Comparison of analyses to predict ruminal fibre degradability and indigestible fibre in temperate grass silages

________________________________________________________________________________ Abstract The objective of this study was to compare the ruminal degradability of neutral detergent fibre (NDF) and indigestible NDF (INDF) between silages (n = 24) that originated from three different temperate grass species, i.e. Dactylis glomerata L., Festuca arundinacea L. and hybrid, Felina – Lolium multiflorum L. × Festuca arundinacea S. The data is used to create prediction equations for the effective degradability of NDF (EDNDF), assuming ruminal solid outflow rates of 0.02 (low), 0.05 (medium) and 0.08/h (high), and INDF. The highest values for the potentially degradable fraction of NDF (b), EDNDF2, EDNDF5 and EDNDF8 (874, 560, 366 and 272 g/kg NDF, respectively) and the lowest for INDF (73.1 g/kg DM) were found for F. arundinacea. These parameters were significantly different from the hybrid Felina, which presented the lowest rate of NDF degradation (c), EDNDF5 and EDNDF8 (0.033/h, 341 and 250 g/kg NDF, respectively). The highest for c (0.038/h) and INDF (86.9 g/kg DM) and the lowest for b (847 g/kg NDF) were detected for D. glomerata. The combination of acid detergent fibre (ADF) and acid detergent lignin (ADL) (R = 0.844) were identified by a backward, stepwise, multiple regression analysis as the most accurate to predict INDF. Crude protein, ADF and NDF were found as the most suitable combination for predicting EDNDF2 (R = 0.715), EDNDF5 (R = 0.669) and EDNDF8 (R = 0.648). Calculated equations found practical application in laboratory analyses to evaluate the nutritional quality of feeds for ruminants. ________________________________________________________________________________


Introduction
Grass is one of the most important sources of nutrients for domesticated ruminants during a large part of the year (Taweel et al., 2005).Yan & Agnew (2004) noted that despite being a major feed for ruminants across the world, the nutritive value of grass silage is extremely variable.The digestibility of the different grass species could be distinctly different, and is also influenced by area of origin, including temperature, light intensity, total rainfall, soil type, fertilization level, and by stage of maturity and preservation method (Huhtanen et al., 2006;Jančík et al., 2009).This is especially evident in the rate and extent of ruminal degradation of neutral detergent fibre (NDF) (Hoffman et al., 1993).Several authors have compared the chemical composition and NDF degradability of different grass species (Cherney et al., 1993;Hoffman et al., 1993;Jensen et al., 2003).However, comparisons of NDF degradability and indigestible neutral detergent fibre (INDF) of grass silages, made from Dactylis glomerata, Festuca arundinacea and grass hybrids (Lolium multiflorum L. × Festuca arundinacea S.) grown under the same conditions, are lacking.
The extent and rate of digestion of structural carbohydrates of the cell wall, which is mainly degraded in the rumen with the aid of ruminal microorganisms (Zhang et al., 2007), are highly related to the nutritive value of ruminant feeds (Andrés et al., 2005).The NDF concentration, which represents cell walls of feeds, could be affected by several factors such as temperature, light intensity, water availability, latitude, maturity, and harvesting and storage methods (Van Soest, 1994).Furthermore, the intrinsic characteristics of the carbohydrate fraction, such as the proportion of starch or cellulose and lignification of the cell wall, could have an influence on the rate of fermentation of feeds (Nagadi et al., 2000).Indigestible neutral detergent fibre is a part of the cell wall that is unavailable to microbial digestion in ruminants, even when the total tract residence time of the fibre could be extended to an infinite time period (Van Soest, 1994).Lund et al. (2004) showed that the NDF concentration and potential digestibility of NDF are important factors to be included in new feed evaluation systems.If the incorporation of degradation characteristics into a forage quality prediction system seems promising, methods to replace laborious conventional in situ or in vitro procedures to estimate these characteristics on a routine basis, should be evaluated (Hackmann et al., 2008).
The objectives of the present study were to (1) compare ruminal degradability parameters of NDF and indigestible NDF (INDF) among silages made from three different grasses, and (2) to develop prediction equations for ruminal degradability parameters of NDF and INDF from their chemical composition.

Materials and Methods
Twenty four silage samples made from three different temperate grasses, i.e. orchardgrass -Dactylis glomerata L. cv.Dana; fescue -Festuca arundinacea L. cv.Prolate and the hybrid Felina -Lolium multiflorum L. × Festuca arundinacea S. cv.Felina were evaluated in the present study.The grass was harvested from the primary growth of monocultured grasses on 19 and 26 May of 2004 and 27 May and 10 June of 2005 at the Větrov Breeding Station, Tábor region, Czech Republic (49° 31' 2.04" N lat, 14° 28' 4.9" E long; 620 m altitude).The growth stages of the grass used for silage making are presented in Table 1.The annual mean temperatures in 2004 and 2005 were 8.4 and 8.3 °C, and the annual rainfall 1020 and 1047 mm, respectively.The grass was wilted for up to 4 h on a table drier with cool air ventilation for 300 to 400 g/kg dry matter (DM), whereafter it was cut with a cutting machine (RP03, OPP Polička a.s., Czech Republic) to a theoretical length of 1.0 to 1.5 cm and ensiled, without additives, in hermetic glass vessels (3 L capacity).For each grass species and harvest time, two vessels were stored in a dark room at a temperature of 15 °C for storage periods of 10 and 20 weeks, respectively.At the opening the silages were analyzed for fermentation quality, i.e. pH (measured with a pH meter, WTW Level 1) and concentration of lactic, acetic and butyric acids, determined by isotachophoresis (Kvasnička, 2000) using an Ionosep 2001 apparatus (RECMAN -Laboratory Systems, Ostrava, Czech Republic).Isotachophoresis is one of the basic electrophoretic techniques used based on injecting a sample solution between leading and terminating solutions present in the separation compartment.The leading electrolyte solution contains an anion or cation which has a higher effective mobility than those of the sample constituents to be analyzed.The terminating electrolyte solution contains an anion or cation of a lower effective mobility.The electric field was applied to separate the sample constituents (Kvasnička, 2000).The silage samples were subsequently oven-dried at 50 °C for 48 h and milled to pass a 1 mm screen.
All ensiled samples (n = 24) were analyzed for DM, crude protein (CP), ash, ether extract (EE), NDF, acid detergent fibre (ADF) and acid detergent lignin (ADL).The residual moisture of the samples was determined at 105 °C for 12 h of oven-drying, and ash content was determined after combustion at 550 °C for 4.5 h (Regulation No. 497/2004, 2004).The EE level was determined, using the Soxtec extraction method with petroleum-ether (AOAC Official Method 920.39;AOAC, 2005).The Kjeldahl method was used for determining of nitrogen (N) (AOAC Official Method 976.05;AOAC, 2005), with the CP content  973.18 (AOAC, 2005).The NDF concentration (Van Soest et al., 1991) of samples and digested residues were analysed in the presence of sodium sulphite, but without α-amylase treatment, and is presented as ashfree.All fibre analyses were performed in an ANKOM 220 Fiber Analyzer (ANKOM Technology Corporation, NY.USA).
In situ experiments complied with the guidelines of the Ministry of Agriculture of the Czech Republic (Act No 246/1992 Coll.) in respect of animal experimentation and the care of animals under study.For in situ analyses, two Holstein steers (live weight 800 kg) fitted with rumen cannulas (120 mm in inside diameter), were used.The animals had ad libitum access to meadow hay, and were fed twice a day with 1 kg of barley meal per animal.The meadow hay was made from growth, containing orchardgrass (Dactylis glomerata L.; 30%), timothy (Phleum pratense L.; 20 %), meadow fescue (Festuca pratensis H.; 25%), ryegrass (Lolium perenne L.; 10%), white clover (Trifolium repens L.; 10%) and herbs (5%).For the in situ determination, the silage samples were oven-dried at 50 °C for 48 h and milled to pass a 1 mm screen.The rumen degradability of the NDF, in the experimental silages, was determined by incubating 1.5 g of each silage (in triplicate) in nylon bags (pore size 42 µm, internal dimensions 50 × 120 mm, Uhelon 130 T, Silk and Progress Moravská Chrastová, Czech Republic) for 6, 12, 24, 48, 72 and 96 h in the rumen of the steers.
Indigestible NDF contents of the silage samples were determined by using a 288 h rumen incubation period (Rinne et al., 1997;Nousiainen et al., 2004).Nylon bags (Swiss Silk Bolting Cloth Ltd., Zurich, Switzerland; external dimensions 60 x 120 mm) with a small pore size (17 µm), which minimize particle inflow and outflow but still ensure sufficient microbial activities inside the bags (Huhtanen et al., 1998;Nousiainen et al., 2003b), were used.Three grams of each sample were weighed into a bag and incubated in three replicates per animal.After incubation, the bags were rinsed by hand with cold water for 30 minutes and dried at 50 °C for 48 h.
Parameters of NDF ruminal degradability (Deg (t) ) and effective degradability of NDF (ED NDF ) were calculated according to equations adapted from Ørskov & McDonald (1979): where b = fraction of NDF potentially degradable in the rumen, c = rate constant of degradation of fraction b, t = time of incubation and k = ruminal outflow rate.ED NDF was estimated for each sample assuming ruminal solid outflow rates of 0.02, 0.05 and 0.08/h (ED NDF2 , ED NDF5 and ED NDF8 , respectively), which represent low, medium and high feed intakes, respectively (Petit & Tremblay, 1992).The soluble fraction was omitted from the equations, as NDF is by definition insoluble.
The results for chemical composition (ash, EE, CP, NDF, ADF and ADL), rumen degradation parameters (c, b, ED NDF2 , ED NDF5 and ED NDF8 ) and INDF were analyzed using the MIXED procedure of SAS (SAS, 2002(SAS, -2003)).For the evaluation of rumen degradation parameters and INDF, grass species, year, storage period and animal were used as fixed effects, and harvest time nested within year used as a random effect.The grass species × year, grass species × storage period, year × storage period, grass species × animal, year × animal and storage period × animal interactions were included in the model.The data for chemical composition was analyzed with a similar model as described above, but with the animal effect excluded.Differences in mean values were evaluated by Scheffe's tests (Scheffé, 1953).
Correlation and simple linear regression were used for the evaluation of relationships within the parameters of chemical composition and utilization of NDF of the grass silages.Backward, stepwise, multiple regression analysis was applied to determine the best combination of chemical components for the prediction of ED NDF and INDF.

Results and Discussion
The chemical composition, fermentation quality, NDF ruminal degradation parameters and INDF of grass silages evaluated are presented in Table 2.The values found in the present study showed a high variability in the chemical composition and degradability parameters, reflecting a wide range in NDF quality in the experimental silages.A wide variability is desired for developing prediction equations.The values of chemical composition are comparable to the results obtained by Huhtanen et al. (2002), Koukolová et al. (2004) and Nousiainen et al. (2004).Wilman et al. (2000) reported mean b and c values of 741 g/kg NDF and 0.048/h, respectively, for NDF degradability of grass silages that contained 68 g/kg DM of lignin, whereas Spanghero et al. (2007) found ED NDF values of grass hay to vary from 239 to 483 g/kg NDF (k = 0.03/h).A mean value of 860 g/kg NDF for the potential digestibility of NDF, and 71.8 g/kg DM for INDF, was obtained for grass silage by Oba & Allen (1999).According to Rinne et al. (2002), the INDF contents of grass silages harvested from 13 June to 4 July in Finland ranged from 48 to 124 g/kg DM, and Lund et al. (2007) reported values of 49 and 137 g/kg DM INDF for early and late cut grass silages, respectively.
The differences in chemical composition of grass species are indicated in Table 3.The highest CP content was found in F. arundinacea and the lowest (P <0.05) in the hybrid Felina.The highest NDF content was detected in hybrid Felina (P <0.05).A significantly higher mean value of ADF was detected in hybrid, Felina compared to F. arundinacea.No significant differences in content of ADL and EE were found in the different grass species.D. glomerata had the lowest (P <0.05) level of ash.Similar to current results, Skládanka et al. (2008) detected a higher NDF content in the hybrid Felina, compared to D. glomerata.The same authors, however, stated a better quality (according to chemical composition) for the hybrid Felina, than for F. arundinacea.With the interactions between grass species × storage period and year × storage period being non-significant, both year and storage period, as main factors, did not present any differences (P >0.05).The interactions between grass species × year were different for ash, CP, NDF and ADF (P <0.05) (Table 6).No significant interactions in grass species × storage period, grass species × animal, year × animal, storage period × animal and the animal effect were found for any of the parameters.Significant differences have been found in grass species regarding NDF degradation parameters (b, c, ED NDF ) and INDF content (Table 4).Dactylis glomerata presented the lowest (P < 0.05) value for b in comparison with F. arundinacea and hybrid Felina.The lowest c value (P <0.05) was found for hybrid Felina.The rate of NDF degradation (c) presented differences between year (P <0.0001) (Table 5) and storage periods (P = 0.019) (Table 5).The ED NDF2 , ED NDF5 and ED NDF8 were significantly higher after 20 weeks compared to 10 weeks of storage (P <0.001) (Table 5).At outflow rates of 0.05 and 0.08/h, hybrid Felina presented the lowest (P <0.05) ED NDF .The lowest (P <0.05) content of INDF was found in F. arundinacea, whereas D. glomerata showed the highest numerical INDF content.However, it only differed significantly from F. arundinacea (Table 4).A faster rate of NDF degradation (c) detected after the longer storage period (20 weeks) (P = 0.001; SE = 0.0006) influenced the calculated ED NDF values (P = 0.0005, 0.0003 and 0.0003; SE = 16.6,11.4 and 8.7 for ED NDF2 , ED NDF5 and ED NDF8 , respectively) (Table 5).This could be attributed to the shortening of the cellulose chain length as a result of extra-cellular cellulase activity from silage microflora and hydrolysis of hemicellulose by organic acids produced during fermentation (Yahaya et al., 2001).Cell walls, subjected to silage processing and microorganisms, may be more susceptible to enzymatic attack, mainly in early periods of fermentation in the rumen.Comparable to current results, NDF digestibility of F. arundinacea was found by Pozdíšek et al. (2003) to be higher than for the festulolium hybrid Hykor.Presented differences of ED NDF in the different grass species are in agreement with our results on the in situ degradation of DM (Jančík et al., 2009), where F. arundinacea had higher mean values of DM effective degradability than D. glomerata and hybrid Felina.Differences in grass species might be related to the differences in maturity of the observed grass (Table 1).D. glomerata and hybrid Felina are early maturing grass species, whereas F. arundinacea is a late-maturing grass.In the present study differences were found for interactions between grass species × year and year × storage period.The differences for interactions between grass species × year detected for all degradability parameters and INDF (P <0.0001) are presented in the Table 6.The interactions between year × storage period were different for parameter c, ED NDF2 , ED NDF5 and ED NDF8 (P <0.001) (Table 7).
The correlation coefficients (r) of the relationships between chemical composition and parameters of NDF degradation (b, c, ED NDF2 , ED NDF5 and ED NDF8 ) and INDF are given in Table 8.Parameter b was the highest (P <0.05) correlated to ADL, and the lowest to ash.The lower ash content was found for silages made from younger grasses.It means that these silages had a higher OM content, and a higher OM content can lead to faster digestibility of feeds, especially with regard to the lower lignification of younger grasses.The correlation coefficient between c and ash (r = -0.73)shows that faster degraded silages had a lower ash content.In contrast, parameter c was the highest (P <0.05) correlated to ash, and the lowest to EE.The highest correlations of ED NDF and INDF were found in ADF (P <0.05), whereas these measurements presented low correlations with EE.In agreement with the current study, Hoffman et al. (1993) presented r-values of -0.32, 0.60 and 0.50 between ADL and CP, ED NDF and CP, and ED NDF and ADL, respectively.Andrés et al. (2005) found a high correlation between b and CP (r = 0.65), and NDF degradability after a 96-h incubation period and CP (r = 0.80).Significant correlations of -0.82, -0.78, -0.78 and 0.88 were reported (Koukolová et al., 2004;Jančík et al., 2008) between b and ADL, c and NDF, c and ADF, and INDF and ADL, respectively.Several authors (Traxler et al., 1998;Jančík et al., 2008) used ADL content as the most reliable variable to predict the INDF content of grass.According to Grenet & Besle (1991) and Nagadi et al. (2000) degradability of cell wall carbohydrates is mainly limited by lignin content, accentuating its influence on feed utilization (Ahmad & Wilman, 2001).Bosch & Bruining (1995) confirmed that grass silages, with a high lignin content, have a lower digestibility compared to silages that contained low levels of lignin.Traxler et al. (1998) found an important relationship (r = 0.81) between INDF and the lignin content in the different type of grass.This was confirmed by a high negative r value (-0.80) between the NDF digestibility and ADL content of legumes and grasses by Jung et al. (1997).
The most reliable combinations of chemical components (CP, NDF, ADF, ADL) to predict ED NDF2 , ED NDF5 , ED NDF8 and INDF, calculated by backward, stepwise, multiple regression analysis, are presented in Table 9.A multiple regression analysis was used because of the low predictability of equations determined with simple linear regression and one predictor (R 2 = 0.272 to 0.689) (Table 10).For ED NDF5 , ED NDF8 and INDF the lowest residual mean square error (RMSE) was obtained with a combination of three or four predictors.The ED NDF2 had the lowest RMSE when a combination of four predictors was used.However, the RMSE value derived from the use of two predictors for INDF and three predictors for ED NDF2 did not differ substantially from the use of four predictors.The combination of CP, ADF and NDF was identified as the most suitable predictor for ED NDF2 , ED NDF5 and ED NDF8 , whereas ADF and NDF also predicted ED NDF2 , ED NDF5 and ED NDF8 with an acceptable degree of accuracy.The combination of ADF and ADL was found to predict INDF accurately.However, an R 2 -value of 0.689 showed that ADF can be used as a single predictor for INDF.Jančík et al. (2010) reported that the combination of ADF and NDF was the most suitable predictor for ED NDF of dried F. arundinacea (R 2 = 0.941, RMSE = 18.3).Chemical parameters as predictors of digestion become complicated when plant types such as legumes or gramineae are studied (Khazaal et al., 1995).Although empirical equations have not been proven to be successful in determining DM digestibility (Khazaal et al., 1995) or OM digestibility (Huhtanen et al., 2006) from fibre components, the current study shows that prediction equations based on chemical components can produce equations with acceptable accuracy for NDF digestibility.However, it should be emphasized that equations have to be restricted to forage type and harvest stage.
The content of ADF was the best single predictor of ED NDF and INDF (Table 10).However, a better prediction equation for INDF was found when using a combination of ADF and ADL (Table 9).This could be attributed to the fact that lignin is regarded as the principal factor limiting digestibility, although it does not affect all feed components.Non-cell wall components are not influenced by lignin, but they can often be highly correlated.Therefore, lignin concentration affects mainly the availability of cell wall polysaccharides (Van Soest, 1994).Cellulose, which along with lignin forms ADF, reduces the digestion rate and extent of digestion which are related to the lignin content.Hemicellulose (presenting NDF along with cellulose and lignin) is closely associated with lignin, and the digestibility of hemicellulose is directly related to that of cellulose and inversely related to lignification (Van Soest, 1994).In the present study, a combination of CP, ADF and NDF gave the best equations for ED NDF based on three predictors.The higher effect of CP on ED NDF was caused by the influence of parameter c in the calculation of ED NDF .A high correlation between CP and parameter c most likely occurred because of a co-association when protein declines as grass matures and lignification proceeds (Van Soest, 1994;Nousiainen et al., 2003a).Neutral detergent fibre represents the total insoluble matrix fibre, and it is better related to rumination and passage compared to other chemical components.Acid detergent fibre does not represent all insoluble fibre, but it is frequently better correlated with digestibility than NDF (Van Soest, 1994).

Conclusions
According to parameters describing NDF utilization, all tested types of grass are suitable forages to be ensiled for ruminant nutrition.However, D. glomerata and hybrid Felina, should be harvested earlier because of their rapid maturing rate.F. arundinacea matures more slowly and thus provides a high-quality forage for a longer vegetative period.This study confirmed that the ED NDF and the INDF content of grass silages could be effectively predicted from a combination of two or three chemical components.Based on the current results, the use of a combination of CP, ADF and NDF for prediction of ED NDF , and ADF and ADL for prediction of INDF, can be suggested.These equations could replace laborious and time-consuming in situ and in vivo techniques to evaluate the nutritional quality of feeds for ruminants.However, more research is needed on the relationships between NDF rumen degradation parameters (including INDF) and chemical components for silages made of legumes, maize, regrowth grass and whole grain cereals.

Table 1
Maturity stages 1 of used grass for silage-making by harvest date

Table 2
Chemical composition, silage fermentation quality and NDF utilization of grass silages (n = 24)

Table 3
Chemical composition (g/kg DM) of silages as influenced by grass species (n = 8 per grass species) a-c Within a column, means without a common superscript differ (P <0.05).DM -dry matter; EE -ether extract; CP -crude protein; ADF -acid detergent fibre; NDF -neutral detergent fibre; ADL -acid detergent lignin; SE -standard error.

Table 4
Rumen neutral detergent fibre (NDF) degradability and indigestible neutral detergent fibre content of silages as influenced by grass species (n = 8 per grass species) a,b Within a column, means without a common superscript differ (P <0.05).b -fraction of NDF potentially degradable in the rumen (g/kg NDF); c -rate constant of disappearance of fraction b (/h); ED NDF2 , ED NDF5 and ED NDF8 -effective degradability of NDF (g/kg NDF) assuming rumen solid outflow rates of 0.02, 0.05 and 0.08/h; INDF -indigestible neutral detergent fibre (g/kg DM); SE -standard error.

Table 5
Differences of parameters describing utilization of silages' neutral detergent fibre (NDF) evaluated between years and between storage periods a,b Within a row, means without a common superscript differ (P <0.05).b -fraction of NDF potentially degradable in the rumen (g/kg NDF); c -rate constant of disappearance of fraction b (/h); ED NDF2 , ED NDF5 and ED NDF8 -effective degradability of NDF (g/kg NDF) assuming rumen solid outflow rates of 0.02, 0.05 and 0.08/h; INDF -indigestible neutral detergent fibre (g/kg DM); SE -standard error.

Table 6
Differences in chemical composition and parameters describing utilization of silages' neutral detergent fibre (NDF) evaluated for grass species × year interaction a-e Within a row means without a common superscript differ (P <0.05).D -Dactylis glomerata L.; F -Festuca arundinacea L.; H -hybrid Felina; ADF -acid detergent fibre; NDF -neutral detergent fibre; ADL -acid detergent lignin; b -fraction of NDF potentially degradable in the rumen (g/kg NDF); crate constant of disappearance of fraction b (/h); ED NDF2 , ED NDF5 and ED NDF8 -effective degradability of NDF (g/kg NDF) assuming rumen solid outflow rates of 0.02, 0.05 and 0.08/h; INDF -indigestible neutral detergent fibre (g/kg DM); SE -standard error.

Table 7
Differences of parameters describing utilization of silages' neutral detergent fibre (NDF) evaluated for year × storage period interaction Within a row, means without a common superscript differ (P < 0.05).b -fraction of NDF potentially degradable in the rumen (g/kg NDF); c -rate constant of disappearance of fraction b (/h); ED NDF2 , ED NDF5 and ED NDF8 -effective degradability of NDF (g/kg NDF) assuming rumen solid outflow rates of 0.02, 0.05 and 0.08/h; INDF -indigestible neutral detergent fibre (g/kg DM); SE -standard error.

Table 8
Linear correlation coefficients between chemical components and fibre-related measurements (n = 24)

Table 9
Best-fit combinations of chemical components to predict neutral detergent fibre effective degradability and indigestible neutral detergent fibre content when applying backward, stepwise, multiple regression (n = 24)