How to train your fish

Fish ranching with sound was among the diverse range of future possibilities presented at the WAS 2015 conference

A floating platform equipped with nets on a reservoir in Israel harvests fish trained to come to the sound of a buzzer.
Photo: Boaz Zion

By Catherine Norwood
People and dogs are not the only creatures to respond to the sound of a dinner bell; fish can also be trained to come to a sound associated with food. Israeli researcher Boaz Zion has taken classic conditioning techniques and applied them to fish, training them to come when called – right into a waiting net.

With the right reinforcement, he says, the fish can remember their training for six months, and possibly longer – long enough to grow to a marketable size. And trained fish are also quick to share what they have learned. In the laboratory and in field trials, they have taught their ‘naive’ companions to follow the sound of the dinner bell, or in this case a dinner buzzer.

Boaz Zion has been working on acoustic training techniques for fish for the past six years as an agricultural engineer with the Volcani Center, part of Israel’s Agricultural Research Organization.

His aim is to determine whether acoustic training can be used as a kind of cageless ranching technique. This might allow trained fish to be released into open water to feed and grow, before being harvested when they respond to a previously learned signal.

A critical factor is the duration of a fish’s memory. In presenting his findings to the World Aquaculture Society conference in South Korea in May, he said 28 days was the optimal initial period identified in laboratory trials to train Tilapia (Sarotherodon galilaeus) to come to the sound of the buzzer, initially through the provision of food.

When the learned behaviour was tested at various intervals, there was some residual memory after three months, but none after six months. However, with regular retraining of six sessions every 17 days, the learned behaviour was maintained – potentially indefinitely. Other species successfully trained included European Carp (Cyprinus carpio), African Catfish (Clarias gariepinus) and Grass Carp (Ctenopharyngodon idella).

Boaz Zion said for producers ranching fish in cages, it would be simple and relatively cheap to provide ongoing training with a simple sound system to generate the buzzer in conjunction with feeding. This might prove an effective way to recapture escapees from caged systems, addressing concerns about the contamination of wild gene pools with farmed species, as well as reclaiming potentially lost income.

He also found that when trained fish were released into the Eshkol Reservoir in the Galilee region, it was possible to continue retraining using the same buzzer signal and food supplies from a floating platform that was also fitted with a net.

During the field trial the fish became familiar with the platform system and when the time came to deploy the nets, they effectively harvested themselves.

But even when the buzzer was used from a small boat across many different locations on the reservoir, the fish would come to where ever the boat was, Boaz Zion said: “It was like whistling for fish.”
He also found that the trained fish brought many other fish with them.

“We recaptured 65 per cent of the carp and 13 per cent of the tilapia we put into the reservoir. But in total, we harvested 260 per cent of the original biomass because of the other fish that came with our trained fish.”

Laboratory modelling indicated that under ideal conditions, fish could respond to acoustic signals from as far as 10 kilometres, which equates to an area of 30,000 hectares.

“Of course real-life conditions are not ideal. But even if we only recaptured 20 per cent of the fish we introduced, that would be a break-even point. Every fish captured after that would be profit.”

The research has also found that fish can distinguish between signals with as little as two hertz difference, and can distinguish a signal with a positive association (food) from other, indifferent, background noises.

He said he was confident the technique had potential for the aquaculture industry, although significant work was still needed to investigate a range of factors including responsive fish species and stocking rates, issues associated with shared water resources such as reservoirs, and the technological requirements of acoustic training and harvesting equipment.

More information

Boaz Zion, boazz@volcani.agri.gov.il