Content uploaded by Chaoyang Wu
Author content
All content in this area was uploaded by Chaoyang Wu on Apr 12, 2019
Content may be subject to copyright.
Glob Change Biol. 2019;1–15. wileyonlinelibrary.com/journal/gcb
|
1
© 2019 John Wiley & Sons Ltd
Received:4Februa ry2019
|
Revised:4February2019
|
Accepted:16March2019
DOI:10.1111/gcb.14627
PRIMARY RESE ARCH ARTICLE
Satellite detection of cumulative and lagged effects of drought
on autumn leaf senescence over the Northern Hemisphere
Jie Peng1,2 | Chaoyang Wu1,2 | Xiaoyang Zhang3 | Xiaoyue Wang1 |
Alemu Gonsamo4
1TheKeyLaborator yofLandSurface
Patter nandSimulation,Instituteof
GeographicalScien cesandNatural
ResourcesResearch,ChineseAc ademyof
Science s,Beijing,China
2UniversityoftheChineseAcad emyof
Science s,Beijing,China
3Depar tmentofGeography,Geosp atial
Science sCenterofExcellence(GSC E),South
DakotaStateUniversity,Brooking s,South
Dakota
4Depar tmentofGeographyand
Plannin g,Unive rsit yofToronto,Toronto,
ON,Canada
Correspondence
Chaoyan gWu,TheKeyL aboratoryofL and
SurfacePatternandS imulation,Ins titute
ofGeogr aphic alSciencesandNatural
ResourcesResearch,ChineseAc ademyof
Science s,Beijing,China.
Email:wucy@igsnrr.ac.cn
Funding information
ChineseAcade myofScien ces,Gr ant/Award
Number :XDA1904 0103;NationalNatural
ScienceFoundationofChin a,Grant/Award
Number :41871225;KeyResea rchProgram
ofFrontierS cience s,Gra nt/AwardNumber:
QYZDB-SSW-DQC011
Abstract
Climatechangehassubstantialinfluencesonautumnleafsenescence,thatis,theend
ofthegrowingseason(EOS).Relativetotheimpactsoftemperatureandprecipitation
onEOS, the influence of drought is not wellunderstood, especiallyconsideringthat
ther ea reapp arentcu mulativea ndlag gedef fec t sofdrough to nplantg rowth.He re,we
inve sti gat edt hec umu lat iveandla g gedef f ectsofd rou ght(inte rms oftheStan dar dized
Precipitation–EvapotranspirationIndex,SPEI)onEOSderivedfromthenormalizeddif-
ferencevegetationindex(NDVI3g)dataovertheNorthernHemisphereextra-tropical
ecosystems(>30°N)during1982–2015.Thecumulativeeffectwasdeterminedbythe
number of antecedent monthsatwhich SPEI showedthe maximum correlation with
EOS(i.e.,Rmax-cml)while the lageffectwasdetermined by a monthduring whichthe
maximumcorrelationbetween1-monthSPEIandEOSoccurred(i.e.,Rmax-lag).Wefo u nd
cumulative ef fect of drought on EOS for 27.2% and lagge d effect for 46 .2% of the
vegetatedland area.For thedominanttimescaleswheretheRmax-cml and Rmax-lag oc-
curred, we observed 1–4 accumulated months for the cumulative effect and 2–6
laggedmonthsforthelaggedeffect.Atthebiomelevel,droughthadstrongerimpacts
onEOSingrasslands,savannas,andshrubsthaninforests,whichmayberelatedtothe
differentrootfunctionaltraitsamongvegetationtypes.Consideringhydrologicalcon-
ditions,themeanvaluesofbothRmax-cml and Rmax-lagdecreasedalongthegradientsof
annual SPEIand its slope,suggestingstrongercumulative and laggedeffectsindrier
regionsaswellasinareaswithdecreasingwateravailabilit y.Furthermore,theaverage
accumulatedandlaggedmonthstendedtodeclinealongtheannualSPEIgradientbut
increasewithincreasingannualSPEI.Ourresultsrevealedthatdroughthasstrongcu-
mulativeandlaggedeffectsonautumnphenology,andconsideringtheseeffectscould
providevaluableinformationonthevegetationresponsetoachangingclimate.
KEY WORDS
autumnphenolog y,cumulativeeffect,drought,laggedeffect
1 | INTRODUCTION
Global climate change has influenced vegetation phenology con-
siderably over the past decades (Chen et al., 2014; Dannenberg,
Song, Hwa ng, & Wise, 2015; Go nsamo, Chen, & O oi, 2018; Piao,
Fang, Zhou, Philippe, & Zhu, 2006; Richardson etal., 2013; Zhang
et al., 2018). While numerous research focused upon the spring
phenology(i.e.,thestart ofgrowingseason,SOS)(Piaoetal.,2011;
2
|
PENG Et al.
Seyednasrollah,Swenson,Domec,&Clark,2018;Shen,Tang,Chen,
Zhu,&Zheng,2011;Whiteetal.,2009),influencesofclimatechange
onautumnleafsenescence(i.e.,theendofthegrowingseason,EOS)
received less attention(Liu,Fu,Zhu,et al., 2016b;Wu,Chen,etal.,
2013;Zhuetal.,2017).Itisthereforeimperativetocomprehendthe
effectofclimaticfac torsonEOS,giventhatthereareincreasingevi-
de nce ssh owi ngt h ei mpor t a nt f unc t ion of aut u mnp hen olo g yin con -
trollingecosystemproductivity,water,andcarboncycles(Balzarolo
etal.,2016;Ganguly,Friedl,Tan,Zhang, &Verma,2010;Piaoetal.,
2012;Wu,Chen,etal.,2013;Wu,Gough,Chen,&Gonsamo,2013).
Previousstudiesfoundthatchangesinautumnphenologycould
be attributed to climatic variable suchas solarradiation, tempera-
ture,andprecipitation(Keskitalo,Gustaf,Per,&Stefan,2005;Kong,
Qiang, Singh,&Shi,2016;Liu,Fu,Zeng,etal.,2016a;Liu, Fu,Zhu,
et al., 2016b; Pia o et al., 200 6; Zu et al., 2018). In temp erate for-
ests,delaysinEOScouldbelinkedtoincreasedtemperaturesinau-
tumn (Del pierre et al., 20 09; Estre lla & Menzel, 20 06; Ge, Wang,
Rutishauser,& Dai,2015; Jeong,Ho, Gim,&Brown, 2011; Liu,Fu,
Zeng, et al., 2016a; Yang, Guan, Shen, Liang, & Jiang, 2015), while
earlier leaf senescence may also relate to the warming-induced
drought(Estrella&Menzel,2006;Liet al.,2018;Liu, Fu,Zhu,etal.,
2016b;Zhu, Zheng,Jiang,&Zhang, 2018).Anincreasein precipita-
tioncoulddelay EOS(Liu,Fu,Zeng,etal.,2016a;Richardson et al.,
2013), but may also a dvance autu mn foliar event s (Cong, Shen , &
Piao, 2017;Xie, Wang, Wilson, &Silander,2018; Zhuetal., 2017),
sug ge st ingco mp lexre sp on se sofautumnp henologic aleve nt stoen -
vironmentalcues.
Drought is one of the more complic ated climatic phenomena
influencingvegetation(Chaves, Maroco, &Pereira, 2003; Vicente-
Serra no et al., 2012). It ca n be defined as p eriods of lower wat er
availability between precipitation and evapotranspiration, relative
to its loc al normal water b alance (Dai, 2011; D'Ora ngeville et al.,
2018;Vicente-Serrano et al., 2013; Wilhite&Glantz, 1985). Under
the background of global warming, the intensity and frequency of
droughteventsenhancedoverthelastdecades,resultinginincreas-
ingly seve re impacts on te rrestrial eco systems (Ciais et a l., 2005;
Zhang,Kong,Singh,&Shi,2017),includingconstraining vegetation
growth (D 'Orangev ille et al., 2018; H uang, Wang, Kee nan, & Piao,
2018; Zheng et al ., 2018), causing veg etation mor tality (F ensham,
Fairfax,&Ward, 2009;Huang,A sner, Barger,Neff,& Floyd,2010;
Martínez-Vilalta & Lloret, 2016), and even inducing wildfire and
biotic disturbance (Huang et al., 2017; Wendler, Conner, Moore,
Shulski,& Stuefer,2011).Recently,several studieshavebeen done
toinvestigate theeffec tsofdroughtonEOS. Forexample,drought
in summer a nd autumn adva nced EOS of grasslan ds in China and
Canad a (Cui, Martz , & Guo, 2017; Kang, Wang, & Liu , 2018; Tao,
Yokozawa,Zhang,Hayashi,&Ishigooka,2008).Conversely,itledto
laterleafsenescenceformostdeciduoustreesinnortheasternUSA
(Xie, Wan g, & Silander, 2015; Xie e t al., 2018). Due to th e limited
research, it is difficult to distinguish whether the inconsistency is
caused by d istinc t environmen t of study are as, diverse v egetation
communit ies, or other reaso ns. Thus, investi gation of drought-in-
duced aut umn phenologic al changes at a regi onal, continent al, or
globalscalewouldbeparticularlyimportanttoenhanceourunder-
standingofautumnphenologyanditsresponsetodrought.
Extensive studies have recognized that the cumulative effects
ofclimaticfactors(e.g.,temperatureandprecipitation)beforephe-
nologicalevenshavestrongimpactsonvegetationgrowth(Liu,Fu,
Zeng,etal.,2016a;Pastor-Guzman,Dash,& Atkinson,2018;Shen
et al., 2011). Recent ly,dr ought was uncove red to have legac y ef-
fectsonvegetationdynamics(D'Orangevilleetal.,2018;Huanget
al., 2018; Vicente-Serrano et al., 2013;Zhang, Kong,et al., 2017).
Thecumulativeandtime-lageffectofmajorclimaticdrivingforces
onphenologyshouldbepaidmoreat tentionforabetterprediction
andevaluationof vegetation response (Andersonet al.,2010; Wu
et al., 2015). G iven all thes e findings , we reasonabl y hypothesize
that drought may play a crucial rolein regulating EOS through its
cumulativeandlagged effects.Therefore,weaimtoquantitatively
investig ate the cumula tive and lagge d effect s of drought on EOS
usingthesatellitederivedNormalizedDifferenceVegetationIndex
(NDVI)dataset(1982–2015)fromtheGlobalInventoryModelingand
Mappin gStudi est hirdge nerat ion(GIMMS3g )andtheStan dardized
Precipitation–EvapotranspirationIndex(SPEI)timeseriesoverthe
NorthernHemisphereextra-tropicalecosystems(>30°N).Ourmain
objectiveswere(a)toanalyzethecumulativeandlaggedeffectsof
recentdroug hto nEOS;(b)tounder standhowt heseeff ec tschange
fordifferentbiometypesandwetnessconditions.
2 | MATERIALS AND METHODS
2.1 | Study area
Our study was concentrated on the Northern Hemisphere extra-
tropical terrestrial ecosystems (>30°N, Figure 1) due to inherent
seasonality of vegetation in tropical regions. Considering that the
seasonalvariation of cropland ismostlycontrolledby humanactivi-
ties, we excl uded the pi xels covered by cro pland using t he classifi-
cationoftheInternationalGeosphere-BiosphereProgramme(IGBP)
intheModerateResolutionImagingSpectroradiometer(MODIS)
Land Cover Climate Modeling Grid Product (MCD12C1).To reduce
thenoisesfromnon-vegetationareas,NDVIpixelswithlowerannual
mean valu e (<0.15), which were usu ally identified as bare land or
sparse ve getation, were al so removed (Jeong e t al., 2011; Liu, Fu,
Zeng,etal.,2016a).
2.2 | Datasets
2.2.1 | Satellite NDVI dataset
Weusedthe15-dayGIMMS3gNDVItimeseriesrangingfrom1982
to 2015 to extract EOS (ht tp://ecocast.arc.nasa.gov/data/pub/
gimms/3g.v1). It has a spatial resolution of approximately8km. As
the longest NDVI record, the GIMMS3g NDVI data have been ex-
tensively a pplied for vege tation dyn amics monito ring (Chen et al .,
2014;Kongetal.,2016;Panet al., 2018;Schut, Ivits,Conijn,Brink,
&Fensholt,2015;Wuetal.,2015;Zhang,Kong,etal.,2017;Zhang,
|
3
PENG Et al .
Song,Band, Sun,&Li,2017).Numerousstudiesonvegetationphe-
nology were also conducted based on this dataset (Buitenwerf,
Rose,&Higgins, 2015;Garonna,De,& Schaepman, 2016;Jeong et
al.,2011;Liu,Fu,Zeng,etal., 2016a;Liu, Fu, Zhu, et al.,2016b). In
order to matc h datasets a t different s patial resolu tions, esp ecially
the GIMMS3g NDVI (~8km) and SPEI(0.5°) (see Section 2.2.2),all
thedata were interpolated into a spatialresolutionof 8km usinga
bilinearinterpolationalgorithm.
2.2.2 | Multi‐scalar SPEI dataset
To quantify drought duration and severity, we acquired the
monthlySPEIat1-to12-monthtimescales(1982–2015)fromthe
SPEIbasev.2.5atConsejoSuperiordeInvestigacionesCientíficas
(CSIC) (http://spei.csic.es/database.html). The dataset contains
monthly SPEI at1-to48-month time scales,covering aperiod of
1901–2015witha0.5°spatialresolution.ForthevalueofSPEIat
each pixel , it was compute d by the differe nce betwee n precipi-
tation and potential evapotr anspiration (PET) with t he Climatic
Research Unit Time series of the University of E ast Anglia, and
PET was calculated based on the FAO-56 Penman–Monteith
equatio n (Allen, Pereir a, Raes, & Smith , 1998) . To comp are the
degree of water availability spatially and temporally, the final
value of SPEI w as a standardi zed variable foll owing a log-logis-
tic distribution, representing deviations from the climatic water
balance (Vicenteserrano, Beguería, & Lópezmoreno, 2010). The
larger po sitive values indi cate excess water af ter satisfyi ng the
concurrent waterdemand, while lower negative valuesindicate
more sever e water defici ency. For the mult i-scalar tr ait of SPEI,
SPEI at n time s cale (i.e., n-month SPEI) denotesthe cumulative
climatic w ater balance for th e previous n months. For example,
the1-monthSPEIprovidest heclimaticwaterconditionatthecur-
rent month , while the 12-mont h SPEI (e.g., at De cember) repr e-
sents the accumulatedclimatic waterbalance for the wholeyear
(e.g., from January to December). Accordingly,this characteristic
ofSPEImakesitpossibletoidentifydifferentdroughttypes(e.g.,
sho rt-,me dium-,andl ong-te rmdrought )andv ar iousi nfluenceson
vegetation (Beguería, Vicenteserrano, & Angulomartínez, 2010;
Martin-Benito, Beeckman, & Cañellas, 2013; Vicenteserrano et
al.,2010; Vicente-Serranoetal.,2013;Zhang, Kong,etal.,2017).
Here, we used the 1-to 12-month SPEI dat asetsfor 1982–2015
assumingthatdroughtmayimpactthewholelifecycleofplantin
thefollowing12months(Gordo&Sanz,2010;Kangetal.,2018).
2.2.3 | Biome data
To investigate the influences of drought on EOS at different bi-
omes,weemployedthebiometypesfromtheTerrestrialecoregions
ofthe world(TEOW),which was developedby Olson etal. (2001)
(https://www.worldwildlife.org/publications/terrestrial-ecoregions-
of-the-world). The T EOW is a biogeograp hic regional ization base d
on terrestrial biodiversity. The biogeographic units within TEOW
are recognized as ecoregions, which were then classified into 14
different biome types, including forests, shrublands, grasslands,
and dese rts. In ord er to explore the c haracteri stics of cumul ative
andlaggedef fectsat differentbiomesintheNorthern Hemisphere
extra-tropicalecosystems, weselected 8of14biomet ypesby ex-
cludingbiomes thatcorrespondtotropicaland subtropical regions
(Table1).
FIGURE 1 Thespatialdistributionofbiometypeswithinourstudyareaobtainedfromtheterrestrialecoregionsoftheworld(TEOW)
data(Olsonetal.,20 01)
TABLE 1 Theselectedeightbiomet ypesfromterrestrial
ecoregionsoftheworlddataset
Value Biome typ e Abbreviation
4Temperatebroadleafandmixed
forests
TBMF
5Temperateconiferousforest s TCF
6Borealforests/taiga BF
8 Temperategrasslands,savannas,
andshrublands
TGSS
10 Montanegrasslandsand
shrublands
MGS
11 Tundra TUN
12 Mediterraneanforests,
woodlands,andscrub
MFWS
13 Deser tsandxericshrublands DXS
4
|
PENG Et al.
2.3 | Determination of EOS from NDVI
As the or iginal GIMM S3g NDVI value s were usually d epressed b y
unstableatmosphericstates,suchasclouds,we adoptedthemodi-
fied Savi tzky-G olay filter to l essen the n oises in NDVI tim e series
beforehand(Chenetal.,2004).Thesmoothingwindowanditeration
timeweresetas4 and 20,respectively.Athighlatitudes(or eleva-
tions),snowcoverisimportantforregionalclimateandarrivesearly
inautumnandpotentiallymasking evergreenvegetation.However,
wesuggestedthatusingaSavitzk y–Golayfiltercouldsolvethenoise
from a "sudd en" change in the tim e series of NDVI due to s now.
Then, to reduce errorsfrom different approaches, two individual
phenologyex trac tionmethodswereemployedtoextractEOSfrom
GIMMS3 g NDVI data (Figu re 2). The firs t method is the m idpoint
method,atwhich EOSis determined asthedaywhenNDVIr atio de-
creases to0.5 in autumn (White, Thornton, & Running, 1997), and
NDVIrat ioiscalculatedby:
where NDVIrepresents the linearly interpolated dailyNDVI after
smoothing,NDVImaxrepresentstheseasonalmaximumNDVIvalue,
andNDVIministheseasonalminimumNDVIvalue.
Thesecondmethodisthedoublelogisticmethod(Elmore,Guinn,
Minsley,& Richardson,2012;Zhangetal.,2003), and it can be de-
scribedas:
whereNDVI(t)is t h efit te d N DVI a tday t.Theannualvariationcurve
of15-dayNDVIwasdividedintotwosec tionsbytheNDVImax:th e
ascendingpart(i.e.,vegetationrecoveryfromnongrowingseason)
andthe descending part(i.e., aprocessof plantsenescencefrom
peak).Forthelatterpart,thesmoothedNDVItimeserieswasfitted
to NDVI (t)usingthedoublelogisticfunction.Theoptimalcoeffi-
cientsofN DVI(t)weredeterminedwhenR2betweenthesmoothed
andsimulatedNDVIreached 0.95or theiterationwaslargerthan
2,500times(Wuetal.,2017).Thereafter,thedateofEOSwasex-
tract ed at the loca l minimum in the f irst deri vative of the fi tted
NDVI(t). Finally,weusedtheaverage valueof two EOSextraction
resultsasthefinalEOSappliedinourstudytoremoveuncertainty
causedbysinglecalculationalgorithm(Wuetal.,2017).
2.4 | Determination of the cumulative and lagged
effects of drought on EOS
ThecumulativeeffectassessmentofdroughtonEOSwasconducted
todetermine on what time scaleof SPEI that EOSyieldedthelarg-
est signi ficant correl ation with SPEI. For e xample, if the n-month
SPEI(ncouldbeanymonthfrom1to12)showedthehighestcor-
relation withEOS,thecumulativeeffectofdrought onEOS forthis
pixelwouldbedeterminedasnmonths,indicatingthattheclimatic
waterbalance duringthecontinuousnmonthsbeforeEOS(includ-
ingthe EOS occurring month)isthe mostimportant cuetotrigger
EOS.Thespecificstepsfordeterminingthecumulativeeffectwere
representedasfollows:
First, at each pixel, we determined the month of EOS (i.e.,
EOSm) for each year(1982–2015)and then extracted 12monthly
SPEIvaluesatthe EOSmmonthfrom1-to12-monthSPEIateach
year.Second,weexploredtheresponseofEOSto1-to12-month
SPEIateach pixelthrough thePearson'scorrelation (significance
levelwassettop<0.05),andallsignificantcorrelationcoefficients
betweenEOSand SPEI wereconsideredas candidates for deter-
miningthecumulativeeffectatthispixel.Third,thecumulativeef-
fectofdroughtonEOSwasdeterminedasnaccumulatedmonths
(i.e.,accumulatedtimescales)whentheabsolutemaximumsignifi-
cantcorrelationcoefficient(i.e.,|Rmax-cml|atp<0.05)occurredbe-
tweenEOSandthen-monthSPEI,andthestrengthofcumulative
effec t was identif ied as Rmax-cml(Rmax-cml could be eitherpositive
ornegative).
ThelaggedeffectassessmentofdroughtonEOSwasconducted
todetermineatwhichmonththat1-monthSPEIhadthelargestsig-
nificantcorrelationwithEOSamongtheprevious12months.Unlike
thecumulativeeffectusing1-to12-monthSPEI,hereonly1-month
SPEIwasused. Toinvestigatethis laggedeffect, wecorrelatedEOS
with1-monthSPEIatprecedingnmonthsbeforeEOS(includingthe
EOSm mont h), and the lagged ef fect n was determined when the
Rmax-lagwasobtained.Forexample,ifthe1-monthSPEIattheEOSm
monthhadthehighestcorrelationwithEOS,thelaggedeffectwas
recordedas1laggedmonth (i.e.,laggedtime scales), indicating no
lagged effectonEOS;however,if EOShappened inOctoberin this
year, and the Rmax-lag was found u sing 1-month SPEI in Novem ber
from the previous year,then the lagged effect was recorded as 12
laggedmonths,andtheactuallagwas11months.Furtherstatistical
(1)
NDVI
ratio =
(
NDVI−NDVImin
)
(
NDVI
max
−NDVI
min)
(2)
NDVI
(t)=a1+a2−a7t×
1
1+exp a3−t
a4−1
1+exp a5−t
a6
FIGURE 2 Anexampleofdeterminationoftheendofthe
growingseason(EOS)usingGlobalInventoryModelingand
MappingStudiesthirdgenerationNDVIdatasetwiththetwoEOS
extractionmethods.Thedatescorrespondingtotheverticallines
ofAandBareEOSvaluesextractedbythemidpointmethodand
thedoublelogisticmethod,respectively
050100 150 200250 300350
0.2
0.4
0.6
0.8
1.0
NDVImin
Firstderivative(×10–2)
NDVI
Dayofyear(DOY)
Observed NDVI
Daily NDVI
Fitted NDVI(t)
NDVImax
–1.0
–0.5
0.0
0.5
1.0
B
Firstderivative line
A
|
5
PENG Et al .
analysiswasconductedmainlybasedontheresultsof1to12lagged
months.
The subse quent analys es across biome an d wetness grad ients
wereinvestigatedprimarilyonthebasisofthemaximumcorrelation
coefficients (Rmax-cml and Rmax-lag,respectively)andthecorrespond-
ingtimescales(accumulatedandlaggedmonths,respectively).Here,
to detect t he influen ce of hydrologic al condition s on these cum u-
lative and l agged effec t, the 12-month S PEI dataset at De cember
(i.e.,annual SPEI) during1982–2015wasadopted, representingthe
annualwaterbalance(namely,annualwetnesssituation).Theslope
ofannualSPEI wasalsoquantifiedbasedontheannualSPEIduring
1982–2015 usingt heline arle ast-squares regression, representing
the chang ing trend of water balance (or wetness) over the thr ee
decades.Furthermore,the wetness gradients weredeterminedby
equalintervalbasedonthemeanvalueofannualSPEIandtheslope
ofannualSPEIoverthewholestudyduration,withinter valvaluesof
0.1and0.01,respectively.
3 | RESULTS
3.1 | The cumulative effect of drought on EOS
3.1.1 | The responses of EOS to drought at different
time scales
The proportions of positive and negative significant correlations
betweenEOSand1-to12-monthSPEI arepresentedinFigure3.
As shown i n Figure 3, the p ercentage s of significa nt correlati on
weremainlypeakedat3–5months,andthepeakvaluewasfound
at 4-month time scale (positive in 9.0% and negative in 3.2%),
whereas i t had the minimu m value at 1month ( positive in 3.6%
and negat ive in 2.5%). Ove rall, the po sitive corre lation bet ween
EOSandSPEIhadlargerproportionthanthenegativecorrelation
ateachtimescale.
3.1. 2 | Spatial patterns of the cumulative effects of
drought on EOS
For the cumu lative effe ct of drought o n EOS, Figure 4 show s the
spatial pa tterns of the Rmax-cml bet ween them at p<0.05 a nd the
correspondingtimescales(i.e.,accumulatedmonths).Overall,27.2%
ofthevegetatedlandsshowed theRmax-cml(Figure4a).Positivecor-
relations(16.3%) were mostlylocated in Southern North America,
SouthernEurope, Central and WesternAsia, Central and Nor thern
Russia, an d Central Ch ina. Negati ve correlatio ns (10.9%)we re pri-
marily observed in Northern North America, Northern Europe,
WesternandEasternRussia,andNor theasternChina.InFigure 4b,
the accumulated months where the Rmax-cmloccurredwasconcen-
tratedatshortertimescales(i.e.,1–4months),totaling63.1%ofthe
areas, am ong which the 3-mo nth time scale occ upied the larges t
area(17.5%).Meanwhile,about13.7%ofthemaximumcorrelations
were occupied by the 1-month SPEI, largely located in middle and
highnorthernlatitudes.
3.1.3 | The cumulative effect of drought on EOS
at the biome level
Differentpatternswereshowninthepercentagesofmaximumsig-
nifica nt correlation bet ween EOS and 1- to 12-month SPEI (both
positive and negative correlations) across biomes(Figure 5).In the
viewof thetotal percentagesateachbiome,weobserved thelarg-
est percentage of maximum significant correlation in temperate
grassla nds, savannas, a nd shrublands ( TGSS) (48.5%), follow ed by
mediterranea nfores ts,w oo dl an ds ,a nd sc ru b(MF WS )(36.7% ),mon-
tanegrasslandsandshrublands(MGS)(36.3%),anddesert sandxeric
shrublands(DXS)(32.5%),whilethesmallestproportion wasshown
in boreal forests/taiga (BF) (21.5%). Meanwhile, the Rmax-cml was
primarilypositive in TGSS,MGS, MFWS,and DXS biomes,andthe
ratiosofnegativeandpositivecorrelationwereapproximatelyclose
attemperatebroadleafandmixedforests(TBMF),temperate conif-
erou sf or est s(TCF ), BF,an dt undra (TU N) bi ome s. Fromth ep ers pec-
tiveoftimescale,thepercentagesofRmax-cmlbetweenSPEIan dEOS
predominated at3–4accumulatedmonths forgrassland,savannas,
andshrubland(i.e.,TGSS,MGS,andDXSbiomes),2monthsforTUN,
and 1–6months for th e MFWS biome, resp ectively. Though EOS
showedslightlylargerproportionswith1-to4-monthSPEIinforests
(i.e.,TBMF,TCF,andBFbiomes),thereseemsnocleardominanttime
scalesinthesebiomes.
The potential relationship between the cumulative effect and
thewetnessacrossbiomeswasinvestigated,andanegativecorrela-
tion between themean Rmax-cmlandmeanSPEIattheEOSmmonth
wasfound(R2=0.75inFigure6).Generally,theaverageRmax-cml de-
creasedfrom0.38 to −0.07 with the increasing wetness,whichin-
dicates thatbiomes with positivewater balanceat theEOSm month
hadlowerco rr el ationbetweenEOSandSP EI(e.g.,BF,TCF,andTUN).
FIGURE 3 Percentagesofsignificantcorrelationbetween
StandardizedPrecipitation–EvapotranspirationIndexandEOS(at
p<0.05)from1-to12-monthtimescales
123456789101112
0
2
4
6
8
10
Percentage (%)
Accumulatedmonths
PositivecorrelationNegative correlation
6
|
PENG Et al.
3.1.4 | The cumulative effect of drought on EOS
along the water balance gradient
Boththe Rmax-cml and corresponding accumulated months showed
correlationswiththewaterbalancegradients(Figure7).Forthetime
scale of cumulative effect, the accumulated month was negatively
correlatedwith the meanannual SPEI (R2=0.36inFigure 7a),sug-
gesting thatdrought had cumulative effect on EOS atshortertime
scales at wet ter regions(i.e., regionswith larger valueofmean an-
nual SPEI). At t he same time, th e accumulated mont hs tended to
increas e from 3.7 to 5.3 with th e increased slo pe of annual SPEI
(R2=0.34 in Figure 7b). As showninFigure 7c, the correlationbe-
tweentheannualSPEIandpositiveRmax-cmlwasnegative(R2=0.57),
butthecorrelationwithnegativeRmax-cmlwas in sig nif ica nt. Alo ngt he
gradientofslopeofannualSPEI,thepositiveRmax-cmlalsoshowedan
obviouslydecliningtrend(R2=0.65inFigure7d),whilethevariation
ofnegativeRma x-cmlwasnotsignific ant.
3.2 | The lagged effect of drought on EOS
3.2.1 | The responses of EOS to drought at different
lagged time scales
We acquired t he percentages of p ositive and negati ve significant
correlations between EOS and 1-month SPEI during 1982–2015
(Figure8).The1-monthSPEIhadvariouscorrelationwithEOSduring
thepreviouslyindividual12months.Generally,thesignificantcorre-
lation betweenthem was mainly focused on short-term lag within
6months(positivein26.5%andnegativein17.0%).Comparedwith
otherlaggedtimescales,thelargestproportionwasobservedinthe
3-month lag,andthepercentage ofpositivecorrelation(8.1%)was
clearlylargerthanthenegativeone(1.7%).
3.2.2 | Spatial patterns of the lagged effects of
drought on EOS
Fromtheperspectiveoflaggedeffect,thespatialpatternsoftheRmax-lag
betweenEOSand1-monthSPEI(atp<0.05)aswellasthecorrespond-
inglagged months are presented inFigure 9.Generally,about 50.4%
of the veget ated pixels showe d the maximu m significa nt correlatio n
betweenEOS andSPEIovertheNorthernHemisphere(>30°N),and
the distributionof the Rmax-lagwasspatiallyheterogeneous.Thepro-
portionsofpositive andnegative correlation were26.6%and23.8%,
respectively. Consideringthe lag time, only4.2% of the area was af-
fected by t he concurre nt SPEI of the EOS month ( Figure 9b), which
sugges ts that 46.2% of t he vegetated lands s howed certain la gged
effectsbetweenEOSandSPEI.Among the pixelsshowing laggedef-
fect,nearly59.1%ofthemoccurredatshorterlaggedtimescales(e.g.,
1–6months),and the Rmax-lagat3laggedmonthsheldthelargestper-
centage (12. 8%). Moreover, in midlat itudes (30 °N–55°N),t he lagged
effectmostlyfellintherangebet ween2and3laggedmonths.
3.2.3 | The lagged effect of drought on EOS
at the biome level
The perce ntages of lagged e ffect varie d substantia lly at different
laggedmonthswithineachbiomeaswellasamongdif ferentbiomes
(Figure 10).Overall, droughtshowedevidentlagged effectonEOS,
FIGURE 4 Spatialdistributionofthecumulativeef fectofdroughtonEOS.(a)Spatialdistributionofthemaximumsignificantcorrelation
coefficients(i.e.,Rmax-cml)betweenStandardizedPrecipitation–EvapotranspirationIndexandEOSduring1982–2015.(b)Spatialpatternof
thecorrespondingtimescales(i.e.,accumulatedmonths)wheretheRmax-cml occurred
|
7
PENG Et al .
with over 45% of the a reas showing m aximum corr elation in ea ch
biome.ThelargestpercentageoftheRmax-lagwas68.7 %inTGSS,fo l-
lowedbyMGS(61.4%),MFWS(57.1%),andDXS(55.3%),while the
smallestproportion was45.5%inBF.Thereinto,themaximumcor-
relationwithconcurrentSPEIonlyaccountedfor3.1%–6.2%among
biomes.EOSinTGSS,MGS,andMF WSbiomessharedsubstantially
largerareas ofpositivecorrelationswithSPEI, which were approxi-
matelytwicethesizeofnegativecorrelation.However,wefoundno
apparentdifference betweentheproportionsof positiveandnega-
tive corre lations in oth er five biome s, and BF was t he only biome
which had a slightly larger percentage of negative correlation. For
thelagged monthsoflagged effect,thehighestproportions ofthe
Rmax-lagwereapparentlyobservedat2laggedmonthsforTUNand3
laggedmonthsforTGSS,MGS,MFWS,andDXSbiomes.Thelagged
effectof droughtonEOS was almostuniformlydistributedateach
laggedmonthfortheremainingbiomes(i.e.,TBMF,TCF,andBF).
Across biomes, we further found a significant correlat ion be-
tweenthebiome-averagedRmax-lagan d m e anSP E Iatt h e EO S mm o n t h
(R2=0.66 in Figure 11). Generally,themean Rmax-lag was smaller if
thebiomepossessedalargervalueofmeanSPEI.ThemeanRmax-lag
FIGURE 5 Percentagesofthe
Rmax-cmlbetweenEOSandStandardized
Precipitation–EvapotranspirationIndex
(SPEI)at1-to12-monthtimescales
ateachbiome.Ineachbiome,the
percentageofpositivecorrelation(P)at
iaccumulatedmonthsrepresentsthe
ratioofpixelsshowingpositiveRmax-cml
betweenEOSandi-monthSPEItoallthe
vegetatedpixelswithinthisbiome,while
thepercentageofnegativecorrelation
(N)representstheratioofpixelsshowing
negativeRmax-cml.Theletters(a–h)
representTBMF(a),TCF(b),BF(c),TGSS
(d),MGS(e),TUN(f),MFWS(g),andDXS
(h)biomes,respectively,anddetailed
descriptionsarefoundinTable1
123456789101112
0
3
6
9
12
15
Percentage (%)Percentage (%)Percentage (%)Percentage (%)
Percentage (%)Percentage (%)Percentage (%)Percentage (%)
(a) TBMF P: 15.3%
N: 12.1%
123456789101112
0
3
6
9
12
15
Positive correlation Negative correlation
(b) TCF P: 14.2%
N: 16.1%
123456789101112
0
3
6
9
12
15 (c) BF P: 9.0%
N: 12.5%
123456789101112
0
3
6
9
12
15 (d) TGSS P: 42.6%
N: 5.9%
123456789101112
0
3
6
9
12
15 (e) MGS P: 29.6%
N: 6.7%
123456789101112
0
3
6
9
12
15 (f) TUN P: 13.8%
N: 10.1%
123456789101112
0
3
6
9
12
15
Accumulatedmonths
(g) MFWS P: 32.5%
N: 4.2%
123456789101112
0
3
6
9
12
15
Accumulatedmonths
(h) DXS P: 25.7%
N: 6.8%
FIGURE 6 RelationshipbetweenthemeanvaluesofRmax-cml
andthemeanStandardizedPrecipitation–EvapotranspirationIndex
(SPEI)attheEOSmmonthacrossbiomes.Thesolidanddashlines
arethelinearfittedcurveandtheconfidencelevelforcurves(95%),
respectively
–0.15 –0.10 –0.05 0.00 0.05 0.100.15 0.20
–0.1
0.0
0.1
0.2
0.3
0.4
0.5
TBMF TCF
BF TGSS
MGS TUN
MFWS DXS
Mean Rmax-cml
Mean SPEI
y=–1.727x+0.253
R2=0.75, p<0.05
8
|
PENG Et al.
wasmorepositiveforTGSS,MFWS,andMGS,whilethevaluewas
negativeforBF.
3.2.4 | The lagged effect of drought on EOS
along the water balance gradient
We explored the patterns of the lagged effect of drought on
EOS along the water balance gradients and found significant
correlations between the mean Rmax-lagandmeantimescales
with the meanannual SPEIand slope of annual SPEI (Figure 12).
From the per spective of lagged time scale, the average lagged
monthshowe daninsignificantdownwardtrendalongtheincreas-
ing mean annual SPEI (Figure 12a), and it decreased from 6.4 to
6.1months. Along the gradient of slopeof annual SPEI, thetime
scalesoflaggedeffectexperiencedasignificantuptrend,increas-
ingfrom 5.3 to 6.7 (R2=0.92 in Figure12b).Inthe view of Rma x-
lag, the mean v alue of positi ve Rmax-lagwas negativelycorrelated
withthemeanannual SPEI (R2=0.76),whilethe relationship was
positivebetweenthenegativeRmax-lagandmeanSPEI(R2 = 0.34 in
Figure 12c).Similarly,the slope of annual SPEI showed negative
correlation with positive Rmax-lag(R2=0.67) and positive correla-
tionwithnegativeRmax-lag(R2=0.43inFigure12d).
4 | DISCUSSION
4.1 | The cumulative effect of drought on EOS
ThecumulativeeffectofdroughtonEOShasbeenquantifiedbased
onthe Pearsoncorrelation between EOS and 1-to12-month SPEI
during 1982–2015.Resultsshowed that thecumulative ef fect sex-
isted in 27.2% of the veget ated lands, wit h dominant time sc ales
beingconcentratedbetween1and4months.Thecumulativeeffect
ofdroughton EOS exhibited spatially heterogeneous distributions
overtheNorthernHemisphere(>30°N).Ourstudyreveals that the
wateravailabilitymayplayalong-termroleonautumnphenologyin
FIGURE 7 Distributionofthe
accumulatedmonths(a,b)andthemean
valuesofRmax-cml(c,d)alongthewater
balancegradients.Thewaterbalance
gradientsarerepresentativesofthe
meanannualStandardizedPrecipitation–
EvapotranspirationIndex(SPEI)(left)and
theslopeofannualSPEI(right).Thesolid
anddashlinesarethelinearfittedcurve
andtheconfidencelevelforcur ves(95%),
respectively,whenthefitissignificantat
p<0.05
0.80.6 0.40.2 0.00.2 0.40.6 0.8
3.5
4.0
4.5
5.0
5.5
0.06 0.04 0.02 0.00 0.02 0.04 0.06
3.5
4.0
4.5
5.0
5.5
0.80.6 0.40.2 0.00.2 0.40.6 0.8
0.45
–
–
––––
–
–
0.40
0.40
0.45
0.50
0.55
0.06 0.04 0.02 0.00 0.02 0.04 0.06
0.45
0.40
0.40
0.45
0.50
0.55
Mean R
max-cml
Mean annual SPEI
(a)
Mean Rmax-cml
SlopeofannualSPEI
(b)
y=–0.276x+4.728
R2=0.36, p<0.05
Accumulatedmonths
Mean annual SPEI
(c)
R2=0.57, p<0.05
Rmax>0:y=–0.024x+0.437
Rmax<0:NS
Rmax>0:y=–0.666x+0.444
Rmax<0:NS
y=7.728x+4.757
R2=0.34, p<0.05
(d)
Accumulatedmonths
SlopeofannualSPEI
R2=0.65, p<0.05
–––
–––– –––
FIGURE 8 PercentagesofsignificantcorrelationbetweenEOS
and1-monthStandardizedPrecipitation–EvapotranspirationIndex
atdifferentlaggedtimescales(atp<0.05)
123456789101112
0
2
4
6
8
10
Percentage (%)
Laggedmonths
Positive correlation Negativecorrelation
|
9
PENG Et al .
terrestrial ecosystems.Thebiological implicationssuggest thatthe
moisture(orwaterdeficiency)duringdifferentperiodsfacilitates(or
inhibits) vegetation growth and development synchronously, and
subsequentlyinfluencescurrentstageofplant'slifecycle.Basically,
the timin g and status of e ach stage d epend on the p revious ones
during th e life cycle (Liu, Fu, Zh u, et al., 2016b; Zu et al., 2018).
Therefore,theimpactsofwateravailabilityoneachstageconjointly
form thecu mu lat iv ee f fe c to nt hen ex tst a ge ,w hic hc aus et hecum u-
lativeeffectofwateronvegetationphenologyinautumn.
The cumulative effects of drought on EOS were substantially di-
vers ea mon gb iom es .InTGSS ,MF WS,MG S ,a ndDX Sbiom es, th epro-
portionsofmaximumcorrelationbetweenSPEIandEOS were larger
thanthose inforestsandtundra(i.e.,TBMF,TCF,BF,and TUN), con-
firmingthekey roleof wateravailability ingrasslands, savannas, and
shrubs(Knapp&Smith,2001;Liu,Fu,Zhu,etal.,2016b;Zhang,Kong,
etal., 2017). Theunderlying mechanism may be linked to the dif fer-
ent root fun ctional t raits (e.g ., root diamet er and root dep th) across
herbaceousandwoodyplants.Herbaceousplantswiththinnerroots
canrespondquicklytoseasonallyinfertile resourcesandunfavorable
conditions(e.g.,seasonal droughtorcold)by exploring soil resources
efficiently and reducing dependence on symbiotic mycorrhizal fungi
(Ma etal .,2018),buttheirshallowr oo tspreven tt he ma bs or bingwa te r
fromdeeper soillayerstosomeextent(Dodd, Lauenroth,&Welker,
1998;Dorjiet al.,2013; Fan, Miguez-Macho, Jobbágy,Jackson,& R.,
&OteroCasal,C .,2017;Wuetal.,2018),whichconjointlymakesEOS
respond rapidly to drought in grasslands. On the contrary, woody
plants w ith thicker roots a re usually more relia nce on mycorrhizae
which ena ble them to comp ete for abund ant resource s and survi ve
inintenseplant–plantcompetition(Maetal.,2018),anddeeperroots
arefavorablefordeeper soilwateruptake (Doddet al.,1998; Fan et
al., 2017; Wu et al., 2018), resulting in less dependent on variations
ofwatervariabilityontheterrestrialsurface.Besidesthat,forwoody
plants,thecapability andprocessesof waterstorage in planttissues
arecomplexandsubstantiallydifferfromthoseofherbaceousplants
(whosewaterstoragechieflydependsonthequantityofleaf),proba-
blycontributingtotheirdroughtresistance(Tianetal.,2018).
Meanwhile, the positive correlation between EOS and SPEI
waspredominant in TGSS, MF WS, MGS ,and DXS, suggestingthat
drought couldadvance thedateof EOS in thesebiomes. However,
theimpactofdroughtonEOSwasmoreambiguousinforests,asthe
propor tionsofnegativeandpositivecorrelationweresimilar.Onone
hand,treegrowthissensitivetoearly-seasondrought(i.e.,periodof
cambial g rowth) (D'Or angeville et al. , 2018), and water defi ciency
couldinduceearlierleafsenescence,especiallyolderleaves,formin-
imizingwaterlossandreallocating nutrients tootherpart s(Chaves
etal.,20 03).Ontheotherhand, for some treespecies, drought-in-
duced changesatgenelevelcould resultinstrongerdrought resis-
tance in leaves, preventingcell death induced by drought, altering
therelationshipsbetweenphotosyntheticsourcesandsinksamong
plantorgans, and thus delayed autumnsenescenceanddormancy
(Xie et al ., 2015). Moreover, the cumulative effect was predomi-
nated at 3- an d 4-month time s cales amo ng grassland s, savannas ,
and shru bs (i.e., TGSS, M GS, and DXS), wh ile there wer e no obvi-
ously dominant time scales for forests (Figure 5). Characterizedby
diverse plantspecies and complexrootsystems,forestsmay adapt
tovariousclimaticandhydrologicalconditionsandrespondtowater
availability at differenttime scales (Ma etal., 2018; Yan, Zhong, &
Shangg uan, 2017). The diverge nt hydraulic strateg ies in different
ecosystems(e.g., temperate andboreal forests) and areaswith dif-
ferenttreecovermaybeanotherreason(Tianetal.,2018).
FIGURE 9 SpatialpatternsofthelaggedeffectsobservedbetweenStandardizedPrecipitation–EvapotranspirationIndex(SPEI)andEOS.
(a)Spatialpat ternofthemaximumsignificantcorrelationcoefficients(i.e.,Rmax-lag)between1-monthSPEIandEOSduring1982–2015.(b)
Spatialdistributionofthecorrespondinglaggedmonths(i.e.,laggedtimescales)wheretheRmax-lagwasobserved
10
|
PENG Et al.
Additionally,inMFWS, thevariationinEOScould be largely
attributed to SPEI, especially at 1–6 accumulated months,
which was su pported by oth er studies tha t summer drought is
the primary environmental driver of the plant phenology at
Mediterraneanregions(Debussche,Garnier,&Thompson,2015;
Montserratmartíetal.,2009;Richardsonetal.,2013).Intundra,
the cumulative effect of drought on EOS was relativelyweaker,
mostly because that temperatureisthemaincontrol on vegeta-
tion in cold r egions, and th e various enviro nmental dri vers, in-
clud ingphotoperiod,insolation,permaf ros tdy nam ics,andlength
ofice/snowfreeperiod,maketheresponseofEOSmorecompli-
cated(Bin,Huang,Chen,&Wang,2018;Forkeletal.,2014,2015;
Khorsandetal.,2016).
Among al l these biom es, water def iciency w hen EOS occurre d
resultedinstrongerimpactsonEOSinaridbiomes(e.g.,MFWSand
TGSS),becausewateravailabilityhascriticalclimaticconstraints on
vegetatio n dynamics in t hese areas ( Wu et al., 2015). Conver sely,
those bio mes who were in favor able water condi tions when EOS
occurred (e.g., TCF and BF) exhibited less influences by water
availability.
FIGURE 10 PercentagesoftheRmax-lag
betweenEOSand1-monthStandardized
Precipitation–EvapotranspirationIndex
(SPEI)at1–12laggedmonthsateach
biome.Ineachbiome,thepercentage
ofpositivecorrelation(P)ati lagged
monthsrepresent stheratioofpixels
showingpositiveRmax-lagbetweenEOS
andtheSPEIatalagofimonthstoallthe
vegetatedpixelswithinthisbiome,while
thepercentageofnegativecorrelation
(N)representstheratioofpixelsshowing
negativeRmax-lag.Theletters(a-h)
representTBMF(a),TCF(b),BF(c),TGSS
(d),MGS(e),TUN(f),MFWS(g),andDXS
(h)biomes,respec tively,andmoredetailed
descriptionsarefoundinTable1
123456789101112
0
5
10
15
20
Percentage (%)Percentage (%)Percentage (%)Percentage (%)
Percentage (%)Percentage (%)Percentage (%)Percentage (%)
(a)TBMF P: 25.9%
N: 22.2%
123456789101112
0
5
10
15
20
Positivecorrelation Negativecorrelation
(b) TCFP: 26.9%
N: 23.3%
123456789101112
0
5
10
15
20 (c) BF P: 21.1%
N: 24.4%
123456789101112
0
5
10
15
20 (d) TGSS P: 42.4%
N: 26.3%
123456789101112
0
5
10
15
20 (e) MGSP: 41.6%
N: 19.8%
123456789101112
0
5
10
15
20 (f) TUNP: 26.4%
N: 23.1%
123456789101112
0
5
10
15
20
Lagged months
(g) MFWS P: 39.0%
N: 18.1%
123456789101112
0
5
10
15
20
Laggedmonths
(h) DXS P: 30.0%
N: 25.3%
FIGURE 11 RelationshipbetweenthemeanvaluesofRmax-lag
andthemeanStandardizedPrecipitation–EvapotranspirationIndex
(SPEI)atEOSmmonthacrossbiomes.Thesolidanddashlinesare
thelinearfittedcurveandtheconfidencelevelforcur ves(95%),
respectively
0.040.02 0.00 0.02 0.040.06 0.08
–
––
0.05
0.00
0.05
0.10
0.15
0.20
TBMF TCF
BF TGSS
MGS TUN
MFWS DXS
Mean Rmax-lag
Mean SPEI
y=–1.809x+0.126
R2=0.66, p<0.05
|
11
PENG Et al .
Theclearpatternsofcumulativeeffectalongthewetnessgradi-
entcouldbeassociatedwithhydrologicalconditions.Vegetation in
aridregions(i.e.,withlowerannualSPEI)usuallyexperiencedstron-
gercumulativeeffectofdroughtonEOS,indicatedbylargerpositive
Rmax-cml(Fig ure 7c) .It is rea s ona ble beca u seveg eta t ion gr owth ina rid
regions issensitiveto availablewater,and droughtin theseregions
wouldaggravatewaterdeficiency,constrainvegetationac tivit y,and
usually l ed to earlier de foliation (Dor man, Svoray, Perevolot sky, &
Sarris , 2013; Vicente-Serran o et al., 2013). Meanwhile , the larger
positive Rmax-cml was also obser ved in regions whe re annual SPEI
decreased during the period of 1982–2015 (Figure 7d), suggesting
that increased water loss exacerbated its influence on EOS, and
theweakerimpactofdroughtcouldbe attributed torelievedwater
stress.Fromthe viewpointofaccumulated time scale, EOS tended
torespondtoSPEIatshortertimescalesindrierregions(Figure7a)
probablydue to theirquick response to limitedwater in water-de-
ficient re gions and the refore mitig ate drought-induced d amage by
meansofvariousphysiologicaladaptationsandfunctionalstrategies
(Chaves et al. , 2003; Dorma n et al., 2013; Vicente-Ser rano et al.,
2013).
4.2 | The lagged effect of drought on EOS
Inthis study,we found that46.2% of the vegetated lands showed
certainlaggedeffects,withtheRmax-lagbetweenEOSandSPEIfrom
antecedent2ndto12thmonths.TheeffectofdroughtonEOSwas
mainly occ urred at short t ime scales withi n 6months, which fur-
ther emphasized that EOS usually responds to shor t-termdrought.
Especially,thepredominant 3 laggedmonths indicate thatdrought
that happ ened 2month s before EOS had th e stronges t impact on
EOS,consistentwiththetime-lageffectofprecipitation(Congetal.,
2017;Liu, Fu,Zhu,etal.,2016b).Biologically,wateravailabilityhas
influenceonconcurrentstageofvegetation'slifecycle,andthenext
stageiscloselyconnectedwiththepreviousones,andconsequently,
thewateravailabilityduring the previousstageshowed indirectef-
fectonthenextstageofvegetationdevelopment.Hence,theavail-
ablewaterduringthepreceding monthsprobablyhaslaggedef fect
onautumnphenology.
Similar tothecumulativeeffect,thelaggedeffec tof droughton
EOSwasapparentlydifferentamongbiomesespeciallybetweenfor-
ests (i.e., TBMF, TCF,andBF) andgrasslands, savannas, andshrubs
(i.e., MGS, DXS,andTGSS).Ingrasslands,savannas,andshrubs,the
laggedeffectpredominatedat3laggedtimescale,inlinewithprevi-
ousresultsshowingcumulativeandlaggedeffectsofprecipitationon
plants based on the 3-month StandardizedPrecipitation Index (Ji &
Peters,2003).Conversely,inforest s,thelaggedeffectsofdroughton
EOSwerem or ec om pl ic at ed an da lm os tu ni fo rm ly di st ributedate ac h
laggedmonth.Thediffe renceoflagge def fectbet weenthes ebi omes
mayberelatedtothespecies-specificsensitivit ytodroughtanddis-
tinctive local e nvironments (Anderson et al., 2010; D u, He, Yang,
Chen,&Zhu,2014;Wuetal.,2015).Consideringthedifferentwater
bal an cewhenEOSocc ur re da crossbi om es ,d roughts ho wedstronger
FIGURE 12 Distributionofthe
laggedtimescales(a,b)andthemean
Rmax-lagbetweenEOSandSt andardized
Precipitation–EvapotranspirationIndex
(SPEI)(c,d)alongthewaterbalance
gradients.Thewaterbalancegradients
areindicatedbythemeanannualSPEI
(left)andtheslopeofannualSPEI(right).
Thesolidanddashlinesarethelinear
fittedcurveandtheconfidencelevelfor
curves(95%),respectively,whenthefitis
significantatp<0.05
–
–
–
–
–
–––
––––
––––––0.80.6 0.40.2 0.00.2 0.40.6 0.8
5.0
5.5
6.0
6.5
7.0
0.06 0.04 0.02 0.00 0.02 0.04 0.06
5.0
5.5
6.0
6.5
7.0
0.80.6 0.40.2 0.00.2 0.40.6 0.8
0.45
0.40
0.40
0.45
0.50
0.55
0.06 0.04 0.02 0.00 0.02 0.04 0.06
0.45
0.40
0.40
0.45
0.50
0.55
Mean Rmax-lag
Mean annualSPEI
(a)
Mean Rmax-lag
Slopeofannual SPEI
(b)
Rmax>0:y=0.006x–0.415
Rmax<0:y=–0.031x+0.424
y=–0.161x+5.986
R2=0.15, p=0.21
Lagged months
Mean annual SPEI
(c)
R2=0.76, p<0.05
R2=0.34, p<0.05
Rmax>0:y=0.139x–0.416
Rmax<0:y=–0.614x+0.428
y=9.795x+5.881
R2=0.92, p<0.05
(d)
Lageed months
Slopeofannual SPEI
R2=0.43, p<0.05
R2=0.67, p<0.05
12
|
PENG Et al.
laggedeffe ct sonEOSindrierbio mes(e.g.,MFWSa ndMGS),whileit
hadweakercontrolinmoisterbiomes(e.g.,BF)(Figure11).
Underthehydrologicalconditions,thegeneralpatternoflagged
effectis similarto thecumulativeeffect.Forexample, droughthad
strongerlaggedeffectsonEOSindrierregionsthanwetterregions,
reflec ted by both posi tive and negat ive Rmax-lag(F igure 12a). From
theperspectiveoflaggedtimescale,clearpatternsthatEOSusually
respondstodrought at longer lagged monthsindrier regionswere
recognized,but respondstodroughtmorequicklywhenexperienc-
ingwaterloss.
Overall , at both the biome level a nd water balance level , the
lag gede ffectofd ro ug htonEOSh asasimi la rpat te rnwiththecumu-
lativeeffect.However,theaveragetimescalesoflaggedeffectwere
1monthlongerthanthe cumulativeeffect(Figure7andFigure12).
It confirm ed our assum ption that EOS was n ot only influe nced by
droughtoftheEOSmonthbutalsoaffectedbythelegacyofdrought
during earlier months. Moreover, the proportions of the maximum
correlat ion betwee n SPEI and EOS in fores ted lands were sm aller
than tho se in the grass lands, sava nnas, and shr ubs both in lag ged
andcumulativeef fect s(Figure 5andFigure 10),indicatingthatau-
tumnphenologyofforest shasstrongerresistancetowaterstress.
Additionally, it should be noted that SPEI is the representa-
tive of climat ic water balan ce between th e precipita tion and PET.
Therefo re, both precip itation and PE T contribute to t he status of
SPEI,consequentlyaffectingtherelationshipbetweenSPEIandEOS.
Studies have confirmed thatincreasedevapotranspiration not only
aggravatesthearidit yinareas with lower precipitation, it mayalso
leadtodroughtinregionswithsufficientandabundantprecipitation
(Cook, Sm erdon, Seage r, & C oats, 2014). Altho ugh SPEI respond s
theoreticallyequaltoprecipitationandevapotranspiration,thecon-
tributionofthesetwofactorstodrought iscomplicatedand varies
withregions(Vicente-Serrano, Schrier,Beguería, Azorin-Molina, &
Lopez-Moreno, 2015). According to the Penman–Monteith equa-
tion, the calculation of evap otranspiration u sed in SPEI database
involves various climatic parameters, such as temperature, relative
humidity,and insolation (Allen etal., 1998).Therefore,theimpacts
of drought on EOS c ould be more com plex due to the v arious in-
fluentialfactors(e.g.precipitation,evapotranspiration,temperature,
andinsolation),probablyrequiringmoreresearchontheinteractions
and mechanisms in thefuture.Moreover, the quantitative cumula-
tive and lagged effects of droughton EOS (e.g.,the dominant 1–4
and2–6months,respectively)provideanalternativetobetter pre-
dict autumnphenological datesbyincorporating theSPEIdatasets
atthe dominanttimescalesintoaphenologicalmodel,whichcould
furt her benefit t he modeling an d evaluating of eco system carb on
flux by using accuratelypredicted autumnphenology andadvance
ourunderstandingofthevegetation–climateinteractions.
ACKNOWLEDGMENTS
ThisworkwasfundedbytheStrategicPriorityResearchProgramof
theChineseAcademyofSciences(XDA19040103),NationalNatural
Science Foundation of China (41871225), and the Key Research
ProgramofFrontierSciences,CAS(QYZDB-SSW-DQC011).
ORCID
Chaoyang Wu https://orcid.org/0000-0001-6163-8209
Alemu Gonsamo https://orcid.org/0000-0002-2461-618X
REFERENCES
Allen,R.G.,Pereir a,L.S.,Raes,D.,&Smith,M.(1998).Crop evapotranspi-
ration: Guidelines for computing crop water requirements. Irrigation and
drainage paper no 56.Rome,Italy:FoodandAgricultureOrganization
oftheUnitedNations(FAO)
Anderson,L.O., Malhi,Y.,Aragão,L .E. O.C.,Ladle,R .,Arai,E .,Barbier,
N., & Phill ips, O. (2010). Remote s ensing detec tion of droug hts in
Amazonianforestcanopies.New Phytologist,187,733–750.https://
doi.org/10.1111/j.1469-8137.2010.03355.x
Balzarolo, M., Vicca,S., Nguy-Rober tson, A. L., Bonal, D., Elbers, J.A.,
Fu,Y.H.,…Veroustr aete, F.(2016).Matching thephenology ofNet
EcosystemExchangeandvegetationindicesestimatedwithMODIS
andFLUXNET in-situ observations. Remote Sensing of Environment,
174,290–300.https://doi.org/10.1016/j.rse.2015.12.017
Beguería, S., V icenteserrano, S. M., & A ngulomartínez, M. (2010). A
multiscalar global drought dataset: The SPEIbase: A new gridded
produc tfor the analysis of drought variabilityand impac ts. Bulletin
of the American Meteorological Society, 91, 1351–1356 . https://doi.
org /10.1175/2010 BAMS2988.1
Bin,H.E.,Huang,L.,Chen ,Z.Y.,&Wang,H .Y.(2018).Weakeningsensi-
tivityofglobalvegetationtolong-termdroughts.Science China Ear th
Sciences,61,60–70.https://doi.org/10.1007/s11430-016-9119-8
Buitenwerf,R .,Rose,L., &Higgins,S. I. (2015). Threedec ades ofmulti-
dimensionalchangeingloballeafphenology.Nature Climate Change,
5,364–368.https://doi.org/10.1038/nclimate2533
Chaves,M.M.,Maroco,J.P.,&Pereira,J.S.(2003).Under standingplant
respon ses to drought—From genes t o the whole plant . Functional
Plant Biology,30,239–264.https://doi.org/10.1071/FP02076
Chen, B.,Xu ,G., Coops, N. C.,Ciais, P.,Innes, J. L ., Wang, G.,…Liu, Y.
(2014).Changesinvegetationphotosyntheticactivitytrendsacross
the Asia-Pacificregion over the last three decades. Remote Sensing
of Environment, 144, 28–41. https://doi.org/10.1016/j.rse.2013.
12.018
Chen, J., Jönsson, P.,Tamura, M., Gu, Z., Matsushita, B., & Eklundh, L.
(2004). A simple method for recons truc ting a high-quality NDVI
time-series dat a set based on the Savitzk y-Golay filter. Remote
Sensing of Environment, 91, 332–344 . ht tps ://doi.org/10.1016/j .
rse.2004.03.014
Ciais, P.H.,Reichstein, M., Viovy,N ., Granier,A ., Ogée, J.,Allard, V., …
Valentini,R . (2005). Europe-wide reductionin primary productiv-
ity caused bythe heat and drought in 2003. Nature,437,529–533.
https://doi.org/10.1038/nature03972
C o ng , N . , S h e n , M . ,& P i a o , S .( 2 0 1 7 ). S p a t i a l v ar i a t i o n s i nr e s p o n s e s o fv e g -
etationautumnphenolog ytoclimatechangeontheTibetanPlateau.
Journal of Plant Ecology, 10, 744 –752 . http s://doi .org/10 .1093/jpe /
rtw084
Cook,B.I.,Smerdon,J.E.,Seager,R.,&Coat s,S.(2014).Globalwarming
and21stcentur ydrying.Climate Dynamics,43, 2607–2627. ht tps://
doi.org/10.10 07/s00 382- 014-2075-y
Cui, T., Mart z, L., & Gu o, X. (2017). Gr assland phe nology re sponse to
droughtintheCanadianPrairies.Remote Sensing,9, 1–21.htt ps ://doi .
org/10.3390/rs9121258
|
13
PENG Et al .
Dai, A. (2011). Drought under global warming: A review. Wiley
Interdisciplinary Reviews: Climate Change, 2, 45–65. https://doi.
org /10.1002/w cc. 81
Dannenberg, M. P.,Song,C., Hwang,T., &Wise, E. K. (2015). Empirical
evidenceofEl Niño-SouthernOscillationinfluenceon landsurface
phenol ogy and prod uctivit y in the wester n United States . Remote
Sensing of Environment, 159, 167–180. https://doi.org/10.1016/j.
rse.2014.11.026
Debussche, M., Garnier, E., & Thompson, J. D. (2015). Exploring
the causes of variation in phenology and morphology in
Mediterranean geophy tes: A genus-wide study of Cyclamen.
Botanical Journal of the Linnean Society, 145, 469–48 4. https://doi.
org /10.1111/j .1095-8 33 9.20 0 4. 00 298.x
Delpie rre, N., Duf rêne, E., S oudani, K ., Ulrich, E ., Cecchini , S., Boé, J.,
& François, C. (20 09). Modelling interannual and spatial variabil-
ity of lea f senescence f or three deci duous tree sp ecies in Franc e.
Agricultural and Forest Meteorology, 149, 938–948. https://doi.
org/10.1016/j.agrformet.2008.11.014
Dodd, M. B., Lauenroth, W. K., & Welker, J. M. (1998). Dif ferential
water resource useby herbaceous and woodyplant life-forms in a
short grass steppe community.Oecologia,117,50 4–512.https://doi.
org/10.1007/s004420050686
D'Ora ngeville, L ., Maxwell , J., Kneesh aw,D., P ederson, N ., Duches ne,
L.,Logan,T.,…Phillips,R.P.(2018).Droughttimingandlocalclimate
determi ne the sens itivit y of easter n tempera te forest s to drought .
Global Change Biology, 24, 2339–23 51. https: //doi.org/10 .1111/
gcb.140 96
Dorji,T.,Totland,O.,Moe,S.R.,Hopping,K.A.,Pan,J.B.,&Klein,J.A.
(2013).Plantfunctionaltraits mediatereproductivephenologyand
success in response to experimental warming and snow addition in
Tibet.Global Change Biology,19, 459– 472. h tt ps://do i.or g/10 .1111/
gcb.12059
Dorman , M., Svoray, T., Perevolotsk y, A., & Sar ris, D. (2013). Forest
performance during two consecutive drought periods: Diverging
long-termtrendsandshort-termresponsesalongaclimaticgradient.
Forest Ecology and Management,310,1–9.htt ps: //doi.or g/10.1016/j.
foreco.2013.08.009
Du,J.,He,Z.,Yang, J.,Chen, L.,&Zhu, X.(2014).Detectingtheef fect s
of climate change on canopy phenolog y in coniferous forests in
semi-aridmountain regionsof China.International Journal of Remote
Sensing,35,6490–6507.https://doi.org/10.1080/01431161.2014.95
5146
Elmore, A . J., Guinn , S. M., Minsl ey,B. J ., & Richards on, A. D. (2012).
Lands cape control s on the timing of sp ring, autum n, and growing
season length in mid-Atlantic forests. Global Change Biology, 18,
656– 674.https://doi .org/10.1111/j.1365 -2486. 2011 .0 2521 .x
Estre lla, N., & Menzel , A. (2006). Re sponses of leaf col ouring in four
deciduoustree speciestoclimate and weather inGermany.Cli mate
Research,32,253–267.https://doi.org/10.3354/cr032253
Fan, Y., Miguez-Ma cho, G., Job bágy, E. G., Jac kson, R. B ., & Otero-
Casal, C. (2017). Hydrologic regulation of plant rooting depth.
Proceedi ngs of the National Academy of Sciences of the United
States of America, 114, 10572–10577. ht tps ://doi.org/10.1073/
pna s.171 2381114
Fensham,R.J.,Fairfax,R.J.,&Ward,D.P.(2009).Drought-inducedtree
death in s avanna. Global Change Biology, 15, 380–387. https://doi.
org /10.1111/j .136 5-2486.200 8. 01718.x
Forkel, M., Migliavacca, M., Reichstein, M., Schaphoff, S., Thonicke,
K., Thurner, M., … Carvalhais, N. (2014). Identifying controls on
vegetation greenness phenology through model-data integration.
Biogeosciences Discussions,11,7025–7050.https://doi.org/10.5194/
bg-11-7025-2014
Forkel, M.,Migliavacca, M.,Thonicke,K., Reichstein,M., Schaphoff,S.,
Weber, U., & Car valhais, N. (2 015). Codominant w ater control on
globalinterannual variability and trendsinland surfacephenology
and greenness. Global Change Biology, 21, 3414–3435. htt ps://doi.
org /10.1111/gcb.12 950
Ganguly,S.,Friedl, M.A., Tan, B., Zhang, X ., & Verma, M.(2010). Land
surfacephenologyfromMODIS:CharacterizationoftheCollection
5globallandcoverdynamicsproduct.Remote Sensing of Environment,
114,1805–1816.https://doi.org/10.1016/j.rse.2010.04.0 05
Garonna,I.,De,J. R.,& Schaepman, M. E. (2016).Variabilit yandevolu-
tion of global land sur face phenology over the past three decades
(1982–2012). Global Change Biology, 22, 1456–1468 . ht tps: //doi.
org /10.1111/gcb.13168
Ge,Q.,Wang,H .,Rutishauser,T.,&Dai,J.(2015).Phenologicalresponse
toclimate changein China: Ameta-analysis.Global Change Biology,
21,265–274.https://doi.org/10.1111/gcb.12648
Gonsa mo, A., Che n, J. M., & Ooi, Y. W. (2018). Peak se ason plant ac-
tivity shifttowards spring isreflec ted by increasing carbon uptake
byextratropicalecosystems. Global Change Biology,24, 2117–2128.
https ://doi.or g/10.1111/gcb.140 01
Gord o,O.,&Sanz,J.J.(2010).Impactofclimatechangeonplan tphe nol-
ogyinMediterranean ecosystems.Global Change Biology,16, 10 82–
1106.https://doi.org/10.1111/j.1365-2486.2009.02084.x
Huang,C.Y.,Asner,G.P.,Barger,N.N.,Neff,J.C.,&Floyd,M.L.(2010).
Regionalabovegroundlivecarbonlossesduetodrought-inducedtree
diebackinpinon-juniperecosystems.Remote Sensing of Environment,
114,1471–1479.https://doi.org/10.1016/j.rse.2010.02.003
Huang, L.,He,B., Han, L.,Liu,J.,Wang, H., & Chen, Z. (2017).A global
examina tion of the re sponse of eco system wa ter-use effic iency to
droughtbasedonMODISdata.Science of the Total Environment,6 01–
602,1097–1107.https://doi.org/10.1016/j.scitotenv.2017.05.084
Huang, M . T.,Wang, X.H., Keenan, T.F.,& Piao, S. L.(2018). Drought
timing influencesthelegacy oftreegrowth recovery.Global Change
Biology,24,3546–3559.https://doi.org/10.1111/gcb.14294
Jeong, S . J., Ho, C. H., G im, H. J., & Brown, M . E. (2011). Phenolog y
shift s at start vs. end of growing season in temperate vege-
tation over the Nor ther n Hemisp here for the period 1982–
2008 . Global Change Biology, 17, 2385–2399. https://doi.
org /10.1111/j .136 5-2486.2011. 023 97.x
Ji, L., & Peters,A. J. (20 03). Assessing vegetation response to drought
in the nor thern G reat Plai ns using veget ation an d drought in dices.
Remote Sensing of Environment,87, 85–98. https://doi.org/10.1016/
S0034-4257(03)00174-3
Kang,W.,Wang,T.,&Liu,S.(2018).Theresponseofvegetationphenology
andproductivit ytodroughtinsemi-aridregionsof Northern China.
Remote Sensing,10,1–15.https://doi.org/10.3390/rs10050727
Keskitalo,J.,Gustaf,B.,Per,G.M.,&Stefan,J.(2005).Acellulartimeta-
bleofautumnsenescence.Plant Physiology,139,1635–1648.https://
doi.org/10.1104/pp.105.066845
Khorsand,R.R.,Oberbauer,S.F.,Starr,G.,Parker,L.P.I.,Pop,E.,Ahlquist,
L.,&Baldwin ,T.(2016).Plantphenologicalresponsestoalong-term
experimentalextension of growing season and soil warming inthe
tussock tundra of Alaska. Global Change Biology, 21, 4520–4532.
https ://doi.or g/10.1111/gcb.130 40
Knapp, A . K., & Smit h, M. D. (2001). Var iation amon g biomes in tem-
poral d ynamics of aboveg round primar y productio n. Science, 291,
481–484.https://doi.org/10.1126/science.291.5503.481
Kong,D., Qiang,Z.,Singh,V.P.,&Shi,P.(2016).Seasonalvegetationre-
sponse to climatechange in the northernhemisphere (1982–2013).
Global and Planetary Change, 148 , 1–8. https://doi.org/10.1016/j.
gloplacha.2016.10.020
Li, P., Peng, C. , Wang, M., Luo , Y.,Li , M., Zhang , K., … Zhu, Q. ( 2018).
Dynamicsofvegetationautumnphenologyanditsresponsetomul-
tiple environmental factors from 1982to2012 on Qinghai-Tibetan
PlateauinChina.Scien ce of the Total Environ ment,637–638,855–86 4.
https://doi.org/10.1016/j.scitotenv.2018.05.031
Liu, Q., Fu, Y. H., Zeng, Z., Huang, M., Li, X., & Piao, S. (2016a).
Temperature, precipitation, and insolation effects on autumn
14
|
PENG Et al.
vegetationphenologyintemperateChina.Global Change Biology,22,
64 4–655.ht tp s://doi.o rg /10.1111 /gcb .13081
Liu, Q., Fu, Y.H., Zhu, Z.,Liu, Y.,Liu, Z., Huang, M.,… Piao, S. (2016b).
DelayedautumnphenologyintheNorthernHemisphereisrelatedto
changeinbothclimateandspringphenology.Global Change Biology,
22,370 2–3711.ht tps: //doi.org/10.1111/gc b.13 311
Ma,Z.,Guo,D.,Xu,X.,Lu,M.,Bardgett,R.D.,Eissenstat,D.M.,…Hedin,
L. O. (2018). Evolutionar y history resolves global organization of
root func tional t raits . Nature, 555, 94–97. htt ps://doi. org /10.1038/
nature25783
Martin-Benito, D., Beeckman, H ., & Cañellas, I. (2013). Influence of
drought on tree ringsandtracheidfeatures of Pinus nigra and Pinus
sylvestrisinamesicMediterraneanforest.European Journal of Forest
Research,132,33–45.ht tp s://doi.org /10.10 07/s10342-012-0652-3
Martínez-Vilalt a,J.,&Lloret,F.(2016).Drought-inducedvegetationshifts
in terrestrial e cosystems: The ke y role of regenerati on dynamics.
Global and Planetary Change,144 ,94–108.ht tps://doi.o rg/10.1016/j.
gloplacha.2016.07.009
Montserratmartí,G ., Camarero, J. J., Palacio,S., Pérezrontomé,C ., Milla,
R., Alb uixech, J. , & Maestro , M. (2009). S ummer-drough t constrai ns
the ph enolog yan dg row thoftwoco existingM edite rr an ea noakswi th
contrastin gle afhab it:Im plicati onsforth eirpe rsisten cea ndrep ro duc-
tion.Tre es ,23,787–799.https://doi.org/10.1007/s00468-009-0320-5
Olson, D. M ., Dinerstein, E., Wik ramanayake, E. D., Burgess, N.
D., Powell, G . V. N., Under wood, E . C., … Kassem, K . R. (2001).
Terrestrial ecoregions of the world: A new map of life on earth.
Bioscience,51 ,933–938.https://doi.org/10.16 41/0006-3568(20 01)
051[0933:TEOTWA]2.0.CO;2
Pan,N.,Feng,X.,Fu,B.,Wang,S.,Ji,F.,&Pan,S.(2018).Increasingglobal
vegetationbrowning hiddenin overallvegetationgre ening:Insights
fromtime-varyingtrends.Remote Sensing of Environment,214,59–72.
https://doi.org/10.1016/j.rse.2018.05.018
Pastor-Guzman,J.,Dash,J.,&Atkinson,P.M.(2018).Remotesensingof
mangroveforest phenology and its environmental drivers.Re mote
Sensing of Environment, 205, 71–84. https://doi.org/10.1016/j.
r se . 2 01 7.1 1. 0 0 9
Piao, S., Cui,M., Chen, A., Wang,X., Ciais, P.,Liu,J., &Tang,Y.(2011).
Altitudeandtemperaturedependenceofchangein thespringveg-
etation green-up date from 1982 to 2006 in the Qinghai-Xizang
Plateau.Agricultural and Forest Meteorology,151 ,159 9–16 0 8. htt p s: //
doi.org/10.1016/j.agrformet.2011.06.016
Piao, S., Fang, J., Zhou, L., Philippe, C., & Zhu, B. (20 06).
Variations in satellite-derived phenology in China's temper-
ate vegetation. Global Change Biology, 12, 672–685. https://doi.
org /10.1111/j .136 5-2486.200 6. 0112 3. x
Piao, S., Tan, K., Nan, H., Ciais,P.,Fang, J.,Wang, T., … Zhu, B.(2012).
Impacts ofclimate andCO2changes on thevegetationgrowthand
carbon balance of Qinghai-Tibetan grasslands over the past five
decades. Global and Planetary Change, 98–99, 73–80. https://doi.
org/10.1016/j.gloplacha.2012.08.009
Richardson,A.D., Keenan,T.F.,Migliavacca,M.,Ryu,Y.,Sonnentag,O.,
&Toomey,M.(2013).Climate change,phenology,andphenological
control of ve getatio nf eedbac ks to the clim ate syste m. Agricultural
and Forest Meteorology, 169, 156–173. ht tp s://doi .or g/10.1016/j.
agrformet.2012.09.012
Schut,A.G .T.,Ivits,E.,Conijn,J.G.,tenBrink,B.,&Fensholt,R .(2015).
Trends in globa l vegetation a ctivit y and climatic d rivers ind icate a
decoupled response to climate change. Plos One, 10, 1–21.https://
doi.org/10.1371/journal.pone.0138013
Seyednasrollah,B.,Swenson,J.J.,Domec,J.-C.,&Clark,J.S.(2018).Leaf
phenology paradox:W hywarming mattersmostwhere itisalready
warm. Remote Sensing of Environment, 209, 4 46–455. ht tps://doi.
org /10.1016/j.r se.2018.02.059
Shen, M ., Tang, Y.,C hen, J., Zhu , X., & Zheng , Y.(20 11).Inf luences of
temperature and precipitationbeforethegrowing seasononspring
phenology ingrasslands ofthecentral and eastern Qinghai-Tibetan
Plateau.Agricultural and Forest Meteorology,1 51,1711–1722.https://
doi.org/10.1016/j.agrformet.2011.07.003
Tao, F., Yokozawa, M., Zhang, Z., Hayashi, Y., & Ishigooka, Y.
(2008). Land surface phenology dynamic s and climate varia-
tions in the Nort h East China Transec t (NECT ), 1982–2000.
International Journal of Remote Sensing,29, 5461–5478. https://doi.
org/10.1080/01431160801908103
Tian, F., Wign eron, J. P., Ciais, P., Chave, J., Oge e, J., Penuela s, J., …
F en s h o l t , R . ( 2 01 8 ) . Co u p l i n g o fe c o s y s t e m - sc a l e p l a n t wa t e r s t o r a ge
andleafphenolog yobservedbysatellite.Nature Ecolog y & Evolution,
2,1428–1435.https://doi.org/10.1038/s41559-018-0630 -3
Vicenteserrano,S.M.,Beguería,S.,&Lópezmoreno,J.I.(2010).Amulti-
scalar drought indexsensitiveto global warming: The Standardized
PrecipitationEvapotranspirationIndex.Journal of Climate,23,1696–
1718. https://doi.or g/10 .1175/20 09JCLI290 9.1
Vicente-Serrano, S. M., Beguería, S., Lorenzo-L acruz, J., Camarero, J.
J., López-Moreno, J. I., Azorin-Molina, C., … Sanchez-Lorenzo, A.
(2012).Performance of drought indicesforecological,agricultural,
andhydrologicalapplications.Earth Interactio ns,16,1–27.ht t ps ://doi .
org /10.1175/2012EI 00 0 43 4.1
Vicente-Serrano,S.m.,Gouveia,C.,Camarero,J.j.,Begueria,S.,Trigo,R.,
Lopez-Moreno,J.i.,…Sanchez-Lorenzo,A .(2013).Responseofveg-
etationtodroughttime-scalesacrossgloballandbiomes.Proceedings
of the Nation al Acade my of Sciences of the United States of America,
110,52–57.https://doi.org/10 .1073/pn as.1207068110
Vicente -Serra no, S. M., Schrie r, G . V. D. , Beguería, S., A zorin-Molina ,
C., & Lop ez-Moren o, J. I. (2015). Co ntribut ion of preci pitatio n and
referenceevapotranspirationtodroughtindicesunderdifferentcli-
mates. Journal of Hydrology, 526, 42–5 4. http s://d oi.org /10.1016/j.
jhydrol.2014.11.025
Wendler, G., C onner, J., Moore, B., S hulski, M., & Stu efer, M. (2011).
Climatol ogy of Alask an wildfire s with specia l emphasis on t he ex-
treme year of 20 04. Theoretical and Applied Climatology,10 4, 459–
472.https://doi.org/10.1007/s00704-010-0357-9
White,M.A .,deBEURS,K .M.,Didan,K .,Inouye, D.W.,Richardson,A.
D.,Jensen,O. P.,…Lauenroth,W.K .(20 09).Intercomparison ,inter-
pretation,andassessmentofspringphenolog yinNor thAmericaes-
timated from remote sensingfor1982–2006.Global Change Biology,
15,2335–2359.https://doi.org/10.1111/j.1365-2486.2009.01910.x
Whi te, M.A .,Th ornton,P.E.,&R unning,S .W.(1997).Aco ntine ntalp he-
nologymodelformonitoringvegetationresponsestointerannualcli-
matic variability.Global Biogeochemical Cycles,11,217–234.ht tps://
doi.org/10.1029/97GB00330
Wilhite,D. A.,& Glantz, M.H.(1985). Under standing the droughtphe-
nomenon:Theroleofdefinitions.Water International,10, 111–120.
https://doi.org/10.1080/02508068508686328
Wu,C.,Chen,J.M.,Black, T.A., Price,D.T.,Kurz,W.A.,Desai,A.R.,…
Bo hr er,G .( 20 13).I ntera nnu al va ria bi li t yo fn et ec os yst empro du c ti v-
ityinforestsisexplainedbycarbonfluxphenologyinautumn.Global
Ecology and Biogeography, 22, 994–1006 . ht tps: //doi.org/10 .1111/
geb.12044
Wu, C., Gough, C . M., Chen, J. M.,& Gonsamo, A.(2013). Evidence of
autumn phenologycontrolonannualnetecosystemproductivityin
twotemper atedeciduousforest s.Ecological Engineering,60,88–95.
https://doi.org/10.1016/j.ecoleng.2013.07.019
Wu,C .,Peng, D.,S oudani,K .,Soudani,K .,Siebicke,L .,G ough,C.M.,…
Ge,Q.(2017).L andsurfacephenolog yderivedfromnormalizeddif-
ferencevegetationindex(NDVI)atglobalFLUXNETsites.Agricultural
and Forest Meteorology, 233, 171–182. htt ps://doi.org /10.1016/j.
agr for met.2016.11.193
Wu,D.,Zhao,X.,Liang,S.,Zhou,T.,Huang,K.,Tang,B.,&Zhao,W.(2015).
Time-lag effects ofglobal vegetation responses to climate change.
Global Change Biology, 21, 3520–3531 . ht tps://doi.org /10.1111/
gcb.12945
|
15
PENG Et al .
Wu, X., Liu, H., Li, X., Ciais, P., Babst, F., Guo, W., … Ma, Y. (2018).
Diffe rentiating drou ght legacy eff ects on veget ation growth over
the temperate Northern Hemisphere. Global Change Biology, 24,
504–516.https://doi.org/10.1111/gcb.13920
Xie,Y.,Wang,X.Jr,&Silander,J.A. (2015).Deciduousforestresponses
to temper ature, precip itation, and dr ought imply comp lex climate
change impacts. Proceedings of the National Academy of Sciences
of the United States of America, 112, 13585–13590. https://doi.
org /10.1073/p nas.1509 99111 2
Xie, Y., Wang, X ., Wilson, A . M. Jr, & Siland er,J. A . (2018). Pred icting
autumnphenology:Howdeciduoustreespeciesrespondtoweather
stressors. Agricultural and Forest Meteorology, 250–251, 127–137.
htt ps://doi.org/10.1016/j. agrfo rmet. 2017.12.259
Yan,W.,Zhong,Y.,&Shangguan,Z.(2017).Responsesofdiffe rentphys-
iologic alparameterthresholdsto soilwateravailability infourplant
speciesduringprolongeddrought.Agricultural and Forest Meteorology,
247,311–319.https://doi.org/10.1016/j.agrformet.2017.08.017
Yang, Y., Guan, H. , Shen, M., L iang, W., & Jiang , L. (2015). Ch anges in
autumn vegetationdormancy onset date and the climatecontrols
across te mperate ecos ystems in Chin a from 1982 to 2010. Global
Change Biology,21,652–665.https://doi.org/10.1111/gcb.12778
Zhang, Q., Kong, D., Singh, V.P.,&Shi, P.(2017). Resp onse of vegeta-
tionto different time-scales droughtacross China: Spatiotemporal
patterns,causesandimplications.Global and Planetary Change,152,
1–11.https://doi.org/10.1016/j.gloplacha.2017.02.008
Zhang, X ., Friedl, M . A., Sc haaf, C. B., Str ahler, A. H ., Hodges, J . C. F.,
Gao,F.,… Huete,A.(2003).Monitoringvegetation phenologyusing
MODIS. Remote Sensing of Environment, 84, 471–475. ht tp s://doi .
org/10.1016/S0034- 4257(02)0 0135-9
Zhang,X .,Liu,L .,Liu,Y.,Jayavelu,S.,Wang,J.,Moon,M.,…Schaaf,C.B.
(2018).GenerationandevaluationoftheVIIRSlandsurfacephenol-
ogy product.Remote Sensing of Environment,216, 2 12–2 29. ht tp s: //
doi.org/10.1016/j.rse.2018.06.047
Zhang,Y.,Song,C.,Band,L.E.,Sun,G.,&Li,J.(2017).Reanalysisofglobal
terrestrial vegetation trends from MODIS products: B rowning or
greening? Remote Sensing of Environment,191 ,145–155.https://doi.
org/10.1016/j.rse.2016.12.018
Zheng, C., Tang,X.,Gu,Q.,Wang ,T.,Wei,J., Song,L.,& Ma,M.(2018).
Climatic anomalyand its impact on vegetation phenology, carbon
seques tration and w ater-use efficie ncy at a humid tem perate for-
est. Journal of Hyd rology, 565, 150–159. htt ps://doi.org/10.1016/j.
jhydrol.2018.08.012
Z h u , W. , J i a n g , N . , C h e n , G . , Z h a n g , D . , Z h e n g , Z . , & F a n , D. ( 2 0 1 7 ) . D i v e r g e n t
shi ftsandre sp on se sof plantau tu mnphenolo gytoclima techa ng eon
theQinghai-TibetanPlateau.Agricultural and Forest Meteorology,239,
166–175.https://doi.org/10.1016/j.agrformet.2017.03.013
Zhu,W.Q.,Zheng,Z.T.,Jiang,N.,&Zhang,D.H.(2018).Acomparative
analysisofthe spatio-tempor al variation in thephenologies of two
herbac eous species a nd associated c limatic drivi ng factors o n the
Tibet an Plateau . Agricultural and Forest Meteorology, 248, 177–184.
https://doi.org/10.1016/j.agrformet.2017.09.021
Zu,J.,Zhang,Y.,Huang,K.,Liu,Y.,Chen,N.,&Cong,N.(2018).Biological
andclimatefactors co-regulatedspatial-temporaldynamicsofveg-
etation autumn phenology on the Tibetan Plateau. International
Journal of Applied Earth Observation and Geoinformation,69,198–205.
https://doi.org/10.1016/j.jag.2018.03.006
How to cite this article:PengJ,WuC,ZhangX,WangX,
GonsamoA.Satellitedetectionofcumulativeandlagged
effectsofdroughtonautumnleafsenescenceoverthe
NorthernHemisphere.Glob Change Biol. 2019;00:1–15.
https://doi .org /10.1111/gcb.14627
