Principles

These multi-species tables are the result of a collaboration between INRA, CIRAD and AFZ. They are a follow-up of the Tables of composition and nutritional values of feed materials published by INRA and AFZ in 2002-2004.

Choice of feeds

The feed materials described in the tables are, in most cases, the ones used in industrial compound feeds. They were chosen according to their frequency of use, estimated from the French Feed Database. We added less common feed materials when reliable nutritional data were available for them.

Some readers may be surprised by the absence of forages in a book containing ruminant, horse and rabbit data. In fact, INRA has undertaken an in-depth review of the concepts and the values for forages, particularly for ruminants, and this work was not yet finished at the time of writing. When the review has been completed, a new edition of this book or specific tables will be published for forages.

Concerning the choice of feed names and categories (e.g. the different types of oil seed meals), we have tried to make sure that each category is representative of actual feed materials and sufficiently individualised from a nutritional point of view. For instance, the barley data refer to a generic barley grain, irrespective of the type (2-row or 6-row): while these two species have a slightly different composition, they are rarely identified as such in the feed market. Another example is the soybean meal, for which there are many different categories available commercially. These categories, often defined by their protein or protein + fat value, are marketed separately and can have specific nutritional properties. They can also change over time or even disappear. Therefore, we grouped these categories into three main soybean meal types that are both representative and nutritionally distinct.

However, the reader should keep in mind that table values, however well designed, are only indicative.

Chemical data

Origin

The chemical composition values have been collected by AFZ using its own database of feed information. Started in 1989, this database contains more than one million values of chemical, physical and nutritional characteristics of feed materials, most of them obtained from the participating laboratories (the list of companies and organisations is presented in the Acknowledgements section).

As often as possible, the chemical compositions were established from original and recent data that were accessible in sufficient numbers to allow critical evaluation. When more than 500 values were available, the averages were calculated for the samples collected (cereal grains, legume and oil seeds, French oil seed meals) or marketed starting from 1995. In addition, the data in the top or bottom 5% were removed from the calculations. A standard deviation was calculated when 5 or more values were available.

Literature values were used in the absence of original data, or in order to complete a data set. This was notably the case for vitamins and for part of the trace elements and fatty acids data.

A list of the tables used in order to produce this book is given in the references list at the end of this chapter. Due to lack of space, not all the references used are cited.

Consistency of composition profiles

For the same feed material, the number of available values can be very different for certain characteristics compared to others. For example, the crude protein content of soft wheat was calculated using more than 7000 original values while the number of results available for tryptophan was only 65. Also, there were 8 times more crude fibre values than ADF values available in our database. This may cause consistency issues when mean values are calculated using different numbers of original data per characteristic. In order to obtain coherent composition vectors, we have calculated regression equations using common characteristics (such as crude fibre) to predict less frequent ones (such as ADF). These equations are specific to a feed material or to a group of feed materials. We have also used equations from the scientific literature. A total of more than 2000 regression equations were established, of which more than half were used to perform adjustments. The crude protein content was usually the “pivot value” from which other values could be predicted in a cascade, depending on the availability of significant equations.

Nutritional value data

Basis

The starting point was to respect scrupulously the concepts of the evaluation systems used for different species. The nutritional values were calculated so that they were consistent with the values of chemical composition for a particular feed material. Although the calculations are described for each species in their respective chapters, the general method is common to all species. Firstly, individual data of nutritional value obtained in vivo were collected from INRA or from the literature. Secondly, relationships were identified between the nutritional values and the chemical composition of the samples. These relationships are applicable for well-defined families of feed materials or for larger groups of feed materials (for example, grains or seeds and their by-products). The choice of which feed materials were to be grouped together was systematically based on statistical analyses. We also used equations from other sources when they were considered reliable.

When relationships could be determined, components of nutritional value, such as the in vivo digestibility of the organic matter, were calculated by regression so that they corresponded with the average values of chemical characteristics given in the table. Many of these equations use cell wall constituents (crude fibre, NDF, ADF or ADL) as predictors, but other parameters were taken into account depending on the species, the feed materials and the characteristics to be predicted.

When pertinent equations were not available, the traditional approach based on average values was used. We then checked the consistency between the samples used in the corresponding trials and the chemical composition presented in the tables. Finally, in some rare cases, the data were taken from previous tables because no well-documented original data were available.

New or recent nutritional characteristics

Several new or recent nutritional characteristics of feed materials are introduced in these tables.

In the case of pigs, net energy content (growing pigs and sows), phosphorus availability and amino acid digestibilities of feed materials have been added. Amino acid digestibilities were derived (with modifications) from the AmiPig tables published in 2000 by different partners, amongst them INRA and AFZ. In the case of rabbits, the tables used the digestibility data (also with some modifications) and the concept of metabolisable energy published by Perez et al. (1998). The fish values were derived from the same sources used to establish the tables in Nutrition et Alimentation des poissons et crustacés (Guillaume et al., 1999).

In the case of ruminants, the tables now include the digestible amino acids in the intestine, calculated according to the method published by Rulquin et al. (1993 and 2001). In addition to updating the in sacco degradability values for nitrogen in the rumen and the intestine, we have indicated the kinetic parameters of in sacco degradation of dry matter and starch in the rumen. We have also included, when the data were available, the levels of absorbed phosphorus for ruminants.

Finally, for mineral sources, the tables present the summary of a literature study performed for the EMFEMA concerning the relative biological value of minerals and trace elements for pigs, poultry and ruminants.