Site icon aviNews International, poultry information

Estimating variation in mixed feed

Conteúdo disponível em: Indonesia (Indonesian) Philipino

A major goal of poultry nutritionists and feed millers is to ensure that each bird receives the nutrients it needs daily. To do that, the feed must be fairly uniform and contain adequate amounts of each required nutrient.

Amy Moss and her team (2021) at the University of New England in Australia showed that if you overestimate the amount of nutrients in feedstuffs, you could lose 63% of your profit or $635,100 for every million broilers.

The variability in feed ingredients is caused by raw ingredients, sampling, and analysis, according to Moss et al. (2021).

FEEDSTUFF VARIABILITY

Once feeds are mixed, they must be adequately sampled to ensure that batches contain what they are expected to. Because variation is inherent in different batches with the same ingredient formula, multiple samples of each batch are necessary to estimate the mean.

The variances of mixtures are calculated as follows from the variances of the ingredients.

Suppose Xi is a feed ingredient that follows a Normal distribution with the mean μi and variance σ2 i, N (μi,σi2), i = 1, . . . , k, and suppose Xi’s (feed ingredient nutrient composition) are independent. Then:

Follows a Normal distribution with mean μ and variance σ2, N(μ,σ2), where:

How variable are nutrients in a feed?

The Microsoft Excel workbook called “FeedVariation.xlsx” was designed by Dr. Pesti to use these formulas. It is available from the Poultry Hub Australia web page under “Research Resources”.

Figure 1 shows a portion of the “Protein Example” worksheet. The ingredients, with their average protein levels and standard deviations, are from samples collected from Australian producers and compiled in the Australian Feed Ingredient Database (AFiD).

Figure 1. A portion of the Microsoft Excel workbook called “FeedVariation.xlsx” shows the formulas to calculate the variation of a mixed feed from the reported variation in the ingredients.

In the right center portion of Figure 2 are some formulas for feeds for different classes of chickens and turkeys, and there are more on the actual worksheet.

If many batches of Broiler Starter, fed generally from 0 to 10 days, were mixed from random samples of Australian ingredients, the average crude protein levels of the feeds would be expected to be 230 g/kg CP. Half of the batches would be expected to contain more and half less than 230 g/kg CP.

Figure 2. A portion of the Microsoft Excel workbook, “FeedVariation.xlsx,” shows the variation in the crude protein of mixed feeds based on the reported variation in Australian ingredients.

The normal distribution (Figure 3), defined by the mean and standard deviation, could be used to estimate the distribution of batches of feed.

Thirty-four percent of batches of this feed would contain between 230 and 230 – 4.48 = 225.52 g/kg CP; 13.5 % of batches of feed would contain between 225.52 and 225.52-4.48 = 221.04; and 2.5 % of batches of feed would contain less than 221.04 g/kg CP.

Poultry producers often purchase ingredients from the same supplier, so the variation in some ingredients may be less than expected from the AFiD database.

Nonetheless, this analysis points out the importance of monitoring ingredients to decrease variation as much as possible.

Figure 3. The Normal Distribution

Analytical variability and NIRS

Energy content, CP, and AA digestibility are rarely determined for distinct batches of feedstuffs in feed mills.

NIRS

For more than thirty years, the feed industry has had an alternative to tracking nutrient composition in feed ingredients. This uses near-infrared reflectance spectroscopy (NIRS), but not everyone agrees with using the data from NIRS. Wet chemistry results are still considered the most reliable in many places.

The NIRS analysis offers several advantages, including:

These factors make NIRS analysis more sustainable than the wet chemistry analysis.

 

There are two ways to make NIRS calibration curves: direct and indirect.

Numerous research groups have evaluated the precision and accuracy of NIRS calibration models for predicting feedstuff nutritional value, yielding results comparable to those obtained through lab wet chemistry and in-vivo approaches.

CONCLUSION

 

PDF
Exit mobile version