Manual of quality analysis - Soya products

MANUAL
of QUALITY
ANALYSES
for SOYBEAN
PRODUCTS
in the FEED
INDUSTRY

MANUAL
of QUALITY
ANALYSES
for SOYBEAN
PRODUCTS
in the FEED
INDUSTRY
J.E. van Eys(1), A. Offner(2) and A. Bach(3)

(1) Global Animal Nutrition Solutions Inc., Corresponding author.

24 Av. de la Guillemotte, 78112 Fourqueux, France ; Jvaneys@cs.com.
(2) Cybelia, 104, Avenue du président Kennedy,
75781 Paris cedex 16, France ; anne.offner@cybelia.fr
(3) ICREA, IRTA-Unitat de Remugants, Edifici V,
Campus Universitari de Bellaterra, 08193 Bellaterra, Spain ; alex.bach@irta.es

CONTENTS
1. Introduction 5
2. Soybeans, soybean products and production processes 7
3. Definitions and applications of soybeans and soy products 9
4. Chemical and nutritional composition of soybean products 17
5. Official standards of some soybean products 21
6. Sampling soy products 27
7. Physical evaluation and equipment 30
8. Chemical analyses 32
8.1. Moisture 32
8.2. Ash 34
8.3. Protein 34
8.4. Protein quality 37
8.4.1. Urease Index 38
8.4.2. KOH protein solubility 39
8.4.3. Protein Dispersibility Index (PDI) 40
8.4.4. Protein quality in ruminants 42
8.4.4.1. In situ technique 42
8.4.4.2. In vitro technique 43
8.5. Amino acids 44
8.6. Crude fiber 45
8.7. Neutral Detergent Fiber (NDF) 46
8.8. Acid Detergent Fiber (ADF) 48
8.9. Lignin 49
8.9.1. Klason lignin 49
8.9.2. Permanganate lignin 50
8.10. Starch 51
8.10.1. Polarimetric starch determination 51
8.10.2. Enzymatic or colorimetric starch determination 53
8.11. Non Starch Polysaccharides (NSP) and Monosaccharides 55
8.12. Ether Extracts 57
8.13. Lipid quality 57
8.13.1. Moisture 58
8.13.2. Insoluble impurities 58
8.13.3. Unsaponifiable matter 59
8.13.4. Iodine value 61

Contents

8.13.5. Acid value 62
8.13.6. Lipid oxidation 62
8.13.6.1. Peroxide value 64
8.13.6.2. Thiobarbituric acid (TBA) 65
8.13.6.3. Anisidine value 65
8.13.6.4. Lipid stability tests 66
8.13.6.4.1. AOM (Active Oxygen Method) 66
8.13.6.4.2. OSI (Oil Stability Index) 66
8.13.7. Fatty acid profile 67
8.14. Minerals 68
8.14.1. Calcium 68
8.14.2. Phosphorus 69
8.14.3. Sodium chloride 70
8.15. Isoflavones 71
8.16. Antinutritional factors 72
8.16.1. Trypsin inhibitors 72
8.16.2. Soy antigens 74
8.16.3. Lectins 75
8.17. Mycotoxins; rapid tests 77
8.17.1. Ochratoxin 78
8.17.2. Zearalenone 78
8.17.3. Fumonisins 78
8.17.4. Aflatoxins 79
8.17.5. Deoxynivalenol 79
8.18. Genetically Modified Organisms (GMO) 80
9. NIR analyses 82
10. Data management 87
10.1. Sample statistics 87
10.2. Quality indicators 91
10.3. Significance of parameter estimates 93
10.4. Control charts 96
10.5. Follow-up and application of analytical results 98
11. References 100
12. Annex 106

1. INTRODUCTION
The use of soybean products in the feed and food industry has increased
steadily over the past decennia. Fifty years ago world soybean production was
estimated to be 17 million tons with China being the major producer (UNEP,
1999). A little more than 50 years later and the production for 2003 is expected
to reach more than 190 million tons with the major centers of production being
the USA, Brazil and Argentina (USDA, 2003). The USA remains the largest
producer of soybeans and soybean meals but its production is leveling off
while Brazilian production and crushing of beans is increasing rapidly.

Of the total world production of soybeans, less than 10 % is directly used for
human consumption. The overwhelming majority is used in animal feed in the
form of various types of soybean meals or specialty soy products. The current
world production of soybean meal is estimated to be in excess of 130 million mt
(USDA, 2003). With global animal feed production estimates approximating
1.100 mt (Speedy, 2002), and compound feed production well above the
600 million mt (Gill, 2003), soybean meals represent the dominant source of
protein in animal diets. However, total use and importance of soybeans or
soybean products is likely to be higher than indicated by major statistics as a
plethora of different soybean products are entering the feed and food chain.

This dominant position of soybeans and their products is no doubt
associated with their high quality especially with respect to protein and amino
acid profile. Following proper treatment or extraction, digestibility of the protein
fraction is high and the amino acid profile provides a close match with cereals
to meet animal requirements. Nevertheless, in their untreated form, soybeans
contain a number of factors that have the potential to seriously diminish their
nutritive value - to the point of decreasing animal performance and health
(Liener, 2000). A treatment of soybeans to eliminate these anti nutritional
factors (ANF) is thus necessary especially in the case of monogastric diets.
These treatments, combined with varietal differences in the production process
of soybean meals or other products lead to potentially large variations in quality.

While basic standard specifications for soybean meals have been established
(NOPA, 1997) no official specifications exist for other soy products that are
routinely used in the feed industry. Furthermore the NOPA specifications only
refer to four chemical characteristics. Current evaluations of soy products are
based on a much larger array of tests allowing a more accurate evaluation of the

5

1. Introduction

nutritive value of the different products. However, under practical conditions of
feed production the choice of tests differ greatly among producers and feed
compounders and not all tests are applied on a regular basis (West, 2002).
It is most likely that in the future more analyses of greater complexity will be
needed. Developments in the technological modification of soybean products,
along with a better understanding of the effects on performance and health
of relatively unknown compounds, such as isoflavones, will add value to soy
products. Accurate analysis to measure the effects of new treatments and the
relatively unexplored compounds will be of great importance.

In order for results of quality tests to have real value and to be comparable
between producers it is important that tests are standardized in method as well
as equipment. This standardization is becoming increasingly important as trade
in soybean products grows more global and competition amongst suppliers
increases. Identity preservation and traceability associated with detailed quality
characterization are issues of major importance in the (future) trade of soy
products. Accurate and consistent quality procedures and analyses along
with precise descriptions of the product are necessary. These tests must be
reproducible at different levels of the supply chain. Furthermore, the increasing
demands of the implemented quality systems (HACCP, ISO or GMP) will dictate
the establishment of more detailed quality procedures and a larger analytical
capacity. For the information that is generated at the various production
stages to be consistent and comparable it is important that a single reference
is available.

This quality manual intends to provide clear directives and explanations for
the quality analysis needed at all stages of the soy protein supply chain in the
feed industry. The objective is to supply information that is applicable at all
levels of operation, from the crusherto the compounder and from the quality
operator in the plant to the nutritionist. Applications of the methods and
analyses presented will enhance the value of soy products through improved
knowledge and application resulting in improved performance and health.

CONTENTS

6

2. SOYBEANS,
SOYBEAN PRODUCTS and
PRODUCTION PROCESSES
A large number of soybean varieties exist, producing soybeans that vary greatly
in shape and color. For the complete range of soybeans shapes vary from flat to
spherical and colors range from yellow to green, brown and black. Modern varieties,
mainly grown for their oil content, are generally spherical in shape with a yellow or
green as the accepted seed coats. These characteristics logically will affect many of
the soybean products obtained from these beans. Official limits have been set on
the minimal size requirements for the beans (see below) but generally soybeans
grown for industrial purposes will weigh between 18 – 20 g per 100 beans.

The soybean consists of two cotyledons which represent approximately 90 % of
the weight, a seed coat or hull (8% of weight), and two much smaller and lighter
structures the hypoctyl and the plumule. The cotyledons contain the proteins and
lipids (oils) that constitute the main nutritional components of the soybean products
obtained from soybeans. They are also the main storage area for the carbohydrates
and various other components of importance, most notably the enzymes
(lipoxygenase, urease) and the ANF. The various soybean products are obtained
through the separation or extraction of the different component of the soybean.

A large array of different manufacturing processes is applied to obtain the many
soy products used in animal and human nutrition (Berk, 1992). Figure 1 provides a
schematic representation of the transformation from soybean into the various
products. In the “crushing” process of soybeans, which includes a series of
preparatory operations, crude oil is obtained as a major product. The crude oil is
refined and separated into lecithin and refined oil used in human as well as animal
nutrition; especially in young animal diets.

The soybean meals, which on a volume basis are the most important products
obtained from soybeans, have the defatted flakes as an intermediary product that
requires further treatment. Two main processes are used to extract the oil and obtain
the defatted flakes: the expeller process (mechanical extraction of the oil by a screw
press) or solvent extraction where non-polar solvents (commonly hexane and
hexane isomers) are used to extract the oil. Solvent extraction is the most efficient
and widely used process at present. In the case of solvent extraction the flakes are
desolventized. All flakes are toasted in order to eliminate the heat-labile anti
nutritional factors. Sometimes the hulls obtained in the preparatory steps are added

7

2. Soybeans, Soybean Products and Production Processes

back to the toasted flakes. This is done in variable degrees resulting in soybean
meals with variable levels of fiber and crude protein. When no hulls are added the
high protein meals are obtained. These are the meals used predominantly in poultry
diets. Flash desolventization or heat vacuum drying of the defatted flakes produces
the white flakes that are higher in protein quality (solubility) and do not have the
undesirable darker color. Through a series of different extraction and precipitation
process soy protein isolates (SPI) or soy protein concentrates (SPC) are produced.
Whereas SPI production is fairly standardized, different methods of extraction are
used to obtain the SPC resulting in slightly different compositional characteristics.
SPC but also the white flakes can be further elaborated (grinding, texturizing;
separation on basis of molecular weight) to obtain a large array of products used in
human nutrition. SPI and SPC are used in animal nutrition but are limited to specialty
diets due to the relatively high cost. The use of these ingredients in animal diets is
mainly as a replacement of high quality protein sources such as animal or milk
proteins or as a replacement of fishmeal in aquaculture diets.

Figure 1
Schematic representation of the manufacturing of soybean products

Drying & tempering*
Soybean cleaning, cracking, dehulling (optional), conditioning, flaking
cooking/toasting

Expelling or solvent extraction • soy hulls
• full-fat soy flour or grits

Crude oil Defatted flakes

• refined oil Desolventizing, toasting
• lecithin Soy hulls added (optional)
Flash de-solventization
or heat vacuum drying
Soybean meals

De-fatted soy flour or grits White flakes

Soluble carbohydrates Extraction Extraction, precipitation

• soy molasses soy protein soy protein
• isoflavones concentrates (SPC) isolates (SPI)

processes intermediary products

italics and green are final soy products * only in the case of solvent extraction

CONTENTS

8

3. DEFINITION and
APPLICATION of SOYBEANS
and SOY PRODUCTS
The number of soy products currently being used in the feed industry is large,
and an exhaustive review is hardly possible. Recent years have seen a dramatic
expansion of specialty products based on soybeans. Classical, commodity products
such as raw soybeans and soybean meals are relatively well defined with thorough
descriptions and specifications. This is not necessarily the case for some of the recent
modifications or adaptations of these products (i.e. Rumen-protected soybean meal)
or further elaborated products (i.e. soy concentrates). These evolved, value-added
products may differ significantly among producers with each producer applying
proprietary knowledge and specialized treatments. Typically, value–added products
must be evaluated on the basis of the entity that produces them taking into account
the guarantees provided by the manufacturer or distributor. Consistent analysis of
these producer-specific products allows classification and the building of a database
along with confidence about the product. This increased level of knowledge will
allow an analysis schedule of decreased intensity and increase inclusion rates in diets.

Commodities as well as the value-added products can be classified in a specific
class or group of products for which a sufficiently specific description can be
developed. For an efficient and correct use - as well as a meaningful interpretation
of analytical results - a precise and generally agreed upon definition of the product
is needed. Trading, purchasing, formulation, and the entire operation of feed
manufacturing depend on the precise referencing of a raw material and the
consistent use of the correct name and description. Also the quality control
mechanisms that have been introduced in the feed industry require a precise
description and classification for all ingredients.

Although many databases and ingredient tables have their own classification
system, the most widely recognized system is probably the IFN system (International
Feed Name and number) (INFIC, 1980). In this system, ingredients have been divided
into eight fairly arbitrary feed classes on the basis of their composition and use (NRC,
1982). The system is widely used in the UK, the US and in Canadian feed composition
tables but less so in other countries.

In the IFN system, ingredients are assigned a six digit code with the first digits
denoting the International Feed Class number. With the exception of soybean hay,
soybean hulls (class 1), lecithin, soybean mill run and soybean mill feed (class 4),
soy products listed in table 2 (page 16) fall in the class of protein supplements (5) defined
as products that contain more than 20 % crude protein on a dry matter basis. The five

9

3. Definition and Application of Soybeans and Soy Products

digits following the class number is the link between the INF and chemical and
biological data in the USA databank (NRC, 1982). The number appears generally on
official US ingredient specifications and, although the system may not be used by all
feed producers or manufacturers, it provides an easy and systematic reference for
quality systems and formulation purposes.

A brief and general description is available for many soy products. This
description has the advantage of providing information that is not generally
captured in compositional tables. It also provides for a general appreciation of the
origin and quality and thus the potential applications or uses in a feed. Although
these definitions might differ slightly between different sources, they are in general
sufficiently similar to use them interchangeably. AAFCO publishes at regular intervals
reference specifications for soybean products (AAFCO, 2001). These definitions have
been used as a basis for the specifications listed in Table 1.

Table 1
Description and classification of soybean products *

1. Condensed Soybean Solubles is the product obtained by washing soy flour or
soybean flakes with water and acid at a pH of 4.2-4.6. The wash water is then
concentrated to a solid content of not less than 60%. IFN 5-09-344.
2. Dried Soybean Solubles is the product resulting from the washing of soy flour
or soybean flakes with water and acid; water, alkali and acid; or water and alcohol.
The wash water is then dried. IFN 5-16-733.
3. Ground Extruded Whole Soybeans is the meal product resulting from
extrusion by friction heat and/or steam of whole soybeans without removing any
of the component parts. It must be sold according to its crude protein, fat and fiber
content. IFN 5-14-005.
4. Ground Soybean Hay is the ground soybean plants including the leaves and
beans. It must be reasonably free of other crop plants and weeds and must contain
no more 33% crude fiber. IFN 1-04-559.
5. Ground Soybeans are obtained by grinding whole soybeans without cooking
or removing any of the oil. IFN 5-04 -596.
6. Heat Processed Soybeans (Dry Roasted Soybeans) is the product resulting from
heating whole soybeans without removing any of the component parts. It may be
ground, pelleted, flaked or powdered. It must be sold according to its crude protein
content. Maybe required to be labeled with guarantees for maximum crude fat,
maximum crude fiber and maximum moisture (CFIA 2003). IFN 5-04-597.
7. Kibbled Soybean Meal is the product obtained by cooking ground solvent
extracted soybean meal, under pressure and extruding from an expeller or other
mechanical pressure device. It must be designated and sold according to its
protein content and shall contain not more than 7% crude fiber. IFN 5-09-343.

* In alphabetical order; adapted from the AAFCO Official Publication 2001 and the CFIA. 2003.

10


8. Protein Modified Soybean is a product that has been processed to primarily
modify the natural protein structure by utilizing acids, alkalies or other chemicals
and without removing significant amounts of any nutrient constituent. The
defined name under section 84 of the applicable soybean product so modified
shall be declared in the product name. IFN 5-26-010.
9. Soy Flour is the finely powdered material resulting from the screened and
graded product after removal of most of the oil from selected sound cleaned
and dehulled soybeans by a mechanical or solvent extraction process. It must
contain not more than 4.0% crude fiber. Some organisms also require labeling
guarantees for minimum crude protein and maximum crude fat and moisture.
IFN 5-12-177.
10. Soy Grits is the granular material resulting from the screened and graded
product after removal of most of the oil from selected, sound, clean and dehulled
soybeans by a mechanical or solvent extraction process. It must contain not
more than 4% crude fiber. Soybean grits mechanical extracted: IFN 5-12-176.
Soybean grits solvent extracted: IFN 5-04-592.
11. Soy Lecithin or Soy Phosphate is the mixed phosphatide product obtained
from soybean oil by a degumming process. It contains lecithin, cephalin and
inositol phosphatides, together with glycerides of soybean oil and traces of
tocopherols, glucosides and pigments. It must be designated and sold according
to conventional descriptive grades with respect to consistence and bleaching.
IFN 4-04-562.
12. Soy Protein Concentrate is prepared from high quality, sound, dehulled
soybean seeds by removing most of the oil and water soluble non-protein
constituents from selected, sound, cleaned, dehulled soybeans (CFIA 2003) and
must contain not less than 65% protein on a moisture-free basis. It shall be
labeled with guarantees for minimum crude protein, maximum crude fat,
maximum crude fiber, maximum ash and maximum moisture.
IFN Number: 5-08-038.
13. Soy Protein Isolate is the major proteinaceous fraction of soybeans prepared
from dehulled soybeans by removing the majority of non-protein components,
and contains not less than 90% protein on a moisture-free basis. The CFIA (2003)
adds that the original material must consist of selected, sound, cleaned, dehulled
soybeans and that it shall be labeled with guarantees for minimum crude
protein (90%), maximum ash and maximum moisture. IFN Number 5-08-038
(CFIA lists this product with the IFN Number 5-24-811.
14. Soybean Feed, Solvent Extracted is the product remaining after the partial
removal of protein and nitrogen free extract from dehulled solvent extracted
soybean flakes. IFN 5-04-613.
15. Soybean Flour Solvent Extracted (or Soy flour) is the finely powdered
material resulting from the screened and graded product after removal of most
of the oil from dehulled soybeans by a solvent extraction process. It shall contain
less than 4 percent crude fiber. It shall be labeled with guarantees for minimum

11


crude protein, maximum crude fat, maximum crude fiber and maximum
moisture. IFN 5-04-593.
16. Soybean Hulls consist primarily of the outer covering of the soybean.
IFN-1-04-560.
17. Soybean Meal, Mechanically Extracted is the product obtained by grinding
the cake or chips which remain after removal of most of the oil from soybeans
by a mechanical extraction process. It must contain not more than 7% crude
fiber. It may contain an inert, non-toxic conditioning agent either nutritive or
non-nutritive or any combination thereof, to reduce caking and improve
flowability in an amount not to exceed that necessary to accomplish its intended
effect and in no-case exceed 0.5%. The name of the conditioning agent must be
shown as an added ingredient. IFN 5-04-600.
18. Soybean Meal, Dehulled, Solvent-Extracted is obtained by grinding the
flakes remaining after removal of most of the oil from dehulled soybeans by a
solvent extraction process. It must contain not more than 3.3% crude fiber.
It may contain an inert non-toxic conditioning agent either nutritive or non-
nutritive or any combination thereof, to reduce caking and improve flowability in
an amount not to exceed that necessary to accomplish its intended effect and in
no-case to exceed 0.5%. The name of the conditioning agent must be shown as
an added ingredient. IFN 5-04-612. It may also be required to be labeled with
guarantees for minimum crude protein, maximum crude fat and maximum
moisture(CFIA 2003).
19. Soybean Meal, Solvent-Extracted, is the product obtained by grinding the
flakes which remain after removal of most of the oil from soybeans by a solvent
extraction process. It must contain not more than 7% crude fiber. It may contain
an inert, non-toxic conditioning agent either nutritive or non-nutritive and any
combination thereof, to reduce caking and improve flowability in an amount not
to exceed that necessary to accomplish its intended effect and in no-case exceed
0.5%. It shall contain less than 7 percent crude fiber. The CFIA (2003) specifies
that it shall be labeled with guarantees for minimum crude protein, maximum
crude fat and maximum moisture. IFN 5-04-604.
20. Soybean Mill Feed is composed of soybean hulls and the offal from the tail
of the mill which results from the manufacture of soy grits or flour. It must
contain not less than 13% crude protein and not more than 32% crude fiber.
IFN 4-04-594.
21. Soybean Mill Run is composed of soybean hulls and such bean meats that
adhere to the hulls which results from normal milling operations in the
production of dehulled soybean meal. It must contain not less than 11% crude
protein and not more than 35% crude fiber. IFN 4-04-595.
22. Soybean Oil consists of the oil from soybean seeds that are commonly
processed for edible purposes. It consists predominantly of glyceride esters of
fatty acids. If an antioxidant(s) is used, the common name or names shall be
indicated on the label. It shall be labeled with guarantees for maximum

12


moisture, maximum insoluble matter, maximum unsaponifiable matter and
maximum free fatty acids. IFN 4-07-983.
23. Soyflour Chemically and Physically modified is the product resulting from
treating soy flour by chemical and physical (heat and pressure) means. It shall be
labeled with guarantees for minimum crude protein, maximum crude fat,
maximum crude fiber and maximum moisture. IFN 5-19-651.

The list in Table 1 gives an overview of the large diversity of soy products and
different methods of producing them. It provides a brief description of how the
product is obtained and for some products, compositional reference points.
The common name and IFN is provided which allows for a consistent and non-
equivocal use of ingredients, important in quality systems. The description gives
an adequate back ground of the products for trading and classification purposes,
references in quality systems and production purposes. It is sufficiently precise to
provide clear reference points for product definition and contract agreements but
general enough to cover a substantial variation in composition and production
processes. For proper use of an ingredient additional analytical data should
complement the information provided in the description. However, for analytical
purposes the descriptions provide general back ground information as to what
can be expected and how analysis should be carried out or what results may be
expected. For formulation objectives the description only serves as a classification
aide and more precise compositional data will be necessary.

The products listed in Table 1 only represent the major soy products produced.
At present, a large number of additional specialty products are marketed and the list
does not adequately reflect the acceleration seen in the development of new soy
products; mostly branded products. Many new, more elaborated products have come
on the market over the past 10 - 20 years. The most important examples of these are
the different types of soy protein concentrates and soy isolates. These products,
characterized by strongly reduced anti-nutritional factors, can effectively be used in
diets for young animals, pets and aquaculture, replacing other protein sources such
as milk or animal proteins (fish meal). Additional new soy products have often been
developed with applications in human or pet food nutrition in mind. In this area
special importance is attached to the functional properties of soybean proteins
which include the ability of the proteins to increase viscosity, emulsify, form gels,
foam, produce films and absorb water and/or fat. Specific applications allow the
production of texturized structures, a much sought after property in certain human
and pet food products. The functional properties of soy proteins are related to the
amino acid composition and sequence (primary structure) as well as the spatial
configuration of the protein molecule and the inter-molecular forces (secondary and

13


tertiary structures). Soybean protein products with unique functional properties may
constitute important tools in the formulation of the so-called specialty diets used in
animal nutrition. However, these techniques and products remain insufficiently
explored in the production of specialty diets for domestic livestock, with economic
considerations probably being the major limiting factor at present.

The most important products in terms of volume of use are soybean meals
(SBM) solvent extracted or dehulled (18 and 19) resulting from the original use of
soybeans i.e. the removal of oil. This is also the case for the mechanically extracted
SBM (17) although this type of SBM is much less common. Fullfat soybeans in
ground, extruded or heated form are defined and their use is increasing due to their
high energy content, especially in formulations where previously animal products
(meat and bone meals and fats) were of interest. Two fiber-rich products are included
in the list: ground soybean hay (4) and soybean hulls (16). While soybean hay has
little application in the compound feed industry, the interest in soybean hulls is
important and increasing. Soy flour and soy grits are primarily products destined for
human consumption although minor amounts may find an application in specialty
diets. Technological modifications of these products have produced different types
of flour and grits. They are further classified and commercialized according to their
application objectives with the main differences being the level of fat content
or heat treatment.

The remaining products are mainly modifications of different types of soybean
meal with the objective of rendering the product more digestible; either through
the modification of the protein structure or a removal of the ANF. The specifications
do not make reference to these factors leaving the decision as to how the product
compares in this respect to the interpretation of the nutritionist or guarantees
provided by the producer. Quality analysis must provide a more precise indication
of the product in terms of these characteristics in order to assure that the diet meets
proper nutritional and animal performance objectives.

With the increased complexity of production processes aimed at removing ANF
and improving protein digestibility, a clear understanding of the products and the
production process becomes more important and adapted quality procedures/
analysis more critical. Quality differences between producers/suppliers for these
products can be substantial, especially for the more evolved products. These
differences need to be verified and understood at the feed manufacturer’s level.
Nevertheless, it remains the responsibility of the user to carry out the needed quality
analysis and classify suppliers and products accordingly. Reliable manufacturer’s
information is, of course, important but verification remains the basis of this tool and

14


of the overall quality assurance program. The quality of the information provided by
the manufacturer must be an integral part of the “supplier classification process”
.

The quality of ingredients play a determining role in the level at which these
ingredients are used in animal diets. Quality criteria used to determine the inclusion
level for an ingredient go beyond the standard nutrient levels, and have often more
to do with residual ANF, storage and contaminations (see Chapter 5) and the
physiological characteristics of the animal. The inherent variation in quality and
chemical characteristics associated with these ingredients make repeated quality
analyses necessary which in turn will determine more precisely the inclusion levels
employed. The nutritionist’s experience and interpretation of the quality analyses
play a major role in defining the final inclusion level used in particular diets. Table 2
gives thus only general estimates of maximum inclusion levels of each product
under practical conditions of diet formulation. The inclusion levels suggested are for
inclusion in complete diets and are thus necessarily general. They will also need to be
adjusted to the specific diet (inclusion of other ingredients) and feeding objectives.
Also, the precise nutrient and ANF concentrations and the diet requirements (the
ability of the animal to use nutrients or deal with anti nutritional factors) will need
to be taken into consideration. Fine tuning of inclusion levels for each product is
very much a company-specific decision reflecting depth of understanding of the
formulation complexities and confidence in proprietary data relative to the
ingredients. The suggestions listed in Table 2 must therefore be regarded as general
recommendations that need to be further defined for each feed manufacturer, the
manufacturing process and the feed being formulated.

Some of the maximums suggested are not defined by any inability of the
animal to use the nutrients in a given product, but rather by the effects of specific
nutrients on carcass or product quality. Such is for instance the case for whole heat
treated soybeans or soybean oil. Other maximums are controlled by economic
considerations. While higher inclusions in diets may be possible, those levels will
inevitably lead to additional costs with no or limited gain in performance.

Some soy products listed in Table 1 are not included in the recommendations for
use in animal diets. This is the case of “protein modified” soybean meal, soy flour or
grits. Although these ingredients could be used in animal diets (and they actually
may be when quality is not sufficient to include in human diets) they are primarily
produced for utilization in human foods. Included in small amounts, they may
convey major nutritional or technological advantages to certain food items (Liu,
1997). Evaluation of these products in pet foods or certain specialty diets merit
consideration.

15


Table 2
Application of soybean products(1)
Species(2)
Product Po Sw R A Pe Level (%)(3)
1. Condensed Soybean Solubles √ 10
2. Dried Soybean Solubles √ 15
3. Ground Extr. Whole Soybeans √ √ √ 35
√ √ 5(4)
4. Ground Soybean Hay √ 20
5. Ground Soybeans √ 15
6. Heat Processed Soybeans √ √ √ 15
7. Kibbled Soybean Meal √ √ 10(Y)
√ √ 7
8. Soy Lecithin or Soy Phosphate √ √ √ √ √ 3
9. Soy Protein Concentrate √ √ √ 7(Y)
√ √ 5(4)
10. Soy Protein Isolate √ √ √ 10(Y)
√ √ 15(4)
11. Soybean Feed, Solvent Extracted √ √ √ 5(Y)
√ √ 3
12. Soybean Flour Solvent Extracted √ √ √ 40
13. Soybean Hulls √ √ 25
14. SBM Mechanically Extracted √ √ √ 30
15. SBM Dehulled Solvent Extracted √ √ √ 35
16. SBM Solvent Extracted √ √ √ 35
17. Soybean Mill Feed √ √ √ 10
18. Soybean Mill Run √ √ √ 10
19. Soybean oil √ √ √(5) √ 8

(1) Suggested upper-use levels in diets of different domestic species; this will vary with age of animal, quality,
composition and analysis of product; does not include young animal diets unless specifically indicated.
Detailed and extensive analyses will allow discretionary changes in usage level.
(2) Species: Production diets (growing/finishing) for Poultry (Po), Swine (Sw), Ruminants (R), Aqua (salmonids) (A);
Pets (dogs) (Pe).
(3) On a diet dry matter basis.“Y” indicates primarily in young animal diets.
(4) Higher levels may be used in salmon and trout grower, finisher diets.
(5) Maximum inclusion of oil in Ruminant diets should not exceed 2%.

CONTENTS

16

4. CHEMICAL and NUTRITIONAL
COMPOSITION of SOYBEAN
PRODUCTS
The compositional data provided in the Tables 3 (p.19) and 4 (p.20) (with
additional details in Appendix Tables 1, 2) are better descriptors of the nutritional
characteristics of soybean products. They require however, a more in-depth
understanding of the chemical, analytical and nutritional aspects of the
products. The composition data also provide an indication of the specific
processes that have been used to obtain the product. This is especially true for
the data in Table 4. Along with the general description provided above, these
data give thus a rather complete picture of the various properties and potential
applications for each product. The total number and types of soybean products
commercialized is clearly much larger that the ones listed in the tables.
The tables only provide values for the main products. A large variety of different
soy products are produced by different companies and for a large number of
specific applications. Soy protein concentrates or heat or formaldehyde treated
products for ruminant diets are an excellent example of this. The nutritional
concentration as analyzed may not differ significantly from an ingredient listed
but the nutritional value (due to a change in digestibility or degradability)
may vary greatly. Since the tables only report composition that can be directly
analyzed, such differences do not show up and are therefore not included.

The nutrient concentration of the different soy products in Tables 3 and 4 are
compiled from a wide range of official sources and publications (NRC 1982, 1998,
2001; INRA-AFZ 2002; CVB 2000; FEDNA, 1994 and others). Besides completing
the descriptive information provided in Table 2 the major purpose of the
composition tables is to provide reference values that can be used to either
evaluate the analytical data that are obtained in the laboratory or to further
classify a specific ingredient. Since the data are obtained from a wide range of
publications, the user may want to refer to the original publications if the sample
corresponds more closely to one of the sources in his region. This is especially
true in the case of soybean meals or soy by-products where crushing and further
handling of the ingredient determine to a large extent the nutrient quality
of the products.

The table values provide means based on a large number of samples
covering many years and a wide range in origin. They cannot be used as
standard values but only as reference points around which analysis of individual
samples should be situated if they are to be identified by the specific ingredient
name. Most individual samples will be within an acceptable statistical range of

17

4. Chemical and Nutritional Composition of Soybean Products

these means (see Chapter 10). This level of precision is adequate for classification,
storage and trading agreements, as those are generally based on a small set of
analyses (proximate analysis or just humidity, protein and fat). More detailed analyses
concerning the more difficult to determine nutrients may show larger variations from
the means and possibly inconsistencies with some values above and others below
the table values. This is often the case for amino acids or micro minerals. As such they
may point to consistent differences in the production process of a given supplier or,
alternatively, reflect problems in the analytical procedure. The experience and
know-how of a lab technician in interpreting the result is here of great value.
Cross-checking of values known to be affected in a similar fashion by a production
process or a laboratory procedure may provide an explanation of a discrepancy
and confirm the true value and classification for the ingredient.

For most users of soy products the detailed nutrient concentrations serve
as a basis to formulate diets and to calculate total nutrient supply to animals.
Since animal performance is determined by nutrient concentration and the
relationship between nutrients, knowing the precise nutritional composition of the
ingredients that make up the diet allows the prediction of animal performance and
thus a detailed estimation of the value of each ingredient. Clear compositional
descriptions of soy products are thus not only necessary for quality control reasons,
but also for the evaluation in a diet or feeding operation. For precise formulations
the analytical data on the ingredient in the plant should be used. The use of the table
values, especially because of the large contribution that soy products make to the
protein and amino acid supply, may lead to significant variations in nutrients
between the formulated value and the real diets.

The compositional data in Table 4 includes nutrients that can be directly analyzed
in a large and well equipped laboratory. Routine analyses, as carried out in standard
quality control procedures or smaller laboratories, mainly concern the proximate
analysis, the van Soest fiber components (with the exception of lignin) and the
minerals calcium and phosphorus. These analyses (especially the proximate) are most
often used to derive other nutrient values such as amino acids or energy. In advanced
formulation systems they are generally combined with estimates of digestibility for
each individual nutrient. No digestibility data are included here as this information is
not necessarily the result of direct observations but rather of literature compilations
and research conducted by feed compounders. Thus digestibility data used in
formulationsystems can differ considerably among users and are generally
considered proprietary information. In the Appendix tables (1, 2) specific energy
values have been included however, because of their importance as descriptive
parameters for individual soy products and because of their importance in classifying
and referencing ingredients.

18


Table 3
Composition of some soy protein ingredients used in animal feeds
Heat SBM SBM SBM
processed SBM solvent solvent solvent Soy Soy
FF soybean mechanically extracted extracted extracted Soybean protein protein
Unit seeds extracted 44 48 50 hulls concentrate isolate

Dry matter % 89.44 89.80 88.08 87.58 88.20 89.76 91.83 93.38
Crude protein % 37.08 43.92 44.02 46.45 48.79 12.04 68.60 85.88
Crude fiber % 5.12 5.50 6.26 5.40 3.42 34.15 1.65 1.32
Ether extracts % 18.38 5.74 1.79 2.1 1.30 2.16 2.00 0.62
Ash % 4.86 5.74 6.34 6.02 5.78 4.53 5.15 3.41
NDF % 12.98 21.35 13.05 11.79 9.95 56.91 13.50 -
ADF % 7.22 10.20 8.76 7.05 5.00 42.05 5.38 -
ADL % 4.30 1.17 0.75 0.90 0.40 2.05 0.40 -
Starch % 4.66 7.00 5.51 5.46 3.28 5.95 - -
Total sugars % - - 9.06 9.17 9.29 1.40 - -
Gross energy kcal/kg 5013 - 4165 4130 4120 3890 4280 5370
Lysine % 2.34 3.50 2.85 2.89 3.00 0.73 4.59 5.26
Threonine % 1.53 2.21 1.80 1.84 1.90 0.73 2.82 3.17
Methionine % 0.52 0.80 0.62 0.63 0.67 0.14 0.87 1.01
Cystine % 0.55 0.77 0.68 0.73 0.73 0.16 0.89 1.19
Tryptophane % 0.49 0.74 0.56 0.63 0.65 0.12 0.81 1.08
Calcium g/kg 2.62 2.96 3.12 3.07 2.68 4.96 2.37 1.50
Phosphorus g/kg 5.70 6.64 6.37 6.37 6.36 1.59 7.63 6.50
Magnesium g/kg 2.80 2.84 2.72 3.03 2.88 2.23 1.85 0.80
Potasium g/kg 15.93 20.28 19.85 22.00 20.84 12.15 12.35 2.75
Sodium g/kg 0.29 0.33 0.18 0.18 0.88 0.10 0.55 2.85
Linoleic acid C18:2 % 9.70 2.87 0.64 0.80 0.56 1.21 - -

FF Soybean = Full Fat Soybean; SBM = Soybean meal. For more detailed compositional data on soybean
products see Appendix table 1, 2.
Source: compilation of NRC, INRA-AFZ, CVB, FEDN and selected suppliers
NDF = Neutral detergent Fiber; ADF= Acid Detergent Fiber; ADL = Acid Detergent Lignin (Klason Lignin)

Protein quality analyses (Urease Index, KOH soluble N, or PDI) are also not
included as these do not generally differ among soy protein products. A number
of these analyses do exist and they are important in evaluating soy protein quality
especially in terms of digestibility of amino acids. Methods and optimal values for
these tests are detailed further in Chapter 8. In many respects they refer to the
residual values for the ANF listed in Table 5 (p. 22) but only the heat labile ones
such as trypsin inhibitors, lectins and goitrogens (Liener, 2000). There is no proven

19


relationship between heat stable ANF and protein quality indexes. For many diets,
especially in the case of diets for young animals, aquatic species and pets, the
application and use of soy products depends to a much larger extent on the residual
ANF than on the nutrient concentration. In such diets the more elaborated soy
products such as SPC or SPI are more frequently used. Accurate analyses for most of
these ANF are difficult to carry out and under most practical conditions the suppliers’
guarantees are accepted. As Table 5 indicates, the range in some of these ANF is
considerable and a thorough supplier classification is thus important. In many cases,
if an analysis for a specific ANF is indicated, the choice to use external laboratories
may be advised. External, specialized, laboratories will provide reliable results and
generally are in a position to give advice as to the quality and level of an ANF relative
to other samples of a similar product. If preference is given to install an analysis for
ANF (generally trypsin inhibitor) in a laboratory the adherence to a ring test or
systematic comparisons of results with a well established laboratory is necessary.

Table 4
Analytical characteristics of common types of soy protein products
Enzyme Alcohol
Soybean treated extracted
Product type Unit seeds SBM SPC SPC SPI
Humidity % 10 - 12 10 - 12 6-7 6-7 6-7
Crude protein % 33 - 17 42 - 50 55 - 60 63 - 67 >85
Fat % 17 - 20 0.9 - 3.5 2.5 0.5 - 3.0 0.1 - 1.5
Ash % 4.5 -5.5 4.5 - 6.5 6.2 - 6.8 4.8 - 6.0 2 - 3.5
Oligosacharides % 14 15 <1 <3.5 <0.4
Stachyose % 4 - 4.5 4.5 - 5 <0.3 1-3 <0.2
Raffinose % 0.8 - 1 1 - 1.2 <0.2 <0.2 <0.1
Trypsin inhibitor TIA mg/g CP 45 - 60 4-8 1-2 2-3 <1
Glycinin mg/g 150 - 200 40 - 70 <0.1 <0.1 <0.01
ß-conglycin mg/g 50 - 100 10 - 40 <0.1 <0.1 <0.005
Lectins ppm 50 - 200 50 - 200 <1 <1 <1
Saponins % 0.5 0.6 0 0 0
Phytic acid bound P % 0.6 0.6 0.6 0.6 -
SBM = defatted soybean meal; SPC = soy protein concentrate; SPI = soy protein isolate.
Adapted from: Hansen (2003) and Peisker (2001)

Anti-nutritional factors decrease in concentration as the elaboration increases
and the soy product becomes richer in protein. The increased concentration of
protein associated with a lower level of ANF increases the value of soy products in
a proportionally greater fashion than the increase in cost of production. They are
therefore much sought after products in specialty diets. However, they remain
uneconomical in diets of older livestock animals as those animals are less sensitive
to the ANF and their protein requirements can be met with lower concentrations
and/or quality of proteins.
CONTENTS
20

5. OFFICIAL STANDARDS
of SOME SOYBEAN
PRODUCTS
While a large number of compositional tables and publications for soybean
products exist, those data cannot be considered as standard values, especially
not for trading purposes. For trading and contractual purposes they are too
detailed and thus unpractical. Furthermore, they do not provide the required
borderline minimum or maximum values for limited number readily identifiable
parameters.

A limited number of official standards have been published to start with the
basic material: whole, untreated soybeans or seeds (IFN 5-04-610). As is the case
for all other grains and seeds the USDA publishes official standards for soybean
grains as defined under the United States Grain Standards Act. These standards
do not generally change much over time and under the act soybeans are
defined as grains that consists of 50 percent or more of whole or broken
soybeans (Glycine max (L) Merr.) that will not pass through an 8/64’’round hole
sieve (3183 microns) and does not contain more than 10.0 percent of other
grains for which standards have been established under the United States Grain
Standards Act (USDA, 1999).

For trading purposes – especially in view of specific applications and export
requirements – additional specifications are provided by dividing soybeans into
classes and grades. Only two classes of soybeans have been defined (yellow
soybeans and mixed soybeans) but 5 grades are specified. The grades and grade
requirements for the major export countries (USA, Brazil and Argentina) are
similar. However, while Brazil and Argentina have a special export grade, the
United States does not define a specific export grade as soybeans are exported
from the US at any pre-defined specification or grade. The USDA (1999)
description of grades is provided in 5.

Next to whole soybeans only three soybean products (two soybean meals
and soybean oil) have standard values. Used as official references standards they
have been developed by the National Oil Processors Association (NOPA, 1997)
and are also published by the American Soybean Association (ASA, 1998) in the
Soy Importers Handbook. These standards are now widely accepted and provide
minimums or maximums on only a few, easily identifiable, key parameters. In the
case of soybean meals their main purpose is the classification of soybean meals
into two main categories: solvent extracted SBM and dehulled, hipro SBM.

21

5. Official Standards of Some Soybean Products

Table 5
US grades and grade requirements for soybeans
Minimum
test weight Maximum limits of:
Damaged kernels Soybeans
per per Heat Foreign of other
bushel hl damaged Total material Splits colors
Grade (lbs) (kg) % % % % %

U.S. No.1 56 72 0.2 2.0 1.0 10.0 1.0
U.S. No.2 54 69 0.5 3.0 2.0 20.0 2.0
U.S. No.3 (1) 52 67 1.0 5.0 3.0 30.0 5.0
U.S. No.4 (2) 49 63 3.0 8.0 5.0 40.0 10.0
U.S. Sample
grade (3)

(1) Soybeans that are purple mottled or stained are graded not higher than U.S. No 4.
(2) Soybeans that are materially weathered are graded not higher than U.S. No 4.
(3) Soybeans that do not meet the requirements for U.S. Nos. 1,2,3 or 4, or
i) Contain 8 or more stones which have an average weight in excess of 0.2% of the sample weight, 2 or more
pieces of glass, 3 or more Crotalaria seeds, 2 or more castor beans, 4 or more particles of an unknown
substance(s), 10 or more rodent pellets, bird droppings or equivalent quantity of other abnormal filth per
1,000 grams of soybeans; or
ii) Have a musty, sour or commercially objectionable foreign odor (except garlic odor);
iii) Are heating or otherwise of distinctly low quality.
See also: USDA, 2001: http://www.usda.gov/gipsa/reference-library/brochures/soyinspection.pdf

For soybean oil the NOPA standards refer to crude degummed soybean oil
mainly with food application purposes in mind. These standards serve as a
general guide for transactions, thus assuring a minimal degree of quality and
consistency in at least the three main types of soy products being traded.
However, the standards and trading guidelines proposed by NOPA are not
binding. Organizations, companies or individuals participating in a transaction
involving soybean meals are free to adopt, modify or disregard the NOPA rules.
They principally serve the trading and marketing of US soybean products within
the USA but due to their wide acceptance, their impact goes well beyond US
meals (and oils) as they are generally applied to compare and benchmark
soybean products from other origins.

22


Solvent extracted soybean meal can be the result of blending back soybean hulls
in the dehulled meal. The blending of different types of soybean meals or
soybean components at the point of shipping is allowed under NOPA regulations
and standards for minimum blending procedures are provided. As a matter of fact,
this can be the source of a significant variation in quality and chemical composition.
However, blending of soybeans is not permitted. For soybean meals only soy hulls,
soybean mill run and soybean mill feed are permitted to be blended with soybean
meals before the point of sampling. The blending must lead to a meal of uniform
quality representative of the contract terms.

Table 6
Specifications for solvent extracted and dehulled soybean meals (%)

Solvent Dehulled
Min/Max extracted SBM SBM

Moisture max. 12 12
Protein min. 44 47.5 - 49
Fat min. 0.5 0.5
Crude fiber max. 7 3.3 - 3.5
Anti-cacking agent max. 0.5 0.5

NOPA, 1997

For SBM, the NOPA standards clearly aim at providing a minimum number of
primary quality characteristics and as such are only a basis for contract specifications
(Table 6). The only characteristics defined are moisture, crude protein, fat and crude
fiber with a maximum tolerance for an anti-caking agent. Beyond purchasing
and possibly storage allocations these specifications have little impact on normal
feed milling operations; neither from a specific quality point of view nor from a
formulation perspective. They do not provide a sufficiently detailed overview of the
nutritional characteristics required for proper quality management or further use.
Meals purchased under NOPA contract specifications will therefore still need
additional analysis. In order to provide greater quality assurances and meet the
nutritional requirements of the feed compounder or nutritionist additional
recommendations have been added by NOPA (Table 7- next page).

23


These are, again, only recommendations that apply in a non-binding manner to
all soybean meals. Rather than guidelines they should be regarded as further sugges-
tions to both producers of soybean meal and buyers, provided in an effort to improve
the quality of US soybean meals. Under practical conditions there remains a large
variation around these recommendations and from a feed compounder’s point of
view, information on quality requirements for SBM needs to be still more detailed.
Also, new parameters have been added and more recently evaluations have changed
slightly. For instance there is a definite tendency for KOH values to shift to the high
end of the established range (close to the 85 % value).

Table 7
Recommended additional specifications for soybean meal

Lysine 2.85 % (basis 88 % dry matter)
Ash < 7.5 %
Acid insoluble ash (silica) <1%

Protein solubility in 0.2 % KOH 73 - 85 %
Urease activity 0.01 - 0.35 pH unit rise
Bulk density 57 - 64 g/100 cc

Screen analysis (mesh) 95 % thru # 10, 45 % thru #20, 6 % thru # 80
Texture Uniform, free flowing, no lumps, cakes, dust
Color Light tan to light brown

Odor Fresh - not musty, sour, ammonia, burned
Contaminants No urea, ammonia, pesticides, grains, seeds,
molds

NOPA, 1997

The Protein Dispersibility Index (PDI), an additional measure of protein quality,
has been added as a routine quality evaluation. This follows the general application
of this method in evaluating protein quality in products for human consumption
(AACC, 1976). The results of this method are considered to be superior to the KOH
solubility especially where it concerns cases of inadequate heat treatments (Batal et
al., 2000). The KOH solubility index is considered better to estimate overheating of
SBMs. Nevertheless, consistent application of the recommendations in Table 7 would
go a long way in meeting product quality and nutritional requirements.

An additional degree of detail is necessary for the regular and detailed
formulation changes that are required to meet the performance guarantees of

24


animal diets and the constant cost-reduction objectives. The generation of this
information is, at present, considered to be the responsibility of the in-house quality
control and analytical services organization of the feed compounder. As a matter
of fact, this is often regarded as part of the proprietary know-how by feed
manufacturers. It does, however, offer the crusher an opportunity to provide a more
consistent and better quality product and therefore a means to add value to a
commodity. As identity preservation (IP) and traceability tools improve, a greater
detail and guarantee on nutritional characteristics will be possible.

The NOPA standards for soybean oil have the same objectives as those for SBM
i.e. providing a framework for trading and contract negotiations. However the
emphasis is on oil for human consumption as the designated types are for edible oil
(officially referred to as crude degummed soy oil). As a matter of fact, no standards
for oil used in animal feed is available and most feed companies or users of oil in
animal feed have developed in-house standards for oils and fats or mixtures of both.
These proprietary standards for animal feed will generally be slightly more relaxed
(see Table 8) but information for additional parameters such as iodine and peroxide
numbers are often required. On the other hand, information on P levels and flash
point are not considered. This difference in standards allows for the use of soy oils
which are rejected for human consumption to be used in animal feed provided they

Table 8
Standards for edible crude degummed soybean oil
and vegetable oils in animal feed
NOPA1 Feed2
Analytical parameter Unit Max Max

Unsaponifiable matter % 1.5 1.5
Free fatty acids (as Oleic acid) % 0.75 1
MIU (Moisture, Isolubles, Volatile matter) % 0.3 1
Flash point °F 250 –
Phosphorous % 0.02 –
Iodine value g/100g EE – 130 - 1363
Peroxide value Meq/kg – 2

1,2 NOPA, 1999; Feed refers to common values for vegetable oil.
3 Range for soybean oil.

25


meet the still stringent formulation and feed quality guarantees. In general, soy oil
usage in animal feed is reserved for specialty feeds often for those diets where highly
digestible energy sources are needed. This is typically the case in young animal diets.

Besides the basic products (soybeans, soybean oil, solvent extracted SBM and
dehulled SBM) there are no published requirements or recommendations for the
large array of other soy products that are marketed in various forms and conditions.
This leaves it up to the user to set internal quality control measures. Those may
include most of the criteria considered for the 3 main (basic) products but they
will need to go beyond this and include a measure of anti-quality components
(anti-nutritional factors – ANF), expanded amino acid profiles, in vitro digestibility
and measures of microbial contamination. It is interesting to notice that no specific
requirements have been published on the degree of microbial presence in soybeans
or SBM. The end user will therefore have to apply industry norms as established by
local governments or organizations.
CONTENTS

26

6. SAMPLING SOY
PRODUCTS
The quality of any analysis carried out on feed or the feed ingredients stands or
falls with the sampling tools and procedures. It seems evident - but is not necessarily
recognized under routine operating conditions - that in order for any subsequent
analytical work and interpretation to make sense, the collection of a correct,
representative sample is fundamental.

The objective of any sampling procedure, no matter what the subject to be
evaluated may be, is collection of a truly representative sampling; a sample that
represents to the greatest possible degree the composition and characteristics or
the material to be analyzed or studied. This always leads to a compromise between
cost of sampling and analysis and the degree or precision or confidence that is
acceptable. Statistical tools have been developed to asses the minimal number of
samples needed to achieve a given level of confidence regarding the composition of
the ingredients (see Chapter 10). As the number of samples that have been collected
and analyzed increases and variation for a particular nutrient and ingredient is better
understood, a more precise number of samples and sampling frequency can be
established. In the animal feed business, separate estimates of the number of
samples per supplier are not only recommended but are routine procedures for
many feed producers.

The sampling techniques and procedures vary with the ingredient, the form or
particle size of the ingredient, the conditioning and size of the consignment,
methods of loading or unloading and storage conditions. The soy products that are
used in animal feeds cover the entire range of physical forms from seeds to flakes
and powder and sampling methods will therefore need to be adapted to the specific
ingredient that enters a feed plant. Details to this extent need to be included in
quality control (QC) procedures and do now appear routinely on QC documents.
These techniques are fairly standard throughout the world and a detailed description
of sampling techniques for grains and seeds have been provided by Herrman (2001).
They apply to the majority of the soy products, in bag or bulk. Also NOPA has
published basic rules for the sampling of soybean meal at vessel loading facilities
using an automatic sampling device (see Appendix 3 - 6). These procedures are
practical and can be implemented under almost any condition or operating
procedure. A small degree of local adaptation may be necessary and may even be
advisable to assure the collection of a truly representative sample. The experience

27

6. Sampling Soy Products

and training of the samplers and persons in charge of the quality program will
determine to a large extent the efficacy of the sampling program and thus the
precise way to sample.

Prior to sampling soybean products a sampling scheme or frequency has to be
established. For a given ingredient this will depend to a large extent on the supplier
and the information received prior to delivery. Additional considerations are
laboratory capacity and availability, analytical cost, size of the consignment and
the use of the soy product (in which feed it will used as an ingredient and at which
percentage). In general, random sampling of different consignments (corrected for
experience or prior knowledge about the supplier and ingredient) is combined
with systematic sampling of the vessel, truck or container. To this purpose a pre-
determined sampling grid is established. Details on the sampling of open containers
with soybean products are taken from Herrman (2001) and GIPSA (1995) and are
summarized in Appendix 3- 6. A first, rapid evaluation of the material to be sampled
and of the sample is considered part of the sampling procedure. The total load (bags,
container or carrier) is evaluated for homogeneity and possible local damage during
loading or transport. In the case of a homogeneous delivery a pre-established
sampling grid is applied and samples are collected accordingly (Appendix 3A).
Separate sampling schemes have been developed to allow sampling of sound
versus damaged areas (Appendix 3B).

The tools that are used to sample depend on the material and form in which the
ingredient has been transported. While automatic sampling of trucks or containers is
increasingly implemented, hand-sampling remains a dominant means of obtaining
sample of soy products. In the case of hand- sampling, slotted grain probes can be
used to correctly sample soy beans and meals from a bag or a container (Appendix 4
- Figure 1A). Tapered bag triers (Appendix 4 - Figure 1B) are used to sample powder
and granular material, such as SPC and SPI from bags. For the sampling of soybeans
or soybean meals from a conveyer belt or a discharging truck, a Pelican Probe
sampler can be used (Appendix 4 - Figure 1D). The sampler is pulled through a
stream of falling grain or meal, collecting a sample into a leather bag. NOPA has
established special procedures for sampling soybean meals at vessel loading facilities
(Appendix 6).

The sampling of oil follows principles of sampling of other liquid feed
ingredients. A bomb or zone sampler (Appendix 4 - Figure 1 C) is used to collect
liquids such as soy oil from bulk containers. This sampler consists of a closed cylinder
(30 to 40 cm long by 4.5 to 7.5 cm in diameter) which is lowered at pre-defined

28

6. Sampling Soy Products

places in an oil tanker and filled with a 100 to 1000 ml sample of oil. Drums are
sampled using a glass or stainless steel tube 1 – 1.5 cm in diameter and 50 – 100 cm
long (Herrmann, 2001). A minimum of 500 ml sample of liquid must be obtained for
storage and sub-sampling.

The size of the sample depends on the homogeneity of the load (or lack thereof ),
and - again - previous experience is of importance. A larger sample should be
collected than that what is ultimately retained for further analysis and storage
(for the minimal legally required period). A minimum sample size of 2 kg is
recommended. In order to reduce the sample to the minimal required size, the
sample is passed through a gated riffle sample splitter (25 mm riffles) or a Boerner
divider (Appendix 6 - Figure 2 A and B respectively). This is done repeatedly until
the sample is homogeneous. A sub-sample (minimum 500 g) is obtained for further
analysis and storage.

The sample obtained prior to reduction as well as the final sample is rapidly
evaluated for test or specific weight and a number of physical and organoleptic
characteristics. The reduced sample is divided in two portions of roughly equal size
(250 g). Both are stored in airtight containers. One container is dispatched to the
laboratory for further analysis; the second container is stored in a dry storage area,
reducing to a minimum any type of chemical changes due to deterioration as the
sample may be used for subsequent analysis in the case of claims.
CONTENTS

29

7. PHYSICAL EVALUATION
and EQUIPMENT
Following the sampling, three types of evaluations are carried out on soybean
products. These are: Physical, Chemical and Microbiological. The Physical examination
of the material aims at establishing the general soundness of the product, its origin
and a rapid, general approximation of nutritive quality. This is a series of tests the
merchandise has to pass in order to be accepted by the buyer. The chemical analysis
will establish the nutritive value of the product. The specific analysis carried out may
differ according to future use (animal species). Results of these analyses aim at
providing the basis for a detailed nutritional profile possibly resulting in adaptations
in the formulation matrix. As such they establish the maximum and minimum level of
use in a feed as well as a precise price: quality relationship for the ingredient and the
individual nutrient supplied by the ingredient. The micro-biological evaluation
intends to reveal any specific microbial, fungal or yeast contamination. It mainly
refers to levels of salmonella and specific mycotoxins (mainly zearalenone and
ochratoxins). Exceeding pre-set (often legal) limits will lead to a rejection of the
material for further use or modifications in the inclusion level and/or the production
technology. All measures - physical, chemical and biological – when found to be
outside the contractual or legal limits may lead to claims and or changes in the
contractual agreement.

Soybean products are evaluated for a number of physical and organoleptic
criteria. A first evaluation of this type is carried out prior to sampling, but is repeated
on the original sample. In general a vessel, container, truck or bag is inspected
before unloading and a sample is taken. Only when the merchandise is considered
acceptable - upon general evaluation and a rapid analysis of the sample - will
unloading proceed. This inspection approaches the more detailed physical evaluation
of the sample and requires a certain level of expertise of the quality control person.
Inspection criteria should be part of a pre-established quality system. Most important
are those referring to the physical characteristics provided in Tables 5 and 7.
More stringent in-house standards or requirements may apply. At this stage the
important criteria for whole soybeans and soybean meals are: contamination or
foreign materials, bulk density, texture, particle size or screen analysis, color and odor.

The latter, color and odor are rapidly evaluated on the entire load by a trained
person. They are the first evaluation but are of crucial importance. Deviations from
the standard colors indicate excessive contamination with foreign material or
excessive or inadequate heat treatment. Deviations from the characteristic odor may

30

7. Physical Evaluation and Equipment

confirm the visual observations but will also provide a first idea of the past storage
conditions, contamination with other substances (especially liquids) and the
excessive presence of molds.

All further physical evaluations should be carried out at a plant laboratory or
special QC area. A first appreciation of the degree of contamination with foreign
material is obtained visually. A detailed count is obtained from the sample by
physically (hand-) separating a sub sample and weighing the various fractions. It is
recommended at this stage to take a sample for light microscopic analysis. Evaluation
of a sub-sample under a microscope permits a more detailed analysis of the material
and the contaminants. In general a wide field stereoscopic microscope with a
magnification of 20 to 40 times is adequate. Additional equipment required for
microscopic evaluation is a microscope-illuminator, forceps or probe and in the case
of large clumps a mortar and pestle. Precise analysis of contamination is possible
with a microscope but requires an experienced operator and may require additional
techniques specific to light microscopy in feed analysis.

Bulk density is measured by taking a liter of material (in an official container –
kettle) and weighing the content. Bulk density (expressed in lbs/bu, g/100 cc or
kg/hl) is a first appreciation of various attributes of the received ingredient namely:
the moisture content, texture and level of damage or contamination. The range of
required bulk density (test weight) for soybeans increases with the grade from
63 kg/hl for grade 4 to 72 kg per hl for grade 1 (Table 5). For soybean meals a single
range of 57 to 64 kg/hl is recommended (Table 7). The importance of this measure
has come under some criticism, especially from foreign operators. While it is widely
used in North America, only a minor number of processors or compounders outside
the USA use test weight on a regular basis. The equipment required for these
measures is relatively simple. Besides the kettle used to measure bulk density, a
balance with a minimum accuracy of + 0.1 grams is required (ASAE, 1993).

Texture may be considered as primarily a visual observation (verifying the
absence of lumps, cakes or coarse particles). A first rapid evaluation can be carried
out by hand-sieving a sample in a 0.525 Tyler (0.530US standard equivalent; 13.5 mm)
sieve. For a more precise and objective estimation of particle size (especially the
presence of small or dust particles) an analysis with an official particle separator
needs to be conducted. Special equipment for particle size separation exists.
Generally, a RoTap Sieve Shaker is used for this purpose. This allows separation of
particles to a size down to 150 micron (0.0059 inch) covering adequately
requirements for standards advised for soybean meal (see Table 7).
CONTENTS

31

8. CHEMICAL ANALYSES
The nutritional quality of a feed ingredient, and thus soybean products, is
dependent on the content of several chemical elements and compounds which
carry a nutritional function. These elements and compounds are referred to as feed
nutrients. When feeding animals, nutritionists select a combination of ingredients
that supply the right amounts of a series of feed nutrients. Therefore, when preparing
rations, ingredients are treated as carriers of feed nutrients. Thus, the quality and
value of a given ingredient will largely depend on the concentration of its nutrients.
Because determining the content of all feed nutrients is extraordinarily time
consuming and almost impossible, nutritionists use different systems for estimating
or approximating the nutritional value of a feed. The most common system is the
so-called Weende system (developed in Germany more than 100 years ago).
The system measures water or humidity, crude protein, crude fat, crude fiber, ash and
nitrogen-free extract. This method has been proven to be useful for assessing the
value of ingredients, however, as with any system, it has a number of shortcomings.
The most important one refers to the crude fiber fraction (and consequently the
nitrogen-free extract which is not directly determined but calculated by difference).
Nowadays, as will be discussed later in this chapter, there are improved methods to
determine nutrients within the fibrous fraction of soybean products.

Soybean meal is one of the most consistent (least variable) and highest quality
protein source for animal nutrition. However, some variation does occur in both
the nutrient concentration (chemical determination) and quality (digestibility or
bioavailability) among different samples and sources of soybean meal. These
variations can be attributed to the different varieties of soybeans, growing
conditions, storage conditions and length, and processing methods. Because
soybean products, especially soybean meals, are such an important fraction of feeds
(in poultry they can account for 35% of the total formula) it is crucial to monitor the
quality of soybean products. Small changes in quality might translate into important
changes in animal performance due to their high inclusion rate in the ration.
CONTENTS

8.1 Moisture
Moisture content is one of the simplest nutrients to determine, but at the same time
is one of the most important. The moisture content of soybean products is important
for three main reasons:

32

8. Chemical Analyses

1. To establish the appropriate acquisition price based on the concentration of the
nutrients on a dry matter basis and thus not paying more than necessary for
water.
2. A wrong determination of moisture will affect the rest of the nutrients when
expressed on a dry matter basis, potentially leading to erroneous
concentrations of nutrients in formulated diets.
3. To assure that mold growth cannot occur.

In general, samples with moisture content above 12.5% present a high risk of
molding, and should be accepted with caution and correspondent penalties for
quality. However, moisture is not evenly distributed across the sample particles.
A sample batch containing an average of 15.5 percent moisture may, for example,
contain some particles with 10 percent moisture and others with 20 percent
moisture. The particles with the highest moisture content are the ones most
susceptible to mold growth. Consequently, at the early stages of development mold
growth is often concentrated in specific areas of a batch of soy products underlining
the importance of good sampling methods. To determine moisture content it is
necessary to have a forced-air drying oven, capable of maintaining 130°C (± 2°C),
porcelain crucibles or aluminum dishes and an analytical balance with a precision
of 0.01 mg.

The official method (AOAC, 1990) to determine the moisture content of soybean
products consists of:
• Hot weighing porcelain crucibles and registering their tare.
• Placing 2 ± 0.01 g of ground sample in a porcelain crucible and drying
at 95-100°C to a constant weight (usually about 5 hours is sufficient).
• Hot weighing crucible and sample.
• Calculating the moisture content as a percentage of original weight:
Original weight – Final weight
Moisture, % = x 100
Original weight
and
Dry matter, % = 100 –moisture, %

An alternative, but less accurate method that has the advantage of being fast and
simple is the determination of moisture with a microwave. In this method a sample of
100 g is placed in a microwave oven for about 5 minutes. It is important not to run
the microwave for more than 5 minutes to avoid burning the sample. Reweigh and
record the weight, and place the sample in the microwave for 2 more minutes.
Repeat the process until the change in weight is less than 0.5 g than the previous
one. This weight would be considered the dry or final weight. The calculations are
performed as indicated above.

33


In feed plants, for routine QC procedures, moisture is often determined by the
Brabender test. Like the microwave method, this test is rapid, simple and considered
less accurate than the oven dried reference method. This test requires a small,
semi-automatic Brabender moisture tester, a scale and aluminum dishes. For most
soy products the thermo-regulator of the Brabender moisture tester is set to 140°C
with the blower on. Allow the unit to stabilize (± 0.5°C). Tare an aluminum dish on
the analytical balance. Weigh ~10 g of sample in the dish and record exact weight.
Place the dish (or dishes, up to 10) in the oven, close door. Start timing when
temperature returns to 140°C and then dry for one hour. Re-weigh the sample hot
after the specified drying time. Calculate moisture with equation above.
Moisture can also be determined by near infrared spectroscopy (see Chapter 9).
CONTENTS

8.2 Ash
Ash determination requires a muffle furnace, porcelain crucibles, and an
analytical balance (precision of 0.01 mg).

The ash content of soybean products is determined by weighing 2 ± 0.1 g of
sample in a tared porcelain crucible and placing it in a furnace at 600°C for 2 hours.
The oven is turned off, allowed to return to room temperature and the crucible plus
ash weighed. To obtain the ash content of the sample, the final weight should be
divided by the initial weight and then multiplied by 100 to express it in a percentage
basis. The ash content is thus calculated as:
Final weight
Ash, % = x 100
Original weight

Monitoring ash content is not only a way to assess the nutritional quality of
soybean products but also to detect possible contaminations, especially soil.
For example, the ash content of soybean meal should not exceed 7%.
CONTENTS

8.3 Protein
Protein is no doubt the most important and frequently analyzed nutrient in soy
products. The protein content of soybean products is estimated as total nitrogen in
the sample multiplied by 6.25. This assumes that protein in soybean products has
16% nitrogen; however, the actual amount of nitrogen in soybean protein is 17.5%.
Nevertheless, like for most other ingredients used in feed formulation, the standard
value of 6.25 is used. Determining crude protein from nitrogen content has the

34


drawback that part of the nitrogen present in soybean products is considered to
be part of proteins (or amino acids), which is not the case as there is nitrogen in
the form of ammonia, vitamins and other non-protein compounds. However, the
nitrogen fraction that is not in the form of amino acids or protein in soybean
products is very small and corrections for the difference in N content in soybean
products relative to other ingredients are carried out at the amino acid level.

The most accurate method for determining the nitrogen content of soybean
products is the Kjeldahl method. This method consists of digesting the sample in
sulfuric acid (H2SO4) and a copper and titanium catalyst to convert all nitrogen into
ammonia (NH3). Then, the NH3 is distilled and titrated with acid. The amount of
nitrogen in the sample is proportional to the amount of acid needed to titrate the
NH3. The Kjeldahl method requires:
• A digestion unit that permits digestion temperatures in the range of 360 – 380°C
for periods up to 3 hours.
• Special Kjeldahl flasks (500 – 800 ml).
• A distillation unit that guarantees air-tight distillation from the flask with the
digested sample into 500 ml Erlenmeyer flasks (distillation receiving flask).
• A buret to measure exactly the acid that needs to be titrated in the receiving flask
to neutralize the collected ammonia hydroxide.
• All Kjeldahl installations require acid-vapor removing devices. This may be by a
fume removal manifold or exhaust-fan system, water re-circulation or a fume
cupboard.

The chemical needs for the procedure are as follows:
• Kjeldahl catalyst: contains 10 g of K2SO4 plus .30 g of CuSO4.
• Reagent grade, concentrated H2SO4
• Mixed indicator solution: 3125g methyl red and .2062 g methylene blue in 250 ml
of 95% ethanol (stirred for 24 hours).
• Boric Acid Solution: 522 g U.S.P. boric acid in 18 l of deionized water. Add 50 ml of
mixed indicator solution and allow stirring overnight.
• Zinc: powdered or granular, 10 mesh.
• Sodium hydroxide: 50% wt/vol. aqueous (saturated).
• Standardized .1 N HCl solution.

The procedure is as follows:
• Weigh a 1 g sample and transfer into an ash free filter paper, and fold it to prevent
loss of sample.
• Introduce one catalyst in the Kjeldahl flask.
• Add 25 ml of reagent grade, concentrated H2SO4 to each Kjeldahl flask.
• Start the digestion by pre-heating the digester block to 370°C, and then place the
Kjeldahl flaks on it for 3 hours.
• After removing flasks from the digester, and once they are cool, add 400 ml of
deionized water.

35


• Prepare the receiving flask for steam distillation by adding 75 ml of prepared boric
acid solution to a clean 500 ml Erlenmeyer flask and place on distillation rack shelf.
Place delivery tube from condenser into the flask.
• Turn the water on the distillation system and all the burners on.
• Prepare the sample for distillation by adding approximately .5 g of powdered zinc
to flask, mix thoroughly and allow to settle.
• After digest has settled, measure 100 ml of saturated, aqueous NaOH (50% wt/vol)
into a graduated cylinder. Slant Kjeldahl flask containing prepared digest solution
about 45° from vertical position. Pour NaOH slowly into flask so that a layer forms
at the bottom. All these operations need to be performed wearing gloves and a
face mask.
• Attach flask to distillation-condenser assembly. Do not mix flask contents until
firmly attached. Holding flask firmly, making sure cork is snugly in place, swirl
contents to mix completely. Immediately set flask on heater. Withdraw receiving
flask from distillation-condenser delivery tube momentarily to allow pressure to
equalize and prevent back suction.
• Continue distillation until approximately 250 ml of distillate has been collected in
receiving flask.
• Turn heater off. Remove receiving flask partially and rinse delivery tube with
deionized water, collecting the rinse water into receiving flask.
• Replace receiving flask with a beaker containing 400 ml of deionized water.
This water will be sucked back into the Kjeldahl flask as it cools, washing out the
condenser tube.
• Titrate green distillate back to original purple using 0.1 N HCl and record volume
of acid used in titration.
• It is recommended to use a couple of blanks and controls or standards on every
run. Blanks - Kjeldahl reagents generally contain small amounts of nitrogen, which
must be measured and corrected for in calculations. Prepare blanks for dry
samples by folding one ash free filter paper and placing it into the Kjeldahl flask.
Treat blanks exactly like samples to be analyzed.

Standards: weigh two 0.1 g samples of urea, transfer into an ash free filter paper and
treat exactly like the rest of samples. Calculate percent recovery of nitrogen from
urea and make sure the obtained result is the one expected.

The calculation is:
(ml of acid – ml of blank) x normality x .014 x 6.25 x100
Crude protein, % = x 100
Original weight

A more recent and alternatively way to determine nitrogen content is by the
Dumas method. The method requires very little sample but the sample size will differ
with the type of ingredient to be analyzed. Sample size depends largely on the
expected level of crude protein in the material. In the case of soybean products a
sample size of 50 – 150 mg is recommended (AOAC, 2000). The sample is placed in a

36


tin foil cup for subsequent burning at 850 - 900°C to determine the amount of N2
by nitrometer. This method has the advantage over the Kjeldahl that is faster, better
suited for automation and creates little residues. However, the Kjeldahl method
continues to be the reference method. Total Dumas nitrogen can be slightly higher
than values obtained with the classical Kjeldahl method. However, for most purposes,
especially in the case of soy products, the difference is extremely small.

Crude protein can also be predicted by NIR, with an acceptable relative standard
deviation of about 0.42% (see Chapter 9).
CONTENTS

8.4 Protein quality
Protein quality is a function of the amino acid profile and the proportion of each
amino acid that is available to the animal. When soybean meals are intended for
monogastric feeding it is well known that proper heat processing has a dramatic
positive effect on amino acid digestibility, consequence of the destruction of anti-
nutritional factors (Table 1). However, over-heating can result in a decrease in both
concentration (Table 9) and digestibility of several amino acids, especially lysine.
The reduction in digestibility is due to the Maillard reaction which binds free amino
acids to free carbonyl groups (i.e., from carbohydrates). The Maillard reaction-end
products are not bio-available for all livestock species.

Table 9
Effect of heat processing on amino acid digestibility of raw
soybeans in poultry (adapted from Anderson-Haferman et al., 1992)
Autoclaving (minutes) Lysine Methionine Threonine
0 73 65 64
9 78 70 68
18 87 86 82

Table 10
Effect of heat-processing soybean meals on amino acid
concentration (adapted from Parsons et al., 1992)
Autoclaving (minutes) Lysine % Methionine % Cystine % Threonine %
0 3.27 0.70 0.71 1.89
20 2.95 0.66 0.71 1.92
40 2.76 0.63 0.71 1.87

There are several methods (Table 12-page 41) to determine protein quality of soybean
products for monogastric species.
CONTENTS

37


8.4.1. Urease Index
The urease index (AOCS, 1980) is the most common test used to evaluate the
quality of the soybean processing treatment. The method requires a pH meter,
volumetric flasks (250 ml), a small water bath that allows maintenance of
temperature at 30°C for at least 30 minutes, test tubes and a pipette.

The method determines the residual urease activity of soybean products as
an indirect indicator to assess whether the anti-nutritional factors, such as trypsin
inhibitors, present in soybeans have been destroyed by heat processing.
Both enzymes, urease and trypsin inhibitor, are deactivated during heating.
The laboratory method for urease involves mixing soybean meal with urea and
water for one minute.
Procedure:
• Place 0.2 g of soybean sample in a test tube.
• Add 10 ml of a urea solution (30 g of urea into 1 l of a buffer solution,
composed of 4.45 g of Na2HPO4 and 3.4 g of KH2PO4).
• Place the test tube in a water bath at 30°C for 30 minutes.
• Determine pH and compare it with the original pH of the urea solution.

The test measures the increase in pH consequence of the release of ammonia,
which is alkaline, into the media arising from the breakdown of urea by the urease
present in soybean products (urea is broken down into ammonia and carbon
dioxide). Depending on the protocol used, the endpoint is determined differently.
In the American Oil Chemists Society (AOCS, 1980) method, the endpoint is
determined by measuring the increase in pH of the sample media. In the EEC
method, the endpoint reflects the amount of acid required to maintain a constant
static pH. Results of these two methods differ slightly from one another.

The optimum pH increase is considered to be between 0.05 (McNaughton et al.,
1980) and 0.20 (Waldroup et al., 1985). Usually, all overheated samples yield urease
indexes below 0.05, but that does not imply that all samples with urease tests
below 0.05 have been overheated. It is recommended that, when using soybean
products for swine or poultry the increase in pH is not greater than 0.35 (Waldroup
et al., 1985). Animal performance is severely impaired with urease indexes above
1.75 pH units.

The urease test is useful to determine whether the soybean has been
sufficiently heated to deactivate anti-nutritional factors, but it is not a good
indicator to assess whether the soybean product has received an excessive heat
treatment.
CONTENTS

38


8.4.2. KOH Protein Solubility
This method consists of determining the percentage of protein that is solubilized
in a potassium hydroxide (KOH) solution (Araba and Dale, 1990). The method
requires volumetric flasks (250 ml), a small magnetic stirrer, filtering funnels or a
centrifuge, and the Kjeldahl equipment to measure nitrogen.
Procedure:
• Determine nitrogen content of soybean sample using official methods.
• Place 1.5 g of soybean sample in 75 ml of a 0.2% KOH solution (.036 N,
pH 12.5) and stir at 8,500 rpm for 20 minutes at a temperature of 22°C.
• Then, about 50 ml is taken and immediately centrifuged at 2500 x g for
15 minutes.
• Take aliquot of about 10 ml to determine nitrogen content in the liquid
fraction by Kjeldahl method.
• The results are expressed as a percentage of the original nitrogen content
of the sample.

The KOH protein solubility is not sensitive enough to gauge the level of heat
processing that a soybean product has undergone, but it is effective in
differentiating overheated products from correctly processed ones.

Table 11
Effect of autoclaving soybean meal on chick performance
(1-18 days), KOH protein solubility and urease activity
(adapted from Araba and Dale, 1990)
Autoclaving Weight KOH protein Urease Index
(120°C) gain Feed : gain solubility (pH units
minutes g ratio % change)
0 450a 1.79c 86.0 0.03
5 445a 1.87bc 76.3 0.02
10 424a 1.83bc 74.0 0.00
20 393b 1.89b 65.4 0.00
40 316c 2.04b 48.1 0.00
80 219d 2.55a 40.8 0.00

a, b, c, d Means within a column with common superscripts are not significantly
different (P < 0.05).

The solubility values have been correlated with growth rates in poultry and swine
(Lee and Garlich, 1992; Araba and Dale, 1990), with a clear decline in performance
with solubility values below 72%. Raw soybeans and well heat-processed soybean
products should have a protein solubility around 90% (that is 90% of the protein
present in the product is solubilized in a KOH solution).
CONTENTS

39


8.4.3. Protein Dispersibility Index (PDI)
Among the available tests for determining protein quality in soybean products,
the PDI is the simplest, most consistent, and most sensitive method. This test
measures the solubility of soybean proteins in water and is probably the best
adapted to all soy products. The PDI method measures the amount of soy protein
dispersed in water after blending a sample with water in a high-speed blender.
The water solubility of soybean protein can also be measured with a technique
called Nitrogen Solubility Index (NSI). Thee two methods differ in the speed and
vigor at which the water containing the soybean product is stirred. In animal
nutrition the PDI method is used.

Both methods require a blender (8,500 ppm), filtering funnels or a centrifuge,
and the routine Kjeldahl equipment for N analysis.
Procedure:
• Determine nitrogen content of soy sample using official methods.
• Place a 20 g sample of a soybean product in a blender.
• Add 300 ml of deionized water at 30°C.
• Stir at 8,500 rpm for 10 minutes (AOCS, 1993a).
• Filter and centrifuge for 10 minutes at 1000g.
• Analyze nitrogen content of the supernatant.
• The results are expressed as a percentage of the original nitrogen content of
the sample.

The NSI method uses a 5 g soybean sample into 200 ml of water at 30°C
stirred at 120 rpm for 120 minutes (AOCS, 1989). With either method, the final step
consists of determining the nitrogen content of the liquid fraction and the results
are expressed as a percentage of the original nitrogen content of the sample.

Nowadays, most soybean producers and users of soy products advocate the
PDI method as the best for assessing protein quality in soybean meals. Because
this test is more recent it is often used as a complement to the urease and KOH
solubility measurements. As a matter of fact, the PDI method has proven to be
especially useful in determining the degree of under heating soybean meals to
remove ANF. Furthermore, Batal et al. (2000) described a greater consistency in
the results of heating of soy flakes obtained with the PDI procedure than those
from urease or protein solubility. Since the work of Batal et al. (2000) which
recommended PDI values below 45 % recommendations have shifted slightly
under the influence of practical experience. Consequently, current
recommendations are for soybean meals with PDI values between 15 and 30 %,
KOH solubilities between 70 and 85 % and a urease index of 0.3 pH unit change
or below. These meals are considered adequately heat processed, without
under- nor over-processing.

40


Table 12
A brief description of available methods to determine
protein quality of soybean meal

Urease Index
1. Mix 0.2 g of soybean meal with 10 ml of urea solution (3% of urea)
2. Place in 30°C water bath for 30 minutes
3. Determine pH
4. Calculate pH increase (final pH - initial pH)

KOH Protein Solubility
1. Mix 1.5 g soybean meal with 75 ml of 0.2% KOH solution and stir for
20 minutes
2. Centrifuge at 2,500 x g for 20 minutes
3. Measure soluble nitrogen in the liquid fraction

Protein Dispersibility Index (PDI)
1. Mix 20 g of soybean meal with 300 ml of deionized distilled water
2. Blend at 8,500 RPM for 20 minutes at a temperature of 22°C.
3. Centrifuge (1000 x g for 10 minutes) or filter and measure nitrogen content
of the liquid fraction

Nitrogen Solubility Index (NSI)
1. Mix 5 g of soybean meal with 200 ml of water
2. Stir at 120 RPM for 120 minutes at 30°C
3. Centrifuge at 1,500 RPM and measure soluble nitrogen in the liquid fraction

Absorbance at 420 nm
1. The supernatant (if centrifuged) or the liquid fraction (if filtered) from the
PDI technique is diluted 80 times.
2. Filter through .2 µm pore size filter.
3. Read the absorbance of the clear filtrate at 420 nm with a spectrophotometer.

(Adapted from Dudley-Cash, W.A, 1999)

All these assays will give slightly different results depending on the particle size
of the sample used, temperature of the solutions and centrifugation speeds and
times. For example, protein solubility indexes will yield greater values as mean
particle size decreases (Parsons et al., 1991; Whitle and Araba, 1992). Therefore, it
is recommended to grind the sample at a consistent mesh size (1 mm), and to
maintain (at least within the same laboratory and company) rigorously the same
duration for treating the samples in the respective solutions and for
centrifugation.
CONTENTS

41


8.4.4. Protein quality in ruminants
For ruminants, protein quality of soybean meals will depend on its rumen
degradation and its intestinal digestion. The trypsin inhibitor factors present in
soybeans are irrelevant in ruminants, as they are mostly inactivated in the rumen
(Caine et al., 1998).

Amino acids are supplied to the duodenum of ruminants by microbial protein
synthesized in the rumen, undegraded dietary protein, and endogenous protein.
Microbial protein usually accounts for a substantial portion of the total amino
acids entering the small intestine. Ruminal degradation of protein from dietary
feed ingredients is one of the most important factors influencing intestinal amino
acid supply to ruminants. Soybean meal is extensively degraded in the rumen,
providing an excellent source of degradable intake protein for the ruminal
microbes, but not enough undegradable protein to meet the demands of high
producing ruminants. Because soybeans contain a high quality protein with a
good amino acid profile and they are highly digestible in the small intestine,
various processing methods and treatments have been used to increase its
undegradable protein value. The most common methods for protecting soybean
proteins from ruminal degradation are heat application, incorporating chemicals
such as formaldehyde or a combination of heat and chemicals such as
lignosulfonate combined with xylose.

To assess the extent of protein degradation of a soybean product several
techniques are available.
CONTENTS

8.4.4.1. In situ technique

Although this technique is relatively expensive, labor intensive, and requires
access to rumen cannulated animals, it is very useful to determine the rate of
degradation of proteins from soybeans. This technique requires consecutive
times of ruminal incubation of the samples under study so that the rate of
protein degradation can be determined. The in situ technique determines
degradation of the insoluble fraction only. The soluble fraction is considered to
be totally and instantaneously degraded. To accurately predict rate of protein
degradation, sufficient time points must be included in early as well as later
stages of degradation (Figure 2).

42


Figure 2
Protein disappearance from soybean meal and curve peeling processa

100 –

70 –
CP remaining, % of CP

Crude protein disappearance
Rapidly degradable pool
50 – Slowly degradable pool
Observed values

20 –

0 –I I I I I
0 20 40 60 80
Time ( hours)

Adapted from Bach et al. (1998).

After ruminal incubation, the data are fitted to different models to determine the
rate of protein degradation in the rumen. Bach et al. (1998) studied the effects of
different mathematical approaches (curve peeling, linear and nonlinear regression)
to estimate the rate of protein degradation in soybean samples and concluded
that using curve peeling (Shipley and Clark, 1972) allowed for the best separation
of the different protein pools in soybean proteins. CONTENTS

8.4.4.2. In vitro technique

There are several in vitro methods that require the use of rumen fluid, such as the
Tilley and Terry (1963) technique, or the in vitro inhibitor technique (Broderick,
1987). Like the in situ technique, these two methods present the disadvantage that
they require access to cannulated animals. The in vitro technique consists of
incubating a small feed or ingredient sample with strained rumen fluid and a
buffer under anaerobic conditions in a test tube or container. The test tube or
container is located in a water bath that is maintained at 37 – 38°C throughout the
incubation.

43


At regular, pre-determined intervals a sample is removed from the incubator,
centrifuged and analyzed for dry matter and nitrogen disappearance (using the
Kjeldahl method). Data are analyzed as described for the in situ technique.

There are a number of enzymatic techniques which have the important
advantage that they are completely independent of the animal, and should result
in less variation, making this technique relatively simple to standardize.

The most common enzymatic techniques are the Ficin technique (Poos-Floyd et
al., 1985) and the Streptomyces griseus technique (Nocek et al., 1983). The biological
value of the results from these techniques may be limited due to incomplete
enzymatic activity compared with the ruminal environment. Mahadevan et al.
(1987) found large differences when comparing digestion of different protein
sources using protease from Streptomyces griseus with an extract of ruminal
microbial enzymes. Chamberlain and Thomas (1979) reported that, although rate
constants can be calculated using these proteases, these results do not always
rank proteins in the same order as degradabilities estimated in vivo. When using
enzymatic techniques to predict microbial fermentation in the rumen, it is crucial
that the enzyme concentration is sufficient to saturate the substrate.
Some researchers have attempted to use near infrared reflectance spectroscopy
(NIR) to estimate protein degradation of feedstuffs in the rumen. Tremblay et al.
(1996) evaluated NIR as a technique for estimating ruminal CP degradability of
roasted soybeans and found a coefficient of determination between NIR and
undegraded protein estimated by the inhibitor in vitro technique of .70. However,
the use of NIR for this purpose would require continuous access to cannulated
animals to maintain the prediction equations.
CONTENTS

8.5. Amino Acids
Determining the amino acid composition of proteins is essential to characterize
their biological value. The greater the proportions of essential amino acids the
greater the biological value of a protein.

The amino acid analysis requires the use of high performance liquid
chromatography (HPLC) or the combination of commercial kits and gas
chromatography (GC). The analysis involves four steps:
• Hydrolysis (using HCl or barium hydroxide); this breaks the peptide bonds and
releases the free amino acids.
• Separation; column chromatography separates amino acids on the basis of their
functional groups.

44


• Derivatization; a chromogenic reagent enhances the separation and spectral
properties of the amino acids and is required for sensitive detection.
• Detection; a data processing system compares the resulting chromatogram,
based on peak area or peak height, to previously known and calibrated standard.

HPLC analysis for amino acids is a highly specialized laboratory procedure
requiring skilled personnel and sophisticated equipment. For amino acid analysis the
sample preparation is critical and differs with the type of ingredient and the amino
acid of major interest. Most amino acids can be hydrolyzed by a 23 or 24 h hydrolysis
in HCl (6 mol/l). For sulfur amino acids hydrolysis should be preceded by performic
oxidation and for tryptophane a hydrolysis with barium hydroxide (1.5 mol/l) for 20 h
is required. In general it is recommended to use a specialized laboratory to conduct
the amino acid analysis.
CONTENTS

8.6. Crude Fiber
The original method was intended to quantify the materials in the feed that form
part of the cell wall and provide relatively low energy as their digestibility is usually
low. Thus, the technique was meant to quantify cellulose, certain hemicelluloses and
lignin. However, later it was shown that crude fiber also included pectines, and that
not all the lignin was recovered in the crude fiber fraction. The major disadvantage
of this technique is that hemi-cellulose, lignin and pectines are inconsistently
accounted for.

The method requires the following reagents:
• Sulfuric acid solution, .255N, 1.25 g of H2SO4/100 ml.
• Sodium hydroxide solution, .313N, 1.25 g of NaOH/100 ml, free of Na2CO3.
• Alcohol - Methanol, isopropyl alcohol, 95% ethanol, reagent ethanol.
• Antifoam agent (n-octanol).

Equipment:
• Digestion apparatus.
• Ashing dishes.
• Desiccator.
• Filtering device (Buchner filter).
• Suction filter: To accommodate filtering devices. Attach suction flask to trap
in line with vacuum source.
• Vacuum source with valve to break or control vacuum.

45


The procedure described by the AOAC (1980) can be summarized as follows:
• Weigh 2 g of sample (A). Remove moisture and fat using ether (removing fat is not
necessary if the sample has less than 1% ether extract).
• Transfer to a 600 ml beaker, avoiding fiber contamination from paper or brush.
Add approximately 1 g of prepared asbestos, 200 ml of boiling 1.25% H2SO4 ,
and 1 drop of diluted antifoam. Avoid using excessive antifoam, as it may
overestimate fiber content.
• Place beaker on digestion apparatus with pre-adjusted hot plate and boil for
30 minutes, rotating beaker periodically to prevent solids from adhering to sides.
• Remove beaker and filter as follows:
– Filter through Buchner filter and rinse beaker with 50 to 75 ml of boiling water.
– Repeat with three 50 ml portions of water and apply vacuum until the sample
is dried. Remove mat and residue by snapping bottom of Buchner against top,
while covering stem with the thumb and replace in beaker.
– Add 200 ml of boiling 1.25% NaOH, and boil 30 more minutes.
• Remove beaker and filter as described above. Wash with 25 ml of boiling 1.25%
H2SO4, three 50 ml portions of H2O, and 25 ml of alcohol.
• Dry mat and residue for 2 h at 130°C.
• Remove, place in desiccator, cool, weigh and record (B).
• Remove mat and residue, and transfer to an ashing dish.
• Ignite for 30 minutes at 600°C. Cool in desiccator and reweigh (C).
• Calculate crude fiber content on dry matter basis as:
weight after acid and base extraction (B) – weight after ashing(C)
Crude fiber, % = x 100
Original weight (A) x % dry matter
CONTENTS

8.7. Neutral Detergent Fiber (NDF)
Neutral detergent fiber (NDF) accounts for the cellulose, hemicellulose and lignin
content of soybean products. These fractions represent, most of the fiber or cell wall
fractions of soybean products, with the exemption that pectines are not included in
the NDF fraction.

The neutral detergent fiber (NDF) was first described by Goering and Van Soest
(1970) and later modified by Van Soest et al. (1991). The NDF determination requires
a refluxing apparatus 600 ml and Berzelius beakers.
The technique is as follows.
Reagents:
• NDF solution: dilute 30 g of sodium lauryl sulfate, 18.61 g of disodium
dihydrogen ethylene diamine tetra acetic dihydrate, 6.81 g of sodium borate

46


decahydrate, 4.56 g of disodium hydrogen phosphate, 10 m of triethylene
glycol 65 in 1 l of deionized water.
• Acetone.

The Goering and Van Soest (1970) procedure for NDF determination is as follows:
• Weigh 0.5 to 1.0 g sample (to precision of ± 0.0001 g) in a 600-ml Berzelius
beaker (A).
• Add 100 ml of neutral detergent fiber solution.
• Heat to boiling (5 to 10 min). Decrease heat as boiling begins.
Boil for 60 minutes.
• After 60 minutes, filter contents onto a pre-weighted, ash-free filter paper (B)
under vacuum. Use low vacuum at first, and increase it as more force is needed.
• Rinse contents with hot water, filter, and repeat twice.
• Wash twice with acetone.
• Dry at 100°C in forced air oven for 24 h.
• Cool filter paper and sample residue in desiccator; weigh and record (C).
• Fold filter paper and place in a pre-weighted aluminum pan.
• Ash in muffle at 500°C for 4 h.
• Cool in desiccator. Weigh and record (D).
The NDF content on a dry matter basis is then calculated as:
(Weight of NDF residue, C – Weight of filter paper, B) - Weight after ashing, D
NDF, % = x100
Original weight of sample, A x % Dry matter

For the Ankom system the following procedure applies:
• Number filter bags.
• Weigh 0.5 g sample in filter bag, record exact weight (± 0.0001 g) (A) and one
blank bag (included in extraction to determine blank bag correction).
• Seal bags within 0.5 cm from the open edge.
• Spread sample uniformly inside the filter bag by shaking and lightly flicking the
bag to eliminate clumping.
• Pre-extract soybean products containing more than10% fat with acetone
• Place bags containing samples in a 500 ml bottle with a screw cap. Fill the bottle
with acetone into bottle to cover bags (at least 15 ml/bag) and secure top.
Swirl gently after 3 and 6 min has elapsed and allow bags to soak for a total of
10 min. Repeat with fresh acetone.
• Pour out acetone, press bags gently between two layers of absorbent paper, and
place bags in a hood to air dry for at least 15 min.
• Place 24 bags in the suspender, putting 3 bags per basket.
• Stack baskets on center post with each basket rotated 120°C.
• Include one standard and one blank.
• Place duplicate samples in separate batches and in reverse order of top to bottom

47


• Bring center post with bags in the vessel and agitate lightly to remove air.
• Close the vessel and boil at 100°C for 60 minutes.
• Drain liquid from vessel.
• Add 2 liter of boiling water to vessel along with 4 ml thermamyl and continue
to boil for 5 minutes. Drain and repeat this part of the procedure with 2 ml of
thermamyl.
• Drain, remove bags and squeeze excess water carefully.
• Clean bags with acetone and again squeezing bags carefully.
• Leave bags to air dry for 30 minutes.
• Dry bags for 8 hours at 103°C and cool afterwards in desiccator. Weigh (B).
• Weigh blank bag (C).
• Ash bags on pre-registered and weighed aluminum pan (D); Db for blank) for
6 hours at 550°C in muffle furnace, cool, place in desiccator and weigh blank (E)
and pans with samples (F).
The NDF content (dry matter basis) is then calculated as:
(B – C) – (F –D) – (E – Db)
NDF, % = x100
A x % Dry matter
CONTENTS

8.8. Acid Detergent Fiber (ADF)
It is recommended that ADF is determined sequentially, that is using the residue
left from NDF determination. If not done sequentially, some fractions of pectines
and hemicellulose could contaminate and overestimate the ADF fraction. For doing
sequential analysis, the Ankom procedure is recommended. Like for the NDF
procedure the ADF analysis requires 600 ml Berzelius beakers, a fiber digestion
apparatus and a filtering flask. Also sintered glass crucibles of 40 to 50 ml with coarse
porosity are required.
Reagents needed are:
• Acid Detergent Solution. For this add 27.84 ml of H2SO4 to a volumetric flask
and bring to 1 l volume with deionized water (it is recommended that before
adding the acid, some water is placed in the volumetric flask). Then add 20 g of
CH3(CH2)15N(CH3)3Br to this solution.
• Acetone.
• 72% H2SO4 standardized to specific gravity of 1.634 at 20°C.
Extraction of sample
• Transfer 1 (± 0.0001) g air-dried sample to Berzelius beaker (A).
• Add 100 ml acid detergent solution.
• Heat to boil (5 to 10 minutes), and then boil for exactly 60 minutes.
• Filter with light suction into previously tared crucibles.

48


• Wash with deionized hot water 2 to 3 times.
• Wash thoroughly with acetone until no further color is removed. Suction dry.
• Dry in oven at 100°C for 24 h.
• Cool in desiccator. Weigh and record weight (B).
• Cool in desiccator. Weigh and record (C).
The ADF content on a dry matter basis is then calculated using the following
equation:
Weight of ADF residue and crucible, B – Weight after ashing, C
ADF, % = x100
Original weight , A x % Dry matter

CONTENTS

8.9. Lignin
Lignin is a polymer of hydroxycinnamyl alcohols that can be linked to phenolic acids,
and also non-phenolic compounds. Lignin acts like a shield that prevents the action
of enzymes and bacteria, by physical means. Lignin, not only is totally indigestible,
but also limits digestion of some nutrients (especially fiber fractions) of soybean
products. The determination of lignin is thus, important to estimate the digestibility
and energy value of certain, fiber-rich, soybean products.
There are two methods to determine lignin, the Klason lignin and the permanganate
lignin. The method of choice is the Klason lignin.
CONTENTS

8.9.1 Klason lignin
Klason lignin requires 72% sulfuric acid and sintered glass crucibles.
The technique consists of adding 25 ml of sulfuric acid to the residue of an ADF
determination (without ashing), filtering and adding distilled water three times.
Procedure:
• Place ADF crucible in a 50 ml beaker on a tray. For the original weight use same
as for ADF analysis (A).
• Cover contents of crucible with 72% H2SO4. (Fill approximately half way
with acid).
• Stir contents with a glass rod to a smooth paste.
• Leave rod in crucible, refill hourly for 3 h, each time stirring the contents of
the crucible.
• After 3 h, filter contents of crucible using low vacuum at first, increasing
progressively as more force is needed.
• Wash contents with hot deionized water until free of acid (minimum of
five times).

49


• Rinse rod and remove.
• Dry crucible in oven at 100°C for 24 h.
• Cool in desiccator. Weigh and record weight (B).
• Cool in desiccator. Weigh and record (C).
Calculate Klason lignin (on dry matter basis) as:
Weight of lignin residue and crucible, B – Weight after ashing, C
Lignin, % = x100
Original weight , A x % Dry matter
CONTENTS

8.9.2. Permanganate lignin
The permanganate lignin requires 80% ethanol, a permanganate buffer
solution, acetone, fiber crucibles and a Fibertec apparatus or a vacuum system.
The permanganate buffer solution consists of 2 parts of potassium permanganate
and one part of lignin buffer solution. The lignin buffer solution in turn is made
up of : 300 ml of distilled water, 18 g of ferric nitrate, .45 g of silver nitrate, 1.5 l of
glacial acetic acid, 15 g of potassium acetate and 1.2 l of tertiary butyl alcohol.
• Determine ADF following the above-described procedure using crucibles (not
Ankom) (B). For the original weight, use same as for ADF analysis (A).
• Place crucibles with ADF digested samples (not ashed) on an enamel pan.
• Fill the pan with distilled water to the bottom of the filter plate of the crucible.
• Place a stirring rod in each crucible and gently break the matt residue with
a little of distilled water.
• Fill the crucibles about half way, with the permanganate-buffer solution.
Stir, and keep filling crucibles as solution drains out.
• Leave the permanganate solution on for 90 minutes, stirring occasionally.
• Filter the permanganate using the vacuum system of the Fibertec.
• Place crucibles on another enamel pan.
• Fill crucibles with distilled water (avoiding overflow) and refill as necessary.
• Add demineralizing solution to the samples and leave until they turn white.
• Place on cold extractor and filter the demineralized solution using vacuum.
• Wash with 80% ethanol 2 to 3 times.
• Rinse with acetone. Air dry.
• Place in a 105°C oven overnight.
• Place in desiccator, cool, weigh and record weights (C).
Calculate Permanganate lignin (on dry matter basis) as:
Weight of ADF residue and crucible, B – Weight after oxidation, C
Lignin, % = x100
Original weight , A x % dry matter
CONTENTS

50


8.10. Starch
Starch occupies only a small part of most soy products but the nitrogen free
extract (NFE) fraction- with which it is often identified – may represent a large part of
the product. Chemically speaking, starch is defined as a polymer of linear alpha-1,4
linked glucose units (amylose) or alpha-1,5 branched chains of alpha-1,4 linked
glucose units (amylopectine).

The starch content of soybean products can be determined with a large number
of methods of which the most common methods are the polarimetric method and
the enzymatic. The polarimateric method, also referred to as the Ewers method, will
recuperate free sugars, pectins and a selection of non-starch polysaccharides.
It is generally recommended not to use this method for samples high in the above
mentioned substances or rich in optically active substances that do not dissolve in
ethanol (40%) (v/v). The most common alternative method of starch determination is
the enzymatic method. This method is based on the selective enzymatic digestion of
amyloses and amylopectins by an amylo-glucosidase.

The polarimatric method and the various enzymatic methods do not generally
provide the same numeric starch value for an ingredient, feed or digesta sample.
The Ewers value being generally higher. However, the enzymatic method(s) are more
accurate and are better in discriminating between true starch and related molecules.
A comparison of starch analysis in the CVB (2000) tables shows that the two
methods give close to identical results for ingredients high in starch. For raw
materials with low to intermediate starch levels and ingredients rich in NSPs or cell
wall components, starch determination is higher with the Ewers method compared
to the enzymatic method. Consequently, for soy products high in (soluble) sugar
content (see appendix Tables 1, 2) the polarimatric method will result in higher
values than the enzymatic method and the enzymatic method should be preferred.
CONTENTS

8.10.1 Polarimatric starch determination
The Polarimetric method requires: Erlenmeyers volumetric flasks, pipettes, filter
paper, a water bath, and a polarimeter or saccharo-meter plus the following
reagents:
• 2.5% HCl.
• 1.128% HCl (this solution must be verified by titration with a 0.1 N NaOH
solution in presence of 0.1% (w/v) methyl red in 94% (v/v) ethanol.
• Carrez solution I: made by dissolving 21.9 g of zinc acetate and 3 g of glacial
acetic acid into 100 ml of water.

51


• Carrez solution II: dissolve 10.6 g of potassium ferro-cyanide in 100 ml of
deionized water.
• 40% (v/v) ethanol.

The polarimetric procedure has two parts, the total optical rotation and the
determination of the optical rotation of the dissolved substances in 40% ethanol:

Total optical rotation determination:
• Weigh 2.5 g of soybean sample previously ground through a 5-mm mesh into
a 100 ml volumetric flask.
• Add 25 ml of HCl and stir to obtain a homogenized solution and add 25
additional milliliters of HCl.
• Immerse and continuously shake the volumetric flask in a boiling water bath
for 15 minutes.
• Remove the flasks from the water bath, add 30 ml of cold water and
immediately cool to 20°C.
• Add 5 ml of Carrez solution I and stir for 1 minute.
• Add 5 ml of Carrez solution II and stir, again, for 1 additional minute.
• Add water to the 100 ml level.
• Measure the optical rotation of the solution in a 200 mm tube with the
polarimeter or saccharo-meter.

Optical rotation determination of dissolved substances in 40% ethanol:
• Weigh 2.5 g of soybean sample previously ground through a 5-mm mesh into a
100 ml volumetric flask.
• Add 80 ml of 40% ethanol and let react for 1 hour at room temperature, stirring
every 10 minutes.
• Complete to volume (100 ml) with ethanol, stir and filter.
• Pipette 50 ml of the filtrate into a 250 ml Erlenmeyer.
• Add 2.1 ml of HCl and shake vigorously.
• Place Erlenmeyer (with cooling device) in a boiling water bath for exactly
15 minutes.
• Transfer the sample into a 100 ml volumetric flask.
• Cool and maintain at room temperature.
• Clarify the sample with Carrez solution I and II and fill to the 100-ml level
with water.
• Filter and measure optical rotation in a 200 mm tube with a polarimeter or
saccharo-meter.
• The starch content of the sample is then calculated using the following
equation:
2000 x (total rotation – dissolved rotation)
Starch, % =
Specific optical rotation of pure starch

52


The specific optical rotation of pure starch will depend on the type of starch
used. Table 13 depicts the generally accepted values for some common starch-rich
ingredients.

Table 13.
Optical rotation of various pure starch sources

Starch source Optical rotation
Rice starch 185.9º
Potato starch 185.4º
Corn starch 184.6º
Wheat starch 182.7º
Barley starch 181.5º
Oat starch 181.3º

CONTENTS

8.10.2. Enzymatic or colorimetric starch determination
The enzymatic method is much longer than the polarimetric one.

• Acetate buffer solution, .2 M at pH 4.5.
• Amyloglucosidase enzyme.
• Glucose reagent kit containing: NAD, ATP, hexokinase, glucose-6-phosphate,
magnesium ions, buffer and non reactive stabilizers and filters.
• Glucose standards. Prepare three solutions of 100 ml each with 100, 300, and
800 mg/dl of glucose, and 10, 30 and 300 mg/dl of urea nitrogen.

The total procedure takes three days.
Day on:
• Weigh 125 Erlenmeyer flaks are record their weight to the nearest tenth
of gram.
• Add 25 ml of distilled water.
• Add .1 g of soybean product and swirl gently.
• Place Erlenmeyers with samples on autoclave at 124°C and 7 kg of pressure,
once these conditions are reached, leave the samples in the autoclave for
90 minutes.
• Turn autoclave to liquid cool and leave sample in the autoclave overnight.

53


Day two:
• Remove from autoclave and cool to room temperature.
• Add 25 ml of acetate buffer and swirl gently.
• Add .2 g of amylo-glucosidase enzyme and swirl.
• Cover tight with aluminum foil caps and put in drying oven at 60°C for 24 hours.
Day three:
• Remove flasks from oven and let to cool at room temperature.
• Remove foil caps and weigh to the nearest tenth of gram and record weight.
• Pour contents into 50 ml centrifuge tubes and centrifuge at 1000 x g for
10 minutes.
• Save supernatant in a plastic scintillation vial.
• Prepare a standard curve using the standard solutions:

Table 14.
Solutions to prepare standard curve for colorimetric
starch determination
Working standards Combined standards
50 1:1 dilution of 100 mg/dl standard and water
100 Use 100 mg/dl standard
200 1:3 dilution of 800 mg/dl standard and water.
300 Use 300 mg/dl standard
400 1:1 dilution of 800 mg/dl standard and water
800 Use 800 mg/dl

• Set up a series of test tubes for the color determination step. Include tubes for
standards and a blank (i.e. glucose reagent only).
• Prepare glucose reagent kit according to the instructions provided by the
supplier of the kit.
• Add 1.5 ml of glucose reagent agent into test tubes.
• Read and record absorbance at 340 nm vs water as a reference. This will be
INITIAL A (the blank) in the calculations.
• Add 10 µl of sample to the test tube. Mix gently.
• Incubate tubes for 5 minutes at 37°C.
• Read and record the absorbance at 340 nm vs water as a reference. This will be
FINAL A in the calculations.
• Subtract INITIAL A from FINAL A to obtain change in absorbance (∆ A in the
calculations).
• Calculate glucose concentration using the following equation:

FINAL A (sample) – INITIAL A (sample)
Glucose, mg/dl = standard concentration x
FINAL A (standard) – INITIAL A (standard)

54


• Calculate the content of alpha linked glucose polymers:
Alpha-linked glucose polymer, mg/g = Glucose concentration in standard x
(V/100) x (1/sample weight)
where, V is the flask volume difference (sample + flask weight - flask weight)

• Calculate starch content as:
Alpha linked glucose polymer, mg/g
Starch, % =
1.111
CONTENTS

8.11. Non starch polysaccharides (NSP) and
monosaccharides
A large part of the NFE fraction of soy products may belong to the group of
non-starch polysaccharides. This group is composed of fairly simple, soluble and
insoluble sugars, most notably raffinose, stachyose, β-mannans and xylans. A major
proportion of these sugars are not readily digested, especially by young animals and
they are thus often considered part of the ANF. Consequently, a correct estimation of
these sugars or the mono-saccharides that make-up these NSP is important when
formulating special diets.

The precise analysis for simple sugars requires HPLC equipment. The first part
of the procedure requires the elimination of starch from the sample. This is
accomplished with the following procedure:
• Weigh 2.5 g of sample in Hungate tubes.
• Add 2.5 ml of acetate buffer (70 ml 0.1 M sodium acetate and 30 ml of 0.1 M
acetic acid).
• Add 2.5 µm of α-amylase.
• Place in boiling water bath for 1 hour, shaking every 10 minutes.
• Cool to 40°C.
• Add 50 µl of glucosidase.
• Place in water bath at 60°C for 6 hours and shake every 30 minutes.
• Cool to room temperature.
• Add 10.5 ml of pure ethanol.
• Place in refrigerator for 1 hour.
• Centrifuge at 1000 x g for 5 minutes.
• Discard the supernatant, rinsing the pellet twice with distilled water.
• Dry overnight at 40°C.
The total NSP fraction can be estimated as follows:

Total NSP, % = 100 – (humidity, % + ash, % + protein,% + lipids,% + NDF,% + starch,%)

55


Once starch has been removed it is necessary to conduct the hydrolysis of sugars.
• Detach the sample from the tube walls.
• Add 1.5 ml of sulfuric acid (75 ml of 96% sulfuric acid and 25 ml of water).
• Vortex.
• Place in water bath 30°C for 1 hour.
• Transfer sample into a 100-ml Erlenmeyer and add 40 ml of distilled water.
• Add 5 ml of myo-inositol (2mg/l) as an internal standard.
• Cover Erlenmeyer with aluminum foil and autoclave (125°C) for 1 hour.
• Filter sample.
• Re-suspend the filtrate into 50 ml of distilled water.

After hydrolysis, the derivatization needs to be performed:
• Place 1 ml of filtrate into a 5-ml plastic test tube.
• Neutralize with 200 µl of 12 M ammonium hydroxide.
• Vortex.
• Add 100 µl of 3 M ammonium hydroxide containing 150 mg/ml of KBH4
(Borate is very toxic; all following steps must be conducted under a hood).
• Place in a 40°C water bath for 1 hour.
• Add 100 µl of glacial acetic acid and vortex.
• Transfer 500 µl into a 30 ml glass tube.
• Add 500 µl of 1-metilimidazol.
• 5 ml acetic acid, vortex and wait 10 minutes.
• Add 1 ml of ethanol, vortex and wait 10 minutes.
• Add 5 ml of distilled water.
• Add 5 ml of 7.5 M KOH, vortex, and wait 3 minutes.
• Add, again, 5 ml of 7.5 M KOH, vortex, and wait 3 minutes.
• Cover tubes.
• Take a 1-ml aliquot and transfer into a 5-ml test tube.
• Add 50 mg of anhydrous sodium sulfate.
• Decant supernatant into a GLC vial.
• Dry at 40°C for 8-10 hours
• Add 0.5 ml of chloroform.

Chromatography:
• Run samples against stand and blank through a gas chromatograph following
equipment-specific procedures.
CONTENTS

56


8.12. Ether Extract
The ether extract (EE) method measures the proportion of a feed that is soluble
in ether. It is equivalent to the total amount of lipids present in a feed and it
represents mostly true fats and oils. However, it also includes some ether-soluble
material that are not true fats, such as fat-soluble vitamins, carotenes, chlorophylls,
sterols, phospholipids, waxes and cutins.

Fatty acids will readily form insoluble complexes with free cations, most notably
calcium. These reactions may occur in soy products that have a relatively high
concentration of positively charged minerals. To assure that all the fat components
are extracted from a mineral rich sample it is recommended to perform an acid
hydrolysis in hot HCl prior to the ether extraction.

The EE technique requires a Soxhlet extraction system, funnels, filter paper, HCl
(3 N), and anhydrous diethyl ether.

• Weight approximately 2 g of sample ground trough 1 mm-mesh into an
Erlenmeyer.
• Add 100 ml of 3 N HCl and boil for 1 h.
• Cool at room temperature.
• Filter through a filter paper and rinse with distilled water to remove all HCl.
• Remove the moisture of the sample by drying it in an oven at 105°C for 24 hours.
(If the sample were not dried the ether would have difficulties penetrating all the
areas of the ingredient).
• Place sample with anhydrous diethyl ether in a Soxhlet extractor. Turn the heater
coil high enough to evaporate 2-3 drops of ether per second in the condenser.
Extract for 24 hours. After that time, the ether should be removed, and replaced
with clean ether, leaving the samples in the Soxhlet for 8 more hours.
• Remove from Soxhlet, air-dry for about 2 hours and oven dry at 105°C for
12 hours.
The calculation of crude fat is as follows:
Final weight after extraction, g
Crude fat, % = x100
Original weight, g

CONTENTS

8.13. Lipid quality
Fat or oil quality depends on the fatty acid profile, specific physical
characteristics and the oxidation level. While fatty acid characteristics and
composition determine the physical and nutritional quality of the true lipid fraction,

57


the physical characteristics and oxidation level are the aspects that are of greatest
importance in the routine QC procedures that are applied when oils or fats enter the
feed production process. Consequently, the two most common physical tests to
assess quality of oils are the insoluble impurities and the unsaponifiable matter.
Along with moisture in the oil or fat sample, these characteristics are collectively
referred to as the MUI (Moisture, Unsaponifiables, Insolubles) value.
CONTENTS

8.13.1. Moisture
Through the crushing and various treatments of soy oil water may settle in oil
samples especially if these samples have undergone significant temperature changes.
Generally the moisture content is small but it may have a large effect on the oil
quality.The procedure is simple but calls for a forced air drying oven capable of
maintaining 130°C ± 2°C, aluminium sample pans with tight fitting covers and a
desiccator. Attention, high temperatures may cause the fat sample to ignite.

The procedure is following:
• Accurately weigh 5.0 ± 01 g of sample into a tared moisture dish that has been
previously dried and cooled in a desiccator.
• Place the dish in a forced air oven and dry it for 30 min at 130°C + 1°C.
Remove from the oven, cool to room temperature in a desiccator and weigh.

Repeat until the loss in weight does not exceed 0.05% per 30 min drying period.
Loss in weight, g
Moisture content, % = x100
Weight of sample, g
CONTENTS

8.13.2. Insoluble impurities
This is a measure of the content of non-lipid compounds in oil. It should be
less than 1 %.

The method is as follows:
• Place 15 ml of sample into a graduate cylinder (if sample is not liquid it should
be liquefied applying a mild increase in temperature using a hot plate). Maintain
in liquid state for the duration of the test. The lower values of the tube should
be clearly identified to ensure easy reading following the procedure.
• Let the sample settle in the graduate cylinder for 24 hours.
• Observe the amount of insolubles that have settled out of the sample and
collected at both at the top and bottom of the tube.

58


• The insoluble impurities are then calculated as:
Reading of settled insolubles, ml
Insoluble impurities, % = x 100
Total sample volume, ml (15)

• If no insoluble matter is seen in the tube, report the insoluble matter as < 0.2%.
CONTENTS

8.13.3. Unsaponifiable matter
The method measures those substances which cannot be saponified by a
caustic alkali treatment. It includes compounds such as aliphatic alcohols, sterols,
pigments and hydrocarbons. They do not have a recognized energy value, and
thus are of little nutritional interest.

The technique (AOCS, 1993b) requires Erlenmeyer or Soxhlet flasks, beakers,
separator funnels, a balance(accuracy of ± .001g), pipettes, a water bath, a reflux
condenser, an explosion-proof hot plate, a 50ml burette with its stand, a Soxhlet
fat cup and Soxhlet HT2 system, and a desiccator.

The reagents for this method are:
• 85% Ethanol.
• Petroleum Ether.
• NaOH, ACS grade.
• Phenolphthalein indicator solution.
• 0.2 M HCL standard.
• Deionized water.
• Accurately weigh 5 ± 0.0001 g of well mixed sample into an extraction flask.
If the sample is fluid at room temperature, shake to mix well before weighing
out sample, and if the sample is solid at room temperature, melt the sample in a
water bath, set at 60°C, until the sample is liquefied. Remove and shake to
mix well.
• Add 30 ml of 85% ethanol to the sample.
• Add 5 ml of 45% aqueous potassium hydroxide.
• Assemble the extractor by turning on the hot plates and the water taps.
Reflux the solution gently but steadily for 1 hour or until completely saponified.
• Quantitatively transfer the solution to a 500 ml separator funnel and rinse the
flask into the funnel with approximately 10 ml of 85% ethanol.
• Wash the flask into the separator funnel with approximately 5ml of warm water
and pour it into the separator funnel.
• Add approximately 5ml of cool distilled water, swirl and pour it into the
separator funnel.

59


• Complete the transfer from the flask by rinsing with approximately 5ml of
petroleum ether.
• Allow the solution to cool to room temperature.
• Add approximately 50 ml of petroleum ether.
• Insert the stopper and shake vigorously by repetitions of inverting for at least
one minute. After every few seconds, release the accumulated pressure in the
funnel by inverting and opening the stopcock.
• Allow to settle until the solution has separated into two layers.
• Transfer the bottom fat layer back into the original flask and transfer the
petroleum ether layer into a clean 250ml Erlenmeyer flask.
• Repeat the former 4 steps until the petroleum ether layer is clear and colorless
(about 6 times).
• Once the washes are completed, discard the fat portion of the sample in a waste
container and transfer all of the petroleum ether to the 500ml separator funnel.
• Add 30ml of 10% ethanol to the petroleum ether.
• Insert the stopper and shake vigorously by repetitions of inverting for at least
one minute. Release any pressure in the funnel by inverting the funnel and
opening the stopcock.
• Allow the mixture to settle until the solution has separated into two layers.
• Draw off the alcohol, or bottom layer, and discard, being careful not to remove
any of the ether layer.
• Continue the alcohol washes until the petroleum ether layer is clear,
approximately 5 or 6 times or until the bottom layer no longer turns into a pink
color after addition of 1 drop of phenolphthalein indicator solution.
• Transfer 60 ml of the ether layer (top layer) to a previously tared Soxhlet fat cup.
• Evaporate the petroleum ether layer.
• Repeat the ether evaporation on the Soxhlet system from the same fat cup until
all petroleum ether has been completely evaporated from the separator funnel.
• Place the cup in the oven, set at 100°C, for approximately 20 minutes.
• Allow to cool to room temperature in a desiccator and weigh.
• After weighing, dissolve the residue in 50 ml of the phenolphthalein indicator
solution. Heat on a hot plate to the point where the alcohol is just starting to
boil, then transfer to a 250 ml Erlenmeyer flask.
• Titrate with standardized 0.02 N sodium hydroxide to a faint pink of the same
intensity as the original indicator solution. No titration is needed if the sample is
already pink when poured into the flask. The amount of ml added times 0.0056
will yield the weight of fatty acids in the sample.
• The unsaponifiable matter is calculated as follows:
(Weight of fat cup plus residue – Weight of fat cup) – Weight of fatty acids
Unsaponifiable matter, % =
Weight of sample

CONTENTS

60


8.13.4. Iodine value
The iodine value is an estimate of the proportion of unsaturated fatty acids
present in a sample. Iodine will bind to unsaturated or double bonds in fatty acids.
The greater the amount of iodine bound to the sample the greater the proportion
of unsaturated fatty acids. The procedure requires the following reagents:
• Glacial acetic acid.
• Carbon tetrachloride.
• Iodine trichloride.
• Iodine.
• Potassium iodide (100 g/l aqueous solution).
• Sodium thiosulfate, 0.1 N (19.76 g of sodium thiosulfate into 230.24 ml of water).
• Potassium iodate, 0.4 N.
• starch solution: 10g/l aqueous dispersion recently prepared from natural
soluble starch.
• Wijs solution: Add 9 g of trichloride into a brown glass bottle (1500 ml capacity).
Dissolve in 1 l of a mixture composed of 700 ml of acetic acid and 300 ml of
carbon tetrachloride.

• Determine the halogen content of the Wijs solution by taking 5 ml of the solution
and adding 5 ml of the potassium iodide and 30 ml of water. Then add 10 ml of
pure iodine and dissolve by shaking. Determine again the halogen content as
previously described.The titer should now be equal one and half times that of the
first determination. If this were not the case, add a small amount of iodine until
the content slightly exceeds the limit of one and half times. Let the solution stand,
then decant the clear liquid into a brown glass bottle.
• Place about 100 g of sample in a flask with 15 ml carbon tetrachloride and
25 ml of Wijs reagent. Insert a stopper and shake gently.
• Let sample sit in a dark location for 60 min for fats with expected iodine
numbers below 150, and for 120 min for fats with expected iodine values
above 150.
• Remove the flask from the dark and add 20 ml of the aqueous potassium iodide
solution and 150 ml of distilled water.
• Titrate the solution with 0.1 N sodium thio-sulfate until the yellow color has
mostly disappeared.
• Add 1 to 2 ml of starch indicator solution and continue the titration until the
blue color has just disappeared after vigorous shaking.

Determine the iodine value using the following equation:

12.69 x 0.1 x (ml titration of blank – ml titration of sample)
Iodine Value =
Weight of original sample, g

61


The iodine test can also be useful as an indicator of lipid oxidation by
comparing the initial iodine value and monitoring it across time. The oxidation
process destroys the double bonds or reduction of di-enoic acids (see later in this
chapter), and thus if the iodine value decreases with time it is an indication of lipid
oxidation in the sample under study.
CONTENTS

8.13.5. Acid value
The acid value is a measurement of the proportion of free fatty acids in a given
sample. It is determined by measuring the milligrams of potassium hydroxide
required to neutralize 1 g of fat. Oxidation is not involved directly in free fatty acid
formation, but in advanced states of oxidation, secondary products such a butyric
acid may contribute to FFA formation (Shermer et al, 1985).

The technique requires the following reagents: Solvent mixture (95%
ethanol/diethyl ether, 1/1, v/v), 0.1 M KOH in ethanol accurately standardized
with 0.1 M HCl (pure ethanol may be also used if aqueous samples are analyzed),
1% phenolphthalein in 95% ethanol.
• Weigh 0.1 to 10 g of oil (according to the expected acid value) in glass vial
and dissolve in at least 50 ml of the solvent mixture (if necessary by gentle
heating).
• Titrate, while shaking, with the KOH solution (in a 25 ml burette, graduated in
0.1 ml) to the end point of the indicator (5 drops of indicator), the pink color
persisting for at least 10 seconds.
• The acid value is calculated by the formula:
ml of KOH
Acid value = 56.1 x KOH x
Weight of original sample, g
CONTENTS

8.13.6. Lipid Oxidation
Lipids, especially oils, can undergo oxidation, leading to deterioration. In feeds,
these reactions can lead to rancidity, loss of nutritional value, destruction of
vitamins (A, D, and E) and essential fatty acids, and the possible formation of toxic
compounds and changes in color of the product.

The important lipids involved in oxidation are the unsaturated fatty acid
moieties, oleic, linoleic, and linolenic. The rate of oxidation of these fatty acids
increases with the degree of unsaturation. The overall mechanism of lipid

62


oxidation consists of three phases: (1) initiation, the formation of free radicals; (2)
propagation, the free-radical chain reactions; and (3) termination, the formation
of non-radical products. Chain branching consists in the degradation of hydro-
peroxides and the formation of new hydroxyl radicals which will then induce a
new oxidation. During the process, there are secondary products being formed
from the decomposition of lipid hydro-peroxides producing a number of
compounds that may have biological effects and cause flavor deterioration in feed.
These compounds include aldehydes, ketones, alcohols, hydrocarbons, esters,
furans and lactones (Figure 3).

Figure 3
Auto-oxidation of linolenic acid

X• XH

14 11
Initiators
O2
Linolenate OO• OO•

α-Tocopherol
A LH

O O
OOH L• OOH

B
LH 12(13) – OOH Monohydroperoxides 9(16) – OOH
(25%) (50%)
O2 LH
L•
L• O2
O O OOH
OOH OOH OOH OOH
+
C
Hydroperoxy Epidioxides
D E
(25%) Dihydroperoxides

Soybean products are relatively sensitive to oxidation because they are rich in
unsaturated FA especially linoleic acid. If soybeans are cracked or ground they
become more susceptible to oxidation, as fat becomes exposed to oxygen and
light. The finer the soybeans are ground, the greater the exposure and thus, the
greater the risk of oxidation. Evidently, soybean oil in its pure form (no additives) is
very susceptible to oxidation.

63


There are several techniques to determine the oxidation state of a soybean
product or soybean oil. These tests can be classified according to the type of
oxidation compound quantified:
• Determination of primary products of oxidation: peroxide value.
• Determination of secondary products of oxidation:
– Colorimetric methods: TBA and anisidine value.
– Volatile compounds determination: Chromatography.
• Stability tests: AOM and OSI.
CONTENTS

8.13.6.1. Peroxide value

The peroxide value is an indicator of the products of primary oxidation
(peroxides). They can be measured by techniques based on their ability to
liberate iodine from potassium iodide, or to oxidize ferrous to ferric ions.

The peroxide value is determined by the amount of iodine liberated from a
saturated potassium iodine solution at room temperature, by fat or oil dissolved
in a mixture of glacial acetic acid and chloroform (2:1). The liberated iodine is
titrated with standard sodium thiosulfate, and the peroxide value is expressed in
milli-equivalents of peroxide oxygen per kilogram of fat.
Procedure:
• Place 5 g of sample in a 250 ml Erlenmeyer flask and add 30 ml of the acetic
acid-dodecane solution.
• Swirl until the sample is dissolved and add 0.5 ml of a saturated potassium
iodide solution (150 g potassium iodide to 100 ml).
• Allow the solution to stand with occasional shaking for exactly one minute,
and then add 30 ml of distilled water.
• Titrate with 0.01N sodium thiosulfate adding it gradually and with constant
and vigorous shaking. Continue the titration until the yellow color has almost
disappeared, and add 1 ml of a starch indicator solution. Continue the
titration until the solution acquires a blue color.
The calculations are as follows:
Peroxide value, milliequivalents/1000 = Titration (ml used) x Acid normality x 1000

Although the peroxide value is applicable to peroxide formation at the early
stages of oxidation, it is, nevertheless, highly empirical. During the course of
oxidation, peroxide values reach a peak and then decline. Consequently the
accuracy of this test is sometimes questionable as the results vary with the
duration of the procedure used. Thus, a single peroxide value cannot be
indicative or the real oxidation state of a product. Also, this test is extremely
sensitive to temperature changes potentially leading to poor repeatability of
this test.
CONTENTS

64


8.13.6.2. Thiobarbituric acid (TBA)

TBA is the most widely used test for measuring the extent of lipid
peroxidation in foods due to its simplicity and because its results are highly
correlated with sensory evaluation scores. The thio-barbituric acid has a high
affinity to carbonyl substances (aldehydes and ketones) and its reaction with
aldehydes (especially with malon-aldehyde, secondary oxidation product of
fatty acids with three or more double bonds) forms a colorimetric complex
with maximum absorbance at 530 nm.

The basic principle of the method is the reaction of one molecule of
malon-aldehyde and two molecules of TBA to form a red malon-aldehyde-TBA
complex , which can be quantified with a spectrophotometer (530nm).
However, this method has been criticized as being nonspecific and insensitive
for the detection of low levels of malon-aldehyde. Other TBA-reactive
substances including sugars and other aldehydes could interfere with the
malon-aldehyde-TBA reaction.

The procedure was first described by Witte et al. (1970). The technique
requires a spectrophotometer, a water bath, pipettes, test tubes and the
following reagents:
• TBA solution: 0.02 M (1.44 g/500 ml of distilled water) 4,
6-dihydroxypyrimidine-2-thiol.
• m-phosphoric acid solution 1.6% (v/v).
• Standard solution: 1,1,3,3-tetraethoxipropyl (TEP) 10.2 M (0.2223 g/100 ml
of TCA solution).
• Construct calibration curve using several dilutions.
• Place 5 g of sample in a beaker and add 50-ml of a 20% tri-chloro-acetic acid
and 1.6% of m-phosphoric acid solution for about 30 minutes.
• Filter the slurry.
• Dilute the residue with 5 ml of freshly prepared 0.02 M (1.44 g in 500 ml of
distilled water) 4, 6-dihydroxypyrimidine-2-thiol and mixed.
• Tubes are then stored in the dark for 15 hours to develop the color.
• The color is measured by a spectrophotometer at a wavelength of 530 nm.
CONTENTS

8.13.6.3. Anisidine value

The principle of this technique is the preparation of a test solution in
2,2,4-trimethylpentane (iso-octane). Reaction with an acetic acid solution of
p-anisidine and measurement of the increase in absorbance at 350 nm.

65 13


The anisidine value is mainly a measure of 2-alkenals. In the presence of acetic
acid, p-anisidine reacts with aldehydes producing a yellowish color and an
absorbance increase if the aldehyde contains a double bond.
CONTENTS

8.13.6.4. Lipid Stability tests

Lipid stability tests are either predictive or indicative tests. They measure
the stability of lipids under conditions that favor oxidative rancidity. The
predictive tests use accelerated conditions to measure the stability of an oil or
fat. Indicator tests are intended to quantify the rancidity of an oil or fat. The
most important tests to determine lipid stability are the Active Oxygen
Method (AOM) and Oxygen Stability Index (OSI).
CONTENTS

8.13.6.4.1. AOM (active oxygen method)

This method predicts the stability of a lipid by bubbling air through a
solution of oil using specific conditions of flow rate, temperature and
concentration. It measures the time required (in hours) for a sample to attain
a predetermined peroxide value (in general 100 mEq/kg oil) under the specific
and controlled conditions of the test. The length of this period of time is
assumed to be an index of resistance to rancidity. Peroxide value is
determines as under 8.13.6.1.

The more stable the lipid (oil) the longer it will take to reach the
predetermined value (100 mEq/kg). For products other than oils such as full
fat soybeans, the oil must first be gently extracted. The method is very time-
consuming since stable oil or fat may take 48 hours or more before reaching
the required peroxide concentration. While still being used today, the AOM
method is being replaced by faster, automated techniques.
CONTENTS

8.13.6.4.2. OSI (oil stability index)

The OSI method is similar in principle to the AOM method, but it is faster
and more automated. Air is passed through a sample held at constant
temperature. After the air passes through the sample, it is bubbled through
a reservoir of deionized water. Volatile acids produced by the lipid oxidation
are dissolved in the water. These organic acids are the stable secondary
reaction products when oils are oxidized by bubbling steam. They are
responsible for an increase in conductivity of the water. This conductivity is
monitored continuously and the OSI value is defined as the hours required
for the rate of conductivity change to reach to pre-determined value.
A major advantage of this method is that multiple samples can be tested
simultaneously.
CONTENTS

66


8.13.7. Fatty acid profile
The fatty acid (FA) profile is, from a nutritional point of view, the most important
characteristic of oils. The FA composition of the oil is often a fingerprint for the
origin, treatment and storage of the oil and it determines largely the quantity that
can be used in specific animal diets. On average, palmitic, stearic, oleic, linoleic and
linolenic acid proportion of total fatty acids in soybeans is about 10, 4, 25, 51.5 and
7.5% respectively. However, there seems to be a recent trend for oil from soybeans
to be richer in palmitic, stearic and oleic acids, and poorer in linoleic and linolenic
acids. Part of this decrease has been attributed to global warming, as high temper-
atures induce a reduction in poli-unsaturated acids in soybeans. However, this
assumption will need further substantiation.

The fatty acid profile can be determined by gas or liquid chromatography. The
most common is the gas liquid chromatography procedure (GLC). For this analysis
a pure sample of oil is used after removal of moisture, insoluble impurities and
unsaponifiable substances. Sample preparation requires the following reagents:
• Metanolic-HCl (5% v/v): Add 10 ml of acetyl chloride into 100 ml of anhydrous
methanol.
• 6% K2CO3: 15 g of K2CO3 into 250 ml of distilled water.

Procedure to prepare samples for GLC (adapted from Sukhija and Palmquist, 1988):
• Weight 0.15 g of sample into 10 ml test tubes.
• Add 0.5 ml of an internal standard (i.e. 2mg of C19 per 1 ml of toluene).
• Add 0.5 ml of toluene.
• Add 1.5 ml of metanolic-HCl.
• Close tubes to avoid sample loses.
• Vortex for 1 min.
• Place in water bath at 70°C for 2 hours.
• Cool at room temperature.
• Add 2.5 ml of the K2CO3 solution.
• Add 1 ml of toluene.
• Vortex 30 for seconds.
• Centrifuge at 3000 rpm for 5 minutes.
• Keep the supernatant and add 0.5 g of anhydrous Na2SO4.
• Vortex for 30 seconds.
• Centrifuge at 4000 rpm for 10 minutes.
• Collect the supernatant and place in gas chromatography (GC) vial
for subsequent C analysis.

67


For operation of the GC equipment and analyses of fatty acids it is
recommended to follow the specific procedure provided by the manufacturer
of the chromatographic equipment. The chromatography methods are based on
the separation and quantitative measurement of specific fractions, such as volatile,
polar, or polymeric compounds or individual components such as pentane
or hexane.
CONTENTS

8.14. Minerals
Mineral composition of soy products can vary considerably among and within
products. The concentration of minerals depends greatly on a number of factors
most notably the origin and crop-growing conditions of the soybean, the variety and
the different types of extraction processes that are applied to obtain the soy product.
Although a measure of the concentration of these minerals is important for most
feed applications, under routine feed production conditions table values are used
to formulate. Generally, in feed production, formulators count on the contribution of
the minerals in the premix to cover mineral requirements of animals. This is especially
the case for the micro-elements. Regular analyses are generally only considered
necessary for the macro minerals calcium and phosphorus. For these elements, rather
than table values analytical values are used to formulate.

In certain regions, especially in areas of intensive animal production, the
regulatory limits on phosphorus use and excretion by animals make a precise
estimate of this element in the feed necessary. Phosphorus concentrations in soy
products are high and with the exception of soybean hulls and soybean mill feed,
P levels in these products are a multiple of Ca levels. This makes analyses for P, both
from a regulatory and nutritional point of view important. In addition to Ca and P,
salt (NaCl) analysis may be carried out on a routine basis for QC purposes.

Routinely, under more sophisticated laboratory conditions, most minerals are
analyzed by atomic absorption or flame emission. However, this requires a
considerable amount of investment and expertise. For normal QC objectives, classical
wet chemistry can be used to estimate the content of the most important minerals.

CONTENTS

8.14.1. Calcium
The determination of calcium by wet chemistry requires a set of porcelain
dishes, volumetric flasks (250 ml), beakers (250 ml), filter paper and funnels, and
a burette.

68


The following reagents are needed:
• Hydrochloric acid (1 to 3 v/v).
• Nitric acid (70%).
• Ammonium hydroxide (1 to 1 v/v).
• Methyl red indicator (Dissolve 1 g in 200 ml alcohol).
• Ammonium oxalate (4.2% solution).
• Sulphuric acid (98%).
• Standard potassium permanganate solution (0.05N).

Ca is determined as follows: weigh 2.5 g finely ground material into a
porcelain dish and ash (see section 8.2; alternatively use residue from ash
determination). Add 40 ml hydrochloric acid and a few drops of nitric acid to
the residue, boil, cool and transfer to a 250 ml volumetric flask. Dilute to volume
and mix.

Pipette a suitable aliquot of the solution (100 ml for cereal feeds; 25 ml for
mineral feeds) into a beaker, dilute to 100 ml and add 2 drops of methyl red.
Add ammonium hydroxide drop-wise until a brownish orange color is obtained,
then add two drops of hydrochloric acid to give a pink color. Dilute with 50 ml
water, boil and add - while stirring - 10 ml of hot 4.2% ammonium oxalate solution.
Adjust pH with acid to bring back pink color if necessary. Allow precipitate to
settle out, and filter, washing precipitate with ammonium hydroxide solution
(1 to 50 v/v). Place the filter paper with precipitate back in beaker and add a
mixture of 125 ml water and 5 ml sulphuric acid. Heat to 70°C and titrate against
the standard permanganate solution.

Calculation:
ml, permanganate solution Aliquot used (ml)
Calcium (%) = x x 0.1
wt. of sample, g 250
CONTENTS

8.14.2. Phosphorus
The method for phosphorus analysis requires a spectrophotometer able to read
at 400 nm, volumetric flasks (100 ml) and the following reagents:
• Molybdo-vanadate reagent. To obtain this dissolve 40 g ammonium molybdate
4H0 in 400 ml hot water and cool. Dissolve 2 g ammonium meta-vanadate in
250 ml hot water, cool and add 450 ml 70% perchloric acid. Gradually add the
molybdate to the vanadate solution with stirring and dilute to 2 liters.
• Phosphorous standards. Prepare stock solution by dissolving 8.788 g potassium
di-hydrogen ortho-phosphate in water and making up to 1 liter. Prepare the

69


working solution by diluting the stock 1 in 20 (working concentration =
0.1 mg P/ml).

To determine phosphorus: pipette an aliquot of the sample solution
prepared as for the calcium determination into a 100 ml flask and add 20 ml of
the molybdo-vanadate reagent. Make up to volume, mix and allow to stand for
10 minutes. Transfer aliquots of the working standard containing 0.5, 0.8, 1.0 and
1.5 mg phosphorus to 100 ml flasks and treat as above. Read sample at 400 nm
setting the 0.5 mg standard at 100% transmission. Determine mg phosphorus
in each sample aliquot from a standard curve.
CONTENTS

8.14.3. Sodium chloride
The reagents used for the determination of salt in feed samples or feed
ingredients are:
• Standard 0.1N silver nitrate solution.
• Standard 0.1N ammonium thio-cyanate solution.
• Ferric indicator - saturated aqueous solution of ferric aluminum.
• Potassium permanganate solution - 6% w/v.
• Urea solution - 5% w/v.
• Acetone (A.R. grade).

The method consists of: weighing a 2 g sample into a 250 ml conical flask.
Moisten the sample with 20 ml water and then pipette, 15 ml 0.1N silver nitrate
solution - mix well. Add 20 ml concentrated nitric acid and 10 ml potassium
permanganate solution and mix. Heat mixture continuously until liquid clears
and nitrous fumes are evolved. Cool. Add 10 ml urea solution and allow to
stand for 10 minutes. Add 10 ml acetone and 5 ml ferric indicator and back titrate
the excess silver nitrate with the 0.1N thio-cyanate solution to the red brown
end point.

Calculation:
15 – ml 0.1 N NH4CNS x 0.585
NaCl(%) =
wt. of sample, g

For rapid, routine QC procedures, Quantabs, a bench-top test kit is used.
CONTENTS

70


8.15. Isoflavones
In many diets, human as well as animals, soybean products are the main dietary
source of isoflavones. These secondary metabolic compounds may play an important
role in preventing cancers and reducing risk of cardiovascular diseases. There is also
an increasing interest in the role and use of isoflavones in animal production as these
compounds have been implicated in enhancing immunity and improving growth
performance and carcass traits (Cook, 1998; Payne et al., 2002; Kerley and Allee, 2003).

Two forms of isoflavones can be determined: the bound glucoside form (genistin,
daidzin, glycitin) and the free aglycone form (genistein, daidzein, glycitein). Lee et al.
(2003) reported that the total isoflavone contents in soybean cultivars grown in
Korea ranged from 110 to 330 mg 100 g–1. The USDA and Iowa State University (2002)
have developed a database on isoflavones from scientific articles. The analysis of
isoflavones was carried out according to the method of Wang and Murphy (1994)
using high-performance liquid chromatography (HPLC).
For the analysis of isoflavones the following reagents are needed:
• Acetonitrile.
• HCl (0.1 N) or phosphoric acid.
• Isoflavone standards (commercial source).

Besides normal laboratory equipment the essay requires an HPLC instrument
with a YMC-pack ODS-AM-323 column (10 µm, 25 cm x 10 mm i.d.).

The procedure consists of an Isoflavone extraction and an HPLC quantification
step. For the extraction two grams of ground soybean products are mixed with 2 ml
of HCl and 10 ml of acetonitrile in a 125 ml flask, stirred for 2 hours and filtered. The
filtrate is dried under vacuum at a temperature below - 30°C and then re-dissolved in
10 ml of 80 % HPLC grade methanol in distilled water. The sample is then filtered
through a 0.45 mm filter unit and then transferred to 1 ml vials.

The HPLC quantification of isoflavones requires a column temperature of 25°C
and a mobile phase employing a linear HPLC gradient using 0.1 % glacial acetic acid
in distilled water (solvent A) and 0.1 % glacial acetic acid in acetonitrile (solvent B).
Following the injection of 20 µL of the sample, solvent B is increased from 15 to 35 %
over 50 min and then held at 35 % for 10 min. The recommended flow rate is 1 ml
min–1 and the detection wavelength: 200 - 350 nm.
The content of each isoflavone is expressed on a w.w–1 basis.
CONTENTS

71


8.16. Antinutritional factors (ANF)
One of the most important restrictions on the use of soybeans and their products
in animal diets is the presence of a relatively large number of antinutritional factors.
The presence of these factors is also the main reason why different technological
treatments are applied to soybeans or their products. The ANF in soybeans include
compounds classified as protease inhibitors, phyto-hemaglutins (lectins), urease,
lipoxygenases and antivitamin factors which can relatively easily be destroyed by
heat application or fermentation (Liener, 2000). The methods referred to under
section 8.4 provide a relative estimate of the effectiveness with which they have
been destroyed. The effect of heat treatment on ANF is a direct function of the
degree and duration of the heat application along with particle size and moisture
level. ANFs that are not destroyed by heat are the poorly digested carbohydrates,
Saponins, Estrogens, Cyanogens and Phytate (Liener, 2000). In the case of soybean
products, the most important and best known ANF is the trypsin inhibitors.
The quality of technological treatment to destroy ANF is mainly related to this
specific factor.

To analyze for any ANF a large number of different methods and procedures
are available ranging from instrumental (HPLC, GC, CE) to thin-layer chromatography
(TLC) and immuno-assays. The reliability and accuracy of results obtained with
these methods varies and no preferred method has been defined for all ANF. When
possible, and for practical routine QC purposes, the use of ELISA (enzyme-linked
immuno-sorbent assay) tests are recommended.

The ELISA tests rest on the principle that the compound called the antigen (in
this case an ANF obtained by extraction from the feed or ingredient) will bind with
enzyme-linked antibodies. Upon this reaction, the enzyme-linked antibodies will be
released from the surface to which they were attached (this maybe a stick, plate or
tube). The enzyme-linked antibodies are then washed away and an enzyme substrate
is added to allow a reaction with the remaining enzyme-linked antibodies.
This procedure results in a color change which is inversely related to the antigen
concentration. Thus, the deeper the color, the smaller the antigen (ANF)
concentration since less antibody-antigen complexes have been formed and
washed away leaving more enzyme-linked antibodies to react with the color
causing enzyme substrate.
CONTENTS

8.16.1. Trypsin inhibitors
The residual trypsin inhibitor in soy products combines with the trypsin in
the small intestine and forms an inactive complex thus reducing digestibility of

72


protein. In addition to the negative effect on protein digestibility, the trypsin
inhibitor induces pancreatic hypertrophy and leads therefore to an increase in
secretion of trypsin (endogenous nitrogen). The combined effect on the animal is a
reduction in nitrogen retention, growth and feed conversion.

The procedure described to determine trypsin inhibitors activity is based on
the ability of the inhibitors to form a complex with the enzyme and thus to reduce
the enzyme activity. Uninhibited trypsin catalyzes the hydrolysis of a synthetic
substrate BAPNA, forming a yellow-colored product and thus producing a change
in absorbance. The reference procedures proposed by the American Oil Chemists'
Society (AOCS) and the French Association for Normalization (AFNOR) are based
upon the work of Kakade et al. (1969, 1974). Here, the AOCS (1997) procedure is
summarized but the only difference with the AFNOR (1997) procedure is the
composition of the extraction buffer, which is alkaline whereas it is acid in the
other case. Still, these procedures are not very well adapted for routine QC use,
and a well equipped lab with skilled technician is necessary.

For practical reasons, the method described measures total trypsin inhibitors.
It reflects thus the concentration and effects of two distinctively different types of
inhibitors namely the KTI (Kunitz trypsin inhibitor) and the BBI (Bowman-Birk
inhibitor).

• Hexane or petroleum ether.
• Sodium hydroxyde solution (0.01 N).
• Tris buffer: dissolve 6.05 g tris (hydroxyl-metyl)-amino-methan and 2.94 g
calcium chloride in 900 ml of water, adjust to pH 8.2 and dilute to 1 L.
Bring to 37°C before using.
• Trypsin solution: dissolve 4 mg, accurately weighed, twice-crystallized, salt-free
trypsin in 200 ml hydrochloric acid (0.001 N).
• BAPNA solution: In a water bath, dissolve 40 mg N α-benzoyl DL-arginine
p-nitroanilide (BAPNA) in 1 ml dimetyl sulfoxide. Dilute to 100 ml with tris buffer
(at 37°C). Prepare new solution daily. Maintain at 37°C for use.
• Acetic acid solution (30 %): mix 30 ml glacial acetic acid and 70 ml water
(caution).

Equipment required: a grinding mill, with screen size 0.15 mm or smaller and
a Spectrophotometer capable to read at 410 nm.

• Samples should be finely ground without excessive heating. Samples with
more than 5 % fat should be defatted with hexane or petroleum ether and
desolventized before grinding.

73


• One gram of ground sample is subsequently weighed into a beaker containing
a magnetic stirring bar. 50 ml sodium hydroxide solution is added and the
suspension is agitated slowly. After 3 hr, the pH is measured; pH should range
between 8.4 and 10.0.
• An aliquot of suspension should be taken with a serological pipette and diluted
with distilled water so that soybean trypsin inhibitor concentration is sufficient
for 40 - 60 % trypsin inhibition. When it is not possible to estimate the expected
trypsin inhibitor units, more than one dilution should be made.
• With serological pipettes, 0, 0.6, 1.0, 1.4 and 1.8 ml of the diluted suspension is
added to duplicate sets of test tubes. Water is then added to bring the volume
to 2 ml in each tube.
• With a regular time interval for the different tubes, 2 ml trypsin solution is
added to each tube and quickly mixed on the Vortex stirrer and placed in the
37°C water bath. 5 ml BAPNA is added to each tube, mixed on Vortex stirrer.
The samples are incubated for 10 min at 37°C. After exactly 10 min, the reaction
is stopped by addition of 1 ml acetic acid solution followed by mixing on the
Vortex stirrer.
• Prepare a blank sample as above, except that trypsin is added after acetic acid.
• The contents of each tube are filtered and absorbance is measured at 410 nm.

Calculation of trypsin inhibitors activity. One trypsin unit is arbitrarily defined
as the amount of enzyme, which will increase absorbance at 410 nm by 0.01 unit
after 10 minutes of reaction for each 10 ml of reaction volume. Trypsin inhibitor
activity is defined as the number of trypsin units inhibited (TIU).
Absorbance blank – absorbance sample
TIU (/ml) =
0.01 x volume of diluted sample solution, ml

TIU is plotted against the volume of the diluted sample solution.
The extrapolated value of the inhibitor volume to 0 ml gives the final TIU /ml.
This value is used to calculate the TIU per g sample:

TIU(/g) = TIU (/ml) x d x 50

where d = dilution factor (final volume divided by the amount of aliquot taken).
The results of this analytical method should not exceed 10 % of the average value
for repeated samples.
CONTENTS

8.16.2. Soy antigens
Immunoassay techniques are used to determine concentrations of soy
antigens (glycinin and ß-conglycinin) in soy products. The ELISA tests require little
training and can be used in small laboratories. Various types of ELISA tests with

74


specific polyclonal antisera (Pabs) or monoclonal antibodies (Mabs) can be used to
assess soy antigens contents (Table 15).

To apply the different ELISA tests, the protein fraction of the soy product is
first extracted in borate buffer (100 mM NaBO3, 0.15 M NaCl, pH 8) for 1.5 hr (Tukur
et al., 1993). The level of glycinin and ß-conglycinin can be measured by a specific
competitive inhibition ELISA using anti-soy globulin Pabs (Heppell et al., 1987).
Serial, four-fold dilutions of the sample are incubated with a standard dilution
of rabbit antiserum to test protein and the residual unbound antibodies are
quantified.

Table 15
ELISA formats used for analysis of soy globulins
ELISA
Antibody format Specificity
Glycinin Pab LJR J4 inhibition intact glycinin
Mab IFRN inhibition binds proteolytic intermediates and
0025 thermally denatured glycinin;
epitope lies within acidic polypeptides
Mab IFRN two-site recognize proteolytic intermediates and
0025 & Pab thermally denatured glycinin
R103b3
ß-conglycinin Pab LJR J2 inhibition intact ß-conglycinin
Mab IFRN inhibition recognizes epitopes in acidic regions of a
0089 and α' subunits of ß-conglycinin
Mab IFRN two-site recognition of thermally denatured
0089 & Pab ß-conglycinin is 3-fold greater than native
R195b3
(from Tukur et al., 1996)

CONTENTS

8.16.3. Lectins
Lectin is a protein with a specific binding affinity for sugar residues. The
lectin-sugar interaction is important at the level of the membrane receptors in the
gut where it is thought to be responsible for agglutination and mitosis. As for most
leguminous plants or seeds of these plants, lectins have been shown to be an
important ANF in raw soy products (Pusztai, 1991).

75


Table 16
Anti nutritional factor contents in various soy products
Product PDI Trypsin inhibitor Lectins Antigens
(%) activity (mg/g) (mg/g) (mg/g)

Untoasted soy flour 90 23.9 7.3 610
Slightly toasted soy flour 70 19.8 4.5 570
Toasted soy flour 20 3.1 0.05 125
Ethanol/water-extracted
soy concentrate 6 2.5 <0.0001 <0.02
(adapted from Huisman and Tolman, 1992)

Lectins are heat sensitive and are therefore only present at residual levels
in soybean products. Heat treatment to inactivate antinutritional factors in soy
products is less efficient for antigens than for trypsin inhibitors or lectins
(Table 16).

The level of soy lectins can be estimated by measuring the hemaglutination
activity. More recently, ELISA (total lectins) and FLIA (functional lectins) tests
have been developed and these methods are more sensitive and selective
(Delort-Laval, 1991). Lectins can vary considerably (chemical structure, molecular
weight a.o.), therefore a specific essay is required for each legume seed tested
(de Lange et al., 2000).

The procedure as presented by Schulze et al. (1995) can be summarized
as follows:
One gram of sample is mixed with 20 ml tris-HCl buffer (50 mM, pH 8.2) and
stirred for 1 hr. Extracts are centrifuged at 7500 x g for 15 min and the
supernatant is used for serial dilutions. Lectins are is determined in the
supernatant.

Polyclonal antibodies against soy-lectins (ELISA) are coated to micro-titer
plates overnight at 4°C. The plates were then blocked with 0.5 % BSA (bovine
serum albumin) and 0.2 % Tween-20 in TBS for 1 hr at 37°C. Subsequently, the
plates are washed and samples are diluted at appropriate concentrations.
A reference soy-lectin sample is run in parallel. All samples are transferred to
micro-titer wells and incubated for 2 hr at 37°C. The plates are washed and
peroxidase-conjugated anti-lectin antibodies are applied and incubated for 2 hr
at 37°C. Finally, the plates are washed again and bound conjugated antibodies
are developed for peroxidase activity using 1,2-phenylendiamine. Absorbance is
read at 492 nm. Data can be evaluated by the parallel line assay using a computer
software package connected to the ELISA reader system. Lectin concentrations are
expressed in w.w–1 on a dry matter basis.
CONTENTS

76


8.17. Mycotoxins; rapid tests
Mycotoxins are a major quality concern for the feed industry. Although soy
products do not generally show the same level or range of mycotoxin contamination
as cereal grains, they do occur occasionally and routine QC methods should be in
place to control their presence. This is especially the case now that regulatory
restrictions on mycotoxin levels are becoming increasingly more stringent. The most
common mycotoxins occurring in feed ingredients are aflatoxins, deoxynivalenol
(DON), zearalenone, ochratoxin and fumonisins. All these mycotoxins can potentially
be found in soy products but the most important mycotoxins in the case of soy
products are ochratoxin (produced by the molds Aspergillus ochraceous or Penicillium
verrucosum under poor storage conditions) and zearalenone (produced by the
fungus Fusarium graminearum).

As in the case of ANFs, the analyses for mycotoxins and their metabolites can
be carried out by a range of methods. No preferred method has been defined for all
mycotoxins. For practical QC purposes, however, the use of the TLC and ELISA tests
are recommended. In the case of mycotoxins, these tests can be separated in
screening and quantitative analysis with the former detecting a simple presence of
the mycotoxin and the later providing rather precise estimates of mycotoxins levels
present in a sample. Qualitative analysis will require additional equipment such as
long-wave microwell strip readers, UV lights or fluorometers.

The precision of these quantitative measures varies with the type and
manufacturer of the test and some prior evaluation and training as to which test
most suitable for a particular laboratory setting is recommended. Minimum
detection levels may vary among producers and types of test kits. However, the
significant improvements in the quantitative ELISA tests over the last 10 to 20 years
have made these tests perfectly suited for routine quality procedures and several
have been validated by the AOAC and received approval (AOAC International, 1995;
Trucksess et al., 1989). Nevertheless, due to the many factors that may affect the
results of the ELISA test kits, the variation between laboratories and analysts may be
considerable. In some instances, limits of detection are also inadequate to meet the
increasingly stringent demands for measurement at low levels. False positive or
negative readings are known to occur and for purposes other than routine quality
procedures, classical instrumental analysis as referred to above will be needed.
Also, test kits have been developed that will qualitatively detect several mycotoxins
in a single test.

General procedure:
Before performing the rapid test, the mycotoxins need to be extracted from the
sample. Most of mycotoxins can be extracted by grinding the sample to 0.6 mm-

77


mesh, then blending 25 g of that sample with 125 ml of a 70% methanol solution
(7 parts of methanol and 3 parts of de-ionized water). Stir vigorously in a high-speed
blender for 2-3 minutes. The ELISA test should be performed as indicated by the
manufacturer of the test kits.

When choosing the ELISA test for mycotoxin analyses it is necessary to make sure
that the kit has been validated for use with soybean products.
CONTENTS

8.17.1. Ochratoxin
This mycotoxin is often considered the most common mycotoxin in soybean
products. It is thought to be principally produced during storage under humid
and warm (>20°C) conditions. Damage to grains by insects or through mechanical
means will provide an entry for the fungi and enhance initial contamination.
Ochratoxin is a mycotoxin produced by several species of the mold genera
Aspergillus and Penicillium. Usually, the rapid tests for ochratoxins have a lower
limit of detection of 0.01 ppm in the case of screening methods while quantitative
tests have a lower detection limit at 0.005 ppm. It seems that at levels of 0.2 ppm
clinical signs associated with ochratoxins will appear in monogastric species.
CONTENTS

8.17.2. Zearalenone
Zearalenone is primarily produced by Fusarium graminearum. By itself,
zearalenone is not toxic, but once metabolized, its end-products have estrogenic
activity, which may cause some reproductive alterations in animals. Sensitivity to
zearalenone differs considerably among livestock species with swine considered
most sensitive. Levels above 1 ppm result in noticeable effects on reproduction
in swine. Usually, the rapid screening tests for zearalenone have a lower limit of
detection of 0.1 ppm with quantitative tests having a lower detection limit of
0.2 ppm.
CONTENTS

8.17.3. Fumonisins
Fumonisins includes a group of mycotoxins produced by Fusarium
moniliforme and Fusarium proliferatum. Horses are especially sensitive to
fumonisins. Usually, the rapid tests for fumonisins have a lower limit of detection
of .2 ppm, a limit of quantification of 1 ppm up to 6 ppm.
CONTENTS

78


8.17.4. Aflatoxins
Aflatoxin is often considered the most common mycotoxin in feeds and
grains. However, the occurrence of this toxin in soy products is relatively rare.
Aflatoxin is a mycotoxin produced by Aspergillus flavus and Aspergillus parasiticus.
Not all strains of these fungi are capable of aflatoxin production. Drought
conditions associated with warm temperatures and physical damage to the grain
strongly increase the probability of aflatoxin occurrence. There are several types of
aflatoxins, with the most common in order of prevalence being B1, B2, G1 and G2.
To date, aflatoxins are the only mycotoxins for which official maximum levels have
been defined. The FDA as well as the EEC has established a maximum level of total
aflatoxins of 20 ppb in ingredients for the feed industry.

Usually, the quantitative rapid tests for aflatoxins have a lower limit of
detection of 1 ppb and a limit of quantification of 5 ppb up to 50 ppb.

In addition to the method described above, aflatoxins can be extracted,
by weighing 10 ml of soybean products into a wide mouthed bottle and
thoroughly mixing it in 10 ml of water. Add 100 ml of chloroform, stopper with
a chloroform resistant bung and shake for 30 minutes. Filter the extract through
diatomaceous earth.
CONTENTS

8.17.5. Deoxynivalenol
Deoxynivalenol (DON), commonly referred to as vomitoxin, is a trichothecene
primarily produced by Fusarium graminearum. Fusarium growth requires a
minimum moisture level of 19 % thus DON levels are not known to develop or
increase during normal storage conditions. The FDA has established advisory
levels for DON. Maximum levels for ingredients other than wheat and wheat
by-products have been set at 5 ppm for swine and 10 ppm for ruminants (with
a 20 % limit at the inclusion rates of these contaminated ingredients in the case
of swine diets).

The extraction of DON from soybean should not be performed with ethanol.
It should be conducted with about 10 g of sample ground to 0.6 mm. Shake
vigorously in 50 ml of de-ionized water for 3 minutes. Then the sample is filtered
and the liquid fraction is kept for subsequent ELISA analyses.

Usually, the rapid tests for DON have a lower limit of detection of 2.0 ppm for
the screening tests and 0.5 ppm for the quantitative tests.
CONTENTS

79


8.18. Genetically modified organisms (GMO)
Some soybeans have been genetically modified. As market demands for
traceability are growing and market demands for non-GMO products are decreasing,
it is important to be able to distinguish between genetically modified and traditional
products. Certain official maximum limits on the presence of GMO material in non-
GMO products exist. In the EEC these levels are now fixed at a maximum of 0.9 %.
Japanese legislation allows food products containing less than 5% of approved
biotech crops, like corn and soybeans, to be labeled as non-GMO. In the presence of
the extensive use of GMO soybean varieties, the risk of commingling and analytical
variability, these minimum levels reflect in part the inability to guarantee complete
absence of GMO material in products labeled as GMO-free.

The GMO varieties are characterized by the insertion of a new, functional gene
(or cluster of genes) into their genomes. The expression of these genes provides the
soybeans with some advantages, such as resistance to insects and herbicides. Several
commonly used GMO testing protocols, including biological tests, as well as ELISA
and PCR (polymerase chain reaction) tests, exist. The ELISA methods are based on the
same principle as described above for the detection of mycotoxins. A popular version
of the ELISA test, used for screening purposes only, uses lateral flow strips that deliver
results in a couple of minutes. This makes this test especially suited for QC purposes
at feed mills. Quantitative ELISA tests also exist. They are normally presented as plate
tests with the degree of color change being indicative of the level of GMO material
present in the sample. An important limitation of the ELISA tests is that they have
limited accuracy when applied to heat-processed ingredients; especially in the case
of high temperature application (extrusion). The limitation applies to all products in
which the application of high temperatures leads to substantial denaturation of the
soy proteins, thereby making detection of proteins difficult.

The PCR tests (more sensitive than the ELISA methods) are based on the
detection of DNA sequences in the genome of the soybean product. The PCR is an
extremely sensitive technique and is able to identify different types of GMOs at very
low levels. It is also the only method that can effectively detect GMOs in heat treated
ingredients and feeds which makes this method the preferred procedure in the case
of most soy products. However, due to the requirements for equipment, the delay
in obtaining results (2 to 3 days) and the level of expertise required, the test is not
suited for routine QC analyses at the feed plant level. This test should be carried out
in a proper laboratory setting. An additional disadvantage of this procedure is its
tendency to give false positives which may require replicate testing.

80


Biological tests are mainly limited to the herbicide resistant soybean varieties and
can only be applied to the untreated bean. The advantage of these tests is that they
are relatively inexpensive and produce clear-cut results. In these tests seeds are
placed in a germination media. The seeds are then moistened with a diluted solution
containing the herbicide against which the seed is thought to be resistant or the
germinated seeds are sprayed with the herbicide in question. Herbicide tolerant
GMO seeds will germinate and/or grow normally while the non-GMO seeds will fail
to germinate or grow. A minimum one week period is needed to carry out this test.
CONTENTS

81

9. NIR ANALYSES
Near infra-red reflectance (NIR) spectroscopy has been used for more than
35 years to rapidly analyze grains, animal feeds and forages. The first application of
NIR spectroscopy was developed by Norris and associates in the early sixties to
measure water content in grains and seeds (Givens et al., 1997). Since these early
developments NIR spectroscopy has matured to a well established and broadly
accepted method to measure a wide array of chemical compounds in feed and food
ingredients or diets.

Concerning soybean products, the largest and most evident application is in the
rapid determination of proximate components previously carried out by the time-
consuming and laborious conventional wet chemistry. The potential of NIR to carry
out more evolved analysis such as protein quality and ANF is a real possibility since
the technique has been used to measure characteristics of similar complexity such
as digestibility of individual amino acids (van Kempen and Bodin, 1998). However,
despite the rapid answers and the major time savings made possible by NIR, the
development of the calibrations required for protein solubility and ANF have as yet
received little attention.

NIR spectroscopy is based on the principle that infra-red radiation of a sample
results in the reflection or transmittance of the radiation that is not absorbed by the
sample. The characteristics of the reflected or transmitted radiation can be used to
describe certain chemical characteristics of the sample. Since this relationship is not
mathematical, the relationship between the reflected radiation and the chemical
compound of interest must be based on a calibration. In this calibration the amount
of light reflected (or absorbed) at one or more wavelengths are related to a specific
chemical compound or compounds. More precisely, it is the chemical bonds and
functional groups of the compound that are related to the reflectance at a specific
wavelength. Consequently, molecules characterized by a repetitive bond and
structure are often more suited for detection by NIR. The choice of a wavelength or
a combination of wavelengths to detect a chemical compound is not necessarily
constant. The optimum choice of wavelengths to correlate with a specific compound
differs not only among ingredients but also among laboratories, equipment and even
years. Also, the scattered reflectance from other compounds leads to interference.

82

9. NIR Analyses.

Consequently, the wavelength best related to the compound of interest is the one at
which absorption by the compound is maximized and interference by reflectance of
other sample constituents is minimized.

A number of items interfere with the near infra-red reflectance spectra. The
reflectance obtained from a sample is characterized by scatter due to instrument
type and function, sample preparation (grinding and thus particle size), temperature,
water content and interference of reflectance from other compounds. Variations in
water content of the sample are important because water absorbs radiation strongly.
In order to increase the precision of NIR analyses the factors interfering with the NIR
spectra need to be standardized when analyzing an ingredient or they need to be
eliminated through the application of mathematical corrections on the spectrum or
calibrations. Since standardization of sample preparations is not always practical and
since it reduces the major benefit of NIR analyses (time savings) preference is given
to mathematical corrections. A series of mathematical tools have been developed to
correct the spectral data and improve the predictive capacity of the calibrations.
The choice and application of these corrections differ considerably among the
constituents to be analyzed. The range of mathematical tools that is available to treat
spectral data is increasing rapidly thus improving the quality of the analysis and the
requirements for sample preparation.

Before routine analyses can be carried out equations need to be developed
for each individual constituent and often the individual ingredients. Sometimes, a
common equation can be developed for ingredients and/or their by-products.
In the case of soy products a single equation can by used for a number of products
if they are sufficiently alike in composition and preparation. This is for instance the
case for all soybean meals. However as a general rule of thumb it may be said that
the larger the physical and chemical differences among ingredients, the greater the
need to develop separate equations.

NIR calibrations are equations developed from a dataset composed of the
component of interest analyzed by a standard reference method (i.e. crude protein)
and the infra-red reflectance spectra. Least square multiple linear regression analysis
are used to develop the prediction equation (calibration) i.e. chose the equation
that provides the best fit between the analytical component and reflectance or
absorption at one or more wavelengths. The calibration data set should include
samples that represent the total chemical, physical and spectral variation normally

83

9. NIR Analyses.

found in the population of samples that will be analyzed with the calibration. For
instance in the case of a calibration to measure crude protein in all soybean meals
the calibration dataset should include samples of SBM ranging from 42 to 50 % crude
protein. Calibration sets should have the widest possible range in composition but
above all they should be representative of all samples to be routinely analyzed with
the equation. It is generally not recommended to include samples with extreme
values (Shenk and Westerhaus, 1991). Extrapolation beyond the range of values
covered in the calibrations is not acceptable. Thus for most soybean products
separate equations will need to be developed for groups of products with similar
characteristics and values (i.e. Full fat soybeans, SBMs, SPCs, oils etc.).

The quality of a calibration depends greatly on the number of samples and
the choice of the samples. The number of samples required to develop a reliable
equation remains a subject of discussion. No definite numbers can be provided as
the size of the calibration dataset is related to the variability within a set and the
range of values that needs to be covered. Under most conditions applicable to
soybean products, the number of samples will be no less than 40. The larger the set
of well prepared and selected samples the stronger the calibration will be. Once the
calibration established, validation of the calibration will be necessary. Samples for
validation are subject to the same criteria for representation and number as those
used for samples to establish the equation. Generally a smaller number are allowed
when samples are representative of the population. Routine procedures to verify
the validity and quality of the equation need to be established. The calibration can
and should be strengthened through a continuous updating and expansion of the
calibration set by adding critically selected samples.

A number of statistical measures are used to describe the quality of a calibration
or evaluate its predictive capacity. Most of these refer directly to the least square
multiple linear regression techniques used to develop the equations. Most common
measures are the regression coefficient (R2), the standard error of prediction or
estimate (SEP) and bias (D). The R2 is a measure of the variability in the reference
data accounted for by the regression equation; the SEP is the variability between
predicted values and reference values when the equation is applied to the data
other than the calibration set, and D is the average difference between the predicted
and reference values. Ideally R2 should be as close as possible to 1.0 while SEP and
D should be as small as possible.

84

9. NIR Analyses.

Analyses obtained by NIR are potentially subject to a large number of errors
related to the equipment, the calibration and validation process or sample
preparation (Williams, 1987). Not all errors are of equal importance and their
occurrence and impact is being reduced by the development and installation of
more sophisticated NIR techniques and equipment. Users have learned to manage
the equipment better and increased their understanding of the special requirements
needed for NIR analysis. While the routine use of the equipment is quite simple, the
maintenance and development of calibrations require a high level of expertise.
For proper operation and in order to reduce errors clear protocols should be drawn
up and implemented at all levels of NIR operations. It is important that these
protocols assure continuity between the use of NIR for routine analytical functions
and the development of new calibrations. When used for routine quality assurance
analyses, it is important to provide a separate dust-free environment. This is often
difficult to realize in operations dealing with commodities and feed production.

An important number of the errors that can occur in NIR are related to the
equipment. There is a relatively large variation between NIR equipments.
Consequently, in the case of monochromatic equipment for instance calibrations
cannot be transferred directly from one NIR to another without adjustments or
corrections followed by a series of validations. Universal calibrations have been
developed to solve the problem of transferability of calibrations. These equations
are based on a larger dataset than normal covering often different regions and years.
Results of these calibrations are often less accurate that those of equipment-specific
calibrations. More recently the concept of cloning or networking NIRs has been
developed. In these networks and through a series of mathematical corrections the
NIRs are calibrated to provide identical spectral results. This of course facilitates
enormously the transfer of calibrations and the verification of the different NIRs in
the network.

While in principle all organic compounds of a feed or feed ingredient can be
analyzed by NIR, for most ingredients and especially for soybean products, best
results in terms of accuracy and precision are obtained for humidity, crude protein
and lipids. NIR results for fiber components and non-fiber carbohydrates (starch,
sugars) normally give larger SEPs and biases and lower R2 values. NIR cannot be used
for the analyses of minerals although a rough estimate for ash and minerals may be
obtained by relating the reflectance at specific wavelengths to the organic matter or
components of the organic matter (Givens et al., 1997). NIR can be used to analyze

85

9. NIR Analyses.

other organic compounds such as amino acids (van Kempen and Bodin, 1998) ANF
or fatty acids in soy products, however, the number of publications on this subject is
limited and more work is needed.

Equipment required for NIR analysis of soy products:
• Drying equipment (force draught oven).
• Wet chemistry laboratory (to conduct analyses for reference values used in
calibration development – see previous sections).
• Grinder (preferably Retch grinder but this is optional; calibrations can be
developed for un-ground, homogeneous material).
• NIR equipment.
Procedure (calibration development).
• Dry sample to constant weight (see Section 8.1).
• Grind (optional).
• Split sample in 2 sub-samples, one for reading on NIR equipment and one for
analysis by the reference method(s) (wet chemistry).
• Fill sample holder (as described in manual).
• Insert sample holder in NIR and read reflectance or analyte concentration.
• Obtain analytical results for analyte of interest by reference method
(see Chapter 8).
• Using a statistical software perform multiple linear regression analysis between
wavelength spectra (independent variable) and results of chemical analysis
(dependent variable).
• Establish regression equation (high R2, low SE); beware of over-parameterization
(use of too many wave lengths).
• Validate equation with samples not used to establish equation.
• Re-evaluate calibration regularly.

Procedure (application):
• Dry sample to constant weight (see Section 8.1).
• Grind (optional).
• Fill sample holder (as described in manual).
• Insert sample holder in NIR and read reflectance or analyte concentration.
(Modern apparatus have integrated computers that will give a direct reading of
the component concentration).

CONTENTS

86

10. DATA MANAGEMENT
10.1 Sample statistics
The physical, chemical and microbiological analyses that are performed on
feed or soybean products provide information on the nutritional or health value of
a selected lot (statistically speaking: the population). The analysis of the whole
population is generally not possible. Therefore, statistical procedures are required to
obtain information from samples to describe the population accurately.

a. Basic assumptions:
The distribution of a measured parameter (X) in the population of size N is
assumed to be normal. In statistical terms, this is expressed as:
Xi ~ N (µ,σ2). Where µ is the population mean and σ2 the population variance).

Figure 4
Example of a density curve describing a normal distribution

0.4 – Parameters:
mean = 48%
std. deviation = 1%
0.3 –
Probability

0.2 –

0.1 –

0–
I I I I I I I I I I I
43 44 45 46 47 48 49 50 51 52 53
Protein content, %

| µ−3σ|µ−2σ| µ±1σ |µ+2σ | µ+3σ|

87

10. Data Management

Note: The area under the curve gives the proportion of observations that falls in a
particular range of values.
Properties of the normal distribution:
68 % of the observations fall within ± 1σ of the mean µ.
95 % of the observations fall within ± 2σ of the mean µ.
99.7 % of the observations fall within ± 3σ of the mean µ.

The population can be characterized by its mean µ and variance σ2 (unknown).
The normal distribution is the most common random distribution about the mean
value. An example of this could be the distribution of crude protein content (CP) in a
load of soybean meal (SBM) guaranteed to contain 48 % of CP (Figure 4).

b. Parameter estimates:
Sample statistics are used to estimate the population parameters from a sample
of smaller size (n). In our SBM example, this would be the estimation of the crude
protein of all SBM in the load on the basis of a set of samples of SBM from that load.
Main parameter estimates (Table 17) can be calculated simply from the measured
results on the samples.
Table 17
Common notation of parameters and parameter estimates

Parameters Parameter estimates
(population) (sample)
Mean µ X
Variance σ2 S2
Standard deviation σ S

Mean
The mean x represents the average value of the analyzed component and is
calculated by taking the sum of the measurements and dividing by the number
of samples.
Σx
Mean (x): x = n i
Where xi: individual sample measurement, n: number of samples

Variability
More important than the mean of a parameter maybe the variability in the
observations on the samples as it provides information about the spread in values
within the population. For our example: how many samples have crude protein
values above or below the mean and how much do they differ from the mean value?
Different parameters can be used as indicators of the variability present in a set of
measurements:

88

10. Data Management

Range (w): w = xmax - xmin
Relative percent difference (RPD) used for duplicates:
W
RPD = x 100%
x
Variance (S2) obtained from at least three replicates:
Σ(xi – x)2 Σxi2 – (Σxi)2/n
S2 = or S2 =
n–1 n–1
Standard deviation (s): square root of the variance. The standard deviation is
often preferably calculated because it is expressed in the same physical unit
as the original data.
Coefficient of variation (CV):
S
CV = x 100%
x
CV is mainly used when the size of the standard deviation changes with the
magnitude of the mean.

c. Presentation of analytical results (example):
A cargo of SBM was sampled and 14 samples were collected (n = 14 replicates)
to determine protein content of the SBM. The sampling was conducted to be
representative on the entire load. The results of the analyses are presented in Table 18.

Table 18.
Protein content of soybean meal: calculation steps to
determine the mean and variance
n° sample measurement: xi xi – x (xi – x)2

1 50.2 1.79 3.19
2 54.0 5.59 31.20
3 48.7 0.29 0.08
4 44.2 -4.21 17.76
5 45.4 -3.01 9.09
6 46.8 -1.61 2.61
7 51.3 2.89 8.33
8 49.7 1.29 1.65
9 47.7 -0.71 0.51
10 47.6 -0.81 0.66
11 42.9 -5.51 30.41
12 48.0 -0.41 0.17
13 52.1 3.69 13.58
14 49.2 0.79 0.62

Sum Σ 677.8 0 119.86
Σ/n x = 48.41 - -
2
Σ / (n-1) - - S = 9.22

89

10. Data Management

In this example, the mean protein content in the sample was of 48.41 % of DM
and the standard deviation of 3.04 % of DM.

The construction of histograms is helpful to visualize the data (average value and
range) and to determine if they follow a normal distribution. Histograms are an
important tool in quality control (QC) because they help to identify the cause of
problems by the shape (i.e., uni- or bimodal) and the width of the distribution.

d. How to construct a histogram?
This procedure was developed from the above example.
- Calculate the range of the values: w = 54.0 – 42.9 = 11.1 % of DM
- Choose a number of intervals (ex. 7).
The size of the interval is equal to: w / 7 (= 1.6)
For practical considerations, it is better to round the interval size (ex. 2 % of DM).
- Calculate the frequency of occurrences for each interval:
Ex. Interval: 41 - 43 > occurrence: 1
Interval: 45 - 47 > occurrence: 2
- Draw the corresponding figure (Figure 5).

Figure 5
Histogram of the data based on seven intervals

0.4

0.3
Frequency

0.2

0.1

0
42 44 46 48 50 52 54
Protein content, %

CONTENTS

90

10. Data Management

10.2. Quality indicators
The reliability of analytical results and thus the quality of our estimations
concerning the population (SBM) depend on critical parameters. First of all, the
analytical method should be specific for the compound to be measured (ex. crude
protein). The method should also be sensitive to variations in the amount of the
compound under study. A small change in CP content should result in a relatively
equivalent change in the instrumental response. Finally, accuracy and precision
of the method are required (Figure 6) and quality indicators can help to evaluate
these two measures.

Figure 6
Definition of accuracy versus precision

Good precision
Poor accurarcy

Good precision
Good accurarcy

Poor precision
Poor accurarcy

(Galyean 1997)

In the above example, the method for crude protein analysis in SBM could
present a poor accuracy (mean value of 46 % of DM when 48 % of DM should be
measured) but a good precision (small range: 45.5 - 46.5 % of DM). On the contrary,
the method could present a good accuracy (48 % of DM) and a poor precision (large
range: 45 - 51 % of DM).

a. Accuracy
Accuracy is a measure of the bias between the analytical results (Xi) and the true
value (Xt). The accuracy can be tested on a sample, when the composition is known.

91

10. Data Management

Accuracy can be determined by: the absolute error (Xi - Xt) or the relative error:
100 x (Xi - Xt) / Xt.
For example, if the CP value of SBM is 48 % of DM and the analytical result yields 50,
the method is not accurate: the absolute error for this result is 2 % of DM and the
relative error is 4.17 %.

How to check the accuracy of a method?
• Certified reference materials (CRM).
When available, CRM are materials issued and certified by an external
organization and whose properties are validated and reliable. The use of CRM
is a powerful tool to assess the good performance in the analytical method.
• Laboratory reference materials (LRM)
Because of the high cost of CRM, in-house reference standards are generally
preferred. The standard recovery is a good indicator of the accuracy of the
method.
• "Spiked" sample.
Accuracy can also be estimated by the ability to measure an amount of
substance in a "spiked" sample. A sample is "spiked" when it contains a precisely
measured amount of substance. This amount is adjusted to a desired and known
level (S).
The percent recovery is then calculated as follows:
QS – QN
% Recovery = x 100
S
where QS is the measured quantity in spiked sample, QN: the measured quantity
in unspiked sample and S: the quantity of substance in spiked sample.
• Blank.
A blank is a QC sample designed to check for contamination into the sampling
and analytical procedure. A method blank should be free of the molecule to be
measured.
• Inter-laboratory comparisons.
Inter-laboratory comparisons programs should be conducted to compare
accuracy of analytical results.

b. Precision
Precision is a measure of the ability to reproduce analytical results.

How to check the precision of a method?
The precision can be estimated with laboratory duplicate samples. These
samples should be collected at the same time and location and analyzed in the same
conditions. Laboratory duplicates intended to be identical to the original sample.
Precision can be determined by calculating the relative percent difference of the

92

10. Data Management

duplicates. It can also be calculated by the standard deviation or coefficient of
variation when three or more replicates are used. High variability (RPD, s, CV) among
duplicates reflects low precision. Table 19 depicts typical and acceptable coefficients
of variation for common analyses.

Table 19
Typical ranges and acceptable coefficients of variation for
proximate analysis in feedstuffs (Galyean, 1997)
Analysis Typical range, % Acceptable CV, %

DM 80 - 100 0.5
Ash 0 - 20 2.0
CP 5 - 50 2.0
ADF 5 - 70 3.0
NDF 10 - 80 3.0
ADL 0 - 20 4.0
EE 1 - 20 4.0

CONTENTS

10.3. Significance of parameter estimates
a. Hypothesis tests
These tests can be performed to address the uncertainty of the sample estimates
and to take decisions about the validity of the data (Feinberg, 1996). For example, it
can help to determine if an observed value of a statistic differs from a hypothesized
value of a parameter. For our example on SBM the question is:“Is the crude protein
analyzed in the sample really different from the population of all SBM in the load?”
To answer this, generally two hypotheses can be tested:
Ho: "null hypothesis". The population mean is equal to a reference value
(µ − µo = 0).
The mean value of crude protein in all SBM is equal to 48 % of DM.
H1: "alternative hypothesis". The population mean is different to the reference
value (µ − µo ≠ 0). The mean value of crude protein in all SBMs differs from
48 % of DM.

Select a level of significance (α):
The level of significance represents the probability to reject the hypothesis Ho.
By convention, α is set at 5 % - sometimes 10 % is accepted but this increases the
probability of being wrong (10 % vs. 5 %).

93

10. Data Management

Calculate the test statistics, in other words test the hypothesis from
the sample data:
The test procedure measures the compatibility between the null hypothesis
and the data. Several statistical tests exist. The choice of the statistical test will
depend on the sample (size), the knowledge on population parameters (ex. variance),
the accepted/assumed probability and the hypotheses under question.
For example: can it be concluded from the sampling procedures that the mean
value of CP in SBM is 48 % of DM? The Student t-test of the population mean is the
test of choice for this case (n small, σ unknown); the following formula can be used
for one-sample testing:
x–µ
t = s o , therefore t = 48.41 – 48 = 0.51
3.04
√n √14
Determine the P-value:
The probability value (P-value) of a statistical hypothesis test is the probability
to obtain results equal to or more "extreme" in future experiments (given that Ho is
true). This probability (P) can be determined using statistical tables to compare the
value of the test statistic (ex. 0.51) with values from the probability distribution
(ex. Student distribution). The Student t-test and the Normal z tables are presented
in Appendix 7 and 8.

In the above example, the lower and upper bounds for a Student-t test statistic
with n-1 = 13 degrees of freedom: (tp13) can be determined with the tables in
appendix 7: t0.4 (13) < 0.51 < t0.25 (13), therefore P ranges from 0.25 to 0.4.
The P-value for a two-sided test is twice the P-value of a one-sided test; consequently,
in the above example P is between 0.50 and 0.80. The computed actual P-value is
equal to 0.62.

Set up decision rules:
P-value ≤ α
The difference is said to be "statistically significant" when P, the probability that
Ho is true given the sample data, is less or equal to the level of significance. In that
case, it can be concluded that results are not due to chance and the hypothesis Ho
can be rejected.
P-value > α
The difference is attributed to chance or to an error of measurement. In that case,
the null hypothesis cannot be rejected; alternatively, Ho is accepted. In the above
example, P-value is 0.62 (p > 0.05) therefore it is concluded that the crude protein
content of SBM is not statistically different from 48 % of DM.
Two types of errors may occur (Table 20-next page). Ho is rejected when it is true
(type I error). Ho is accepted when H1 is true (type II error). The probability α represents
the "producer's risk" whereas β represents the "consumer's risk". For example, α is the
risk of rejecting a "good" lot and β, the risk of accepting a "bad" lot.

94

10. Data Management

Table 20
Error types in hypothesis testing

Actual situation
H0 H1
Type I error Correct
Reject Ho
(P : α) (P : 1 – β)
Decision
Correct Type II error
Retain Ho
(P : 1 – α) (P : β)

The results of the tests should always be applied with caution. It is particularly
important to choose an appropriate sample size to answer the question and detect
differences. The ability of the test to detect differences (P = 1 – β), called power of
the test, depends on the size of the difference, the sample size and the level of
significance. The test's power increases as sample size increases but decreases as the
level of significance increases. Typical power probabilities are set at 0.80, the sample
size needed to reach this value can then be estimated.

b. Confidence interval
The sample mean and the population mean are rarely exactly the same but
sometimes we like to be able to say that we are pretty sure that the population is
within a given amount of our sample mean. Statistically it is possible to calculate an
interval around the sample mean with a given level of confidence (probability).
Interval estimates are dependent on the heterogeneity or variance associated with
the measured variable (s2), the number of samples (n) and the probability of being
wrong (α).
The confidence interval for the mean µ of the population (σ unknown) can be
determined with the z or t values in statistical tables (Appendix 7, 8):
- in the case of a small sample from a normal population:
s
x ± t α/2 (n –1)
√n
- in the case of a large sample from a normal population (n > 30):
s
x ± z α/2
√n
For our example:
s = 48.41 ± t (13) 3.04
x ± t α/2 (n –1)
√n √14
0.025

3.04
= 48.41 ± 2.16 x
√14
= [46.66 - 50.16]

Thus we are 95 % confident that the average crude protein content in SBM is
between 46.7 and 50.2 % of DM.

95

10. Data Management

c. Sample size determination
Sampling is costly and time-consuming, therefore it is important to know what
sample size should be selected to obtain a desired precision. The sample size can be
determined if we know the confidence required (P-value; ex. α = 0.05), the variability
in the population and the precision required. The precision is expressed as H,
representing half the width of the confidence interval. The answer should be
rounded up to next following whole number.

- unknown population variance: n = ( t αH s )2
( /2)

- known population variance (σ): n = (z αH σ)2
( /2)

CONTENTS

10.4. Control charts
Control charts are efficient devices to control an analytical method and to
check its stability over time (Daudin and Tapiero, 1996). They are used to indicate the
range of variability of a process and to decide whether the process is under statistical
control. In certification schemes (HCAPP, ISO, GMP) and solid quality control
programs, they have become fundamental tools. For routine QC procedures, different
types of charts are developed depending on the controlled parameter (average or
range) and the number of replicates per sample.
- Measurements in group (X or range chart)
- Individual measurements (individual X or moving range chart)
Historical data and experience are generally used to establish the specific charts.

Basic Principles
A control chart is composed of:
- A centerline:
This value is calculated as the average value of a large number of samples
plotted (n > 30).
- Horizontal lines:
These lines represent the upper control limits (UCL) and the lower control
limits (LCL). Typically, these limits are calculated based on the mean and
standard deviation:
• warning control limits: mean ± 2s
• action control limits: mean ± 3s
The data is plotted over time.

96

10. Data Management

The results of the analytical measurements are plotted in chronological order on
the control chart. If the process is in control, the sample points will fall between the
control limits. However, points that plot outside of the control limits are interpreted
as evidence that the process is out of control. Exceeding a warning control limit
generally means that the process is not operating properly. The analyst can try to
assess the source of errors, however, no action is needed, providing that next results
fall within the warning limits. Exceeding an action control limit leads to the necessary
identification and elimination of the causes of errors.

How to develop an individual control chart?
When samples are individual measurements, control charts can be drawn up very
simply. In this case, the moving range xi – xi-1 can be calculated for each pair of
data (see Table 21).
The lines are then defined as follows:
Centerline: x = 48.41
The standard deviation of the process is estimated from the average moving
range (MR) divided by 1.128 (conversion factor d2 for n = 2).
Action control limits:
MR
x ± 3 1.128
(LCL = 39.98 and UCL = 56.84).

Table 21
Protein content (% of DM) in soybean meal samples

n° sample measurement: xi moving range
1 50.2 -
2 54 3.8
3 48.7 5.3
4 44.2 4.5
5 45.4 1.2
6 46.8 1.4
7 51.3 4.5
8 49.7 1.6
9 47.7 2
10 47.6 0.1
11 42.9 4.7
12 48 5.1
13 52.1 4.1
14 49.2 2.9
Average x = 48.41 MR = 3.17

97

10. Data Management

Figure 7
Control chart for protein content (% of DM) analyses in soybean meal samples

58 – Upper control limit
–
54 –
–
Protein content

50 –
–
46 –
–
42 –
Lower control limit
–
38 –I I I I I I I I I I I I I I
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Sampling number

The process can be said to be “in control” since none of the points fall outside the control
limits (Figure 7).
CONTENTS

10.5. Follow-up and application of analytical results
Analyses of any type are always associated with uncertainty. Indeed, both systematic
and random errors can occur. Therefore, it is important to evaluate the size of the errors and
to have an estimation of the reliability of the analytical results. This procedure should be part
of a standard quality control procedure and needs to be developed through a joint effort
between analysts and nutritionists. Each has a specific responsibility/task, which can be
summarized as follows:

98

10. Data Management

a. Analyst:
• Perform the sampling and analysis correctly.
• Use proper QC measures to validate the data and to keep systematic and
random errors under control: calibration standards, controls, duplicate field
samples and blanks to estimate sampling errors, laboratory duplicates to
estimate analytical errors.
• Establish quality objectives (precision, accuracy) or quality acceptance limits.
The acceptable level of variation between duplicates varies by test and by
concentration of nutrient (Table 3).
• Propose corrective actions (re-sampling, re-calibration...) if needed.

b. Nutritionist:
• Define the parameters that need to be analyzed.
• Include ingredient quality specifications in the purchasing agreement
and provide this information to the analyst.
• Adjust formulation.
• Find alternative ingredient if quality specifications are not met.

The objectives of the analyst and the nutritionist may be to reduce variation
(increase quality of the results) but also to maximize the value of a raw material.
There is a subjective judgment associated with quality control. The risk of type I or
type II errors exists. It is possible to reduce these risks (higher significance level,
higher power of the tests, and larger number of samples) but this is generally
associated with a higher cost.
CONTENTS

99

11. REFERENCES
AACC, 1976. American Association of Cereal Chemists, Cereal Laboratory Methods,
St. Paul, Minn., USA, 1968.
AAFCO, 2001. Official Publication 2001. Association of American Feed Control
Officials Incorporated. Atlanta, GA.
AFNOR XP V18-202, 1997. Aliments des animaux: dosage des inhibiteurs
trypsiques. Agence française de normalisation, Paris, France.
Anderson-Haferman, J. C., Y. Zhang, C. M. Parsons, and T. Hymoitz. 1992. Effect
of heating on the nutritional quality of kunitz-trypsin-inhibitor-free and
conventional soybeans for chickens. Poultry Sci. 71:1700-1709.
AOAC, 1980. Official Methods of Analysis of the Association of Official Analytical
Chemists.
AOAC, 1990. Official methods of analysis. 15th edition. AOAC, Arlington, VA.
AOAC, 1995. Animal Feed. Chapter 4 in “Official methods of analysis” 16th ed. AOAC
.
International, Arlington, VI, USA.
AOAC, 2000. Official Methods of Analysis of AOAC International, 17th d. Official
method 968.06.
AOCS, 1980. Urease Activity. Official Method Ba 9-58. American Oil Chemists Society.,
Champagne, IL.
AOCS, 1989. Official method Ba 11-54. Nitrogen Solubility Index (NSI).
AOCS, 1993a. Official method Ba 10-65. Protein dispersibility index (PDI).
AOCS, 1993b. Volume 1, Official Method Ca 6c – 40.
AOCS Method 1997. Ba 12-75. In Official methods and recommended practices of
the American Oil Chemists Society. American Oil Chemists Society, Champaign,
Illinois.
Araba and Dale, 1990. Evaluation of protein solubility as an indicator of over-
processing of soybean meal . Poultry Sci. 69: 76-83.
ASA, 1998. Soy Importers Handbook, 2nd ed. American Soybean Association, United
Soybean Board, Missouri Soybean Merchandising Council.
ASAE, 1993. American Society of Agricultural Engineers. Methods of determining
and expressing fineness of feed materials by sieving. ASAE Standard ASAE S319.2.
Bach, A., M. D. Stern, N. R. Merchen, and J. K. Drackley. 1998. Evaluation of
selected mathematical approaches used to describe the kinetics of protein
degradation in situ. Anim. Sci. 76:2885-2893.
Batal, A.B., M.W. Douglas, A.E. Engram, and C.M. Parsons. 2000. Protein
dispersibility index as an indicator of adequately processed soybean meal.
Poultry Sci., 79: 1592-1596.

100

11. References

Berk, Z., 1992.“Technology of production of edible flours and protein products from
soybeans” FAO Agricultural services Bulletin No. 97.
.
Broderick, G. A. 1987. Determination of protein degradation rates using a rumen
in vitro system containing inhibitors of microbial nitrogen metabolism.
Br. J. Nutr. 58:463-475.
Caine, W. R., M. W. A. Verstegen, W. C. Sauer, S. Tamminga and H. Schulze.
1998. Effect of protease treatment of soybean meal on content of total soluble
matter and crude protein and level of soybean trypsin inhibitors. Anim. Feed.
Sci. Technol. 71:177-183.
CFIA, 2003. Canadian Food Inspection Agency, Approved Feed Ingredients, Schedule
IV Part 1, List of approved feed ingredients, Class 5.
http://www.inspection.gc.ca/english/anima/feebet/sched4/tab_ae.shtml
Chamberlain, D. G., and P. C. Thomas. 1979. Ruminal nitrogen metabolism and
the passage of amino acids to the duodenum in sheep receiving diets containing
hay and concentrates in various proportions. J. Sci. Food Agric. 30: 677-686.
Cook, D.R. 1998. The effect of dietary soybean isoflavones on the rate and efficiency
of growth and carcass muscle content in pigs and rats. Ph.D. dissertation. Iowa
State Univ., Ames
CVB, 2000. Centraal Veevoederbureau; Veevoedertabel 2000; chemische
samenstelling, verteerbaarheid en voederwaarde van voedermiddelen.
Daudin J-J. and Tapiero C.S., 1996. Les outils et les contrôles de la qualité.
Economica (ed.), Paris, France, 112p.
de Lange, C.F.M., Nyachoti, C.M., Verstegen, M.W.A., 2000. The significance of
anti-nutritional factors in feedstuffs for monogastric animals. In: Feed evaluation
principles and practice, Moughan, P.J., Verstegen, M.W.A. and Visser-Reyneveld,
M.I. (eds). Wageningen Press, 169-188.
Delort-Laval, J., 1991. Dosage des facteurs anti-nutritionnels. In: Techniques
d'analyse et de contrôle dans les industries agro-alimentaires. Vol. 4: Analyse des
constituants alimentaires, Tec & Doc, Lavoisier, 371-387.
Dudley-Cash, W.A. 1999. Methods for determining quality of soybean meal protein
important. Feedstuffs. Vol 71. N°1 January 4 1999.
FEDNA, 1994.“Normas FEDNA para la Formulacion de Piensos Compuestos; valor
nutritivo de los concentrados de proteina vegetal” G. de Blas, G.G. Mateos y
.
P.G. Rebollar.
Feinberg M., 1996. La validation des méthodes d'analyse. Masson (ed.), Paris,
France, 397p.
Frankel, E. N. 1984. Lipid oxidation: mechanism, products and biological
significance. JAOCS. 61:1908-1917.
Galyean M.L., 1997. Laboratory procedures in animal nutrition research, 193p.
Web: http://www.asft.ttu.edu/home/mgalyean/

101

11. References

Gill, C., 2003.“World Feed Panorama: Back to basics of growth” Feed International
.
p. 6.
GIPSA, 1995. Grain Inspection, Processors and Stockyard Administration.
Grain Inspection Handbook 1. Federal Grain Inspection Service, USDA.
Givens, D.I., J.L. De Boever and E.R. Deauville. 1997. The principles , practices
and some future applications of near infrared reflectance spectroscopy for
predicting the nutritive value of foods for animals and humans. Nutrition
Research Reviews, 10:83-114.
Goering, H. K. and P. J. Van Soest. 1970. Forage fiber analyses (apparatus,
reagents, procedures and some applications). USDA-ARS Agricultural Handbook
379. U.S. Government Printing Office., Washington, DC.
Hanssen, O.K., 2003.“Soya is no soya”: (soya proteins for feed products).
Feed International 23 (9):14-18.
Heppell, L.M.J., Sissons, J.W., Pederson, H.E., 1987. A comparison of the anti-
genicity of soya-bean-based infant formulas. Br. J. Nut., 58, 393-403.
Herrman, T., 2001. Sampling: Procedures for Feed. Kansas State University
Agricultural Experiment Station and Cooperative Extension Service. MF-2036.
Feed Manufacturing.
Huisman, J. and Tolman, G.H., 1992. Antinutritional factors in the plant proteins of
diets for non-ruminants. In: Recent advances in Animal Nutrition, Garnsworthy,
P.C., Haresign, W. and Cole, D.J.A. (eds). Butterworth Heinemann, 3-32.
INFIC, 1980.“International Feed Descriptions, International Feed Names and Country
Feed Names” International Network of Feed Information Centres, Publication
.
No. 5, Utah State University, Logan.
INRA-AFZ, 2002. Tables de composition et de valeur nutritive des matières
premières destinées aux animaux d’élevage. Sauvant,D, J.-M Perez et G. Tran, Eds.
Kakade, M.L., N. Simons and I.E. Liener, 1969. An evaluation of natural versus
synthetic substrates for measuring the antitryptic activity of soyabeans samples.
Cereal Chem. 46, 518-526.
Kakade, M.L., J.J. Rackis, J.E. McGhee and G. Puski, 1974. Determination of
trypsin inhibitor activity of soy products: a collaborative analysis of an improved
procedure. Cereal Chem. 51, 376-382.
Kerley, M.S., & Allee, G.L. (2003). Modifications in soybean seed composition to
enhance animal feed use and value: Moving from a dietary ingredient to a
functional dietary component. AgBioForum, 6(1&2), 14-17. Available on the
World Wide Web: http://www.agbioforum.org.
Lee, H. and J. D. Garlich. 1992. Effect of overcooked soybean meal on chicken
performance and amino acid availability. Poultry Sci. 71:499-508.
Lee, S.J., Ahn, J.K., Kim, S.H., Kim, J.T., Han, S.J., Jung, M.Y., Chung, I.M., 2003.
Variation in isoflavone of soybean cultivars with location and storage duration.
J. Agric. Food Chem., 51, 3382-3389.

102

11. References

Liener, I.E., 2000.“Non-Nutritive Factors and Bioactive Compounds in Soy. In: Soy in
Animal Nutrition” Ed. J.D. Drackley. Federation of Animal Science Societies.
.
Liu, K., 1997.“Soybeans: Chemistry, Technology, and Utilization” Aspen Publishers,
,
New York.
Mahadevan, S., F. D. Sauer, and J. D. Erfle. 1987. Preparation of protease from
mixed rumen microorganisms and its use for the in vitro determination of the
degradability of true protein in feedstuffs. Can. J. Anim. Sci. 67: 55-64.
McNaughton, J. M., F. N. Reece, and J. W. Deaton. 1980. Effect of moisture
content and cooking time on soybean meal urease index, trypsin inhibitor
content, and broiler performance. Poultry Sci. 59:2300-2306.
Nocek, J. E., J. H. Herbein, and C. E. Polan. 1983. Total amino acid release rates of
soluble and insoluble protein fractions of concentrate feedstuffs by Streptomyces
griseus. J. Dairy Sci. 66:1663-1667.
NOPA, 1997. National Oil Producers Association. Yearbook 1997. 1225 Twenty-third
street N.W., Washington, D.C. 20037, U.S.A.
NOPA, 1999. 1997-1999 Yearbook and Trading Rules. National Oilseed Processors
Association, Washington, D.C.
NRC, 1982. National Research Council.“United States-Canadian Tables of Feed
Composition”; Third Revision. National Academy Press, Washington D.C.
NRC, 1998. Nutrient requirements of swine; Tenth Revised Edition, National Academy
of Sciences 1998.
NRC, 2001. Nutrient requirements of dairy cattle; Seventh Revised Edition, National
Academy of Sciences 2001.
Parsons, C. M., K. Hashimoto, K. J. Wedekind, and D. H. Baker. 1991. Soybean
protein solubility in potassium hydroxide: an in vitro test of in vivo protein
quality. J. Anim. Sci. 69:2918-2924.
Parsons, C. M., K. Hashimoto, K. J. Wedekind, Y. Han and D. H. Baker. 1992.
Effect of over-processing on availability of amino acids and energy in soybean
meal. Poultry Sci., 71:133-140.
Payne, R.L., T.D. Bidner, L.L. Southern and J.P. Geaghan. 2000. Effects of dietary
soy isoflavones on growth , carcass traits, and meat quality in growing-finishing
pigs. The Louisiana Agric. Exp. Sta. No.: 2000-11-0323.
Peisker, M., 2001. Manufacturing of soy protein concentrate for animal Nutrition.
Cahiers Options Mediterraneennes, 54, 103 -107.
Poos-Floyd, M., T. Klopfenstein, and R. A. Britton. 1985. Evaluation of
laboratory techniques for predicting ruminal protein degradation.
J. Dairy Sci. 68:829-839.
Pusztai, A., 1991. Plant lectins. Cambridge University Press, Cambridge.
Schulze, H., Saini, H.S., Huisman, J., Hessing, M., van der Berg, W.,
Verstegen, M.W.A., 1995. Increased N secretion by inclusion of soya lectin in
the diets of pigs. J. Sci. food Agric., 69, 501-510.

103

11. References

Shenk, J.S. and M.O. Westerhaus, 1991. Population definition, sample selection,
and calibration procedures for near infrared reflectance spectroscopy.
Crop Science, 31:469-474.
Shermer,W.D et al. 1985. Oxidation of feed: How much has occurred.
Feedstuff, 4:19-20
Shipley, R. A., and R. E. Clark. 1972. Tracer Methods for In Vivo Kinetics: Theory
and Application. Academic Press, New York.
Speedy, A.W. 2002.“Overview of World Feed Protein Needs and Supply” FAO Expert
.
Consultation and Workshop on Protein Sources for the Animal Feed Industry.
Bangkok, Thailand.
Sukhija, P.S., and D.L. Palmquist, 1988. Dissociation of calcium soaps of long-
chain fatty acids in rumen fluid. J Dairy Sci. 1990. 73:1784-7.
Tilley, J. M. A., and R. A. Terry. 1963. A two-stage technique for the in vitro
digestion of forage crops. J. Br. Grassl. Soc. 18:104.
Tremblay, G. F., G. A. Broderick, and S. M. Abrams. 1996. Estimating ruminal
protein degradability of roasted soybeans using near infrared reflectance
spectroscopy. J. Dairy Sci. 79: 276-282.
Trucksess, M.W., M.E. Stack, S. Nesheim, D.L. Park and A.E. Pohland. 1989.
Enzyme-linked immunosorbent assay of aflatoxins B1,B2 and G1 in corn,
cottonseed, peanuts, peanut butter and poultry feed: collaborative study.
J. of the Assoc. of Official Chemists 72, 957-962.
Tukur, H.M., Lallès, J.P., Mathis, C., Caugant, I., Toullec, R., 1993. Digestion of
soybean globulins, glycinin, α-conglycinin and β-conglycinin in the preruminant
and the ruminant calf. Can. J. Anim. Sci., 73, 891-905.
Tukur, H.M., Lallès, J.P., Plumb, G.W., Mills, E.N.C., Morgan, M.R.A., Toullec, R.,
1996. Investigation of the relationship between in vitro ELISA measures of
immunoreactive soy globulins and in vivo effects of soy products.
J. Agr. Food Chem., 44, 2155-2161
UNEP, 1999.“Agri Food Production; Facts and Figures” UNEP Industry and
.
Environment, p 4 – 31.
USDA, 1999.“United States Standards for Soybeans (Revised edition June 1999)”
.
Subpart A – General Provisions Terms Defined.
USDA, 2001.“Grain inspection, packers and stockyards administration”Technical
services administration.
http://www.usda.gov/gipsa/reference-library/brochures/soyinspection.pdf
USDA, 2003.“Oilseeds: World Markets and Trades” Circular series FOP 02 – 03 Foreign
.
Agricultural Service. U.S. Department of Agriculture.
USDA-IOWA State University, 2002. Database on the isoflavone contents of food,
release 1.3.
Web: http://www.nal.usda.gov/fnic/foodcomp/Data/isoflav/isoflav.html

104

11. References

Van Kempen, T.A. and J.C. Bodin, 1998. Near infrared reflectance spectroscopy
(NIRS) seems to be superior to nitrogen-based regressions as a rapid tool in
predicting the poultry digestible amino acid content of commonly used
feedstuffs. Anim. Feed Sci. Technol.76:139-147.
Van Soest, P. J., J. B. Robertson, and B. A. Lewis. 1991. Methods for dietary fiber,
neutral detergent fiber, and nonstarch polysaccharides in relation to animal
nutrition. J. Dairy Sci. 74: 3583-3597.
Waldroup, P. W., B. E. Ransey, H. M. Hellwig and N. K. Smith. 1985. Optimum
processing for soybean meal used in broiler diets. Poultry Sci. 64:2314-2320.
Wang, H.-J. and Murphy, P.A., 1994. Isoflavone content in commercial soybean
foods. J. Agric. Food Chem., 42, 1666-1673.
West, B. 2002.“EU soybean meal purchasing patterns and quality assurance”
.
Final report. American Soybean Association (ASA)
Whittle, E., and M. Araba. 1992. Sources of variation in the protein solubility assay
for soybena meal. J. Appl. Poultry Res. 1:221-225.
Williams, P.C., 1987. Variables affecting near-infrared reflectance spectroscopic
analysis. In: P. Williams and K. Norris (eds.) Near-Infrared Technology in the
Agricultural and Food Industries. St. Paul, MN; American Association of Cereal
Chemists Inc., pp. 143-167.
Witte, V. C., G. F. Krause, and M. E. Bailey. 1970. A new extraction method for
determining 2-Thiobarbituric acid values of pork and beef during storage.
J. Food Sci. 35:582—585.
CONTENTS

105

12. ANNEX
Appendix 1
Average nutrient composition of soy protein products used in livestock diets
Heat
Soybean Ground processed Soy Soy
seeds FF soybean FF soybean protein protein Soybean
Nutrient Unit extruded seeds seeds concentrate isolate hulls

Dry matter % 88.10 89.72 89.44 91.83 93.38 89.76
Crude protein % 34.80 37.50 37.08 68.60 85.88 12.04
Crude fiber % 5.20 5.03 5.12 1.65 1.32 34.15
Ether extracts % 17.90 18.04 18.38 2.00 0.62 2.16
Ash % 5.20 4.77 4.86 5.15 3.41 4.53
NDF % 11.00 12.50 12.98 13.50 – 56.91
ADF % 6.40 8.80 7.22 5.38 – 42.05
ADL % 1.00 4.10 4.30 0.40 – 2.05
Starch % 0.00 4.59 4.66 – – 5.95
Total sugars % 7.70 – – – – 1.40
Gross energy kcal/kg 4870 5027 5013 – 5370 3890
DE-swine kcal/kg 3800 4157 4088 4517 4545 1944
ME energy-swine kcal/kg 3560 3739 3714 3661 3955 1687
NE-swine kcal/kg 2560 2920 2803 2000 2000 1074
App. ME-broiler kcal/kg 3350 3475 3332 – 4060 –
App. ME-adults kcal/kg 3450 3621 3564 2472 3945 334
ME-ruminants kcal/kg 3400 3373 3400 2690 – 1241
NE-dairy kcal/kg 2159 1986 2097 1600 – 1544
NE-beef kcal/kg 2311 2080 2230 1610 – 1618

Amino acids

Lysine % 2.16 2.39 2.34 4.59 5.26 0.73
Threonine % 1.40 1.54 1.53 2.82 3.17 0.73
Methionine % 0.53 0.51 0.52 0.87 1.01 0.14
Cystine % 0.56 0.55 0.55 0.89 1.19 0.16
Tryptophane % 0.44 0.49 0.49 0.81 1.08 0.12
Isoleucine % 1.61 1.78 1.79 3.68 4.25 0.41
Valine % 1.66 1.88 1.85 3.69 4.21 0.49
Leucine % 2.59 2.81 2.76 5.41 6.65 0.75
Phenylalanine % 1.74 1.84 1.87 3.60 4.35 0.47
Tyrosine % 1.23 1.18 1.22 1.55 3.14 0.43
Histidine % 0.93 0.95 0.96 2.41 2.25 0.29
Arginine % 2.57 2.73 2.71 7.34 6.87 0.62
Alanine % 1.41 1.54 1.52 – 3.33 0.51
Aspartic acid % 3.88 4.09 4.06 – 8.29 1.14
Glutamine % 6.17 6.37 6.35 – 12.0 1.49
Glycine % 1.48 1.52 1.59 3.32 3.38 0.85
Serine % 1.78 1.94 1.89 5.19 4.81 0.67
Proline % 1.83 1.89 1.86 – 4.45 0.55

106

12. Annex

Heat
Soybean Ground processed Soy Soy
seeds FF soybean FF soybean protein protein Soybean
Nutrient Unit extruded seeds seeds concentrate isolate hulls

Minerals
Calcium g/kg 3.10 2.58 2.62 2.37 1.50 4.96
Phosphorus g/kg 5.50 5.83 5.70 7.63 6.50 1.59
Magnesium g/kg 2.30 2.98 2.80 1.85 0.80 2.23
Potasium g/kg 18.50 14.60 15.93 12.35 2.75 12.15
Sodium g/kg 0.80 0.20 0.29 0.55 2.85 0.10
Chloride g/kg 0.40 0.25 0.33 0.20 0.20 0.25
Sulfur g/kg 2.80 2.20 2.43 0.70 7.00 0.95
Manganese mg/kg 23.00 36.00 31.79 27.50 5.00 10.67
Zinc mg/kg 40.00 57.00 47.80 47.00 34.00 37.75
Copper mg/kg 34.00 6.90 15.17 17.00 14.00 10.68
Iron mg/kg 146.00 84.00 128.01 137.00 137.00 437.50
Selenium mg/kg 0.28 0.11 0.17 0.14 0.14 0.16
Cobalt mg/kg – – 0 – – 0.06
Molybdenum mg/kg 4.00 – 2.00 – – 0.60

Fatty acids
Myristic acid-C14:0 % 0.01 0.03 0.03 – – 0
Palmitic acid-C16:0 % 1.05 1.47 1.95 – – 0.24
Palmitoleic acid-C16:1 % 0.02 0.03 0.04 – – 0
Stearic acid-C18:0 % 0.38 0.53 0.71 – – 0.09
Oleic acid-C18:1 % 2.17 2.97 3.96 – – 0.50
Linoleic acid-C18:2 % 5.31 7.28 9.70 – – 1.21
Linolenic acid-C18:3 % 0.74 1.06 1.40 – – 0.17

FF Soybean is Full Fat Soybean; SBM = Soybean meal.
Source: compilation on NRC, INRA-AFZ, CVB, SRTNA and selected suppliers.

107

12. Annex

Appendix 2
Average nutrient composition of soy protein products used in livestock diets
SBM SBM SBM
SBM solvent solvent solvent Soybean
mechanically extracted extracted extracted mill Soybean
Nutrient Unit extracted 44 48 50 feed oil

Dry matter % 89.80 88.08 87.58 88.20 89.70 99.25
Crude protein % 43.92 44.02 46.45 48.79 12.93 1.40
Crude fiber % 5.50 6.26 5.40 3.42 33.47 –
Ether extracts % 5.74 1.79 2.13 1.30 1.70 97.20
Ash % 5.74 6.34 6.02 5.78 4.73 0.40
NDF % 21.35 13.05 11.79 9.95 – –
ADF % 10.20 8.76 7.05 5.00 41.40 –
ADL % 1.17 0.75 0.90 0.40 – –
Starch % 7.00 5.51 5.46 3.28 – –
Total sugars % – 9.06 9.17 9.29 – –
Gross energy kcal/kg – 4165 4130 4120 – –
DE-swine kcal/kg – 3394 3446 3776 1167 8915
ME energy-swine kcal/kg – 2986 3210 3299 925 8400
NE-swine kcal/kg – 1903 1955 1992 – 6760
App. ME-broiler kcal/kg – 1929 1973 2147 – 8600
App. ME-adult kcal/kg – 2171 2208 2464 774 8805
ME-ruminants kcal/kg – 2831 2840 3010 1630 8180
NE-dairy kcal/kg – 1706 1748 1826 1001 4520
NE-beef kcal/kg – 1838 1847 1993 965 5022

Amino acids

Lysine % 3.50 2.85 2.89 3.00 0.65 –
Threonine % 2.21 1.80 1.84 1.90 0.30 –
Methionine % 0.80 0.62 0.63 0.67 0.13 –
Cystine % 0.77 0.68 0.73 0.73 0.14 –
Tryptophane % 0.74 0.56 0.63 0.65 0.13 –
Isoleucine % 2.88 2.26 2.17 2.30 041 –
Valine % 2.73 2.19 2.30 2.38 0.38 –
Leucine % 4.29 3.42 3.60 3.60 0.58 –
Phenylalanine % 2.79 2.16 2.37 2.37 0.38 –
Tyrosine % 1.52 1.61 1.68 1.64 0.23 –
Histidine % 1.44 1.64 1.21 1.21 0.18 –
Arginine % 3.98 2.99 3.48 3.53 0.75 –
Alanine % 1.85 2.53 2.05 2.04 – –
Aspartic acid % 5.16 4.03 5.49 5.55 – –
Glutamine % 8.18 6.29 8.62 8.52 – –
Glycine % 2.29 3.46 1.97 2.09 0.48 –
Serine % 2.20 2.13 2.38 2.49 0.30 –
Proline % 2.35 2.17 2.37 2.43 – –

108

12. Annex

SBM SBM SBM
SBM solvent solvent solvent Soybean
mechanically extracted extracted extracted mill Soybean
Nutrient Unit extracted 44 48 50 feed oil

Minerals

Calcium g/kg 2.96 3.12 3.07 2.68 4.05 –
Phosphorus g/kg 6.64 6.37 6.37 6.36 1.75 –
Magnesium g/kg 2.84 2.72 3.03 2.88 3.20 –
Potasium g/kg 20.28 19.85 22.00 20.84 15.20 –
Sodium g/kg 0.33 0.18 0.18 0.88 2.50 –
Chloride g/kg 0.72 0.42 0.35 0.53 – –
Sulfur g/kg 3.37 3.51 2.76 4.30 0.55 –
Manganese mg/kg 40.86 37.85 43.11 33.92 290.00 –
Zinc mg/kg 58.98 53.56 50.12 53.70 – –
Copper mg/kg 21.90 20.03 18.04 17.10 – –
Iron mg/kg 218.45 304.84 319.43 190.75 – –
Selenium mg/kg 0.10 0.31 0.30 0.30 – –
Cobalt mg/kg 0.18 0.14 0.17 0.09 – –
Molybdenum mg/kg 3.80 – 2.45 3.56 – –

Fatty acids

Myristic acid-C14:0 % – 0.00 0.00 0.01 – 0.10
Palmitic acid-C16:0 % – 0.77 0.14 1.05 – 10.50
Palmitoleic acid-C16:1 % – 0.00 0.00 0.02 – 0.15
Stearic acid-C18:0 % – 0.28 0.05 0.38 – 4.20
Oleic acid-C18:1 % – 0.28 0.27 0.21 – 23.30
Linoleic acid-C18:2 % 2.87 0.64 0.80 0.56 – 52.00
Linolenic acid-C18:3 % 0.42 0.55 0.12 0.08 – 6.90

FF Soybean is Full Fat Soybean; SBM = Soybean meal.
Source: compilation on NRC, INRA-AFZ, CVB, SRTNA and selected suppliers.

109

12. Annex

Appendix 3
Sampling patterns for bulk carriers
(From: Herrman, 2001)

A. Sampling pattern for bulk carriers containing a
homogeneous load
Sampling pattern as
A C F
recommended by GIPSA (1995) for
the sampling of bulk truck or rail D
shipments of soybean seeds or
soybean meals using a hand-held B E G
sampling devise or an automatic
sampler. Site A: Probe the grain approx. 0.6 m. from the front and side. Site B: Probe
approx. half-way between the front and center; Site C: Probe approx. three-quarter of
the way between the front and center; Site D: Probe grain in the center of the carrier.
Site E,F,G: follow a similar pattern described above for the back part of the carrier.

B. Sampling pattern for bulk carriers containing areas
with damaged material
Recommended stratified
sampling patterns for carriers
containing inferior or damaged
portions of soybean seeds or
soybean meals. In this case a three A B C
step procedure is recommended. A: Probe the carrier as a whole (inferior and sound
portions) as if the load was homogeneous. B: Probe the portion or portions
containing the inferior grain thoroughly so as a representative cross section is
obtained of the damaged or inferior material. C: Probe the portion or portions
with the sound material to collect a representative sample. The sample of each
step should be a minimum of 2 kg. Samples should be analyzed individually and
proportions of sound to inferior material noted.

110

12. Annex

Appendix 4
Sampling devices for soy bean products
(From: Hermann, 2001)
Figure 1A Figure 1B Figure 1C Figure 1D

Grain probes Tapered bag triers Bomb Pelican grain probe sampler
sampler
Appendix 5
Sampling guidelines for bagged material

Sampling of 1 bag: Stand bag up and insert sampling probe in top corner of the bag.
Lower the probe diagonally through the bag to reach the opposite corner and
withdraw sample.

For lots up to 10 bags, each bag should be samples.

Sampling of more than 10 bags: sample 10 bags selected at random.

Enough material should be collected to perform the necessary assays and retain a
sample. Generally a 0.5 kg sample is adequate.

Appendix 6
Devices for splitting of samples
(From: Hermann, 2001)
Figure 2A Figure 2B

Riffle sample splitter Boerner divider

111

12. Annex

Appendix 7
Student's t test: tp(df)
Degrees
of freedom Probability, p

0.40 0.25 0.10 0.05 0.025 0.01 0.005 0.0005
1 0.324920 1.000000 3.077684 6.313752 12.70620 31.82052 63.65674 636.6192
2 0.288675 0.816497 1.885618 2.919986 4.30265 6.96456 9.92484 31.5991
3 0.276671 0.764892 1.637744 2.353363 3.18245 4.54070 5.84091 12.9240
4 0.270722 0.740697 1.533206 2.131847 2.77645 3.74695 4.60409 8.6103
5 0.267181 0.726687 1.475884 2.015048 2.57058 3.36493 4.03214 6.8688
6 0.264835 0.717558 1.439756 1.943180 2.44691 3.14267 3.70743 5.9588
7 0.263167 0.711142 1.414924 1.894579 2.36462 2.99795 3.49948 5.4079
8 0.261921 0.706387 1.396815 1.859548 2.30600 2.89646 3.35539 5.0413
9 0.260955 0.702722 1.383029 1.833113 2.26216 2.82144 3.24984 4.7809
10 0.260185 0.699812 1.372184 1.812461 2.22814 2.76377 3.16927 4.5869
11 0.259556 0.697445 1.363430 1.795885 2.20099 2.71808 3.10581 4.4370
12 0.259033 0.695483 1.356217 1.782288 2.17881 2.68100 3.05454 4.3178
13 0.258591 0.693829 1.350171 1.770933 2.16037 2.65031 3.01228 4.2208
14 0.258213 0.692417 1.345030 1.761310 2.14479 2.62449 2.97684 4.1405
15 0.257885 0.691197 1.340606 1.753050 2.13145 2.60248 2.94671 4.0728
16 0.257599 0.690132 1.336757 1.745884 2.11991 2.58349 2.92078 4.0150
17 0.257347 0.689195 1.333379 1.739607 2.10982 2.56693 2.89823 3.9651
18 0.257123 0.688364 1.330391 1.734064 2.10092 2.55238 2.87844 3.9216
19 0.256923 0.687621 1.327728 1.729133 2.09302 2.53948 2.86093 3.8834
20 0.256743 0.686954 1.325341 1.724718 2.08596 2.52798 2.84534 3.8495
21 0.256580 0.686352 1.323188 1.720743 2.07961 2.51765 2.83136 3.8193
22 0.256432 0.685805 1.321237 1.717144 2.07387 2.50832 2.81876 3.7921
23 0.256297 0.685306 1.319460 1.713872 2.06866 2.49987 2.80734 3.7676
24 0.256173 0.684850 1.317836 1.710882 2.06390 2.49216 2.79694 3.7454
25 0.256060 0.684430 1.316345 1.708141 2.05954 2.48511 2.78744 3.7251
26 0.255955 0.684043 1.314972 1.705618 2.05553 2.47863 2.77871 3.7066
27 0.255858 0.683685 1.313703 1.703288 2.05183 2.47266 2.77068 3.6896
28 0.255768 0.683353 1.312527 1.701131 2.04841 2.46714 2.76326 3.6739
29 0.255684 0.683044 1.311434 1.699127 2.04523 2.46202 2.75639 3.6594
30 0.255605 0.682756 1.310415 1.697261 2.04227 2.45726 2.75000 3.6460
∞ 0.253347 0.674490 1.281552 1.644854 1.95996 2.32635 2.57583 3.2905

112

12. Annex

Appendix 8
Standard normal Z table

z 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09

0.0 0.0000 0.0040 0.0080 0.0120 0.0160 0.0199 0.0239 0.0279 0.0319 0.0359
0.1 0.0398 0.0438 0.0478 0.0517 0.0557 0.0596 0.0636 0.0675 0.0714 0.0753
0.2 0.0793 0.0832 0.0871 0.0910 0.0948 0.0987 0.1026 0.1064 0.1103 0.1141
0.3 0.1179 0.1217 0.1255 0.1293 0.1331 0.1368 0.1406 0.1443 0.1480 0.1517
0.4 0.1554 0.1591 0.1628 0.1664 0.1700 0.1736 0.1772 0.1808 0.1844 0.1879
0.5 0.1915 0.1950 0.1985 0.2019 0.2054 0.2088 0.2123 0.2157 0.2190 0.2224
0.6 0.2257 0.2291 0.2324 0.2357 0.2389 0.2422 0.2454 0.2486 0.2517 0.2549
0.7 0.2580 0.2611 0.2642 0.2673 0.2704 0.2734 0.2764 0.2794 0.2823 0.2852
0.8 0.2881 0.2910 0.2939 0.2967 0.2995 0.3023 0.3051 0.3078 0.3106 0.3133
0.9 0.3159 0.3186 0.3212 0.3238 0.3264 0.3289 0.3315 0.3340 0.3365 0.3389
1.0 0.3413 0.3438 0.3461 0.3485 0.3508 0.3531 0.3554 0.3577 0.3599 0.3621
1.1 0.3643 0.3665 0.3686 0.3708 0.3729 0.3749 0.3770 0.3790 0.3810 0.3830
1.2 0.3849 0.3869 0.3888 0.3907 0.3925 0.3944 0.3962 0.3980 0.3997 0.4015
1.3 0.4032 0.4049 0.4066 0.4082 0.4099 0.4115 0.4131 0.4147 0.4162 0.4177
1.4 0.4192 0.4207 0.4222 0.4236 0.4251 0.4265 0.4279 0.4292 0.4306 0.4319
1.5 0.4332 0.4345 0.4357 0.4370 0.4382 0.4394 0.4406 0.4418 0.4429 0.4441
1.6 0.4452 0.4463 0.4474 0.4484 0.4495 0.4505 0.4515 0.4525 0.4535 0.4545
1.7 0.4554 0.4564 0.4573 0.4582 0.4591 0.4599 0.4608 0.4616 0.4625 0.4633
1.8 0.4641 0.4649 0.4656 0.4664 0.4671 0.4678 0.4686 0.4693 0.4699 0.4706
1.9 0.4713 0.4719 0.4726 0.4732 0.4738 0.4744 0.4750 0.4756 0.4761 0.4767
2.0 0.4772 0.4778 0.4783 0.4788 0.4793 0.4798 0.4803 0.4808 0.4812 0.4817
2.1 0.4821 0.4826 0.4830 0.4834 0.4838 0.4842 0.4846 0.4850 0.4854 0.4857
2.2 0.4861 0.4864 0.4868 0.4871 0.4875 0.4878 0.4881 0.4884 0.4887 0.4890
2.3 0.4893 0.4896 0.4898 0.4901 0.4904 0.4906 0.4909 0.4911 0.4913 0.4916
2.4 0.4918 0.4920 0.4922 0.4925 0.4927 0.4929 0.4931 0.4932 0.4934 0.4936
2.5 0.4938 0.4940 0.4941 0.4943 0.4945 0.4946 0.4948 0.4949 0.4951 0.4952
2.6 0.4953 0.4955 0.4956 0.4957 0.4959 0.4960 0.4961 0.4962 0.4963 0.4964
2.7 0.4965 0.4966 0.4967 0.4968 0.4969 0.4970 0.4971 0.4972 0.4973 0.4974
2.8 0.4974 0.4975 0.4976 0.4977 0.4977 0.4978 0.4979 0.4979 0.4980 0.4981
2.9 0.4981 0.4982 0.4982 0.4983 0.4984 0.4984 0.4985 0.4985 0.4986 0.4986
3.0 0.4987 0.4987 0.4987 0.4988 0.4988 0.4989 0.4989 0.4989 0.4990 0.4990

CONTENTS
113

Disclaimer.
Products and services referred to in this publication
are for identification and as a general example only.
No endorsement of any type is intended, nor is criticism
of similar products or services not mentioned. Persons
using products referred to in this publication assume
full responsibility for their use in accordance with label
directions provided by the manufacturer or supplier.

CONTENTS

F04CX14804-042004-500

Rue du Luxembourg 16-B • 1000 Brussels - Belgium
Tel: +32 2 548 93 80 • Fax: +32 2 502 68 66
Email: soyabru@coditel.net • www.asa-europe.org

CONTENTS

Manual of quality analysis - Soya products

More Related Content

What's hot (20)

Viewers also liked (20)

Similar to Manual of quality analysis - Soya products (20)

More from Vasu Chithiravelu (9)

Manual of quality analysis - Soya products