Researchers from São Paulo State University (UNESP) in Jaboticabal, in partnership with EMBRAPA (Brazilian Agricultural Research Corporation), have developed an artificial intelligence (AI) tool capable of measuring stress levels in tambaqui (Colossoma macropomum), the most widely farmed native fish species in Brazil.
The findings, published in the journal Aquaculture, mark an important step toward improving animal welfare and enabling the genetic selection of stress-tolerant fish for aquaculture.
Why Tambaqui Stress Monitoring Matters
Tambaqui is an Amazonian freshwater species and a cornerstone of Brazilian aquaculture, with 110,000 tons produced in 2022. Stress management is critical in fish farming because it directly affects:
- Growth performance
- Disease resistance
- Overall animal welfare
The research team found that tambaqui exposed to confined conditions or treated with stress hormones displayed darker body coloration. This visible trait became the basis for training AI software to detect stress automatically.
How the AI Tool Works
The scientists used 3,780 images of tambaqui from two populations:
- 1,280 fish from CAUNESP
- 2,500 fish from EMBRAPA in Tocantins
By marking the lower half of the body in each photo, the team trained a deep learning model to analyze the ratio of black to white pixels. This allowed the system to identify a threshold for stress detection.
Interestingly, since the Tocantins fish had known ancestry records, the researchers were also able to demonstrate that stress tolerance is heritable, meaning selective breeding programs could produce more resilient generations of farmed tambaqui.
Physiological Mechanisms Behind Stress
Fish often display color changes when stressed, a phenomenon also seen in tilapia (Oreochromis niloticus). The process is triggered by stress hormones like α-MSH (melanocyte-stimulating hormone), which expand melanophores (black pigment cells) in the skin and scales.
To confirm this in tambaqui, the team conducted two experiments:
- Hormone exposure test – Scales soaked in α-MSH solution darkened significantly after 30 minutes.
- Confinement study – Fish transferred from large 200 m² tanks to smaller 2,000-liter reservoirs developed darker coloration after 10 days.
These findings proved that darker pigmentation is a reliable stress indicator in tambaqui.
Applications for Sustainable Aquaculture
The AI-based stress detection tool offers several benefits for fish farming:
- Real-time monitoring of animal welfare using simple photographs
- Guidance for farm management, such as adjusting stocking density
- Support for selective breeding of stress-tolerant fish
- Contribution to better productivity and disease resistance in aquaculture
According to project coordinator Diogo Hashimoto, the goal is to ensure future generations of tambaqui show improved well-being and performance in farming environments.
Study Reference
Celma G. Lemos et al. Deep learning approach for genetic selection of stress response in the Amazon fish Colossoma macropomum. Aquaculture (2025).
DOI: 10.1016/j.aquaculture.2025.742848