Tag Archives: Thomas Seeley

Quick thinking & second thoughts

I gave my last talk of the winter season on Tuesday to a lovely group at Chalfont Beekeepers Society. The talk 1 was all about nest site selection and how we can exploit it when setting out bait hives to capture swarms.

It’s an enjoyable talk 2 as it includes a mix of science, DIY and practical beekeeping.

Nest sites, bait hives and evolution

The science would be familiar to anyone who has read Honeybee Democracy by Thomas Seeley. This describes his studies of the features considered important by the scout bees in their search for a new nest site 3.

Under offer ...

Under offer …

The most important of these are:

  • a 40 litre cavity (shape unimportant)
  • a small entrance of 10-15cm2
  • south facing
  • shaded but in full view
  • over 5m above ground level
  • smelling of bees

All of which can easily be replicated using a National brood box with a solid floor. Or two stacked supers.

And – before you ask – a spare nuc box is too small to be optimal.

That doesn’t mean it won’t work as a bait hive, just that it won’t work as well as one with a volume of 40 litres 4.

Evolution has shaped the nest site selection process of honey bees. They have evolved to preferentially occupy cavities of about 40 litres.

Presumably, colonies choosing to occupy a smaller space (or those that didn’t choose a larger space 5 ) were restricted in the amount of brood they could raise, the consequent strength of the colony and the weight of stores they could lay down for the winter.

Get these things wrong and it doesn’t end well 🙁

A swarm occupying a nuc box-sized cavity would either outgrow it before the end of the season, potentially triggering another round of swarming, or fail to store sufficient honey.

Or both.

Over thousands of colonies and thousands of years, swarms from colonies with genetics that chose smaller cavities would tend to do less well. In good years they might do OK, but in bad winters they would inevitably perish.

Bait hive compromises

If you set out a nuc box as a bait hive, you’re probably not intending to leave the swarm in that box.

But the bees don’t know that. Their choices have been crafted over millenia to give them the best chance of survival.

All other things being equal they are less likely to occupy a nuc box than a National brood box.

Another day, another bait hive, another swarm …

For this reason I don’t use nuc boxes as bait hives.

However, I don’t recapitulate all the features the scout bees look for in a ‘des res’.

I studiously ignore the fact that bees prefer to occupy nest sites that are more than 5 metres above ground level.

This is a pragmatic compromise I’m prepared to make for reasons of convenience, safety and enjoyment.

Bees have probably evolved to favour nest sites more than 5 metres above ground level to avoid attention from bears. The fact that there are no bears in Britain, and haven’t been since the Middle Ages 6, is irrelevant.

The preference for high altitude nest sites was ‘baked into’ the genetics of honey bees over the millenia before we hunted bears 7 to extinction.

However, I ignore it for the following reasons:

  • convenience – I usually move occupied bait hives within 48 hours of a swarm arriving. It’s easier to do this from a knee height hive stand than from a roof ladder.
  • safety – I often move the bait hive late in the evening. Rather than risk disturbing a virgin queen on her mating or orientation flights (assuming it’s a cast that has occupied the bait hive) I move them late in the day. In the ‘bad old days’ when I often didn’t return from the office until late, this was sometimes in the semi-dark. Easy and safe to do at knee height … appreciably less so at the top of a ladder.
  • enjoyment – I can see the scout bees going about their business at a hive near ground level without having to get the binoculars out. Their behaviour is fascinating. If you’ve not watched them I thoroughly recommend it.

Scout bee activity

The swarming of honey bees is a biphasic process. In the first phase the colony swarms and forms a temporary bivouac nearby to the original nest site.

The two stage process of swarming

The scout bees search an area ~25 km2 around the bivouacked swarm for suitable nest sites. They communicate the quality and location of new nest sites by performing a waggle dance on the surface of the bivouac.

Once sufficient scouts have been convinced of the suitability of one of the identified nest sites the second phase of swarming – the relocation of the swarm – takes place.

Swarm of bees

Swarm of bees

However, logic dictates that the scout bees are likely to have already identified several potential new nest sites, even before the colony swarms and clusters in a bivouac.

There are only a few hundred scout bees in the swarmed colony, perhaps 2-3% of the swarm.

Could just a few hundred scouts both survey the area and reach a quorum decision on the best location within a reasonable length of time?

What’s a reasonable length of time?

The bivouacked swarm contains a significant amount of honey stores (40% by weight) but does not forage. It’s also exposed to the elements. If finding sites and reaching a decision on the best nest site isn’t completed within a few days the swarm may perish.

Which is why I think that scout bees are active well before the colony actually swarms.

Early warning systems

If scout bees are active before a colony swarms they could be expected to find and scrutinize my bait hive(s).

If I see them doing this I’m forewarned that a colony within ~3 km (the radius over which scout bees operate) is potentially making swarm preparations.

Since I’ll always have a bait hive or two within 3 km of my own apiaries I’ll check these hives at the earliest opportunity, looking for recently started queen cells.

Whether they’re my colonies or not, it’s always worth knowing that swarming activity has started. Within a particular geographic area, with similar weather and forage, there’s usually a distinct swarming period.

If it’s not one of my colonies then it soon might be 😉

So, in addition to just having the enjoyment of watching the scout bees at work, a clearly visible – ground level – bait hive provides a useful early warning system that swarming activity has, or soon will, start.

Questions and answers

Although talking about swarms and bait hives is enjoyable, as I’ve written before, the part of the talk I enjoy the most is the question and answer session.

And Tuesday was no exception.

I explained previously that the Q&A sessions are enjoyable and helpful:

Enjoyable, because I’m directly answering a question that was presumably asked because someone wanted or needed to know the answer 8.

Helpful, because over time these will drive the evolution of the talk so that it better explains things for more of the audience.

Actually, there’s another reason in addition to these … it’s a challenge.

A caffeine-fueled Q&A Zoom session

It’s fun to be ‘put on the spot’ and have to come up with a reasonable answer.

Many questions are rather predictable.

That’s not a criticism. It simply reflects the normal range of topics that the audience either feels comfortable asking about, or are interested in. Sometimes even a seemingly ‘left field’ question, when re-phrased, is one for which there is a standard answer. The skill in this instance is deciphering the question and doing the re-phrasing.

But sometimes there are questions that make you think afresh about a topic, or they force you to think about something you’ve never considered before.

And there was one of those on Tuesday which involved biphasic swarming and scout bee activity.

Do all swarms bivouac?

That wasn’t the question, but it’s an abbreviated form of the question.

I think the original wording was something like:

Do all swarms cluster in a bivouac or do some go directly from the original hive/location to the new nest site?

And I didn’t know the answer.

I could have made a trite joke 9 about not observing this because my own colonies swarm so infrequently 🙄

I could have simply answered “I don’t know”.

Brutally honest, 100% accurate and unchallengeable 10.

But it’s an interesting question and it deserved better than that.

So, thinking about it, I gave the following answer.

I didn’t know, but thought it would be unlikely. For a swarm to relocate directly from the original nest site the scout bees would need to have already reached a quorum decision on the best location. To do this they would need to have found the new nest site (which wouldn’t be a problem) and then communicate it to other scout bees, so that they could – in turn – find the site. Since this communication involves the waggle dance it would, by definition, occur within the original hive. Lots of foragers will also be waggle dancing about good patches of pollen and nectar so I thought there would be confusion … perhaps they always need to form a bivouac on which the scout bees can dance? Which explains why I think it’s unlikely.

In a Zoom talk you can’t ponder too long before giving an answer or the audience will assume the internet has crashed and they’ll drift off to make tea 11.

An attentive beekeeping audience … I’d better think fast or look stupid

You therefore tend to mentally throw together a few relevant facts and assemble a reasonable answer quite quickly.

And then you spend the rest of the week thinking about it in more detail …

Second thoughts

I still don’t know the answer to the question Do all swarms bivouac?”, but I now realise my answer made some assumptions which might be wrong.

I’ll come to these in a minute, but first let me address the question again with the help of the people who actually did the work.

I’ve briefly looked back through the relevant literature by Seeley and Lindauer and cannot find any mention of swarms relocating without going via a bivouac. I may well have missed something, it wouldn’t be the first time 12.

However, their studies are a little self-selecting and may have overlooked swarms that behaved like this.

Both were primarily interested in the waggle dance and the decision making process, they therefore needed to be able to observe it … most easily this is on the surface of the bivouac.

Martin Lindauer mainly studied colonies that had naturally swarmed, naming them after the location of the bivouac, and then studied the waggle dancing on the surface of the clustered swarm. In contrast, Tom Seeley created swarms by caging the queen and adding thousands of very well fed bees.

Absence of evidence is not evidence of absence.

So, what were the assumptions I made?

There were two and they both relate to confusion between waggle dancing foragers and scout bees.

  1. Swarming usually occurs during a strong nectar flow. Therefore there are likely to be lots of waggle dancing foragers in the hive at the same time the scouts are trying to persuade each other – using their own fundamentally similar – waggle dances.
  2. Bees ‘watching’ are unable to distinguish between scouts bees and foragers.

So, what’s wrong with these assumptions?

A noisy, smelly dance floor

Foragers perform the waggle dance on the ‘dance floor’. This is an area of vertical comb near the hive entrance. It’s position is not fixed and can move – further into the hive if the weather is cold, or even out onto a landing board (outside the hive) in very hot weather 13.

So, although the dance floor occupied by foragers isn’t immovable, it is defined. There’s lots of other regions of the comb that scouts could use for their communication i.e. there could be spatial separation between the forager and scout bee waggle dances.

Secondly, foragers provide both directional and olfactory clues about the identity and location of good sources of pollen and nectar. In addition to two alkanes and two alkenes produced by dancing foragers 14 they also carry back scents “acquired from the environment at or en route to the floral food source” which are presumed to aid foragers recruited by the waggle dancer to pinpoint the food source.

Importantly, non-dancing returning foragers do not produce these alkanes and alkenes. Perhaps the dancing scouts don’t either?

A dancing scout would also lack specific scents from a food source.

Therefore, at least theoretically, there’s probably a good chance that scout bees could communicate within the hive. Using spatially distant dances and a unique combination of olfactory clues (or their absence) scouts may well be able to recruit other scouts to check likely new nest sites.

All of which would support my view that bait hives provide a useful early warning system for colonies that are in the very earliest stages of swarm preparations … rather than just an indicator that there’s a bivouacked swarm in the vicinity.


All this of course then begs the question … if the scout bees can communicate within the hive, why does the swarm need to bivouac at all?

The bivouac must be a risky stage in the already precarious process of swarming. 80% of wild swarms perish. At the very least it’s subject to the vagaries of the weather. Surely it would be advantageous to stay within the warm, dry hive until a new nest site is identified?

Apple blossom ...

Apple blossom … and signs that a bivouacked swarm perished here

This suggests to me that the bivouac serves additional purposes within the swarming process. A couple of possibilities come to mind:

  • the gravity-independent, sun-orientated waggle dancing 15 on the surface of the bivouac may be a key part of the decision making process, not possible (for reasons that are unclear to me) within the confines of the hive.
  • the bivouac acts to temporally coordinate the swarm. A swarm takes quite a long time to settle at the bivouac. Many bees leave the hive during the excitement of swarming but not all settle in the bivouac. Perhaps it acts as a sorting mechanism to bring together all the bees that are going to relocate, separate from those remaining in the swarmed colony?

Clearly this requires a bit more thought and research.

If your association invites me to discuss swarms and bait hives next winter I might even have an answer.

But, as with so many things to do with bees, knowing that answer will only spawn additional questions 😉


Weight for spring

I’m currently reading The Lives of Bees by Thomas Seeley. It’s a very good account of honey bee colonies in the wild.

In the book Seeley describes studies he conducted in the early-80’s on the changes in weight of unmanaged colonies during the season 1.

One particular figure caught my attention as I was off to the apiary to heft 2 some hives and check on the levels of stores.

Colony weight (top) and weekly weight change (lower). Black arrows midwinter, red arrows early spring.

The upper panel shows the overall hive weight over a period of ~30 months, including three successive winters. I’ve butchered annotated the figure with black arrows to mark the approximate position of the winter solstice and added red arrows to indicate early spring (approximately mid-February i.e. about now) 3.

Look carefully at the slope of the line. In each year it steepens in early spring.

Shedding pounds

During the winter the colony survives on its reserves. There’s no forage available and/or it’s too cold to fly anyway, so the colony has to use honey stores to keep the worker bees alive.

In late autumn and early winter they can get by using minimal amounts of stores, just sufficient to keep metabolic activity of the bees high enough to maintain a cluster temperature of ~10°C.

All of this uses stores and so the colony gradually loses weight.

Frosty apiary

Frosty apiary

The weekly weight gained and lost is shown in the lower panel. In 1981/82, with the exception of a tiny weight gain in late August, the colony lost weight every week from mid-July until mid-April (9 months!).

But from mid-April to early-July the colony literally piles on the pounds 4.

To achieve this they need a strong population of worker bees. It’s not possible to collect that much nectar with just a few thousand bees. The colony must undergo a large population expansion from the 5,000-10,000 bees that overwintered the colony to a summer workforce of 30,000+.

This expansion is not simply the addition of a further 20,000 bees. At the same time as the new workers are emerging the winter bees are dying off. The colony therefore needs to rear significantly more than 20,000 bees to be ready to exploit the summer nectar flow.

And, since it takes bees to make bees this means that the colony must rear repeated cycles of new workers starting in very early spring.

But you can’t rear brood at 10°C

Brood rearing requires a cluster temperature of ~35°C. This is achieved by the bees raising their metabolic activity, repeatedly flexing their flight muscles and generating heat.

All of which uses lots of energy … which, in turn, is derived from the honey stores.

Which explains why, in early spring, the rate at which stores are consumed suddenly increases. And the rate at which the colony uses the stores is the weight lost per unit time i.e. the slope of the graph shown above, or below for emphasis.

Colony weight in early spring

Of course, some of the honey stores are also consumed by the developing larvae. How much presumably depends upon the amount of brood being reared and the external temperature.

Less brood requires less honey stores. Lower temperatures mean more energy must be expended to keep the brood at 35°C, so more stores are used.

Rearing brood also requires protein (pollen). Nearly 90 years ago Clayton Farrar studied the spring weight loss of colonies in Wisconsin maintained with or without pollen stores. Colonies unable to rear brood because they lacked pollen used ~50% less stores over the same period.

Thermoregulation is energetically costly and colonies must raise the temperature of the cluster to ~35°C in very early spring to rear sufficient brood to exploit the late spring nectar sources 5. They need to maintain these elevated temperatures – using yet more stores up – until the spring nectar flows start.

The danger zone

All of which means that we’re currently approaching the danger zone when colonies are much more likely to starve to death if they have insufficient stores.

The next six to eight weeks or so are critical.

If they have ample stores they will rear plenty of brood.

Clear evidence for brood rearing on trays under colonies in the bee shed

If they have borderline levels of stores they might be able to maintain viability of the colony, but they’ll only achieve this by not rearing brood 6.

If they start brood rearing and then run out of stores they will likely starve to death.

Winter chores

Every couple of weeks I check all of my colonies.

I confirm two things:

  • the entrance is clear i.e. not blocked with the corpses of dead bees.
  • the colony has sufficient stores for another few weeks.

This takes no more than one minute per colony and can be done whatever the weather, or even at night if you cannot get to the apiary in the short daylight hours.

Bent bicycle spoke to keep entrances clear

The floors I favour have an L-shaped entrance tunnel which during extreme periods of confinement can get blocked with dead bees. A quick scrape with a bent piece of stiff wire clears them away.

Even ‘normal’ entrances should be checked as it’s surprising the number of corpses that can accumulate, particularly after a long period of very adverse weather when no undertaker bees are flying.

As an aside, assuming no brood rearing, a colony entering the winter with 25,000 bees will likely lose an average of ~150 bees a day before brood rearing starts again in earnest. They will not be lost at the same rate throughout the winter. Do not worry about the corpses (though it’s worth also noting that a strong, healthy colony should clear these).

Hefting the hive

The weight of the hive can be determined in at least three different ways:

  1. wealthy beekeepers will use an electronic hive monitoring system with integral scales. No need even to visit the apiary to check these 😉 Where’s the fun in that?
  2. thorough beekeepers will use set of digital luggage scales to weigh each side of the hive, summing the two figures and noting the total carefully in their meticulous hive records.
  3. experienced beekeepers will briefly lift the back of the hive off the stand and decide “Hmmm … OK” or “Hmmm … too light”. This is termed hefting the hive.

If this is your first winter I strongly recommend doing the second and the third method every time you visit your apiary.

The second method will provide confidence and real numbers. These are what really count.

Hefting the hive for comparison will, over time, provide the ‘feel’ needed to judge things without a set of scales.

Over time you’ll find you can judge things pretty well simply by hefting. I do this 7, but only after removing the hive roof. I’ve got a variety of roofs in use – deep cedar monstrosities, dayglo polystyrene and lots of almost-weightless folded Correx – which, coupled with the variation in the number of boxes and the material they’re made from, complicates things too much.

Without the roof I find it a lot easier to judge.

Hmmm … too light

Anything that feels too light needs a fondant topup as soon as possible.

If you even think it feels too light it’s probably wise to add a block of fondant.

You need to use fondant as it’s probably too cold for bees to take down syrup. Fondant works, whatever the weather.

How much fondant should you add?

Look again at the hive weight loss per week in the early months of the year in the lower panel of the first graph. These colonies lost at least 1-1½ lb per week.

When will you next check them?

Do the maths as they say …

If you check them fortnightly you really need to add 1-2 kilograms which should be sufficient to get them through to the next check. Do not mess around with pathetic little 250g blocks of clingfilm-wrapped fondant. They might use that in three days …

You also don’t want to be opening the hive unnecessarily. Add a good-sized block and let them get on with things.


Loads of supermarket foods are supplied in a variety of clear or semi-translucent plastic trays – chicken, mushrooms, tortellini, curry ready meals etc 8. Throughout the season I wash these out and save them for use with the bees.

I stressed clear and semi-translucent as it helps to be able to tell how much of the fondant the bees have used up when you’re trying to judge whether they need any more.

Waste not, want not

Many of these trays are about 6″ x 4″ x 2″ which when packed with fondant conveniently weighs about a kilogram. I fill a range of these with fondant, cover them with a single sheet of clingfilm and write the weight on the clingfilm with a black marker pen.

Location, location, location

In the winter the goal should be to locate the fondant block as close as possible to the cluster.

This means directly over the cluster.

Not way off to one side because there are fewer bees on the top bars that are in the way.

There’s no point in adding fondant if you also force the bees to move to reach it.

You’ll need an eke (or a nice reversible, insulated crownboard) to provide the ‘headspace’ to accommodate the fondant block 9.

Fondant block directly over the top of the cluster (in this case on a queen excluder)

Don’t delay.

Don’t wait for a ‘nice’ day.

You’ve decided the colony is worryingly light so deal with it there and then.

Remove the roof and the crownboard. If it’s cold, windy or wet the bees are going to be reluctant to fly. Don’t worry, you’re helping them. You’ll do more harm by not feeding them than by exposing them to the elements for 30 seconds 10.

Remove the clingfilm entirely 11 and invert the plastic tub directly over the top of the cluster.

Add the eke. Replace the crownboard, the top insulation and the roof.

Job done.

Crownboards with holes and queen excluders

Some crownboards have holes in them. Often these in the centre of the board.

It’s often recommended to add the fondant block above the hole in the crownboard. I think it’s better to place the fondant directly onto the top bars for the following reasons:

  • the central hole in the crownboard is probably not above the cluster 12. Why give them more work to do to collect the stores you’re providing?
  • it’s cold above the crownboard. The bees are less likely to venture up there if it’s chilly and uninviting.
  • fondant deliquesces (absorbs moisture) and can get distinctly sloppy when located in the humid headspace above the crownboard. In contrast, if the fondant is on the top bars of the frames any moisture absorbed softens the fondant surface at exactly the point where the bees are going to eat it anyway.

Finally, if there’s any reason you need to go through the brood box (before the fondant is finished) place the fondant on top of a queen excluder directly over the frames. Fondant has a tendency to stick down firmly to the top bars and it’s a nightmare to remove it to get to the frames. In contrast, if it is on a queen excluder you can easily lift it off.

You might not need to do this, but I learnt the hard way 🙁


Polyandry and colony fitness

Honey bees are polyandrous. The queen mates with multiple drones during her mating flight(s). Consequently, her daughters are of mixed paternity.

In naturally mated queens there is a relationship between the number of patrilines (genetically distinct offspring fathered by different drones) and the ‘fitness’ of a colony.

Colony fitness

A ‘fit’ colony is one that demonstrates one or more desirable traits (those that benefit the colony … and potentially the beekeeper) such as better population growth, weight gain, resistance to pathogens or survival.

If you analyse the molecular genotype of the worker offspring you can determine which patriline they belong to. If you genotype enough workers you start to see the same patrilines appearing again and again. The more patrilines, the more drones the queen mated with 1.

Shallow depth of field

One of many …

Naturally mated queens mate with ~13 drones. Depending upon the study a range from as low as 1 to as high as 40 (and exceptionally into the high 50’s) has been demonstrated, though different studies all tend to produce an average in the low- to mid-teens.

There is a well-established link between polyandry and colony fitness 2. Essentially, the more genetically diverse a colony i.e. the larger the number of patrilines, the fitter that colony is.

The benefits of polyandry

Why should colonies with increased genetic diversity be fitter?

There are a number of hypotheses that attempt to explain why intracolonial genetic diversity is beneficial. These include the increased behavioural repertoire of the worker bees, a reduced production of diploid drones (which would otherwise be produced due to the single-locus sex determination system) and an increased resistance to a wide range of parasites and pathogens 3.

Parasites and pathogens are an extremely effective evolutionary selective pressure. Several studies from David Tarpy and Thomas Seeley have shown that increased polyandry results in better resistance to chalkbrood and American foulbrood.

But what about Varroa? It’s a new pathogen (evolutionarily speaking) to honey bees and there is evidence that the resistance mechanisms observed are genetically determined 4.

Does polyandry contribute to Varroa resistance? 

Would increased polyandry result in improved resistance to mites?

Limits of polyandry and natural resistance

Why is the average number of drone matings in the low teens?

If polyandry is beneficial – and there’s no doubt it is – then surely more patrilines (hyperpolyandry) would be even more beneficial?

How could this be tested?

Naturally mated queens only very rarely exhibit 30+ drone matings. Not only are these colonies hard to find, but they are so rare that doing any sort of statistical analysis of the improved (or otherwise) fitness is probably a non-starter.

Perhaps there’s an alternative way to approach the question? Rather than look at individual colonies within a mixed population, why not study the overall level of polyandry within a population that demonstrates resistance?

For example, do queens that head colonies of untreated feral bees that exhibit a demonstrated enhanced resistance to Varroa, the most important pathogen of honey bees, exhibit higher levels of polyandry?

Two relatively recent scientific papers have tackled these questions. Both have produced clear answers.

Drones : if more is better, is lots more better still?


Keith Delaplane and colleagues used instrumental insemination (II) of virgin queens to produce queens ‘mated’ with 15, 30 or 60 drones. Sperm was collected from 1, 2 or 4 drones from 15 donor colonies, mixed thoroughly and used for queen insemination.

Full-sized colonies were requeened with the II queens and left for 6 weeks 5 after which sampling started. Over two seasons a total of 37 colonies (with 11, 13 and 13 colonies respectively headed by queens ‘mated’ with 15, 30 and 60 drones) were tested at approximately monthly intervals.

Testing involved visual analysis of colony strength 6 and comb construction. Mite levels were measured using standard alcohol wash of ~300 bees at mid- or late-summer timepoints.

Brood frame with a good laying pattern

The results of this study are commendably brief … just 8 lines of text and two tables. I’ll summarise them in just a couple of sentences.

Colonies headed by queens ‘mated’ with 30 or 60 drones produced significantly more brood than the colony headed by the queen ‘mated’ with only 15 drones. Conversely, significantly more colonies headed by queens mated with only 15 drones had a higher level of mite infestation 7.

Natural Varroa resistance and polyandry

One of the best studied populations of feral bees co-existing with Varroa are those in the Arnot Forest in New York State. These are the bees Thomas Seeley and colleagues study.

These colonies live in natural holes in trees at low density through the forest. The colonies are small and they swarm frequently. Their spatial distribution, size and swarminess (is that a word?) are all evolutionary traits that enable resistance, or at least tolerance, to Varroa and the pathogenic viruses the mite transmits.

I’ve discussed Seeley’s studies of the importance of colony size and swarming previously. I don’t think I’ve discussed his work on spatial separation of colonies, but I have described related studies by Delaplane and colleagues.

Essentially, by being well-separated, mite transmission between colonies (e.g. during robbing) is minimised. Similarly, by existing as small colonies that swarm frequently Iwith concomitant brood breaks) the mite population is maintained at a manageable level.

Marked queen surrounded by a retinue of workers.

Her majesty …

Do the Arnot Forest Varroa-resistant 8 bees exhibit especially high levels of polyandry, suggesting that this contributes their survival?


Seeley and colleagues determined the number of patrilines in 10 Arnot Forest colonies using the same type of genotyping analysis described earlier. They compared these results to a similar analysis of 20 managed honey bee colonies located nearby.

On average, Arnot Forest queens had mated with ~18 drones (17.8 ± 9.8) each. In contrast, queens in managed colonies in two nearby apiaries had mated with ~16 and ~21 drones. These figures are not statistically different from each other or from the natural mating frequencies reported for honey bees in other studies.

Hyperpolyandry and colony fitness

The first of the studies confirms and extends earlier work demonstrating the polyandry (and in this instance hyperpolyandry i.e. at an even greater level than seen normally) increases colony fitness – at least in terms of colony strength and Varroa resistance.

Delaplane and colleagues hypothesise that the increased mite resistance in hyperpolyandrous (30 or 60 drones) colonies may be explained by either:

  • the importance of extremely rare alleles (gene variants), which would only be present in colonies in which the queen had mated with a very large number of drones.
  • the presence of beneficial non-additive interactions between genetically-determined traits e.g. grooming and hygienic behaviour and reduced mite reproduction.

Neither of which are mutually exclusive and both fit at least some of the extant data on natural mite resistance. Discriminating between these two hypotheses and teasing apart the variables will not be straightforward.

Absence of hyperpolyandry in naturally mite-resistant colonies

At first glance, the absence of the hyperpolyandry in the mite-resistant Arnot Forest bees studied by Thomas Seeley and colleagues appears to contradict the studies using the instrumentally inseminated queens.

The Arnot Forest bees exhibit the same level of polyandry as nearby managed colonies, and for that matter, as colonies studied elsewhere. They are mite-resistant but the queen has not mated with an increased number of drones.

In other studies 9, naturally mated colonies exhibiting different levels of polyandry (within the normal range) showed no correlation between Varroa levels and queen mating frequency.

Perhaps it’s surprising that the Arnot Forest queens hadn’t mated with fewer drones considering the extreme separation of the colonies (when compared with managed colonies). The colony density within the Forest is approximately one per square kilometre.

However, at least during the peak swarming and mating period in the season, drone availability is rarely limiting.

This is because drones are not evenly spread in the environment. Instead, they accumulate in drone congregation areas (DCA) to which the queen flies for mating.

What limits polyandry?

Polyandry is beneficial and, apparently, hyperpolyandry is more beneficial. However, queens mate with 10 – 20 drones, rather than 50 or more. Why is this?

Queen mating is a risky business. The queen has to fly to the DCA, mate with multiple drones and then return to the hive. She may make one or several mating flights.

I’ve discussed how far drones and queens fly to reach the DCA previously. Most drones fly less than 3 miles and 90% of matings occur within about 5 miles of the virgin queen’s hive. The queen probably flies further to the DCA.

All the time she is travelling to and from the DCA, and all the time she is present within it mating, she’s potentially at risk from hungry house martins, swallows, bee eaters (!) or from thunderstorms.

Or simply from getting lost.

Additionally, a number of honey bee pathogens are transmitted between drones and queens during mating. Hyperpolyandrous queens 10 are therefore at risk from these sexually transmitted diseases 11.

It’s therefore likely that the level of polyandry observed in honey bees has evolved as a consequence of the beneficial pressures polyandry brings balanced by the risks associated with mating multiple times.

Practical beekeeping

Although the two studies described here don’t have an immediate relevance to day-to-day practical beekeeping, it’s worth remembering that poor queen mating is regularly blamed for queen failures e.g. queens that develop into drone layers during the winter.

I’m going to write about drones later this year so for the moment will just make these points:

  • drone production is maximised to generate sexually mature drones for the swarming season
  • after eclosion, drones need to mature before being able to mate
  • drones live about 30 days and their sperm volume, though not necessarily viability, decreases as they age

Together this means that late in the season – perhaps late July or early August (though this will vary depending upon location) – the number of drones will decrease.

More significantly, the drones will be ageing.

In turn this means that late-mated queens may not mate with as many drones, or that the matings may not result in insemination.

Most beekeepers will be aware of queens that apparently ‘run out of sperm’ and become drone layers.

However, there may be less obvious problems with late-mated queens. I’m not aware of any studies on seasonality of queen mating and polyandry. However, I would not be at all surprised if they exhibited a reduced level of polyandry.

And, as described above, these colonies are likely to exhibit reduced fitness.

Something else to consider when deciding whether to unite a colony late in the season or hope the last of your virgin queens mates successfully …


Midwinter, no; mites, yes

There’s a certain irony that the more conscientious you are in protecting your winter bees from the ravages of Varroa in late summer, the more necessary it is to apply a miticide in the winter.

Winter bees are the ones that are in your hives now 1.

They have a very different physiology to the midsummer foragers that fill your supers with nectar. Winter bees have low levels of juvenile hormone and high levels of vitellogenin. They are long-lived – up to 8 months – and they form an efficient thermoregulating cluster when the external temperature plummets.

Winter bees production

In the temperate northern hemisphere, winter bees are reared from late summer/early autumn onwards. The combination of reductions in the photoperiod (day length), temperature and forage availability triggers changes in brood and forager pheromones.

Factors that influence winter bee production

Together these induce the production of winter bees.

For more details see Overwintering honey bees: biology and management by Döke et al., (2010).

Day length reduces predictably as summer changes to autumn. In contrast, temperature and forage availability (which itself is influenced by temperature and rainfall … and day length) are much more variable (so less predictable).

All of which means that you cannot be sure when the winter bees are produced.

If there’s an “Indian summer“, with warm temperatures stretching into late October, the bees will be out working the ivy and rearing good amounts of brood late into the year. The busy foragers and high(er) levels of brood pheromone will then delay the production of winter bees.

Conversely, low temperatures and early frosts reduce foraging and brood production, so bringing forward winter bee production.

It’s an inexact science.

You cannot be sure when the winter bees will be produced, but you can be sure that they will be reared.

Protect your winter bees

And if they are being reared, you must protect them from Varroa and the viral payload it delivers to developing pupae. Most important of these viruses is deformed wing virus (DWV).

Worker bee with DWV symptoms

Worker bee with DWV symptoms

Aside from “doing what it says on the tin” i.e. causing wing deformities and other developmental defects in some brood, DWV also reduces the longevity of winter bees.

And that’s a problem.

If they die sooner than they should they cannot help in thermoregulating the winter cluster.

And that results in the cluster having to work harder to keep warm as it gets smaller … and smaller … and smaller …

Until it’s so small it cannot reach its food reserves (isolation starvation) or freezes to death 2.

So, to protect your winter bees, you need to treat with an appropriate miticide in late summer. This reduces the mite load in the hive by up to 95% and so gives the winter bees a very good chance of leading a long and happy life 😉

Time of treatment and mite numbers

Time of treatment and mite numbers

I discussed this in excruciating detail in 2016 in a post titled When to treat?.

The figure above was taken from that post and is described more fully there. The arrow indicates when winter bees are produced and the variously coloured solid lines indicate mite numbers when treated in mid-July to mid-November.

The earlier you treat (indicated by the sudden drop in the mite count) the lower the peak mite numbers when the winter bees are being reared.

Note that the mite numbers indicated on the right hand vertical axis are not ‘real’ figures. They depend on the number present at the start of the year. In the figure above I “primed” the in silico modelled colony with just 20 mites. This will become very important in a few paragraphs.

Late season brood rearing

Compare the blue line (mid-August treatment) with the cyan line 3 (mid-October treatment) in the figure above.

The mid-October treatment really hammers the mite number down and they remain low until the end of the year 4.

The reason the mite numbers remain low after a mid-October treatment is that there is little or no brood being reared in the colony during this period.

Mites need brood, and specifically sealed brood, to reproduce on.

In the absence of brood the mites ‘colony surf‘, riding around as phoretic mites on nurse bees (or any bees if there aren’t the nurse bees they prefer).

And that late season brood rearing is the reason the end-of-year mite number for the colony treated in mid-August (the blue line) remains significantly higher.

Mites that survive the miticide in August simply carry on with their sordid little destructive lives, infesting the ample brood available (which could even include some highly mite-attractive and productive drone brood) and reproducing busily.

So, the earlier you treat, the more mites remain in the hive at the end of the year.

Weird, but true.

Early season brood rearing

The winter bees don’t ‘just’ get the colony through the winter.

As the day length increases and the temperature rises the colony starts rearing brood again. Depending upon your latitude it might never stop, but the rate at which it rears brood certainly increases in early spring.

Or, more correctly, in mid- to late-winter.

And it’s the winter bees that do this brood rearing. As Grozinger and colleagues state Once brood rearing re-initiates in late winter/early spring, the division of labor resumes among overwintered worker bees.”

Some winter bees revert to nurse bee activity, to rear the next generation of bees.

And this is another reason why strong colonies overwinter better … not because they (also) survive the cold better 5, but because there are more bees available to take on these brood rearing activities.

Strong, healthy colonies build up better in early spring.

Colonies that are weak in spring and stagger through the first few months of the year, never getting close to swarming, are of little use for honey production, more likely to get robbed out and may not build up enough for the following winter.

Midwinter mite treatments

Which brings us back to the need for miticide treatment in midwinter.

The BEEHAVE modelled colony shown in the graph above was ‘primed’ at the beginning of the season with 20 mites. These reproduced and generated almost 800 mites over the next 10-11 months.

What do you think would happen if you start the year with 200 mites, rather than 20?

Like the 200 remaining at the year end when you treat in mid-August?

Lots of mites … probably approaching 8000 … that’s almost as many mites as bees by the end of the season.

So, one reason to treat in the middle of winter is to reduce mite levels later in the season. The smaller the number you start with, the less you have later.

Vapour leaks out ...

Vaporisation … oxalic acid vapour leaks out …

But at the beginning of the season these elevated levels of mites could cause problems. High levels of mites and low levels of brood is not a good mix.

There’s the potential for those tiny patches of brood to become mite-infested very early in the season … this helps the mites but hinders the bees.

Logically, the more mites present at the start of brood rearing, the more likely it is that colony build up will be retarded.

So that’s two reasons to treat with miticides – usually an oxalic-acid containing treatment – in midwinter.

Midwinter? Or earlier?

When does the colony start brood rearing again in earnest?

This is important as the ‘midwinter’ treatment should be timed for a period before this when the colony is broodless. This is to ensure that all the mites are phoretic and ‘easy to reach’ with a well-timed dribble of Api-Bioxal.

In studies over 30 years ago Seeley and Visscher demonstrated that colonies have to start brood rearing in midwinter to build up enough to have the opportunity to swarm in late spring. These were colonies in cold climates, but the conditions – and season length – aren’t dramatically different to much of the UK.

Low temperatures regularly extend into January or February. The temperature is also variable year on year. It therefore seems (to me) that the most likely trigger for new brood rearing is increasing day length 6.

The apiary in winter ...

The apiary in winter …

I therefore assume that colonies may well be rearing brood very soon after the winter solstice.

I’m also aware that my colonies are almost always broodless earlier in the winter … or even what is still technically late autumn.

This is from experience of both direct (opening hives) or indirect (fresh brood mappings on the Varroa tray) observation.

Hence the “Midwinter, no” title of this post.

Don’t delay

I therefore treat with a dribbled or vaporised oxalic acid-containing miticide in late November or early December. In 2016 and 2017 it was the first week in December. Last year it was a week  later because we had heavy snow.

This year it was today … the 28th of November. With another apiary destined for treatment this weekend.

If colonies are broodless there is nothing to be gained by delaying treatment until later in the winter.

Most beekeepers treat between Christmas and New Year. It’s convenient. They’re probably on holiday and it is a good excuse to escape the family/mince pies/rubbish on the TV (delete as appropriate).

But it might be too late … don’t delay.

If colonies are broodless treat them now.

If you don’t and they start rearing brood the mites will hide away and be unreachable … but their daughters and granddaughters will cause you and your bees problems later in the season.

Finally, it’s worth noting that there’s no need to coordinate winter treatments. The bees aren’t flying and the possibility of mites being transferred – through robbing or drifting – from treated to untreated colonies is minimal.


Crime doesn’t pay

At least, sometimes it doesn’t.

In particular, the crime of robbery can have unintended and catastrophic consequences.

The Varroa mite was introduced to England in 1992. Since then it has spread throughout most of the UK.

Inevitably some of this spread has been through the activities of beekeepers physically relocating colonies from one site to another.

However, it is also very clear that mites can move from colony to colony through one or more routes.

Last week I described the indirect transmission of a mite ‘left’ by one bee on something in the environment – like a flower – and how it could climb onto the back of another passing bee from a different colony.

Mite transmission routes

As a consequence colony to colony transmission could occur. Remember that a single mite (assuming she is a mated female, which are the only type of phoretic mites) is sufficient to infest a mite-free hive.

However, this indirect route is unlikely to be very efficient. It depends upon a range of rather infrequent or inefficient events 1. In fact, I’m unaware of any formal proof that this mechanism is of any real relevance in inter-hive transmission.

Just because it could happened does not mean it does happen … and just because it does happen doesn’t mean it’s a significant route for mite transmission.

This week we’ll look at the direct transmission routes of drifting and robbing. This is timely as:

  • The early autumn (i.e. now) is the most important time of year for direct transmission.
  • Thomas Seeley has recently published a comparative study of the two processes 2. As usual it is a simple and rather elegant set of experiments based upon clear hypotheses.

Studying phoretic mite transmission routes

There have been several previous studies of mite transmission.

Usually these involve a ‘bait’ or ‘acceptor’ hive that is continuously treated with miticides. Once the initial mite infestation is cleared any new dead mites appearing on the tray underneath the open mesh floor must have been introduced from outside the hive.

All perfectly logical and a satisfactory way of studying mite acquisition.

However, this is not a practical way of distinguishing between mites acquired passively through drifting, with those acquired actively by robbing.

  • Drifting being the process by which bees originating from other (donor) hives arrive at and enter the acceptor hive.
  • Robbing being the process by which bees from the acceptor hive force entry into a donor hive to steal stores.

To achieve this Peck and Seeley established a donor apiary containing three heavily mite-infested hives of yellow bees (headed by Italian queens). These are labelled MDC (mite donor ccolony) A, B and C in the figure below. This apiary was situated in a largely bee-free area.

They then introduced six mite-free receptor colonies (MRC) to the area. Three were located to the east of the donor hives, at 0.5m, 50m and 300m distance. Three more were located – at the same distances – to the west of the donor apiary. These hives contained dark-coloured bees headed by Carniolan queens.

Apiary setup containing mite donor colonies (MDR) and location of mite receptor colonies (MRC).

Peck and Seeley monitored mite acquisition by the acceptor hives over time, fighting and robbing dynamics, drifting workers (and drones) and colony survival.

Test a simple hypothesis

The underlying hypothesis on the relative importance of robbing or drifting for mite acquisition was this:

If drifting is the primary mechanism of mite transmission you would expect to see a gradual increase of mites in acceptor colonies. Since it is mainly bees on orientation flights that drift (and assuming the egg laying rate of the queen is constant) this gradual acquisition of motes would be expected to occur at a constant rate.

Conversely, if robbing is the primary mechanism of mite transmission from mite-infested to mite-free colonies you would expect to see a sudden increase in mite number in the acceptor hives. This would coincide with the onset of robbing.

Graphically this could (at enormous personal expense and sacrifice) be represented like this.

Mite acquisition by drifting (dashed line) or robbing (solid line) over time (t) – hypothesis.

X indicates the time at which the mite-free acceptor colonies are introduced to the environment containing the mite-riddled donor hives.

These studies were conducted in late summer/early autumn at Ithaca in New York State (latitude 42° N). The MDC’s were established with high mite loads (1-3 mites/300 bees in mid-May) and moved to the donor apiary in mid-August. At the same time the MRC’s were moved to their experimental locations. Colonies were then monitored throughout the autumn (fall) and into the winter.

So what happened?

Simplistically, the three mite donor colonies (MDC … remember?) all collapsed and died between early October and early November. In addition, by mid-February the following year four of the six MRC’s had also died.

In every case, colony death was attributed to mites and mite-transmitted viruses. For example, there was no evidence for starvation, queen failure or moisture damage.

But ‘counting the corpses‘ doesn’t tell us anything about how the mites were acquired by the acceptor colonies, or whether worker drifting and/or robbing was implicated. For this we need to look in more detail at the results.

Mite counts

Mite counts in donor (A) and receptor (B, C) colonies.

There’s a lot of detail in this figure. In donor colonies (A, top panel) phoretic mite counts increased through August and September, dropping precipitously from mid/late September.

This drop neatly coincided with the onset of fighting at colony entrances (black dotted and dashed vertical lines). The fact that yellow and black bees were fighting is clear evidence that these donor colonies were being robbed, with the robbing intensity peaking at the end of September (black dashed line). I’ll return to robbing below.

In the receptor colonies the significant increase in mite numbers (B and C) coincided with a) the onset of robbing and b) the drop in mite numbers in the donor colonies.

Phoretic mite numbers in receptor colonies then dropped to intermediate levels in October before rising again towards the end of the year.

The authors do loads of statistical analysis – one-way ANOVA’s, post-hoc Wilcoxon Signed-Rank tests and all the rest 3 and the data, despite involving relatively small numbers of colonies and observations, is pretty compelling.


So this looks like robbing is the route by which mites are transmitted.

A policeman would still want to demonstrate the criminal was at the scene of the crime.

Just because the robbing bees were dark doesn’t ‘prove’ they were the Carniolans from the MRC’s 4. Peck and Seeley used a 400+ year old ‘trick’ to investigate this.

To identify the source of the robbers the authors dusted all the bees at the hive entrance with powdered sugar. They did this on a day of intense robbing and then monitored the hive entrances of the MRC’s. When tested, 1-2% of the returning bees had evidence of sugar dusting.

Returning robbers were identified at all the MRC’s. Numbers (percentages) were small, but there appeared to be no significant differences between nearby and distant MRC’s..

Drifting workers and drones

The evidence above suggests that robbing is a major cause of mite acquisition during the autumn.

However, it does not exclude drifting from also contributing to the process. Since the bees in the MDC and MRC were different colours this could also be monitored.

Yellow bees recorded at the entrances of the dark bee mite receptor colonies.

Before the onset of significant robbing (mid-September) relatively few yellow bees had drifted to the mite receptor colonies (~1-2% of bees at the entrances of the MRC’s). The intense robbing in late September coincided with with a significant increase in yellow bees drifting to the MRC’s.

Drifting over at least 50 metres was observed, with ~6% of workers entering the MRC’s being derived from the MDC’s.

If you refer back to the phoretic mite load in the donor colonies by late September (15-25%, see above) it suggests that perhaps 1% of all 5 the bees entering the mite receptor colonies may have been carrying mites.

And this is in addition to the returning robbers carrying an extra payload.

Since the drones were also distinctively coloured, their drifting could also be recorded.

Drones drifted bi-directionally. Between 12 and 22% of drones at hive entrances were of a different colour morph to the workers in the colony. Over 90% of this drone drifting was over short distances, with fewer than 1% of drones at the receptor colonies 50 or 300 m away from the donor apiary being yellow.

Discussion and conclusions

This was a simple and elegant experiment. It provides compelling evidence that robbing of weak, collapsing colonies is likely to be the primary source of mite acquisition in late summer/early autumn.

It also demonstrates that drifting, particularly over short distances, is likely to contribute significant levels of mite transmission before robbing in earnest starts. However, once collapsing colonies are subjected to intense robbing this become the predominant route of mite transmission.

There were a few surprises in the paper (in my view).

One of the characteristics of colonies being intensely robbed is the maelstrom of bees fighting at the hive entrance. This is not a few bees having a stramash 6 on the landing board. Instead it involves hundreds of bees fighting until the robbed colony is depleted of guards and the robbers move in mob handed.

As a beekeeper it’s a rather distressing sight (and must be much worse for the overwhelmed guards … ).

I was therefore surprised that only 1-2% of the bees returning to the mite receptor colonies carried evidence (dusted sugar) that they’d been involved in robbing. Of course, this could still be very many bees if the robbing colonies were very strong. Nevertheless, it still seemed like a small proportion to me.

It’s long been known that mites and viruses kill colonies. However, notice how quickly they kill the mite receptor colonies in these studies.

The MRC’s were established in May with very low mite numbers. By the start of the experiment (mid-August) they had <1% phoretic mites. By the following spring two thirds of them were dead after they had acquired mites by robbing (and drifting) from nearby collapsing colonies 7.

It doesn’t take long

The science and practical beekeeping

This paper confirms and reinforces several previous studies, and provides additional evidence of the importance of robbing in mite transmission.

What does this mean for practical beekeeping?

It suggests that the late-season colonies bulging with hungry bees that are likely to initiate robbing are perhaps most at risk of acquiring mites from nearby collapsing colonies.

This is ironic as most beekeepers put emphasis on having strong colonies going into the winter for good overwintering success. Two-thirds of the colonies that did the robbing died overwinter.

The paper emphasises the impact of hive separation. Drifting of drones and workers was predominantly over short distances, at least until the robbing frenzy started.

This suggests that colonies closely situated within an apiary are ‘at risk’ should one of them have high mite levels (irrespective of the level of robbing).

If you treat with a miticide, treat all co-located colonies.

However, drifting over 300 m was also observed. This implies that apiaries need to be well separated. If your neighbour has bees in the next field they are at risk if you don’t minimise your mite levels … or vice versa of course.

And this robbing occurred over at least 300 m and has been reported to occur over longer distances 8. This again emphasises both the need to separate apiaries and to treat all colonies in a geographic area coordinately.

Most beekeepers are aware of strategies to reduce robbing i.e. to stop colonies being robbed. This includes keeping strong colonies, reduced entrances or entrance screens.

But how do you stop a strong colony from robbing nearby weak colonies?

Does feeding early help?

I don’t know, but it’s perhaps worth considering. I don’t see how it could be harmful.

I feed within a few days of the summer honey supers coming off. I don’t bother waiting for the bees to exploit local late season forage. They might anyway, but I give them a huge lump of fondant to keep them occupied.

Do my colonies benefit, not only from the fondant, but also from a reduced need to rob nearby weak colonies?

Who knows?

But it’s an interesting thought …

Note there’s an additional route of mite transmission not covered in this or the last post. If you transfer frames of brood from a mite-infested to a low mite colony – for example, to strengthen a colony in preparation for winter – you also transfer the mites. Be careful.


The idiom “Crime doesn’t pay” was, at one time, the motto of the FBI and was popularised by the cartoon character Dick Tracy.

Woody Allen in Take the Money and Run used the quote “I think crime pays. The hours are good, you travel a lot.”