Diets and diet ingredients contain materials other than nutrients that may affect metabolism in a positive or negative way. These components may occur naturally in feedstuffs or may be added to meet physiological needs, to improve or preserve the quality of the diet, or to fulfill economic purposes. Included are such substances as water, hormones, antibiotics, fiber, pellet binding agents, synthetic antioxidants, and feeding stimulants.

Diets contain water as an added ingredient, a natural constituent of the feedstuffs, or an element adsorbed from the atmosphere. Moisture derived from the atmosphere is generally low, 6 to 10 percent by weight, and little significance is attributed to these low ''air-dry" concentrations in diets or ingredients. Low-moisture concentrations permit relative ease of storage and handling. In contrast, feeds with 12 percent or more moisture are generally more susceptible to spoilage by microorganisms.

Moist (10 to 40 percent) diets have been used at first feeding for Pacific salmon (Hublou et al., 1959; Hublou, 1963; Fowler and Burrows, 1971; Crawford et al., 1973) and Atlantic salmon (Lemm and Hendrix, 1981; Lemm, 1983) because these fish prefer moist feeds under hatchery conditions, particularly at cold-water temperatures. Thus, in recent years, commercial semimoist (15 to 20 percent) diets have been introduced that do not require refrigeration. To reduce moisture loss during processing and storage and to improve feed texture, polyhydric alcohols (propylene glycol, glycerol, and sorbitol, for example) are incorporated into these diets. The addition of compounds that inhibit molds is also required. Although many reports attribute the higher palatability of semimoist and moist diets to the higher moisture content (Ghittino, 1979), these claims have not been supported by experiments where proper comparisons were made. Studies on Atlantic salmon, chinook salmon, coho salmon, pink salmon, brown trout, and turbot fingerlings show no apparent benefit from the addition of water to dry diets (Poston, 1974; Bromley, 1980; Higgs et al., 1985; Hughes, 1989). Other factors such as pellet hardness or feeding strategy may be more important in improving feed intake than moisture.

Fiber refers to indigestible plant matter such as cellulose, hemicellulose, lignin, pentosans, and other complex carbohydrates found in feedstuffs. These components are indigestible unless bacterial action occurs within the digestive tract. Fish do not secrete cellulase (Lindsay and Harris, 1980; Bergot, 1981), therefore cellulose digestion does not play an important role in their nutrition.

Fiber provides physical bulk to the feed. Cellulose and hemicellulose have been used as diluents and binders in experimental fish diets. Dietary fiber improved gastric evacuation time of rainbow trout fed purified diets (Hilton et al., 1983). Buhler and Halver (1961) reported that small amounts of supplemental cellulose increased growth and the efficiency of protein utilization in laboratory diets. Most fish can tolerate up to 8 percent fiber in their diets, but higher concentrations (8 to 30 percent) depress growth (Buhler and Halver, 1961; Leary and Lovell, 1975; Edwards et al., 1977; Hilton et al., 1983; Poston, 1986). The poor performance of salmonids fed certain types of fiber may result from a combination of factors including poor digestion and faster gastric emptying rates, which in turn affect feed intake and utilization of nutrients (Davies, 1988). In natural ingredient diets that contain 3 to 5 percent fiber (derived mainly from plant ingredients), adding fiber is unlikely to have any measurable benefit. In most cases the concern is to formulate diets without excessive fiber content, which may reduce the

nutrient intake and increase fecal waste production. To limit environmental pollution from aquaculture waste, an important strategy is to use highly digestible feed ingredients and limit the fiber content of the diet.

Various natural and synthetic hormones have been evaluated in fish growth experiments, including growth hormone, thyroid hormones, gonadotropin (GnH), prolactin, insulin, and various steroids. Experimental feeding of synthetic androgens has enhanced growth and improved feed conversion in some species, especially in salmonids (Donaldson et al., 1979; Higgs et al., 1982; Matty, 1986). Approximately 20 fish species have shown anabolic responses to steroids (Donaldson et al., 1979; Matty, 1986). Some warm-water species, however, such as channel catfish (Gannam and Lovell, 1991a,b), have responded negatively to the feeding of androgens. Prolonged steroid treatment for growth acceleration may cause detrimental side-effects including early gonadal development, skeletal deformity, increased susceptibility to infections, and pathological changes in the liver, kidney, and digestive tract (Zohar, 1989; Gannam and Lovell, 1991a,b). None of these hormones has been approved by the U.S. Food and Drug Administration (FDA) for growth enhancement in food fish.

Hormones have been successfully used to induce or synchronize ovulation and the stimulation of spermiation. To increase gamete availability and fry production throughout the year, hormones have some application for the initiation and stimulation of oogenesis and spermatogenesis. The most commonly used preparations are pituitary extracts and human chorionic gonadotropin (Lam, 1982; Donaldson and Hunter, 1983). Failure of fish to release GnHs may be responsible for the lack of final oocyte maturation, ovulation, and spawning (Zohar, 1988). GnH-releasing hormones (GnRH) have been effective in inducing ovulation and spawning in salmonids, cyprinids, Indian catfish, winter flounder, plaice, grey mullet, milkfish, sea bass, red sea bream, sablefish, and herring (Donaldson and Hunter, 1983; Zohar, 1988).

Sex steroids have also been used to reverse the sex of some species of salmonids, carps, and tilapias. The objectives are to produce monosex and sterile fish of the faster growing sex, achieve better somatic growth, and prevent sexual maturation and the accompanying deterioration of flesh quality. This subject has been extensively reviewed by Hunter and Donaldson (1983). Feminization can be achieved by feeding estrogenic steroids (ethynylestradiol, esterone, and diethylstilbestrol) to tilapia fry and 17-β-estradiol to salmonid fry. Production of all-male populations of tilapia by feeding androgenic steroids to the fry is practiced in many countries. Generally, ethynyltestosterone or methyltestosterone (30 to 60 mg/kg of diet) is incorporated in the first feed of tilapia fry and fed for 14 to 21 days. This sex reversal method produces 90 to 100 percent male tilapia.

A wide range of antibiotics are used for therapeutic purposes in livestock production; however, only two compounds, sulfadimethoxine/ormetoprim and oxytetracycline, have been approved by the FDA for use in fish. Generally these compounds are stable during compression pellet processing and storage. Extrusion processing, however, will inactivate some of the oxytetracycline but very little of the sulfadimethoxine/ormetoprim. The quantity of antibiotic fed must be controlled, and proper feeding rate and withdrawal time must be practiced to reduce the entry of such compounds into the tissues of food fish or into the surrounding water. Antibiotics may only be added to feeds in the United States by a licensed manufacturer.

Subtherapeutic concentrations of antibiotics in the diets of poultry, swine, and other farm animals influence bacterial microflora of the gut and stimulate growth and feed efficiency (National Research Council, 1980). However, oxytetracycline and chlortetracycline in the diets of salmonid fish showed no appreciable benefit (Wolf, 1952; Herman, 1969). In red sea bream, however, administration of a subtherapeutic concentration (0.01 percent) of furazolidone, a nitrofuran derivative used against salmonellosis and protozoan diseases, improved growth and feed efficiency (Yone, 1968). Chemotherapeutic compounds may also be toxic when administered for an extended period or at high doses. Hicks and Geraci (1984) found that rainbow trout fed therapeutic concentrations of erythromycin (110 mg/kg daily) for 10 weeks showed vascular degeneration of proximal renal tubules. Prolonged sulfonamide therapy in salmonids caused growth retardation (Gutsell and Snieszko, 1949), renal tubular casts, focal hepatic necrosis, and visceral arterial sclerosis (Wood et al., 1957).

Antioxidants are commonly used in fish feeds that contain a high concentration of polyenic fatty acids to prevent the oxidation of lipids. Oxidative rancidity, or lipid peroxidation, affects the nutritional value of lipids, oxidation sensitive vitamins, and other feed components. The breakdown products of lipid peroxidation can react with the epsilon amino group of lysine and reduce its nutritional value. Natural tocopherols (vitamin E) have antioxidant activity; however, synthetic vitamin E is usually supplied in the diet in ester form, which has little antioxidant activity until it is hydrolyzed in the gut to the alcohol form. Thus, synthetic

vitamin E has little antioxidant activity in the diet. More vitamin E is required alone than when used in combination with antioxidants. Murai and Andrews (1974) showed that the commercial antioxidant ethoxyquin (1,2-dihydro-6-ethoxy-2,2,4-trimethyl-quinoline) could physiological replace vitamin E in channel catfish. Lipoid degeneration of liver and other pathologies have been associated with the feeding of rancid fat and/or the absence of vitamin E from the diet (Smith, 1979; Moccia et al., 1984; Tacon, 1985).

Synthetic antioxidants used in fish feeds are ethoxyquin, BHT (3,5-di-tert-butyl-4-hydroxytoluene), BHA (2(3)-tertbutyl-4-hydroxyanisole), and propyl gallate. Several excellent reviews of antioxidant types and mechanisms have been published (Uri, 1961; Stuckey, 1968; Porter, 1980). The maximum concentration of BHA and BHT permitted by the FDA is 0.02 percent of the fat content; for ethoxyquin, it is 150 mg/kg diet (21 C.F.R. § 573.380, 582.3169, 582.3173 [1987]).

Many plants and animals contain a variety of natural pigments that impart yellow, orange, and red colors to the flesh, skin, and eggs of fish. One of the most important groups of natural pigments in the plant and animal kingdom is the carotenoids. Fish and birds use oxygenated carotenoids (xanthophylls) for pigmentation of skin, flesh, and plumage. Fish cannot synthesize these pigments; therefore, they must be present in the diet. In salmonids, two oxycarotenoids, astaxanthin (3,3'-dihydroxy-4,4'-diketo-β-carotene) and canthaxanthin (4-4'-diketo-β-carotene) are responsible for the red to orange coloring of the flesh, skin, and fins. Astaxanthin is the main carotenoid pigment of wild salmonids, and is derived mainly from zooplankton. Feedstuffs of plant origin contain pigments that do not produce the desired color of salmon flesh. The major plant carotenoids are lutein (3,3'-dihydroxy-a-carotene) and zeaxanthin (3R,R'-β,β-carotene-3,3'-diol), as found in alfalfa, yellow corn, and algae. Lutein produces a yellow color whereas zeaxanthin imparts a yellow-orange color. Carotenoid concentration of some animal and plant sources are presented in Tables 2-1 and 2-2.

The retention of carotenoids in tissues depends on absorption, transport, metabolism, and excretion of these compounds (as reviewed by Torrissen et al., 1989). The digestibility of astaxanthin found in yeast and shrimp waste meal is low. However, ensiling of shrimp by-products improves the digestibility of astaxanthin by degrading the chitincalcium-protein-carotenoid complex in shrimp shells (Torrissen et al., 1981). Free astaxanthin is absorbed more efficiently than the astaxanthin ester (Torrissen and Braekkan, 1979; Schiedt et al., 1985). It appears that the rate of hydrolysis of the astaxanthin ester to free astaxanthin in the digestive tract of salmonids is limited. Approximately 90 percent of astaxanthin in fish flesh is located in free form, while the ester form predominates in skin. Salmonids are not able to oxygenate carotenoids, but deposit ingested oxygenated carotenoids without modification. Wide differences in the accumulation of carotenoids may be due to the differences in absorption of these compounds. Apparently, absorption is enhanced by the incorporation of hydroxyl groups into the carotene skeleton because astaxanthin is deposited at significantly higher concentrations than canthaxanthin in both Atlantic salmon and rainbow trout (Torrissen, 1986, 1989; Choubert and Storebakken, 1989). In salmonids, the absorption of astaxanthin and canthaxanthin is 10 to 20 times more efficient than lutein and zeaxanthin, while chickens absorb zeaxanthin at three times the rate of astaxanthin (Schiedt et al., 1985).

Yellow pigment deposition in the flesh of channel catfish, which is produced by zeaxanthin and lutein (Lee, 1987), is considered undesirable. Lee (1987) found that a concentration of 0.6 g carotenoid/g of flesh produced a distinguished yellow color of the fillet. A discernible concentration of carotenoid was deposited in catfish flesh from feeds containing 11 mg xanthophyll/kg.

Skin pigmentation is important in cultured yellowtail and red sea bream. These fish convert dietary astaxanthin largely into tunaxanthin and deposit it in their skin. Goldfish and fancy red carp are similar to the chicken in their absorption preference: zeaxanthin-astaxanthin-lutein. Hata and Hata (1972, 1973, 1976) showed that the yellow pigment, zeaxanthin, is readily metabolized to astaxanthin in goldfish and fancy red carp, which imparts red coloration. Goldfish metabolize little β-carotene and no lutein to astaxanthin (Hata and Hata, 1972).

The function of carotenoids other than as precursors of vitamin A in fish is not well defined and mostly speculative (Tacon, 1981). Although the mobilization of carotenoids from the flesh to skin and ovaries of salmonids during maturation is well documented, their role in reproduction is not clear. Schiedt et al. (1985) reported a biological function of astaxanthin, canthaxanthin, and zeaxanthin as vitamin A (retinol and 3,4-dihydroretinol) precursors for vitamin A-depleted rainbow trout.

Binders are incorporated into fish feeds to improve stability in water, increase pellet firmness, and reduce the amount of fines produced during processing and handling. Among the most widely used binders are sodium and calcium bentonites, lignosulfonates, hemicellulose, carboxymethylcellulose, alginate, and guar gum. More recently, some inert polymeric binders have been introduced, but limited information is available on their composition or toxicity to commonly cultured fish. Cereal grains provide starch that, when gelatinized, gives a durable, water-stable pellet. Certain feed ingredients such as whey, wheat gluten, pregelatinized starches, and molasses will permit the production of good-quality pellets. Most binders are considered to be inert and have limited or no nutritional value. However, incorporation of alginate and guar gum in rainbow trout diets reduced feed intake, increased moisture content of feces, and lowered the digestibility of protein and lipids (Storebakken, 1985). Wood et al. (1954) showed that carboxymethylcellulose at 2 percent in the diet of trout caused no growth depression.

The primary modes of feed detection by fish are through olfaction or sight, but the taste of the item is the key factor in determining whether the item is swallowed or rejected (Adron and Mackie, 1978). There appears to be a well-defined and species-specific tuning of the taste receptors of fish for the particular cues present in their feed items (Goh and Tamura, 1980). Many researchers and feed manufacturers have attempted to add substances to fish feeds to enhance palatability and feed acceptance. This focus has taken particular importance in the production of larval and starter feeds, where feed acceptability is a major concern.

Carr (1982) identified four major characteristics of feeding stimulants for fish that were derived from animal tissues: (1) they have a low molecular weight (<1,000), (2) they contain nitrogen, (3) they are nonvolatile and water-soluble, and (4) they are amphoteric (have both acid and base properties simultaneously). Several substances or groups of substances for which these generalizations apply, such as amino acids, betaine, and inosine, have improved feeding behavior in carnivorous and omnivorous species (as reviewed by Atema [1980], Carr [1982], Mackie [1982], Adams and Johnsen [1986a], Rumsey [1986]). Harada (1989) has shown that some dipeptides elicit a greater feeding response than either of the constituent amino acids presented alone for abalone. Few data exist on feeding stimulants for herbivorous species, but in four studies using Zillii's tilapia (Adams and Johnsen, 1986a,b; Johnsen and Adams, 1986; Adams et al., 1988), organic acids along with certain amino acids were found to be stimulatory. Feeding was stimulated by the organic acid, dimethyl-β-propiothetin, in goldfish, common carp, and tilapia (Nakajima et al., 1989).

When data on the effectiveness of the various feeding stimulants containing amino nitrogen are considered, a pattern seems to emerge relating to the feeding behavior of the fish and the type of compounds that are stimulatory. In general, carnivores show the greatest positive response to alkaline and neutral substances, such as glycine, proline, taurine, valine, and betaine, while herbivores respond more

to acidic substances, such as aspartic acid and glutamic acid. The pattern is reasonable considering the chemical characteristics of the feed items the fish would seek in the natural environment (Mackie, 1982; Adams and Johnsen, 1986a).

The presence of certain compounds can also act as feeding deterrents. This phenomenon has been shown to occur with certain combinations of amino acids (Adron and Mackie, 1978; Mackie and Adron, 1978; Mackie, 1982). Trimethylamine or its oxidation products, which are produced in decaying fish flesh, were shown to cause a decrease in feed consumption in turbot (Mackie and Adron, 1978), plaice (Mackie, 1982), and chinook salmon (Hughes, 1991) when these compounds were added to the diet. Salmonids (Hung and Slinger, 1980; Ketola et al., 1989) and yellowtail (Murai et al., 1988) show aversion to highly oxidized oils and fishmeals.