• Tidak ada hasil yang ditemukan

Characteristics of wood CO2 efflux in a Bornean tropical rainforest

N/A
N/A
Protected

Academic year: 2024

Membagikan "Characteristics of wood CO2 efflux in a Bornean tropical rainforest"

Copied!
10
0
0

Teks penuh

(1)

ContentslistsavailableatScienceDirect

Agricultural and Forest Meteorology

jou rn al h om ep a ge :w w w . e l s e v i e r . c o m / l o c a t e / a g r f o r m e t

Characteristics of wood CO 2 efflux in a Bornean tropical rainforest

Ayumi Katayama

a,∗

, Tomonori Kume

b

, Mizue Ohashi

c

, Kazuho Matsumoto

d

, Michiko Nakagawa

e

, Takami Saito

f

, Tomo’omi Kumagai

f

, Kyoichi Otsuki

a

aKasuyaResearchForest,KyushuUniversity,Sasaguri,Fukuoka,811-2415,Japan

bSchoolofForestryandResourceConservation,NationalTaiwanUniversity,Taipei,106-17,Taiwan

cSchoolofHumanScienceandEnvironment,UniversityofHyogo,Himeji,Hyogo,670-0092,Japan

dFacultyofAgriculture,UniversityoftheRyukyus,Okinawa,903-0213,Japan

eGraduateSchoolofBioagriculturalSciences,NagoyaUniversity,Chikusa-ku,Nagoya,464-8601,Japan

fInstituteforSpace-EarthEnvironmentalResearch,NagoyaUniversity,Chikusa-ku,Nagoya,464-8601,Japan

a r t i c l e i n f o

Articlehistory:

Received13September2015

Receivedinrevisedform16January2016 Accepted20January2016

Keywords:

Emergenttrees Ecosystemscale Inter-individualvariation Scaling

StemCO2efflux

a b s t r a c t

WoodCO2efflux(Rwood)isanimportantcarboncyclingprocessinforestecosystemsandislessunderstood inBorneantropicalrainforestscomparedwiththatinNeotropics,whichhavedifferingbioticandabiotic characteristics.Borneantropicalrainforestshaveparticularlyhigherabovegroundbiomassbecauseof highstanddensitiesoftallemergenttrees,andnoregularseasonaldryperiod.Thesepropertiesmay reflectspecificRwoodvariationandhigherannualRwoodattheecosystemscale(annual-Rwood)inBorneo.

Hence,inthisstudy,weinvestigatedfactorsthataffectinter-individualvariationinchamber-basedRwood

onstemsurfacesatbreastheight(chamber-Rwood)andestimatedRwoodattheecosystemscale(ecosystem- Rwood)inaBorneantropicalrainforest.Subsequently,weexaminedtemporalvariationinecosystem-Rwood

withenvironmentalfactors,estimatedannual-Rwood,andthenevaluatedtheeffectsoflargetreeson ecosystem-Rwood.

Stemgrowthrateswerethemostsignificantpredictorofinter-individualvariationinchamber-Rwood

duringallfivemeasurementperiodsbetween2012and2014.Accordingly,stemgrowthratesallowed accurateestimatesofecosystem-Rwood,althoughchamber-Rwoodvariedwithdiameteratbreastheight (DBH)andtreespeciesinsomemeasurementcampaigns.Ecosystem-Rwooddecreasedwithsoilmoisture.

Consideringtheinter-individualandtemporalvariation,annual-Rwoodwasestimatedat7.06±2.09MgC ha−1year−1.ThisvaluewascomparabletothosedeterminedinNeotropicalforests,eventhoughabove- groundbiomassinthepresentstudysitewasapproximatelytwiceofthoseinNeotropicalsites.Large treeswithDBH>70cmcomprised38%ofabovegroundbiomassbutaccountedforonly23%ofecosystem- Rwoodbecauseofthesmallportionofwoodytissuesurfaceareaofthelargetrees.Thesedataindicatethat thestanddensityoflargetreescanconsiderablyaffectabovegroundbiomassbutexertlessinfluenceon variationinecosystem-Rwoodamongvariousforests.

©2016ElsevierB.V.Allrightsreserved.

1. Introduction

Tropicalforestsplayimportantrolesinglobalcarbon cycling becauseofthehighestphotosynthesisandrespirationratesoffor- estecosystems(Beeretal.,2010;Malhi,2012).WoodCO2efflux (Rwood)isameasureofcarbonreleasefromstemsandbranchesand isanimportantindicatorofforestecosystemproductivity,because itaccountsforabout20–25%ofplantrespiration(Chambersetal., 2004;Malhietal.,2009)and13–27%ofcarbonassimilationfrom photosynthesis(Malhietal.,2009;daCostaetal.,2013;Doughty

Correspondingauthor.Tel.:+81929483101;fax:+81929483127.

E-mailaddress:[email protected](A.Katayama).

etal.,2013).RecentstudieshaveestimatedRwoodattheecosystem- scale(ecosystem-Rwood,␮molm2s1)inNeotropicalrainforests (Chambersetal.,2004;Cavalerietal.,2006;daCostaetal.,2013;

Rowlandetal.,2014),althoughstudiesonecosystem-Rwoodoutside Neotropicalareasremainlimited(MeirandGrace,2002).Measure- mentsofecosystem-Rwoodinvariouskindsoftropicalrainforests areimportantforinvestigationsofglobalcarboncycling,because variationofecosystem-Rwoodwithbioticandabioticfactorsremains poorlyunderstood.

Rwoodonwoodysurfaceareas(chamber-Rwood,␮molm2s1) isgenerallyestimatedusingthechambermethod,andecosystem- Rwoodisusuallyestimatedbymultiplyingchamber-Rwoodatbreast height by woody tissue surface areas or volumes (Meir and Grace,2002;Chambersetal.,2004).Estimatesofecosystem-Rwood http://dx.doi.org/10.1016/j.agrformet.2016.01.140

0168-1923/©2016ElsevierB.V.Allrightsreserved.

(2)

still remainchallengedby intra-andinter-individualvariations in chamber-Rwood that can lead to errors in scaled estimates (Chambers et al.,2004; Malhi etal., 2009).Recently,Katayama etal.(2014)showedlimitedverticalvariationinchamber-Rwood basedonmeasurementsofupto40-mhightreesin aBornean tropicalrainforest,suggestingthatintra-individualvariation has limitedeffectsonestimatesofwholetreestemCO2efflux.However, theeffectsofinter-individualvariationremainunclear,warranting carefulexaminationespeciallyinold-growthtropicalrainforests wheretreespeciesandsizesvary.

Multiple factors have been shown to affect inter-individual variation in chamber-Rwood, and various scaling methods have beensuggested.Inpreviousstudiesoftropicalrainforests,inter- individualvariation in chamber-Rwoodwas related todiameters atbreast height(DBH;Cavaleri etal.,2006), stemgrowthrates (Ryan et al.,1994),and both (Chambers etal.,2004; Robertson etal.,2010).Subsequently,thesefactorswereusedtoextrapolate chamber-Rwoodvaluesinindividualswithoutdirectmeasurements andthentoestimateecosystem-Rwood.InrecentstudyofNeotrop- icalforests,ecosystem-Rwoodvalueswereestimatedbymeansof measuredchamber-Rwoodvalues(Rowlandetal.,2014).Previous studiesalsoreportedthatecosystem-Rwoodvarytemporallywith soilmoisture,reflectingchangesinstemgrowthinresponsetosea- sonaldryperiodsinNeotropicalrainforests(Chambersetal.,2004;

Malhietal.,2014).Thesestudieswarrantinvestigationoffactors that affecttemporal variationin ecosystem-Rwood.Furthermore, thefactorsaffectinginter-individualvariation inchamber-Rwood mayvarytemporally.Thus,examinationsofinter-individualvari- ationinchamber-Rwoodacrossmeasurementperiodsarerequired toinvestigatetemporalvariationinecosystem-Rwoodandtoesti- mate annual Rwood at the ecosystem scale (annual-Rwood, MgC ha1year1).

It iswellknownthat Borneantropicalrainforests have high aboveground biomass. Accordingly, two-fold higher stand den- sities of large trees (DBH>70cm) resulted in 60% increases in abovegroundbiomasscomparedwiththoseinNeotropicalforests (Paoli etal., 2008;Sliket al.,2010).Moreover,largetrees con- tributedconsiderablytoabovegroundbiomassandaboveground biomassincrementsattheecosystemscaleinBorneo(Paoliand Curran,2007).Inarecentstudy,higherstemgrowthrateswere shown in larger trees than in small trees according to global analyses(Stephensonetal.,2014).Thesedataindicatethatlarge trees can have higher individual Rwood because of both effects of highchamber- Rwood caused by highstem growthrates and DBH, and large surface area or volume. Thus, large trees may alsocontributeconsiderablytoecosystem-Rwood,potentiallylead- ingtogreaterecosystem-Rwoodvalues inBorneanthanthose in Neotropicalforests.Higherannual-RwoodinBorneoisalsoexpected withconsiderationoftemporalvariation.Thereisnoregularsea- sonaldryperiodbutirregularshort-termdroughtperiodoccurs inBorneo,whilethereareseveralseasonaldrymonthsinmost Neotropicalarea(Malhietal.,2006)wherelowerchamber-Rwood valueswereobservedduringdrymonths(Chambersetal.,2004;

Malhietal.,2014).Althoughhigherannual-Rwoodisexpectedin Borneo,thestudyonannual-Rwoodhasnotbeenconductedinsuch forests.

Inthisstudy,wemeasuredchamber-Rwoodatbreastheightsof morethan50individualtreesof36speciesinaBorneantropical rainforestwithhighabovegroundbiomassandnoregularseasonal dryperiod.Measurementswereperformedfivetimesundervari- oussoilwaterconditions.Insubsequentanalyses,factorsaffecting inter-individualvariationinchamber-Rwoodforeachmeasurement periodwereidentified,andecosystem-Rwoodwassubsequentlycal- culated.Wealsodeterminedtheeffectsofenvironmentalfactors, suchassoilmoistureandairtemperature,ontemporalvariationin ecosystem-Rwood,andannual-Rwood,anditsratiotogrossprimary

production(GPP)wereestimatedaccordingly.Finally,weexam- inedthecontributionoflargetreestoecosystem-Rwood.

2. Materialsandmethods 2.1. Studysite

This study was conducted in Lambir Hills National Park, Sarawak,Malaysia(412N,11402E)atabout20kmsouthwestof Miri,SarawakontheislandofBorneo.Lowlandmixed-dipterocarp forest,whichisatypicalmaturelowlandtropicalrainforesttypein SoutheastAsia,covers85%ofthis6949hapark.Thecontinuoustree layerisapproximately40mabovetheground,buttheheightsof treeswithDBH>100cmoftenexceed50m(Katayamaetal.,2014).

Theleafareaindexrangesspatiallybetween4.8and6.8m2m2, withameanof6.2m2m−2(Kumagaietal.,2004).Between2000 and2009,themeanannualtemperatureatthis sitewas25.8C and rainfall wasdistributed throughout the year with a mean annualrainfallof2600mm(Kumeetal.,2011).Themonthlyrain- fallbetween1958and2001averagedmorethan100mm(Kumagai etal.,2005).Althoughthereisnoclearseasonalityintempera- tureandprecipitation,unpredictabledryperiodssometimesoccur throughouttheyear.

A 4ha study plot was established surrounding a crane for ecologicalobservation.Previously,weperformedcarboncycling measurements,includingeddycovariancemeasurements,carbon allocation(Katayamaetal.,2013),andsoilrespiration(Katayama etal.,2009)atthissite.

2.2. Rwoodmeasurements

Chamber-Rwoodwasmeasuredfivetimesatbreastheightduring 9–28January2012,5–8May2013,13–14July2013,24December 2013–2January2014,and29June–7July2014,mainlyinthe4- ha plot.Measuredtrees were selectedrandomly from52 trees duringeachmeasurementperiod,andthenumbersofmeasured treeswere48,48,47,50,and38inJanuary2012,May2013,July 2013,December2013,andJune2014,respectively(TableA1).DBH rangedfrom10 to158cm.The52 samplesincluded36species, and24oftheseweretheDipterocarpaceaefamily.Tominimizethe effectsofdiurnalvariationinRwood,measurementswereconducted between8:30and15:00accordingtoapreviousstudy(Katayama etal.,2014).Diurnalmeasurementschamber-Rwoodoffivetrees wereperformedevery2hbetween7:30and17:30onJanuary1, 2014,toexaminetheeffectofdiurnalvariationonthedailyRwood estimatebasedonone-timemeasurementsduringdaytime.

Chamber-Rwood was measured using a closed-static system withaninfraredgasanalyzer(GMP343,Vaisala,Helsinki,Finland) attachedinsideachamber.Thechamberandcollarsweremadeof polyvinylchloride.Asmall8-Vbattery-operatedfanwasinstalled tocirculateairinsidethechamber,collarswereattachedtostems usingglueandsiliconsealant,andgapsbetweenstemsandcollars werefilledwithnondryingclay.Leakswerecheckedaftermeasure- mentsbybreathingaroundthechamber.For allmeasurements, incubationperiodsrangedfrom60to180s,andCO2 concentra- tionswithinthechamberweremeasuredevery5sandrecorded usingamonitor(M170,Vaisala).Chamber-Rwood(␮molm−2s−1) wascalculatedusingEq.(1)asfollows:

Rwood=

273

.2 273.2+Ta

× V 22.4×dc

dt ×106×1

A, (1)

where V is the chamber volume (0.389l), Ta is air tempera- ture inside the chamber (C), A is the projected area of the chamber(3.849×10−3m2),anddCistheincrementofCO2 con- centration (ppm) during the incubation time (s, dt). Eq. (1) assumesthatrespirationismeasuredunderstandardbarometric

(3)

pressure. Allmeasurements at each point were replicated two orthreetimesandmeanswerecalculated.Systemaccuracywas determined by measuring Rwood of some trees using a com- mercial respirometer(LI-6400, Li-CorInc., Lincoln, NE, USA) at thesame timepoints.TherelationshipbetweenRwood observed usingthesystem(Rwood-system)andLI-6400(Rwood-LI6400)waslinear (Rwood-LI6400=1.1394×Rwood-system,R2=0.97);thus,chamber-Rwood wasestimatedbycorrectingRwood-systemusingthisequation.

Stemsurfaceand airtemperaturesweremeasuredconcomi- tantlyusingathermometer(Ondotori;T&DCo.Ltd.,Nagano,Japan).

Toeliminatetheeffectsoftemperature,chamber-Rwoodvalueswere correctedtoareferencetemperatureof25CusingEq.(2)asfol- lows:

R25=R(st)Q10(25−st)/10, (2)

whereR25ischamber-Rwoodat25C,stisthestemsurfacetemper- ature(C)whereRwood wasmeasured,andQ10isatemperature coefficient of Rwood. Because Q10 was not calculated from the presentdata,itwasassumedthatQ10wasequalto2.0,asshown previouslybyCavalerietal.(2006)intropicalrainforestsinLaSelva, CostaRica.Inthepresentstudy,chamber-RwoodwasusedasR25 basedonsurfacearea,whichwascorrectedat25C.

Environmentalobservations wereincorporatedintoanalyses of temporal variation in ecosystem-Rwood. Air temperaturewas monitoredonthecraneat75mabovethegroundusingather- mohygrograph(HMP35A,Visala)andthedatawasloggedusinga programmabledatalogger(CR10X,CampbellScientific,Utah,USA).

Rainfallwasmonitoredusingatippingbucketraingauge(RS102, OgasawaraKeiki,Tokyo,Japan)withadatalogger(HOBOevent, OnsetComputer,USA)atthetopofthecrane(85.8mabovethe ground).Soilwatercontentwasdeterminedatdepthsof10,30, 60,and100cmusingtimedomainrefrectometrysensors(CS616, Champbell) and a data logger nearthe crane, and mean daily soilwater content(SWC, m3m3)wascalculated ata depthof 0–120cm.Measurementswereinitiatedon3rdSeptember2012, andsoilmoisturecontentswereinterpolatedbetween1stJanuary 2011and2ndSeptember2012(SWCestimate)usingtherelationship betweencumulative30dayrainfall(Rcum)andmeasuredsoilwater contents(SWCestimate=0.1631×Rcum0.107,R2=0.67).

2.3. Treecensus

Atreecensuswasconductedinthe4-haplotin2012and2013.

DBHvaluesfortreeswithDBH>10cmweremeasuredwithareso- lutionof1mm,anddifferencesindiameterbetween2012and2013 werecalculatedasstemcircumferencegrowthrates(sg,cmy1)for alltrees.AbovegroundbiomassinBorneowasestimatedusingDBH andapreviouslydescribedallometricequation(Yamakuraetal., 1986).

2.4. Scalingmethodandstatisticalanalysis

Toestimateannual-Rwood,ecosystem-Rwoodforeachmeasure- mentserieswasestimatedfrommeasuredchamber-Rwood(spatial scaling),and dailyecosystem-Rwoodin 2013wasalsoestimated from the relationship between the estimated ecosystem-Rwood andenvironmentalfactors(temporalscaling).Forspatialscaling, ecosystem-Rwood in the 4-haplot wasestimated by two meth- ods.Inone,ecosystem-Rwoodwasestimatedbymultiplyingmean measuredchamber-Rwoodvaluesbywoodytissuesurfaceareasin the4-haplotandthendividingbythegroundarea,asdefinedby Model0.Intheother,ecosystem-Rwoodwasestimatedasthesum ofindividualRwood(individual-Rwood,␮moltree−1s−1)foralltrees inthe4-haplotanddividedbythegroundarea.Individual-Rwood wasestimatedbymultiplyingindividualchamber-Rwoodvaluesby woody tissue surface areas. Chamber-Rwood values for all trees

were estimated by regression models. Predictors of the model includedsg(continuousvariable),DBH(continuousvariable),and treespecies(categoricalvariable;DipterocarpaceaeDryobalanops aromatica,DipterocarpaceaeShoreabeccariana,MoraceaeArtocar- pus anisophyllus, and others) (Table A1). The three species for whichmorethanthreeindividualsweremeasuredwereselectedas variables.Toexaminetheeffectofpredictorsonecosystem-Rwood estimates,weconstructedthreemodelsforchamber-Rwood esti- mate,thebest-fitmodel,Model1,andModel2.Thebest-fitmodel wasdefinedasthemodelforwhichpredictorsweredetermined accordingtothelowestAICvaluefromthefullmodel.Predictor variablesinModel1weresgandDBHandthepredictorinModel2 wassg. Thesame coefficients of sgand DBH wereused across species.Woodytissuesurfaceareaswereestimatedusingtherela- tionshipbetweenDBHandsurfacearea,aspreviouslydescribed (Chambersetal.,2004)forNeotropicalforests,andsurfaceareaof woodytissueperunitgroundarea(stemareaindex,SAI,m2m2) wascalculated for comparison withprevious studies.Errors in ecosystem-RwoodforModel0werestandarddeviationsofmeasured chamber-Rwood,anderrorsforothermodelswerederivedfrom95%

confidencelimitsforeachmodel.

Temporalscalingwasperformedusingaregressionmodelwith environmentalfactorssuchasSWCandairtemperature.Onlypre- dictivefactorswithsignificantcorrelationwereusedforthemodel.

Annualecosystem-Rwoodwereestimatedusingthebest-fitmodel, Model1,Model2, andModel0. Errorsinannualecosystem-Rwood werederivedfrom95%confidencelimitsforeachmodelfortempo- ralscaling.AllstatisticalanalyseswereconductedusingRversion 3.1.3.

Althoughsomepreviousstudiesdemonstrateverticalvariation inRwoodforscaling(Yodaetal.,1965;Damesinetal.,2002),we assumedconstantRwoodwithheightbecausepreviousstudiesshow limitedverticalvariationinRwoodbasedonsurfaceareainthisstudy site(Katayamaetal.,2014).Inthepresentstudy,theuseofsurface areaforscalingindividual-Rwoodfromchamber-Rwoodwasdeter- minedbasedontheresultofagraphicaltechnique(LevyandJarvis, 1998;Cavalerietal.,2006)usingverticalRwooddatafromaprevious study(Katayamaetal.,2014;Fig.A1).

2.5. GPPestimates

GPPwasdefinedasthesumofabovegroundplantrespiration (Rabove), abovegroundnet primaryproduction (ANPP), and total belowground carbon flux (TBCF) according to a previousstudy (Littonetal.,2007).Raboveisthesumofecosystem-Rwoodandfoliage CO2efflux(Rfoliage).Ecosystem-Rwoodwasestimatedinthepresent study(seeSection2.4).Ecosystem-Rfoliagewasestimatedasthesum ofleafareadensitiesmultipliedbyRfoliageatevery5m.Leafarea densitieswereestimatedusingtheplantareaindex(PAI)profile andtotalleafareas.ThePAIprofilewasreportedbyKumagaietal.

(2006),whomeasuredverticalvariationintheplantareaindexat fourpointsusingapairofcanopyanalyzers(LAI-2000,Li-Cor,Lin- coln,Nebraska)inthisstudysite.Subsequently,totalleafareasat theecosystem-scalewereestimatedusingtheequationdescribed byYamakura etal. (1986). Rfoliage wasreported byKenzo etal.

(2015),whomeasuredfoliage darkrespirationratesin24 fam- ilies,62 species,and 123individualtreesfromtheground toa heightof52.5minthisstudysiteusingacommercialrespirometer (LI-6400,LI-CORInc.,Lincoln,NE).Rfoliagewascorrectedtotheres- pirationrateat25CusingQ10.Q10wasnotmeasuredeitherinthe presentstudyorinpreviousstudiesinAsiantropicalrainforests.

AlthoughtherewashighvariationinQ10amongtreespeciesinthe Neotropics(Asaoetal.,2015),Q10wasnotdifferentamongcanopy positionsandthevalueofQ10averagedamongvariousspecieswas 2.0inAustralia(Weerasingheetal.,2014)and2.3inCostaRica (Cavalerietal.,2008).Inthepresentstudy,Q10wasassumedto

(4)

be2.3,avaluereportedintropicalrainforestsinLaSelva(Cavaleri etal.,2008)thatwasobtainedbecauseofahighernumberofmea- surements.SeveralstudiesshowedlowerRfoliageinthelightthan inthedark(Atkinetal.,2000).Thus,toaccommodatelightinhi- bitionof Rfoliage, daytimeRfoliage wascalculated as 67%of night timeRfoliageaccordingtopreviousstudies(Malhietal.,2009,2014;

daCostaetal.,2013;Doughtyetal.,2013).AnnualRfoliage atthe ecosystemscale(annual-Rfoliage,MgCha−1year−1)wasestimated assuming12hofdaytimeandnighttimeandaconstanttempera- tureof25Ceachdaythroughouttheyear,andtheerrorinannual Rfoliage representsspatialvariationaccordingtoapreviousstudy (Metcalfeet al.,2010).ANPP(6.76±0.57 MgCha−1year−1)and TBCF(19.63±6.37MgCha1year1)werepreviouslyestimatedin thisstudysite(Katayamaetal.,2013).ANPPwasestimatedbasedon treecensusandlitterfallmeasurementsbetween2001and2006in the4-haplot,andtheerrorwasthestandarddeviationforthefive years.TBCFwasestimatedasthedifferencebetweenannualsoil respirationandannuallitterfallCusingasteady-stateassumption forsoilC(RaichandNadelhoffer,1989).Soilrespirationmeasure- mentswereperformedbetween2002and2006,andtheerrorsfor soilrespirationandlitterfall representedspatialvariation. Thus, eacherrorofannualfluxforGPPestimatewasestimatedbasedon measurementsormodels,resultingindifferentrepresentationsof variation.Errorpropagationwasappliedforallcombinedquanti- tiesusingstandardrulesofquadratureaccordingtoMalhietal.

(2014),assumingthat uncertaintieswereindependentandnor- mallydistributed.Uncertaintiesintheestimatesaredescribedin Section4.

2.6. Effectsoflargetreesonecosystem-Rwoodandaboveground biomass

Ecosystem-Rwoodandabovegroundbiomassinresponsetostand densitiesoflargetreeswereinvestigatedtoassessthepotential effectsoflargetreesonecosystem-Rwoodandabovegroundbiomass.

Large trees were defined as those with DBH>70cm according toSliketal.(2013).Ecosystem-Rwood forvarious fictionalstand densitiesoflargetreeswascalculatedasthesumofecosystem- Rwood for large trees (ecosystem-Rwoodlarge, ␮molm−2s−1) and ecosystem-RwoodfortreeswithDBH<70cm(ecosystem-Rwoodsmall,

␮molm2s1).Ecosystem-Rwoodlarge wasestimatedbymultiply- ing mean individual-Rwood values for large trees obtained by the best-fit model in January/2012with standdensities. Stand density of large trees differed between zero and actual den- sities (16.5 trees ha−1). Ecosystem-Rwoodsmall was the sum of individual-RwoodvaluesfortreeswithDBH<70cm.Aboveground biomasswassimilarlycalculatedforvariousstanddensitiesoflarge trees.

3. Results

3.1. Inter-individualvariationinRwood

Average±standard deviation (S.D.) chamber-Rwood values at breastheightwere1.39±0.74,0.99±0.58,0.81±0.52,1.23±0.87, and0.91±0.65␮molm−2s−1inJanuary2012,May2013,July2013, December2013,andJuly2014,respectively.Multipleregression analysesofeachmeasurementperiodshowedthatsgandDBHwere positivelycorrelatedwithchamber-Rwood(Fig.1),althoughselected variablesdifferedforthebest-fitmodelofchamber-Rwood.Among variables,sgwasthemostpredictiveofchamber-Rwood(Table1), althoughDBHandspeciessignificantlyaffectedchamber-Rwoodin somecases.Specifically,chamber-RwoodwashigherforD.aromat- icaandlowerforS.beccarianathantheotherspecieswhenselected asthebest-fitmodel.Diurnalvariationpatternsdifferedbetween

0 1 2 3 4 5

0 1 2 3 4 5

0 1 2 3 4 5

0 1 2 3 4 5

0 50 100 150 0

1 2 3 4 5

-1 0 1 2 3 4 5 Chamber_Rwood(μmolm-2s-1)

sg (cm yr-1) DBH (cm) Jan-12

May-13

Jul-13

Dec-13

Jul-14

Fig.1.Relationshipsofchamber-Rwood duringeachmeasurementperiod,with diametersatbreastheight(DBH;right)andstemgrowthrates(sg;left).Sym- bolsindicatetreespecies:circles,DIPTEROCARPACEAEShoreabeccariana;squares, DIPTEROCARPACEAEDryobalanopsaromatic;cross,MORACEAEArtocarpusaniso- phyllus;closeddiamond,otherspecies.

individuals(Fig.A2).Chamber-Rwoodpeakedintheearlyafternoon forsometreesbutinthelateafternoonforothers.ThemeanCVof diurnalvariationwas13±7%,anddailychamber-Rwoodwithcon- siderationofdiurnalvariationwas1.09±0.17higherthanthose basedononlyone-timemeasurementsinthedaytime,suggest- inganapproximately10%underestimateofchamberRwoodinthis study.

(5)

Table1

Statisticalresultofmultipleregressionanalysisforeachmeasurementsperiod.Theobjectivevariableischamber-Rwood,predictorvariablesforthebest-fitmodelwere determinedaccordingtothelowestAICvaluefromallpredictorsincludingstemgrowthrate(sg),dbhandtreespecies.PredictorvariablesofModel1andModel2aresgand DBH,andsg,respectively.

Jan-12 May-13 Jul-13 Dec-13 Jul-14

Best-fit model

Model1 Mode2 Best-fit model

Model1 Mode2 Best-fit model

Model1 Mode2 Best-fit model

Model1 Mode2 Best-fit model

Model1 Mode2

Coefficientsofindependentvariables

sg 0.28* 0.28* 0.47*** 0.36*** 0.36*** 0.49*** 0.32*** 0.32*** 0.41*** 0.26* 0.30* 0.55*** 0.37*** 0.33** 0.37***

DBH 0.007* 0.007* 0.005* 0.005* n.s. 0.003 0.007* 0.008* n.s. 0.002

D.beccariana n.s. −0.28 n.s. n.s. n.s.

D.aromatica n.s. 0.44 0.63** 0.58

A.anisophyllus n.s. n.s. n.s. n.s. n.s.

Intercept 0.84*** 0.84*** 1.03*** 0.53*** 0.51*** 0.64*** 0.53*** 0.44*** 0.52*** 0.61*** 0.58*** 0.79*** 0.61*** 0.56*** 0.61***

AIC 84.34 84.34 87.67 34.19 45.31 47.75 36.23 43.52 43.58 106.37 107.33 110.00 63.63 65.42 63.63

R2 0.47 0.47 0.40 0.72 0.62 0.58 0.60 0.53 0.51 0.46 0.43 0.37 0.36 0.36 0.36

N 48 48 47 50 38

***P<0.001.

** P<0.01.

* P<0.1.

“n.s.”meansthatthefactorswerenotselectedbythebest-fitmodel.“–”meansthatthefactorwerenotusedforthemodel.

Fig.2.Temporalvariationsin(a)airtemperature,(b)soilmoisture,and(c)Rwood betweenJanuary2012andJuly2014.Linesin(c)showecosystem-Rwoodvaluesesti- matedusingalinearregressionmodelwithsoilmoisture.Circlesin(c)indicate ecosystem-Rwoodvaluesthatwereestimatedusingthebest-fitmodelanderrorbars indicatethe95%confidencelineofthemodel.

3.2. Temporalvariationsinecosystem-Rwood,Rfoliage,andGPP Seasonalvariationsinairtemperatureandsoilmoisturewere notclear,althoughairtemperaturedecreasedinJanuary2014,and soilmoisture decreased in June 2013and March 2014(Fig.2).

Onthe otherhand,hightemporal variationin ecosystem-Rwood wasobserved(Fig.2),withbest-fitmodelestimatesof2.48±0.34, 1.68±0.20,1.32±0.21,2.08±0.40,and1.44±0.43␮molm−2s−1 duringJanuary2012,May2013,July2013,December2013,andJuly 2014,respectively (Table2).Estimatesofecosystem-Rwoodfrom eachmodelweresignificantlyrelatedtoSWCexceptforModel2 (Table3)butwerenot relatedtoair temperature.Annual-Rwood between2012and2013estimatedfromthelinearregressionmodel withthepredictorofSWCforBest-fitmodel,Model1,Model2,and Model0were7.06±2.09,7.07±2.18,6.68±2.26,and8.56±2.20 MgCha1year1,respectively.Ecosystem-Rwoodandannual-Rwood werecomparablebetweenthebest-fitmodel,Model1,andModel2,

butModel0gavehighervaluesthantheothermodels(Table2), suggesting that sg is necessary for scaling, and that DBH and treespeciesarenot.Annual-RfoliageandGPPwere9.54±0.46and 42.99±6.74MgCha−1year−1,respectively,andratiosofRwoodand Rfoliage toGPPwere0.17±0.05 and0.23±0.01MgCha1year1 (MgCha1year1)1,respectively.

3.3. EffectoflargetreesonRwoodandabovegroundbiomass The number of large trees strongly influenced aboveground biomass,butlessstronglyinfluencedecosystem-Rwood.Largetrees werepresentinthestudysiteat16.5treesha1,accountingfor only3%ofalltrees(567treesha1),theycomprised21%oftotal ecosystem-Rwood,38%oftotalabovegroundbiomass,and16%of totalwoodysurfacearea(Fig.3).Thehighercontributionoflarge trees(38%)toabovegroundbiomassthantosurfacearea(16%)was causedbythedifferencesinallometricequations.Theallometric equationfor aboveground biomasshasan exponentialfunction (Yamakuraetal.,1986),whereastheallometricequationforsur- faceareahasalogarithmicfunction(Chambersetal.,2004).Thus, abovegroundbiomassexponentiallyincreasedwithDBH,whereas theincrease ratio ofa woody surface areawithDBHgradually decreasedwithDBHandalmostsaturatedathighDBHontheindi- vidualscale.Aschamber-RwoodincreasedwithDBH,theratioof largetreestoecosystem-Rwood(21%)wasslightlyhigherthanthe ratiotowoodytissuesurfacearea(16%).Accordingly,decreasein standdensityoflargetreesmorestronglyaffectedaboveground biomassthanecosystem-Rwood(Fig.4).Whenstanddensitiesof largetreesweredecreasedby55%tolevelsthatarecomparable tothose inNeotropicalareas(7.5treesha−1;Paolietal.,2008), abovegroundbiomassdecreasedto79%,andecosystem-Rwoodwas decreasedto87%.Accordingly,AGBwasdecreasedto62%inthe absence of large trees in this forest and ecosystem-Rwood was decreasedto79%,suggestingthatwhereasstanddensitiesoflarge trees affectaboveground biomass considerably,theireffects on ecosystem-Rwoodaremoderate.

4. Discussion

4.1. Inter-individualandtemporalvariationinRwood

Sgwasthemostimportantfactoraffectinginter-individualvari- ationinchamber-Rwoodforallmeasurementperiods,ratherthan DBHortreespecies(Table1).DBHand/orsghavebeenusedfor scalinginpreviousstudiesintropicalareas(MeirandGrace,2002;

(6)

Table2

Ecosystem-Rwoodandannualecosystem-Rwoodwithsoilmoistureestimatedbydifferentmodels.Multi-regressionanalyseswereconductedforecosystem-Rwoodestimates.

Predictorvariablesforthebest-fitmodelweredeterminedaccordingtothelowestAICvaluefromallpredictorsincludingstemgrowthrate(sg),dbhandtreespecies.

PredictorvariablesofModel1andModel2aresgandDBH,andsg,respectively.Annualecosystem-Rwoodwasestimatedbylinearregressionmodelwhichobjectivevariables wereecosystem-RwoodforeachmodelandpredictorvariableisSWCforeachmeasurementsperiod.

Measuredvalue Ecosystem-Rwood(␮molm−2s−1) Annualecosystem-Rwood(MgCha−1year−1)

SWC(m3m−3) Bestmodel Model1 Model2 Model0 BestModel Model1 Model2 Model0

Jan-12 0.303±0.004 2.48±0.34 2.48±0.34 2.37±0.41 2.85±1.5 7.06±2.09 7.07±2.18 6.68±2.26 8.56±2.20 May-13 0.287±0.004 1.68±0.20 1.65±0.22 1.56±0.27 2.04±1.20

Jul-13 0.239±0.001 1.32±0.21 1.34±0.22 1.27±0.27 1.67±1.10 Dec-13 0.309±0.005 2.08±0.40 2.08±0.40 1.91±0.49 2.53±1.80 Jun-14 0.253±0.003 1.44±0.43 1.47±0.37 1.44±0.43 1.88±1.34

Table3

Statisticalresultoflinearregressionanalysisfortemporalvariationinecosystem- Rwood.

Bestmodel Model1 Model2 Model0 Coefficientsofindependentvariables

SWC 18.17* 17.46* 15.43 18.16*

Intercept −3.36 −3.16 −2.68 −2.97

R2 0.75 0.71 0.65 0.72

*P<0.1.

Fig.3. RatioofeachDBHclassestoecosystem-Rwood,abovegroundbiomass(AGB), surfaceareasofwoodytissue(SA),andnumbersoftrees(number).

Chambersetal.,2004;Robertsonetal.,2010).DBHisanindica- torofsapwoodvolume,andchamber-RwoodincreaseswithDBH becausesapwoodvolumedirectlyaffectsmaintenancerespiration, reflecting increases in livecell numberswith sapwoodvolume (Ryanetal.,1994).Thus,strongrelationshipsbetweenchamber- RwoodandDBHhavebeenshownpreviously.However,thepresent studysuggestedthatDBHwasnotparticularlyimportantforscal- inginthisstudy,potentiallyreflectingdifferencesinmeasuredtree sizes.WemeasuredmultipletreeswithDBHof>100cm,whereas chamber-RwoodwasmeasuredfortreeswithDBH<100cminpre- viousstudies.For suchlargetrees,sapwoodvolumesunderthe collarmaynotincreasewithDBH,probablyowingtolimitedsap- wooddepthof<10cm(Ichie,unpublisheddata).Atthestudysite, sapwooddepths,whichweremeasuredbydyeexperiments(Ichie etal.,unpublisheddata),didnotcorrelatewithDBH(n=34,P=0.66, 32<DBH<106cm),suggestingthatsgisthemostsignificantpre- dictorofinter-individualvariationinforestswithverylargetrees.

Inthepresentbest-fitmodel,chamber-RwoodforD.aromatica washigherthanthatforS.beccariana,especiallyinJuly2013(Fig.

A3).However,DBHandsgofmeasuredtreesdidnotdifferbetween the two species (P>0.1, respectively). Differences in chamber- Rwoodmaybereflectedbythedifferenceinsapwooddepthsand wooddensitybetweenthetwospecies.SapwooddepthsforD.aro- matica(5.4±1.4cmfortreeswithDBHof59–96cm,n=3)were

50%

60%

70%

80%

90%

100%

0% 50% 100%

Decreased ratio in ecosystem-Rwood and AGB

Decreased ratio in stand denstiy Rwood AGB

Fig.4.Decreasedratiosofecosystem-Rwoodandabovegroundbiomass(AGB)with differentstanddensityoflargetrees;100%,actualstanddensityoflargetrees(16.5 treeha−1).

significantlyhigherthanthose forS.beccariana(3.2±0.4cmfor treeswithDBHof93–102cm,n=3,P<0.05;Ichieetal.,unpublished data,Fig.A4).Wood densityforD.aromatica(0.796gcm−3)was higherthanthatforS.beccariana(0.531gcm3,Inoueunpublished data).Thedifferencemayresultinhigherconstructionrespiration forD.aromaticathanforS.beccarianawithcomparablesgbecause constructionrespirationcanbeestimatedbysgandwoodydensity (e.g.,Ryan,1991;Ryanetal.,1994).Hence,highersapwoodvolumes and wood density mayhave affectedspecies-specific chamber- Rwoodinthisstudy,eventhoughthesespeciesbelongtothesame dipterocarpfamily.

Thetemporalvariationinecosystem-Rwoodinthepresentstudy wasassociatedwithsoilmoisturedespitetheabsenceofaregu- larseasonaldryperiod(Table3,Fig.2).Thisfindingmaybedueto growthphenology,giventhatRwoodseasonallyincreasedingrow- ingstems(e.g.,Ryan and Yoder,1997)and stemthegrowthof severaltreespeciesofDepterocarpaceawasobservedafterdrought (Ichieetal.,2004).Rwoodalsorespondedtosoilmoisturecondi- tionsandlikelyreflectedstemgrowthphenologyinNeotropical rainforests,wherestemgrowthishighestatthestartofthewet seasonandlowestatthestartofthedryseason(Chambersetal., 2004;Malhietal.,2014).Incontrast,Cavalerietal.(2006)reported noseasonalityinRwood inCostaRica,where therewerenodis- tinctdryseasons,asisthecaseinBorneanforests.Cavalerietal.

(2006)attributedtheabsenceofseasonalvariationinRwoodtothe asynchronyofthephenologiesofallspecies.Thus,stemgrowth phenologyinresponsetosoilwateravailabilityislikelylinkedwith temporal variationin Rwood intropical rainforests.Thereis also anotherpossibility:moreCO2respiredforrootsmightbetrans- ported bythetranspirationstreamand releasedfromthestem becauseoflowdiffusioninwetsoil(Teskeyetal.,2008).Continuous

(7)

measurementsofchamber-Rwoodandstemgrowthratesinaddition tosapfluxmeasurementsarenecessaryforaninvestigationofthe causeofincreaseinecosystem-Rwoodinawetperiod.

Therangeof temporalvariation inecosystem-Rwood(ratioof highesttolowestecosystem-Rwood,1.89)washigherinthisstudy sitethanthatobservedinNeotropicalforests(1.30,Chambersetal., 2004).Moreover,thelowestecosystem-Rwoodwasobservedduring aseveredroughtinJune2013,andwasfollowedby“generalflow- ering,”withmastingatthecommunitylevelatirregularintervals ofonetoseveralyears(Sakaietal.,2006;Nakagawaetal.,2012).

Thephenomenon of generalflowering is reported in aseasonal tropicalrainforestsinAsiaandislikelytriggeredbyunexpected droughtsinLambir(Sakaietal.,2006).Strongtemporalvariation innetprimaryproductionwasalsoreportedinastudyinLambir thatincludedaperiodofgeneralflowering(Khoetal.,2013),and annualabovegroundbiomassincrementsdecreasedintheyearof generalflowering,whereaslitterfallofflowersandseedssignifi- cantlyincreased(Nakagawaetal.,2012).Thesestudiesindicatethat generalfloweringfollowsdynamictemporalvariationinnetpri- maryproductionandRwood,reflectingincreasedcarbonallocation tothecanopyforreproductiveorgansduringgeneralflowering.

4.2. Annual-Rwoodandimpactoflargetrees

Annual-Rwoodestimatesbasedonthebest-fitmodelwerecom- parabletothosebased onModel1andModel2, suggesting that DBHand treespeciesarenotalwaysnecessaryfor estimatesof ecosystem-Rwoodinthisstudysite.Incontrast,annual-Rwoodbased onModel0 was higher than that from the best-fit model, and thedifference(1.50±3.03MgCha1year1)washigherthanthe netecosystemproductioninaNeotropicalforest(0.71±9.34MgC ha1year1,Phillips etal.,1998)andwasmorethanhalfofthe netecosystemproductioninanold-growthforest(2.4±0.8MgC ha−1year−1,Luyssaertetal.,2007).Thesedatasuggestthatscaling byaveragesonlymayleadtolargeerrors.Nonetheless,thisdiffer- encelikelyreflectsdifferencesinsgdistributionbetweensampled treesandthoseinthepresent4-haplot.Accordingly,averagemea- suredsg(0.80±0.95cmy−1)wasmuchhigherthanthatinthe4-ha plot(0.2±0.3cmy1).Thus,sgdistributionsshouldbeconsidered duringscalingofaveragedchamber-Rwoodvalues.

Annual-Rwoodvaluesdifferedamongthestudysites(Table4).

This variation was not explained by aboveground biomass, althoughsomestudieshaveshownpositiverelationshipsbetween plantrespirationandbiomass(KiraandShidei,1967;Deluciaetal., 2007).Inthepresentstudy,possiblefactorsaffectingvariationin annual-RwoodamongforestssuchasSAI,chamber-Rwood,andsamp- lingmethodarediscussedusingcomparisonsofannual-Rwoodand biomasswiththoseofpreviousstudies(Table4).Comparedwith annual-RwoodvaluesobservedinPeru(Malhietal.,2014),annual- Rwood values in thepresent studysite were similardespitethe twicehigherabovegroundbiomass.ComparableSAIandchamber- Rwoodledtocomparableannual-Rwoodbetweenthetwostudysites.

Higherstanddensityoflargetreesin thepresentstudysiteled tohigherbiomassbutcomparableSAI.ComparedwithbothBrazil studies(daCostaetal.,2013;Doughtyetal.,2013),annual-Rwoodin thisstudysitewaslowerdespitehigherbiomass.Lowerchamber- RwoodandslightlyhigherSAIinthisstudysiteresultedinlower annual-Rwood.Cavalerietal.(2006)reportedmuchlowerabove- groundbiomassbutslightlylowerannual-RwoodinCostaRica.The reasonfortheseresultsmaybeattributedtothedifferenceinthe samplingmethod;theymeasuredchamber-Rwoodfor lianas and palmsinadditiontoverticalvariationfortrees.Lianaandpalms accountedfor24and 9%oftotalRwoodintheforest,andunder- estimatesusingdatawithnoverticalmeasurementshaveoften beenreported(e.g.,Katayamaetal.,2014).Theseresultssuggest thatannual-Rwoodinthepresentstudyandpreviousstudiesmay

beunderestimated.Considerationoftreesizedistributions,inaddi- tiontodifferencesinSAI,chamber-Rwoodandsamplingdesign,may bethekeytounderstandingtherelationshipbetweenecosystem- Rwoodandbiomassacrossvariousforests.

GPPandannual-Rfoliage werethehighestandsecondhighest, respectively,amongtropicalrainforestsreportedintheliterature, whereasratiosofRwoodandRfoliagetoGPPwerewithintherangeof thoseshowninNeotropicalforests(Table4).MostratiosofRwood andRfoliagetoGPPinthisstudyandothertropicalrainforestswere higherthantheaveragesforvariousforestecosystems,including borealandtemporal,andyoungandold-growthforests(0.16and 0.16,respectively,Littonetal.,2007).Hence,ratiosofRabovetoGPP (0.38–0.51)were higherthantheaverageamong variousforest ecosystems(0.32,Littonetal.,2007).Inagreement,previousstud- iesshowedthatratiosofplantrespirationtoGPPwerehigherin tropicalforeststhanintemperateforests(Luyssaertetal.,2007;

Piaoetal.,2010).Althoughthecausesofthesedifferencesremain unclear,hightemperatures(Piaoetal.,2010),biomass(Kiraand Shidei,1967), andstandages(Deluciaetal.,2007)mayleadto higherratiosinmaturetropicalrainforests.

We estimated annual-Rwood with consideration of inter- individualandtemporalvariationinthepresentstudy,butthereare stilllargeuncertaintiesineachcomponentoftheGPPestimate.For annual-Rwood,therearefourmajorsourcesofuncertainty.First,ver- ticalvariationforsmallertreesandbrancheswasnotinvestigated, andcanleadtounderestimateswhenmeasurementsareconducted onlyatbreastheight(e.g.,Katayamaetal.,2014).Second,tempo- ralvariationwasinvestigatedinthepresentstudy,butonlyfor fivemeasurementscampaign,resultinginlargeerrorsinannual- Rwood.Toimproveaccuracy,relationshipsofenvironmentalfactors withtreephenologyshouldbeexaminedbymoremeasurement timeseries.Third,thesurfaceareaofwoodytissuewasestimated byanequationobtainedinthecentralAmazon(Chambersetal., 2004).Itisreportedthatheight–diameterrelationshipsweredif- ferentamongcontinents;foragivendiameter,stemsweretallest inAsia,followedbyAfrica,America,andAustralia,basedonameta- analysis(Baninetal.,2012).Thisresultsuggeststhatanequationfor woodytissuesurfaceareausingDBHmaybedifferentandunder- estimatedecosystem-Rwoodinthepresentstudy.Finally,thereis considerableuncertaintywhetherthesourceofchamber-Rwoodcan include root-respiredCO2 and stem-respiredCO2 can betrans- portedtothecanopy(Teskeyetal.,2008).Thisisoftenassociated withdiurnalvariationinchamber-Rwood.Differentdiurnalvariation inindividualtreesmaybeaffectedbythisuncertainty.

ThereareuncertaintiesinothercarbonfluxesusedforGPPesti- mation.Forannual Rfoliage,seasonalvariationsinRfoliage and LAI werenotconsideredinthepresentstudyandthevariationsmay leadtounder-oroverestimationalthoughRfoliage attheecosys- temscaledidnotvaryseasonallyinPeru(Malhietal.,2014)and Brazil(Metcalfeetal.,2010).Inthepresentstudy,Rfoliageat25C wasestimatedusingQ10of2.3reportedinCostaRica.IfQ10value of2.0,whichwasreportedinAustralia(Weerasingheetal.,2014), isusedfortheestimateinthepresentstudy,annual-Rfoliageisesti- matedas10.23MgCha−1year−1,suggestingunderestimationinthe presentstudy.Aswell,therearelargeuncertaintiesinTBCFinthe presentstudysite.TBCFwasestimatedasthedifferencebetween annualsoilCO2 effluxandannuallitterfall-C,basedonasteady- stateassumption(RaichandNadelhoffer,1989),whereasTBCFwas estimatedasasumofbelowgroundrespirationandnetprimary productioninpreviousstudiesinTable4.Furthermore,annualsoil CO2 effluxwasestimatedbymeasurementsonlyinthedaytime andtemporalvariationwasnotconsidered.Thispracticemaylead tooverestimationduetodiurnalvariationinsoilCO2efflux.

Climatechangeispredictedtocausemorefrequentdroughtsof greaterseverityintropicalareasMalhiandPhillips(2004).Drought mayaffectphenologyandcarbonallocationinthepresentstudy

(8)

Table4

Abovegroundbiomass,annual-Rwood,annual-Rfoliage,annual-Rabove,GPPandratiostoGPPinvarioustropicalrainforests.

Site Malaysia CostaRica French

Guiana

Brazil Brazil Peru

Borneo LaSelva Paracou Caxiuana

tower

Caxiuana Terrapreta

Caxiuana Control

Caxiuana Dry

TAM-05 (Nutri- entpoor)

TAM-06 (Nutri- entrich) Aboveground

biomass (MgCha−1)

272.4 81 228 179.5 110.9 213.9 200.6 138.2 115.6

Annual-Rwood

(MgCha−1 year−1)

7.06±2.09 5.08±1.35 5.44±0.99 8.71±1.07 8.46±2.82 10.21±4.49 11.17±4.96 5.43±1.77 7.62±2.48

Surfacearea index(m2m−2)

2.01 1.1(diame-

ter<10cm)

1.3 1.65 1.63 1.98 2.09

AverageCham- berRwood

0.81±0.52 (dry) 1.39±0.74 (wet) 1.07±0.48 (average)

0.60(DBH of10cm) 0.83(DBH of80cm)

1.13±0.55 (dry) 1.56±0.65 (wet)

2.08±0.15 2.11±0.17 1.61±0.12 1.94±0.19 0.95±0.03 1.03±0.08

Scalingfactors Growth, dbh, species

Growth, diameter, Species, height, slope

Average Growth Growth Growth

Samples DBH>10cm Alltrees, Liana,Palm

DBH>10cm Alltrees Alltrees Alltrees

Annual-Rfoliage

(MgCha−1 year−1)

9.54±0.46 11.59±0.54 5.02±1.58 5.09±1.67 5.69±2.14 9.26±3.63 8.86±2.84 6.43±2.07

Annual-Rabove

(MgCha−1 year−1)

16.60±2.14 17.03±0.55 13.73±4.4 13.55±4.49 15.9±6.63 20.43±8.59 14.29±4.61 14.05±4.55

ANPP 6.76±0.57 9.35±0.49 8.75±0.59 6.78±0.14 5.71±0.20 9.96±0.83 11.59±1.62

TBCF 19.63±6.37 9.01±1.22 13.3±1.62 16.5±4.03 14.3±4.20 11.26±1.11 8.85±1.06

GPP(MgCha−1 year−1)

42.99±6.74 37.57 32.08±3.46 35.68±3.65 39.18±1.81 40.44±6.39 35.47±3.55 34.47±3.53 Rwood/GPP 0.17±0.05 0.14 0.27±0.09 0.24±0.08 0.26±0.12 0.28±0.13 0.15±0.05 0.22±0.08

Rfoliage/GPP 0.23±0.01 0.31 0.16±0.05 0.14±0.05 0.15±0.06 0.23±0.10 0.25±0.08 0.19±0.06

Rabove/GPP 0.40±0.06 0.45 0.43±0.11 0.38±0.10 0.41±0.14 0.51±0.17 0.40±0.10 0.41±0.10

Reference Thisstudy Cavaleri etal.,2006

Stahletal., 2011

Doughty etal.,2013

daCosta etal.,2013

Malhietal., 2014 Rowland

etal.,2014

daCosta etal.,2010 site,andseveredroughtofteninduceshightreemortality(Allen

etal.,2010).InLambirHillsNationalPark,severedroughtcaused byElNinoin1997–1998inducedhightreemortality,especially amonglargetrees(Nakagawaetal.,2000).Recentstudiesalsoshow thatlargetreesdriveforestabovegroundbiomassvariationinmoist lowlandforestsacrossthetropics (Slik etal.,2013)and higher stemgrowthattheplotlevelinBorneo(Baninetal.,2014).These observationssuggestthatlargetreesareimportantcarbonstores.

PreviousstudiesinLambirshowedthatpartialvariationinsoilCO2 effluxwasaffectedbylargetrees(Katayamaetal.,2009),andsoil CO2 effluxwashigherinthepresenceofemergenttrees(Ohashi etal.,2014).Moreover,photosyntheticabilityreportedlyincreases linearlywithheight(Kenzoetal.,2015),suggestinghigherpho- tosynthesisratesforlargertrees.Hence,soilCO2effluxandGPP mightdecreasewiththelossoflargetreesduetoseveredroughts, whereasRwoodmightbeconserved.Takentogether,futureclimate changecanaltercarboncyclingthroughthechangeincarbonallo- cationnotonlybythephysiologicalandphenologicalresponseto droughtforlargetrees,butalsochangesinnumberoflargetrees.

5. Conclusions

Wemeasuredchamber-Rwood intrees ofvariousspecies and sizes in a Bornean tropical rainforest. All measurements were performedfive times between2012and 2014,and subsequent analyses indicated that the factor of stem growth rate was

themostimportantdeterminantofinter-individualvariationin chamber-Rwood,followedbyDBHandtreespecies.Inthis study, ecosystem-Rwoodwasestimatedbasedonthemodelforindividual chamber-Rwoodusingthepredictivefactorsofstemgrowthrate, DBH,andtreespecies,andtheaveragemeasuredchamber-Rwood value.Estimatesofecosystem-Rwoodusingstemgrowthrateonly weresimilartothoseestimatedusingtheallfactors,whereasmean measuredecosystem-Rwoodwashigherthanthatestimatedusing regressionmodels,suggestingthatspatialscalingshouldbecon- ducted withconsideration of factorsthat affectchamber-Rwood. Moreover,ecosystem-Rwoodvariedtemporally,reflectingchanges insoilmoisturedespitetheabsenceofaregularseasonaldryperiod.

Thesedatasuggestthatcarbonallocationmaydynamicallyvary withtimeandchangesinsoilmoisture.Annual-Rwoodvalueswere comparabletothoseinNeotropicalforests,despitehigherabove- groundbiomassinthepresentBorneanstudysite.Higherstand density of largetrees caused higher aboveground biomass, but comparableecosystem-RwoodcomparedwithNeotropicalforests.

Althoughtheeffectsofstandageorbiomassonautotrophicrespira- tionremainunclear,thepresentdatasuggeststhatstandstructure maymediatetheseeffectsonrespirationattheecosystemscale.

Acknowledgments

WethanktheForestDepartmentofSarawakandtheSarawak ForestryCorporationfortheirpermissionandkindassistancein

Referensi

Dokumen terkait