The insect detector that listens through a soybean stem

Inside a soybean plant, a stem-borer larva can work its way towards a harvest problem. As the crop matures, larvae can girdle the stalk from within. The weakened stem may later fall over, making the plant harder to harvest. Inspecting the damage ordinarily means splitting stems and looking inside.[5]

Fieldwork reported by Kansas State University on 24 September follows master’s student Emily Sur and her team using plant-mounted microphones to study Dectes texanus, the soybean stem borer. Their project is part of the wider Insect Eavesdropper collaboration.[1]

The device begins with a useful piece of repurposing: an inexpensive clip-on guitar microphone. Inventor Emily Bick, an entomologist at the University of Wisconsin–Madison, describes sensors that pick up feeding on leaves, boring inside stalks and chewing around roots, with machine learning used to identify patterns and develop warnings for growers.[2]

Its listening surface is the plant itself. A contact microphone senses vibrations travelling through solid material; it is not waiting for an insect to make a loud airborne call. Feeding that is concealed from view can still disturb the tissue connected to the sensor. The plant carries a signal from the activity taking place on or inside it.[3]

Teaching the machine what the plant contains

Sur’s team records continuously for several days, removes the equipment and then examines the plants for evidence of insect activity. Bick’s group develops the machine learning, while partners supply field data and help make the equipment dependable. The wider Kansas work includes soybean fields and cotton-filled greenhouses.[1]

Opening the plant provides the crucial label for the recording: what was actually there. Without that check, a distinctive vibration is only a recognisable pattern, with no reliable biological explanation.

Continuous recording also changes the sampling question. A visit gives a scout a view of a field at one moment. A working sensor can preserve what happened between visits. The potential gain is a history of activity: whether it appeared, persisted or changed. Making that history useful requires knowing which changes deserve attention.

A companion tool shows the less glamorous work involved. In a July 2026 paper, Bick and colleagues describe BickGraphing, browser software for inspecting waveforms and spectrograms locally, including offline use. Waveforms show amplitude over time; spectrograms show how energy at different frequencies changes. Researchers can use them to check recordings and whether a sensor is functioning in the field.[4]

The software paper reports that Insect Eavesdropper had been evaluated on 27 insect species across 17 cropping systems by more than 30 academic and government researchers. Those figures describe the breadth of investigation. They do not supply a single identification accuracy that applies across all the plants, insects and conditions.[4]

Sur’s students also troubleshoot equipment in the field; Kansas State documents undergraduate Olivia Herold working on a device. Reliable recordings depend on this practical work with sensors and electronics.[1]

The useful warning may change the harvest queue

The soybean stem borer makes the project especially interesting because finding an insect does not automatically point towards spraying. K-State’s current soybean management guide says foliar insecticide control of this pest is not feasible. It identifies timely harvest as the best way to avert losses and advises prioritising fields with high percentages of infested stems, before weakened stalks fall.[5]

That suggests a specific future role for reliable sensing. If recordings can identify activity that corresponds to a meaningful infestation, they could help decide where to inspect and which fields need closer attention as harvest approaches. Better information might change the order in which work happens. The value would come from connecting an alert to a useful decision in the crop’s life cycle.

The September field account does not publish detection accuracy, sample totals or a measured yield benefit. The present contribution is the fieldwork building that evidence. Its proposed use is to alert growers to investigate changes, alongside existing scouting.[1]

Commercial development is also under way: BugBug presents the Insect Eavesdropper as a combination of contact microphones and proprietary algorithms for detecting and identifying pests. Its site lists initial detection examples spanning insects on tobacco, maize, trees and other plants. A broad range of detectable activity creates an opportunity; each crop still needs an interpretation that fits its own management decisions.[6]

For the soybean grower, the most valuable message could eventually be very ordinary: take a closer look at this field before choosing where to send the combine. Producing that message starts with learning to hear an insect through the stalk that hides it.

Report an error or suggest a correction