The future guardians of our orchards

In the basement of the Laimburg Research Centre, a swarm of tiny flies flit about inside a display case. The insects move frantically, bathed in bluish light and the hum of scientific equipment. Bodies as small as sesame seeds, large red eyes, they are specimens of Drosophila suzukii, a plague on fruit production in Italy. Something is watching them, scrutinizing their every wing beat, but it’s not a human. “Observing” the flight of the insects is an infrared sensor capable of catching even the slightest peculiarity. Each time the sensor picks up the passage of one of the insects, the system acquires new data. But it doesn’t end there. The sensor is just the eye of a far more complex technology still under development: an algorithm. This is an artificial intelligence that, once it has completed its training, will be able to recognize crop-damaging species and report their presence in real time. The algorithm is set to become one of the future guardians of our orchards, the final frontier in the fight against agricultural pests.

For Drosophila suzukii, a cherry is not just a delicious meal: it is their key to the future. The females of the species entrust their offspring to its juicy flesh. And unlike other species, which are only capable of affecting soft rotting fruit, Drosophila suzukii lay their eggs inside unripe fruit. The larvae then devour the fruit from the inside, compromising the fruit’s marketability. Drosophila suzukii has been present in Italy since 2009, when the species was accidentally introduced from Asia. Since then, the insect has spread like wildfire due to its marked adaptability to temperate environments and an absence of natural predators. All to the detriment of the cultivation of cherries and other small fruits, such as blueberries and raspberries.

In attempts to curb the problem, a variety of countermeasures have been taken – nets to cover orchards and hinder access, pesticides, the release of predatory organisms antagonistic to the drosophilas. Each of these methods, while capable of containing the damage caused by infestations, has limitations and side effects. In addition to preventing Drosophila suzukii specimens from reaching the cherries, nets do not allow pollinating insects to pass through and they also alter the microclimate of cherry orchards; pesticides can cause damage to the ecosystem and our health; and antagonistic species curb drosophila proliferation but are unable to eradicate them in the event of an orchard infestation.

“A big step forward would be to know the exact moment when an infestation is about to take place,” says Silvia Schmidt, entomologist at the Laimburg Research Centre. “That way, nets and pesticides could be used only when actually needed.” That’s where sensors and artificial intelligence come in. Silvia Schmidt is leading a research project aimed at developing innovative solutions to combat orchard-damaging species. The project is called INSTINCT and involves the Laimburg Research Center, Eurac Research’s Center for Sensing Solutions, Unibz and two companies which specialize in the development of artificial intelligence-based technologies. “One of our goals is to develop insect traps that can identify Drosophila suzukii specimens using artificial intelligence. Whenever a drosophila falls into the trap, it will send an alert to the farmer, who can take appropriate action,” the project leader explains. For the traps to recognize what species the fruit flies belong to, however, they need to be “trained.” “We are currently collecting data with which to train the algorithm to distinguish Drosophila suzukii from similar species,” says Silvia Schmidt. To do this, the project team is using a sensor that detects how each species interacts with infrared beams during flight. And once the sensor has made thousands of detections, the data will be fed to artificial intelligence, which will use it as a guide for insect recognition.

Currently, to monitor the presence of Drosophila suzukii in cherry orchards, it is necessary to collect a sample of cherries and to scrutinize them one by one, looking for the hole produced by the insects when they laid their eggs.

To simplify cherry inspection, the Center for Sensing Solutions is testing the use of special optical sensors. “These are devices that emit a beam of light and allow us to study how light is reflected from the fruit,” says Gianluca Scuri, a researcher at the center. “Our hypothesis is that healthy cherries reflect light differently from cherries that contain drosophila eggs".

As part of the INSTINCT project, cherries are analyzed using the spectroradiometer, an instrument capable of emitting and capturing not only visible light but also infrared and ultraviolet. The goal is to identify the type of light reflected from infected cherries. “This will allow the development of instruments that emit and detect only those wavelengths,” explains Center for Sensing Solutions director Roberto Monsorno. “Narrowing the scope of investigation of these sensors means making them more accessible and practical to use in the field.” A similar instrument, the hyperspectral camera, is already being tested by Monsorno’s team in the Laimburg Research Center’s cherry orchards.

Drosophila suzukii is not the only species that affects the productivity of orchards in Italy – Cydia pomonella, also known as apple carpocapsa, is a moth which is native to European wild apple trees. Its larvae penetrate fruits, especially apples and pears, damaging them and causing them to fall from the tree To date, the most widely used technique to curb the proliferation of carpocapsa is that of so-called “sexual confusion.” This involves the use of pheromones, volatile chemicals similar to those that moths use to communicate at a distance. Through the use of these compounds, males of the species are prevented from finding females, limiting the number of matings.

The technique of sexual confusion, however, does not always work. Pheromones do not have long-range efficacy, and to be effective, they must be released throughout the orchards. An alternative is to use a virus that targets the larvae, killing them. This virus, however, must be employed at a specific time in the moth’s biological cycle: the time when most of the females lay their eggs. If you miss this window of time, intervening is useless. “We are targeting Cydia pomonella from two angles: smart traps like the ones used for drosophila, and environmental sensing,” says Roberto Monsorno. “We use sensors for humidity, temperature, solar irradiance, rainfall, and wind strength and direction to monitor environmental conditions in the orchard. This information will be analyzed along with the number of captures, looking for a relationship between environmental parameters and moth abundance.” The final step of the study will be to develop a computer system that, based on real-time sensor readings and taking advantage of artificial intelligence, will alert farmers and orchardists to an impending infestation.