Nutritional value to farm livestock of wheat of low specific weight




Six million tonnes of wheat per year are used in animal feeds in the UK, making the

livestock feed industry the largest national user of wheat. Of this tonnage more than 50% is used for poultry, almost 30% for pigs and less than 20% for ruminants.

Specific weight of wheat is the trading standard used by the feed industry to determine purchase price and is therefore assumed to be an indicator of nutritive value. However evidence suggests that specific weight is not an appropriate measure of wheat quality for animal feed. Work investigating the effects of low specific weight wheat on its nutritive value for livestock is limited and generally confounded with variety. In a review of the literature on the relationship between specific weight and nutritional value of wheat when used as a livestock feed, Miller and Wilkinson (1998) concluded:

Specific weight appears to be a poor measure of wheat quality for animal feed.
The results of previous studies were often compromised because wheat samples had been selected on the basis of specific weight without regard for variety
Further studies were required to develop a better predictor of nutritional value.

The main objective of this work was therefore to establish whether specific weight of wheat does indicate nutritive value by conducting a comprehensive series of trials at four research centres utilising three species of farm livestock.

When wheat was traded by volume, specific weight was a sensible measure on which to base the price since it reflected the weight of grain that was being purchased, indicating that our ancestors realised that it was the weight of grain that mattered! It was logical to pay less for a lighter bushel weight. Somewhat surprisingly, now that grain is traded on a weight basis this logic has become reinterpreted such that relative density of wheat is assumed to have merit as a measure of quality in its own right. One reason that specific weight has persisted as a trading standard for feed wheat is because, so far, we have failed to find an adequate rapid assessment of nutritive value that can be made at point of purchase which would form a justifiable basis for payment. Therefore the second objective of this work was to try to identify such a technique.

In order to meet these objectives we aimed to:

Collect wheat samples of known variety by a range of specific weights
Measure the physical and chemical components of these wheats
Measure animal performance in a range of species when fed these wheats
Relate wheat composition and specific weight to variation in nutritive value as indicated by animal performance
Develop a rapid technique to predict animal performance.
Key criteria used in experimentation

This project exceeded the scope of previous work in this area by meeting the following key criteria:

Wheat variety of all samples used in the trials was confirmed by electrophoresis performed by the National Institute of Agricultural Botany (NIAB).
Wheat was incorporated in all the diets at a minimum of 65%. This level was chosen to maximise the effect of the wheat in the diet whilst remaining within the extremes of commercial practice.
Within each set of trials identical ingredient inclusion levels were utilised and, with the exception of the test wheats, the same batches of ingredients were used.
Diets were formulated to have marginally excess protein content and were balanced for lysine, methionine, threonine and tryptophan so that protein inadequacy could not be a factor affecting measured nutritional value (does not apply to Chapters 3 or 4).
All diets were prepared at the same mill, namely Roslin Institute, thus ensuring that all diets were prepared to the same standards. This was particularly important for the poultry trials which were conducted at more than one site but using identical diets.
The main matrix of 16 wheat samples were tested across a range of species, namely broiler chickens, pigs and sheep. In addition a further 46 wheat samples were used in broiler trials (Chapters 3, 4 and 5).
Experimentation occurred at 4 research centres.
Animal performance was measured as the key indicator of nutritive value.
The wheat samples were tested under commercial conditions in all species.
Statistical analysis and interpretation was independently confirmed by Biomathematics and Statistics Scotland (BioSS).
Organisation of trial work

The research reported here was a co-operative programme of work conducted by four institutions: The University of Leeds Centre for Animal Sciences, Queen's University Belfast, The Roslin Institute and Harper Adams University College. Each group was responsible for separate areas of activity as follows:

Leeds University - Performance trials with grower pigs, determination of digestibility and DE.

Assessment of digestibility and estimated ME in sheep.

Queen's University- Cage trials with broiler chickens, determination of AME and performance.

Roslin Institute - TMEn and amino acid availability assessment in cockerels.

Harper Adams - Cage trials and larger scale floor trials to assess broiler chicken performance.

Collection of samples

It proved to be surprisingly difficult to collect a suitable matrix of wheat samples for use in this project. We had originally hoped to use thirty samples comprising 6 varieties by 5 specific weights, however in the end we had to be content with 16 samples; 4 varieties by 4 specific weights, of which 3 samples were generated by gravity separation. These sixteen samples were used for trials in all test species and at all centres.

We actually collected 85 wheat samples in total but we took the precaution of confirming both variety and specific weight after purchase and discovered that by far the majority of wheat samples we had procured had been incorrectly described. The most common problems were:

Specific weight was incorrect, in particular initial assessments of specific weight on the farm of origin were frequently much lower than subsequent determinations in the store or laboratory.
Variety was either completely different to that stated or the wheat sample proved to be a blend of two or more varieties.
These problems in obtaining correctly described wheat throw doubt on the validity of studies in which specific weights and variety have not been confirmed.

The 16 commercially sourced wheat samples used in all trials were augmented by an additional 46 wheat samples supplied by the Northern Ireland Department of Agriculture and Rural Development. These samples had been grown in Recommended List trials and therefore only sufficient wheat was available to run evaluation trials with poultry and not the other two animal species.

Effect of specific weight on chemical composition

Starch concentration increased with increasing specific weight of wheat, whilst modified acid detergent fibre (MADF) decreased thus indicating a shift in the composition of the complex carbohydrate fraction of the grain with changing specific weight. Such changes in composition have been reported elsewhere (see Chapter 2). There was a trend for crude protein content to decrease with increasing specific weight. In the matrix of 16 wheats used across the whole study there was no change in fat content with changing specific weight.

Effect of specific weight on energy value

TMEn (Poultry) - There was a small but significant fall in TMEn value with decreasing specific weight (r2= 0.31). When Haven 66 was excluded from the analysis the correlation coefficient for the whole data set was improved (r2= 0.64). The regression relationship so derived indicated that for every 10 kg/hl increase in specific weight there should be a 0.3 MJ/kg increase in TMEn of the wheat. This is likely to be of limited commercial significance. It is interesting to note that Haven 66 (the third lowest specific weight) had the highest TMEn value.

AME (Poultry) - In the first study, utilising non-commercial high wheat and casein diets, (Chapter 3) there was a weak positive relationship between wheat AME and SW (r2 = 0.16). In the second study the weak linear relationship was negative but there was a significant quadratic relationship between wheat AME and specific weight (r2= 0.31) associated with low values for a few of the intermediate specific weight samples!

DE (Pigs) - There was a significant, though extremely weak (r2= 0.06), negative correlation between DE and specific weight. For every 10kg/hl increase in specific weight there was a 0.2 MJ/kg decrease in DE.

ME (Sheep) - There was no effect of specific weight on ME value.

Across all experiments it was evident that there was no consistent effect of specific weight on energy value.

Effect of specific weight on animal performance

In all trials animal performance was normal for the appropriate facility. In fact Belfast reported exceptionally high growth rates in their caged broiler trials (Chapter 5).

Although there were some changes in chemical composition with changing specific weight of wheat this was not reflected in animal performance. Indeed the low specific weight wheats performed just as well as the other wheats and there were no consistent effects of specific weight on animal performance in any of the three species studied. It was quite clear that performance differences between individual wheats were far greater than those between different specific weights. This finding emphasises the need for a good measurable indicator of nutritive value of wheat since specific weight clearly is not.

Alternative methods of rapid assessment

Unfortunately the extensive work performed here was unable to yield a rapid method of assessment of the nutritive value of wheat. NIR was considered to be the most encouraging method investigated but although reasonable calibrations were obtained these proved ineffective in predicting values for wheats not used in the initial calibration. In vitro viscosity was also a possible candidate but was only correlated with performance in certain studies and was not able to predict performance across studies. Starch content was reasonably correlated with specific weight but not with performance, illustrating the inadequacy of either measure as an indicator of nutritive value.

In addition to measuring specific weight we also performed two other assessments of wheat density, pour density and tap density. We perceived the advantages with these measures would be that they would remove any differences in density relating purely to unusual morphology of grains. They might therefore be better correlated to grain composition and hence nutritive value. However since no differences in grain composition were correlated to animal performance it is unsurprising that these measures were equally unrelated to performance.


Specific weight of wheat does not indicate its nutritive value.
Currently no rapid technique has been identified for assessing nutritive value of wheat.

Cereals & Oilseeds
Project code:
01 January 1999 - 31 December 2001
AHDB Cereals & Oilseeds.
AHDB sector cost:
£300,000 from HGCA (project no. 2073).
Project leader:
HELEN M MILLER1, J M WILKINSON1, K J McCRACKEN2, ANNE KNOX3, J McNAB3 and P ROSE4 1The University of Leeds, Woodhouse Lane, Leeds, West Yorkshire LS2 9JT 2The Queen's University, Newforge Lane, Belfast, County Antrim BT9 5PX 3The Roslin Institute, Roslin, Midlothian EH25 9PS 4Harper Adams University College, Edgemond, Newport, Shropshire TF10 8NB



About this project


The primary aim of this project was to determine whether specific weight of wheat indicates its nutritive value to farm livestock. To achieve this a matrix of 4 known varieties of wheat each at 4 specific weights was fed to poultry, pigs and sheep in a series of feeding trials at 4 research centres: University of Leeds, Queen's University Belfast, the Roslin Institute and Harper Adams University College. Specific weights ranged between 60 and 78 kg/hl. In addition a further 46 wheat samples, comprising known varieties each at least 2 specific weights, were tested in poultry.

Parameters measured included true metabolisable energy (TME), apparent metabolisable energy (AME), amino acid availability, growth rate and feed conversion ratio in poultry; digestible energy (DE), digestibility, growth rate and feed conversion ratio in growing pigs and metabolisable energy (ME) and digestibility in adult sheep.

There was a significant regression relationship between specific weight and TME measured in adult cockerels; however this amounted to 0.3 MJ/kg per 10 points increase in specific weight and is therefore of no commercial significance. In contrast linear relationships between AME (broiler chickens) or DE (grower pigs) and specific weight were only just significant and were negative. There was no relationship between ME measured in sheep and specific weight. There were no differences in animal performance due to wheat specific weight in any of the trials conducted. We concluded that specific weight, at least between 60 and 78 kg/hl, does not indicate the nutritive value of wheat.

The second aim of the project was to determine a characteristic of wheat which could be used to predict its nutritional value. With this end in view all wheats used in the trials were extensively analysed for physical and compositional traits. These traits were then correlated against animal performance. Despite the comprehensive range of analyses performed no factor of the wheat was consistently found to correlate with animal performance. The best results were found for near infrared reflectance spectroscopy but the equations derived were not effective in predicting the nutritive value of wheats not used in their derivation.