Category Archives: Bees

Summer doldrums

Synopsis : Weather, nucs, queens, swarms and wax. A potpourri of topics that have entertained me while waiting for the heather flow to start … if it starts. 

Introduction

Maybe it’s old age 1 but the calendar seems to be speeding up these days. The dawn chorus is just a chirrup or two, there are mushrooms appearing everywhere, and I can sense the end of the season galloping towards me.

Or, maybe this is just a bit of a weird season.

It barely seems to have got started before it feels like it’s about to end.

The peak swarming period in my part(s) of Scotland has long gone 2 and the ‘June gap’ didn’t really happen, perhaps because the summer flowers started early. It’s now late July and the lime and blackberry are over. There’s some rosebay willow herb left, though not much and the heather has yet to properly start.

Erratic, just flowering heather

These are the summer doldrums; the time between the end of swarming and taking the summer honey off. It’s a relatively quiet time for my beekeeping. It’s too late (in Scotland, at least in my experience) for dependable queen rearing and it’s too early for any serious winter preparations.

There’s no longer a need to conduct weekly colony inspections. I’m reasonably confident that my colonies won’t swarm now, though I expect a few might supersede 3.

That doesn’t mean they won’t swarm … it just means my confidence might be misplaced 😉 .

Of course, that doesn’t mean that there’s no beekeeping to do … it’s just that I’ve got a little more time to complete what I need to do, and a bit of spare time to do a few other related activities.

Continue reading

Coffee, constancy and fidelity

Synopsis : Why do bees collect pollen of only one type when foraging? Why do they forage repeatedly in the same area? What has coffee got to do with this?

Introduction

Foraging is what my bees should be doing now. The summer nectar flow should be strong – lime, blackberry, rosebay willow herb (RBWH, fireweed) then heather – it’s bonanza time.

But note the qualifier ’should’.

So far, it’s not looking promising.

The lime was hopeless, the blackberry flowered well but doesn’t appear to have yielded much, the fireweed is nearly over (early) and the heather … well, let’s not prejudge anything, but I’m not hopeful.

Going, going, gone … rosebay willow herb, mid-July 2023

Not only do those four plants/trees yield nectar, but they also produce pollen and you can often tell what the bees are foraging on by the colour of the packed corbiculae on the hind legs of returning workers.

Despite their overlapping flowering periods the pollen baskets are almost always a single colour. For example, you don’t get deep purple pollen baskets from RBWH speckled with much paler borage pollen, despite the fact you can find both flowering – in a field and its margins – simultaneously.

This is because honey bees tend to forage on one plant species 1 on any foraging trip. This feature of the foraging habit of honey bees is termed constancy.

If you marked a foraging worker on a patch of RBWH, watched it fly off to the hive and waited a bit you might well see the marked bee return to the same patch of RBWH and start collecting pollen or nectar again.

This is not constancy but is instead termed fidelity.

Both fidelity and constancy have consequences for plant pollination. It’s therefore unsurprising to discover that some plants have evolved to influence these foraging habits of bees … which is where the coffee comes in.

Continue reading

More droning on …

Synopsis : Drones are now being evicted from colonies. How and why does a honey bee colony regulate drone numbers?

Introduction

Over the course of the last eight years posts on The Apiarist have got longer. This year, posts are now five times the length of the 2014 average. I’ve written – and hopefully, you’ll have already read – more words this year than are in The Hobbit.

If this continues until the end of the year we’ll have exceeded the word count in Tolkien’s The Two Towers.

This is probably unsustainable 1.

The increase is explained in part by the complexity of some topics. It’s compounded by the need to provide some contextual information … and by my prolixity 2. The latter is unavoidable, the former is probably necessary, not least because of the significant churn in new beekeepers.

A topic needs to be introduced, explained, justified and concluded.

Without this contextual information a post on oxalic acid trickling could be just:

5 ml of 3.2% w/v per seam when they’re broodless.

And where’s the fun in reading that?

Or writing it?

Furthermore, it’s probably of little use to a beginner who might not know what w/v means. Or what a seam is … or for that matter why being broodless is critical.

Keeping it topical

To maximise the income from site advertising I need to keep readers returning. This means the choice of topics should be important.

However, although some topics are chosen because they’re key concepts in the art and science of beekeeping, the majority are picked simply because I find them interesting.

And this week is one of the latter as I’m going to be droning on about … drones.

Specifically about drone numbers in the colony.

This was prompted by seeing the first drones of the season turfed out of the hive.

Dead drone at hive entrance

Another one bites the dust

Seeing this coincided with me discovering an interesting paper on how the queen’s laying history influences whether she produces drone or worker brood. This, inevitably, led me to other papers on drone production and discussions of how the colony controls drone numbers 3.

Drones are topical now because their days in your colonies are limited.

Already the colony will be producing significantly less drone brood than three months ago. The drones the colony has already produced will still 4 be flying strongly on good days.

However, when in the hive they will be being increasingly harassed by the workers.

Herding drones

If you open a colony very gently in the next few weeks 5 you might find the corners of the box contain high numbers of drones. The photo above was taken in late August and I’ve seen it several times late in the season. My interpretation is that it’s the only location in the hive in which drones can escape harassment by the workers.

Drone eviction

Drones ‘cost’ the colony a lot to maintain. A drone consumes about four times the amount of food than a worker (Winston, 1987). Therefore, once the fitness benefit of keeping drones falls below the expected costs needed to keep them they become ’surplus to requirements’. At this point the workers turf them out of the hive.

Evicted drones cannot feed themselves, so they perish.

It’s a tough life.

Interestingly, workers preferentially evict old drones. Presumably younger drones are more likely to fly strongly and mate with a virgin queen. Additionally, sperm viability in older drones is reduced, so their genes (and therefore those of the colony) are less likely to be passed on.

This ‘cost’ of maintaining drones is influenced by both the colony and the environment. For example, queenright colonies (which, by definition, have less need for drones) evict more drones than queenless colonies in the autumn, as nectar becomes limiting.

Although most beekeepers associate drone eviction with late summer/early autumn it also occurs when nectar is in short supply e.g. during the ‘June gap’.

It has also been suggested that drone eviction rates are related to colony size. Small queenright nucs, which have less need for drones, are more likely to evict than a full colony.

There’s still a lot we don’t know about drone eviction. For example, since drones tend to accumulate in queenless colonies, do these preferentially evict related drones to maintain potential genetic diversity in the population? 6

Hannibal the cannibal

Allowing an unfertilised egg to hatch, feeding the larva, incubating the pupa to emergence and then maintaining the resulting drone is a waste of resources if conditions are not appropriate. For example, doing this during a nectar dearth – particularly when drones are unlikely to be required for mating – makes no sense 7.

Therefore, in early spring and late autumn, workers cannibalise developing drone larvae. Effectively they are recycling colony resources. They preferentially cannibalise young larvae rather than older larvae. This makes sense as young larvae are going to need more food to reach maturity.

As above, queenless colonies cannibalise less queen-laid 8 drone larvae than queenright colonies.

In addition, some studies have shown that colonies with abundant adult drones cannibalise a greater proportion of developing drones. Again, this makes reasonable sense. Why rear more if you’ve got enough already?

However, to me it makes ‘reasonable’, but not ‘complete’ sense. Drones being reared as larvae are genetically related to the colony, adult drones may well not be. Drones that have drifted in from adjacent hives may therefore reduce the likelihood of the colony passing on its genes under environmental conditions which favour larval destruction but not eviction of adult drones.

Someone needs to look into this in a bit more detail 😉 .

There are lots of other aspects of larval cannibalism that are not understood. For example, how do workers discriminate between drone and worker larvae? Can they – as the queen can – measure the cell dimensions? Drone and worker brood pheromones differ from day 3 or 4. This seems a bit late to account for the cannibalisation of young larvae?

The influence of the queen

Since workers may cannibalise developing larvae 9 at different rates (drone vs. worker) it’s necessary to measure the colony’s egg sex allocation to see how the queen may influence drone numbers.

Only a few studies have done this …

There are experiments that suggest (they’re not definitive) that queens in continuously fed colonies lay more drone eggs in spring and summer than in autumn. This implies that day length or temperature may influence the queen, but it could also be a response to colony strength i.e. the queen lays more unfertilised eggs in a colony increasing in size, than in one decreasing.

In addition – and this is where I started down this rabbit hole in the first place – the egg laying history of the queen influences her current egg laying activity.

This easy-to-understand study was conducted by Katie Wharton and colleagues (Wharton, 2007). They confined queens for a period on either drone (DC) or worker comb (WC) – ensuring the queen could only lay drone or worker eggs for 4 days. They then transferred the queens to frames containing a 50:50 mix of drone and worker comb and recorded the amount of drone or worker eggs laid over 24 hours.

WC queens laid more drone eggs but the same amount of worker eggs as DC queens

There was a marked difference in the egg laying activity of the DC or WC queens when given the choice of laying drone or worker eggs. Although both the DC and WC queens laid similar amounts of worker eggs, the WC queens produced significantly more drone eggs as well.

Egg laying history or drone brood quantification?

This is a good study. The authors controlled for a variety of factors including season, colony size and food availability.

They additionally excluded the possibility that the egg laying activity of the queen was influenced by preferential cleaning of particular cells types by the workers, or by the workers backfilling certain cell types with nectar.

Finally, Wharton and colleagues allowed the colony to rear the eggs laid to pupation. The bias already observed was retained i.e. colonies headed by WC queens reared significantly more drone pupae than those headed by DC queens. The workers did not ‘correct’ the negative feedback exhibited by the WC queen, for example by preferentially cannibalising drone brood.

Although I termed this the ‘egg laying history’ of the queen a few paragraphs ago there is another interpretation.

The worker or drone comb already laid up by the queen – during the 4 day confinement period – remained in the colony. It’s therefore possible that the egg laying activity of the queen was influenced by the amount of drone brood already present in the colony.

Either explanation is intriguing.

How does the queen ‘count’ the number of drone or worker eggs she’s laid in the recent past? Alternatively, how does she quantify the amount of drone brood in the colony?

But what about the workers?

The Wharton study largely excluded the possibility that there was preferential cleaning of drone or worker cells by the workers in the hive. In fact, earlier studies have indicated that cell cleaning continues almost constantly and workers were equally likely to clean worker or drone cells.

The Wharton study also addressed – and excluded – the possibility that differences in the backfilling of drone or worker cells might influence egg laying.

However, it’s not understood what determines whether drone cells get backfilled with nectar by workers. My colonies are starting to do this now. Do the workers fill drone cells with nectar because the queen hasn’t laid in these cells or because they only backfill drone cells late in the season?

The former suggests that there is some sort of competition between the egg laying activity by the queen and nectar storage by workers. In contrast, the latter suggests that there are environmental triggers that influence this worker activity.

Or both … of course 😉 .

Comb building

In contrast to some of the studies outlined above, comb building is easy to monitor and – perhaps consequently 10 – has been well studied.

I’ve already discussed comb building in a recent post about queenless colonies. These preferentially build drone comb (Smith, 2018).

What else influences drone comb production?

Probably the two strongest determinants are the amount of drone comb already in the nest and the season.

Drone comb production is reduced in colonies that already contain lots of drone comb. Many beekeepers never observe this as they only use frames containing worker foundation. The workers squeeze in little patches of drone comb – often in the corners of the frame – but it never exceeds 5% of the total.

Colonies often prefer to build drone comb when given the choice

In contrast, natural nests contain 15-20% drone comb. That’s equivalent to two full frames in a National-sized hive. Once drone comb approaches this level a negative feedback loop operates and the workers build less drone comb. The negative influence of this drone comb (on building more drone comb) is enhanced if the comb contains drone brood.

Colonies drawing comb now (and certainly in the next month or two) will build worker comb. Some beekeepers exploit this to get lovely new worker frames drawn – nucs are particularly good at this 11. In contrast, drone comb is drawn in spring and early summer. The season – presumably day length and temperature – therefore influences drone comb production, and hence drone production.

A thousand words

Well, nearer 2000.

As we near the end of the season we start to see drones evicted from our colonies. It’s interesting to think about the interplay of events that resulted in the colony producing those drones in the first place … and how and why the colony regulates drone production throughout the season.

Wharton (2007) neatly summarised the five stages from comb building to adult drone eviction.

Drone production and maintenance in a honeybee colony

I’ve dealt with these in reverse order because that was the best fit with the photo of the dead drone on the landing board that I started with.

There’s a lot we still don’t understand about the regulation of drone numbers. In particular, I think the majority of studies have ignored the influence of adult drone numbers on any of five stages illustrated above.

This is an important omission as drones move more or less freely between hives. That means that adult drones may well be genetically unrelated to the colony.

Perhaps this means that adult drones do not influence drone production? After all, if they did negatively influence drone production – as suggested above – it would potentially limit the ability of a colony to reproduce its genes. Evolutionarily this doesn’t make sense (at least, to me).

There are a couple of studies that have tried to determine the influence of adult drones, but they have produced conflicting results. Rinderer (1985) added drones to a colony which consequently reduced drone brood production. However, Henderson (1994) did the opposite and showed that removal of adult drones had no effect on drone brood production.

There’s clearly lots more to do …


Notes

I wrote this late on Thursday night. While doing so I watched the page views of my four year old post on Mad honey go ‘off the scale’ (which for a beekeeping site means hundreds of views per hour). The interest wasn’t sparked by my erudite description of grayanotoxin intoxication. Instead it was related to a video of a ‘stoned’ Turkish brown bear cub rescued after eating honey produced from rhododendron nectar.

It’s now abundantly clear that if I want to maintain my outrageous advertising income I should probably write more about hallucinogenic honey and less about the evolutionary subtleties of honey bee sex ratio determination.

That’ll teach me 😉

References

Boes, K.E. (2010) ‘Honeybee colony drone production and maintenance in accordance with environmental factors: an interplay of queen and worker decisions’, Insectes Sociaux, 57(1), pp. 1–9. Available at: https://doi.org/10.1007/s00040-009-0046-9.
Henderson, C.E. (1994) ‘Influence of the presence of adult drones on the further production of drones in honey bee (Apis mellifera L) colonies’, Apidologie, 25(1), pp. 31–37. Available at: https://doi.org/10.1051/apido:19940104.
Rinderer, T.E. et al. (1985) ‘Male Reproductive Parasitism: A Factor in the Africanization of European Honey-Bee Populations’, Science, 228(4703), pp. 1119–1121. Available at: https://doi.org/10.1126/science.228.4703.1119.
Smith, M.L. (2018) ‘Queenless honey bees build infrastructure for direct reproduction until their new queen proves her worth’, Evolution, 72(12), pp. 2810–2817. Available at: https://doi.org/10.1111/evo.13628.
Wharton, K.E. et al. (2007) ‘The honeybee queen influences the regulation of colony drone production’, Behavioral Ecology, 18(6), pp. 1092–1099. Available at: https://doi.org/10.1093/beheco/arm086.
Winston, M.L. (1987) The Biology of the Honey Bee. Cambridge, Massachusetts: Harvard University Press.

Battlefield bees

Synopsis: For millennia bees were used as weapons of war. They are now being developed as ‘weapons of peace’ to help clear the millions of anti-personnel mines left after active conflict ends. Their legendary scent detection abilities combined with high-tech ‘bee detection’ methods show promise and may help reduce the thousands of civilian casualties that occur decades after the war ends.

Introduction

The weekly posts on this website are about bees and beekeeping.

In the same way that I deliberately shun sponsorship and avoid advertising 1 I also try and avoid politics, law and other divisive subjects. These cause enough problems without adding my opinions into the mix … and worse, having to moderate the opinions of others in the subsequent comments.

So, although I might write about the detrimental effects of thiamethoxam on bees, I would focus on the science of neurotoxins rather than politics of lifting the ban of the use of this neonicotinoid in the UK .

It is harmful to bees, but the aphid-transmitted yellows viruses may otherwise decimate our sugar beet crop, and might not the alternative pesticides be more harmful to bees?

Do we even need that much sugar beet?

And what about the livelihoods involved?

You can see how quickly it gets very messy.

But, at the same time, I strive to make posts relevant and even topical. When viewed retrospectively – even if just by me – the posts represent a snapshot of my jumbled thoughts of what’s current at the time.

There are ageing posts on oxalic acid treatment that don’t reference ApiBioxal (the good old days as I like to call them), and others that make passing mention of maximising the oil seed rape (OSR) crop, or processing OSR honey 2.

All of which means I cannot really avoid mentioning the recent Russian invasion of Ukraine 3.

Six-legged soldiers and one-legged civilians

Two years ago I wrote a post about the use of bees in warfare. For many hundreds of years they were an effective tactical weapon to be dropped on or fired at the enemy.

Not individually – that would be just silly – but a hive at a time.

I’ve been present when someone has dropped a full brood box. The whirling cloud of angry bees was reminiscent in shape, though nothing else, to the mushroom-shaped cloud over Hiroshima. Lots of people got stung in the training apiary during that session.

Unfortunately, the ingenuity of man knows no bounds when it comes to killing and maiming others.

The thermobaric weapons employed today are very different from the torsion-powered ballista siege engines used by the Ancient Greeks over two millennia ago 4.

Soldiers now wear carbon and kevlar rather than Corinthian helmets and short-sleeved tunics.

Although you can buy a camouflaged bee suit, it’s not designed for the battlefield, and is likely to be about as much use as the hoodies, jeans and T-shirts that the innocent civilians inevitably caught up – or deliberately targeted – in today’s wars are wearing.

The legacy of war

And long after the battle has ended – years or even decades later – those surviving civilians continue to be maimed and killed by unexploded ordinance and, particularly, by anti-personnel mines.

Minefield sign, Cyprus

So, rather than dwell on the horrors of the present – which is about as far removed from beekeeping as it’s possible to get – I’m going to discuss some more hopeful stories of how bees might help reduce the deaths and injuries caused by anti-personnel mines.

Those readers expecting the (un)usual humour may be disappointed this week … this is not a topic that lends itself well to jokes.

The solutions that scientists are developing to detect anti-personnel mines use a clever combination of the truly awesome scent detection capabilities of honey bees coupled with some very clever technology.

The problem

In 2021 there were 61 countries which were ‘contaminated’ with anti-personnel mines. These mines are typically produced for a few dollars, are perhaps 30 cm in diameter and are buried just sub-surface … and often forgotten.

One definition of the word minefield is ‘a situation or subject presenting unseen hazards’. The plural hazards, and use of ‘field’, indicate that lots of anti-personnel mines are usually buried across an area, thereby rendering it too dangerous to enter.

Farming, trade and communication is inhibited as people have to avoid the minefield(s) … and, if they don’t then the consequences can be devastating.

The International Campaign to Ban Landmines recorded over 7,000 casualties in 2020, 2,500 of whom were killed.

80% of these casualties were civilians and – where the age was known – over 50% of the casualties were children.

Aside from banning the use of anti-personnel mines in the first place, a priority must therefore be to find and removes mines from these areas.

Typically this involves using metal detectors or sniffer dogs for detection, followed by manual clearance. Inevitably, there are considerable risks involved in both the detection and – to a lesser extent – the clearance. These risks make the process time-consuming and expensive.

Which is where honey bees come in …

”I love the smell of Napalm in the morning”

So said Lieutenant Colonel Kilgore (Robert Duvall) in the film Apocalypse Now.

Landmines don’t contain Napalm, but about 90% if them contain TNT (trinitrotoluene).

And not only does TNT have an odour, but landmines continue to emit this odour for years after they have been buried. On a calm day there are vapour plumes of TNT above each of the buried mines 5.

The vapour concentration of TNT is measured in parts per trillion (pptr) and is usually in the range 0.01 – 100 pptr.

Just as the usual way to express areas is by comparison to Wales 6, the ‘volume comparison’ is typically made to an Olympic swimming pool.

1 part per trillion is about the same as half a teaspoon added to an Olympic swimming pool. Not very concentrated …

… but well within the detection capabilities of honey bees. For comparison, this is similar to that of sniffer dogs 7.

To cut a long story short, scientists 8 trained bees to feed on syrup laced with trace quantities of TNT. They then tested the ability of these bees to detect targets emitting field-realistic amounts of TNT.

The results were very encouraging. 97-99% of targets were detected with 1-2.5% of false-positives.

More importantly, the false-negatives (targets that were missed) were less than 1%. It’s much more important to not miss any than to ‘find’ some that aren’t there.

Lidar

The authors neatly sum up the benefits and principles of the study:

Bees do not cause mines to explode, do not require a handler, and can be trained more rapidly than dogs. This technique makes use of the natural foraging behavior of bees, which frequently cover ranges up to several km around a hive. The bees identify the sample location by their increased dwell time while flying in its vicinity.

And it’s that last sentence that should give you pause for thought.

How do you detect the “increased dwell time” – a fancy term meaning spending more time flying in one small area than the remainder of the study area – if you’re trying to find mines in an area about the size of a couple of football pitches 9.

Remember, as if you’d forgotten, you cannot enter the minefield because of those unseen hazards that are lurking just under the surface waiting to blow your legs off.

Bees are pretty small. I can see them against the clear blue sky at 20-30 metres range, but I can’t see them against mixed foliage at anything like that distance.

One potential solution is light detection and ranging (lidar) technology. Essentially this involves shining a scanning laser across an area and detecting the light scattered when it ‘hits’ objects – such as flying bees 10. For additional discrimination, changes in the polarisation of the scattered light has been used to distinguish between bees and what the scientists termed ‘clutter’, which I take to mean foliage.

And it works.

Lidar detection of bees; a) bee heatmap, b) chemical detection (5 is a false positive), and c) visual mapping of bees.

Not only in theory but also in practice … lidar has been used to detect bees detecting mines in an active ‘minefield’. Mines were buried in known locations, trained bees were released and their ‘dwell times’ were recorded using lidar (with the detector about 80 metres away).

Football fields and minefields

But there’s a problem.

Lidar involves a laser scanning horizontally 30-60 cm above the ground. Anything lower than this and the foliage prevents accurate (or any) detection of the bees.

And, even though a minefield might be the size of a football field, it doesn’t look like a football field … either when mined in the first place and definitely not after a few years.

Minefield in the Golan Heights

Unfortunately, there’s also an additional problem.

Tragically the sign above was probably erected after someone inadvertently stepped on a mine. Until that fateful day it might have just been ‘that scrubby bit of field bordering the river’.

Detecting the location of mines in a minefield is one problem.

Detecting whether a field is a minefield is a different – albeit related – problem.

And it turns out that bees might be able to help us discriminate between minefields and football fields, or any other sort of equally harmless fields.

I’ll discuss this before returning to the detection of individual mines in a minefield.

REST (and be thankful)

REST is an acronym for Remote Explosive Accent Tracing 11 .

Since those buried anti-personnel and landmines give off a vapour plume of TNT there are methods of sampling the air and testing it for the presence of trace amounts of explosives.

The ‘sampling’ is highly technical and outside the scope of this post 12.

The ‘testing’ involves sniffer dogs and lots of doggy treat-type rewards. Consequently it is a time-consuming and therefore costly procedure.

However, honey bees are covered in tiny hairs. Through electrostatic interactions, these pick up molecules while they are out foraging. Graham Turnbull 13 and colleagues have shown that flying bees can pick up molecules of TNT (from buried mines) which can subsequently be detected.

The bees are therefore used for wide area sampling, but how is the TNT detected? Graham is a physicist whose speciality is organic semiconductor sensing films … essentially thin films that change fluorescence when certain chemicals are deposited on their surface.

The hive entrances were modified so that the bees passed through a tube. This was made of a special material to pick up – and since lots of bees were making the trips, to also concentrate – the molecules that adhered to their hairs whilst out foraging.

Quenched photoluminescence (red bars) compared to negative control areas (black line)

To cut another long story short, the concentrated molecules were then transferred to the semiconductor sensing film and the photoluminescence quantified. A reduction in the photoluminescence (quenching) was indicative of TNT detection 14.

So honey bees can be used to discriminate between minefields and, er, other fields.

Detecting mines not minefields

I should add that the bees used in the study above were not trained 15 in any way. They simply placed feeders on the opposite side of the uncontaminated test or mined field to encourage them to sample in a reasonably defined area. The bees were not searching for mines, or the TNT vapour plumes, they were just flying back and forth ‘doing their stuff’ and foraging.

Bees are fast learners and can – as described briefly above – readily be trained to associated particular scents (such as TNT) with rewards (i.e. syrup). When you release these trained bees into an area with TNT vapour plumes they home in on these looking for the rewards … and hence exhibit those previously mentioned ’increased dwell times’ near buried mines.

But there’s still the problem of how the bees can be detected.

Huge advances have been made in unmanned aerial vehicles (UAV’s or drones), GPS and video technology in the 15+ years since the original use of lidar to detect bees detecting landmines.

The combination of these technologies now provides a way to detect individual bees, and consequently anti-personnel mines, within an area.

Using drones to monitor bees

A drone flying 10 m above the minefield 16 was used to record high resolution video from which the locations of individual bees could be (computationally) determined.

By detecting bees on individual frames (from video taken at 45 fps), rather than tracking each bee, it was (again computationally) relatively straightforward 17 to generate spatial density maps showing where the bees preferentially concentrated.

Video stills (left) and bee location heat maps (right). The blue circles show mine locations. Bee numbers on scale.

The accurate spatial location was ensured by using a modification (RTK) to GPS which involved an additional ground base station. This increased the standard GPS resolution to provide a horizontal accuracy of 5 cm. The bright coloured foci of ‘increased dwell times’ were less than 0.5 x 0.5 m.

Conclusions and problems

These studies are encouraging. They suggest that a biohybrid system 18, combining advanced physics and area-wide sampling of bees, with the exquisite scent detection of trained foragers coupled with highly accurate video monitoring, might help reduce the number of victims of landmines that occur long after the original conflict ended.

However, there are a few problems that remain to be resolved.

Bees learn fast, but they also forget fast. The authors developed some reinforcement training exercises to ‘remind’ them that they were searching for things scented with TNT.

Bees are not ideal chemical biosensors. They are potentially easily distracted by a strong nearby nectar flow, they don’t forage 19 in poor weather and their availability might be seasonal, depending upon the latitude.

Drones have limited flight times and long vegetation still impedes accurate video detection. Either the bees fly at lower altitudes or the foliage is disturbed by rotor downwash and so increases background noise.

Nevertheless, the expected costs and time involved in both the wide-area sampling and mine detection are lower, and the mine detection per se is much safer.

The mines still have to be destroyed, but knowing precisely where they are – and where they are not – is much more than half the battle … in solving the lethal legacy of a possibly long-forgotten battle.


 

Radar love

The average beefarmer in the UK is probably somewhere in their mid-60’s 1. This means that in 1973, when the Dutch rock band Golden Earring had their only notable chart success Radar love, they were about 18.

Bear with me …

As 18 year olds they probably wore denim flares and loud shirts with spearpoint collars. They would go to the local disco to meet similarly-attired members of the opposite sex (whose shorter hair may have been their only distinguishing feature).

They knew when and where to meet … the weekly Saturday night (obviously 2 ) disco.

There was no point in turning up at 10 in the morning … the disco was closed 🙁

Similarly, despite their ‘cool threads’, wearing them to the launderette would have resulted in almost certain disappointment … the dance partners they were seeking weren’t likely to be found doing the laundry 🙁

No, the disco was the place to go. 

Radar love would have been on the playlist. It reached the top 10 in the charts in many countries.

Hold that thought … we’ll return to Radar love in a few minutes … 3

The birds and the bees

Of course, these young beefarmers didn’t just go to the disco to dance

Oh no.

They had an ulterior motive 😉

They knew that they had a good chance of meeting a like-minded (and similarly attired) member of the opposite sex who was also ‘looking for love’.

These meetings were effectively ritualised … a particular time and place.

Let’s forget the bell bottoms and hippie shirts now … I only added that detail so that any readers who know an ageing beefarmers can have a little giggle imagining them dressed for the disco 😉 

OK, back to the disco … metaphorically.

The disco is not fundamentally dissimilar to the lek used by male grouse 4

Greater sage-grouse at a lek, with multiple males displaying for the less conspicuous females

A lek is defined as a location where males congregate to compete and mate with females. Importantly, there are no direct benefits – such as food or territory – that the females gain from attending the lek 5.

How do the males know where to congregate?

Grouse tend to live for several years 6. Older grouse know where the lek is because they attended last season. Juveniles probably tag along and learn from their elders despite the fact they are too immature to mate, or lack the social dominance (or plumage 7 ) to compete.

As a consequence of this male hierarchy the location of the lek is invariant.

The birds congregate at the same place each year.

One of the features of leks is that males show high levels of fidelity to a single lekking site.

So now we know something about the birds … what about the bees?

Drones congregate in particular – rather ill defined – landscape features called drone congregation areas (DCA’s).

These, like a black grouse lek, are stable from day to day and year to year.

The drones compete (for the queen, though not directly with each other by displaying) and offer the queen no territorial or food benefits … meaning that DCA’s are effectively insect leks 8

Drone congregation areas

There are studies going back well over 50 years on DCA’s. There are no hard and fast rules that define their location (at least to humans … thankfully virgin queens have no problems finding them). However, you can sometimes hear them; they sound like a small swarm, the noise caused by thousands of drones circling 5-40 metres above the ground in a swirling, traffic cone-shaped, perhaps a 100 metres or more in diameter.

How do drones know where to congregate? There is no male hierarchy 9. An individual drone lives for just a few weeks and perishes before winter. 

The location must be somehow ‘hard-coded’ in the environment. Effectively a set of features that – once located – attract the drones back repeatedly until they either mate with a queen, or die trying 10.

Many studies have attempted to identify DCA’s – geographic features on the ground, sheltered from strong winds, a dip in the horizon etc. These have tended to produce rather mixed results.

I don’t think we’re anywhere close to being able to point to an intersection of two hedges and say “Over there … that’s where drones will congregate”.

An alternative approach is to go fishing for DCA’s.

Literally. 

Having identified a number of potential DCA’s from landscape analysis, you can dangle a virgin queen from a helium balloon and sample the drone density in each of the areas.

It sounds a lot simpler than it is … there’s a nice account by Aude Sorel in Bee Culture if you’re interested.

By definition, the drone congregation areas are the ones you trap the most drones in.

Right?

Well, possibly not.

Perhaps the very method used to sample the drones attracted them there in the first place? 

It’s been known since the 1960’s that high concentrations of queen mandibular pheromone can attract drones to almost any location – in one notable example, even 800 metres out to sea 11.

If you use bait, how can you be certain that the areas you define are ‘real’. 

A better way to define a DCA would be to observe individual drones accumulating in a particular area … to watch them leaving the hive, fly the tens or hundreds of metres to the same place they flew to yesterday, and record them ‘strutting their funky stuff’.

Have you ever tried to follow a drone in flight?

They’re strong and fast. They need to be to outcompete other drones when chasing the queen.

It’s almost impossible to track them across the apiary, let alone over the hedge, across two fields and into the lee of a copse.

But scientists can now do exactly that … using a technique called harmonic radar tracking.

The title of this post should now make a bit more sense … it’s the use of radar to find where drones go ‘looking for love’ 😉

Harmonic radar tracking

A harmonic radar system emits a stimulus signal. This signal is picked up by a harmonic tag (the transponder) which uses the low frequency stimulus energy to generate a second harmonic which is then re-radiated back out to a receiving system.

The harmonic signal emitter/receiver is portable … if you’ve got a lorry.

Harmonic radar emitter and detector – with Rothamsted Manor in the background.

Fortunately, the transponder is tiny … small and light enough to be glued to the back of a bee.

Drone with harmonic radar transponder attached.

Harmonic radar has been used to study orientation flights in honey bees 12, to track Asian hornets, and to follow butterfly flight paths 13 (amongst other things).

And now it’s been used to map drone congregation areas by tracking the flights of individual drones from the hive.

Harmonic radar is a relatively short range system. You can’t track transponder-tagged insects flying miles away. The effective range is just a few hundred metres for most systems.

However, for drone congregation areas this shouldn’t be a major limitation. Drones generally fly shorter distances to mate than queens (an evolutionary mechanism to avoid inbreeding) and DCA’s have often been found near to apiaries 14.

Tracking drones by harmonic radar

The study, by Woodgate et al., was published a couple of weeks ago in iScience. The full reference is:

Woodgate et al., Harmonic radar tracking reveals that honeybee drones navigate between multiple aerial leks, iScience (2021), https://doi.org/10.1016/j.isci.2021.102499

It’s available under open access (i.e. free, for anyone) and I recommend you read it if you’re interested.

I’m just going to pick out a few highlights.

During two sequential seasons the authors tracked over 600 flights by at least 78 drones. These included 19 first flights (orientation flights) and – for four drones – 6-8 consecutive flights, including their first ever orientation flight.

Orientation flights were typically observed as multiple loops in different directions, centred on the hive from which the drone originated.

Drone orientation flights

The average duration of these orientation flights was ~13 minutes and the drones observed only took one or two before changing their flight pattern (see below) and seeking drone congregation areas.

Worker bees typically take more (~6) orientation flights than drones. Presumably foragers need to ‘map’ the hive location better because they may end up returning to it (and they’ve failed if they don’t) from any location.

As we’ll see in a minute, drones tend to use particular ‘flyways’ which are probably determined by landscape features. Drones also may return to a different hive to the one they set out from.

Identifying drone congregation areas by harmonic radar tracking

Scientists love ‘heat maps’.

These are a graphical way of depicting levels of activity of one kind or another.

If you overlay the flights by every transponder-tagged drone in each of the two years of this study you generate a map (like C and E shown below). In this study they used a ‘white to red’ scale where the paler the colouration, the more drones were detected in that particular point on the map.

You can easily see the hive location (points 1, 2 and 3) as all flights originated there.

Heat map of the landscape used by drones.

Actually, C and E are a bit confusing because they include the orientation flights which are centred on the hives. If you exclude these you end up with the heat maps D and F on the right.

From these the authors could detect particular areas where the drones tended to concentrate … these are proposed to be the drone congregation areas. There were four within range of the harmonic radar system – A-D above (confusingly labelled on images D and F).

There are a few obvious features of these proposed DCAs:

  1. They are in approximately (but not exactly) the same position in the two study years.
  2. The frequency with which they were visited changes. A is visited less frequently in the second year (panel F) than in the first (panel D).
  3. The most distant DCA (at least that could be mapped in this study) was ~600 metres from the hive. 
  4. Each DCA had a roughly symmetrical ‘core’ of 30-50 metres, significantly smaller than many drone trapping studies suggest..

One thing that was noticeable by comparison of the orientation flights and the proposed DCAs was that they did not overlap.

So how do the drones ‘find’ the DCA if they don’t discover them on an orientation flight?

Flyways, straight and convoluted flights

Heat maps are cumulative data.

It was also possible to look at the individual flight paths of drones on their way to and from a DCA (in exactly the same way as they mapped orientation flights).

Analysis of these showed that drones adopted two distinct types of flight – an approximately straight, direct flight interspersed with periods of convoluted, looping flight. There are lots of pictures of these in the paper but, rather than showing another published image, here’s my “no expense made spared” diagram of these two patterns of flight.

Drone flight paths showing distinct direct and convoluted elements.

The convoluted flight defines the drone congregation areas. In these the drones showed very distinctive behaviour – the further they were from the centre of the DCA the more strongly they accelerated back towards the centre. 

Drone flight paths (inevitably) overlapped in DCAs.

However, they also overlapped in the straight line flight. Drones tended to use particular flyways from the hives to, and between, the DCAs.

Scientists have previously identified (or at least suggested the existence of) these flyways that drones use to travel to and from the hive and the DCAs 15

However, what they had previously not identified was that drones often visit more than one DCA in a single (potential) mating flight.

In 20% of the flights analysed drones visited more than one DCA. 

Finally, drones tended to only spend about 2 minutes flying around very fast (at ~5 m/s rather than the sedate ~3 m/s they fly around the hive at 16 ) within the proposed DCA.

This suggests that drones might routinely patrol several DCAs in a single flight, moving on unless a queen is present.

Harmonic radar mapping the flights of virgin queens

I’ve often preceded the term ‘drone congregation area’ in the text above with the word ‘proposed’. A DCA has a very specific meaning that describes the places where drones congregate to attempt to mate with a virgin queen.

None of the studies above showed queen mating, or even the presence of a queen.

But, of course, the authors tried that as well.

They transponder-tagged queens (94 in total) and tracked their orientation flights and mating flights (26 in total). The orientation flights were remarkably similar to those of the drones; the average number of these flights was 3 and no queen went on more than 6 orientation flights.

Unfortunately the tracking of queen mating flights was less successful 17.

Queens flew out of range (I’ll return to this shortly), the transponder fell off, or parts of the flight were not picked up by radar. Some of the queens ‘followed’ (or for which tracking was attempted) did get mated, but not apparently in the DCAs identified during the flight tracking of drones.

This type of study clearly needs further work …

Conclusions

Drone congregation areas could be detected using harmonic radar tracking of transponder-tagged drones. Unlike other well-studied lekking areas, males (drones) did not display lek fidelity, but instead visited several in rotation 18.

The DCAs are a consequence of drones exhibiting a convoluted flight pattern in particular locations. The conservation of the flyways – the routes taken by the drones – between DCAs suggest they might contribute to the location of the DCAs.

Understanding what defines these flyways might allow better prediction of DCA locations.

Previous studies have shown that queens tend to fly further to DCAs than drones, presumably to avoid inbreeding. One possibility is that tagged queens in this study might have been more likely to visit the four DCAs identified if they were placed in mating nucs situated further away from this study site.

But, of course, they could have then flown off in a different direction altogether 🙁

Finally, it’s worth noting that a different pattern of queen mating activity had been described for dark, native (Apis mellifera mellifera) and near-native bees. This is apiary vicinity mating (AVM), and is nicely described by Jon Getty on his website

I now have some native black bees. I’m also experiencing the worst spring of my entire beekeeping career for queen mating. I am increasingly interested in AVM as a mechanism for saving the queen from drowning or freezing to death while attempting to reach a DCA 🙁