TABLE OF CONTENTS

I. INTRODUCTION

II AMINO ACID DIGESTIBILITY
1. Procedures to estimate amino acid digestibility
2. Surgical techniques
3. Apparent ileal amino acid digestibility
4. Estimating endogenous protein and amino acid losses
5. Factors affecting endogenous losses

III DEAL PROTEIN FOR SWINE AND AMINO ACID LIMITATIONS
1 Definition and practical importance in diet formulation
2 Designing an ideal protein for different categories of swine

IV ESTIMATING DIETARY AMINO ACID REQUIREMENTS FOR SWINE: EMPIRICAL AND FACTORIAL APPROACHES

V AMINO ACID REQUIREMENTS
1. Maintenance
2. Mammary development
3. Fetal Growth and Gestation
4. Lactation
5. Litter size
6. Replacement gilts and boars
7. Starter, grower and finisher pig
7.1 Amino acid essentiality and antagonisms in the growing pig
7.2 Low protein-amino acid fortified diets

Tables

Literature Cited

I. INTRODUCTION

Amino acid nutrition of swine is an intense and growing research area and has been for the past 50 years. Consequently, there is a plethora of information on amino acid nutrition of the pig. Therefore, in this nutrition guide, the most current knowledge on practical amino acid nutrition is emphasized and summarized. The main goal is to provide the reader with an understanding of the basic principles on which amino acid requirements are built.

II. AMINO ACID DIGESTIBILITY

1. Procedures to estimate amino acid digestibility

Initially, the fecal collection procedure, developed by Kuiken and Lyman (1948), was used to determine amino acid digestibility. Fecal digestible amino acid values represented the amount of amino acid in the feedstuff that disappeared over the total digestive tract of the pig. It is now recognized that the ileal collection of digesta from the terminal ileum is more accurate than the traditional fecal approach (Sauer and Ozimek, 1986). Estimates of amino acid digestibility in the pig obtained by fecal collection are confounded by digestive processes in the lower tract, particularly the microbial activity in the cecum and large intestine (Easter and Tanksley, 1973). Amino acid digestibilities obtained by the fecal analysis method are, for most amino acids in feedstuffs, higher than those obtained by the ileal digesta collection method. Therefore, depending on the Amino acid and on the feedstuff, digestibilities obtained by the fecal analysis method overestimate those obtained by the ileal digesta collection method (Sauer et al., 1989). It is now widely recognized that the use of ileal digesta is the most accurate method to determine digestibility (Moughan and Smith, 1985; Sauer et al., 1989).

2. Surgical techniques

There are several techniques used to collect digesta from pigs. The simple "T"-piece cannula is the most commonly used cannulation technique in the U.S. and the Netherlands. The "T" cannula is inserted into the terminal ileum about 5-10 cm anterior of the ileocecal valve (Sauer et al., 1989). Digesta is collected from the cannula over a 24-hour period after the pigs have been on the test diets for 5-10 days. This technique offers the advantage that the surgery is relatively simple and that fecal sampling can also be undertaken. On the other hand, because this is a partial collection of digesta, an indigestible marker must be included in the test diet. Variability in digestibility coefficient estimates often arise due to analytical difficulty of the marker and lack of homogenous mixing of the marker in the diet.

Due to the uncertainties associated with obtaining a representative sample and the shortcomings of indigestible markers, total collection of digesta can be performed in pigs fitted with re-entrant cannulas. This type of cannula is inserted into either the terminal ileum, or either side of the ileocecal valve, or posterior to the ileocecal valve (Sauer et al., 1989). The advantage of this technique is that total collection of digesta is possible without feeding an indigestible marker. The technique has some limitations depending on what feedstuff is fed to the animal. Blockage of the cannula can occur and the extent of blockage increases with increasing particle size, crude fiber content, feed intake, and factors that increase the viscosity of the digesta (Sauer et al., 1989). In addition, because of a total diversion of the digesta, the small intestine function may be affected (Easter and Tanksley, 1973) such as disruption of the electrochemical pulses (Sauer et al., 1989).

The "ileal-rectal shunt" is a technique commonly used in France. The terminal ileum is anastomosed to the rectum, thereby allowing the digesta to bypass the large intestine and be collected from the anus. This technique allows for complete collection of digesta over the growth span of the pig and is suitable for routine collections. The shortcomings of this technique include possible mineral and water imbalances resulting from the by-passed large intestine (Sauer et al., 1989) and skin irritation due to the fact that the pig has little control over rectal function.

The various use of the available surgical techniques contribute to the variability in amino acid digestibility coefficient estimates of feed ingredients. However, an assessment of this area is beyond the scope of this review.

3. Apparent ileal amino acid digestibility

Apparent ileal digestibility measures the difference between the amino acid intake and Amino acid in the digesta. Depending on the test ingredient fed, the digestibility can be measured by one of two methods, i.e., the direct or indirect difference method. The direct method usually involves cereal grains that pigs willingly consume when formulated into single feed ingredient diets. The single ingredient supplies the only source of dietary protein, therefore the digestibility can be directly determined by the difference between the amount ingested and the amount at the terminal ileum. The indirect difference method is usually used with feedstuffs that are unpalatable or cannot be investigated with the direct method. In this case, digestibility is measured indirectly by the addition of a cereal grain to make a complete diet along with using a diet that is primarily comprised of the cereal grain. Then, by difference, the digestibility of the test feedstuff can be determined. Digestibility estimation require the presence of an indigestible marker in the diet. Chromic oxide (0.25% of diet) is often used for determination of amino acid digestibility (Fan et al., 1995). The following equation is used to estimate the digestibility of each amino acid.

Simply stated amino acid digestibility = (AA_f - AA_d) / AA_f
Where:
AAf is the amino acid concentration in diet
AAd is the amino acid concentration in digesta

The equation used to estimate digestibility coefficient with the use of a marker is as follow:
AD = (100- [(AA_d /AA_f) x (Cr_f /Cr_d)] x 100%
Where:
Crf is the analyzed chromium concentration in diet
Crd is the analyzed chromium concentration in digesta

The main disadvantage of measuring apparent digestibility is that it underestimates the actual digestibility of a feedstuff by not adjusting for endogenous nitrogen (N) or amino acid losses. Endogenous secretions are proteins or Amino acid that originate from various regions within the body. These secretions include saliva, gastric juices, bile acids, intestinal juices, sloughed off epithelial cells, and mucin (Soufrant, 1991; Grala et al., 1998). In addition, apparent digestibility coefficients are dependent on the protein and amino acid concentrations in the test diet. Thus, apparent ileal Amino acid digestibility coefficients increase curvilinearly with increasing Amino acid levels in the diet (Furuya and Kaji, 1989) and, as such, underestimate the digestibility of feedstuffs containing lower protein concentration. This results from basal endogenous losses not accounted for in the collection and the calculation of apparent amino acid digestibility. Basal endogenous losses are assumed to be equal in pigs fed either low or high levels of dietary protein. Therefore, in pigs fed feedstuffs of low protein concentration, the basal endogenous protein losses represent a greater proportion of the total protein in the digesta compared to pigs fed feedstuffs of higher protein concentration. As the level of protein in the feedstuff increases, this proportion decreases, resulting in this curvilinear relationship. In the past, many researchers reported apparent digestibility due to the labor and complications associated with attempts to estimate endogenous nitrogen (Leterme et al., 1991; Southern, 1991). However, today, measuring endogenous losses has become an essential component of amino acid nutrition work.

4. Estimating endogenous protein and amino acid losses

The endogenous nitrogen and Amino acid losses are divided into two origins: basal (minimum) and specific. The basal loss is non-specific and related to dry matter intake, whereas the specific loss is related to inherent factors in the feedstuff, e.g. fiber and anti-nutritive factors (Rademacher et al., 1999). Several methods can be used to measure basal amino acid endogenous loss. Feeding a protein-free diet is a common method, where all nitrogen-containing compounds in the digesta are of endogenous origin (de Lange et al., 1989; Donkoh and Moughan, 1994). The problem with this method is that the absence of protein per se may not be reflective of the endogenous secretions produced by a protein-containing diet. Therefore, it assumes that there is no relationship between protein intake and endogenous losses (Southern, 1991). Protein-free diets may overestimate endogenous losses of certain amino acid and, as such, overestimates total endogenous protein loss. For instance, proline recovered from a protein-free diet can make up 30% or more of the total amount of Amino acid in the digesta (de Lange et al., 1989). Other amino acids, such as glycine and arginine were also found to be elevated in the digesta compared to feeding a casein diet (Otto, 2001). Another commonly used method is feeding a highly digestible protein source such as casein (Chung and Baker, 1992) or wheat gluten. Apparent casein amino acid digestibility is close to 98% (Chung and Baker, 1992). The advantage of using casein is that while the amino acid digestibility is almost complete prior to reaching the hind-gut, the animal maintains a close to normal plane of nutrition (Stein, 1998).

The following equation is used to determine basal endogenous amino acid losses (EAL):
EAL = AA_d · (Cr_f / Cr_d)

Where:
AA_d is the analyzed amino acid content in digesta
Cr_f is the analyzed chromium concentration in diet
Cr_d is the analyzed chromium concentration in digesta

Thus, the standardized amino acid digestibility (SD) can be estimated using the apparent amino acid digestibility (AD) value and the corresponding endogenous amino acid loss (EAL):
SD = AD + (EAL / AA_f )
Where AA_f is the amino acid concentration in diet

A regression approach has also been used, but it is time consuming. Pigs are fed graded levels of dietary protein and the recovery of N and amino acids at the terminal ileum is related to N and amino acid intake. Via mathematical extrapolation, the recovery of N and amino acid at zero N and amino acid intake can be estimated (Fan and Sauer, 1994; Fan et al., 1995). This method assumes that endogenous losses are constant with varying dietary protein concentrations (de Lange et al., 1989). A final method that is not as commonly used as the above methods involves feeding a diet in which the sole source of N is enzymatically hydrolyzed casein. This hydrolyzed casein contains approximately 6% free amino acids and 41% di- and tri-peptides with a molecular weight of less than 5,000 Dalton (Da) (Leterme et al., 1994). It is assumed that all endogenous N is found in the ultrafiltrated fraction of the digesta with molecular weights greater than 10,000 Da. Thus, by separating the low (<5,000 Da) and the high (>10,000 Da) molecular weight N-containing compounds in ileal digesta, a distinction is made between non-absorbed exogenous and endogenous nitrogen.

5. Factors affecting endogenous losses

Secretion of endogenous N is mainly affected by dietary factors (Nyachoti et al., 1997). While dry matter mainly affects the basal endogenous losses (Butts et al., 1993), the diet specific losses are related to the level and quality of dietary protein (Sauer et al., 1989; de Lange et al., 1990; Donkoh and Moughan , 1994) , and the presence of anti-nutritional factors, such as trypsin inhibitors and lectins (Schulze, 1995). Dietary protein intake has been shown to be positively related to endogenous nitrogen secretions (den Hartog et al., 1989). Increased poorly digested dietary proteins result in increased enzyme secretion (den Hartog et al., 1989). Grala et al. (1998) found that a high trypsin inhibitor activity in soybean products fed to pigs was associated with high losses of endogenous N recovered at the ileum. Fiber can also influence the secretion of endogenous gut proteins and N. Increased protein and fiber levels were shown to augment endogenous nitrogen secretion (Sauer et al., 1977). Schulze et al. (1995) demonstrated that increased dietary neutral detergent fiber content increased ileal DM and N flow, reduced N utilization, and resulted in an increased endogenous N loss. Increased fiber in a protein free diet was found to increase the endogenous nitrogen at the terminal ileum as a result of sloughing off of mucosal cells and additional mucus production (den Hartog, 1989). Decreasing particle size increases protein digestibility (Wondra et al.,1995), and whether it decreases endogenous losses is unknown. There is a dearth of information on the effect of age and physiological status on amino acid digestibility. Thus, amino acid digestibility coefficients determined in the growing pig are used for all categories of swine. To date, only one study has determined amino acid digestibility coefficients in gestating and lactating sows, and only limited number of cereal feed ingredients have been studied (Stein et al., 1998). Apparent and standardized amino acid digestibility is similar between growing pigs and lactating sows, but lower compared to gestating sows (Stein et al., 1998). Stein et al. (1998) demonstrated that the difference in amino acid digestibility coefficient between growing pigs and gestating sows is due to the dry matter intake level, where gestating sows provided ad libitum access to feed had similar amino acid digestibility when compared to growing pigs and lactating sows. Thus restricting feed intake during gestation improved amino acid digestibility. Further studies are needed to determine amino acid digestibility in gestating sows fed various feed ingredients.

Estimation of dietary ingredient specific losses is extremely difficult due to the lack of reliable techniques and cost. For this reason, apparent amino acid digestibility can only be corrected for basal endogenous losses, and is thus referred to as standardized amino acid digestibility as opposed to what was considered "true" amino acid digestibility. At the Technical Institute for Cereals and Forages (ITCF, Paris, France), non-specific endogenous losses are estimated by feeding a protein free diet and the feedstuff specific endogenous losses are estimated with a model approach (Jondreville, 1995). Consequently, the standardized amino acid digestibility values may vary from one source to another, depending on whether the ingredient specific endogenous losses are included in the estimation. The new NRC (1998) provides regression equations to convert digestibility values from apparent to true and total to true. However, these equations are valid for diets that are speficically corn and soybean meal-based as a source of dietary proteins.

The availability of digestibility coefficient estimates allows dietary formulation to be more precise and the estimation of dietary amino acid requirements to be universal when based on a true digestibility measure.

III. IDEAL PROTEIN FOR SWINE AND AMINO ACID LIMITATIONS

1. Definition and practical importance in diet formulation

The traditional approach to determine amino acid requirement, the empirical dose-response assay, is extremely costly and time consuming (Bryden, 1991). In addition, empirical studies thus far have been based on titration of single amino acids (Trottier and Guan, 2000). With the advent of molecular biology techniques and new livestock genetics, conventional empirical studies may not keep pace with the need for new nutritional requirements (Trottier and Guan, 2000). An alternative approach, the "ideal protein," originates from the earlier observation that feedstuffs of high protein quality contain an amino acid profile similar to the amino acid profile found in the body protein of animals consuming them (Mitchell, 1950). Thus, in theory, the amino acid requirements of the pig could be derived from the amino acid composition and amount of the protein synthesized. A high correlation between the empirically derived amino acid requirement estimates and those based on the ideal protein approach (Fuller et al., 1979) led to the current amino acid requirement estimates published in the National Research Council Nutrient Requirements for Swine(NRC, 1998). The ideal protein is indeed a practical approach for dietary formulation. It allows one to only test for the requirements of the most limiting amino acid in the diet and simply derive the requirements for the remainder amino acids using the first limiting amino acid as a reference. One limitation is that the requirements based on this approach are not universal since they depend on the diet. In maize and soybean meal based diet, lysine is the first limiting amino acid for swine. Hence, requirements for the remainder amino acids are appropriate for pigs fed diets containing maize and soybean meal as their main feed ingredients or diets containing feed ingredients limiting in lysine. The ideal protein may also not apply to all categories of swine as discussed later in this review. In the lactating sow for instance, the loss of body weight during lactation complicates the "practical" application of the ideal protein.

2. Designing an ideal protein for different categories of swine

By definition, the ideal protein consists in an optimum dietary balance of indispensable amino acids supplied with sufficient nitrogen for the synthesis of dispensable amino acids that matches the animal requirements for growth and minimizes nitrogen loss (Cole and Van Lunen, 1994). One must assume that the amino acid composition of the swine species does not vary and that variation in the amino acid composition of the pig of similar age and/or physiological status is most likely resulting from analytical error. Thus, the foundation on which the ideal protein is based is universal. However, recent work has demonstrated that the amino acid composition of growing pigs is affected by protein and energy intake. This aspect will be discussed later in this review. Carcass protein accretion represents the main pool from which the dietary amino acid profile is derived in both the growing pig and the gestating sow. Milk represents the main protein accretion pool in the lactating sow. For all physiological states of swine, the same amino acid profile for maintenance requirements is used. The following table (Table 2) shows the amino acid composition of the various protein pools used in building an ideal protein for each category of swine. In Table 2, the first limiting amino acid, in this case lysine and threonine, is used as the reference amino acid. Hence, each amino acid is expressed as a ratio to the reference amino acid. If another amino acid is deemed first limiting, again, depending on the diet being fed and the respective production phase, this amino acid in question becomes the new reference, and a new set of ratios is determined. In Table 2, threonine is chosen as the first reference amino acid during pregnancy since it is often first limiting in gestating sows (King and Brown, 1993).

IV. ESTIMATING DIETARY AMINO ACID REQUIREMENTS FOR SWINE: EMPIRICAL AND FACTORIAL APPROACHES

While both the empirical and factorial approaches have been used, the empirical approach has been preferentially employed. The empirical approach involves the addition of graded concentrations of the test amino acid in its crystalline form to a diet deficient in that amino acid (D'Mello, 1982). The response curve is then used to determine the optimal dose for given levels of the response criteria chosen. Requirements derived from the empirical approach vary depending on the response criteria chosen and the feed ingredients used. In the growing pig, requirements have been based on lean growth, weight gain and feed efficiency, and nitrogen balance. Requirements for the gestating sow have been based mainly on nitrogen balance and litter characteristics at birth. Litter weight gain has been the main response criteria used to determined amino acid requirements for the lactating sow.

Accurate measurement of feed intake represents an essential prerequisite for satisfactory interpretation of the data in dose response experiments (D'Mello, 1994). One obstacle in the growing pig is that feed intake potential varies depending on environment and genetics. One obstacle in using the empirical approach for the lactating sow is the limited feed intake capacity of the sow and, to some extent, the accuracy of feed intake measurement. Body weight loss during lactation is essential to buffer nutrient demand imposed by the nursing litter (Weldon et al., 1994), consequently contributing to the wide variation in dietary lysine requirement estimates among studies.

In the factorial approach, amino acid requirements are determined from the sum of its physiological components (Fuller, 1994) and are directly estimated on a digestible basis. One advantage of the factorial approach is that the requirement estimates are independent of the diet being fed. Requirement estimates obtained from the empirical approach are affected by the ingredients fed, thus contributing to the variability in estimates. Expressing requirements on a true digestible basis decreases such variability, and, therefore, one must recognize the importance of precise digestibility coefficient estimates.

V. AMINO ACID REQUIREMENTS

1. Maintenance

Nutrient requirements for maintenance are influenced primarily by the body weight. Requirements are based on metabolic body size. Several exponential functions have been proposed (NRC, 1998) to describe metabolic body size but body weight raised to the 0.75 power is the most widely accepted. Protein requirements for maintenance are based on metabolic body weight just like maintenance energy requirements. The true ileal digestible lysine requirement for maintenance was estimated by NRC (1998) to be 36 mg/kg BW^0.75. Current maintenance requirements are most likely overestimated for the gestating and lactating sow since they have been assessed with growing pigs (Fuller et al., 1989) and mature gilts (Baker et al., 1966a, b, c; Baker and Allee, 1970). Conversely, it is unknown whether gestation or lactation increase maintenance requirements. In this review, maintenance requirements for growing pigs estimated by Fuller et al. (1989) are used for the gestating and lactating sow. Table 1 shows the amino acid estimates for maintenance requirement on a true digestible basis used for all phases of the pig's life cycle.

2. Mammary development

Little is known regarding amino acid nutrition and mammary development in gestation and lactation. Body composition in late gestation may affect mammary development and ultimately milk production in the subsequent lactation. Very little mammary development occurs during gestation until d 75 postmating. After d 75, there is an exponential growth in the concentration (Kensinger et al., 1982) and total quantity (Weldon et al., 1991) of DNA and RNA in the mammary gland. Altering protein intake of gilts during late gestation has no influence on mammary development (Weldon et al., 1991; King et al., 1996). Increasing lysine intake from severely deficient (4 g/d) to liberal (16 g/d) levels (NRC, 1998) had no effect on number of milk secreting cells in the mammary gland (Kusina et al., 1999b) but did increase milk production (Kusina et al., 1999a) in the subsequent lactation. Increased milk production may have resulted from elevated synthetic capacity of the mammary tissue (Trottier and Johnston, 2001). On the other hand, Kim et al. (1999) found that 1 g lysine per day was required for mammary growth during lactation.

3. Fetal growth and gestation

Limited information is available on the proper amino acid nutrition of the pregnant sow. Growth of the products of conception and the associated nutrient needs for that growth are fairly resistant to nutritional manipulations at feed intakes typical of production settings (Trottier and Johnston, 2000). For example, feeding level has little (Noblet et al., 1990) or no (Noblet et al., 1985) influence on body composition of fetuses. Pregnant sows can be fed a diet containing as low as 0.5% protein before the birth weight of her progeny is negatively affected (Antinmo et al., 1976). Litter size was not reduced in sows fed a diet containing 0.5% protein compared to sows fed a diet containing 17.7% protein (Antinmo et al., 1976), suggesting that protein restriction in pregnancy does not affect embryo survival. Protein restriction throughout pregnancy reduces postnatal weight gain (Pond et al., 1969; Antinmo et al., 1976), while protein restriction after the first months of pregnancy does not reduce birth weight or postnatal growth rate (Pond et al., 1968). In an other study, lactation failure occurred when pregnant gilts were fed low protein diet throughout gestation, but was optimal when pregnant gilts were fed adequate protein diets during the last trimester of gestation (Baker et al., 19xx). The results of these earlier studies, combined with the knowledge that nitrogen retention is higher during the last trimester of gestation (Noblet et al., 1985; King and Brown, 1993) indicate that amino acid requirements during the last trimester of gestation is higher. Nitrogen retention is also positively related to the number of fetuses in the uterus of pregnant swine (Rippel et al, 1965a). However, the maximum nitrogen retention for a given level of energy and amino acid intake is poorly associated with optimum reproductive and lactational performances (Miller et al., 1969). Amino acid requirement during pregnancy is still a matter of debate and remains subjective due to the dearth of information on the effect of body composition during pregnancy on lactational performance (Pettigrew and Yang, 1997).

Amino acid requirements during pregnancy are derived from total lean tissue accretion, which include maternal protein accretion (maintenance plus weight gain), and protein accretion in the products of conception (mainly fetuses). Because threonine is the first limiting amino acid during pregnancy in sows fed a corn-soybean meal based diet (Pettigrew, 1993), all the dispensable amino acid requirement estimates are derived from threonine and assumed to have the same utilization efficiency for protein synthesis. Table 2 displays the amino acid profile used to derive amino acid requirement from threonine. In practice, if another amino acid is determined to be first limiting, another profile could be calculated from Table 2. For instance, lysine may also be used, as lysine and threonine are near co-limiting in corn soybean meal based diets (Pettigrew and Yang, 1997).

Threonine requirement for maintenance has been estimated by nitrogen balance and extrapolation to the "Y axis" (King, 1991; Pettigrew, 1993). Across stages of pregnancy, on a true digestible basis, threonine is required at 38 mg/kg.75 per day for maintenance and .45 g per g of nitrogen retained (.53 g total converted to true digestible basis according to regression equation in NRC (1998)).

The amino acid requirements for pregnancy can be estimated using the following approach. First, a prediction for body protein accretion is made according to the following equation where fat tissue accretion = -9.08 + (0.638 x gestation weight gain) (Beyer et al., 1994). Thus, primiparous sows may gain 60 kg during the first pregnancy while a second and third (and above) parity sow may gain 40 and 30 kg, respectively. The sow parity will also affect the breeding weight; hence, maintenance requirement must be adjusted. With the assumption that lean tissue is 23% protein and that protein is 16% nitrogen, each kilogram of lean tissue accretion will correspond to 36.8 g nitrogen. Therefore, a sow predicted to gain a total of 43 kg of lean tissue during a 115-day gestation period will deposit 13.76 g nitrogen per day (43 kg x 36.8 g N/115 days). Threonine requirement can be derived, where .45 g total threonine is required for each g nitrogen deposited: .45 g digestible threonine x 13.76 g N per day = 6.19 g true digestible threonine. The other indispensable amino acids are derived from threonine. If one assumes lean gain to be associated mainly with fetal gain during the second stage of pregnancy, then lysine represents 1.82% of threonine, thus 1.82 x 5.76 = 10.48 lysine required per day. During the first stage of pregnancy where minimum fetal growth occurs, the sow body protein amino acid profile is used, and then lysine requirement decreases, i.e., 1.73 x 6.19 = 10.71 g required per day. If lysine is used as the first limiting amino acid, then a value of true digestible lysine requirement of .807 g per g N deposited is used. This represents, using the example above, a requirement of 10.33 g true digestible lysine per day. Therefore, either using threonine or lysine to determine lysine requirement gives nearly the same results. In Table 2, amino acid requirement for fetal gorwth are considered in stage 2. Stage 1 amino acid requirements are based on the body protein amino acid profile. It should be noted that while requirements for arginine are indicated in Table 2, arginine has long been known to be a dispensable amino acid for pregnant sows (Easter et al., 1974). Amino acid requirement estimates shown in Table 2 are in agreement and/or higher when compared to empirical estimates from earlier studies. The higher estimates may stem from the fact that protein deposition potential in sows is higher today compared to sows from 10 to 30 years ago. Easter and Baker (1977) demonstrated a tendency for improved nitrogen retention in gilts fed diets containing .12% histidine or 2.4 g histidine per day, compared to gilts fed diets containing 0% histidine. In addition, in the same study, histidine deprivation during gestation resulted in decreased hemoglobin concentration of the offspring. Thus, histidine is classified as an indispensable amino acid for pregnant sows. Phenylalanine requirement estimate from Speer et al. (1990) varied between 5 and 7 g per day, which is in agreement with the estimate in Table 2, when total aromatic amino acid concentration was low. In the presence of higher tyrosine concentration, phenylalanine requirement estimate decreased to approximately 3.63 g per day, which grossly overestimate the factorially derived phenylalanine requirement. Total aromatic amino acid requirement was estimated to be 9 g per day, which agrees more closely with Table 2. Considering that phenylalanine should be provided to meet 42% of the total aromatic amino acid requirement (Speer et al., 1990), an adjustment is made in Table 2 to account for the overestimation in the factorially derived phenylalanine requirement. Factorially derived phenylalanine requirement represents approximately 57% of the total aromatic amino acid requirements across level of production and parity. Isoleucine requirement was estimated to vary between 4.2 and 6 g per day (Speer et al., 1990), which is close to the factorially derived isoleucine requirement shown in Table 2. Previously published requirement for threonine of 5.4 g per day (Leonard and Speer, 1983) underestimates the threonine requirement shown in Table 2. More recent empirical requirement estimates of 9.5 g threonine per day (Dourmad and Etienne, 1998) and 8.4 g per day (Kusina et al., 1999) are is closer agreement with factorially derived values shown in Table 2. Similarly, recent requirement estimates of 12.3 g digestible lysine per day for the second stage of gestation averaged for primiparous and multiparous sows (Dourmad and Etienne, 1998) are equal to the factorially derived lysine requirement (Table 2). Similarly, tryptophan requirement estimated at 1.52 g per day (Meisinger and Speer, 1979) is in close agreement to the factorially derived tryptophan requirements in Table 2.

4. Lactation

Milk production

Milk production in the lactating sow is largely governed by demand from the litter (Sauber et al., 1994; Boyd et al., 1995) and nutrients availability from the sow. Nutrients arise from two sources, dietary and endogenous. Direct measurement of milk production in the lactating sow is not possible; hence, various indirect methods have been used to estimate milk yield in nutritional research such as litter weight gain (3.88 g milk:1 g of pig gain; Clowes et al., 1998), deuterium oxide (Pettigrew et al., 1987; Pluske et al., 1997), and weigh-suckle-weigh (Speer and Cox, 1984). Recent research indicates that milk yield peaks as early as between d 15 and 18 of lactation (Guan, 2000; Toner et al., 1996). This is in contrast to earlier studies in which lactation occurred between the third and fifth week (Elsley, 1971).

Sows usually not only satisfy nutrient need for milk production primarily from dietary nutrients but also from body protein and adipose tissue storage, where lean and fat represent 95% of the total body weight change during lactation (Mullan and Williams, 1990). Body weight loss is becoming a key issue and an essential parameter to consider when estimating amino acid requirements and predicting reproductive success (Trottier and Johnston, 2001). Today, sows require a greater amino acids supply than in the past because they have larger litters (Johnston et al., 1993), and the understanding behind body stores catabolism in support of milk production is becoming more important as feed intake may not keep pace with litter growth potential. Selection for leanness and for genotypes with high lean growth capacity has led to a reduced capacity for feed intake in sows (Riley, 1989).

Defining the amino acid requirements of the lactating sow is essential to optimize dietary formulation for ensuring maximum milk production and litter weight gain. Lysine requirements have been extensively researched for the past 20 years (NRC, 1998), and a wide variety of estimates found (Pettigrew, 1993). This variability has permitted development of models that allow prediction of lysine requirement to be made based on litter weight gain or production level and feed intake and body weight loss of sows during lactation (NRC, 1998). Thus, recommendations for lysine requirements are now based on predicted production levels and, therefore, are a better reflection of individual biological variation. There is a large gap of information, however, on the empirically derived estimates of essential amino acid requirements other than lysine.

For instance, requirements of a few essential amino acids such as valine (Richert et al., 1996; Richert et al., 1997a), methionine (Schneider et al., 1992a), and tryptophan (Libal et al., 1997) have only recently been determined through empirical studies. As previously mentioned, the most widely used approach to determining amino acid requirements, the empirical dose-response assay, is costly and time consuming. Alternative approaches such as the factorial and ideal protein are examined later in this review.

A substantial challenge behind designing nutritional studies with lactating sows has been selecting the response criteria that would most accurately reflect lactational performance. Average daily litter gain has been the response criterion of choice for measurement of lactational performance in sows. Blood urea nitrogen (BUN) (Coma et al., 1996 ) and nitrogen balance (Dourmad, 1998) are other response criteria that have been used to a limited extent to estimate lysine requirements. In the BUN approach, requirement of the amino acid in question is determined at nadir where it is assumed that utilization of all amino acids for protein synthesis is maximized and body protein degradation is minimized (Trottier and Guan, 1999). Coma et al. (1996) determined lysine requirement for a sow nursing a litter with a daily growth rate of 2.2 kg at 57 g per day. The BUN approach may not be applicable to all amino acids. For example, while increasing dietary valine level increases litter weight gain, it simultaneously increases BUN (Richert et al., 1996), suggesting valine may be oxidized rather than incorporated into milk protein.

Composition of weight loss

The composition of weight loss is important in determining milk production requirements for protein and amino acids. Typically, the body compositional loss during lactation is affected by the dietary protein and energy intake, the body composition of the sow, and the metabolic need imposed by lactation (Trottier and Johnston, 2001). Compositional weight loss in sows fed typical lactation diets has been assumed to be 13% protein (NRC, 1988). More recent studies have found that, as a proportion of body weight loss, protein loss varies from 0.09 to 0.14, showing that the loss of lean tissue is not constant and is diet dependent (Mullan, 1991). Energy-restricted sows have elevated blood urea nitrogen concentration throughout lactation, indicating greater muscle catabolism (Reese et al., 1982; Nelssen et al., 1985). Lactating sows restricted if fed during lactation can mobilize up to 25 to 30% of their body protein stores to maintain milk production (Mullan and Williams, 1990). The study of compositional weight loss of lactating sows has been of interest for many years because of the extended postweaning anestrous period in sows losing excessive amounts of body weight. Recently, composition of weight loss, particularly regarding body protein, has provided important insights in building an amino acid requirement model for the lactating sow (NRC, 1998; Guan, 2000). Accounting for body weight and protein loss during lactation has permitted the estimation of dietary amino acid requirements. The wide variability of amino acid requirement estimates has mainly resulted from the variability in body weight loss of the experimental sows. In other words, amino acids arising from endogenous sources such as muscle and internal organs protein degradation confound the requirement estimates of amino acid from dietary origin. Thus, the majority of dietary amino acid requirement estimates published to date are underestimated since the majority of the sows have limited feed intake capacity and, as such, lose body weight during lactation to keep up with lactation demand.

Ideal protein approach

Requirements for all essential amino acids are based in part on the ideal protein concept and on a series of selected empirical observations from the literature (Pettigrew, 1993). For the lactating sow, both milk and body protein amino acid pattern relative to lysine are used as the basis to establish the ideal protein (Pettigrew, 1993). The following steps are used in estimating requirements. First, from a regression analysis obtained by combining selected empirical studies, Pettigrew (1993) determined that 26 g of lysine is required for each kilogram of litter weight gain per day (Figure 1).

Figure 1

The efficiency of lysine utilization is unknown but assumed to be constant across milk production levels, litter sizes, stages of lactation, and feed intake level. Notice that the negative intercept represents the difference between maintenance requirement and lysine mobilization from body protein stores. Second, the amounts of amino acids present in milk and body protein, both expressed as grams per 16 g of nitrogen and as proportion of lysine, are used to estimate their requirements for milk production (Table 2). Table 4 shows an example for a sow weighing 160 kg post farrow with an expected litter growth rate of 2,000 g/d (10 pigs x 200 g body weight gain/day), lysine requirement is calculated first, where 26 g lysine per kg of litter gain is multiplied by 2 kg of expected litter gain per day, for a total of 52 g of lysine required per day. Assuming no body protein losses occur, the maintenance requirement for lysine is added to litter growth requirements, i.e., 2.09 g (maintenance) + 52.00 g lysine (litter growth), to obtain a total of 54.09 g lysine per day. Requirements for other amino acids are derived from their ratio relative to lysine in milk. Because all other indispensable amino acids are based on lysine concentration in milk, the efficiency of utilization of these amino acids is also assumed constant. For example, the threonine requirement is 0.58 of lysine requirement because the threonine to lysine ratio in milk is 0.58 (see Table 2). Hence, the threonine requirement for a sow nursing a litter with a growth rate of 2 kg per day, would be 30.16 g/d (52 g lysine x 0.58), plus the maintenance requirement for threonine of 1.76 g/d, for a total of 32.12 g/d . Table 4 shows the total amino acid requirement for this particular sow. Initially, lysine requirement was estimated on a total basis from studies using corn and soybean meal based diets. Amino acid requirements on an ileal true digestible basis are calculated in the same manner, using true digestible amino acids required for maintenance and true lysine required for litter growth. The current NRC (1998) amino acid requirement estimates are based on true digestibility estimates and on Pettigrew's (1993) approach with the exception of valine. Valine requirement has been increased relative to its requirement derived from its ratio to lysine in milk protein. The requirement is still somewhat arbitrary as the dietary ratio of valine to lysine of 0.85:1 in NRC (1998) lies between the ratio of 0.71:1 found in milk and a ratio of 1:1 suggested by Richert et al. (1996).

Factorial approach

Perhaps the easiest approach to use when estimating amino acid requirements for lactation is the factorial method. Table 4 also shows the total amino acid requirement estimates factorially derived compared to the original regression approach of Pettigrew (1993). Calculating the lysine requirement for milk production in a sow nursing a litter gaining 2 kg per day, using a factorial approach, results in similar estimates to those found using the original regression approach. In the case of the factorial approach, accuracy of requirement estimates is primarily dependent on knowledge of utilization efficiency for each amino acid. Because these coefficients are unknown, a single value derived from the efficiency of digestible nitrogen utilization is used for all amino acids. The factorial approach used herein assumes that all post-gut amino acids utilization for milk protein synthesis is 70%. Looking at both approaches illustrated in Table 4, requirement estimates are virtually the same for all amino acids, indicating that lysine requirement obtained by regression corresponds closely to its daily output in milk. This, in turn, suggests lysine is used in major part for milk protein synthesis during lactation.

Because the factorial approach yields requirements first on a true digestible basis, the amino acid requirement estimates for the sow recommended in this review are factorially derived. Table 5 shows true digestible amino acid requirement estimates for sow lactation with three different milk production or litter weight gain potentials. Requirement estimates largely agree with the latest estimates published by NRC (1998), except for arginine and valine. When compared to the NRC (1998), the factorial approach reveals that arginine requirement is higher, i.e., 30 g vs 23 g for a sow with a litter growth rate of 2 kg per day, and 37.5 g vs 31 g for a sow with a litter growth rate of 2.5 kg per day. No explanation can be offered for this discrepancy since the milk amino acid profile published in this review closely resembles that of the NRC (1998). In contrast, valine requirements determined by the factorial approach are underestimated when compared to NRC (1998), i.e, 31.8 g vs 35.8 gand 39.5 g vs 45.8 g for a sow with a litter growth rate of 2.0 and 2.5 kg per day, respectively. As mentioned previously, the NRC (1998) did not derive valine requirement from milk profile but assumed the requirement to lie between its ratio relative to lysine in the milk and empirical estimates from Richert et al. (1996).

Arteriovenous difference model approach

Arterio-venous differences of certain amino acids such as lysine, threonine, methionine, phenylalanine, tryptophan, and valine plateau with an increase in dietary concentrations of these amino acids, while the arterio-venous differences of leucine and isoleucine seem to be unregulated (Trottier, 1997; Guan et al., 1998). The nutritional significance of mammary amino acid uptake lies behind the fact that the amino acid uptake profile does not match the amino acid profile in the milk. This implies that amino acid requirements during lactation may be higher than currently thought. Much interest has been given to valine since Richert et al. (1996) demonstrated a response in litter weight gain for lactating sows fed graded levels of crystalline valine. Current productive demands may in fact increase the need of certain amino acids beyond their output reflected in milk, suggesting the amino acid requirements determined from the factorial method are underestimated, such as in the case of valine. The metabolic fate of amino acids in the mammary gland of mammals is virtually unknown, but maintenance costs associated with cellular remodeling and milk protein synthesis should be factored into the amino acid requirement estimates for lactation (Trottier, 1997). The arteriovenous difference model approach can predict amino acid requirement estimates for arginine, histidine, lysine, threonine, methionine, phenylalanine, tryptophan, and valine (Table 6). When comparing the AV approach to the factorial approach or the NRC (1998) model, amino acid requirement estimates are slightly higher, except for arginine where requirement estimates are two-fold higher. Valine requirement estimates are approximately 30% and 20% higher when compared to the factorial model presented in this nutrition guide and the NRC (1998) model, respectively. Because of the lack of research on the nutritional role of valine and arginine in the lactating sow, feeding level recommendations for these amino acids remain unclear. The factorially derived estimate for valine with an efficiency factor of 0.7 most likely underestimates its dietary requirement. Knowledge of valine utilization efficiency into milk protein synthesis may be much lower than 70%. In fact, recent work indicates that 30% of valine is oxidized into the mammary gland (Guan et al., 2001). When correcting for valine oxidation, thus using a utilization efficiency of 49% [0.7 x (1-0.3)=0.49], the factorially derived valine requirement estimates are 45.0, 50.5, and 56.0 g per day, for a sow supporting a LWG of 2, 2.25 and 2.5 kg per day. These requirement estimates exactly match those derived from the A-V model, i.e, 45.74, 51.34, and 56.94 g true digestible valine per day for the same respective LWG. Comparison of amino acid requirement estimates shown in Table 5 with published values is limited due to the paucity of information on recent empirically derived amino acid estimates other than lysine. Phenylalanine and total aromatic amino acid daily requirement of 18.17 and 41.2, respectively estimated by Lellis and Speer (1987 and 1985) grossly underestimates the current factorially derived requirements (Table 5). These studies used pigs with an approximate daily litter weight gain of 1100 g, hence explaining the discrepancy. On the other hand, when comparing more recent studies with similar litter gain potential and assuming a digestibility coefficient of 80%, daily requirement estimates are nearly identical to the factorially derived requirements shown in Table 5. Recent estimates are, for total sulfur amino acids, 33.2 g total (27 g digestible) (Schneider et al., 1992), for tryptophan, 11.05 g total (8.8 g digestible) (Libal et al., 1997), for threonine, 28 g digestible (Cooper et al., 2000) ,and for valine, 72 g total (58 g digestible) (Richert et al., 1996).

Considering endogenous amino acid contribution

As discussed above, body protein loss during lactation contributes to the underestimation on dietary requirements. A correction can be made when the knowledge of the amino acid contribution from body protein loss is known. In the NRC (1998) model, lysine contribution from body weight loss is estimated from the regression equation intercept (Figure 1) and lysine content in protein tissue. The other dispensable amino acids are derived from their ratio relative to lysine in body protein. The major pitfall with this method is that the intercept is not expressed per unit body weight loss and may vary in response to the energy intake. In addition, the rationale for using both the intercept value and the amount of lysine in body tissue remains unclear. Perhaps a simplified approach to this important question is to derive the amino acid contribution from body protein by using the amino acid profile in the body protein and assume that each kilogram of body weight loss contains 13% protein in sows fed balanced diet. Table 7 shows the amino acid contribution per kilogram of body weight loss during lactation. While these values closely match the NRC (1998) model values, their origin is more easily explained.

5. Litter size

Nutrition during lactation can influence subsequent litter size. Tritton et al. (1996) reported a positive relationship between lysine intake during the first lactation and number of pigs born in the second litter, where the number of pigs born live in the second litter increased from a low of 9.48 to a high of 10.92 as dietary lysine intake increased from 37 to 60 g/d. The lysine intake necessary to maximize subsequent litter size was higher than that needed to maximize milk production. In contrast, Touchette et al. (1998) reported that incremental additions of synthetic lysine to diets for lactating primiparous sows yielded a linear decline in number of total and live born pigs in the second litter. Yang et al. (2000) studied the effects of dietary lysine intake during lactation on subsequent litter size over four parities. Increases in dietary lysine were achieved without the use of synthetic lysine. They found detrimental effects of relatively high lactational lysine intakes on size of the subsequent litter in parities 2 and 3. In parity 4, there seemed to be no effect of previous lactational lysine intake on litter size at farrowing. Additionally, they found no difference between the amount of lysine intake needed to maximize milk production and that required to maximize subsequent litter size in sows. Therefore, current knowledge on the effect of amino acid nutrition on litter size is not sufficient to formulate any recommendations. Long-term sow experiments are necessary to fully understand the effects of lactation nutrient intake on subsequent litter size and lactational performance (Trottier and Johnston, 2001).

6. Replacement gilts and boars for the breeding herd

There is a real dearth of studies examining the amino acid requirements of the replacing gilt and boar. Therefore, the optimum level of amino acid nutrition for gilts and boar is unknown. The reason for this is well justified in that one must identify the response criteria that would reflect the amino acid requirements for one or more specific productive functions. In both the gilt and the boar, no specific productive function exists. Rather, requirements are based on "preparing" the animal for future life functions. Thus the amino acid requirements of the growing gilt and boar have not been objectively and scientifically measured. The amino acid requirements of the growing boar based on lean accretion are higher than that of the growing barrow (Williams et al., 1984). However, one must question whether optimizing protein accretion during growth will maximize semen production and quality in the mature boar. Because there is no published information regarding empirically or factorially derived amino acid requirement estimates of the growing boar on semen production and quality of the mature boar, the amino acid requirements are based on lean accretion only. Amino acid requirements of the growing and mature boar are shown in Table 8a.

Gilts are fed to maximize lean gain during the first six months, consequently the amino acid needs during that period will reflect the lean gain potential of the gilt (Trottier and Johnston, 2001). During the pre-pubertal development phase of the replacement gilt, amino acid requirements for maintenance and growth can be extrapolated from the amino acid requirement of a high-lean gain grower-finishing pig (Tri-State, 1998). Thus, until approximately 5 to 6 months of age, protein deposition is emphasized (Table 8b). This feeding strategy is recommended to maximize body protein stores and consequently reproductive longevity. However, a positive relationship between body protein stores and reproductive performance in later production cycles has yet to be established.

7. Starter, grower and finisher pig

A review of the literature on the amino acid nutrition of the growing-finishing pig, especially regarding lysine, methionine, tryptophan, and threonine, reveals a tremendous variability in dietary amino acid requirement estimates for the growing-finishing pig. Amino acid requirements of the growing pig are influenced by many factors, including dietary protein level, dietary energy density, environmental temperature, and genotype (Lewis, 1991). Additionally, as previously mentioned, the method chosen to quantitate amino acid requirements will contribute to this variation. For instance, the dose-response approach has been the method of choice to estimate amino acid requirements of the growing pig. Furthermore, requirements are affected both by the criterion of assessment, such as maximal leanness, rate of gain or feed efficiency, and blood urea nitrogen, and by the approaches used in establishing requirements. Again, variation has been useful to some extent in developing models that allow prediction of amino acid requirements. Hence, the most recent dietary amino acid requirement estimates in growing-finishing pig are based on lysine and protein deposition, which is closely related to the accretion of lean tissue (Möhn and de Lange, 1998). Thus, boars have higher amino acid requirements than gilts, which in turn have higher requirements than barrows (Lewis, 1991). Tables 9, 10, and 11 show the amino acid requirement estimates for the grower and finisher pig, respectively. These requirement estimates were derived from the NRC (1998) model, which is based on whole body lysine and protein accretion rates, and the amino acid profile for maintenance and protein gain.

7.1. Amino acid antagonisms in the growing pig

While amino acids must be present in sufficient amount in the diet to optimize protein accretion, some amino acids may antagonize each other when present in excess and, consequently, reduce growth rate. The interaction occurs usually between amino acids of structural similarities. Although very rare in practice, amino acid antagonism is important to recognize. Leucine given in excess decreases isoleucine and/or valine utilization for protein synthesis. The branched-chain amino acids leucine, isoleucine, and valine share a common catabolism pathway. They are first transaminated by the branched-chain aminotransferase ensyme to their keto-analogues, which are further decarboxylated by the branched-chain a-keto acid (BCKA) dehydrogenase enzyme. Leucine enhances BCKA dehydrogenase, hence increasing the oxidation rate of isoleucine and valine (Harper et al., 1984). Methionine also shares the same catabolism pathway with the BCAA (Benevenga, 1984), however, excesses in dietary BCAA seems to have very little influence on methionine utilization in the growing pig (Langer and Fuller, 2000). Lysine, in considerable excess, induces arginase, which in turn increases arginine catabolism in the rat (Jones et al., 1966). Large excesses (3.5% diet) of lysine are required to reduce plasma arginine and reduce feed efficiency in the pig; hence, a specific antagonism between lysine and arginine in the pig has been disproved (Edmonds and Baker, 1987).

7.2. Low protein-amino acid fortified diets

Amino acid nutrition in pigs extends beyond meeting requirements for maximum growth and/or performance. Dietary ingredients directly influence the excretion of potential environmental pollutants found in pig manure (Gatel and Grosjean, 1992; Hobbs et al., 1996; Sutton et al., 1999) such as nitrogen, ammonia and noxious odors. Ammonia and noxious odors from swine operations are a major concern of producers and the general public (Jongbloed and Lenis, 1998). Ammonia (NH₃) reduces pig health by causing nasal and eye irritation, resulting in rhinitis and epiphora, depression of somatic cell count, and respiratory and reproductive problems. Ammonia also reduces productivity (Malayer et al., 1988) and may affect meat quality. Ammonia emission from buildings contributes to environmental pollution (Nielsen et al., 1991) such as eutrophication and acidification of water and soil resulting in reduced plant diversity. Unionized ammonia or ammonium (NH₄⁺) and urea concentration, pH, and dry matter content of waste are the main factors influencing ammonia ionization and volatilization (Freney et al., 1983; Jongbloed and Lenis, 1998). Nutrition research designed to reduce ammonia concentrations in swine waste has focused primarily on feeding less dietary protein to more closely match supply of amino acid to requirements. This can be achieved by supplementing synthetic indispensable amino acids to replace a portion of the intact dietary protein. Results so far are promising; for every percentage unit reduction of dietary protein (from 16% to 10%), NH₄⁺ concentration and NH₃ emissions are reduced by approximately 11% (Canh et al., 1998). However, discrepancy exists in the literature regarding whether feeding low protein diets with synthetic amino acid can maintain optimum growth performance of growing-finishing pigs.

Origin of ammonia. Ammonia in swine waste originates in part from bacterial breakdown of protein components in the pig hind-gut. The main portion of NH₃ emission originates from urea in urine, which in turns originates from amino acid catabolism. Amino acids are catabolized by oxidative deamination (removal of a-amino group NH₂⁺) in the liver. The resulting NH₄⁺ is converted into urea for transport into blood to the kidneys and excreted. The manure handling system in the majority of swine operations do not permit effective separation of fecal and urine fractions. This has detrimental consequences on NH₃ emission: when urine comes into contact with feces, urinary urea is converted into NH4+ and carbon dioxide by microbial urease present in feces (Canh et al., 1997) (Figure 2).

Figure 2

The ionized form of ammonia, i.e, NH₄⁺ , remains in liquid fraction while the un-ionized form, i.e, NH₃, can become volatile. The proportion of the two forms is pH and temperature dependent. With the combined high pH, low DM content, and high concentration of NH₄⁺ in the slurry, NH₃ emission is increased. When pH is maintained below 4.5, there is no measurable form of free NH₃.

Crystalline amino acid supplementation. The extent to which amino acids are catabolized is largely determined by the balance of dietary amino acids relative to the requirements of amino acids for maintenance and growth functions. Thus, when amino acids are fed in excess of requirements, they are catabolized by the animal, contributing to the excess urinary urea excretion and NH₄⁺ formation. Feeding diets with balanced amino acids is now critical to minimize post absorptive catabolism of amino acids. Growing-pigs are fed diets containing feed ingredients that complement each other in providing adequate supply of limiting amino acids to the animal for growth, however, excessive amounts of 'non-limiting' amino acids and dispensable amino acids are also provided. Reducing the provision of these non-limiting amino acids in diets for growing pigs is essential to attain an ideal balance of amino acids and reduce nitrogen excretion. Ammonium concentration and the pH in the slurry decrease linearly as dietary protein is reduced (Cromwell et al., 1999). However, when reducing dietary protein, dietary concentrations of lysine, threonine, methionine, tryptophan, and isoleucine become limiting for growth. The order of limitation and the kind of amino acid will change as the protein is reduced and will vary with feed ingredients used. For example, in a corn-soybean meal based diet with a protein reduction from 15 to 12%, lysine will be the limiting amino acid and will need to be provided in its crystalline form in the diet. With a further reduction to 9% protein, lysine will be more limiting; thus, more crystalline lysine will be required in the diet. Furthermore, threonine, valine, isoleucine, methionine, and tryptophan will also be limiting, and thus, supplementation in their crystalline form will be necessary in order to meet their dietary requirements to maximize growth performance.

The type and amount of synthetic amino acid supplementation depends upon the extent to which the level of dietary crude protein is reduced and the type of feed ingredient used. These amino acids must be supplemented in synthetic form to maintain the rate of growth at the same level as when feeding an intact protein source. The result is that the animal receives a more "ideal" profile of amino acids with a reduced supply of those amino acids (e.g. some indispensable and dispensable amino acids), which would have been surplus to requirements. A further advantage of supplementing a reduced protein diet with synthetic amino acids is that synthetic amino acids are 100% absorbed prior to reaching the hind-gut, hence minimizing fecal nitrogen excretion. Thus, in dietary formulation, a digestibility coefficient of 100% is attributed to any synthetic amino acid used.

Current research and recommendations. Apart from its detrimental effect on the environment, excess feeding of dietary protein to growing pigs decreases efficiency of nitrogen retention and decreases energy retention and utilization. On the other hand, feeding diets with reduced dietary protein concentration to growing pigs may also reduce growth performance even when synthetic amino acids are provided in the diet to match requirements. While there is general agreement in the literature that reducing the protein concentration of growing and finishing pig diets by 2 to 3% (from 15 or 16% protein to 13 or 14% protein, respectively) and supplementing with synthetic amino acids does not affect growth rate, disagreement exists where the level of protein has been reduced by more than 3 percentage units, regardless of whether indispensable amino acids are supplemented or not. Reducing dietary protein from 17 to 10% with dietary supplementation of lysine, tryptophan, isoleucine, methionine, and valine, with or without glutamic acid supplementation in early growing pigs (20 to 24 kg) reduced gain, feed efficiency, and nitrogen retention (Kephart and Sheriff, 1990). Reducing dietary protein from 16 to 12% with dietary supplementation of lysine, threonine, tryptophan, and dispensable amino acids such as glutamic acid and glycine to early growing pigs (22 kg) did not reduce nitrogen retention (Kerr and Easter, 1995). Removal of glutamic and glycine reduced nitrogen retention. However, no reduction in urinary nitrogen was found, indicating that dispensable and possibly the indispensable amino acids were provided in excess. Formulation was based on a total basis rather than digestible, thus contributing to excess amount of dietary amino acids. Kerr and Easter (1995) also found that while growth performance such as average daily gain, average daily feed intake, gain to feed, longissimus muscle area, percent muscle and marbling is not compromised in starter, grower, and finisher pigs fed reduced dietary protein from 19 to 15%, 16 to 12% and 14 to 11%, respectively, mean backfat thickness is increased. Similarly, growth performance such as ADG, ADFI, G;F, and days on trial was not compromised in finishing gilts (70 to 100 kg) fed diets reduced from 12.74% protein to 8.69% protein with lysine, threonine, isoleucine, and tryptophan supplementation. Furthermore, in that study, nitrogen efficiency was improved, and mean backfat thickness and percent carcass lipid not affected in gilts fed diet containing reduced concentration of protein. Reducing dietary protein from 16 to 14%, 15 to 13% and 13 to 11% in starting, growing, and finishing period, respectively, did not affect the overall growth performance such as ADG, ADFI, G:F, or mean backfat thickness, but reduced ADG, ADFI, G:F and increases mean backfat thickness in the finishing period (Tuitoek et al., 1997). Pierce et al. (1994) monitored nitrogen balance in pigs (body weight of 70 kg) fed either a 14, 12 or a 10% protein diet. The 12 and 10% protein diets were supplemented with synthetic amino acids to meet requirements. While feeding the 12 and 10% protein diets resulted in significant reduction in the amount of nitrogen excreted, nitrogen retention was reduced compared to the 14% protein diet. Similar findings were reported for early growing pigs fed a 12% protein + synthetic amino acids compared to an intact 16% protein diet (Kendall et al., 1999). In contrast to the above studies, Hahn et al. (1995) found no adverse effect on growth performance or body lipid accretion of early and late finishing pigs fed diets reduced by 3.5% but formulated to ideal concentrations of digestible amino acids. The reduced nitrogen balance observed in some of the studies can stem from three possible mechanisms. First, the reduced loss of N in the urine yet accompanied by a decrease N retention indicates that gut endogenous N losses may have been increased. Second, supplementation of indispensable amino acids to meet requirements may not take into account the use of these amino acids for synthesis of certain dispensable amino acids. Third, a shortage of glutamic acid may arise as a result of increased glutamic acid utilization as a source of N and/or keto acid for synthesis of other dispensable amino acids. Consequently, dispensable amino acids may not be synthesized efficiently to achieve maximum growth and N retention. Glutamate may serve several critical roles, including as a source of energy by the gut, serving as a precursor for glutathione and glutamine synthesis. In addition, glutamate addition to growing pig diets seems to conserve threonine by limiting threonine catabolism.

Although the efficiency of dietary nitrogen deposition into body protein has improved by approximately 20% over the past 20 years with a concomitant reduction of 12% in nitrogen excretion (Jongbloed and Lewis, 1998), more reduction in nitrogen are urgently needed in order to lessen the negative environmental impact. Reducing dietary protein by 3 to 4 percentage units in the growing and finishing phase can be safely recommended given the commercial availability of amino acids in their crystalline forms and the economical feasibility of using these amino acids do not represent obstacles. There is no doubt that more research is needed to determine the balance between the minimum level of intact protein and the level of crystalline amino acid substitution necessary to maximize or maintain growth performance. The variability in growth response to feeding diets with reduced dietary crude protein results from a number of factors such as dietary formulation based on total rather that digestible basis, the supplementation of dispensable amino acids on a total nitrogen basis as opposed to an appropriate ratio of indispensable amino nitrogen to dispensable amino nitrogen (Lenis et al., 1999), and the supplementation of dispensable amino acids in diets already in excess of nitrogen (as shown by lysine as a percent of total nitrogen of less than 6%).

Table 1. Estimates of amino acid requirement for maintenance¹

¹Adapted from Fuller et al. (1989). Estimated in growing pigs using casein based diets and dietary crystalline amino acid supplementation.

Table 2. Amino acid composition of body protein, sow milk, and fetuses