Nutrition, Energetics and Growth of Fish:

Current Challenges and Approaches

C. Young Cho^1,2 and Dominique P. Bureau²

1. Fish Nutrition Research Laboratory Ontario Ministry of Natural Resources Guelph, Ontario, Canada, N1G 2W1
2Department of Animal and Poultry Science University of Guelph, Guelph, Ontario Canada, N1G 2W1

Perspective

In culturing fish in captivity, nothing is more important than sound nutrition and adequate feeding. The cost of feeding is the major cost in fish culture. The production of cost-effective nutritionally balanced diets for fish requires efforts in research, quality control, and biological evaluation.

The nutrition of fish has been studied for more than 50 years, and has received generous funding by governments and other organisations worldwide. As a result, tangible progress has been made in defining the nutrient requirements of fish, improving feed manufacturing practices, and in establishing improved feeding practices. At the same time it must be said that the field of fish nutrition is in chaos and our real knowledge of fish nutrition is much less than it should be as we enter the 21st Century.

Unfortunately, high standards of technical and intellectual rigour, which are the norm in other fields of nutritional science, continue to be found only rarely in fish nutrition research. In fact, the field continues to be dominated by some of the poorest quality research ever published in the field of nutritional science. It is most unfortunate, for example, that the U.S. NRC Sub-Committee on Fish Nutrition continues to propagate the untenable notion of interspecies differences in nutrient requirements among salmonids in the face of rigorous evidence that it is most unlikely. Likewise, it is simply unacceptable that this committee perpetuates the confusion among fish nutritionists over the very simple distinction between nutrient requirements and recommended levels and continues its failure to understand the need to consider energy requirements in order to quantify nutrient requirements in a meaningful way. Such ignorance is not permitted to prevail in any other area of nutritional science, but it has been dominant in fish nutrition for at least 40 years. The greatest challenge for fish nutrition as we enter the 21st Century is to effect an about-face to become driven by scientific rigour rather than by unquestioned acceptance of research claims that amount to little more than folklore. To this end, the greatest need is for a critical mass of fish nutritionists with both bona fide expertise in nutritional science and the strength of character to think and act with independently.

Since 1969, the University of Guelph Fish Nutrition Research Laboratory has carried out research in the field of nutrient requirements, digestibility, bioenergetics, physiology, feed formulation, feed manufacturing practices, feeding systems and waste management. This research always attempted to take full advantage of current "at large" knowledge of nutrition and physiology. It has also put in practice what it preached by using available scientific information when formulating cost-effective open feed formulae used by all the governmental fish culture stations in Ontario for the past 20 years and by developing and using rational bioenergetic, feed requirement and waste output models. The Fish Nutrition Research Laboratory has, therefore, been a significant contributor to the establishment of scientific fish nutrition and to the transfer of technology to the feed and aquaculture industries. As such, it was invited to discuss here its views of current challenges in fish nutrition and present rational approaches to meet these challenges.

Note: Perspective was prepared in consultation with Dr. W.D. (Bill) Woodward, Department of Human Biology and Nutritional Sciences, University of Guelph.

Challenges & Approaches

There are numerous challenges for fish nutritionists, feed manufacturers and fish culturist as we enter the new millennium. One of the challenge is to develop less wasteful, cost-effective, diets based on use of economical by-products and feedstuffs which are less suitable for human consumption. Another is to better understand and describe growth and nutrient and energy utilization and make wise use of this knowledge when planning research, developing diets and feeding systems, and managing aquaculture operations.

1. Less Wasteful, Less Fish Meal-Dependent, Feeds
Fish and shrimp feeds are the most expensive type of animal feed on the market. This is due to the fact that fish and shrimp feeds are, in general, of high nutrient density (high protein, high energy), composed of high quality, nutrient-dense, highly digestible ingredients and produced using costly manufacturing processes (steam pelleting, extrusion). Part of this high cost can also be attributed to formulation to excessive nutrient levels (excessive safety margins) and liberal use of various luxurious or "magic" ingredients (krill, protein hydrolyzate, yeast, betaine, attractants, immunostimulants, vitamin mega-doses, amino acid supplements), the need for which being highly questionable.

Good quality fish meal has been and remains a mandatory component of successful salmonid feeds. Fish meal and other protein-rich fisheries products currently makes up anywhere between 30 to 75% of most commercial salmonid, marine fish and shrimp feeds. Well-conducted, systematic research has shown these high fish meal incorporation levels are excessive and could be reduced considerably. Nevertheless, distrust in any protein sources that is not of fish or marine origin is widespread among aquaculturists, feed manufacturers, and even nutritionists. The reasons for this distrust are not always rational. For example, it is still widely believed in the industry that fish, like salmon and high values marine fish species must eat fish to adequately meet their nutritional requirements. The questionable quality or usefulness of a large proportion of the research on the topic and the reluctance of fish nutritionists themselves to use lower fish meal levels in their own experimental feeds are also partly responsible for this distrust and widespread use of high fish meal levels in commercial feeds.

The result is a majority of feeds being wasteful (overly luxurious), more expensive and often more polluting than they need to be. Salmon, trout, marine fish and shrimp aquaculture, the beneficiaries of most research efforts, are still merely converting fish to fish. Aquaculture which produces a kilogram of farmed salmonid fish or shrimp from 3-4 kg of high quality herring, capelin, sardine or anchovy should be a thing of the past. Changing this situation represents a significant challenge. This can only be done by generating reliable information on nutrient requirements and ingredient characteristics (composition, digestibility, limiting factors), conducting rigorous studies, and preaching by good example.

1.1 Formulating with reasonable safety margins
There is a need to move from ‘folkloric’ to scientific formulation of feeds. Precise formulation based on current estimates of nutrient requirements is practised by the poultry and swine feed industry worldwide. The aquafeed industry has been slow to adopt this kind of practice and feeds are too often formulated to nutrient levels that are much too high to be justified.

In the case of vitamins the problem of over-fortification is most widespread. Old NRC (1981) recommendations still appear to be the base for the information distributed by vitamin manufacturers and a large proportion of the vitamin packages used in experimental feeds, and possibly, commercial feeds. NRC (1981) recommendations were based on early vitamin requirements studies with Pacific salmon. Woodward (1994) in a critical review of the evidence pointed out that the water-soluble nature of the purified diet (H440, Halver, 1989) used in early studies on vitamin requirements of salmonids probably lead to extensive leaching of water-soluble vitamins and overestimation of requirements. In fact, the diet was not designed for quantification of nutrient requirements (Halver, 1966). An improved purified diet (Guelph purified C101; Cho and Cowey, 1991) was devised in the mid-1970s by our laboratory. This diet was shown to support good growth rate and was water-stable and, therefore, much less subject to leaching than the H440 diet. Studies on water-soluble vitamin requirement of rainbow trout fed this water-stable purified diet and showing high growth rates have shown that the vitamin requirement estimates generated by the early studies with salmonids, the base for NRC (1981) recommendations, were, indeed, grossly overestimated (Woodward, 1994). In addition to the Guelph purified C101, a number of other diet formulae have been developed and successfully used by our laboratory for studying the requirement of nutrient, including arginine (Cho et al., 1992), leucine (Choo et al., 1991), and methionine (Cowey et al., 1992).

It is frequently argued that the NRC (1981) vitamin recommendations contained safety margins and that under practical conditions, the overestimation of requirements by NRC (1981) may not be so significant when one accounts for occasional extreme losses during processing, storage, or for possible differences in vitamin requirements of fish reared under "secure" experimental conditions and those of fish reared under a "stressful" commercial environment (Hardy, 1999). This argument is not helpful in that it is an attempt to defend recommendations that are based on a flawed dietary model. This is risky and defeats the purpose of scientific research. Moreover, it propagates the belief that estimates of nutrient requirements generated under controlled laboratory conditions have nothing to do with reality, which is false.

The critical review of Woodward (1994) was echoed in the vitamin requirement values proposed by NRC (1993) for rainbow trout (Table 1) but not in those for Pacific salmon, which is a scientific aberration. Kaushik et al. (1998) showed that supplementation of practical diets for rainbow trout with a vitamin premix containing the minimal requirement levels of the NRC (1993) was sufficient to support rapid and maximal weight gain of rainbow trout, chinook salmon and European seabass fed practical diets. These authors suggested that a safety margin of about 50% above NRC (1993) vitamin requirement values was sufficient under most circumstances to meet the requirements of the fish. It is also our experience that this safety margin is sufficient for fast growing rainbow trout (thermal-unit growth coefficient (TGC)= 0.240) reared at the Alma Aquaculture Research Station (University of Guelph) under conditions that are fairly representative of those found on commercial operations.

Establishing requirement values for every single essential nutrient for the dozens of fish species cultured around the world under a large variety of environmental conditions would be, to say the least, extremely costly and probably a waste of research efforts. The solution lies is approaching the problem rationally. Woodward (1994) suggested that based on the similarity in the biochemical machinery used to sustain intermediary metabolism, differences in dietary vitamin requirement levels among fish are probably the exception rather than the rule. Experimental evidence presented by Kaushik et al. (1998) support this hypothesis. The same is probably true in terms of amino acid requirements (Cowey, 1994). From this perspective, the approach of Kaushik et al. (1998) is probably the most efficient. Briefly, this approach is to feed diets containing various levels of a nutrient package (package designed based on critical evaluation of the nutritional information available) to various species reared under different conditions.

Table 1. Some vitamin B requirements (mg/kg diet) for maximal weight gain of young rainbow trout (Oncorhynchus mykiss) proposed by the National Research Council (1993) in comparison with the National Research Council (1981) recommended levels and with those of chick and piglet.

	Rainbow trout NRC 1993	Fish NRC 1981	Chick NRC 1984	Piglet NRC 1988
Riboflavin	4	20	3.6	3.0
Pyridoxine	3	10	3.0	1.5
Biotin	0.15	10	0.15	0.10
Pantothenic acid	20	40	10	13
Thiamine	1	10	1.8	1.3
Vitamin B12	0.01	0.02	0.009	0.02
Folic acid	1	5	0.55	0.30

1.2 Relying Less on Fish Meal

Good quality fish meal remains a mandatory component of successful commercial salmonid feeds. General distrust in any protein sources that is not of fish or marine origin is still common and the result is widespread use of high levels of fish meal in commercial feeds. This situation does not make any sense from an economical point of view. Moreover, it is difficult to defend the use of 3 to 4 kg of perfectly good fish products (herring, anchovy, sardine, etc.) to produce on 1 kg of another fish products (salmon, shrimp). The use of more economical protein sources, especially those used at very wide scale by the other animal feed industries makes more sense and may help insure the long term economical sustainability of aquaculture.

Good quality fish meal made from whole fish, is an ingredient that has several nutritional qualities. It is rich in highly digestible protein (if it has been carefully processed) of excellent quality since it contains high levels of most essential amino acids in proportions that resembled the requirement of fish. Fish meal also contains numerous essential or conditionally essential nutrients, such polyunsaturated fatty acids, minerals, vitamins, vitamin-like compounds (choline, inositol), phospholipids, cholesterol, etc. Finally, it possess excellent organoleptic properties for most fish species.

All these factors must be taken into account when increasing level of other protein sources at the expense of fish meal in the diet. Nutrient deficiencies, lower digestible nutrient contents, suboptimal amino acid balance, excessive levels of antinutritional factors or bulk agents (fibre, ash), or lower palatability are factors that can explain the decrease in performance of the fish fed lower fish meal diets in several studies. Properly conducted studies using nutritionally complete, balanced and palatable experimental diets with very low levels, or no fish meal at all, have shown that the problem is only one of compensation for all the factors described above when replacing fish meal with other protein sources. Economical salmonid feed formula based on economical protein sources (soybean meal, corn gluten meal, poultry by-product meal, blood meal) and 20% fish meal or less were shown to support as good growth performance as diet with much higher fish meal levels (Table 2, 3, and 4). One such formulae, the MNR-89G formula has been used with much success for several years by the Ontario Ministry of Natural Resources fish culture stations as well as many fish producers.

The popular quest for the formulation of successful fish meal-free diets may not be more rational than the other extreme of using excessive levels of fish meal in the feed. The formulation of successful fish meal free feed requires the use of highly processed and expensive protein ingredients (soy protein concentrate, canola protein concentrate, spray-dried blood meal, krill meal) with very low antinutrient contents and good amino acid profile, and various supplements (amino acids, feed stimulants) to improve nutritional balance and palatability of the feed (Kaushik et al., 1995; Higgs et al., 1996; Watanabe et al., 1997). While able to support good growth performance, these diets are more expensive to produce than practical diets containing fish meal because of the prohibitive cost of some of the ingredients required. The formulation of fish meal-free fish feeds is a feat in itself as it shows that "fish don’t absolutely need to eat fish". However, it does little for solving the problem of the high cost of salmonid feeds. Maintenance of a small amount of fish meal (10-20%) in salmonid feeds appears to make economical and practical sense at this point in time. It is more rational, for both fish nutritionists and feed manufacturers, to gradually reduce fish meal levels in feeds, become gradually more comfortable with a riskier feed formulation process, slowly learn by trial and error, than attempt to move straight away from high fish meal to fish meal-free feeds.

Focused and systematic research and effective transfer of knowledge are essential to meeting the challenge of lesser reliance on fish meal by the aquaculture industry. Researchers must reliably determine nutrient requirements and other factors important to consider when formulating fish feeds (nutrient density of diet, palatability, tolerance to anti-nutritional factors, fibre level, etc.). Better knowledge of the composition, availability of nutrients, variability, and limitations of feed ingredients is equally essential. Preaching by example is by far the most effective mean of technology transfer. Research trials should, therefore, be conducted with the most rigorous manner so that the results are credible. Feeds used in these trials should also be formulated making full use of available scientific information. Growth performance should exceed a certain yardstick which should ideally be established based on the genetic potential of the fish stock used. In that perspective, sustained use of reliable growth model, such as the thermal-unit growth coefficient (TGC, Iwama and Tauz, 1981; Cho, 1992) allowing meaningful comparisons among trials, fish size, and water temperature, is most helpful.

Table 2. Ingredient composition of the diets used in a feeding trial comparing various economical salmonid diet formulae against a high fish meal formula.

Ingredients	Fish Meal Based	C203	MNR-89G	MNR-95HG
Fish meal, anchovy, 68% CP	60	30	20	18
Corn gluten meal, 60% CP		13	17	49
Soybean meal, 48% CP		17	12
Blood meal, ring-dried			9
Brewer’ dried yeast				6
Wheat middlings	16	23	18	0
Whey	7	-	7	10
Vitamin & mineral premixes	3	3	3	3
Fish oil	14	14	14	14
Total	100	100	100	100

Table 3. Growth performance of the rainbow trout (Spring-spawning strain) fed the various salmonid diet formulae over a 16-week period.

Description	IBW/100 g	FBW/100 g	Gain/100 g	Feed/100 g	FE gain:feed	TGC %
FM	1738	10598 a	8860	8999	0.984 a	0.127
C203	1724	9879 a	8155	8911	0.915 ab	0.121
MNR-89G	1731	10164 a	8433	8977	0.939 ab	0.124
MNR95HG	1751	8407 b	6656	7858	0.847 b	0.106

Values in a same column not sharing the same subscript are not significantly different (P>0.05) according to Tukey's Honestly Significant Different Test.

Legends:

IBW/100	Average initial body weight per 100 fish (g)
FBW/100	Average final body weight per 100 fish (g)
Gain/100	Total weight gain per 100 fish
Feed/100	Feed served (fish fed to near-satiety, 3 meals/d) per 100 fish
FE	Feed efficiency
TGC	Thermal growth coefficient (Cho, 1992)
	Formula: FBW1/3-IBW1/3                    S(T(°C)*days)

Table 4. Composition of the grower formulae used by the OMNR Fish Culture Stations over the past 10 years.

	Formulae
Ingredients	MNR89G	MNR91H	MNR95HG	MNR98HG
	%
Fish meal, herring, 68% CP	20	35	18	18
Blood meal, spray-dried, 80% CP	9	9	-	-
Corn gluten meal, 60% CP	17	15	49	37.6
Soybean meal, 48% CP	12	14	-	-
Poultry meal, 68% CP	-	-	-	13
Brewer’s dried yeast, 45% CP	-	-	6	-
Wheat middlings, 17% CP	20	-	-	-
Whey, 12% CP	8	10	11	9
Vitamin premix	0.5	0.5	1	0.5
Mineral premix	0.5	0.5	1	0.5
L-Lysine	-	-	-	1.4
Fish oil	13	16	14	20
Digestible Composition
Digestible protein, %	37	44	44	42
Digestible energy, MJ/kg	17	20	20	21
DP/DE, g/MJ	22	22	22	20

1.2.1 Proper Knowledge of Nutritive Value of Ingredient

The first consideration for formulation and production of cost-effective diets is the quality of the feed ingredients. Diets produced with poor quality raw materials and under adverse processing conditions have inferior nutritive value and adverse effects on fish health. Quality criteria for the ingredients must be respected to insure that the final product is of consistent quality and that deleterious effects are avoided. The chemical composition (nutrient, energy, antinutrients, contaminants) of the ingredient obviously plays a determinant role on quality. However, biological aspects, such as digestibility and utilization of nutrients are most important and often overlooked.

The loss of undigestible matter from the diet as feces is the primary reason for variation in the nutritional value of feed ingredients. Measurement of digestibility provides, in general, a good indication of the availability of energy and nutrients, thus providing a rational basis upon which diets can be formulated to meet specific standards of available nutrient levels (Gomes da Silva and Oliva-Teles, 1998). Formulation of isoproteic and isoenergetic diets of a crude protein or gross energy basis is still widely practised by fish nutritionists and this must become a thing of the past! Differences in digestible nutrient contents of experimental diets often explain the difference in performance of fish fed different diets in many studies. It is inappropriate to compare the value of a protein source vs. that of fish meal if this comparison is not done on a levelled playing field (digestible nutrient basis).

When examining the nutritive value of an ingredients, a rational approach is to first conduct a digestibility trial to estimate digestible nutrient composition of the ingredient. A digestibility trial may also allow the investigator to get insights as to the acceptability of this ingredient for the target fish species. Secondly, growth trial(s) should be conducted using diet formulated with various levels ingredient under investigation.

1.2.2 Complementarity of ingredients

It is common in studies to attempt to formulate diets relying on a single protein source. In recent years, the production of "all soy" or "all canola" protein feed has been a goal of many studies. In practice, relying on a single protein source may not be economically sound and is risky. Unexpected variation in the quality of the protein source used may result in grave consequences in terms of growth and health of the fish. The use of two or three protein sources may damper the potential negative impact of such occurrence. Composition in individual amino acids is variable among protein sources. Individual essential amino acid levels in a specific different protein sources may be in excess or below what is required by the fish. Different protein sources may be complementary, meaning that the composition of a mixture of these protein source may closely match or exceed the amino acid requirement of the fish. Imperfect protein sources, supporting only suboptimal performance when use as major protein source, may be combined with other economical imperfect protein sources to produce a mixture that support optimal performance.

Two protein sources with complementary amino acid profiles are corn gluten meal and soybean meal. Recent results from our laboratory showed that corn gluten meal and soybean meal complement, indeed, each other very well nutritionally for Atlantic salmon. A similar observation was also made by Pongmaneerat and Watanabe (1992) for rainbow trout. This complementary is also exploited very extensively by other animal feed industries. Corn and soybean meal make up the bulk of poultry and swine feed produced world wide (Parson, 1998). Corn gluten meal is a highly palatable protein source for fish and it does not contain anti-nutritional factors. Diets containing high levels of corn gluten meal may be marginally deficient in lysine and incorporation of a certain amount of soybean at the expense of corn gluten meal may alleviate this lysine deficiency. Alternatively, high corn gluten meal feed can be supplemented with L-Lysine, an economical crystalline amino acid commonly used in poultry and swine feeds. The incorporation of corn gluten meal has been limited in feeds for food fish production because of fear that its high concentration in xanthophylls could produce undesirable pigmentation of the flesh or decrease pigmentation efficiency as these pigments may compete with expensive synthetic pigment added in the feed. Recent experimental evidence from our laboratory does not support this hypothesis. Feeding diets containing between 22 to 34% corn gluten meal and 50 PPM astaxanthin to rainbow trout growing from 270 to 950 g in 24 weeks resulted in pigmentation that well exceeded minimum pigmentation levels required of rainbow trout fillets in North America (acceptability of pigmentation of the fillets was assessed by commercial fish processors).

There are numerous other economical feed ingredients whose complementarity can be exploited. In a recent trial in our laboratory showed that poultry by-product meal, feather meal and blood meal complemented each other well and that feeds with 2/3 digestible protein provided by a mixture of these ingredient and only 20% fish meal performed as well as fish fed diets containing fish meal, soybean meal and blood meal as the major protein sources.

1.2.3 Don’t reinvent the wheel: Understand the basis. The example of soybean meal

A major problem in fish nutrition research is the endless repetition of studies on the same topics. Slight variations of dietary design, use of different fish species, distrust of results from other laboratory, or simple ignorance of the literature are generally what justify these repetitions. A typical example is that of research on soybean meal. In the 1970’s, soybean meal and other soy products (full-fat soybean) were considered new alternative protein sources for fish feeds and their nutritive value and their potential as protein sources for fish feeds was examined in numerous studies (Cho et al., 1974, Spinelli et al., 1979; Fowler, 1980). More than twenty-five years have passed and soybean meal is still being presented as a novel protein source for fish feeds! Studies closely resembling those conducted many years ago are still being published as "new" information.

A good review of the literature indicates that soybean meal is a good quality, highly digestible protein sources for fish. However, incorporation levels exceeding 25-30% in the diet of young salmonid result in growth depression which can partly be explained by a reduction in feed intake (Pongmaneerat and Watanabe, 1992). The reason for the decrease in performance and feed intake of the fish fed high levels of soybean still remain very much a mystery despite the enormous of amount of research effort devoted to evaluating the potential of soy products as protein sources for many fish species. In recent years, the focus of most studies has been the comparison of soy ingredients produced with different techniques or processed under different conditions. There has been much speculations, each author blaming a different fraction of soy for the decrease in performance of the fish. Trypsin inhibitors, oligosaccharides, phytoestrogens, allergens, lectins, phytate are all factors that have been suggested as having potential negative effects of performance of fish. Much of the speculation has been based on simple comparison of results between types of ingredients differing very significantly in term of chemical composition. For example, aqueous ethanol washing and high heat treatment used in the production of most soy protein concentrates remove a large number of chemical compounds, such as oligosaccharides, isoflavones, soyasaponins, lectins, allergens, etc. It is, therefore, impossible to conclude, that improved performance observed when fish are fed soy protein concentrates compared to soybean meal is due to removal of this or that compound. It is difficult to justify so much speculations when most of the compounds (trypsin inhibitors, oligosaccharides (raffinose, stachyose), lectins, phytoestrogens (isoflavones), putative allergens (glycinin, b -conglycinin), etc.) identified as possibly responsible for the decreased performance of the fish can be purchased in a pure form from many suppliers.

It is essential to move from simply descriptive approaches or feed formulae testing to more explanatory (mechanistic) ones to improve efficiency of the effort invested in fish nutrition research. A noteworthy example of mechanistic studies with practical implications is the work of Geurden et al. (1995, 1998) on phospholipid nutrition of fish larvae. Using simple dietary treatments these authors determined the basis of the essentiality of phospholipid for common carp larvae. These studies show that conducting mechanistic studies does not always implies fancy analyses or experimental manipulations.

A classical nutrition approach used to identify essential or, conversely, deleterious factors, is to isolate various fractions from the ingredient studied. This approach has been used for the discovery of vitamins and other essential nutrients. It was recently used in our laboratory (Bureau et al., 1998) in an attempt to identify component(s) of soy affecting the performance of salmonid fish based on preliminary work by van den Ingh (1996). Extracts were prepared from soybean meal and soy protein isolate by a series of sequential solvent extraction process aiming the isolation of soyasaponins, a fraction suspected to be responsible for the negative effects of high dietary levels of soy on fish based on examination of results from trials comparing different soy ingredients. The two studies showed that the extracts produced contained a potent feeding deterrent for chinook salmon and possibly also rainbow trout. Preliminary observations on chinook salmon and concurrent work with the same extracts in a mammalian species (Philbrick et al., 1999) suggest that the mechanism is not related to an effect on palatability but rather to other biological effects. More work is required to isolate, identify and study the effect of individual components of the extracts. However, the components with negative effects on performance of salmonids has been pinpointed to a small fraction of soy and this will greatly simplify further investigations.

In this search for the understanding, investigators should not go to extreme and lose sight of the potential usefulness of the information generated. Many studies focus on metabolic and physiological parameters (e.g. plasma amino acids) using experimental conditions (fish fasted or fed diets of undefined composition, fish restrained from swimming, etc.) which have no or little relevance to the conditions for which the information may be useful. In addition, the usefulness to investigators, feed manufacturers or aquaculture producers of some of the information gathered at great cost is also not always clear. Investigation of these very specific parameters too often take precedence over basic understanding of nutrient utilization (digestibility, nutrient budget at various intakes) and this is not wise.

2. Understanding Growth and Nutrient Utilization

Basic understanding of growth and nutrient utilization by fish is definitely one of the area for which we know much less than what we should as we enter the 21^st Century. Hundreds of studies have focused on diet composition (% nutrient in diet) but few have rationally look at quantitative aspect of nutrient utilization (g nutrient/unit of weight gain). Moreover, it is very difficult to go back to published studies and examine nutrient utilization a posteriori because the inadequate experimental approaches used in many studies. Inadequate feeding practices, such as feeding fish at a certain fixed feeding levels regardless of whether it correspond to the need of the fish or whether feed wastage is occurring or not, lack of definition of the digestible nutrient composition of the experimental diets, and failure to determine the chemical composition of the experimental animals, are all too common practices in fish nutrition studies.

An area of considerable controversy is the effect of feed or nutrient intake level on live weight gain, feed efficiency and nutrients deposition of fish (Azevedo et al., 1998). Despite decades of research, opinions and hypotheses are plentiful but little solid experimental data are available. Studies have suggested that maximum feed efficiency is achieved at feeding levels below that required for maximum growth (G_max) (Brett, 1979). Conversely, it has been suggested that maximum feed efficiency is attained at maximum feed intake (R_max) and G_max (Talbot, 1993). The use of physical, relative or semi-quantitative parameters, such as weight of feed, G_max, R_max and feed efficiency, inappropriate growth models, such as specific growth rate (SGR), and dubious methodological approaches, such as the gastric emptying rate, radiographic assessment of feed intake of individual fish, are all factors that contributes to the ongoing controversy and lack of tangible progress over the past 40 years. More pragmatic approaches must be used if a better understanding the dynamic of growth and nutrients deposition in fish and, consequently, feed and nutrient utilization is to be achieved.

Better understanding of growth and nutrient utilization by the animal can only be achieved through use of methodological approaches based on nutrient and energy intake, catabolism, and retention. This type of approach has been used in domestic animals for decades and has been adopted for a limited number of fish nutrition studies (e.g. Cho et al., 1982; Meyer-Burgorff et al., 1989a &b; Cho and Kaushik, 1990; Schwartz and Kirchgessner, 1995; Azevedo et al., 1998; Lupatsch et al., 1998; Ohta and Watanabe, 1998; Rodehutscord and Pfeffer, 1999). These studies have generated valuable information (e.g. Figure 1) which allows us to progressively better understand of growth and nutrient utilization processes of fish. The definition of the energy requirement and nutrient utilization of the fish species cultured around the world must become a research priority for fish nutritionists.

Legends:
RE Recovered energy (carcass)
ME Metabolizable energy intake
HeE Estimate of basal metabolism

The reduction of feeding cost is the aim of a large proportion of the research efforts devoted to aquaculture. There is often confusion between low-cost feeds and low feeding costs or cost-effective feeds. Minimizing feeding cost does not necessarily implies minimizing the cost of the feed per unit of weight but rather minimizing the cost of supplying the amount of nutrient required per kg weight gain,. This cannot be truthfully achieved unless attention is paid to the amount of feed consumed by the animal, and the digestible nutrient composition of the feed. Comparison between feeds with different protein contents, for example, cannot meaningful made unless digestible protein and energy content of the feed is assessed and that the amount feed consumed by the fish is known with reasonable accuracy. Certain fish species, e.g. channel catfish, may perform well when fed feeds with low protein content (e.g. 26% digestible protein), but may need to consume a greater amount of a lower protein feed than a feed with higher protein content (e.g. 32% digestible protein) to achieve the same growth (e.g. Lim and Klesius, 1998). Feed with lower protein content may be less expensive per unit of weight but total feed cost may be greater with that type of feed since a greater amount of that feed is needed to achieve the same performance.

A better understanding of nutrient and energy utilization may allow fish nutritionists, feed manufacturers and fish producers to look at feed cost under a new light. When presented with accurate nutrient and energy utilization data, one may reconsider, for example, the use of low nutrient and energy density feeds (low cost feeds but not necessarily cost-effective feed) for the rearing of warm water omnivorous fish (channel catfish, tilapia, carp). Based on current knowledge on energy requirement and nutrient utilization, it is clear that the use of low nutrient and energy density feeds are the main reason for the very poor feed conversion ratio (feed/gain, between 1.5 and 3) seen for most studies with this type of fish. Production cost with such feeds may not be so advantageous as often touted when one accounts for amount of feed needed per kg weight gain, feed manufacturing (e.g. extrusion) and transport costs, and potential negative impact on the productive capacity of rearing environment of the high of organic waste output associated with feeding low digestible nutrient density feed.

It is imperative to move from a narrow vision of nutrient requirement or utilization as "% of the diet" to a more integral vision which takes into account diet composition (e.g. digestible protein to digestible energy ratio, digestible nutrient and energy levels), nutrient and energy requirement per unit of live weight, nutrient or energy gain. The challenges are numerous, the task very significant but tangible progress can rapidly be made if investigators adopt effective methodological approaches.

3. Making Good Use of Scientific Information : Feed Requirement and Waste Management Models.

Greater understanding of growth and nutrient utilization processes should ideally be translated into concrete applications so that the whole aquaculture can benefit from it. Optimization of the composition of practical feed formulae to improve cost-effectiveness is one type of technology transfer to the fish producer. Another type is the development of practical models.

Practical models have been developed at the Fish Nutrition Research Laboratory to rationally predict feed requirement and waste output of salmonids in a situation or farm-specific manner (Cho, 1992; Cho and Bureau, 1998). The models based on the principle that feed requirement is generally governed by how much protein, fat and minerals the animal deposit in its body and the biological cost of depositing these body components and that waste output is determine by digestibility of the feed used and the efficiency of nutrient utilization. Calculation of feed requirement and waste output using these models involves five steps:

1) DIET Selection

The amount of feed required by a fish depends firstly on the composition of the feed used. In general, a greater amount will be required of a lower nutrient density feed than a higher nutrient density feed to achieve the same performance level. Composition of the feed also affect the composition of the fish produced which, in turn, also affects the amount of feed required. Composition of the feed is the main determinant of digestibility and nutrient retention by fish and, therefore, waste output.

2) GROWTH Prediction

The accurate prediction of the growth of the animal over the period for which the feed requirement is calculated is probably the most critical factor for the accurate prediction of feed requirements and waste output. Growth involves the deposition of nutrients (accretion of body components) which is the main factor determining feed requirement. Fish growing at different rates will, therefore, have different feed requirements. Production records are very valuable starting points when trying to predict the growth of the fish for which one wants to calculate ration allowance. A growth model, the thermal-unit growth coefficient (TGC), has been developed to help predict growth of fish based on previous production records and current water temperature profile. It is important to note that the instantaneous growth rate model, better known as specific growth rate (SGR), is not up to the task and its use should be avoided.

Live weight gain is the result of deposition of water, protein, fat and minerals. The amount of these components deposited per unit of live weight gain is not constant but rather changes with fish species and size, feed used, etc. For this reason, knowledge of the composition of the fish reared is another key factor for the accurate determination of feed requirement and waste output. Research aiming at the development of models to predict rainbow trout composition at various sizes depending on the composition of the feed fed is underway. It might be faster and more reliable, however, for fish producers to invest in the determination of the chemical composition and energy content of their own fish. A number of private laboratories can perform, at very reasonable cost, the basic chemical analyses required (moisture, protein, fat, energy, phosphorus) on fish samples.

3) WASTE Estimation

The maintenance of life processes (integrity of the tissues of the animal, osmoregulation, respiration, circulation, swimming, etc.) and the deposition body components have costs in terms of nutrient and feed energy. Basic and practical research projects have allowed the development of simple, yet reliable, models (equations) to calculate these costs or wastes. Studies involving the rearing of fish under the variety of conditions (water temperature, feeding level, fish size, etc.) have shown that these biological costs are, surprisingly, fairly constant and, consequently, fairly easily predicted.

4) RATION Allowance

. Calculation of the ration can be done by simply to adding all the different components calculated above (body components deposited + waste produced) per unit of time (day, week) to calculate total cost. This cost is generally expressed as "digestible energy". Knowledge of the digestible energy content of the feed used allows calculation of the amount of feed to be served over a certain period of time or the feeding level to be used. This calculated amount of feed generally represents the minimum amount of feed required to achieve the predicted growth of the fish.

5) FEEDING Strategies

Any model, as good as it is, cannot replace common sense when feeding fish and it is up to the producer to determine how much feed to serve and how to serve it depending on the prevailing conditions. Feed should be served in manner that allows adequate opportunity (time or space-wise) for the fish to consume the determined ration and achieve their growth potential while minimising feed wastage.

This approach to rationally feed requirement has been used with much success by OMNR fish culture stations and for numerous studies with salmonid fish species at the Fish Nutrition Research Laboratory and the Alma Aquaculture Research Station. Used properly, it could become a valuable management tool for commercial fish culture operations, notably by providing yardsticks to compare current performance (example: feed conversion ratio, FCR) with what is estimated to be biologically achievable. The models are evolving as more accurate information becomes available and as our understanding of nutrient and energy utilization improves. This type of approach could be used for a number of aquaculture production provided the necessary information is generated (Kaushik, 1998). Other practical applications could be developed using this or other rational approaches.

References

Azevedo, P.A., C.Y. Cho, S. Leeson and D. P. Bureau. 1998. Effects of feeding level and water temperature on growth, nutrient and energy utilization and waste outputs of rainbow trout (Oncorhynchus mykiss). Aquatic Living Resources, 11: 227-238.

Brett, J.R. 1979. Environmental factors and growth. In Hoar, W.S., D.J. Randall and JR. Brett. Fish Physiology, Vol. VIII, Bioenergetics and Growth. Academic Press, New York, pp. 599-667 .

Bureau, D.P., A.M. Harris and C.Y. Cho. 1998. The effects of purified alcohol extracts from soy products on feed intake and growth of chinook salmon and rainbow trout. Aquaculture, 161: 27-43.

Cho C.Y. 1992. Feeding systems for rainbow trout and other salmonids with reference to current estimates of energy and protein requirements. Aquaculture, 100: 107-123.

Cho, C.Y. and C.B. Cowey. 1991. Rainbow trout, Oncorhynchus mykiss. In: Wilson, R.P. (Editor). Handbook of Nutrient Requirements of Finfish. CRC Press, Boca Raton, pp.131-143.

Cho, C.Y. and D.P. Bureau. 1998. Development of bioenergetic models and the Fish-PrFEQ software to estimate production, feeding ration and waste output in aquaculture. Aquatic Living Resources, 11: 199-210.

Cho C.Y. and S.J. Kaushik. 1990. Nutritional energetics in fish: energy and protein utilization in rainbow trout (Salmo gairdneri), World Rev. Nutr. Diet., 61: 132-172.

Cho, C.Y., H.S. Bayley and S.J. Slinger. 1974. Partial replacement of herring meal with soybean meal and other changes in a diet for rainbow trout (Salmo gairdneri). J. Fish. Res. Board Can., 31: 1523-1528.

Cho C.Y., S.J. Slinger, and H.S. Bayley, 1982. Bioenergetics of salmonid fishes: Energy intake, expenditure and productivity, Comp. Biochem. Physiol., 73B: 25-41.

Cho, C.Y., S.J. Kaushik and B. Woodward. 1992. Dietary arginine requirement of young rainbow trout (Oncorhynchus mykiss). Comp. Biochem. Physiol. 102A: 211-216.

Choo, P.S., T.K. Smith, C.Y. Cho and H.W. Ferguson. 1991. Dietary excess of leucine influence growth and body composition of rainbow trout. J. Nutr. 121: 1932-1939.

Cowey, C.B. 1994. Amino acid requirements of fish: a critical appraisal of present values. Aquaculture, 124: 1-11.

Cowey, C.B., C.Y. Cho, S.G. Sivak, J.A. Weerheim and D. D. Stuart. 1992. Methionine intake in rainbow trout (oncorhynchus mykiss), relationship to cataract formation and the metabolism of methionine. J. Nutr. 122: 1154-1163.

Fowler, L.G. 1980. Substitution of soybean and cottonseed products for fish meal in diets fed to chinook and coho salmon. Prog. Fish-Cult., 42 : 87-91.

Gomes da Silva, J. and A. Oliva-Teles. 1998. Apparent digestibility coefficients of feedstuffs in seabass (Dicentrarchus labrax) juveniles. Aquat. Living Resour., 11: 187-191.

Geurden, I., D. Marion, N. Charlon, P. Coutteau and P. Bergot. 1998. Comparison of different soybean phospholipidic fractions as dietary supplements for common carp, Cyprinus carpio, larvae. Aquaculture, 161: 225-235.

Geurden, I., J. Radunz-Neto and P. Bergot. 1995. Essentiality of dietary phospholipids for carp (Cyprinus carpio L.) larvae. Aquaculture, 131: 303-314.

Halver, J.E. 1966. Vitamin and amino acid requirements of Pacific salmon (Oncorhynchus), Eur. Inland Fish. Advis. Comm. Techn. Paper, 3: 61-68.

Halver, J.E. 1989. The vitamins. In: Halver, J.E. (Ed.). Fish Nutrition, 2^nd edn., Academic Press, Inc. New York, pp. 31-109.

Hardy, R.W. 1999. Problems and opportunities in fish feed formulation. Aquaculture Magazine, July/August 1999, p.56-60.

Higgs, D.A., B.S. Donsanjh, R.M. Beames, A.F. Prendergast, S.A. Mwachireya, and G. Deacon. 1996. Nutritive value of rapeseed/canola protein products for salmonids. Proceedings of the Eastern Nutrition Conference, 15-17 May 1996, Darmouth/Halifax, Nova Scotia. pp.187-196.

Iwama G.K., and A.F. Tautz. 1981. A simple growth model for salmonids in hatcheries, Can. J. Fish. Aquat. Sci., 38: 649-656.

Kaushik, S.J. 1998. Nutritional bioenergetics and estimation of waste production in non-salmonids. Aquatic Living Resources, 11: 211-217.

Kaushik, S.J., M.F. Gouillou-Coustans and C.Y. Cho. 1998. Application of the recommendations on vitamin requirements of finfish by NRC (1993) to salmonids and sea bass using practical and purified diets. Aquaculture 161: 463-474.

Kaushik, S.J., J.P. Cravedi, J.P. Lalles, J. Sumpter, B. Fauconneau and M. Laroche. 1995. Partial or total replacement of fish meal by soybean protein on growth, protein utilization, potential estrogenic or antigenic effects, cholesterolemia and flesh quality in rainbow trout, Oncorhynchus mykiss. Aquaculture, 133: 257-274.

Lim C. and P.H. Klesius. 1998. Immune response and resistance of channel catfish to Edwardsiella ictaluri challenge when fed various protein levels of dietary protein. 27^th Fish Feed & Nutrition Workshop, September 13-15, 1998, Pine Bluff, Arkansas, USA.

Lupatsch, I., G. Wm. Kissil, D. Sklan, E. Pfeffer. 1998. Energy and protein requirement for maintenance and growth in gilthead seabream (Sparus aurata L.). Aquaculture Nutrition, 4: 165-173.

Meyer-Burgorff, K.H., C. Muller, K. Becker and K.D. Gunther. 1989. Determination of the energy metabolism in mirror carp. J. Anim. Physiol. And Anim. Nutr., 62: 75-84.

Meyer-Burgorff, K.H., M.F. Osma, and K.D. Gunther. 1989. Energy metabolism in Oreochromis niloticus. Aquaculture, 79: 283-291.

National Research Council (NRC). 1981. Nutrient Requirements of Cold Water Fishes, National Academy Press, Washington.

National Research Council (NRC). 1993. Nutrient Requirements of Fish, National Academy Press, Washington.

Ohta, M., and T. Watanabe. 1998. Effect of feed preparation methods on dietary energy budgets in carp and rainbow trout. Fisheries Science, 64: 99-114.

Parsons, C. M., 1998. Variation in protein quality of soybean meal for poultry. Proceeding

Arkansas Nutrition Conference, September 15-17, Fayetteville, Arkansas, USA.

Philbrick, D.J., D.P. Bureau, F. W. Collins, B. Sarr, M.R. Ogborn and B.J. Holub. 1999. Effect of a soyasaponin-enriched alcohol extract from soy protein isolate on disease progression in mice with polycystic kidney disease. Amercian Society of Nephrology Annual Meeting, 5-8 November 1999, Miami Beach, Florida, USA.

Pongmaneerat, J. and Watanabe, T., 1992. Utilization of soybean meal as protein source in diets for rainbow trout. Nippon Suisan Gakkaishi, 58 : 1983-1990.

Rodehutscord, M. and E. Pfeffer. 1999. Maintenance requirement for digestible energy and efficiency of utilisation of digestible energy for retention in rainbow trout, Oncorhynchus mykiss. Aquaculture, 179: 95-107.

Schwartz. F.J. and M. Kirchgessner. 1995. Effects of different diets and levels of feeding on retention and efficiency of utilization of energy and protein by carp. J. Appl. Ichthyol., 11: 363-366.

Spinelli, J., C. Mahnken and M. Steinberg. 1979. Alternate sources of proteins for fish meal in salmonid diets. In: Halver, J.E. and K. Tiews. Finfish Nutrition and Fishfeed Technology. Proc. World Symp. on Finfish Nutrition and Fishfeed Technology, Hamburg, 20-23 June 1978. pp. 131-142.

Talbot, C. 1993. Some aspects of the biology of feeding and growth in fish. Proceedings of the Nutrition Society, 52: 403-416.

van den Ingh, T.S.G.A.M., Olli, J. and Krogdahl, A., 1996. Alcohol-soluble components in soybeans cause morphological changes in the distal intestine of Atlantic salmon, Salmo salar L. J. Fish Dis., 19: 47-53.

Watanabe, T., V. Verakunpiriya, K. Watanabe, V. Kiron and S, Satoh. 1997. Feeding of rainbow trout with non-fish meal diets. Fisheries Science, 63: 258-266.

Woodward, W.D. 1994. Dietary vitamin requirements of cultured young fish, with emphasis on quantitative estimates for salmonids. Aquaculture, 124: 133-168.