**3. Evolution of poultry amino acid nutrition**

In the 1990s, poultry amino acid nutrition reports placed more emphasis on practical aspects than those of previous decades. Large advancements in least-cost formulation strategies for broiler integrators, brought about by linear programing and personal computers that could conduct it, occurred in the mid 1980s, allowing nutritionists to rapidly produce mock formulas [59]. Lack of experience with this technology caused a distrust with feed-grade amino acids limiting their use, which would later be overcome with the widespread adoption of L-threonine in the 1990s [59]. This allowed for dietary crude protein to settle on the 4th limiting amino acid which varied depending on the ingredients included in broiler diets [60]. Therefore, research during the 1990s and early 2000s largely shifted to the determination of amino acid requirements, although some antagonism work remained.

By the 1990s research evaluating the lysine-arginine antagonism had largely come to an end. Mendes et al. [61] failed to observe any response to variations in dietary lysine or the arginine to lysine ratio when feeding broilers three to six weeks of age. The classic responses observed were largely the result of the ingredients used in nonpractical diets (i.e., casein) and not something that would typically occur in poultry production. Similarly, studies determining the arginine requirement began to produce relatively consistent requirement estimates, likely resulting from the constraints placed on lysine during formulation (**Table 2**). In addition to arginine's role in animal growth, research into its influence on animal health gained popularity and was added to requirement parameters [67–70].

Conversely for the branched-chain amino acids, Farran and Thomas [71] implemented central-composite, rotatable design to model the branched-chain amino acids


*1 Ratio of arginine to lysine.*

*2 Select trials used due to experimental design.*

*3 Non-heat stressed.*

#### **Table 2.**

*Estimations of the arginine requirement for broiler chickens of various age, strain, and sex.*

and determine the requirements of the three simultaneously. Farran and Thomas [71] found significant interactions between valine and isoleucine, but were unable to identify any effect of leucine, differing from historic data. Due to the lack of effect of leucine, Farran and Thomas [71] eliminated leucine from their model, only determining requirements for valine and isoleucine, and began working with valine instead of continuing antagonism work [72, 73].

Also in the early 1990s, Burnham et al. [74] implemented a dilution technique in order to assess the effects of increasing isoleucine at different dietary valine and leucine levels. Burnham et al. [74] found that valine had no effect on the isoleucine requirement, and that leucine only depressed body weight when isoleucine was at the lowest tested levels. These findings resulted in Burnham et al. [74] postulating that the negative influences of leucine would not be an issue in practical diets if the ingredients used contained adequate amounts of isoleucine. Barbour and Latshaw [75] also evaluated the influence of valine and leucine on broiler isoleucine requirements but implemented practical type diets. No influence of valine nor leucine were observed on the isoleucine requirement. Barbour and Latshaw [75] indicated that the lack of a response was due to their experimental design in which not only were basal diets formulated with practical ingredients but adjustments in valine and leucine were brought about by practical ingredients available to the broiler industry. The final experiment of this era was conducted by Waldroup et al. [76]. Similar to the design of Barbour and Latshaw [75], Waldroup et al. [76] tested the effect of excess leucine by varying the amount of corn gluten meal in the diet. No negative effects were observed as a result of the excessive leucine levels, reaching over 3.5% of the diet. Waldroup et al. [76] indicated that the lack of response was driven by the increasing levels of isoleucine and valine that accompanied the excess leucine levels as a result of using intact protein sources to drive the leucine level. These universal excesses among the branched-chain amino acids allowed for the bird to account for potential losses of valine and isoleucine associated with the antagonism. Waldroup et al. [76] concluded their report theorizing that as more feed-grade amino acids entered poultry formulation, branched-chain amino acid antagonism may become a practical concern due to the elimination of excess valine and isoleucine in broiler diets.

## **4. The return of antagonism research**

The doctoral work of I.C. Ospina-Rojas, resulted in three papers investigating interactions between valine and leucine [77–80]. To evaluate the relationship between valine and leucine and its influence on live performance and carcass traits, Ospina-Rojas [77] conducted two 5 × 5 factorials, after which results were displayed via response surface graphs to allow for visual observations of trends. During a 1–21 day starter phase, Ospina-Rojas et al. [80] observed valine × leucine interactions for fed intake and feed conversion. Ospina-Rojas et al. [80] was able to determine leucine and valine to lysine requirement values of 104 and 77 and 102 and 73 for feed intake and feed conversion, respectively. Feed intake was most severely impacted when valine levels were low and leucine levels were high, whereas feed conversion spiked when both amino acids were fed at low levels.

When varying valine and leucine levels were fed during a 21–42 day period, Ospina-Rojas [78] observed significant valine by leucine interactions for feed intake and body weight gain. Unlike with the previous growth phase, requirement values could not be estimated for maximal feed intake as a ridge occurred for feed intake between valine to lysine ratios of 82 and 91 for the entire range of leucine. Feed intake values remained relatively constant across leucine levels but it was again minimized when valine levels were low and leucine levels were high. For body weight gain, a requirement estimate was determined at a valine and leucine ratio to lysine of 111 and 83, respectively. As with feed intake, body weight gain was lowest when dietary valine was low and leucine was high.

Zeitz et al. [81] evaluated the influence of excess leucine on broiler performance and carcass traits when branched-chain amino acid levels were either fixed [82] or allowed to drop in relation to leucine level [81]. When branched-chain amino acid ratios were fixed, no differences were observed in growth performance over a 1–35 day period, but breast yields were decreased when leucine was increased by approximately 60%. However, no differences were observed for a 1–34 day period nor day 34 carcass traits when levels valine and isoleucine ratios in relation were allowed to drop when leucine increased.

Ospina-Rojas et al. [83] evaluated the influence of high leucine levels on the valine and isoleucine requirements for a starter (1–14 day), grower (14–28 day), and finisher periods (28–42 day) through the implementation of central-composite, rotatable design. Ospna-Rojas et al. [83] observed consistent influence of branched-chain amino acids on feed conversion across all three feeding phases, but body weight gain was not affected until the finisher phase. Unlike previous experiments, Ospina-Rojas et al. [83] did not generate response surface graphs, but did report regression equations. The lack of response surface graphs was due to the significant effect of three factors that cannot be displayed on a three-dimensional graph. Requirements estimates needed for optimal body weight gain reported by Ospina-Rojas et al. [83] generally showed that valine and isoleucine requirements decrease as the bird ages, but leucine needs increase.

A pair of studies published in 2021 implemented the use of Box-Behnken design to characterize the broilers response to various branched-chain amino acid levels [84, 85]. The studies were completed as part of a ring study and followed the same experimental design, with the only difference being the type of birds used (i.e., strain and sex). Maynard et al. [85] found significant interactions between valine and isoleucine for body weight gain, feed conversion, and breast meat yield when branched-chain amino acid levels were varied in diets fed to Cobb MV × 500 broilers. The effect of leucine was

#### *Broiler Amino Acid Research: Then and Now DOI: http://dx.doi.org/10.5772/intechopen.101896*

limited to an interaction between leucine and valine on breast meat yield. Maynard et al. [85] came to a similar conclusion to that of Farran and Thomas [71] that leucine may not be a significant factor under practical conditions and eliminated it from their model, replacing it with glycine + serine due to the potential limitation of glycine or nonessential nitrogen in the reduced crude protein diets implemented. When leucine was removed from the model, the interactions between valine and isoleucine were virtually eliminated, indicating that the "real" effect of leucine may be a "shadow effect" that does not present as a traditional significant response Maynard et al. [85]. Kidd et al. [84] focused on the branched-chain amino acids in their study but conducted it in male and female Lohman Indian River broilers. Contrary to the findings of Maynard et al. [85], Kidd et al. [84] did observe significant influence of leucine, citing interactions between leucine and isoleucine for body weight gain and feed conversion and leucine × valine interactions for carcass and breast meat yield. Furthermore, Kidd et al. [84] found that female broilers were more responsive to branched-chain amino acid supplementation than males.

Maynard [86] followed up the findings from the Maynard et al. [85] studies through the implementation of factorial designs meant to confirm the modeling responses. The first factorial study conducted by Maynard et al. [86] sought to determine the shift in the valine requirement when high and low levels of isoleucine and leucine were fed in practical type diets. Interestingly, a three-way interaction was observed for feed conversion but two sub interactions, valine × leucine and valine × isoleucine, were observed for body weight gain. The body weight gain responses observed by Maynard [86] (**Figure 4**) closely resembled those observed by D'Mello and Lewis [34] 50 years ago, without using purified diets or large swings in dietary leucine. Maynard [86] then attempted to characterize the sub interactions, valine × leucine and valine × isoleucine, observed in the larger study but failed to get a response. This lack of response again highlighted and expanded upon the observations Maynard et al. [85], that investigations into the branched-chain amino acid antagonism require testing of all three due to the complex nature of the antagonism.

The most recent study evaluating the branched-chain amino acids was a centralcomposite, rotatable design presented by Corzo and Silva [87]. Corzo and Silva [87] observed significant three-way interactions for body weight gain, feed conversion, carcass yield, and breast meat yield. General trends showed that increased isoleucine and valine were needed when leucine was fed in excess, but potentially more important, that the negative effects of leucine could be overcome for all parameters, except

#### **Figure 4.**

*Influence of titrating valine on average daily gain at high (solid line) and low (dashed line) leucine (left) and isoleucine (right). Adapted from Maynard [86].*

carcass yield, when leucine levels continued to increase with proper isoleucine and valine supplementation.

These recent studies have shown that unlike the lysine-arginine antagonism, branched-chain amino acid antagonism presents in practical type diets and will continue to be an issue for practical broiler production as crude protein levels decrease. The work of Corzo and Silva [87] is promising as it appears the effects of this antagonism could be turned from a negative to a positive. Maynard et al. [88] indicated that in the future this phenomenon may be referred to as the branched-chain amino acid synergism based on a meta-analysis conducted on branched-chain amino acid research conducted from 2000 to 2021.

## **5. Future research**

In the modern era, broiler amino acid research is centered around more complex problems as opposed to the simple strategies of the past. Titration studies used to determine amino acids requirements will remain the gold standard, as laid out by Lewis [89], but further refinement of these requirements will require researchers to consider test diet nutrient profiles compared to those observed in commercial practice. With the present known, and potentially unknown, antagonisms influencing amino acid requirements, generated values from test diets may not accurately represent those that produce optimal performance under commercial or practical conditions. Likewise, differentiated responses to branched-chain amino acid levels were observed when broiler strain or sex was changed under similar experimental conditions.

The double-edged sword of evaluating these complex interactions is the need for larger research facilities to achieve necessary experimental unit and replication. Another more manageable approach is the use of modeling. Previous researchers have shown that modeling research (i.e., Box-Behnken design) can be used in order to reduce the treatments necessary to characterize large scale interactions [90]. By effectively halving the number of treatments necessary to test a 3 × 3 × 3 interaction, the number of replicates can be doubled without increasing the number of necessary pens. This approach can be used over a broad range of inclusion levels in order to map general responses, then if a significant response is observed, treatment ranges can be reduced to reflect those observed in commercial practice to allow for a targeted approach. While Maynard [86] largely failed in the attempt to follow this strategy, the larger valine titration factorial was successful in observing a shift in valine requirements.

It is important to note that the collective work of Kidd et al. [84], Maynard [86], and Maynard et al. [85] used P-values ≥0.10 to identify significant interactions due to repeated observance of these levels. Originally, Kidd et al. [84] and Maynard et al. [85] set significance levels at P ≥ 0.10 due to the modeling approach used in their studies, but subsequent work by Maynard [86] observed similar P-values in their factorial approaches. P-values for the three-way interaction observed by Maynard [86] for feed conversion were found to be between 0.05 and 0.10, but when the data was broken into the individual titrations, P-values were found to be highly significant (i.e., P < 0.01). Relative consistency in responses to the branched-chain amino acids have been historically observed and noted by previous researchers [91].

While the current body of literature does not allow for concrete formulation strategies, promising studies have been recently conducted and the prevalence of this

## *Broiler Amino Acid Research: Then and Now DOI: http://dx.doi.org/10.5772/intechopen.101896*

style of research is increasing. The original observations of these antagonisms were brought about through the use varying ingredients, which changed the amino acid profiles of the diets implemented. The implications of how these discoveries were made are still relevant today with the ability of nutritionists to simply monitor the levels of nutrients in diets through the addition of nutrients to formulation software. While requirement minimums or formulation constraints will not necessarily be added for these nutrients, their inclusion in matrices allow for monitoring that can be reevaluated if negative performance or responses are observed in the field.
