Estimation of indigestible NDF in forages and concentrates from cell wall composition

Estimation of indigestible NDF in forages and concentrates from cell wall composition

Animal Feed Science and Technology 177 (2012) 40–51 Contents lists available at SciVerse ScienceDirect Animal Feed Science and Technology journal ho...

453KB Sizes 2 Downloads 58 Views

Animal Feed Science and Technology 177 (2012) 40–51

Contents lists available at SciVerse ScienceDirect

Animal Feed Science and Technology journal homepage: www.elsevier.com/locate/anifeedsci

Estimation of indigestible NDF in forages and concentrates from cell wall composition M. Krämer a,∗ , M.R. Weisbjerg a , P. Lund a , C.S. Jensen b , M.G. Pedersen b a b

Department of Animal Science, Aarhus University, AU Foulum, Blichers Allé 20, P.O. Box 50, DK-8830 Tjele, Denmark Research Division, DLF-Trifolium A/S, Højerupvej 31, DK-4660 Store Heddinge, Denmark

a r t i c l e

i n f o

Article history: Received 19 October 2011 Received in revised form 5 July 2012 Accepted 13 July 2012

Keywords: Fiber Lignin Legume Grass Ruminant INDF

a b s t r a c t This study examined the potential of plant cell wall fractions as predictors of indigestible neutral detergent fibre (INDF) in forages with respect to species within plant type, cut number and stage of maturity (harvest time) within primary growth, and for concentrates with respect to species within plant type, where INDF is defined as the portion of plant cell walls not digested after 288 h rumen incubation in Dacron bags with 12 ␮m pore size. INDF is one of the more important parameters determining the net energy (NE) value of a diet in some recently developed ruminant feed evaluation systems. Effects of maturity and cut number on INDF in three legumes and 18 grasses were determined based on an experiment in which each forage was cut at three times of primary growth and once in each of the following three regrowths. These data were supplemented with data from earlier experiments to develop regression equations for INDF intended for use in practice based on a total of 321 samples. Plant type and species within plant type affected (P<0.001) all cell wall fractions. The INDF/lignin(sa) ratio varied substantially from the 2.4 factor used in the Cornell Net Carbohydrate and Protein System (CNCPS) to predict INDF, averaging 2.6 for legumes, grains and grain byproducts, 2.7 for grasses and 1.0 for oilseeds including byproducts. The INDF/IOM ratio varied less among plant species within plant type than among plant types. Multiple linear regression analysis revealed higher INDF prediction accuracy compared to simple regression equations. Estimation of INDF in DM was more accurate than estimation of INDF in ash-free neutral detergent fibre (aNDFom). Results indicate that universal regression equations among plant types are not possible, and, within plant types, prediction equations are not promising for grasses and legumes if only lignin(sa) is included. Comparing plant types, INDF prediction was best for grains and byproducts in simple linear regression and best for grass/clover in multiple linear regression equations. Variations within and among plant type were related to stage of maturity and cut number. © 2012 Elsevier B.V. All rights reserved.

1. Introduction Indigestible neutral detergent fibre (INDF) is one of the more important chemical and digestibility measurements for the determination of the net energy (NE) value of a diet in some recently developed ruminant feed evaluation systems. Ellis

Abbreviations: aNDFom, NDF assayed with a heat stable amylase and expressed exclusive of residual ash; CNCPS, Cornell Net Carbohydrate and Protein System; DM, dry matter; INDF, indigestible NDF; IOM, indigestibility of OM; IVOMD, in vitro OM digestibility; lignin(pm), lignin determined by oxidation of lignin with permanganate; lignin(sa), lignin determined by solubilisation of cellulose with sulphuric acid; NE, net energy; NIRS, near infrared reflectance spectroscopy; OM, organic matter; OMD, OM digestibility. ∗ Corresponding author. Tel.: +45 8715 7859; fax: +45 8715 4249. E-mail address: [email protected] (M. Krämer). 0377-8401/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.anifeedsci.2012.07.027

M. Krämer et al. / Animal Feed Science and Technology 177 (2012) 40–51

41

et al. (1999) define INDF as an ideal nutritional entity as it is digested at a rate of zero, and it is an input factor in feed evaluation systems such as NorFor (Volden, 2011) and the Cornell Net Carbohydrate and Protein System (CNCPS; Fox et al., 2004). INDF is defined by in situ analysis in the NorFor system (Volden, 2011) and in vitro analysis in the CNCPS (Fox et al., 2004). However, the INDF concentration in both approaches depends on methods of NDF analysis and on the methodological approach in relation to incubation, such as grinding, mill type, bag pore size, incubation time in the rumen, cow, and the ratio between incubated dry matter (DM) and bag surface area. Grinding size, for instance, is a compromise between allowing rumen microbes access to particles and prevention of particle loss. Especially for forages, the digestibility of plant cell walls is of great importance, as cell contents are almost completely digestible, making cell wall concentration and digestibility a key determinant of the NE of a diet. In practice, near infrared reflectance spectroscopy (NIRS) is used in NorFor to determine the INDF concentration in forages based on a calibration developed from ruminal in situ INDF values. As in situ INDF determination is extensive in labour and costs, and depends on the availability of ruminally cannulated animals, reliable and fast laboratory methods are needed to estimate INDF in commonly used ruminant feedstuffs. Lignin(sa) concentration has been used in CNCPS to calculate INDF (Van Soest et al., 1991). Plant species (Rinne et al., 2006), as well as cut number and maturity (Kuoppala et al., 2004) influence the INDF concentration of forages. Legumes have a higher INDF/aNDFom ratio (Rinne et al., 2006) and a different degradation rate of potentially digestible aNDFom compared to grasses (Weisbjerg and Søegaard, 2008) which can be related to morphological and chemical differences (Van Soest, 1994). Huhtanen et al. (2006) emphasized the contradictory results when different lignin analyses were used to predict INDF concentration in different datasets. Preliminary research from our institute on a small dataset (Krämer et al., 2010; Weisbjerg et al., 2010), consistent with Spanghero et al. (2003), indicated that the association between lignin(sa) and INDF concentration is not uniform among plant types and that universal prediction equations are not accurate for estimation of INDF concentration. The objective was to elucidate effects of cut number and maturity within primary growth on INDF concentration of several forages. Further aims were to develop prediction equations for INDF concentration based on chemically defined plant cell wall fractions for forages and concentrate feedstuffs. 2. Material and methods 2.1. Data 2.1.1. Dataset 1 The first dataset included 21 forages of which 13 were perennial ryegrasses, two festuloliums, two hybrid ryegrasses, one cocksfoot, one lucerne, one red clover and one white clover. Forages were sown in plots on a fine Cambisol soil (pH 6.8) at Bredeløkke, Denmark (55◦ 20 N, 12◦ 23 E) on September 4, 2007 and harvested in 2008. Each variety was sown in two fields. The first field was cut four times (June 2; July 2; July 29; September 3, 2008) for primary growth and three regrowths, with a medium harvest time of the primary growth. The second field was divided into two parts, giving the opportunity to cut it twice (May 19; June 9, 2008) in primary growth which represented an early and late primary growth harvest. Samples analysed as “harvest time 2” correspond to samples of primary growth grown on the first field. Plots were cut 6 cm above ground level using a Haldrup plot harvester (Haldrup, Løgstør, Denmark). Sample amounts of 2 to 4 kg of fresh forages were chopped and a part was dried at 60 ◦ C for analysis. Further details on forage management and cultivars are in Bertram et al. (2010). The 126 forage samples were grouped into grasses and legumes in order to compile plant species with similar biological and chemical properties (i.e., changes in morphology and chemically defined plant cell wall fractions during growth and flowering). 2.1.2. Dataset 2 This dataset consists of 321 samples combined from dataset 1 and samples collected since 2000 at AU Foulum, Tjele (Denmark). The dataset included 170 samples for which INDF was estimated according to the method used in NorFor (Volden, 2011) with an endpoint of 288 h rumen incubation (288 h method; including the 21 forages cut 6 times resulting in 126 samples from dataset 1), 127 samples with INDF values analysed using an in situ method ending at 504 h rumen incubation (504 h method) and 24 samples for which INDF was analysed using both methods, as introduction of the NorFor standard changed the 504 h method to the 288 h method in Denmark. Samples for which INDF was analysed by both the 288 and 504 h method were grouped as grasses, legumes, seeds, and grains and related byproducts. Dataset 2 contained samples with information on aNDFom, lignin(sa) and INDF. Data not belonging to dataset 1 which were in dataset 2 covered a wide range of plant types, allowing a grouping into grasses, legumes, oilseeds and their byproducts, grains and their byproducts, and other byproducts. The group “other byproducts” included those samples which could not be allocated to any other group. 2.2. In situ degradability, in vitro digestibility and chemical analysis Samples were dried at 60 ◦ C for 48 h and milled on a 1 mm screen in a hammer mill (Årslev Maskinfabrik, Årslev, Denmark). Analysis of in vitro organic matter (OM) digestibility (IVOMD) followed the method of Tilley and Terry (1963). Samples were

42

M. Krämer et al. / Animal Feed Science and Technology 177 (2012) 40–51

analysed in duplicate, with each duplicate in a separate run. Samples were incubated in rumen fluid for 48 h followed by 48 h digestion in pepsin and HCl according to Tilley and Terry. Residues were combusted to determine IVOMD. Rumen fluid was collected from three non-lactating Danish Holstein cows also used for in situ determination of INDF and fed at a maintenance metabolizable energy level of a diet consisting of grass/clover hay, barley straw and a pelleted concentrate mixture consisting of 40 kg barley grain, 40 kg oat grain, 10 kg soybean meal, 3 kg rapeseed meal, 3 kg sugar beet molasses and 4 kg of a commercial mineral mixture (6 g/100 g Ca, 10 g/100 g P, 12 g/100 g Mg, 5 g/100 g Na; Type 3, Vitfoss, Gråsten, Denmark) per 100 kg fresh mixture. The forage to concentrate ratio was 670:330 on DM basis and the crude protein concentration of the diet 139 g/kg DM. The daily ration was divided into two meals of equal size. Lignin(sa) was determined according to ISO method 13908 (2008) and aNDFom analysis used a FibertecTM 2010 (Foss, Hillerød, Denmark) including amylase treatment according to Mertens et al. (2002). In situ INDF estimation used Åkerlind et al. (2011), where INDF concentration was determined after 288 h rumen incubation in Dacron bags with 12 ␮m pore size after grinding samples on a 1.5 mm screen using a cutter mill (Fritsch, Idar-Oberstein, Germany). Feed samples of 2 g were incubated in the bags. Polyester cloth (Saatifil PES 12/6, Saatitech S.p.A., Como, Italy) with 12 ␮m pore size and 0.06 cm2 /cm2 open surface area was used. A maximum of six bags were mounted on one rubber stopper. Bags were pre-soaked for 20 min in 39 ◦ C warm tap water without agitation prior to rumen incubation. Subsequently, a maximum of nine rubber stoppers were gathered in a household washing net with two 912 g weights. Each feedstuff was incubated in the ventral rumen of three cows and the INDF concentration was reported as the mean of the three. Cows and feeding were similar to the above description for IVOMD analysis. After rumen incubation, each bag on the rubber stopper was rinsed with cold tap water before bags were machine-washed (AEG, Fredericia, Denmark) twice for 5 min with 22 L water at 25 ◦ C. The washing program did not include a spinning cycle. Subsequently, bags were frozen at −20 ◦ C until analysis in which residues were transferred quantitatively to crucibles. INDF concentration using the 504 h method was determined in Dacron bags with 36 ␮m pore size. The method followed the above described steps for determination of INDF concentration after 288 h rumen incubation, but included simultaneous measurement of aNDFom washing losses to correct for initial particle loss (Hvelplund and Weisbjerg, 2000) assuming particle loss of aNDFom to be representative for the whole aNDFom fraction incubated. 2.3. Calculations and statistical analysis All calculations and statistical analyses used the software package R (11-07-2011). All calculations on IVOMD were on a sub-dataset as IVOMD was only available for a total of 233 samples in dataset 2. 2.3.1. Dataset 1 Effects of forage type and species within forage type on chemical and digestibility measurements were analysed using a general linear model including the factors forage type and forage species nested within forage type. As there were differences (P<0.05) among forage types, data for legumes and grasses were analysed separately. Ratios between INDF concentration in DM or in aNDFom and lignin(sa) concentration as well as between INDF and indigestibility of OM (IOM; IOM = 1000 − IVOMD in g/kg OM) were calculated within forage type. Two general linear models were used. The first model included the factors harvest time for primary growth and forage species. The second model included the two factors cut number and forage species. The models were: Yij =  + ˛i + ˇj + ˛ˇij + εij where Yij = cell wall fraction (e.g., INDF, lignin(sa) concentrations in aNDFom or in DM, IOM),  = overall mean, ˛i = effect of harvest time (i = 1 to 3 within primary growth (model 1)) or cut number (i = 1 to 4 (model 2)), ˇj = effect of forage species (j = 1 to 3 for legumes; j = 1 to 18 for grasses), ˛ˇij = interaction between harvest time within primary growth and forage species (model 1) or between cut number (model 2) and forage species, εij = residual error assumed to be normally distributed. Differences between each of the species, cut numbers or harvest times within the primary growth for grasses and legumes were tested using orthogonal contrasts. 2.3.2. Dataset 2 In order to include the samples with INDF concentration determined with the 504 h method, the 24 measurements with measured INDF concentration according to both 288 and 504 h methods were analysed using two-way ANOVA. Two models were used of which the first included effects of plant type and incubation time whereas the second included effects of plant species and incubation time. The models were: Yij =  + ˛i + ˇj + ˛ˇij + εij where Yij = aNDFom concentration in DM degraded,  = overall mean, ˛i = effect of plant type (i = 1 to 4, grains and byproducts, grasses, legumes, seeds; model 1) or plant species (i = 1 to 11; model 2), ˇj = effect of method (j = 288 or 504 h method), ˛ˇij = interaction between plant type and incubation time (model 1) or plant species and incubation time (model 2), εij = residual error assumed to be normally distributed.

M. Krämer et al. / Animal Feed Science and Technology 177 (2012) 40–51

43

Effects of plant type and plant species within plant type on plant cell wall characteristics were tested using two-way ANOVA based on the model: Yij =  + ˛i + ˇj + εij where Yij = cell wall fraction (e.g., INDF, lignin(sa) concentrations in aNDFom or in DM, IOM),  = overall mean, ˛i = effect of plant type (i = 1 to 6, grasses, grass/clover, legumes, grains and byproducts, oilseeds and byproducts, other byproducts), ˇj = effect of plant species within plant type (j = variable numbers, see Tables 3 and 4), εij = residual error assumed to be normally distributed. Differences between each plant type and between plant types among forages and plant types among concentrates were analysed with orthogonal contrasts. The heterogeneous group of “other byproducts” was omitted from contrast analysis, since it contained feedstuff samples that could not be allocated to any of the other concentrate types. Simple linear regressions were completed between INDF concentration in DM or in aNDFom and each of the explanatory cell wall fractions such as lignin(sa) concentration in DM and lignin(sa) concentration in aNDFom. Stepwise multiple linear regression analyses with aNDFom and lignin(sa) concentrations in DM and in aNDFom as explanatory variables in the model were completed for INDF concentration in aNDFom or in DM, for the full dataset and within plant type. For stepwise multiple linear regression analyses, a backwards selection procedure was chosen. The criterion used was Akaike’s information criterion. For simple regression analyses, differences among means with P<0.05 were accepted as statistically significant and differences with 0.05
44

M. Krämer et al. / Animal Feed Science and Technology 177 (2012) 40–51

Table 1 Plant cell wall fractions and in vitro indigestibility of organic matter and ratios between those for grasses and legumes in the primary growth (Cut number 1) and three regrowths (Cut numbers 2 – 4) in dataset 1.

Grasses Cut number 1 2 3 4 Pe Cut numberf Speciesg Cut × Species RSE Legumes Cut number 1 2 3 4 P Cut numberh Speciesi RSE

na

INDF (g/kg aNDFom)

aNDFom (g/kg DM)

18 18 18 18

156.8 79.0 93.2 71.9

501.4 436.0 487.8 474.4

<0.001 0.064 0.294 1.95

3 3 3 3

46.5 35.3 34.3 28.3

<0.001 <0.001 0.419 2.16

247.0 273.5 255.2 148.8

283.2 344.8 353.7 353.6

0.133 0.665 5.83

lignin(sa)b (g/kg aNDFom)

292.7 256.5 284.0 274.4

<0.001 <0.001 0.580 0.64

<0.001 <0.001 0.210 2.11

89.0 130.3 113.6 79.0

0.003 <0.001 1.45

IOM (g/kg OM)c

234.6 313.5 317.0 301.1

0.271 0.064 3.13

0.037 0.405 2.85

INDF/lignin(sa)

3.4 2.3 2.8 2.7 <0.001 0.012 0.921 0.52

INDF/IOMd

0.5 0.3 0.3 0.3 <0.001 0.562 0.147 0.04

2.8 2.3 2.3 2.0

1.1 0.9 0.8 0.5

0.122 0.011 0.36

0.028 0.369 0.16

Data for the primary growth are from the medium harvest time (i.e., harvest time 2). Effect of interaction between cut number and species for legumes could not be tested due to missing degrees of freedom. a Number of samples. b Lignin determined by solubilisation of cellulose with sulphuric acid. c In vitro indigestibility of organic matter in g/kg OM (Tilley and Terry, 1963). d INDF in g/kg aNDFom; IOM in g/kg OM. e Significance on respective plant cell wall parameter. f Cut numbers 2 and 4 did not differ, but they differed from 1 and 3, 1 differed from 3 (P<0.05) in INDF; cut numbers 3 and 4 did not differ, but they differed from 1 and 2, 1 differed from 2 in aNDFom and in INDF/lignin(sa); cut numbers 2 and 3 did not differ, but they differed from 1 and 4, 1 differed from 4 in lignin(sa) and in INDF/IOM; cut numbers 1, 3 and 4 did not differ from each other but they differed from 2 in IOM. g All 4 species differed (P<0.05) from each other in aNDFom; no differences between perennial ryegrass and hybrid ryegrass, no differences between festulolium and cocksfoot but the first two differed from the other two species in IOM; perennial ryegrass differed from the other species, but hybrid ryegrass, festulolium and cocksfoot did not differ from each other in INDF/lignin(sa). h Cut number 1 differed (P<0.05) from the others, but 2, 3 and 4 did not differ from each other in aNDFom and in IOM; cut numbers 2 and 3 did not differ, but they differed from 1 and 4, cut numbers 1 and 4 differed in INDF/IOM. i Differences (P<0.05) in aNDFom and in INDF/lignin(sa) between the species lucerne and white clover, between white clover and red clover, but no differences between lucerne and red clover.

Later harvest time within primary growth affected cell wall parameter concentration more than regrowth number. There was a linear correlation between lignin(sa) and INDF concentrations (both in DM) across harvest time in primary growth (RSE = 18.9 g/kg; Fig. 1) and a linear but much lower correlation among cut number (RSE = 7.1 g/kg; Fig. 1). Comparing IOM with INDF concentration in DM for perennial ryegrass revealed that chemically defined cell wall contents in primary growth were distinct from any of the subsequent regrowths (Fig. 2). 160

140

INDF (g/kg DM)

120 Primary growth, 1. Harvest

100

Primary growth, 2. Harvest Primary growth, 3. Harvest

80

1. Regrowth 2. Regrowth

60

3. Regrowth Linear (primary growth)

40

Linear (regrowths) 20 5

15

25 lignin(sa) (g/kg DM)

35

45

Fig. 1. Regression of INDF (y) in g/kg DM to lignin(sa) (x) in g/kg DM for perennial ryegrass harvested at three stages of maturity in the primary growth and once in each regrowth. Regression equation among harvest times in primary growth, where y = 3.77 x − 15.8 (slope: SE = 0.39; P<0.001; intercept: SE = 9.54; P=0.106; RSE = 18.9 g/kg). Regression equation among regrowths: y = 1.21 x + 18.5 (slope: SE = 0.41; P=0.005; intercept: SE = 6.31; P=0.006; RSE = 7.13 g/kg).

M. Krämer et al. / Animal Feed Science and Technology 177 (2012) 40–51

45

Table 2 Plant cell wall fractions and in vitro indigestibility of organic matter for grasses and legumes cut in different stages of maturity (harvest times 1, 2 and 3) in the primary growth in dataset 1. na Grasses Harvest time 1 18 2 18 3 18 Pe Harvest timef Speciesg Harvest × Species RSE Legumes Harvest time 1 2 3 P Harvest timeh Speciesi RSE

3 3 3

INDF (g/kg aNDFom)

aNDFom (g/kg DM)

100.3 156.8 210.1

407.5 501.4 527.8

<0.001 0.008 0.999 3.40

lignin(sa)b (g/kg aNDFom)

43.9 46.5 57.7

<0.001 <0.001 0.649 2.93

241.3 283.2 313.7

238.2 247.0 341.3 0.032 0.006 3.27

194.1 292.7 331.3

<0.001 0.646 0.738 1.06

<0.001 0.016 0.883 3.06

113.3 89.0 127.0

0.012 0.001 1.56

IOM (g/kg OM)c

222.0 234.6 281.0

0.093 0.060 1.56

0.107 0.022 2.65

INDF/lignin(sa)

2.4 3.4 3.7 <0.001 0.035 0.366 0.52

INDF/IOMd

0.5 0.5 0.6 <0.001 0.001 0.968 0.07

2.2 2.8 2.7

1.0 1.1 1.2

0.166 0.007 0.32

0.244 0.036 0.10

Data shown are means across plant species. Effect of interaction between harvest time and species for legumes could not be tested due to missing degrees of freedom. a Number of samples. b Lignin determined by solubilisation of cellulose with sulphuric acid. c In vitro indigestibility of organic matter in g/kg OM (Tilley and Terry, 1963). d INDF in g/kg aNDFom; IOM in g/kg OM. e Significance on respective cell wall parameter. f Differences (P<0.05) in INDF, in IOM and in INDF/IOM between each of the harvest times 1, 2 and 3; no differences between harvest times 1 and 2 but they differed from 3 in aNDFom and in lignin(sa); harvest times 2 and 3 did not differ, but each differed from harvest time 1 in INDF/lignin(sa); no differences between harvest times 1 and 2, but 1 and 2 differed from 3 in INDF/IOM. g Hybrid ryegrass, festulolium and cocksfoot did not differ from each other, but they differed (P<0.05) from perennial ryegrass in INDF, in IOM, in INDF/lignin(sa) and in INDF/IOM; in aNDFom no differences between festulolium and cocksfoot, no differences between perennial ryegrass and hybrid ryegrass, but the first two differed from the other two. h Harvest times 2 and 3 did not differ, but they differed (P<0.05) from 1 in INDF and in aNDFom. i Lucerne and red clover did not differ, but differed (P<0.05) from white clover in INDF and in aNDFom, in IOM and in INDF/lignin(sa); no differences between white clover and red clover, but these differed from lucerne in INDF/IOM.

3.1.2.2. Legumes. Higher INDF and aNDFom concentrations among legumes occurred in the second harvest than in the first, but there were no differences between the second and third harvest in primary growth (Table 2). No changes in lignin(sa) or IOM concentrations, INDF/lignin(sa) or INDF/IOM ratios occured. INDF, aNDFom, IOM concentrations and INDF/lignin(sa) ratio were similar between lucerne and red clover, but different between these two and white clover. In contrast, white clover and red clover had similar INDF/IOM ratios which differed from lucerne (Table 2). The INDF/lignin(sa) ratio varied 27% and the INDF/IOM ratio 15% among harvest times.

160 140

INDF (g/kg DM)

120

Primary growth, 1. Harvest Primary growth, 2. Harvest

100

Primary growth, 3. Harvest 1. Regrowth

80

2. Regrowth 60

3. Regrowth Expon. (primary growth)

40

Linear (regrowths) 20 150

200

250 300 IOM (g/kg OM)

350

400

Fig. 2. Regression of INDF (y) in g/kg DM to indigestibility of organic matter (IOM) (x) in g/kg organic matter (OM) for perennial ryegrass in three maturity stages in the primary growth and in three regrowths. Regression equation among harvest times in primary growth, where y = 8.323 e (0.0076 x) + 0.94 (slope: SE = 0.25; P<0.001; intercept: SE = 2.35; P=0.691; RSE = 6.71 g/kg). Regression equation among regrowths, where y = 0.249 x − 29.5 (slope: SE = 0.05; P<0.001; intercept: SE = 12.6; P=0.025; RSE = 6.06 g/kg).

46

M. Krämer et al. / Animal Feed Science and Technology 177 (2012) 40–51

aNDFom degraded (g/kg aNDFom) after 288 h

1000 Grains + Byproducts Grasses

900

Legumes Seeds

800

y=x Linear regression

700

600

500

400 400

500 600 700 800 900 aNDFom degraded (g/kg aNDFom) after 504 h

1000

Fig. 3. Regression of aNDFom degraded (g/kg aNDFom) determined by the 288 h method (y) on aNDFom degraded (g/kg aNDFom) determined by the 504 h method (x). The linear regression equation is y = 0.997 x − 15.5 (slope: SE = 0.04; P<0.001; intercept: SE = 32.4; P=0.637; RSE = 28.9 g/kg).

Lucerne and red clover had similar chemically defined cell wall contents (e.g., aNDFom among cut number and INDF, aNDFom among harvest times), IOM among harvest times and INDF/lignin(sa) ratio among both cut numbers and harvest times, which were each distinct from white clover (Tables 1 and 2).

3.2. Dataset 2 Effect of method used for INDF determination was examined on 24 feedstuff samples including grains and byproducts, grasses, legumes and seeds. The INDF concentrations determined by the 288 and 504 h methods differed in bag pore size (12 ␮m versus 36 ␮m), bag surface area relative to sample weight, and correction for immediate washing losses was only included in the 504 h INDF concentration calculation. However, the slope did not differ from 1 (P=0.64; Fig. 3) when INDF concentrations determined by the two methods were compared. This regression was nevertheless used to estimate degradability after 288 h for those feedstuff samples for which only the 504 h INDF concentration was available. Dataset 2 was analysed groupwise by plant type, as it had effects (P<0.05) on all cell wall fraction levels and the INDF/IOM ratio. Plant species within plant type affected INDF, lignin(sa), IOM concentrations and the INDF/IOM ratio (P<0.05). Among forages, forage type influenced all cell wall fraction levels and the INDF/lignin(sa) and INDF/IOM ratios, whereas there were no differences in IOM between forage types (Table 3). The aNDFom concentration and the INDF/lignin(sa) ratio were highest in grasses, which also contained the lowest lignin(sa) concentration and INDF/IOM ratio, and were highest in legumes and intermediate in grass/clover (P<0.05; Table 3). The highest INDF concentration was in legumes and the lowest in grass/clover. Among concentrates, lignin(sa) concentration was highest (P<0.05) in oilseeds and their byproducts (290 g/kg aNDFom). Average INDF/lignin(sa) ratio was higher (P<0.05) in grains and byproducts (2.6 g/g DM) than in oilseeds and byproducts (1.0 g/g DM), a value similar to that of grasses and legumes. There were no differences between concentrate types in INDF and aNDFom concentrations (Table 4). Simple linear regression analysis showed a better fit of chemically defined plant cell wall concentrations and INDF concentration in DM than INDF in aNDFom. Best-fit models for INDF concentration in DM and related measurements for each plant type and among plant types are in Table 5. Regression equations for concentrates (grains and byproducts, oilseeds and byproducts) are based on a limited number of samples. For grasses, legumes, oilseeds and byproducts, lignin(sa) concentration in DM was the most important predictor of INDF concentration in simple linear regression analysis in the dataset (Table 5). The aNDFom concentration was the best prediction variable for INDF concentration in DM in grass/clover and grains including the respective byproducts. Using simple linear regression analysis (Table 5), the linear regression equation for grains and byproducts fitted best (RSE = 17.3 g/kg) comparing the datasets. Table 6 shows multiple linear regression equations for prediction of INDF concentration in DM from chemical measurements, resulting from the stepwise procedure. Multiple regression analysis was completed to consider combined effects of cell wall fractions.

Table 3 Means of plant cell wall fractions and in vitro organic matter indigestibility of the forage samples grouped according to morphological and biological similarities into grasses and legumes and ratios between them in dataset 2. na

186 (152) 4 11 6 (6) 18 (18) 25 (25) 4 12 (12) 4 (4) 3 89 (87) 10 41 (30) 19 (15) 6 16 (15) 56 (51) 21 (19) 5 (3) 16 (15) 14 (14)

aNDFom (g/kg DM)

lignin(sa)b (g/kg aNDFom)

IOM (g/kg OM)c

INDF/lignin(sa)

INDF/IOMd

Mean

SEM

Mean

SEM

Mean

SEM

Mean

Mean

SEM

Mean

SEM

151.31 185.5 205.3 131.2 153.7 182.0 175.7 129.3 213.5 199.8 121.7 227.1 150.32 139.2 206.8 142.3 333.83 416.6 375.6 282.4 253.5

5.52 48.1 14.3 23.4 19.5 13.0 27.5 18.6 35.1 5.10 7.35 9.02 9.27 15.8 6.62 11.4 16.2 26.1 41.9 20.3 27.6

468.61 520.1 406.7 545.3 503.1 467.1 502.5 474.4 419.9 399.5 473.6 396.6 422.12 340.4 637.7 438.3 334.33 378.5 326.2 337.5 268.0

4.81 30.2 8.76 21.7 14.3 14.7 56.5 12.2 21.4 17.1 6.29 4.93 19.4 16.6 19.4 20.6 9.98 15.2 25.3 15.3 18.1

56.91 60.6 85.2 46.1 43.6 82.2 96.7 41.4 59.5 79.3 42.7 97.7 76.92 72.1 113.0 69.1 131.53 144.0 131.9 100.2 144.3

1.99 7.77 2.54 2.97 3.63 3.07 7.04 3.92 7.71 10.6 5.84 2.88 5.28 6.40 14.1 8.29 5.52 6.72 8.36 54.8 2.29

294.21

2.71 2.9 2.4 2.8 3.4 2.2 1.9 3.0 3.5 2.6 2.8 2.3 2.02 1.9 2.0 2.2 2.63 2.9 2.8 3.0 1.8

0.06 0.48 0.12 0.37 0.22 0.10 0.40 0.17 0.14 0.31 0.08 0.12 0.09 0.13 0.19 0.15 0.09 0.10 0.21 0.17 0.08

0.41

0.01

0.4 0.5 0.5

0.06 0.04 0.02

0.5 0.6

0.06 0.07

0.4

0.01

0.52 0.5

0.03 0.04

0.5 1.13 1.2 1.2 1.0 1.0

0.04 0.04 0.06 0.10 0.06 0.06

SEM 6.32

297.3 302.6 386.4

21.2 16.4 19.9

276.5 335.0

13.9 23.3

266.4

5.88

276.31 231.1

18.3 18.5

321.5 299.11 336.9 359.7 284.2 250.8

27.3 10.2 15.6 30.9 14.4 17.7

Different superscripts within the same column indicate that groups differ (P<0.05). Samples are fresh, unensiled if not otherwise indicated. a Number of samples, values in brackets indicate number of samples in the sub-dataset used for calculation of IOM. b Lignin determined by solubilisation of cellulose with sulphuric acid. c In vitro indigestibility of organic matter in g/kg OM (Tilley and Terry, 1963), calculated from a sub-dataset. d INDF in g/kg aNDFom, IOM in g/kg OM. e Unspecified pasture grass samples.

M. Krämer et al. / Animal Feed Science and Technology 177 (2012) 40–51

Grasses Barley, green Barley, whole crop Cocksfoot Festulolium Grasse Grass, silagee Ryegrass, hybrid Maize, whole crop Maize, silage Ryegrass, perennial Wheat, whole crop Grass/clover Grass/clover Grass/clover, hay Grass/clover, silage Legumes Lucerne Pea, whole crop Clover, red Clover, white

INDF (g/kg aNDFom)

47

48

M. Krämer et al. / Animal Feed Science and Technology 177 (2012) 40–51

Table 4 Means of plant cell wall fractions and in vitro organic matter indigestibility of the concentrate samples grouped according to morphological and biological characteristics and ratios between them in dataset 2. na

Grains and Byproducts Barley, grain Corn, distillers grain Maize, dried Oat, grain Rye, grain Sorghum, grain Wheat, grain Wheat, bran Oilseeds and Byproducts Coconut, cakec Cotton seed, cake Palm kernel, cake Rapeseed, cake Rapeseed, meald Soybean, meal Soypasse Sunflower, cake Other Byproducts Sugar beet pulp, dried Guar, meal Soybean, hulls

15 2 1 4 1 1 2 3 1 15 1 2 1 2 1 5 1 2 5 2 1 2

INDF (g/kg aNDFom)

aNDFom (g/kg DM)

lignin(sa)b (g/kg aNDFom)

INDF/lignin(sa)

Mean

SEM

Mean

Mean

SEM

Mean

SEM

240.31 320.9 184.4 146.6 343.4 328.4 260.8 239.4 281.4 303.71 248.4 353.0 246.4 555.4 571.4 96.7 95.4 546.4 78.2 125.9 69.4 34.9

20.2 6.50

226.11 178.0 353.0 248.4 286.0 154.0 360.2 60.0 349.0 255.01 426.0 257.0 563.0 248.0 234.0 136.5 336.0 286.5 478.4 433.0 256.0 635.0

147.31 67.6 327.1 63.7 117.3 100.9 459.2 82.6 107.7 290.22 201.1 371.2 187.6 563.7 516.5 168.0 170.2 284.3 95.8 163.0 50.2 51.3

36.9 13.9

2.61 5.0 0.6 2.3 2.9 3.3 0.6 3.0 2.6 1.02 1.2 1.0 1.3 1.0 1.1 0.7 0.6 1.9 0.9 0.8 1.4 0.7

0.38 1.12

20.9

9.92 35.7 53.8 18.5 6.00 20.5 84.0 20.6 5.50 6.50

SEM 29.6 2.00 32.2

30.2 33.0 102 18.0 14.2 4.50 72.4 23.0 25.0

28.9

4.23 42.2 62.9 1.7 20.0 9.56 29.4 43.4 4.96

0.35

0.03 0.56 0.13 0.22 0.11 0.18 0.16 0.14 0.13 0.10

Different superscripts within the same column indicate that groups differ (P<0.05). Other byproducts were omitted from the comparison due to the inhomogeneity of this group. a Number of samples. b Lignin determined by solubilisation of cellulose with sulphuric acid. c Byproduct after oil extraction by pressure. d Byproduct after solvent oil extraction. e Xylose and heat treated soybean meal.

Table 5 Simple linear regressions for prediction of INDF concentration in DM and chemical measurements in dataset 2. na

All Grasses Grass/clover Legumes Grains + Byproducts Oilseeds + Byproducts

321 187 41 58 15 15

Parameter estimates Intercept

SE

P

Slope

SE

P

RSE

35.39

3.702

<0.001

11.811

<0.001

1.28 lignin(sa)b 2.73 lignin(sa) 0.26 aNDFom 2.67 lignin(sa) 0.24 aNDFom 1.07 lignin(sa)

0.087 0.061 0.027 0.086 0.018 0.101

<0.001 <0.001 <0.001 <0.001 <0.001 <0.001

39.35 24.6 21.1 31.9 17.3 35.4

c

−43.42 c c c

Values are given in g/kg DM, if not otherwise indicated. a Number of samples. b Lignin determined by solubilisation of cellulose with sulphuric acid. c Intercept not different from zero; model without intercept used.

Table 6 Multiple linear regressions for prediction of INDF concentration in DM based on chemical measurements.

All Grasses Grass/clover Legumes Grains + Byproducts Oilseeds + Byproducts

na

Regression equation

RSE

321 187 41 58 15 15

30.56 + 2.48 lignin(sa) − 0.44 lignin(sa)b −46.66 + 0.12 aNDFom + 3.99 lignin(sa) − 0.75 lignin(sa)b −93.58 + 0.30 aNDFom − 1.04 lignin(sa) + 0.909 lignin(sa)b 169.81 − 0.38 aNDFom + 7.47 lignin(sa) − 1.95 lignin(sa)b 0.50 + 0.24 aNDFom −48.86 + 0.27 aNDFom + 0.22 lignin(sa)b

35.8 19.6 16.3 22.2 18.0 36.2

Values are given in g/kg DM if not otherwise indicated. a Number of samples. b Lignin determined by solubilisation of cellulose with sulphuric acid, in g/kg aNDFom.

M. Krämer et al. / Animal Feed Science and Technology 177 (2012) 40–51

49

4. Discussion 4.1. Effect of method used to determine INDF concentration There are numerous methods to determine INDF, despite that INDF is one of the more important parameters for the determination of the NE value of ruminant feeds in many modern feed evaluation programs (e.g., CNCPS, NorFor). The regression equation comparing aNDFom not degraded by the 288 h versus that degraded using the 504 h method showed that a similar proportion of aNDFom was degraded in both. In situ and in vitro methods used to estimate INDF have been shown to differ (Huhtanen et al., 2006). An in vitro method has been used in the United States (Mertens, 1993), but an in situ method is the basis for estimation of INDF in NorFor (Åkerlind et al., 2011). Bossen et al. (2008) compared INDF concentrations estimated by two in vitro approaches, one at pH 6 and the other at pH 6.8 and an in situ method, and found that differences in aNDFom residue after in situ or in vitro incubations were due to technical and biological conditions during incubation in nylon bags. The in situ method yielded lower INDF than in vitro methods which the authors partly explained by particle losses during incubation in nylon bags. Methods for INDF analysis mimic in vivo NDF digestion to some degree. In situ INDF analysis requires nylon bag incubation of feedstuffs which could lead to differences in microbial access to particles and pH between the material inside and outside the bags. In situ and in vivo methods, moreover, depend on an initial drying and grinding of the material, and do not allow material to be re-chewed as would occur in vivo. The dependence of INDF values on the method of analysis should be considered when comparing INDF values produced in different studies. Reported INDF values, moreover, should be accompanied by a description of the analytical approach as, to an important degree, INDF is defined by the method of analysis and its relationship to what may occur in vivo is unknown. Indeed, the value of INDF is solely to drive mathematical models, such as NorFor or CNCPS, and the importance of INDF values is dependent upon the accuracy of the assumptions within such models, which are largely unknown. 4.2. Grasses The INDF concentration in DM correlated better with chemical measurements for all plant types than did INDF concentration in aNDFom, mainly due to a higher variation in INDF concentration in DM among plant species. Grasses and legumes had a numerically similar, but statistically different, average INDF/lignin(sa) ratio (2.7 and 2.6 g/g DM) which was higher than the 2.4 ratio used among plant types in CNCPS (Fox et al., 2004). The INDF/lignin(sa) ratio varied among grass species, cut numbers and harvest times within primary growth, making a universal prediction equation based on only lignin(sa) concentration, as used in CNCPS, impossible. Nevertheless, for grasses, simple linear regression analysis revealed lignin(sa) concentration in DM to be the best single predictor of INDF concentration in DM. Lignin(sa) concentration of grasses differed among grass species, consistent with Huhtanen et al. (2006) who found that plant species affected organic matter digestibility (OMD), and its variation. When analysing the full dataset, as well as when choosing a groupwise approach in analysis, slopes of the simple linear regression equations in Table 5 had high standard errors. Decreasing concentrations of INDF, aNDFom, lignin(sa) and IOM from primary growth to third regrowth in grass species were contrasted to the trends analysed in primary growth and regrowth silages by Kuoppala et al. (2010), who reported lower concentrations of aNDFom and OMD, and higher INDF concentrations, in regrowth compared with primary growth silages of a mixed timothy-meadow fescue sward as well as a decreasing digestibility of plant cell walls in regrowth indicated by higher lignin(pm) and INDF concentrations in aNDFom. Our study differs from Kuoppala et al. (2010) in the choice of grass species, as Kuoppala et al. (2010) used mixed timothy-meadow fescue silages, and in the length of regrowth periods (i.e., 70 d). Indeed, increased heterogeneity of plant material in regrowth plots partly consisting of dead tissues and other plant species (Kuoppala et al., 2010) and different environmental conditions during growth, could have led to the differences in chemically defined plant cell wall values among cut numbers. While IOM did not differ among cuts 1, 3 and 4, they differed from cut number 2, and lignin(sa) decreased from cut number 1 to 4 but not between cut numbers 2 and 3. This suggests that it is not the extent of lignification, but rather its distribution and lignin linkages with polysaccharides, which influences IOM and may also explain the decrease in IOM in spite of increasing fibre concentrations. This is consistent with the linear relationship between INDF concentration in DM and IOM among regrowths in grasses, which was also described by Nousiainen et al. (2003). In relation to maturity, Rinne et al. (2002) found a linear decrease in OMD and aNDFom digestibility with increasing maturity in grass silage. A quadratic trend for NDF digestibility was observed, in so far as the greatest change occurred between the third and fourth harvest. Our results confirmed the better fit of a non-linear regression line among harvest times in primary growth (Fig. 2). 4.3. Legumes For legumes, the INDF/lignin(sa) ratio had a high variation among species making it an unacceptable single predictor of INDF concentration among legume species. The high standard errors associated with the cell wall concentrations for legumes may reflect the low number of legumes used. High standard errors, moreover, may be the reason for the lesser effects of cut number and forage type on concentrations of plant cell wall fractions. Growth number and forage species did not affect

50

M. Krämer et al. / Animal Feed Science and Technology 177 (2012) 40–51

most of chemical and digestibility measurements, probably due to the limited number of samples. Chemical and digestibility measurements should, therefore, be regarded with care. 4.4. Comparison of grasses and legumes The decreasing concentrations of INDF, aNDFom and lignin(sa), and similar IOM, in grass regrowths were not consistent with legumes, which had increasing aNDFom concentration and IOM from primary to third regrowth. The increase in IOM in association with decreasing INDF and lignin(sa) concentrations in aNDFom underlines effects of forage type and species within forage type on development of cell wall fractions among growth number and is probably due to the morphological location of lignin(sa) in legumes, which makes microbial access and degradation of legume cell walls difficult (Wilson and Kennedy, 1996). According to their study, lignin is mainly found in the xylem in which the lignin concentrations reach levels which render cells completely indigestible. In contrast, lignin concentrations in other tissues are low making them almost completely digestible. Grasses contain relatively less lignin which is distributed on all tissues except phloem (Wilson and Kennedy, 1996). This explains why lower concentrations of indigestible fractions can limit digestion of cell walls by rumen microbes in grasses. Plant cell wall concentrations and IOM varied among cut numbers, probably because increasing maturity increases the ratio between stem and leaves as well as lignification (Kuoppala et al., 2009), which increases because stems contain more lignin(sa) than leaves. Lignification, moreover, depends on weather conditions (e.g., the higher the temperature, the more pronounced the lignification). The pattern of lignification is also affected by temperature and drought stress. Increasing temperature and drought stress decrease OMD (Van Soest, 1994). Judged by the increased INDF/IOM ratio with later harvest, the increase in INDF concentration was higher than the decrease in digestible OM. This phenomenon was also observed by Huhtanen and Jaakkola (1994), who concluded that increased maturity in grasses affected cell wall degradation more than cell wall concentration, indicating that cell wall digestibility decreased with increasing maturity in spite of small changes in cell wall concentration. 4.5. Regression equations for INDF In the full dataset, inclusion of more parameters improved model fit for all plant types, except for grains and byproducts and oilseeds and their byproducts. In the sub-dataset (Tables 5 and 6), model fit for legumes was better than for grasses with stepwise multiple linear regression analysis, but better for grasses in simple linear regression equations. Multiple regression equations improved model fit except for grains and byproducts and oilseeds and byproducts as lignin(sa) concentration might identify differences between legumes, grasses and grass/clover. When considering the full dataset it was shown that inclusion of more parameters in the model was not sufficient to remove systematic residuals. Intercepts and slopes of the regression equations varied among plant types. Comparing forage and concentrate types revealed lowest INDF concentrations in aNDFom for grasses and legumes, intermediate concentrations for concentrates and highest values for legumes. The aNDFom concentration was higher in forages compared to concentrates. Lignin(sa) concentration in aNDFom was higher in concentrates than in forages. Highest INDF/lignin(sa) ratios were in grasses, followed by grains and byproducts, legumes, grass/clover, and oilseeds and byproducts. Variations in chemical and digestibility measurements may partly be due to the heterogeneous environmental conditions during growth of species included in dataset 2. Distinct development of plant material caused by varying weather and soil conditions led to chemical and morphological differences in cell walls among plant samples, even when plant samples were harvested at the same stage of maturity, growth period or growth number. The grass species belonged to different flowering groups and were therefore not necessarily harvested at their optimal harvest time. This could impact the cell wall composition and concentrations, as cutting vernalized grasses in which the meristem is removed with the cut is followed by extensive vegetative development in regrowth due to the induced breakdown of the apical dominance. Plant mass which then mainly consists of leaves leads to the higher OMD in regrowth. In contrast, vernalized grasses of which the meristem is not removed with the cut have higher proportion of stem in total plant material in regrowth and thereby a reduced OMD in regrowth (Søegaard, 1994). The impact of morphological structure of cell walls on their digestibility was partly considered by inclusion of more parameters in multiple linear regression analysis which improved model fit. Therefore, an approach to developing practical prediction equations could be to group plant species according to similarities in the morphological structure of their cell walls. 5. Conclusions Plant cell wall fractions and IOM varied widely among plant types, and plant species within plant type. The ratio between INDF and lignin(sa), moreover, varied among plant types indicating that accurate prediction of INDF concentration based on only lignin(sa) concentration is not possible. Combination of several chemical measurements in multiple linear regression equations improved model fit compared to simple regression analysis for forages, with INDF concentration in DM having a higher predictive potential than INDF concentration in aNDFom. Universal prediction equations for INDF among plant types are not accurate due to variable ratios in cell wall concentrations among plant types.

M. Krämer et al. / Animal Feed Science and Technology 177 (2012) 40–51

51

Acknowledgements This work was completed with financial support from the Commission of the European Communities, FP7, KBB-2007-1; The Danish Ministry of Food, Agriculture and Fisheries, “Fødevareforskningsprogrammet 2007”, 3304-FVFP-07-766-01; and “Mælkeafgiftsfonden” through the project “Development of methods for estimation of iNDF in feedstuffs”. References Åkerlind, M., Weisbjerg, M.R., Eriksson, T., Thøgersen, R., Udén, P., Ólafsson, B.L., Harstad, O.M., Volden, H., 2011. Feed analyses and digestion methods. In: Volden, H. (Ed.), NorFor – The Nordic Feed Evaluation System. Wageningen Academic Publishers, Wageningen, The Netherlands, pp. 41–54. Bertram, H.C., Weisbjerg, M.R., Jensen, C.S., Pedersen, M.G., Didion, T., Petersen, B.O., Duus, J.O., Larsen, M.K., Nielsen, J.H., 2010. Seasonal changes in the metabolic fingerprint of 21 grass and legume cultivars studied by nuclear magnetic resonance-based metabolomics. J. Agric. Food Chem. 58, 4336–4341. Bossen, D., Mertens, D.R., Weisbjerg, M.R., 2008. Influence of fermentation methods on neutral detergent fiber degradation parameters. J. Dairy Sci. 91, 1464–1476. Ellis, W.C., Poppi, D.P., Matis, J.H., Lippke, H., Hill, T.M., Rouquette, F.M., 1999. Dietary-digestive-metabolic interactions determine the nutritive potential of ruminant diets. In: Jung, H.J.G., Fahey Jr., G.C. (Eds.), Nutrition and Ecology of Herbivores. American Society of Animal Science, pp. 423–481. Fox, D.G., Tedeschi, L.O., Tylutki, T.P., Russell, J.B., Van Amburgh, M.E., Chase, L.E., Pell, A.N., Overton, T.R., 2004. The Cornell Net Carbohydrate and Protein System model for evaluating herd nutrition and nutrient excretion. Anim. Feed Sci. Technol. 112, 29–78. Huhtanen, P., Jaakkola, S., 1994. Influence of grass maturity and diet on ruminal dry-matter and neutral detergent fiber digestion kinetics. Arch. Anim. Nutr. 47, 153–167. Huhtanen, P., Ahvenjärvi, S., Weisbjerg, M.R., Nørgaard, P., 2006. Digestion and passage of fibre in ruminants. In: Sejrsen, K., Hvelplund, T., Nielsen, M.O. (Eds.), Ruminant Physiology: Digestion, Metabolism and Impact of Nutrition on Gene Expression, Immunology and Stress. Wageningen Press, Wageningen, The Netherlands, pp. 87–135. Hvelplund, T., Weisbjerg, M.R., 2000. In situ techniques for the estimation of protein degradability and postrumen availability. In: Givens, D.I., Owen, E., Axford, R.F.E., Omed, H.M. (Eds.), Forage Evaluation in Ruminant Nutrition. CABI Publishing, New York, USA, pp. 233–258. ISO, 2008. Animal feeding stuffs – determination of acid detergent fibre (ADF) and acid detergent lignin (ADL) contents. ISO 13906:2008. Krämer, M., Weisbjerg, M.R., Lund, P., 2010. Estimation of indigestible NDF in feedstuffs for ruminants. In: Udén, P., Eriksson, T., Müller, C.E., Spörndly, R., Liljeholm, M. (Eds.), 1st Nordic Feed Science Conference. Swedish University of Agricultural Sciences, Uppsala, Sweden, pp. 14–19. Kuoppala, K., Rinne, M., Ahvenjärvi, S., Nousiainen, J., Huhtanen, P., 2004. Digestion kinetics of NDF in dairy cows fed silages from primary growth and regrowth of grass. J. Anim. Feed Sci. 13, 127–130. Kuoppala, K., Ahvenjärvi, S., Rinne, M., Vanhatalo, A., 2009. Effects of feeding grass or red clover silage cut at two maturity stages in dairy cows. 2. Dry matter intake and cell wall digestion kinetics. J. Dairy Sci. 92, 5634–5644. Kuoppala, K., Rinne, M., Ahvenjärvi, S., Nousiainen, J., Huhtanen, P., 2010. The effect of harvesting strategy of grass silage on digestion and nutrient supply in dairy cows. J. Dairy Sci. 93, 3253–3263. Mertens, D.R., 1993. Kinetics of cell wall digestion and passage in ruminants. In: Jung, H.G., Buxton, D.R., Hatfield, R.D., Ralph, J. (Eds.), Forage Cell Wall Structure and Digestibility. Am. Soc. Agron. Inc., Crop Sci. Soc. Am. Inc., Soil Sci. Am. Inc., Madison, WI, USA, pp. 535–570. Mertens, D.R., Allen, M., Carmany, J., Clegg, J., Davidowicz, A., Drouches, M., Frank, K., Gambin, D., Garkie, M., Gildemeister, B., Jeffress, D., Jeon, C.S., Jones, D., Kaplan, D., Kim, G.N., Kobata, S., Main, D., Moua, X., Paul, B., Robertson, J., Taysom, D., Thiex, N., Williams, J., Wolf, M., 2002. Gravimetric determination of amylase-treated neutral detergent fiber in feeds with refluxing in beakers or crucibles: collaborative study. J. AOAC Int. 85, 1217–1240. Nousiainen, J., Rinne, M., Hellämäki, M., Huhtanen, P., 2003. Prediction of the digestibility of primary growth and regrowth grass silages from chemical composition, pepsin-cellulase solubility and indigestible cell wall concentration. Anim. Feed Sci. Technol. 110, 61–74. R version 2.11.0 Crawley, M.J., 2007. The R Book. John Wiley & Sons Ltd., Chichester, West Sussex, England. Rinne, M., Huhtanen, P., Jaakkola, S., 2002. Digestive processes of dairy cows fed silages harvested at four stages of grass maturity. J. Anim. Sci. 80, 1986–1998. Rinne, M., Olt, A., Nousiainen, J., Seppala, A., Tuori, M., Paul, C., Fraser, M.D., Huhtanen, P., 2006. Prediction of legume silage digestibility from various laboratory methods. Grass Forage Sci. 61, 354–362. Søegaard, K., 1994. Combinations of cut and grazing of grass and grass-clover [Kombinationer af slæt og afgræsning i græs og kløvergræs.] Statens Planteavlsforsøg, SP-rapport nr. 4, Tjele, Denmark. Spanghero, M., Boccalon, S., Gracco, L., Gruber, L., 2003. NDF degradability of hays measured in situ and in vitro. Anim. Feed Sci. Technol. 104, 201–208. Tilley, J.M.A., Terry, R.A., 1963. A two-stage technique for the in vitro digestion of forage crops. Grass Forage Sci. 18, 104–111. Van Soest, P.J., 1994. Nutritional Ecology of the Ruminant, second ed. Comstock Publishing Associates, Cornell University Press, Ithaca, New York, USA. 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. Volden, H., 2011. NorFor – The Nordic Feed Evaluation System. Wageningen Academic Publishers, Wageningen, The Netherlands. Weisbjerg, M.R., Søegaard, K., 2008. Feeding value of legumes and grasses at different harvest times. In: Hopkins, A., Gustafsson, T., Bertilsson, J., Dalin, G., Nilsdotter-Linde, N., Spörndly, R. (Eds.), Biodiversity and Animal Feed. Future Challenges for Grassland Production. Swedish University of Agricultural Sciences, Uppsala, Sweden, pp. 513–515. Weisbjerg, M.R., Lund, P., Chrenkova, M., Larsen, M.,2010. Estimation of indigestible NDF (iNDF) in forages. In: 3rd EAAP International Symposium on Energy and Protein Metabolism and Nutrition. Wageningen Academic Publishers, Wageningen, The Netherlands, pp. 721–722. Wilson, J.R., Kennedy, P.M., 1996. Plant and animal constraints to voluntary feed intake associated with fibre characteristics and particle breakdown and passage in ruminants. Aust. J. Agric. Res. 47, 199–225.