A comparison of in vitro rumen fluid and enzymatic
methods to predict digestibility and energy
value of grass and alfalfa hay
N. Iantcheva
a,*, H. Steingass
b, N. Todorov
a, D. Pavlov
a aDepartment of Animal Nutrition, Thracian University, 6000 Stara Zagora, BulgariabInstitute for Animal Nutrition, Hohenheim University, 70599 Stuttgart, Germany
Received 23 September 1997; received in revised form 3 July 1998; accepted 23 February 1999
Abstract
Relationships between in vivo organic matter digestibility (DOM, %) or metabolizable energy (ME MJ/kg OM and different laboratory measurements have been calculated for 22 samples of grass hay and 20 samples of alfalfa hay. Laboratory measurements included Weende constituents (crude protein, CP; crude fiber, CF; ether extract, EE; and ash), pepsin±cellulase digestible organic matter (CDOM, %), gas production (GP, ml/24 h) and a combination of in vitro rumen fluid fermentation in 100 ml glass syringes and determination of the undigested neutral detergent residue. Digestible dry matter (DDM, %) was calculated from dry matter quantity of the feed minus total residue after incubation. On the basis of the quantity of organic matter (OM) in the original samples and the NDF residue true digestible OM (TDOM, %) was calculated. All in vitro procedures are significantly correlated to in vivo DOM and ME. Prediction of energy value of feeds using gas production alone and DDM is less accurate compared to other methods. Combination of GP with CP, NDF, ADF and ash improves prediction accuracy. Rumen fluid±neutral detergent method (TDOM) and cellulase method (CDOM) predict OM digestibility and energy value of hay with similar accuracy. The best regression equations for prediction of ME (MJ) obtained are the following:
ME MJ=kg OM 7:40:07TDOMÿ0:004NDF0:02EE
R0:92;SEE0:25;n42
ME MJ=kg OM 12:3ÿ2001=CDOMÿ0:003ADF0:01EE0:002NDF
R0:91;SEE0:27;n42
Using glass syringes instead of the centrifuge tubes used in the method of Tilley and Terry [Tilley, J.M.A., Terry, R.A., 1963. A two-stage technique for the in vitro digestion of forage crops. J. Br. Grassl. Soc. 18, pp. 104±111] simplifies the procedure from an operative point of view. The
81 (1999) 333±344
* Corresponding author. Tel.: +359-42-76186; fax: +359-42-56102
E-mail address:[email protected] (N. Iantcheva)
combination of the two methods (gas production and two-stage method with rumen fluid) could give information either of digestion kinetics or undigested products.#1999 Elsevier Science B.V. All rights reserved.
Keywords: Digestibility; Energy value; In vitro technique; Gas production; Enzymatic methods; Hay
1. Introduction
The nutritive value of hay varies considerably, especially when harvested at different stages of maturity. Therefore, the prediction of the quality of roughages is important for the prediction of animal performance.
The in vivo measurement of digestibility as a basis for the calculation of nutritive value requires animals, relatively large quantities of test feed and time. These factors limit its use and several attempts have been made to develop simple techniques for predicting in vivo digestibility of organic matter.
There are several laboratory procedures used to predict organic matter (OM) digestibility or energy value of feedstuffs. These methods have advantages because they are rapid and inexpensive.
Chemical constituents such as crude fibre and some cell wall fractions were used in the prediction of the nutritive value of forages, but they are not accurate enough (Aerts et al., 1977; De Boever et al., 1986; Andrighetto et al., 1992). Another possibility is in vitro techniques based on incubation of forages with rumen fluid (Tilley and Terry, 1963; Goering and van Soest, 1970; Moore, 1970; Troelsen, 1970; Coelcho et al., 1988). There are indications that gas production from incubation of forages with rumen liquor can predict the digestibility and energy value of a wide range of feeds (Menke et al., 1979; Menke and Steingass, 1988). This method is basically similar to the Tilley and Terry method, but measures the amount of fermented substrate instead of dry matter loss. Methods using rumen fluid are accurate, but cumbersome, only moderately reproducible and require cannulated animals. Another drawback is their sensitivity to the diet of the donor animals. These disadvantages can be avoided by the in vitro estimation of the OM digestibility using cell-free cellulase-type enzymes (Aufrere, 1982; Dowman and Collins, 1982; De Boever et al., 1986; Aufrere and Michalet-Doreau, 1988; Cottyn et al., 1993). The objective of our study was to compare different laboratory methods to predict in vivo digestibility and energy value of forages and to test the potential of combining the gas production technique with neutral detergent treatment in order to improve predicting accuracy.
2. Material and methods
2.1. Forage samples and chemical analysis
The investigation was carried out with 20 alfalfa hays and 22 grass hays. The samples were collected from four research institutes in Bulgaria, mainly from the Department of
Animal Nutrition of the Thracian University, Stara Zagora. All hays were field-dried. Alfalfa hays were prepared with an experimental purpose from early vegetation (approximately 20 cm height of the plant) until the full bloom stage, four of them got rain during wilting. Predominant forage species of the grass hays were Phleum pratense, Agrostis alba, Dactylis glomerata, Bromus intermis, Festuca rubraand small percentages of legumes such asLotus corniculatus, Trifolium pratenseandMedicago sativa.
Forage and faecal samples were analysed for crude protein (CP), ether extracts (EE), crude fiber (CF) and ash by the Weende methods as described by AOAC (1980). Cell walls (NDF and ADF) were analyzed according to van Soest et al. (1991).
2.2. In vivo measurements and calculation of the energy value of forages
In vivo digestibility of each forage was determined with four mature wethers, 2±5 years old. The animals were fed at near maintenance with forage as the sole feedstuff, supplemented with the required minerals and vitamins. Digestibility measurements were done for 7 days after a preliminary period of 10 days. During the experimental period total faeces were collected and sampled for later chemical analyses. Digestible organic matter (DOM) was expressed in percent of the OM content.
Metabolizable energy (ME) was calculated according to the energy system, introduced in Bulgaria by Todorov (1995), which is similar to the one proposed by van Es (1978).
2.3. In vitro rumen fluid techniques
Rumen liquor was obtained from two lactating ruminally fistulated dairy cows, maintained on a 60% good quality grass hay and 40% concentrate diet according to their requirements. It was collected from the ventral sac of the rumen before morning feeding. The strained rumen fluid was mixed with the buffer medium in a ratio of 1 : 2 and continuously flushed with CO2. The rumen fluid preparation procedure and the
composition of the buffer solution were as described by Menke and Steingass (1988). All incubations were completed in 100 ml calibrated glass syringes. Six syringes from each sample were incubated in two runs. The first run was carried out with approximately 200 mg forage according to Menke and Steingass (1988) and gas production was recorded at 2, 4, 6, 8, 12, 24, 32, 48, 72 and 96 h. The weight of the incubated samples in the second run was about 500 mg. This quantity was chosen to decrease analytical error but at the same time to avoid production of more than 90 ml gas, as suggested by Steingass (1983). Gas production was recorded at 2, 4, 6, 8, 12, 24, 32 and 48 h. After 48 h the contents of the syringes were transferred into glass crucibles (Porosity 1), washed with distilled water and dried. Indigestible DM was calculated at 48 h by dividing residual DM by sample's DM. Subtracting the percentage of indigestible DM from 100, equalled the digestible dry matter (DDM, %). Residues were then treated with neutral detergent solution for 1 h as described by Goering and van Soest (1970), following the procedure suggested by BluÈmmel et al. (1997). After drying and weighing, samples were ashed at 5508C for 3 h. In each run corrections were made for blank, forage-free incubations. On the basis of the quantity of OM in the original samples and the residue of
OM true digestibility of OM (TDOM, %) was calculated. Gas production (GP) from all parallels was recalculated to 200 mg DM. Correction with standards (hay and concentrate) was made only for gas production at 24 h.
2.4. In vitro enzymatic technique
The cellulase-type used was Onozuka R-10 extracted fromTrichoderma viride(Merck, Darmstadt, Germany). All procedures were carried out according to De Boever et al. (1986, 1988). The results are expressed as the cellulase digestibility of the OM (CDOM, %).
2.5. Statistical analyses
The computer programmeSTATISTICA FOR WINDOWS (Release 4.3, Stat. Soft Inc., 1994)
was used for the regression procedures. The coefficients of correlation (r) and the standard error of the estimate were calculated as they indicate the strength of the association and the accuracy of the equations, respectively. Simple and forward stepwise multiple regressions were performed between DOM or ME and Weende constituents, GP24 (ml/24 h), GP48 (ml/48 h), DDM, TDOM or CDOM.
3. Results
3.1. Database
Table 1 lists mean values, ranges and standard deviations for chemical composition, energy value, in vivo and in vitro digestibility of the hays. The nutrient composition and energy value of hay are within the range of values reported for similar feedstuffs (Todorov, 1995). Moreover, the variation in all data is considered to be large enough to calculate relationships between the parameters.
Mean in vivo DOM (%) is quite similar for alfalfa and grass hay. There is a larger range of about 20% units in DOM for alfalfa hay because the four samples, which were rained on during field-drying were included. The range for CDOM, TDOM, and first-stage ruminal DDM for alfalfa hay are also substantial. The ranges and standard deviations of DOM, CDOM, TDOM, and DDM for grass hays are relatively lower. Gas production from 200 mg DM, corrected only with blank is shown in Fig. 1. There is tendency for faster fermentation of alfalfa hay during the first 24 h of incubation compared to grass hay. However, fermentation of alfalfa hay was almost complete at 48 h and was very slow thereafter while fermentation of grass hay continued up to 96 h. Total gas production at 96 h was significantly higher for grass compared to alfalfa hay.
Mean values of TDOM agreed well with mean in vivo values within both forage groups, whereas mean CDOM values only agree well with mean in vivo values for alfalfa hay. This seems consistent with De Boever et al. (1986), who reported that cellulase digestibility of OM for good quality forages and concentrates was higher than in vivo values. The opposite is true for forages with in vivo DOM less than about 70%, and the
difference increases with decreasing quality. In spite of having the same in vivo digestibility, pepsin±cellulase solubilised more DM from legumes than from grasses, because grasses contain more cell walls and crude fiber (Aufrere, 1982).
Table 1
Chemical composition, energy value and digestibility of the hays
Items Alfalfa hay (n20) Grass hay (n22)
Mean Range SDc Mean Range SDc
Chemical composition (g kgÿ1DM)
Crude protein (CP) 182 126±242 37 96a 56±160 32
Ether extracts (EE) 19 11±27 5 18 10±33 6
Crude fiber (CF) 321 151±436 60 341 292±393 27
Neutral detergent fiber (NDF) 692 585±803 53 793a 756±882 37
Acid detergent fiber (ADF) 418 200±580 78 468a 439±587 37
Ash 93 65±107 18 62a 41±95 16
Metabolizable energy (ME, MJ)
ME/kg DMb 8.31 6.79±9.09 0.53 8.11 7.36±9.00 0.45
ME/kg OMb 9.17 7.5±10.59 0.62 8.65 7.92±9.50 0.50
In vivo digestibility of OM,% (DOM) 60.4 49.2±69.0 3.9 58.0 51.6±63.3 3.52
In vitro digestibility (%) of:
DM ruminal first stage (DDM) 47.1 29.9±57.8 6.3 52.9a 44.1±63.4 4.6
OM ruminalND solution (TDOM) 60.4 50.2±76.4 4.9 59.9 54.2±68.7 4.6
OM cellulase (CDOM) 60.9 42.7±72.7 6.9 51.1a 43.1±59.1 4.7
Gas production (GP24-ml/200 mg DM/24 h) 33.9 25.8±39.0 3.8 34.9 31.2±40.5 3.1
Gas production (GP48-ml/200 mg DM/48 h) 39.2 31.1±47.7 4.8 43.0 40.4-47.9 1.8
aDifference between means for alfalfa hay and grass hay is significant atp< 0.05. bCalculated on the basis of in vivo digestibility.
cSD, standard deviation.
Fig. 1. Mean gas production over 96 h incubation of 200 mg DM of hay (ml).
3.2. Prediction of digestibility and energy value with simple regression
The simple regression statistics for the relationships between in vivo DOM or ME and the various laboratory measurements are shown in Table 2. Relationships for Weende constituents are significant for crude fiber and crude protein for alfalfa hay and crude protein for grass hay. Cell walls improve prediction accuracy, especially in the case of grass hay. All in vitro procedures correlate significantly with in vivo data. However, the coefficients of correlation are highest, and standard error of the estimate lowest and almost equal, for TDOM and CDOM. Correlation of DOM or ME with GP or DDM gives a higher standard error of the estimate compared to other methods. Square and cubic values of the variables (data not presented) could not improve neitherRnor SEE in the prediction of DOM, and ME, which agree with the results of Menke and Steingass (1988). A slight improvement in prediction of digestibility and energy value was obtained with the reciprocal value of CDOM in the case of alfalfa hay.
Digestibility and energy values of grass hay correlate best with GP, compared to alfalfa hay, but the statistical data do not reach the accuracy reported by Menke and Steingass (1988). Relationships between gas production at different incubation times and in vivo DOM are in Fig. 2. The correlation coefficients are lowest at 8 h for alfalfa hay and at 12 h for grass hay, highest at 24 h for both hays, and similar thereafter.
Table 2
Prediction of in vivo dry matter digestibility (DOM,%) and ME (MJ/kg OM) of alfalfa hay (n20) and grass hay (n22) with simple regression
Variables Alfalfa hay Grass hay
1/GP24, 1/TDOM, 1/CDOMÐreciprocal function of GP, TDOM, CDOM.
r, coefficient of correlation. SEE, standard error of estimate.
aGP24 and GP48, gas production (ml) for 24 and 48 h, respectively.
For other abbreviations see Table 1.
*p< 0.05;**p< 0.01;***p< 0.001.
3.3. Prediction of digestibility and energy value by forward stepwise multiple regression
The best regression equations were obtained using forward stepwise multiple regressions between DOM and ME, as dependent variables and a combination of the chemical constituents, plus some of the in vitro procedures, as independent variables (Tables 3 and 4).
Including crude protein and NDF in the case of alfalfa hay and ADF and ash in the case of grass hay increased the prediction accuracy of the gas production method. Fig. 2. Correlation coefficient (r) of the relationship between gas production after different incubation periods and in vivo DOM.
Table 3
Regression equations for prediction of in vivo DOM (%) and ME (MJ/kg OM) of alfalfa hay (n20)
Independent variables
Dependent variable (y)
DOM ME
Regression coefficients for independent variables
Intercept 82.7 66.4 84.0 32.9 29.6 78.5 13.1 10.0 6.4 5.8 12.5
CP 0.05 0.05 0.05 0.02 0.03 0.02 0.006 0.007 0.004 0.002 0.003
CF 0.03 ÿ0.003
NDF ÿ0.05 ÿ0.03 ÿ0.05 ÿ0.03 ÿ0.006 ÿ0.005ÿ0.004
ADF ÿ0.002 ÿ0.001 ÿ0.002
EE 0.01 0.02 0.02
Ash ÿ0.06 0.03 ÿ0.01
GP24 0.23 0.04
DDM ÿ0.02
TDOM 0.68 0.09
CDOM 0.42 0.06
1/CDOM ÿ1439 ÿ199
r 0.83 0.84 0.83 0.92 0.89 0.91 0.86 0.87 0.93 0.90 0.92
SEE 2.31 2.28 2.38 1.81 1.89 1.71 0.348 0.337 0.265 0.305 0.274
r, coefficient of correlation. SEE, standard error of estimate. For other abbreviations see Table 1.
Improvement in prediction accuracy of the other in vitro methods could also be achieved by the inclusion of crude protein and cell walls and to some extent the other Weende constituents. The mean deviation between the measured values of alfalfa hay DOM and those estimated by TDOM, NDF, CP, CF and ash is 1.81 (Table 3) or expressed as a percent of the mean DOM (1.81/60.41) is 3%. With similar accuracy TDOM predicts ME of alfalfa hay. SEE of ME using TDOM, NDF, CP, EE, Ash in percent of mean ME (coefficient of variation, CV) is 2.9%. Equations using CDOMCP and CDOM combined with CP, ADF and EE predict DOM or ME with coefficients of variation, respectively, 3.1 and 3.3%.
TDOM or CDOM plus chemical constituents (Table 4) predict DOM of grass hay with SEE 1.43 and 1.53 or expressed in percent of the mean DOM as 2.5 and 2.6%, respectively. Almost the same precision was obtained for prediction of ME of grass hay. Figs. 3 and 4 show plots of actual versus predicted ME (MJ) of all hays by the best regression equations using TDOM or the reciprocal value of CDOM plus chemical constituents, respectively.
4. Discussion
4.1. Chemical constituents as a predictors
Our study confirms the results obtained from other references (Aerts et al., 1977; De Boever et al., 1986; Andrighetto et al., 1992) that proximate Weende crude nutrients have poor capacity to predict in vivo digestibility and energy value of forages. The equations Table 4
Regression equations for prediction of in vivo DOM (%) and ME (MJ/kg OM) of grass hay (n22)
Independent variables
Dependent variable (y)
DOM ME
Regression coefficients for independent variables
Intercept 71.5 49.9 65.5 32.6 24.7 11.2 9.7 6.7 4.0
CP 0.05 0.04 0.02 0.008 0.002
SEE 2.04 1.87 2.02 1.43 1.53 0.311 0.274 0.236 0.220
r, coefficient of correlation. SEE, standard error of estimate. For other abbreviations see Table 1.
based on crude protein and crude fiber are not accurate enough to predict digestibility and nutritive value of forages. Some proximate constituents such as EE and ash have only a small impact on prediction accuracy of forage quality.
Fig. 3. Relationship between observed and predicted value of ME of hays using TDOM.
Fig. 4. Relationship between observed and predicted value of ME of hays using CDOM.
NDF and ADF are better predictors of digestibility and energy value, but they also give a relatively high difference between observed and predicted values.
According to van Soest (1996) the digestibility of the cellulosic carbohydrates is so variable that the NDF and ADF content is not well related to digestibility. This is due to the different environmental factors promoting lignification as opposed to cell wall content. As it is pointed out by van Soest (1996) forages cut after 21 June are apt not to show any reliable relationship between fibre and digestibility. Although forage samples in this study were prepared in spring, the temperatures and soil moisture were not measured.
4.2. Gas production and DM disappearance during rumen fluid fermentation
Gas production and first-step rumen fluid DDM were derived from the same procedure. In spite of the same origin the relationship between the two measurements is not significant (the coefficients of determination between GP48 and DDM are below 0.50). The reason for this is that the amount of gas produced is reduced by the formation of NH4HCO3when NH3is liberated from protein degradation, as indicated by Menke and
Steingass (1988). As the CP content and hence the amount of NH3varies considerably in
these samples, especially in alfalfa hays, there is no doubt that gas production alone is poor correlated with in vivo or other in vitro digestibility parameters.
The mean gas production in this study was calculated using six replicates. In three of them the proportion of rumen fluid to substrate was 10 to 500 mg sample and gas production was recalculated to 200 mg sample. The other three parallels were incubated following original Hohenheim gas test procedure. As was pointed out by BluÈmmel and Becker (1997), changing the proportion of rumen microbes to substrate may change the fermentation patterns. In this study the highest accuracy in prediction of energy value was obtained from gas production at 24 h, which agrees with Menke and Steingass (1988). A little lower are correlations with static gas production between DM intake of roughages and gas production at 8 h incubation using a ratio of 10 ml rumen fluid : 200 mg sample. DDM is calculated from the total residue after incubation, which consists of unfermented substrate and microbial matter. As microbes are calculated as unfermented feed, it is evident, assuming a positive correlation between true substrate degradation and microbial growth, that microbial matter interferes leading to poor prediction accuracy.
A combination of GP24 with CP, NDF, ADF and ash improved the prediction accuracy and coefficients of correlation with DOM rise from 0.58 (Table 2) to 0.83 (Table 3) for alfalfa hay and from 0.78 (Table 2) to 0.87 (Table 4) for grass hay. SEE drops from 3.24 to 2.28 for alfalfa hay and from 2.26 to 1.87 for grass hay. Our equation for predicting ME of grass hay is similar to the Eq. 12c, suggested by Menke and Steingass (1988). In the case of alfalfa hay an improvement in the statistical data can be seen when crude ash and NDF are taken into account. The reason why gas production has to be completed with CP for accurate prediction of digestibility has been discussed above. Nevertheless, multiple equations using gas production and chemical constituents do not reach accuracy of TDOM or CDOM methods, probably as only NH3 from degraded protein interferes
with gas and as the protein of the samples might be of different degradability. Despite lower accuracy compared to the other methods studied the gas production method gives information of the kinetics of digestion (Table 2; Fig. 1). Since digestion is a dynamic
process, this method could play an important role in the evaluation of intake and nutritive value of feedstuffs.
4.3. Rumen fluid±neutral detergent and pepsin±cellulase digestibility
The success of any rumen in vitro system depends on the degree to which it reflects rumen events and the sequential processess of the rumen digestive tract (van Soest, 1994). The Tilley and Terry (1963) method is largely used, because it reproduces well the ruminant digestion sequence. The main modification of the Tilley and Terry procedure done by Goering and van Soest (1970) is that it shortens the duration of the second step to 1 h boiling the residue with neutral detergent compared with a 48 h incubation with pepsin. Besides the long time to complete it, the procedure of Tilley and Terry required a number of steps to do the analysis. Changing centrifuge tubes used in the original method with glass syringes simplifies the procedure from the operative point of view, due to the lack of any decanting and centrifuging which can cause sample losses. However, the modification of the first step does not alter the adequacy of the Tilley and Terry system. This modified method combines two methods (gas production and two-stage method with rumen fluid) and could give information either of digestion kinetics and undigested product.
Treatments of the residue with neutral detergent after measuring gas production and the use of TDOM increased the prediction accuracy of apparent in vivo digestibility of hays. The pepsin±cellulase method (CDOM) predicted OMD and energy value with similar accuracy.
A slight decrease of the SEE in prediction of ME could be observed using CDOM instead of TDOM in the case of grass hay (Table 4), but the opposite is true for alfalfa hay (Table 3). There are very small differences between CDOM and TDOM in the prediction of DOM of grass hay (Table 4). For rapid estimation of digestibility and energy value of hays needed for calculation of ruminant rations, the equations based on CDOM and TDOM presented in Tables 3 and 4 and Figs. 3 and 4 could be used for similar feedstuffs.
Acknowledgements
N.I. would like to thank to Prof W. Drochner who approved her fellowship program. The comments and suggestions of Dr J. De Boever on the manuscript are also gratefully acknowledged.
References
Aerts, J.V., De Brabander, D.L., Cottyn, B.G., Buyse, F.X., 1977. Comparison of laboratory methods for predicting the organic matter digestibility of forages. Anim. Feed Sci. Technol. 2, 337±349.
Andrighetto, I., Gruber, L., Cozzi, G., Uray, G., Guidetti, G., Buchgraber, K., 1992. Prediction of digestible organic matter in dry matter in vivo from the chemical composition, in vitro and in situ measurements on native mountain forages. Anim. Feed Sci. Technol. 39, 323±333.
AOAC, 1980. Official Methods of Analysis, 13th edn., Association of Official Agricultural Chemists, Washington, DC, pp. 125±142.
Aufrere, J., 1982. Etude de la prevision de la digestibilite des fourrages par une methode enzymatique. Ann. Zootech. 31, 111±130.
Aufrere, J., Michalet-Doreau, B., 1988. Comparison of methods for predicting digestibility of feeds. Anim. Feed. Sci. Technol. 20, 203±218.
BluÈmmel, M., Becker, K., 1997. The degradability characteristics of fifty-four roughages and roughage neutral detergent fibres as described by in vitro gas production and their relationship to voluntary feed intake. Br. J. Nutr. 77, 757±768.
BluÈmmel, M., Steingass, H., Becker, K., 1997. The relationship between in vitro gas production, in vitro microbial biomass yield and 15N incorporation and its implication for the prediction of voluntary feed intake of roughages. Br. J. Nutr. 77, 911±921.
Coelcho, M., Hembry, F.G., Barton, F.E., Saxton, A.M., 1988. A comparison of microbial, enzymatic, chemical and near-infrared reflectance spectroscopy methods in forage evaluation. Anim. Feed. Sci. Technol. 20, 219±231.
Cottyn, B.G., De Boever, J.L., Vanacker, J.M., Boucque, Ch.V., 1993. Chemical and enzymatic methods for predicting the energy value of feedstuffs for dairy cattle. In: Proc. 44th Annu. Meeting of EAAP, AÊ rchus, Denmark, 16±19 August 1993, vol.1, p. 271.
De Boever, J.L., Cottyn, B.G., Buysse, F.X., Wainman, F.W., Vanacker, J.M., 1986. The use of an enzymatic technique to predict digestibility, metabolizable and net energy of compound feedstuffs for ruminants. Anim. Feed Sci. Technol. 14, 203±214.
De Boever, J.L., Cottyn, B.G., Andries, J.I., Buysse, F.X., Vanacker, J.M., 1988. The use of cellulase technique to predict digestibility, metabolizable and net energy of forages. Anim. Feed Sci. Technol. 19, 247±260. Dowman, M.G., Collins, F.C., 1982. The use of enzymes to predict the digestibility of animal feeds. J. Sci. Food
Agric. 33, 689±696.
Goering, H.K., Van Soest, P.J., 1970. Forage fiber analyses (Apparatus, Reagents, Procedures and some Applications), Agriculture Handbook no. 379, Agric. Res. Serv., USDA, Washington, DC, 20 pp. Menke, K.H., Raab, L., Salewski, A., Steingass, H., Fritz, D., Schneider, W., 1979. The estimation of the
digestibility and metabolizable energy content of ruminant feedingstuffs from the gas production when they are incubated with rumen liquor in vitro. J. Agric. Sci. Camb. 93, 217±222.
Menke, K., Steingass, H., 1988. Estimation of the energetic feed value obtained from chemical analysis and in vitro gas production using rumen fluid. Anim. Res. Dev. 28, 7±55.
Moore, J.E., 1970. In vitro dry matter or organic mater digestion. In: Harris, L.E. (Ed.), Nutrition Research Techniques for Domestic and Wild Animals, Logan, UT, pp. 5001±5005.
Stat. Soft Inc., 1994.STATISTICA FOR WINDOWS, General Convention and Statistics I, Stat Soft Inc., Tusla, UK.
Steingass, H., 1983. Bestimung des energetischen Futterwertes von wirtschaftseigenen Futtermiteln aus der Gasbildung bei der Pansenfermentation in vitro, Hohenheim University, Dissertation, Fak. IV.
Tilley, J.M.A., Terry, R.A., 1963. A two-stage technique for the in vitro digestion of forage crops. J. Br. Grassl. Soc. 18, 104±111.
Todorov, N., 1995. Nutrient Requirements of Cattle and Buffalo, (in Bulgarian, with English summary) NIS-UZVM, Stara Zagora, 216 pp.
Troelsen, J.E., 1970. In vitro digestion of forage samples. In: Harris, L.E. (Ed.), Nutrition Research Techniques for Domestic and Wild Animals, Logan, UT, pp. 5053±5056.
Van Es, A.J.H., 1978. Feed evaluation for ruminants. I. The system in use from May 1977 onwards in The Netherlands. Livest. Prod. Sci. 5, 331±345.
van Soest, P.J., 1994. Nutritional Ecology of the Ruminants, 2nd edn., Cornell University Press, Ithaca, NY, 476 pp.
van Soest, P.J., 1996. A critique upon problems of predicting feed quality for ruminants, Proc. 47th Annu. Meeting of the Eur. Assoc. Anim. Prod. Littlehamer, Norway, 26 August 1996.
van Soest, P.J., Robertson, J.B., Lewis, B.A., 1991. Methods for dietary fiber, neutral detergent fiber, and nonstarch polysaccharides in relation to animal nutrition. J. Dairy Sci. 74, 3583±3597.
