Predicting Pollinator Sounds from Morphology: AI-driven integration of photogrammetry and bioacoustics for scalable insect monitoring

Pollinators are vital to global biodiversity and food security but are declining due to climate change, habitat loss, and pesticide exposure. Acoustic monitoring offers a non-invasive way to track pollinators in the field, yet most insect species lack recorded sounds, limiting the reach of this method. This PhD will develop an AI-powered framework to predict and identify insect flight and pollination sounds from morphology, enabling scalable, automated biodiversity monitoring.

Using high-resolution 3D models of insect specimens generated through photogrammetry, the student will quantify morphological traits such as wing shape, venation, and body size. These data will be linked to flight and pollination sounds collected under controlled and field conditions. Deep learning models will then be trained to predict acoustic signatures from morphology, allowing inference of sounds for species with no existing recordings.

The student will explore:
    •    Building and analysing a 3D morphological dataset of diverse bee and hoverfly species.
    •    Recording and characterising insect flight and buzz pollination acoustics.
    •    Developing and validating AI models linking morphology and sound production.
    •    Applying models to improve species detection in passive acoustic monitoring networks.

This project sits at the intersection of ecology, bioacoustics, and machine learning, offering training in computational modelling, 3D imaging, and AI. The outcomes will advance scalable monitoring tools for pollinator conservation, enabling us to “hear” the species that science has yet to record.