The value of large-scale field trials for determining the effects of pesticides on the non-target arthropod fauna of cereal crops

Abstract

The short-term (i.e. within-season) risk to non-target arthropods posed by broad-spectrum insecticides is arguably best evaluated in large-scale field trials, as only these can provide quantitative assessments of the impact of products at the population level. However, although they provide realism, field trials have their limitations. From a biological viewpoint, the most important of these is that the arthropod populations being sampled need to be both bountiful and relatively homogeneous across a trial site. If they are not, it is often difficult to demonstrate a statistically significant treatment effect, even with a harmful product. Unfortunately, the density and heterogeneity of populations cannot be predicted accurately, even with extensive pre-treatment sampling. Furthermore, since there may be considerable variation in the numbers of arthropods present between sites and between years, there is a need for caution when interpreting data from single-site studies. From a purely practical viewpoint, the greatest limitation on the design of field trials is often the labour resources required to take, process and analyse a large number of samples. In designing a sampling strategy, it is often necessary to compromise between what is considered ideal (i.e. high sample replication to measure leve;s of within-plot variance more accurately), and what is feasible in terms of the time and labour available.