Modern society depends on a safe, reliable food supply. This supply is a complex system that itself depends on a wide range of other systems, from farming practices to distribution networks. Much work goes into keeping these systems robust.
But in 2007, an unexpected threat to the food supply emerged. Up to 70 percent of the European honeybee colonies in North America suddenly died. Honeybees pollinate a wide range of edible plants, and their demise could quickly and severely reduce food productivity. Biologists have puzzled over possible causes, such as pesticides called neonicotinoids and a disease triggered by the Israeli acute paralysis virus. But exactly how honeybee colonies can be protected in the future isn’t clear.
One thing is certain, however. Biologists need better ways to study these creatures and the way they pollinate plants in their neighborhood. It turns out that bees are good at communicating exactly where they have been. Their famous waggle dance encodes the direction and distance of food sources so that other bees can also take advantage of these sources.
One of the great breakthroughs in animal behavior was the decoding of this dance in the 1920s by the German biologist Karl von Frisch, who won a Nobel prize for his efforts. But this decoding process requires a human to measure the orientation and duration of the dance.
And while videorecording techniques have made this easier, it is still a time-consuming task that can only be done on a small scale for just a few bees. So biologists would dearly love to have a better way to decode honeybee waggle dances.
Enter Tim Landgraf and pals at the Free University of Berlin in Germany. These guys have developed a neural network that can automatically decode honeybee waggle dances. “We have developed a system capable of automatically detecting, decoding, and mapping communication dances in real time,” they say. The new method has the potential revolutionize the study of honeybee foraging.
The waggle dance decoding system is simple in principle. It consists of a video camera that records the movement of bees on a honeycomb. This may look like a seething mass of random motion but there is considerable order here.
The waggle dances themselves are highly ordered. A honey bee begins its dance by waggling its body from side to side at a rate of about 13 Hz while moving forward in a straight line. This is followed by a return phase in which the bee circles back to its starting point. The orientation of this dance relative to the sun encodes the direction of the food source while the length of the dance encodes the distance.
The waggle dance decoder is machine vision system that first searches the video images for the characteristic 13 Hz waggles. Once these have been found, a neural network isolates the bee and its dance. Other algorithms then calculate the dance orientation and duration and finally work out the position of the food source.
In tests on bees trained to visit a known food source some 300 meters from a hive, Landgraf and co say their system accurately identified the position of the food source over 90 percent of the time. That’s just as good as human observers. But crucially the machine vison system can work on much larger groups of bees over much longer time scales.
Landgraf and co say they want to make some improvements to their system, such as making it even more accurate and making it more user friendly. These should be straightforward.
Beyond that, it’s not hard to see how such a system could be deployed on a large scale at relatively low cost. That will allow biologists to study the behavior of bees in much more detail and at much larger scales than ever before. It may even be feasible for farmers to watch in real time as their crops are pollinated and so make efforts to help when things aren’t going to plan.
That could help tackle the problems of colony collapse disorder as it happens and help safeguard the food supplies we all rely on.
Ref: arxiv.org/abs/1708.06590: Automatic Detection and Decoding of Honeybee Waggle Dances