Relationships between chemical composition and voluntary intake of feeds by sheep and cattle

Introduction

The Texas A&M University (TAMU) herd model (Sanders and Cartwright, 1979) and its derivatives, such as the ILCA model (Konandreas and Anderson, 1982; Konandreas et al 1983) estimate voluntary dry matter (DM) intake as being primarily a function of DM digestibility, down to the point at which protein content falls below 6%, when it becomes a function of feed protein percentage.

Predicted intake is subject also to either a physiological upper limit depending mainly on energy requirements at high values of digestibility (Conrad, 1966), an availability limit or a physical limit, the last being imposed on low-quality feeds by the apparent limitation to daily throughput of faecal DM of cattle. This has been estimated variously as about 1.07 kg/100 kg liveweight (Conrad et al, 1964) for dairy breeds in USA, or from 4.2 to 4.5 kg/100 kg metabolic weight for dry cows up to 4.9 kg/100 kg metabolic weight for lactating beef cows in Southern Africa (Elliott et al, 1961, as cited by Konandreas and Anderson, 1982).

Recent work by ILCA, using Ethiopian highland sheep and both highland Zebu and Friesian/Zebu cross cattle fed on local forages and crop residues, suggests that there is much variation in these relationships and that they may need to be re-examined.

Methods

Sheep experiments

sheep were fed in metabolism cages to measure voluntary intake and digestibility of four cereal-crop residues fed in conjunction with different proportions of a clover hay (Trifoliun tembense) (Mosi and Butterworth, 1985).

Cattle experiments

Zebu oxen and Friesian/Zebu cross cows, four animals per subgroup, were fed individually ad libitum (approximately 20% rejection), on a range of local hays and experimental forages for 10 days, after preperiods of 5–7 days. Digestibility was determined by total faecal collection.

Results and discussion

Sheep experiments

Feed DM intake expressed as g/kg ^0.75 or as g/head/d generally showed positive correlations with crude protein (CP) content and with DM digestibility, and negative correlations with neutral-detergent fibre (NDF) and acid-detergent fibre (ADF) content of the feed mixtures. Figure 1 illustrates the relationship between DM intake and the nitrogen (N) percentage (Figure 1 a) and the digestibility (Figure 1 b) of the feeds based on the four crop residues fed alone (left-hand points), with increasing proportions of clover hay, and the clover hay itself (right-hand point). Each data point is the mean for five sheep per feed over two periods of 10 days. Two extra points are shown, for similar sheep fed T. tembense harvested on different dates in earlier work (van Eeghen, 1984).

Figure 1. Relationships between (a) dry matter intake (DMl, g/ka ^0.75) and nitrogen concentration (N%) and (b) dry matter intake and dry matter digestibility (DM D%), for sheep fed different rations

DM intake was generally more closely correlated with N% than with digestibility and the negative correlations with NDF% were generally closer than with ADF% (Table 1). None of the correlations was significant with the oat/trifolium mixtures despite the apparently good agreement of the means in Figure 1. In the case of the maize stover/trifolium feeds, too, the correlation between intake and DM digestibility was not significant. Intakes of all the maize stover diets were significantly lower than the intakes of the other diets.

Table 1. Relationships between dry-matter intake (DMI) and nitrogen percentage (N%), neutral-detergent fibre (NDF), acid-detergent fibre (ADF) and dry matter digestibility (DMD%) for diets containing cereal crop residues and Trifolium tembense.

Cereal		r	R2	Significance
Wheat	DMI x N%	0.77	0:59	0.01
	DMI x NDF	–0.73	0.53	0.01
	DMI x ADF	–0.66	0.44	0.01
	DMI x DMD%	0.50	0.25	0.01
Oats	DMI x N%	–	–	NS
	DMI x NDF	–	–	NS
	DMI x ADF	–	–	NS
	DMI x DMD%	–	–	NS
Maize	DMI x N%	0.26	0.07	0.1
	DMI x NDF	–0.63	0.40	0.01
	DMI x ADF	0.57	0.26	0.01
	DMI x DMD%	0.15	0.02	NS
Teff	DMI x N%	0.77	0.60	0.01
	DMI x NDF	–0.58	0.34	0.01
	DMI x ADF	–0.24	0.06	NS
	DMI x DMD%	0.69	0.48	0.01

The results over all diets were in general agreement with the theory of Mertens (1973) as developed by Bywater (1984) that intake of NDF tends to be constant at about 40g/kg ^0.75 regardless of the NDF% in the feed. In this case NDF intake was more nearly constant within feeds than was faecal DM output, although there were also significant differences in NDF intake among the crop residues. Recent research in this laboratory (Reed, unpublished) indicates that in vitro digestibility is influenced considerably by the concentration of tannins in different parts of the sorghum plant, and it is possible that similar effects may account for the differences among other cereals reported here.

Cattle experiments

1. Figure 2 shows the results of an initial comparison of voluntary intake of four grass hays of diminishing quality fed to Zebu oxen and dry crossbred cows.

Figure 2. Dry matter intake (DM I) as a function of (a) nitrogen concentration (N%) and (b) dry matter digestibility (DM D%), for Friesian/Zebu crosses and Zebus

Expressed as g/ kg ^0.75, DM intake of the smaller (275–350 kg) Zebus was similar to that of the heavier (400–450 kg) crossbreds for each of the four hays. However, the Zebus gave 5–10% higher digestion coefficients on the poorer hays, so that the relationship in Figure 2 b indicates that their DM intake was lower in relation to DM digestibility than that of the crossbreds. This is misleading, since on a given feed DM intake by both Zebus and crossbreds was the same. The relationship between DM intake and N% of the hays over the range 0.5–1.25% N was more consistent between breeds than was its relationships to DM digestibility and would provide a better basis for prediction of feed intake.

2. Figure 3 shows the liner regressions for the crossbreds from Figure 2, without data points, and the means (each again for four cows over 10 days) from another experiment comparing a good grass hay with a Trifolium tembense/grass hay of much higher N% but similar (64–68%) digestibility (van Eeghen, 1984).

Figure 3. Dry matter intake (DM) as a function of (a) nitrogen concentration (N%) and (b) dry matter digestibility (DM D%) for cows fed Trifolium or grass hay (van Eeghen, 1984). Regression lines from Figure 2

Dry matter intake in relation to N% conformed very closely to the relationship for crossbreds from Figure 2, but DM intake of the trifolium hay was excessively high in relation to its modest digestibility. Thus, in both cases and over a range of N% frown 1.0 to 1.75%, N% proved a better predictor of intake than did digestibility.

3. The results of experiments with crossbred cows fed oat hay cut early, oat/vetch hay and a tall, fast-growing grass common in wetter parts of the Ethiopian highlands — Snowdenia polystacha — are shown in Figure 4, superimposed on the linear regressions from Figure 2 for crossbreds (Lambourne and Askabe, unpublished data). The digestibility of all these feeds was about 68 to 75%, but they contained only 1.0 to 1.2% N.

Figure 4. Dry matter intake (DM I) as a function of (a) nitrogen concentration (N%) and (b) dry matter digestibility (DM D%) for crossbred cows fed three feeds. Regression lines are from figure 2

In contrast to the data in Figure 3, DM intakes were very close to the original regression for crossbreds of intake on digestibility but the intakes of oat hay and S. polystacha were nearly 10 g/kg^0.75 higher than the original regression of intake on N%. The oat/vetch mixture, which had somewhat higher N%, fell 5 g/kg^0.75 below this line.

4. The data in Table 2 show the mean daily faecal DM output of each group of cattle on each of the nine feeds used in the trials mentioned above.

Table 2. Faecal DM output (% Lwt.) of Zebu and Friesian/Zebu crosses given a range of feeds.

Zebu and Friesian crosses
			Faecal DM output (% Lwt.)
	Feed:		1	2	3	4	all
Zebu			0.54	0.48	0.53	0.72	0.55
Friesian crosses			0.63	0.60	0.70	0.67	0.65*
Friesian crosses
Feed:
Snowdenia			0.57
Oats/vetch			0.68
Oat hay			0.69	0.65
Grass hay			0.66		**
T. tembense		(a)	0.90
		(b)	0.96	0.93

Feeds 1–4 were grass hays of diminishing quality.

* = significant at P < 0.05.

** = significant at P < 0.01.

There were significant differences (P < 0.05) in daily faecal DM output between the Zebus (0.55% Lwt.) and the crossbreds (0.65% Lwt.) over the four grass hays and among the hays within the zebus (0.48–0.72% Lwt., Table 2a). The difference between the daily faecal DM output of crossbreds fed Trifolium tembense (0.93% Lwt.) and those fed the other feeds (0.65% Lwt.) was highly significant (P < 0.01, Table 2b) . The values found in this study are lower than that found by Conrad et al (1964) of 1.07% Lwt. and that found by Kahn and Speeding (1984) for non-lactating cattle of 0.8–0.9%. The values for the Zebus are significantly lower for all the feeds studied than even the lowest of these values, which suggests that faecal DM output varies so much between closely related breeds and among a small sample of forages fed without supplements that its use as a constant to set limits to predicted DM intake should perhaps be reexamined.

General discussion

When a general simulation model is to be applied to local livestock breeds in a particular pastoral setting, it is desirable to check that the more important of the model's basic algorithm really apply in these new conditions. The results of this study suggest that for some native African livestock, and for at least one of the exotic/indigenous crosses being studied as a possible means of increasing production, the relationships derived from experiments in temperate production systems may need modification before they can be used with confidence (Wagenaar and Kontrohr, 1986). This is doubly important if the results of simulation modelling may be used as a guide for development and investment plans, rather than for purely academic purposes.

Specifically, it may be wise to reconsider present algorithms which place prime emphasis on feed digestibility, and to include feed protein (N) content as an equally-ranked second predictor over the full range of feed protein (N) values, rather than just at their lower end. This will become particularly important as forage legumes become more widely used to improve productivity of pastoral livestock systems, since intake of legume forage appears to be a function of their higher N content rather than of their normal DM digestibility.

It seems sensible also to study further the possibility of replacing daily faecal DM output by NDF intake in g/kg ^0.75 or some similar unit, as an upper limiting factor. Sanders and Cartwright (1979) considered using intake of indigestible DM in this way. Use of NDF would have the merit also of providing a third predictor based on chemical analysis of the feed, but inversely related to N% and digestibility. These last two factors both reflect the content of soluble, largely digestible, cell contents, and are often closely correlated. The NDF content, on the other hand, measures the generally indigestible structural carbohydrate material of the plant. The compensating nature of their relationships might well make the joint use of N%, digestibility and NDF intake a fairly robust approach.

Several authors have suggested that animal production is more closely related to the amount of green material available than to chemical composition of a whole pasture sample, and changes in nutritive value and voluntary intake with senescence of the green and dry fractions of a sward have been described and modelled (see Lambourne, 1986). The broader approach argued from the present results might help to embrace these ideas also— perhaps in time it may seen worthwhile to determine, say, chlorophyll content as a proxy for greenness and for carotene content and vitamin A adequacy. The widespread availability of oesophagal fistulation means that analyses can be carried out on samples of feed on offer or of feed eaten.

Acknowledgements

These experiments were carried out at the IICA HQ experimental animal barn and the assistance of Aklilu Askabe, Fikre Teku and their staff is gratefully acknowledged. Thanks are given also to Mebratu Ogbai and his laboratory staff for skilled analytical support, and to R Sayers for assistance with statistical analysis.

References

Bywater A C. 1984. A generalized model of feed intake and digestion in dairy cows. Agric. Systems 13:167.

Conrad H R. 1966. Physiological and physical factors limiting feed intake. J. Anim. Sci. 25:226.

Conrad H R, Pratt A D and Hibbs J W. 1964. Regulation of feed intake in dairy cows. 1. Change in importance of physical and physiological factors with increasing digestibility. J. Dairy Sci. 47:54.

van Eeghen M. 1984. Trifolium tembense: an exploratory study on its value as a fodder crop. Dissertation submitted to the Agricultural University of Wageningen in Tropical Animal Husbandry and Grassland Science. August 1984.

Elliott R C, Fokkema K and French C H. 1961. Herbage consumption studies on beef cattle. Rhodesian Agric. J. 58:124.

Kahn H E and Spedding C R W. 1984. A dynamic model for the simulation of cattle herd production systems. 2. An investigation of various factors influencing the voluntary intake of dry matter and the use of the model in their validation. Agric. Systems 13:63–82.

Konandreas P A and Anderson F M. 1982. Cattle herd dynamics: an integer and stochastic model for evaluating production alternatives. ILCA Research Report No.2 ILCA, Addis Ababa.

Konandreas P A, Anderson F M and Trail, J C M. 1983. Economic trade-offs between milk and meat production under various supplementation levels in Botswana. ILCA Research Report No. 10. ILCA, Addis Ababa.

Lambourne L J. 1986. ILCA's policy towards modelling in the framework of livestock systems research. In: N de Ridder, H van Keulen, N G Seligman and P J H Neater eds. Modelling of extensive livestock production systems. Proceedings of the ILCA/ARO/CABO workshop, 5–9 Feb 1985. Bet Dagan, Israel. ILCA, Addis Ababa.

Mertens D R. 1973. Application of theoretical mathematical models to cell wall digestion and forage intake in ruminants. Ph.D. thesis, Cornell Univ., Ithaca, USA.

Mosi A K and Butterworth M H. 1985. The voluntary intake and digestibility of combinations of cereal crop residues and legume hay for sheep. Anim. Feed Sci. Technol. (in press).

Sanders J 0 and Cartwright T C. 1979. A general cattle production systems model. Part II — Procedures used for simulating animal performance. Agric.Systems 4:289.

Wagenaar K T and Kontrohr E. 1986. Appraisal of the ILCA cattle herd dynamics model using data from pastoral systems in Mali and Kenya. In: N de Ridder, H van Keulen, N G Seligman and P J H Neate, eds. Modelling of extensive livestock production systems. Proceedings of the ILCA/ ARO/CABO workshop, 5–9 Feb 1985, Bet Dagan, Israel. ILCA, Addis Ababa.

Discussion

Comment – The digestive capacity of the Zebus is, I understand, about 15% higher than that of the Friesian crosses.

Reply – The difference between them was more noticeable on feed of lower quality; on feed of 70% digestibility, the digestion coefficient was the same for the two breeds. But where the Zebu consistently gave figures of around 60 or 62%, the digestibility with Friesians dropped back to the low fifties.

Question – That would be of advantage to the Zebus, but on the other- hand, with equal intake and limited faecal output, would the two factors cancel each other out?

Answer – On those same feeds, the intake of the Zebu was a little lower than that of the Friesians. Here we have two confounded effects; was the digestibility higher because the intake was lower, or did they digest it better and therefore need to take in less? We think that one of the reasons that the correlation between dry matter intake and dry matter digestibility was not as close as expected was the way the experiment was done. The animals were fed a fixed amount of Trifolium and then fed teff, wheat, oats, or maize ad libitum. Having eaten a fixed amount of Trifolium, an animal which eats more teff straw has a higher intake, which will automatically have a slightly lower digestion coefficient, because of the smaller proportion of highly digestible Trifolium in its intake. So that among the animals in the trial a negative relationship is obtained between dry matter intake and digestibility. Between means there is undoubtedly a positive relation: higher digestibility, higher intake.

Question – What were the weights of the cattle used in those experiments?

Answer – There was quite a difference between them. The Zebu weighed on average between 270 and about 330 kg and the Friesian crosses were normally 350 to 450 kg.

Comment – My impression from the literature is that differences in digestive capacity between Zebus and Friesians, or rather between Bos indicus and Bos taurus are not very large, unless you compare animals which differ substantially in weight. The heavier animals have, in general, higher digestive capacity.

Comment – What strikes me in the discussion we have just heard is its statistical naivete. Not only are the two variables, digestibility and nitrogen content, highly correlated, which makes it very difficult to use one as a predictor of the other, but there are probably block effects and interactions between other independent variables and the experimental animals. With a very small sample size, you would rapidly reduce the confidence with which you can make inferences from this kind of data.

You are attempting to use some variables to predict intake, digestibility and nitrogen. Now, it does not make too much difference which is more important. What you are really interested in is the prediction of the dependent variable, intake, and you are not so much interested in assigning it to the one or the other. The question is then, how much better is the prediction of the dependent variable? The methods presented here do not tell you that.

Reply – Well, there is a difference of course, between statistical regression models and what Jan Ketelaars has been trying to explain, which is why differences in intake take place, rather than actually using those relations for intake predictions. That is the ultimate goal of course.

At the moment the weakness in the intake formulation of the ILCA model and the TAMU model is that nitrogen percentage comes in only at a lower limit, when in fact it is closely correlated with dry-matter intake over the whole range of nitrogen contents. I think, therefore, that it should not come in simply as a lower limit. We should use both digestibility arid nitrogen content, and perhaps NDF content, over the full range of intake. Your criticism is perfectly valid, but I do not have enough data for a sophisticated analysis. It does seen to me that there are at least three or four variables which influence dry-matter intake over the full range of each variable, and that it is not a hierarchy in which we have one variable over a certain range and others over a different range.

Comment – A strong correlation between digestibility and protein only exists if you look at a range of feeds at a given location. But if you extend the sample to more feeds, frown different sites, then the correlation is very low, as you can see in some of the grasses.

Reply – That probably just means you have to exclude variables, because the data are based on research from experiments done in different countries and different years. The fact that the correlation breaks down when you use data from different countries, different breeds, different experimenters, and different years, is not surprising. There are differences in the ways the experiments were conducted and one has to be very careful when making such inferences using data collected in different ways.

Comment– Yes, but even if you look at a very homogeneous sample like the experiments done by French workers, you will find a large range of protein concentrations within each digestibility level.

Reply – From our rather narrow view point, which is restricted to animals on the range of pastures in African grazing system, there is probably a very close correlation between nitrogen content and digestibility. Granted, this relationship breaks down when you look at cattle fed on straw, citrus waste, brewers' grains, and chicken manure, but we don't have that many options in Africa. For the broad class of forages, I think the relationship is pretty close. My example shows that you can get sane feeds which have an abnormally high intake in relation to either their protein content or their digestibility. I would therefore prefer to have a predictor which includes both those variables, and a fibre component as well, because if we rely only on one of them in an individual feed we can be out by 20– 30%. I think a mixture of the three would give us a greater robustness.

Comment – I tried to see if, within a given digestibility class, there were species characteristics related to high or low intake. I couldn't find any species characteristic or any species differences which lead to higher or lower intakes that could not already be derived from digestibility and nitrogen content alone.

Comment – Could I just suggest that the problem is not necessarily just a question of running a multiple regression package? The objective of Fetelaars' paper is to arrive at a new structural model. That should be the first step, and then the statistical analysis should be used to test whether that new structural model gives better fit than the simpler models.

Comment – You can expect that including protein concentration in addition to digestibility will certainly improve predictive power.

Comment– I still feel that digestibility per se is a very poor independent variable because of the inherent errors in its determination.

Dr Alms Goldman of ARO, Bet Degan, received about 450 feed samples from all over the world for which there were data for in vivo digestibility. He did a detailed multiple regression analysis of the data and concluded that in vitro digestibility gave a more reliable estimate than in vivo determination, because a very large part of the variation was animal variation. If you really want to blow the digestibility of a feed, do not put it into an animal, because it will just make a mess of it!

Reply – What we need to know, for predictive purposes, is the digestibility of the feed which the animal is given. So I think we have to rely on some sort of in vitro procedure or a chemical determination, or a bio-assay of the intrinsic features of the pasture. We use a double enzyme assay, because it is quick and easy It is not as precise as the nitrogen, but it is probably better than an in vivo assay with all of its inherent errors. `