Researchers from IRTA and the University Rovira i Virgili, in Spain, have demonstrated that viable larvae of Senegalese sole (Solea senegalensis) can be produced using hormone-free in vitro fertilisation supported by artificial intelligence capable of predicting ovulation.
The study combines automated behavioural monitoring using underwater cameras with machine-learning models that anticipate ovulation events. This allows technicians to collect gametes at the optimal time and perform fertilisation without hormonal induction.
Results show that hormone-free fertilisation can reach fertilisation rates of up to 44% and hatching rates of 18%, producing viable larvae for aquaculture production.
A key limitation is timing. In Senegalese sole, ovulated eggs remain viable for only about three hours, making accurate identification of ovulation essential for successful fertilisation.
The monitoring system analyses behavioural patterns linked to reproduction, particularly locomotor activity and a courtship interaction known as Rest the Head, in which one fish rests its head on another. Both behavioural increase in the hours preceding spawning.
The predictive model achieved an overall accuracy of 82-85% in identifying ovulation nights, enabling aquaculture technicians to anticipate when fertilisation procedures should take place.
According to the authors, this predictive capability turns hormone-free fertilisation from an opportunistic process into a manageable protocol suitable for aquaculture operations.
The approach could also support the development of aquaculture protocols compatible with organic certification schemes, where hormonal treatments are restricted.
The work builds on previous research by the same group showing that artificial intelligence can detect spawning nights in Senegalese sole based on behavioural analysis.
Referencia:
Qadir, A., Salcedo Martínez, S., Serratosa, F., & Duncan, N. (2026). In vitro fertilisation procedure assisted with computer vision models for organic Senegalese sole culture