It’s a fact that farmed fish can and do escape. Sometimes these escapes happen because storms rip up and damage the cages, or when predators attack the nets to get at a tasty meal, leaving a hole for the fish to escape through. Sometimes escapes can happen from unexpected incidences – like when a cargo ship hit a rainbow trout farm off Assens, Denmark, releasing 80,000 individuals into the western Baltic Sea. Less dramatically, simple human error may be the cause – as is the more serious issue of negligence. When some 250,000 Atlantic salmon broke out of Cooke Aquaculture's Cypress Island farm into the Pacific in 2017, Cooke initially blamed a combination of currents, unusually high tides, and a solar eclipse for the escape. It later emerged failure to properly clean the pens, poor condition of parts of the infrastructure, and engineering issues were the most likely cause.

Getting a handle on just how many escapees there are is no easy task. Not all countries require aquaculturalists to report escapism. In some cases, aquaculturalists under-report the number of escapees, or simply may not know how many have broken free. Cooke Aquaculture, for example, initially reported some 160,000 salmon escaped from Cypress Island in 2017, but an independent investigation put that number between 243,000 and 263,000. Despite such accounting issues, the number of escapees is large. In the 1990s, between 20% and 40% of Faroe Islands salmon fisher catches consisted of farmed Atlantic salmon. Meanwhile in Europe, over just three years some 8.9 million finfish escaped farms located in just six countries. Regardless of how it happens, for the aquaculturalist escapees come at a cost. Damaged equipment needs to be repaired or replaced. Every individual lost is revenue lost. When gusts of up to 100 mph tore twelve salmon cages off their mooring in Scotland on Christmas day in 2011, they were washed out into the North Sea - along with £3 million worth of fully-grown salmon. Fines may also be issued. In April this year, Cooke Aquaculture reached a settlement with the Washington Department of Ecology for U$322,000. With Cooke Aquaculture being the largest seafood company in the world and bringing in an estimated C$2.4 billion of revenue in 2018 alone, the settlement amount has invariably received some criticism.

Economic impacts on aquaculturalists are easier to measure than impacts escapees the natural world. Nevertheless, it is clear that impacts can - and do - occur.

One of the biggest concerns comes from the potential for farmed and wild individuals to interbreed. Whilst wild individuals are subject to natural evolution, resulting in individuals that are adapted to living in the wild, farmed individuals aren’t. Particularly for species like salmon, breeding is heavily controlled by aquaculturalists. Eggs from the females and milt from the males are manually taken from farmed fish selected for breeding. Once fertilized, the eggs hatch and resulting fry are reared in one facility before being transferred to another, and perhaps another as they continue to grow. This careful control is primarily designed to create individuals ideal for a farm setting, like fast growth, large size, quality meat, and even resistance to disease. The more generations of farmed fish that are produced, the more they differ from their wild counterparts, and the less likely they are to be able to survive outside a farmed environment.

For those fish that do survive to the breeding season – and make it to a breeding location – successful mating isn’t guaranteed. Alongside losing traits for survival in the wild, farmed individuals can also lose traits that improve their chances of breeding. For male salmon, these traits include a rather interesting nub on the end of their lower lip called a kype.

A salmon’s kype is known as a secondary sexual trait, meaning it is used indirectly in breeding. In this case, the kype is used as a signal of fitness for females (females prefer fitter males to fertilize their eggs), and to fight with males. “The fighting can be pretty horrific” William Perry, a PhD researcher at Bangor University in Wales explained as he recounted a story his supervisor told him of salmon fighting in a stream in Norway “The one salmon that came off worse had his entire lower jaw ripped off”. Whilst a kype is useful for a wild salmon, for farmed salmon subject to multiple generations of artificial selection, it is less useful. Indeed, when Perry compared kypes from wild salmon with those of 12th generation farmed salmon, the results were clear. “We found you get a reduction in kype length relative to body size in farmed individuals compared to wild”. Combined with other studies that have shown farmed salmon also don’t exhibit the appropriate mating behavior, farmed fish are less likely to breed in the first place.

However, we know that wild and farmed salmon are breeding – and producing fertile offspring. One of the reasons is down to the sheer volume of escapees. “In 2006, for example, almost 900,000 farmed salmon escaped in Norway alone. The Norwegian wild population is about half a million”, Perry said. For salmon at least, the chance of successfully breeding with wild individuals higher for farmed females than farmed males. When a female lays her eggs, she is almost guaranteed to have them fertilized. The female ‘success’ story does not end there; “Egg production is correlated with body size. Because farmed females have been well fed and have been bred for their larger body size, it means that they can produce more eggs than wild females”, Perry explained.

The resulting hybrid offspring are more suited to life in the wild than their farmed parent, but less so than their wild parent. Farmed populations tend to have lower genetic diversity than wild populations, reducing the chance of evolutionary adaptation to changing environmental conditions. Farmed parents can pass on other undesirable traits, like reduced anti-predator responses and indeed smaller kypes. If interbreeding happens frequently, farmed fish genes can swamp the wild genes, potentially lowering the survival chances of what largely becomes a hybrid population.

To minimize the risk of escapees producing hybrid offspring, some aquaculturalists are choosing to (and in some cases required to) sterilize their stock. By exposing eggs to high pressure or high heat, the offspring receive one set of chromosomes from its father (as normal) but from the mother, it receives two. The result – a triploid animal whose eggs or sperm are not viable. Triploidy is not a new technique - In fact is already widely used in agriculture to produce seedless fruits, for example. Nevertheless, the technique is not 100% fool-proof. A small number of individuals can remain fertile and issues like skeleton deformities and reduced growth have been reported in some aquatic animals.

Despite these issues, inducing triploidy is currently the only viable method of sterilizing many individuals and for the most part, has resulted in success. AquaBounty Technologies, Inc. AquAdvantage® salmon – the World’s first transgenic animal to gain approval for sale – induce triploidy in their animals using high pressure; “The results are very efficient. We routinely achieve 100% egg batch sterility; our data presented to the FDA and Health Canada averaged 99.8%” Dave Conley, AquaBounty’s Director of Corporate Communications explained. To improve these averages, AquaBounty is collaborating with the Center for Aquaculture Technology to develop a gene-edited solution, which when finalized, should ensure 100% sterility.

Whilst a triploid salmon cannot produce offspring, a triploid individual isn’t necessarily aware of this. In experimental tanks, male Atlantic salmon still show spawning behaviors and wild females will respond – albeit with the end result of unfertilized eggs. Fortunately, triploid farmed Atlantic farmed salmon may be less likely to enter freshwater spawning grounds and disrupt spawning of wild fish in the first place, but again the sheer volume of escapees could be sufficient to ensure that disruption will occur. Even if breeding isn’t disrupted, or hybrids produced, farmed fish can cause other issues. When the rainbow trout escaped from their cage in Denmark, it was just in time for the native brown trout breeding season. Rainbow trout eat brown trout eggs, prompting Environmental group Vandpleje Fyn’ Soeren Knabe to encourage fisher to "grab their rods and catch as many [rainbow trout] as possible”.

Despite best efforts, once a fish has escaped in the sea, the chance of capturing it again is extremely low, with some studies showing recapture rates as low as 8%. Rivers, on the other hand, are a closed environment offering a unique opportunity to capture marine escapees like salmon and trout that breed in freshwater. Current efforts to remove farmed individuals from rivers are extremely labor-intensive, often involving divers or people with nets to identify and capture farmed individuals whilst leaving wild individuals to continue with their journey. Correctly identifying a farmed from wild fish can be done based on morphological features, but for humans it is not necessarily easy. For Artificial Intelligence systems, Geir Stang Hauge, CEO of Norwegian-based company BioSort AS notes, identification can be much easier and efficient.

BioSort Elv is a prototype system for automatically detecting and crucially capturing farmed salmon in rivers. In essence, the Elv acts as a fence that forces fish to swim through a chamber to continue their journey upstream to the breeding grounds. It is in this chamber that the magic happens. A suite of sensors that takes morphological measurements from the fish, like the size and shape of the jaw and the placement of the fins, for a computer vision system to assess. If the system determines the fish to be wild, the fish passes through. If it is farmed, then it is guided into a holding area. Tests have proven successful; “We were able to recognize and remove about 90% of escaped salmon,” Geir said. As for misclassification of wild individuals, this came in at below 3%. What’s more, when it came to hybrids, where the morphological differences are not as pronounced as between wild and farmed fish, the system had a 50% success rate.

Developing such systems – from the hardware to the software, to the integration of all the components – is a complex process. The fence itself, for example, has been created to allow river flow and small aquatic animal movement to continue as well as allow large items such as branches or sea ice to pass over, whilst the AI needs to be trained to differentiate wild from farmed fish. All this comes at a cost that needs to be met. “The development costs to commercialize this type of product can’t be supported by an initiative to equip a few rivers, it has to be bigger to develop a good system”, Geir explained. With funding options currently low, BioSort Elv currently remains a prototype. Instead, BioSort is currently focusing on its iFarm system that uses Fish Facial Recognition to build medical records of all farmed fish the pens. In the future, Geir says, this Fish ID system can also be used to trace the origin of escaped salmon, and even acquit individual fish farmers of being the source of a specific escape.

Capturing escapees is one thing - reducing their occurrence in the first place is another. Operational changes and better training may help reduce escapes caused by human error. Farms can be designed to withstand strong storms, and nets created from materials that reduce the chances of predators like sharks creating holes for the fish to escape through in the first place. For many though, there is one clear solution – land-based aquaculture. “Obviously, on land, you're not going to get anywhere near the escapees that you get if your pen is out in at sea,” Perry said. AquaBounty agrees – which is why it employs Recirculating Aquaculture Systems (RAS) for the entire life cycle of their salmon.

RAS are particularly efficient forms of land-based aquaculture, allowing water to be recycled in the tanks, waste removed and disposed of properly, and conditions carefully controlled. “Having complete control of the environment in which our fish grow enables us to optimize their growing conditions so that we get the best performance,” Conley explained. “Unlike in sea-pen salmon farming, our salmon are not exposed to parasites and pathogens requiring the use of antibiotics or chemical treatments”.

Whilst the chance of an individual escaping from a land-based tank is already very low, to minimize the risk further AquaBounty have also installed a series of barriers; “These barriers include various types of netting, screens, filters, sumps, and chemical disinfectants in effluent discharges to kill all life stages. In-line pumps used to move the water around the system also act as physical barriers as no fish or egg would survive the impellers, which act like a blender on any solid” Conley said.

Land-based systems are the exception rather than the norm, largely because the costs of such farms are greater than those in the sea. Today, with the drive for innovation and improvement in technology, those costs are falling. “Historically, the cost of land-based RAS farms was a barrier, but that cost has become competitive relative to other production costs in the traditional sea-pen farming”, Colney said. With start-ups like Atlantic Sapphire Whole Oceans gaining support, land-based aquaculture looks set to grow. With ever-increasing demands for seafood, whether it will result in a decline in sea-based farming is another matter.

This story appeared in Eco Magazine.