisture content just below –33 kPa, 15°C, in the dark). Technical grade pesticide solution was applied dropwise to the equivalent of 200 g of dry soil (three replicates) to reach an initial concentration of 2 mg a.s. kg-1 (1 mg a.s. kg-1 for the two sulfonylureas). Soil was thoroughly mixed and the moisture content was adjusted by weight to exactly –33 kPa. The soil was then transferred to a 500 mL glass flask and incubated at 15°C in the dark. During the incubation, moisture content was maintained by weight twice a week and lids were only loosely closed to avoid anaerobic conditions being created. At appropriate time intervals, samples of 20 g of soil were weighed and immediately frozen. Nine samples were taken during the incubation period. The duration of incubation was chosen according to half-lives previously reported in the literature. Soil bioactivity was evaluated by measuring dehydrogenase activity after 2 and 6 weeks of incubation (triplicates).

SID 5 (2/05) Page 8 of 28

Compound pKaa Koc (ml/g)b DT50 (d)c Log Pd

2,4-D carboxylic acid 2.64 5 - 212 8 - 40 2.72Dicamba carboxylic acid 1.97 3.5 - 21 2 - 6 2.30Fluroxypyr carboxylic acid 2.94 51 - 81 6 - 21 2.20Fluazifop-P carboxylic acid 2.98 39 - 84 6 - 16 1.24Metsulfuron-methyl sulfonylurea 3.80 4 - 60 14 - 51 1.23Flupyrsulfuron-methyl sulfonylurea 4.90 15 - 23 6 - 25 4.31Metribuzin triazine 1.00 9 - 95 6 - 377 1.60 (pH 5.6)

Fenpromimorph morpholine 6.98 2772 - 5943 15 - 127 4.40 (pH 9)

Terbutryn triazine 4.30 2000 14 - 50 3.65Pirimicarb carbamate 4.44 290 7 - 234 1.7

This enzyme is only active in living organisms and thus is an indicator for soil microbial activity. Extraction and quantification of pesticide remaining in soil was undertaken at the end of the respective incubation period. Degradation was measured through the relative decline of residues extracted with an appropriate organic solvent. Analysis was by HPLC and GC-MS detection. Non-extractable residues were considered to be degraded and the extraction efficiency was assumed to remain constant over the course of the experiment.

Three kinetic models were fitted to the degradation curves and parameters were optimised according to recommendations by FOCUS (2006). The simple first-order kinetic model always described the data adequately. The first-order rate of degradation and the DT50 (time required for 50% of the initial dose of pesticide to be degraded) of each compound in each soil were determined with the following equations:

and

where Ct, is the concentration of pesticide remaining in soil (mg kg -1) after t (days), Co, the initial concentration of pesticide (mg kg-1) and r, the rate of degradation (day-1).

3.1.5 Statistical analysis

Relationships between sorption/degradation parameters and soil/pesticide descriptors were investigated using two software packages. The aim was to identify the best combination of properties to describe the variation in sorption/degradation. The three best properties to include in the regression equations were selected with (i) a forward stepwise search with Genstat (starting with no terms in the model, variables are added or dropped according to the residual mean square; Genstat for Windows, 7th edition, Rothamsted Research, UK) and (ii) MobyDigs (software designed to identify an optimal regression model where a large number of potential parameters are available, using a genetic algorithm approach coupled with ordinary least squares regression; MobyDigs Version 1.0, Talete srl.). Each pesticide and soil was first considered individually. The data for the acids and bases were then integrated and the software were run again. Finally, the whole dataset was considered. The same approach was followed for the descriptors with separate analysis with soil descriptors, pesticide descriptors and finally all descriptors considered together. Variance and correlation analyses were performed using Genstat.

3.2 Results of batch screening measurementsIn general, sorption of acids was weak compared to bases and followed the order: dicamba < metsulfuron-methyl < fluazifop-P < metribuzin < 2,4-D < flupyrsulfuron-methyl < fluroxypyr < terbutryn < pirimicarb < fenpropimorph. Sorption of ionisable pesticides tends to be stronger in soils with lower pH and containing more organic carbon (Figure 2 and Figure 3). These trends are much clearer for acids than for bases. Results confirmed that approaches consisting in the normalisation of sorption coefficients to the organic carbon (Koc) or clay content (Kclay) are not suitable for ionisable compounds. Statistical analyses against a wide range of soil and pesticide descriptors were used to identify the best combination of properties that describes the variation in sorption. Main results are given in Table 3. Regression coefficients are generally large even when only soil properties are considered. The low level of sorption of dicamba makes the determination of sorption difficult and prone to error and this probably explains the low regression coefficients obtained for this compound. Results obtained with the Genstat package were generally similar to those obtained with MobyDigs (results not shown). Better predictions were generally obtained for Log Kd than for Kd probably because Log Kd gives a narrower range of values. Different combinations of properties were generally selected for the different soils and pesticides but after a careful analysis of the results, some trends clearly appeared (Table 3). Results obtained for acidic and basic compounds are discussed separately below.

SID 5 (2/05) Page 9 of 28

Figure 2. Plots of Kd (mL g-1) against OC (g kg-1)(a) and Koc (mL g-1) against soil pH KCl (b) for the six acids.

-0.5

0.5

1 .5

2 .5

3 .5

4 .5

2,4-D dicamba flupyrsulfuron-methyl fluroxypyr metsulfuron-methyl fluazifop-P

0

1

2

3

4

5 10 15 20 25 30 35OC (g kg-1)

Kd (m

L/g)

0

40

80

120

160

4 5 6 7 8

pH KCl

Koc

(mL/

g)

SID 5 (2/05) Page 10 of 28

(a)

(b)

Figure 3. Plots of Kd (mL g-1) against OC (g kg-1) (a) and Koc (mL g-1) against soil pH KCl (b) for the four basic compounds. Outlier for pirimicarb: Kd=105 (mL g-1), Koc= 3262 (mL g-1) in soil 8 (pH = 4.87, OC=32.3) and fenpropimorph: Koc= 4026 (mL g-1) in soil 2 (pH=7.41, OC=10.8)

0

1 0

2 0

4 4 .5 5 5 .5 6 6 .5 7 7 .5 8 8 .5

p H K C l

metribuzin pirimicarb fenpropimorph terbutryn

0

10

20

30

40

4 9 14 19 24 29 34

OC (g kg-1)

Kd (m

L/g)

0

200

400

600

800

1000

1200

4 5 6 7 8pH KCl

Koc

(mL/

g)

3.3 Factors influencing sorption of acidic pesticides When all descriptors were considered together, Log D and OC were selected as the best predictors for the sorption of acids (Table 3). Better predictions were obtained for Log Kd than for Kd and to be consistent with linear free energy relationships, the OC content was transformed to logarithmic values as well. The neutral and ionic forms of ionisable compounds have different polarities. Since their ratio varies with pH, the lipophilicity of ionisable pesticides is pH dependent. For acids, Log D decreases with increasing pH as the proportion of anionic species increases. For bases, Log D increases with pH since the dominant form at pH>pKa is neutral. The parameter Log D describes two sources of variability in sorption. When Log D is selected for a single compound, it describes the shift in concentrations of the neutral and ionic forms and the difference in their strength of sorption. The correlation between Kd and Log D is thus positive for acids and negative for bases (for which the cation sorbs more strongly than the neutral form). When several pesticides are considered together, Log D also allows ranking of pesticides according to their intrinsic tendency for hydrophobic partitioning. In this case, the relationship between Kd and Log D is always positive. Log D can thus be used to predict the sorption of acids that mainly occurs through hydrophobic partitioning, but it is unsuitable for bases.

SID 5 (2/05) Page 11 of 28

(a)

(b)

Table 3. Best predictors for sorption variability (expressed as Kd and Log Kd) selected by the data-mining package.

Soil properties selected for Kd r2 Soil properties selected for Log Kd r2

2,4-D Log D, CaCO3, CEC 0.961 P2O5, CEC, Al 0.953dicamba Log D, OC 0.161 CaCO3, CEC, Al 0.418Fluroxypyr Log D, CEC, K 0.946 pH, CEC, Al 0.969fluazifop-P Log D, OC, Ca 0.952 P2O5, CEC, Al 0.957metsulfuron-methyl P2O5, Mg, Na 0.911 sand, pH, Ca 0.953flupyrsulfuron-methyl P2O5, Mg, Na 0.961 CaCO3, CEC, K 0.939

All acids Log D, Ca 0.119 Log D, OC 0.397

metribuzin pH, CEC, K 0.931 clay, OC, Na 0.922pirimicarb C/N, P2O5, Mg 0.979 Log D, Al, Fe 0.972fenpropimorph pH, C/N, Mg 0.791 C/N, Mg, Na 0.862terbutryn CaCO3, P2O5, Mg 0.912 OC, CaCO3, Si 0.919

All bases Mg, K 0.303 Log D, OC, K 0.412

All pesticides Log D, CEC 0.159 Log D, OC, Al 0.554

All properties for Kd r2 All properties for Log Kd r2

All acids Log D, OC, nCs 0.710 Log D, OC, GATS7v 0.906All bases CaCO3, Mg, BELp1 0.458 OC, Mg, BELv3 0.821All pesticides Mg, ATS1m, ATS8e 0.360 OC, CIC2, JGI3 0.838

ATS1m

ATS8e

BELp1 lowest eigenvalue n. 1 of Burden matrix / weighted by atomic polarizabilitiesBELv3 lowest eigenvalue n. 3 of Burden matrix / weighted by atomic van der Waals volumesCIC2 complementary information content (neighborhood symmetry of 2-order)GATS7v Geary autocorrelation - lag 7 / weighted by atomic van der Waals volumesJGI3 mean topological charge index of order3nCs number of total secondary C(sp3)

Broto-Moreau autocorrelation of a topological structure - lag 1 / weighted by atomic massesBroto-Moreau autocorrelation of a topological structure - lag 8 / weighted by atomic Sanderson electronegativities

As a final output from the statistical analysis, two regression equations were proposed to predict the sorption of acids in soils:

(Equation 1)

(Equation 2)

When tested on our set of data, equation (1) that includes Log D and Log OC, described a large part of the variation in sorption (Figure 4a). The lines represent the trend lines fitting the data for each pesticide. It clearly appears that trend lines are approximately parallel to the 1:1 line. This indicates that variations in sorption due to differences in soil pH and OC content can be well described for different acidic pesticides in various temperate soils by a unique equation. For modelling purposes, the following assumptions can hence be reasonably made: (i) variations due to pH are a direct consequence of pesticide dissociation (variations in surface charges can be neglected in temperate soils), (ii) differences in quality of soil organic matter in the different soils can be neglected, at least for these soils from the UK, (iii) sorption to the clay fraction can be neglected.

The trend lines fitting data for 2,4-D and fluroxypyr are offset compared to the 1:1 line (Figure 4a). The intersection with the Y-axis needs therefore to be corrected for the pesticide considered. Two pesticide parameters are already included in equation (1) through the parameter Log D (i.e. pKa, the dissociation constant and Log P, the lipophilicity of the neutral form). Reliable pKa values can be obtained from various methods (e.g titration, spectrophotometric, conductometric methods). On the other hand, the determination of the lipophilicity of ionisable compounds is still a problem. Although numerous methods are available to measure lipophilicity, only a few are appropriate for ionisable pesticides. The offsets observed for 2,4-D and fluroxypyr could thus indicate that Log P has not been correctly evaluated for these two compounds. However, the lipophilicity of the pesticides studied was determined with a method suitable for ionisable compounds (pH-metric method) and although errors

SID 5 (2/05) Page 12 of 28

are possible, the method is expected to give consistent results for different compounds. Another possible explanation for the offset observed for 2,4-D and fluroxypyr is that an additional characteristic of the compound is required to describe its affinity with soil organic matter. Ionisable pesticides may sorb on soil particles through various mechanisms and experiments measuring the influence of ionic strength on sorption (see Section 4.3) confirmed differences between 2,4-D and flupyrsulfuron-methyl. The inclusion of the pesticide parameter GATS7v (Geary autocorrelation - lag 7 / weighted by atomic van der Waals volumes) significantly improved the description (equation (2), Figure 4b) by setting the intersection of trend lines with the Y-axis to 0. Although it is not fully understood, the GATS7v parameter might describe the propensity of an organic compound to sorb to OM by van der Waals interaction and thus explain that part of sorption that is not explained by the Log D term.

The two equations were tested on an independent dataset of sorption coefficients measured by BRGM (Surdyk et al., 2006). The dataset comprised seven acidic pesticides (three common to both datasets: 2,4-D, metsulfuron-methyl, dicamba, and four independent phenoxy acids: MCPA, 2,4,5-T, dichlorprop, mecoporop-P) and 36 temperate soils sampled in France and the UK (2.13 < OC (g kg-1) < 47.9 and 3.43 < pH KCl < 8.02; Figure 5). Log D for MCPA, 2,4,5-T, dichlorprop and mecoprop-P were calculated based on the lipophilicity of the neutral form (Kow) estimated by KowWin v.1.67 (2000) and using the equation:

with Log Pn, the lipophilicity of the neutral form. Better predictions were again observed with equation (2) that includes the pesticide parameter related to van der Waals volume (equation (1): r2=0.451; equation (2): r2=0.721; Figure 5).

Independent work undertaken by a visiting researcher to CSL showed that the model developed here could not be applied to predict sorption of dicamba, metsulfuron-methyl, 2,4-D and flupyrsulfuron-methyl in soils with small organic carbon content from Spain, where the clay fraction appeared to be a significant influence on sorption (Villaverde et al., 2008). The equations were also tested on sorption data obtained from disparate literature sources. Although variation in sorption was well predicted in some cases, relatively poor predictions were generally obtained. This can be partly explained by differences in the experimental protocols. It is also likely that specific equations are required to account for sorption mechanisms occurring for some pesticides and not for others. It cannot be ruled out that the parameter related to atomic van der Waal’s volume was selected from the statistical analysis by chance.

SID 5 (2/05) Page 13 of 28

Figure 4. Sorption coefficients measured for the acids (Log Kd) are plotted against the values predicted with two regression equations including the lipophilicity of the compound corrected for soil pH (Log D), the soil organic carbon (Log OC) and the pesticide descriptor GATS7v (Geary autocorrelation - lag 7 / weighted by atomic van der Waals volumes). The dashed line is the 1:1 line. Trend lines for each pesticide were added (a).

-0.5

0.5

1 .5

2 .5

3 .5

4 .5

2,4-D dicamba flupyrsulfuron-methyl fluroxypyr metsulfuron-methyl fluazifop-P

(1) Log Kd = 0.13 Log D + 1.02 Log OC – 1.51r2 = 0.392, slope = 0.949, intercept = -0.005

-1.5

-1

-0.5

0

0.5

1

-1.5 -1 -0.5 0 0.5 1

measured

pred

icte

d

(2) Log Kd = 0.06 LogD +1.07 LogOC + 0.99 GATS7v - 2.45r2 = 0.872, slope = 0.914, intercept = -0.006

-1.5

-1

-0.5

0

0.5

1

-1.5 -1 -0.5 0 0.5 1

measured

pred

icte

d

SID 5 (2/05) Page 14 of 28

(a)

(b)

Figure 5. External validation of the regression equation including three parameters: lipophilicity of the compound corrected for soil pH (Log D), soil organic carbon content (Log OC) and pesticide descriptor (GATS7v). The dataset comprises sorption coefficients (Log Kd) measured for seven acids in 36 temperate soils by BRGM (France). The dashed line is the 1:1 line.

-2

-1

0

1

2,4-D dicamba metsulfuron-methyl dichlorprop 2,4,5 T MCPA mecoprop-P

(2) Log Kd = 0.06 Log D + 1.07 Log OC + 0.99 GATS7v – 2.45r2 = 0.721, slope = 0.972, intercept = 0.092

-2

-1

0

1

-2 -1 0 1 2

measured

pred

icte

d

3.4 Factors influencing sorption of basic pesticidesThe sorption behaviour of basic pesticides is more complex than that of acids and the influence of OC content and pH on their sorption was less clear (Figure 3). This is probably due to the variety of mechanisms likely to retain basic compounds on soil particles. When the four bases were considered together, the data mining software selected the parameters OC, exchangeable magnesium and BELv3 (a parameter related to the atomic van der Waals volumes) to predict sorption (Table 3). The inferred influence of magnesium content on sorption probably resulted from the set of soils studied. Magnesium content was particularly large in soil 8 and this soil was an outlier for pirimicarb sorption and gave the largest sorption for fenpropimorph and terbutryn. Since this property is rarely reported in the literature, its relevance is difficult to test on external datasets.

Fenpropimorph had a different behaviour relative to the other bases (no significant influence of pH or OC). When this compound was excluded from the dataset, the best properties selected for the three bases remaining were then pH, CEC and BELm8 (lowest eigenvalue n. 8 of Burden matrix/weighted by atomic masses). The equation that only included the soil properties pH and CEC was tested on data reported in the literature. It gave a good match to sorption data for terbutryn but failed to predict sorption of metribuzin. In contrast to results for the acids, the inclusion of the pesticide parameter (BELm8) did not improve the match to independent datasets for the bases.

Several authors previously observed high correlations between the sorption of basic compounds and soil pH, OC and clay content. However, an equation applicable to a range of basic compounds has not been proposed to date and is not supported by results from the current study. Basic compounds can bind to soil organic and clay fractions through many different mechanisms. The relative importance of one mechanism over another depends on the soil constituents, the molecule and the chemical environment of the soil. Relatively little experimental evidence is available and the balance between these processes is not fully understood. Equations specific to a particular compound are thus preferred at present. More experimental data are needed to achieve a greater level of accuracy and to provide equations for other basic compounds.

3.5 Factors influencing degradation of ionisable pesticidesResults regarding degradation confirm some marked differences between soils in their ability to degrade different pesticides. Results were first submitted to the same statistical analysis as for sorption data. The parameters selected to explain variations in rates of degradation strongly depended on the soil-pesticide combination. The lack of consistent behaviour renders a global approach to prediction of degradation unrealistic.

Despite this, a correlation analysis was able to identify distinct types of behaviour (Table 4). Metsulfuron-methyl, pirimicarb (and perhaps dicamba) seemed mainly degraded by abiotic acidic hydrolysis. The degradation rates of these three pesticides were positively influenced by soil OC content and negatively influenced by soil pH. A

SID 5 (2/05) Page 15 of 28

positive relationship linked their sorption and degradation parameters, possibly as a result of a catalyzed hydrolysis after sorption onto soil organic matter (Henriksen et al., 2004). In contrast, microbial degradation seemed to dominate the breakdown of 2,4-D, fluazifop-P, flupyrsulfuron-methyl, metribuzin, fenpropimorph and terbutryn. As a consequence, degradation rates of these pesticides were very sensitive to soil bioactivity level, positively influenced by soil pH and not related to sorption. Fluroxypyr had an intermediate behaviour and the influence of soil properties on its degradation was unclear. The dominance of one route of degradation over another strongly depends on the characteristic of the pesticide. Pesticides with similar structures may also behave differently as shown for the two sulfonylureas.

Table 4. Correlations coefficients between degradation rates, some soil properties and sorption coefficients. *, ** and *** indicate a significance at p<0.05, 0.01 and 0.001 levels, respectively.

3.6 Link between sorption and degradation processesThere was no statistical relationship between sorption and degradation for most of the pesticides (Table 4). Significant correlations were only observed for three ionisable pesticides out of ten, with faster degradation in soils with stronger sorption. Although a negative relationship between these two processes has been observed in other studies, the results cannot be generalised. The six acidic compounds studied here all had relatively weak sorption and this may contribute to the lack of relationship between sorption and degradation.

4 DETAILED INVESTIGATIONS OF FACTORS INFLUENCING SORPTION OF IONISABLE PESTICIDES

4.1 Competition in sorption for basic compoundsBasic pesticides are generally present as cations in the soil environment (pH>pKa). Competition for sorption onto anionic sorption sites of soil particles is therefore possible when several basic pesticides coexist in the soil solution.

Methodology. Four basic pesticides were either studied alone or paired: metribuzin with pirimicarb and fenpropimorph with terbutryn. Sorption coefficients were determined at one concentration (2 mg kg -1) in soil 2 and soil 9 using the standard batch method described in Section 3.1.3. Sorption coefficients obtained with single and paired pesticides were compared to test for any competition effects.

Results. Sorption coefficients were not statistically different and there were no consistent differences when pesticides were studied alone or paired. This suggests that competition effects do not occur for the combination of basic pesticides studied and are unlikely to occur for similar molecules applied at similar or lower concentrations. Further experiments are required to generalise this result to other soil-pesticide combinations and higher concentrations.

SID 5 (2/05) Page 16 of 28

OC bioactivity pH KCl clay Kd

2,4-D  0.361 0.618*** 0.379 0.280 - 0.014dicamba 0.902*** 0.348 - 0.210 0.534** 0.387*fluroxypyr 0.479* 0.266 - 0.289 0.205 0.287fluazifop-P 0.380 0.758*** 0.657*** 0.606*** 0.102metsulfuron-methyl 0.830*** 0.132 - 0.500** 0.468* 0.824***flupyrsulfuron-methyl 0.552** 0.454* 0.312 0.621*** 0.377

All  acidic compounds 0.232** 0.229** 0.090 0.194* 0.295***

metribuzin 0.455* 0.808*** 0.520** 0.518** 0.084pirimicarb 0.600*** - 0.028 - 0.402* 0.508** 0.668***fenpropimorph 0.241 0.749*** 0.681*** 0.295 - 0.026terbutryn 0.639*** 0.864*** 0.592** 0.639*** - 0.019

All basic compounds 0.458*** 0.537*** 0.303** 0.458*** 0.265*

All  ionisable pesticides 0.259*** 0.275*** 0.128* 0.233*** - 0.011

4.2 Competition in sorption between acidic pesticides and fertiliser (phosphate) Phosphate ions have been shown to compete for sorption with ionisable pesticides such as glyphosate (Gimsing and Borggaard, 2002). A sorption-desorption experiment was carried out to test for competition effects between phosphate ions and two acidic pesticides (2,4-D and flupyrsulfuron-methyl) and one basic pesticide (pirimicarb).

Methodology. Sorption coefficients were determined at one concentration (2 mg a.s. kg -1) using the standard batch method as described in Section 3.1.3. Three consecutive desorption steps (24 h steps) were then performed by replacing 6.5 mL of supernatant by either 6.5 mL of pesticide-free water, or 1 or 10 g L -1 NH4H2PO4

solutions. The pH of each combination of soil and phosphate solution was measured at each desorption step. Desorption profiles obtained with the three solutions were compared to test for any competition effects between pesticides and phosphate ions.

Results. The addition of phosphate did not induce any significant change in the sorption-desorption behaviour of the three pesticides studied. Competition is likely to occur for pesticides that sorb to soil particles through similar mechanisms as phosphate (e.g. ligand exchange). Competition effects are therefore unlikely to occur between phosphate and most ionisable pesticides.

Differences in desorption behaviour could be observed between pesticides. Desorption profiles approximately followed the sorption line for 2,4-D, whereas significant non-singularity (hysteresis) was observed for flupyrsulfuron-methyl and pirimicarb. This difference in desorption behaviour could indicate that flupyrsulfuron-methyl and pirimicarb sorb to soil particles through mechanisms that are less reversible than those of 2,4-D. More experiments are needed to investigate whether significant irreversibility in sorption might occur with other ionisable pesticides, how they depend on pesticide and soil properties and whether they should be taken into account within risk assessment procedures.

4.3 Influence of ionic strength on sorption: differences between 2,4-D and flupyrsulfuron-methylFlupyrsufuron-methyl has a greater pKa and hydrophobicity than 2,4-D (Table 2) and this suggests a stronger sorption than 2,4-D in soils. However, similar sorption coefficients were measured for these two pesticides in eight out of nine arable soils (Figure 2). In one soil only (soil 3, pH=7.41 in 1M KCl), sorption of 2,4-D was much weaker than that of flupyrsulfuron-methyl. It was thus hypothesised that a specific sorption mechanism was operating for 2,4-D but not for flupyrsulfuron-methyl, that this mechanism was cation bridging and that cation bridging was either not operating or was much reduced in soil 3. The aim of this experiment was to characterise the sorption mechanism operating for 2,4-D and not for flupyrsulfuron-methyl and to understand how it was influenced by soil properties.

Methodology. Three soils with similar pH were selected from the screening experiment (soils 2, 3 and 4; Table 1). In soils 2 and 4, sorption of 2,4-D and flupyrsulfuron-methyl was similar when measured in 0.01M CaCl2 whereas sorption of 2,4-D was much weaker than that of flupyrsulfuron-methyl in soil 3. Sorption isotherms for the two acids were determined at four concentrations of CaCl2. The objectives were (i) to investigate any influence on sorption mechanisms from adding CaCl2 as an electrolyte, and (ii) to identify the soil and/or pesticide characteristics that could explain differences in behaviour. Properties of soils 2, 3 and 4 are presented in Table 1. Additionally, the clay fraction of each soil was extracted and characterised by X-ray diffraction. Soils 2, 3 and 4 contain 10.3, 17.9 and 10.4 g of strongly charged clays per 100g of soil (smectite, vermicullite and illite) respectively. Sorption of flupyrsulfuron-methyl and 2,4-D was measured for four pesticide concentrations (0.1, 0.4, 1 and 4 mg kg-1) and with four background electrolytes (milliQ water, 0.01, 0.1 or 1M CaCl2) using the batch equilibrium method described in Section 3.1.3 (three replicates).

Results. The Freundlich equation fitted the data very well (r2>0.99). The curvature of the isotherms (1/n) ranged from 0.81 to 1.10 and indicated that sorption generally decreased as concentration increased. The curvature was not consistently related to the soil type or concentration of CaCl2. On the other hand, a significantly greater curvature (smaller 1/n) was observed for 2,4-D and this could indicate that more specific sorption mechanisms were operative. Strength of sorption for both pesticides generally followed the order soil 2 < soil 4 < soil 3 and this can be partly explained by differences in the soil organic matter content (organic carbon content 3.24, 2.00 and 1.08% for soils 2, 4 and 3, respectively).

Sorption (Kf values) increased significantly with CaCl2 concentration for both pesticides in all soils particularly up to 0.1M CaCl2 (Figure 6). A positive influence of ionic strength on the sorption of acidic pesticides has often been observed previously (Kah and Brown, 2006). It is known to result in part from a decrease in pH (pH decreased by approximately one pH unit between soil suspensions in water and 1M CaCl2).

Measured sorption coefficients (Kf) were compared to coefficients predicted using equation (1) proposed in Section 3.3. The regression equation takes into account the shift in hydrophobicity of acidic compounds as a function of pH (Log D) and the OC content of the soil. This equation can thus be used to predict the increase in sorption due to the decrease in pH occurring when CaCl2 is added. The comparison of the observed and

SID 5 (2/05) Page 17 of 28

predicted increase in sorption between water and CaCl2 permits therefore to distinguish that part of the change in sorption attributable to factors other than the shift in pH (Figure 7).

Figure 6. Freundlich sorption coefficients (mg mL-1) for 2,4-D (●) and flupyrsulfuron-methyl (○) in three soils as a function of CaCl2 concentration. Error bars show standard deviations

0.0

0.5

1.0

1.5

2.0

2.5

0 0.02 0.04 0.06 0.08 0.1Molarity of CaCl2 solution (mol L-1)

Kf

Figure 7. Part of the increase in sorption, observed between 0.01M and 1M CaCl2, attributable to other factors than the shift in pH

The increase in sorption observed for 2,4-D was much greater than what would be predicted based on the shift in pH and this supports the occurrence of sorption mechanisms directly dependent on Ca2+ concentration (such as cation bridging). Sorption of flupyrsulfuron-methyl was much less affected by the addition of CaCl2 than 2,4-D (Figure 7). The formation of Ca-pesticide complexes depends on the type of anion involved in the formation (Clausen et al., 2001). The negative charge on the sulfonylurea bridge of flupyrsulfuron-methyl might be too sterically hindered to allow inner-sphere complex formation, as suggested by Ukrainczyk and Ajwa (1996) for primisulfuron. When measured in 0.01M CaCl2, sorption of flupyrsulfuron-methyl was much stronger than that of 2,4-D in soil 3, whereas the two pesticides have similar sorption levels in the other soils (Figure 2). Since soil 3

SID 5 (2/05) Page 18 of 28

contains almost twice as much strongly charged clay (smectite, vermicullite and illite) as the other soils, a specific interaction between flupyrsulfuron-methyl and the clays in soil 3 might have occurred.

These results give evidence for differences in the mechanisms of sorption between two acidic pesticides. The application of more advanced techniques such as nuclear magnetic resonance, electron spin resonance, or fluorescence spectroscopies would help to characterise the type of bonds formed and to explain differences observed for different soil-pesticide combinations. This experiment confirms that ionic composition of the soil solution (or background electrolyte used in batch experiments) can have a significant impact on the sorption of acidic pesticides. The effect depends on the ionic strength, nature of ions and on the pesticide-soil combination. The ionic strength of natural soil solution does not normally exceed 0.001M, so that effects of ionic strength on sorption can usually be neglected. Nevertheless, the choice to use 0.01M CaCl2 in standardised batch experiments (OECD, 1997) will affect the sorption coefficients of ionisable pesticides. This places a constraint on the use of results from these standardised tests to predict sorption behaviour of ionisable compounds in the field.

4.4 Centrifugation technique and time-dependency of sorption for acidic pesticidesThe results on sorption discussed so far were obtained with the standard batch method. Although this technique is the most commonly used for direct measurement of the sorption of organic molecules in soil, the results may not adequately reflect sorption processes in field-moist or unsaturated soil (Walker and Jurado-Exposito, 1998). Results obtained using the batch technique for the six acidic pesticides were compared to sorption coefficients measured using a centrifugation technique that allows measurement at realistic soil to solution ratios. The technique was also applied to investigate changes in sorption with time. Results of this experiment have been published and a more detailed description of the procedure and results can be found in Kah and Brown (2007b).

Methodology. A 130 g sample of each soil was brought to a moisture content close to -33 kPa and pre-incubated for 24 h at 4ºC in the dark. Pesticide in 0.01M CaCl2 was then applied dropwise and the moisture content adjusted by weighing to exactly that at -33 kPa. The soils were then incubated at 4ºC and in the dark. After 1 and 7 days incubation, the moisture content was readjusted by weighing. Samples were taken to calculate the sorption coefficient. For each soil, 10 g of incubated soil (4 replicates) was weighed into the insert of a 50-mL centrifuge tube (VectaSpin 20, PVDF, Whatman International Ltd., Maidstone, UK). The original filter was removed and replaced by a glass microfibre filter (MF 300, 25 mm diameter, 75 g m -2, 0.45 μm thickness, Fisher Scientific, UK) pre-wetted with 100 µL of deionised water. The tubes and contents were then centrifuged at 1500g for 30 min to collect an aliquot of soil solution that was directly analysed for pesticide concentration by HPLC or LSC. The centrifugation force applied was such that the soil was subjected to a tension of -200 kPa to extract soil solution. This corresponds to the boundary between “mobile” and “immobile” water proposed by Tom Addiscott and others (Walker and Jurado-Exposito, 1998). To determine the total concentration of pesticides present in the soil, soil samples were extracted by adding an appropriate organic solvent after 1 and 7 d (soil to solvent ratio 1:2) and pesticide residues were quantified by HPLC or LSC.

Results. Although the batch method gave significantly higher values of Kd than the centrifugation method after 1 day for the more strongly sorbed molecules in the more sorptive soils, it tended to give lower sorption coefficients compared to the centrifugation method when sorption was weaker (Figure 8, 9). Discrepancies between the two methods were probably attributable to the vigorous shaking applied in the batch technique that artificially enhances the availability of sorption sites. This implies that shortly after application, more pesticide may be present in the soil solution and thus be available for degradation, plant uptake or leaching than will be predicted using a sorption coefficient determined with the batch method.

Sorption of the six acidic compounds increased significantly between one and seven days and the extractability of total residues decreased with time. The relative increase in sorption was not directly related to the level of sorption although the increase in sorption with time tended to be larger in soils containing more organic carbon. These results confirm the importance of time-dependent processes and the necessity to include them in risk assessment procedures. The centrifugation technique is a useful method to measure sorption of pesticides at realistic soil moisture contents and seems to be an adequate technique to characterise the fraction of pesticide that is available for leaching at a given time after application.

SID 5 (2/05) Page 19 of 28

Figure 8. Relative difference between sorption coefficients determined by batch (Kdb) and centrifugation after one day (Kdc1) plotted against strength of batch sorption (Kdb).

Figure 9. Ratio of sorption coefficients obtained by centrifugation and batch techniques (values less than one indicate that batch results are larger). Each column is the mean of ratios measured in nine soils. *, **, *** indicate a significant difference between the two methods at 0.05, 0.01, 0.001 probability level respectively; NS indicates no significant difference.

0

0.5

1

1.5

2

2.5

2,4-D dicamba fluroxypyr fluazifop-P metsulfuron-methyl

flupyrsulfuron-methyl

Cen

trifu

gatio

n K

d / B

atch

Kd

1 day vs batch 7 day vs batch

*** ****** ******

**NS

***

*NS

NS***

5 ANALYSIS OF EXISTING PROCEDURES FOR THE EVALUATION OF FATE OF IONISABLE PESTICIDES

5.1 Approaches to derive sorption parameters for ionisable compounds and consequences for model outputs

Among pesticide fate models used in the first tier risk assessment, PELMO and PEARL can account for the influence of pH on the sorption of ionisable pesticides. However, since the pH throughout the profile of the soils selected for the FOCUS scenarios varies by less than a pH unit, FOCUS (2000) advised to input experimental data obtained at a pH similar to the soil selected in the scenario rather than relying on some theoretical relationship. The same procedure is recommended when using the models MACRO and PELMO, which do not take the influence of pH on sorption into account explicitly. The degree of protection associated with this approach has not been evaluated to date

SID 5 (2/05) Page 20 of 28

5.1.1 Losses by leaching

The PEARL model was used to predict concentrations in leachate for the seven ionisable pesticides for which a good/acceptable prediction of variation in sorption was obtained (2,4-D, fluroxypyr, fluazifop-P, metsulfuron-methyl, flupyrsulfuron-methyl, terbutryn and pirimicarb). The four FOCUS ground water scenarios applied in the UK (Châteaudun, Hamburg, Okehampton and Kremsmünster) were used with winter cereals. Pesticides were applied at realistic rates on the 1 November. DT50s were derived from our experimental dataset and corrected for moisture content when necessary following FOCUS recommendations. Two approaches to derive sorption inputs were compared. Sorption was either (i) measured, using Kd obtained in the soil with the pH closest to that of the FOCUS soil scenario, or (ii) predicted using regression equations derived from our experimental dataset; equations specific to each pesticide had previously been derived to predict sorption as a function of Log (%OC) and Log D (hydrophobicity corrected for pH) (see regression lines on Figure 4a).

No leaching was predicted for five pesticides out of seven and this can be partly explained by the relatively short DT50 of the pesticides studied. For the two pesticides that leached, more leaching was generally predicted when sorption was based on our equation than on the FOCUS guidance (Table 5), although this was not always the case. The largest differences were observed in the Châteaudun scenario that has the highest pH (maximum a factor of seven discrepancy).

Table 5. Concentrations in leachate predicted with PEARL for two ionisable pesticides. No losses by leaching were predicted for the five other pesticides studied.

5.1.2 Losses by drainage

The same procedure was followed to compare the two approaches to derive Kd when predicting concentrations in drainflow with MACRO. Losses by drainage were predicted for 2,4-D and metsulfuron-methyl (which have the smallest sorption coefficients) and the six FOCUS drainage scenarios. The FOCUS methodology was protective for three scenarios (i.e. for D2, D3 and D5, lower or similar concentrations were predicted when using the FOCUS approach). However, for the three scenarios with highest pH, higher concentrations in drainflow were obtained when sorption was based on our equations (Table 6). The maximum discrepancy was a factor of six.

Table 6. Maximum concentration predicted with MACRO for two ionisable pesticides in the three less protective surface water scenarios.

D1 D4 D6

2,4-D Kd predicted 0.08 1.39 4.11FOCUS 0.07 0.25 3.64

Metsulfuron-methyl Kd predicted 2.50 0.10 0.03FOCUS 2.49 0.08 0.03

Maximum concentration in drainflow (ug/L)

Based on these two modelling exercises, the pragmatic approach recommended by FOCUS seems to be reasonably protective for scenarios having low to medium pH but it might be less precautionary for soils with high pH. Considering that minimum regulatory datasets comprise three sorption measurements, the selection of a relevant value for each scenario is likely to be problematic. The scarcity of experimental data is probably the most important limitation of the FOCUS approach to take the influence of pH on sorption into consideration.

5.2 Extent to which the FOCUS scenarios are representative for the pH of agricultural soils in England and Wales

5.2.1 Soils used to grow cereals in England and Wales

Experimental results have confirmed that sorption of ionisable pesticides decreases with increasing pH. Alkaline soils should therefore represent the worst-case scenario for losses of ionisable pesticides by leaching and drainage. The SEISMIC database reporting the soil properties and land use of soils in England and Wales (Hollis et al., 1993, cropping data are from 2000) was used to investigate the extent to which the FOCUS scenarios are

SID 5 (2/05) Page 21 of 28

Chateaudun Hamburg Kremsmunster Okehampton

Fluazifop-P Kd Predicted 0.00003 0.00014 0.00004 0.00018FOCUS 0.00000 0.00002 0.00003 0.00667

Metsulfuron-methyl Kd Predicted 0.04142 0.04875 0.04443 0.04277FOCUS 0.00611 0.01720 0.04675 0.09702

Concentration closest to the 80th percentile (ug/L)

representative of high pH soils in the UK. Figure 10 shows the cumulative distribution of soils used to grow cereals as a function of pH. Around 40% of soils used for growing cereals have a pH above 7 and 25% have a pH above 7.5. The comparison of the distribution of the FOCUS scenarios for leaching and drainage shows that soils with high pH tend to be reasonably represented between pH 6.4 and 7.5 for the groundwater and drainage scenarios. However, the highest pH in the groundwater scenarios is 8.0 and that in the drainage scenarios is 7.5. The FOCUS runoff scenarios have a greater prevalence of high pH soils than is present in cereals soils in England and Wales.

Figure 10. Cumulative distribution of soils used to grow cereals as a function of pH in England and Wales.

0

10

20

30

40

50

60

70

80

90

1004.

4

4.8

5.2

5.6

6.0

6.4

6.8

7.2

7.6

8.0

8.4

pH value

Cum

ulat

ive

dist

ribut

ion

freq

uenc

y (%

) Whole cereal areaDrained cereal areaFOCUS GWSFOCUS SWS drainageFOCUS SWS runoff

5.2.2 Relative vulnerability of the FOCUS scenarios and their pH

The relative vulnerability of the FOCUS scenarios has been evaluated previously based on the loss predicted for sets of dummy compounds (FOCUS, 2000; 2002). Results from these simulations were used to assess the overall vulnerability of the scenarios relative to each other. Scores obtained are reported in Figure 11 together with the pH of each scenario. Figure 11a shows that groundwater scenarios with high pH are not particularly vulnerable regarding their other characteristics. The highest pH present in any of the three most intrinsically vulnerable groundwater scenarios is pH 7.0. The FOCUS scenarios have not been designed to represent the whole range of possible soil pH. It is thus possible that these scenarios do not cover the worse case situation for losses of ionisable compounds by leaching in the UK. Figure 11b presents similar data for scenarios used to predict losses by runoff. Among the four runoff scenarios, three have a pH >7 and one has a pH of 8.4. Worse case situations for ionisable compounds are therefore likely to be covered. Figure 11c shows the same data for the scenarios used to assess the risk of losses in drainflow. The most alkaline soil has a pH of 7.5 whereas Figure 10 shows that more than 25% of the soils where cereals are grown in England and Wales have a pH >7.5. As with the ground water scenarios, there seems to be a gap that does not cover the most vulnerable situations regarding losses of ionisable pesticides by drainflow.

SID 5 (2/05) Page 22 of 28

Figure 11. Vulnerability ranking of the FOCUS scenarios for ground water (a), surface water (b) and drainflow (c). Topsoil pH for each soil profile is indicated at the top of the graph. The larger a column, the greater the vulnerability of a scenario. Y-axis scale is arbitrary and cannot be compared between analyses for groundwater, drainage and runoff scenarios.

0

4

8

12

16

20

24

28

32

36

C H J K N P O S T

Groundwater scenario

Aggr

egat

ed v

ulne

rabi

lity

8.0 5.7 6.2 7.0 5.8 7.0 4.9 7.3 7.0

0

4

8

12

16

20

R1 R2 R3 R4

FOCUS runoff scenario

Aggr

egat

ed v

ulne

rabi

lity

Autumn

Spring

Summer

7.3 4.5 7.9 8.4

0

5

10

15

20

25

30

D1 D2 D3 D4 D5 D6

FOCUS drainage scenario

Aggr

egat

ed v

ulne

rabi

lity

Autumn

Spring

Summer

7.2 5.3 6.5 7.57.0 6.9

5.2.3 Prediction of leaching through two extremely vulnerable soil profilesBased on the previous observations, a final modelling exercise was carried out. Two extremely vulnerable soils were defined for leaching to groundwater under UK conditions. These soils were Wisbech (pH 7.8-8.8, OC 1.8%, 14% clay, 36% sand) and Elveden (pH 8.1-8.8, OC 1.0%, 14% clay, 36% sand). Selection was based on identifying soils with both very high pH and small organic carbon content. Parameters for the PEARL model were selected using mean data for these two soil types from the SEISMIC database (the lowest subsoil layer was extended to 450 cm depth) and combining these soil parameters with cropping and weather parameters for

SID 5 (2/05) Page 23 of 28

(b)

(c)

(a)

Kremsmünster. Losses by leaching were predicted for the seven pesticides previously tested on the FOCUS scenarios. The modelling approach was the same as used in section 5.1.1 with the FOCUS scenarios.

Concentrations in leachate predicted for the two extremely vulnerable UK soil profiles were up to 300 times larger than the largest concentration predicted when using the FOCUS scenarios (Table 7). This comparison gives an upper limit to the additional vulnerability of extreme worst-case soils in the UK for leaching of ionisable pesticides compared to the soils used in the FOCUS scenarios. Wisbech soils are intensively used for arable crops including sugar beet, potatoes, vegetables, winter cereals and oilseed rape. Similarly, Elveden soils are widely cropped with sugar beet and cereals.

Table 7. Comparison between leaching of three compounds predicted with PEARL for the FOCUS scenarios and for two extreme worst-case soils from the UK. Sorption parameters are selected either from regressions of measured Kd against soil properties (Kd predicted) or from the soil with pH closest to that in the scenario (FOCUS).

Wisbech Elveden Max. FOCUS scenario Factor of difference

Fluazifop-P Kd Predicted 0.0003 0.0580 0.0002 324FOCUS 0.0002 0.0368 0.0067 6

Metsulfuron-methyl Kd Predicted 0.0684 0.1576 0.0488 3FOCUS 0.0552 0.1181 0.0970 1

Concentration closest to the 80th percentile (ug/L)

6 CONCLUSIONS6.1 Experimental results and implications1. Differences in sorption of all six acidic compounds in nine contrasting soils could be described using a

combination of soil organic carbon content and Log D (the logarithm of the octanol-water partition coefficient corrected for the pH of a particular soil). Current regulatory procedures do not facilitate the identification of relationships between sorption and more than one soil property. Derivation of a relationship including both pH and organic carbon content would require measurement of sorption on a larger number of soils than is generally measured for regulatory datasets (8-10 contrasting soils seems pragmatic).

2. Introduction of a pesticide parameter related to the atomic van der Waal’s volume allowed both relative differences in sorption of the acidic compounds in different soils and the magnitude of that sorption to be predicted. The resulting equation was successfully validated on an independent dataset of seven pesticides and 36 soils provided by BRGM, but predictions were relatively poor for sorption data obtained from disparate literature sources. The latter failures may partly result from discrepancies in methodology between sources. It cannot be ruled out that the parameter related to atomic van der Waal’s volume was selected from the statistical analysis by chance. A generally applicable model seems out of reach at the present time.

3. Independent work undertaken by a visiting researcher to CSL showed that the model developed here could not be applied to predict sorption in low organic carbon soils from Spain where the clay fraction appeared to be a significant influence on sorption.

4. The sorption of four basic pesticides was more complex than that of acids, partly because of the larger number of sorption mechanisms that may be implicated in the process. Soil pH, organic carbon content, clay content and cation exchange capacity have all been proposed as important influences on sorption of basic compounds. The results presented here suggest that relationships between sorption and soil properties should continue to be determined on the basis of individual chemicals.

5. An important element of the work was to determine the extent to which sorption and degradation processes are correlated. Significant correlations between sorption and degradation were only observed for three of the ten pesticides. In each case, the correlation was negative (i.e. faster degradation in soils with stronger sorption), thus giving the opposite relationship to that which had been expected. No generalisation on the extent of correlation was possible despite the extensive dataset generated (all combinations of ten pesticides and nine soils).

6. There was no evidence of competition for sorption sites either between pairs of basic pesticides with relatively strong sorption or between acidic or basic pesticides and phosphate in soil. The latter phenomenon has been reported for a few pesticides, particularly glyphosate, but is most likely related to a specific mechanism of sorption.

7. The ionic composition of the soil solution (or background electrolyte used in batch experiments) was shown to have a significant impact on the sorption of acidic pesticides. Measured sorption coefficients generally

SID 5 (2/05) Page 24 of 28

increased with ionic strength of the background electrolyte. As standard tests are carried out with 0.01M CaCl2 but the ionic strength of soil solution rarely exceeds 0.001M, this may influence the applicability of laboratory measurements of sorption to the field situation. The effect depends on the ionic strength, nature of ions and on the pesticide-soil combination. Differences in the effect were used to infer a cation-bridging mechanism as being responsible for part of the sorption of 2,4-D.

8. A centrifugation technique allowed measurement of sorption at realistic soil moisture contents. Strength of sorption was generally smaller than the batch value when measured by centrifugation after one day of incubation, but very similar to the batch value when measured after seven days of incubation. The discrepancy between one-day measurement via centrifugation and the batch value was larger for soil-pesticide combinations showing stronger sorption. Time-dependent sorption was also assessed. The increase in sorption between one and seven days was not directly related to the level of sorption although it was more important in soils containing more organic carbon.

6.2 Implications for regulatory proceduresImplications for regulatory procedures have been determined based solely on data from work carried out for this project. Experimental investigations for the sister project in France are continuing and are significantly behind schedule. This section will be revisited based on the combined datasets and outputs once work in France is completed.

The FOCUS groundwater scenarios work group recommended that where there is evidence of a relationship between pH and sorption, the sorption value for a particular model run should be chosen as that from the test soil with pH closest to that of the soil in the groundwater scenario (FOCUS, 2000). This recommendation also applies to parameter selection for the FOCUS surface water scenarios (FOCUS, 2002). This is a pragmatic approach that appears suitable for determining the overall potential for leaching of a compound. Nevertheless, the recommendation raises several questions: (1) does the approach account for the major influences on sorption? and (2) is the nominal 90th percentile worst-case nature of the FOCUS scenarios retained for ionisable compounds?

The statistical analysis of factors influencing sorption carried out within the current project demonstrated that both pH and organic carbon generally influenced sorption of acidic compounds. The factors influencing sorption of basic pesticides were less consistent, but might include pH, organic carbon, clay content and cation exchange capacity. Clearly, where there is more than one influence on sorption, the selection of sorption parameters from the soil with the most similar pH to the modelling scenario might under- or over-estimate sorption in the model scenario depending on discrepancies in the other factor(s). A further constraint is where the pH of the tested soils does not extend across the full range of pH in the model scenarios.

Modelling investigations demonstrated that predicted environmental concentrations (PEC’s) for both groundwater and drainflow were generally larger when sorption parameters were selected on the basis of the statistical relationship between sorption and pH/%OC than when the parameter was taken from the soil with the closest pH. The maximum discrepancy was a factor of seven. Thus it is recommended that for borderline cases involving ionisable compounds (e.g. maximum PECgw >0.01 μg/L), sorption should be measured in sufficient soils to determine any relationship between sorption and soil properties (8 to 10 soils seem a pragmatic definition for ‘sufficient soils’). This relationship should then be used to select input parameters for exposure modelling.

Where a pH relationship with sorption exists, sorption of ionisable pesticides is likely to be weakest in soils with high pH. It is thus important to understand the extent to which the FOCUS scenarios encompass high pH as a possible factor determining vulnerability to leaching, or transport in drainflow or surface runoff. An analysis of cereal soils in England and Wales showed that 25% have topsoil pH >7.5. The maximum pH in the FOCUS groundwater scenarios is 8.0 for Châteaudun. This scenario is simulated using the MACRO model to assess leaching via preferential flow for all compounds with Koc >100 mL/g. Given the worst-case nature of the preferential flow assessment, the Châteaudun scenario simulated with MACRO should give conservative leaching estimates for ionisable compounds with Koc >100 mL/g. Châteaudun is not a particularly vulnerable scenario when simulated with PEARL, PRZM or PELMO (FOCUS, 2000). As the pH of the remaining groundwater scenarios simulated for UK submissions are 5.7 (Hamburg), 7.0 (Kremsmünster) and 4.9 (Okehampton), it is not certain that these scenarios would be protective for leaching of relatively mobile ionisable compounds (Koc <100 mL/g) proposed for use on alkaline soils in the UK. Two extreme UK soil scenarios were simulated, resulting in leaching concentrations that were up to a factor of 300 larger than those calculated using the FOCUS groundwater scenarios.

Work for Defra project PS2220 indicated that scenarios D2, D3 and D4 from the FOCUS surface water scenarios are directly relevant to drained land in the UK. Further, it was proposed that scenarios D1 and D5 should also be considered when generating exposure estimates for UK submissions (note that the proposals in the report have not been taken up by PSD to date). The highest pH of any of these scenarios is pH 7.2 for the D1 scenario (D6 with pH 7.5 is not considered relevant for the UK). Approximately 35% of drained cereal land in England and Wales has pH >7.2, so the required degree of protection associated with exposure estimates from the FOCUS drainage scenarios cannot be guaranteed for ionisable pesticides.

SID 5 (2/05) Page 25 of 28

FOCUS runoff scenario R1 (pH 7.3) is directly relevant for conditions in the UK, whilst Defra project PS2220 proposes that scenario R3 (pH 7.9) should also be considered when generating exposure estimates for UK submissions (note that the proposals in the report have not been taken up by PSD to date). High pH soils thus appear to be reasonably covered by those FOCUS runoff scenarios relevant to the UK.

6.3 Requirements for future research1. Implications of this research for regulatory assessment procedures will be revisited once results from the

sister project at BRGM, France become available.

2. There is a need to consider the degree of protection associated with exposure estimates from the FOCUS scenario tools. This work should account for any recommendation on UK surface water scenarios should these be introduced at a future date. Results should be used to determine whether revised procedures are required for ionisable pesticides with strong relationships between sorption and pH and with proposed use on high pH soils.

3. Fundamental research is required into the mechanisms of pesticide sorption to soil, especially in relation to behaviour of basic pesticides. Whilst such studies are challenging and time-consuming, a range of approaches are now available that could improve our understanding of the sorption process. The lack of universal relationships at an empirical level, particularly for basic compounds, suggests that further progress in predicting extent of sorption in different soils is unlikely without a better quantification of the mechanisms involved.

SID 5 (2/05) Page 26 of 28

7 PUBLICATIONS ARISING OUT OF THE PROJECT

7.1 Refereed journal publicationsKah, M. and Brown, C.D. (2006). sorption of ionisable pesticides in soils. Reviews of Environmental

Contamination and Toxicology, 188:149-218.

Kah, M. and Brown, C.D. (2007). Prediction of the sorption of ionisable in soils. Journal of Food and Agricultural Chemistry, 55:2312-2322.

Kah, M., Beulke, S. and Brown, C.D. (2007). Factors influencing degradation of pesticides in soils. Journal of Food and Agricultural Chemistry 55:4487-4492.

Kah, M. and Brown, C.D. (2007). Changes in pesticides sorption with time at high soil to solution ratios. Chemosphere, 68:1335-1343.

7.2 Conference proceedingsKah, M. and Brown, C.D. (2006). Factors controlling the behaviour of ionisable pesticide in soils. Platform

presentation at the SCI international conference “Pesticide behaviour in soils, water and air” Warwick, UK, March 2006.

Kah, M. and Brown, C.D. (2006). Factors controlling the behaviour of ionisable pesticide in soils. Platform presentation at the SCI student meeting organised by the Pest Management Group, Syngenta, Bracknell, UK, April 2006.

Kah, M. and Brown, C.D. (2007). Behaviour of acidic pesticides in soils. Poster presentation at the international conference “Diffuse inputs into the ground water: Monitoring-Modelling-Management” Graz, Austria, January 2007.

Kah, M. and Brown, C.D. (2007). Behaviour of ionisable pesticides in soils. Proceedings of the XIII Symposium on Pesticide Chemistry: Environmental fate and ecological effects of pesticides, September 3-6, 2007, Piacenza, Italy, p. 80-89 (platform presentation).

References to published material9. This section should be used to record links (hypertext links where possible) or references to other

published material generated by, or relating to this project.

BioWinTM, component of the EPI SuiteTM package (Estimation Program Interface), U.S. Environmental Protection Agency, available at http://www.epa.gov/oppt/exposure/pubs/episuite.htm

Clausen L., Fabricius I., Madsen L. (2001). sorption of pesticides onto quartz, calcite, kaolinite, and alpha-alumina. Journal of Environmental Quality 30:846-857.

Dubus I.G., Brown C.D. and Beulke S. (2000). Sensitivity analyses for leaching models used for pesticide registration in Europe. SSLRC report for DEFRA PL0532, Silsoe, Beds. UK, 85p.

EU (European Union) (2002) Existing active substances decisions and review reports. http://europa.eu.int/comm/food/fs/ph_ps/pro/eva/existing/index_en.htm.

FOCUS (2000). FOCUS groundwater scenarios in the EU review of active substances. Report of the FOCUS Groundwater Scenarios Workgroup, EC Doc. Reference SANCO/321/2000 rev.2, 202pp.

FOCUS (2002). FOCUS Surface Water Scenarios in the EU Evaluation Process under 91/414/EEC. Report of the FOCUS Working Group on Surface Water Scenarios, EC Document Reference SANCO/4802/2001-rev.2. Brussels. Belgium.

FOCUS (2006) Guidance Document on Estimating Persistence and Degradation Kinetics from Environmental Fate Studies on Pesticides in EU Registration; Report of the FOCUS Work Group on Degradation Kinetics, EC Document Reference Sanco/ 10058/2005, version 2.0; FOCUS: Brussels, Belgium, 2006

FOOTPRINT (2006). The FOOTPRINT Pesticide properties dataBase. Database collated by the University of Hertfordshire as part of the EU-funded FOOTPRINT project (FP6-SSP-022704).www.eu-footprint.org/ppdb.html.

SID 5 (2/05) Page 27 of 28

Gimsing A. L. and Borggaard O. K. (2002). Competitive sorption and desorption of glyphosate and phosphate on clay silicates and oxides. Clay Mineralogy 37:509-515.

Henriksen T., Svensmark B., Juhler R. K. (2004). Degradation and sorption of metribuzin and primary metabolites in a sandy soil. Journal of Environmental Quality 33:619-628.

Hollis, J.M., Hallett, S.H. and Keay, C.A. (1993). The Development and Application of an Integrated Database for Modeling the Environmental Fate of Herbicides. Proceedings of the BCPC Conference - Weeds 1993, 3:1355 1364.

Kah, M. and Brown, C.D. (2006). Sorption of ionisable pesticides in soils. Reviews of Environmental Contamination and Toxicology, 188:149-218.

Kah, M. and Brown, C.D. (2007a). Prediction of the sorption of ionisable in soils. Journal of Food and Agricultural Chemistry, 55:2312-2322.

Kah, M., Beulke, S. and Brown, C.D. (2007). Factors influencing degradation of pesticides in soils. Journal of Food and Agricultural Chemistry 55:4487-4492.

Kah, M. and Brown, C.D. (2007b). Changes in pesticides sorption with time at high soil to solution ratios. Chemosphere, 68:1335-1343.

KowWin v.1.67 (2000) Component of the EPI SuiteTM package proposed by the U.S. Environmental Protection Agency, available at http://www.epa.gov/oppt/exposure/pubs/episuite.htm

OECD (1997) Guidelines for the Testing of Chemicals Test No. 106: sorption - Desorption Using a Batch Equilibrium Method. Organisation for Economic Co-Operation and Development.

OECD (2002) Guidelines for the Testing of Chemicals Test No. 307: Aerobic and anaerobic transformation in soils. Organisation for Economic Co-operation and Development.

Surdyk N., Dubus I. G., Crouzet C., Gautier A., Flehoc C. (2006) Estimation de la mobilité dans les sols de molécules ioniques à caractere faible: application a l'évaluation des risques environementaux dans le cadre de l'homologation de produits phytosanitaires. Rapport d'avancement du projet BRGM PDR04EAU19, mai 2006, 54 pages.

Todeschini, R.; Consonni, V.; Mauri, A.; Pavan, M. MobyDigs: software for regression and classification models by genetic algorithms. In Nature-inspired methods in chemometrics: genetic algorithms and artificial neural network. Leardi, R. Ed., Elsevier, 2004, pp 141-167.

Tomlin C.D.S. (1997). The Pesticide Manual, 11th edition. BCPC, Farnham, Surrey.

Ukrainczyk L. and Ajwa H. A. (1996). Primisulfuron sorption on minerals and soils. Soil Science Society of America Journal 60:460-467.

Villaverde J., Kah M. and Brown C.D. (2008). Adsorption and degradation of four acidic herbicides in soils from southern Spain. Pest Management Science (in press).

Walker A. and Jurado-Exposito M. (1998). Sorption of isoproturon, diuron and metsulfuron-methyl in two soils at high soil:solution ratios. Weed Research 38:229-238.

SID 5 (2/05) Page 28 of 28