This article originally appeared at Forbes and has been republished here with the author’s permission.
Researchers in the U.K. this week have provided us the most recent example of a scourge about which I and others have written previously: science by press release. It often includes almost everything that is wrong with what passes for science today: ex post facto cherry-picking of data to support an agenda-driven conclusion (a form of “confirmation bias”); hyping of questionable results to garner headlines; the failure of large parts of the science community to call their colleagues on what are clear and egregious distortions; and shortcuts by “science writers” who substitute parroting the press release for critical thinking.
The most recent hype surrounds an article, “Country-specific effects of neonicotinoid pesticides on honeybees and wild bees,” by B.A. Woodcock of the U.K.’s Center for Ecology and Hydrology (CEH) and his colleagues, published on Thursday in the journal Science.
For those coming late to the saga, neonicotinoids–or neonics for short–are an innovative group of insecticides that are used primarily as a coating on seeds, reducing the total amount of pesticide applied to a tiny fraction of what would otherwise be used in spray applications of older pesticides. This is far more benign to beneficial insects (e.g. bees) than the older, sprayed pesticides that neonics replaced–and that farmers are forced to revert to when neonic use is restricted or eliminated. Neonics are also safe to humans and other mammals, which is an advantage for farmers–and for pet owners, who apply neonics directly to the skin of their beloved dogs and cats to keep them free of fleas and ticks.
Not surprisingly, neonics have become the most popular, widely used insecticide on the planet–which has put them squarely in the cross hairs of environmental activists and the organic industry. Claiming (inaccurately) that neonics were causing a collapse in honeybee populations, threatening their imminent extinction and jeopardizing the world’s food supply, bad-faith activists have managed to get them mostly banned in Europe.
Only after the ban was in place did a few science writers in the United States start pointing out that bee populations weren’t collapsing as claimed by the activists and EU regulators. Bee populations are, in fact, increasing–in Europe, the U.S., Canada and indeed on every continent except Antarctica–and have been rising since the mid-1990s, when neonics first came on the market.
That realization probably prevented a copycat ban in this country under the Obama Administration, but it hasn’t caused the EU regulators to skip a beat. They’ve doubled down on the need for a ban, basing their conclusions on shoddy lab studies that subject bees to massive overdoses and find, unsurprisingly, that the bees don’t like being overdosed.
Until now, however, every large-scale field study conducted has demonstrated what we know from real-world experience–honeybee colonies actually thrive on neonic-treated crops. That’s why the study just published in Science, expensively underwritten by industry (an irony I will return to) and one of the largest ever conducted, will certainly gain the headlines its authors so obviously covet.
On its face, the CEH study looks impressive and important. Commercially grown oilseed rape fields were treated with two different neonic pesticides, clothianidin (CLO) and thiamethoxam (THX), in three different countries–Germany, Hungary and the U.K. Scientists measured the effects of field-realistic exposures to these treated crops on three different bee species: domesticated honeybees and two wild species–bumble bees (Bombus terrestris) and “red mason” solitary bees (Osmia bicornis).
According to the press release for this study, the researchers found that “exposure to [neonicotinoid] treated crops reduced overwintering success of honeybee colonies…in two of the three countries.” And “[l]ower reproductive success–reflected in queen number (bumblebees) and egg production (red mason bee)–was linked with increasing levels of neonicotinoid residues in the nests of [the two] wild bee species…across all three countries.”
That prompted the study’s lead author, Dr. Ben Woodcock, to summarize: “The neonicotinoids investigated caused a reduced capacity for all three bee species to establish new populations in the following year, at least in the U.K. and Hungary.”
What neither the press release nor the two-page published article conveyed, however, is that the authors are basing their study’s conclusions on a handful of outlier results–effects so small that they could be occurring by chance–amid a much larger amount of experimental data they generated that points to precisely the opposite conclusions. Inexplicably, this was not caught and required to be corrected by the peer-reviewers.
Because the larger part of the study on honeybees was actually commissioned by Bayer CropScience and Syngenta, the two largest manufacturers of neonic pesticides, the companies received a far more extensive report on the contracted experiments. In his Wednesday appearance at the Science Media Centre in the U.K., Syngenta scientist Dr. Peter Campbell pointed out to the press that this full data set runs to around 1000 pages, encompassing 258 separate statistical analyses of the primary and secondary scientific endpoints (measurable results) of the study. Needless to say, this full data set dwarfs the two-page study published in Science and the 50 pages of published “supplementary materials.”
Here’s what this mass of data looks like laid out in a matrix that was provided to the public by Dr. Campbell during his talk.
CEH Endpoint Analysis (Source: Peter Campbell)
The nine red squares–about 3% of the results–indicate a negative effect on honeybee colonies for those measured endpoints (compared to bees exposed to untreated control fields). The seven dark green squares–also about 3% of the results–indicate positive, beneficial effects on honeybee colonies, compared to control.
These results are striking: Beneficial, positive effects on honeybees–from exposure to a pesticide–were observed on almost as many endpoints as negative effects. That’s unexpected–especially in a study claiming that neonics are harming honeybees. One might expect that these unanticipated positive effects–observed in both the U.K. and Germany only slightly less frequently than the negative effects seen in the U.K. and Hungary–would merit equally prominent billing and inspire deeper investigation. Instead, the CEH team chose to bury this result, at least in its press release, noting blandly: “No harmful effects on overwintering honeybees were found in Germany.”
The other obvious feature of this matrix is the broad expanses of light green squares– 238 of them, the other 94% of the data, each denoting experimental endpoints on which neonic exposure had no effect on honeybee colonies. This is of course exactly what farmers and home gardeners want in a pesticide: something that kills pests but leaves bees and other beneficial insects unharmed.
This widespread “no effect” finding is consistent with the results of about a dozen previous large-scale field studies of honeybees foraging in neonic seed-treated crops, which found no adverse effects at the colony level.
The CEH article virtually ignores this mass of experimental data (not even bothering to publish the “no effect” results for nine measured sub-lethal endpoint sets in their study’s supplemental data). Instead, they base their conclusions on nine endpoint results observed in two countries–three percent of the study’s data. These results may well be the result of chance.
In short, it looks as though the CEH team has built its conclusion that neonics harm honeybee colonies’ ability to survive the winter on a handful of possibly random effects amid a plethora of other data in their study that point to the opposite conclusion–that neonics had no adverse effects on honeybee colonies.
The CEH team also failed to mention that all of the honeybee colonies negatively affected by neonics had recovered from those effects before winter–and so entered their over-wintering period unimpaired. This is not a minor omission. It is a deficiency that other, independent scientists need to take into account when assessing the overall validity of this paper.
But what of the CEH team’s finding that neonic exposure harmed the two wild bee species–the ones they studied “on their own nickel?”
It turns out that once again the CEH scientists omitted some important qualifiers. First, as Dr. Peter Campbell, the Syngenta scientist, pointed out, “circa 95% of the time, no neonic residues” were found in CEH’s pollen and nectar residue analyses, “even in samples taken directly from the treated crop. Therefore, bees in these trials were hardly ever exposed to any neonic residues.”
Emeritus Professor Rob Smith of the University of Huddersfield pointed to a second confounding factor in the CEH paper’s wild bee data. It turns out that the two wild bee species studied were unaffected by neonic residues in the pollen and nectar directly from the treated crops. The negative effects came from accumulated residues in their nests–and those effects were dominated by a third neonic (imidacloprid), which was not used in the study but apparently was encountered by the wild bees during their travels.
The CEH team was no more transparent or forthcoming about these confounding features of their wild bee findings than about their honey bee results. This reinforces the appearance of agenda-driven findings in search of confirmation from limited, inconclusive data. CEH was unable to be reached for comment.
When scientists cherry-pick their data and magnify the significance of sparse results while burying or underplaying contrary indications in their own data, they do science–and society–a disservice. Before this article can influence public policy, it will be essential for various experts to weigh in on it.
Henry I. Miller, a physician, is the Robert Wesson Fellow in Scientific Philosophy & Public Policy at Stanford University’s Hoover Institution. He was the founding director of the FDA’s Office of Biotechnology. Follow him on Twitter @henryimiller.
