Are Neonicotinoids the Sole Factor Responsible for Colony Collapse Disorder?

posted in: Science | 83
Frank explores a bee hive. Credit: KJHvM. See the whole album here.

A recent paper published in The Bulletin of Insectology claiming that neonicotinoids are the sole cause of CCD has been circulating in the media. The author, Chensheng Lu, has a history of doing research that makes spurious claims about the relationship between CCD and a specific group of pesticides. In this post, I am going to discuss Lu’s research, and use it as a stepping stone to discuss the role that pesticides play in honeybee health.

Why are honeybees exposed to pesticides?

Bees are insects which are raised as livestock, and kept around farms in order to pollinate crops. In order to combat mites which damage adults and spread diseases, beekeepeers use a variety of pesticides. The two most widely used are a pyrethroid called Fluvalinate and an organophosphate called Coumaphos. It is easy to forget that we treat these mites with insecticides, and many popular media reports neglect to mention this completely and instead focus on the agricultural pesticide angle. However, Fluvalinate and Coumaphos are found in virtually all pollen and wax samples. They are frequently found with chlorotalonil, which will synergize the activity of pyrethroids. Coumaphos is the only pesticide found more frequently in non-CCD afflicted colonies. These pesticides are an important part of the honeybee health story.

In 2010, a team lead by Christopher Mullin did a broad survey of agrochemicals found in American beehives that is (to my knowledge) the only wide-scale survey of agrochemicals found in American beehives. The dataset they generated is so immense that it is impossible to properly discuss in a single blog post, so I will need to relegate myself to a very narrow discussion. In general, pesticide levels were well below the lethal limit (Mullin et. al, 2010). It’s important to note that there were occasional exceptions where some insecticides, namely a variety of pyrethroids, organophosphates and the neonicotinoid imidacloprid, did approach lethal limits (Mullin et. al, 2010). Because Lu’s paper focuses on neonicotinoids I will focus the majority of my discussion on this group of pesticides. I will, however, return to discuss the broader implications of Mullin’s work.

How do neonicotinoids affect bees?

Honeybees are in a unique position in the world of insects which are important to agriculture. They’re social insects who must leave their home, find food by recognizing specific plants, return to a specific area, and communicate to their nestmates the location of food. In addition to this, there is a division of labor within the nest. Newly hatched bees will care for young, and perform pest control. As they age, they gradually move into a forager type position and guard the nest. This sort of lifestyle takes a lot of brainpower. Thus there is a concern about neurotoxins which affect the behavior of bees.

Alcohol affects human social behavior at sublethal levels, as well.
Alcohol affects human social behavior at sublethal levels as well.

The question of how toxic a pesticide can be is figured by determining which dose kills 50% of bees in a test group, a dose called the LD50. This dose is figured using individual bees, which is all well and good, but bees are a superorganism and this might not capture effects that effect colony health at lower than lethal levels. You need the superorganism LD50, which isn’t nearly as straightforward due to space limitations and variation between colonies.

Because bees rely on a complex set of behaviors, levels of pesticides that disrupt these behaviors represent a particular concern. This is actually pretty intuitive, and we can draw parallels with alcohol. A lot of important human activities, like driving and social interaction, also depend on complicated behaviors which can be disrupted by neurotoxins well below lethal levels. I like to draw parallels with alcohol, because it’s something everyone will understand. The NSFW website Texts From Last Night compiles textbook humorous examples of human social behavior which have been disrupted by the neurotoxin alcohol. It could be said that bees can potentially become drunk on neonicotinoids.

Neonicotinoids are neurotoxic pesticides which incorporate themselves into plants, and can be found in the pollen and nectar of treated plants. Neonics can be coated on the seed (known as seed treatment) or simply dumped in the soil around the plants (known as drench treatment), or injected right into the plant (usually reserved for ornamental crops). The Xerces Society has a good review on Neonicotinoid concentrations in plant tissues. Concentrations in pollen can vary widely and depends on the crop they’re used on, the application rate, and how the pesticides are applied. Seed treatments, which represent most neonicotinoid use, likely present few problems because these application rates are very low and result in pollen neonic levels of 1-2 ppb. Soil drenches may present problems, because neonic levels can exceed 50 ppb in pollen in some cases. The real problems with neonics lie with ornamental crops because these can have application rates 10 times those used in agriculture, and neonic levels in pollen can reach fatal levels.

Any response to Lu’s study shouldn’t shy away from discussing potential problems with neonicotinoids, however, in various interviews Lu makes it very apparent he thinks neonic seed-treatments are the cause of CCD. He frequently singles out corn pollen in his interviews, and claims (but never demonstrates) that the corn syrup beekeepers use is contaminated with the insecticides. He also frequently claims in these interviews that the levels he exposes bees to are lower than field rates.

Before long, I’ll demonstrate that Lu’s claims are false. But first things first…

How did Lu conduct his experiment?

Lu experimental setup, a rough pictorial description.Lu’s experimental design was a bit complicated. Lu began with 18 colonies, and split them into two groups of nine. One group of colonies was fed sucrose syrup; the other, High Fructose Corn Syrup (HFCS). These groups were subdivided into groups of three, and fed with one of two neonicotinoid pesticides (Imidacloprid or Clothiandin); another was fed only syrup. In essence, he had 3 pairs of triads fed either different syrup or different pesticides. One representative from each pair of triads was shuttled to one of three apiary locations.

He divided three of these groups among three sites, let them forage for the summer and began to feed the treatment groups pesticides once they were ready to overwinter. He treated the colonies at .74 ng/bee/day for the pesticide treated groups. Treatment time was 13 weeks during the overwintering stage.

If we take Lu’s interpretations at face value, he believes that he managed to replicate CCD because his neonicotinoid treated hives shrunk during the winter, without dead bees accumulating at the bottom of the hive. According to Lu’s analysis of his team’s results:

One of the defining symptomatic observations of CCD colonies is the emptiness of hives in which the amount of dead bees found inside the hives do not account for the total numbers of bees present prior to winter when they were alive. On the contrary, when hives die in the winter due to pathogen infection, like the only control colony that died in the present study, tens of thousands of dead bees are typically found inside the hives. The absence of dead bees in the neonicotinoid-treated colonies is remarkable and consistent with CCD symptoms.

Did Lu manage to replicate CCD using neonicotinoid treatments?

Unfortunately there’s a lot of areas where Lu went wrong, both in his methods and in his interpretation of his results. His main finding is that he believes that he replicated CCD, and this has been widely reported in the media. So let’s tackle his take-home message first, before moving onto the nitty-gritty details.

Remember that Colony Collapse Disorder is a very specific set of symptoms, and that dead and abandoned hives aren’t neccessarily afflicted with CCD. In general, beekeepers lose about 1/3rd of their hives to things like parasites, pathogens, pesticide poisoning, and CCD. CCD accounts of about 1/3rd of the lost hives, or about 1/9 of the total colonies. This is how the USDA defines CCD:

  1. Sudden loss of the adult bee populations with very few bees found near the dead colonies.
  2. Several frames with healthy, capped brood
  3. Low levels of parasitic mites, and absence of nest-damaging kleptoparasites (e.g. wax moths, hive beetles).
  4. Avoidance of hive by other bees
  5. Laying queen present, with a small number of newly emerged adult bees.

So…did Lu recreate CCD as he claimed?


LuCCDfailAt best, Lu’s results in both his 2012 and 2014 papers can be interpreted to demonstrate that honeybees can have trouble overwintering if fed high amounts of neonicotinoids. This point is well taken, but his claim of demonstrating that neonicotinoids are the cause of CCD is a giant overstep in interpretation.

Lu did report declines in levels of bees which seem to indicate that insecticides made a difference in adult survival during overwintering. However, Lu also reported that of the honeybee hives which survived the insecticide treatment, all had either no queen or no brood. Hives afflicted with CCD must have both queens and brood. He did, to his credit, report low levels of mites and didn’t mention the kleptoparasites so he’s got that working in his favor. However, the symptoms he reported did not match the case definition CCD. The biggest problem with Lu’s interpretation of his data is that he conflates hive abandonment with CCD.

To understand this result, you need to know a little bit about social insect biology. While hive abandonment is a part of CCD, hive abandonment is not unique to CCD. Hive abandonment in social insects is, in essence, suicidal quarantine. They leave the hive in order to prevent spread of disease to nestmates. Hive abandonment can be triggered by pathogens (e.g. Malpighamoeba mellificae), sublethal doses of toxic chemicals other than neonicotinoids (e.g. pyrethroids), even CO2 narcosis (Evans & Schwarz, 2011, vanDame et. al, 2009, Cox & William, 1987, Rueppell et. al, 2010). Hive abandonment is a generalized response to sickness in social insects, and by itself does not indicate CCD.

The picture in the next paragraph allows a side-by-side comparison between Lu’s hives and a hive which has CCD. Suffice to say, Lu’s pictures are very different from those which have been affected by CCD. There are very important differences between the colonies Lu poisoned with insecticide and those which have been affected by CCD. Despite these differences, Lu claims he has replicated CCD. However, his data demonstrates that he did not replicate CCD.

What did Lu do wrong?

To the left is a picture of a representative frame from the honeybee colonies Lu poisoned with imidacloprid in his 2012 study. On the right are colonies afflicted with CCD. Note that the frames from a CCD colony (Oldroyd, 2007) consist almost entirely of sealed brood, while the frame Lu is claiming is afflicted by CCD (Fig. 3 in his 2012 paper) consists entirely of honey, and no sealed brood. The pictures Lu shows in his papers do not resemble those of hives affected by CCD, yet despite this he still claims he has replicated CCD.
To the left is a picture of a representative frame from the honeybee colonies Lu poisoned with imidacloprid in his 2012 study. On the right are colonies afflicted with CCD (Oldroyd, 2007). Note that the frames from a CCD colony consist almost entirely of sealed brood, while the frame Lu is claiming is afflicted by CCD (Fig. 3 in his 2012 paper) consists entirely of honey, and no sealed brood. The pictures Lu shows in his papers do not resemble those of hives affected by CCD, yet despite this he still claims he has replicated CCD.

Lu’s tests were not precisely performed, and suffered from a small sample size of 18 hives. In essence, he had six treatment regimes but treated them as three by merging the two different types of separately prepared syrup for his analysis. While this might not have had a huge effect, it probably still introduced some variability.  Many other variables were completely unaccounted for. For example, it wasn’t discussed if the dead colonies were spread evenly through the sites, or if they were found in the same site. Instead of measuring the temperature at his colonies, he instead looked at NOAA measurements taken at the local airport. I could discuss these problems in detail, but I think the problem which is most worthy of discussion is the fact that Lu is claiming that this experiment demonstrates that neonicotinoids are responsible for wide-scale problems with honeybee health.

Lu’s experiment was a no-choice feeding test, which is kind of similar to a cell-culture test. The tests occur in an environment which may not match real-world scenarios. Basically, he dosed the bees with neonicotinoids by pumping them straight into the colony in corn syrup which they’ll consume because it’s the closest food source.

No-choice tests don’t take into account the fact that bees will feed preferentially from different sources in real-world situations. So while levels of neonicotinoids in pollen from seed-treated crops will vary from 1 ppb to more than 10 ppb, the levels they actually eat are very different for a number of reasons. At a certain level of contamination (~20 ppb in corn syrup), it’s likely they’ll avoid pesticide contaminated pollen (Blacquiere et. al, 2012). They’ll also gather pollen from uncontaminated crops, and mix it in with the contaminated pollen. So neonicotinoid residues in pollen don’t necessarily reflect what they are in bee colonies.
Image courtesy of Randall Munroe, at XKCD

Doses in honey aren’t nearly as straightforward because there’s scarce data for the United States. However, they appear to be in the neighborhood of 1 ppb in other countries (Blacquiere et. al, 2012). Based on levels found in actual bee colonies a field-realistic dose of neonicotinoid in pollen is probably 1-3 ppb, although significantly higher levels can occur (Mullin et. al, 2010, Blacquiere et. al 2012). However, the neonicotinoids are not found ubiquitously in bee pollen. Mullin et. al 2010, for example, found imidacloprid in less than 3% of pollen samples taken from colonies. Lu treated his bees at 136 ug/l of syrup. Assuming a liter of syrup weighs 1375 grams (based on density of HFCS), his syrup contained 99 ug/kg of the neonicotinoid insecticides, which is a 5x higher dose than even the most contaminated pollen bees are likely to encounter in crops with seeds treated as per the pesticide label. More importantly, it’s also a dose 33-fold higher than the neonicotinoid contaminated pollen which is found in typical honeybee colonies. Bottom line: he appears to have overdosed the colonies compared to what they are encountering in the real world.

There are also some severe issues with his dosing schedule. He claimed that he dosed the bees at .74 ng/bee/day, but the paper seems to indicate that he did not change the dosing schedule as the populations declined during the winter. As the bees declined, the dosage per bee increased. He also neglected to measure the bee populations to determine his initial dose. He merely assumed a starting point of 50,000 bees.

Lu does not cite literature which undermine his Hypothesis

Just as telling as Lu’s misinterpretation of his results, and his questionable methods, are Lu’s citations in the article. He cites a handful of popular media reports, which is unusual but not completely unheard of in the scientific literature. However, there are a lot of citations which should be there but aren’t. For instance, there is no discussion of field-realistic levels of agrochemicals found in honeybee colonies. This information can be easily found in open-access journals (Mullin et. al 2010, Blacquiere et. al, 2012). Perhaps the most damning omission from Lu’s citation list are field trials. There have been many field trials which have attempted to look at the effects of neonicotinoids on bees (Pohorecka et. al, 2012, Schmuck et. al 2012, Cutler & Scott-Dupree, 2007, Nguyen et. al, 2009, just to name a few). While they all represent different scales, and while each has its own quirks and intricacies, they overwhelmingly indicate that neonicotinoids do not affect colony health under field conditions and proper use. These are high profile papers which are easy to find and which cast doubt on Lu’s claims, but Lu makes no attempt to reconcile his results with these tests.

Pesticides and Bees: the Bigger Picture

The relationship between pesticides and bees is extremely complex, and would probably take several dozen posts to fully discuss. Earlier, I mentioned detections of specific pesticides in Mullin, 2010 and I’d like to return to that point. This study reported a high amounts of imidacloprid in one sample, but neonicotinoids were found in less than 10% of the samples tested. When they were found, they were far below the levels which caused harm (Blacquiere et. al, 2012). While neonicotinoids didn’t appear in the concentrations or frequency which could cause harm, the team found that multiple types of pesticides (mainly pyrethroids and organophosphates) can approach LD50 levels in honeybee colonies. We’d reasonably expect sublethal effects in the colonies where they reach these levels. Often, these are found with fungicides which will synergize their activities by blocking the enzymes bees use to detoxify them. The effects of the miticides on bees are similarly in question, given their frequent and high detections in colonies. A lot of pesticides are found in honeybee colonies and based on pesticide survey data the USDA actually suggests that pyrethroid insecticides are a higher risk to honeybee colonies than are neonicotinoids. Ironically, Lu cited this report but did not discuss this conflict between his data and the views of the larger community.

Frank loves bees. Credit: KJHvM

Neonicotinoids are a small subsection of the pesticide story. Pesticides are a very small part of the CCD story, and are a small part of the overall honeybee health story. Landscape changes due to agricultural intensification and urbanization can change the diversity of the available food, which can change the physiological status of the bees. Bees are kept in crowded, stressed conditions and are driven cross-country where they will come into contact with other bees. This creates an opportunity for rapid disease transmission to new populations, and creates conditions which favors strains of existing diseases evolving to become more virulent. Pesticides act as a stressor on top of all these. Because there are so many interacting factors, it is generally believed that there is no ‘One True Cause’ of CCD.

The story of CCD is a serious one, and it should be discussed in the public sphere. What disturbs me about this discussion is that the Lu paper discussed above has managed to go viral among the media outlets not because it’s quality science but because it fits an anti-pesticide narrative that the media has become increasingly comfortable with. The standard neonicotinoid narrative is convenient because it makes the situation simpler…a single problem, and a single solution which involves banning a single substance. However the real pesticide story involves dozens of compounds with wildly different uses, which interact with biological and environmental factors which are still poorly understood at best. The neonicotinoid story is just as complex because they likely don’t cause problems in all crops, but issues with proper use and application rates still need to be sorted out. There’s also a human component in some systems which is never discussed, where neonicotinoids frequently replace pesticides more toxic to people like organophosphates. Unfortunately, Lu’s research does nothing to highlight legitimate issues with these pesticides in particular. The thing that perhaps makes people the most uncomfortable, is that unlike climate change or evolution, the issues discussed here are not a case of settled science and continue to evolve as we better understand these factors.


Lu, Chensheng. Warchol, K. Callahan, R. (2014) Sub-lethal exposure to neonicotinoids impaired honey bees winterization before proceeding to colony collapse disorder. Bulletin of Insectology 67 (1) 125-130

Lu, C. Warchol, K. Callahan, R. (2012) In-situ replication of honey bee colony collapse disorder. Bulletin of Insectology (65) (Online)

Mullin, C. Frazier, M. Frazier, J. Ashcraft, S. Simonds, R. vanEngelsdorp, D. Pettis, J. (2010) High Levels of Miticides and Agrochemicals in North American Apiaries: Implications for Honey Bee Health. PLOS ONE. 5(3)

Blacquiere, T. Smagghe, G. van Gestel, CAM. Mommaerts, V. (2012) Neonicotinoids in bees: a review on concentrations, side-effects and risk assessment. Ecotoxicology. 21: 973-992

Stoner KA, Eitzer BD (2012) Movement of Soil-Applied Imidacloprid and Thiamethoxam into Nectar and Pollen of Squash (Cucurbita pepo). PLoS ONE 7(6): e39114. doi:10.1371/journal.pone.0039114

Evans JD. Schwarz, RS. (2011) Bees brought to their knees: microbes affecting honey bee health. Trends in Microbiology. 12: 614-620.

van Dame, R. Meled, M. Colin, ME. Belzumces, L. (1995) Alteration of the Homing-Flight in the Honey Bee Apis mellifera Exposed to Sublethal Dose of Deltamethrin. Environmental Toxicology and Chemistry 14(5): 855-860.

Cox, R. Wilson, W. (1984) Effects of Permethrin on the Behavior of Individually Tagged Honey Bees, Apis mellifera (Hymenoptera: Apidae). Environmental Entomology 13(2): 375-378

Rueppell, O. Hayworth, MK. Ross, NP. (2010) Altruistic Self-Removal of Health-Compromised Honey Bee Workers From Their Hive. Journal of Evolutionary Biology 23(7): 1538-1546

Pohorecka, K. Skubida, P. Miszczak, A. Semkiw, P. Sikorski, P. Zagibajlo, K. Teper, D. Zbigniew, K. Skubida, M. Zdanska, D. Bober, A. Residues of Neonicotinoid Insecticides in Bee Collected Plant Materials From Oilseed Rape Crops and Their Effect on Bee Colonies. Journal of Apicultural Science 56(2): 115-134

Schmuck, R. Schoning, R. Stork, A. Schramel, O. (2001) Risk Posed to Honeybees (Apis mellifera, Hymenoptera) by an Imidacloprid Seed Dressing of Sunflowers. Pest Management Science 57(3): 225-238

Cutler, C. Scott-Dupree, C. (2007) Exposure to Clothianidin Seed-Treated Canola Has No Long Term Impact on Honey Bees. Journal of Economic Entomology 100(3): 765-772

Nguyen, BK. Saegerman, C. Pirard, C. Mignin, J. Widart, J. Thirionet, B. Verheggen, F. Berkvens, D. DePauw, E. Haubruge, E. (2009) Does Imidacloprid Seed-Treated Maize Have an Impact on Honey Bee Mortality? Journal of Economic Entomology 102(2): 616-623

USDA (2013) Report on the National Stakeholders Conference on Honey Bee Health. Online.

Oldroyd BP (2007) What’s Killing American Honey Bees? PLoS Biol 5(6): e168. doi:10.1371/journal.pbio.0050168

Follow Joe Ballenger:
Joe Ballenger is an entomologist who works in the biotech industry as a contractor. In his spare time, he helps answer questions about bugs at Ask an Entomologist.