For the past eight years or so, I’ve been lucky enough to be one of the few people in the world who can follow a bee as it flies through the landscape. I have a longstanding interest in how small-brained insects can efficiently obtain and process information about the world to allow them to move around the world without getting lost; and in how they use the impressive abilities we’ve uncovered to forage, breed and survive. Most studies into these questions have to be done at small scales, often in unrealistic laboratory conditions, because flying insects move too fast, over long distances and often over difficult terrain, making it almost impossible for humans to keep track of them in their usual habitats. To try to get more accurate information about how bees move around large-scale environments, I have been collaborating with Rothamsted Research as the primary researcher using one of only a very few entomologic harmonic radars in the world.

 

The basic principle behind radar is that you use a powerful radio transmitter to “illuminate” the area in front of you with a beam of radio waves. Just as a torch-beam of visible light bounces off objects and back to your eyes, where your retina and brain interpret the signals to work out what objects the light was reflected from, the radar signal bounces off any objects in its path allowing us to look out for reflections. Because radio waves travel at the speed of light, measuring the time between emitting the signal and receiving the reflection allows us to calculate how far it travelled and thus the range of the target it bounced off. If you scan back and forth with the radar, you can also work out the position of the target. This works very well if you are trying to detect targets in an uncluttered environment, like planes in the sky or ships in the open ocean, but flying insects like bees have an inconvenient habit of flying around in landscapes cluttered by trees, hedges, buildings and so on. Trying to detect the reflection from a tiny bee among all that clutter is like looking for the reflection of a needle in a haystack.

 

That’s where the “harmonic” part of harmonic radar comes in. When I want to track a bee, I first catch it and attach a little piece of tech called a transponder to its back: the transponder looks like a little aerial and is very lightweight: it weighs less than the pollen a foraging bee typically collects to bring back to the nest. When our outgoing radar signal hits the transponder, it absorbs some of the energy and uses it to emit a signal of its own and double the original frequency. Our detector is tuned to look out for this “harmonic” signal, so we can safely ignore any reflections with the original frequency and only pay attention to the harmonic frequency. In theory, the transponder should be the only thing in the landscape that is emitting a signal at this frequency, although in practice there are quite a few sources of unwanted noise that need to be filtered out.

 

Considering that honeybees are widely kept by humans across the world – many people consider them a domesticated species, although there is some debate about exactly what counts as “domestication” – it may be a surprise to learn that some crucial aspects of their mating behaviour remain mysterious to us. You probably know that a honeybee colony has a single queen who is the mother of all the tens or hundreds of thousands of female workers who perform all the work of the colony, from building honeycomb, to taking care of the larvae, to foraging for pollen and nectar. Male honeybees are called drones and have just one purpose in life: to mate with a queen. They cannot even feed themselves without assistance from the workers and they die as soon as they mate. New queens are born only if the old queen dies or when the colony is large and successful enough to split into two, and they mate only during a small number of “nuptial flights” undertaken early in life. Drones are typically produced in larger numbers and undertake several flights each day, trying desperately to find a queen to mate with. Meanwhile, the nuptial flights are the riskiest time, not only for individual queens but for their entire colonies which will die out without a mated queen to lay eggs. So, it’s of paramount importance that drones and queens find each other efficiently and safely. Bees mate in flight, making it almost impossible to observe them, so we still have huge gaps in our knowledge about this critical juncture in the honeybee life cycle.

 

Where do drones actually go and what do they actually do during their daily excursions? Since at least the 1920s, many scientists and beekeepers have believed that they congregate in large numbers in particular areas, known imaginatively as drone congregation areas, or DCAs. The first hint came from people who heard loud buzzing in certain locations and concluded that huge numbers of bees must be flying above their heads, too high to be seen. Later, scientists sent up balloons with queen bees dangling from them, or just cotton wool soaked in the scent of a queen, and observed that they would acquire little “comet tails” of drones, attracted by the queen or her pheromones. When lots of drones gathered in certain places but few or none in other parts of the landscape – and amazingly, they could attract lots of drones in these particular areas year after year – the obvious answer seemed to be that huge numbers of drones must be hanging out in a few, select places, presumably waiting for a queen to come along and mate with them.

 

This DCA hypothesis seems to be pretty well accepted by scientists nowadays (although many beekeepers I’ve met were still sceptical), so I was surprised, when I dug into the published research, to find that the picture was much less clear. There are two main problems with the picture of drone activity based on traps baited with pheromones or queens: first, it’s been shown that drones will repeatedly return to a place they’ve previously encountered a queen or her scent, so there is a risk that researchers are accidentally creating the very phenomenon that they are trying to investigate and attracting drones to a particular area who hang around when they would normally leave pretty quickly. The second is that not all researchers have found the tell-tale pattern in which drones can be easily and consistently attracted to lures in some places and not others. The influential bee scientist Colin Butler, who worked at my field site at Rothamsted Research back in the sixties, reported that he could attract drones to a lure with equal ease over a very large area, even 800m out to sea! Drones can only detect the queen’s scent within a relatively short distance, so Butler concluded that drones must be pretty evenly spread throughout the landscape in order to find and respond to his lures so quickly. There haven’t been enough sufficiently rigorous surveys to draw a strong conclusion, but one way to interpret the accumulated evidence is that apparently distinct, stable congregation areas can be commonly found in mountainous areas but drones appear to be more evenly spread in flatter, open areas.

 

The best evidence for congregation areas, until now, comes from a lovely study carried out by Gerry Loper and colleagues in Arizona, USA, in the early nineties. They used a radar – not a harmonic one, but the traditional sort, that looks out for reflected signals – to monitor the skies around a large apiary and found clear evidence of drones gathering in large numbers in many areas. But the radar results appeared to show a pattern of behaviour that was strikingly different from the conventional account of drone congregations. Studies with pheromone lures have concluded that congregation areas are pretty big – up to several hundred meters wide – and absolutely packed with drones; some scientists believe that tens of thousands of drones may gather simultaneously. By contrast, the radar data appeared to show large numbers of closely packed areas in which drones would mill around. Some of these were stable, reforming in the same places day after day, but others were ephemeral, bubbling up for brief periods then dissipating again. Loper and colleagues monitored one such congregation for ana afternoon and never counted more than 68 drones present at once. More striking than the possible congregation areas was a network of “flyways” – relatively narrow routes connecting the hives that most drones apparently stuck to when moving around the landscape.

 

So, do drones disperse widely, searching for queens or do they stick to a congested “road” network in the sky? Do they gather together in congregation areas, and if so, what do they do up there? Given that individual drones are relatively short-lived, and none survive from one season to the next to teach their brothers, how do individual drones go about finding congregations? James Makinson and I really wanted to know, too, so we spent two summers attaching transponders to drones and tracking their flights with radar. Our harmonic radar allowed us to see the actual flight paths taken by 78 drones over hundreds of flights and start to gain insight into their behaviour that no-one had ever been able to observe before.

 

The first thing we did with our data was to ask where in our field site they went. I created a heatmap in which the areas most visited by bees are lit up in brighter colours than those in which bees were seldom detected. The immediately striking thing was that bees had been observed in single every part of the landscape visible to the radar (the radar requires a direct line of sight to operate so the hedges and trees that formed the boundaries in the farming landscape in which we were working also defined the boundaries of our trackable area). These visitations were far from uniform, though; instead, clear bright lines connected the three different hives we used, and extended beyond them to the north-east and south-east, apparently threading their way between hedges and stands of trees. Apparently both sides were right: drones manage to disperse across the whole landscape, yet primarily make use of “flyways” to get around. Examining the tracks of individual drones clearly showed bees from different colonies, and even in different years, flying along remarkably similar flight paths, even taking kinked or dog-legged routes past the other hives rather than a more direct “bee-line”!

 

Next, we looked more carefully at the flight paths of individuals. Most flights moved straight and fast across the landscape, but occasionally drones showed a profound change in behaviour, switching abruptly to a slower, more convoluted pattern of flight, in which they appeared to circle or fly back and forth in a relatively restricted area. I came up with a simple algorithm to detect these periods of convoluted flight, using just the change in direction from one radar fix to the next. Looking at a map of where these convoluted flights occur, several hotspots of activity became apparent: drones showed this pattern of flight near the hives and in four other spots, all located on flyways. The convoluted flying near hives cannot represent congregation areas (lure studies have found that drones will not approach a queen near a hive, perhaps to ensure that drones do not mate with their own sisters), so we are probably seeing drones learning the position of the nest entrance. Closer examination showed that there were telltale differences between the hive flight and the other four spots in various aspects of the flight dynamics, such as the speed of flight, that collectively suggest that there was a distinctly different sort of movement going on in those four areas. These places, far from home or any food sources, in which many drones, at different times and over several years, have abruptly stopped their flight through the landscape and switched behaviour completely must surely be the fabled drone congregation areas.

 

We found that drone flight at these apparent congregation areas had a particular pattern of movement, in which the further they flew from the center of the area, the harder they accelerated back toward it. Think of a group of marbles being swirled around in the bottom of a bowl: at first they move away from the center, but as they roll up the sloping sides they slow and eventually reverse direction, gathering speed as they rush back to the middle. The marbles are in constant, chaotic motion and yet, over time, a distinct picture of a “marble congregation area” emerges, centered on the middle of the bowl. This same pattern of accelerations can be found in insect swarms, such as midges, and is the force behind swarm cohesion. Individuals may come and go but those that remain appear bound to a particular area by an invisible force. This explains how a congregation can remain in the same place even as the individual drones in the swarm come and go. In fact, we found that individual drones only spend around two minutes at a time in convoluted flight before resuming their journey, yet the four congregation areas we identified persisted over at least two years.

 

So, what exactly is the purpose of gathering in congregation areas? Drones’ only role in the life of the hive is to mate with a virgin queen, so congregations must surely facilitate mating in some way. In fact, gatherings of large numbers of males hoping to mate are a common phenomenon across many animal groups and are known as leks. You might be familiar with fallow deer fighting over territories, for example, or with male black grouse, each holding a small territory, calling loudly and puffing themselves up to attract females. Many other animals, including various fish species, also aggregate at leks, and males of many insect species collect in swarms that are generally considered to be a form of lek. Honeybee congregations share a number of key characteristics with these leks: the males hold no useful resources and contribute nothing but their genes to the females they mate with. On the other hand, once key feature of vertebrate leks is that a small number of very successful males generally monopolise access to females so that there is a huge disparity in mating success, with most males failing to mate at all while others mate with many females. Because drones die immediately after copulation, it is simply impossible for a few successful males to mate with all the females. On the other hand, there are hundreds if not thousands of times as many drones as virgin queens, and it seems impossible that most males will successfully mate at all, so even mating just once would make a drone far more successful than the majority of his rivals.

 

One major outstanding question regards female choice: in classic lek mating systems, the females choose between males (which is why they fight to hold the best territories or compete with songs and colourful displays for the females’ attention), but it is received wisdom that honeybee queens do not choose at all. Instead, males compete in a “scramble competition” to catch any females who fly by, with the fastest males physically catching the female and copulating with her. Upon further investigation, we cannot find any good evidence for this view at all. It seems unlikely that drones perform any sort of display for the benefit of queens but there is no reason to suppose that queens wouldn’t allow some males to catch up while trying to avoid others. Even when caught, copulation only occurs when the queen opens her sting chamber which is something she can control, so it may be possible for her to refuse to mate with a substandard male. Finally, queens mate with multiple drones and are able to control which sperm is used to fertilise their eggs, so it is possible, even likely, that they exercise “post-copulatory” mate choice, choosing which drones father their offspring.

 

Although insect mating swarms are not identical to vertebrate mating systems, it seems pretty clear that they form part of a continuum of lek-like mating systems that span the animal kingdom. When we tracked drones performing convoluted flights at our congregation areas, they were almost certainly taking part in a communal lek and hoping to chase down a passing virgin queen on her “nuptial flight”.  One important observation from our data was that it was very common for a single flight to contain several of these episodes of convoluted flight, at different congregation areas. Vertebrate studies have universally found that males are faithful to a single lek location (at least, within a mating system; occasionally males will switch their allegiance from one lek to another between years), so our study is the first ever to describe a lekking system in which males patrol between multiple sites. Of course, this might not be that honeybees are unique, just that no-one has previously known what to look for. I suspect that mating swarms of various other insect species probably operate in a similar way, albeit at smaller scales than honeybees.

 

Identifying that drone congregation areas are a form of lek doesn’t really explain why they gather in this way, but it does establish that whatever forces shaped this behaviour, they are not unique to honeybees but are common to many other animal groups. There are a number of competing (but not mutually exclusive) ideas about why males gather in leks, the leading ones being the “hotshot” theory – that some males are just so sexy that the best chance other males have is to hang around nearby and hope to mate with the females who show up hoping for a chance with the  “hotshots” – and the “hotspot” theory – that all the males hang around in places that they think females are likely to be visiting, in the hope of distracting them from their other tasks and persuading them to mate instead. Neither is a great fit for honeybees: since no male can mate more than once, any “hotshots” are likely to be too short-lived to attract a group of hangers on; and the only reason that virgin queens ever have to eave the nest is for their mating flights, so there are no obvious places to intercept them as they go about their other business.

 

I think this difficulty in otherwise locating females is the key:  the nuptial flight is the most dangerous part, not just of a queen’s life, but that of her entire colony – by the time she tries to mate, there are no backup queens to take her place if she fails or gets killed – so she is strongly motivated to go wherever she stands the best chance of quickly mating with multiple males. Essentially, the congregation area operates as a rendezvous, almost like a bee nightclub: the boys go there because they think it’s the best place to meet girls and the girls go there because they think it’s the best place to meet boys. Sure, there’s a lot of competition between males for the few females who show up, but that’s still better than trying to search the entire landscape alone: there are so few queens and they come out for such short periods, that the chances of locating a mate that way are minuscule. (I mentioned earlier that drones have been observed returning repeatedly to places they have encountered a queen or her pheromone: having an idea where you might be able to meet a female without thousands of your mates cramping your style changes the cost-benefit calculation and makes joining the lek somewhat less attractive. This also likely explains the fact that drones appear to fly everywhere, but with a strong bias for the flyways and congregations: it’s worth spending some time looking for mating opportunities away from the lek since the potential rewards are so great; but since the probability of success is low, it’s also rational to expend most of your effort on joining the lek with everyone else.)

 

The big unsolved mystery is how drones and queens find these congregation areas in order to rendezvous. People have made various suggestions over the years (albeit with very little evidence to back them up), but they are all things like “low points on the horizon”, “south facing slopes” and so on. It’s obvious from a cursory examination of the landscape that none of them are precise enough to specify a particular location. In other words, even if congregation areas only form in areas with these properties, that is not enough to tell you where in a given landscape the congregations are going to be. We were able to track a number of drones taking their first ever flights. These bees, that had never been outside the nest before, tended to fly in short loops centred on the nest, rarely going further than 100m from the nest and not once visiting a congregation area. Interestingly, the shape of these flights was more reminiscent of what bumblebees do than of the flight of honeybee workers. When we looked at the sequence of flights undertaken by drones, starting with no knowledge of their environment, we found that they made a small number of these early, short-range flights, then switched immediately to making congregation area visits with no evidence of any systematic searching at all. So, whatever the “signature” of a drone congregation area turns out to be, apparently it can be detected from relatively close to the hive.

 

On the face of it, this is extraordinary: honeybees are found all over the world, in a huge variety of different landscapes and a bee has no way to know where its nest is located until it undertakes that first, exploratory flight. How can it possibly guarantee that it will see the way to a congregation wherever its nest happens to be? In fact, when you think about it, any rules that are specific enough to lead you to a single site are going to be fraught with risk: what if your colony turns out to be in a landscape with no sites meeting that description? What if there are too many potential rendezvous? Building on Gerry Loper’s work, I suggest that the way out of this dilemma is that there are no rules about where congregations can occur, but that the congregations emerge from the organization of flyways in the landscape.

 

Drones on mating flights probably follow a small number of simple flight rules. These might include a preference for having more visual “mass” at the edges of the visual field than the center (which will have the effect of making drones turn away from high or wooded areas and fly toward low spots on  the horizon); a desire to balance the “optic flow” – the speed at which things move across the retina – in each eye (which will tend to make them fly down the middle of corridors between trees and so on); or a preference for placing the horizon at a particular angle relative to the eye (which will tend to make them fly at a constant distance from linear features like hedgerows or woodland edges). These rules are general enough that they will operate in any landscape – wherever your nest turns out to be located, you will know which direction to fly in – and they will tend to funnel all bees toward a common network of flyways, no matter what their starting point, just as the shape of a watershed channels raindrops toward a network of rivers and streams, no matter where individual drops fall and no matter what shape the landscape is.

 

This explains how drones from different hives end up sharing a set of common flyways, but what about congregation areas? My suggestion is that congregations form wherever drones meet one another in large enough numbers. Usually, this is likely to be where multiple flyways cross, or at dead ends where large numbers of bees turn around and go back, only to meet those flying up behind them; but if the drone population becomes very large, new congregations will also start to form in the middle of flyways, probably at bottlenecks, where drones are funneled into a restricted passageway. This system would be very flexible: flyways can occur in any type of landscape; drones will join them without any searching becoming necessary; and whenever the drone population in the flyways becomes large enough, congregations will form. The number of active congregations will depend on the encounter rate in the flyways, so more congregations will form in densely populated areas; but in sparsely populated areas, drones will not end up split between several equally good locations, instead a congregation will form wherever in the flyway network the encounter rate is highest. This also explains the appearance and disappearance of ephemeral “bubbles” of drones noted by Loper and colleagues, which appeared otherwise similar to the more stable congregation areas: at times of high activity, rising encounter rates will trigger convoluted flight behaviour;  if the population of drones is not large enough to sustain an extra congregation, the encounter rate in the flyway will fall and the bubble will dissipate; but if the population were to rise, the ephemeral bubble might become a permanent congregation.

 

This was one of the biggest challenges I’ve undertaken, with a huge dataset, collected over several field seasons, and a very complex analysis. It has thrown a lot of light on the previously mysterious behaviour of drones. It now seems certain that they really do congregate together in large groups in specific locations which can remain stable over long periods of time, far exceeding the life span of any of the individuals that make up the congregation. In fact, we have shown that even within a day, individual drones only spend a few minutes at a time in the congregation, and yet the lek itself persists. Many past advocates for the congregation area hypothesis believed that these were huge swarms, several hundred meters across and containing tens of thousands of drones. Instead, we’ve shown that they have a complex internal structure, with a much greater number of far smaller leks, connected by a network of linked flyways. It is likely that scientists who have used pheromone traps to catch drones at congregation areas have gradually depleted the entire flyway system – catching every drones that passes through so that drones keep on arriving but none can leave – giving the impression of huge swarms, although that any drones are never really there at one time.

 

Importantly, this work dismisses the idea that honeybee mating is a completely unique system. In fact, drones use similar flight dynamics to midges and other swarming insects to form a cohesive swarm, and along with many other flying insects, they are part of a whole suite of lek-like mating systems, demonstrating how similar ecological challenges lead to similar solutions across the animal kingdom. Our discovery that drones appear to routinely move between multiple mating swarms opens up a fascinating new variation on lek mating. Future work will reveal what selection pressures gave rise to such behaviour, and which other animal groups have hit upon similar strategies.