Evaluation of lignin contents in tropical forages using different analytical methods and their correlations with degradation of insoluble fiber

(1)

ContentslistsavailableatScienceDirect

Animal

Feed

Science

and

Technology

journalhomepage:www.elsevier.com/locate/anifeedsci

Evaluation

of

lignin

in

tropical

forages

using

different

analytical

methods

and

their

correlations

with

degradation

of

insoluble

ﬁber

Daiany

I. Gomes

a

_{, Edenio}

_Detmann

a,∗

_{, Sebastião}

_de

_C.

_Valadares

_Filho

a

_,

Romualdo

S. Fukushima

b

_,

_Marjorrie

_A.

_de

_Souza

a

_,

_Tiago

_N.P.

_Valente

a

_,

_Mário

_F.

_Paulino

a

_,

Augusto

C. de

Queiroz

a

a_Departamento_de_Zootecnia,_Universidade_Federal_de_Vic¸_osa,_Vic¸_osa_36571-000,_Minas_Gerais,_Brazil b_Faculdade_de_Medicina_Veterinária_e_Zootecnia,_Universidade_de_São_Paulo,_{Pirassununga,}_São_Paulo,_Brazil

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received21October2010

Receivedinrevisedform11April2011 Accepted4May2011

Keywords:

Acetylbromidelignin

Indigestibleneutraldetergentﬁber Klasonlignin

Potassiumpermanganatelignin Sulfuricacidlignin

a

b

s

t

r

a

c

t

Wecomparedthelignincontentsoftropicalforagesbydifferentanalyticalmethodsand evaluatedtheircorrelationswithparametersrelatedtothedegradationofneutraldetergent fiber(NDF).Thelignincontentwasevaluatedbyfivemethods:cellulosesolubilizationin sulfuricacid[Lignin(sa)],oxidationwithpotassiumpermanganate[Lignin(pm)],theKlason ligninmethod(KL),solubilizationinacetylbromidefromaciddetergentfiber(ABLadf)and solubilizationinacetylbromidefromthecellwall(ABLcw).Samplesfromtengrassesand tenlegumeswereused.Thelignincontentvaluesobtainedbygravimetricmethodswere alsocorrectedforproteincontamination,andthecorrectedvalueswerereferredtoasLignin (sa)p,Lignin(pm)pandKLp.TheindigestiblefractionofNDF(iNDF),thediscretelag(LAG) andthefractionalrateofdegradation(kd)ofNDFwereestimatedusinganinvitroassay. Correctingforproteinresultedinreductions(P<0.05)inthelignincontentsasmeasured bytheLignin(sa),Lignin(pm)and,especially,theKLmethods.Therewasaninteraction (P<0.05)ofanalyticalmethodandforagegroupforlignincontent.Ingeneral,LKpmethod providedthehigher(P<0.05)lignincontents.Theestimatesoflignincontentobtainedbythe Lignin(sa)p,Lignin(pm)pandLKpmethodswereassociated(P>0.05)withalloftheNDF degradationparameters.However,thestrongestcorrelationcoefficientsforallmethods evaluatedwereobtainedwithLignin(pm)pandKLp.Thelignincontentestimatedbythe ABLcwmethoddidnotcorrelate(P>0.05)withanyparametersofNDFdegradation.There wasacorrelation(P<0.05)betweenthelignincontentestimatedbytheABLadfmethodand iNDFcontent.Nonetheless,thiscorrelationwasweakerthanthosefoundwithgravimetric methods.Fromtheseresults,weconcludedthatthegravimetricmethodsproduceresidues

Abbreviations:ABLadf,lignindeterminedbysolubilizationwithacetylbromidefromtheaciddetergentfiber;ABLcw,lignindeterminedbysolubilization withacetylbromidefromthecellwallmatrix;CP,crudeprotein;DM,drymatter;iNDF,indigestiblefractionofneutraldetergentfiber;kd,fractional degradationrateofpotentiallydegradableneutraldetergentfiber;KL,lignindeterminedbyKlasonmethod;KLp,lignindeterminedbyKlasonmethodand correctedforprotein;LAG,discretelagforfiberdegradation;Lignin(sa),lignindeterminedbysolubilizationofcellulosewithsulfuricacid;Lignin(sa)p, lignindeterminedbysolubilizationofcellulosewithsulfuricacidandcorrectedforprotein;Lignin(pm),lignindeterminedbyoxidationwithpotassium permanganate;Lignin(pm)p,lignindeterminedbyoxidationwithpotassiumpermanganateandcorrectedforprotein;NDF,neutraldetergentfiber;pdNDF, potentiallydegradablefractionofneutraldetergentfiber.

∗ Correspondingauthor.Tel.:+553138992252;fax:+553138992252. E-mailaddress:detmann@ufv.br(E.Detmann).

(2)

thatarecontaminatedbynitrogenouscompounds.Adjustmentforthesecontaminantsis suggested,particularlyfortheKLmethod,toexpresslignincontentwithgreateraccuracy. TherelationshipsbetweenlignincontentmeasurementsandNDFdegradationparameters canbebetterdeterminedusingKLpandLignin(pm)pmethods.

1. Introduction

Thequantificationofthenutritionalvalueofruminantfeedrequiresstudiesthatevaluatethefiberfractionofforages, whichisoffundamentalimportanceinthetropicsandsubtropicsasitprovidessignificantenergyatlowcost.Thefiber fractionshouldoccupyacrucialpositioninenergyevaluation,asitisnaturallymorevariablethantheotherchemical components,suchascellcontents(Detmannetal.,2008).

Severalfactorshavebeenstudiedinanattempttoclarifytheuseofforagesbyanimalstooptimizeruminantperformance inthetropics.Factorsassociatedwiththecompositionofthecellwallhavebeenfoundtoberesponsibleforthelowerforage intakeandanimalperformanceinthetropics.Ofthecomponentsofthecellwall,ligninisconsideredthemainlimiting factorofthedegradationofﬁbrouspolysaccharidesintherumen(VanSoest,1994).

Degradationofthecellwallrequiresanactivemicrobialpopulationthatiscapableofutilizingitscomponents.Therefore, itdependsonacomplexinteractionofmicrobialenzymesandsubstrate,whichwilldeterminetheeffectivenessofthe degradationprocess(Detmannetal.,2009).Thus,thecomponentsofthecellwallthatpreventthecolonizationandutilization oftheﬁbrousfractionsoffeedsmustbestudiedcarefullynotonlyasabsolutechemicalstructures,butalsoinregardtotheir inﬂuenceonthedynamicprocessofruminaldegradation.

Inthemajorityofnutritionalevaluations,thecomplexitiesofthebiologicalmechanismsofdigestionhavebeenignored becauselignincontentiscorrelatedwiththepunctualdigestibilityofforages(VanSoest,1963;JungandVogel,1986;Jung andVarel,1988;FukushimaandHatifield,2004).However,theprocessofsubstrateuseinthegastrointestinaltractofthe ruminantencompassesseveralmorecomplexfactors.Thus,theassociationsbetweenthelignincontentandvariablesrelated tothepotentialforandeffectivenessofmicrobialdegradationofinsolublefiberareimportantfromabiologicalstandpoint. Itisunclearwhichmethodproducesmostaccurateestimatesoflignincontentsintropicalforages.Nevertheless,from anutritionalandfunctionalstandpoint,themethodsforthequantificationofligninmustbeseenasameanstocheckthe portionofthefeedthatcorrespondstodeleteriouseffectsonthedigestionoffibrouscarbohydrates,whichseemstobemore importantnutritionallywhencomparedtoevaluationofabsolutechemicalvalues.

Nomeasureofforagequalitycanbefullycorrect,buttheusefulnessofthedataresultingfromeffortsaimedatthe improvementofforagequalityislimitedbyhowthematerialischaracterized(JungandAllen,1995).Therefore,the mea-surementmethodthatestablishesthebestrelationshipbetweenlignincontentandtheﬁberruminaldegradationoftropical foragesremainsundeﬁned.

Inthisstudy,weaimedtoevaluatelignincontentintropicalforagesbydifferentanalyticalmethodsandtoevaluateits relationshipwithneutraldetergentﬁber(NDF)degradationparametersintropicalgrassesandlegumes.

2. Materialsandmethods 2.1. Locationandsamples

TheexperimentwasconductedattheAnimalNutritionLaboratoryoftheAnimalScienceDepartmentoftheUniversidade FederaldeVic¸osainVic¸osa,Brazil,andattheLigninLaboratoryoftheSchoolofVeterinaryMedicineandAnimalScienceof theUniversidadedeSãoPauloinPirassununga,Brazil.

Tengrasses,Pennisetumpurpureum,Brachiariadecumbens,Panicumrepens,Brachiariahumidicula,Andropogongayanus, Panicummaximumcv.Aruana,Panicummaximumcv.Mombac¸a,Brachiariabrizanthacv.Xaraés,tifton85bermudagrass (Cynodonsp.)andPanicummaximumcv.Massai,andtenlegumes,Arachispintoi,Medicagosativa,Leucaenaleucocephala, Galactiastriata,Dolichoslablab,Centrosemapubescens,Glycineweghtii,Gliricidiasepium,StylosantesguianenensisandCajanus cajan,wereevaluated.Theforageswerecultivatedin2-m×4-mplots.AllsampleswerecutatgroundlevelinDecember 2008.Theplantshadapproximately45daysofregrowth.

Thesampleswereoven-driedat60◦Candprocessedinaknifemill(1-mm). 2.2. Laboratoryanalyses

2.2.1. Chemicalcompositionofsamples

Thedrymatter(DM,indexno.934.01),organicmatter(indexno.942.05),crudeprotein(CP,indexno.954.01),andether extract(indexno.920.39)contentsofthesampleswereanalyzedaccordingtothemethodsoftheAOAC(1990).FortheNDF analysis,thesamplesweretreatedwithaheat-stablealphaamylasewithoutusingsodiumsulﬁteandcorrectedforresidual ash(Mertens,2002)andprotein(Licitraetal.,1996)(Table1).

(3)

Table1

Chemicalcompositionofforages.

Forage DMa,b _OMa,c _CPa,c _EEa,c _aNDFom,pa,c _NFCa,c,d Grasses P.purpureum 200.2 886.6 113.7 21.3 658.0 93.6 B.decumbens 305.8 920.6 78.5 36.9 710.3 94.9 P.repens 310.7 951.5 97.3 25.4 714.8 114.0 B.humidicola 290.2 930.9 70.2 23.1 723.0 114.5 A.gayanus 290.5 940.4 87.9 23.2 736.7 92.5 P.maximumcvAruana 347.5 942.1 78.0 25.1 736.5 102.5 P.maximumcvMombac¸a 294.6 905.6 103.9 12.8 690.4 98.8 B.brizanthacvXaraés 272.1 933.1 86.8 23.8 719.8 102.7 Cynodonsp. 312.5 946.5 107.4 24.0 692.3 122.8 P.maximumcvMassai 321.0 932.9 72.6 21.5 799.5 39.0 Legumes Arachispintoi 198.7 918.8 176.1 18.8 403.6 320.4 Medicagosativa 256.9 916.7 219.6 48.5 398.7 249.7 Leucenaleucocephala 342.5 934.2 220.8 39.7 408.9 264.8 Galactiastriata 332.9 924.8 198.7 26.9 407.1 292.1 Dolichoslablab 195.8 914.3 168.7 29.8 443.6 272.2 Centrozemapubescens 262.7 932.6 156.9 20.8 680.9 74.0 Glicinewightti 243.2 904.9 178.2 36.5 512.3 177.8 Gliricidiasepium 176.2 934.7 196.8 15.5 598.0 124.4 Stylosantesguianenensis 304.2 942.4 127.5 27.0 546.9 241.0 Cajanuscajan 293.8 949.2 169.8 29.0 692.5 57.9

a_DM,_dry_matter;_OM,_organic_matter;_CP,_crude_protein;_EE,_ether_extract;_aNDFom,p,_neutral_detergent_ﬁber_assayed_with_a_heat_stable_amylase_and

expressedexclusiveforresidualashandprotein;NFC,non-ﬁbrouscarbohydrates.

b_g/kg. c_g/kg_DM.

d_NFC₌_OM₋_(CP₊_EE₊_aNDFom,p)₍_Detmann_and_Valadares_Filho,₂₀₁₀_).

2.2.2. Ligninanalyses

Thelignincontentsofforageswerequantiﬁedbyﬁvedifferentmethods:cellulosesolubilizationbysulfuricacidafter extractionwithaciddetergent[Lignin(sa)];oxidationbypotassiumpermanganateafterextractionwithaciddetergent [Lignin(pm)];theKlasonligninmethod(KL);thesolublelignininacetylbromidemethodafterextractionwithaciddetergent (ABLadf)andthesolublelignininacetylbromidemethodusingthecellwallresidue(ABLcw).

Toquantifythelignin contentbytheLignin(sa) method,approximately1gof samplewasconditionedin120-mL polyethylenescrewcappedbottles,and100mLofaciddetergentwasadded(VanSoestandRobertson,1985).Aftersealing, thebottleswereautoclavedat105◦Cfor1h(PellandSchofield,1993).Theaciddetergentinsolubleresiduewasretained byvacuumfiltrationinafiltercrucible,washedsequentiallywithhotwaterandacetone,andoven-driedat105◦Cfor16h. Afterward,thefiltercruciblescontainingtheresidueswereconditionedinpolyethyleneflasksandtreatedwith12Msulfuric acidfor3hasdescribedbyVanSoestandRobertson(1985).Afterthat,thecruciblesweresubjectedtovacuumfiltrationand washedwithhotwatertocompletelyremovetheacid.Thematerialwasoven-driedat105◦Cfor16handthenweighedto obtainthemassoftheresiduecomposedofligninandminerals.Then,thecruciblesweretransferredtoamufflefurnaceat 500◦Cfor3h.Theywereweighedagain,andthemassofligninwascalculatedbytheweightlossafterincineration.

Toquantifythecontentofresidualprotein(N×6.25)associatedwithlignin,aliquotsofresiduesobtainedaftertreatment withsulfuricacidwereevaluatedaccordingtotheKjeldahlmethod(AOAC,1990).

TheKLmethodfordetermininglignincontentisbasedontheacidhydrolysisofthewater-insolublefraction(Theander andWesterlund,1986).Inthismethod,thematerialwasnotsubjectedtoextractionwithaciddetergent.Approximately 250mgofsamplewasconditionedin120-mLpolyethylenescrewcappedbottles.Threemillilitersof12Msulfuricacid wasaddedtothesample,whichwasstirredwithaglassrod.Thebottleswerekeptinawaterbathat30◦Cfor30min. Subsequently,80mLofdistilledwaterwasaddedtoeachpot,andthebottleswerethensealedandautoclavedat105◦Cfor 1h.Afterautoclaving,whilethecontentswerestillwarm,theinsolublematerialwasquantitativelyvacuum-transferredto ﬁltercruciblesandthenwashedwithhotwateranddriedat105◦Cfor16h.Subsequently,thecrucibleswereheatedinthe mufﬂefurnaceat500◦Cfor3h.Theweightafterincinerationwassubtractedfromtheweightoftheresidueinsolublein sulfuricacidtocalculatethelignincontent.TheresidualproteinwasevaluatedsuchdescribedforLignin(sa).

TheLignin(pm)methodwasperformedbyfirstobtainingtheaciddetergentinsolubleresidueasdescribedfortheLignin (sa)method.Then,thefiltercruciblescontainingtheresiduewereplacedinapolyethylenetraywitha2–3-cmlayerof distilledwaterandsequentiallyextractedwithasaturatedKMnO4solutionandademineralizingsolutionasdescribedby VanSoestandWine(1968).Afterthat,theresiduewasvacuumfiltered,andwashedwithanethanolsolution(800mL/L) andacetone.Thecruciblescontainingtheresiduewereoven-driedat105◦Cfor16h.Theligninmasswascalculatedbythe differencebetweenthemassoftheaciddetergentinsolubleresidueandtheresidualmassafterthetreatment.

(4)

Subsequently,aliquotsoftheresiduesobtainedaftertreatmentwiththepermanganateanddemineralizingsolutions wereevaluatedaccordingtotheKjeldahlmethod(AOAC,1990).Theproteincontentassociatedwithlignin(N_{× 6.25)}was calculatedbysubtractingtheproteinintheresidueobtainedfromtheaciddetergentinsolubleprotein(Licitraetal.,1996). ToisolatethecellwallfortheevaluationofABLcw,10-galiquotswereconditionedinnon-woventextileﬁlterbags (100g/m2_;₁₅_cm_×₁₀_cm),_which_were_conditioned_in_a_Soxhlet_extractor_equipped_with_a_heating_blanket._The_samples_were

subjectedtosequentialextractionwithwater,ethanol(960mL/L),chloroform:methanol(2:1)andacetone.Approximately 250mLofeachsolutionwereusedandtheextractionsweresupposedtobecompletedwhentheresidualliquidwascolorless. Afterextractions,thematerialwasoven-driedat60◦Cfor72h.

MaterialforABLadfquantiﬁcationwaspreparedinasimilarmannertothatdescribedfortheevaluationofLignin(sa). TheresiduesobtainedfortheevaluationofABLcwandABLadfcontentwereanalyzedaccordingtotherecommendations ofFukushimaandKerley(2011).

A100-mgsampleofcellwalloraciddetergentresiduewasweighedinaglasscentrifugetubewithaTeﬂoncap.Then, 10mLofasolutionofacetylbromideinaceticacid(250mL/L)wasadded,andthesamplewasslowlyhomogenized. Subse-quently,thetubeswerekeptinawaterbathat50◦Cfor2h.Thecontentswerestirredevery30min.Ablankcontrolwas setupwiththesameseriesoftubes.Afterthetubescooleddown,thematerialwascentrifugedat2000_×gfor15min.Then, 0.5-mLaliquotsofthesolutions,eachcontainingaround5mgofresidues,werepipettedintotesttubescontaining6.5mL ofglacialaceticacidand2mLof0.3MNaOH.Thematerialwasstirred.Onemilliliterof0.5Mhydroxylaminehydrochloride solutionwasadded,andthecontentswerefurtherstirred.

Theabsorbanceofthesolutionwasreadat280nmandconvertedintoconcentrationsaccordingtotheequationsuggested byFukushimaandKerley(2011):

L=A−₂₃0_.077.0009 (1)

whereListheligninconcentration(mg/mL)andAistheabsorbance.

ThelignincontentsdeterminedbytheABLadfandABLcpmethodswereconsideredfreefromproteincontamination (Morrison,1972);therefore,nocorrectionwasmade.

2.3. EvaluationofNDFdegradation

FortheinvitroevaluationoftheNDFdegradationdynamics,aliquotsofforage(350mgofDM)wereconditionedin50-mL glassﬂasks.Subsequently,28mLofabuffersolutionwhosepHhadpreviouslybeenadjustedto6.8byﬂushingwithCO2

wereadded(McDougall,1949).

Theflasksweremaintainedinaclimate-controlledroom(39◦C)forpriorhydrationofthesamples.Duringthehydration process,ruminalfluidwascollectedfromarumen-fistulatedsteerthatwaskeptclosetotheincubationroom.Theanimal wasfedadlibitumwithamixeddiet(80:20forage:concentrateratio)andcompletemineralmixture.

Theliquidwascollectedintheliquid:solidinterfaceoftheruminalenvironment,filteredbyatriplelayerofcheesecloth, conditionedinanthermalcontainerandimmediatelytransported totheincubationroom.Sevenmillilitersofruminal inoculawasaddedperflask.TheincubationenvironmentwasimmediatelysaturatedwithCO2andtheflaskswerequickly

sealed.Theﬂasksweremaintainedat39◦Cwithorbitalshaking(40rpm).Gasesarisingfromfermentationwereremoved every3husingneedles.

Incubationtimesof0,3,6,9,12,24,36,48,72and96hwereevaluated.Theincubationprocedurewasrepeatedfour times,resultinginatotaloffourevaluationsperincubationperiodforeachforagesample.Attheendofeachincubation period,theflaskswereremovedfromtheclimate-controlledroom,andthecontentswerevacuumfilteredinfiltercrucibles. Thecrucibleswerethenconditionedinpolyethyleneflasks(120mL)towhich50mLofneutraldetergentwasadded. Afterbeingsealed,theflaskswereautoclavedat105◦Cfor1h(PellandSchofield,1993).Afterthat,residueswerevacuum filtered,sequentiallywashedwithhotwaterandacetone,andoven-driedat105◦Cfor16h.

TheNDFresiduesweresubjectedtoadjustmentbythenon-linearlogisticmodeldescribedbyVanMilgenetal.(1991) throughtheGauss-NewtonalgorithmimplementedinthePROCNLINofSAS:

Rt=pdNDF×(1+×t)×exp(−×t)+iNDF (2)

whereRtisthenon-degradedNDFresidueattime“t”(g/100gNDF);pdNDFisthepotentiallydigestibleNDFfraction(g/100g NDF);iNDFistheindigestibleNDF(g/100gNDF);isthecombinedfractionalrateoflaganddegradationofpdNDF(h−1); andtisthetime(h).

Giventhattheparameterrepresentsboththelaganddegradationrates,thefractionalrateofdegradationrateofpdNDF (kd,h−1)wasestimatedfromusingthegamma-2distributionproperties(Ellisetal.,1994):

kd=0.59635× (3)

TheestimatesofdiscretelagtimewereobtainedaccordingtoVieiraetal.(1997): LAG=R(0)−R(ti)

(5)

whereLAGisthediscretelagtime(h);R(0)isthenon-degradedNDFresidueatt=0(g/100gNDF);R(ti)isthenon-degraded

NDFresidueobtainedatthepointofinﬂectionofthedegradationcurve(g/100gNDF);R(ti)isthederivativefromtheﬁtted

degradationcurvetothepointofinﬂection(maximumrateofsubstratedegradation)(h−1);andtiisthetimeequivalentto

thepointofinﬂectiononthedegradationcurve(h).

ThetivalueswereobtainedaccordingtoVanMilgenetal.(1991):

ti= 1 (5)

2.4. Statisticalanalyses

ThevaluesforcrudeligninandlignincorrectedforproteincontaminantswerecomparedindependentlyfortheLignin (sa),Lignin(pm)andKLmethodsbyadjustingthesimplelinearregressionequationofcorrectedvalues(dependentvariable) oncrudevalues(independentvariable);thestatisticalanalysiswasconductedunderthenullhypothesis:

H0: ˇ0=0 and ˇ1=1 (6)

Correctedandcrudelignincontentswereconsideredtobesimilarwhenthenullhypothesiswasnotrejected.

Thelignincontentsobtainedbythedifferentmethodsweredirectlycomparedbetweenthedifferentspeciesgroups (grassesorlegumes)accordingtothemodel:

Yijk=+Gi+S(i)j+Mk+GMik+εijk (7)

whereisthegeneralconstant;Giistheeffectofthespeciesgroupi(grassorlegume;ﬁxedeffect);S(i)jistheeffectof

speciesjnestedwithingroupi(randomeffect);Mkistheeffectofthekthmethodofanalysis(ﬁxedeffect);GMikisthe

interactioneffectofthespeciesgroupiandthemethodk;andεijkistherandomerror.

Thetotalnumberofobservationsusedintheanalysisofvariancewas100,consistingofplantgroups(2),specieswithin groups(10)andligninmethods(5).

TherelationshipsbetweenthelignincontentsobtainedbythedifferentmethodsandthecharacteristicsofNDF degra-dation(kd,LAGandiNDF)wereevaluatedbylinearregressionusingadummyvariable(DraperandSmith,1966)according tothebasicmodel:

Yij=ˇ0+ˇ1×D+ˇ2×Lij+ˇ3×(D×Lij)+eij (8)

whereYijisthedependentvariableobservedinspeciesjofgroupi;Lij isthelignincontent(g/kgNDF);Disthedummy

variablecorrespondingtothegroupofspecies,withD=0forgrassesandD=1forlegumes;andeijistherandomerror.

Thebestmodelforthedescriptionofrelationshipswaschosenviathebackwardregressionmethod(DraperandSmith, 1966).

AllstatisticalprocedureswereperformedusingSAS(StatisticalAnalysisSystem)(PROCMIXEDandPROCNLIN)(_˛=0.05). 3. Results

CorrectingforcontaminantproteinreducedthelignincontentsestimatedbytheLignin(sa),KLandLignin(pm)methods, onaverage,by0.8,19.9and2.8g/kgDM,respectively,ingrasses.Forlegumes,thecorrectionreducedtheestimatedlignin contentsby2.6,38.9and7.9g/kgDM,respectively(Table2).Therewashigherproteincontaminationinthelignincontents evaluatedgravimetricallyinlegumesthantherewasingrasses.

Consideringbothunitsofexpressionoflignincontent(g/kgDMandg/kgNDF),theadjustmentforproteincontamination resultedinareduction(P<0.05)ofthelignincontentsobtainedbytheLignin(sa),KLandLignin(pm)methods(Table3).

Duetothereduction(P<0.05)inthelignincontentsobtainedbytheLignin(sa),KLandLignin(pm)methodsasaresult ofthecorrections,weusedtheadjustedvalues,Lignin(sa)p,KLpandLignin(pm)p,asthebasisforfurtheranalysisand discussion.Thisdecisionwasbasedonthefactthatthenitrogenouscompoundsofthecellwalldonotexertinhibitory effectsonthedegradationofﬁbrouscarbohydrates,thusallowingamoreaccuratecomparisonwiththecontentsevaluated bytheABLadfandABLcwmethods,whichareconsideredtobefreefromproteincontamination.

Therewasaninteractioneffect(P<0.05)oftheanalyticalmethodandtheforagegrouponlignincontentconsideringboth evaluatedunits(g/kgDMandg/kgNDF;Table4).Ingeneral,KLpproducesthehigherlignincontents(P<0.05).Thepattern oflignincontentswasquitevariableamongmethodsandbetweenforagegroups(Table4).

Despitethedifferencesbetweenthemethods(Table4),theLignin(sa)p,KLpandLignin(pm)pmeasurementswere correlatedwitheachother(P<0.05),withmoderatetostrongcorrelationcoefﬁcientestimates(Table5).Ontheotherhand, themeasurementsobtainedbythespectrophotometricmethods,ABLadfandABLcw,werenotcorrelatedwitheachother (P>0.05)orwiththemeasurementsobtainedbythegravimetricmethods(Table5).

Consideringthattheinhibitoryeffectsofligninondegradationareonlyobservedintheﬁbrousfraction,therelationships withtheNDFdegradationparameterswereinterpretedonlyintermsoftheunitg/kgNDF.Similarlytowhatwaspreviously described,lignincontentobtainedbygravimetricmethodsandadjustedforproteincontaminationwasconsidered.

TheestimatesoflignincontentobtainedbytheLignin(sa)pandLignin(pm)pmethodswererelated(P<0.05)totheiNDF contentaswellastokdandLAG(Table6andFigs.1and2).However,forbothmethods,therewerenodifferencesinbehavior

(6)

Table2

Descriptivestatisticsfortheevaluatedvariables.

Itema _Grasses _Legumes

Mean SD Mean SD g/kgDM NDFap 568.5 42.3 412.4 62.3 Lignin(sa) 63.4 11.1 92.3 21.5 Lignin(as)p 62.6 10.7 89.7 18.6 KL 154.2 20.9 171.3 56.2 KLp 134.3 18.6 132.4 47.1 Lignin(pm) 83.9 7.0 116.8 16.4 Lignin(PM)p 81.1 6.6 108.9 17.4 ABLcw 79.7 16.7 53.1 49.9 ABLadf 46.9 11.1 41.6 6.6 g/kgNDF Lignin(sa) 111.5 18.3 224.1 43.3 Lignin(as)p 110.3 17.7 217.7 37.6 KL 271.8 38.9 413.8 122.3 KLp 236.8 35.1 319.3 101.8 Lignin(pm) 147.8 11.1 286.1 37.9 Lignin(PM)p 142.8 10.7 266.6 40.1 ABLcw 141.4 34.1 131.8 25.0 ABLadf 82.8 19.9 101.3 9.8 Degradationparameters iNDF 369.3 48.7 510.4 84.7 LAG 3.43 0.47 4.71 1.20 kd 0.0498 0.0070 0.0376 0.0088

a_NDFap,_neutral_detergent_ﬁber_corrected_for_ash_and_protein;_Lignin_(sa),_lignin_determined_by_{solubilization}_of_cellulose_with_sulfuric_acid;_Lignin

(sa)p,lignindeterminedbysolubilizationofcellulosewithsulfuricacidandcorrectdforprotein;KL,Klasonlignin;KLp,Klasonlignincorrectedforprotein; Lignin(pm),lignindeterminedbyoxidationwithpotassiumpermanganate;Lignin(pm)p,lignindeterminedbyoxidationwithpotassiumpermanganate andcorrectedforprotein;ABLadf,lignindeterminedbysolubilizationwithacetylbromidefromtheaciddetergentfiber;ABLcw,lignindeterminedby solubilizationwithacetylbromidefromthecellwallmatrix;iNDF,indigestiblefractionofneutraldetergentfiber(g/kgNDF);LAG,discretelag(h);kd, fractionaldegradationrateofpotentiallydegradableneutraldetergentfiber(h−1).

Table3

Estimatesofregressionparametersofrelationshipbetweencrude(X)andproteincorrected(Y)lignincontentsaccordingtodifferentgravimetricmethods.

Methodsa _Regression_parameters

Intercept Slope r2 _s xy P-valueb g/kgDM Lignin(sa) 4.8238 0.9163 0.9941 1.61 <0.0001 KL 5.4141 0.7859 0.9010 11.34 <0.0001 Lignin(pm) 3.5644 0.9109 0.9772 2.98 <0.0001 g/kgNDF Lignin(sa) 6.7810 0.9370 0.9960 4.06 <0.0001 KL 31.3182 0.7196 0.9345 22.44 <0.0001 Lignin(pm) 6.5452 0.9135 0.9920 6.42 <0.0001

a_Lignin_(sa),_lignin_determined_by_{solubilization}_of_cellulose_with_sulfuric_acid;_KL,_Klason_lignin;_Lignin_(pm),_lignin_determined_by_oxidation_with

potassiumpermanganate.

b _H

0:ˇ0=0andˇ1=1(Eq.(6)).

Table4

Evaluationofinteractioneffectofanalyticalmethodandforagegroupsonaveragelignincontents.

Forage Methoda

Lignin(as)p KLp Lignin(pm)p ABLcw ABLadf SEM g/kgDMb

Grasses 62.6Bc 134.3Aa 81.1Bb 79.7Ab 46.9Ad 6.24 Legumes 89.7Ac 132.4Aa 108.9Ab 53.1Bd 41.6Ad

g/kgNDFb

Grasses 110.3Bc 236.8Ba 142.8Bb 141.4Ab 82.8Ac 13.18 Legumes 217.7Ac 319.3Aa 266.6Ab 131.8Ad 101.3Ad

a_Lignin_(sa)p,_lignin_determined_by_{solubilization}_of_cellulose_with_sulfuric_acid_and_corrected_for_protein;_KLp,_Klason_lignin_corrected_for_protein;_Lignin

(pm)p,lignindeterminedbyoxidationwithpotassiumpermanganateandcorrectedforprotein;ABLadf,lignindeterminedbysolubilizationwithacetyl bromidefromtheaciddetergentﬁber;ABLcw,lignindeterminedbysolubilizationwithacetylbromidefromthecellwallmatrix.

b _Means_in_the_column_followed_by_different_capital_letters_or_in_the_row_followed_by_different_lower_case_letters_are_different_according_to_Fishers’_LSD

(7)

Table5

Partialcorrelationsbetweenlignincontents(g/kgNDF)obtainedbydifferentmethods.

Method Methoda,b,c

Lignin(as)p KLp Lignin(pm)p ABLcw

KLp 0.7397 – (0.0003) Lignin(pm)p 0.5954 0.6545 – (0.0072) (0.0024) ABLcw −0.0046 0.0743 0.3760 – (0.9850) (0.7625) (0.1126) ABLadf 0.2034 0.1044 0.1550 −0.0103 (0.4036) (0.6705) (0.5263) (0.9667)

a_Lignin_(sa)p,_lignin_determined_by_{solubilization}_of_cellulose_with_sulfuric_acid_and_corrected_for_protein;_KLp,_Klason_lignin_corrected_for_protein;_Lignin

(pm)p,lignindeterminedbyoxidationwithpotassiumpermanganateandcorrectedforprotein;ABLadf,lignindeterminedbysolubilizationwithacetyl bromidefromtheaciddetergent.

b_The_correlation_estimates_are_adjusted_regarding_forage_groups_effect.

c_The_values_between_parenthesis_correspond_to_descriptive_level_of_probability_for_type_I_error_associated_with_H 0:=0.

betweengrassesandlegumesintermsoftheinterceptortheslopeoftheﬁttedline(P>0.05).Lignincontentestimatesmade usingtheLignin(pm)pmethodweremorestronglycorrelatedwiththeNDFdegradationparametersthanwerethosemade bytheLignin(sa)pmethod(Table6).

ThelignincontentsestimatedbyKLpwereassociated(P<0.05)withallNDFdegradationparameters(Table6). Neverthe-less,unliketheLignin(sa)pandLignin(pm)pmethods,therewasadifferencebetweengrassesandlegumesintermsofthe interceptoftheﬁttedline(P<0.05);higheriNDFandLAGvaluesandlowerkdvalueswereobservedforlegumes,regardless oftheligninconcentration(Fig.3).Theinclusionofanadditionalparameter(interceptadjustmentbyusingthedummy variable;Eq.(8)),basedonwhichpeculiaritiesoftheforagegroupswereconsidered,resultedintheKLpmethodhaving

Table6

Estimatesofregressionparametersforrelationshipbetweenparametersofneutraldetergentﬁberdegradationandlignincontentsobtainedbydifferent methods.

Itema _Methodb,c

Lignin(sa)p KLp Lignin(pm)p ABLcw ABLadf iNDF(g/kgNDF) ˇ0 242.3 255.9 187.7 369.3 179.8 ˇ1 – 111.3 – 141.1 – ˇ2 1.2292 0.4790 1.2817 – 2.8263 ˇ3 – – – – – P-Value 0.0001 <0.0001 0.0004 0.0002 0.0208 r 0.7698 0.8534 0.8237 – 0.5209 rd _0.7541 _0.8333 _0.8121 _– _0.4779 sxy 67.20 58.68 71.56 69.29 87.27 kd(h−1) ˇ0 0.0590 0.0605 0.0637 0.0499 0.0499 ˇ1 – −0.0067 – −0.0106 −0.0106 ˇ2 −0.000087 −0.000045 −0.000095 – – ˇ3 – – – – – P-Value 0.0046 0.0030 0.0010 0.0056 0.0056 r −0.6200 −0.7179 −0.6934 – – rd _−0.5901 _−0.6744 _−0.6710 _– _– sxy 0.0072 0.0066 0.0066 0.0073 0.0073 LAG(h) ˇ0 2.60 2.45 2.16 3.43 3.43 ˇ1 – 0.63 – 0.98 0.98 ˇ2 0.0080 0.0041 0.0088 – – ˇ3 – – – – – P-Value 0.0038 0.0019 0.0007 0.0040 0.0040 r 0.6301 0.7372 0.7077 – – rd _0.6013 _0.6974 _0.6867 _– _– sxy 0.64 0.58 0.58 0.64 0.64

a_See_more_details_about_parameters_in_Eq.₍₈₎_.

b_Lignin_(sa)p,_lignin_determined_by_{solubilization}_of_cellulose_with_sulphuric_acid_and_corrected_for_protein;_KLp,_Klason_lignin_corrected_for_protein;

Lignin(pm)p,lignindeterminedbyoxidationwithpotassiumpermanganateandcorrectedforprotein;ABLadf,lignindeterminedbysolubilizationwith acetylbromidefromtheaciddetergentﬁber;ABLcw,lignindeterminedbysolubilizationwithacetylbromidefromthecellwallmatrix.

c_iNDF,_indigestible_fraction_of_neutral_detergent_fiber;_LAG,_discrete_lag;_kd,_fractional_degradation_rate_of_potentially_degradable_neutral_detergent_fiber. d_Correlations_adjusted_for_the_number_of_parameters_in_the_fitted_model₍_Draper_and_Smith,₁₉₆₆_).

(8)

200 300 400 500 600 700 50 100 150 200 250 300 350 Lignin (as)p (g/kg NDF) iNDF (g/kg NDF) 0.02 0.03 0.04 0.05 0.06 50 100 150 200 250 300 350 Lignin (sa)p (g/kg NDF) kd (/h) 2 3 4 5 6 50 100 150 200 250 300 350 Lignin (sa)p (g/kg NDF) LAG (h)

Fig.1. Relationshipbetweenligninobtainedbysolubilizationwithsulfuricacidandcorrectedforprotein[Lignin(sa)p]andtheindigestibleneutral detergentﬁber(iNDF),thefractionaldegradationrateofpotentiallydegradableneutraldetergentﬁber(kd)andthediscretelag(LAG)(+=grasses; =legumes).

strongercorrelationcoefﬁcientsthanalloftheothermethodsevaluated.However,whenthecorrelationcoefﬁcientswere adjustedforthenumberofparametersinthemodel,theyweresimilartothoseobtainedwiththeLignin(pm)pmethod (Table6).

ThelignincontentsestimatedbytheABLcwmethodshowednofunctionalrelationship(P>0.05)toanyoftheparameters ofNDFdegradationdynamics.For alloftheseparameters,onlytheaveragedifferencebetweentheforagegroupswas obtained(Fig.4).

Therewasanassociation(P<0.05)betweenthelignincontentsestimatedbytheABLadfmethodandtheiNDFcontents; however,therewerenodifferencesbetweengrassesandlegumes(P>0.05).Nevertheless,thecorrelationcoefﬁcientwas weakerthanthosefoundfortheLignin(sa)p,KLpandLignin(pm)pmethods(Table6).Inaddition,asfortheABLcwmethod, nofunctionalrelationships(P>0.05)werefoundbetweenlignincontentandthekdandLAGparameters(Table6),inwhich onlyanaveragedifferencebetweenthegroupsofforageswasdetected(Fig.5).

(9)

200 300 400 500 600 700 100 150 200 250 300 350 Lignin (pm)p (g/kg NDF) iNDF(g/kg NDF) 0.02 0.03 0.04 0.05 0.06 100 150 200 250 300 350 Lignin (pm)p (g/kg NDF) kd(/h) 2 3 4 5 6 100 150 200 250 300 350 Lignin (pm)p (g/kg NDF) LAG (h)

Fig.2. Relationshipbetweenligninobtainedbyoxidationwithpotassiumpermanganateandcorrectedforprotein[Lignin(pm)p]andtheindigestible neutraldetergentﬁber(iNDF),thefractionaldegradationrateofpotentiallydegradableneutraldetergentﬁber(kd)andthediscretelag(LAG)(+=grasses; =legumes).

4. Discussion

Thisstudyshowedthatproteincontaminationsigniﬁcantlyaffectedtheestimatesoflignincontentbyallgravimetric methods.However,thiscontaminationwasmoreevidentfortheKLmethodthanfortheLignin(sa)andLignin(pm)methods (Tables2and3).

Amongthegravimetricmethodsofligninanalysis,KLhasthehighestproteincontamination(FukushimaandHatifield, 2001),whichconstitutesoneofthemainbiasesintheestimatesobtainedbythismethod.KLwasinitiallydevelopedforthe evaluationoflignincontentinwood.Inthistypeofmaterial,proteincontaminationproblemsarenotexpectedduetothe lowproteinconcentrationinthesamples(WhiteheadandQuicke,1964;TheanderandWesterlund,1986).However,theuse ofKLwithnoadjustmenttoquantifythelignincontentinforagesbecomescomplicatedbecauseofthesignificantpresence ofnitrogenouscompoundsintheinsolubleresiduethatcouldbeincorrectlyclassifiedaslignin(VanSoest,1994).

(10)

200 300 400 500 600 700 150 200 250 300 350 400 450 500 550 600 KLp (g/kg NDF) iNDF(g/kg NDF) 0.02 0.03 0.04 0.05 0.06 150 200 250 300 350 400 450 500 550 600 KLp (g/kg NDF) kd(/h) 2 3 4 5 6 150 200 250 300 350 400 450 500 550 600 KLp (g/kg NDF) LAG (h)

Fig.3.RelationshipbetweenligninobtainedbyKlasonmethodandcorrectedforprotein(KLp)andtheindigestibleneutraldetergentﬁber(iNDF),the fractionaldegradationrateofpotentiallydegradableneutraldetergentﬁber(kd)andthediscretelag(LAG)(+=grasses;=legumes).

Thenitrogenouscompoundsassociatedwithligninresiduecanoriginatefromfourpotentialsources:nitrogenthatis naturallyassociatedwiththecellwall,nitrogenattachedtoMaillardartifacts,nitrogenlinkedtotannins,andkeratinsof animalorigin(VanSoest,1994).Thelastsourceisnotapplicabletotheresultsofthisstudy.

Itispresumedthatlegumesshowhigherproteincontaminationthandograsses(FukushimaandHatiﬁeld,2001),as foundinthisstudy(Table2).Althoughlegumesnaturallyhavehigherproteincontentsthangrassesdo,thelevelofCPina samplewasnotcorrelatedwiththedegreeofproteincontaminationinligninresidues(Hatiﬁeldetal.,1994).However,as previouslyshown,thecorrectionoftheKLvaluesforproteincontaminationreducedtheaverageestimatedlignincontentsof legumesby38.9g/kg,comparedto19.9g/kgDMforgrasses(Table2).Atleastpartofthisgreatercontaminationinlegumes canbetracedtotheirgreatertannincontent,whichwouldresultintheformationofinsolublecomplexeswiththeprotein componentsofforages(VanSoest,1994).

Theanalysisoflignincontentislaboriousandtimeconsuming,anditcanproduceresiduesconsistingpartlyofartifacts thatoverestimatelignincontent(VanSoest,1963).Partoftheproteincontaminationrelatedtolignincouldbeattributedto theformationofartifactsbytheMaillardreaction.However,theirformationoccursmainlyduringdryingattemperatures ofatleast65◦C,whichwereavoidedinthisstudy(samplesweredriedat60◦C).Theuseofadequatetemperaturesand

(11)

200 300 400 500 600 700 50 100 150 200 250 ABLcw (g/kg NDF) iNDF (g/kg NDF) 0.02 0.03 0.04 0.05 0.06 50 100 150 200 250 ABLcw (g/kg NDF) kd (/h) 2 3 4 5 6 50 100 150 200 250 ABLcw (g/kg NDF) LAG (h)

Fig.4. Relationshipbetweenligninobtainedbysolubilizationwithacetylbromidefromcellwall(ABLcw)andtheindigestibleneutraldetergentﬁber (iNDF),thefractionaldegradationrateofpotentiallydegradableneutraldetergentﬁber(kd)andthediscretelag(LAG)(+=grasses;=legumes).

ventilatedovensreducestheformationoftheseartifactsbyacceleratingtheremovalofhumidityfromthematerial,which isnecessaryfornon-enzymaticreactionstooccur(VanSoest,1994).

Inaddition,theuseofhightemperaturesduringtheextractionprocesscouldberelatedtotheformationof nitrogen-containingartifacts.However,theMaillardreactionisnotfavoredunderacidconditions(Eskinetal.,1971);thereisno evidencethatinsolubleartifactsformbetweenthecellularproteinandtheproductsofcellwallcarbohydratehydrolysis (Hatiﬁeldetal.,1994).

Therefore,despitethepossibilityofprotein–tannincomplexformationinlegumes,mostoftheproteincontamination associatedwithKLappearstoberelatedtoretentionofnitrogenouscompoundsnaturallypresentinthecellwallininsoluble residues.

Thecellwallcontainsproteinsthathaveastructuralroleinthematrix,whichmayhavecross-linkswithlignin(Whitmore, 1982).However,itisnotclearwhetherthenitrogenouscompoundspresentintheinsolubleresidueofKLrepresentintact proteins,proteinfragments,modiﬁedproteinsornucleicacids(Hatiﬁeldetal.,1994).Nonetheless,regardlessoftheoriginof

(12)

200 300 400 500 600 700 50 70 90 110 130 150 ABLadf(g/kg NDF) iNDF(g/kg NDF) 0.02 0.03 0.04 0.05 0.06 50 70 90 110 130 150 ABLadf(g/kg NDF) kd(/h) 2 3 4 5 6 50 70 90 110 130 150 ABLadf(g/kg NDF) LAG (h)

Fig.5. Relationshipbetweenligninobtainedbysolubilizationwithacetylbromidefromaciddetergentfiber(ABLadf)andtheindigestibleneutraldetergent fiber(iNDF),thefractionaldegradationrateofpotentiallydegradableneutraldetergentfiber(kd)andthediscretelag(LAG)(+=grasses;=legumes).

thenitrogenouscompounds,themainprobleminthechemicalfractionationoftheﬁbroustissueoftheplantistheefﬁcient separationofproteinsandlignin(VanSoest,1963).

ThemainanalyticaldifferencebetweentheKLandLignin(sa)methodsisthesequenceinwhichthedifferent concentra-tionsofsulfuricacidandextractiontemperaturesareused,whichcausesdifferenteffectsonthehydrolysisofpolysaccharides (Hatiﬁeldetal.,1994).However,itmustbenotedthatcationicdetergents(e.g.,cetyltrimethylammoniumbromide,CTAB) arenotusedintheKLmethodasanaccessoryintheremovalofmaterialstobesolubilized,whichallowsthecleaning ofthematerialthatwillbesubjectedtoacidhydrolysis.Thus,theaciddetergentsolution(20g/LCTABin0.5MH2SO4)is

responsibleforobtainingaresiduethatisnearlyfreefromproteininterference(VanSoestandRobertson,1985),promoting solubilizationofmuchoftheproteinassociatedwiththecellwall(VanSoest,1994).

Becauseofthepreviousextractionwithaciddetergent,Lignin(pm)showedlowerlevelsofproteincontaminationthan KLbutslightlyhigherproteincontaminationlevelsthanwereobservedwithLignin(sa)(Table2).Thisfindingseemstobe justifiedbythefactthatsomeproteinsthatarenotremovedbyaciddetergentcanreactwiththepermanganateandbe quantifiedaslignin(VanSoestandWine,1968).

(13)

AccordingtoVanSoest(1994),evenwhenligninanalysesareperformedcarefully,contaminationwillstillexistregardless ofthegravimetricmethodused,whichcorroboratestheresultsfoundinthisstudy(Table3).Therefore,tocorrectlyexpress theconcentrationsofligninobtainedbygravimetricmethods,especiallyKL,proteincontaminationmustbecorrectedfor (Henriquesetal.,2007).

Inthiscontext,themethodsweredirectlycompared,emphasizingthecorrectionforproteincontaminationintheresidues measuredgravimetricallyandthusavoidingconfusionbecausetherearedifferencesbetweenthesemethodsintermsof con-taminationintensity(Table2).Theresultsobtainedspectrophotometricallyareconsideredfreefromthistypeofcontaminant (FukushimaandKerley,2011).

Lignincontentsinforagesamplesvaryaccordingtothemethodofchemicalisolationofthepolyphenolicmolecule. Thephysicalpropertiesofligninaregenerallychanged bystrongacids,whichpromote polymerizationandadditional condensationandcanconvertpartoftheoriginallysolublematerialintoinsolubleproducts(VanSoest,1994).

Thedifferenceinlignincontentsmeasuredinthesamesamplebydifferentmethodsmayresultfromdifferencesinthe mechanismsofactionofthereagents.Thisﬁndingimpliesthatdifferentanalyticalproceduresprovidedifferentestimates oflignincontent(FukushimaandDehority,2000).

Themethodsforligninanalysiscanbedividedintothreemaincategories:gravimetricmethodslikeLignin(sa)andKL thatremovecellwallconstituentsbutleaveligninbehind;gravimetricmethodslikeLignin(pm)thatoxidizeligninfrom thecellwallmatrix;andsolubilizationmethodsemployingspectrophotometricquantiﬁcation,suchasABLadfandABLcw.

However,inadditiontothechemicalpeculiaritiesofeachmethod,differencesintheresultscanalsobeobtainedbasedon differencesinsamplepreparationsteps.Theuseofaciddetergent,whichreducesproteincontamination(VanSoest,1994), alsogenerallyunderestimatesthelignincontentsofforages,especiallygrasses,asaresultofthepartialsolubilizationof phenoliccompounds(Lowryetal.,1994;FukushimaandHatiﬁeld,2001;Fukushimaetal.,2009).Ingeneral,theseaspects supporttheﬁndingoflowerligninvalueswithLignin(sa)pandLignin(pm)pthanwithKLpandwithABLadfincomparison toABLcw(Table4).

AlthoughLignin(pm)pandLignin(sa)parebothgravimetricmethodsthatareprecededbyaciddetergentextraction, higherlignincontentestimateswerefoundwithLignin(pm)pthanwithLignin(sa)p(Table4).AccordingtoVanSoestand Wine(1968),theLignin(pm):Lignin(sa)ratioisapproximately1.2:1.Inagreementwiththeirﬁndings,weobtainedratios of1.30:1and1.23:1forgrassesandlegumes,respectively(basedontheg/kgNDFvaluesinTable4).

Theevaluationoflignincontentsbyoxidationinpotassiumpermanganatecanbeaffectedbysomesamplecomponents, suchasphenolsandotherunsaturatedsubstances,includingtanninsandpigments,thatarenotcompletelyremovedduring aciddetergentextraction.Thesesubstancesreactwiththepermanganatesolutionandarethuscountedaslignin,especially inimmaturegrasses(VanSoestandWine,1968),increasingthetotallignincontentofthesample.Thiscouldjustify,atleast partly,thehigherestimatesgivenbytheLignin(pm)pmethodcomparedtotheLignin(sa)pmethod(Table4).

Generally,themainlimitationintheuseoftheKLmethodisproteincontamination,whichgeneratesapositivebiasin thelignincontentestimates(VanSoestandRobertson,1985;Kondoetal.,1987).However,aftercorrection,thehighvalues estimatedbyKLpcannotbeattributedtoproteincontamination.

AlthoughtheKLmethodwasdevelopedtoextractligninfromwood(VanSoest,1994),itcanbeusedtoquantifyligninfrom feedsusedforruminantnutrition.However,evenaftermodificationsinwhichdirectheatingwasincorporated(Theander andWesterlund,1986),themethodcontinuestobeappliedtointactsamples.Therefore,itisspeculatedthatsulfuricacid cansolubilizepartofthehemi-cellulosecontainedinthecellwall,whichprecipitateswiththedilutionofacidwithwater, leadingtoitsquantificationaslignin(VanSoest,1967).Compoundssuchasd-galacturonicacid,d-glucuronicacidand d-xylosecanbeconvertedintoaromaticcompoundsinheatedandslightlyacidifiedaqueousmedia(PopoffandTheander, 1976),similartothesecondstageoftheKLmethod.

Nevertheless,Hatifieldetal.(1994)foundthatthecontaminationbycarbohydratesintheligninobtainedbytheKL methodcouldbeconsideredsmallenoughnottocontributesignificantlytotheresidue.Ontheotherhand,theseauthors foundthatthesyringyl:guaiacylratiosofligninresiduesobtainedbytheLignin(sa)andKLmethodsweresimilar.Therefore, thereseemstobenosignificantcontaminationwithphenoliccompoundsformedfromcarbohydratesthatcouldalterthe syringyl:guaiacylratio.

Thus,thedifferencebetweentheKLpandLignin(sa)pmethodsand,consequently,betweentheKLpandLignin(pm)p methodsisinthesolubilizationofsomeoftheligninbyaciddetergent(Hatiﬁeldetal.,1994;Lowryetal.,1994).Inthis case,thedissolutionofthehemicellulosematrixbyaciddetergentwouldleavepartoftheligninwithoutthesupportgiven byhemicellulose,allowingitssolubilization(Lowryetal.,1994).Insupportofthisidea,thereisevidencethatKLpprovides moreaccurateestimatesofthetotallignincontentinforagesamplesthandoesLignin(sa)p,especiallyingrasses(Kondo etal.,1987;Hatiﬁeldetal.,1994;Jungetal.,1997).

Theacetylbromide-solubleligninmethod,aspectrophotometricmethoddevelopedtoquantifylignincontentsinsmall woodsamples,isbasedonmeasurementofthesample’sabsorbanceat280nmaftersolubilizationinasolutionofacetyl bromideinaceticacid(Johnsonetal.,1961).Thismethodwasmodifiedforuseinforagesamples,whichhaveaconsiderable amountofproteinthatsignificantlyinterfereswiththemeasurementoflignincontents(Morrison,1972).Fukushimaetal. (2009)found,however,thattheeffluentfromtheaciddetergentsolutioninseveralgrassesshowedpeaksofabsorbance similartothoseofligninretainedintheaciddetergentinsolubleresidue.Hence,thelignincontentinacetylbromidewas evaluatedbasedonthecellwall,providingmorecredibleestimatesofthelignincontentofforages(IiyamaandWallis,1990; FukushimaandDehority,2000;FukushimaandHatifield,2001,2004;FukushimaandSavioli,2001;Changetal.,2008).

(14)

Theisolationofthecellwallbysequentialwashingwithwater,ethanol,chloroformandacetoneaimstocrediblyrepresent theamountofpolysaccharidesinthecellwall(FukushimaandHatiﬁeld,2004).Therefore,thispreparationofthecellwall wouldbecharacterizedbyhigherligninvalueswhencomparedtoothermethodsofisolationofcellwallcomponents,such asADF.However,ifthenon-lignincomponentsandotherphenoliccompounds,liketannins,arenotremovedduringthecell wallpreparationstep,theycanbedissolvedintheacetylbromidesolution,leadingtointerferenceduringsamplereading (Morrison,1972).

Thelackofstandardsforspectrophotometercalibrationisthemainlimitingfactorfortheroutineuseofthemethod (Saviolietal.,2000).Therefore,inarecentstudy,ligninextractedwithacetylbromideand subsequentlycorrectedfor carbohydrate,ash,waterandproteincontentswereusedtomakeauniversalstandardcurveinanattempttoshortenthe analyticalproceduretosampledigestionandspectrophotometricreadingonly(FukushimaandKerley,2011).Thisuniversal standardcurvewasusedinthisstudy.

ThelignincontentsobtainedforgrassesbytheABLcwmethodweregreaterthanwerethoseobtainedbytheLignin(sa)p method(Table4).Thisresultisexpectedgiventhepartialsolubilizationofligninduringtheaciddetergentextractionstepof theLignin(sa)pmethod,corroboratingtheresultsofotherauthors(FukushimaandDehority,2000;FukushimaandSavioli, 2001;FukushimaandHatiﬁeld,2004).

Forlegumes,however,thelignincontentsobtainedwiththeABLadfandABLcwmethodsweresimilar,whereasABLcw producedlowerestimatesthanLignin(sa)p.Thispatternofresultsisdifferentfromthatobservedforgrasses(Table4)and contradictstheproposedexplanationsforthedifferencesbetweenthemethodspresentedpreviously.

Generally,lignincontentsobtainedbytheABLcwmethodwerehigherthanthoseobtainedwiththeLignin(sa)pmethod (FukushimaandDehority,2000).However,differingresultscanbefoundintheliterature(FukushimaandSavioli,2001).

Oneofthedifficultiesofquantifyinglignininacetylbromideliesinobtainingasatisfactoryspectrophotometricstandard forlignin(Savioliet al.,2000;Hatfieldand Fukushima,2005).Generally,evaluationsoflignininacetylbromideusing grasssampleshavebeenmoreintenselystudiedcomparedtoevaluationsoflegumesamples.Thesetofsamplesusedby FukushimaandKerley(2011)forauniversalstandardcurvehadonly3legumesinasetofsamplesfrom14speciesthatalso includedgrasses,treespeciesand3commerciallignins.Consideringthatthechemicalcompositionsofligninfromgrasses andlegumesdiffer(VanSoest,1994),theapparentdistortionsfoundinthisstudymayindicatethatthepredictionefficiency oftheuniversalstandardcurve(Eq.(1)),althoughapparentlyhighforgrasses,islowforlegumes.

SimplyknowingtheNDFcontentofafeedisnotenoughtogenerateinformationaboutthepotentialforinsolubleﬁberto beutilizedinthegastrointestinaltractinruminants.Inotherwords,twofeedscanhavesimilarNDFcontentsbutdifferent potentialsforutilization.ThisﬁndinglimitsinferencesbasedsolelyonNDFcontentfromanutritionalstandpoint.Knowledge ofthedegradationdynamicsofdifferentNDFsourcesintheruminalecosystemisnecessarytoincreasetheknowledgeabout effectivedigestibilityandthepotentialfortheimplementationofphysicallyrestrictiveeffectsonvoluntaryintake(Detmann etal.,2008,2009).

However,infrastructureandtimelimitationsconstraindatacollectiontocharacterizeNDFdegradationdynamics.Thus, itiscrucialtoﬁndacharacteristicthatiscapableofgeneratinginformationthatquicklydetermines,withrelativeprecision, thecapacitiesofdifferentNDFsourcestobeutilizedbyruminants.

Becauseitisindigestibleandreducesthepotentiallydegradablefibrousfraction(Traxleretal.,1998),ligninisprimarily responsibleforthelimitationofdegradationoffibrousforagecomponents(VanSoest,1994).Thelaboratoryestimateof itsconcentrationisfastandrequireslessinfrastructurethandoinsituorinvitrostudiesofNDFdegradationdynamics. Thus,itisnecessarytoidentifywhichoftheanalyticalmethodsbestdiscriminatesforageswithregardtoNDFruminal degradationaspects.AccordingtoLowryetal.(1994),rumenfermentationcharacteristicsthatdefinethecellwallfraction thatisdeleterioustomicroorganismsaremoreimportanttotheevaluationoffeedsforruminantsthanisdefiningexact chemicalfractions.

Generally,theLignin(sa)pandLignin(pm)pestimatesweresimilarintermsoftheirrelationshipswithNDFdegradation parameters(Table6),showingbiologicallycoherentresultsconsideringthepositiveassociationsbetweenlignincontent andiNDFandLAGcontentsandthenegativeassociationsbetweenligninandkd(Table6andFigs.1and2).However,even thoughtheyareinterrelated(Table5),strongercorrelationswereobservedwiththeLignin(pm)pmethodthanwithLignin (sa)p(Table6).TheseresultsconﬁrmandextendthoseobtainedbyTraxleretal.(1998)andClipes(2007),whoobserved greateraccuracyofiNDFpredictionfromLignin(pm)incomparisontoLignin(sa)estimatesingrassesandlegumes.

Aspreviouslydiscussed,theroleofsulfuricacidintheLignin(sa)methodistooxidizethecellulosiccomponentsofthe plantcellwallafterextractionwithaciddetergent(VanSoestandRobertson,1985),maintainingthephenoliccomponents asresidue;theroleofpotassiumpermanganateistosolubilizethephenoliccompoundsonthecellwall,alsoaftertreatment withaciddetergent,producingaresidueinwhichthecellulosiccompoundsareconcentrated.

AlthoughtheLignin(sa)andLignin(pm)methodsmayseemperfectlycomplementary,problemslieintheexact deﬁ-nitionsofthelimitsofactionofeachofthereagents(VanSoest,1994),causingthelignincontentestimatesobtainedby thetwomethodstodiffer,asseeninthisstudy(Table4).Thegreatestdifferencebetweentheactionsofsulfuricacidand ofpotassiumpermanganatecanbeseenintheboundaryregionsbetweencellulosicandphenoliccompounds,aregionin whichthereisgreaterinhibitoryactionofligninonmicrobialdegradation(Traxleretal.,1998;Clipes,2007).

Therefore,thedifferentactionsoftheabove-mentionedreagentscanleadtoqualitativedifferentiationsinthegravimetric estimatesofthelignincontentsoffeeds.Thatis,theinhibitoryactionofthephenoliccompoundsretainedasligninandtheir extractionpeculiaritiesintheareasadjacenttocellwallcarbohydratescouldhavedifferentassociationswiththeinsoluble

(15)

-200 -100 0 100 200 -200 -100 0 100 200 Centered iNDF Residue -0.02 -0.01 0.00 0.01 0.02 -0.02 -0.01 0.00 0.01 0.02 Centered kd Residue

(a)

(b)

(c)

Fig.6. OrdinaryresidualplotsagainstthecenteredpredictedvaluesofiNDF(g/kgNDF)andkd(h−1)[(a)Klasonlignincorrectedforprotein;(b)lignin determinedbyoxidationwithpotassiumpermanganateandcorrectedforprotein;(c)lignindeterminedbysolubilizationofcellulosewithsulfuricacid andcorrectedforprotein].

(16)

NDFfraction(Clipes,2007)suchthatthelignincontentsdeterminedbyLignin(pm)phaveabetterassociationwiththe insolubleNDFfractionthandothosedeterminedbyLignin(sa)p(Table6).

ThedifferencesbetweengrassesandlegumesindicatedbyLignin(sa)pandLignin(pm)pareonlyattributedtothelignin concentrationsofthesamplesbecausenospeciﬁcparameterfortheirdifferentiationwassigniﬁcant(Table6).

Ontheotherhand,therelationshipsbetweentheNDFdegradationparametersandtheKLpcontentsconsidered,in additiontothedifferencesinsampleconcentrations,aparameterrelatedtothedifferentiationoftheinterceptofthefunction forgrassesandlegumes(Table6andFig.3).ThisadditionaldiscriminationinrelationtoLignin(sa)pseemstobeassociated withthelossofthesolubleligninfractioninaciddetergent,aspreviouslydiscussed.Inthiscontext,theinclusionofan additionalparameterinthefunctionprovidedKLpwithstrongercorrelationsinrelationtoLignin(sa)p,evenconsidering theadjustmentforthenumberofparametersofthemodel(Table6).TheseresultscontradictthoseobtainedbyJungetal. (1997),whofoundsimilarcorrelationsbetweenKLandLignin(sa)andtheinvitroandinvivodigestibilityofNDF.

Consideringthecorrelationcoefficientsadjustedforthenumberofparametersinthemodel,theLignin(pm)pandKLp estimateswereequallystronglyassociatedwiththeNDFdegradationparameters(Table6).Thesemethodsshowedhigher lignincontentsthanLignin(sa)pdid(Table4).Inthiscontext,thehigherlignincontentsestimatedbytheKLpandLignin (pm)pmethodsmaycontainelementsthathaveasignificantroleinthedegradationofinsolublefiber,whichwouldnotbe quantifiedinLignin(sa)p,thusjustifyingtheweakercorrelationsobservedwiththismethod(Table6).

Theresidualevaluationofgravimetricmethodsdoesnotrevealpatternsthatgiveevidenceofmodelunder-speciﬁcation orheterogeneousvarianceandtherewerenosystematictrend(P>0.05)ofordinaryresidues(P>0.05;Fig.6).Theresidual plotwasslightlymorehomogeneousforKLpcomparedtoLignin(sa)pandLignin(pm)p,whichseemstoreﬂectthemodel discriminationwithregardgrassesandlegumes(Table6).Actually,thedatasetevaluatedinthisworkwouldnotbe con-sideredcompletelyadequatetosuggestanaccuratemodeltopredictNDFdegradationparameters.However,theevidences presentedinTable6andFig.6indicatethatKLpshouldbeconsideredwhenfurthermodelswillbeadjustedtoestimate suchcharacteristicsfromlignincontents.

Generally,withtheexceptionoftherelationshipbetweenABLadfandiNDF,therewerenoassociationsbetweenthelignin contentsobtainedbythespectrophotometricmethodsandtheNDFdegradationparameters(Table6andFigs.4and5).This ﬁndingisreinforcedbythelackofcorrelationbetweenthelignincontentsobtainedbythesemethodsandthegravimetric methods(Table5).

Recentresearchontheevaluationofsolublelignininacetylbromidehasraisedthepossibilityofobtainingmoreaccurate estimatesoflignincontentinfeedbyminimizingtheinterferencebyothercompounds,mainlyusingthecellwallasa baseindetrimenttotheaciddetergentinsolubleresidue(FukushimaandDehority,2000;FukushimaandHatiﬁeld,2004; Fukushimaetal.,2009).

Insomestudies,thecorrelationsbetweentheABLcwcontentsandinvitrodigestibilityofDMorothercellwall compo-nentswerestrongerthanwerethoseobtainedwithLignin(sa)orLignin(pm)(FukushimaandDehority,2000;Fukushima andHatiﬁeld,2004).However,inthesestudies,thespectrophotometricstandardswereobtainedindependentlyforeach evaluatedmaterialratherthanusingtheuniversalstandardcurveproposedbyFukushimaandKerley(2011)andadopted inthisstudy.

AccordingtoFukushimaandDehority(2000),theligninextractedfromaforagesamplecannotbeutilizedasa stan-dardfortheanalysisofsamplesobtainedfromdifferentspeciesorforagesatdifferentstagesofmaturation.Empirical regressionequations,likethatproposedbyFukushimaandKerley(2011),arepopulation-dependent.Thesemodelsare basedexclusivelyonexperimentalinformationratherthanonatheoreticalorbiologicalbasis.Therefore,evenwithgood dataadjustment,themodelmustbeconsideredtobespeciﬁctotheconditionsinwhichthedatawereobtained,andits predictivevaluewillbelimited(ForbesandFrance,1993).

Therefore,consideringpreviousinformation(FukushimaandDehority,2000;FukushimaandHatifield,2004),thelack ofassociationbetweenthelignincontentsdeterminedbyABLadfandABLcwmayberelatedtotheinefficiencyofstandard curvepredictionsuggestedbyFukushimaandKerley(2011).Thisfindingconfirmsthattheevaluationofsolubleligninin acetylbromideisrestrictedbytheneedtoobtainspecificstandardsforeachtypeofmaterialevaluated.

5. Conclusions

ThelignincontentsobtainedbytheKlasonmethod,bycellulosesolubilizationinsulfuricacidandbyoxidationwith potassiumpermanganatepresentproteincontamination.Therefore,proteincorrectionissuggested,particularlyforthe Klasonligninmethod.Legumesproducemoreprominentproteincontaminationthandograsses.

Webetterestablishedrelationshipsbetweenligninandtheparametersofruminaldegradationofneutraldetergentﬁber usingtheestimatesproducedbytheKlasonmethodandbyoxidationinpotassiumpermanganate.

Acknowledgments

TheauthorswishtothanktheConselhoNacionaldeDesenvolvimentoCientíficoeTecnológico(CNPq),theFundaçãode AmparoàPesquisadoEstadodeMinasGerais(FAPEMIG-PPM),andtheINCTCiênciaAnimalforfinancialsupport.

(17)

References

AssociationofOfﬁcialAnalyticalChemistry(AOAC),1990.OfﬁcialMethodsofAnalysis,15thed.AOACInternational,Arlington.

Chang,X.F.,Chandra,R.,Berleth,T.,Beatson,R.P.,2008.Rapid,microscale,acetylbromide-basedmethodforhigh-throughputdeterminationoflignin contentinArabidopsisthaliana.J.Agric.FoodChem.56,6825–6834.

Clipes,R.C.,2007.Evaluationofparameterstoestimatethedegradationpotentialofﬁbrousandnitrogenouscompoundsintropicalgrasses.D.S.Thesis. UniversidadeEstadualdoNorteFluminense,CamposdosGoytacazes(inPortuguesewithEnglishabstract).

Detmann,E.,ValadaresFilho,S.C.,2010.Ontheestimationofnon-ﬁbrouscarbohydratesinfeedsanddiets.Arq.Bras.Med.Vet.Zootec.62,980–984. Detmann,E.,Paulino,M.F.,ValadaresFilho,S.C.,2008.Avaliac¸ãonutricionaldealimentosoudedietas?Umaabordagemconceitual.In:Proceedingsof2nd

InternationalSymposiumonBeefCattleProduction,Vic¸osa,Brazil,pp.21–52.

Detmann,E.,Paulino,M.F.,Mantovani,H.C.,ValadaresFilho,S.C.,Sampaio,C.B.,Souza,M.A.,Lazzarini,I.,Detmann,K.S.C.,2009.Parameterizationofruminal ﬁbredegradationinlow-qualitytropicalforageusingMichaelis–Mentenkinetics.Liv.Sci.126,136–146.

Draper,N.,Smith,H.,1966.AppliedRegressionAnalysis.JohnWilleyandSons,NewYork.

Ellis,W.C.,Matis,J.H.,Hill,T.M.,Murphy,M.R.,1994.Methodologyforestimatingdigestionandpassagekineticsofforages.In:FaheyJr,G.C.(Ed.),Forage Quality,Evaluation,andUtilization.AmericanSocietyofAgronomy,Madison,pp.682–756.

Eskin,N.A.M.,Henderson,H.M.,Townsend,R.J.,1971.BiochemistryofFoods.AcademicPress,NewYork.

Forbes,J.M.,France,J.,1993.Introduction.In:Forbes,J.M.,France,J.(Eds.),QuantitativeAspectsofRuminantDigestionandMetabolism.CABInternational, Wallingford,pp.1–12.

Fukushima,R.S.,Dehority,B.A.,2000.Feasibilityofusingligninisolatedfromforagesbysolubilizationinacetylbromideasastandardforligninanalyses. J.Anim.Sci.78,3135–3143.

Fukushima,R.S.,Hatiﬁeld,R.D.,2001.Extractionandisolationofligninforutilizationasastandardtodetermineligninconcentrationusingtheacetyl bromidespectrophotometricmethod.J.Agric.FoodChem.49,3133–3139.

Fukushima,R.S.,Hatiﬁeld,R.D.,2004.Comparisonoftheacetylbromidespectrophotometricmethodwithotheranalyticalligninmethodsfordetermining ligninconcentrationinforagesamples.J.Agric.FoodChem.52,3713–3720.

Fukushima,R.S.,Kerley,M.S.,2011.Useofligninextractedfromdifferentplantsourcesasstandardsinthespectrophotometricacetylbromidelignin method.J.Agric.FoodChem.,doi:10.1021/jf104826n.

Fukushima,R.S.,Savioli,N.M.F.,2001.Correlationbetweeninvitrocellwalldigestibilityandthreeanalyticalmethodsforquantifyinglignin.R.Bras.Zootec. 30,302–309(inPortuguesewithEnglishabstract).

Fukushima,R.S.,Kerley,M.S.,Porter,J.H.,Kallenbach,R.,2009.Theacetylbromideligninmethodtoquantifylignincontentin.In:Proceedingsof46th MeetingofBrazilianSocietyofAnimalScience,Maringá,Brazil(eletronicproceedings;inPortuguesewithEnglishabstract).

Hatﬁeld,R.,Fukushima,R.S.,2005.Canligninbeaccuratelymeasured?CropSci.45,832–839.

Hatiﬁeld,R.D.,Jung,H.G.,Raplh,J.,Buxton,D.R.,Weimer,P.J.,1994.ComparisonoftheinsolubleresiduesproducedbytheKlasonligninandaciddetergent ligninprocedures.J.Agric.FoodChem.65,51–58.

Henriques,L.T.,Detmann,E.,Queiroz,A.C.,ValadaresFilho,S.C.,Leão,M.I.,Paulino,M.F.,2007.Fractionsofcellwallnitrogenouscompoundsintropical forages.Arq.Bras.Med.Vet.Zootec.59,258–263(inPortuguesewithEnglishabstract).

Iiyama,K.,Wallis,A.F.,1990.Determinationoflignininherbaceousplantsbyanimprovedacetylbromideprocedure.J.Sci.FoodAgric.51,145–161. Johnson,D.B.,Moore,W.E.,Zank,L.C.,1961.Thespectrophotometricdeterminationoflignininsmallwoodsamples.Tappi44,793–798.

Jung,H.G.,Allen,M.S.,1995.Characteristicsofplantcellwallsaffectingintakeanddigestibilityofforagesbyruminants.J.Anim.Sci.73,2774–2790. Jung,H.G.,Varel,V.H.,1988.Influenceofforagetypeonruminalbacterialpopulationsandsubsequentinvitrofiberdigeston.J.DairySci.71,1526–1535. Jung,H.G.,Vogel,K.P.,1986.Influenceofligninondigestibilityofforagecellwallmaterial.J.Anim.Sci.62,1703–1712.

Jung,H.G.,Mertens,D.R.,Payne,A.J.,1997.CorrelationofaciddetergentligninandKlasonligninwithdigestibilityofforagedrymatterandneutraldetergent ﬁber.J.DairySci.80,1622–1628.

Kondo,T.,Mizuno,K.,Kato,T.,1987.Somecharacteristicsofforageplantlignin.Jpn.Agric.Res.Quart.21,47–52.

Licitra,G.,Hernandez,T.M.,VanSoest,P.J.,1996.Standardizationofproceduresfornitrogenfractionationofruminantfeeds.Anim.FeedSci.Technol.57, 347–358.

Lowry,J.B.,Conlan,A.C.,Shlink,A.C.,Mcsweeney,C.S.,1994.Aciddetergentdispersibleligninintropicalgrasses.J.Sci.FoodAgric.65,41–49. McDougall,E.I.,1949.Studiesonruminalsaliva.1.Thecompositionandoutputofsheep’ssaliva.J.Biochem.43,99–109.

Mertens,D.R.,2002.Gravimetricdeterminationofamylase-treatedneutraldetergentﬁberinfeedswithreﬂuxinginbeakersorcrucibles:collaborative study.J.AOACInt.85,1217–1240.

Morrison,I.M.,1972.Improvementsintheacetylbromidetechniquetodetermineligninanddigestibilityanditsapplicationtolegumes.J.Sci.FoodAgric. 23,1463–1469.

Pell,A.N.,Schoﬁeld,P.,1993.Computerizedmonitoringofgasproductiontomeasureforagedigestioninvitro.J.DairySci.76,1063–1073.

Popoff,T.,Theander,O.,1976.Formationofaromaticcompoundsfromcarbohydrates.IV.Chromonesfromreactionofhexuronicacidsinslightlyacid, aqueoussolution.ActaChem.Scand.30,705–710.

Savioli,N.M.,Fukushima,R.S.,Lima,C.G.,Gomide,C.A.,2000.Yieldandspectrophotometricpatternofligninextractedfromcellwall,neutraldetergent ﬁberoraciddetergentﬁber.R.Bras.Zootec.29,988–996(inPortuguesewithEnglishabstract).

Theander,O.,Westerlund,E.A.,1986.Studiesondietaryﬁber.3.Improvedproceduresforanalysisofdietaryﬁber.J.Agric.FoodChem.34,330–336. Traxler,M.J.,Fox,D.G.,VanSoest,P.J.,Pell,A.N.,Lascano,C.E.,Lanna,D.P.D.,Moore,J.E.,Lana,R.P.,Vélez,M.,Flores,A.,1998.Predictingforageindigestible

NDFfromligninconcentration.J.Anim.Sci.76,1469–1480.

VanMilgen,J.,Murphy,L.L.,Berger,L.L.,1991.Acompartmentalmodeltoanalyzeruminaldigestion.J.DairySci.74,2515–2529.

VanSoest,P.J.,1963.Symposiumonnutritionandforageandpastures:newchemicalproceduresforevaluatingforages.J.Anim.Sci.22,838–845. VanSoest,P.J.,1967.Developmentofcomprehensivesystemoffeedanalysesandapplicationtoforages.J.Anim.Sci.26,119–128.

VanSoest,P.J.,1994.NutritionalEcologyoftheRuminant,2nded.CornellUniversityPress,Ithaca. VanSoest,P.J.,Robertson,J.B.,1985.AnalysisofForagesandFibrousFoods.CornellUniversityPress,Ithaca.

VanSoest,P.J.,Wine,R.H.,1968.Thedeterminationofligninandcelluloseinacid-detergentﬁbrewithpermanganate.J.Assoc.Off.Anal.Chem.51,780–785. Vieira,R.A.M.,Pereira,J.C.,Malafaia,P.A.M.,Queiroz,A.C.,1997.Theinﬂuenceofelephantgrass(PennisetumpurpuremSchum.Mineirovariety)growthon

thenutrientkineticsintherumen.Anim.FeedSci.Technol.66,197–210.

Whitehead,D.L.,Quicke,G.V.,1964.Acomparisonofsixmethodsofestimatingligniningrasshay.J.Sci.FoodAgric.15,417–422. Whitmore,F.W.,1982.Lignin-proteincomplexincellwallsofPinuselliottii:aminoacidconstituents.Phytochemistry21,315–318.