Available via license: CC BY 4.0
Content may be subject to copyright.
RemoteSens.2024,16,2393.https://doi.org/10.3390/rs16132393www.mdpi.com/journal/remotesensing
Article
ImpactofFragmentationonCarbonUptakeinSubtropical
ForestLandscapesinZhejiangProvince,China
JiejieJiao
1,2
,YanCheng
3
,PinghuaHong
4
,JunMa
5
,LiangjinYao
2
,BoJiang
2
,XiaXu
1
andChupingWu
2,
*
1
StateKeyLaboratoryofSubtropicalSilviculture,SchoolofEnvironmentalandResourcesScience,Zhejiang
A&FUniversity,Lin’an311300,China;2023202012007@stu.zafu.edu.cn(J.J.);xuxia1982@zafu.edu.cn(X.X.)
2
ZhejiangHangzhouUrbanEcosystemResearchStation,ZhejiangAcademyofForestry,
Hangzhou310023,China;lj890caf@hotmail.com(L.Y.);jiangbo2246@hotmail.com(B.J.)
3
HangzhouImmigrationInspectionStation,Hangzhou311223,China;hxj0201@hotmail.com
4
YuhangEcologicalandEnvironmentalMonitoringStationofHangzhou,Hangzhou311100,China;
hongpinghua87@gmail.com
5
MinistryofEducationKeyLaboratoryforBiodiversityScienceandEcologicalEngineering,
SchoolofLifeSciences,FudanUniversity,Shanghai200433,China;ma_jun@fudan.edu.cn
*Correspondence:wcp1117@hotmail.com
Abstract:Globalchangescausewidespreadforestfragmentation,which,inturn,hasgivenriseto
manyecologicalproblems;thisisespeciallytrueiftheforestcarbonstockisprofoundlyimpacted
byfragmentationlevels.However,thewayinwhichforestcarbonuptakechangeswithdifferent
fragmentationlevelsandthemainpathwaythroughwhichfragmentationaffectsforestcarbonup-
takearestillunclear.Remotesensingdata,vegetationphotosynthesismodels,andfragmentation
modelswereemployedtogenerateatimeseriesGPP(grossprimaryproductivity)dataset,aswell
asforestfragmentationlevelsforforestlandscapesinZhejiangprovince,China.WeanalyzedGPP
variationwithforestfragmentationlevelsandidentifiedtherelativeimportanceofthephenology
(carbonuptakeperiod—CUP)andphysiology(maximumdailyGPP—GPP
max
)controlpathwaysof
GPPunderdifferentfragmentationlevels.Theresultsshowedthatthenormalizedmeanannual
GPPdataofhighlyfragmentedforestsduringtheperiodfrom2000to2018weresignificantlyhigher
thanthoseofotherfragmentationlevels,whiletherewasalmostnosignificantdifferenceinthe
annualGPPtrendofforestlandscapeswithallfragmentationlevels.Moreover,thepercentagearea
ofthecontrolvariable,GPP
max
,graduallyincreasedwithfragmentationlevels;themeanGPP
max
between2000and2018ofhigh-levelfragmentationwashigherthanthatofotherfragmentation
levels.Ourresultsdemonstratethatthecarbonuptakecapacityperunitareawasenhancedin
highlyfragmentedforestareas,andthemaximumphotosyntheticcapacity(physiology-basedpro-
cess)playedanimportantroleincontrollingcarbonuptake,especiallyinhighlyfragmentedforest
landscapes.Ourstudycallsforabeeranddeeperunderstandingofthepotentialofforestcarbon
uptake,anditisnecessarytoexplorethemechanismbywhichforestfragmentationchangesthe
vegetationphotosyntheticprocess.
Keywords:forestcarbonsequestration;vegetationgrowth;landscapepaerns;grossprimary
productivity;phenology
1.Introduction
Forestsareakeycomponentinterrestrialecosystems[1];globally,theyplayanim-
portantroleinprovidingavarietyofecologicalservicesformankind,suchasabsorbing
andstoringcarbon,maintainingsoilandwater,andpurifyingair[2–4].However,the
world’sterrestrialbiomeshavebeensignificantlyaffectedbyhuman-inducedglobal
changes[5].Theglobalforestareahasshrunkbyaboutone-thirdsincethe1850s[6],most
ofwhichwasinducedbynaturaldisturbancesandtheexpansionoftheagriculturaland
Citation:Jiao,J.;Cheng,Y.;Hong,P.;
Ma,J.;Yao , L.;Jiang,B.;Xu,X.;Wu,
C.ImpactofFragmentationon
CarbonUptakeinSubtropicalForest
LandscapesinZhejiangProvince,
China.RemoteSens.2024,16,2393.
hps://doi.org/10.3390/rs16132393
AcademicEditors:ZhixinQi,LeYu,
LeiFang,KasturiDeviKanniahand
BrianAlanJohnson
Received:22May2024
Revised:25June2024
Accepted:28June2024
Published:29June2024
Copyright:©2024bytheauthors.
LicenseeMDPI,Basel,Swierland.
Thisarticleisanopenaccessarticle
distributedunderthetermsand
conditionsoftheCreativeCommons
Aribution(CCBY)license
(hps://creativecommons.org/license
s/by/4.0/).
RemoteSens.2024,16,23932of14
urbanindustries[7].Therapidandvastlossofforestresultsinthespliingofspatially
contiguousdistributedforestlandscapeintosmallpatches,leadingtoworldwideforest
fragmentation[8].Consideringthatsomekeyfunctionsofforestscanbedamagedbyfrag-
mentationthroughthedegradationofvegetationbiomassandbiodiversity[9,10],itisnec-
essarytocomprehensivelyunderstandtheimpactsofforestfragmentationonecosystems.
Aseriesofecologicalproblems,particularlycarbonlossandtheintensificationofthe
greenhouseeffect,aredirectlyinducedbyforestfragmentation[11–13].Severalstudies
haveindicatedthathumanactivitiesinducemoreforestedgeareas,aggravatingfrag-
mentationanddecreasingcarbonstorage[14,15].However,carbonuptakeisalsoproven
tobeenhancedinhighlyfragmentedforests[16,17].Thecomplexvariationincarbonup-
takeunderdifferentfragmentationlevelsproposesademandforassessingtherelation-
shipandtheinfluencepathwaybetweenforestfragmentationandvegetationcarbonup-
take,whichcanenhanceourinsightsintothefuturecarbonuptakepotentialofforestsin
thecontextofwidespreadforestfragmentation.Inparticular,inthecontextofintenseland
usecausedbyurbanization,forestsshowahighdegreeoffragmentation[8].Inaddition
tothedirectimpactofforestareachangeonvegetationcarbonsequestration,itisunclear
howforestdistributionpaernsaffectforestcarbonsequestration;thiswillgreatlylimit
theeffectivenessofurbanforestmanagementandecologicalconservationmeasures,and
willalsolimitourunderstandingofregionalcarboncycledynamics.
Anumberofmethodshavebeenusedtomeasureforestfragmentationlevels,with
thepatchorclasslevel’sfragmentation-relatedlandscapepaernmetricsbeingadopted
bymoststudiestoquantifythelevelsofforestfragmentation[12,16,18].However,these
approacheshaveneglectedthefactthatforestfragmentationisalandscape-scaleissue
containingcomplexspatialprocesses[19].Althoughtherearesomehierarchicalap-
proaches,whicharebasedonspatialprocesses,tomeasuringthelevelsofforestfragmen-
tation[20–22],theseweremainlyusedtoestimatethestatusofforestfragmentation,and
haverarelybeenusedtostudytheimpactsoffragmentationonecologicalfunctionsdue
tothedifficultyofaccessingfieldobservationdataforrelevantfragmentationlevels.In
reality,theclassificationresultsofthefragmentationlevelsfromthesemethodsmayhave
greatpotentialinstudiesinvestigatingtheecologicaleffectsofforestfragmentation,par-
ticularlythosewiththeassistanceoftheremotesensingapproach,toestimateecological
functionindicatorsatlargespatialscales.
Theprocessesofforestcarbonuptakecanbesimplysummarizedastheabsorption
ofatmosphericcarbondioxidebyplantsthroughphotosynthesis[23];grossprimary
productivity(GPP),adirectmeasurementofthetotalamountofcarbonassimilatedby
vegetation,isgenerallyconsideredasareflectionofvegetationcarbonuptake.Inaddition,
thevegetationcarbonstockishighlyrelatedtotheaccumulationofannualGPPovera
longperiod,whichismainlycontrolledbytwofactors—thecarbonuptakeperiod(CUP)
andthemaximumdailyGPP(GPPmax),whichareassociatedwithvegetationphenology
andphysiology,respectively[24,25].Bothofthesefactorsareaffectedbytheclimateand
soilnutrientchangecausedbyforestfragmentation[26–28];therefore,detectingthevari-
ationinCUPandGPPmaxunderdifferentfragmentationlevelsiscrucialforexplainingthe
pathwayorthemechanismoftheimpactoffragmentationonforestcarbonuptake.
Inthisstudy,forestlandscapesinZhejiangprovince,China,wereselectedasthe
studyareainwhichtoexaminetherelationshipbetweenforestfragmentationandGPP.
Multiplesourcesofremotesensingdatawereappliedtocalculateforestfragmentation
levelsandtosimulatevegetationGPP.WehaveproposedabasicassumptionthattheGPP
anditstwocomponents—CUPandGPPmax—varybetweenforestfragmentationlevels.
BasedontheanalysisofthevariationinGPP,CUP,andGPPmaxwithfragmentationlevels,
wewouldliketoanswerthefollowingtwoscientificquestions:(1)Howdoesforestcarbon
uptakechangewithdifferentfragmentationlevels?(2)Whatisthemainpathwayaffecting
forestcarbonuptakeunderdifferentfragmentationlevels?
RemoteSens.2024,16,23933of14
2.MaterialsandMethods
2.1.StudyArea
Zhejiangprovince,China,whichextendsacross27.03°–31.18°Nand118.02°–
123.17°E,wasselectedasthestudyarea(Figure1).TheforestcoverageofZhejiangprov-
inceisabout59.4%,mostofwhichisdistributedinthemountainsandhills.Thesubtrop-
icalmonsoonclimateisthedominantclimateofZhejiangprovince;ithashot,rainysum-
mersandcold,drywinters,andhasobviouscharacteristicsofseasonalchange.Themean
annualtemperatureofZhejiangisbetween15and18°C,whiletheaverageannualpre-
cipitationrangesbetween980and2000mm.ThegrowingseasonusuallystartsinMarch
andendsinOctober.Theevergreenbroadleafforestsandmixedconiferous–broadleaffor-
estsarethemainvegetationtypesofthestudyarea(Figure1);therearemorethanthirty
treespeciesintheforestlandscapesofZhejiangprovince.
Inthisstudy,thetropicalforestlandscapesofZhejiangwereselectedasthestudy
areaprimarilyforthefollowingreasons:ontheonehand,Zhejianghasadevelopedecon-
omy,wherebyurbanizationandotherhumanactivitiesinZhejianghavecauseddrastic
landusechanges.Inaddition,Zhejianghasalargeforestareaandahighforestcoverage;
thesetwofactorsresultinthedistributionpaernofsubtropicalforestsinZhejiangcon-
tainingalltypesofforestfragmentation,aswellashavingtypicalgradientsofforestfrag-
mentation.Ontheotherhand,Zhejianghasturnedtothestageofecosystemprotection
afteritsearlydevelopment;thechangeinforestcoverafter2000isquitelimitedcompared
tothatofitsneighboringregions,whichmakesitconvenientforthisstudytocomparethe
productivityanditstrendamongstaticfragmentationtypes.
Figure1.Location,foresttypedistribution,andclassifiedforestfragmentationlevelsofforestland-
scapesinZhejiang,China.(a)ThelocationofZhejiang,China.(b)Distributionofmainforesttypes
inZhejiang.(c)ForestfragmentationlevelsinZhejiang.UNC:unclassifiedforest,PAT:patchforest,
TRA:transitionalforest,PER:perforatedforest,UND:undeterminedforest,EDG:edgeforest,INT:
interiorforest.TheforesttypedatawerederivedfromtheMODISMCD12Q1landcoverproduct
(from2010),whilethemapoffragmentationlevelswascalculatedusingtheforestfragmentation
model(seeSection2.2)inthisstudy.
RemoteSens.2024,16,23934of14
2.2.ForestMapDataandFragmentationLevelsClassification
Consideringthefactthatforestcoverchangemayaltertheforestfragmentationcat-
egorieswithtime,thisstudyrequiresatemporalstablestatusofforestcover.Inorderto
excludetheinfluencefromforestcoveragechangeontheaccuracyoftheclassificationof
forestfragmentationlevels,ourstudyonlyfocusedontheforestareasinZhejiangprov-
incethathavenotchangedsince2000.Weusedgridsof500minsizetocalculatethe
proportionofforestlossbetween2000and2018(theratiooftotalforestlosspixelstototal
pixelsina500mgrid)basedontheGlobalForestChangedataset[29];thegridswitha
forestlossproportionoflessthan5%wereretainedandwererepresentativeofunchanged
forestlandscapes.Moreover,a50mresolutionforestmapfrom2010,generatedusing
ALOSPAL SAR- 2,MODIS,andLandsat-7/8satellitedata[30],wasselectedasthebasic
forestcoverdatatoconductaclassificationofforestfragmentationlevelsbasedonthe
retainedgrids(unchangedforestlandscapes).
Aforestfragmentationmodel[20,21]thatiswidelyusedtoestimatethestatusof
forestfragmentationbasedonsatellite-derivedforestmapswasemployedinthisstudyto
classifyforestfragmentationlevels.Thisfragmentationmodelwasdevelopedbasedon
theconsiderationofcausingforestlossasaresultofspatialdisturbances;ithasbeen
widelyappliedtoestimateforestfragmentationbyanalyzingforestdistributionmaps
[31,32].Additionally,forestareadensity(Pf)andforestconnectivity(Pff)aretwokeyindi-
catorsinthismodelfordefiningthefragmentationtypeofagiven“window”or“land-
scape”basedonaforest/non-forestbinarymap[20].Thesizeofthe“window”reflects
differentspatialscalesintheestimationofforestfragmentation.PfandPffarecalculated
usingthefollowingequations:
𝑃
(1)
𝑃
𝐷
𝐷
(2)
wherePfistheproportionofforestpixelsinagivenwindow,calculatedbydividingthe
numberofforestpixels(Nf)bythetotalnumberofpixels(Nw)inthewindow;Pffisthe
forestconnectivity,calculatedbydividingthepixelpairnumberthatincludesatleastone
forestpixel(Dff)bythepixelpairnumberthatincludestwoforestpixelsincardinaldirec-
tions(Df).
Alook-uptable(Table1)containingthecriteriaofthetwoindicators(DffandDf)was
usedtoconducttheclassificationofforestfragmentationlevels.Sixfragmentationtypes,
includingpatch(PAT),transitional(TRA),perforated(PER),undetermined(UND),edge
(EDG),andinterior(INT),wereobtained.Thefragmentationlevelsofthesesixtypesare
shownasPAT>TRA>PER>UND>EDG>INT.However,UNDwasexcludedfromour
studyduetothefactthatitonlyexistsinextremecases.Inaddition,inordertomatchthe
resolution(~500m)oftheGPPdata,a9×9size“window”wasadoptedinthisstudy.
Tab l e1.Classificationcriteria,area,andareapercentageofeachforestfragmentationlevelofthe
forestfragmentationmodelinforestlandscapesinZhejiang,China.PAT:patchforest,TRA:transi-
tionalforest,PER:perforatedforest,UND:undetermined,EDG:edgeforest,andINT:interiorforest.
Fragmentation
Categories
Classification
Criteria
Area
(104ha)
AreaPercentage
(%)
PATPf<0.4263.033.7
TRA0.4<Pf<0.6129.516.6
PER0.6<=Pf<1andPf>Pff226.128.9
UND0.6<=Pf<1andPf=Pff0.70.1
EDG0.6<=Pf<1andPf<Pff153.019.6
INTPff=19.01.2
RemoteSens.2024,16,23935of14
2.3.TimeSeriesGPPDataset
Asatellite-basedlight-useefficiencymodel—thevegetationphotosynthesismodel
(VPM)—wasusedtosimulatedtimeseriesGPPdatasetfrom2000to2018forthestudy
area.TheVPMwasdrivenusingtheMODISMOD09A1surfacereflectancedataset,the
MCD12Q1landcoverdataset,theMYD11A2landsurfacetemperaturedataset,andNCEP
reanalysisIIclimatedata.Thefinal500mresolutionGPPofeach8-dayperiodduring
2000–2018wasobtained.ThedetailsofthesimulationprocessesusingtheVPMhavebeen
reportedinpreviousstudies[33,34].
Basedonthe500m8-dayperiodGPPdataset,theannualGPPforeachyearbetween
2000and2018wasobtainedbyaggregatingallthedatainarelevantyear.Inaddition,the
GPPmaxforeachyearbetween2000and2018wasobtainedbyextractingthepeakvaluefor
eachpixelineachyear.Sincethevegetationinourstudyareausuallyhasobviouschar-
acteristicsofseasonalchange,thetemporaldynamiccurveofGPPgenerallyshowsauni-
modalshape.Therefore,inthisstudy,thegrowingseasonisdefinedusingarelative
thresholdmethod,assuggestedinpreviousstudies[35,36];thethresholdissetas20%of
theGPPmaxinagivenyear.Thecarbonuptakeperiod(CUP)foreachyearbetween2000
and2018wasobtainedbycountingthenumberofdaysinwhichtheGPPwashigherthan
thevalueof20%oftheGPPmaxineachyear.
2.4.DataAnalysis
2.4.1.ComparisonofGPPTrendamongDifferentFragmentationLevels
Inordertoexploretheimpactofforestfragmentationonvegetationproductivity,we
comparedthemeanannualGPPandthenormalizedmeanannualGPPamongdifferent
fragmentationlevelsfrom2000to2018.ThenormalizedmeanannualGPPthroughout
2000–2018wascalculatedbydividingthemeanannualGPPfrom2000to2018bythePf
foreachpixelinordertoexcludetheinfluenceofforestareasonGPPvaluesforapixel.A
one-wayANOVAwasadoptedtotestthesignificanceofthedifferencebetweenmeanan-
nualandmeannormalizedannualGPPdatabetween2000and2018amongforestfrag-
mentationlevels.Meanwhile,theLSDmultiplecomparisonmethodwasusedtodetect
thedifferencebetweenmeanannualandnormalizedmeanannualGPPdatabetween2000
and2018undervariousfragmentationlevels.
ThelineartrendoftheannualGPPduring2000–2018foreachpixelandtheentire
Zhejiangprovincewascalculated,andtheslopevalueofthelinearregressionlinewas
representedasthemagnitudeofthetrend.TheannualGPPtrendofdifferentfragmenta-
tionlevelsthroughout2000–2018wasalsocomparedusingaone-wayANOVA.
2.4.2.DetectingtheRelativeImportanceoftheControlPathwaysofGPP
InordertoexplorethecontrolpathwaysofGPPinourstudyarea,weconducted
multiplelinearregressionbetweenGPPandeitherCUP,GPPmax,oracombinationofCUP
andGPPmaxforeachpixeloftheperiodbetween2000and2018.Ametricassessingmeth-
odinthe“relaimpo”packageintheRsoftware[37]wasemployedtoquantifytherelative
importanceofeachindividualvariableinlinearmodelsforeachpixelinthisstudy;the
variablewiththehighestrelativeimportancevaluewasregardedasthecontrolvariable
ofGPP.Therelativeimportanceofeachindividualcontrolvariablewascalculatedtore-
flectitsdegreeofcontrol,andthevalueswerenormalizedfrom0to100%.Inaddition,the
pixelnumberpercentagesofthethreecontrolvariableswerealsocalculatedandwere
statisticallycomparedforeachforestfragmentationlevel.
RemoteSens.2024,16,23936of14
2.4.3.ComparisonofGPPControlPathwaysamongDifferentFragmentationLevels
Inordertoanswerthequestionofwhichpathwayofforestfragmentationaffectsveg-
etationproductivity,theCUP,GPPmax,andCUP×GPPmaxwerecalculatedandcompared
amongvariousforestfragmentationlevels.Aone-wayANOVAwasadoptedtotestthe
significanceofthedifferencesinthethreecontrolvariables,andtheLSDmultiplecom-
parisonmethodwasusedtodetectthedifferencebetweenthethreecontrolvariablesun-
dervariousfragmentationlevels.
2.4.4.StatisticalAnalysisandDataProcessingMethods
Therasterdataprocessing,mapping,andstatisticalanalysiswereconductedusing
ArcGIS10.3andRsoftware[38];thesignificanceleveloftheone-wayANOVAwassetto
0.05.
3.Results
3.1.GPPDynamicsofDifferentFragmentationLevels
Therewasasignificant(p<0.05)differenceinthemeanannualGPPin2000and2018
forforestlandscapeswithvariousfragmentationlevelsinZhejiang(Table2).Generally,
forestlandscapeswithfragmentationlevelsofPATandINThadthelowestandhighest
meanannualGPPvalues,respectively,whilenosignificantdifferenceinthemeanannual
GPPwasidentifiedbetweenTRA,PER,andEDG.Althoughasignificant(p<0.05)differ-
enceinthenormalizedmeanannualGPPfor2000and2018inZhejiangwasfoundamong
allforestfragmentationlevels(Table2),thehighestandlowestnormalizedmeanannual
GPPvaluesinforestlandscapeswererepresentedbythePATandINTfragmentationlev-
els,respectively.Similarly,therewasnosignificantdifferenceinthenormalizedmeanan-
nualGPPbetweenPERandEDG.
Tab l e2.ComparisonofmeanannualGPPandmeannormalizedannualGPPamongdifferentforest
fragmentationcategoriesfor2000and2018.ThemeannormalizedannualGPPwascalculatedasthe
ratioofthemeanannualGPPtotheproportionofforestpixelsinaforestlandscape.Thecharacters
indicatetheresultsofmultiplecomparisonsbetweenanytwocategories.PAT:patchforest,TRA:
transitionalforest,PER:perforatedforest,EDG:edgeforest,andINT:interiorforest.
Fragmentation
Categories
MeanAnnualGPP(gCm−2)MeanNormalizedAnnualGPP(gCm−2)
2000201820002018
PAT1040.4±319.1a1101.6±449.4a11,611.3±17,514.1d11,337.4±16,718.4d
TRA1268.7±256.4b1516.8±351.1b2979.0±4071.6c3505.2±4629.4c
PER1366.5±193.1bc1643.4±231.5bc1870.6±1276.2b2246.4±1487.0b
EDG1349.8±202.4bc1635.2±246.3bc2199.1±2681.0b2656.5±3158.8b
INT1421.4±197.6c1693.9±213.9c1657.0±1222.9a1972.6±1478.9a
Significant(p<0.05)increasingtrendswerefoundinallfragmentationlevelsandall
typesofforestlandscapesinZhejiang,exceptforthePATlandscape(Figure2a).Thein-
creaseintherateofannualGPPwithregardtovariousfragmentationlevelsfluctuated
from10.6gCm−2year−1inTRAto13.1gCm−2year−1inPER.Spatially,theannualGPP
trendinZhejianggenerallyshowedpositivevaluesformostpixels(Figure2b).Theannual
GPPtrendinPATwassignificantlylowerthanthoseinforestlandscapeswithotherfrag-
mentationlevels,buttherewasnosignificantdifferenceintheannualGPPtrendamong
TRA,PER,EDG,andINT.Thehighervalues(>30gCm−2year−1)oftheannualGPPtrend
weremainlydistributedinthewestandcentralpartofZhejiang,whilethelowervalues
(<−45gCm−2year−1)oftheannualGPPtrendweremainlydistributedinthenorthand
coastalareasofZhejiang.
RemoteSens.2024,16,23937of14
Figure2.(a)AnnualGPPdynamicsofdifferentfragmentationlevelsand(b)lineartrenddistribu-
tionofannualGPPofforestlandscapesinZhejiangbetween2000and2018,andthecomparisonof
theannualGPPtrendvaluesamongdifferentfragmentationlevels.Theredasteriskindicatesthe
annualGPPtrendofPATissignificantly(p<0.05)thanthatofotherfragmentationcategories.PAT:
patchforest,TRA:transitionalforest,PER:perforatedforest,EDG:edgeforest,andINT:interior
forest.
3.2.RelativeImportanceofDifferentGPPControlPathways
CUP×GPP
max
wasidentifiedasbeingthecontrolvariableforabout61.8%oftheen-
tireZhejiangprovince;theareapercentagesofthecontrolvariablesforCUPandGPP
max
were19.7%and18.5%,respectively(Figure3a).ThecontrolvariableforCUPwasmainly
distributedinsomepartsofwesternandcentralZhejiang,whilethecontrolvariablefor
GPP
max
wasmainlydistributedincentralandeasternZhejiang.Therelativeimportanceof
thethreecontrolvariablesrangedfrom0to82%,andmostofthemweredistributedinthe
rangeof5~35%(Figure3b).TheareapercentagesofthecontrolvariablesforCUP×GPP
max
weremaintainedatapproximately60%amongvariousfragmentationlevels(Figure3c).
However,thecontrolvariableforCUP(13.2~27.8%)andGPP
max
(11.1~25.0%)stillac-
countedforacertainproportionofareaamongvariousfragmentationlevels.Therewas
nosignificant(p>0.05)differenceintherelativeimportancebetweenanytwofragmenta-
tionlevels,andthemeanrelativeimportanceofvariousfragmentationlevelsrangedbe-
tween25%and28%(Figure3d).
RemoteSens.2024,16,23938of14
Figure3.Distributionofdifferentcontrolvariablesandtheirrelativeimportance.(a)Spatialdistri-
butionofdifferentcontrolvariables,(b)spatialdistributionoftheirrelativeimportanceonannual
GPPduring2000–2018,(c)comparisonofrelativeimportanceamongthreecontrolvariables,and
(d)pixelnumberpercentagesofthreecontrolvariablesfordifferentfragmentationlevels.PAT:patch
forest,TRA:transitionalforest,PER:perforatedforest,EDG:edgeforest,andINT:interiorforest.
RemoteSens.2024,16,23939of14
3.3.GPPControlPathwaysofDifferentFragmentationLevels
Asignificant(p<0.05)differenceinthemeanCUPofforestlandscapesinZhejiang
during2000–2018wasdetectedamongallfragmentationlevels(Table3).Thehighestand
lowestmeanCUPwererepresentedbytheINTandPATforestlandscapes,respectively,
andtherewasnosignificant(p>0.05)differenceinthemeanCUPbetweenTRA,PER,
EDG,andINT.Moreover,themeanGPP
max
ofINTforestlandscapeswassignificantly(p
<0.05)higherthanthatforthePATandTRAlandscapes;nosignificant(p>0.05)difference
inthemeanGPP
max
existedinforestlandscapesinZhejiangbetweenTRA,PER,andEDG,
orbetweenPER,EDG,andINT.Similarly,asignificant(p<0.05)differenceinthemean
CUP×GPP
max
offorestlandscapesinZhejiangduring2000–2018wasdetectedamongall
forestfragmentationlevels(Table3).ThehighestandlowestmeanCUP×GPP
max
values
existedintheINTandPATlandscapes,respectively,andtherewasnosignificant(p>0.05)
differenceinthemeanCUP×GPP
max
betweenPERandEDG.
Spatially,themeanvaluesofCUPandGPP
max
intheforestlandscapesofZhejiang
during2000–2018bothhavehighheterogeneities.ThehighermeanCUPvalues(>330
days)weregenerallyfoundinthesouthandeastpartsofZhejiang,whilethelowermean
CUPvalues(<210days)weregenerallyfoundinthenorthpartofZhejiang(Figure4a).
Amongtheregionswithahighforestcoverage,therewasliledifferenceinthemean
CUPthroughout2000–2018,butthemeanGPP
max
during2000–2018presentedobvious
spatialvariations.Inparticular,thehighestvalues(>12.5gCday
−1
)ofthemeanGPP
max
during2000–2018wereconcentratedindenseforestlandscapesintheeastandnorthwest
partsofZhejiang(Figure4b);however,thevaluesofmeanGPP
max
during2000–2018were
generallylessthan10gCday
−1
inthesouthdenseforestlandscapesinZhejiang.
Figure4.Spatialdistributionofthemeanvaluesofcarbonuptakeperiod(CUP)andmaximumdaily
GPP(GPP
max
)forsubtropicalforestlandscapesinZhejiangProvinceduring2000–2018.(a)Thespa-
tialmapofthemeanCUPduring2000–2018.(b)ThespatialmapofthemeanGPP
max
during2000–
2018.
RemoteSens.2024,16,239310of14
Tab l e3.ComparisonsofmeanCUP,meanGPPmax,andmeanCUP×GPPmaxofforestlandscapesin
Zhejiangamongdifferentfragmentationcategoriesduring2000–2018.Thecharactersindicatethe
resultsofmultiplecomparisonsbetweenanytwocategories.PAT:patchforest,TRA:transitional
forest,PER:perforatedforest,EDG:edgeforest,andINT:interiorforest.
Fragmentation
Categories
CUP(days)GPPmax(gCm−2day−1)CUP×GPPmax
PAT256.1±73.4a8.7±2.2a2339.4±949.7a
TRA310.7±49.7b9.6±1.6b3041.2±737.1b
PER329.8±28.5b10.2±1.3bc3369.1±548.5c
EDG327.8±30.3b10.1±1.3bc3320.6±551.2bc
INT332.9±28.1b10.7±1.5c3547.7±524.7d
4.Discussion
4.1.ImpactsofForestFragmentationonCarbonUptake
ThemeanannualGPPthroughout2000–2018generallydeclinedfromINTtoPAT
forestlandscapes(Table2);thisisinlinewithpreviousstudies,whichsuggestthatcarbon
lossinforestsisaggravatedbyfragmentationlevel[39–41].Thecarbonlossisdirectly
aributedtotheforestlossunderforestfragmentation,whichmaybemainlycausedby
naturalandhumandisturbances,suchaslogging,fire,andagricultureandurbanexpan-
sion[5,27,42].However,ourresultsalsoshowedthatthenormalizedmeanannualGPPof
PATforestlandscapesduring2000–2018wassignificantlyhigherthanthatofotherfrag-
mentationlevels(Table2),whichimpliesthatthecarbonuptakeofhigherfragmented
forestareaswasenhancedwhenexcludingtheinfluenceofforestareadensity.Asimilar
resultwasfoundinatemperatebroadleafforestofnorthAmerica[16],wheretheforest
growthratedecreasedwiththedistancefromforestedge.Thevariationsinlightavaila-
bility[43]andthesensitivitytoclimate[44,45]oftheforestswithdifferentfragmentation
levelsarelikelytoberesponsiblefortheirdifferencesincarbonuptake.
Inthisstudy,exceptforthePATforestlandscapes,therewasnosignificantdifference
intheannualGPPtrendthroughout2000–2018amongfragmentationlevels(Figure3);
thiscouldbeaributedtotwopossiblereasons.Ontheonehand,forestlandscapeswith
ahighfragmentationlevelaremorelikelytobedistributednearurbanareas,andthe
changedenvironmentwithahighertemperatureandCO2concentrationandbeerhu-
manmanagementandprotection(e.g.,artificialirrigation)hadapositiveinfluenceonfor-
estcarbonuptake.Ontheotherhand,forestlandscapeswithdifferentforestfragmenta-
tionlevels(exceptPAT)wereevenlydistributedinZhejiang(Figure1),whichledtopar-
allelforestagestructuresexistingindifferentfragmentationlevels;thisresultsuggests
thattheimpactofforestfragmentationonthechangerateofvegetationproductivityis
verylimited.
4.2.ImpactsofForestFragmentationonControlPathwaysofCarbonUptake
Sincevegetationcarbonuptakeprocessescanbedividedintophenologicalandphys-
iologicalfactors,thecontrolpathwayofGPPiseitherphenological,physiological,orboth.
OurresultsshowedthatCUP×GPPmaxwasthemostwidespreadcontrolvariableofGPP,
atmostlymorethan60%,during2000–2018forallfragmentationlevels(Figure3),which
isconsistentwithapreviousstudythatreportedthatGPPwasjointlycontrolledbyphe-
nologyandphysiologyprocesses[24].However,theindividualCUPorGPPmaxstillcon-
trolledtheGPPinsomeforestlandscapesinZhejiang.Forexample,forestcarbonuptake
wasmainlycontrolledbytheCUPinmountainareas,whilethecontrollingroleofGPPmax
onGPPusuallyoccurredinthecentralandcoastalareaswithaflatterrain(Figure3).Our
studyindicatesthatthedominantpathwaysofforestcarbonuptakearecloselyrelatedto
topographicconditions.
Moreover,ourresultsshowedthatthepercentageareaoftheGPPmaxcontrolvariable
graduallyincreaseswithfragmentationlevel;themeanGPPmaxofPATduring2000–2018
RemoteSens.2024,16,239311of14
wasfoundtobehigherthanthatofotherfragmentationlevels(Table3).Thismeansthat
themaximumphotosyntheticcapacityofvegetation,mainlyreflectedbyGPPmax,isim-
portantincontrollingcarbonuptakeinhighlyfragmentedforestlandscapes.Thefactors
oflessinter-individualcompetition[46,47]andmorecarbondioxidefertilization[48,49]
arelikelypossibledriversoftheobservedimportanceofGPPmaxathigherfragmentation
levels.
Inaddition,thecontrolvariablesoftheCUPandCUP×GPPmaxofINTforestland-
scapeswereobviouslyhigherthanthoseofotherfragmentationlevels.Thissuggeststhat
phenology-basedprocessesbenefitfromlessfragmentedforestlandscapesandaremost
likelytobeaributedtothewarmingconditionscausedbytheedgeeffect[50]inhighly
fragmentedforests.
4.3.InsightsonLocalForestManagementunderForestFragmentation
Historically,forestsinZhejiang,China,haveexperiencedsevereloggingdisturb-
ances,causingagrimstatusofforestfragmentation.However,theresultsofthisstudy
maystillprovideusefulinsightsintolocalforestmanagement.Firstly,ahighercarbon
uptakecapacityperunitareaexistsinhighlyfragmentedforestlandscapes,remindingus
toexplorethepotentialofforestcarbonuptakeinthecontextofthewidespreadexistence
offorestfragmentation.Inparticular,asoneofthemostdevelopedregionsinChina,
Zhejiangissubjecttointensehumanactivities,suchasrapidurbanizationandeconomic
development,whichhavesignificantlyaffectedthedistributionpaernofnaturalecosys-
tems,suchasforests.Thepositiverelationshipbetweenforestfragmentationandcarbon
sequestrationprovidesanopportunityforZhejiangtobalancedevelopmentandecologi-
calprotection.Secondly,althoughtheforestcarbonuptakemaybeenhancedunderfrag-
mentation,thereisstillagreatneedtoprotecttheforestsduetothefacthatanylossof
forestcoveragewillleadtoafurtherlossofcarbonstock.Finally,thisrelationshipisalso
basedontheimplicationthattheincreaseinthemaximumphotosyntheticcapacitymay
inducetheenhancementofcarbonuptakeinhighlyfragmentedforestlandscapes.There-
fore,itisnecessarytoexplorethemechanismbywhichforestfragmentationchangesthe
vegetationphotosyntheticprocess,whichwillplayanimportantroleinformulatingforest
managementmeasures.Inaddition,althoughsubtropicalforestsinZhejiangProvince
wereselectedasbeingrepresentativeinthisstudy,consideringtheuniversalityofre-
searchapproachesandthewidespreadavailabilityofresearchobjects,ourresultshave
greatpotentialvaluesforunderstandinghowthecarbonuptakeofsubtropicalforestsre-
spondstoforestfragmentationinothersubtropicalregions.
4.4.ResearchUncertaintiesandProspects
Inthisstudy,weexploredhowforestcarbonuptakechangeswithdifferentfragmen-
tationlevelsandwhatthemainpathwayisinwhichfragmentationaffectedforestcarbon
uptakethroughout2000–2018inforestlandscapesinZhejiang,China.Someuncertainties
mayexistinourstudy.Firstly,althoughforestprotectionandsuspendedloggingdisturb-
anceshavebeenimplementedsince2000inthestudyarea[51,52],weassumedthatthe
forestcoveragewaskeptstablewhentheforestlossproportionwaslessthan5%inthe
periodbetween2000and2018;thismaystillinducesomeuncertainties,sincethesucces-
sionofforeststhemselvesmayhavecausedsomechangesinforestcover[53].Secondly,
thedatasetsofforestcoverandGPPwereproducedonalargescale(thewholeofChina)
andhaveahighaccuracy.However,weonlyusedsomepartsofthedatasets,inducing
uncertaintiesintheaccuracyofthedatasets.Lastly,althoughtheforestfragmentation
levelclassificationandGPPestimationwereobtainedbasedoncertifiedmethods
[20,29,54],theaccuracyofthesedatamayalsohavesomeimpactsonthestudyresults.
RemoteSens.2024,16,239312of14
5.Conclusions
Comparedtopreviousstudies,thisstudyhasrevealedthevariationincarbonuptake
withforestfragmentationlevelsatbothlandscapeandregionalscales,andtheforestland-
scapepaernwasfoundtohavecertainimpactsonforestproductivity.Inaddition,the
controlpathwaysofcarbonuptakeunderforestfragmentationwerealsoidentified.These
findingshaveenhancedtheunderstandingofvegetationcarbonuptakeprocessesinthe
contextofwidespreadforestfragmentation.
Ourresultsdemonstratethatcarbonuptakecapacityperunitareawasenhancedin
highlyfragmentedforestlandscapes.However,therewasalmostnosignificantdifference
intheannualGPPtrendthroughout2000–2018amongforestlandscapeswithdifferent
fragmentationlevels.Additionally,althoughforestcarbonuptakewasjointlycontrolled
byphenology-andphysiology-basedprocesses,themaximumphotosyntheticcapacityof
vegetation—thephysiology-basedprocess—playedanimportantroleincontrollingcar-
bonuptake,particularlyinhighlyfragmentedforestlandscapes.Hence,theresultsand
findingsfromthisstudycallforabeeranddeeperunderstandingofthepotentialof
forestcarbonuptakeinconnectiontofragmentation.Thereisanemergingneedtopay
seriousaentiontosustainableforestconservationandprotectioninthecontextofthe
widespreadexistenceofforestfragmentation.Moreover,inordertoformulateappropri-
ateforestmanagementandplanningactivities,itisnecessarytoexplorethemechanism
inwhichforestfragmentationchangesthevegetationphotosyntheticprocessinconnec-
tiontocarbonuptakeandrelease.
Aut horContributions:Conceptualization,J.J.,J.M.andC.W.;methodology,J.J.,P.H.,andJ.M.;for-
malanalysis,J.J.,Y.C.,P.H.,B.J.andC.W.;datacuration,J.J.,Y.C.,P.H.,B.J.andC.W.;writing—
originaldraftpreparation,J.J.;writing—reviewandediting,Y.C.,P.H.,J.M.,L.Y.,B.J.,X.X.andC.W.;
visualization,J.J.andY.C.;fundingacquisition,C.W.andJ.M.Allauthorshavereadandagreedto
thepublishedversionofthemanuscript.
Funding:.ThisresearchwassupportedbytheresearchinstitutesupportprojectofZhejiangProv-
ince(2024F1065-1),“Pioneer”and“LeadingGoose”R&DProgramofZhejiang(2024C03227),Nat-
uralScienceFoundationofChina(32271659),andscientificresearchprogramofShanghaiscience
andtechnologycommission(23DZ1204504).
DataAvailabilityStatement:.ThetimeseriesGPPdatasetwasdownloadedfromhps://doi.pan-
gaea.de/10.1594/PANGAEA.879558?format=html#download.Theforestcovermapisavailableon
requestfromDr.YuanweiQinfromtheUniversityofOklahoma.TheCUP,GPPmax,andforestfrag-
mentationcategoriesmapareavailableonrequestfromthecorrespondingauthors.
Acknowledgments:WethanksYuanweiQinfromtheUniversityofOklahomaforhisshareofthe
forestcovermapofChina.WealsothankJijiaLiufromFudanUniversityforhisparticipateinthe
discussionofthisstudy.
ConflictsofInterest:Theauthorsdeclarenoconflictsofinterest.
References
1. Pan,Y.;Birdsey,R.A.;Fang,J.;Houghton,R.;Kauppi,P.E.;Kurz,W.A.;Phillips,O.L.;Shvidenko,A.;Lewis,S.L.;Canadell,J.G.;
etal.ALargeandPersistentCarbonSinkintheWorld’sForests.Science2011,333,988–993.hps://doi.org/10.1126/sci-
ence.1201609.
2. Canadell,J.G.;Raupach,M.R.Managingforestsforclimatechangemitigation.Science2008,320,1456–1457.
hps://doi.org/10.1126/science.1155458.
3. Fang,J.Y.;Guo,Z.D.;Hu,H.F.;Kato,T.;Muraoka,H.;Son,Y.ForestbiomasscarbonsinksinEastAsia,withspecialreference
totherelativecontributionsofforestexpansionandforestgrowth.Glob.ChangeBiol.2014,20,2019–2030.
hps://doi.org/10.1111/gcb.12512.
4. Kumar,P.;Pirjola,L.;Keel,M.;Harrison,R.M.Nanoparticleemissionsfrom11non-vehicleexhaustsources—Areview.Atmos.
Environ.2013,67,252–277.hps://doi.org/10.1016/j.atmosenv.2012.11.011.
5. Hansen,M.C.;Stehman,S.V.;Potapov,P.V.Quantificationofglobalgrossforestcoverloss.Proc.Natl.Acad.Sci.USA2010,107,
8650–8655.hps://doi.org/10.1073/pnas.0912668107.
6. Arora,V.K.;Boer,G.J.Uncertaintiesinthe20thcenturycarbonbudgetassociatedwithlandusechange.Glob.ChangeBiol.2010,
16,3327–3348.hps://doi.org/10.1111/j.1365-2486.2010.02202.x.
RemoteSens.2024,16,239313of14
7. Curtis,P.G.;Slay,C.M.;Harris,N.L.;Tyukavina,A.;Hansen,M.C.Classifyingdriversofglobalforestloss.Science2018,361,
1108–1111.hps://doi.org/10.1126/science.aau3445.
8. Ma,J.;Li,J.;Wu,W.;Liu,J.Globalforestfragmentationchangefrom2000to2020.Nat.Commun.2023,14,3752.
9. Haddad,N.M.;Brudvig,L.A.;Clobert,J.;Davies,K.F.;Gonzalez,A.;Holt,R.D.;Lovejoy,T.E.;Sexton,J.O.;Aus tin ,M.P.;Collins,
C.D.;etal.HabitatfragmentationanditslastingimpactonEarth’secosystems.Sci.Adv.2015,1,e1500052.
hps://doi.org/10.1126/sciadv.1500052.
10. Collins,C.D.;Banks-Leite,C.;Brudvig,L.A.;Foster,B.L.;Cook,W.M.;Damschen,E.I.;Andrade,A.;Aust in,M.;Camargo,J.L.;
Driscoll,D.A.;etal.Fragmentationaffectsplantcommunitycompositionovertime.Ecography2017,40,119–130.
hps://doi.org/10.1111/ecog.02607.
11. Bello,C.;Galei,M.;Pizo,M.A.;Magnago,L.F.S.;Rocha,M.F.;Lima,R.A.F.;Peres,C.A.;Ovaskainen,O.;Jordano,P.De-
faunationaffectscarbonstorageintropicalforests.Sci.Adv.2015,1,e1501105.hps://doi.org/10.1126/sciadv.1501105.
12. Ma,L.;Shen,C.;Lou,D.;Fu,S.;Guan,D.Ecosystemcarbonstorageinforestfragmentsofdifferingpatchsize.Sci.Rep.2017,7,
13173.hps://doi.org/10.1038/s41598-017-13598-4.
13. Smith,I.A.;Hutyra,L.R.;Reinmann,A.B.;Marrs,J.K.;Thompson,J.R.Piecingtogetherthefragments:Elucidatingedgeeffects
onforestcarbondynamics.Front.Ecol.Environ.2018,16,213–221.hps://doi.org/10.1002/fee.1793.
14. Chaplin-Kramer,R.;Ramler,I.;Sharp,R.;Haddad,N.M.;Gerber,J.S.;Wes t,P.C.;Mandle,L.;Engstrom,P.;Baccini,A.;Sim,S.;
etal.Degradationincarbonstocksneartropicalforestedges.Nat.Commun.2015,6,10158.hps://doi.org/10.1038/ncomms10158.
15. Ordway,E.M.;Asner,G.P.Carbondeclinesalongtropicalforestedgescorrespondtoheterogeneouseffectsoncanopystructure
andfunction.Proc.Natl.Acad.Sci.USA2020,117,7863–7870.hps://doi.org/10.1073/pnas.1914420117.
16. Reinmann,A.B.;Hutyra,L.R.Edgeeffectsenhancecarbonuptakeanditsvulnerabilitytoclimatechangeintemperatebroadleaf
forests.Proc.Natl.Acad.Sci.USA2017,114,107–112.hps://doi.org/10.1073/pnas.1612369114.
17. Morreale,L.L.;Thompson,J.R.;Tang ,X.;Reinmann,A.B.;Hutyra,L.R.Elevatedgrowthandbiomassalongtemperateforest
edges.Nat.Commun.2021,12,7181.hps://doi.org/10.1038/s41467-021-27373-7.
18. Remy,E.;Wuyts,K.;Boeckx,P.;Ginzburg,S.;Gundersen,P.;Demey,A.;Van DenBulcke,J.;Van Acker,J.;Verheyen,K.Strong
gradientsinnitrogenandcarbonstocksattemperateforestedges.For.Ecol.Manag.2016,376,45–58.
hps://doi.org/10.1016/j.foreco.2016.05.040.
19. Fahrig,L.Effectsofhabitatfragmentationonbiodiversity.Annu.Rev.Ecol.Evol.Syst.2003,34,487–515.
hps://doi.org/10.1146/annurev.ecolsys.34.011802.132419.
20. Riiers,K.;Wickham,J.;O’Neill,R.;Jones,B.;Smith,E.Global-scalepaernsofforestfragmentation.Conserv.Ecol.2000,4,30.
21. Riiers,K.H.;Wickham,J.D.;O’Neill,R.V.;Jones,K.B.;Smith,E.R.;Coulston,J.W.;Wade,T.G.;Smith,J.H.Fragmentationof
continentalUnitedStatesforests.Ecosystems2002,5,815–822.hps://doi.org/10.1007/s10021002-0209-2.
22. Hermosilla,T.;Wulder,M.A.;White,J.C.;Coops,N.C.;Pickell,P.D.;Bolton,D.K.Impactoftimeoninterpretationsofforest
fragmentation:Three-decadesoffragmentationdynamicsoverCanada.RemoteSens.Environ.2019,222,65–77.
hps://doi.org/10.1016/j.rse.2018.12.027.
23. Monteith,J.L.Solar-radiationandproductivityintropicalecosystems.J.Appl.Ecol.1972,9,747–766.
hps://doi.org/10.2307/2401901.
24. Xia,J.;Niu,S.;Ciais,P.;Janssens,I.A.;Chen,J.;Ammann,C.;Arain,A.;Blanken,P.D.;Cescai,A.;Bonal,D.;etal.Jointcontrol
ofterrestrialgrossprimaryproductivitybyplantphenologyandphysiology.Proc.Natl.Acad.Sci.USA2015,112,2788–2793.
hps://doi.org/10.1073/pnas.1413090112.
25. Zhou,S.;Zhang,Y.;Ciais,P.;Xiao,X.;Luo,Y.;Caylor,K.K.;Huang,Y.;Wang, G.Dominantroleofplantphysiologyintrend
andvariabilityofgrossprimaryproductivityinNorthAmerica.Sci.Rep.2017,7,41366.hps://doi.org/10.1038/srep41366.
26. Keenan,T.F.;Gray,J.;Friedl,M.A.;Toomey,M.;Bohrer,G.;Hollinger,D.Y.;Munger,J.W.;O’Keefe,J.;Schmid,H.P.;SueWing,
I.;etal.Netcarbonuptakehasincreasedthroughwarming-inducedchangesintemperateforestphenology.Nat.Clim.Change
2014,4,598–604.hps://doi.org/10.1038/nclimate2253.
27. Arroyo-Rodriguez,V.;Melo,F.P.L.;Martinez-Ramos,M.;Bongers,F.;Chazdon,R.L.;Meave,J.A.;Norden,N.;Santos,B.A.;
Leal,I.R.;Tabarelli,M.Multiplesuccessionalpathwaysinhuman-modifiedtropicallandscapes:Newinsightsfromforestsuc-
cession,forestfragmentationandlandscapeecologyresearch.Biol.Rev.2017,92,326–340.hps://doi.org/10.1111/brv.12231.
28. Grilli,G.;Urcelay,C.;Galeo,L.Forestfragmentsizeandnutrientavailability:Complexresponsesofmycorrhizalfungiin
native-exotichosts.PlantEcol.2012,213,155–165.hps://doi.org/10.1007/s11258-011-9966-3.
29. Hansen,M.C.;Potapov,P.V.;Moore,R.;Hancher,M.;Turuban ova,S.A.;Tyukavina,A.;Thau,D.;Stehman,S.V.;Goe,S.J.;
Loveland,T.R.;etal.High-ResolutionGlobalMapsof21st-CenturyForestCoverChange.Science2013,342,850–853.
hps://doi.org/10.1126/science.1244693.
30. Qin,Y.W.;Xiao,X.M.;Dong,J.W.;Zhang,G.L.;Shimada,M.;Liu,J.Y.;Li,C.G.;Kou,W.L.;Moore,B.ForestcovermapsofChina
in2010frommultipleapproachesanddatasources:PALSAR,Landsat,MODIS,FRA,andNFI.ISPRSJ.Photogramm.Remote
Sens.2015,109,1–16.hps://doi.org/10.1016/j.isprsjprs.2015.08.010.
31. Dong,J.;Xiao,X.;Sheldon,S.;Biradar,C.;Zhang,G.;NguyenDinh,D.;Hazarika,M.;Wikantika,K.;Takeu hci, W.;Moore,B.,
III.A50-mForestCoverMapinSoutheastAsiafromALOS/PALSARandItsApplicationonForestFragmentationAssessment.
PLoSONE2014,9,e85801.hps://doi.org/10.1371/journal.pone.0085801.
32. Li,Y.;Xiao,X.;Li,X.;Ma,J.;Chen,B.;Qin,Y.;Dong,J.;Zhao,B.Multi-scaleassessmentsofforestfragmentationinChina.
Biodiversity2017,25,372–381.(InChinese)
RemoteSens.2024,16,239314of14
33. Ma,J.;Xiao,X.M.;Zhang,Y.;Doughty,R.;Chen,B.Q.;Zhao,B.Spatial-temporalconsistencybetweengrossprimaryproductiv-
ityandsolar-inducedchlorophyllfluorescenceofvegetationinChinaduring2007-2014.Sci.TotalEnviron.2018,639,1241–1253.
34. Zhang,Y.;Xiao,X.M.;Wu,X.C.;Zhou,S.;Zhang,G.L.;Qin,Y.W.;Dong,J.W.DataDescriptor:Aglobalmoderateresolution
datasetofgrossprimaryproductionofvegetationfor2000-2016.Sci.Data2017,4,13.hps://doi.org/10.1038/sdata.2017.165.
35. Zhao,J.-J.;Liu,L.-Y.ExtractionoftemperatevegetationphenologythresholdsinNorthAmericabasedonfluxtowerobservation
data.Chin.J.Appl.Ecol.2013,24,311–318.
36. White,M.A.;Thornton,P.E.;Running,S.W.Acontinentalphenologymodelformonitoringvegetationresponsestointerannual
climaticvariability.Glob.Biogeochem.Cycles1997,11,217–234.hps://doi.org/10.1029/97gb00330.
37. Groemping,U.RelativeimportanceforlinearregressioninR:Thepackagerelaimpo.J.Stat.Softw.2006,17,1–27.
38. RDevelopmentCoreTeam.R:ALanguageandEnvironmentforStatisticalComputing;RFoundationforStatisticalComputing:
Vienna,Austr ia, 2013.
39. Berenguer,E.;Ferreira,J.;Gardner,T.A.;Aragao,L.;DeCamargo,P.B.;Cerri,C.E.;Durigan,M.;DeOliveira,R.C.;Vieira,I.C.G.;
Barlow,J.Alarge-scalefieldassessmentofcarbonstocksinhuman-modifiedtropicalforests.Glob.ChangeBiol.2014,20,3713–
3726.hps://doi.org/10.1111/gcb.12627.
40. Silverio,D.V.;Brando,P.M.;Bustamante,M.M.C.;Pu,F.E.;Marra,D.M.;Levick,S.R.;Trumbore,S.E.Fire,fragmentation,and
windstorms:Arecipefortropicalforestdegradation.J.Ecol.2019,107,656–667.hps://doi.org/10.1111/1365-2745.13076.
41. Zeng,Z.Z.;Gower,D.B.;Wood,E.F.AcceleratingforestlossinSoutheastAsianMassifinthe21stcentury:AcasestudyinNan
Province,Thailand.Glob.ChangeBiol.2018,24,4682–4695.hps://doi.org/10.1111/gcb.14366.
42. Barber,C.P.;Cochrane,M.A.;Souza,C.M.;Laurance,W.F.Roads,deforestation,andthemitigatingeffectofprotectedareasin
theAmazon.Biol.Conserv.2014,177,203–209.hps://doi.org/10.1016/j.biocon.2014.07.004.
43. Briber,B.M.;Hutyra,L.R.;Reinmann,A.B.;Raciti,S.M.;Dearborn,V.K .;Holden,C.E.;Dunn,A.L.TreeProductivityEnhanced
withConversionfromForesttoUrbanLandCovers.PLoSONE2015,10,e0136237.hps://doi.org/10.1371/journal.pone.0136237.
44. Martin-Benito,D.;Pederson,N.Convergenceindroughtstress,butadivergenceofclimaticdriversacrossalatitudinalgradient
inatemperatebroadleafforest.J.Biogeogr.2015,42,925–937.hps://doi.org/10.1111/jbi.12462.
45. Tan g ,G.P.;Beckage,B.;Smith,B.;Miller,P.A.EstimatingpotentialforestNPP,biomassandtheirclimaticsensitivityinNew
Englandusingadynamicecosystemmodel.Ecosphere2010,1,1–20.hps://doi.org/10.1890/es10-00087.1.
46. Campbell,M.J.;Edwards,W.;Magrach,A.;Alamgir,M.;Porolak,G.;Mohandass,D.;Laurance,W.F.Edgedisturbancedrives
lianaabundanceincreaseandalterationofliana-hosttreeinteractionsintropicalforestfragments.Ecol.Evol.2018,8,4237–4251.
hps://doi.org/10.1002/ece3.3959.
47. Fischer,R.;Bohn,F.;dePaula,M.D.;Dislich,C.;Groeneveld,J.;Gutierrez,A.G.;Kazmierczak,M.;Knapp,N.;Lehmann,S.;
Paulick,S.;etal.Lessonslearnedfromapplyingaforestgapmodeltounderstandecosystemandcarbondynamicsofcomplex
tropicalforests.Ecol.Model.2016,326,124–133.hps://doi.org/10.1016/j.ecolmodel.2015.11.018.
48. Malhi,Y.;Aragao,L.;Galbraith,D.;Huntingford,C.;Fisher,R.;Zelazowski,P.;Sitch,S.;McSweeney,C.;Meir,P.Exploringthe
likelihoodandmechanismofaclimate-change-induceddiebackoftheAmazonrainforest.Proc.Natl.Acad.Sci.USA2009,106,
20610–20615.hps://doi.org/10.1073/pnas.0804619106.
49. Qie,L.;Lewis,S.L.;Sullivan,M.J.P.;Lopez-Gonzalez,G.;Pickavance,G.C.;Sunderland,T.;Ashton,P.;Hubau,W.;AbuSalim,
K.;Aiba,S.;etal.Long-termcarbonsinkinBorneo’sforestshaltedbydroughtandvulnerabletoedgeeffects.Nat.Commun.
2017,8,1966.hps://doi.org/10.1038/s41467-017-01997-0.
50. Schmidt,M.;Jochheim,H.;Kersebaum,K.C.;Lischeid,G.;Nendel,C.Gradientsofmicroclimate,carbonandnitrogenintransi-
tionzonesoffragmentedlandscapes—Areview.Agric.For.Meteorol.2017,232,659–671.
hps://doi.org/10.1016/j.agrformet.2016.10.022.
51. Ding,L.;Lu,J.;Xu,G.;Wu,J.EffectsofecologicalprotectionanddevelopmentonlandscapepaernintheThousand-Island
Lakeregion,ZhejiangProvince.Biodivers.Sci.2004,12,473.
52. Ren,G.;Youn g ,S.S.;Wan g,L.;Wan g,W.;Long,Y.;Wu,R.;Li,J.;Zhu,J.;Yu,D.W.EffectivenessofChina’snationalforest
protectionprogramandnaturereserves.Conserv.Biol.2015,29,1368–1377.
53. Lu,B.;He,Y.H.SpeciesclassificationusingUnmannedAerialVehi cle (UAV)-acquiredhighspatialresolutionimageryina
heterogeneousgrassland.ISPRSJ.Photogramm.RemoteSens.2017,128,73–85.hps://doi.org/10.1016/j.isprsjprs.2017.03.011.
54. Xiao,X.M.;Hollinger,D.;Aber,J.;Gol,M.;Davidson,E.A.;Zhang,Q.Y.;Moore,B.Satellite-basedmodelingofgrossprimary
productioninanevergreenneedleleafforest.RemoteSens.Environ.2004,89,519–534.hps://doi.org/10.1016/j.rse.2003.11.008.
Disclaimer/Publisher’sNote:Thestatements,opinionsanddatacontainedinallpublicationsaresolelythoseoftheindividualau-
thor(s)andcontributor(s)andnotofMDPIand/ortheeditor(s).MDPIand/ortheeditor(s)disclaimresponsibilityforanyinjuryto
peopleorpropertyresultingfromanyideas,methods,instructionsorproductsreferredtointhecontent.