Submerged cage aquaculture of marine fish: A review of the biological challenges and opportunities

Industrial marine fish farming is a relatively young phenomenon but has grown to be a major industry in many regions of the world, producing some 6.6 million tons of fish per year.1 The standard production units, seacage fish farms, are variations on a common theme,floating, surfacebased structures holding large nets which contain thousands to hundreds of thousands of fish. The genesis of this technology came from the first Atlantic salmon farms in the 1960s and 1970s in Norway and Scotland, where nylon trawl nets were hung from wooden or polyethylene pipe structures.2,3 Although more archaic forms of caged aquaculture have long been practised elsewhere, such as Asia,4 shifts Received: 12 November 2020 | Revised: 3 March 2021 | Accepted: 16 May 2021 DOI: 10.1111/raq.12587

to commercial-scale marine cages didn't occur here until the late 1970s -the early 1980s. 5 Stepwise innovation of this technology has generated the modern, highly engineered structures which dominate production today, with nets hung from either steel platforms or circular plastic rings ( Figure 1). Most major commercial marine finfish aquaculture operations worldwide have adopted this production system because it is proven to be effective and comes production-ready 'off the shelf'.
Despite their widespread use, a range of issues are associated with surface-based production, including net deformations and cage breakdowns from storms which can lead to escape events, parasites and diseases, algal and jellyfish blooms, and the presence of less-than-optimal culture conditions such as high temperatures, low oxygen levels and contaminants from freshwater inputs (see Table 1 for a full list of problems). Further, several commercially TA B L E 1 Hazards, depth of influence (the experience of the hazard within a pen), estimated duration of unsuitable surface conditions, the production problems caused for finish aquaculture in sea-cages, and example source references

Depth of influence (m) Duration Production problem Source
Storm 0-10 Hours-weeks Cage and or net rupture and subsequent escapes 7,8 Current/net deformation 0-20 Hours-weeks Net deformations leading to excessive crowding of fish 105 Ice Surface structures Hours-days Cage damage leading to escapes 106 Algal bloom 0-20 Days-weeks Fish mortality and sub-lethal effects on welfare 107,108 Jellyfish bloom 0-10 Hours-weeks Fish mortality and sub-lethal effects on welfare 109,110 Parasitic lice larvae L. salmonis on salmonids 0-5 Persistent Infestation, leading to reduce growth when severe, and lethal and sub-lethal effects due to treatments 111 Parasitic lice larvae C. rogercresseyi on salmonids 0-10 Persistent Infestation, leading to reduce growth when severe, and lethal and sub-lethal effects due to treatments 112 Parasitic skin fluke N. girellae on farmed kingfish 0-5 Persistent 12 Amoebic gill disease 0-5 Weeks Reduced gill health 13 Tapeworms (Eubothrium sp.) 0-10 Weeks Growth reduction 10 Reduced oxygen Variable Hours-weeks Loss of appetite, reduced growth rates 113 Unsuitable temperature 0-10 a Hours-weeks Loss of appetite, reduced growth rates 114 High aluminium levels 0-2 Days-weeks 115 Biofouling 0-10 a Summer, autumn Low oxygen levels when severe; poor water quality after cleaning of cages 103,116 a Variable, but usually greatest in surface waters.
important species such as sea bream (Sparus aurata), Atlantic cod (Gadus morhua) and cobia (Rachycentron canadum) are benthopelagic or benthic in nature, so production in surface sea cages may not provide ideal conditions. The production inefficiencies caused by these problems can be substantial, and the broader environmental costs of parasite transmission to wild stocks 6 and escaped fish from net breakdowns 7,8 create much of the controversy surrounding the industry and erode its public perception.
Culture in submerged cages, whether temporary or permanent, could alleviate the extent or severity of many of these problems.
Deeper environments typically have more stable temperatures and salinities, largely avoid the full impact of storms, and are less favoured by the infectious stage of problematic parasites. [9][10][11][12][13] The adoption of submerged cages may also unlock new areas for production where surface-based sea-cage technologies are inappropriate due to surface wind and waves, or by social constraints such as space conflicts with other coastal users. 14 Perhaps due to the dominance of surface-based sea-cages in the marketplace, the question of whether alternate marine production units, such as submerged cages, provide production advantages remains largely unanswered for most marine species. In addition, a range of biological and technical challenges associated with submerged culture (Table 1) have proven difficult to solve thus far, except for some species with physiologies more accepting of longterm submergence (e.g. cobia). As a result, submerged culture as a commercial method is still very much in its infancy. There are few sufficiently replicated trials that have assessed the effects of submerged culture on key production and welfare parameters, and most trials rely on data from one or few submerged cages with no or few control cages (i.e. traditional surface cages; Table 2). Such experimental designs provide minimal power to detect effects, and results generated are largely inadequate to properly assess whether submerged culture provides production advantages or disadvantages.
Still, the small but growing body of literature ( Figure 2) provides critical knowledge to further the development and application of sub-

| THE S TATUS OF SUB MERG ED C AG E AQ UACU LTU RE
Submerging cultured fish has occurred since at least the 1970s, with early experiments on rainbow trout 15 and more comprehensive trials in the 1980s with Atlantic salmon. 16,17 These were largely either short-term submergence or shallow depths (Table 2) and were often attempts to avoid temporary hazardous surface conditions (e.g. extreme winter surface cooling). In the last decade, there has been a considerable surge in research into submerged culture ( Figure 2).
To date, at least 11 finfish species that have been produced, largely experimentally, in submerged cages of various sizes, at different depths, and over various submergence durations (Table 2; it is probable that additional species have been trialled, but published research on these is not available or were not identified).
Several species appear to cope and grow well in submerged cages, yet few species have been produced at truly commercial scales in submerged cages. Collaboration between industry and research to develop a submerged culture in Costa Rica has resulted in the successful start-up of a submerged culture industry for cobia, now produced at commercial scales. [18][19][20][21][22] These cobia sustain high growth rates when reared in submerged cages 19 with relatively low ecological impacts on the surrounding environment. 22  gence as a viable production method. Salmonids grow poorly when held in submerged cages for longer than a month in the on-growing phase in seawater. 32,33 Even when continuous lighting reduced some of the negative side effects of submergence, growth rates were still TA B L E 2 Research on finfish production within submerged cages, including species information, level of replication, production parameters and location. Research identified using the search terms outlined in Figure 2 and bibliographies of those papers lower relative to surface cages. 34 In contrast, shorter-term submergence for periods less than 21 days appear to have relatively little effect on growth rates [35][36][37] and have been promoted as an effective way to avoid temporary negative surface events such as storms. 36 However, integrating an air dome into the ceiling of a submerged cage to enable salmonids to refill their swim bladders underwater 38 led to sustained good growth rates over submergence periods up

| THE B I OLOG I C AL OUTCOME S , CONS IDER ATIONS AND CHALLENG E S OF SUB MERG ED FIS H FARMING
One of the main biological considerations surrounding the adoption and success of submerged cages is centred around fish buoyancy regulation. Swim bladders make up 3%-6% of the body volume in marine fish species, and reduce the metabolic cost of maintaining buoyancy by around 90% compared to hydrodynamic compensation alone. 39 Buoyancy problems can arise in multiple ways in submerged cages, with swim bladders becoming either too full or too empty, dependent upon the physiological system a fish species possesses to fill and empty their swim bladder. Swim bladder anatomy and mechanisms for regulating volume differ among species. 40

| Swim bladder, buoyancy and maximum neutral buoyancy depth
The swim bladder in physostomous species is connected to the oesophagus via a short pneumatic duct. 41 Physostomes need to refill their swim bladder periodically by snapping and swallowing air during 'porpoising' rolls or jumps out of the water. 42 For all physostomes, achieving neutral buoyancy reduces the energetic cost of horizontal swimming and sustaining vertical position in the water column. 39 The maximum depth at which physostomous fish attain neutral buoyancy is likely an important influence of swimming depth behaviour. That fish were provided surface access weekly during submergence.

TA B L E 2 (Continued)
For example, wild Atlantic salmon spend >80% of their time in the upper 10 m of the ocean, 43 which may in part be explained by their ability to fill their swim bladder at the surface and achieve neutral buoyancy at shallow depths, but not deeper. Forcing physostomous fish to swim deeper than the maximum depth at which they are neutrally buoyant results in negative buoyancy. Therefore, determining this depth threshold is important for farmed fish that will be forcibly submerged. The extent to which a fish can fill their swim bladder will influence this neutral buoyancy depth limit.

| Growth and welfare
Achieving comparable fish growth and welfare is essential if submerged culture is to become a viable alternative to surface-based cage production. Based on the published research on physostomous fishes (mainly salmonids), comparable growth has not been achieved for a full production cycle, although most research has been shortterm (i.e. <56 days; Table 2). Short-term periods of submergence (7-22 days) of Atlantic salmon, without access to air, generally has no negative effect on growth or welfare, 36,37 but this may be due to the submergence period not being long enough for the acute effects of negative buoyancy to result in a measurable reduction in growth. One short-term submergence trial reported lower SGRs in submerged fish, but this was likely due to lower temperatures in submerged compared to surface cages. 35 Submergence for longer periods (>40 days) without access to air, led to sub-optimal growth rates and some fin and snout erosion. 32,34 The recent addition of air domes to submerged cages to resolve fish buoyancy issues resulted in a submergence trial run for ~40 days reporting no negative effect on growth or welfare on salmon. 38

| Swim bladder and buoyancy
Like physostomous fish, physoclistous fish also fill their swim bladder by swallowing air, but only when larvae. 56,57 During development, the connection between the swim bladder and gut disappears, resulting in a closed swim bladder disconnected from the external environment ( Figure 3). 41 The impact of a sudden ascent for physoclistous fish depends on the degree of pressure reduction. If the vertical distance is within the FVR and the fish can retain behavioural control (e.g. by downward swimming), any stress will likely be short-term and diminish as gas is released from the swim bladder via the oval organ and reabsorptive capillary network. Extending beyond the FVR will lead to an uncontrolled and highly stressful experience, where the lift force of the expanding swim bladder will accelerate the movement of fish towards the surface, creating a negative feedback buoyancy loop.
If a fish is unable to swim forcefully downward to a depth where swim bladder pressure is safe, it will quickly surface with an overinflated swim bladder and may experience symptoms of barotrauma, which can be lethal. 67 If rapid surfacing causes a pressure reduction greater than approximately 70%, a cod's swim bladder can rupture 68 and gas releases out the anal opening. 60 This bursting mechanism functions as a safety valve preventing a total loss of buoyancy control, with some individuals able to recover under optimal conditions. 60 Whether recovery would occur under commercial settings, however, is unclear. Other physoclists, such as red snapper (Lutjanus campechanus, Poey 1860), do not have this safety valve and the expanding gas in their swimming bladder following rapid changes in pressure often causes catastrophic decompression, which everts the stomach and bulges the eyes, leading to mortality. 61 The lifting of submerged cages with physoclistous fish must therefore be done slowly to reduce stress and limit mortality. Since sea-caged cod voluntarily ascend to depths representing a maximum of 40% pressure reduction, raising submerged cages would ideally involve lifting stages each representing a 40% pressure reduction or less with a pause of at least 10 h between each lift. 28 The vertical distance that and appetite was reduced for several days. 28 Therefore, as with cage lifting, cage lowering should be done slowly, with the FVR in mind to avoid these problems.
Future research should attempt to quantify swim bladder gas resorption rates for other physoclistous species that might be suited to aquaculture, as they likely differ from cod, and thus, differ in their tolerance to submergence and surfacing speeds.

| Swimming behaviours
The vertical movements of wild physoclistous fish are thought to depend on temperature, depth, season and ontogenetic stage. 63,69,70,71 Cultured Atlantic cod distribute shallower than wild cod, particularly wild males (~40 m depth compared to farmed fish (~20-30 m)). 71 Further, when submerged, swimming speeds (1.3-2.3 times) and tail beat frequencies (1.4-2.3 times) increase immediately, and fish swim with an average 30-degree head-up swimming angle. 28 However, cod return to normal swimming angles after 16-60 h. 28 Although comparative research on swimming behaviours of submerged physoclists is scarce, there is no evidence suggesting compromised production as a result of altered swimming behaviours under submerged conditions.

| Growth and welfare
Current evidence suggests that cultured, physoclistous fish have high growth and welfare under submerged conditions. For example, Atlantic cod submerged below 20 m for 14 months had very high survival (~99%), grew faster than estimated rates based on empirical models, 72 and had negligible problems during sexual maturation with or without artificial light. 29 Maricchiolo et al. 73 also documented similar growth rates of seabass between surface-based and submerged cages, with those reared in submerged cages also having lower stress levels (measured as higher haemolytic activity and lysozyme levels).
Finally, red porgy in submerged cages displayed more natural skin colours and had lower skin melatonin content than in surface-based cages, indicative of more optimal rearing conditions. 31

| Fish without swim bladder
Fish without swim bladders are always negatively buoyant, and cope by either continuous swimming and/or by utilising hydrodynamically efficient body shapes, and large fins and tails that generate lift with forward swimming (e.g. mackerel, tuna and cobia; Figure 3). There is no constraint with regards to vertical migration as no gas expansion or compression occurs. Consequently, these fish utilise a large depth range. Wild cobia, for example, freely swim anywhere within 100 m of the surface. 74 Cobia (and likely other species without swim bladders) do not suffer the same issues from long term submergence as physostomous and physoclistous fish, such as lacking surface access or rapid lifting of sea-cages. As such, submerged culture may be well suited to fish without swim bladders. Indeed, as mentioned, submerged cobia grew more rapidly than surface-reared cobia, suggesting submerged culture can provide a perfect match between low stress and optimal water quality. 19 In fact, cobia stocked in submerged cages are often observed spawning naturally, and several commercial submerged culture facilities are in operation. 75

| Broader challenges and bottlenecks
There are a suite of broader challenges or bottlenecks for the commercial adoption of a submerged culture of finfish, that are more related to technological, social or financial factors rather than biological. Since these are still inherently related to the specific biology of the cultured species, we briefly discuss several of these here (also

| Optimisation of environmental conditions
Fish have optimal environmental conditions at which survival, growth and condition are maximised. 77,78 In some, but not all locations, deeper waters can provide more stable or appropriate temperatures for production, salinities and oxygen levels, as they are often below thermoclines and haloclines. Increased risk of poor oxygen availability, suboptimal growth and increased mortalities occurs at the higher end of surface temperatures for salmon (>12℃), which occur in several Atlantic salmon producing countries during summer and early autumn. 48,79,80,81,82,83 Conversely, during winter when surface water is coldest, growth rates slow. Submergence to find better temperatures may provide better growth performance during these times, and short-term periodic submergence can be a solution to avoid negative surface events such as heat waves, storms or swell. As an additional benefit, less frequent or severe damage to sea-cages from storms events will lessen the number of farmed fish escaping into the wild. 8

| Reduced interaction with harmful organisms
Submergence can be an effective measure for parasite and disease control. Salmon lice (Lepeophtheirus salmonis), often regarded as the greatest threat for the sustainability, growth and social perception of much of the Atlantic salmon industry 11,84 distribute predominantly in surface layers, 9,85 so attracting or keeping fish deeper can lower infestation rates. 34

| Unlocking new production areas
The adoption of submerged cages could unlock new areas for production where surface-based sea-cage technologies are inappropriate due to surface wind and waves, or due to space conflicts with other coastal users. 14 Further, given the growing interest in offshore aquaculture, submerged cages will likely be crucial to reduce expensive construction costs and avoid large swells and extreme offshore weather events such as hurricanes. Offshore production sites have the added advantage of greater waste dispersal leading to limited benthic impacts beneath below cages, 22 which can occur in nearshore, shallow water culture. 104

| CON CLUS I ON S AND FUTURE RE S E ARCH
Submerging aquaculture cages hold the promise of providing relief from periods of less than optimal environmental conditions, reducing fish interactions with harmful organisms, and unlocking new production areas devoid of conflict with other coastal users. However, not all fish species will be similarly suited to submerged culture, and a suite of key challenges and bottlenecks stand in the way of commercial production of several species. Based on the available evidence, fewer issues exist for the submerged culture of physoclists and fish without swim bladders. Finding optimal culture sites based on the biology of the species, focusing on streamlining operational techniques, and documenting behavioural and welfare responses to long-term submergence at commercial scales will ground-truth the projected benefits of submerged culture.
Physostomous fish present unique and complex challenges for submerged culture; recent advances have overcome many of these issues. Recent developments in technologies that allow fish to refill their swim bladders while submerged via an underwater air dome 38,53 means fish can be grown in submerged cages for a full production cycle. Concerted testing at industry scale is required to unlock the potential of submerged cages for salmonids and resolve remaining production and welfare issues. The use of dynamic submergence, where cage depth is manipulated to maintain fish in the most optimal conditions in the water column year-round may reduce some of these issues.
If submerged culture is to mature and fulfil its promise, research to empirically document production and environmental benefits, and issues surrounding fish welfare throughout the production cycle needs to lead the way. Robust, industry-scale experiments of new production technologies are difficult to conduct, but possible 76,88 and significant co-investment from government and industry is required to achieve them. Conducting meaningful scientific research will thus assist in ensuring the successful adoption of submerged culture, where possible. For physostomous fish, in particular, this requires a shift from the typically short-term, unreplicated, and uncontrolled trials found in the current literature (see Table 2), towards long-term (preferably over the full production cycle), replicated and controlled trials (e.g., Warren-Myers et al. in prep). Once these are established, the technological developments required to realise functioning submergence systems that integrate the myriad of procedures (e.g. net handling, feeding, sorting, harvesting, depth preference) required in modern aquaculture can follow.

ACK N OWLED G EM ENTS
Funding was provided by the Research Council of Norway to the project "Environmental requirements and welfare indicators for new cage farming locations and systems" (Future Welfare; project no. 267800).

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.