There are around 20 different kinds of bumblebees that can be discovered throughout the British Isles. However, several of these species have become extinct since the end of the Second World War, and others are in danger of going extinct.
It is now envisaged that by teaching computers how to automatically detect endangered bees in their natural habitat, it will be possible to contribute to the preservation of those bees.
The researchers will need to train the system to recognize the unique buzzes that are produced by each species.
They accomplish this by recording the sound and adding information about the bee that created it to the recording.
Alixandra Prybyla, who is currently pursuing a doctoral degree at Edinburgh University, and Eva Nelson, who is a fellow undergraduate student at the university, have been looking for bees in the abundant wildflower meadows that surround the peninsula of Brodgar in Orkney.
The region is home to a diverse collection of species, one of which is the Big Yellow Bumble bee, which is considered to be among the rarest of all bumblebees found in the British Isles.
It is important to consider the thorax of a bee to be “like an insignificant musical instrument,” as Alixandra Prybyla stated in an interview with BBC Radio Orkney.
According to her, the buzzing of a bumblebee is analogous to the sound produced when an instrument string of a guitar is plucked, which causes a vibration to travel down the string through the internal structure of the instrument.
Its body is designed to shape the vibrations it creates when it flies, and these vibrations are then exhaled as a distinct sound.
Therefore, the human ear is able to distinguish, for example, the distinction between the faint buzz generated by the relatively slow winged beats of a buff-tailed bee queen as well as the significantly higher — almost whining — noise made by a little early bumblebee worker. This is because the queen’s wings beat more slowly than the worker’s.
However, in order to recognize distinctions that are more nuanced or intricate, one must make use of scientific gear that analyzes the frequencies that insects emit.
Studying through all of the alternatives would require a human investigator “millions of hours.” Therefore, Alixandra Prybyla and the students under her instruction are utilizing artificial intelligence.
We are training an artificial intelligence program that we are programming by capturing thousands upon thousands of bee noises, gathering a wide variety of environmental data, and collecting a wide variety of morphological data. We are then providing all of this information to the algorithm. Instructing it on what to search for.
That implies that when they have recorded an individual, the next step is to live to catch it in a net, cool it, and put it into torpor, which is a state similar to hibernation, and then handle it securely. The animal is subsequently identified, weighed, measured, tagged, and then rewarmed before being released.
It is intended that a database will be developed that will link the noises with all of those specifics, in addition to environmental parameters such as the current temperature, the amount of pollen that the bees are carrying, and the current season.
According to her, if scientists measure bees for the entirety of their season, they should be able to gain a comprehensive view of all the factors that can influence the way a bee sounds.
The buzzes that bees create can be caused by a wide variety of factors. They include aggression buzzes, which are signaling buzzes, and sonication, which is a booming buzz that is used to encourage flowers to release their pollen.
But in order to ensure that the methodology is comparing similar, scientists are concentrating on recording foraging buzzes, which is the noise that a bee makes as it travels from bloom to flower in search of nectar and pollen.
It is envisaged that in the not-too-distant future, the system will have the ability to recognize and analyze recordings that have been automatically created by microphones that have been left in bee homes.
According to Alixandra Prybyla, it can be challenging and time intensive to have specialists go to every one of the different locations in order to identify the bees, conduct population examinations, and obtain the kind of data that is required to execute conservation actions.
Therefore, we are keeping our fingers crossed that this technology may be utilized to create something that is known as a “remote acoustic surveillance station.”
You would be setting up microphone arrays that would take in all different types of buzzes along with automatically identify them based on their characteristics. And if we observe changes that are cause for concern or anything that a researcher out in the field would call attention to, we will report them. It is possible for us to go outside and investigate the situation.