Feeds and feedstuffs contain nutrients and energy sources essential for fish growth, reproduction, and health. Deficiencies of these substances can reduce growth rates or lead to diseases, and, in some cases, excesses can cause a reduction in growth rate. Dietary requirements can be established for energy, protein and amino acids, lipids, minerals, and vitamins.

Energy is not a nutrient—it is released during metabolic oxidation of carbohydrates, fats, and amino acids. Absolute energy requirements of the animal can be quantified by measuring either oxygen consumption or heat production. However, estimates of dietary allowances must be determined by equating animal performance with feed materials in which the amount of available energy is accountable.

This section familiarizes the user with those aspects of nutritional energetics that deal with feed energy use by the animal, energy value of feedstuffs, and dietary energy requirements. Readers who require more detailed information on physiological energetics of fish may refer to the review of Brett and Groves (1979). Those who wish to read further on nutritional energetics, with an emphasis on determining dietary energy allowances for captive fish, should refer to the reviews of Smith (1989) and Cho and Kaushik (1990).

The energy of ingested feed is divided into many components in the animal's body. An illustration of energy flow in the animal with accepted abbreviations of energy metabolism terms (National Research Council, 1981) is shown in Figure 1-1. There are many places where energy is lost between intake and recovered products. Losses occur in feces, in urine and gill excretions, and as heat. Ideally, the fish feeder needs to minimize these losses and thereby obtain maximum return as useful products. The magnitude of these losses depends primarily on characteristics of the diet and the level of feeding. The difference between intake energy (IE) and digestible energy (DE) is energy lost in the feces (FE). The inclusion of fibrous materials that are poorly digested by fish will increase the FE loss. Metabolizable energy (ME) represents DE corrected for energy lost by

FIGURE 1-1 Schematic presentation of the fate of dietary energy for fish, categorizing the losses that occur as feed is digested and metabolized, leaving a fraction of the energy to be retained as new tissue. Source: Adapted from National Research Council. 1981. Nutritional Energetics of Domestic Animals and Glossary of Energy Terms. Washington, D.C.: National Academy Press.

excretion through the gills (ZE) and urine (UE). The difference between ME and energy recovered as growth and/or reproductive products (RE) is energy lost as heat (HE). Heat loss occurs primarily by two processes: the heat increment of feeding (HiE) and maintenance heat loss (HEm).

The HiE is the increase in heat production subsequent to ingestion of feed. The factors contributing to HiE are the digestion and absorption processes (HdE), the transformation and interconversion of the substrates and their retention in tissues (HrE), and the formation and excretion of metabolic wastes (HwE). The main biochemical basis for HiE in mammals and birds is the energy required for the ingested amino nitrogen (N) to be deaminated and excreted (Kleiber, 1975); however, this represents less of an energy loss in fish because they can eliminate and products of protein metabolism (ammonia, bicarbonate, and carbon dioxide) without the need to synthesize urea, uric acid, or other similar compounds. Energy expenditures associated with diet ingestion and digestion are small compared with that associated with metabolic work (Brody, 1945). This conclusion has been reinforced by the observation that intravenous infusion of amino acids increases heat production to the same extent as does the oral administration of the amino acids (Benedict and Emmes, 1912; Borsook, 1936). HiE depends to a large extent on the balance of dietary nutrients and the plane of nutrition (Brody, 1945) and, in fish, the water temperature (Cho and Slinger, 1979). Thus, measurement of HiE for balanced feeds is more meaningful than measurement of the HiE of individual feed ingredients, because the metabolic fate of absorbed nutrients depends on the mixture absorbed and, hence, the variety of metabolic processes that are possible.

The HiE in fish is greater for diets with a high protein content than for diets with a low protein content (Cho, 1982). In mammals and birds, however, the effect of high dietary protein on heat increment is even more marked, partly because of the energy expenditure during synthesis of urea or uric acid from the deaminated nitrogen. The energy cost of synthesis for urea and uric acid is 3.1 and 2.4 kcal/g N, respectively (Martin and Blaxter, 1965). In contrast, ammonia is the primary nitrogenous waste product of protein catabolism in fish (Goldstein and Forster, 1970). Because this form of nitrogen can be readily released into the water, energy expenditure on urea or uric acid synthesis is not needed (Cowey, 1975). Cho et al. (1982) found that HiE for rainbow trout at 15°C was 5 to 15 percent of the gross energy consumed (IE) and fell as the ratio of protein to energy decreased. The HiE for livestock can be as much as 20 to 30 percent of the IE (Farrell, 1974; National Research Council, 1984). Thus, because of the lower heat increment of fish, the net energy (NE), which is the energy that is useful to the animal for maintenance and growth, in production diets is higher for fish than for warm-blooded animals.

Maintenance energy (HEm) is that required to maintain those functions of the body immediately essential to life. A major portion of this maintenance energy is spent for basal metabolism (HeE), such as respiration, transport of ions and metabolites, body constituent turnover, and circulation. A smaller portion is spent for voluntary or resting activity (HjE) and, in the case of homeothermic animals, thermoregulation of body temperature. Since fish do not regulate body temperature and they expend less energy in maintaining position in the water than do terrestrial animals in maintaining their posture, the HEm requirement of fish is lower than for homeotherms. The fasting heat production (HEf) is an approximation of the HEm. Cho and Kaushik (1990) measured oxygen consumption of fasting rainbow trout weighing 96 to 145 g at 15°C and calculated their HEf, in kcal/fish/day to be 8.85 W^0.82 where W is body weight in kilograms. Smith (1989) reported an HEf value of 4.41 W^0.63 for rainbow trout weighing 4 to 50 g at 15°C where fasting heat production was measured directly by placing the fish in a calorimeter. Brett and Groves (1979) recommended the exponent 0.8 for metabolic body size for fish. When these HEf values for fish are compared with 70 W^0.75 for mammals and 83 W^0.75 for birds (Brody, 1945), it is apparent that the fasting heat production of fish is much lower. The maintenance energy requirements of fish are one-tenth to one-twentieth of those of homeothermic animals of similar size in a thermoneutral environment (Brett, 1973). The lower maintenance requirement for fish means that the percentage of net energy that is not dissipated as heat but retained within the body as new tissue or recovered energy is greater.

The energy content of a diet depends on its chemical composition, with the mean values of heat of combustion of protein, lipid, and carbohydrate being 5.64, 9.44, and 4.11 kcal/g, respectively. However, the chemical makeup of the diet influences only its heat of combustion, or gross energy, and yields no information on whether the energy and nutrients are available to fish through the digestive process. Prior to formulating diets, therefore, it is necessary to know the bioavailability of the energy in the feedstuffs for the animal being fed.

Available energy values for feedstuffs for fish have been determined on a DE and ME basis. ME, where applicable, is a more exact measure of the energy value for a complete diet that becomes available for metabolism by the animal. Practically, ME offers little advantage over DE in evaluating useful energy in feedstuffs for fish because FE accounts for most of the excretory losses. Energy losses through ZE and UE by fish are smaller than nonfecal energy losses by mammals and birds, and they do not vary among feedstuffs as much as do FE losses. Furthermore, determining ME values with fish is difficult because of the need to force feed

and restrain the fish in metabolism chambers with the aid of a collar for simultaneous collection of fecal, gill, and urinary excretions (Smith, 1976). DE values are generally easier to determine and the fish feed voluntarily (Page and Andrews, 1973; Cruz, 1975; Cho and Slinger, 1979; Takeuchi et al., 1979). However, the use of proper techniques is necessary to give reliable DE values for fish. The collection of feces without the leaching of nutrients is important in determining DE with fish. Early studies (Smith and Lovell, 1973; Windell et al., 1978) showed that improper collection of feces, such as allowing feces to remain in the fish tank too long, caused serious overestimation of digestion coefficients. Methods for determining DE and ME in fish are discussed in Chapter 4.

Both proteins and lipids are highly available energy sources for fish (Cruz, 1975; Smith, 1976; Popma, 1982). The value of carbohydrate as an energy source is variable among species. Nile tilapia (Popma, 1982) and channel catfish (Wilson and Poe, 1985), which are warm-water omnivorous species, digest over 70 percent of the gross energy in noncooked starch while rainbow trout, a cold-water carnivore, may digest less than 50 percent (Cho and Slinger, 1979). Cooking, as in extrusion processing of feeds, increases digestibility of starch for fish. Extrusion processed corn had a 38 percent higher DE for channel catfish than compression pelleted corn (Wilson and Poe, 1985) and gelatinized starch had a 75 percent higher DE for rainbow trout than raw starch (Cho and Slinger, 1979).

Energy intake is a basic nutritional requirement because maintenance of life processes takes priority over growth and other functions. Thus, energy concentration should be the

first nutritional consideration in diet formulation for fish. In practice, however, protein is usually given first priority because it is more expensive than other energy yielding components. Protein and energy should be kept in balance. A dietary deficiency or an excess of DE can reduce growth rates of fish. A diet deficient in energy in relation to protein will mean that protein is used for energy to satisfy maintenance before growth. In contrast, a diet containing excess energy can reduce feed consumption and thus lower the intake of the necessary amount of protein and other essential nutrients for maximum growth. Excessively high ratios of energy to nutrients can also lead to deposition of large amounts of body fat, which can be undesirable in food fish.

Ratios of digestible protein to DE (mg/kcal) for maximum weight gain for several fish species have been measured in growth studies (Table 1-1). Values range from 81 mg/kcal to 117 mg/kcal and are substantially higher than protein-energy ratios for swine and poultry, which range from 40 to 60 mg/kcal (National Research Council, 1984, 1988). The reason the protein-energy ratio for fish is higher than that for farm animals is not because fish have a higher protein requirement (fish convert dietary protein into tissue protein about as efficiently as warm-blooded animals [Smith, 1989]) but because fish require less energy for maintenance and the synthesis of uric acid.

Since lipid is the primary nonprotein energy source in salmonid diets, the protein-energy allowance for these diets is sometimes reported as the ratio of protein to lipid. The optimum combination for weight gain for rainbow trout was 35 to 36 percent protein and 15 to 16 percent lipid (Watanabe et al., 1979; Cho, 1982).

Empirical calculation of energy requirements of fish based on energy losses and expected energy recovery are possible with reliable information on energy balances in the

animal under a given set of conditions. Energy balances for rainbow trout have been established under laboratory conditions, as discussed previously. Cho and Kaushik (1990) constructed a model for calculating the DE required to grow 1 kg of rainbow trout, from 1 g to 100 g size at 15°C, based on derived heat and excretory losses and estimated recovery of energy in the fish. The model indicated that 3.56 Mcal of DE would be required to produce 1.91 Mcal of recovered energy in 1 kg of fish biomass with an RE:DE efficiency ratio of 0.54, which is comparable to a value of 0.56 reported for channel catfish (Gatlin et al., 1986). However, several factors significantly affect energy balance in fish, such as diet composition, feeding rate, and composition of body gain. Therefore, this approach to calculating energy requirements for production diets must be used cautiously until sufficient information is available to establish reliable energy budgets for a variety of production conditions for a specific aquaculture species.

Proteins are composed of up to 20 α-amino acids linked into chains by peptide bonds. The chains are cross-linked by disulfide bridges, hydrogen bonds, and van der Waals forces. The amino acid content of proteins, particularly feed proteins, may differ markedly. Some, such as gelatin (a mixture of proteins derived from collagen) or zein (a protein from maize gluten), are largely, or even entirely, deficient in one or more amino acids. Others, such as fishmeal, have a balance of amino acids that more closely meets the requirements of fish. Consequently the capacity of different feed proteins to meet the amino acid needs of the fish will differ considerably. Ingested protein is hydrolyzed to free amino acids, dipeptides, and tripeptides by digestive enzymes secreted into the gastrointestinal tract. These products are absorbed by the mucosal cells where intracellular digestion of small peptides occurs; thus only amino acids appear to be released into the portal vein as products of protein digestion (Murai et al., 1987). Some evidence has shown that small amounts of certain whole proteins may be absorbed through the wall of the gastrointestinal tract, but the quantities involved have not been confirmed as being of any quantitative significance (Ash, 1985).

In the context of animal feeding, protein generally refers to crude protein (CP); that is, N × 6.25, a definition based on the assumption that proteins contain 16 percent N. The requirement for dietary protein has two components:

1. a need for indispensable amino acids that the fish cannot synthesize either at all or at a rate commensurate with its need for protein deposition or commensurate with the synthesis of a variety of other compounds with metabolic functions and

2. a supply of either dispensable amino acids or sufficient amino nitrogen to enable the fish to synthesize them.

Insofar as synthesis of dispensable amino acids requires expenditure of energy, feeding dietary proteins that most nearly meet the needs of fish for both indispensable and dispensable amino acids will result in the most efficient growth by the fish. Thus, the concept of balance or pattern of amino acids is basic to protein requirement.

The protein requirements, meaning the minimum amount needed to meet requirements for amino acids and to achieve maximum growth, have now been measured in juvenile fish of many species (see Tables 1-2 to 1-13). They have been obtained mainly from dose-response curves in which graded amounts of high-quality protein were fed in partially defined diets. The response measured was weight gain. The values are expressed as a percentage of dry diet. Although the expression of protein as a proportion of dietary energy would have focused attention on protein as a substantial source of dietary energy, this approach was not possible for many of the data because in formation on the DE content of the diets was unavailable and values used for the energy density of dietary components varied between authors.

The protein allowances in fish diets are appreciably higher than those in the diets of terrestrial warm-blooded animals. The methods used to determine protein requirements, however, may overestimate requirements, in that excess dietary protein or amino acids, which cannot be stored, are catabolized preferentially over carbohydrates and fats and used for energy by some fishes (Wilson, 1989). In addition, adequate consideration has not always been given to factors such as concentration of DE in the diet, amino acid composition of the dietary protein, and digestibility of the dietary protein (Wilson and Halver, 1986; Wilson, 1989). Understanding the nutritional constraints and limitations used in arriving at these reported protein requirements is important for their proper application.

Protein requirements, as a proportion of the diet, decrease as fish approach maturity. For example, 25 percent protein was adequate in the diet of channel catfish of 114 to 500 g, but 35 percent protein produced faster gains than did 25 percent protein in 14- to 100-g fish (Page and Andrews, 1973). Somewhat similar results have been obtained with salmonids, common carp, and tilapia (Wilson and Halver, 1986).

Little convincing evidence exists to show that protein requirement, expressed as a percentage of dry matter, is affected by water temperature. In general, all feeding and growth functions increase in parallel as water temperature rises, although growth rate may increase more rapidly because of an increased feed conversion efficiency coupled

with a higher intake per meal (Brett, 1979). Protein requirement for rainbow trout was unchanged from 35 percent (in diets containing 3,580 kcal DE/kg) at water temperatures ranging from 9° to 18°C (see Figure 5 in National Research Council, 1981).

The high concentrations of dietary protein necessary

for maximal growth rates of fish do not mean that they use more protein as an energy source than is the case with homeothermic vertebrates. Values for net protein retention are in the range of 20 to 50 percent for both types of vertebrate; Bowen (1987) summarized a number of data that showed a median value for fish of 31 percent and for other

vertebrates of 29 percent. Broadly similar proportions of dietary protein are therefore used as an energy source in fish as in warm-blooded terrestrial vertebrates; this, notwithstanding the fact that fish have lower presumed energy requirements than do homeotherms. Attempts have been made to compare absolute protein intake rates (mg protein ingested/g body weight/day); this is a difficult undertaking both because accurate measurement of feed intake by fish is in itself difficult and because the use of data for fish of different physiological ages, held under different conditions of temperature and photoperiod, introduces considerable variation.

An absolute requirement for 10 amino acids (arginine, histidine, isoleucine, leucine, lysine, methionine, phenylalanine, threonine, tryptophan, and valine) has been demonstrated in all fish species examined so far. Quantification of essential amino acid requirements has relied largely on dose-response curves in which the response measured has been weight gain. Various types of chemically defined, purified, and natural ingredient diets have been used to provide graded increments of the amino acid under test. Most studies have used test diets in which the nitrogen component consisted of either amino acids or a mixture of amino acids,

casein, and gelatin formulated to provide an indispensable amino acid composition identical with some reference protein (such as whole hen's egg protein or fish body protein) minus the amino acid under test. For many fish species, growth rates produced by diets with large amounts of free amino acids are inferior to diets of similar amino acid composition in which the nitrogen component is protein (Wilson et al., 1978; Robinson et al., 1981; Walton et al., 1982, 1986). Thus amino acid requirements obtained in this way are based on growth rates below the optimum.

Other approaches to quantifying indispensable amino acid requirements have included using proteins with poor amino acid patterns that differ substantially from that required, such as zein (Dabrowski, 1981) or maize gluten (Halver et al., 1958; Ketola, 1983). Comparatively small amounts of crystalline amino acids are then added to balance

the protein component, leaving it deficient only in one amino acid. Concerns about this approach center on protein digestibility, amino acid availability and rate of transit, and absorption of supplemented free amino acids compared with those from dietary protein. In addition, imbalanced proteins may have high percentages of certain amino acids, such as leucine, and these may depress the assimilation of other amino acids.

Ogino (1980) measured the retention of indispensable amino acids in the whole body protein of carp and rainbow trout and used the increase in indispensable amino acid content measured over periods of 14 to 28 days to estimate requirements. This method assumes that the maintenance requirements of young growing fish are low (although it is not easy to reconcile this view with the fact that only 30 to 40 percent of dietary nitrogen is retained by growing fish), so

that the pattern of amino acids deposited in body weight gain is the main determinant of patterns of amino acids required.

In warm-blooded animals, a constant relationship was shown between indispensable amino acid requirements and protein intake up to the level of protein required for maximum growth (Almquist, 1972). For several indispensable amino acids, intake and weight gain were apparently linearly related and this relationship was presumed to hold for all indispensable amino acids. On this basis amino acid requirements of fish were expressed as a percentage of dietary protein as well as on a dry matter basis (National Research Council, 1981, 1983).

Later studies bear on the finding of Almquist (1972) in

that they show the relationship to be not a linear but an exponential function (Finke et al., 1987). The response of an animal to dietary increments of a limiting nutrient does not break at one particular point. An accurate representation of the so-called ''diminishing returns" area of the response curve is claimed (Finke et al., 1989) to be critical in assessing the efficiency of incremental increases of dietary amino acid concentration as the response approaches the maximum The use of a logistic model supports a more accurate assessment, than that provided by broken-line analysis, of the diminishing returns area of the response curve and of the maximum response (Finke et al., 1989).

The implication of these later studies is that indispensable amino acid requirements are not best expressed as a percentage of dietary protein. Nevertheless, because the dose-response relationship is, for all practical purposes, linear for much of its length (Gahl et al., 1991), amino acid

requirements in Tables 1-3 to 1-13 have again been expressed both as a percentage of dietary protein and on a dry matter basis.

Diets in which the nitrogen component is made up of casein, gelatin, and crystalline amino acids have been referred to in the tables as chemically defined diets. Purified diets are those in which proteins, with an amino acid pattern (g amino acid/16 g nitrogen) that differs substantially from that required, supply the bulk of the nitrogen together with some supplementary amino acids. Natural ingredient diets use normal feed ingredients such as fishmeal, soya meal, blood meal, and wheat middlings.

The values in Tables 1-3 to 1-13 suggest that large differences exist among fish species in their requirements for certain amino acids. Where several estimates are available for one amino acid in a single species, as in the case of rainbow trout (Table 1-8), marked discrepancies occur. Some of these may be due to differences in growth rate, amino acid sources, feed intake, and other aspects of methodology.

For most indispensable amino acids, deficiency is manifest as a reduction in weight gain. In certain species of fish, however, a deficiency of methionine or tryptophan leads to

pathologies, because these amino acids are not only incorporated into proteins but also used for the synthesis of other essential compounds.

Salmonids, including rainbow trout, Atlantic salmon (Salmo salar), and lake trout (Salvelinus namaycush), suffer from cataracts when given a diet deficient in methionine (Poston et al., 1977). The lens begins to become opaque after 2 to 3 months, depending on the extent to which the fish are deficient in sulfur amino acids. As the deficiency increases, lens opacity gradually progresses, causing a large reduction in light transmission. Cataracts also occur as a consequence of tryptophan deficiency in rainbow trout (Poston and Rumsey, 1983; Walton et al., 1984b); the developmental pattern of the cataracts is similar to that occurring in methionine deficiency (Poston and Rumsey, 1983).

Tryptophan deficiency leads to scoliosis (lateral curvature of the vertebral column) and to a derangement of mineral metabolism in certain salmonids, including rainbow trout (Walton et al., 1984b), sockeye salmon (Oncorhynchus nerka) (Halver and Shanks, 1960), and chum salmon (Oncorhynchus keta) (Akiyama et al., 1986). Scoliosis in chum salmon may be reversed by restoring tryptophan to normal concentrations in the diet. The condition may be related to a decline in levels of the brain neurotransmitter serotonin, which is formed from tryptophan. Thus, inclusion of serotonin

in tryptophan-deficient diets greatly reduces the incidence of scoliosis (Akiyama et al., 1986).

Changes in mineral metabolism were observed in tryptophan-deficient rainbow trout (Walton et al., 1984b). Significantly greater concentrations of calcium (Ca) (a fourfold increase over control trout), sodium (Na), and potassium (K) were found in the kidneys of tryptophan-deficient trout. Concentrations of Ca, magnesium (Mg), Na, and K in the livers of tryptophan-deficient trout were also significantly greater than in normal trout. The metabolic lesion(s) responsible for these changes have not been resolved.

Cystine can be formed metabolically from dietary methionine at a rate sufficient to meet the requirements of fish. The reverse sequence of reactions does not occur, however, and fish have an absolute requirement for methionine. Methionine can thus meet the total sulfur amino acid requirement of fish, although some of this requirement may be met by cystine.

Rainbow trout can use D-methionine to replace L-methionine on an equimolar basis (Kim et al., 1992). D-methionine is deaminated by D-amino acid oxidase and subsequently reaminated to L-methionine. This metabolic capacity is probably also characteristic of other fish.

A similar relationship exists between aromatic amino acids. Fish readily convert phenylalanine to tyrosine so that phenylalanine alone can meet requirements for aromatic

amino acids. However, the presence of tyrosine in the diet will reduce some of the requirement for phenylalanine.

Some adverse interactions may occur between amino acids that are structurally related when their concentrations in the diet are imbalanced. Well-known examples in homeotherms are antagonisms arising from dietary imbalances of lysine-arginine and of leucine-valine. No convincing evidence exists, however, for lysine-arginine antagonism in fish. Robinson et al. (1981) could not demonstrate any effects when diets with excess lysine in the presence of adequate or marginal arginine were fed to channel catfish; diets containing excess arginine in the presence of adequate or marginal lysine similarly failed to show any antagonistic effect. Nor did excess lysine affect the growth rates of rainbow trout fed low concentrations of arginine (Kim et al., 1983).

Antagonism between branched-chain amino acids generally arises in mammals from an excess of leucine over isoleucine and valine; the first two steps of the catabolic breakdown of all three branched-chain amino acids are catalyzed by the same enzymes. Data on antagonisms among branched-chain amino acids in fish are not clear-cut and are inconsistent between species. Thus the isoleucine requirement of chinook salmon (Oncorhynchus tshawytscha) increased slightly with increasing concentrations of dietary leucine (Chance et al., 1964). Hughes et al. (1983) observed changes in concentrations of branched-chain amino acids in lake trout given diets containing increasing amounts of valine. Plasma isoleucine and leucine were both elevated in valine-deficient fish, and their concentrations decreased as

dietary valine was increased. No changes in plasma valine concentration occurred until valine in the diet reached the required concentration, after which the plasma valine increased about 2.5-fold. In contrast, rainbow trout showed a high tolerance for dietary leucine; no growth depression occurred with concentrations as high as 9.2 percent. Even with excessive dietary leucine concentrations (13.4 percent), which were overtly toxic, the concentrations of free valine and isoleucine in plasma, liver, and muscle were not depressed (Choo, 1990).

Another interaction characteristic of some homeotherms, and referred to as an imbalance, occurs when diets are supplemented with the second most limiting amino acid, or with all indispensable amino acids other than the first limiting amino acid. This leads to a fall in the concentration of the first limiting amino acid in the blood and eventually to reduced feed intake even though retention of the first limiting amino acid is not affected. No data are available on such interactions in fish. Nevertheless, oversupplementation with the second most limiting amino acid should be avoided as it may exacerbate a primary deficiency. There is considerable information available on amino acid interrelationships in mammals. Further information on these relationships can be found in Czarnecki et al. (1985), Baker (1987), and May et al. (1991).

Dietary lipids are important sources of energy and of essential fatty acids (EFA) that are needed for normal growth and development. They also assist in the absorption of fat-soluble vitamins. Dietary lipids, mainly in the form of triacylglycerols, are hydrolyzed by digestive enzymes to a mixture of free fatty acids and 2-monoglycerides. These compounds are then absorbed and either used for the synthesis of various cellular components or catabolized for energy.

Dietary lipids contain both saturated and unsaturated fatty acids. Fatty acids may be designated by numbering either from the methyl or carboxyl terminal. The notation from the methyl terminal is most convenient for many nutritional purposes and is used here. It involves three numbers given in sequence, the first denoting the number of carbon atoms; the second, following a colon, the number of double bonds; and the third, designated as (n-) indicates the number of carbon atoms between the methyl terminal and the first double bond. The term polyunsaturated fatty acid (PUFA) normally refers to fatty acids with 18 or more carbon atoms and two or more double bonds.

In common with other vertebrates, fish cannot synthesize either 18:2(n-6) or 18:3(n-3) de novo. Hence one or both of these fatty acids must be supplied preformed in the diet, depending on the EFA requirements. In addition, fish vary considerably in their ability to convert 18-carbon unsaturated fatty acids to longer-chain, more highly unsaturated fatty acids of the same series (Owen et al., 1975). The EFA requirement of the fish is thus related, to some extent, to their ability to modify these fatty acids metabolically.

The quantitative EFA requirements of several fish species are summarized in Table 1-14. A major difference appears to exist between freshwater and stenohaline marine fish (those unable to withstand a wide variation in water salinity). In general, freshwater fish require either dietary linoleic acid, 18:2(n-6), or linolenic acid, 18:3(n-3), or both, whereas stenohaline marine fish require dietary eicosapentaenoic acid (EPA), 20:5(n-3), and/or docosahexaenoic acid (DHA), 22:6(n-3).

Among the freshwater species, the ayu, channel catfish, coho salmon, and rainbow trout require 18:3(n-3) or EPA and/or DHA. Chum salmon, common carp, and Japanese eel require an equal mixture of 18:2(n-6) and 18:3(n-3); whereas, Nile tilapia and Zillii's tilapia require only 18:2(n-6) for maximum growth and feed efficiency. Striped bass, however, require n-3 PUFA and cannot chain elongate 18:3(n-3) (Webster, 1989; Webster and Lovell, 1990).

The principal gross signs of EFA deficiency reported for various fishes are dermal signs (fin rot), a shock syndrome, myocarditis, reduced growth rate, reduced feed efficiency, and increased mortality (Castell et al., 1972; Takeuchi and Watanabe, 1977a,b; Takeuchi et al., 1980; Satoh et al., 1989). Essential fatty acid deficiency has also been shown to reduce the reproductive performance of common carp (Shimma et al., 1977), rainbow trout (Watanabe, 1982; Watanabe et al., 1984c; Leray et al., 1985) and red sea bream (Watanabe et al., 1984a,b).

In fish species that can further desaturate and chain elongate 18:2(n-6) or 18:3(n-3), an absence of either of these fatty acids in the diet leads to the desaturation and chain elongation of oleic acid, 18:1(n-9), to 20:3(n-9), which is characteristic of an EFA deficiency in many terrestrial animals. Thus when EFAs are deficient, increased concentrations of 20:3(n-9) are incorporated into tissue polar lipids in place of 20:4(n-6), 20:5(n-3), or 22:6(n-3). Castell et al. (1972) suggested that the ratio of 20:3(n-9)/20:5(n-3) in polar lipids from the liver of rainbow trout might be a useful index of EFA status. By analogy with mammals, the diet is considered satisfactory with respect to EFA if this ratio is not greater than 0.4.

Watanabe et al. (1983) have reported that n-3 PUFAs such as EPA and DHA, are required for normal growth and development of ayu and red sea bream larvae. High mortalities and abnormalities, such as underdeveloped swim bladder and scoliosis, have been observed in red sea bream larvae reared on rotifers and Artemia spp., either devoid of n-3 PUFAs or containing only low concentrations of n-3 PUFAs

(Watanabe et al., 1980; Kitajima et al., 1980a,b; Oka et al., 1980, 1982; Watanabe et al., 1982; Fukusho et al., 1984, 1985). Subsequently, EPA and/or DHA have been shown to be essential for various marine fish larvae, such as ayu (Kanazawa et al., 1981), red sea bream (Izquierdo et al., 1989; Watanabe et al., 1989b), striped jack (Watanabe et al., 1989c), and gilthead sea bream (Koven et al., 1989), as well as for the larvae of one freshwater fish, striped bass (Webster and Lovell, 1990).

The essential fatty acids function as components of phospholipids in all biomembranes and as precursors for eicosanoids that fulfill a variety of metabolic functions. Biomembranes must be in a fluid state to function properly at various temperatures. Membrane fluidity depends on the proper balance of saturated and unsaturated fatty acids as components of membrane phospholipids (Bell et al., 1986). The role of dietary n-3 PUFA during homeoviscous regulation, whereby fish alter their biomembrane phospholipid composition in response to changes in environmental temperature, has been demonstrated by Hazel (1984). During acclimation to cold-water temperatures the total amount of phospholipid in the fish biomembrane does not change, however changes occur in the relative proportion of individual phospholipids, in the fatty acid composition of the phospholipids, and in the distribution of fatty acids within the phospholipid molecules. For example, in rainbow trout transferred from 20° to 5°C, the proportion of phosphatidylethanolamine increases with a corresponding decrease in phosphatidylcholine in both liver and gills (Hazel, 1979, 1985). Similar changes have been observed during cold adaptation in membranes of goldfish intestine (Miller et al., 1976) and carp muscle (Wodtke, 1981). Conversely, during adaptation to higher temperatures, the proportion of phosphatidylethanolamine decreases in trout gill membranes and phosphatidylcholine increases (Hazel and Carpenter, 1985). Thus changes in the ratio of the two major membrane phospholipids, phosphatidylcholine and phosphatidylethanolamine, can be used in fish as an index of proper adaptation to changes in environmental temperatures.

During cold adaptation, the relative amount of n-3 PUFA increases in trout liver membrane phospholipids, whereas the amount of saturated fatty acids decreases and the amount

of monounsaturated fatty acids remains relatively constant (Hazel, 1979; Sellner and Hazel, 1982). A large increase in 22:6(n-3) occurs in phosphatidylcholine, whereas 20:5(n-3) instead of 22:6(n-3) increases in phosphatidylethanolamine in response to lower temperature in trout membranes (Hazel, 1979).

Lipids serve as an important source of dietary energy for all fish, but perhaps to a greater extent for cold-water and marine fish, which have a limited ability to use dietary carbohydrates for energy. Takeuchi et al. (1978) found that the protein content of rainbow trout diets could be reduced from 48 to 35 percent without any reduction in weight gain if the lipid concentration was increased from 15 to 20 percent. Takeda et al. (1975) were able to decrease the protein concentration in yellowtail diets from 70 to 55 percent without any reduction in growth rate by increasing the lipid content. These observations support the recommendation that all diets should be formulated not only to meet the optimum ratio of energy to protein for that species, but also to contain an adequate amount of lipid.

No definite percentage of dietary lipids can be given for fish diets without considering the type of lipid as well as the protein and energy content of the diet. Lipid concentrations of up to 20 percent give optimum results with some species (Yone et al., 1971; Stickney and Andrews, 1972; Lee and Putman, 1973; Adron et al., 1976; Takeuchi et al., 1978). However, too much dietary lipid may result in an imbalance of the DE/CP ratio and in excessive fat deposition in the visceral cavity and tissues, which would adversely affect yield, product quality, and storage. The fatty acid composition of the dietary lipid has a significant influence on the tissue fatty acid composition of the fish (Watanabe, 1982; Henderson and Tocher, 1987; Sargent et al., 1989).

The use of lipids in fish diets also requires the use of appropriate antioxidants. The proper use of antioxidants and the deleterious effects of lipid autoxidation in fish diets are discussed in Chapter 2.

The nutritional value of carbohydrates varies among fish. Warm-water fish can use much greater amounts of dietary carbohydrate than cold-water and marine fish. No dietary requirement for carbohydrates has been demonstrated in fish; however, if carbohydrates are not provided in the diet, other compounds, such as protein and lipids, are catabolized for energy and for the synthesis of various biologically important compounds usually derived from carbohydrates. Thus, it is important to provide the appropriate concentration of carbohydrate in the diet of the fish species being cultured.

Enzymes for carbohydrate digestion are apparently present in fish. The enzymes for the major carbohydrate metabolic pathways, such as glycolysis, tricarboxylic acid cycle, pentose phosphate shunt, gluconeogenesis, and glycogen synthesis, have been demonstrated (Shimeno, 1974). Only a limited number of studies have reported on the kinetic properties of the various enzymes, and the results are consistent with evolutionary development. Even though the various enzymes and pathways for glucose metabolism have been detected, the role of dietary carbohydrates and the contribution of glucose to the total energy requirement of fish remain unclear. Studies have indicated that the hormonal and metabolic regulation of carbohydrate and energy metabolism varies among fish and may be somewhat different than in mammals (Shimeno, 1974; Cowey and Walton, 1989).

The relative use of dietary carbohydrates by fish varies and appears to be associated with the complexity of the carbohydrate. Glucose, maltose, and sucrose resulted in the best growth rates, followed in descending order by dextrin and fructose, galactose and potato starch, and glucosamine when various carbohydrate sources were fed to young chinook salmon at a concentration of 10 percent of the diet (Buhler and Halver, 1961). Rainbow trout have been reported to use 30 percent glucose in a 45 percent protein diet, whereas a concentration of 30 percent glucose in a 30 percent protein diet had a negative effect on growth and feed efficiency (Bergot, 1979a,b). A similar trend has been reported for sucrose use by rainbow trout when 35 and 55 percent protein diets were compared (Luquet, 1971). Fifty-seven and 64 percent of the gross energy of glucose and sucrose, respectively, were used by rainbow trout when these carbohydrates were included at a concentration of 30 percent in a 48 percent protein diet (Pieper and Pfeffer, 1979). In another study with rainbow trout, the replacement of dietary lipid with glucose at concentrations from 2.5 to 18.3 percent in natural ingredient diets containing 40 percent CP and similar ME content resulted in a significant linear decrease in weight gain as glucose increased (Hilton and Atkinson, 1982).

Feeding high concentrations of digestible carbohydrates has been reported to result in an increase in liver size and glycogen content in salmonids (Phillips et al., 1948; Buhler and Halver, 1961; Lee and Putnam, 1973; Bergot, 1979a,b; Pieper and Pfeffer, 1979). Similar effects have been reported in red sea bream (Furuichi and Yone, 1971a), plaice (Cowey et al., 1975), and yellowtail (Shimeno et al., 1979).

The relative utilization of dietary glucose, dextrin, and gelatinized starch has been compared in carp and red sea bream. Growth and feed efficiency of carp were highest when fed the gelatinized starch diet, followed by the dextrin

and glucose diets in decreasing order, whereas red sea bream did not show any significant difference in growth rates for the various carbohydrate sources (Furuichi and Yone, 1982a). Channel catfish used dextrin or starch for growth but not mono- and disaccharides (Wilson and Poe, 1987). Chum salmon fry used glucose, maltose, sucrose, dextrin, and gelatinized starch but not fructose, galactose, or lactose for growth (Akiyama et al., 1982). White sturgeon, however, used glucose and maltose better than dextrin or starch (Hung et al., 1989).

The ability of fish to use dietary carbohydrates differs among species. Studies have indicated that common carp (Shimeno et al., 1977; Takeuchi et al., 1979; Furuichi and Yone, 1980; Shimeno et al., 1981), channel catfish (Garling and Wilson, 1976, 1977), red sea bream (Furuichi and Yone, 1971a, 1980), and tilapia (Anderson et al., 1984; El-Sayed and Garling, 1988) use higher levels than yellowtail (Furuichi and Yone, 1981) and salmonids (Buhler and Halver, 1961; Edwards et al., 1977; Atkinson and Hilton, 1981). In general, a concentration of less than 25 percent dextrin or gelatinized starch appears to be used as an energy source by rainbow trout (Lee and Putnam, 1973), plaice (Cowey et al., 1975), and yellowtail (Takeda et al., 1975), whereas channel catfish (Page and Andrews, 1973; Garling and Wilson, 1977) and common carp (Takeuchi et al., 1979) can use a higher percentage.

Oral glucose tolerance tests have been conducted with brook trout (Phillips et al., 1948), rainbow trout (Palmer and Ryman, 1972), common carp (Furuichi and Yone, 1981), channel catfish (Wilson and Poe, 1987), red sea bream (Furuichi and Yone, 1971b, 1981), and yellowtail (Furuichi and Yone, 1981). In each case, the oral administration of glucose resulted in a persistent hyperglycemia. A similar outcome was observed when rainbow trout were fed diets containing 15 and 30 percent glucose (Bergot, 1979c). Furuichi and Yone (1981) determined the change in plasma insulin levels during glucose tolerance tests in common carp, red sea bream, and yellowtail. Insulin was measured by a radioimmunoassay procedure using antiskipjack insulin serum (Furuichi et al., 1980). The plasma insulin level reached a maximum about 2 hours after oral glucose administration in each species, paralleling the level of plasma glucose. The researchers point out that the plasma insulin pattern, with respect to both the time to reach maximum level and the maximum activity, was very similar to that observed for a diabetic human.

The prolonged hyperglycemia observed in fish following glucose tolerance tests and the relative inability of fish to utilize high concentrations of dietary carbohydrates has been assumed to be the result of low levels of endogenous insulin (Palmer and Ryman, 1972: Furuichi and Yone, 1982b; Wilson and Poe, 1987). However, the development of radioimmunoassay methods for the determination of insulin levels in fish have shown that these levels are similar to or often higher than those observed in mammals (Plisetskaya, 1990; Mommsen and Plisetskaya, 1991). The relative intolerance of fish to large doses of exogenous glucose despite the high levels of circulating insulin has been suggested to resemble conditions known as non-insulin-dependent diabetes mellitus rather than insulin-dependent diabetes (Hilton et al., 1987; Hertz et al., 1989). Rainbow trout muscle tissue has been shown to have from 3 to 10 percent of the insulin receptors per microgram of protein compared with those in either the white or red skeletal muscle of rats, with the overall insulin-receptor binding capacity in trout being lower than that reported for mammals (Gutierrez et al., 1991). However, these workers could not demonstrate a difference in insulin-receptor binding in skeletal muscle of trout fed a high-carbohydrate diet as compared to those fed a low-carbohydrate diet. Thus, they concluded that the high glycemic levels observed in trout fed the high-carbohydrate diet were not due to impaired binding of insulin to its receptors in skeletal muscle. It is apparent from this latest information that the hyperglycemia in fish fed high concentrations of carbohydrates is not solely due to impaired insulin release or receptor binding as previously thought.

Although no specific carbohydrate requirement has been established for fish, some form of digestible carbohydrate should be included in the diet. For example, the growth rate of channel catfish fingerlings was greater when their diets contained some carbohydrates rather than only lipids for all the nonprotein energy (Garling and Wilson, 1977). Carbohydrates may serve as precursors for the dispensable amino acids and nucleic acids, which are metabolic intermediates necessary for growth. Because carbohydrate is the least expensive source of dietary energy, the maximum tolerable concentration should be used with regard to the fish species. Cereal grains serve as inexpensive sources of carbohydrates for warm-water fish, but their use in cold-water fish feeds is limited. Starch is also important for the binding properties of extruded and pelleted feeds.

Fish, unlike most terrestrial animals, can absorb some minerals (inorganic elements) not only from their diets but also from their external aquatic environment. Calcium (Ca), magnesium (Mg), sodium (Na), potassium (K), iron (Fe), zinc (Zn), copper (Cu), and selenium (Se) are generally derived from the water to satisfy part of the nutritional requirements of fish. Phosphates and sulfates, however, are more effectively obtained from feed sources (as reviewed by Lall, 1989). Inorganic elements are required for the normal life processes of fish. Their main functions include the

formation of skeletal structure, electron transfer, regulation of acid-base equilibrium, and osmoregulation. Minerals are also important components of hormones and enzymes, and they activate enzymes. Complex biochemical mechanisms control and regulate the uptake, storage, and excretion of various inorganic elements, allowing fish to live in a dynamic equilibrium with their aquatic medium. The electrolytes Na⁺, K⁺, Mg²⁺, Ca²⁺, Cl^-, and HCO₃^- play a major role in the osmotic and ionic regulation of extra- and intracellular fluids in fish.

The exchange of ions from the surrounding water across the gills and skin of fish complicates the measurement of mineral requirements. Although most essential elements known for terrestrial animals are also considered important for fish, quantitative requirements have been reported for only nine minerals (calcium, phosphorus, magnesium, iron, copper, manganese, zinc, selenium, and iodine) for selected fish species. The role of macrominerals (calcium, phosphorus, magnesium, sodium, potassium, and chlorine) and trace elements (iron, copper, zinc, manganese, selenium, iodine, fluorine, and chromium) in fish nutrition are discussed in this section and sources of the minerals are given in Table 8-3.

Calcium and phosphorus are directly involved in the development and maintenance of the skeletal system and participate in several physiological processes. Fish scales are also an important site of calcium metabolism and deposition. In addition to its structural functions, calcium plays an important role in muscle contraction, blood clot formation, nerve impulse transmission, the maintenance of cell integrity and acid-base equilibrium, and activation of several important enzymes.

Fish absorb calcium directly from their environment (Phillips et al., 1959) and rely entirely on calcium present in water during dietary calcium deprivation (Ogino and Takeda, 1976,1978; Ichii and Mugiya, 1983). The uptake of calcium occurs through gills, fins and oral epithelia, however gills are considered the most important site for calcium regulation. In marine fish, which drink copiously, the gut is not a major site of calcium absorption (Simkiss, 1974). Limited calcium homeostasis occurs in bone, which is acellular (bone without enclosed osteocytes) in most fish (Moss, 1963). There is little exchange of bone calcium with body fluids in marine fish (Simmons et al., 1970); but in a low-calcium environment, like fresh water, where fish must extract calcium against a steep gradient, mobilization of calcium stores from bones and scales may be necessary under certain conditions (Ichii and Mugiya, 1983). Decalcification of scales and bones occurs during ovarian maturation, starvation, and spawning migration (Crichton, 1935; Ichikawa, 1953; Yamada, 1956; Garrod and Newell, 1958; Mugiya and Watabe, 1977). The calcium exchange rate of fish scales is three times that of bones (Berg, 1968). Although fish bone is responsive to calcium-regulating hormones, sufficient evidence does not exist to suggest that bone has a major role in blood calcium regulation.

The calcium requirement of fish is met in large part by absorption through gills and skin in fresh water and by drinking sea water. The calcium requirement is affected by the water chemistry and species differences. The concentration of dietary calcium rarely seems critical for salmonids, and a dietary requirement has not been demonstrated. A low concentration of calcium (0.34 percent or less) is required in the diet of carp and eel (Ogino and Takeda, 1976) for optimum growth. Catfish and tilapia reared in calcium-free water require 0.45 percent and 0.7 percent calcium in the diet, respectively, for optimum growth (Robinson et al., 1986, 1987). Atlantic salmon derive calcium from sea water, thus making dietary supplementation unnecessary (Lall and Bishop, 1977). The uptake of calcium from sea water is not sufficient to meet the calcium requirement of red sea bream (Sakamoto and Yone, 1973, 1976b), which requires 0.34 percent calcium in the diet.

Calcium deficiency has not been detected in carp and catfish in fresh water (Andrews et al., 1973; Ogino and Takeda, 1976; Lovell, 1978) or in Atlantic salmon in sea water (Lall and Bishop, 1977). Generally, calcium from the feed ingredients of natural ingredient diets supplies sufficient calcium to meet the requirements of most finfish.

Phosphorus is an important constituent of nucleic acids and cell membranes, and is directly involved in all energy-producing cellular reactions. The role of phosphorus in carbohydrate, lipid, and amino acid metabolism, as well as in various metabolic processes involving buffers in body fluids, is also well established. Feed is the main source of phosphate for fish because the concentration of phosphate is low in natural waters. The dietary supply of phosphate is more critical than that of calcium because fish must effectively absorb, store, mobilize, and conserve phosphate in both freshwater and sea water environments.

In most fish, the main phosphorus deficiency signs include poor growth, feed efficiency, and bone mineralization. Other signs of deficiency in carp include increase in the activity of certain gluconeogenic enzymes in liver, increase in carcass fat with decrease in carcass water content, reduced blood phosphate levels, deformed head, and deformed vertebrae (Ogino and Takeda, 1976; Onishi et al., 1981; Takeuchi and Nakazoe, 1981). A reduction in hematocrit level of catfish may also occur (Andrews et al., 1973). A low-phosphorus intake by red sea bream also causes curved, enlarged vertebrae; increased serum alkaline phosphatase activity; higher lipid deposition in muscle, liver, and vertebrae; and reduction in liver glycogen content (Sakamoto and Yone, 1980).

Dietary phosphorus requirements ranging from 0.5 to 0.8 percent have been reported for rainbow trout (Ogino and

Takeda, 1978), Atlantic salmon (Ketola, 1975; Lall and Bishop, 1977), chum salmon (Watanabe et al., 1980), carp (Ogino and Takeda, 1976), and red sea bream (Sakamoto and Yone, 1978a). The phosphorus requirement of Atlantic salmon in either fresh water (Ketola, 1975) or sea water (Lall and Bishop, 1977) is 0.6 percent of the diet. Andrews et al. (1973) reported the phosphorus requirement of catfish is 0.8 percent of available phosphorus in a natural ingredient diet. Lovell (1978) and Wilson et al. (1982) reevaluated the phosphorus requirement using chemically defined diets and estimated it to be approximately 0.42 percent available phosphorus. The phosphorus requirement of Japanese eel is relatively low (0.29 percent) compared with that of other finfish (Nose and Arai, 1979). Several workers have reported that the concentration of dietary calcium has no effect on phosphorus requirement of catfish, carp, and rainbow trout (Andrews et al., 1973; Nose and Arai, 1976; Lall and Bishop, 1977; Lovell, 1978; Ogino and Takeda, 1978). However, optimum Ca:P ratio is important in the diet of red sea bream, 1:2 (Sakamoto and Yone, 1973), and eel, 1:1 (Nose and Arai, 1979).

Wide differences in the availability of phosphorus from different sources have been reported. In general, the more soluble the salt, the higher the availability of phosphorus for fish. Thus, the phosphorus of monocalcium or dicalcium phosphates is more readily available than that from tricalcium phosphate (Ogino et al., 1979; Sakamoto and Yone, 1979a). The availability of phosphorus in fishmeal for tilapia is low compared to that for rainbow trout and chum salmon (Watanabe et al., 1980). Also, the availability of phosphorus in fishmeal is significantly lower for carp than for rainbow trout (Ogino et al., 1979). The differences in the availability of phosphorus to salmonids and to carp and tilapia is probably due to the limited secretion of gastric juices by these warm-water species (Ogino et al., 1979; Yone and Toshima, 1979). About 60 percent of the phosphorus in anchovy and menhaden fishmeals was available to channel catfish (Lovell, 1978).

Feedstuffs that originate from seeds contain phosphorus primarily as the calcium-magnesium salt of phytic acid known as phytin. Phytin phosphorus is unavailable to animals with simple stomachs because they lack the enzyme phytase in the gastrointestinal tract. Phytic acid also forms insoluble salts with free calcium in the digestive tract. Hence, the availability of phosphorus in most plant products is low; for example, that of soybean meal is between 29 and 54 percent (Lovell, 1978; Wilson et al., 1982).

Magnesium is an essential cofactor in many enzymatic reactions in intermediary metabolism. These enzymes include phosphokinases, thiokinases, phosphatases, pyrophosphatases, and amino acyl synthetases. Magnesium plays an important role in the respiratory adaptation of freshwater fish (Houston, 1985). It is also required in skeletal tissue metabolism, osmoregulation, and neuromuscular transmission. The quantitative dietary magnesium requirements of rainbow trout (Ogino et al., 1978; Knox et al., 1981a; Shearer, 1989), carp (Ogino and Chiou, 1976), channel catfish (Gatlin et al., 1982), eel (Nose and Arai, 1979), and guppy (Shim and Ng, 1988) have been estimated to range from 0.04 to 0.06 percent of the diet. A dietary magnesium content of 0.06 to 0.08 percent was required for tilapia (Dabrowska et al., 1989).

The magnesium requirement of fish can be met either from the diet or water. Shearer and Asgard (1992) found that in fresh water, a waterborne concentration of 46 mg per liter was sufficient to meet the magnesium requirement of rainbow trout fed a magnesium-free diet. In the marine environment, magnesium supplementation of diet may not be necessary (Lall and Bishop, 1977; Sakamoto and Yone, 1979b). The magnesium requirement of rainbow trout was not influenced by an increase in dietary calcium or phosphorus (Knox et al., 1981a). Inorganic magnesium salts are efficiently used by rainbow trout with an apparent magnesium retention of 76 percent from these compounds, but magnesium retention from the bone fraction of fishmeal is only 54 percent (Shearer and Asgard, 1990).

Magnesium deficiency causes anorexia, reduced growth, lethargy, and reduced tissue magnesium content in fish. In rainbow trout, magnesium deficiency also caused calcinosis of the kidney, vertebrae deformity, and degeneration of muscle fibers and epithelial cells of the pyloric cecum and gill filaments (Cowey et al., 1977; Ogino et al., 1978). Catfish and rainbow trout fed magnesium-deficient diets show flaccid appearance of their muscle (Knox et al., 1981a; Gatlin et al., 1982). Carp maintained on a low-magnesium diet also develop convulsions and cataracts (Ogino and Chiou, 1976). Magnesium deficiency has not been demonstrated in fish in a sea water environment, where they obtain magnesium by drinking the water. An interaction between dietary protein and magnesium concentrations has been demonstrated in tilapia (Dabrowska et al., 1989), where excess magnesium (0.32 percent) in a low-protein (24 percent) diet produced toxicity signs and where magnesium deficiency in a high-protein (44 percent) diet caused whole-body hypercalcinosis.

Sodium, potassium, and chlorine are the most abundant electrolytes in the body. Sodium and chlorine are the principal cation and anion, respectively, in the extracellular fluid of the body; whereas, potassium is the major monovalent intracellular cation. Chloride ion is the major anion of gastric juice and blood. The deficiency signs of these elements are difficult to produce because fish readily absorb these elements from the surrounding aquatic medium.

The abundance of these elements in common feedstuffs used in fish diets means they need not be supplemented in most natural ingredient diets. However, potassium supplementation was found necessary in chemically defined diets for chinook salmon. Juvenile chinook salmon reared in fresh water required 0.8 percent potassium in their diet for maximum growth, and the whole-body potassium saturation was reached at a potassium concentration between 0.6 and 1.2 percent of the diet (Shearer, 1988). Red sea bream reared in sea water, where potassium concentration is much higher, did not require a dietary potassium supplement (Sakamoto and Yone, 1978b). The signs of potassium deficiency in chinook salmon included anorexia, convulsions, tetany, and death (Shearer, 1988).

Supplements of 1 to 4 percent of sodium chloride in natural ingredient diets had no beneficial effect on growth of rainbow trout (Salman and Eddy, 1988), coho salmon (Zaugg and McLain, 1969), Atlantic salmon (Basulto, 1976), channel catfish (Murray and Andrews, 1979), or red sea bream (Sakamoto and Yone, 1978b). However, higher supplements of salt adversely affected growth and feed efficiency of coho salmon and rainbow trout (Zaugg and McLain, 1969; Salman and Eddy, 1988). Atlantic salmon and coho salmon fed salt-enriched diets adapted well to seawater, with few mortalities (Zaugg and McLain, 1969; Basulto, 1976). The Na⁺- and K⁺-stimulated ATPase activity of gill microsomes is elevated by dietary salt supplementation, thus making saltwater adaptation easier physiologically (Zaugg and McLain, 1969).

Iron is an essential element in the cellular respiratory process through its oxidation-reduction activity and electron transfer. It is found in the body mainly in the complex form bound to proteins, such as heme compounds (hemoglobin and myoglobin), heme enzymes (cytochromes, catalase, peroxidase, and so on), and nonheme compounds (transferrin, ferritin, and iron-containing flavoproteins [ferredoxins, dehydrogenases]). Feed is considered the major source of iron for fish because natural waters usually contain low amounts of soluble iron. Fish can absorb soluble iron from the water through the gills because the addition of ferrous sulfate to water enhanced growth and hemoglobin level in sword tail and platyfish (Roeder and Roeder, 1966). Iron was absorbed from the peritoneal cavity in rainbow trout and stored in the liver, spleen, and anterior kidney (Walker and Fromm, 1976).

The iron requirements have been reported for catfish (Gatlin and Wilson, 1986a), Atlantic salmon (Lall and Hines, 1987), and eel (Nose and Arai, 1979), and they are 30, 60, and 170 mg/kg of diet, respectively. Iron deficiency causes characteristic microcytic anemia in brook trout (Kawatsu, 1972), red sea bream (Sakamoto and Yone, 1976a, 1978c), yellowtail (Ikeda et al., 1973), eel (Nose and Arai, 1979), and carp (Sakamoto and Yone, 1978d). In most cases, growth was not influenced by iron deficiency. The normal liver color changed to yellowish-white during iron deficiency in carp (Sakamoto and Yone, 1978d). In catfish, iron deficiency suppressed hematocrit, hemoglobin, and plasma iron concentrations and transferrin saturation (Gatlin and Wilson, 1986a). Sakamoto and Yone (1979c) found that ferrous chloride and ferrous sulfate were equally effective in preventing anemia in red sea bream; however, a somewhat higher concentration of ferric citrate was required. Dietary iron toxicity signs develop in rainbow trout fed more than 1,380 mg Fe/kg (Desjardins et al., 1987). The major effects of iron toxicity include reduced growth, increased mortality, diarrhea, and histopathological damage to liver cells.

Copper is a constituent of many enzymes and is essential for their activities. It is associated with cytochrome c oxidase of the electron transport chain in the cell. Other cuproenzymes found in fish tissues include superoxide dismutase, tyrosinase, lysyl oxidase, ceruloplasmin, and dopamine β-hydroxylase. Although Syed and Coombs (1982) found that the distribution of copper and copper-dependent enzymes is similar in plaice and mammals, copper metabolism of most fish is poorly defined. High concentrations of copper are found in the heart, liver, brain, and eyes. Copper is present as the copper-protein complex, ceruloplasmin, in plasma.

The dietary copper requirements of selected fish species have been reported: rainbow trout and common carp require 3 mg/kg (Ogino and Yang, 1980); channel catfish require 5 mg/kg (Gatlin and Wilson, 1986b); and Atlantic salmon require 5 mg/kg (Lall and Hines, 1987). Gatlin and Wilson (1986b) observed reduced heart cytochrome c oxidase and liver copper-zinc superoxide dismutase activities in copper-deficient catfish. Carp fed diets containing high-ash fishmeal without copper supplement showed reduced growth and cataract formation (Satoh et al., 1983a). A low concentration of copper was found in Atlantic salmon suffering from Hitra disease (Poppe et al., 1986), which is a cold-water bacterial disease caused by Vibro salmonicida.

Fish appear to be more tolerant of copper in the diet than of dissolved copper in the water. Concentrations of 0.8 to 1.0 mg copper per liter as copper sulfate in water are toxic to many species of fish (Friedman and Shibko, 1972). Ashley (1972), however, found that coho salmon tolerated copper at 1,000 mg/kg in the diet with only retarded growth and impaired pigmentation. Knox et al. (1982) found no deleterious effects of feeding rainbow trout diets containing 150 mg copper/kg for 20 weeks.

Copper toxicity has been experimentally produced in rainbow trout fed 730 mg Cu/kg of diet for 24 weeks (Lanno et al., 1985). The toxicity signs include reduced growth and

feed efficiency and elevated liver copper levels. However, a diet with up to 665 mg Cu/kg did not produce toxicity signs (Knox et al., 1982; Lanno et al., 1985). Wild and cultured salmonids accumulate exceptionally high copper levels in the liver without being exposed to elevated waterborne or dietary copper (Poppe, 1986).

The essential function of zine for living organisms is based on its role as an integral part of a number (more than 70) of metalloenzymes, including dehydrogenases, aldolases, peptidases, and phosphatases. Fish accumulate zinc from both water and dietary sources; however, dietary zinc is more efficiently absorbed than waterborne zinc. The gills in rainbow trout play a major role in excretion of zinc (Hardy et al., 1987).

The zinc requirement of young rainbow trout and carp is 15 to 30 mg/kg of diet (Ogino and Yang, 1978, 1979), whereas channel catfish and blue tilapia require 20 mg/kg of diet (Gatlin and Wilson, 1983; McClain and Gatlin, 1988). Dietary calcium and phosphorus concentrations, presence of phytic acid, protein source, and form of zinc and calcium affect zinc absorption and use in fish (Takeda and Shimma, 1977; Gatlin and Wilson, 1984a; Hardy and Shearer, 1985; Richardson et al., 1985; Wekell et al., 1986; Satoh et al., 1987; McClain and Gatlin, 1988; Satoh et al., 1989). Phytate forms a complex with transitional elements, such as zinc, iron, and manganese, in the gastrointestinal tract and prevents their absorption. Calcium promotes the complexing of zinc to phytates. The bioavailability of zinc in fishmeal is inversely related to the tricalcium phosphate content (Satoh et al., 1987). This is presumably caused by absorption of zinc into insoluble calcium phosphate complexes in the intestine that are passed through the gut unabsorbed and excreted. Higher supplements of zinc should be included in natural ingredient fish diets to compensate for reduced zinc bioavailability caused by dietary phytate and tricalcium phosphate.

Rainbow trout and common carp tolerated 1,700 to 1,900 mg Zn/kg of diet without adverse effect on growth or survival (Jeng and Sun, 1981; Wekell et al., 1983). However, Knox et al. (1982, 1984) fed rainbow trout elevated concentrations of zinc to 1,000 mg/kg diet and observed reduced hemoglobin, hematocrit, and hepatic copper concentrations. Copper status of channel catfish was not impaired by diets containing 200 mg Zn/kg (Gatlin et al., 1989).

In rainbow trout, zinc deficiency caused growth suppression, mortality, lens cataracts, erosion of fins and skin, and short-body dwarfism (Ogino and Yang, 1979; Satoh et al., 1983b). High-ash (white) fishmeal may affect zinc absorption and use, resulting in lens cataracts (Ketola, 1979). When zinc supplements (40 mg/kg) were added to rainbow trout diets containing white fishmeal, dwarfism and cataract problems were alleviated (Satoh et al., 1987).

Caudal fin zinc concentration is a good indicator of zinc status in rainbow trout (Wekell et al., 1986). In channel catfish, diets low in zinc reduced growth rate, appetite, and bone zinc, calcium, and serum zinc concentrations (Gatlin and Wilson, 1983). Broodstock diets low in zinc reduced egg production and hatchability (Takeuchi et al., 1981).

Manganese functions either as a cofactor that activates metal-enzyme complexes or as an integral part of certain metalloenzymes in carbohydrate, lipid, and protein metabolism. Many kinases, transferases, hydrolases, and decarboxylases can be activated by either manganese or other divalent cations, such as magnesium, and their activity is not specific for manganese. However, enzymes such as glycosyl transferase are highly specific for manganese activation. Two important manganese metalloenzymes are pyruvate carboxylase and superoxide dismutase.

Although the uptake of manganese from water has been demonstrated (Miller et al., 1980; Srivastava and Agarwal, 1983), it is more efficiently absorbed from feed. The manganese requirements have been demonstrated for channel catfish, 2.4 mg/kg (Gatlin and Wilson, 1984b), and common carp and rainbow trout, 13 mg/kg (Ogino and Yang, 1980).

Manganese deficiency caused reduced growth and skeletal abnormalities in rainbow trout, carp, and tilapia (Ishak and Dollar, 1968; Ogino and Yang, 1980; Yamamoto et al., 1983). In rainbow trout, low manganese intake both decreased the activities of copper-zinc superoxide dismutase and manganese-superoxide dismutase in cardiac muscle and liver and suppressed manganese and calcium concentrations of the vertebrae (Knox et al., 1981b). Liver manganese-superoxide dismutase activity was not influenced in catfish fed a diet containing 2.4 mg Mn/kg of diet (Gatlin and Wilson, 1984b). In broodstock rainbow trout, a fishmeal-based diet without manganese supplement caused poor hatchability and low manganese concentration in the eggs (Takeuchi et al., 1981).

Selenium is an integral part of the enzyme glutathione peroxidase (Rostruck et al., 1973). This enzyme can reduce hydrogen peroxide and fatty acyl hydroperoxides in water and fatty acyl alcohols, respectively, thereby protecting cells and membranes against peroxide damage.

Selenium deficiency causes growth depression in rainbow trout (Hilton et al., 1980) and catfish (Gatlin and Wilson, 1984c), but selenium deprivation alone does not produce pathological signs in these fish. Both selenium and vitamin E are required to prevent muscular dystrophy in Atlantic salmon (Poston et al., 1976) and exudative diathesis

in rainbow trout (Bell et al., 1985). Glutathione peroxidase activity in plasma and liver is a useful index of selenium status in fish (Poston et al., 1976; Hilton et al., 1980: Gatlin and Wilson, 1984c; Bell et al., 1985, 1986, 1987).

The selenium requirement of fish varies with the form of selenium ingested, polyunsaturated fatty acid and vitamin E content of the diet, and concentration of waterborne selenium. The selenium requirement determined on the basis of optimum growth and maximum plasma glutathione peroxidase activity was estimated to be 0.15 to 0.38 mg Se/kg diet for rainbow trout (Hilton et al., 1980) and 0.25 mg Se/kg for channel catfish (Gatlin and Wilson, 1984c). The biological availability of selenium differs in various selenium compounds and feed supplements. Bell and Cowey (1989) reported that selenium present in fishmeal has low digestibility, whereas selenomethionine is highly digestible. A low concentration of selenium is found extensively in aquatic ecosystems. The uptake of selenium across gills is very efficient at low-waterborne concentrations (Hodson and Hilton, 1983).

Selenium toxicity occurred in rainbow trout and catfish when dietary selenium exceeded 13 and 15 mg/kg dry feed, respectively (Hilton et al., 1980; Gatlin and Wilson, 1984c). Reduced growth, poor feed efficiency, and high mortality were the major effects. Trout reared on high-selenium diets (10 mg/kg) also showed renal calcinosis (Hilton and Hodson, 1983).

Iodine is essential for the biosynthesis of the thyroid hormones, thyroxine and triiodothyronine. Fish obtain iodine from water via branchial pumps and from feed sources (Leloup, 1970). The total uptake of iodine depends on the iodine content of the feed and water (Gregory and Eales, 1975). Under laboratory conditions, rainbow trout derive 80 percent of their iodide from water, 19 percent from diet, and less than 1 percent by recycling iodide from thyroid hormone degradation (Hunt and Eales, 1979).

Iodine deficiency caused thyroid hyperplasia in brook trout (Marine, 1914). Thyroid hormone deficiency has been induced by glucosinolates in the diet (Higgs and Eales, 1978). A deficiency of ascorbic acid caused hypoactivity of the thyroid gland as demonstrated by a reduction in accumulation of ¹³¹I by thyroid glands in scorbutic snakehead (Agrawal and Mahajan, 1981). The minimum iodine requirement of most fish species has not been established. Woodall and LaRoche (1964) reported higher iodine requirements for advanced chinook salmon parr compared with fingerlings due to increased thyroid activity during smoltification. Lall et al. (1985) observed that relatively high concentrations of iodine and fluorine (4.5 mg/kg of diet of each) were essential to protect Atlantic salmon from bacterial kidney disease infections.

Information on the dietary requirements of other trace elements is limited. Increased dietary fluoride enhanced fluoride accumulation in the vertebrae of rainbow trout (Tiews et al., 1982; Bowser et al., 1988). Evidence also indicated that elevated concentrations of fluoride may reduce the prevalence of bacterial kidney disease (Lall et al., 1985; Bowser et al., 1988). The importance of chromium and other trace elements essential for other animals and humans is also recognized in fish nutrition, although the effects of their deficiencies have not been reported. Tacon and Beveridge (1982) found that rainbow trout fed a low-chromium chemically defined diet did not show deficiency signs or change in tissue chromium distribution.

Vitamins are organic compounds, distinct from amino acids, carbohydrates, and lipids, that are required in trace amounts from an exogenous source (usually the diet) for normal growth, reproduction, and health. Vitamins are classified as water-soluble and fat soluble. Eight of the water-soluble vitamins are required in relatively small amounts, have primarily coenzyme functions, and are known as the vitamin B complex. Three of the water-soluble vitamins, choline, inositol, and vitamin C, are required in larger quantities and have functions other than coenzymes. Vitamins A, D, E, and K are the fat soluble vitamins that function independent of enzymes or, in some cases such as vitamin K, may have coenzyme roles. In mammals the absence of vitamins leads to characteristic deficiency diseases, but in fish such diseases are less specifically identified.

Some vitamins may be synthesized from other essential nutrients to spare a portion of the dietary requirement. For example, channel catfish appear to synthesize choline if adequate methyl donors such as methionine are present in the diet; however, if the concentration of dietary methionine is limiting, a choline requirement can be demonstrated (Wilson and Poe, 1988). An exogenous source of some water-soluble vitamins for certain warm-water fish has been shown to be derived from microorganisms in the gastrointestinal tract (Limsuwan and Lovell, 1981; Lovell and Limsuwan, 1982; Burtle and Lovell, 1989). In cold-water carnivorous fish microorganisms are not a significant source of vitamins (Hepher, 1988).

Both qualitative and quantitative vitamin requirements of fish have been determined by feeding chemically defined diets deficient in a specific vitamin. The quantitative requirements for most of the vitamins have been established for chinook salmon, rainbow trout, common carp, channel catfish, and yellowtail, while only some of the requirements are known for red sea bream and tilapia. Qualitative requirements

have been identified in several other species. The requirements are affected by size, age, and growth rates as well as by various environmental factors and nutrient interrelationships. Thus, different researchers have reported fairly wide ranges in requirement values for growth in the same species (see Table 1-15). Recent studies with spring chinook salmon indicate that the dietary requirements for certain vitamins may be lower than previously reported for this species (Leith et al., 1990). In addition, the requirement values listed in Table 1-15, as determined by maximum liver storage or based on certain enzyme data, are often much higher than the requirement values based on weight gain and absence of deficiency signs; therefore, professional judgment must be used in selecting which requirement value best fits the user's needs. Thus, more studies are needed to refine the requirements for various species for normal growth, health, and enhancement of defense mechanisms, as suggested by Ikeda (1985). A summary of vitamin deficiency signs reported in several cultured fishes are presented in Appendix Table A-3. Further information on vitamin nutrition research in fishes is discussed by Halver (1989).

The fat-soluble vitamins, A, D, E, and K, are absorbed in the intestine along with dietary fats; therefore, conditions favorable for fat absorption also enhance the absorption of fat-soluble vitamins. Fat-soluble vitamins are stored by animals if dietary intake exceeds metabolic needs. Thus, animals can accumulate enough fat-soluble vitamins in their tissues to produce a toxic condition (hypervitaminosis). This has been demonstrated in the laboratory with trout for vitamins A, D, and E, but it is unlikely to occur under practical conditions (Poston et al., 1966; Poston, 1969a; Poston and Livingston, 1969).

Since fat-soluble vitamins can be stored in the body, the nutritional history of experimental fish prior to their use in requirement studies becomes critical. The time required to deplete fish of their stored fat-soluble vitamins is highly variable. Differences in vitamin intake prior to an experiment may be responsible for some of the conflicting findings on the induction and severity of deficiency signs.

Vitamin A is required in vertebrates for the regeneration of the light-sensitive compound rhodopsin in the retina of the eye. Vitamin A has also been shown to be essential for proper growth, reproduction, resistance to infection, and the maintenance of differentiated epithelia and mucus secretions. Blomhoff et al. (1992) have presented a recent review of metabolic functions of vitamin A in vertebrates.

Vitamin A occurs in three forms: as an alcohol (retinol), an aldehyde (retinal), and an acid (retinoic acid). Vitamin A₁ (retinol) is found in mammals and marine fishes, whereas both vitamin A₁ and vitamin A₂ (3-dehydroretinol) are found in freshwater fishes (Braekkan et al., 1969; Lee, 1987). In freshwater fish, the oxidative conversion of retinol to 3-dehydroretinol occurs (Goswami, 1984) as well as the reversible oxidation and reduction reactions of retinol to retinal and of 3-dehydroretinol to 3-dehydroretinal (Wald, 1945-1946). For example, tilapia has been shown to convert dietary retinol into 3-dehydroretinol and retinal into 3-dehydroretinal (Katsuyama and Matsuno, 1988). Channel catfish were found to convert β-carotene to vitamin A₁ and A₂ in about a 1:1 ratio (Lee, 1987).

Cold-water fish can use β-carotene as a vitamin A precursor (Poston et al., 1977). Dupree (1970) found that channel catfish could use β-carotene as a vitamin A source only if the dietary concentration exceeded 2,000 international units per kilogram (IU/kg). It has recently been shown that β-carotene and canthaxanthin can be biotransformed in the liver of tilapia into vitamin A₁ and that dihydroxycarotenoids such as astaxanthin, zeaxanthin, lutein, and tunaxanthin were directly bioconverted into vitamin A₂ (Katsuyama and Matsuno, 1988). In mammals, carotenoids have been found to fulfill various biological functions independent of vitamin A (Olson, 1989). Thus, more studies are needed on the metabolic role of carotenoids in fish and on the possibility that carotenoids serve as a provitamin A.

Vitamin A deficiency in rainbow trout causes anemia, twisted gill opercula, and hemorrhages in the eyes and base of fins (Kitamura et al., 1967a). Brook trout exhibited poor growth, high mortality, and eye lesions, such as edematous eyes, displaced lens, and degeneration of the retina, when fed a vitamin A-deficient, purified diet from first feeding (Poston et al., 1977). Channel catfish fed 0.4 mg of β-carotene/kg of diet for 3 years developed exophthalmia, edema, and hemorrhagic kidney (Dupree, 1966). Anorexia, pale body color, hemorrhagic skin and fins, exophthalmia, and twisted gill opercula occurred in common carp fed a vitamin A-deficient diet after 8 to 11 weeks (Aoe et al., 1968). Rapidly growing yellowtail fingerlings fed a vitamin A-deficient diet developed deficiency signs in 20 days including arrested growth of gill opercula, dark pigmentation, anemia, and hemorrhage in the eyes and liver, accompanied by high mortality (Hosokawa, 1989).

High dietary intake (2.2 million IU/kg diet) of retinyl palmitate caused slow growth, anemia, and severe necrosis of the caudal fin of brook trout at 8.3°C (Poston et al., 1966). Feeding up to 2.5 million IU retinyl palmitate to trout at 12.4°C also reduced body fat and liver size (Poston, 1971a). A high intake of dietary protein (Poston and Livingston, 1971) or methionine (Eckhert and Kemmerer, 1974) by young trout reduced the toxicity of excess dietary vitamin A observed in fish fed a low-protein diet.

Vitamin A is added to fish feeds as the acetate, palmitate, or propionate ester in the form of free-flowing beadlets in a multivitamin premix.

The two major natural sources of vitamin D are ergocalciferol (vitamin D₂, which occurs predominantly in plants) and cholecalciferol (vitamin D₃, which occurs in animals). Both forms of vitamin D are hydroxylated in the liver to the 25-hydroxy forms. The 25-hydroxy-D₃ is further hydroxylated in the kidney to 1,25-dihydroxyvitamin D₃, which is the biologically active form of vitamin D responsible for facilitating mobilization, transport, absorption, and use of calcium and phosphorus in concert with the actions of parathyroid hormone and calcitonin.

Cholecalciferol has been shown to be at least three times more effective than ergocalciferol in meeting the vitamin D requirement of rainbow trout (Barnett et al., 1982a). Andrews et al. (1980) found that vitamin D₃ was used more effectively by catfish than vitamin D₂ at dietary concentrations of 2,000 IU/kg of diet and that high concentrations of vitamin D₃ (20,000 to 50,000 IU/kg of diet) reduced weight gain. Brown (1988), however, found that vitamin D₂ was utilized as well as vitamin D₃ up to 1,500 IU/kg of diet, but higher concentrations of vitamin D₂ depressed weight gain and feed efficiency in channel catfish reared in calcium-free water.

Rainbow trout fed a vitamin D-deficient diet exhibited poor growth, elevated liver lipid content, impaired calcium homeostasis manifested by tetany of white skeletal muscles, and ultrastructural changes in the white muscle fibers of the epaxial musculature (George et al., 1981). However, in a similar study also with rainbow trout, no hypocalcemia or changes in bone ash were observed (Barnett et al., 1982a). A lordosis-like droopy tail syndrome observed in vitamin D-deficient trout (Barnett et al., 1982b) was suggested to be related to an epaxial muscle weakness. Channel catfish fed a vitamin D-deficient diet for 16 weeks showed poor growth, lowered body calcium and phosphorus levels, and lowered total body ash (Lovell and Li, 1978). Andrews et al. (1980) reported that vertebral ash level in channel catfish was not significantly affected by vitamin D deficiency.

Fingerling brook trout fed 3.75 × 10⁶ IU vitamin D3/kg diet for 40 weeks had hypercalcemia and increased hematocrit levels but no difference in rates of growth and survival (Poston, 1969a). However, Hilton and Ferguson (1982) did not detect any incidence of renal calcinosis in rainbow trout

fed a diet containing up to 1 × 10⁶ IU vitamin D₃/kg diet. Supplementation of 50,000 IU vitamin D₃/kg diet significantly depressed the growth rate of channel catfish (Andrews et al., 1980). By contrast, a diet of 1 × 10⁶ IU vitamin D₃/kg has been reported to show no toxic effects in channel catfish reared in calcium-free water for 14 weeks (Brown, 1988).

Vitamin D₃ is added to fish feeds either in a beadlet with vitamin A or as a spray or drum-dried powder in a multivitamin premix.

Vitamin E is a generic descriptor for all the molecules that possess the biological activity of α-tocopherol. Natural forms of vitamin E are all d-stereoisomers and consist of a substituted aromatic ring and a long isoprenoid side chain. There are eight naturally occurring compounds with vitamin E activity: d-α-; d-β-; d-γ-; d-δ-tocopherols, which differ in the number and position of the methyl groups in the aromatic ring; and their corresponding tocotrienols. The compound with the highest biopotency is d-α-tocopherol. The other tocopherol isomers have some, but very low, biological activity. No interconversion between a-tocopherol and the other tocopherol forms has been detected in liver or muscle tissue of rainbow trout (Watanabe et al., 1981c). The free tocopherol form of vitamin E is unstable to oxidizing conditions; whereas the acetate and succinate esters are quite stable. These ester forms possess no antioxidant activity, but they are readily hydrolyzed in the digestive tract to the biologically active free tocopherol. One IU of vitamin E is defined as the biological activity of 1 mg of DL-α-tocopheryl.

Vitamin E functions in vitro as a very good antioxidant in a manner similar to several synthetic antioxidants. In vivo, vitamin E and selenium (via glutathione peroxidase) function as parts of a multicomponent antioxidant defense system. This system protects the cell against the adverse effects of reactive oxygen and other free radical initiators of the oxidation of polyunsaturated membrane phospholipids, critical proteins, or both.

Vitamin E deficiency signs have been described for chinook salmon (Woodall et al., 1964), Atlantic salmon (Poston et al., 1976), channel catfish (Dupree, 1968; Murai and Andrews, 1974; Lovell et al., 1984; Wilson et al., 1984), common carp (Watanabe et al., 1970a,b, 1981a), rainbow trout (Cowey et al., 1981, 1983; Hung et al., 1981; Watanabe et al., 1981b; Moccia et al., 1984) and yellowtail (Toyoda, 1985). The deficiency signs of vitamin E in various fishes are similar and include muscular dystrophy involving atrophy and necrosis of white muscle fibers; edema of heart, muscle, and other tissues due to increased capillary permeability allowing exudates to escape and accumulate, which are often green in color as a result of hemoglobin breakdown; anemia and impaired erythropoiesis; depigmentation; and ceroid pigment in the liver. The incidence and severity of these deficiency signs have been shown to be enhanced when diets deficient in both vitamin E and selenium were fed to Atlantic salmon (Poston et al., 1976), rainbow trout (Bell et al., 1985), and channel catfish (Gatlin et al., 1986). These latter observations demonstrated a significant interaction between selenium and vitamin E in the nutrition of fish.

Erythrocyte fragility has been used as an indicator of vitamin E status in some animals (Draper and Csallany, 1969). Peroxide hemolysis of red blood cells has been used to determine vitamin E deficiency in rainbow trout (Hung et al., 1981); however, this procedure was not sensitive enough to aid in determining the vitamin E requirement in rainbow trout (Cowey et al., 1981) and channel catfish (Wilson et al., 1984). Cowey et al. (1981) found that in vitro ascorbic acidstimulated lipid peroxidation in liver microsomes of rainbow trout accurately reflected -tocopherol status. This latter procedure has also been used to assess vitamin E status in channel catfish (Wilson et al., 1984; Gatlin et al., 1986).

When high concentrations of dietary polyunsaturated fatty acids are involved in the diets of common carp (Watanabe et al., 1981a) and rainbow trout (Watanabe et al., 1981b; Cowey et al., 1983), the requirement for vitamin E is increased. Vitamin E-deficient rainbow trout have been reported to have significantly reduced immune and nonspecific responses to infection (Blazer and Wolke, 1984a); however, Salte et al. (1988) could show no beneficial effect of dietary vitamin E supplementation alone or in combination with selenium as a prophylaxis for Hitra disease in Atlantic salmon.

High dietary concentrations of vitamin E (5,000 mg of DL-α-tocopherol/kg of diet) have been shown to cause reduced concentrations of erythrocytes in trout blood (Poston and Livingston, 1969).

Vitamin E is added to fish feeds as a dry powder form of DL-α-tocopheryl acetate.

Vitamin K is required for stimulation of prothrombin activity in plasma and synthesis of blood clotting factors VII, IX, and X. The metabolic role of vitamin K involves the vitamin K-dependent carboxylase, which carries out the posttranslational conversion of specific glutamyl residues in the vitamin K-dependent plasma proteins to γ-carboxy-glutamyl residues. These residues are essential for the normal, Ca²⁺-dependent, interaction of the vitamin K-dependent clotting factors with phospholipid surfaces (Suttie, 1985).

The term vitamin K is used as a generic descriptor for both 2-methyl-1,4-naphthoquinone and all 3-substituted derivatives of this compound, which exhibit an antihemorrhagic activity in animals fed a vitamin K-deficient diet. The three major forms of vitamin K include: vitamin K₁ or

phylloquinone, which can be isolated from plants; vitamin K₂ or the menaquinones, which are synthesized by bacteria; and vitamin K₃ or menadione which is a synthetic product.

Many animals do not require vitamin K in the diet because of bacterial synthesis in the intestinal tract, but intestinal vitamin K-synthesizing microflora have not been described in fish (Margolis, 1953). Supplementation of sulfaguanidine to a vitamin K-deficient diet and low water temperature caused prolonged blood coagulation time and low hematocrit values without affecting growth performance of trout (Poston, 1964). Dupree (1966) reported hemorrhages in channel catfish fed a vitamin K-deficient diet. However, Murai and Andrews (1977) failed to detect any deficiency signs in channel catfish fed a diet devoid of vitamin K and supplemented with sulfaguanidine. The addition of dicumarol, a vitamin K antagonist, did not increase prothrombin time in catfish. The addition of pivalyl, a stronger (20 times) vitamin K antagonist than dicumarol, completely blocked the blood coagulation of channel catfish (Murai and Andrews, 1977). High-dietary concentrations of menadione sodium bisulfite (2,400 mg/kg of diet) had no adverse affect on growth, survival, blood coagulation, or the number of erythrocytes of young trout (Poston, 1971b).

Vitamin K is added to fish feeds as a menadione salt—menadione sodium bisulfite (50 percent K₃), menadione sodium bisulfite complex (33 percent K₃), or menadione dimethylpyrimidinol bisulfite (45.5 percent K₃).

The water-soluble vitamins, with the exception of two water-soluble growth factors (choline and myoinositol) and ascorbic acid, have unique coenzyme functions in cellular metabolism. Yet, it is not always possible to correlate a sign of deficiency with a diminished function of an enzyme system for which that vitamin is essential. For some warm-water fishes, intestinal synthesis by microorganisms supplies the requirement for certain vitamins. Thus, deficiency signs result only in those cases when antibiotics are fed along with a deficient diet. A constant supply of essential water-soluble vitamins is required to prevent deficiency signs in fish, since these vitamins are not stored in body tissues.

The coenzyme form of thiamin is thiamin pyrophosphate. Thiamin pyrophosphate functions in the oxidative decarboxylation of α-keto acids, such as pyruvate and α-ketoglutarate, and in the transketolase reaction in the pentose shunt.

Dietary thiamin deficiency has been shown to result in neurological disorders such as hyperirritability in salmonids (Halver, 1957; Coates and Halver, 1958; Kitamura et al., 1967b; Lehmitz and Spannhof, 1977), channel catfish (Dupree, 1966; Comacho, 1978), Japanese eel (Hashimoto et al., 1970), and Japanese parrotfish (Ikeda et al., 1988). However, Murai and Andrews (1978a) did not observe neurological disorders in thiamin-deficient channel catfish. Arai et al. (1972) found only subcutaneous hemorrhages and congested fins in subadult Japanese eels, and Hashimoto et al. (1970) observed neurological disorders in small Japanese eels. Similar deficiency signs with varying degrees of mortality have been reported in common carp (Aoe et al., 1969), red sea bream (Yone and Fujii, 1974), turbot (Cowey et al., 1975), and yellowtail (Hosokawa, 1989).

Erythrocyte transketolase activity has been used as a specific indicator of thiamin status in the turbot (Cowey et al., 1975). Kidney or liver transketolase activity in rainbow trout (Lehmitz and Spannhof, 1977; Masumoto et al., 1987) and thiamin content in the blood of yellowtail (Hosokawa, 1989) also have been shown to decrease much earlier than the appearance of external deficiency signs.

Thiamin is added to fish feeds as thiamin mononitrate, which is 91.9 percent thiamin. Thiamin mononitrate is stable in vitamin premixes that do not contain trace minerals and choline chloride.

Riboflavin functions in the intermediary transfer of electrons in metabolic oxidation-reduction reactions as a component of two coenzymes, flavin monouncleotide (FMN) and flavin adenine dinucleotide (FAD). These coenzymes serve as prosthetic groups of oxidation-reduction enzymes involved in the metabolism of keto-acids, fatty acids, and amino acids in the mitochondrial electron transport system.

Species-specific deficiency signs are found in fish. The only common signs are anorexia and poor growth. The first sign of riboflavin deficiency observed in salmonids (McLaren et al., 1947; Halver, 1957; Steffens, 1970; Takeuchi et al., 1980; Hughes et al., 1981a,b) appeared in the eyes and included photophobia, cataracts, corneal vascularization, and hemorrhages. Lack of coordinated swimming and dark skin coloration have also been reported for riboflavin-deficient chinook salmon (Halver, 1957) and rainbow trout (Kitamura et al., 1967b; Steffens, 1970). In contrast, Woodward (1984) did not observe cataracts or corneal occlusion in riboflavin-deficient rainbow trout fry and fingerlings; however, severe fin erosion and light skin coloration accompanied by high mortality were observed. The eye lesions and dark skin coloration followed by high mortality have also been observed in riboflavin-deficient yellowtail fingerlings (Hosokawa, 1989). Riboflavin-deficient common carp (Aoe et al., 1967c; Ogino, 1967; Takeuchi et al., 1980) and Japanese eel (Arai et al., 1972) exhibited hemorrhages in various parts of the body, nervousness, and photophobia but no evidence of cataract development. Monolateral or bilateral

cataracts have been reported in riboflavin-deficient channel catfish (Dupree, 1966), but Murai and Andrews (1978b) found only poor growth and short-body dwarfism in two independent feeding trials with channel catfish. Lethargy and high mortality have been reported in Japanese parrotfish fed riboflavin-deficient diets (Ikeda et al., 1988).

Hughes et al. (1981a) used the activation coefficient (ratio of activity following preincubation with FAD:basal activity) of erythrocyte glutathione reductase to measure the riboflavin status of rainbow trout. However, Woodward (1983) found the activity of D-amino acid oxidase to be a more sensitive indicator of the riboflavin status in rainbow trout, since the low activity of erythrocyte glutathione reductase made its quantification difficult. Amezaga and Knox (1990) also found that hepatic D-amino acid oxidase was a reliable indicator of riboflavin status in rainbow trout. They pointed out, however, that an assay for glutathione reductase activity in erythrocytes would be advantageous since it could be used on live fish. Woodward (1985) reported that the riboflavin requirement was not affected by temperature or by genetic differences in growth rate. This might be one reason why the riboflavin requirement values shown in Table 1-15 agree fairly well even among different species.

Hughes (1984) found that feeding high concentrations of riboflavin (up to 600 mg/kg diet) had no adverse effects on growth of rainbow trout. These results were expected since riboflavin has not been shown to cause hypervitaminosis in other animals. However, two previous studies (McLaren et al., 1947; Woodward, 1982) had reported depressed growth in rainbow trout fed moderate concentrations of riboflavin. It was concluded that the growth depression observed in the earlier studies must have resulted from some factor other than riboflavin.

Riboflavin is added to fish feeds as a dry powder in a multivitamin premix.

The term vitamin B₆ is the generic descriptor for the 2-methylpyridine derivatives that have the biological activity of pyridoxine. Pyridoxine is the main form found in plant products, whereas pyridoxal and pyridoxamine are the principal forms found in animal tissue. All three forms are readily converted in animal tissue to the coenzyme forms, pyridoxal phosphate and pyridoxamine phosphate. Pyridoxal phosphate is required for many enzymatic reactions involving amino acids such as transamination, decarboxylation, and dehydration. Pyridoxal phosphate also functions in the biosynthesis of porphyrins and in the catabolism of glycogen.

Pyridoxal phosphate is required for the synthesis of the neurotransmitters—5-hydroxytryptamine and serotonin—from tryptophan. Consequently, signs of pyridoxine deficiency include nervous disorders—erratic swimming, hyperirritability, and convulsions—that have been observed in salmonids (Halver, 1957; Coates and Halver, 1958), gilthead sea bream (Kissil et al., 1981), channel catfish (Andrews and Murai, 1979), common carp (Ogino, 1965), yellowtail (Sakaguchi et al., 1969), and Japanese eel (Arai et al., 1972).

Other deficiency signs such as anorexia and poor growth usually appear in the fish within 3 to 6 weeks after being fed a pyridoxine-deficient diet. Pyridoxine deficiency has been reported to cause various histopathological changes in rainbow trout liver (Jurss and Jonas, 1981) and kidney (Smith et al., 1974) and in the intestinal tissue of both rainbow trout (Smith et al., 1974) and gilthead sea bream (Kissil et al., 1981).

The activity of certain aminotransferase enzymes that require pyridoxal phosphate as a coenzyme has been used as an index of pyridoxine status in fish. Serum or tissue alanine and/or aspartate aminotransferase activities have been used to evaluate pyridoxine status in common carp (Ogino, 1965), rainbow trout (Smith et al., 1974; Jurss, 1978), chinook salmon (Hardy et al., 1979), turbot (Adron et al., 1978), and gilthead sea bream (Kissil et al., 1981).

Vitamin B₆ is added to fish feeds as pyridoxine hydrochloride in a dry form as part of a multivitamin premix.

Pantothenic acid is a component of coenzyme A (CoA), acyl CoA synthetase, and acyl carrier protein. The coenzyme form of the vitamin is therefore responsible for acyl group transfer reactions. Coenzyme A is required in reactions in which the carbon skeletons of glucose, fatty acids, and amino acids enter into the energy-yielding tricarboxylic acid cycle. Acyl carrier protein is required for fatty acid synthesis.

A deficiency of this vitamin impairs the metabolism of mitochondria-rich cells that undergo rapid mitosis and high-energy expenditure. Thus, deficiency signs have been found to appear within 10 to 14 days in rapidly growing fish such as fingerling yellowtail (Hosokawa, 1989). Gill lamellar hyperplasia or clubbed gills is a characteristic sign of pantothenic acid deficiency in most fish. In addition to clubbed gills, anemia and high mortality have been observed in pantothenic acid-deficient salmonids (Phillips et al., 1945; McLaren et al., 1947; Coates and Halver, 1958; Kitamura et al., 1967b; Poston and Page, 1982; Karges and Woodward, 1984), channel catfish (Dupree, 1966; Murai and Andrews, 1979; Brunson et al., 1983; Wilson et al., 1983), and yellowtail (Hosokawa, 1989). Pantothenic acid-deficient Japanese parrotfish exhibited anorexia, convulsions, and cessation of growth followed by high mortality (Ikeda et al., 1988). Similar deficiency signs were observed in red sea bream (Yone and Fujii, 1974). Slow growth, anorexia, lethargy, and anemia were observed in common carp (Ogino, 1967). Poor growth, hemorrhage, skin lesions, and abnormal swimming were found in Japanese eel (Arai et al., 1972) fed pantothenic acid-deficient diets.

Pantothenic acid is added to fish feeds as either calcium d-pantothenate (92 percent activity) or calcium DL-pantothenate (46 percent activity) as a dry powder in a multivitamin premix.

Niacin is used as the generic descriptor of pyridine 3-carboxylic acids and their derivatives that exhibit the biological activity of nicotinamide (the amide of nicotinic acid). Of the compounds with niacin activity, nicotinic acid and nicotinamide have the greatest biological activity. Niacin is widely distributed in both plant and animal tissue. Much of the niacin in plant material, however, is present in bound forms that have limited availability to fish.

Niacin is a component of the two coenzymes nicotinamide adenine dinucleotide (NAD) and nicotinamide adenine dinucleotide phosphate (NADP). These coenzymes are essential for several oxidation-reduction reactions involving the transfer of hydrogen and electrons in carbohydrate, lipid, and amino acid metabolism. They are also involved in various energy yielding and biosynthetic pathways including the mitochondrial electron transport system. Tryptophan can be metabolically converted to niacin in many animals, but not in certain salmonid fish (Poston and DiLorenzo, 1973; Poston and Combs, 1980). The fact that niacin deficiency can readily be induced in various fish indicates that most if not all fish lack the capacity for niacin synthesis.

Trout and salmon fed niacin-deficient diets exhibited anorexia, poor growth, poor feed conversion, photosensitivity or sunburn, intestinal lesions, abdominal edema, muscular weakness, spasms, and increased mortality (McLaren et al., 1947; Phillips and Brockway, 1947; Halver, 1957). Channel catfish (Andrews and Murai, 1978) and common carp (Aoe et al., 1967b) showed skin and fin lesions, high mortality, skin hemorrhages, anemia, and deformed jaws when fed niacin-deficient diets for 2 to 6 weeks. Skin hemorrhages, dermatitis, anemia, abnormal swimming, and ataxia were observed in Japanese eels fed a niacin-deficient diet for 14 weeks (Arai et al., 1972).

Poston and Wolfe (1985) have experimentally demonstrated the interaction between the occurrence of dermal lesions and niacin deficiency. Two weeks after exposure of niacin-deficient rainbow trout to ultraviolet radiation, a total loss of mucus-producing cells was observed in histopathological sections of the epidermis.

High dietary intake of niacin (10,000 mg/kg) increased liver fat, decreased body fat, and tended to reduce growth rate in fingerling brook trout (Poston, 1969b).

Niacin is added to fish feeds as either nicotinic acid or niacinamide; both have similar biological activity. Nicotinic acid or niacinamide is added to the multivitamin premix in a dry form.

Biotin acts in certain metabolic reactions as an intermediate carrier of carbon dioxide during carboxylation and decarboxylation reactions. Specific enzymes that require biotin include acetyl-CoA carboxylase, pyruvate carboxylase, and propionyl-CoA carboxylase. Metabolic pathways requiring biotin include the biosynthesis of long-chain fatty acids and the synthesis of purines.

In many animals, a biotin deficiency can only be induced by feeding avidin, a glycoprotein found in raw chicken egg white that binds biotin and prevents absorption of the vitamin from the intestine. Robinson and Lovell (1978) fed avidin in a biotin-free chemically defined diet to channel catfish and noted a growth suppression that led them to suggest some biotin synthesis by intestinal microflora in this species. However, in a later study by Lovell and Buston (1984) no synthesis of biotin by the intestinal microflora in channel catfish could be detected.

Common carp required 8 to 12 weeks (Ogino et al., 1970a) and channel catfish took 11 weeks (Lovell and Buston, 1984) to show growth depression when fed biotindeficient diets. A similar effect in rainbow trout took only 4 to 8 weeks in water temperatures of 15°C (Woodward and Frigg, 1989). Anorexia, reduced weight gain, and higher feed conversion were more noticeable in smaller than in larger rainbow trout fed biotin-deficient diets (Walton et al., 1984). Biotin-deficient channel catfish exhibited skin depigmentation (Robinson and Lovell, 1978), whereas biotin-deficient Japanese eels had darker skin coloration (Arai et al., 1972). Histological signs of biotin deficiency were not detected after 12 weeks in rainbow trout having an initial weight of 25 g (Walton et al., 1984). However, severe deficiency signs were produced in rainbow trout and lake trout having initial weights of 1.3 and 6.7 g, respectively (Poston and Page, 1982; Woodward and Frigg, 1989). Rainbow trout and lake trout developed biotin-related histopathological signs in the gills (Castledine et al., 1978; Poston and Page, 1982), liver (Poston, 1976b; Poston and Page, 1982), and kidney (Poston and Page, 1982).

Hepatic pyruvate carboxylase activity in rainbow trout fed a lipid-free and biotin-deficient diet decreased to 3.3 percent of that in fish fed a diet sufficient in lipid and biotin, although the enzyme activity was restored to about 50 percent of normal following the addition of lipid to the diet (Walton et al., 1984). In contrast, lipid supplementation of biotin-deficient diets did not increase hepatic pyruvate carboxylase activity in channel catfish (Robinson and Lovell, 1978).

Signs of biotin deficiency were not detected in rainbow trout (Castledine et al., 1978) or channel catfish (Lovell and Buston, 1984) fed natural ingredient diets without supplemented biotin for 24 and 17 weeks, respectively. These studies concluded that adequate biotin was available in the

various feed ingredients in the natural ingredient diets used to meet the requirements of the fish.

Biotin is added to fish feeds when necessary as D-biotin in a dry form in the multivitamin premix.

The term folate is used as the generic descriptror for folic acid and related compounds exhibiting qualitatively the biological activity of folic acid. Folic acid is composed of a pteridine ring linked through a methylene bridge to p-aminobenzoic acid to form pteroic acid, which is in turn linked as an amide to glutamic acid. Folic acid undergoes enzymatic reduction in the tissues to its active coenzyme form, tetrahydrofolic acid. It functions as an intermediate carrier of one-carbon groups in a number of complex enzymatic reactions. In these reactions, methyl, methylene, and other one-carbon groups are transferred from one molecule to another. These reactions are found in the metabolism of certain amino acids and the biosynthesis of purines and pyrimidines along with the nucleotides found in DNA and RNA.

Trout and salmon fed folate-deficient diets exhibited anorexia; reduced growth; poor feed conversion; and macrocytic normochromic, megaloblastic anemia (Smith, 1968; Smith and Halver, 1969) characterized by pale gills, anisocytosis, and poikilocytosis. The erythrocytes were large with abnormally segmented and constricted nuclei, and a large number of megaloblastic proerythrocytes were present in the erythropoietic tissue of the anterior kidney. Production of erythrocytes decreased with time in fish fed the folate-deficient diet. Some of these signs have also been observed in the rohu (John and Mahajan, 1979).

Poor growth and dark skin coloration were noted in Japanese eels fed a folate-deficient diet for 10 weeks (Arai et al., 1972). Folate-deficient yellowtail fingerlings also showed congestion in fins and bronchial mantle, dark skin coloration, and anemia (Hosokawa, 1989). Folate deficiency signs in channel catfish included reduced growth, anemia, and increased sensitivity to bacterial infection (Duncan and Lovell, 1991). Deficiency signs were not observed in common carp (Aoe et al., 1967a) fed a folate-free diet, presumably due to bacterial synthesis of folate in the intestine (Kashiwada et al., 1971).

Folate is added to fish feeds as folic acid as a dry powder in a multivitamin premix.

The term vitamin B₁₂ should be used as the generic descriptor for all corrinoids exhibiting qualitatively the biological activity of cyanocobalamin. This vitamin was previously known as vitamin B₁₂ or cyanocobalamin. Vitamin B₁₂ is a large molecule (molecular weight 1355) that contains a cobalt atom. Neither higher plants nor animals can synthesize vitamin B₁₂, but both depend on certain microorganisms for the trace amounts required. Vitamin B₁₂ is required for normal maturation and development of erythrocytes, for the metabolism of fatty acids, in the methylation of homocysteine to methionine, and for the normal recycling of tetrahydrofolic acid. Thus, a deficiency of vitamin B₁₂ can result in signs similar to folate deficiency.

Salmon (Halver, 1957) and trout (Phillips et al., 1964) fed low amounts of vitamin B₁₂ showed a high variability in numbers of fragmented erythrocytes and in hemoglobin values, with a tendency for a microcytic, hypochromic anemia. Channel catfish fed a vitamin B₁₂-deficient diet for 36 weeks exhibited reduced growth but no other clinical deficiency signs (Dupree, 1966). John and Mahajan (1979) observed reduced growth and lower hematocrit in rohu fed a vitamin B₁₂-deficient diet. Japanese eel were found to require vitamin B₁₂ for normal appetite and growth (Arai et al., 1972).

Intestinal microfloral synthesis appeared to satisfy the B₁₂ requirement of Nile tilapia (Lovell and Limsuwan, 1982), but channel catfish required dietary supplementation of B₁₂ to prevent anemia (Limsuwan and Lovell, 1981). Intestinal microfloral synthesis of vitamin B₁₂ has been demonstrated in common carp (Kashiwada et al., 1970; Sugita et al., 1991a), channel catfish (Limsuwan and Lovell, 1981; Sugita et al., 1990, 1991a), Nile tilapia (Lovell and Limsuwan, 1982; Sugita et al., 1990, 1991a), rainbow trout (Sugita et al., 1991b), and ayu and goldfish (Sugita et al., 1991a). Sugita et al. (1991a) found a close relationship between the amount of vitamin B₁₂ and the viable counts of Bacteroides type A in the intestinal contents of the various fish studied. They found that this bacterium was present in the intestinal contents of fish that do not require vitamin B₁₂ and absent in those fish that do require vitamin B₁₂.

Vitamin B₁₂ is added to fish feeds when necessary in a dry form as part of a multivitamin premix.

Unlike the other water-soluble vitamins, choline has no known coenzyme function. Choline has three major metabolic functions: as a component of phosphatidylcholine, which has structural functions in biological membranes and in tissue lipid utilization; as a precursor of the neurotransmitter acetylcholine; and as a precursor of betaine, which serves as a source of labile methyl groups for methylation reactions such as the formation of methionine from homocysteine and creatine from guanidoacetic acid.

Rainbow trout fed a choline-deficient diet developed light yellow-colored livers, protruded eyes, anemia, and extended abdomens (kitamura et al., 1967a). Lake trout fed a choline-deficient diet for 12 weeks had depressed growth rate and increased liver fat content (Ketola, 1976). Depressed

growth, loss of appetite, and white-gray colored intestines were observed in Japanese eels fed a choline-deficient diet (Arai et al., 1972). Increased liver lipid content has been observed in common carp and channel catfish fed choline-deficient diets (Ogino et al., 1970b; Wilson and Poe, 1988). In addition, common carp developed vacuolization of hepatic cells after being on such a diet for 10 weeks (Ogino et al., 1970b). A thinning of the intestinal wall muscle and focal degeneration of the exocrine pancreas were observed in choline-deficient sturgeon (Hung, 1989).

Channel catfish fed casein-gelatin diets containing excess methionine did not develop signs of choline deficiency; however, catfish fed diets adequate but not excessive in methionine did develop deficiency signs (Wilson and Poe, 1988). Rumsey (1991) has suggested that 50 percent of the choline requirement of rainbow trout can be met from betaine. These observations indicate that certain fish can meet a part of their choline needs through the synthesis of choline by the methylation of ethanolamine, which uses methyl groups from S-adenosyl methionine.

Choline is added to fish feeds as a 70 percent choline chloride solution or a 25 to 60 percent dry powder. Choline chloride can decrease the stability of other vitamins in a multivitamin premix during prolonged storage.

Inositol may exist in one of seven optically inactive forms and as one pair of optically active isomers. Only one of these forms, myoinositol, possesses biological activity. Inositol is a biologically active cyclohexitol and occurs as a structural component in biological membranes as phosphatidylinositol. Recently, phosphatidylinositol was shown to be involved in signal transduction of several metabolic processes (Mathews and van Holde, 1990). Although similar in many respects to the adenylate cyclase transduction system, the phosphoinositide system is distinctive in that the hormonal stimulus activates a reaction that generates two second messengers. Membrane bound phosphatidylinositol 4,5-bisphosphate is cleaved to release sn-1,2-diacylglycerol and inositol 1,4,5-triphosphate, following the interaction of a hormone or agonist with the receptor on the cell membrane. Inositol 1,4,5-triphosphate stimulates the release of calcium from its intracellular stores in the endoplasmic reticulum, and sn-1,2-diacylglycerol activates protein kinase C to phosphorylate specific target proteins. Examples of cellular processes controlled by the phosphoinositide second messenger system include amylase secretion, insulin release, smooth muscle contraction, liver glycogenolysis, platelet aggregation, histamine secretion, and DNA synthesis in fibroblasts and lymphoblasts.

Signs of inositol deficiency have been reported to include poor appetite, anemia, poor growth, fin erosion, dark skin coloration, slow gastric emptying, and decreased cholinesterase and certain aminotransferase activities in trout (McLaren et al., 1947; Kitamura et al., 1967b), red sea bream (Yone et al., 1971), Japanese eel (Arai et al., 1972), Japanese parrotfish (Ikeda et al., 1988), and yellowtail (Hosokawa, 1989). Rainbow trout fed a diet devoid of inositol had large accumulations of neutral lipids in the liver, increased levels of cholesterol and triglycerides, but decreased amounts of total phospholipid, phosphotidylcholine, phosphotidylethanolamine, and phosphotidylinositol (Holub et al., 1982).

Inositol appears to be synthesized in common carp intestine (Aoe and Masuda, 1967), but not in amounts sufficient to sustain normal growth of young fish without an exogenous source of this vitamin, because younger carp require a higher level of inositol than older fish. Burtle and Lovell (1989) demonstrated de novo synthesis of inositol in the liver of channel catfish, as well as intestinal synthesis. High concentrations of dietary glucose may increase the need for inositol in some fish (Yone et al., 1971).

Myoinositol is added to fish feeds when necessary as a dry powder in a multivitamin premix.

Most animals can synthesize vitamin C, or L-ascorbic acid, from D-glucose, but many fish cannot (Kitamura et al., 1965; Poston, 1967; Halver et al., 1969; Wilson, 1973; Dabrowski, 1990). Ascorbic acid is a strong reducing agent and is readily oxidized to dehydroascorbic acid. Dehydroascorbic acid can be enzymatically reduced back to ascorbic acid in animal tissue with glutathione or reduced NADP. Ascorbic acid is a cofactor in the hydroxylation of proline and lysine to hydroxyproline and hydroxylysine in procollagen, which is the precursor of collagen and thus is necessary for the formation of connective tissues, scar tissue in wound repair, and bone matrix (Sandel and Daniel, 1988). Ascorbic acid also facilitates the absorption of iron, thus preventing the anemia often observed in ascorbic acid-deficient fish. In addition, ascorbic acid functions with vitamin E to minimize peroxidation of lipids in fish tissues (Heikkila and Manzino, 1987).

Vitamin C-deficient salmon and trout exhibited structural deformities (scoliosis, lordosis, and abnormal support cartilage of the eye, gill, and fins) and internal hemorrhaging usually preceded by nonspecific signs such as anorexia and lethargy (Halver et al., 1969; Hilton et al., 1978; Tsujimura et al., 1978; Sato et al., 1983), ascites and hemorrhagic exophthalmia (Poston, 1967), and high level of plasma triglycerides and cholesterol (John et al., 1979). Similar structural deformities such as scoliosis and lordosis due to vitamin C deficiency have been observed in channel catfish (Wilson and Poe, 1973; Andrews and Murai, 1974; Lim and Lovell, 1978; Wilson et al., 1989), Indian major carp (Agrawal and Mahajan, 1980), common carp and roach (Dabrowski et al., 1988, 1989), blue tilapia (Stickney et al., 1984),

Nile tilapia (Soliman et al., 1986a,b), and yellowtail (Sakaguchi et al., 1969). Japanese eels fed a vitamin C-deficient diet showed reduced growth after 10 weeks and hemorrhage in the head and fins after 14 weeks (Arai et al., 1972). Opacity of the cornea and kidney granulomatosis associated with hypertyrosinemia have been described as signs of vitamin C deficiency in turbot (Messager, 1986; Messager et al., 1986).

Phagocytic activity of cells of the immune system in fish produce reactive oxygen radicals that are potent microbicidal factors, but also autotoxic to fish macrophages (Secombes et al., 1988). Vitamin C appears to protect phagocytic cells and surrounding tissues from oxidative damage. An increased immune response due to high concentrations of vitamin C supplementation has been demonstrated in channel catfish (Durve and Lovell, 1982; Li and Lovell, 1985) and rainbow trout (Blazer and Wolke 1984b; Wahli et al., 1986; Navarre and Halver, 1989). However, Lall et al. (1990) observed no differences in humoral response and the complement system in Atlantic salmon fed diets containing 0 to 2,000 mg of vitamin C/kg after vaccination and subsequent live challenge with Aeromonas salmonicida and Vibrio anguillarum. Dietary and environmental contaminants, such as heavy metals (Yamamoto and Inoue, 1985) and chlorinated hydrocarbon pesticides (Mayer et al., 1978), increase the vitamin C requirements of fish.

Reproduction appears to increase maternal demands for vitamin C. Female tilapia fed vitamin C-free diets for 21 weeks produced eggs and fry containing no detectable ascorbic acid (Soliman et al., 1986b). Reduced reproductive performance has also been reported in rainbow trout fed vitamin C-deficient diets (Sandnes et al., 1984). Ascorbic acid reserves are rapidly depleted during embryonic (Sato et al., 1987) and larval development of certain fish (Dabrowski et al., 1988, 1989; Dabrowski, 1990), suggesting that requirements during early life stages may be higher than for fingerlings or adults.

Liver (Hilton et al., 1977; Sato et al., 1983) and kidney (Halver et al., 1969) ascorbic acid concentrations of less than 20 µg/g have been suggested as an indicator of vitamin C deficiency in salmonid fish. A similar value of less than 26 µg/g of liver has been suggested to indicate vitamin C deficiency in channel catfish (Lim and Lovell, 1978). A much higher value of 100 µg/g of kidney coincided with signs of vitamin C deficiency in snakehead (Mahajan and Agrawal, 1979).

Vertebral collagen levels have been shown to be a sensitive index of vitamin C status in channel catfish (Wilson and Poe, 1973; Lim and Lovell, 1978; El Naggar and Lovell, 1991) and rainbow trout (Sato et al., 1978).

Various derivatives of ascorbic acid, which are more stable than the parent compound, have been shown to provide antiscorbutic activity in fish. These include L-ascorbate-2-sulfate in rainbow trout (Halver et al., 1975; Grant et al., 1989), channel catfish (Murai et al., 1978; Brandt et al., 1985; Wilson et al., 1989), and tilapia (Soliman et al., 1986a); L-ascorbyl-2-monophosphate in channel catfish (Brandt et al., 1985; Lovell and El Naggar, 1990); and L-ascorbyl-2-polyphosphate in rainbow trout (Grant et al., 1989) and channel catfish (Wilson et al., 1989). Ascorbate-2-sulfate does not appear to be used as well as other more stable forms of ascorbic acid by certain fish (Murai et al., 1978; Soliman et al., 1986a; Dabrowski and Kock, 1989; Dabrowski et al., 1990), and in channel catfish it accounted for only 7 percent as much vitamin C activity as L-ascorbic acid or L-ascorbyl-2-monophosphate (Lovell and El Naggar, 1990).

Ascorbic acid is very labile and thus readily destroyed in the manufacturing process, especially in extruded feeds. Therefore it is not usually added to multivitamin premixes for fish feeds. Various coated forms of ascorbic acid, such as ethylcellulose or fat-coated products, have been used to increase retention of the vitamin in fish feeds. Nevertheless, approximately 50 percent of the supplemental ascorbic acid is destroyed during the manufacture of extruded catfish feeds (Lovell and Lim, 1978), and excess ascorbic acid is added to commercial formulations to ensure that an adequate concentration of the vitamin is retained during processing. Phosphorylated ascorbic acid, which is stable during extrusion processing (El Naggar and Lovell, 1991), is available for use in fish feeds but is presently relatively expensive. The form of the vitamin selected depends on how the fish feed is to be manufactured and how long it is to be stored before being fed to the fish. At present, it is still more economical to overfortify channel catfish feeds with the ethylcellulose coated product than to use the phosphate derivatives of ascorbic acid.