In contrast to extensive and semi-intensive farming systems where the cultured species derive all or a substantial part of their nutrient needs from naturally available pond food organisms, fish and shrimp maintained under intensive clear-water culture conditions (ie. cement tanks or raceways, and cages suspended in open water bodies) are totally dependent on the external provision of a nutritionally ‘complete’ diet throughout their culture cycle (for definition see 1.1.4).
The development of artificial diets (ie. complete pelleted feeds) for use within intensive aquaculture production systems has been based largely on the nutritional and manufacturing techniques pioneered and developed by the intensive poultry industry. Whilst this ‘borrowed’ technology may suffice for the development of complete diets intended for use within intensive clear-water aquaculture systems and for fish species which rapidly consume their food (ie. such as salmonids), the aquatic environment poses many technological and nutritional difficulties for the aquaculture feed technologist; for example, developing artificial diets for use within semi-intensive earth pond production systems (due to the presence of natural live food organisms and the difficulties of assessing their role in the overall nutritional balance of the cultivated species), and developing rations for shrimp and fish species which have very slow demersal feeding habits and who require to masticate their food externally prior to ingestion (due to difficulties of feed disintegration and loss of water-soluble nutrients through leaching). Furthermore, in contrast to the intensive poultry industry, where basic dietary nutrient requirements (including those for metabolizable energy) are well established, there is scant information on the basic dietary nutrient requirements of the major cultivated aquaculture species (Tacon, 1987). At present, these short comings are overcome in-part by formulating diets with a high inherent nutrient safety factor; the use of which being economically justifiable only because of the generally high market value of the major commercially cultivated aquaculture species (ie. marine shrimp and fish, salmonids, freshwater prawns). Clearly, this situation must be rectified if maximum economic benefit is to be gained from the farming system employed.
It must be stated at the outset that the success of a complete diet feeding strategy based on the use of a dry or semi-moist pelleted feed is dependent on five important factors, namely:
The nutritional characteristics of the diet formulated (ie. ingredient selection, nutrient level, digestibility and quality control).
The manufacturing process used to produce the formulated ration and the physical characteristics of the resultant diet (ie. cold pelleting, steam pelleting, grinding, microencapsulation, air dried, freeze dried, sun dried, pellet, feed mash, feed size, shape, colour, texture, bouyancy or spacial behaviour within the water column, and water stability).
The handling and storage of the manufactured diet prior to usage on the farm (ie. length of storage, environmental storage conditions - temperature, humidity, irradiation, ventilation, and packing material used).
The feeding method employed (ie. hand or mechanised feeding, feeding frequency, feeding rate - feeding tables, satiation feeding, or demand feeding).
The water quality of the culture system (ie. temperature, photoperiod, dissolved oxygen and mineral concentration, salinity, turbidity and water circulation pattern).
Each of the above mentioned factors hold equal importance; the failure of one reducing the effectiveness of the others. Clearly, the nutritionist must work in tandem with the feed processor and aquaculture biologist if maximum nutritional and economic benefit is to be gained from complete diet feeding.
In this review emphasis will therefore be placed on the multi-disciplinary approach toward complete diet feeding, and in particular concerning the resolution of those nutritional and technological problems unique to aquaculture, rather than reciting well established animal feed formulation and manufacturing techniques. Readers requiring detailed information on the commercial formulation and fabrication of complete feeds should consult the excellent reviews of ADCP (1980, 1983), Cho, Cowey and Watanabe (1985), Coll Morales (1983), Crampton (1985), Halver and Tiews (1979), New (1987), Palmer-Jones and Halliday (1971), Pfost and Pickering (1976), and Robinson and Wilson (1985).
The following factors should be considered when formulating a practical fish or shrimp ration for use within a intensive aquaculture system:
Market value of the aquaculture species to be farmed: as a rule of thumb feed cost should not exceed 20–25% of the farm gate value of the cultured fish or shrimp (ADCP, 1983; Crampton, 1985); a high product market value justifying (if so required) the selection and use of more expensive and higher quality feed ingredients and feed manufacturing techniques (ie. such as the use of rehydratable expanded pelleted feeds for shrimp and marine flatfish, Cadena Roa et. al., 1982).
Feeding behaviour and digestive capacity of the species to be farmed: is the aquaculture species in question a carnivore, omnivore or herbivore; a benthic, pelagic or surface feeder; a day-time, twilight or nocturnal feeder; a visual or olfactory feeder; a moist or dry diet feeder; a rapid or slow feeder; and does the species in question have an acid secreting stomach and possess a full compliment of digestive enzymes. These factors, together with the proposed production unit (ie. tank, cage or earth pond) will dictate if a floating, slow-sinking or sinking feed is required, and will also determine the physical characteristics of the feed to be produced (ie. size, bouyancy, colour, texture, palatability and desired water stability). Similarly, the formulation of a diet for a stomachless fish species or a shrimp will necessitate the use of dietary phosphorus and calcium sources which release their native elements within an alkaline digestive environment (ie. such as the use of plant phytates, whey products and skim milk powder; Tacon, 1987).
Intended feed manufacturing process to be used: straight mixing, cold pelleting, conventional steam pelleting, expansion steam pelleting, flaking, crumbling or microencapsulation. For example, expansion steam pelleting requires the presence of relatively high amounts of starch containing cereal grains within the formulation so as to facilitate starch gelatinization and obtain the desired expansion texture; cold pelleting techniques require the use of special binders which do not have to be activated by heat (ie. such as the use of alginate binders within semi-moist pelleted fish feeds); and microencapsulation techniques for larval shrimp require the use of soluble and highly digestible dietary protein sources such as egg proteins and invertebrate tissue homogenates (Cho, Cowey and Watanabe, 1985).
Dietary nutrient requirements of the species to be farmed: including the dietary protein, essential amino acid, essential fatty acid, vitamin, mineral and energy (if known) requirements for all stages of the life cycle. For suggested dietary nutrient specifications see ADCP (1983), New (1987), NRC (1983) and Tacon (1987; Tables 15–18).
Available feed ingredient sources: nutrient content of available feed ingredients, including quality control and cost (at source and with transportation). The availability, nutritional quality and cost of individual feedstuffs (including micro-nutrient sources such as vitamins, minerals, amino acids, antioxidants and mould inhibitors) will dictate the type of ration which can be formulated. General quality control guides for feed ingredients intended for use within complete diets have been given by ANFAR (1985), Cho, Cowey and Watanabe (1985), Cockerell, Halliday and Morgan (1975), Cooley (1976), Deyoe (1976), Smith (1986) and Tacon (1987a). For example, Table 1 presents the recommended nutritional quality of fish meal and fish oil for salmonid rations.
Table 1. Recommended quality of fish meal and oil for salmonid diets 1
| Compound | Level |
| Fish meal 2 | |
| Crude protein | >68 % |
| Crude lipid | <10 % |
| Ash | <13 % |
| Salt | <3 % |
| Ammonia-N | <0.2 % |
| Moisture | <10 % |
| Antioxidant (sprayed liquid form) | 200 ppm |
| Steam processed, ground finer than 0.25 mm | |
| Fish oil | |
| Peroxide value | <5 meq/kg |
| Anisidine value | <10 |
| Total pesticides | <0.4 ppm |
| Polychlorinatedbiphenyls (PCBs) | <0.6 ppm |
| Nitrogen | <1 % |
| Moisture | <1 % |
| Antioxidant (liquid) | 500 ppm |
1 From Cho (1980)
2 Heavy metal content of the meal should also be checked
Digestibility of the ingredient sources to the fish or shrimp species: the biological availability of the individual nutrients (ie. protein, amino acids, lipid, carbohydrate, minerals, vitamins and energy) contained within ingredient sources to the fish or shrimp species in question. For example, Table 2 summarises the apparent digestibility coefficients of some common feed ingredient sources for rainbow trout (Salmo gairdneri).
Table 2. Apparent digestibility coefficients of selected ingredients for rainbow trout 1
| Ingredient | Dry matter (%) | Crude protein (%) | Lipid (%) | Gross energy (%) |
| Alfalfa | 39 | 87 | 71 | 43 |
| Blood meal (spray dried) | 91 | 99 | - | 89 |
| Blood meal (flame dried) | 55 | 16 | - | 50 |
| Brewers dried yeast | 76 | 91 | - | 77 |
| Corn, yellow | 23 | 95 | - | 39 |
| Corn gluten feed | 23 | 92 | - | 29 |
| Corn gluten meal | 80 | 96 | - | 83 |
| Corn distillers dried soluble | 46 | 85 | 71 | 51 |
| Fish meal, herring | 85 | 92 | 97 | 91 |
| Hydrolysed feather meal | 75 | 58 | - | 70 |
| Meat and bone meal | 78 | 85 | 73 | 85 |
| Poultry by-product meal | 52 | 68 | 79 | 71 |
| Rapeseed meal | 35 | 77 | - | 45 |
| Soybean, full fat, cooked | 78 | 96 | 94 | 85 |
| Soybean meal, dehulled | 74 | 96 | - | 75 |
| Wheat middlings | 35 | 92 | - | 46 |
| Whey, dehydrated | 97 | 96 | - | 94 |
| Fish protein concentrate | 90 | 95 | - | 94 |
| Soybean protein concentrate | 77 | 97 | - | 84 |
1 From Cho, Cowey and Watanabe (1985)
It must be emphasised here that the digestibility of individual feed ingredients will vary depending on 1) the physical and nutritional characteristics of the material under test, 2) the manufacturing process employed in the preparation of the feed ingredient, 3) the dietary inclusion level of the feed ingredient envisaged, 4) the developmental status and digestive capacity of the fish or shrimp species in question, 5) the feeding method employed (ie. force feeding, satiation feeding, or use of a restricted feeding table), and 6) the experimental technique employed for estimating nutrient digestibility (for review see Austreng, 1978; Cho, Slinger and Bayley, 1982; Cho, Cowey and Watanabe, 1985; Choubert, De la Noüe and Luquet, 1982; Jobling, 1983; NRC, 1983; Tacon and Rodrigues, 1984; Talbot, 1985; Wilson and Poe, 1985; Smith, Peterson and Allred, 1980; Kirchgessner, Kürzinger and Schwarz, 1986; De la Noüe and Choubert, 1986; Vens-Cappell, 1985; Atkinson, Hilton and Slinger, 1984; Buddington, 1980). In view of the above factors it is clear that each ingredient (from what ever source) must be considered as being unique and evaluated (in terms of nutrient digestibility) on its own merits. In view of the difficulties encountered with the quantitative collection of faeces within an aquatic environment and the different methods employed by individual nutrition laboratories for estimating nutrient digestibility, further research is required in this subject area before confidence can be given to the digestibility coefficients obtained.
Ingredient handling/processing requirement prior to mixing or pelleting: including service (ie. electricity, water, fuel) and equipment (grinder, mincer, vibrating sieve, freezer) cost. For example, the high energy and equipment cost of fine grinding ingredients on site prior to use within fry or larval diet formulations may be reduced in-part by selecting pre-ground feed ingredient sources. Similarly, the relatively high cost of storing whole trash fish on site by freezing may be overcome by the use of less expensive acid silage techniques (Jackson, Kerr and Cowey, 1984).
Maximum/minimum dietary feed ingredient constraints: in terms of available nutrient level, possible dietary interactions with other ingredient sources (antagonisms), known endogenous anti-nutritional factors, milling and pelleting constraints, and dietary tolerance to the fish or shrimp species in question. For example, Table 3 shows the formulation constraints imposed by Robinette (1984) and Crampton (1985) for a practical pelleted diet for channel catfish (Ictalurus punctatus) and a salmonid, respectively. However, it must be stated that it is up to the user (ie. the person formulating the diet) to set his or her own dietary nutrient and ingredient constraints. There are no hard and fast rules or ‘magic formula’; the nutritional quality and cost of individual ingredient sources varying considerably from one factory or region to another, and fish and shrimp species differing in their dietary tolerance to individual feed ingredient sources (Crampton, 1985; Cho, Cowey and Watanabe, 1985). For the most part dietary nutrient and ingredient constraints are set either through practical trial and error, or through the ‘experience’ or ‘slimy’ fingers of the aquaculture feed technologist; the formulation of complete aquaculture diets still being an art rather than a science. Despite this, some dietary constraints are concessions to milling/pelleting considerations, and consequently are well understood. For example, the use of at least 20% digestible carbohydrate within expanded pelleted feeds (ie. such as corn; expansion pelleting requiring the presence of adequate amounts of starch for sufficient gelatinization, Table 3), the restriction of total dietary lipid during pelleting to 8% (ie. lipid levels above this reducing the binding quality of the pelleted feed; additional lipid may be added if required after the pellet has been formed by spraying on to the outside of the freshly pelleted feed), or the use of dietary binding agents to maintain pellet quality and reduce the dust content of the final pelleted feed (Robinson and Wilson, 1985; Cho, Cowey and Watanabe, 1985). For a review of the nutritive value of individual feed ingredients for fish and shrimp see Deshimaru (1981), Kanazawa (1983), New (1987), Spinelli (1980), Tacon and Jackson (1985), and Tacon (1987a).
Performance of the formulated feed: in terms of handling characteristics (pellet quality and durability), cost, effect on water quality (pollution effect), effect on fish or shrimp carcass quality (ie. pigmentation, lipid content, taste), and effect on fish or shrimp growth and survival (ie. food conversion efficiency, growth rate, and food to fish/shrimp cost).
Table 3. Formulation constraints imposed for a 32% crude protein extrusion processed (expanded) pelleted feed for channel catfish and a 45% crude protein conventional steam pelleted feed for a salmonid (rainbow trout) 1
| Nutrient/ingredient | Dietary constraint | |
| Minimum (%) | Maximum (%) | |
| Channel catfish pellet 2 | ||
| Crude protein | 32 | - |
| Lipid | - | 6 |
| Crude fibre | - | 7 |
| Calcium | - | 1.5 |
| Phosphorus (available) | 0.5 | 0.7 |
| Digestible energy (kcal/kg) | 2618 | - |
| Lysine | 1.63 | - |
| Methionine | 0.30 | - |
| Methionine + cystine | 0.74 | - |
| Yellow corn | 25 | - |
| Wheat | 2 | - |
| Cottonseed meal 3 | - | 10 |
| Menhaden fish meal | 8 | - |
| Vitamin premix | 0.1 | 0.1 |
| Mineral premix | 0.1 | 0.1 |
| Salmonid pellet 4 | ||
| Crude protein | 45 | 100 |
| Lipid | 15 | 100 |
| Carbohydrate | 5 | 20 |
| Crude fibre | 0 | 100 |
| Ash | 5 | 18 |
| Methionine | 0.75 | 100 |
| Methionine + cystine | 1.75 | 100 |
| Lysine | 2.3 | 100 |
| Available lysine | 2 | 100 |
| Metabolizable energy (kcal/kg) | 3000 | 100 |
| Fish meal content | 45 | 100 |
| Fish meal, Scottish | 15 | 100 |
| Fish meal, Danish | 15 | 100 |
| Fish meal, spray dried | 0 | 15 |
| Blood meal, spray dried | 0 | 5 |
| Soybean meal, extracted | 0 | 15 |
| Meat and bone meal | 0 | 10 |
| Fish oil | 2 | 13 |
| Wheat | 0 | 20 |
| Poultry by-product meal | 0 | 5 |
| Distillers by-product meal | 0 | 5 |
| Vitamin mix | 1 | 1 |
| Mineral mix | 1 | 1 |
| Pellet binder | 0.1 | 0.1 |
1 Least-cost formulation constraints imposed by linear programming
2 Data from Robinette (1984)
3 Amount used depends on level of free-gossypol in the meal
4 Data from Crampton (1985)
The object of feed formulation is to mix ingredients of differing nutritional quality so as to obtain a balanced diet whose biologically available nutrient profile approximates to the dietary needs of the animal in question. However, the formulation of a practical diet is largely a compromise between what would be ideal from a nutritional viewpoint (ie. such as the production of a diet in which the protein component is entirely supplied by a high quality fish meal), and what is practical and economical. Ideally the formulation ought to reflect the nutrient requirements of the animal exactly without surpluses (but allowing for manufacturing losses), but in practice the nutrient requirements are not known with certainty and the biological availability of nutrients in feed ingredients is very often unknown. In industry there is a further object, namely, to derive a balanced diet at least cost. This is the diet that is the least expensive of the possible dietary formulae which satisfy the criteria and will meet the standards at least cost to the farmer.
In aquaculture feeding the number of possible ingredient sources and the number of nutrients in each for which requirements have been estimated are extensive. It follows therefore that a large number of arithmetric calculations are required to produce a least-cost diet. Although ‘hand’ formulations using a simple calculator may be sufficient (if not a little tedious and time consuming) for the scientist or farmer wishing to formulate a feed for use within his or her own laboratory or farm, this is not the case for the commercial feed manufacturer where time is money and profits are made or lost in the area of ingredient buying, negotiating contracts and raw material handling; feed ingredient costs currently representing about 80% of the finished feed price. To meet this demand the animal feed manufacturing industry has been employing the computer technique of linear programming since the mid-nineteen fifties. Linear programming is essentially a mathematical tool by which resources are evaluated or selected to achieve an optimal solution to a problem. The value of linear programming is in the number of ingredients and number of requirements or restrictions which can be handled in a short period of time.
Certain data must be supplied to the computer together with the programme to be executed. These data are:
the detailed feeding standard which has to be satisfied (ie. dietary nutrient requirements), together with any permitted deviations in each nutrient;
the detailed nutrient composition of each potential ingredient;
any restriction on the proportion of the final mix that any one ingredient may represent, this can be a maximal or minimal value and in computer jargon is known as a constraint;
the cost per unit weight of each ingredient.
The formula calculated by the computer will be that meeting the specification at the least cost of ingredients, hence any extra mixing or handling charge due to the inclusion of a certain ingredient must be added to the cost per unit weight of that ingredient or else added to the cost of the formula.
When using a computer it must be borne in mind that the adequacy of a diet compounded to the resultant formula will be affected by two main factors: the extent to which the feeding standards (ie. nutrient requirements) adopted adequately represent the biological needs of the fish or shrimp and the accuracy with which the amounts of nutrients in the ingredients available to the fish or shrimp are known.
Readers requiring detailed information on ‘hand’ or ‘linear programming’ formulation techniques should consult the reviews of Cho, Cowey and Watanabe (1985) and New (1987), and Barbieri and Cuzon (1980), Cho, Rumsey and Waldroup (1980), Crampton (1985), Thong (1985) and Poornan (1987), respectively. The basic procedure for formulating a complete diet has been described by Cho, Cowey and Watanabe (1985) and is summarized in Figure 1.
| ESSENTIAL NUTRIENTS REQUIRED (Proteins, amino acids, fatty acids, carbohydrates, vitamins, minerals, energy, energy/protein balance) |
| ↓ |
| SELECTION OF INGREDIENTS (Composition, digestibility, quality control, and cost) |