R M Gous
University of KwaZulu-Natal, Pietermaritzburg, South Africa
Paper presented at the Poultry Science Association meeting in San Antonio, Texas (2022)
Conventionally, the overwhelming emphasis when formulating any poultry feed is to ensure that the minimum levels of all nutrients are met at least cost. These minimum levels are obtained either from tables of nutrient requirements or from the results of experiments designed essentially to measure the response of broilers, turkeys, laying hens or broiler breeders to a range of feeds that are, hopefully, first-limiting in the nutrient under test. The interpretation of the results of such response trials is all-important in defining the minimum levels of nutrients to be used in the formulation program. Generally, the choice has been between the nutrient content that maximises growth rate or minimises feed conversion ratio. These two objectives do not take account of the cost of input or the value of the output, both of which are fundamental when determining profitability for the business.
Although not the subject of this paper, it is nevertheless important to note that, when conducting a response trial, it is essential that the nutrient under test remains first limiting over the full range of inputs, which is often not the case when the graded supplementation technique is used to measure the response (Gous & Morris, 1985; Gous, 1986). If this is not the case, the response will be incorrectly defined and hence the optimum intake will also be incorrectly chosen, no matter what objective function is chosen. This is an important issue when deciding upon an optimum level of a nutrient to be included in a feed.
The primary objective of most businesses is to make a profit. When using tables of nutrient requirements, or nutrient contents that maximise growth rate or feed efficiency, to define the lower bounds for nutrients in a feed formulation, the cost of feeding and revenue are ignored. Given the wide diversity in raw material costs, husbandry conditions, ways in which broilers are sold and markets for broiler meat, it is impossible to expect a fixed set of nutrient requirements to result in maximum profit for any business. The lower bounds need to be dynamically chosen to reflect these differences, and to take account of changes in input costs and revenue. Possibly, the main reason for using fixed requirement values is that the nutritionist feels safe when justifying the levels used, as these are usually from reputable sources. Moving away from these fixed levels involves risk, the responsibility for which might be beyond the remit of the nutritionist. Yet, if management were made aware of the opportunity cost of using fixed nutrient requirements, they would surely approve of a more dynamic approach to defining the minimum bounds to be used.
Defining the optimum using response experiments
Assuming that the amino acid balance for each phase of broiler growth, published by breeding companies or Institutions (e.g., Rostagno et al., 2017; Cobb, 2018; Aviagen, 2019) is correct, this balance could be used to formulate a range of balanced protein levels from, for example, 0.8 to 1.2 times the level recommended. By retaining the relative levels of amino acids in each phase of the feeding program, growth rate, feed intake and body composition could be measured over the range of balanced protein levels chosen. The resultant data may be used to calculate feeding costs and returns under different scenarios, and would remain relevant as long as the potential growth rate of the genotype remains constant. An example of the use of this approach is given below (from Azevedo et al., 2022).
Table 1. Estimated margin over feeding cost a (US$/bird) of male and female broilers offered a range of balanced protein (BP) levels when sold live, dressed or further processed b on days 42 and 56
| BALANCED PROTEIN c | DAY 42 | DAY 56 | ||||
|---|---|---|---|---|---|---|
| ROSTAGNO | COBB | AVIAGEN | MALE | FEMALE | MALE | FEMALE |
| Margin from sale of live bird | ||||||
| 0.60 | 0.66 | 0.64 | 0.245 | 0.243 | 0.347 | 0.320 |
| 0.70 | 0.77 | 0.74 | 0.390 | 0.327 | 0.523 | 0.424 |
| 0.85 | 0.94 | 0.90 | 0.474 | 0.363 | 0.553 | 0.416 |
| 1.00 | 1.10 | 1.06 | 0.456 | 0.334 | 0.493 | 0.349 |
| Margin from sale of dressed bird | ||||||
| 0.60 | 0.66 | 0.64 | 0.387 | 0.373 | 0.647 | 0.586 |
| 0.70 | 0.77 | 0.74 | 0.629 | 0.523 | 0.934 | 0.779 |
| 0.85 | 0.94 | 0.90 | 0.781 | 0.606 | 1.135 | 0.884 |
| 1.00 | 1.10 | 1.06 | 0.790 | 0.595 | 1.142 | 0.852 |
| Margin from sale of further processed bird | ||||||
| 0.60 | 0.66 | 0.64 | 0.748 | 0.690 | 1.183 | 1.049 |
| 0.70 | 0.77 | 0.74 | 1.123 | 0.955 | 1.706 | 1.434 |
| 0.85 | 0.94 | 0.90 | 1.432 | 1.159 | 2.096 | 1.695 |
| 1.00 | 1.10 | 1.06 | 1.531 | 1.212 | 2.201 | 1.736 |
a Margin calculated using base prices for protein-containing ingredients and product sold.
b Breast, legs (thighs plus drums), wings and remainder.
c Proportion of balanced protein levels recommended by Rostagno et al. (2017), Cobb (2018) and Aviagen (2019) based on lysine recommendation in first two phases.
Because the amino acid recommendations of the three sources in Table 1 differ, the maximum margin over feed cost is in some cases below the recommendation and in others above, within a sale category. But in all cases the optimum is higher for the more expensive product thereby illustrating the folly of not taking account of the way in which the bird is sold when deciding on the lower bounds for the amino acids when formulating the feeds to be used.
In Table 2 the effect of changes in the cost of protein-containing ingredients on the optimum balanced protein level to use in broiler feeds is presented. Once again, the value of using a higher balanced protein content when selling further-processed birds is evident, but it is also apparent that when the cost of protein-containing ingredients change, so does the balanced protein content that maximises margin over feed cost. The opportunity cost, of using fixed amino acid lower bounds instead of reacting to changes in raw material costs and the way in which broilers are marketed, is substantial in many cases. Opportunities arising from changes in input costs or product value need to be siezed immediately if maximum benefit is to be gained, and this is only possible if performance can be predicted, rather than measured in the field. Predicting nutritional responses so that the process of optimization may be speeded up may be achieved with the use of nutritional modelling.
Table 2. Effect of increases or decreases of 25 % in the cost of protein-containing ingredients on the proportion of recommended a balanced dietary protein level that maximises margin over feed cost for male and female broilers sold live, dressed or further processed b at 42 and 56 d of age (from Azevedo et al., 2022)
| SALES CATEGORIES | DAY 42 | DAY 56 | ||
|---|---|---|---|---|
| MALE | FEMALE | MALE | FEMALE | |
| Base price | ||||
| Live bird | 0.92 | 0.88 | 0.88 | 0.82 |
| Dressed bird | 0.97 | 0.94 | 0.95 | 0.91 |
| Further processed | 1.04 | 1.01 | 1.02 | 0.99 |
| Protein-ingredient prices -25 % | ||||
| Live bird | 1.00 | 0.96 | 0.98 | 0.93 |
| Dressed bird | 1.03 | 1.00 | 1.01 | 0.98 |
| Further processed | 1.08 | 1.06 | 1.07 | 1.04 |
| Protein-ingredient prices +25 % | ||||
| Live bird | 0.85 | 0.80 | 0.77 | 0.71 |
| Dressed bird | 0.92 | 0.87 | 0.89 | 0.85 |
| Further processed | 1.00 | 0.97 | 0.98 | 0.95 |
a Balanced protein levels recommended by Rostagno et al. (2017).
b Breast, legs (thigh plus drum), wings and remainder.
The application of systems thinking and modelling to the problem of feed formulation leads to the replacement of the conventional approach with one in which nutritional decisions are made entirely in terms of the objectives of the business. Nutrient specifications are chosen that will maximise a desired objective function such as margin over feed cost, margin/m2 per annum or number of hatching eggs per flock. Feeding animals to achieve some company objective is not the same as feeding them to meet a ‘requirement’. Genetic potential as well as economic circumstances will change over time and different nutritional strategies will be needed to maximise margins. Also, nutritional decisions will depend on the stage in the production process at which margin is to be assessed. These are real differences, each requiring specific nutritional decisions. Simulation modelling enables these differences to be accommodated.
Nutritional models should thus be seen as making use of the nutrient requirements of an individual, which are relatively easy to calculate, to simulate the key outputs in performance of a population, such as feed intake, weight gain and carcass yield, such that the economic optimum feeds and feeding programme may be determined for each unique set of circumstances.
Defining the optimum using simulation models
The most important criterion when evaluating a simulation model to be used to optimise the feeding of broiler chickens is that it must be able to predict feed intake. In many simulation models, feed intake is an input to the model and not an output, and it is obvious that such a model would be incapable of finding an economic optimum under different circumstances. Feed intake is not constant over the range of balanced protein levels applied in a response experiment, as is often assumed, but increases as the level of protein becomes marginally deficient and then declines at low levels. Emmans (1981; 1987) described these changes as an attempt by the bird to consume as much of a given feed that was needed to enable it to grow at its potential, hence the increase in intake as the protein content falls. It is not always possible for the bird to consume sufficient of the given feed due to constraints such as gut capacity or the inability to lose sufficient heat to the environment. Some simulation models enable these issues to be accurately accounted for, resulting in accurate predictions of feed intake for different strains, feeds and environmental conditions.
Determining the value of the output, or revenue, might appear to be straightforward, being the product of the weight of the bird and the price per unit weight, but broilers are sold in many different ways in different markets, and the optimum amino acid level will differ depending on whether the bird is sold live, dressed or further processed, as the price and the weight of the portion being sold differs in each case. The weights of different physical parts of broilers and turkeys can now be accurately predicted on the basis of the actual amount of a given feed that is consumed each day (Danisman & Gous, 2011; 2013; Gous et al., 2019) and hence revenue may be accurately predicted under different circumstances.
Some nutritionists account for the variability in raw material composition using stochastic programming (Roush et al., 1996), which results in the defined minimum levels of nutrients being increased to different extents. It is still possible to make use of this technique when determining the optimum economic level of balanced protein to use under different circumstances, but it is questionable whether this is as important as is changing the optimum level for different economic, biological or management circumstances.
Choosing the objective function
As long as feed intake can be accurately predicted, it is possible, in turn, to predict the growth and carcass composition of a given genotype in a given environment on a given feed and feeding programme, which leads to the possibility of being able to optimize the way in which the birds should be fed to maximize or minimize a given objective function. The objective function to be optimized can be defined in terms of any output from a broiler growth model.
Some producers are intent on minimising feed conversion ratio (FCR), whilst others may need to minimise N excretion. Realistically, if the objective of the business is to make a profit, the objective function would be an economic index of some sort, such as margin over feed cost or margin per m2/year.
To illustrate the effect of choosing to maximise margin over feed cost or to minimise FCR or N excretion as the objective function, an exercise was conducted using the EFG Broiler Growth Model (EFG Software, 2022). The Aviagen (2019) balanced amino acid recommendations were used as the basis of these comparisons, with lysine being used as the reference amino acid (Table 3). It is clear that considerable differences in performance and profitability may be achieved by changing the amino acid supply to achieve different objective functions.
Table 3. Consequences on lower bound for lysine and on performance variables of maximising margin over feed cost or minimising feed conversion ratio (FCR) or N excretion
| BASE1 | MAXIMISE MARGIN OVER FEED COST |
MINIMISE FCR | MINIMISE N EXCRETION | |
|---|---|---|---|---|
| Feed | Lysine content, g/kg | |||
| Starter | 12.8 | 13.2 | 14.0 | 11.0 |
| Grower | 11.8 | 11.5 | 12.8 | 9.0 |
| Finisher | 10.0 | 8.0 | 12.0 | 8.0 |
| Performance variable | ||||
| Live weight at 35 d (g) | 2020 | 2102 | 2007 | 1969 |
| Feed intake to 35 d (g) | 3507 | 3725 | 3417 | 3823 |
| FCR | 1.58 | 1.61 | 1.55 | 1.76 |
| N excretion (g/bird) | 96.0 | 94.0 | 102 | 83.3 |
| Cost of feeding (c/bird) | 1622 | 1620 | 1620 | 1556 |
| Revenue (c/bird) | 2224 | 2331 | 2208 | 2183 |
| Margin over feed cost (c/bird) | 602 | 711 | 588 | 627 |
| Cost of production (c/kg) | 1094 | 1043 | 1101 | 1070 |
Now that tools are available for predicting feed intake and the weights of different saleable parts of a broiler, it seems sensible to choose the levels of dietary amino acids to be used in a formulation program based on the objectives of the business (maximise profit) rather than to use fixed table values, no matter how these are derived.
References
- Aviagen (2019)
Ross Broiler Nutrition Specifications [Online]. Available at: http://eu.aviagen.com/assets/Tech_Center/Ross_Broiler/RossBroilerNutritionSpecs2019-EN.pdf. (Verified 5 July, 2020) - Azevedo, J.M., Reis, M.P., Gous, R.M., Dorigam, J.C.P., Lizana, R.R. & Sakomura, N.K. (2021)
Response of broilers to dietary balanced protein. 2. Determining the optimum economic level of protein. Animal Production Science: 61: 1435-1441. - Cobb (2018)
Cobb 500 Broiler performance and nutrition supplement [Online]. Available at: https://cobbstorage.blob.core.windows.net/guides/3914ccf0-6500-11e8-9602-256ac3ce03b1. (Verified 5 July, 2020) - Danisman, R. & Gous, R.M. (2011)
Effect of dietary protein on the allometric relationships between some carcass portions and body protein in three broiler strains. South African Journal of Animal Science 41: 194-208. - Danisman, R. & Gous, R.M. (2013)
Effect of dietary protein on performance of four broiler strains and on the allometric relationships between carcass portions and body protein. South African Journal of Animal Science 43: 25-37. - Emmans, G.C. (1981)
A model of the growth and feed intake of ad libitum fed animals, particularly poultry. In: Hillyer, G.M., Whittemore, C.T. and Gunn, R.G. (eds) Computers in Animal Production. Occasional Publication No. 5. British Society of Animal Production, Edinburgh, UK, pp. 103-110. - Emmans, G.C. (1987)
Growth, body composition and feed intake. World’s Poultry Science Journal 43: 208-227. - Gous, R.M. (1986)
Measurement of response in nutritional experiments. In: Nutrient requirements of Poultry and Nutritional Research, C. Fisher and K.N. Boorman (Eds.) pp 41 57 British Poultry Science Symposium Number Nineteen, Butterworths. - Gous, R.M. & Morris, T.R. (1985)
Evaluation of a diet dilution technique for measuring the response of broiler chickens to increasing concentrations of lysine. British Poultry Science, 26: 147-161. - Gous, R. M., Fisher, C., Tůmová, E., Machander, V., Chodová, D., Vlčková, J., Uhlířová, I. & Ketta, M. (2019)
The growth of turkeys. 2. Body components and allometric relationships. British Poultry Science 60: 548-553. - Rostagno, H.S., Albino, L.F.T., Hannas, M.I., Donzele, J.L., Sakomura, N.K. & Costa, F.G.P. (Eds) (2017)
Brazilian tables for poultry and swine (Eds). (UFV: Viçosa) - Roush, W.B., Cravener, T. L. & Zhang, F. (1996)
Computer formulation observations and caveats. Journal of Applied Poultry Research 5, 116-125.
