IndustrialCropsandProducts95(2017)404–415
ContentslistsavailableatScienceDirect
Industrial
Crops
and
Products
j ourna l h o m e pa g e : w w w . e l s e v i e r . c o m / l o c a t e / i n d c r o p
Catechin-based
extract
optimization
obtained
from
Arbutus
unedo
L.
fruits
using
maceration/microwave/ultrasound
extraction
techniques
Bianca
R.
Albuquerque
a,b,
Prieto
M.A.
c,
Maria
Filomena
Barreiro
d,
Alírio
Rodrigues
e,
Thomas
P.
Curran
f,
Lillian
Barros
a,d,
Isabel
C.F.R.
Ferreira
a,∗aCentrodeInvestigac¸ãodeMontanha(CIMO),ESA,InstitutoPolitécnicodeBraganc¸a,CampusdeSantaApolónia,1172,5300-253Braganc¸a,Portugal bDepartamentodeAlimentos,UniversidadeTecnológicaFederaldoParaná−CampusdeMedianeira,AvenidaBrasil,4232CEP85884-000,CaixaPostal
271,Medianeira,Brasil,Brazil
cNutritionandBromatologyGroup,FacultyofFoodScienceandTechnology,UniversityofVigo,OurenseCampus,E32004Ourense,Spain
dLaboratoryofSeparationandReactionEngineering−LaboratoryofCatalysisandMaterials(LSRE-LCM),PolytechnicInstituteofBraganc¸a,Campusde
SantaApolónia,1134,5301-857Braganc¸a,Portugal
eLaboratoryofSeparationandReactionEngineering(LSRE)−AssociateLaboratoryLSRE/LCM,FacultyofEngineering,UniversityofPorto,Porto,Portugal fUCDSchoolofBiosystemsandFoodEngineering,UniversityCollegeDublin,Belfield,Dublin4,Ireland
a
r
t
i
c
l
e
i
n
f
o
Articlehistory:
Received31August2016
Receivedinrevisedform4October2016 Accepted27October2016
Availableonline5November2016 Keywords:
ArbutusunedoL.fruits Catechin
Valorisation
Maceration/microwave/ultrasoundassisted extraction
Responsesurfacemethodology
a
b
s
t
r
a
c
t
Thisstudycomparesthreeextractiontechniques(maceration,microwaveandultrasound)forcatechin recoverfromArbutusunedofruitextracts.Toobtaintheconditionsthatmaximizecatechin extrac-tionyield, aresponsesurfacemethodology wasapplied usinga3-levelfull factorialBox–Behnken designinwhichtheprocessingtime(t),temperature(T),ultrasonicpower(W)andethanolpercentage (Et%)weretherelevantindependentvariableswiththeresponse(catechincontent,mg/gdw) mea-suredbyHPLC-PDA.Afixedsolid/solventratioof50g/Lwasusedinalltechniques.Macerationand microwaveextractionswerefoundtobethemosteffectivemethods,capableofyielding1.38±0.1and 1.70±0.3mg/gdwofcatechin,respectivelyattheoptimalextractionconditions.Theoptimalconditions formacerationwere93.2±3.7min,79.6±5.2◦Cand23.1±3.7%ofethanol,whileforthemicrowave
extractionwere42.2±4.1min,137.1±8.1◦Cand12.1±1.1%ofethanol.Comparativelywith
macera-tion,themicrowavesystemwasafastersolution,conductingtoslightlyhighercatechinyields,butusing highertemperaturestoreachsimilarvalues.Theultrasoundmethodwastheleasteffectivesolution, yielding0.71±0.1mg/gdwofcatechinat42.4±3.6min,314.9±21.2Wand40.3±3.8%ethanol.The resultshighlightthepotentialofusingA.unedofruitsbio-residuesasaproductivesourceofcatechin.
©2016ElsevierB.V.Allrightsreserved.
1. Introduction
ThesmalltreeofArbutusunedoL.(knownasstrawberrytree), belongingto theEricaceae family, is a native species fromthe Mediterraneanregion.It produces an ediblereddish sweet and tastyberrywhichis,whenfullymatured,richinnutritional prop-erties and has several medicinaleffects, as astringent, diuretic andantisepticproperties(Ziyyatetal.,2002).Someauthorshave reportedthepresenceofphenoliccompoundsinA. unedofruits (Alarcão-E-Silva et al., 2001;Fortalezas et al., 2010;Guimarães et al., 2013; Pabuc¸cuo˘glu et al., 2003; Pawlowska et al., 2006; Ruiz-Rodríguezetal.,2011), inparticularcatechins(monomeric flavan-3-olsandprocyanidinspolymericflavan-3-ols)(Gadkariand
∗ Correspondingauthor.
E-mailaddress:[email protected](I.C.F.R.Ferreira).
Balaraman,2015;Guimarãesetal.,2013;Pallaufetal.,2008). Cat-echinsareflavan-3-olsthathaveattractedattentionparticularly duetotheirrelativehighantioxidantcapacity(AronandKennedy, 2008).Severalstudiespointedouttheinterestofusingcatechinsfor healthbenefits,suchascancerpreventionandplasmaoxidation, aswellasobesitycontrol(HigdonandFrei,2003;Hirasawaand Takada,2004;LotitoandFraga,1998;Nagaoetal.,2009).However, flavan-3-ols,and particularlycatechin, aresusceptibleto degra-dationundervariousconditions,highlightingtheimportanceof optimizingtheextractionconditionstomaximizetheyieldinthese compounds(Ananingsihetal.,2013).
Abroadspectrumofsolid-liquidproceduresisavailableforthe extractionandisolationofnewfunctionalingredients(Galanakis, 2012).However,sometechniquescomprisedisadvantages requir-inglongtimes,largesolventconsumptionandleadingtothermal degradationofphenoliccompounds(Alonso-Salcesetal.,2001;Dai andMumper,2010;Inceetal.,2013).Thechoiceoftheextraction http://dx.doi.org/10.1016/j.indcrop.2016.10.050
solventhasalsoimpactinextractionyield.Catechinsaregenerally extractedwithwater,polarorganicsolvents,andaqueousorganic solventmixtures(Pi ˜neiroetal.,2004;Vuongetal.,2010). Mac-erationextraction(ME)isaconventionalmethodfrequentlyused intheextractionofbioactivecompounds.Theprocedureconsists in stirringthe samplein a solvent for a certainperiod of time andataspecifictemperature.Itisa simpletechnique,butvery oftenrequireslongtimeperiodsandhightemperatures.To over-comethesedrawbacks,alternativeextractionmethodsarebeing proposed, suchas microwaveand ultrasound basedtechniques (Galanakis,2013).Ultrasound-assistedextraction(UAE)isa tech-niquethat isincreasingly usedin chemical and foodindustries (Chematetal.,2017).Thistechniquehasadvantagessuchasbeing fasterthanconventionalextractionmethodologies,itis energeti-callylessdemandingandoftenpermitsthereductioninthesolvent consumption.Addingtothis,itgenerallyresultsinextractswith improvedpurityandyield.Suchadvantagesresultfromthe pro-cessprinciplethatisbasedoncavitationeffectscausingtherupture ofplantcellwalls,thusincreasingthecontactareabetweenthe solidandsolvent(Ghasemzadehetal.,2014;HerreraandLuqueDe Castro,2004;Pingretetal.,2012).Ontheotherhand, microwave-assistedextraction(MAE)isaprocessthatfacilitatesthepartition ofthesamplecompoundsintothesolvent,decreasingthe extrac-tiontimeandtemperature,andincreasingtheprocessefficiency usingloweramountsofsolvent(Lietal.,2013a).Thismethodhas beenemployedintheextractionofmedicinalherbalcompounds (Chenetal.,2008;DaiandMumper,2010;ProestosandKomaitis, 2008).
ME,MAEandUAEdependonseveralprocessvariableswhose valuescannotbegeneralizedforallmatricesduetotheirspecificity intermsofcompositionandtargetcompounds(Jacotet-Navarro etal.,2016).Thusoptimizationofprocessvariablesisneededto selectthebestconditionstoensureamaximumyield,minimum timeconsumption, energyandsolvent,obtainingthemaximum benefitfromthetechnique(Lietal.,2013b).Traditionally, opti-mizationisachievedbymonitoringtheinfluenceofonefactorata time.However,byusingtheresponsesurfacemethodology(RSM), optimizationisdonesimultaneouslyandinamultivariableform; theinteractioneffectsbetweenthefactorscanbeassessed allow-ingamuchmorepreciseidentificationoftheoptimalconditions. TheRSM,bymeansofmathematicalequations,candescribethe behaviourofthevariousvariablesandforecasttheresultsforthe system(Bezerraetal.,2008;Ferreiraetal.,2007;KalilandMaugeri, 2000).
Thepresentstudyaimstooptimizecatechinextractionyield fromA.unedofruitstobeconsideredforfood,pharmaceuticaland cosmeticindustries. Different extractionmethodologiessuchas ME,MAEand UAEwerestudiedandcompared.Thejointeffect oftherelevantvariablesforeachtechnique,tomaximizecatechin extractionyield,wasdescribedthroughRSM,contributingtothe understandingoftherealpotentialofcatechinobtainmentfromA. unedoforindustrialapplications.
2. Materialsandmethods 2.1. Sourcematerial
ThefruitsofArbutusunedoL.(strawberrytree)fromEricaceae weregatheredintheNaturalParkofMontesinhoterritory,in Trás-os-Montes, North-eastern Portugal. The botanical identification was confirmed by Dr. Ana Maria Carvalho (School of Agricul-ture,PolytechnicInstituteofBraganc¸a,Trás-os-Montes,Portugal) accordingwithapreviousreportoftheauthors(Guimarãesetal., 2013).Thefruitswerelyophilized(FreeZone4.5,Labconco,Kansas
City,MO,USA)andstoredinadeep-freezerat−20◦Cforsubsequent
analyses.
2.2. Standardsandreagents
FormicacidandacetonitrileofHPLCgradefromFisher Scien-tific(Lisbon,Portugal)wereused.Catechinstandardwaspurchased fromSigma(St.Louis,MO,USA).WaterwastreatedinaMilli-Q waterpurificationsystem(TGIPureWaterSystems,Greenville,SC, USA).Allotherchemicalsandsolventswereofanalyticalgradeand purchasedfromcommonsuppliers.
2.3. Extractiontechniques
Fromacombinationofsinglevariablepreliminaryexperiments, previousextractionsperformedinourlaboratoryandbibliographic survey,therelevantvariablesandtheappropriatetestedrangesfor eachofthestudiedextractiontechniqueswereselectedandtested. Adetaileddescriptionofthestudyrangesfortheselectedvariables ineachtechnique(RSMdesign)aredescribedinTableA1 (Supple-mentalmaterialsection).Thesolid/solventratiowaskeptconstant (50g/L)foralltechniques.Theusedsolventwasanethanol/water mixturecharacterizedintermsofethanolcontent.
2.3.1. Macerationextraction(ME)
Thelyophilizedpowdered fruitsamples(1g)wereplaced in a beakerwith 20mL of solvent in order to obtainthe desired solid/liquidratio(50g/L).Thebeakerwasplacedinathermostated water bath under continuous electro-magnetic stirring for the requiredtimeperiod.Thevariablesandrangestestedwere:time (torX1,20–150min),temperature(TorX2,20–90◦C)andethanol
percentage(SorX3,0–100%).
2.3.2. Microwave-assistedextraction(MAE)
MAEprocesswasperformedusingaBiotageInitiatorMicrowave (Biotage® Initiator+,Uppsala,Sweden)using closedvessels.The
lyophilized powderedsamples (1g) wereextracted with20mL ofsolvent(solid/solventratio50g/L).Inmicrowavesystemsthe pressureandTarecorrelatedandtheappliedpowerlinkedtothe neededttoreachtheselectedTorpressure.Inconsequence,Twas selectedasthemainvariableandthemicrowavepowerwasset to400W.Undertheselectedconditions,theneededttoreachthe selectedTwasalwayslessthan20sthusguarantyingafastheating process(thistimecanbeneglectedfacetothestudiedextraction timerange).Therefore,thefinalvariablesandrangestestedweret (X1,1.6–45min),T(X2,50–145◦C)andS(X3,0–100%).
2.3.3. Ultrasound-assistedextraction(UAE)
The UAE was carried out in an ultrasonic device (QSonica sonicators, model CL-334, Newtown, CT, USA). The lyophilized powderedsamples(2.5g)wereextractedwith50mL(solid/solvent ratio50g/L)bytheultrasonicdeviceatdifferenttimes(torX1,
5to55min) andatdifferentultrasound powerranges(P orX2,
100–400W)accordingtoanethanolcontent(SorX3,0–100%)while
temperaturewasmonitoredinordertobebelow30–35◦C. 2.4. Extractpurification
ThecollectedextractswerefilteredthroughaWhatmanpaper filtern◦4.Then,thefilteredmaterialwasdriedat40◦Cinarotary evaporatorBüchiR-210(Flawil,Switzerland).Forpurification,aC18
SepPak® Vac3cccartridge(Phenomenex)wasused.Afterbeing
activatedwithethanolfollowedbywater;sugarsandmorepolar substanceswereremovedbypassingthecolumnwith10–20mLof water.Thenthepurifiedextractwasfurtherelutedwith10–15mL
406 B.R.Albuquerqueetal./IndustrialCropsandProducts95(2017)404–415 of ethanol. The purified extract was dried at 40◦ C toremove
ethanol.
2.5. CatechinquantificationbyHPLC-PDA
Thesamplesobtainedduringtheextractionoptimization stud-ieswereanalysedusinga Shimadzu20AseriesUFLC(Shimadzu Corporation,Kyoto,Japan) witha quaternarypumpand a pho-todiodearraydetector(PDA)coupledtoanLCsolutionsoftware data-processingstation.SeparationwasachievedusingaWaters SpherisorbS3ODS-2C18,(3m,4.6mm×150mm)column
oper-atingat35◦C.Theusedmobilephasewasamixtureofformicacid inwater0.1%(A)and100%ofacetonitrile(B),andtheestablished elutiongradientwasasfollows:15%Bfor5min,15%Bto20%B over5min,20–25%Bover10min,25–35%Bover10min,35–50%B for10min,andcolumnre-equilibration(15min),usingaflowrate of0.5mL/min.DetectionwascarriedoutinthePDAat280nmas preferredwavelength.CatechinwasidentifiedbycomparingitsUV spectraandretentiontimeswiththeonesofacommercialstandard asreportedpreviously(Guimarãesetal.,2013).Thequantitative analysiswasperformedusingacalibrationcurvebasedoncatechin (y=66243x-343411;R2=0.999).Resultswereexpressedinmgof
catechinpergofdryfruitweight(mg/gdw).
2.6. Responsesurfacemethodology 2.6.1. Experimentaldesign
Foreachextractiontechniquethreevariableswereselectedas therelevantones.Thosevariableswerestudiedinconjunctionwith astructuredexperimentaldesigncriteria(BoxandHunter,1957) usinga responsesurfacemethodology.Initially,threeRSM vari-ableswereappliedforeachtechniquetooptimizetheextracting conditions.If thetested experimentalrangefailed toprovide a globaloptimuminanyofthethreevariables,arelativeoptimum withinthetestedrangewasattemptedthroughanotherRSMdesign involvingtheunresolvedvariablecombinedwiththeother rele-vantvariablesoftheextractionsysteminacomplementarytwo variablesRSMdesign.Therefore,forcomplexscenarios,thetwo differentexperimentaldesignsusedfor theoptimizationofthe extractionconditionsineachtestedtechniquewereasfollows:
a)Fortheanalysisofthreevariables(X1-3):acircumscribedcentral
compositedesign(CCCD)wasused.Inthisdesignthe experimen-talpointsaregeneratedonaspherearoundthecentrepoint. Thisdesignrequires5levelsforeachfactorand3replicatesper coordinate.
b)For theanalysisof twovariables(X1-2):a fullfactorial design
(FFD)withthreereplicatesperconditionwasused.Thestructure ofaFFDimpliesthatallcombinationsofthreevalues,foreach factor,arestudied(minimum,meanandmaximum).
ForbothRSMdesign,thecentrepointisassumedasavalueclose totheoptimumpositionfortheresponse,beingrepeatedinorder tomaximizethepredictionprecision(Boxetal.,2005). Experimen-talrunswererandomizedtominimizetheeffectsofunexpected variabilityintheobservedresponses.Adetaileddescriptionofthe mathematicalexpressionstocalculatethedesigndistributionand todecodeandcodethetestedvariable’srangescanbefoundinthe Appendixsection.
2.6.2. Mathematicalmodelling
Independently ofthe RSMused(two orthree variables)the model for the analysis of the produced responses follows this second-orderpolynomialequation:
Y=b0+ n
i=1 biXi+ n−1 i=1 j>i n j=2 bijXiXj+ n i=1 biiXi2 (1)whereYisthedependentvariable(responsevariable)tobe mod-elled,XiandXjdefinetheindependentvariables,b0istheconstant
coefficient,biisthecoefficientoflineareffect,bijisthecoefficient ofinteractioneffect,biithecoefficientsofquadraticeffectandnis thenumberofvariables.Althoughthestatisticalconsistentmodel parametersobtainedareempiricalandcannotbeassociatedwith amechanisticmeaning,theyareusefultopredicttheresultsof untestedoperationconditions(Pinelaetal.,2016).Thesignofthe effectmarkstheresponseperformance.Inthisway,whenafactor hasapositiveeffect,theresponseishigheratthehighleveland whenafactorhasanegativeeffect,theresponseisloweratthe highlevel.Thehighertheabsolutevalueofacoefficient,themore importanttheweightofthecorrespondingvariable(Helenoetal., 2016).
2.6.3. Proceduretooptimizethevariablestoamaximumresponse Foroptimizationofcatechinextraction,amaximizedprocess ofthemodel producedresponses wasachieved,usinga simple methodtool tosolvenon-linear problems(Helenoet al.,2016; Pinelaetal.,2016).Limitationsweremadetothevariablecoded valuestoavoidunnaturalconditions(i.e.,timeslowerthan0). 2.7. Numericalmethods,statisticalanalysisandgraphical illustrations
Allfittingprocedures,coefficientestimationsandstatistical cal-culations were performed using a Microsoft Excel spreadsheet andthepresentedgraphicalillustrationsweredevelopedinthe softwareDeltaGraphV6.Fittingandstatisticalanalysisofthe exper-imentalresults,accordingtothedisplayedequations,werecarried outinfourphases:
-Coefficientsdetermination: Parametricestimates wereobtained byminimization of thesumofquadratic differences between observedandmodel-predictedvalues,usingthenonlinear least-square(quasi-Newton)methodprovidedbythemacroSolverin MicrosoftExcel2003(KemmerandKeller,2010), whichallows aquicktestingofhypothesesandanalysisofitsconsequences (MuradoandPrieto,2013).
-Coefficientssignificance:Determinationoftheparametric confi-denceintervalsusingthe‘SolverAid’(Prikler,2009).Themodel wassimplifiedbyexcludingthevalueswhichwerenot statisti-callysignificantat␣=0.05.
-Modelconsistency:TheFisherFtest(␣=0.05)wasusedto deter-minewhethertheconstructedmodelswereadequatetodescribe theobserveddata(ShiandTsai,2002).
-Otherstatisticalassessmentcriteria:Forconfirmationofthe uni-formity of the model, the following criteria were applied: a) The‘SolverStat’macro(Comuzzietal.,2003)whichisusedfor theassessmentofuncertaintiesrelatedtoparameterandmodel predictions;b)R2 whichisinterpretedastheproportionofthe
variabilityofthedependentvariableexplainedbythemodel;c) Adjustedcoefficientsofmultipledetermination(R2
adj),whichis
acorrectiontoR2 takingintoaccountthenumberofvariables
usedinthemodel;d)Biasandaccuracyfactorsofallequations werecalculatedtoevaluatethequalityoffittingsto
experimen-taldata,suchastheMeanSquaredError(MSE),theRootMean SquareoftheErrors(RMSE)andtheMeanAbsolutePercentage Error(MAPE);e)theDurbin-Watsoncoefficient(DW)totest,ifthe residualsofthemodelarenotauto-correlated;andf)theAnalysis ofVariancetable(ANOVA)toevaluatetheexplanatorypowerof thevariables.
3. Resultsanddiscussion
3.1. Preliminaryexperimentstoselecttherelevantvariablesand instrumentalparameterstocentretheirexperimentaldomain previoustotheRSMapplication
Althoughtheexistingpreviousreportsdealingwiththe opti-mizationofcatechinextractionfromnatural matrices(Table1), noreports couldbefounddescribingtheconditionsofcatechin extractionfromArbutusunedoL.fruits.Inaddition,duetothe com-positionaldiversityofthematerialsourcesdescribedinTable1, thetestedconditionscannotbedirectlyextrapolatedforcatechin extractionfromA.unedofruits.Therefore,tofindtheconditions thatmaximizecatechinextractionfromArbutusunedoL.,itis nec-essarytotakeintoaccountthevariablesthataffectsolid/solvent systemtechniquesbehaviour.Thesevariablescanbedividedinto non-intrinsicfactors(solventtype,Sandsolidtoliquidratio)and intrinsicfactors(tandTfortheMEandMAEsystems,andtand PforUAEsystem).Preliminarytestswereexaminedindividually todeterminetheirexperimentaldomain(keepingotherones con-stant)inordertoobtainaproperRSMdesignbyanalyzingtheir generalpatternresponses.
Inconsequence,inallextractingsystems,thenon-intrinsic vari-ablesandrangeswereselectedasfollows:
1)Theextractingsolventtypeisakeyfactorfortheseparationof thedesiredcompounds.Duetothecatechinchemicalstructure, differentsolventmixtureswithwaterwereusedtomaximize extractionyields;mainly,waterwithmethanol,ethanolor ace-tonedifferentcontents(moredetailsinTable1).Duetogreen chemistry principles,binary mixtures of ethanol with water were selected as the extraction solvent. In all systems, the ethanolcontent inthewater/ethanol mixture(S) wastested from0to100%andconfirmedasimpactingsignificantlythe cat-echinextractionyieldand,therefore,selectedintheappropriate range.
2)Withregardtosolid/liquidratio,thetestedrangewas1–60g/L lowervaluesleadtoanenhancedextractionyield,butalso con-tributetoasignificantwasteofsolvent.Ahighersolid/liquid ratiowill resultin lower catechin extractionyields but in a betterrationalizationofrawmaterialsconsumption.However, lowerdifferenceswerefound,discardingthesolid/liquidratio andselectingthe50g/L asthevaluetobeusedin alltested extractiontechniques.
ConcerningtheintrinsicvariablesfromtheME,MAEandUAE,a literaturesurveyconcerningthemainranges,asstudiedinsimilar processes(Table1), wascarriedout.Althoughgoodconclusions canbederivedfromthisreport,resultsmaybehighlydependent onvariationsnotforeseeninthesestudieswherecertainvariables thatremainedconstant,togetherwiththevariabilityintheused rawmaterialstoextractcatechin,canhighlyinfluencetheprocess. Inconclusion,thefirstapproachtooptimizetheefficiencyofthe ME,MAEandUAEprocessesforcatechinextraction,wasperformed bytheapplicationofaRSMofthreevariablesinaCCCD.Five lev-elsofvariationfortheindependentvariablesoft(20–120min),T (20–90◦C),andS(0–100%)forMEandoft(1–20min),T(50–120◦C), andS(0–100%)forMAEandt(5–55min),P(100–400W),andS
(0–100%)forUAEwereused.Adetaileddescriptionofthecoded andnaturalvaluesoftheselectedvariablesforeachtechniquein theCCCDwithtreevariablesispresentedinTableA1(Supplemental materialsection).
3.2. RSMoutputforaCCCDwiththreevariables
TheresultsobtainedaccordingtothestatisticalCCCDareshown in thefirst partof Table2for eachof thecomputedextraction techniques.AfterfittingEq.(1)totheresponseresultsofTable2 usinganon-linearleast-squaresprocedure,theestimated paramet-ricvalues,parametricintervalsandnumericalstatisticalcriteria wereobtainedandpresented inthefirstpartofTable 3.Those coefficients,whichshowedeffectswithcoefficientintervalvalues (␣=0.05)higherthantheparametervalue,wereconsideras non-significant(ns)andwerenotponderedforthemodeldevelopment. Therefore,mathematicalmodelswerebuilt,obtainingthe fol-lowingsecond-orderpolynomialequationsaccordingtoEq.(1)for eachoftheassessedextractiontechniques:
forME:YME=1.1+0.05t+0.24T−0.23S/L−0.11t2−0.15T2
−0.13S/L2
−0.1tT (2)
forMAE:YMAE=0.66+0.09t+0.08T
−0.04S/L−0.01t2−0.03T2−0.11S/L2
+0.03tT−0.04tS/L−0.03TS/L
(3)
forUAE:YUAE=0.69+0.02t+0.02T−0.8S/
L−0.02t2−0.03T2−0.13S/L2
(4)
TheEqs.(2)–(4)translatetheresponsepatternsforeach extrac-tiontechniqueshowinghighlycomplexsceneries(Table3).Linear andquadraticeffectsarefoundplayinganimportantand signifi-cantroleinallextractingsystems.Regardingtheinteractiveeffects, forMEsystemonlytheinteractionbetweent&Twassignificant inapositivemode;forMAEallthevariableinteractionscauseda significanteffect(positivefort&T,andnegativefort&SandT& S);andforUAEnosignificanteffectswerefound.
Fig.1showstheextractionresultsinmg/gdwofcatechin,which isdividedinthreecolumnsforeachoneofthetestedtechniques. Eachcolumnisdividedintotwosubsections(AandB).The subsec-tionAshowsthecombinationofthethree-dimensionalresponse surfaceplotspredictedwiththeirrespectivesecondorder polyno-mialequationdescribedbyEqs.(2)–(4)asafunctionofeachone oftheinvolvedvariables.Thebinaryactionbetweenvariablesis presentedwhentheexcludedvariableispositionedatthecentre oftheexperimentaldomain(seeTableA1).SubsectionBillustrates thecapabilitytopredicttheobtainedresultsandtheresidual dis-tributionasafunctionofeachoneoftheconsideredvariables.
Inalmostallcombinatory3DresponsesofFig.1,theamount of extracted material increases to an optimum value and then decreasesasafunctionofeachoneoftheassessedindependent variables.Therefore,inalmostallcombinationstheoptimumcan befoundatone singlepoint alongwiththeresponse, allowing computingtheconditionsthatleadtotheabsolutemaximum.
By applying a simple procedure considering restrictions to theexperimental ranges,optimalconditions are found, as well as themaximal response values (firstpartof Table 4).For the MEsystem,therelativeoptimal(*)orabsoluteconditionsfound were at 88.3±31.8min, 79.2±15.7◦C and 23.1±3.7% ethanol, producing a maximum response value of 1.36±0.5mg of cat-echin/g dw. For MAE response, the optimal condition values wereat*18.4±1.7min,*118.6±21.3◦Cand12.1±1.1%ethanol,
408 B.R. Albuquerque et al. / Industrial Crops and Products 95 (2017) 404–415 Table1
Bibliographicsummaryofcatechincontentfromdifferentsourcematerialsusingdifferentextractiontechniquesandconditions.
TECHNIQUEAPPLIED SOURCEMATERIAL PLANTPART EXTRACTIONCONDITIONS CATECHINCONTENT REFERENCE Solvent Temperature Poweror
Frequency(kHzor W) pH Time – (oC) – (min) (mg/gdw) ULTRASOUNDASSISTED EXTRACTION
Apple Pulp Water 40 25kHz – 45to90 0.030to0.075 Mieszczakowska-Fr ˛acetal.(2015)
Apple Pomace Water 40 150W 3.8 40 0.15 Pingretetal.(2012)
Curry Leaves Methanol:Water
(80:20)
56 145W – 20 0.48 Ghasemzadehetal.(2014)
Grape Seeds Methanol:Water
(75:25)
– – – 15 0.41 García-Marinoetal.(2006)
Grape Seeds Methanol 60 – – 10 0.23 Pi ˜neiroetal.(2004)
Grape Seeds Methanol:Water
(10:90)
30 – – 30x2 0.65 PalmaandTaylor(1999)
Maritimepine Plant Water 40 – 3.8 43 3.5 Meullemiestreetal.(2016)
Melissa Leaves Water – 150W – 20 2.01 Inceetal.(2013)
Mushroon Ethanol:Water
(60:40)
25 – – 30 0.1 Zhangetal.(2012)
Pistachio Nut Water – 35kHz – 30 0.05 Garavandetal.(2015)
Strawberrie Fruit Acetone – 100W <3 0.5x3 0.015 HerreraandLuqueDeCastro
(2004) MACERATIONASSISTED
EXTRACTION
Apple Pulp Water 40 – – 45–90 0.040to0.050 Mieszczakowska-Fr ˛acetal.(2015)
Apple Pomace Water 40 – 3.8 40 0.115 Pingretetal.(2012)
Foliumeriobotryae Fruit Methanol – 700W – 3 12.1 Chenetal.(2008)
Grape Seeds Methanol:Water
(4:1)
30 – – 960 0.7 PalmaandTaylor(1999)
Greentea Leaves Water 25to80 – <6 30to120 0.600to7.100 Vuongetal.(2011)
Melissa Leaves Water 40 – – 1440 3.45 Inceetal.(2013)
Mushroon Ethanol:Water
(60:40)
25 – – 720 0.1019 Zhangetal.(2012)
Grape Seeds Methanol 60 – – 10 0.27 Pi ˜neiroetal.(2004)
HEATREFLUXEXTRACTION F.eriobotryae Fruit Methanol 80 – – 3 7.34 Chenetal.(2008)
Rosemary Plant Methanol:Water
(60:40)
90 – – 120 0.019 ProestosandKomaitis(2008)
Mushroon Ethanol:Water
(60:40)
90 – – 50 0.1023 Zhangetal.(2012)
MICROWAVEASSISTED EXTRACTION
Rosemary Plant Methanol:Water
(60:40)
– 750W – 4 0.025 ProestosandKomaitis(2008)
Pistachio Nut Water – 800W – 0.5 0.0467 Garavandetal.(2015)
Melissa Leaves Water – 407W – 5 1.353 (Inceetal.(2013)
Mushroon Ethanol:Water
(60:40)
110 500W – 10 0.1079 Zhangetal.(2012)
PRESSURIZEDLIQUID EXTRACTION
Apple Peel Methanol 40 – – 5 0.043 Alonso-Salcesetal.(2001)
Apple Pulp Methanol 40 – – 5 0.018 Alonso-Salcesetal.(2001)
Grape Seeds Methanol 130 – – 10 1.82 Pi ˜neiroetal.(2004)
SUPRECRITICAL EXTRACTION
Grape Seeds Methanol:Water
(10:90)
55 – – 60 0.865 PalmaandTaylor(1999)
Fig.1.ShowsthegraphicalresultsintermsoftheextractionbehaviourfortheCCCD.PartA:Showsthejointgraphical3Danalysisasafunctionofeachthevariablesinvolved. Eachofthenetsurfacesrepresentsthetheoreticalthree-dimensionalresponsesurfacepredictedwiththesecondorderpolynomialofEqs.(2)–(4).Thebinaryactionsbetween variablesarepresentedwhentheexcludedvariableispositionedatthecentreoftheexperimentaldomain(TableA1).Thestatisticaldesignandresultsaredescribedin Table2.EstimatedparametricvaluesareshowninTable3.PartB:Toillustratethegoodnessoffit,twobasicgraphicalstatisticcriteriaareused.Thefirstone,theabilityto simulatethechangesoftheresponsebetweenthepredictedandobserveddata;andthesecondone,theresidualdistributionasafunctionofeachofthevariables.Noteall thedifferencesintheaxesscales.
410 B.R.Albuquerqueetal./IndustrialCropsandProducts95(2017)404–415 Table2
PartAshowstheRSMresultsoftheCCCDforthefirstapproachfortheoptimizationofthethreemainvariablesinvolved(X1,X2andX3)intheME,MAEandUAE.PartB
showstheRSMresultsoftheFFDfortheoptimizationofthetwovariables(X1andX2)involvedfortheMEandMAE(variables,naturalvaluesandrangesinTableA1).Three
replicates(r1-3)wereperformedforeachconditionforeachtechnique.
VARIABLECODEDVALUES CATECHINCONTENT(mg/gdw)
MACERATION MICROWAVE ULTRASOUND
X1 X2 X3 r1 r2 r3 r1 r2 r3 r1 r2 r3
A)CIRCUMSCRIBEDCENTRALCOMPOSITEDESIGN(CCCD)
−1 −1 −1 0.84 0.86 0.87 0.31 0.32 0.33 0.66 0.65 0.65 1 −1 −1 0.74 0.76 0.77 0.55 0.56 0.57 0.67 0.67 0.67 −1 1 −1 1.12 1.15 1.16 0.50 0.50 0.51 0.68 0.69 0.69 1 1 −1 1.32 1.35 1.36 0.95 0.98 0.99 0.69 0.69 0.69 −1 −1 1 0.38 0.39 0.40 0.38 0.39 0.40 0.42 0.42 0.42 1 −1 1 0.30 0.30 0.30 0.45 0.46 0.47 0.45 0.47 0.46 −1 1 1 0.65 0.67 0.68 0.47 0.47 0.48 0.47 0.45 0.46 1 1 1 0.94 0.97 0.99 0.61 0.61 0.62 0.48 0.50 0.49 −1.68 0 0 0.81 0.84 0.85 0.49 0.46 0.47 0.61 0.60 0.60 1.68 0 0 0.99 1.03 1.04 0.74 0.76 0.77 0.71 0.71 0.71 0 −1.68 0 0.40 0.41 0.41 0.49 0.52 0.53 0.58 0.60 0.59 0 1.68 0 1.18 1.21 1.23 0.72 0.72 0.73 0.70 0.70 0.70 0 0 −1.68 1.14 1.17 1.19 0.39 0.41 0.41 0.39 0.40 0.39 0 0 1.68 0.39 0.39 0.4 0.30 0.31 0.32 0.23 0.23 0.23 0 0 0 1.17 1.21 1.23 0.63 0.64 0.65 0.69 0.69 0.69 0 0 0 1.17 1.21 1.23 0.65 0.67 0.68 0.68 0.68 0.68 0 0 0 1.17 1.21 1.23 0.66 0.70 0.71 0.71 0.71 0.71 0 0 0 1.17 1.21 1.23 0.66 0.68 0.69 0.69 0.70 0.70 0 0 0 1.17 1.21 1.23 0.67 0.68 0.69 0.71 0.71 0.71 0 0 0 1.17 1.21 1.23 0.67 0.68 0.69 0.69 0.70 0.70 B)FULLFACTORIALDESIGN(FFD)
−1 −1 − 1.16 1.16 1.16 0.93 0.94 0.94 − − − 0 −1 − 1.24 1.25 1.25 1.35 1.36 1.37 − − − 1 −1 − 0.83 0.83 0.83 1.51 1.52 1.53 − − − −1 0 − 1.30 1.30 1.30 1.22 1.23 1.24 − − − 0 0 − 1.40 1.40 1.40 1.60 1.61 1.62 − − − 1 0 − 1.00 1.01 1.00 1.72 1.74 1.75 − − − −1 1 − 1.27 1.27 1.27 1.19 1.20 1.20 − − − −1 −1 − 1.38 1.38 1.38 1.53 1.54 1.54 − − − Table3
Parametricresultsofthesecond-orderpolynomialequation(Eq.(1))foreachoftheextractingtechniqueassessedaccordingtotheCCCDwith5rangelevels(partA)andFFD with3rangelevels(partB).Theparametricsubscript1,2and3standsforthevariablesinvolvedt,TandS,respectively.Analysisofsignificanceoftheparameters(␣=0.05) arepresentedinnaturalvalues.Additionally,thestatisticalinformationofthefittingproceduretothemodelispresented.
COEFFICIENTS RESPONSES
CENTRALCOMPOSITEDESIGN FULLFACTORIALDESIGN
MACERATION MICROWAVE ULTRASOUND MACERATION MICROWAVE Fittingcoefficientsobtained
Intercept b0 1.126±0.10 0.668±0.37 0.6960.016 1.351±0.02 1.613±0.02 Lineareffect b1 0.051±0.01 0.091±0.02 0.0190.01 0.076±0.01 0.089±0.01 b2 0.241±0.06 0.076±0.02 0.0220.01 −0.114±0.01 0.219±0.01 b3 −0.235±0.06 −0.034±0.02 −0.0840.01 Quadraticeffect b11 −0.112±0.03 −0.009±0.00 −0.0170.01 −0.079±0.01 −0.184±0.02 b22 −0.154±0.06 −0.025±0.00 −0.030.01 −0.227±0.01 −0.15±0.02 b33 −0.13±0.06 −0.111±0.02 −0.1290.01 Interactiveeffect b12 0.09±0.08 0.025±0.02 ns 0.017±0.01 −0.041±0.01 b13 ns −0.042±0.02 ns b23 ns −0.025±0.02 ns
Statisticalinformationofthefittinganalysis
Obs 60 60 60 27 27 df 51 49 52 20 20 R2 0.967 0.9655 0.9479 0.986 0.9879 R2adj 0.9326 0.9593 0.9386 0.9625 0.9631 MEC 0.131 0.0311 0.022 0.0864 0.049 RMSE 0.362 0.1762 0.1483 0.294 0.2213 MAPE 2.0099 4.0771 4.6536 0.3814 0.0917 DW 1.0598 2.5446 1.6565 0.236 2.9929 ns:nonsignificantcoefficient;Obs:Numberofobservations;df:Numberofdegreesoffreedom;R2:Correlationcoefficient;R2adj:Theadjusteddeterminationcoefficientfor
themodel;MSE:TheMeanSquareoftheError;RMSE:TheRootMeanSquareoftheErrors;MAPE:TheMeanAbsolutePercentageError;andDW:TheDurbin-Watsonstatistic.
producinga maximumresponseof0.97±0.2mg catechin/gdw.
For UAE, the optimal conditions were found at 42.4±4.1min,
314.9±21.2W and 40.3±3.8% ethanol obtaining a maximum
Table4
VariableconditionsinnaturalvaluesthatleadtooptimalresponsevaluesforthefirstapproximationRSMusingaCCCDandforthesecondusingaFFDforeachofthe extractingtechniquesassessed.
CRITERIA OPTIMALVARIABLECONDITIONS OPTIMUMRESPONSE X1:t(min) X2:T(
o
C)orP(W) X3:S(%)
IndividualoptimalvariableconditionsfortheCCCD:
Maceration 88.3±31.8 79.2±15.7* 23.1±3.7 1.36±0.5 mg/gdw Microwave 18.4±1.7* 118.6±21.3* 12.1±1.1 0.97±0.2 mg/gdw Ultrasound 42.4±4.1 314.9±21.2 40.3±3.8 0.71±0.1 mg/gdw IndividualoptimalvariableconditionsfortheFFD:
Maceration 93.2±3.7 79.6±5.2 – 1.38±0.1 mg/gdw Microwave 42.2±4.1 137.1±8.1 – 1.70±0.3 mg/gdw GlobaloptimalvariableconditionsforthecombinationoftheCCCDandFFDresponses:
Maceration 93.2±3.7 79.6±5.2 23.1±3.7 1.38±0.1 mg/gdw Microwave 42.2±4.1 137.1±8.1 12.1±1.1 1.70±0.3 mg/gdw Ultrasound 42.4±3.6 314.9±21.2 40.3±3.8 0.71±0.1 mg/gdw
AlthoughtheCCCDwasbasedonpreliminarytestsand
biblio-graphicresults,intheproducedresponses,itwasnotpossibleto
findtheoptimalconditionsforallvariablesinthetestedextraction
techniques.Themainreasonisduetothefactthattheexperiments
wereconductedatonefactorofthetimeanalysisandtheobtained
patternsdonottakeintoaccounttheinteractiveeffects.Only
exper-imentaldesignsbasedonmultivariableanalysis(suchastheRSM)
canproducepatternsthatintegratetheinteractionsbetweenthe
variables.Thepositiveinteractionsbetweenthet&TinMEand
MAEsystemsproduced anadditionaleffectsincetheresponses
werenoteitherconclusiveenough(MEcase)orabsolutely
opti-mized(MAEcase)withinthetestedvariable’srange,findinglarge
confidenceintervalsforsomeoptimalvalues(MEcase)orrelative
optimumconditions(MAEcase)forthevariablestandT.Thelack
ofaclearabsoluteoptimumintheprovidedsolution,forcesthe
acceptanceofoneofthefollowingsolutions:1)anunreliable
opti-mumand/orarelativeoptimum;2)tousethepredictingabsolute
optimumvaluesofthedevelopedmathematicalmodel;or3)to
re-designasecondRSMaroundtherangesthatseemtobethe
opti-malonesinordertofindtheexperimentalvaluesthatwouldhelp
tofindtheabsoluteoptimumofthesevariablesinwhichthefirst
optimizationapproachfailed.Inthisstudy,solution(3)waschosen
andfurtherexperimentswereperformedusingaRSMbasedina
FFDforthespecificanalysisoftheinteractionoftandTvariables
inMEandMAEsystems.
3.3. FinaloptimizationofMEandMAEusingaRSMbasedinFFD
withthetandTvariables
ThevariablesrangeoftandTfortheFFDwereexpanded
accord-ingtotheCCCDresults(experimentaldomaininsecondpartof
TableA1,supplementalmaterial).Theobtainedresults,according
tothestatisticalFFDforMEandMAE,areshowninthesecondpart
ofTable2foreachofthecomputedextractiontechnique.Identically tothepreviousRSMapproach,Eq.(1)wasusedtofittheresultsof Table2usinganon-linearleast-squaresprocedure.Theestimated parametricvalues,parametricintervalsand numericalstatistical criteriawereobtainedandpresentedinthesecondpartofTable3. Mathematicalmodelswerebuilt,obtainingthefollowing second-orderpolynomialequationsaccordingtoEq.(1)foreachoneofthe assessedextractiontechnique:
forME:YME=1.35+0.08t−0.11T−0.08t2−0.23T2+0.08tT (5)
forMAE:YMAE=1.6+0.09t+0.2T−0.18t2−0.15T2−0.04tT (6)
Eqs. (5) and (6) complete the response patterns for each extractiontechniqueshowingnearlyidenticalsceneriestothose previously found for the CCCD approach, but covering a more
extensiverangeofthevariables,allowingtofindtheextraction conditionsthatleadtoareliableabsoluteoptimumforcatechin extraction. Linearand quadratic effects were foundto play an importantandsignificantroleinallextractionsystems.Regarding theinteractiveeffectsoftandT,forMEandMAE,significanteffects werecorroboratedinapositiveandnegativeform,respectively.
Fig.2 shows the catechin extractionresultsfor each one of thetestedtechniques(MEandMAE).ForeachtechniqueFig.2is dividedintothreesections:
–SectionAshows thecatechin extractionyield(mg/gdw)asa functionoftandTvariables.Points(䊉)representtheobtained experimentalresultsaccordingtothedescribedstatisticaldesign. The net surface representsthe theoretical three-dimensional response surfacepredictedwiththesecondorderpolynomial Eqs.(5)and(6).Estimatedparametricvaluesareshowninthe secondpartofTable3.Thebinaryactionbetweenvariablesis presentedwhentheexcludedvariableispositionedatthecentre oftheexperimentaldomain.
–SectionBpresentstwo-dimensionalrepresentationofthefitting resultsofEqs.(5)and(6)(solidline)totheexperimentalpoints (䊐minimum,♦mediumandmaximumvariablevalues)ofthe combinedeffectoftandTonthecatechinextractionyield(mg/g dw).
–SectionCshowsanillustrationforthestatisticalrobustnessofthe reachedsolution.Twobasicgraphicalcriteriaareused:theability tosimulatetheresponsechangesandtheresidualdistributionas afunctionofeachofoneofthevariables.
Inbothtechniques(MEandMAE)theTandtvariables signifi-cantlyaffectedcatechinextraction.Catechinextractionefficiency increasedwiththeincreaseofTandtuntilanabsoluteoptimum from which it decreased. By applyinga simple procedurewith restrictionstothetestedexperimentalranges,theoptimal condi-tionresultscanbefound,aswellas,themaximalcatechinyield responsevaluesforeachtechnique,beingpresentedinthesecond partofTable4.FortheMEsystem,theoptimalabsoluteconditions foundwereat93.2±3.7minand79.6±5.2◦C(ataconstant24%of ethanol)producingamaximumresponsevalueof1.38±0.1mgof catechin/gdw.FortheMAEresponse,theoptimalcondition val-ueswereat42.2±4.1minand137.1±8.1◦C(ataconstant12%of ethanol)producingamaximumresponseof1.70±0.3mgof cate-chin/gdw.
412 B.R.Albuquerqueetal./IndustrialCropsandProducts95(2017)404–415
Fig.2. ShowsthefinaloptimizationextractingresultsofMEandMAEtechniquesinaFFD.A:Catechinextractionyield(mg/gdw)asafunctionofextractingtime(t)and temperature(T).Points(䊉)representtheobtainedexperimentalresults(secondpartofTable2)accordingtothestatisticaldesigndescribed(secondpartofTableA1).The netsurfacerepresentsthetheoreticalthree-dimensionalresponsesurfacepredictedwiththesecondorderpolynomialEqs.(5)and(6).Estimatedparametricvaluesofare showninsecondpartofTable3.B:Two-dimensionalrepresentationofthefittingresultsofEqs.(5)and(6)(solidline)totheexperimentalpoints(䊐minimum,♦medium andmaximumvariablevalues)ofthecombinedeffectofPandtoncatechinextractionyield(mg/gdw).C:Toillustratethestatisticaldescription,twobasicgraphical criteriaareused:theabilitytosimulatethechangesoftheresponseandtheresidualdistributionasafunctionofeachofthevariables.
3.4. Extractiontechniquescomparison,numericaloptimal conditionsthatmaximizetheextraction,statisticalanalysisand experimentalverificationofpredictivemodels
ME,MAEandUAEextractiontechniqueshavebeenoptimized andcomparedconcerningtherecoveryofcatechinrichextracts fromA.unedofruits.Thesesolid/liquidextractionmethodshave beenappliedtotheextractionofcatechinfromdifferentsource materials(García-Marinoetal.,2006;Ghasemzadehetal.,2014; Meullemiestre et al., 2016; Mieszczakowska-Fr ˛ac et al., 2015; PalmaandTaylor,1999;Pingretetal.,2012;Pi ˜neiroetal.,2004).
Althoughascientificsurveyfocusingcatechincontentin differ-entsourcematerialswasconducted,pointedouttheuseofseveral extractiontechniques,asfarasweknow,therearenoreferences intheliteraturedescribingtheoptimizationofcatechinextraction fromA.unedofruits.Comparingtheobtainedyieldsofthepresent workwiththeonesavailableinliterature,A.unedofruitscanbe considereda suitablesourceforcatechin extractionobtainment (Ananingsihetal.,2013;GadkariandBalaraman,2015).
WhencombiningtheinformationproducedfromtheCCCDand FFDapproaches,thecompletebehaviourofeachrelevantvariable influencingcatechinextractionisdefinedinabsoluteterms.Forall techniquestheconditionsthatleadtotheoptimalvalueswere re-checkedinordertoensuretheaccuracyofthepresentedresults. Fig.3showsthesummarizedindividual2Dresponsesasafunction ofthedefined variables for ME,MAEand UAEextraction tech-niquestoguidetheselectionofthemostfavourableconditions. Thelinerepresentsthevariableresponsepatternwhenthe oth-ersarelocatedattheoptimalvaluespresentedinthethirdpart ofTable 4. Thedots () presented alongsidethe linehighlight thelocationoftheoptimalvalue.Comparingtheresultsof
extrac-tionefficienciesamongtechniques,MEandMAEgavesignificantly highervalues,whileUAEextractiongeneratedlowervalues prob-ablyduetodegradationflavan-3-olsandparticularlycatechinasit occursinothernaturalcompounds(Jacotet-Navarroetal.,2016;Li etal.,2013b).Regardingtheextractiontime,MAEwasthefastest extractionmethodwith∼45minwhileMEneeded∼95min. Con-sideringextractionefficiency,macerationgivesimilarresultsto MAE.UAEwasfoundnotadequatecatechinextractionduetoits lowextractionefficiency.
Theperformedcharacterizationtooptimizecatechinextraction yieldinME,MAEandUAEwiththeRSMprovidesastrongsolution thatminimizestheerrorswithashortnumberof experimental trialsasithasbenndemonstratedelsewhere(Roselló-Sotoetal., 2015;Wongetal.,2015).Themultivariablefittingdecreasesthe numberofparametersneededtoanalyzetheresponseleadingto betterestimationsandreducingtheirintervalofconfidence.
The lack-off-fit test used to assess the competence of the modelsshowedthatthenon-significantparametersofbothRSM approaches (Table 3) did not statistically improve the reached solutionand,in contrast,allsignificantparameterswerehighly consistent(p<0.01).ThiswasalsoverifiedbytheachievedhighR2
andR2
adjvalues,indicatingthepercentageofvariabilityexplained
bythemodel(Table3).Thedistributionoftheresidualspresented inFig.1andFig.2wasarbitrarilyaroundzeroandnogroupof valuesorautocorrelationswereobserved.Additionally,the agree-mentbetweentheexperimentalandpredictedvaluesimpliesan acceptableexplanationoftheresultsobtainedbytheindependent variablesused.Therefore, themodelsdevelopedinEqs.(2)–(6), eitherfortheCCCDorFFD,arecompletelyfunctionalandadequate tobeusedforpredictionandprocessoptimization.
Fig.3. Individual2DresponsesofME,MAEandUAEforeachvariable.Eachgraphshowsalineandadot.Thelinerepresentstheresponseofthevariablewhentheothers arepositionedattheoptimalconditionsfound(thirdpartofTable4).Thedots()presentedalongsideeachlinehighlightsthelocationoftheoptimumvalue.Linesanddots aregeneratedbythetheoreticalsecondorderpolynomialmodelsofEqs.(2)–(6).ParametricfittingvaluesobtainedarepresentedinTable3.
414 B.R.Albuquerqueetal./IndustrialCropsandProducts95(2017)404–415 4. Conclusions
TheextractionprocessforME,MAEandUAEwassuccessfully optimizedbyapplyingtheRSM.Theresultsshowedthatextraction time,temperatureandethanolcontentintheusedwater/ethanol mixtureshavesignificanteffectsonthecatechinextractionyield. MEandMAEwerefoundtobethemosteffectivemethodscapable ofyielding1.38±0.1and1.70±0.3mgofcatechin/gdwattheir optimalextractionconditions.MAEwasfoundtobefasterandmore effectiveincomparisonwithotherstudiedtechniques,butlower temperaturewasappliedinMEwithnearlyidenticalextraction yields,whichcanbetranslatedineconomicbenefits.TheUAEwas thelesseffectivesolutionintermsofcatechinextractionyield.
Thisworkoffersanoverviewthroughenvironmental compati-bleextractionprocesses,inwhichthevalorisationofA.unedofruits isperformedby‘clean’technologiesabletointegrateapotential industrialsectorinasustainableapproach(Boukroufaetal.,2015). TheobtainedresultsindicatetheviabilityofusingA.unedofruitsas aproductivesourceofcatechinbyanyoftheassessedtechniques andprovideevidenceoftheadvantagesofmaceration,microwave andultrasoundextractiontechniquesfortheirindustrial produc-tion(Achatetal.,2012).Inaddition,thepresentworkcanreinforce theproductionofA.unedofruitstoserveasasourceofbioactive compoundstobeusedasnaturaladditivesinfunctionalfoods. Acknowledgements
Theauthors thank the Foundation for Scienceand Technol-ogy (FCT, Portugal) and FEDER under Programme PT2020 for financialsupport toCIMO (UID/AGR/00690/2013)and L. Barros (SFRH/BPD/107855/2015)grant.ToPOCI-01-0145-FEDER-006984 (LALSRE-LCM),fundedbyFEDER,throughPOCI-COMPETE2020and FCT.ToXuntadeGaliciaforfinancialsupportforthepost-doctoral researcherofM.A.Prieto.B.AlbuquerquethanksCeleidePereira (UTFPR,Brazil)forhermastercosupervision.
AppendixA. Supplementarydata
Supplementarydataassociatedwiththisarticlecanbefound,in theonlineversion,athttp://dx.doi.org/10.1016/j.indcrop.2016.10. 050.
References
Achat,S.,Tomao,V.,Madani,K.,Chibane,M.,Elmaataoui,M.,Dangles,O.,Chemat, F.,2012.Directenrichmentofoliveoilinoleuropeinbyultrasound-assisted macerationatlaboratoryandpilotplantscale.Ultrason.Sonochem.19, 777–786.
Alarcão-E-Silva,M.L.C.M.M.,Leitão,A.E.B.,Azinheira,H.G.,Leitão,M.C.A.,2001.The arbutusberry:studiesonitscolorandchemicalcharacteristicsattwomature stages.J.FoodCompos.Anal.14,27–35.
Alonso-Salces,R.M.,Korta,E.,Barranco,A.,Berrueta,L.A.,Gallo,B.,Vicente,F.,2001. Pressurizedliquidextractionforthedeterminationofpolyphenolsinapple.J. Chromatogr.A933,37–43.
Ananingsih,V.K.,Sharma,A.,Zhou,W.,2013.Greenteacatechinsduringfood processingandstorage:areviewonstabilityanddetection.FoodRes.Int.50, 469–479.
Aron,P.M.,Kennedy,J.A.,2008.Flavan-3-ols:nature,occurrenceandbiological activity.Mol.Nutr.FoodRes.52,79–104.
Bezerra,M.A.,Santelli,R.E.,Oliveira,E.P.,Villar,L.S.,Escaleira,E.A.,2008.Response surfacemethodology(RSM)asatoolforoptimizationinanalyticalchemistry. Talanta76,965–977.
Boukroufa,M.,Boutekedjiret,C.,Petigny,L.,Rakotomanomana,N.,Chemat,F., 2015.Bio-refineryoforangepeelswaste:anewconceptbasedonintegrated greenandsolventfreeextractionprocessesusingultrasoundandmicrowave techniquestoobtainessentialoil,polyphenolsandpectin.Ultrason. Sonochem.24,72–79.
Box,G.,Hunter,J.,1957.Multi-factorexperimentaldesignsforexploringresponse surfaces.Ann.Math.Stat.28,195–241.
Box,G.E.P.,Hunter,J.S.,WilliamHunter,G.,2005.StatisticsforExperimenters: Design,InnovationandDiscovery,2nded.Wiley.
Chemat,F.,Rombaut,N.,Sicaire,A.,Meullemiestre,A.,Abert-vian,M.,2017. Ultrasonicssonochemistryultrasoundassistedextractionoffoodandnatural products.Mechanismstechniques,combinations,protocolsandapplications.A review.Ultrason.−Sonochem.34,540–560.
Chen,Z.,Zhang,L.,Chen,G.,2008.Microwave-assistedextractionfollowedby capillaryelectrophoresis-amperometricdetectionforthedeterminationof antioxidantconstituentsinFoliumEriobotryae.J.Chromatogr.A1193, 178–181.
Comuzzi,C.,Polese,P.,Melchior,A.,Portanova,R.,Tolazzi,M.,2003.SOLVERSTAT:a newutilityformultipurposeanalysis.Anapplicationtotheinvestigationof dioxygenatedCo(II)complexformationindimethylsulfoxidesolution.Talanta 59,67–80.
Dai,J.,Mumper,R.J.,2010.Plantphenolics:extraction,analysisandtheir antioxidantandanticancerproperties.Molecules15,7313–7352.
Ferreira,S.L.C.,Bruns,R.E.,Ferreira,H.S.,Matos,G.D.,David,J.M.,Brandão,G.C.,da Silva,E.G.P.,Portugal,L.A.,dosReis,P.S.,Souza,A.S.,dosSantos,W.N.L.,2007. Box-Behnkendesign:analternativefortheoptimizationofanalyticalmethods. Anal.Chim.Acta597,179–186.
Fortalezas,S.,Tavares,L.,Pimpão,R.,Tyagi,M.,Pontes,V.,Alves,P.M.,Mcdougall, G.,Stewart,D.,Ferreira,R.B.,Santos,C.N.,2010.Antioxidantpropertiesand neuroprotectivecapacityofstrawberrytreefruit(Arbutusunedo).Nutrients2, 214–229.
Gadkari,P.V.,Balaraman,M.,2015.Catechins:sources,extractionand encapsulation:areview.FoodBioprod.Process.93,122–138. Galanakis,C.M.,2012.Recoveryofhighadded-valuecomponentsfromfood
wastes:conventional,emergingtechnologiesandcommercialized applications.TrendsFoodSci.Technol.28,68–87.
Galanakis,C.M.,2013.Emergingtechnologiesfortheproductionofnutraceuticals fromagriculturalby-products:aviewpointofopportunitiesandchallenges. FoodBioprod.Process.91,575–579.
Garavand,F.,Madadlou,A.,Moini,S.,2015.Determinationofphenolicprofileand antioxidantactivityofpistachiohullusingHPLC-DAD-ESI–MSasaffectedby ultrasoundandmicrowave.Int.J.FoodProp.2,23–30.
García-Marino,M.,Rivas-Gonzalo,J.C.,Ib ˜nez,E.,García-Moreno,C.,2006.Recovery ofcatechinsandproanthocyanidinsfromwineryby-productsusingsubcritical waterextraction.Anal.Chim.Acta563,44–50.
Ghasemzadeh,A.,Jaafar,H.Z.E.,Karimi,E.,Rahmat,A.,2014.Optimizationof ultrasound-assistedextractionofflavonoidcompoundsandtheir pharmaceuticalactivityfromcurryleaf(MurrayakoenigiiL.)usingresponse surfacemethodology.BMCComplement.Altern.Med.14,1–10.
Guimarães,R.,Barros,L.,Due ˜nas,M.,Carvalho,A.M.,Queiroz,M.J.R.P., Santos-Buelga,C.,Ferreira,I.C.F.R.,2013.Characterisationofphenolic compoundsinwildfruitsfromNortheasternPortugal.FoodChem.141, 3721–3730.
Heleno,S.A.,Diz,P.,Prieto,M.A.,Barros,L.,Rodrigues,A.,Barreiro,M.F.,Ferreira, I.C.F.R.,2016.Optimizationofultrasound-assistedextractiontoobtain mycosterolsfromAgaricusbisporusL.byresponsesurfacemethodologyand comparisonwithconventionalSoxhletextraction.FoodChem.197, 1054–1063,PtB.
Herrera,M.C.,LuqueDeCastro,M.D.,2004.Ultrasound-assistedextractionforthe analysisofphenoliccompoundsinstrawberries.Anal.Bioanal.Chem.379, 1106–1112.
Higdon,J.V.,Frei,B.,2003.Teacatechinsandpolyphenols:healtheffects, metabolism,andantioxidantfunctions.Crit.Rev.FoodSci.Nutr.43,89–143. Hirasawa,M.,Takada,K.,2004.Multipleeffectsofgreenteacatechinonthe
antifungalactivityofantimycoticsagainstCandidaalbicans.J.Antimicrob. Chemother.53,225–229.
Ince,A.E.,S¸ahin,S.,S¸ümnü,S.G.,2013.Extractionofphenoliccompoundsfrom melissausingmicrowaveandultrasound.Turk.J.Agric.For.37,69–75. Jacotet-Navarro,M.,Rombaut,N.,Deslis,S.,Fabiano-Tixier,A.-S.,Pierre,F.-X.,Bily,
A.,Chemat,F.,2016.Towardsadrybio-refinerywithoutsolventsoradded waterusingmicrowavesandultrasoundfortotalvalorizationoffruitand vegetableby-products.GreenChem.18,3106–3115.
Kalil,S.,Maugeri,F.,2000.Responsesurfaceanalysisandsimulationasatoolfor bioprocessdesignandoptimization.ProcessBiochem.35,539–550. Kemmer,G.,Keller,S.,2010.Nonlinearleast-squaresdatafittinginExcel
spreadsheets.Nat.Protoc.5,267–281.
Li,Y.,Fabiano-Tixier,A.-S.,Vian,M.,Chem,F.,2013a.Solvent-freemicrowave extractionofbioactivecompoundsprovidesatoolforgreenanalytical chemistry.TrendsAnal.Chem.47,1–11.
Li,Y.,Fabiano-Tixier,A.S.,Tomao,V.,Cravotto,G.,Chemat,F.,2013b.Green ultrasound-assistedextractionofcarotenoidsbasedonthebio-refinery conceptusingsunfloweroilasanalternativesolvent.Ultrason.Sonochem.20, 12–18.
Lotito,S.B.,Fraga,C.G.,1998.(+)-Catechinpreventshumanplasmaoxidation.Free Radic.Biol.Med.24,435–441.
Meullemiestre,A.,Petitcolas,E.,Maache-Rezzoug,Z.,Chemat,F.,Rezzoug,S.A., 2016.Impactofultrasoundonsolid-liquidextractionofphenoliccompounds frommaritimepinesawdustwaste.Kinetics,optimizationandlargescale experiments.Ultrason.Sonochem.28,230–239.
Mieszczakowska-Fr ˛ac,M.,Dyki,B.,Konopacka,D.,2015.Effectsofultrasoundon polyphenolretentioninapplesaftertheapplicationofpredryingtreatmentsin liquidmedium.FoodBioprocessTechnol.9,1–10.
Murado,M.A.,Prieto,M.A.,2013.Dose-responseanalysisinthejointactionoftwo effectorsanewapproachtosimulation,identificationandmodellingofsome basicinteractions.PLoSOne8,e61391.
Nagao,T.,Meguro,S.,Hase,T.,Otsuka,K.,Komikado,M.,Tokimitsu,I.,Yamamoto, T.,Yamamoto,K.,2009.Acatechin-richbeverageimprovesobesityandblood glucosecontrolinpatientswithtype2diabetes.Obesity(SilverSpring)17, 310–317.
Pabuc¸cuo˘glu,A.,Kıvc¸ak,B.,Bas¸,M.,Mert,T.,2003.AntioxidantactivityofArbutus unedoleaves.Fitoterapia74,597–599.
Pallauf,K.,Rivas-Gonzalo,J.C.,delCastillo,M.D.,Cano,M.P.,dePascual-Teresa,S., 2008.Characterizationoftheantioxidantcompositionofstrawberrytree (ArbutusunedoL.)fruits.J.FoodCompos.Anal.21,273–281.
Palma,M.,Taylor,L.T.,1999.Extractionofpolyphenoliccompoundsfromgrape seedswithnearcriticalcarbondioxide.J.Chromatogr.A849,117–124. Pawlowska,A.M.,DeLeo,M.,Braca,A.,2006.PhenolicsofArbutusunedoL.
(Ericaceae)fruits:identificationofanthocyaninsandgallicacidderivatives.J. Agric.FoodChem.54,10234–10238.
Pi ˜neiro,Z.,Palma,M.,Barroso,C.G.,2004.Determinationofcatechinsbymeansof extractionwithpressurizedliquids.J.Chromatogr.A1026,19–23.
Pinela,J.,Prieto,M.A.,Barreiro,M.F.,Carvalho,A.M.,Oliveira,M.B.P.P.,Vázquez,J.A., Ferreira,I.C.F.R.,2016.Optimizationofmicrowave-assistedextractionof hydrophilicandlipophilicantioxidantsfromasurplustomatocropby responsesurfacemethodology.FoodBioprod.Process.98,283–298. Pingret,D.,Fabiano-Tixier,A.-S.,Bourvellec,C.,LeRenard,C.M.G.C.,Chemat,F.,
2012.Labandpilot-scaleultrasound-assistedwaterextractionofpolyphenols fromapplepomace.J.FoodEng.111,73–81.
Prikler,S.,2009.AdvancedExcelforscientificdataanalysis,2nded.ed.Robertde LevieEd.
Proestos,C.,Komaitis,M.,2008.Applicationofmicrowave-assistedextractionto thefastextractionofplantphenoliccompounds.LWT−FoodSci.Technol.41, 652–659.
Roselló-Soto,E.,Galanakis,C.M.,Brnˇci ´c,M.,Orlien,V.,Trujillo,F.J.,Mawson,R., Knoerzer,K.,Tiwari,B.K.,Barba,F.J.,2015.Cleanrecoveryofantioxidant compoundsfromplantfoods,by-productsandalgaeassistedbyultrasounds processing.Modelingapproachestooptimizeprocessingconditions.Trends FoodSci.Technol.42,134–149.
Ruiz-Rodríguez,B.-M.,Morales,P.,Fernández-Ruiz,V.,Sánchez-Mata,M.-C., Cámara,M.,Díez-Marqués,C.,Pardo-de-Santayana,M.,Molina,M.,Tardío,J., 2011.Valorizationofwildstrawberry-treefruits(ArbutusunedoL.)through nutritionalassessmentandnaturalproductiondata.FoodRes.Int.44, 1244–1253.
Vuong,Q.V.,Golding,J.B.,Nguyen,M.,Roach,P.D.,2010.Extractionandisolationof catechinsfromtea.J.Sep.Sci.33,3415–3428.
Vuong,Q.V.,Golding,J.B.,Stathopoulos,C.E.,Nguyen,M.H.,Roach,P.D.,2011. Optimizingconditionsfortheextractionofcatechinsfromgreenteausinghot water.J.Sep.Sci.34,3099–3106.
Wong,W.H.,Lee,W.X.,Ramanan,R.N.,Tee,L.H.,Kong,K.W.,Galanakis,C.M.,Sun,J., Prasad,K.N.,2015.Twolevelhalffactorialdesignfortheextractionof phenolics,flavonoidsandantioxidantsrecoveryfrompalmkernelby-product. Ind.CropsProd.63,238–248.
Zhang,Z.,Lv,G.,Pan,H.,Fan,L.,2012.Optimisationofthemicrowave-assisted extractionprocessforsixphenoliccompoundsinAgaricusblazeimurrill.Int.J. FoodSci.Technol.47,24–31.
Ziyyat,A.,Mekhfi,H.,Bnouham,M.,Tahri,A.,Legssyer,A.,Hoerter,J.,Fischmeister, R.,2002.Arbutusunedoinducesendothelium-dependentrelaxationofthe isolatedrataorta.Phyther.Res.16,572–575.