Relationship between chemical composition and in situ rumen degradation characteristics of grass silages in dairy cows

Abstract

The DVE/OEB₂₀₁₀ system in the Netherlands uses a large database of in situ rumen incubations with grass silage and grass hay samples to derive prediction formulas to estimate the rumen degradation characteristics of a number of feed value parameters. These in situ rumen incubations were not performed for this specific purpose and the data were generated at different research institutes over more than 40 years, using different grass management and fertilization practices and using different protocols. The objectives of this study were to 1) generate a new database on the rumen degradability of dry matter (DM), organic matter (OM), crude protein (CP) and neutral detergent fibre (NDF) of grass silages, 2) compare this new database with the old database used in the DVE/OEB₂₀₁₀ system, and 3) derive regression equations using the new database to investigate the relationships between chemical composition and in situ ruminal degradation characteristics of DM, OM, CP and NDF of the grass silages. Sixty nine grass silages, with a broad range in chemical composition and quality parameters, were selected and incubated using the nylon bag technique in the rumen of three lactating Holstein Friesian cows for 2, 4, 8, 16, 32, 72 and 336 h. There was a large range in the rumen degradable fractions of DM, OM, CP and NDF of the grass silages at each rumen incubation period. The data on the rumen degradation characteristics of DM, OM, CP and NDF in the present study were determined using the same standard incubation protocol, the same cows, and the same chemical analysis procedures for all the grass silage samples. Regression analysis, using the new database, showed relationships between the washable (W) fraction, rumen undegradable (U) fraction, potentially rumen degradable (D) fraction and effective rumen degradation (ED) of DM, OM, CP and NDF, respectively, and the chemical composition of the grass silages.

Keywords:

• In situ nylon bag technique
• Chemical composition
• Effective rumen degradation
• Regression equations
• Grass silages

Abbreviations:

• ADF
• acid detergent fibre
• ADL
• acid detergent lignin
• AAT-PBV
• Danish protein evaluation system
• CP
• crude protein
• D
• potentially rumen degradable fraction
• DM
• dry matter
• DVE/OEB₂₀₁₀
• Dutch protein evaluation system
• ED
• effective rumen degradation
• k _d
• fractional degradation rate
• k _p
• fractional passage rate
• MP
• British metabolizable protein system
• NDF
• neutral detergent fibre
• OM
• organic matter
• PDI
• French protein evaluation system
• REP
• rumen escape protein
• SD
• standard deviation
• TMR
• total mixed ration
• U
• rumen undegradable fraction
• W
• washout fraction

1 Introduction

Grass silage is a forage biomass that is mainly used as a winter fodder for dairy cows and is the most important form of conserved forage for ruminants in many regions of Europe [1]. Protein in grass silage is more degradable in the rumen than protein in grass hay and fresh grass due to the fermentation processes in the silos [2]. Ruminal degradation of different components of grass silages is influenced by many factors such as stage of maturity, preservation method, forage species and cultivars [3–5[3] [4] [5]].

To optimize diet formulation in terms of performance, nutrient losses, animal well-being and economical profitability, information is required on the nutrient requirements of animals and nutrient availability of various feedstuffs. In ruminants, the nutrient availability can be measured using various in situ, in vitro and in vivo techniques. The in situ nylon bag technique is the most frequently used technique for the determination of degradability parameters of chemical feed components such as dry matter (DM), crude protein (CP), neutral detergent fiber (NDF), minerals and trace elements [6–8[6] [7] [8]] and is considered a reference method to determine the rumen degradation characteristics of feedstuffs [9].

Various ruminant feed evaluation systems, such as the DVE/OEB₂₀₁₀ system [10] in the Netherlands and the PDI system [11] in France, use prediction formulas to estimate the rumen degradation characteristics of a number of feed value parameters. The Dutch DVE/OEB₂₀₁₀ feed evaluation system [10] uses a database compiled from different in situ rumen incubations studies with grass silage and grass hay samples. These studies were conducted at various research institutes in the Netherlands and Belgium employing different protocols to estimate the degradation characteristics of certain feed components spanning several decades. These in situ rumen incubations studies were not performed for the specific purpose to derive prediction formulas for the DVE/OEB₂₀₁₀ feed evaluation system but were conducted to investigate other specific scientific hypotheses. At present, these results can be considered outdated as most of these experiments were conducted before 1995 during which time a large variety in crop species and cultivars were used in the Netherlands. Nitrogen fertilization level, ensiling methods and incubation protocols have changed, compared to when the studies were conducted and which data have been used in the DVE/OEB₂₀₁₀ database. Different subsets of the DVE/OEB₂₀₁₀ database were used to derive prediction formulas for degradation characteristics of nutrients for grass silage and grass hay samples. For example, data of a subset of 97 grass silage and grass hay samples were used to derive prediction formulas for the degradation characteristics of CP while the data of a subset of 50 samples were used to derive prediction formulas for the degradation characteristics of NDF.

The main objective of this study was to improve the prediction of the feeding value of grass silages. This study reports the 1) generating of a new database on the rumen degradation characteristics of DM, OM, CP and NDF of grass silages, 2) comparison of this new database with the old database used in the DVE/OEB₂₀₁₀ system, and 3) regression equations derived using new database to investigate the relationships between chemical composition and the in situ ruminal degradation characteristics of DM, OM, CP and NDF of grass silages.

2 Materials and methods

2.1 Selection of grass silages

More than one hundred grass (mainly Lolium perenne) silage samples (∼10 kg per silage) were obtained during 2007, 2008 and 2009 from various Dutch commercial farms located in different regions in the Netherlands. The samples were collected by trained technicians from a feed analysis laboratory (Blgg AgroXpertus, Wageningen, The Netherlands) using a hollow drill. After collection, individual silages were homogenized, divided into roughly two equal parts with one half air dried (70 °C for 16 h) and subjected to chemical analyses while the other half was stored at -20 °C. After chemical analyses of the air dried material, a table was developed containing information on chemical composition and quality parameters of grass silages. A total of 69 grass silage samples were selected on the basis of broad range in the contents of DM, CP and NDF. The selected samples were stored at -20 °C. Each selected grass silage stored at -20 °C was located and cut using a bread slicer (JAC Duro BEL 450; ABO, Leek, The Netherlands) having a distance of 11 mm between the discs, thoroughly mixed by hand and divided into three parts; one part (∼2.5 kg) was subjected to wet chemical analyses after freeze drying, another part (∼1.5 kg) stored (-20 °C) for later nylon bag incubations, and the third part (∼1 kg) was stored (-20 °C) as a reserve for possible future reanalysis.

2.2 In situ rumen incubations

Three multiparous (second lactation) Holstein Friesian cows, producing >15 kg milk per day, fitted with permanent ruminal cannulas were used in this experiment. Cows were fed total mixed ration (Table 1) twice a day and had 24 h/d access to fresh water. The grass silage samples were incubated in the rumen of three cows which were housed at the research farm “Waiboerhoeve” (Wageningen UR Livestock Research, Lelystad, The Netherlands) according to the procedure described by Cone et al. [12]. The fresh grass silage samples (∼5 g DM) were weighed into 10 cm x 19 cm nylon bags (pore size 37 μm; porosity 24%; Nybolt, Zürich, Switzerland) and the bags were stored at -20 °C. The frozen bags were incubated in the rumen for 2, 4, 8, 16, 32, 72 and 336 h. The 0 h bags were washed in the washing machine (AEG-Electrolux Öko Turnamat 2800, Stockholm, Sweden) for 40 min using tap water at 25 °C without incubation in the rumen and residues were used to calculate the W fraction. Six bags of each grass silage sample, combined with 2 reference samples per series, were incubated in the rumen of the three cows (2 bags per cow per incubation time) for 2, 4, 8, 16, and 32 h. Because of a low recovery of incubated residue per nylon bag for the 72 and 336 h incubation periods, 9 bags of each grass silage and 2 reference samples were incubated in the rumen of the three cows (3 bags per cow per incubation time) for these incubation periods. Four incubation series were carried out and all the rumen incubations were performed within 2 months. The same three cows were used for all the rumen incubations. After removal from the rumen, bags were placed in ice water after which the bags were washed to remove the adhering stuffs. Then the bags were stored in freezer at -20 °C for at least 24 h, after which the bags were thawed and washed in the washing machine as described above. The washed bags were stored at -20 °C and subsequently freeze dried. For each grass silage sample, rumen incubation residues from the three cows at each rumen incubation period were pooled and the contents were ground over a 1 mm sieve, using a hammer mill (Pepping, 200 AN-797002, Deventer, The Netherlands).

Table 1 Diet components of the total mixed ration (TMR) fed to the cows during rumen incubations.

Feed ingredient	g/kg DM in TMR
Maize silage	237.4
Grass silage	395.8
Grass hay/wheat straw	13.5
Soybean meal	47.4
Sweet syrup	9.5
Wet distillers grains with solubles	84.3
Soybean meal (rumen protected)	18.6
Minerals and vitamin premix ^a	10.9
Concentrate ^b	182.6

a Premix composition: calcium 175 g/kg; phosphorous 0.2 g/kg; magnesium 130 g/kg; sodium 50 g/kg; chloride 78 g/kg; vitamin A 600,000 IE/kg; vitamin D 120,000 IE/kg; vitamin E 8,000 IE/kg.

b Contained (%): maize, 30; palm kernel, 22; rapeseed solvent extract, 21.3; citrus pulp, 10; soybean meal (rumen protected), 5; beet molasses, 5; molasses, 2; wheat, 2; chalk (CaO), 0.6; urea, 0.6; salt, 0.5; magnesium oxide (80% MgO), 0.5; palm oil, 0.2; vitamin and minerals premix, 0.3.

2.3 Chemical analyses

The ground (1 mm) freeze dried grass silage samples were analyzed for DM, ash, CP, crude fat (CFat), crude fibre (CF), sugar, NDF, acid detergent fibre (ADF), acid detergent lignin (ADL), neutral detergent insoluble nitrogen (NDIN) and acid detergent insoluble nitrogen (ADIN). The pooled, ground rumen incubated residues were analysed for DM, ash, CP and NDF. The DM content was determined by oven drying at 103 °C for 4 h (ISO 6496) and ash content by incineration at 550 °C for 4 h (ISO 5984). The NDF was determined according to the modified method [13] with the use of amylase (ISO 16472) and expressed without residual ash. The ADF was determined by boiling with acid detergent reagent and expressed without residual ash (ISO 13906:2008). The ADL was determined after boiling with acid detergent reagent and a treatment with sulphuric acid (ISO 13906:2008). Crude fat (CFat) was determined by ISO 6492 and sugar content was determined by the Luff-Schoorl method (NEN 3571:1947nl). The N was determined using the Kjeldahl method (ISO 5983) and CP was calculated as N × 6.25.

2.4 Calculations

2.4.1 Effective rumen degradation (ED)

The ED of DM (ED_DM), NDF (ED_NDF) and OM (ED_OM) was calculated according to equation of Ørskov and McDonald [14]:(1) $ED=W+(k_d/(k_d+k_p))\times D$(1) where W = washable fraction, k _d = fractional degradation rate (h⁻¹), k _p = fractional passage rate (h⁻¹) and D = potentially rumen degradable insoluble fraction. For silages, 5% of W fraction of CP was assumed to be rumen escape protein (REP) [10]; therefore the ED of CP (ED_CP) was calculated by the modified formula in the present study;(2) $E{D}_{CP}=0.95\times W+(k_d/(k_d+k_p))\times D$(2)

2.4.2 Rumen escape protein (REP)

The REP was calculated using the following modified model [15]:(3) $REP=U+(k_p/(k_d+k_p))\times D+0.05\times W$(3) where U = rumen undegradable fraction.

2.4.3 Fractional degradation rate (k_d)

The k _d of DM, OM and CP was calculated according to the first order model of Robinson et al. [16] including U, D and kd:(4) $Y_t=U+D_t\times \exp (-k_d\times t)$(4) where Y_t = degradation at time t, D _t = potentially degradable fraction in the rumen at time t, The k _d of NDF was calculated using model 4 including a lag time [17]:(5) $Y_t=U+D_t\times \exp (-k_d\times (t-L))$(5) where L = lag time (h).

2.4.4 Fractional passage rate (k_p)

Jančík et al. [18] used three k _p values for calculating the ED of DM; 0.02, 0.05 and 0.08 h⁻¹ representing low, medium and high feeding amounts respectively The ED of DM in the present study was calculated using three corresponding k _p values of 0.02 h⁻¹ (ED_DM2), 0.05 h⁻¹ (ED_DM5) and 0.08 h⁻¹ (ED_DM8). In the present study, the ED of OM of grass silages was calculated using the same three k _p values as used for DM; 0.02 h⁻¹ (ED_OM2), 0.05 h⁻¹ (ED_OM5) and 0.08 h⁻¹ (ED_OM8). An average k _p value for NDF of 0.025 h⁻¹ (range 0.018 and 0.029 h⁻¹) was used based on data for fresh grass and grass silages [19]. The k _p value (0.045 h⁻¹) for CP was adopted from the DVE/OEB₂₀₁₀ system [10].

2.5 Statistical Analysis

Rumen degradation data of DM, CP and NDF after different rumen incubation times (2, 4, 8, 16, 32, 72, and 336 h) were summarized by descriptive statistics. The k _d was calculated using model 4 and 5 in Genstat (14th edition). Regression equations were derived to determine the relationships between rumen degradation characteristics of DM, OM, CP and NDF and the chemical composition of grass silage samples using the PROC REG procedure of SAS 9.2 [20]. The backward stepwise procedure was followed to derive the regression equations with significant predictors (P < 0.05). For k _d of DM and CP, no regression equation was presented as none of the predictors proved significant.

3 Results

Chemical composition of grass silages (n = 69) used in this experiment and, grass silage (n = 102) and grass hay (n = 14) samples in the database used by the DVE/OEB₂₀₁₀ system is shown in Table 2. The DVE/OEB₂₀₁₀ system uses the combined data of the grass and hay samples to show the large range in the chemical composition and the rumen degradation characteristics. It was not possible to split the DVE/OEB₂₀₁₀ database into grass silage and grass hay samples. The grass silage/hay samples in the DVE/OEB₂₀₁₀ database had a larger range in DM content (143 to 909 g/kg fresh matter) than the silages evaluated in the present study (201 to 680 g/kg fresh matter). The average grass silage/hay CP content (198 g/kg DM) was higher in the samples of the DVE/OEB₂₀₁₀ database compared to the present study (170 g/kg DM) while the reverse was observed for the average sugar content (91 vs. 55 g/kg DM). No information was available on the quality parameters of the grass silages used in the DVE/OEB₂₀₁₀ database.

Table 2 Chemical composition of grass silages (n = 69) used in this experiment and, grass silage (n = 102) and grass hay (n = 14) samples in the database used by the DVE/OEB₂₀₁₀ system.

		Grass silages used in this experiment				DVE/OEB2010 database ^a
Variable		Mean	SD ^b	Minimum	Maximum	Mean	SD	Minimum	Maximum
Chemical composition (g/kg DM) ^#
	Dry matter (g/kg fresh matter)	454	112	201	680	473	194	143	909
	Ash	102	18	70	192	113	24	31	250
	Crude protein	170	29	109	222	198	48	99	305
	Crude fat	42	7	28	63	40	7	17	62
	Crude fiber	257	26	162	323	265	36	201	342
	Sugar	91	45	12	246	55	49	1	228
	Neutral detergent fibre	491	53	326	611	470	68	330	669
	Acid detergent fibre	268	29	157	332	287	42	220	355
	Acid detergent lignin	19	7	10	40	27	13	10	57
	NDIN ^c (g N/kg DM)	23.5	14.5	6.4	67.0	nd ^d	nd	nd	nd
	ADIN ^e (g N/kg DM)	1.5	0.7	-0.1	4.6	nd	nd	nd	nd
Silage quality parameters ^*
	pH	5.01	0.63	4.00	6.67	nd	nd	nd	nd
	Ammonia fraction	0.07	0.02	0.03	0.13	nd	nd	nd	nd

# Chemical composition of grass silages used in this experiment was determined on freeze dried material;

* Silage quality parameters were determined on air dried material.

a The data were compiled from different in situ experiments that were performed 17 to 50 years ago for different purposes.

b Standard deviation.

c Neutral detergent insoluble nitrogen.

d Not determined.

e Acid detergent insoluble nitrogen.

There was a large range in the values for rumen degradable fractions of DM, OM, CP and NDF of the grass silage samples for each incubation period (Table 3). The range in the values for rumen degradable fractions of DM, OM and CP was larger in the case of short (2, 4, 8, and 16 h) compared to longer rumen incubation periods (32, 72 and 336 h). In case of NDF, the range in the rumen degradable fraction was, however, larger after 16, 32, 72 and 336 h. The reference samples were used to see the difference in the apparent rumen degradation between the series. There was a small difference between the series in the apparent degradation of the chemical components of reference samples. The series effect was not statistically significant and therefore series were not corrected. The mean and range in values for W, U, D, k _d and ED of DM, OM, CP and NDF are shown in Table 4. The average values for the W fraction of DM (W _DM), OM (W _OM), CP (W _CP) and NDF (W _NDF) were 0.240, 0.205, 0.405 and 0.007, respectively. The REP fraction of the grass silages ranged from 0.271 to 0.522, with an average value of 0.390 (Table 4). The lag time for NDF ranged from 3.3 to 8.3 h with an average value of 4.9 h (Table 4). Negative relationships were found between the NDF content and the ED_DM (R² = 0.20) and ED_CP (R² = 0.25) of the grass silage samples (Figs. 1 and 2) whereas a positive linear relationship (R²= 0.10) was found between the CP content and the ED_CP of the grass silages (Fig. 3).

Fig. 1 Relationship between neutral detergent fibre (g/kg DM) and effective rumen degradable fraction of dry matter of grass silages.

Fig. 2 Relationship between neutral detergent fibre (g/kg DM) and effective rumen degradable fraction of crude protein of grass silages.

Fig. 3 Relationship between crude protein (g/kg DM) and effective rumen degradable fraction of crude protein of grass silages.

Table 3 Average rumen degradable fractions of dry matter, organic matter, crude protein and neutral detergent fibre of grass silages (n = 69) incubated for different time periods (2, 4, 8, 16, 32, 72 and 336 h).

Time periods (h)	Dry matter				Organic matter				Crude protein				Neutral detergent fibre (NDF)
	Mean	SD ^a	Minimum	Maximum	Mean	SD	Minimum	Maximum	Mean	SD	Minimum	Maximum	Mean	SD	Minimum	Maximum
2	0.306	0.067	0.158	0.505	0.273	0.064	0.132	0.479	0.522	0.087	0.343	0.727	-0.001	0.043	-0.094	0.087
4	0.323	0.070	0.156	0.523	0.292	0.065	0.143	0.493	0.531	0.091	0.342	0.723	0.020	0.045	-0.081	0.115
8	0.389	0.076	0.226	0.564	0.367	0.074	0.209	0.547	0.560	0.090	0.341	0.743	0.129	0.070	-0.015	0.443
16	0.528	0.075	0.369	0.715	0.514	0.074	0.338	0.702	0.662	0.082	0.454	0.830	0.326	0.074	0.180	0.512
32	0.698	0.066	0.544	0.838	0.701	0.060	0.544	0.831	0.740	0.071	0.547	0.891	0.605	0.066	0.442	0.728
72	0.797	0.054	0.660	0.877	0.801	0.046	0.658	0.881	0.773	0.070	0.576	0.894	0.764	0.052	0.624	0.863
336	0.841	0.036	0.742	0.916	0.847	0.031	0.761	0.930	0.777	0.065	0.539	0.878	0.846	0.031	0.738	0.905

a Standard deviation.

Table 4 The average, minimum and maximum dry matter, organic matter, crude protein and neutral detergent fiber values for the washout fraction ^a (W), rumen undegradable fraction ^b (U), potentially rumen degradable fraction ^c (D), degradation rate (k _d, h⁻¹) and effective degradability (ED) ^d of 69 grass silages.

Variable	Mean	SD ^e	Minimum	Maximum
Dry matter (DM)
W DM	0.240	0.062	0.101	0.412
U DM	0.159	0.036	0.084	0.258
D DM	0.601	0.054	0.469	0.742
k d	0.043	0.007	0.030	0.064
EDDM2	0.648	0.048	0.528	0.750
EDDM5	0.516	0.055	0.380	0.647
EDDM8	0.449	0.057	0.314	0.592
Organic matter (OM)
W OM	0.205	0.061	0.086	0.389
U OM	0.153	0.031	0.070	0.239
D OM	0.642	0.056	0.498	0.780
k d	0.043	0.007	0.029	0.062
EDOM2	0.640	0.047	0.507	0.760
EDOM5	0.499	0.056	0.351	0.646
EDOM8	0.427	0.058	0.281	0.588
Crude protein (CP)
W CP	0.405	0.087	0.205	0.598
U CP	0.222	0.065	0.122	0.461
D CP	0.373	0.103	0.143	0.584
k d	0.068	0.025	0.024	0.136
EDCP	0.610	0.061	0.478	0.729
REP ^f	0.390	0.061	0.271	0.522
Neutral detergent fibre (NDF)
W NDF	0.007	0.046	0.000	0.090
U NDF	0.154	0.031	0.095	0.262
D NDF	0.839	0.054	0.746	0.951
k d	0.046	0.009	0.022	0.083
Lag time (h)	4.892	1.070	3.320	8.280
EDNDF	0.545	0.046	0.386	0.647

a The fraction disappeared by washing with tap water in a washing machine at 25 °C for 40 min.

b The residual fraction determined after 336 h incubation in the rumen.

c The D fraction was calculated as D = 1-(W + U).

d The ED of DM was calculated by using three passage rates; 0.02 h⁻¹(ED_DM2), 0.05 h⁻¹ (ED_DM5) and 0.08 h⁻¹ (ED_DM8). The ED of OM was also calculated by using three passage rates; 0.02 h⁻¹(ED_OM2), 0.05 h⁻¹ (ED_OM5) and 0.08 h⁻¹ (ED_OM8). The ED_CP was calculated using 0.045 h⁻¹ value of passage rate while the ED_NDF was calculated using 0.025 h⁻¹ value of passage rate.

e Standard deviation.

f Rumen escape protein. 5% of W was added in REP.

The results of the multiple linear regression analyses between the rumen degradation characteristics of the DM, OM, CP and NDF on the one hand and the chemical composition of the grass silages on the other hand are reported in Table 5. The W _DM, U _DM, D _DM, ED_DM2, ED_DM5 and ED_DM8 were influenced by the contents of ash and NDF while the W _OM, U _OM, D _OM, ED_OM2, ED_OM5 and ED_OM8 were mainly influenced by the contents of NDF and ADF in the grass silages. The W _CP, U _CP, D _CP and ED_CP were influenced by the contents of DM, CP, sugar, NDF and ADF in the grass silages. The relationship between REP and the chemical composition was explained by the DM, CP and sugar contents in grass silages (Table 5). The D _NDF and ED_NDF were influenced by the content of DM and ADF while the U _NDF fraction was influenced by the contents of DM and ADL.

Table 5 Relationship between the washable fraction (W), the rumen undegradable fraction (U), the potentially rumen degradable fraction (D), and the effective rumen degradable fraction (ED) of dry matter (DM), organic matter (OM), crude protein (CP) and neutral detergent fibre (NDF), and the chemical composition of grass silages.

Regression equation	R^2	RMSE ^a
Dry matter
W DM = 0.76 (± 0.09) - 0.76 (± 0.37) ash - 0.35 (± 0.16) sugar - 0.83 (± 0.13) NDF	0.40	0.05
U DM= - 0.03 (± 0.01) + 0.62 (± 0.22) ash + 0.47 (± 0.14) ADF ^b	0.23	0.03
D DM = 0.28 (± 0.06) + 0.37 (± 0.14) sugar + 0.59 (± 0.11) NDF	0.30	0.05
EDDM2 = 0.94 (± 0.06) - 0.86 (± 0.29) ash - 0.42 (± 0.10) NDF	0.26	0.04
EDDM5 = 0.86 (± 0.07) - 0.83 (± 0.32) ash - 0.52 (± 0.11) NDF	0.27	0.05
EDDM8 = 0.81 (± 0.07) - 0.77 (± 0.33) ash - 0.57 (± 0.11) NDF	0.29	0.05
Organic matter
W OM = 0.64 (± 0.04) - 0.89 (± 0.09) NDF	0.57	0.04
U OM = 0.25 (± 0.02) - 0.52 (± 0.09) CP - 0.40 (± 0.06) sugar + 1.46 (± 0.40) NDF	0.57	0.02
D OM = 0.38 (± 0.06) + 0.53 (± 0.11) NDF	0.25	0.05
k d-OM = 0.08 (± 0.01) - 0.14 (± 0.03) ADF	0.30	0.01
EDOM2 = 1.04 (± 0.03) - 0.41 (± 0.09) NDF - 0.72 (± 0.16) ADF	0.75	0.02
EDOM5 = 0.96 (± 0.03) - 0.58 (± 0.10) NDF - 0.67 (± 0.20) ADF	0.74	0.03
EDOM8 = 0.91 (± 0.04) - 0.66 (± 0.11) NDF - 0.59 (± 0.21) ADF	0.73	0.03
Crude protein
W CP = 0.83 (± 0.08) - 0.31 (± 0.08) DM - 1.15 (± 0.28) NDF + 1.15 (± 0.47) ADF	0.56	0.06
U CP = 0.57 (± 0.04) - 1.74 (± 0.19) CP - 0.62 (± 0.12) sugar	0.59	0.04
D CP= - 0.08 (± 0.07) + 0.52 (± 0.08) DM + 1.18 (± 0.35) CP	0.44	0.08
EDCP = 0.47 (± 0.04) - 0.28 (± 0.06) DM + 1.21 (± 0.23) CP + 0.62 (± 0.15) sugar	0.44	0.05
REP ^c  = 0.53 (± 0.05) + 0.28 (± 0.06) DM - 1.21 (± 0.23) CP - 0.62 (± 0.15) sugar	0.44	0.05
Neutral detergent fibre
U NDF = 0.09 (± 0.03) - 0.29 (± 0.11) DM + 0.52 (± 0.12) ADL ^d	0.21	0.03
D NDF = 0.80 (± 0.01) + 0.45 (± 0.14) NDF	0.13	0.05
k d-NDF = 0.05 (± 0.01) + 0.24 (± 0.03) DM - 0.14 (± 0.01) ADF	0.54	0.01
EDNDF = 0.54 (± 0.01) + 1.83 (± 0.07) DM - 0.99 (± 0.05) ADF	0.91	0.01

a Root mean square error.

b Acid detergent fibre.

c Rumen escape protein. 5% of W was also added in REP.

d Acid detergent lignin.

4 Discussion

The DVE/OEB₂₀₁₀ database consisted of data obtained from different in situ experiments performed at different institutes in the Netherlands and Belgium, 17 to 50 years ago. These studies used different grass species, ensiling methods, fertilization practices and incubation protocols compared to the present study. In addition, different numbers of samples were used to derive prediction formulas for the degradation characteristics of CP (n = 97) and NDF (n = 50). Moreover, different chemical analyses procedures were used across the different in situ experiments and these experiments were performed to investigate other specific scientific hypotheses. In contrast, the database generated in the present study included 69 grass silages which were selected on the base of a broad range in chemical composition and quality parameters and purposely used to derive regression equations for the prediction of the feeding value of grass silages. The same standard incubation protocol, same cows, and same chemical analysis procedures were used for all the grass silages in the new database. In addition, the regression equations were developed to predict the rumen degradation characteristics of DM, OM, CP and NDF of grass silages. The main disadvantage of DVE/OEB₂₀₁₀ database is that it is combined database of grass silage and grass hay samples. The chemical composition of grass silage and grass hay samples is different. This variation in chemical composition due to combination of grass silage and hay samples makes this database unreliable.

The large range in the DM content of the samples in the DVE/OEB₂₀₁₀ database was due to the inclusion of grass silage and grass hay samples, which vary highly in their DM content. The average CP content in the DVE/OEB₂₀₁₀ database was higher than the grass silages used in the present study, which is likely caused by a much higher N fertilization level on grasslands 30 years ago. Nowadays, the N fertilization level is much lower due to effective legislation implemented in 1984 to reduce environmental pollution [21]. The higher average sugar content in the grass silage samples used in the present study might be due to lower N fertilization and lower NDF and CF contents in combination with higher DM content of the samples. The average U _CP fraction (0.222) in the new database obtained in the present study was higher than the U _CP fraction (0.109) in the DVE/OEB₂₀₁₀ database. The higher U _CP fraction might be due to different CP content of samples (170 g/kg DM) used in the present study compared to the samples (198 g/kg DM) used in DVE/OEB₂₀₁₀ database. It also might be due to microbial contamination of the rumen incubated residues. The results of this study were not corrected for microbial contamination. The average k _d value (0.068) of CP was also higher in the new database compared to the k _d value (0.058) of DVE/OEB₂₀₁₀ database. This may be due to the use of only grass silages in the present study. De Boever et al. [2] also concluded that the CP content in grass silage is more degradable in the rumen than that of grass hay and fresh grass due to the fermentation processes in the silos. The average U _NDF fraction (0.154) of grass silages used in the new database was close to the average U _NDF fraction (0.180) of grass silages and grass hay used in the DVE/OEB₂₀₁₀ database but the range in U _NDF fraction was larger (0.088 to 0.383 vs. 0.095 to 0.262) in the DVE/OEB₂₀₁₀ database than the new database. This larger range in the DVE/OEB₂₀₁₀ database may be due to the use of grass silage and grass hay samples together, different incubation protocols, and different number of grass silage samples compared to the new database.

The range in the rumen degradable fractions of DM, OM, CP and NDF of the grass silages after different incubation periods in the rumen was due to the large range in the chemical composition of the grass silage samples. In the present study, the average values for W _DM, W _OM, W _CP and W _NDF fractions were 0.240, 0.205, 0.405 and 0.007, respectively. Gosselink et al. [22] reported a W _DM fraction of 0.250 which is almost similar to the results obtained in the present study. The W _CP fraction reported by Cone at al. [12] was 0.535 for low DM grass silages (± 250 g DM/kg) and 0.408 for high DM grass silages (± 450 g DM/kg). The average U _CP fraction in the present study was 0.222, whereas Cone et al. [12] reported average value of 0.181 for U _CP fraction of low DM grass silages and average value of 0.184 for U _CP fraction of high DM grass silages. The difference in W _CP and U _CP fractions might be due to the high CP content present in the grass silages (208 g/kg DM for low DM grass silages and 209 g/kg DM for high DM grass silages) used by Cone et al. [12] compared to the present study (170 g/kg DM).

The large range in the ED values of DM, OM, CP and NDF was also due to larger range in the chemical composition of the grass silages. In the present study, average ED_DM2, ED_DM5 and ED_DM8 values were 0.648, 0.516 and 0.449, respectively. Jančík et al. [18] reported values of 0.665, 0.539 and 0.478 for ED_DM2, ED_DM5 and ED_DM8, respectively, for grass silages. The difference in the values could be due to the low average ash content (84.5 g/kg DM) of grass silages used by Jančík et al. [18] compared to the ash content (102 g/kg DM) of the grass silages used in the present study and the different number of grass silages (n = 40) studied by Jančík et al. [18]. The average value for ED_NDF in the present study was 0.545, which is lower than the ED_NDF value of 0.660 reported for grass cubes [17]. Lag time varied between the different samples of grass silages. In the present study, there was no relationship found between the lag time and k _d of NDF of grass silages. Varga and Hoover [23] also did not find relationship between the lag time and k _d of NDF. The average ED_CP was 0.610 in the present study which is comparable to the result (0.629) reported by Von Keyserlingk et al. [24]. Castillo et al. [25] reported a lower value (0.571) for ED_CP of grass silages. The higher ED_CP value in the present study might be due to the higher average CP content (170 g/kg DM) in grass silages compared to the grass silages (123 g/kg DM) investigated by Castillo et al. [23]. Cone et al. [12] calculated the REP, using the same REP formula as used in the present study, and reported average values of 0.270 and 0.316, for low and high DM grass silages, respectively. These values are lower than the average REP value of 0.390 obtained in the present study. The difference in REP values is likely to be due to the larger range in CP content (128-305 g/kg DM) in the grass silages used by Cone et al. [12] and the grass silages (108-222 g CP/kg DM) used in the present study. De Boever et al. [2] reported an average REP fraction of 0.244 for grass silages. The low REP fraction in case of De Boever et al. [2] might be due to the variation in CP content of grass silages.

The regression analyses showed that relationships were found between the W, U, D and ED of DM, OM, CP and NDF and the chemical composition of the grass silages (Table 5). The regression equations show that the W _DM and D _DM fractions were influenced by the DM and NDF content of the grass silage samples. The ED_DM at different rumen passage rates was negatively related to the NDF content of the grass silages, indicating that the presence of insoluble fibre restricts the DM degradation. The ED_CP and REP fractions were mainly influenced by the contents of DM, CP and sugar in the grass silages. A similar relationship between the CP content and rumen degradable fraction of CP was reported by Jančík et al. [17] for grass silages. The ED_CP is affected by the CP content in the forage, with a higher CP content resulting in a higher value for ED_CP [26]. The regression analysis showed that the D _CP fraction was mainly influenced by the content of DM and CP while the U _CP was influenced by the contents of CP and sugar in the grass silages. The U _CP fraction was negatively related to the CP content in grass silages. High CP content of grass silage samples means that more N was present inside the cell. That N was readily degradable in the rumen and leaving small fraction for U _CP. The D _NDF fraction was influenced by the NDF content in the grass silages while the U _NDF content was influenced by the contents of DM and ADL in the grass silages. The ED_NDF and k _d-NDF were positively related to DM content in the grass silages. This was related to maturity stage of grass at cut. The new regression equations can be used for the estimation of W, U, D and ED of DM, OM, CP and NDF of the grass silages. The old regression equations were derived for only CP and NDF (Table 6). Information about the R² values of those equations was also missing. A number of regression equations derived from the new database with high R² values can be used for the rapid estimation of rumen degradation characteristics of DM, OM, CP and NDF of the grass silages.

Table 6 Regression equations ^a derived using DVE/OEB₂₀₁₀ database (Personal communication, Product Board Animal Feed, The Netherlands).

Regression equations
Crude protein (CP, %)
% W = 94.35 - 0.10 DM ^b
% U = - 0.03 CP + 0.03 NDF
k d = 38.40 + 0.02 CP - 0.14 NDF + 0.0001 NDF^2
Neutral detergent fibre (NDF, %)
% U = 40.90 - 0.65 DOM ^c  + 0.05 NDF
k d = 6.13 - 0.006 NDF

a These are unpublished regression equation received from the Product Board Animal Feed, the Netherlands. No regression equation was derived for dry matter and organic matter. Information about R² values and root mean square error is not available.

b Dry matter.

c Digestion of organic matter.

5 Conclusions

The new database, consisting of grass silages with a broad range in chemical composition and quality, is more adequate than the old one and more representative for the grass silages used in practice. Regression equations, obtained by relating the in situ degradation characteristics of DM, OM, CP and NDF with the chemical composition of the grass silages, can be used for a rapid estimation of the rumen degradation characteristics of grass silages.

Acknowledgements

The authors are grateful to the Higher Education Commission (HEC, Pakistan), the Dutch Product Board Animal Feed (PDV, The Hague, The Netherlands) and the Dutch Dairy Board (PZ, Zoetermeer, The Netherlands) for financial support and to BLGG AgroXpertus (Wageningen, The Netherlands) for performing the chemical analysis.