In the last few months, there’s been a steady stream of new publications about honey bees and pesticides. One news item ran with the headline: Mystery of the Disappearing Bees: Solved!”
Um. NO. Not even close.
I got a bajillion emails about this paper from a Harvard researcher:
“In Situ Replication of Honey Bee Colony Collapse Disorder,” Chensheng Lu, Kenneth M. Warchol, Richard A. Callahan. Bulletin of Insectology. June 2012.
This paper claims two things we’re already conditioned to think are evil cause CCD: high fructose corn syrup and pesticides. It’s from a big name, widely respected university. It is also, sadly, a flawed paper. (The fact that it’s in the “Journal of Insectology” should be a tip-off. Is that even a word?)
A fair amount of bashing is going on in the bugosphere:
“My reading of the paper suggests that the author knows little about bees, little about pesticides, nothing about HFCS, and had no understanding of the distribution of systemic pesticides in plants. “
Ouch! And that was from one of the nicer reviews. What immediately jumped out at me when I read the paper was that the control hives failed.
One consistent criticism is that the levels of pesticide were increased in the middle of the trial when they didn’t immediately see bee deaths. That isn’t how you do science–especially when you increase the levels of pesticide to well over 40 times what is normally seen in the field. Oddly enough, that still didn’t kill all the bees. Randy points this out:
“…it appears that the data from this study actually support an alternative hypothesis–that field realistic doses of imidacloprid had no measurable adverse effects upon the colonies. And even patently toxic doses had little immediate effect.” [emphasis original author]
This paper is a good example of the problem I mentioned in an earlier post, of some research being given more weight simply because it fits within a narrative of what we believe ought to happen–even if the evidence doesn’t support it.
Do pesticides kill bees? Oh Hell Yes. Pesticides are bad for bees, and there is no doubt about that. That is not the same question, though, as “do pesticides cause Colony Collapse Disorder?”, which is a distinct syndrome with clear signs. And it’s also a different question from “Which pesticides harm bees most?” and “How do pesticides harm bees?”
The answers to all of these questions based on the scientific literature from the last few years is incredibly mixed. Lab trials don’t match field trials. Very wide dose ranges and modes of delivery are used, and results are inconsistent from study to study. Something is going on, certainly, but if pesticides–in particular the neonicotinoid class of pesticides–were the primary factor in bee deaths, I would expect more coherence in the literature.
This is why I think the paper that should really be getting all the press is this one:
Cresswell, J., Desneux, N., & vanEngelsdorp, D. (2012). Dietary traces of neonicotinoid pesticides as a cause of population declines in honey bees: an evaluation by Hill’s epidemiological criteria Pest Management Science, 68 (6), 819-827 DOI: 10.1002/ps.3290
Science already has a lot of tools for picking apart complex causes of diseases and environmental changes. In this paper, three leading bee scientists from the three countries most involved in CCD research–USA, UK, and France–evaluate CCD using one of those tools. From the paper:
“Normally, the results of manipulative experiments are the hard currency of decisions about causality in natural science. In situations involving public concern over environmental change, however, decisions about causes sometimes must be made under political pressure, in spite of scientific uncertainties, which may include the lack of experimental evidence. In such circumstances, a scientific evaluation is nevertheless possible, but it uses a different process to manage uncertainty and to validate its conclusions.”
They evaluate if there is enough evidence to implicate neonicotiniod pesticides in bee deaths using a standard epidemiological method developed by Hill. It looks at nine different criteria, and assigns a weight to each one (1-Slight, 2-Reasonable, 3-Substantial, 4-Clear and 5-Certain) with regard to the cause-effect hypothesis. The value (+/-) of each score relates to the nature of the effect; positive if it causes a population decline, and negative if it has little effect on the population.
Because it’s a complex paper, I’ve reproduced their analysis table here, and will only discuss some of the criteria they used. The first criteria is experimental evidence–they assigned this a value of -1, since there isn’t a consistent pattern to support the hypothesis that neonicotinoid pesticides cause CCD.
The second criteria is coherence, which evaluates whether identifying a factor as the cause of a particular phenomenon conflicts with established knowledge. It’s certainly a reasonable assumption that these pesticides harm bees, so the value of the score is positive (+3, Substantially supports the hypothesis). However, it does not receive a higher score because despite lots of research, we still don’t have a numeric value or dose that is a meaningful (or consistent!) threshold of harm.
The temporality criterion asks whether “the cause precedes the consequence.” As you can see from the graph I’ve reproduced here, bees were already starting to decline before the introduction of neonicotinoid pesticides. And despite pesticide use increasing steadily, the bee declines are far slower.
Since trace dietary neonicotinoids neither preceded nor intensified honey bee decline, this factor had a score of -4; Clearly lacking evidence to support the hypothesis.
The next factor, consistency, asks whether the association between the hypothesized cause and consequence is repeated in space and time. It isn’t, within the US or worldwide. From the paper:
“Worldwide, honey bee declines are not ubiquitous, and, according to figures produced by the United Nations Food and Agriculture Organization, the global stock of managed colonies has increased by 45% in the last 50 years, in spite of the declines in North America and Europe. Even in Europe, stocks of colonies have increased in some countries, such as Spain, where the numbers have risen by over 50%”
The specificity criterion asks whether the consequence is uniquely associated with the hypothesized cause. As I have detailed before, pretty much everything wants to kill your bees. A whole bunch of other factors clearly cause declines in bees, including:
- increased losses due to Varroa mite;
- diseases such as Israeli Acute Paralysis virus and the gut parasite Nosema;
- pesticide poisoning through exposure to pesticides for in-hive fungal or mite control (NOT neonicotinoid pestides; these are different)
- habitat loss for foraging; inadequate forage/poor nutrition;
- poor nutrition and migratory stress brought about by the increased need to move bee colonies long distances to provide pollination services.
When you add this all together, you come up with the conclusion that neonicotinoid pesticides are not the cause of honey bee declines. Because non-entomologists mostly see just a few papers that are covered by the media, it creates the illusion that there is far more evidence for pesticide causes of CCD than actually exists.
Is it possible that neonicotinoid pesticides in the environment harm bees? It’s not only possible, it’s likely. But we can’t just assume that pesticides did it (“I knew it! Those bastards had it in for bees from the beginning.”). By focusing on just one of the many threats that harm bees, we are missing the real story.
An incredibly complex host of factors makes it hard out there for bees of all sorts–and not just honey bees. We have to evaluate all of them, and how they might interact, to try to help bees make a comeback.
Bug Gwen 
I’m a docent at a wildlife refuge with endangered plants where pollination is a topic on every tour. Recently I’ve had a rash of self-appointed experts who want to explain to the group how neonicotinoids have been identified as the cause of CCD. This post will certainly help me address that misconception.
I really appreciated the post and the explanations that accompanied it. When something is outside your field and you rely on doubtful scientific journalism without reading the original paper it is difficult to realise how far off some assertions can be.
Thank you, a thousand times over for a clear discussion. Two additional comments: The number of managed colonies are biased (Fig. 1) to a higher number than they might have been because beekeepers can replace colonies fairly readily (but at great expense). This obscures the total impact of bee colony losses from all causes and weakens the value of the relationship between colony losses and neonicotinoid use pattern. Internationally, there is much stronger negative evidence regarding bee losses and neonics being the cause of bee losses. For example, Australia uses neonics, does not have Varroa mites that vector viruses, and lacks CCD. Secondly, there is another hypothesis (Berenbaum and Robinson’s work) that lack of genetic diversity within hives (possibly due to artificial insemination rather than mating during nuptial flights) predisposes managed bees to being less resilient to stresses (diseases, in particular).