VIII SIMPOSIO INTERNAZIONALE ICPBR

HAZARDS OF PESTICIDES TO BEES

Bologna, 4-6 Settembre 2002

Metodologie dei test e studi sugli effetti dei pesticidi sulle api

A semi-field approach to testing effects of fresh or aged pesticide residues on bees in multiple rate test designs

Frank Bakker (1) and Johan Calis (2)

(1) MITOX Trial Management BV, P.O. Box 92260, 1090 AG Amsterdam, NL. E-mail: fbakker@science.uva.nl
(2) INBUZZ Beekeepers, Ijsbaanweg 8, 1251 VV Laren, NL

We describe a semi-field cage test that involves the use of standardised mini-beehives. The test can be used to evaluate the effects of plant protection products using several exposure scenarios, such as direct contact resulting from applications performed during bee flight and exposure to aged residues from applications performed at pre-determined intervals before the initiation of exposure. The latter feature makes the test also appropriate to evaluate the effectiveness of proposed risk mitigation strategies such as evening applications or the delayed introduction of bees.

The principle of the trials is to prepare two groups of potted test plants per treatment. The first group of plants remains untreated, while the second group is treated at the desired rate and interval before exposure. Standardised honeybee colonies, individually enclosed in meshed cages of 20 m2, are confined to the untreated plants for a period of four days before the start of the exposure phase. In this period foraging activity and mortality are monitored daily. In the evening before the initiation of exposure, the untreated plants are exchanged with treated plants. During the subsequent four days monitoring of foraging activity and mortality is continued. The trial is concluded by inspecting the colony for brood development, presence of the queen, food/water and by determining weight loss of the colony.

Honeybee colonies are standardised with respect to age structure and total honeybee weight shortly before the start of the experiment. To enable a straightforward assessment of mortality, the colonies were manipulated such that no new adult honeybees would emerge during the trial period.

The relatively small size of the test units and the high degree of standardisation achieved with the set-up make the test highly reproducible and allows for the simultaneous testing of several treatment groups in a replicated test design. In our set-up the level of replication is normally four units per treatment and up to eight treatments can be tested simultaneously. We illustrate this using data from trials with commercially available insecticides.

With this level of replication and the pre-post test approach a powerful statistical analysis is possible. Effects on mortality are analysed by comparing the number of dead bees found in the different treatment groups to the water control using a covariance alternative to repeated measures ANOVA, using pre-treatment mortality as a predictor variable to account for potential intrinsic among colony differences in mortality. Effects on foraging activity are analysed by comparing the total number of foraging bees observed at the final assessments of each day to the water control. The number of flowers present in each cage at the end of the trial may be included as a covariate in case inhomogeneities in the numbers of flowers among cages are apparent.