Listening in on the hive
A honeybee flies back to the hive after an unsuccessful hunt for food. The nearby field of flowers – usually a popular and safe place to collect nectar and pollen – has proved otherwise: This time, the bee narrowly escaped a crab spider that had been hiding in the blossoms of a goldenrod.
Flying off in a hurry, the bee quickly arrives at the hive just a few hundred metres away. At first, the bee takes a moment to collect itself before noticing that another bee is performing its waggle dance for a group of bees on the honeycomb. A moment later, the recently arrived bee is on a collision course with the dancing bee. The resulting crash is unavoidable.
Dance as a form of communication
To understand what has just taken place, we start by looking at the unsuspecting target of the collision. Why was the bee dancing across the honeycomb dance floor anyway? Giovanni Galizia, who has been studying the neurobiology of insects for many years, now, knows the answer: "Honeybees are social insects that exchange information about food sources. Colonies then repeatedly return to particularly prolific fields of flowers. They share this information using a one-of-a-kind form of communication: the waggle dance." When a bee returns from a successful hunt for food, it shares taste samples of the nectar with other bees in the hive and starts to perform its dance. Afterwards, colony members who are convinced by tasting the sample join the dancing bee and learn from its steps what direction and how far they must fly to reach the corresponding food source.
How the waggle dance works
The steps of the waggle dance go like this: The bee waggles its abdomen while taking some steps step straight ahead, then turns right in a semicircle until it returns to the starting point; it then repeats the same movement, this time turning left. Afterwards, the process starts all over again.
The decisive information comes from the straight stretch: the more steps the bee takes, the greater the distance from the hive to the food source. The angle of this stretch relative to the vertical honeycomb corresponds to the angle relative to the sun that a bee must follow when leaving the hive in order to reach the food source.
This way, although only a small number of bees receives this information at any one time in the busy hive environment, word spreads quickly. It only takes a few bees to start following the dance. Once they return to the hive with the same sweet nectar, they spread the word through their own corresponding dances.
"What starts with one bee, rapidly becomes ten, a hundred, a thousand bees – and, after a few repetitions, the entire colony knows about the new food source", Galizia explains. The bee dancing on the honeycomb was thus intent on telling the other bees in the hive about a lucrative food source – exactly the flowery field where the other bee had just narrowly escaped from the dangerous spider.
Stopping the snowball
So, what happens when a popular source of nectar suddenly becomes a dangerous location? Does the whole colony continue to fly towards its own destruction? Or is there a way to stop the snowball of faulty information in its tracks? "This is possible!" Galizia confirms and adds: "If a bee has had a bad experience with a food source, it can interrupt another bee while it is dancing to advertise for this particular source. And it does this in a rather brutal way – by crashing into the dancing bee, producing a loud sound." This is exactly the scene described above.
© Inka Reiter"The stop signal is an amazing cognitive feat for an insect. To do this, a bee not only has to correctly interpret another bee's waggle dance but also link the information to its own negative experience at the corresponding location."
Giovanni Galizia
Since this behaviour occurs quite rarely and research into the stop signal is still in a fairly early stage, we do not know much more than its basic structure and function. To finally unravel the secrets of the signal, a team of experts from different disciplines has come together at the Konstanz CASCB – including biologists, an audio engineer, a computer scientist, and a robotics expert. Their initial goal is to reliably recognize and document the stop signal amidst the bustle of thousands of bees moving around the hive.
A sound is key
Watching and waiting are critical in this case. "It would be way too difficult to simply plant cameras in a hive, record the activities there, and wait for stop signals to appear", says project member Andreagiovanni Reina, whose research focuses on bio-inspired robotic systems. Instead, the team of researchers in Konstanz wants to automate observation of the signal. "With regard to the waggle dance, we have already come a long way. We have trained our artificial intelligence system to automatically recognize this conspicuous type of dance in camera images and even to decode the corresponding navigational information for us", Reina explains.
Neurobiologist James Foster in protective clothing captures bees as they leave the research hives on the University of Konstanz campus.
A look inside an experimental beehive
The team of experts studying the stop signal of bees (from left to right): Frida Hildebrandt (behavioural experiments), James Foster (bee sensors), Jacob Davidson (computer science), Andreagiovanni Reina (robotics), Valery Kuklovsky (bee learning), Alberto Doimo (audio engineering) and Giovanni Galizia (neurobiology)
Yet even AI image analysis has its limits when it comes to recognizing the more inconspicuous stop signal in the midst of the beehive hubbub. This is why the project team uses another aspect of this behaviour to their advantage: When bees collide, the bee interrupting the dancing bee produces a clearly audible, high-frequency sound. "We don't know whether this sound is simply a side effect, or whether the interrupted bee – and others following its dance – gain information from it. Either way, the sound is so characteristic of the stop signal that it can help us to reliably detect it", Reina says.
Listening instead of just watching
With this idea in mind, the team is developing a new kind of observation beehive in close collaboration with the university's Scientific Engineering Services. In addition to containing cameras, the high-tech beehive also features a grid of microphones that register each time there is the signature sound for a stop signal. But there is more: Using triangulation – based on the recordings of individual microphones – it is possible to also calculate the approximate location in the hive where the bee displays this behaviour.
"The acoustic information provides the location and time of the stop signals so we can automatically find the camera recordings from the hive that contain corresponding images of the behaviour. As a result, we can gather the number of observations much faster that we need to describe and characterize the signal in detail", Galizia says. The researchers are initially interested in answering some basic questions: How often do bees display the behaviour under natural conditions? Who observes the stop signal – just the dancing bee, or those following it, too? And how effectively and sustainably does it really change the colony's foraging behaviour?
One researcher could not manage a project of this scale on their own – it requires addressing too many challenges in multiple different disciplines. The central research question of the CASCB thus also applies to its own work: How can different individuals work together to complete a task that they are not able to do on their own? In the centre, people from different disciplines come together to put complex, interdisciplinary projects into action. This also means considering the impacts of results from basic research for applications outside a specific field.
The findings from studying the bees' stop signals will help refine mathematical models of the dynamics of foraging behaviour in bee colonies. While existing models account for the amplifying effect of the waggle dance on bee traffic to specific food sources, they mostly do not factor in the stop signal as a type of negative feedback. Yet the mixture of positive and negative feedback is precisely what makes it possible for a colony to flexibly adapt to new environmental conditions – a skill that is also valuable in other systems.
"Good mathematical models of how bee colonies organize themselves are of great interest in the field of robotics. Basically, bee colonies and systems of robots often face similar problems – like building consensus, for example. If evolution has come up with an effective solution for bees, the field of robotics is open to taking inspiration from nature."