Director of Technical Services
National By-Products, LLC
Des Moines, U.S.A.
Introduction
Mammalian meat and bone meal (MBM) has been successfully incorporated into poultry diets for decades. In fact, of all segments of animal husbandry, the poultry industry has been the largest user of MBM. Although its successful use has been dependent upon its economical contribution of available nutrients such as essential amino acids, calcium, phosphorus, and energy, it has historically been traded based solely on its protein content. More recently, inquiries about its nutrient content, availability and variability, oxidative stability, and safety have emerged. This paper will explore how individual companies and the rendering industry have addressed these queries, how these initiatives impact the value of MBMs, and suggestions for assessing and purchasing good quality MBM.
Nutrient content, variability, and relative value
The 1994 edition of Nutrient Requirements of Poultry reports a complete list of the nutrient contents of MBM and meat meal (MM). These two products are distinguished by their protein, calcium, and phosphorus levels where MM contains more protein and less bone. This is consistent with the definitions for these two products created by the American Association of Feed Control Officials. MBM must exceed 4% phosphorus content while MM has no phosphorus minimum. The feed industry typical trades the 50% MBM product. What may surprise many seasoned veterans of animal agriculture is the fact that renderers do not typically produce a MBM containing 50% protein. The protein content is entirely dependent upon the proportion of soft tissues and bone that comprise the rendered mix. Protein content of MBM/MM can range from 40% to 65% depending upon the proportion of these mammalian tissues. A survey of 11 Midwest MBM/MM rendering sources illustrates this range of proteins (Figure 1). The average for each source represent 15-16 samples.
Consequently, the first step in purchasing MBM/MM is to determine the preferred level of protein and mineral content. The best method for making this determination is by offering a number of MBM/MM sources with their nutrient specifications and prices to a diet formulation process to assess their value relative to each other and competing protein and mineral sources. This can be a powerful tool for identifying the best-valued sources of MBM or MM for your application. Table 1 contains three MBM/MM examples with their nutrient specifications that could be included in such an evaluation.
This table illustrates that as the bone content (ash, calcium, phosphorus) decreases, the protein, energy, and amino acids increase. Moreover, the quantity of essential amino acids increase as does their proportion in the protein.
Literally there are thousands of potential diet formulations for layers in egg production that could be envisioned that differ by bird strain, feed intake, stage of production, nutrient requirements, and nutrient contributions from feed ingredients. For the following comparison of relative MBM/MM values, a series of diets were created. The first step was to design a simple corn:soy layer diet with the following nutrient requirements (Roland, 2004): lysine 0.92%; cystine:lysine ratio of 0.75; calcium 4.0%; available phosphorus 0.45%; sodium 0.19%; and metabolizable energy of 2816 kcal/kg (Figure 2).
Ingredient costs assumed a delivered price to Central Iowa on January 14th, 2005. The amino acid profile of this solution provided the minimum requirements for the subsequent solutions when the individual MBM/MMs were offered. Table 2 summarizes the outcomes.
In the solution that included 57% MM, it remained in the diet at 210 lbs/ton until it reached a price that was 116.9% of the soybean meal (SBM) price. For 50% MBM, it remained in the diet at 197 lbs/ton until it reached a price that was 107.6% of SBM. For 44% MBM, it remained in the diet at 127 lbs/ton until it reached a price that was 108% of SBM. The lower inclusion of the 44% MBM is due to its heavier content of minerals and the upper limit of calcium set for the diet. The different price points where the MBM/MMs dropped from the formula also represented their relative value to each other.
In this example, the 57% MM had a relative value that exceeded both MBMs by approximately 108%. Consequently, the outcome of this exercise would encourage the use of 57% MM over the two MBMs, while all MBM/MM sources provided cost savings over a corn:soy diet that excluded MBM/MM. Such information aids buyers in making better-informed purchasing decisions. Reputable suppliers of MBM/MMs should be able to provide sufficient nutrient information such as amino acid, calcium, phosphorus, and energy content to support such an evaluation.
Further examination of the amino acid profiles for the eleven MBM/MM sources helps explain why the 57% MM was favored over the two MBMs. Figure 3 expresses the essential amino acid content for each source as a percent of the crude protein as well as showing the ash content of each source.
This data illustrates a strong inverse relationship (r2 = 0.79) between contents of bone and essential amino acids in MBM/MM. In other words, one can expect a greater proportion of essential amino acids in the protein of a MM coming from soft tissues than in the protein of a MBM coming from bone. Using broiler chicks, Johnson (1997) showed that MBMs with greater bone content supported less protein gain per unit of protein intake (PER). It was determined this was not due to a dilution in energy by the ash content, but rather by a poorer amino acid profile. Using cecectomized roosters, Johnson (1998) also determined there was no significant relationship between the amino acid digestibilities and the ash content further substantiating that the lower PER values were not due to poorer digestibility but due to the protein containing a greater proportion of non-essential amino acids.
A sense for nutrient variability is also important. Ingredients having consistent nutrient contents permit nutritionists to formulate less expensive diets having smaller safety margins. Since each rendering facility has its own unique composition of raw materials, some variation in the nutrient contents of the resulting MBM can be expected. Nevertheless, this variability within facilities should not be excessive. The notion that MBM/MMs are highly variable products comes from comparisons made across plants producing very diverse products. Indeed, the CV for protein across all eleven rendering sources above was large at 11.2% (Figure 4). However, the CVs for protein within each facility were much smaller ranging from 2.8% to 5.0%.
These values are supported by other data reported by Kirby (1993) where 264 samples of MBM shipped to four broiler-feed manufacturing sites in Georgia and Kentucky gave protein CVs of 3.8%. Knowledge of nutrient variability generated through repeated sampling by the supplier or buyer is indispensable to further help identify consistent suppliers.
Often buyers may be looking for more product than can be purchased from a single rendering facility. In these cases it is important to identify a number of suppliers producing MBM/MMs having similar nutrient profiles. Among the eleven rendering sources found in Figure 1, sources 1, 3, 8, 9, and 11, could be selected as potential sources having similar protein content. In fact, their average in this survey was 55.6% with a combined low CV of 4.3%.
Protein blending facilities often have an additional advantage in managing the nutrient variation in MBM/MMs. Some of these facilities measure the protein on the in-bound material and then segregate it based on its content. This creates the ability to blend MBM/MM sources to a targeted value. Because the protein and mineral contents of MBM/MMs have a strong negative correlation, any segregation of inbound product on the basis of protein also causes a corresponding segregation of phosphorus and calcium. The result is a blended MBM with reduced nutrient variations. A study by Kirstein (2002) of two Midwest protein-blending facilities included 663 samples. The proteins from these facilities averaged 54.8% and 51.8% (Figure 5).
Their protein CVs were 1.9% and 2.1% and were lower than those generated by the individual MBM/MMs from which they were produced (Figure 4). They were also comparable to those found in dehulled soybean meal (SBM). Studies reported by Kirby (1993) and Cromwell (1999) found that sources of SBM they each surveyed to have similar CVs of 1.7 % and 1.6%, respectively (Figure 5). Blending of MBM/MM can be a valid approach to minimizing nutrient variation. Renderers utilizing this approach should be strongly considered as potential suppliers.
Recommended tolerance for peroxide value in meat and bone meal
Rancidity refers to the oxidative state of fats, which is a characteristic that may have nutritional relevance and, in extreme cases, may reduce the feed consumption and growth rate of poultry and livestock. The relevance of lesser-oxidized fats is poorly understood and defined.
Oxidative rancidity is a complex process that is thought to occur in phases: (1) initiation, (2) auto-oxidation and (3) termination. During each phase, the formation of products increase and decrease over time, as shown in Figure 6.
Hydroperoxides form when oxygen and unsaturated fatty acids combine in the presence of a catalyst (such as iron, copper, heat, light, enzymes, etc.) during the initiation phase. These peroxides are reactive and can combine with other fats to form additional reactive products during auto-oxidation. Hydrocarbons, aldehydes and ketones are formed during the termination or final phase. These compounds are volatile, but relatively unreactive. These compounds, and not peroxides, are largely responsible for the characteristic odor of rancid fat. The human threshold for detecting these compounds in soybean oil that is undergoing the initial stages of oxidation correlates with peroxide measurements of around 10 meq/kg. For fats or oils that also carry other aromatic compounds that can mask these odors it is around 20 meq/kg for olive oil and 20-40 meq/kg is estimated for rendered animal fats (Warner, 2005). N-PAL Laboratories (2005) describes the various methods used to predict or interpret the rancidity of fats.
Of these methods, peroxide value (PV) is the most widely used indicator of fat oxidation. A peroxide value is required only for USDA certified edible animal fats, such as tallow or lard. However, the feed industry also uses PV to assess the stability or rancidity of fats used as feed ingredients, by measuring lipid peroxides and hydroperoxides formed during the initial stages of oxidation. It should be noted that difficulties in the titration end point are common for low PV levels, which may account for a portion of the high variability generally associated with the method.
Determining Peroxide Value
Peroxide value is most often determined using the AOCS Official Method Cd 8-53 (same as AOAC Official Method 965.33). An alternative procedure that is sometimes used is AOCS Official Method Cd 8b-90, which uses isooctane instead of chloroform as a solvent. These procedures all measure the milliequivalents of peroxide per kilogram of fat (meq O2/kg). In order to determine the PV on a sample of protein meal (such as meat and bone meal) it is necessary to first extract the 5 grams of fat needed for the procedure.
Unfortunately, there is no standard extraction procedure for obtaining fat from protein meals on which to subsequently perform a PV test. One area of disagreement is the solvent used for the extraction. Some of these solvents include hexane, hexane:methanol, and pet ether. One could argue that the hexane:methanol solvent provides the best mixture to most efficiently extract tri-, di-, and mono-glycerides as well as more polar free fatty acids. However, the use of a standard solvent has not been established. PV results can be expected to vary when comparing fats extracted from meals using different solvents.
Another important issue involves how the solvent is evaporated from the fat. It is critical the solvent is evaporated without using elevated temperatures and doing so under a stream of nitrogen gas. Overlooking this step in the extraction process will itself cause a rise in the PV. Laboratories in the United States have not universally adopted this step.
It is not surprising that The American Oil Chemists' Society (AOCS) notes that the official PV method is "highly empirical, and any variation in the test procedure may result in variation of the results" (AOCS, 1997). Variation in the sample size of extracted fat appears to be a good example of this. Although the method calls for a standard 5 grams of fat, when labs run short on meal volume, they may run the PV test on whatever amount of fat they were able to extract. Other results from National By-Products' (NBP) lab appeared to confirm AOCS's warning that comparing samples of different sizes will cause variation in the results.
Given the empirical nature of the PV test and no official method for extracting fat from protein meals, it is important that buyers and sellers agree upon the methodology. Prior agreements should be made to have involved laboratories follow identical protocols. Arrangements should also be made for a referee lab to follow the same procedure in order to arbitrate any conflicting results.
Interpreting Peroxide Value
Nutritionists and buyers have arbitrarily established maximum initial PV levels of between 5 and 20 meq O2/kg of fat as acceptable. The origin of these standards is unknown. Carpenter et al. (1966) suggested the source to be from the footnote to a table in a paper published in 1941 (Gray and Robinson). The footnote gave analytical data for a series of meat meal samples and included the comment: "a fat with a peroxide value of more than 20 is definitely rancid", even though they concluded later in the paper that such "rancid" meat meals may be fed to animals without harm.
Lipid chemists consider a maximum of 5 meq O2/kg of fat as too low for production animal diets. Even human snack foods have been shown to contain >10 meq O2/kg of fat (Warner, 2005).
In fact, poultry and livestock are relatively resistant to all but excessively oxidized fats. Peroxide values up to 100 meq O2/kg of fat did not depress feed consumption or growth rate in broiler chickens (L'Estrange et al., 1966; Cabel and Waldroup, 1988) or turkeys (Lea et al., 1966). This is illustrated in Table 3.
Pesti et al. (2002) fed diets containing either 3 % or 6 % poultry fat, restaurant grease, choice white grease, an animal/vegetable blend, palm oil, yellow grease or edible soybean oil to broiler chickens. All fat sources had similar initial PV levels, but their oxidative stability, measured by the Active Oxygen Method, varied from 2 to 370 meq O2/kg of fat. Fat source and oxidative stability did not affect growth rate, feed consumption or feed efficiency of the birds in their study. The authors further concluded that rancidity indicators such as peroxide value, are not correlated (P>0.60) with broiler chick performance.
Because MBM/MMs can contain between 8 and 14 % fat, some nutritionist and regulatory officials have been concerned about their oxidative status. Only the fat portion is subject to oxidation. As with using pure fat, animals are quite resistant to the presence of peroxides in the fat of MBM/MMs. Growth rate, feed consumption and feed efficiency were not affected when broiler chickens were fed poultry byproduct meal (Kirkland and Fuller, 1971) or fish meal (Lea et al., 1966) containing oxidized fat having peroxide values of from 32 to 110 meq O2/kg of fat.
Establishing an acceptable tolerance for peroxides in MBM/MM
Oxidized poultry fat (PV of 175 meq O2/kg of fat; Cabel and Waldroup, 1988) and oxidized vegetable oil (PV of 156 meq O2/kg of fat; Engberg et al., 1996) reduced the growth rate of broilers when included at 5 % and 11 % of the diets, respectively. Based on the inclusions of fat in these diets, the negative responses occurred when peroxides totaled 8.75 and 17.2 meq O2/kg of complete feed.
In two other studies, peroxide levels of 5.0 (Cabel and Waldroup, 1988) and 5.45 meq O2/kg of complete feed (L'Estrange at al., 1966) did not affect broiler performance and were considered safe levels.
The studies sited in this paper suggest that diets containing 4 meq O2/kg of diet or less will not effect poultry performance and will provide a safety margin. NBP recommends that a maximum PV tolerance of 80 meq O2/kg of fat be established for MBM/MM. This level will provide more than an eight–fold safety margin as shown in the example below:
Dietary inclusion rate for MBM/MM = 5 %
Level of fat in MBM/MM = 12 %
Level of fat in diet coming from MBM/MM = 0.6 %
Fat level (0.6 %) * PV (80 meq O2/kg of fat) = 0.48 meq O2/kg of feed.
4.0/0.48 meq O2/kg of feed = safety factor of 833 %
Recommendation for stabilizing MBM/MM
The concern of feeding moderately oxidized fats may be of less importance today than the time Gray and Robinson (1941) made their comments about rancid fats because of advances made in commercial antioxidants and fat-soluble vitamins. Adding 125-ppm ethoxyquin alleviated the effects on broiler performance when they were fed fat having 175 meq O2/kg (Cable and Waldroup, 1988).
Nevertheless, preventing more excessive oxidation of this fat is still important. MBM/MM naturally creates a pro-oxidant environment for its residual fat content due to the minerals present and the large surface area that is exposed to oxygen. Generally cattle fat is fairly stable and slow to oxidize since it contains few poly-unsaturated fatty acids to serve as a substrate. However, pork fat can contain as high as 14% linoleic acid. So it is important that MBM/MM suppliers stabilize the residual fat with antioxidants proven to provide protection in this type of matrix. Most importantly, the stabilizer should contain an effective chelator that will render pro-oxidant metals like iron and copper ineffective as catalysts (Wagner, 2005). MBM/MM suppliers and antioxidant manufacturers should be able to demonstrate the effectiveness of the stabilizer they have chosen and the rate at which it is being applied.
Some performance reductions observed in commercial feeding situations that are attributed to fat stability may in fact have their roots in over-supplementation of the diet with copper or iron. Special attention should also be paid to antioxidant levels and fat-soluble vitamin levels in the final feed whenever fat sources are used in diets fortified with elevated levels of trace minerals and stored for prolonged periods in hot climates.
MBM safety
A discussion about the many government, industry, and proprietary company initiatives regarding MBM/MM safety could easily be the subject of another paper. A few steps that progressive inedible renderers have taken to ensure the safety of their products should be noted here. One of those steps involved the voluntary design and implementation of process control plans. These plans are designed to address physical, chemical, or biological risks that, after examination, have been identified as potential hazards. Pesticide residues and food-born pathogens are common hazards usually addressed by these plans. These renderers have also adopted a set of prerequisite programs designed to create an environment where the process control plans can effectively function and safe products can be produced. Prerequisite programs are commonly identified as Good Manufacturing Practices (GMPs), Standard Operating Procedures (SOPs), or Standard Sanitary Operating Procedures (SSOPs). To further encourage the uniform production of safe rendered products, the Animal Protein Producers Industry (APPI) adopted an industry-wide Code of Practice in 2004. The Code of Practice sets minimum industry practices and an accreditation process to enhance consumer confidence and facilitate domestic and global trade. The Facility Certification Institute conducts the third-party audits and issues the certifications. While MBM/MM suppliers may not be willing to share copies of their programs, since many are considered proprietary, they should be willing to discuss what hazards they address and the outcomes of those efforts.
One area commonly discussed concerns the biological safety of MBM/MM. It is well known that food born bacteria are common inhabitants in meat producing animals regardless of their health status. So it is critical that the rendering process effectively pasteurize those raw animal by-products when converting them into proteins and fats. In 1999-2000, the prevalence of Salmonella, Coliforms, Listeria monocytogenes, Campylobacter jejuni, and Clostridia perfringens in animal proteins was studied in the raw material, the cooked solids, and the solids at load-out (Troutt, 2004). Seventeen Midwest rendering plants were sampled during the summer and winter seasons. As expected, the raw materials were contaminated with the five food born pathogens used as the index organisms (Table 4). However, following the cooking process, virtually all of the index organisms were undetectable in the cooked solids (Table 5). As with all feed ingredients, some environmental recontamination occurred with the conveying and storage of the solids when they were sampled at load-out (Table 6). This study verifies that the rendering process is highly effective in inactivating food born pathogens, even the more heat resistant, spore-forming Clostridia perfringens organism.
Environmental contamination is a feed ingredient issue, not an animal protein issue. A survey conducted by the Food and Drug Administration (FDA, 1993) of animal and vegetable protein sources for Salmonella contamination showed a 56.4% and 36.0% frequency of positives, respectively. The data clearly showed that all types of protein meals could be contaminated with Salmonella. As a result, the FDA stated that the removal of MBM/MMs from animal diets is not a good strategy for reducing the incidence of Salmonellosis (FDA, 1993). A similar survey by the Canadian Food Inspection Agency (CFIA, 1999) confirmed those findings. The frequency of Salmonella positives for oilseeds, animal by-products, fishmeal, and feed grains were 18.0, 20.5, 22.0, and 5.0%, respectively.
Environmental recontamination of feed ingredients including MBM/MMs should not be surprising due to the ubiquitous nature of Salmonella. However, for those samples of MBM/MM that do test positive, the levels are very low. The median Most Probable Number (MPN) of Salmonella organisms found in a survey of 197 MBM/MMs was 0.09 cfu/g (Franco, 2005). Over 75% of the samples contained < 1.0 cfu/gm. The geometric mean was 0.24 cfu/gm. By comparison, the USDA baseline geometric means for Salmonella in raw ground chicken and turkey were 1.27 and 2.63 cfu/gm, respectively (USDA-FSIS, 1996).
It is also important to remember that not all of the 2,300+ Salmonella serovars currently known to exist are associated with disease in animals or humans. In fact, only 30-40 serovars are routinely reported to have clinical significance (Franco, 1997). To further investigate the relevance of serovars in animal proteins, the serovar identities were also determined for the same samples used in the MPN study (Franco, 2005). The study found only four serovars of clinical interest: S. Typhimurium, S. Enteriditis, S. Infantis, and S. Agona. Ninety-two percent of the serovars had no pertinence to animal disease. These findings agree with the results of earlier surveys of animal proteins in the U.S., Japan, and the United Kingdom (Franco, 2005). Moreover, experts in veterinary medicine support this view. Poor correlations between serovars detected in feed and serovars from diseased pigs continue to be a regular observation in field studies (Davies, 2004).
Although feed sources can be one of many potential sources of Salmonella introduction to farms, the risk of infection from non-feed sources appears to greatly exceed the risk presented by feed (Davies, 2004). Studies of two modern US swine multiplier sites confirmed that feed played an insignificant role in Salmonella contamination (Harris, 1997). Similarly, a European study reported no correlation between contamination of feed and Salmonella seroprevalence (Davies, 1997).
Take home messages
Feed formulation software remains the best tool for assessing the value of MBM/MM sources relative to each other, as well as to competing sources of amino acids, minerals, and energy. MBM/MM suppliers should be able to provide necessary nutrient profiles complete with variability estimates to aid in this process. There are a number of strategies buyers of MBM/MM can employ to minimize product variation. MBM/MMs containing less bone have a greater proportion of essential amino acids.
Arbitrarily low PV limits of 5-10 meq O2/kg of fat in MBM/MM are too low. Moderately oxidized fats do not compromise poultry performance. The total PV from all dietary fat sources should not exceed 4 meq O2/kg of feed. Even MBM/MMs with fat residues measuring 80 meq O2/kg of fat contribute only a small portion towards that total. When limits are established, buyers and sellers should agree to a common testing methodology. Sensory assessment may be used to help interpret quantitative measurements. Stabilizers are recommended to prevent excessive oxidation of the residual fat in MBM/MMs.
The rendering process effectively pasteurizes MBM/MM. Environmental recontamination of feed ingredients by food born pathogens, like Salmonella, is a common occurrence. Therefore, whatever criteria are used for assessing the biological safety of the major dietary components, like corn and soybean meal, should also be applied to the other smaller dietary constituents like MBM/MM.
References
AOCS, 1997. The Official Methods and Recommended Practices of the American Oil Chemists’ Society, 4th edition. AOCS Official Method Cd 8-53: Peroxide Value – Acetic Acid-Chloroform Method, Re-approved 1997.
Cabel, M. C. and P. W. Waldroup, 1988. Poultry Science. 67:1725-1730.
Carpenter, K. J., J. L. L’Estrange and C. H. Lea, 1966. Proc. Nutr. Soc. 25:25-31.
Cromwell, G. L. et al., 1999. Journal of Animal Science 77:3262-3273.
Davies, P. R., W. E. M. Morrow, F. T. Jones, J. Deen, P. J. Fedorka-Cray, I. T. Harris. 1997. Epidemiol. Infect. 119:237-244.
Davies, P. R., 2004. Proceedings from the CDC Animal Feed Workshop, Atlanta, GA.
Engberg, R. M., C. Lauridsen, S. K. Jensen and K. Jakobsen, 1996. Poultry Science 75:1003-1011.
Franco, D. A., 1997. National Renderers Association Sanitation Handbook.
Franco, D. A., 2005. J. of Environmental Health, Vol 67, No. 6, p. 18-22.
Gray, R. E. and H. E. Robinson, 1941. Poultry Science 20:36.
Harris, I. T., P. J. Fedorka-Cray, J. T. Gray, L. A. Thomas, 1997. JAm Vet Med Assoc 210:328-385.
Johnson, M. L. and C. M. Parsons, 1997. Poultry Science 76:1722-1727.
Johnson, M. L., C. M. Parsons, G. C. Fahey Jr., N. R. Merchen, and C. G. Aldrich, 1998. Journal of Animal Science 76:1112-1122.
Kirby, S. R., 1993. Poultry Science. 72:2294-2298.
Kirkland, W. M. and H. L. Fuller, 1971. Poultry Science. 50:137-143.
Kirstein, D. D., 2002. Proceedings from the II Latin American Rendered Products Nutrition Conference, Guadalajara, Mexico.
L’Estrange, J. L., K. J. Carpenter, C. H. Lea and L. J. Parr, 1966. British Journal of Nutrition 20:113-122
Lea, C. H., L. J. Parr, L’Estrange and K. J. Carpenter, 1966. British Journal of Nutrition 20:123.
N-PAL, 2001. N. P. Analytical Laboratories. http://www.ralstonanalytical.com/npal2/.
Pesti, G. M., R. I. Bakalli, M. Qiao and K. G. Sterling, 2002. Poultry Science. 81:382-390.
Roland, D. A., 2004. Dept of Poul Sci. - Auburn University. Personal communication.
Troutt, F. H., 2004. Proceedings from the CDC Animal Feeds Workshop, Atlanta, GA.
USDA-FSIS, 1996. National Raw Ground Chicken or Turkey Microbiological Survey
www.fsis.usda.gov//Frame/FrameRedirect.asp?main=/OPHS/haccp/salm6year.htm
Warner, Kathleen, 2005. National Center for Agricultural Utilization Research – USDA, Peoria, IL. Personal communication.
From Proceedings of the “Midwest Poultry Federation Convention”, St. Paul, Minnesota, U.S.A.
