David's blog posts for September 2013

One more unto the breach; a look at Defra’s stance on neonicotinoids



On 27 March 2013, Defra released “An assessment of key evidence about neonicotinoids and bees”. They recently reiterated their conclusion that the ban was 'unnecessary and unjustified'. Defra’s chief scientist, Prof Ian Boyd, can be viewed on YouTube expounding the points made in their March document. I’d like to draw your attention to something which strikes me as very odd.

The report flags up three key studies that suggest there may be a strong link between exposure to neonicotinoids and bee ill-health. They then say that five field studies provide contradictory evidence, suggesting little or no impact of these pesticides on bees in the real world. The remainder of the document is a detailed critique of the three studies that did find a link – suggesting that they gave the bees unrealistic doses of pesticide, and that they were “lab” studies and so not representative of the real world (neither of which is true – all of the studies were performed largely or entirely outside, with free-flying bees - but that is for another day).

What I find extremely odd is that the five field studies are not subject to the same examination. Why on Earth not? If we are to weigh up the evidence, surely we should examine each study equally carefully? They are not mentioned again, although the document ultimately concludes that they were correct in their findings.

Let me fill in the gap, and provide a brief evaluation of these five documents, held up by Defra as “field studies in which bees were allowed to forage naturally in the presence of crops treated with neonicotinoids”:

1)      Blacquiere et al., (2012), Ecotoxicology 21: 973-992. There is no field study in this paper. It is a broad review of the subject, but presents no new data at all.

2)      Chauzat et al., (2009), Environmental Entomology 38: 514-523. This study examines mortality in 125 honeybee hives in France over three years, in relation to levels of pesticides found in their pollen, honey, and in the bees themselves. Low levels of imidacloprid were detected in the hives, but hive mortality was low, and imidacloprid exposure did not predict mortality. However, no experiment was conducted. There were no control nests, and we do not know if any nests were near treated crops. This study appears to be perfectly valid as far as it goes, but is not a good test of whether neonicotinoid pesticides harm bees.    

3)      Hecht-Rost S (2009) Eurofins-GAB GmBH. This is an industry report, presumably funded by an agrochemical company. It is not published in a scientific journal, and hence has not been subject to peer review. I cannot obtain a copy, so it is impossible to evaluate.

4)      Thompson H et al. (2013). This is also unpublished, and so has not been subject to peer review. It is a study conducted by Fera (the research wing of Defra) in which bumblebee colonies were placed next to an untreated, control field of oilseed rape, or next to fields treated with either imidacloprid or clothianidin. It is unreplicated – there was only one control field – and the nests by the control field were all exposed to neonicotiniods, so there were no longer any controls. In short, it was a disaster.

5)      Genersch E et al. (2010). Apidologie 41: 332-352. At a quick glance, this one looks quite good – similar to no. 2, but on a bigger scale. There is no controlled experiment, but they followed the fates of over 4,000 honeybee hives over time. They conclude that Varroa and other parasites are important in causing honeybee ill-health, but not pesticides. However, if you look a little deeper this is somewhat misleading. Only 210 of the colonies were actually screened for pesticides, while all were screened for Varroa. In their analysis of the impacts of pesticides, the authors sum exposure to all pesticides (including fungicides and herbicides), and then use this value to try to predict winter colony loss. This is a very odd analysis to do, since herbicides and fungicides mostly have very low toxicity to insects. Not surprisingly, they find no relationship between total pesticide exposure and colony death. They don’t actually perform ANY analysis of the impacts of the insecticides alone, or of neonicotinoids in particular. The study appears to have been funded by the agrochemical industry.

In summary, these five studies tell us almost nothing about the impacts of neonicotinoids on bees. There is not one successful experiment, with replication and proper controls, to be found amongst them. Two of the five aren’t actually published studies at all.

Compare this to the three key studies that Defra are at pains to dismiss. They are published in the best journals in the world (two in Science and one in Nature). They will have been through an excruciatingly tough peer review process. All three have good levels of replication, and proper controls. Which group of studies would you trust?

Have Defra actually read the five studies in which they place their faith? I'm not convinced that they have.

Dave Goulson's research lab website