The winter cluster

We had our first snow of the year last night and the temperature hasn’t climbed above 3°C all day. The hills look lovely and, unsurprisingly, I’ve not seen a single bee venturing out of the hives.

Winter wonderland

If you crouch down close to the hive entrance and listen very carefully you’ll be able to hear …

… absolutely nothing.

Oh no! Are they still alive? Maybe the cold has killed them already?

If you rap your knuckles against the sidewall of the brood chamber you’ll hear a brief agitated buzz that will quickly die back down to silence.

Don’t do that 😯

Don’t disturb them unless you absolutely have to. They’re very busy in there, huddling together, clustering to maintain a very carefully regulated temperature.

Bees and degrees

Any bee that did venture forth at 3°C would get chilled very rapidly. Although the wing muscles generate a lot of heat (see below), this uses a large amount of energy.

If the body temperature of an individual bee dips below ~5.5°C they become semi-comatose. They lose the ability to move, or warm themselves up again. Below -2°C the tissues and haemolymph starts to freeze.

However, as long as they’re not exposed to prolonged chilling (more than 1 hour) they can recover if the environmental temperature increases 1.

An individual bee has a large surface area to volume ratio, so rapidly loses heat. Their hairy little bodies help, but it’s no match for prolonged exposure to a cold environment.

But the bees in your hives are not individuals. Now, perhaps more than any other time in the season, they function as a colony. Survival, even for a few minutes at these temperatures, is dependent upon the insulation and thermoregulation provided by the cluster.

All for one, one for all

The temperature in the clustered colony is always above the coma-inducing 5.5°C threshold, even for the bees that form the outer surface layer, which is termed the mantle.

And the temperature in the core of the cluster is much warmer still, and if they’re rearing brood (as they soon will be 2) is maintained very accurately.

The mantle

The temperature inside the hive entrance, some distance from the cluster, is the same as the external ambient temperature. On a cold winter night that might be -5°C (in Fife), or -35°C (in Manitoba).

Studies have shown that clustered colonies can survive -80°C for 12 hours, so just a few degrees below freezing is almost balmy.

The winter cluster

Due to thermal radiation from the clustered colony, the temperature of the airspace around the colony increases as you get nearer the cluster. Draught free hives – and beekeepers that refrain from rapping on the brood box sidewall – will reduce movement of this air, so reducing thermal losses from convection.

The clustered colony is not a uniform ‘ball’ of bees. It has two distinct layers. The outer layer is termed the mantle and is very tightly packed with bees facing inwards. These bees are packed in so tightly that their hairy bodies trap air between them, effectively forming an insulating quilt.

To reduce heat loss further these mantle bees have a countercurrent heat exchanger (between the abdomen and the thorax) that reduces heat loss from the haemolymph circulating through their projecting abdomens.

The mantle temperature is maintained no lower than about 8°C, safely above coma-inducing lower temperatures.

Penguins and flight muscles

I’ve seen it suggested that the mantle bees circulate back into the centre of the cluster to warm up again, but have been unable to find published evidence supporting this. It’s an attractive idea, and it’s exactly what penguins do on the Antarctic ice sheet … but that doesn’t mean it’s what bees do.

Penguins, not bees

Although bees can cope with temperatures of 8°C, they cannot survive this temperature for extended periods. If bees are chilled to below 10°C for 48 hours they usually die. This would support periodically recirculating into the centre of the cluster to warm up.

Bees do have the ability to warm themselves by isometric flexing of their flight muscles. Essentially they flex the opposing muscles that raise and lower the wings, without actually moving the wings at all.

This generates a substantial amount of heat. On a cool day, bees warm their flight muscles by this isometric flexing before leaving on foraging flights. They have to do this as the flight muscles must reach 27°C to generate the wing frequency to actually achieve flight. Since bees will happily forage above ~10°C this demonstrates that the isometric wing flexing can raise the thoracic flight muscle temperature by at least 15-17°C.

But, briefly back to the penguin-like behaviour of bees, neuronal activity is reduced at lower temperatures. In fact, at temperatures below 18°C bees don’t have sufficient neuronal activity to activate the flight muscles for heat generation. This again suggests there is a periodic recycling of bees from the mantle to the centre of the cluster.

How can bees fly on cool days if it’s below this 18°C threshold? The day might be cooler, but the bee isn’t. The colony temperatures are high enough to allow sufficient neuronal activity for the foragers to pre-warm their flight muscles to forage on cool days.

Anyway, enough of a digression about flight muscles, onward and inward.

The core

Inside the mantle is the core. This is less densely occupied by bees, meaning that they have space to move around for essential activities such as brood rearing or feeding.

The temperature of the core varies according to whether the colony is rearing brood or not. If the colony is broodless the core temperature is maintained around 18°C.

The tightly packed mantle bees reduce airflow to the core. As a consequence of this the CO2 levels rise and the O2 levels fall, to about 5% and 15% respectively (from 0.04% CO2 and 21% O2 in air). A consequence of this is that the metabolic rate of bees in the core is decreased, so reducing food consumption and minimising the heat losses from respiration.

Brood rearing

My clustered winter colonies are probably just thinking about starting to rear brood 3.

Bees cannot rear brood at 18°C. Brood rearing is very temperature sensitive and occurs optimally at 34.5-35.5°C.

Outside that narrow temperature band things start to go a bit haywire.

Pupae reared at 32°C emerge looking normal (albeit a day or so later than the expected 21 days for a worker bee), but show aberrant behaviour. For example, they perform the waggle dance less enthusiastically and less accurately 4. In comparison to bees reared at 35°C, the ‘cool’ bees performed only 20% of the circuits and the ‘waggle run’ component was a less accurate predictor of distance to the food source.

Neurological examination of bees reared at 35°C showed they had increased neuronal connections to the mushroom bodies in the brain, when compared with those reared as little as 1°C warmer or cooler. This, and the behavioural consequences, shows how critical the brood nest temperature is.

The cluster position

The cartoon above shows the cluster located centrally in the hive. This isn’t unusual, though the cluster does tend to move about within the volume available as they utilise the stores.

You can readily determine the location of the cluster. Either insert a Varroa tray underneath an open mesh floor for a few days …

All is well ...

Tell tale signs of a brood-rearing cluster …

… or by using a perspex crownboard. I have these on many of my colonies and it’s a convenient way of determining the size and location of the cluster with minimal disturbance to the colony.

Perspex crownboard

Perspex crownboard …

Though you don’t need to check on them like this at all.

The photograph above was from late November (6 years ago). The brood box is cedar and therefore provides relatively poor insulation.

While checking the post-treatment Varroa drop in my colonies this winter it was obvious that cluster position varied significantly between cedar and poly hive types.

In poly hives (all my poly hives are either Abelo or Swienty) it wasn’t unusual to find the cluster tight up against one of the exterior side walls. In contrast, colonies hived in cedar brood boxes tended to be much more central.

This must be due to the better insulation of polystyrene compared with cedar.


Although I don’t think I’ve noticed this previously in the winter, it’s not uncommon in summer to find a colony in a poly hive rearing brood on the outer side of the frames adjacent to the hive wall. This is relatively rare in cedar boxes, other than perhaps at the peak of the summer.

If you’re interested in hive insulation, colony clustering and humidity I can recommend trying to read this paper by Derek Mitchell.

I don’t provide additional insulation to my colonies in the winter. It’s worth noting that all my hives have open mesh floors. In addition, the crownboard is topped by a 5 cm thick block of insulation throughout the year, either integrated into the crownboard or just stacked on top.

Perspex crownboard with integrated insulation

If you use perspex crownboards you must have insulation immediately above them. If you don’t you get significant amounts of condensation forming on the underside which then drips down onto the cluster.

The winter cluster and miticide treatment

The only time you’re likely to see the winter cluster is when treating with an oxalic acid-containing miticide. And only then when trickle treating.

With the choice between vaporising or trickle treating, I tend to be influenced by the ambient temperature.

If the cluster is very tightly clustered (because it’s cold) I tend to trickle treat.

If it is more loosely clustered I’m more likely to vaporise.

The threshold temperature is probably about 8°C, but I’m not precious about this. The logic – what little is applied – is that the oxalic acid crystals permeate the open cluster better than they would a closed cluster.

I’ve got zero evidence that this actually happens 😉

However, it’s worth reiterating the point I made earlier about airflow through the mantle. Since this is restricted in a tightly clustered colony – evidenced by the reduced O2 and elevated CO2 levels – then it seems reasonable to think that OA crystals are less likely to penetrate it either.

Of course, there’s an assumption that the trickled treatment can penetrate the cluster, and doesn’t just coat the mantle bees with a sticky OA solution.

Which neatly brings us back to penguins … if these mantle bees do recirculate through the cluster core they’ll take some of the OA with them, even if it didn’t get there directly.

Finally, it’s worth noting that cluster formation starts at about 14°C. As the temperature drops the cluster packs together more tightly. Between 14°C and -10°C the volume of the cluster reduces by five-fold.

By my calculations 5, at 2°C and 8°C the cluster is three and four times it’s minimal volume respectively, so perhaps both OA vapour and trickled solution could permeate perfectly well.



  1. We take advantage of this in the lab and chill bees briefly on ice to anaesthetise them for certain things. Soon after you return them to the incubator they become fully active again.
  2. Here at least …
  3. Actually, they’re not thinking about it at all. That’s me being anthropomorphic. They’ve been broodless for the last 4-5 weeks and I fully expect them to have capped brood shortly after the winter solstice. So, although they’re not thinking about it, it’s going to happen anyway. Isn’t evolution a wonderful thing?
  4. Note to self … I really should write something about the waggle dance … the only bee-related Nobel Prize deserves a post of its own.
  5. Of the back of an envelope type which assume there’s a linear relationship between cluster size and the ambient temperature.

    Relationship between cluster size and ambient temperature

24 thoughts on “The winter cluster

  1. Matthew Harris

    Interesting article. Thanks.

    Regarding penguin-like behaviour, I had always thought that the existence of the combs would make it hard for bees to commute from the outer crust right to the centre. That’s a problem penguins don’t have to contend with. I mean, I know combs can have holes in, but even so.

    1. David Post author

      Hi Matthew

      Even without the need to move about, simply organising a three-dimensional shell-like mantle with the combs and spaces between them is quite an achievement. The mantle bees in the comb are head down in the cells. What you can be sure about is that evolution will have created a solution that works, whether it involves penguin-like recycling, ‘heater’ bees or tiny little radiators 😉


  2. James Reid

    Hi David
    I treated my freebie swarm with OA in the summer, just after their arrival, and have also just removed Apivar strips. Should I give them another dose of OA? I seem to think it’s a once a year thing.

    1. David Post author

      Hi James

      You can treat them again. However, if you’ve only just removed the Apivar after the ‘proper’ 6-10 weeks and the colony is currently broodless then there’s little chance the mites will have had an opportunity to replicate further, so it might not be needed.


  3. Dave Stokes

    Another interesting article, Thanks David.
    I’m playing with thermocouples on top of the frames, just above the cluster and note that the temperature there is a steady 10 deg C above the outside temperature. This even applied a couple of weeks ago when the ambient was 8 deg C and the bees were bringing home pollen.
    I also noticed that when the colony was disturbed, the inside temperature shot up an extra 20 deg and took 36 hours to return to normal. There’s now a “Do not disturb” sign on the hive!
    I don’t suppose my efforts will win that second Nobel prize 😉

    1. David Post author

      Hello Dave

      Very interesting comment re. the increase in temperature after disturbance. I’ve noticed that colonies continue to buzz rather loudly some time after doing a vaporised OA treatment. Any bees that fly after removing the entrance block soon return, but it’s clear that the hive continues to be in a mildly agitated state.

      I too have dabbled with temperature probes, both using a commercial Arnia system and a homegrown Raspberry Pi hooked up to a few DS18B20’s.

      If you work out how to keep the temperature probe within the centre of the (moving) winter cluster you’ll be able to determine when they start and stop rearing brood … if you manage that it’s time to start practising Swedish.


        1. David Post author

          Very wise … possibly more exclusive as well. I’ve only met one Ignobel winner (the award for the ponytail study in 2012), but have met several Nobel laureates during the day job.

          I certainly think the Ignobel ceremony would be a lot more fun.


          1. David Post author

            It’s somehow reassuring that the discoverer of the Higgs boson still takes the bus … however, having experienced the traffic and roadworks in Edinburgh, it’s probably the fastest way to travel.


  4. Mandy

    This is my first winter with my first colony that I am lucky enough to have in my garden. I listen every day and have heard absolutely nothing over the last couple of days..but reassured that that’s a good thing right?
    Really informative article but with my inexperience (but s brilliant mentor) I am putting my faith in the bees themselves. Fingers crossed 🙏🏻

    1. David Post author

      Hi Mandy

      Yes, don’t expect to hear anything … particularly on cold days. If you minimised Varroa levels and gave them sufficient stores they should get through the winter without a problem. I’ll have a post in a week or two on midwinter checks.


  5. Julian Cox

    You in Scotland, me in Devon. I could really use a model of what my bees do in the warm and wet autumn leading into the cooler and wet winter and how this affects timing of treatment. If I assume that my bees keep a “summer” level of activity going for longer than yours, as someone else said, it is hard to squeeze in two treatments – Apivar and then almost immediately OA. It rarely snows here but the temperature can get down to as low as -5 on a handful of nights in the winter so just how long is their overwintering phase before brood starts again, or does it ever stop? (I don’t want to start poking about to find out!). Perhaps you could come down here to study how the bees annual calendar of activity differs compared to Scotland…?

    1. David Post author

      Hello Julian

      The BEEHAVE software for modelling hive activity (brood rearing, Varroa levels etc.) includes a climate parameter that influences the output. The default settings are for Rothamsted. However, it shouldn’t be too difficult – if you had access to climate data in the appropriate (rather simple) format – to use this instead. How accurate or otherwise would need to be determined.

      In previous years I’ve gone with the tried and trusted “end of the first protracted cold period” and it’s worked pretty well, both here in Scotland and in Warwickshire. However, we opened a couple of hives in mid-October and noted they were broodless, so treated relatively soon after that. Colonies were broodless when treated (though in fairness, we didn’t formally check every one). This was earlier than I’d expected and well before the first protracted cold spell.

      Do you have nectar flows that extend well into September? If not, you could consider treating with Apivar a bit earlier to protect the winter bee population and then having a bit more flexibility in choosing the time to treat in the winter.

      Remember that the strain of bee also probably influences things. Most of mine are relatively dark and tend to shut up shop quite early, or when there is a dearth of nectar or pollen.

      Look carefully at the Varroa trays, clean them regularly and re-check. It’s not a disaster if you get the midwinter timing wrong, at least not an immediate disaster. It just means you have let more mites survive than should have, which stores up problems for the future, so necessitating more checks through the season ahead.


  6. Alan Jenyon

    I notice from my hive monitoring that ‘de-clustering’ occurs as the ambient temperature rises above 13ºC but that an increase in the core temperature towards 34ºC seems to be triggered by a midday peak of 15ºC+. This year this has happened in my hives in the Welsh Marches on 30th October and again on 10th November.

    Whether this results in any successfully reared brood after the ambient temperature has quickly fallen again I cannot confirm as, unlike you, I have not been opening-up my hives to check! However a small brood nest could be located out-of-reach of the temperature sensor, as you have pointed-out.

    But I just wondered whether the numbers days since the last ambient midday peak of 15ºC or more could also be used to estimate the onset of brood-less periods?

    1. David Post author

      Hello Alan

      I don’t know … I wish I did. I suspect it’s not as straightforward as that and is also influenced by daylength and/or other factors. It’s pretty clear that a single factor cannot solely determine the onset of broodlessness, or the restarting of brood rearing. If it was as simple as that then aberrant weather in any year would induce dramatic changes in the colony that would have long-term consequences.

      An abnormally early warm period would induce brood rearing … only for a subsequent return to a normal lower temperature to force the colony to use stores at an advanced rate leading to starvation before foraging could begin (or sacrificing the brood, which they also do).

      In the same way, if broodlessness (or, more specifically, the queen stopping egg laying – broodlessness follows 21 days later) was induced solely by a dropping temperature, then a very cold early autumn would curtail winter brood rearing, condemning the colony to an early death as there were too few winter bees. Look at the factors known to contribute to winter bee production … foraging, daylength, nutrition, all mediated through pheromones. It’s likely to be something similar for brood rearing.

      An interesting topic for certain 😉

      Don’t think I’ve been opening hives all autumn and early winter … we opened some research colonies in mid-October to harvest some brood for some experiments, only to find it was present in vanishingly small amounts. Since then I’ve only looked in two or three colonies (all of which were broodless). Tempted as I am to know what’s going on, I’m happy enough to leave them essentially untouched (though I did slide a Varroa tray under one today).

      Etienne Tardif on Bee-L has recently been posting some interesting temperature measurements from his colonies. He’s recently emailed me and I’m slowly reading through the details. He also has hive monitors so might have some useful comparable data to look at.

      Monitoring cluster temperature in mid-season is relatively easy as the broodnest is so big you can dangle the probe more of less anywhere and it will be against brood. In contrast, in the winter the broodnest might be the size of golf ball (which still has ~100 worker cells by my back of an envelope calculations) and so is really easy to miss.


  7. Michael

    i noticed that you leave the 5cm insulation on top of your (presumably cedar) hives all year. I do someting very similar but only from October to March. Why leave them on through the warm period – would that not mean the colonies overheat on hot days?

    1. David Post author

      Hello Michael

      I have 5 cm of insulation on all of my hives all year, whether cedar or poly. It’s there for two reasons:

      1. Insulation keeps things cool when the external temperature is high
      2. Laziness Convenience

      All my hives have open mesh floors. I don’t think colonies need additional ventilation, as evidenced by the speed with which they seal up small holes in crownboards given the chance.

      I’ve done this for years and it works for me … and, more importantly, my bees.


  8. Sandra Wilson

    Hi again
    You can tell I’m catching up on some winter reading! 🙂
    If temperatures remain less than 18 degrees over winter which presumably is the norm in Scotland would it be right to assume brood rearing doesn’t take place for much of this period? I am a bit confused by considering open brood with winter treatments but the reality of average temperatures prevent this. Am I missing something?
    Thanks again


    1. David Post author

      Hello Sandra

      Some summers average less than 18 degrees (C) here! My colonies were broodless this season from mid/late October. I suspect they may just be starting again now but don’t intend to look as I’ve already treated them and don’t want to disturb them.

      For winter treatments – which usually means some form of oxalic acid – the key things to remember are 1) that sealed brood will contain mites and is untouched by the treatment, meaning mites survive, and 2) that dribbled oxalic acid damages open brood.

      The ideal time to treat is when there is no brood at all in the colony. This might not be possible. If that’s the case you have to judge the period when there’s minimal brood present. I’ve suggested ways that this can be investigated here and in previous posts on oxalic acid treatments.


  9. Alex

    Hi David – Another really informative article and a great read! You state that “Between 14°C and -10°C the volume of the cluster reduces by five-fold”. I wondered where you got this from – is there a paper I can read? I found this interesting paper by Charles Owens ( and wondered if anything else had been done more recently or expanded on the topic.

    1. David Post author

      Hello Alex

      Off the top of my head I can’t remember where that figure originates and I currently don’t have time to search through my browser cache to find it again … or worse, go through my bookshelf. On a related point Severson & Erickson looked at the relationship between cluster volume and temperature ( and quote an ~2 fold decrease in cluster volume (actually 55% decrease in volume and 40% decrease in surface area) when the temperature drops from 4C to -23C. Since the cluster is already, er, pretty well-clustered at 4C I don’t think it’s unreasonable to think that there’s likely to be significant expansion between 4C and 14C (or contraction, if the temperature is dropping).

      Have a look at the recent activity on the Bee-L mailing lists for a lot of discussion of the winter cluster and temperature. For example, look at recent threads that Etienne Tardif has started or contributed to.

      Finally, there’s a nice review of the winter cluster by Meghan Milbrath in the January 2020 edition of American Bee Journal … and, amazingly, I’ve just looked up the page number (p.33) and see that this is the source of the 5-fold decrease in cluster volume 🙂 Meghan’s article has 8-10 references quoted at the end, but the specific source is not cited.


      1. Alex

        Thanks David. I’ve downloaded the Severson & Erickson paper and will have a read with a nice cuppa and slice of cake. Thanks for the pointers on Bee-L – another one to tune in to 🙂 Hopefully when the Moir Library is open again I can call in and have a read of the Meghan Milbrath article in the ABJ. Since she’s based in Michigan i’m guessing she’ll know a thing or two about getting bees through extreme winters!


Leave a Reply

Your email address will not be published. Required fields are marked *