How AI and Robotics Are Transforming Fisheries Management

Across the eastern Pacific Ocean, an uncrewed surface vessel (USV) recently completed a remarkable 4,200-km journey, collecting environmental DNA (eDNA) samples throughout its transit. This pioneering project showed how autonomous ocean-going platforms can conduct research in remote regions without the logistical challenges and high costs associated with crewed ship-based monitoring.

The USV Surveyor, a 22-meter autonomous sailing vessel, carried the Environmental Sample Processor (ESP): A robotic device that automates water sample collection and processing. During its 29-day journey from California to Hawaiʻi, the system collected and preserved surface eDNA samples that provided critical information about marine biodiversity across different ocean zones.

The vastness and inaccessibility of much of the world's oceans make it difficult to obtain representative samples on meaningful scales. This technology presents a solution, enabling sample collection at scales previously unattainable through traditional methods.

Another technological breakthrough is transforming fisheries management across Pacific Island nations. The Pacific Community (SPC) has developed a comprehensive AI-enabled monitoring system named Ikasavea (fish-survey in Tongan, Wallisian, etc.) that harnesses deep learning and computer vision technologies.

In many Pacific Island languages, fish names are highly specific and often differ from island to island. For example, in Samoan, various species of tuna might be differentiated with terms like Atu (Skipjack Tuna) versus Asiasi, while neighboring islands might use completely different terminology. This linguistic diversity has historically complicated fisheries monitoring, but AI technology is helping bridge these gaps.

The Ikasavea system automates data extraction and analysis processes that allow fisheries officers to identify over 600 nearshore finfish species and measure more than 80,000 specimens simply by taking photographs. The technology significantly reduces the level of taxonomic knowledge required at the point of data acquisition, which makes fisheries monitoring more accessible and less disruptive to fishers and retailers.

In the Clarion-Clipperton Fracture Zone (CCZ) of the Pacific, researchers are evaluating the economic feasibility of mining polymetallic nodules and rare earth elements from the deep seabed. This area contains significant deposits of critical metals such as cobalt, nickel, copper, and manganese, along with rare earth elements essential for modern technologies.

Different mining approaches have been proposed, including a combined method that simultaneously mines polymetallic nodules and extracts rare earth elements from deep-sea sediments. Economic analysis indicates that combined mining operations could yield higher economic benefits than separate mining of either resource type.

However, environmental considerations are paramount. The method of separating rare earths from the sediments must ensure that acid solutions used do not directly discharge into the seabed environment and undergo strict environmental treatment.

Language technology plays an important role in these monitoring systems, particularly in the Ikasavea application. Many Pacific languages have rich vocabularies for describing marine life, with subtle distinctions that don't always align with scientific taxonomy. For instance, a single fish species might have different names based on its life stage or size in languages like Tahitian, Samoan, Tongan or Fijian.

The AI system helps reconcile these differences by providing standardized identification while still allowing data collection in local contexts. Users can customize the system through a web portal to adapt survey designs to their specific needs, including setting administrative regions and assigning spatial management areas to sampling hierarchies.

As of recent reports, the system has been implemented across 11 national fisheries authorities in the Pacific, with several more currently being onboarded. This widespread adoption shows the system's scalability and performance in diverse geographical contexts.

These technological advancements however face certain limitations. The AI species identification models show declining accuracy when confronted with new species or specimens with varied colorations. Regular model retraining is required to maintain performance, especially following large influxes of new species.

For the autonomous eDNA collection system, sample degradation is a concern when stored at ambient temperatures for extended periods. Environmental DNA stored on vessels can experience significant temperature fluctuations, resulting in substantial degradation of certain genetic markers.

These technological innovations are addressing historically entrenched technical and financial barriers in fisheries management and environmental monitoring across Pacific island communities. By automating data collection and analysis processes, they enable timely, evidence-based adaptive management of marine resources.

The integration of autonomous vessels, AI systems, and environmental DNA analysis is creating unprecedented opportunities for sustainable resource management in the region. These systems enable near real-time transfer of data to diverse, data-poor coastal fisheries management contexts for evidence-based adaptive management.

As these systems continue to evolve and expand across the Pacific region, they change how institutions and communities communicate and use environmental information. The commitment by national fisheries authorities and stakeholders engaged with these systems demonstrates their value and potential for addressing critical barriers to achieving sustainable development in Small Island Developing States.