This paper describes an industrial research approach to fish farming and illustrates where appropriate how it differs from the research strategy employed by universities and governmental bodies. To illustrate this the Unilever experience of the establishment of salmon farming from a research concept to a production and marketing operation is discussed in detail. Three separate phases in the process of establishing industrial fish farming were undertaken and were entitled ‘Biological, Technological and Commercial’. The research programme involved in each of these phases is described in detail.

The salmon farming cycle which was arrived at after the biological phase is described in detail. The technological phase involved the translation of this biological phase into a viable farming operation. This enabled a computerised cost model of the salmon farming process to be built. This identified the six or seven crucial areas which had to be improved if the enterprise was to be profitable—the commercial phase.

To demonstrate the way the nature of the research programme has changed during the period 1968–76 a graph is given showing the main discipline involved in each year and how these have changed over that period.

The distributions of the four variables—temperature, salinity, currents and dissolved oxygen—are discussed with particular reference to the Scottish west coast. In fjords, the distributions are related to the sources of change—tides, freshwater runoff, winds and, in the case of currents, to friction, which makes currents very variable over small areas. Current speed in floating cages has been found less than that outside. A model of the oxygen concentration in such a cage has been used to examine the conditions of current and fish stocking density under which the cage can be operated without reaching dangerously low oxygen concentrations.

A 3-year survey of wild fish in Loch Awe and the surrounding watershed has shown a restricted distribution and low prevalence of infection with infectious pancreatic necrosis virus. No signs of clinical disease were observed in infected fish. The restricted distribution close to a rainbow trout farm is consistent with the virus originating from this farm, where virus infected stocks have been found each year since 1971. The continuing low prevalence is also consistent with virus in wild fish resulting from the spread of infection from the farm. If infection is spreading between wild fish, it is slow within the time scale of the survey. The possible mechanisms of transmission of infection from the farm to wild fish are discussed.

Two behavioural responses were considered. In the first, rainbow trout were trained to press a trigger to obtain food, and in the second they were conditioned to respond to a sound signal in anticipation of food delivery. The trigger pressing, demand feeding, response was quickly learned and the pattern of feeding analysed. There was a dusk peak of feeding as well as appreciable nocturnal trigger pressing. Feeds were highly aggregated into bouts, more so in the case of groups of trout than with individuals. This suggests two behavioural mechanisms, one operating within individuals and a second social effect. The preferred time between trigger presses was 8–16 min. Total daily food intake could be controlled by adjusting the reward level per trigger press so that demand feeding was not necessarily ad lib feeding. Rainbow trout would respond to a sound signal of 100 Hz played for 30 seconds prior to a food delivery by showing a feeding type of behaviour in response to the sound only. Experiments are planned to try and use this sort of response to recall fish from a relatively free-swimming environment to a central point for feeding and eventual recapture.

Administration, by dietary supplement, of ethylestrenol to rainbow trout and to salmon parr results in significant increases in growth rate; further treatment of smolts retained in fresh water reduces the growth rate. Treated parr develop into 1 + smolts in significantly lower proportion than controls, but have a higher condition factor.

Stress induces in rainbow trout long-lasting reductions in plasma thyroxine levels and increases in serum cortisol concentrations. The origin and significance of these changes is discussed.

The administration of 17β-oestradiol at 20 mg/kg of food for the first 30 days of feeding causes sex-reversal of male salmonids. Treatment with 17α-methyltestosterone at 3 mg/kg food for 90 days following first-feeding causes sex-reversal of female salmonids.

There is considerable interest in the cultivation of salmonids in several countries. In Norway it is now a question of rate of expansion rather than a question of whether this industry will develop or not. Interest is turning more and more towards sea farming based on Atlantic salmon rather than on rainbow trout.

One of the most important economic factors in fish fanning is the producing ability of the fish. It is essential for the future expansion of fish farming that one develops a more productive animal which is better adapted to captivity.

Progress in the development of marine flatfish farming suggests that, for two species, commercial viability is becoming a real possibility.

Dover sole can be reared in tanks on shore supplied with warmed water from the cooling of coastal power stations. Turbot can be reared likewise, or in cages floating in the sea.

The main area in which problems remain is in the controlled production of juveniles in hatcheries on shore, but recent progress has been good.

There is a good potential in the UK for marine fish farming, with probable sites available for an annual output of all species of up to around 50,000 tonnes p.a. At this level, or even at 25% of this level, the requirement for the catching of fish as farm fodder would become significant to the fishing industry.

This paper describes only those aspects of progress or outstanding problems which, in the author's opinion, are of major significance in economic terms.