Tell us whether you accept cookies

We use necessary cookies to make our website work. We also use cookies to collect information about how you use so we can improve our services.

Beta This is a new way of showing guidance - your feedback will help us improve it.

PN0928 - UK Case Studies on Quantitative Risk Assessment

Brexit: Transition period

The UK has now left the EU. Your health and safety responsibilities have not changed in the transition period.

Find the latest information on our Brexit pages

PN0928 - Rothamsted Research - 2002

Executive Summary

In 1994, standard procedures to be adopted for regulatory testing for the effects of Plant Protection Products (PPPs) on non-target arthropods within the European Union (EU) were agreed at a workshop entitled “European Standard Characteristics Of beneficial Regulatory Testing” (ESCORT). These involved a 3 tier testing system starting with laboratory ‘worst case’ assays on inert surfaces and proceeding through more realistic extended laboratory/semi-field tests to large scale field trials. These guidelines were modified at a second workshop in 2000, at which it was decided to use only two sensitive indicator species in Tier I and initial Tier II tests and to use multiple dose assays at Tier I to generate a median lethal rate (LR50), from which a Hazard Quotient (HQ) could be calculated as a trigger for proceeding to higher tier tests. This project examined the robustness and repeatability of existing protocols being used by commercial testing laboratories to assess the effects of PPPs on the two indicator species, the aphid parasitoid Aphidius rhopalosiphi and the predatory mite Typhlodromus pyri, designated for generating Tier I and Tier II regulatory data. Factors chosen for examination were partly determined by discussions with chemical companies and independent contract laboratories during visits to their testing laboratories and attendance at European workshops.

An organophosphate (dimethoate), a pyrethroid (l-cyhalothrin) and a nicotinamide (imidacloprid) were used as representative insecticides throughout the project.

Aphidius Tier I assays

In Tier I tests using hymenopteran parasitoids the pesticide is usually applied to glass plates on which adult wasps are exposed to fresh residues. The wasps are confined on the glass plates by a metal arena, through which air is pumped to avoid the build up of toxic vapour. Our consultations revealed that interpretations of this method vary significantly. We therefore examined the following experimental variables that could potentially influence the observed sensitivity of A. rhopalosiphi to a degree where the LR50 – and hence HQ – would be affected: arena ventilation, sex of the parasitoids used, species and size of Aphidius used.

Ventilation prevents the build up of pesticide vapours within the arena and thus ensures that mortality is due to the contact toxicity of the pesticide residues rather than due to inhalation or asphyxiation. The dose response of A. rhopalosiphi was similar whether ventilation was used or not in the case of all three chemicals. However, the rate of airflow used in our study was relatively low compared to that described by some other workers. Thus, although results indicate that ventilation does not have an effect on the dose response of Aphidius, we still recommend the use of ventilation in these assays until more extensive comparative data are available.

Assays were done using Aphidius colemani and A. ervi in addition to A. rhopalosiphi to compare the sensitivity of the two sexes. The dose response patterns for males and females were very similar and the sex of the animals did not significantly affect mortality, so it seems unlikely that variations in sex ratio would significantly influence the estimated LR50.

Although A. rhopalosiphi has been designated as a sensitive indicator species to be used in Tier I tests, other Aphidius species are more widely available commercially and are less expensive. We compared the sensitivity of A. rhopalosiphi, A. colemani and A. ervi to all three insecticides. For all three chemicals, there were significant differences in sensitivity between species. However, A. rhopalosiphi was the most – or equal most - sensitive of the three species for all three PPPs, justifying its choice as a most sensitive indicator species. Parasitoids of the same species but obtained from different suppliers did not show significant differences in sensitivity to any chemical. These three species vary significantly in size but there was no apparent relationship between size and sensitivity. However, estimates for the LR50 for the same species and compound did show considerable variation, as reported in previous work where ¡Ö 20-fold differences in estimates of A. rhopalosiphi LR50 between laboratories have been recorded. This suggests that repeat assays or the simultaneous consideration of data from different laboratories, may be required before declaring a compound harmless with any confidence.

Aphidius Tier II tests

Tier II tests for A. rhopalosiphi usually expose adult wasps to residues on leaf surfaces and mortality is assessed after 48 hours. Sublethal effects can then be assessed by examining fecundity of the surviving females. The recommended method, accepted by registration authorities, uses potted barley plants treated in the laboratory and confined in transparent acetate cylinders. Again, considerable variation in the details of the test protocol exist amongst laboratories and we investigated a number of concerns that arose during discussion workshops: the influence of ventilation and spray volume, the influence of sugar solution applied to the leaves as food for the parasitoids and the effect of applying the test chemical to the inert sand layer applied to the soil surface of the pots.

Application rates of active ingredient/formulated product equivalent to 200 l ha-1 and 400 l ha-1 were compared in tests using dimethoate and l-cyhalothrin against A. rhopalosiphi and A. ervi, with and without ventilation of the arenas. There were no significant effects on parasitoid mortality of either application rate or ventilation. Therefore, there was no evidence that changing the total volume of pesticide or using ventilation in these assays would have a strong effect on the estimation of LR50 values. However, if the higher rate of 400 l ha-1 consistently provides a more even coverage of the test foliage, this should become the standard application rate in these tests.

It has been suggested that an application of sugar solution to the test plants would provide food for the test parasitoids and maximise their contact with residues on the leaf surface. This is controversial, as it provides a ‘synthetic’ attractant to the wasps and provides a second route of exposure via the ingestion of toxins, although it could be viewed as mimicking the natural role of honeydew as a searching stimulant and food source. Our results indicate that using different types of sugar solution is unlikely to influence the average dose of chemical the wasps receive, and sugar solution did not increase mortality compared to the water control, suggesting that any effects of ingesting sugar solution sprayed with pesticide were negligible.

In Tier II Aphidius assays, a layer of fine sand is applied to the surface of the soil holding the barley seedlings. The parasitoids stand out against this layer of sand, which aids behavioural observations and locating dead animals. Some laboratories apply the test chemical after the sand has been placed in the pots which would create a relatively large area of inert material with a covering of unbound chemical. In a series of replicated tests using different insecticides and Aphidius species combinations, the sand layer was applied to the soil surface either before or 30 minutes after the plants were sprayed. Although somewhat ambiguous, the results indicated that, for some chemicals, applying the sand prior to spraying can increase mortality and thus influence the estimate of the LR50.

Initial range-finder tests involving four or five widely-spread doses with one or two replicates per dose were done in order to find an approximate location of the LR50. Subsequent definitive assays used 7 doses and a water control. Mortality data after 48 h exposure were corrected for background mortality and LR50s estimated using probit models. A. rhopalosiphi was the most sensitive species to all three chemicals, further supporting its use as an indicator species. For l-cyhalothrin the LR50 values for all Aphidius species were over 10x the highest recommended field rate. We feel that testing to this level of pesticide application is inappropriate and some upper limit should be placed on the highest concentration to be tested. LR50s were much higher at Tier II than at Tier I but the ranking of LR50s for the 3 species and 3 insecticides were the same at both tiers. At Tier I, all three chemicals gave HQs greater than 2 with A. rhopalosiphi but at Tier II, only dimethoate exceeded this trigger value, although imidacloprid had an HQ of 1.9, which would possibly induce further testing. In both Tier I and Tier II tests, 24 h and 48 h mortality checks gave significantly different LR50 estimates which would affect HQ values. Test durations should therefore be standardised, probably at 48 h. LR50 estimates based on probit and logit models were compared for Aphidius assays. There was a very close correlation between estimates from the two methods for both Tier I and Tier II data. However, whatever method of LR50 estimation is used in regulatory testing, a measure of the goodness-of-fit should be reported and a graphical representation of the dose response given. As the HQ for Tier I tests has now been set at two, it seems sensible that a geometric series of doses based on the field rate and a factor of two should be used. Such a recommendation would also mean that different laboratories are using exactly the same doses and thus make inter-laboratory comparisons easier.

Typhlodromus Tier I assays

The recommended ‘open plate’ method for Tier I assays with the predatory mite Typhlodromus pyri was assessed. This involves confining the mites in a defined area on glass cover slips using a sticky gel barrier. LR50 estimates for the three test chemicals differed significantly: l-cyhalothrin was approximately ten times more toxic than dimethoate, which in turn was ten times more toxic than imidacloprid.

One of the major problems with this open cell mite assay is the number of mites that are categorized as missing or become stuck in the sticky gel barrier, so that it is difficult to determine whether the assay is measuring toxicity or repellency. We investigated the effect of increasing the size of the defined area, comparing 1cm2 and 4cm2 areas. In dimethoate assays, more mites became stuck in the smaller arenas than in the larger ones, but for imidacloprid and l-cyhalothrin there was no significant effect of arena size. Attempts to use closed cell arenas based on circular washers of various materials, placed between glass cover slips, gave poor results due to condensation problems.

Calculated Hazard Quotients from our Tier I assays were compared for the two indicator species. Both species produced HQs over the trigger value of two for all three chemicals. However, the relative toxicities of the compounds were different: T. pyri was much less susceptible to the dimethoate than was A. rhopalosiphi and whereas l-cyhalothrin was the least toxic compound to A. rhopalosiphi in terms of its field rate, it was the most toxic to T. pyri, probably because a repellent effect of the pyrethroid greatly increased the number of mites trapped in the sticky barrier. Although repellent effects are important, it is undesirable to categorize a compound as highly toxic to beneficials when in fact it only has a repellent effect.

The main conclusions and recommendations from the work are summarised at the end of the full report. Figures referred to in the text of the main report are supplied as an appendix.

Full Report

UK Case Studies on Quantitative Risk Assessment (pdf, 39 pages).

Further information

Updated 2016-12-21