Category Archives: Science

Does DWV infect bumble bees?

Covid (the disease) is caused by a virus called SARS-Cov-2. SARS is an abbreviation of severe acute respiratory syndrome and the suffix ‘Cov’ indicates that it’s a coronavirus. The final digit (2) shows that it’s the second of this type of virus that has caused a pandemic. The first was in 2003, and was caused by a virus we now call SARS-Cov-1. That virus had a case fatality rate of 11%, but only infected ~8500 people worldwide. 

SARS-Cov-2 is not a human virus, by which I mean it’s not a virus that has been present in the human population for a long time. It’s actually a virus that most probably originated in bats.

We’re still not sure how SARS-Cov-2 jumped species from bats to humans.

SARS-Cov-1 made the same transition from bats and we do have a pretty good idea how this happened. Before the virus was found in bats it was detected in palm civets and raccoon dogs, both of which are farmed for food and sold in live game markets. Neither animal shows any symptoms when infected with SARS-Cov-1.

Bats were also sold in the same live game markets in Guangdong province in China and it seems likely that the virus either crossed directly from bats to humans, or went via a third species such as the palm civet.

Pathogen spillover

It’s likely that SARS-Cov-2 followed the same route. We are entering a second – likely extended – period of geographic lockdown due to Covid. Since SARS-Cov-2 made its cross-species jump from bats to humans it has infected at least 41 million people and killed over 1.13 million.

Pathogens that jump from one species to another can have catastrophic consequences for the recipient population 1 or, as in the case of palm civets, might cause no harm whatsoever.

The term ‘pathogen spillover’ is often used to describe the event when a pathogen – whether viral, bacterial or a parasite – escapes or spills over from its natural host to another species. 

CSI in the apiary … motive, opportunity, means

To make the species jump a number of criteria must occur. The pathogen has to be present at a high enough level to be infectious in the ‘donor’ species. For convenience, let’s consider this as the motive to jump species 2.

Secondly, the pathogen needs to have an opportunity to jump species. For example, because the donor and ‘recipient’ species share the same habitat or regularly come into contact.

Finally, it has to have the means to replicate in the recipient. If it cannot replicate it can never get established in the recipient population or cause disease.

In fact, this is an oversimplification. It also needs to be transmissible between individuals of the recipient species (or it will never spread in the recipient species).

So what has all this to do with the bumble bees in the title of this post?

Honey bees get a bad press from some scientists and environmentalists. They compete with native solitary and other bees and pollinators for environmental resources – like pollen and nectar. Increasingly, particularly in agricultural areas, these can be in limiting supply at certain times of the season. For examples, because all the hedgerows have been grubbed up and the wildflower meadows obliterated.

In addition, there are a number of studies that have suggested that honey bee viruses have spilled over into other pollinators, in particular bumble bees, and that this pathogen spillover has contributed to the decline in free-living bee populations.

Do honey bee viruses have the motive, opportunity and means to cause disease in bumble bees?

Are honey bee viruses responsible for the decline in bumble bee populations?

Correlation and causation

There are numerous studies showing that the most widespread honey bee virus, deformed wing virus (DWV), can be detected in wild-caught bumble bees.

Let me pose a quick question … does detection mean ‘replication’?

Deformed wing virus “does what it says on the tin” in honey bees. When transmitted by Varroa it causes developmental defects in pupae that appear as wing deformities in newly emerged workers.

DWV symptoms

DWV symptoms

There’s also one study that implicates DWV in directly causing disease in bumble bees.

This influential paper, published 15 years ago, demonstrated that some bumble bees had the characteristic crippled wings seen in symptomatic emerging honey bee workers. The ‘smoking gun’ was that DWV was also detected in these bumble bees.

Exhibit A : Evidence for DWV infection in bumble bees – click image for full legend.

This is the only figure in the paper and there have been no substantive follow-up papers. For some reason they showed ‘symptomatic’ Bombus terrestris (the buff-tailed bumble bee; panel A, left) and PCR detection data for B. pascorum (the common carder bee).

Nevertheless, this association between presence and symptoms was sufficient for the authors to conclude “we demonstrated that DWV is pathogenic to at least two bumble bee species … causing wing deformity similar to clinically DWV-infected honey bees”.

Here’s a second important question … does the detection of DWV in bumble bees demonstrate it is responsible for the symptoms observed? 3

As a virologist interested in the evolution, replication and transmission of viruses – and a beekeeper – this seemed like a worthwhile topic to explore in a bit more detail.

There might be a correlation between the presence of DWV in bumble bees, but does this account for causation of the DWV-like symptoms seen in the bumble bees?

Motive and opportunity

Let’s get these two out of the way quickly (though we’ll return to opportunity in the notes at the end).

Remember, viruses don’t want to do anything. I’m using motive as a hideously contrived reference to whether the pathogen, DWV, is present at high and infectious levels in honey bees.

It is.

Healthy, mite-naive (but reared in a hive with Varroa) workers can carry as little as ~1000 DWV viruses. Similar levels of DWV are also present in hives from Varroa-free regions – at least of the UK 4. In contrast, after parasitisation by Varroa, symptomatic or asymptomatic adult worker honey bees regularly have more that 109 (one trillion) viruses coursing through their little bodies. 

This virus is highly infectious. When injected into honey bees, as few as 10 viruses are sufficient to start a new infection and are replicated several million-fold within 24-48 hours.

Let’s agree that they have the ‘motive’ … to spread to other hosts.

They also have the opportunity, at least in the broadest sense of the word. Honey bees and bumble bees share the same environment. They collect nectar and pollen from the same flower species. They can even regularly be seen visiting the same flower simultaneously. In addition, it’s not unusual to see bumble bees trying to access honey bee hives to steal nectar.

I’d argue that the virus appears to have ample opportunity to move from one species to the other.

Does DWV replicate in bumble bees?

This is an important question. The data figure (‘Exhibit A’) shown above does not answer this question. Their assay simply detects the presence of DWV, with no indication of whether the virus is replicating.

And the reason this is important is really explained in the section on motive and opportunity above. DWV is ubiquitous and present at extraordinarily high levels in many honey bees. It’s present in honey bee faeces. It can be detected on flowers that honey bees have visited, or in pollen collected from those flowers.

DWV is absolutely everywhere.

So, if it is everywhere, there’s a good chance it might simply contaminate things … like bumble bees that look a bit sick for other reasons.

However, if DWV is involved in causing disease in bumble bees then the virus must replicate in bumble bees.

Virus replication – select for full size and legend.

I’ve used this figure before. To be certain that the virus is replicating you need to identify the intermediate replication products (the negative strand RNA shown as red arrows above).

Almost none of the large number of papers that have reported “DWV infected bumble bees” have identified these intermediate replication products.

Many studies didn’t even bother looking for these critical intermediate products.

So we did

There are two ways we could have investigated this. We could have collected large numbers of bumble bees from the fields and screened them in the lab for these replication intermediates. The problem with this approach is that DWV might be very rare in bumble bees, or might be common, but only replicate rarely. The additional problem is that it involves going out and collecting bees from the environment – that’s hard work 😉 5

An alternative approach is to buy a nest of bumble bees 6 and to inject a few bumble bee pupae and adults with DWV. This is what we did. In parallel, to ‘prove’ the virus used can replicate in honey bees, we injected a few honey bee pupae.

We injected the bees as it’s the definitive way we have of being sure that the virus was present.

‘Gene jockeys’

I’ve got some gifted molecular virologists in my lab. These are scientists who use genetic engineering or biotechnology techniques to address tricky questions (gene jockeys). They are particularly skilled at manipulating nucleic acids, such as the genomes of viruses.

If DWV is so common (it is – see above) how would we know that the replication intermediates were from the virus we injected into the bumble bees, rather than from a virus that was possibly already present?

To solve this puzzle Alex (the lead author on our recent paper) did two things. She engineered a unique genetic marker into the virus which we could look for (and that knew was absent from other similar viruses that might already be present in bumble bees). In addition, she ensured that the virus she injected into the bumble bees contained none of the negative strand RNA that is produced as the intermediate when the virus replicates. 

And … to cut a long story short, we could detect the negative strand RNA replication intermediate in injected bumble bees. In addition, the virus contained the unique genetic marker Alex had engineered into the virus genome, so we were absolutely certain it was the injected virus that was replicating. All of the controls worked exactly as expected.

DWV does replicate in bumble bees … at least under the conditions used in our experiment.

But it does not replicate very fast

The bumble bees we used for these experiments are commercially produced under very clean conditions. When we tested control bumble bees they contained no DWV at all, even using our most sensitive assays.

In contrast to injected honey bee pupae, DWV replicates rather slowly in bumble bees. In honey bees it will amplify a million-fold in 48 hours. However, in bumble bees, in 48 hours we only observed a 100-10,000 fold increase in DWV levels. It was only when we injected large amounts of DWV to bumble bees that we could recapitulate the virus levels seen in symptomatic honey bees.

This was a little puzzling as we’ve assumed that the very rapid replication of DWV in honey bees contributes to pathogenesis. The virus needs to replicate fast to spread through the pupa and infect the particular tissues and organs that, when damaged, result in the characteristic symptoms seen.

If it only replicates slowly in bumble bees how can it cause symptoms?

But hold on … does it cause symptoms?

Not as far as we can tell. 

We never found any injected bumble bees with deformed wings, or any that looked anything like the symptoms seen with DWV in honey bees. The actual quote from the paper is:

Strikingly, none of the eclosed bumble bees showed any signs of the wing deformities that are characteristic of DWV infection of honey bees”. 

Morbidity of DWV in bumble bees

Injected pupae developed until eclosion and were either non-viable, discoloured or apparently normal in appearance (see (c) above). We observed similar numbers of these three types whether the pupae were injected with DWV or mock-injected with buffer alone (see (b) above). What’s more, of those injected with the virus, the level of the virus was the same irrespective of the appearance (or viability) of the bumble bee (see (a) above).

We know that these bumble bees contain replicating DWV but see no evidence for overt disease. I acknowledge that they may have invisible symptoms. However, we see no evidence for the wing deformities reported in the 2005 paper from Genersch and colleagues (‘Exhibit A’, above).

Heavy going? But I’ve only just started … 

So, let’s briefly return to our “motive, opportunity, means” crime analogy and summarise where we’ve got to so far. 

DWV is present at very high levels in at least some honey bees. In addition, bumble bees are likely to regularly come into contact with DWV in the environment. This could happen through contaminated pollen, when attempting to rob hives, or by direct interaction with honey bees when both visit flowers. Finally, and most tellingly, DWV replicates in bumble bees.

So, if I was Peter Falk as Lieutenant Columbo, I’d argue that DWV has the motive, opportunity and means to potentially cause disease in bumble bees.

But, as is apparent from the figure above, it appears not to actually cause disease.

Which is puzzling. 

Just one more thing 7

There’s a major problem with the experiments I’ve discussed so far.

Like all experiments 8 they were tightly controlled, with single variables and lots of statistical analysis to demonstrate our confidence in their reproducibility.

The problem wasn’t technical, it was how well they recapitulated potential transmission in the environment.

We’re not aware of anything that goes around “injecting” bumble bee pupae or adults 9. They are not parasitised by Varroa and, although there are bumble bee mites (such as Parasitellus), they don’t feed on bumble bees in the same way that Varroa feeds on honey bees.

How else might bumble bees acquire DWV?

The obvious route is orally, while feeding. 

There are a few issues with feeding bees DWV. The method is simple enough … you add the virus to sugar syrup and they slurp it down. However, you have little control over how much individual bees consume. Do some bees feed directly and other acquire food when fed by other bees (trophallaxis)? It gets rather difficult control.

In addition, we knew from our studies of honey bees that they are far less susceptible to infection per os (by mouth) than by injection. And by far less I mean tens of thousands of fold less susceptible, at least as adults.

Feeding bumble bees DWV

We chose to investigate two routes of feeding.

The first was to directly feed individual bees in the laboratory. Using this approach we failed to detect any evidence for infection or DWV replication in fed adult bumble bees, even when fed 100 million DWV viruses.

Not an encouraging start. However, larvae generally have increased susceptibility to pathogens, so we also investigated feeding larvae in the laboratory. In these studies we did manage to establish infection. However, to do so we had to add 100 million DWV viruses to he food and only achieved a 50% infection rate. 

The second approach we used was direct feeding of complete bumble bee colonies maintained in the lab.

Bumble bees are easier to keep in the lab than honey bees as they don’t need to be free-flying. Each colony occupies a 30cm3 perforated plastic box, supplied with pollen and syrup. We fed three colonies 100 million DWV per bee per day for 4-6 weeks 10.

That’s a huge amount of virus for a protracted period. Our reasoning here was straightforward. Perhaps there was a particular developmental stage that had increased susceptibility? Perhaps adult bees feeding larvae would – for whatever reason – reduce their resistance to infection via the oral route?

We screened every egg, larva, pupa and adult bee for replicating DWV at the end of the experiment.

There was none present 🙁 ( or perhaps 🙂 , depending upon your viewpoint )

Summary and conclusions

We generated unequivocal evidence that DWV replicates in bumble bees. Specifically in Bombus terrestris, the buff tailed bumble bee. I’d be surprised if it did not replicate in other Bombus species, but that will need to be investigated. 

Infection of adult bees was only possible by injection, with no evidence for infection during feeding.

Bumble bee larvae can be infected with DWV while feeding, but only when fed very large amounts of virus directly. When larvae were reared by a colony supplied with DWV-laced food for several weeks the larvae did not become infected.

These results were recently published in Scientific Reports (Gusachenko et al., 2020) and are freely available. The title of the paper neatly sums up the study “Evidence for and against deformed wing virus spillover from honey bees to bumble bees: a reverse genetic analysis”

What does this mean in terms of our understanding of pathogen spillover from managed honey bee colonies to free living bees? 

If ecologists and environmental scientists are going to make the case that honey bees are threatening the survival of free-living solitary or bumble bees (due to pathogen spillover) they need to:

  1. formally demonstrate that the honey bee virus replicates in the free-living bee
  2. show that this replication is detrimental to the free-living bee
  3. provide evidence for a natural route of transmission by which infection can occur

In my view, and using legal terms again, it’s case proven for the first point in Bombus terrestris 11. In contrast, if I was the judge I’d throw out the other cases due to lack of evidence.

Monkey puzzles

There are viruses everywhere. Every living species has one or more viruses that infect it. Inevitably, because viruses replicate to very high levels, species other than the natural host may become exposed.

But almost always this has no consequences at all …

And to illustrate that I’ll briefly describe a study of monkey viruses infecting humans in Cameroon from several years ago. HIV is one of a very large group of viruses called immunodeficiency viruses. The ‘H’ stands for human, though the virus originated in chimpanzees and is very closely related to the simian immunodeficiency viruses (SIV).

A study of hunters living in the forests of Cameroon showed evidence for exposure to multiple different SIV isolates in tribespeople who hunted non-human primates or who were involved in butchering them or preparing them for market or cooking. There was no evidence these viruses were spreading in the human population, or that there was any sickness or disease associated with prior exposure. 

Environmental exposure happens all the time, and is relatively easy to detect. That’s exactly what had happened with these monkey viruses.

But evidence for environmental exposure, whilst easy to get, is not sufficient to support a claim that the virus causes disease at the level of the individual, or that it threatens an entire population.

As a virologist I think it’s interesting that DWV replicates in Bombus terrestris. However, I’ve yet to see any convincing evidence that DWV spillover from honey bees is responsible for the decline in wild bee populations.

Circumstantial evidence is not the same as convincing evidence.


What are the missing experiments that we didn’t do?

A key one is to determine whether long-term infection of bumble bees with DWV results in disease. We didn’t see overt deformed wing disease, but it’s possible that infection for weeks could have caused this or other symptoms. 

Although we fed bumble bee nests for weeks with DWV we saw no evidence of larval infection. This was despite previously demonstrating that larvae could be infected with high levels of DWV orally. One possibility is that DWV is inactivated by adult bees. I think it would be interesting to look at the level of infectious DWV in larval food.

Are there natural routes of exposure to DWV that result in bumble bee infection?

Does DWV ever cause environmental contamination at a high enough level to naturally infect bumble bees? For example, is the level of DWV in honey bee faeces high enough and does it ever contaminate pollen?

We are not going to do these studies so it will be interesting to see if others do … or whether simply the presence of the virus (whether replicating or not) will be proof’ that honey bee viruses are responsible for the decline in free-living bee populations.

Diutinus bees

Diutinus is Latin for long-lasting.

Diutinus bees are therefore long-lasting bees. These are the bees that, in temperate regions, maintain the colony through the winter to the warmer days of spring.

I’ve discussed the importance of these bees recently., and I’ve regularly made the case that protecting these ‘long-lived’ bees from the ravages of Varroa-vectored viruses is critical to reduce overwintering colony losses.

Winter is coming …

In most cases the adjective diutinus is replaced with ‘winter’, as in winter bees; it’s a more familiar term and emphasises the time of year these bees are present in the hive. I’ll generally use the terms interchangeably in this post.

Diutinus does not mean winter

From a scientific standpoint, the key feature of these bees is that they can live for up to 8 months, in contrast to the ~30 days a worker bee lives in spring or summer. If you are interested in what induces the production of long-lived bees and the fate of these bees, then the important feature is their longevity … not the season.

Furthermore, a proper understanding of the environmental triggers that induce the production of long-lived bees might mean they could be produced at other times of the season … a point with no obvious practical beekeeping relevance, but one we’ll return to in passing.

It’s worth emphasising that diutinus bees are genetically similar to the spring/summer bees (which for convenience I’ll refer to as ‘summer bees’ for the rest of the post). Despite this similarity, they have unique physiological features that contribute to their ability to thermoregulate the winter cluster for months and to facilitate spring build-up as the season transitions to spring.

What induces the production of winter bees? Is it a single environmental trigger, or a combination of factors? Does summer bee production stop and winter bee production start? What happens at the end of the winter to the winter bees?

Segueing into winter bee production 

The graph below shows the numbers of bees of a particular age present in the hive between the end of August and early December.

Colony age structure from August to December – see text for details

Each distinct colour represents bees reared in a particular 12 day ‘window’. All bees present before the 31st of August are blue. The next 12 day cohort of bees are yellow etc. The area occupied by each colour indicates the number of bees of a particular age cohort.

Note that egg laying (black) is negligible between early October and late November when it restarts.

The graph shows that that there is no abrupt change from production of summer bees to production of winter bees.

For example, about 95% of the blue bees have disappeared by December 1. Of the yellow bees, which first appeared in mid-September, about 33% are present in December. Finally, the majority of the lime coloured bees, that first put in an appearance in early October, are present at the end of December.

The colony does not abruptly stop producing short-lived summer bees on a particular date and switch to generating long-lived ‘diutinus’ winter bees. Instead, as late summer segues into early autumn, fewer short lived bees and more long lived bees are produced. 

Note that each cohort emerge from eggs laid 24 days earlier. The orange cohort emerging from 24/09 to 05/10 were laid within the first two weeks of September. This emphasises the need to treat early to reduce mite levels sufficiently to protect the winter bees.

Winter bees are like nurse bees but different

Before we consider what triggers the production of diutinus bees we need to discuss how they differ from summer bees, both nurses and foragers.

Other than being long-lived what are their characteristics?

Interaction of key physiological factors in nurse (green), forager (red) and winter bees (blue). Colored disks indicate the relative abundance of each factor.

The four key physiological factors to be considered are the levels of juvenile hormone (JH), vitellogenin (Vg) and hemolymph proteins and the size of the hypopharyngeal gland (HPG).

As summer nurse bees transition to foragers the levels of JH increases and Vg decreases. This forms a negative feedback loop; as Vg levels decrease, JH levels increase. Nurse bees have high levels of hemolymph proteins and large HPG, the latter is involved in the production of brood food fed to larvae.

So if that describes the summer nurse bees and foragers, what about the winter bees?

Winter bees resemble nurse bees in having low JH levels, high levels of VG and hemolymph proteins and large HPG’s. 

Winter bees differ from nurse bees in being long lived. A nurse bee will mature into a forager after ~3 weeks. A winter bee will stay in a physiologically similar state for months.

There have also been time course studies of JH and Vg levels through the winter. In these, JH levels decrease rapidly through October and November and are at a minimum in mid-January, before rising steeply in February and March.

As JH levels rise, levels of Vg and hemolymph proteins decrease and the size of the HPG decreases i.e. as winter changes to early spring winter bees transition to foragers.

Now we know what to look for (JH, Vg levels etc) we can return to think about the environmental triggers that cause these changes.

No single trigger

In temperate regions what distinguishes winter from autumn or spring? 

Temperatures are lower in winter.

Daylength (photoperiod) is shorter in winter.

There is less pollen and nectar (forage) available in winter.

Under experimental conditions it’s quite difficult to change one of these variables without altering others. For example, shifting a colony to a cold room (i.e. lowering the ambient temperature to <10°C) leads to a rapid decrease in JH levels (more winter bee-like). However, the cold room was dark, so perhaps it was daylength that induced the change? Alternatively, a secondary consequence of moving the colony is that external forage was no longer available, which could account for the changes observed.

And forage availability will, inevitably, influence brood rearing.


Reducing photoperiod alone does induce some winter bee-like characteristics, such as increases in the protein and lipid content of the fat bodies. It also increases resistance to cold and starvation. It can even cause clustering at elevated (~19°C) temperatures. However, critically, a reduced photoperiod alone does not appear to make the bees long lived. 

Remember also that a reduced photoperiod will limit foraging, so reducing the nutritional status of the colony. This is not insignificant; pollen trapping 2 in the autumn accelerates the production of winter bees.

But again, this may be an indirect effect. Reduced pollen input will lead to a reduction in brood rearing. Feeding pollen to broodless winter colonies induces egg-laying by the queen.

Brood, brood pheromones and ethyl oleate

One of the strongest clues about what factor(s) induces winter bee production comes from studies of free-flying summer colonies from which the brood is removed. In these, the workers rapidly change to physiologically resemble winter bees 3.

How does the presence of brood prevent the generation of diutinus bees?

There are some studies which demonstrate that the micro-climate generated in the colony by the presence of brood – elevated temperature (35°C) and 1.5% CO2 – can influence JH levels. 

However, brood also produces pheromones – imaginately termed brood pheromone – which does all sorts of things in the colony. I’ve discussed brood pheromone previously in the context of laying workers. The brood pheromone inhibits egg laying by worker bees.

Brood pheromone also contributes to a enhancement loop; it induces foraging which results in increased brood rearing and, consequently, the production of more brood pheromone.

One way brood pheromone induces foraging is by speeding the maturation of middle-aged hive bees into foragers. Conversely, when raised in the absence of brood, bees have higher Vg levels, start foraging later and live longer.

But it’s not only brood that produces pheromones.

Workers also produce ethyl oleate, a pheromone that slows the maturation of nurse bees, so reducing their transition to foragers.

A picture is worth a thousand words

All of the above is quite complicated.

Individual factors, both environmental and in the hive, have direct and indirect effects. Experimentally it is difficult to disentangle these. However, Christina Grozinger and colleagues have produced a model which encapsulates much of the above and suggests how the production of winter bees is regulated. 

Proposed model for regulation of production of winter bees.

During autumn there is a reduction in forage available coupled with a reduced daylength and lower environmental temperatures. Consequently, there is less foraging by the colony. 

Since more foragers are present within the hive, the nurse bees are exposed to higher levels of ethyl oleate, so slowing their maturation.

There’s less pollen being brought into the colony (reduced nutrition), so brood production decreases and so does the level of brood pheromone. The reduced levels of brood pheromone also reduces nurse bee maturation.

As shown in the diagram, all of these events are in a feedback loop. The reduction in levels of brood pheromone further reduces the level of foraging … meaning more foragers are ‘at home’, so increasing the effects of ethyl oleate.

All of these events have the effect of retarding worker bee maturation. The workers remain as ‘nurse-like’ long-lived winter bees.

Is that all?

The difference between nurse bees and winter bees is their longevity … or is it?

In the description above, and in most of the experiments conducted to date, the key markers of the levels of JH, Vg and hemolymph proteins, and the size of the HPG, are what has been studied. 

I’d be astounded if there are not many additional changes. 

Comparison of workers and queen bees have shown a large range of epigenetic changes induced by the differences in the diet of young larvae 4. Epigenetic changes are modifications to the genetic material that change gene expression.

I would not be surprised if there were epigenetic changes in winter bees, perhaps induced by alteration of the protein content of their diet as larvae, that influence gene expression and subsequent longevity. Two recent papers suggest that this may indeed happen; the expression of the DNA methyltransferases (the enzymes that cause the epigenetic modifications) differs depending upon the demography of the colony 5 and there are epigenetic changes between the HPG in winter bees and bees in spring 6.

Clearly there is a lot more work required to fully understand the characteristics of winter bees and how they are determined.

Don’t forget …

It’s worth emphasising that the local environment (forage and weather in particular) and the strain of the bees 7 will have an influence on the timing of winter bee production.

Last week I discussed a colony in my bee shed that had very little brood on the 2nd of October (less than one side of one frame). When I checked the colonies last weekend (11th) there were almost no bees flying and no pollen coming in. A colleague checked an adjacent colony on Monday (13th) and reported it was completely broodless. These bees are ‘local mongrels’, selected over several years to suit my beekeeping.

Early autumn colonies

In contrast, my colonies on the west coast are still busy. These are native black bees. On the 14th they were still collecting pollen and were still rearing brood. 

The calendar dates in the second figure (above) are therefore largely irrelevant.

The transition from summer bees to the diutinus winter bees will be happening in your colonies, sooner or later. I suspect it’s already completed in my Fife bees.

Whether genetics or environment has a greater influence on winter bee production remains to be determined. However, I have previously described the good evidence that local bees are better adapted to overwintering colony survival.

To me, this suggests that the two are inextricably linked; locally selected bees are better able to exploit the environment in a timely manner to ensure the colony has the winter bees needed to get the colony through to spring.


Winter bee production

There are big changes going on in your colonies at the moment.

The summer foragers that have been working tirelessly over the last few weeks are slowly but surely being replaced. As they die off – whether from old age or by being eaten by the last of the migrating swallows – they are being replaced by the winter bees.

Between August and late November almost the entire population of bees will have changed. The strong colonies you have now (or should have) will contain a totally different workforce by the end of the year.

Forever young

The winter bees are the ones responsible for getting the colony from mid/late autumn through to the following spring. They are sometimes termed diutinus bees from the Latin for “long lived”.

These are the bees that thermoregulate the winter cluster, protecting the queen, and rearing the small amounts of brood during the cold, dark winter to keep the colony ticking over.

Midwinter cluster

A midwinter colony

Physiologically they share some striking similarities with so-called hive or nurse bees 1 early in the summer.

Both hive bees and winter bees have low levels of juvenile hormone (JH) and active hypopharyngeal glands. Both types of bee also have high levels of vitellogenin, high oxidative stress resistance and corpulent little bodies.

But early summer nurse bees mature over a 2-3 week period. Their JH levels increase and vitellogenin levels decrease. This induces additional physiological changes which results in the nurse bee changing into a forager. They sally forth, collecting nectar, pollen and water …

And about three weeks later they’re worn out and die.

Live fast, die young

And this is where winter bees differ. They don’t age.

Or, more accurately, they age   v  e  r  y    s  l  o  w  l  y.

In the hive, winter bees can live for 6 months if needed. Under laboratory conditions they have been recorded as living for up to 9 months.

They effectively stay, as Bob Dylan mumbled, forever young.

Why are winter bees important?

Although not quite eternal youth … staying forever young is useful as their longevity ensures that the colony does not dwindle and perish in the middle of winter.

With little or no nectar or pollen available in the environment the colony reduces brood rearing, and often stops altogether for a period.

But what about the kilograms of stores and cells filled with pollen in the hive? Why can’t they use that?

Whilst both are present, there’s nothing like enough to maintain the usual rate of brood rearing. If they tried the colony would very quickly starve.

Evolution has a very effective way of selecting against such rash behaviour 🙁

If you doubt this, think how quickly hives get dangerously light during the June gap. With no nectar coming in and thousands of hungry (larval) mouths to feed the colony can easily starve to death during a fortnight of poor weather in June.

The winter bees ‘hold the fort’, protecting the queen and rearing small amounts of brood until the days lengthen and the early season pollen and nectars become available again.

And, just as the winter bees look after the viability of the colony, the beekeeper in turn needs to look after the winter bees … we rely on them to get the colony through to spring.

Lots of bees

Can you identify the winter bees?

But before we discuss that, how do you identify and count the winter bees? How can you tell they are present? After all, as the picture above shows 2, all bees look rather similar …

Counting the long lived winter bees

The physiological changes in winter bees, such as the JH and vitellogenin levels, are only identifiable once you’ve done some rather devastating things to the bee. These have the unfortunate side effect of preventing it completing any further bee-type activities 🙁

Even before you subject them to that, their fat little bodies aren’t really sufficiently different to identify them visually.

But what is different is their longevity.

By definition, the diutinus or winter bees are long lived.

Therefore, if you record the date when the bee emerged you can effectively count back and determine how old it is. If it is more than ~6 weeks old then it’s a winter bee.

Or the queen 😉

And, it should be obvious, if you extrapolate back to the time the first long lived bees appear in the hive you will have determined when the colony starts rearing winter bees.

The obvious way to determine the age of a bee is to mark it upon emergence and keep a record of which marks were used when. Some scientists use numbered dots stuck to the thorax, some use combinations of Humbrol-type paint colours.

I’m not aware that anyone has yet used the barcoding system I discussed recently, though it could be used. The winter bee studies I’m aware of pre-date this type of technology.

Actually, some of these studies date back almost 50 years, though the resulting papers were published much more recently.

This is painstaking and mind-numbingly repetitive work and science owes a debt of gratitude to Floyd Harris who conducted many of the studies.

Colony age structure – autumn to winter

Here is some data showing the age structure of a colony transitioning from late summer into autumn and winter. There’s a lot in this graph so bear with me …

Colony age structure from August to December - see text for details

Colony age structure from August to December – see text for details

The graph shows the numbers and ages of bees in the colony.

The ages of the bees is indicated on the vertical axis – with eggs and brood (the youngest) at the bottom, coloured black and brown respectively. The adult bees can be aged between 1 and ~100 days old 3. The number of bees is indicated by the width of the coloured bar at each of the nine 12 day intervals shown.

All of the adult bees present in the hive at the end of August are coloured blue, irrespective of their age. There are a lot of these bees at the end of August and almost all of them have disappeared (died) by mid-November 4.

The remaining colours indicate all the bee that emerge within a particular 12 day interval. For example, all the bees that emerge between the 31st of August and the 12th of September are coloured yellow.  Going by the width (i.e. the numbers of bees of that age) of the yellow bars it’s clear that half to two-thirds of these bees die by mid October, with the rest just getting older gracefully.

But look at the cohort that emerge between the end of September and early October, coloured like this 5. The number of these bees barely changes between emergence and early December. By this time they are 72 days old i.e. an age that most summer bees never achieve.

Brood breaks and climate

In the colony shown above the queen continued laying reduced numbers of eggs – the black bars – until mid-October and then didn’t start again until the end of November. During this period the average age of the bees in the colony increased from ~36 days to ~72 days and the strength of the colony barely changed.

The figure above comes from a BeeCulture article by Floyd Harris. The original data isn’t directly referenced, but I suspect it comes from studies Harris conducted in the late 70’s in Manitoba, some of which was subsequently published in the Journal of Apicultural Research. In addition, Harris co-authored a paper presenting similar data in a different format in Insectes Sociaux which describes the Manitoban climate as having moderate/hot summers and long, cold winters.

My hives in Scotland, or your hives in Devon, or Denmark or wherever, will experience a different climate 6.

However, if you live in a temperate region the overall pattern will be similar. The summer bees will be replaced during the early autumn by a completely new population of winter bees. These maintain the colony through to the following spring.

The dates will be different and the speed of the transition from one population to the other may differ. The timing of the onset of a brood break is likely to also differ.

However, the population changes will be broadly similar.

And, it should be noted, the dates may differ slightly in Manitoba (and everywhere else) from year to year, depending upon temperature and forage availability.

Colony size and overwintering survival

Regular readers might be thinking back to a couple of posts on colony size and overwintering survival from last year.

One measured colony weight, showing that heavier colonies overwintered better 7. A second discussed the better performance of local bees in a Europe-wide study of overwintering survival. In this, I quoted a key sentence from the discussion:

“colonies of local origin had significantly higher numbers of bees than colonies placed outside their area of origin”

I can’t remember when during the season those studies recorded colony size, but I’m well aware that large colonies in the winter survive better.

The colonies that perish first in the winter are the pathetic grapefruit-sized 8 colonies with ageing queens or high pathogen loads.

In contrast, the medicine ball-sized ‘boomers’ go on and on, emerging from the winter strongly and building up rapidly to exploit the early season nectar.

But what the graph above shows is that the bees in a strong colony in late summer are a completely different population from the bees in the colony in midwinter.

The strength of the midwinter colony is determined entirely by when winter bee rearing starts and the laying rate of the queen, although of course both may be indirectly influenced by summer colony strength.

The influence of the queen

Other than this potential indirect influence, it’s possibly irrelevant how large the summer colony is in terms of winter colony size (and hence survival).

After all, even if the summer bees were three times as numerous, their fate is sealed. They are all going to perish six weeks or so after emergence.

Are there ways that beekeepers can influence the size of the overwinter colony to increase its chances of survival?

I wouldn’t pose the question if the answer wasn’t a resounding yes.

It has been known for a long time 9 that older queens stop laying earlier in the autumn than younger queens. As explained above, the longer the queen lays into the autumn the more winter bees are going to be produced.

Mattila et al., 10 looked at the consequences of late season (post summer honey harvest) requeening of colonies. In these they removed the old queen and replaced her with either a new mated or virgin queen, or allowed the colony to requeen naturally.

Using the ’12 day cohort’ populations explained above, the authors looked at when the majority of the winter bees were produced in the colony, and estimated the overall size of the winter colony.

The influence of new queens on winter bee production.

The influence of new queens on winter bee production. Note shift to the right in B, C and D, with new queens.

With the original old queen, 53% of winter bees were produced in the first two cohorts of winter bees. With the requeened colonies 54-64% of the winter bees were produced on average 36 days later, in the third and fourth cohorts of winter bees.

This indicates that young queens produce winter bees later into the autumn.

This is a good thing™.

In addition, though the results were not statistically significant, there was a trend for colonies headed by new queens to have a larger population of bees overwinter.

Perhaps one reason the requeened colonies weren’t significantly larger was that the new queens delay the onset of winter bee rearing. I’ll return to this at the end.

The influence of deformed wing virus (DWV)

Regular readers will know that this topic has been covered extensively, and possibly exhaustively, elsewhere on this site … so I’ll cut to the chase.

DWV is the most important virus of honey bees. When transmitted by Varroa destructor there is unequivocal evidence that it is associated with overwintering colony losses. The reason DWV causes overwintering losses is that it reduces the longevity of the winter bees.

The virus might also reduce the longevity of summer bees but,

  1. there’s so many of them to start with
  2. there’s loads more emerging every day, and
  3. they only survive a few weeks anyway,

that this is probably irrelevant in terms of colony survival.

Dainat et al., (2012) produced compelling evidence showing that DWV reduces the longevity of winter bees 11. The lifespan was reduced by ~20%.

A consequence of this is that the winter bees die off a little faster and the colony shrinks a little more. At some point it crosses a threshold below which it cannot thermoregulate the cluster properly, further limiting the ability of the colony to rear replacement bees (assuming the queen is able to lay at a low rate).

This colony is doomed.

Even if they stagger through to the longer days of spring they contain too few bees to build up fast. They’re not dead … but they’re hardly flourishing.

Winter bees and practical beekeeping

I think there are three ways in which our understanding of the timing of winter bee production should influence practical beekeeping:

Firstly … The obvious take-home message is that winter bees must be protected from the ravages of DWV. The only way to do this is to minimise the mite population in the colony before the winter bee rearing starts.

The logical way to do this is to treat using an approved miticide as soon as practical after the summer honey is removed 12.

I discuss the importance of the timing of this treatment in When to treat?, which remains one of the most-read posts on this site.

Secondly … Avoid use of miticides (or other colony manipulations) that reduce the laying rate of the queen in early autumn.

When I used to live at lower latitudes I would sometimes use Apiguard. This thymol-containing miticide is very effective if used when the temperature is high enough. However, in my experience a significant proportion of queens stop laying when it is being used. Not all, but certainly more than 50%.

I don’t know why some stop and others don’t. Is it genetic? Temperature-dependent?

Whatever the reason, they stop at exactly the time of the season you want them to be laying strongly.

Thirdly … consider requeening colonies with young queens after the summer honey is removed. This delays the onset of winter bee production and results in the new queen laying later into the year. The later start to winter bee production gives more time for miticides to work.

A win-win situation.


The new normal

For many, beekeeping associations provide the bookends that bracket the practical beekeeping season. In meetings during the dark, wet, cold winter months we can at least discuss bees, reminisce about the season just gone or plan for the season ahead.

Usually with tea and biscuits 🙂

Or in the more civilised associations (and a quick plug here for the Fortingall & District BKA) with fantastic homemade cakes 😉

Elderflower lemon drizzle cake

Beekeeping associations, through the training and social events that they organise and the contacts that they enable, provide an important support framework for beekeepers, both new and old.

Training new beekeepers is one important function associations provide, but more experienced beekeepers also benefit from co-operative purchasing schemes for foundation or fondant 1 and – of course – from the winter seminar programmes.

Double whammy

The Covid-19 pandemic has dealt a double whammy to many associations.

Training events, necessitating flagrant breaches of social distancing during hands-on practical beekeeping demonstrations, are a problem. Many associations delivered the theoretical coursework before lockdown was imposed, but were subsequently unable to provide the practical component of the training for beginners.

It’s difficult to spot the queen from 50 centimetres sometimes, let alone 2 metres.

Returning a marked and clipped queen – tricky to do at arm’s length

The independent first inspections for 2020 beginners are likely to have been a pretty tough challenge for many. Congratulations to those who got through them and the rest of the season with little support.

I’m hearing that some associations have cancelled or postponed all training events for the ’20/’21 winter season.

The imposition of lockdown 2 in March probably had little impact on the ’19/’20 winter seminar programmes, but they’re likely to have a significant impact going forwards.

I give quite a few talks on science and practical beekeeping in most winters. Audiences and venues vary, depending on the association. I’ve talked in drafty church halls to groups of 15, or swanky conference centres to ten times that number.

There is always a good turnout by new beekeepers, or even by those who have yet to start keeping bees.

Not your typical beekeeping audience … or church hall

However, there is generally a gender imbalance, with more men than women attending. And – and I’m afraid there’s no gentler way to write this – there’s an age imbalance as well, with the enthusiastic young ‘uns outnumbered by older, and in some cases old, beekeepers.


This age and gender imbalance inevitably make the ’20/’21 winter seminar programmes an endangered species, at least in the format we’ve grown used to over past seasons.

If you look at the statistics for serious Covid-19 cases it is clear that there is a strong bias towards elderly males. There are other biases as well … underlying medical complications and ethnicity also have a major influence, though whether the latter is socio-economic, genetic or due to the presence of comorbidities remains unclear.

All of which means that spending an hour in a drafty church hall listening to a talk on bait hives is probably unwise … not least because the social distancing needed precludes any chance to huddle together for warmth when the one bar electric heater blows a fuse.

Zoom …

In the brave new, socially distanced, world we’re currently inhabiting, drafty church halls and excellent homemade cakes are now just a distant memory. 

Instead we have Teams talks, Demio demonstrations and WebEx webinars. 

And Zoom, but I can’t think of a suitable alliteration to go with Zoom 🙁

For many office-based workers, lockdown resulted in the substitution of boardroom meetings with spare bedroom virtual meetings. 

Hastily repurposed guest bedrooms have become home offices. The combination of IKEA furniture, a reasonably recent laptop and a fast internet connection has enabled ‘business as usual’.


All of my meetings – with administrators, colleagues, my research team and students – have been online since late March (and in certain cases since early March).

Academics are used to collaborating globally and so were already familiar with Zoom, Teams or Skype for conference calls and job interviews. These have just continued 3, and been extended to now include all the in-person meetings that used to happen.

One or two colleagues have embraced this expanded use of the technology to have their own ‘green screens’. This allows their head and upper torso to be projected in front of a selected image – of a tropical beach, their favourite golf course or local boozer.

The Maldives … the perfect backdrop for a dull committee meeting

The really professional ones even change out of their pyjamas before calls … 😉

But many beekeepers will be largely unfamiliar with the technology and the advantages it offers … and disadvantages it imposes.

Online beekeeping talks

I’ve both attended and delivered online beekeeping talks. Not a huge number, but enough to have a fair idea of what works and what doesn’t. In addition, I’ve taken part – as audience or presenter – in hundreds of non-beekeeping online events.

For readers who have yet to take part here’s a general guide of what to expect.

The speaker and topic are advertised in advance and those interested in listening/watching register to attend. The talk is hosted by the beekeeping association who provide a ‘chairperson’ or ‘master of ceremonies’. This person has the unenviable task of dealing with the speaker, the technology and the audience. 

And two of these three might do something incomprehensibly stupid … and the internet can break.

On the evening of the talk 4 you login via a website (using a username/password provided on registration) and launch the necessary software to take part in the event.

Eventbrite beekeeping talk

Sometimes this can be through the web browser, but – more usually – it involves downloading and installing software onto your computer. Which might be an issue for some people wanting to take part. Do this in advance of the start time of the talk, not in the last 2 minutes before kickoff.

After an introduction by the chairperson, control of the graphics is usually handed to the speaker who delivers the talk. To avoid awkward ‘noises off’ 5 the chairperson usually mutes all other microphones

Why unenviable?

I previously described the chairpersons role as unenviable.

While the speaker blathers away the chair is probably:

  • dealing with email enquiries about how to launch the software
  • justifying why there isn’t a video of the speaker actually speaking (it’s turned off to save bandwidth), and
  • telling someone that they are the only person unable to hear the presenter. Therefore, it must be their audio output settings that are wrong.

And if that isn’t enough, the chair will be collecting and collating questions during and after the talk, for reasons I’ll discuss shortly.

Finally, it’s not unusual for the chair to also ensure that the talk is recorded so that those who couldn’t download the software or hear the presenter can attempt to listen to it in the future.

That’s a lot to deal with.

Questions and answers

Good talks generate questions.

As a speaker, there’s nothing worse than a talk being met by an echoing wall of silence.

Hello? Is there anybody [out] there? Just nod if you can hear me 6.

Some are points of clarification, others are after elaboration or explanation of a contrary view.

Some questions are nothing whatsoever to do with the talk 😉

They might not even be about beekeeping.

All require an answer of some kind.

And this is where the technology gets in the way of communication. 

Questions from the floor, in which the audience member switches on their microphone, clearly enunciates the question, and turns off the mike returning ‘control’ to the host and speaker cause delays.

Often significant delays. However, even short breaks interrupt good communication – think back to the lag on transatlantic satellite phone calls. 

The speaker asks the question clearly … but omits to turn on the mike.

Or they fail to the turn off the mike, so the entire audience hears the follow up “and I hope he answers quickly as Strictly’s on in a few minutes”.

For a couple of questions this is just about acceptable. For twenty or thirty it is not.

So, the beleaguered chair takes written questions from the audience, collates them, removes duplicates … and then asks the presenter on behalf of the audience.

I refer you back to the word ‘unenviable’. If you take part, cut the chair some slack …

Disadvantages of online talks

Unfortunately a subset of beekeepers who would have attended a gathering on the second Tuesday of the month in the church hall will never attend an online beekeeping talk.

For a start, they might not even own a computer. 

They might – and I have considerable sympathy for this view – mainly attend talks for the craic, the opportunity to catch up with friends and the chance of some homemade lemon drizzle cake.

All of those are good reasons to attend a talk in person … and in the case of lemon drizzle cake I’d say a compelling reason to attend 😉

As a regular speaker at associations I’d add here that the craic and the homemade cake are the parts of the evening I enjoy the most. After all, I’ve heard the talk before. I might even have heard the questions before 😉

None of these more social things are achievable online. Everyone listens in their own little bubble, isolated from the shared experience.

If they can’t bake it’s going to be a long evening 🙁

For others, the technology will continue to be a problem. They can see the pictures but can’t hear the words. Or vice versa. Or worse …

No Zoom for you …

The fact that 190 others don’t have the same problems just makes it a more frustrating and unrewarding experience. Being live, there’s no real chance of resolving these ‘local’ problems without delaying the talk and irritating the rest of the audience.

Over time the numbers unable to handle the technology will reduce.

In some cases it’ll be because they have learned to master it – either by perseverance, or by the beekeeping association providing some sort of training sessions.

In other cases it’ll be because they simply gave up 🙁

Like those who don’t have a computer in the first place, this means online talks are serving a different audience and some association members are likely to be excluded by the switch to online talks.

Advantages of online talks

But it’s not all bad news. I can see some benefits for both the speaker and audience from online presentations.

Associations can invite speakers from anywhere.

They don’t have to be from the same county.

Or the same country.

This broadens the topics that can be covered and provides the opportunity to discover different beekeeping practices from other areas (always remembering that this might simply confuse beginners).

There are some good speakers out there – just look through past programmes for the BBKA, SBA or WBKA Annual Conventions or the National Honey Show.

Associations can ‘share’ speakers by running joint events or inviting neighbouring association members to register.

As a speaker, this means that audiences tend to be larger. Bigger audiences are almost always better 7. Since everyone is logging on, rather than driving across the county to the venue, there’s less chance a spot of bad weather will put people off.

This is a huge advantage for the speaker as well … I’ve regularly talked, answered questions, drunk tea, chatted, eaten lemon drizzle cake, drunk more tea, said my goodbyes and then driven for three hours to get home 8.

The Beast from the East ...

The Beast from the East …

I’ve also had to cancel talks at relatively short notice due to ‘adverse driving conditions’ – which in Scotland means a bit more than a dusting of snow. 

None of that happens in our brave new digital world.

Is this the new normal?

For the foreseeable future I think it is. The national lockdown is being replaced by local restrictions where virus transmission is increasing. However, school and university students have yet to return and this will likely lead to increased transmission in some areas (in Scotland, we’re already seeing this, though transmission is usually ascribed to “unregulated house parties” rather than within the school 9 ).

A vaccine remains some way off. It’ll be even longer until we have vaccinated a large enough proportion of the population to interrupt transmission.

Or to know how long immunity lasts.

All of which means that indoor social events, like talks about bait hives or swarm control, are likely to be undesirable, unattractive or simply not allowed.

What can we do to improve things?

Delivering a talk online is a very much less rewarding experience than doing so in person.

There’s no ability to properly engage with the audience – no banter, no eye contact, no jokey comments.

You can’t tell whether the old boy in the back row has switched off or just nodded off.

Or perhaps he’s simply cheesed off because he disagrees with everything I’m saying.

You don’t necessarily know who is in the audience – you might know overall numbers, but not whether the local bee inspector or beefarmer is logged on. Knowledge of the audience can influence the way you pitch a talk.

It’s an oddly sterile undertaking. This makes judging the pacing and content of the talk very much more difficult. With a live audience it’s usually possible to tell whether they’re ‘keeping up’ or ‘tuning out’. You can’t do this online.

If the audience is present you can ask questions and get immediate answers … anyone who has heard me talk will know I ask about drifting and ‘how many of your bees are your bees?’. There are ways of doing this online, but it requires familiarity with more software (or additional features of the current software).

Starting materials ...

Practical demonstrations and online presentations – tricky

In the meantime …

  1. Associations can help their members embrace the technology by providing limited training where it is needed. 
  2. Think creatively about topics that can be covered within the limitations of the technology. For example, some of the talks I’ve been to in person have had a practical component. I’ve attended excellent candle dipping and skep making workshops 10. Likewise, I talk about DIY for beekeeping which involves handing round examples of my hastily cobbled together beautifully crafted floors or roofs. These sorts of things might be achievable entirely online, perhaps with more videos. However, preparing these will certainly require a lot more work by the presenter to be effective.
  3. Provide feedback to speakers – what worked and what didn’t work?

Welcome to the new normal … I hope to “see” you online sometime 🙂


The new normal means “a previously atypical or unfamiliar situation, behaviour, etc., which has become standard, usual, or expected” (OED). Although now associated with the Covid-19 pandemic, it’s usage can be traced back over 200 years. The big increase in usage was in reference to the 2008 financial crisis, and – historically – it is often used in reference to economic events. 

The new normal – Google ngram results

Orientation flights

Part of the reason for the success of honey bees is the division of labour between workers of different ages. Young workers (hive bees) clean the cells, nurse larvae and look after the queen. Older workers are foragers, collecting pollen and nectar (and water) from across the landscape.

To be successful, foragers need to know where to look and how to return.

The ‘where to look’ is partly accounted for by the well-known waggle dance 1.

In this post I’m going to discuss the second component of successful foraging – the homing ability of foragers.

More specifically, I’m going to discuss how the bee first learns about the location of the hive. 

Orientation flights

Bees do not instinctively know where the hive (or the tree they are nesting in for a wild colony) is located. They have to learn this before embarking on foraging trips to collect nectar or pollen.

This learning takes the form of one (or usually several – as we shall see) orientation flights. These enable the bee to memorise the precise location of the hive with relation to geographic landmarks. On subsequent foraging flights the bees use these landmarks to return to the hive.

Orientation flight have a characteristic appearance …

… and are very nicely described in the introduction to a paper by Capaldi and Dyer 2:

An orientation flight at the nest entrance begins as a departing bee turns and hovers back and forth, turning in short arcs, apparently looking at the hive entrance. Then, the bee increases the size of the arcs until, after a few seconds, she flies in circles while ascending to heights of 5–10 metres above the ground. This spiraling flight takes the bee out of sight of human observers. She returns a few minutes later, always without nectar or pollen.

Which I couldn’t have written any better, so have reproduced verbatim.

There are a number of features of the orientation flight that are immediately obvious from this description (which all beekeepers will recognize). These include:

  • A ‘local’ component, in the immediate vicinity of the hive
  • Wider ranging flight at a greater altitude and a longer distance
  • Direct observation does not allow the location, duration or track of these distant flights to be monitored
  • The bee returning from the orientation flight does not bring pollen or nectar with her

Do orientation flights allow orientation?

How do we know that these flights enable the bee to learn where the hive is located?

Early studies conducted by Becker (1958) showed that bees captured after a single orientation flight and then re-released up to 700 metres away from the hive could find their way ‘home’. In contrast, bees that had not gone on an orientation flight before were, by definition, disoriented and did not return to the hive.

However, the percentage that returned after undertaking a single orientation flight was related to the distance of the release point, and was never more than ~60% (at 200 metres). 

In contrast, reorienting older foragers (for example, as happens after moving a hive to a new location) were much better (~90%) at returning to the hive after a single reorientation flight. 

Capaldi and Dyer extended these early studies by Becker to investigate the impact of the visibility of local landmarks on orientation and reorientation, and also measured the speed with which bees returned after being displaced.


These studies showed that a single orientation flight allowed bees to identify the landmarks in the immediate vicinity (100 – 200 m) of the hive. When released from more distant locations, returning flights were faster and more successful (i.e. fewer lost bees) when the bees had sight of the landmarks in the vicinity of the hive.

The hive itself was effectively invisible except at very short ranges. This makes sense for a tree-nesting animal. One tree looks much the same as another 3, but if you learn that the nest is in the tree between the very tall conifer and the long straight hedgerow – two features visible from hundreds of metres distant – then orientation is straightforward.

This suggests that apiaries located near distinctive landscape features may be preferable in terms of increased returning forager rates.

“Distinctive” as far as an orienting worker bee is concerned, which may not be the same as distinctive to the beekeeper of course 😉

Reorienting bees (compared to first flight bees) took longer to explore the environment and were better at returning. Either these bees learn differently (a distinct possibility) or their prior experience in the wider landscape gives them an advantage when the hive is relocated.

Where do you go to my lovely?

The early studies by Becker and those by Dyer and colleagues defined many of the parameters that characterise orientation flights. What they did not do is show where the bees actually go during the orientation flights?

Do they just zoom around randomly?

Do they fly ever-increasing spirals?

Perhaps they perform some sort of grid search, exploring individual landscape features carefully for future reference?

Recent developments with harmonic radar have allowed tracking of individual bees during orientation flights over hundreds of metres. These have provided further insights into the process.

Because harmonic radar is also relevant to other studies of honey bee flight – for example, the impact of neonicotinoids on foraging ability – I’ll digress slightly from orientation flights to describe the technology.

Harmonic radar

Harmonic radar has revolutionised tracking studies of insects in much the same way as GPS tags have provided unique insights into bird migration (or, for that matter, shark migration).

The radar system has two components. The insect is tagged with a tiny antenna attached to a Schottky diode (together termed the transponder). The transmitter/detector is a ground-based scanner that transmits the radar signal. This is used as the energy source by the diode which re-emits a harmonic of the original signal which can then be detected.

Tagged bumble bee (left) and harmonic radar detector (right)

The transponder weighs less than a normal pollen load, though presumably there is some wind resistance from the antenna. In studies of bees with and without transponders fitted the orientation flights were of a similar duration, suggesting any wind resistance didn’t appreciably impact the flying ability of the bee.

Orientation (a-c) and foraging (d) flights monitored by harmonic radar.


Orientation flights 4 were taken by bees between 3 and 14 days post-emergence, with the mean onset of foraging being 14 days post-emergence.

Bees took between 1 and 18 orientation flights, though there wasn’t a direct relationship between the number of flights and the age of the bee, suggesting they may learn at different rates.

Initial orientation flights were generally in the immediate vicinity of the hive. Older workers – pre-foraging – ventured further afield. More recent studies have addressed this in greater detail (see below).

Orientation flights were distinctly different from foraging flights. The former were slower and less direct. The ground speed of orienting bees was ~3.6 m/s in contrast to foragers who flew at ~5.6 m/s and, as shown in D above, foraging flights were very much ‘there and back’ straight lines.

Venturing forth …

A more recent study 5 has used harmonic radar to investigate multiple orientation flights by individual bees, effectively analysing how the bee explores the landscape as it ages towards a forager.

This was a remarkable study. It involved addition and subsequent removal of the transponder from 115 individual bees during 184 orientation flights. When the orienting bee returned they recaptured it, removed the transponder and allowed it to reenter the hive. When it reappeared for another orientation flight they reattached the transponder.

Anyone complaining about their inability to mark queens 6 should do this as a training exercise 😉

The scientists also recorded several foraging flights of a smaller number of the same bees, to allow comparison with their behaviour during orientation flights.

Orientation and foraging flights of five individual bees.

Flights were defined as short or long range, but long range orientation flights were still significantly shorter than foraging flights.

  • Short range flights were made in poor weather and familiarised the bees with the immediate vicinity of the hive.
  • Consecutive long range flights reduced in duration as the bees learnt about the immediate hive vicinity i.e. the long range flights included some local exploring at first as well.
  • Orientation flights explored different areas of the landscape, rather than focusing on one sector.
  • Subsequent foraging flights involved areas that the bees may have never visited during orientation flights.
  • Some very long duration foraging flights may involve a degree of exploration, though it’s not clear whether this is truly orientation, or actual scouting activity.

Not all bees performed short and long range flights though early long range lights did involve local exploration as well. 

Ground clues and conclusions

The final part of this study investigated the influence of visual landscape features on orientation flights. This deserves a post in its own right as the techniques are quite involved.

Essentially they generated heat maps of the flights overlaid onto the geography. Using this approach they determined that some features visible from the air e.g. borders between grassland and a track, influenced the direction of flight and hence the orientation flights. 

There are additional studies of the influence of visible landmarks on bee flight which I’ll return to at some point in the future.

Again, like the comment made above about visible landmarks, it suggests to me that apiaries situated near such distinctive features may aid orientation and subsequent homing flights by honey bees.

When you next stand by your hive entrance on a warm, sunny afternoon and watch young workers flying to and fro across the entrance before spiralling up and away out of sight, you’ll know that it is an essential component of their training to be effective foragers.

They don’t forage for long – perhaps three weeks at most – but they are very effective, partly because they know where to return to.


Where do you go to my lovely? was the title of a rather syrupy (my blog, my opinion! … Apologies if it’s a favourite of yours 😉 ) song by Peter Sarstedt in 1969. It’s notable for some quite clever rhyming lyrics and a particularly dodgy mustache he sports in the YouTube video. (I’m just linking it, rather than embedding just because of the mustache).

It has nothing to do with bees.


A virus that changes bee behaviour

Particle physicists might not agree, but I think that evolution is the most powerful force in the universe. It is responsible for the fabulous diversity of life, for everything from the 6,000,000 kg Pando clonal colony of quaking aspen covering 43 hectares of the Fishlake National Forest in Utah, to the teeniest of tiniest of viruses.

As a microbiologist I’m acutely aware of the role evolution has played in the genetic arms race between hosts and pathogens. This is what is responsible for the multi-faceted immune system higher organisms carry – the antibodies, the lymphocytes, the complement and interferon responses, and everything else.

In turn, the fast replicating bacteria and viruses have evolved countermeasures to subvert these immune mechanisms, to switch them off entirely or to decoy them into targeting the wrong thing. This ‘arms race’ has gone on since well before the evolution of multicellular organisms (~600 million years ago) … and continues unabated.

Evolution is powerful for one simple reason; if a particular genetic combination 1 ‘works’ it will be passed on to the progeny. If a virus evolves a way to resist the immune response of the host, or to spread between hosts more efficiently, then the trait will be inherited.

Molecular mechanisms and behavioural changes

Some of changes work at the molecular level, invisible without exquisitely sensitive in vitro analysis; protein A binds to protein B and, in doing so, stops protein B from doing whatever it should have be doing. These are important but often very subtle.

Rabies: Slaying a mad dog, 1566 illustration from Wellcome Images

Other changes are far more obvious. Take rabies for example (or don’t, it’s not recommended as it has a near-100% case fatality rate) … the primary host of the rabies virus are carnivorous mammals. Infection causes gross behavioural changes that facilitate virus transmission. The animal becomes bolder and much more aggressive, resulting in virus transmission through biting.

Recent studies have elegantly demonstrated that this (also) is an example of protein A binding to protein B at the molecular level, it’s just that the phenotype 2 is very much more marked.

A protein on the surface of the virus resembles a snake venom toxin and has the ability to bind to nicotinic acetylcholine receptors present in the central nervous system of the mammalian host. These receptors ‘do what they say on the tin’ and bind the neurotransmitter acetylcholine. If the virus protein binds the receptor the response to acetylcholine is blunted and this, in turn, leads to hyperactivity, one of the key behavioural responses caused by rabies viruses.

That’s enough about mad dogs.

If virus-induced behavioural changes are so obvious, why haven’t lots of different examples already been identified and characterised?


Part of the problem is the blurring of distinctions between overt behavioural changes and the direct symptoms induced due to the virus replicating.

Human rhinovirus, the aetiological agent of the common cold, causes upper respiratory tract infections. You get a runny nose and you sneeze a lot.


Your behaviour changes.

However, it’s generally accepted that the sneezing and runny nose are a result of the physiological response to infection, rather than a virus-induced behavioural response to facilitate transmission.

It’s worth noting that all that “stuff” that comes out of your nose contains infectious virus, so it’s perhaps an artificial distinction between the general symptoms of sickness and evolutionarily-selected host behavioural changes caused by the virus.

Which in a roundabout way …

… allows me to finally introduce the topic of bee viruses that cause host behavioural changes involved in their transmission.

Or, rather, one bee virus that does this … though I’m certain that there will be more.

Israeli Acute Paralysis Virus (IAPV) is an RNA virus transmitted horizontally by direct contact between bees, or while feeding on developing pupae, by the parasitic mite Varroa destructor. It was implicated as a causative agent of Colony Collapse Disorder (CCD), though really compelling evidence supporting it as the primary cause never materialised.

It’s a virus UK beekeepers should be aware of, but unworried by, as it is extremely rare in the UK.

A recent paper has shown two behavioural changes in response to IAPV infection in honey bees. One of them – that facilitates horizontal transmission between colonies – is also partially explained at the molecular level.

The paper was published a couple of months ago:

Geffre et al., (2020) Honey bee virus causes context-dependent changes in host social behavior. Proc. Natl. Acad. Sci. USA 117:10406-10413. 3

I’m going to focus on the results, rather than the methods, though the methods are rather cool. They used barcoded bees to allow the automated image analysis of every bee in a colony for some of the studies where they had introduced known IAPV infected individuals.

Responses of nestmates to IAPV infected bees

Imagine watching a few hundred waggle dances and being able to recount the position, distance and response of every bee ‘watching’ 4 the dance, and then being able to summarise the results.

Over five days.


Including nights (and yes, bees do still waggle dance at night – a subject for the future).

The scientists orally infected groups of 30 bees with a sub-lethal dose of IAPV, marked them and released them into an observation hive. They then recorded their movements around the hive and their interactions with other bees in the colony. In particular, they focussed on trophallaxis interactions where one bee ‘feeds’ another.

Trophallaxis is also considered to be a method of communication in the hive and has been implicated in disease transmission.

The authors love their whisker plots and statistical analysis.

Who doesn’t? 😉

However, they generally make for rather underwhelming images in a bee blog for entertaining reading. Here .. see what I mean …

Number of trophallaxis interactions per hour.

Suffice to say that the results obtained were statistically significant.

They showed that the infected bees in the colony actually moved about the colony more than their nestmates. Conversely, they were engaged in fewer trophallaxis interactions i.e. it appeared as though they were being ‘ignored’ by their nestmates.

Were they really being ignored altogether or did their nestmates approach them, detect something was amiss and move away?


Antennation is the mechanism by which bees recognise nestmates. They use the sensitive chemoreceptors on their antenna to detect cuticular hydrocarbons (CHC) which are distinctive between bees from different hives.

Antennation is a precursor to trophallaxis.

After all, bees do not want to feed a foreigner, or exchange chemicals involved in communications, or even potentially risk being exposed to a new pathogen.

Good as the barcoding and camera system is, it’s not good enough to record antennation within the observation hive. To do this they manually 5 recorded antennation events between IAPV-infected bees and nestmates in cages in the laboratory 6.

In these studies IAPV-infected bees were engaged in the same number of antennation events as control bees. This strongly suggests that the nestmate could detect there was something ‘wrong’ with the IAPV-infected individuals. In support of this conclusion, the authors also demonstrated that bees inoculated with a double stranded RNA (dsRNA) stimulator of the honey bee immune response were also also antennated equally, but engaged in less trophallaxis interactions.

Therefore, these studies appear to show that nestmates exhibit a behavioural response to IAPV-infected bees (and bees with elevated immune responses, recapitulating their response to pathogen infection) that is likely to be protective, reducing the transmission of horizontally acquired viruses.

It’s worth noting two things here.

  1. There were no virus transmission studies conducted. It’s assumed that the lack of trophallaxis reduces virus transmission. That still needs to be demonstrated.
  2. This response is not induced by the virus on the host. It’s a response by nestmates of the host to virus infected individuals (or individuals that present as ‘sick’). As such it’s not the same as the rabies example I started this post with.

Virus-induced behavioural responses

But do the IAPV-infected bees behave differently when they come into contact with other bees who are not their nestmates?

After all, IAPV is a pathogenic virus and its continuing presence within a population (not just a single hive) depends upon it being spread from hive to hive.

For a highly pathogenic virus this is very important. If you spread from bee to bee within a hive and kill the lot you also go extinct … this partly explains the mechanism by which highly virulent viruses become less virulent over time.

But back to IAPV. What happens when IAPV-inoculated bees interact with bees from a different hive?

For example, what would happen if they drifted from one hive to another in a densely populated apiary? Drifting is a significant contributor to the spread of bees between adjacent colonies – studies show that 1% of marked bees drift to adjacent hives over a 3 day window. This partially accounts for the genetic mix of workers (up to 40% are unrelated to the queen that heads the colony) in a hive, a fact generally unappreciated by beekeepers.

The authors first showed that IAPV-infected bees could apparently leave and return to the hive with a similar frequency as uninfected foragers. Their flying was not compromised.

They then resorted again to recording interactions in the laboratory between IAPV-infected bees or control dsRNA-inoculated bees and workers from a different hive.

This was where it gets particularly interesting.

Hello stranger

The dsRNA-immunostimulated bees (remember, these induce a generalised immune response characteristic of a ‘sick’ bee) were treated aggressively by unmatched workers from a different hive.

In contrast, the IAPV-infected bees (which were ‘sick’ and would have been undergoing immunestimulation caused by the IAPV infection) experienced significantly less aggression than both uninoculated workers (which induced an intermediate response) and the dsRNA-inoculated.

This strongly suggests that IAPV is somehow able to modulate the appearance or behaviour (and one often determines the other) of the host to make it more acceptable to an unmatched worker.

They extended this study to conduct “field-based assays at the entrances of three normally managed honey bee colonies”, monitoring whether IAPV-infected bees were more likely to be accepted by the guard bees at the entrance of the hive.

They were. The IAPV-infected bees received a less aggressive reception and/or entered the hives much more easily than the controls.

But what about proteins A and B?

Good question.

The behavioural alterations described above must be explainable in terms of the molecular changes that IAPV induces in the bees. By that I mean that the virus must make, or induce the making of, a chemical or protein or other molecule, the presence of which explains their acceptance by the foreign guard bees.

And the obvious candidates are the cuticular hydrocarbons (CHC) that are recognized during antennation, which I introduced earlier.

And here the story leaves us with some tantalising clues, but no definitive answer.

The scientists demonstrate that there were marked differences between the CHC profile of IAPV-infected and control bees. Again, they used their favoured whisker plots to show this, but collated all of the CHC data into an even more difficult to explain scatter plot of linear discriminant analysis.

CHC profiles (relative abundance) shown using linear discriminant analysis.

The key take home message here is that for each of the CHC’s analysed there were differences in both the quality and relative abundance between the control bees, bees immunestimulated with dsRNA and the IAPV-infected bees.

These differences were so marked that you can see distinct clustering of points in the analysis above … these bees ‘look’ 7 different to the guard bees that antennate them.

This is a great story.

It’s as yet incomplete. To complete the understanding we will need to know which of those CHC’s, or which combination, when suppressed (or overrepresented) induce the guard bees to say “Welcome, step this way … “.

We’ll then of course need to find out how IAPV induces the change in CHC profile, which takes us right back to protein A and protein B again.

Ever the pedant

Much as I like this science I’d perhap argue that, again, the virus isn’t directly inducing a behavioural change in the host.

What it’s doing is inducing a behavioural change in the response to the infected host (by the guard bee). So perhaps this again isn’t quite the same as the rabies example we kicked off with.

A behavioural change in the host might include IAPV-infected bees drifting more, or drifting further. Alternatively, perhaps a colony with widespread IAPV infection could more easily indulge in robbing neighbouring colonies as they would experience less aggression from guard bees.

Smaller is better ...

Reduced entrance to prevent robbing …

I can see immediate evolutionary benefits to a virus that induced these types of behavioural changes. It’s not an original idea … the late Ingemar Fries suggested it in a paper two decades ago 8.

I’m also certain that researchers are looking for evidence supporting these types of directly-induced behavioural changes caused by viral pathogens in honey bees.

All religion, my friend, is simply evolved out of fraud, fear, greed, imagination, and poetry

Edgar Allen Poe may or may not have said this.

However, while we’re on the thorny subject of pathogen-induced behavioural changes in the host, it might be worth mentioning a couple of more controversial areas in which it has been proposed.

In the snappily titled paper “Assortative sociality, limited dispersal, infectious disease and the genesis of the global pattern of religion diversity” Fincher and Thornhill argue 9 that the wide diversity of religions in the tropics (compared to temperate regions) is driven by infectious disease selecting for three anti-contagion behaviours; in-group assortative sociality; out-group avoidance; and limited dispersal. It’s an interesting idea and I’m pleased I don’t have to test it experimentally. Their argument is that these three behavioural changes select for fractionation, isolation and diversification of the original culture … and hence the evolution of religions.

Conversely, perhaps microorganisms induce religious behaviours (rather than religion per se) that facilitate their transmission. This is exemplified in the entertainingly titled paper “Midichlorians – the biomeme hypothesis: is there a microbial component to religious rituals?” by Panchin et al., (2014). They argue that microbes – and they are really thinking about the gut microbiota here – might be able to influence their hosts (humans) to gather for religious rituals at which both ideas (memes) and infections are more easily transmitted.

Perhaps something to think about when mindlessly spinning out all that summer honey in the next few weeks?

Party, party

I think it’s fair to say that both the papers in the section above have some way to go until they achieve mainstream acceptance … if they ever do.

Furthermore, the general area in which parasites, bacteria and viruses, induce changes in the behaviour of their hosts’ is really in its infancy. We are aware of a lot of behavioural changes, but few are understood at the molecular level 10. As such, we often don’t know whether the association is correlative or causative.

Evolution is certainly a powerful enough selective force to ensure that even extremely subtle benefits to the pathogen may become a genetically-fixed feature of the complex interaction it has with the host.

Respiratory viruses, such as the common cold, Covid-19 and influenza infect millions of people globally and are readily transmitted by direct or indirect contact.

That’s why most of the readers of this post have a face mask nearby and a bottle of hand sanitizer ‘at the ready’. Or should.

Direct transmission benefits the virus as it does not have to survive on a door handle, milk bottle or petrol filling pump.

But direct transmission requires that people meet and are in close contact.

And a paper 10 years ago demonstrated that infection with influenza virus resulted in increased social interactions in the 48 hours post-exposure, compared with the same period pre-exposure 11.

It’s amazing what viruses can do … or might do … or (just look around you) are doing.


Barcoding bees

Every jar of honey I prepare carries a square 20mm label that identifies the apiary, batch, bucket and the date on which is was jarred. The customer can scan it to find out about local honey … and hopefully order some more.

The label looks a bit like this:

Scan me!

This is a QR code.

You’ll find QR codes on many packaged goods in the supermarket, on bus stop adverts, on … well, just about anything these days.  QR codes were first used in 1994 and are now ubiquitous.

QR is an abbreviation of quick response.

It’s a machine-readable two-dimensional barcode that is used to provide information about the thing it’s attached to.

QR codes contain positional and informational content. In the image above the three corners containing large squares allow the orientation to be unambiguously determined.

Within the mass of other, much smaller, black and white squares are several alignment points, an indication of the encoding 1 and the ‘information payload’. 

Large QR codes can contain more information and more error correction (so they can be read if damaged 2 ). Conversely, small QR codes contain reduced amounts of information and less error correction, but can still be used to uniquely identify individual things in a machine-readable manner.

A barcoded bee and barcode diagram.

And those ‘things’ include bees.

I am not a number 3

I had intended to write a post on how pathogens alter honey bee behaviour. This has been known about in general terms for some time, but only at a rather crude or generic level. 

To understand behavioural changes in more detail you need to do two things:

  • observe bees in a ‘natural setting’ (or at least as natural as can be achieved in the laboratory)
  • record hundreds or thousands of interactions between bees to be able to discriminate between normal and abnormal behaviour. 

And that isn’t easy because they tend to all look rather similar.

Lots of bees

How many of the bees above are engaging in trophallaxis?

Does the number increase or decrease over the next five minutes? What about the next hour?

And is it the same bees now and in an hour?

And what is trophallaxis anyway? 

I’ll address the last point after describing the technology that enables these questions to be answered.

And, since it’s the same technology that has been used to monitor the behavioural changes induced by pathogens, I’ll have to return to that topic in a week or two. 

Gene Robinson and colleagues from the University of Illinois at Urbana–Champaign have developed a system for barcoding bees to enable their unique identification 4.

Not just a few bees … not just a couple of dozen bees … every bee in the colony.

Though, admittedly, the colonies are rather small 😉

Each barcode carries a unique number, readable by computer, that can be tracked in real time.

So, unlike Patrick McGoohan, these bees are a number.


The scientists designed a derivative of the QR code that could be printed small enough to be superglued to the thorax of a worker bee. They termed these mini-QR-like codes bCodes 5. The information content of a bCode was limited by its size and the reference points it had to carry that allowed the orientation of the bee to be determined.

In total the bCode could carry 27 bits of data. Eleven bits (each essentially on or off, indicated by a black or white square) encoded the identification number, allowing up to 2048 bees to be uniquely numbered. The remaining 16 bits were the error-correction parity bits that had to be present to ensure the number could be accurately decoded.

If you’re thinking ahead you’ll realise that the maximum number of bees they could therefore simultaneously study was 2048. That’s about 1/25th of a very strong colony at the peak of the season, or the number of bees covering both sides of a two-thirds full frame of sealed brood.

It’s enough bees to start a one frame nucleus hive, which will behave like a mini-colony 6 and, in due course, expand to be a much larger colony.

And if you’re thinking a long way ahead you’ll realise the every barcode must be affixed to each bee in the same orientation. How otherwise would you determine whether the bees were head to head or abdomen to abdomen?

Labelling bees

This is the easy bit.

Each bCode was 2.1mm square and weighed 0.6mg i.e. ~0.7% of the weight of a worker bee. Honey bees can ‘carry’ a lot more than that. When they gorge themselves before swarming they ingest ~35mg of honey. 

The bCode therefore should not be an encumbrance to the bee (and they confirmed this in an exhaustive series of control studies).

A single frame of sealed brood was incubated and the bees labelled within a few hours of emergence. Typically, two batches of ~700 bees each were labelled from a single frame for a single experiment.

Each bee was anaesthetised by chilling on ice, the bCode glued in place (remember … in the same orientation on every bee) and the bee allowed to recover.

Labelling a single bee took 1-2 minutes.

Labelling 1400 bees takes several people a long time.

I said it was easy.

I didn’t say it was interesting.

Smile for the camera

I’ve not yet discussed the goal of the study that needed barcoded bees. It’s not really important while I’m focusing on the technology. Suffice to say the scientists wanted to observe bees under near natural conditions.

Which means a free-flying colony, on a frame of comb … in the dark.

Free-flying because caged bees do not behave normally.

On a frame of comb because they were interested in the interactions between bees under conditions in which they would normally interact.

And in the dark because that’s what it’s like inside a beehive (and it’s one of the features that scout bees favour when selecting a site for a swarm).

Camera and hive setup.

The scientists used an observation hive with a difference. It had an entrance to allow the bees to fly and forage freely and it contained a single sided, single frame. In front of the frame was a sheet of glass separated by 8mm from the comb. This prevented the bees from clambering over each other, which would have obscured the bCodes 7. Behind the frame was an 850nm infrared lamp to increase contrast, and the front was illuminated by several additional infrared lamps.

Bees cannot see light in the infrared range, so they were effectively in the dark.

The camera used (an Allied Vision Prosilica GX6600 … not your typical point and shoot) recorded ~29MP images every second. A typical experiment would involve the collection of about a million images occupying 4-6 terabytes of hard drive space 8.

The recorded images were processed to determine the temporal location of every bee with a visible (and readable) bCode. This was a computationally interesting challenge and involved discarding some data – e.g. barcodes that moved faster than a bee can walk or barcodes that fell to the bottom of the hive and remained motionless for days (i.e. dead bees). About 6% of the data was discarded during this post-processing analysis.


Which finally gets us to the point where we can discuss trophallaxis. 

Honey bees and other social insects engage it trophallaxis.

It involves two insects touching each other with their antennae while orally transferring liquid food. It occurs more frequently than would be required for just feeding and it has been implicated in communication and disease transmission

bCoded bees and trophallaxis

So, if you are interested in trophallaxis, how do you determine which bees are engaging in it, and which are just facing each other head to head?

In the image above the two bees in the center horizontally of the insert 9 are engaged in trophallaxis. The others are not, even those immediately adjacent to the central pair.

Image processing to detect trophallaxis – head detection.

This required yet more image processing. The image was screened for bees that were close enough together and aligned correctly. An additional set of custom computer-vision algorithms then determined the shape, size, position and orientation of the bees’ heads. To be defined as trophallaxis the heads had to be connected by thin shapes representing the antennae or proboscis.

And when I say the image … I mean all million or so images.

Bursty behaviour

And after all that the authors weren’t really interested in trophallaxis at all.

What they were really interested in was the characteristics of interactions in social networks, and the consequences of those interactions.

This is getting us into network theory which is defined as “Well out of my depth”

Transmission of things in a network depends upon interactions between the individuals in the network.

Think about pheromones, or honey, or email … or Covid-19.

It’s only when two individuals interact that these can be transmitted between the individuals. And the interaction of individuals is often characterised by intermittency and unpredictable timing. 

Those in the know – and I repeat, I’m not one of them – call this burstiness. 

If you model the spread of ‘stuff’ (information, food, disease) through a bursty human communication network it is slower than expected.

Is this an inherent characteristic of bursty networks?

Are there real bursty networks that can be analysed.

By analysing trophallaxis Gene Robinson and colleagues showed that honey bee communication networks were also bursty (i.e. displayed intermittent and unpredictable interactions), closely resembling those seen in humans.

However, since they had identified every trophallaxis interaction over several days they could follow the spread of ‘stuff’ through the interacting network.

By simply overlaying the real records of millions of interactions over several days of an entire functional community with an event transmitted during trophallaxis they could investigate this spread..  

For example, “infect” (in silico) bee 874 in the initial second and follow the spread of the “infection” from bee to bee through the real network of known interactions.

In doing this they showed that in a real bursty network, interactions between honey bees spread ‘stuff’ about 50% faster than in randomised reference networks. 

Why isn’t entirely clear (certainly to me 10, and seemingly to the authors as well). One obvious possibility is that the topology of the network i.e. the contacts within it, are not random. Another is that the temporal features of a bursty network influence real transmission events. 

Scientists involved in network theory will have to work this out, but at least they have a tractable model to test things on …

… and at a time when some remain in lockdown, when others think it’s all a hoax, when social distancing is 2m 11, when some are wearing masks and when prior infection may not provide protective immunity anyway, you’ll appreciate that ‘how stuff spreads’ through a network is actually rather important.

Stay safe


If it quacks like a duck …


… it might be a trapped virgin queen.

I discussed the audio monitoring of colonies and swarm prediction last week. Whilst interesting, I remain unconvinced that it is going to be a useful way to predict swarming. 

And, more importantly, that replacing the manual aspects of hive inspections is desirable. I’m sure it will appeal to hands off beekeepers, though I’m not sure that’s what beekeeping is about.

However there was a second component to what was a long and convoluted publication 1 which I found much more interesting.

Listening in

If you remember, the researchers fitted hives with sensitive accelerometers and recorded the sounds within the hive for two years. Of about 25 colonies monitored, half swarmed during this period, generating 11 prime swarms and 19 casts.

In addition to the background sounds of the hive, with changes in frequency and volume depending upon activity, some colonies produced a series of very un-bee-like toots and quacks.

Have a listen …

The audio starts with tooting, the quacking starts around 8-9s, and there’s overlapping tooting and quacking from near the 21s mark.

Queen communication

I’ve previously introduced the concept of pheromone-based communication within the hive. For example, the mated queen produces the queen mandibular and queen footprint pheromones, the concentrations of which influence the preparation and development of new queen cells.

Tooting and quacking is another form of queen communication, this time by virgin queens in the colony.

It’s not unusual to hear some of these sounds during normal hive inspections, but only during the swarming season and only when the colony is in the process of requeening.

If you rear queens, and in my experience particularly if you use mini-mating nucs, you will regularly hear “queen piping” – another term for the tooting sound – a day or so after placing a mature charged queen cell into the small colony.

But we’re getting ahead of ourselves. 

How does the queen make these sounds?

Queen piping or tooting

Queen tooting has been observed. The queen presses her thorax tight down against the comb and vibrates her strong thoracic wing muscles. Her wings remain closed. The comb acts as a sounding board, amplifying the sound in the hive (and presumably transmitting the vibrations through the comb as well).

This doesn’t happen just anywhere … the virgin queen is usually near the cell she has recently emerged from. 

And this swarm cell is usually on the periphery of a frame.

This is because the laying queen only rarely ventures to the edges of frames, so the concentration of her footprint pheromone is lower in this area, eventually resulting in queen cells being produced there

In their study, accelerometers embedded in the periphery of comb were able to detect much stronger tooting and quacking signals, supporting the conclusions of Grooters (1987) 2 who had first published studies on the location of piping queens.

Queen tooting and quacking

Queen piping is usually recorded at around 400 Hz and consists of one or more 1 second long pulses, followed by a number of much shorter pulses. In previous studies the frequency of tooting had been shown to be age-related. It starts at ~350 Hz and rises in frequency to around 500 Hz as the virgin queen matures over several days.

Compare the image above with the audio file linked further up the post. The tooting is followed by an extended period of quacking, and then both sounds occur at the same time.

Going quackers

The duck-like quacking is presumably also made by queens vibrating their flight muscles while pressed up against the comb.

I say ‘presumably’ as I don’t think it has been observed, as opposed to heard.

The reason for this is straightforward, the queens that are quacking are still within the closed queen cell.

Quacking is a lower frequency sound (is this because of the confines of the queen cell, the way the sound is produced, or the ‘maturity’ of the queen’s musculature?) but has also been shown to increase in frequency – from ~200 Hz to ~350 Hz – the longer the queen remains within the cell.

Afterswarms = casts

Before discussing the timing of tooting and quacking we need to quickly revisit the process of swarming. I’ve covered some of this before when discussing the practicalities of swarm control, so will be brief.

  1. Having “decided” to swarm the colony produces swarm cells. Usually several.
  2. Weather permitting, the prime swarm headed by the original laying queen leaves the hive, on or around the day that the first of the maturing queen cells is capped.
  3. Seven days after the cell was capped the first of the newly developed virgin queens emerges. 
  4. If the colony is strong, this virgin also swarms (a cast swarm). Some texts, including the publication being discussed, call these afterswarms.
  5. Over the following hours or days, successively smaller cast swarms may leave the hive, each headed by a newly emerged virgin queen.

Not all colonies produce multiple cast swarms, but initially strong colonies often do.

From a beekeeping point of view this is bad news™. It can leave the remnants of the original colony too weak to survive and potentially litters the neighbourhood with grapefruit, orange and satsuma-sized cast swarms. 

Irritating 🙁

Whether it’s good for the bees depends upon the likelihood of casts surviving. The very fact that evolution has generated this behaviour suggests it can be beneficial. I might return to this point at the end of the post.

Tooting timing

The Grooters paper referred to earlier is probably the definitive study of queen tooting or piping. The recent Ramsey publication appears to largely confirm the earlier results 3, but has some additional insights on colony disturbance during inspections 4.

Here is the acoustic trace of an undisturbed colony producing a prime swarm and two casts.

Timing of tooting and quaking in a swarming colony

I’ve added some visible labels to the image above indicating the occurrence of tooting and quacking in an undisturbed naturally swarming colony.

  • The prime swarm exited the hive on the afternoon of the 13th. No tooting had been recorded before that date.
  • On the 17th tooting starts and increases in frequency over the next two days.
  • Quacking starts 6 hours after the tooting starts.
  • The first cast swarm (afterswarm) exits the hive on the 19th and is followed by a three hour break in tooting.
  • Tooting and quacking then continue until the second cast swarm on the afternoon of the 21st.

So, in summary, tooting starts after the prime swarm leaves and stops temporarily when the first cast leaves the hive. Quacking starts after the tooting starts and then continues until the last swarm leaves the hive.

Why all the tooting and quacking?

The timing of queen tooting is consistent with it being made by a virgin queen that has emerged from the cell. The cessation of tooting upon swarming (the first afterswarm) suggests that the virgin left with the swarm. The restarting of tooting a few hours later suggests a new virgin queen has been released from another cell and is announcing her presence to the colony.

In previous studies, Grooters had shown that replaying the tooting sound to mature virgin queens actively chewing their way out of a queen cell delayed their emergence by several hours. This delay allowed the attendant workers to reseal the cell and obstruct her emergence for several days.

These timings and the behaviour(s) they are associated with suggest they are a colony-level communication strategy to reduce competition between queens. 

The newly emerged virgin queen toots (pipes) to inform the workers that there is ‘free’ queen in the colony. The workers respond by holding back emergence of other mature queens. 

If all (or several) of the virgin queens emerged and ran around the hive simultaneously they would effectively be ‘competing’ for the hive resources needed for successful swarming i.e. the workers. 

By controlling and coordinating a succession of queen emergence, a strong colony has the opportunity to generate one, two or more cast swarms whilst sufficient workers remain in the hive. It presumably helps ensure the casts are of a sufficient size to give them the best chance of survival.

At what point does this succession stop or break down? One possibility is that this happens when there are insufficient workers to prevent additional virgin queens from emerging.

Unanswered questions

Why do mature virgin queens within the cell quack? It is clearly a response to tooting, rather than being standard behaviour of a soon-to-emerge queen. 

Hear! Hear the pipes are calling, Loudly and proudly calling (from Scotland the Brave)

Is the quacking to attract workers to help reseal the cell?

I suspect not. At least, I suspect there is a more pressing need to attract the workers. After all, wouldn’t it be easier for the queen to simply stop chewing her way out for a few hours? 

Isn’t there a risk that a quacking cell-bound queen might attract the virgin queen running around ‘up top’ who might attempt to slaughter her captive half-sister? 

Possibly, so perhaps the workers that are attracted to the quacking cell also protect the cell, preventing the loose virgin queen from damaging the yet-to-emerge queens.

This would make sense … if the virgin leaves with a cast, the workers that will remain must be sure that there will be a queen available to head the colony

And finally, back to the tooting. I also wonder if this has additional roles in colony communication. For example, what other responses does it induce in the workers? 

Does the increasing frequency of tooting inform the workers that the virgin is maturing and that they should ready themselves for swarming? Perhaps tooting above a certain frequency induces workers to gorge themselves with honey to ensure the swarm has sufficient stores?

In support of this last suggestion, studies conducted almost half a century ago by Simpson and Greenwood 5 concluded that a 650 Hz artificial piping sound induced swarming in colonies containing a single mobile (i.e. free) virgin queen.


The apparently self-destructive swarming where a colony generates a series of smaller and smaller casts seems to be a daft choice from an evolutionary point of view.

Several studies, in particular from Thomas Seeley, have shown that swarming is a risky business for a colony … and that the majority of the risk is borne by the swarm, not the parental colony. 

87% of swarmed colonies will rear a new queen and successfully overwinter, but only 25% of swarms survive. And the latter figure must only get smaller as the size of the swarms decrease. 

One possibility is that under entirely natural conditions a colony will not undergo this type of self-destructive swarming. Perhaps it is a consequence of the strength of colonies beekeepers favour for good nectar collection or pollination?

Alternatively, perhaps it reflects the way we manage our colonies. Ramsey and colleagues also record tooting and quacking from colonies disturbed during hive inspections. In at least one of these their interpretation was that there were multiple queens ‘free’ in the hive simultaneously, presumably because workers had failed to restrict the emergence of at least one virgin queen.

So, perhaps hive inspections that (inadvertently) result in the release of multiple virgin queens are the colonies that subsequently slice’n’dice themselves to oblivion by producing lots of casts.

I can only remember one colony of mine doing this … and it started days after the previous inspection, but that doesn’t mean the disturbance I created during the inspection wasn’t the cause.

I’d be interested to know of your experience or thoughts.


The title of this post is derived from the Duck Test:

If it looks like a duck, swims like a duck, and quacks like a duck, then it probably is a duck.

This probably dates back to the end of the 19th Century. It’s a form of abductive 6 reasoning or logical inference. It starts with an observation or set of observations and then seeks to find the simplest and most likely conclusion from those observations. In comparison to deductive reasoning, logical inference does not lead to a logically certain conclusion. 

Inevitably, Monty Python stretched the logical inference a little too far in the Witch Logic scene from Monty Python and the Holy Grail:

What do you do with witches? Burn them! And what do you burn apart from witches? Wood! So, why do witches burn? ‘cos they’re made of wood? So; how do we tell if she is made of wood? Build a bridge out of ‘er! Ah, but can you not also make bridges out of stone? Oh yeah. Does wood sink in water? No, it floats! It floats! Throw her into the pond! What also floats in water? Bread! Apples! Very small rocks? Cider! Gra-Gravy! Cherries! Mud! Churches? Churches! Lead, Lead. A Duck! Exactly. So, logically… If she weighs the same as a duck, she’s made of wood… and therefore… a witch!

Does DWV replicate in mites?

Buckle up.

After a gentle tale of bad beekeeping last week 1 I think it’s time for a bit of science.

Does deformed wing virus replicate in Varroa mites?

Varroa destructor – focus stacked image

Actually, do any pathogenic virus of honey bees replicate in mites?

Does it matter?

Not really … in comparison to the 2.1 billion people who don’t have access to safe drinking water 2, it’s unimportant.

But if you’re interested in how viruses are transmitted between bees, or how the virus population evolves, then knowing whether the virus replicates in Varroa is important.

For example, perhaps replication in Varroa amplifies a particular sub-population of virus that are more virulent, or more transmissable?

Or causes bees to drift more, so spreading the virus more widely in the environment?

Behaviour-altering viruses are interesting. There has been recent a report that IAPV 3 changes the social behaviour of honey bees to potentially increase transmission 4.

So, if it does matter …

Why now?

Three things prompted me to write this post now:

  1. A recent paper on Black Queen Cell Virus replicating in Varroa was briefly discussed on the Bee-L mailing list. This paper repeated many of the errors of assumption in the historical literature  on virus replication in Varroa 5.
  2. The experiments done to address this question over the last couple of decades have been either contradictory or uninformative. Or both. Or just not very good 🙁
  3. We have just published a study 6 that specifically addresses whether deformed wing virus can replicate in mites. Inevitably 7 I think it’s the most complete answer to date but, as will become clearer, it raises more questions than it answers and I know we don’t yet understand the full story.

To understand the problem you need to have a little background on both viruses and Varroa. Our paper is on deformed wing virus (DWV) and I’ll almost exclusively be using this as an example. If I use the word ‘virus’ I mean DWV.

However, the generalities of the shortcomings in some of the historical studies and the way we tackled the problem is relevant to any virus of honey bees. In that regard, when I use the word ‘virus’ it could probably mean any virus of honey bees.

Background to viruses

Viruses are obligate intracellular parasites. They can only replicate inside a living cell. The majority of pathogenic viruses of honey bees have a genome composed of single stranded positive sense RNA.

I did remind you to buckle up? … here we go.

All single strand positive sense RNA viruses share a broadly similar replication strategy.

Virus replication – select for full size and legend.

The virus enters the cell and the genome is translated to make proteins. There are essentially two types of proteins made:

  • the structural proteins – these form the new progeny virus particles. They are what package the new virus genomes into particles (thereby protecting it from damage) to infect the next cell, or host.
  • the non-structural proteins – these proteins are the ‘motor’ that makes new virus genomes 8. They do not form part of the virus particle.

After translation the non-structural proteins replicate the virus genome. To do this they first make a single stranded negative-sense template from which they eventually copy hundreds or thousands of more positive-strands.

The positive strands (only) associate with the structural proteins to form new virus particles. These leave the cell and go on to infect another cell, or another bee.

It’s as simple as that 🙂

A few important things to remember:

  • negative strands are never present in the virus particle. Since they would be non-infectious the virus has evolved all sorts of elegant ways of excluding them from the particle.
  • there are lots of positive strands produced from each negative strand. A ratio of 1000:1 or higher is not unusual 9.
  • without the initial production of the non-structural proteins the virus cannot replicate.

Background to Varroa

I’ve discussed Varroa extensively in previous posts. A full description of the life cycle was presented in the post Know your enemy. For lots of juicy background detail please refer back to that post.

As far as this post is concerned the important facts are as follows:

  • Varroa feeds on developing honey bee pupae and, during the (misnamed) phoretic stage of the life cycle, on adult bees.
  • the mite ingests a rich ‘soup’ of cells and proteins from the bee. For a long time it was thought that the mite fed on haemolymph (effectively bee blood), but some currently favour the evidence that Varroa feeds on the fat body of bees. As far as this post goes, it’s irrelevant whether the mite feeds on haemolymph or the fat body 10.
  • when the mite feeds on a virus-infected bee is also ingests the virus.
  • when the mite subsequently feeds on a different bee (for example, to initiate another round of incestuous replication) it transmits some of the virus in its saliva to the new bee.

There, that wasn’t so bad was it?

Evidence for virus replication in Varroa

I’m not going to provide an exhaustive review the extensive literature here. It goes back to the mid-90’s and is pretty dull and rather hard going in places.

There are also lots of papers on DWV replication in bumble bees … does it?

Essentially the evidence for replication of viruses in mites is the presence of virus negative strands in the mite. Time and again this has been presented as “definitive proof” that the virus replicates in Varroa.

Does that seem reasonable?

You already know enough Varroa and virus biology to see the flaw in this conclusion.

Q. What did that mite last feed on?

A. A virus-infected pupa.

Q. And what was therefore present in abundance in that pupa?

A. Negative strands of the virus genome.

Just because viral negative strands are detectable in the mite does not mean that they were produced in the mite when (if?) the virus replicated.

Perhaps they were simply ingested with the soup of proteins and cells that the mite eats?

This is an error based upon a positive result. There is an assumption that the positive result tells you something … but it doesn’t, or at least it might not.

Evidence against virus replication in Varroa

This one is more subtle and – in contrast to the above –  is an error based on a negative result.

It’s always dangerous drawing scientific conclusions from negative results. Perhaps the experiment just didn’t work? Perhaps it wasn’t sensitive enough? This is why lots of controls are needed.

Scientists looked in detail at all of the proteins present in Varroa mites. They used a method called mass spectroscopy that detects tiny fragmented proteins which they identified by comparison with a database of known proteins.

Orbitrap ID-X Tribrid Mass Spectrometer

If one of the tiny fragments matches, there’s a reasonable chance that the entire protein was present. If lots of different tiny fragments match one protein there’s compelling evidence that the entire protein was present in the sample 11.

Using mass spectroscopy scientists detected the structural proteins of deformed wing virus. However, using the same methods, they could not detect the non-structural proteins.

Since the non-structural proteins were ‘absent’ the virus could not have been replicating in the mite as they are a prerequisite for replication.

Seems logical, but is again potentially wrong.

Perhaps the tiny protein fragments of the non-structural proteins of DWV are ‘sticky’ and remain associated with cellular fats and membranes 12 during the purification process?

What about controls? Did the authors demonstrate they could detect the non-structural proteins of a virus that does replicate in Varroa? Was the sensitivity of their assay sufficient to detect the levels of non-structural proteins expected to be present? Did they provide evidence they could detect other similarly low abundance proteins?


Should honey bee viruses replicate in Varroa?

Viruses are obligate intracellular parasites. They depend upon proteins and processes in the cell for their replication. This often means that viruses are restricted to a very narrow range of hosts. Sometimes viruses will only replicate in certain tissues or cell types of a single host species.

Varroa and honey bees are both Arthropods, but belong to different classes (Arachnida and Insecta respectively) within this phylum. They are therefore rather distantly related.

This does not mean that a virus of honey bees cannot replicate in the mite. There are several human viruses that also replicate in the mosquito (Dengue, Yellow Fever, Zika for example) during transmission. However, it’s an interesting philosophical cul-de-sac which is distracting us from the main event.

Genetically engineered deformed wing virus

Keeping up?


Now we’re getting to the more exciting stuff 13.

For some time now my lab have been able to genetically engineer deformed wing virus. This means we can make changes to the virus that help us do interesting or informative experiments.

For example, Olesya ‘Alex’ Gusachenko in my team built a virus that expresses a protein that turns bees green.

Green bees

This is allowing us to look in detail at the specific tissues the virus replicates in, or at transmission of the virus between bees. I’ll discuss green bees again in the future.

To investigate whether deformed wing virus replicates in Varroa we used a virus that had been engineered to contain a tiny genetic tag. It was equivalent to altering 0.05% of the virus genetic material. This genetic modification had three really important features:

  • it was unique and was not present in any other strains of deformed wing virus
  • although present in the virus genome it did not change any of the characteristics of the virus 14.
  • it was easy to detect. Alex developed an assay that could detect really tiny amounts (probably less than 100 copies) of genomes that contained this modification.

An artificial diet for Varroa

The final thing I need to introduce is how to keep Varroa mites in the laboratory 15. The experts can maintain Varroa in incubators, feeding them on little drops of honey bee haemolymph in what are called feed packets.

We’re not experts. But we know people who are.

We collaborated with our friends Alan Bowman, Ewan Campbell and Craig Christie in Aberdeen University who helped us with the Varroa feed packet part of the experiment.

A mite-infested apiary in Aberdeen with Luke, from my lab, and Craig discussing craft beers.

Critically, Alan’s lab have developed a feed packet system that contains no honey bee material. They can maintain Varroa in the lab without feeding them on mushed up bees. This means we had total control over what our Varroa could eat.

The feed packet contents are a closely guarded secret 16 but will be published soon.

Enough! Enough! Does DWV replicate in mites?

This is how we did the experiment. We grew large amounts of our genetically-tagged virus in honey bee pupae.

We purified the virus and treated it with an enzyme called RNAse.

As described above, the genome of DWV is made from the chemical RNA. RNAse degrades and destroys RNA.

However, viruses are resistant to RNAse because the RNA genome is protected inside the structural protein coat that forms outer layer of the virus particle.

We treated the purified virus preparation with RNAse to remove all RNA that was not inside the virus particles. This contaminating RNA could otherwise compromise the experiment.

Importantly, RNAse degrades both positive and negative strand RNA molecules. Therefore, we could be certain that our virus input into the experiment contained no contaminating negative strand RNA (and confirmed this in several controls).

We then added our virus to the feed packets being used to maintain groups of ten Varroa. After 4 days we harvested the Varroa and extracted all the RNA from them. We also had control mites which were not fed our tagged virus, but were otherwise treated in an identical manner.

All the mites contained very large amount of DWV-specific positive strand RNA. This was unsurprising. Even if our tagged DWV did not replicate in mites they had been feeding on honey bee pupae packed with DWV only four days ago.

We then looked for the presence of negative strand RNA. Again, there were reasonable amounts present, entirely consistent with their previous diet of DWV-infected honey bees. The control mites contained good levels of negative strand RNA.

DWV replicates in Varroa

However, only the mites fed the tagged virus contained a variably weak but consistent signal for the unique genetic tag we had introduced.

We interpret that as very good evidence that DWV does replicate in Varroa.

Is this level of replication significant?

Significant in terms of what?

It’s significant in that this experiment demonstrates for the first time the de novo production of replication intermediates (the negative strand RNA) in mites.

I’m not unbiased ( 🙂 ) but I think this is some of the best evidence supporting DWV replication in Varroa. It also demonstrates the types of experiments that are probably needed to determine whether other honey bee viruses replicate in the mite. The presence of negative strands is not enough. As the figure above shows, at least in the case of DWV, the majority come from the bee pupa the mite last fed on and we have no way to determine whether they are newly replicated or not.

But is the replication significant in terms of the amplification of the virus?

The presence of this signal confirms that there is some replication of DWV in Varroa. The fact that the signal is weak suggests that virus replication is not very extensive or very fast.

For comparison, if we inject the equivalent of 10 DWV virus particles into a honey bee pupa 17 they will be replicate to produce 1,000,000,000 (one billion) viruses within 48 hours. Assuming a conservative ratio of positive to negative strands, these individual pupae would each contain a million copies of negative strand RNA two days after infection.

We have yet to fully quantify specific negative strand levels replicated in Varroa, but it looks a whole lot less than that to me.

These are technically very demanding experiments and will take some time to complete. In the meantime we can and should ask a more relevant question.

Is this replication significant in terms of the biology of the virus?

By ‘biology of the virus’ I mean things like differential amplification of specific variants of the virus in the mite, so changing the virus population. For example, does virus adaptation occur, selecting for DWV variants that are more transmissable by Varroa?

We don’t know yet, but at least we have a system which will allow us to test this.

Although we have additional genetically-tagged variants of DWV we can test in the future, in mixed infections one will only be consistently selected over the others if it has a growth advantage.

We don’t yet know such a growth advantage even exists and there might be better ways to determine if it does. These are not straightforward experiments to undertake.

Based upon the limited replication we see in mites I do not think that the replication is significant in terms of increasing the amount of infectious virus in the mite. In contrast, the very high level of replication in honey bees may well allow the evolution of particular strains with particular characteristics.

Where is the replicating virus in Varroa?

We treated the mite like a simple bag of RNA.

We didn’t determine whether the virus was present in muscle, or brain or the ovaries.

The location of the virus is really important.

If the virus is replicating in a tissue that facilitates transmission to the next bee 18, particularly if certain types of virus are being selected because they grow better there, then it might be significant.

Some tissues allow viruses to be transmitted, others do not.

For example, poliovirus is a very distant cousin of DWV. Poliovirus is transmitted faecal-orally 19 in contaminated water supplies. After ingestion it replicates really well in the small intestine. It then has a short distance to travel (!) until it is excreted again. So, the primary site of poliovirus replication (the gut), is ideal as it ensures an obvious route for transmission.

Actually, poliovirus replicates so well in the gut that in about 0.5% of infected people it escapes from the gut and enters the peripheral nervous system. It then travels to the spinal cord and finally reaches the brain.

Poliovirus replicates pretty well in the brain as well … so well that it destroys some of the grey matter causing paralytic poliomyelitis 20.

But, as far as the virus is concerned, the brain is a dead-end. There’s no way out. Virus that replicates in the brain can never infect another person. How could it get from the brain to a water supply?

What about DWV? Is DWV replicating in tissues that enables transmission to another host?

Replication in the gut, in salivary glands or possibly the ovaries would all be consistent with transmission (with the ovaries as a source for vertical transmission to progeny mites – though they have already fed at the same spot on the pupa so determining this might be difficult). However, if DWV is replicating in mite muscle or brain it may well be irrelevant in terms of honey bee virus biology as there’s no obvious way out of those tissues.

Questions and answers

We attempted to answer a seemingly simple question “Can DWV replicate in mites?

We think the answer is yes.

But, as you can see from the few paragraphs above, that short question and even shorter answer has generated lots more questions.

And some of them are much more difficult to answer than the one we tackled 🙁

Identifying the site(s) of virus replication in Varroa would normally require microdissection of a variety of different tissues. Is the virus present? Is it replicating?

But a virus that expresses a fluorescent green protein during replication should make things a lot more straightforward … but more on that some other time.


The paper containing this study is available – open access (i.e. free) – the full title and journal details are: Gusachenko et al., (2020) Green Bees: Reverse Genetic Analysis of Deformed Wing Virus Transmission, Replication, and Tropism. Viruses 12: 532.

The memory of swarms

I’m writing this waiting for the drizzle to clear so I can go to the apiary and make up some nucs for swarm control. Without implementing some form of swarm control it’s inevitable that my large colonies will swarm 1.

Swarming is an inherently risky process for a colony. Over 75% of natural swarms perish, often because they do not build up strongly enough to overwinter successfully.

As a mechanism for reproduction swarming is somewhat unusual in that the intact colony is split into two not fully functional ‘halves’ 2.

By not fully functional I mean that neither the swarmed colony, nor the swarm are guaranteed to survive.

The swarmed colony lacks a queen, but has ample stores.

The swarm has a queen but has only the stores carried in the bellies of the workers.

The swarmed colony needs to rear a new queen. The swarm needs to find a new nest site, move there, build comb, rear brood, forage etc.

That seems like the very opposite of intelligent design, but it’s the way evolution has made things work. This being the case it involves a whole range of compromises and quick fixes that make it work.

One of these involves the memory of worker bees, which is what this post is about.

Two-stage swarming

A range of events within the hive – which for reasons that will become obvious I will term the original nest site – trigger the urge to swarm. I discussed some of these when covering swarm prevention. Swarming is then essentially a two-stage process. 

The two stage process of swarming

The first stage is the swarm leaving the original nest site and establishing a bivouac nearby. This is the classic cluster of bees hanging from a branch.

The bivouac sends out scout bees to search the nearby area for potential new nest sites. After ‘discussion’ (comprehensively covered by Thomas Seeley in Honeybee Democracy) between the scouts they reach a consensus of the best site.

The second stage is the relocation of the bivouacked colony to the new nest site. For example, this could be the church tower, a hollow tree or a bait hive. This site is likely to be within a few hundred metres of the original nest site, but can be further away.

All of which should raise some questions in the minds of beekeepers who are familiar with the “less than 3 feet or more than 3 miles” rule.

Have these bees not read the rules?

If you want to move a hived colony of bees you’ll often be told, or have read, that you need to move them either less than three feet or more than 3 miles.

Worker bees have a foraging range of about 3 miles. Within this range they have an uncanny ability to return to the hive location using features of the landscape to orientate themselves. The ‘final approach’ uses scent from the hive entrance.

Therefore, if you move a colony 3 feet they’ll still find the general location using landscape features, and then orientate to the hive entrance using scent.

If you move a colony 10 miles away everything is new to them and they’ll embark on some orientation flights to learn the new landscape features.

But if you move the colony a mile they’ll use the landscape features to return to the site of the original hive … to find it gone 🙁 3

Swarms break all these rules.

The bivouac is (in my experience) always more than 3 feet from the hive entrance. If the scout bees make the choice (e.g. selecting a bait hive to occupy), the swarm always relocates to a new nest site less than three miles from the site it left 4.

And a beekeeper who drops a bivouacked colony into a skep can move it wherever she wants, even back to the same hive stand it recently vacated.

If the swarm followed the rules, the majority of the workers would return from the bivouacked swarm to the original nest site.

At least they would if they had orientated to the original nest site in the first instance.

Are the bees naive?

About half of a workers life is spent as a forager collecting water, pollen or nectar. But before they venture out of the hive, the first half of a worker bees life is spent building comb, nursing larvae or cleaning cells.

Therefore, one possibility is that the bees present in a swarm have no knowledge of the hive location because they’ve never before left the hive.

We know that the proportion of workers that leave the colony when it swarms is about 75%. This has been determined in a number of independent studies and is remarkably consistent, irrespective of the size of the colony that swarms.

If 75% of the workers leave the colony when it swarms it is mathematically impossible for the swarm not to include older foragers (assuming the laying rate of the queen is steady).

In fact, we don’t need to resort to any underhand mathematics as the age classes of bees in a swarm have been measured. I’ve discussed this before when comparing natural and artificial swarms.

Age distribution of bees in swarms

Age distribution of bees in swarms

The median age of adult bees in the hive is 19 days. The median age of bees in a swarm is 10 days. Therefore swarms do contain younger bees, but not exclusively so.

One of the reasons for this bias towards younger bees must be to do with the relatively short lifespan of foragers. Many of the older bees in the swarm will have perished long before the new brood laid by the queen emerges.

Permanent amnesia?

The bivouacked swarm doesn’t dwindle in size as the older foragers drift back to the original nest site. Other than a few hundred scout bees, the majority of the bivouacked swarm huddle together to protect the queen, buried somewhere in the centre, from the elements.

They don’t fly or forage … they’re waiting for the signal from the scout bees that a new nest site has been located.

And, once they relocate to the church tower, the hollow tree or a bait hive, the older foragers stay in the new nest location. It’s as though the bees in a swarm that previously knew where the original nest site was have amnesia.

And this makes sense. If they did return to the original nest site the swarm (whether bivouacked or relocated) would shrink in size and it’s chances of surviving would be severely diminished. Other than a full belly of honey a swarm can rely on nothing. They need as many bees as possible to take on all the roles needed to establish a new colony – comb builders, nurses, foragers etc.

But have they really forgotten the original nest site?

Temporary amnesia?

It turns out that swarms do retain a memory of their original nest site.

In 1993 Gene Robinson and colleagues demonstrated that a swarm shaken out from its new nest site preferentially returns to the original nest site, rather than to an equidistant alternate 5.

This ability must rely on the memory of the foragers in the swarm. Therefore it is likely to be lost in a relatively short time (days, not weeks) 6.

Firstly, the foragers will be busy reorienting to the new nest site, effectively overwriting the memory of the original nest location. In good weather this takes just a couple of days.

Secondly, these ageing bees don’t have long to live, so there will be ever-decreasing numbers of them to lead a shaken out swarm back to the original location.

Rain stops play

Sometimes the bivouacked colony never relocates to a new nest site. Either the scouts never achieve a consensus or – more likely – bad weather forces the swarm to hunker down.

When you hive a bivouacked swarm you will often find a small crescent or two of new wax on the branch they were clinging to. If the bees get trapped by bad weather I think the comb building continues. It’s not unusual to find comb in hedgerows near apiaries where bees that have got trapped have ended up trying to make a new nest.

Natural comb

Natural comb …

What does the memory – or lack of it – of swarms mean for practical beekeeping?

The (temporary) amnesia of swarms means you can collect a bivouacked swarm and move it wherever you want. A swarm that relocates to your bait hive can also be moved, but don’t wait too long. Within just a few days of a swarm arriving the bees will have reoriented to their new location. I always try and move bait hives to their final location within three days of a swarm appearing.


The drizzle stopped and I spent the entire day finding queens and making up nucs.

Note to self … a super-strong colony with no queen cells, wall-to-wall brood and no very young larvae or eggs probably has a faulty queen excluder 🙁

Second note to self … Sod’s law dictates that the colony with the faulty queen excluder probably has supers filled with drone comb 🙁