and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
ContentslistsavailableatSciVerseScienceDirect
Agriculture,
Ecosystems
and
Environment
j ou rna l h o m e p a g e :w w w . e l s e v i e r . c o m / l oc a t e / a g e e
Modelling
the
impact
of
climate
change
on
the
interaction
between
grapevine
and
its
pests
and
pathogens:
European
grapevine
moth
and
powdery
mildew
Amelia
Caffarra
a,∗,
Monica
Rinaldi
a,
Emanuele
Eccel
a,
Vittorio
Rossi
b,
Ilaria
Pertot
aaIASMAResearchandInnovationCentre,SanMicheleall’Adige,Trento,Italy bUniversitàCattolicadelSacroCuore,Piacenza,Italy
a
r
t
i
c
l
e
i
n
f
o
Articlehistory: Received18April2011 Receivedinrevisedform 18November2011 Accepted23November2011 Keywords: Chardonnay Erysiphenecator Host Lobesiabotrana NorthernItaly Pathogen Pest Projections Synchrony Viticulture
a
b
s
t
r
a
c
t
Climatechangemayimpactpatternsofplantdiseasesandarthropoddevelopmentinmorecomplex waysthanexpected.Infact,whereasbothcropsandcroppathogensandpestsareaffectedbyclimatic variables,theymightbeinfluencedbydifferentcombinationsofdrivingfactors,andtheymightrespondto theirchangeatdifferentrates.Inordertoseparatetheseeffects,weneedtoimproveourunderstanding ofthehost-pest/pathogensystem,andconsidertheirinteraction.Theaimofthisstudywastorefine currentassessmentsofclimatechangeimpactsonpestanddiseasepressureongrapevinesbyconsidering pest/pathogen–hostinteractions.Thisresearch(i)combineddetailedphenologicalmodelsofgrapevine withphenologicalmodelsofoneofitskeyinsectpests(Europeangrapevinemoth)andoneofitskey pathogens(powderymildew),(ii)appliedthemodelstoclimatechangescenariosforaselectedstudyarea intheeasternItalianAlps,and(iii)consideredpotentialchangesintheirinteractions.Thesesimulations suggestthatinthewarmer,moreprofitableviticulturalareasofthestudyregionincreasingtemperature mighthaveadetrimentalimpactoncropyieldduetoincreasedasynchronybetweenthelarvae-resistant growthstagesofgrapevineandlarvaeoftheEuropeangrapevinemoth.Ontheotherhand,theincreasein pestpressureduetotheincreasednumberofgenerationsmightnotbeassevereasexpectedonthebasis ofthepestmodelonly,duetotheadvanceinharvestdateslimitingdamagesfromlate-seasongenerations. Simulationsforpowderymildewhighlightedadecreaseinsimulateddiseaseseverity,especiallyinyears withalateronsetofthediseasesymptomsandintheclimatescenariowithhighertemperatureincreases. © 2011 Elsevier B.V. All rights reserved.
1. Introduction
Majorshiftsintemperatureandchangesintheseasonalpattern ofrainfalldistributionarecurrentlyaffectingmostoftheworld. Climatic projections suggest that these trends will continue in thecomingdecades,affectingbothmeanandextremevaluesof thesevariables(Easterlingetal.,2007).Inthelatestreportofthe IntergovernmentalPanelonClimateChange(IPCC),meanglobal temperatureisestimatedtoincreasebetween1.8and4.0◦C(with alikely range of1.1–6.4◦C),by theendof thepresentcentury, dependingonthegreenhousegasemissionscenario (Easterling etal.,2007).Thecombinationofclimatechange,associated dis-turbancesandotherglobalchangedriversisexpectedtoexceed theresilienceofmanyagro-ecosystems.Asaconsequence,climate changecouldsubstantiallyimpactagricultureandfoodproduction
(OlesenandBindi,2002;Fuhrer,2003;Maracchietal.,2005;Kang
etal.,2009).Theresultofclimaticchangeshouldnotbealways
∗ Correspondingauthor.
E-mailaddress:amelia.caffarra@gmail.com(A.Caffarra).
seenasathreattofarmproductivity,especiallywherewaterisnot alimiting factor.However,Olesenetal.(2011)pointedoutthat theperceivedoutcomesofclimatechangeexpectedbyEuropean farmers remainmostly negative, and in particular, interviewed farmersdisclosedthefeelingthattheriskfrompestsanddiseases forgrapevinewillincreaseintheAlps(bothnorthandsouth).
However,consequencesofclimate-drivenchangesarenot eas-ilypredictableincomplexagro-ecosystems,asthebiologyofpests andpathogensandthatoftheirhostplantsareinterdependent.For example,manypests/pathogensaffecttheirhostplantonlyduring specificvulnerableperiodsoftheplantlife-cycle.Thisisthecase ofthepathogensthatinfectplantsthroughtheirflowers,asthe bacteriumErwiniaamylovora,whichcanpenetrateitshosts(e.g., appleand pear)duringflowering (Thomson, 2000).Otherpests mightbeabletoattacktheirhostthroughouttheirgrowth sea-sonbut causehigherdamageduringspecificgrowthstages.For example,thelarvaeofEuropeangrapevinemoth(Lobesiabotrana) arelessharmfulduringflowering(GabelandRoehrich,1995),but producemoredamageinthepost-veraisonperiod,whenthey influ-encegreymould(Botrytiscinerea)infections(Moschosetal.,2004). Manyplantspeciesprogressivelyincreasetheirresistancetopests
0167-8809/$–seefrontmatter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.agee.2011.11.017
andpathogensastheyagebydeveloping“ontogenic”resistance, whichmaybeactiveonthewholeplantorinspecificorgansor tissues(PanterandJones,2002;Gadouryetal.,2003).For exam-ple,grapeberriesarereportedtobesusceptibletoErysiphenecator (thepowderymildewfungus)infectionsuntilsolublesolidslevels reach8%(8◦Brix),andtheestablishedfungalcoloniesarereported tosporulateuntilsolublesolidslevelsreach15%(15◦Brix)(Delp,
1954;ChellemiandMarois,1991).
Thus, inpresenceofpests,infestationswilloccuronlyunder specificenvironmental conditions and only if the hostplant is in a susceptible growth stage (Chakraborty et al., 2000). For pathogens,thisinteractionhasbeenfrequentlyrepresentedbythe “plant–diseasetriangle”,whichismadeupbythethreeelements requiredfortheinfectiontodevelop:asusceptiblehost,the pres-enceofthepathogen,andaconduciveenvironment(Chakraborty
etal.,2000).
Inadditiontotheabovedescribed“susceptibilitywindows”,one shouldconsiderthedurationoftheproductivecycleofeachcrop. Infact,whilehighertemperaturesmightfavourthedevelopment ofcertainpests,theycouldalsoshortenthelengthofcropcycles, thusbalancingoutapotentialincreaseinpestpressure.For exam-ple,highertemperaturesmightcauseanincreaseinthenumberof generationsofinsectspeciesthatareabletoproduceseveralbroods peryear(multivoltinespecies).Thiswouldimplyanincreasein thenumberofreproductiveeventsperyear,leadingtoanincrease inpopulation,andincreasedlevelsofinfestation(Yamamuraand
Yokozawa,2002;Dukesetal.,2009).However,ifthelast
gener-ationsemergeafteracropisharvested,theycannotimpactcrop yield,andpestpopulationmightdecreaseinsizeduetotheabsence ofsuitablefood.Somepathogensareabletoinfectitshostswhen theplants areincertaindevelopmentalstages.Thismeansthat inordertomaximizetheirchanceof infection,thelife cycleof pathogenpopulationsmust beinsynchronywithhost develop-ment.Sinceclimatechangecaninfluencetherateofbothhostand pathogendevelopment,itcouldaffectthedevelopmentandimpact ofplantdiseases.Somepathogenspeciesmaybeabletomaintain theirsynchronywithtargethosttissue,andothersmaybecomeout ofsynchrony(Garrettetal.,2009).
Whileitisclearthatallthesefactorsrespondtoclimatic vari-ables,theymightbecontrolledbydifferentcombinationsofdriving factors,orrespondtotheirchangeatdifferentrates.Inorderto separatethese effects, we need tobetter assess the dynamical interactionsofthehost-pest/pathogensystem.Indeed,meaningful projectionsofclimatechangeimpactsondisease/infestation pres-surecanbeobtainedonlybycouplinghostphenologywithpatterns ofpestdevelopmentandinfestation(Grulke,2011).Atpresent,only afewmodellingstudieshaveconsideredtheseinteractionsforthe projectionofclimatechangeimpactsonagriculture(Bakeretal.,
2005;Calonnecetal.,2008;Pontietal.,2009;Gutierrezetal.,2009).
Infact,mostresearchhasconcentratedontheeffectsofclimate changeoneitherthephysiology/phenologyofsinglecrops(see,for exampleWebbetal.,2008;HallandJones,2008;Ecceletal.,2009;
CaffarraandEccel,2011)orpestsalone(see,forexamplePorter
etal.,1991;Woiwodetal.,1997;Baleetal.,2002;Salinarietal.,
2006;Estayetal.,2009).Whereasthepressureofpest/pathogenon
theirhostplantwillprobablyalsodependonfactorsotherthanthe directeffectoftemperatureontheirdevelopment,suchasgenetic adaptation,thesimulationoftheeffectsofclimatechangeontheir phenologicalinteractionisnonethelessusefultohighlightpossible trendsinfuturedisease/infestationpatterns.
Theaimofthisstudywastorefinecurrentassessmentsof cli-matechangeimpactsonpest/pathogenoccurrenceongrapevines bysimulatingpest/pathogen–hostinteractions.Thisresearch(i) combinesdetailedphenologicalmodelsofgrapevinewith pheno-logicalmodelsofoneofitskeypests(Lobesiabotrana,Den.and Schiff.,Lepidoptera:Tortricidae)andoneofitskeypathogens[E.
Fig.1. Locationofthemeteorologicalstationsusedforthestudy,thesimulation sitesandthesitesofprovenanceofthegrapevinephenology,pestandpathogen observationsusedformodelcalibrationandvalidation.Themeteorological sta-tionsusedtocalibrateandvalidatethemodelsforthesitesofCembra,SanMichele, Besagno,Cognola,FaedoandTennawereincloseproximity(lessthan1km)tothe phenology/pestanddiseasemonitoringsitesandarenotrepresentedinthemap forclarity.Betweenparenthesis():elevationabovesealevelofeachsite.Inset:The Trentoprovinceishighlightedinboldatthecentreofthesquareencompassingthe studyarea.
necator,(Schw.)Burr.],(ii)applythemodelstoclimatechange sce-nariosforaselectedstudyarea(intheeasternItalianAlps(Fig.1), and(iii)considerpotentialchangesintheinteractionsinthesetwo systems.
TheEuropeanGrapevinemoth(Lobesiabotrana)isoneofthe mostnoxiousvineyard-pestsintheEuropeanandMediterranean areas(Delbacet al.,2010).Itslarvaefeedongrapevineflowers andberries,withafacultativediapauseandavariablenumberof generationsperyear,dependingontemperatureandphotoperiod
(Pavanet al.,2010).It is usuallyreportedas beingtrivoltinein
Mediterraneanareasalthough,inthewarmestyears,afourth par-tialgenerationhasbeenreported(Torres-Vilaetal.,2004).Thefirst adultsofL.botranaappearinthespringandareshortlyfollowedby thefirstgenerationoflarvaewhichfeedoninflorescencesandbuds; inNorthernItalythisoccursbetweenMayandJune.Subsequent generationsfeedonberriesandusuallycauseconsiderable
dam-age(Moschosetal.,2004).However,thesensitivityofgrapevines
toinfestationbythispestvariesduringthegrapegrowingseason
(GabelandRoehrich,1995;Pavanetal.,2009).GabelandRoehrich
(1995)comparedthedamageproducedbylarvaeinfestationat
dif-ferentgrowthstagesondifferentgrapevinecultivarsandobserved forallstagesaperiodinwhichfructiferousorgans(flowersand berries)wereunsuitableforinfestationbyfreshlyhatchedlarvae, i.e.floweringandfruitset.Duringthis“resistant”phenological win-dow,thelevelofdamagecausedbylarvaewassignificantlylower comparedtoearlierandlatergrowthstages.
Powderymildew(E.necator)is oneof themajordiseasesin grapevine(Gadouryetal.,2003;Bendeketal.,2007;Caffietal., 2011). It affects green leaves and fruit and reduces the yield of grapes and the quality of must and wine (Gadoury et al.,
asexualcyclesduringtheyear,overwintersasmyceliumininfected budsorchasmotheciainthebarkofvines(GadouryandPearson,
1988;Cortesietal.,1995).Theprimaryinfectionsareusuallycaused
bytheascospores;afterwards,thediseaseprogressesduringthe seasonbyasexual,secondary infectioncyclesdrivenby conidia
(Carisseetal.,2009).Theseverityofthediseaseisrelatedtothe
numberofdiseasecyclesperseason,byairhumidity(a moder-atelyhighairhumiditypromotesthegerminationofconidia)and rainfall(rainpreventsgerminationofconidia)(CarrollandWilcox,
2003;Bendeketal.,2007).Asexualreproductionandrateof
epi-demicdevelopmentofpowderymildewaremainlycontrolledby temperature(Delp,1954;Sall,1980;ChellemiandMarois,1991). Thelengthofthe“latencyperiod”ofeachcycle(i.e.,thetimeperiod betweensporedepositionoftheplantsurfaceandsporulationof theresultantcolony)isthemaindriverofthenumberofdisease cycles,whichisinturnaffectedbytemperature(Calonnecetal., 2008).Thislatencyperiod is minimumwhen temperaturesare withintheoptimalrangebetween20and28◦C(Caffietal.,2011).
2. Methods
2.1. Studyareaandsites
Thestudyareaisinthepre-alpineandalpineviticulturalregions along the Adige River in the central-eastern Italian Alps, and includespartoftheprovincesofVerona,TrentoandBolzano(Fig.1). Thismountainousareaconsistsofasystemofminorvalleys con-vergingtothelargestandlongestofthem,theAdigeValley.
Viticultureis animportantsourceof revenuefor this region andformspartofitshistoricalandsocialidentity,withevidence ofvine-growingdatingbacktotheRomanperiod.Chardonnayis oneofthemostwidelygrownvarieties.IntheprovinceofTrento, atthecoreofthestudyarea,thisvarietyisgrownonabout one-thirdofthetotalviticulturalareaandisusedtoproducestilland “classic-method”,sparklingwine,called“spumante”.Commercial vineyardsofChardonnayaregrownbetweenabout100and700m a.s.l.
Forthiswork,twostudysiteswerechosenintheTrentoprovince (Fig.1).ThefirstisinSanMicheleall’Adigeandislocatedinthe Adigevalleyfloor;thesecondisinCembraandislocatedatahigher elevationona mountainslope. Thesetwo sitesare representa-tiveofthelowandmid-highelevationviticulturalenvironment, respectively.TheAdigevalleyfloorisoneofthemostprofitable viticulturalareasoftheregionbecauseofthequalityofgrapesand therelativeeaseofviticulturalpracticesduetoitsflatterrain.
2.2. Hostandpest/pathogenmodels
2.2.1. Grapevine
2.2.1.1. Modeldescription. Inordertosimulatesusceptibility win-dowsandontogenicresistancetopestswe adoptedanexisting phenologicalmodel,namedFENOVITIS(CaffarraandEccel,2010), whichsimulatesthefollowinggrowthstages:(i)budburst(stage BBCH08);(ii)flowering(stageBBCH65),and(iii)veraison(stage BBCH81).TheFENOVITISmodelconsiderstheactionofcool (chill-ing)temperaturesfordormancyrelease,andtheactionofwarm (forcing) temperaturesfor subsequent budgrowth. It describes plantdevelopment in terms of developmentalunits, calculated throughfittedorexperimentallymeasuredrelationships:chilling unitsareaccumulateduptoacriticalchillingthresholdsimulating dormancyrelease,whichisfollowedbytheaccumulationofheat (forcing)unitsuptoacriticalforcingthreshold,simulating bud-burst.Theinclusionofchillingandtheuseofanexperimentally establishedrelationshipforquantifyingtheactionofwarm tem-peraturesongrowth,makestheFENOVITISmodelcomparatively
moreprocess-basedthanthephenologicalmodelsbasedon grow-ingdegree-daysorotherbioclimaticindices.Whenthismodelwas validatedondatasetsfromfourdifferentsitesfromNorthernItaly,it yieldedsmallerpredictionerrorsthanthemodelbasedongrowing degreedays,andshowedabetterperformanceduringwarmyears, suggestingitsreliabilitywhenappliedtoclimatechangescenarios
(CaffarraandEccel,2010).
2.2.1.2. Model calibration andvalidation. In order to extendthe FENOVITISmodeltoincludethegrowthstageswhichdefinethe beginningandendoftheplantsusceptibilitytoEuropeangrapevine mothandpowderymildew,wecollecteddatathroughamonitoring surveyin2009and2010.DataontheprevalentBBCHgrowthstage werecollectedweeklyorbi-weekly,onnineChardonnayvineyards sitedatdifferentaltitudes(Fig.1),inordertocoverawideclimatic gradient. Temperature data wereobtained frommeteorological stationsclosetothevineyards(<2kmaway).Meantemperature duringtheperiodApril–September(during2000–2010)atthese sitesrangedfrom19◦CinAvioto15.8◦CinFaedo.Startingfromthe beginningofJuly,onceaweekapproximately60berries(randomly collected)weresampledineachvineyardtoanalysethesoluble solids.
Werandomlyselectedhalfofthedatacollectedduringthe mon-itoringsurveysof2009and2010tocalibratetheheatrequirements ofthegrowthstagesofinterest,fixingsimulatedbudburstasthe startingpoint.Heatunits(HU)werecalculatedusingtheFENOVITIS model,asfollows:
HU= 1
1+e−0.26(Tm−16.06) (1)
whereTm=meandailytemperature.
With this approach, wequantified theheat requirementsof BBCH phases:(i) 61 (beginningof flowering),(ii)71 (fruitset), (iii)8◦Brix,and(iv)21◦Brix(whichwasconsideredasan indica-torof the“berries ripeforharvest” phenophase,i.e.,BBCH89). Thesephase-specificheatrequirementswerecalibratedusingthe Metropolisalgorithm,alreadyusedforthispurposebyCaffarraand
Eccel(2010).TheMetropolisalgorithm(Metropolisetal.,1953),
incontrasttotraditional numericalmethodssuchasthe Down-hillsimplex andtheNewtonmethods, enablesa moreeffective explorationoftheparameterspaceinfunctionswithmanylocal “minima”, as is thecasein phenologicalmodels (Chuine et al., 1998).Theremaininghalfofthedatawasusedtovalidatethese sub-models.Theadoptedindicatorofmodelperformancewasthe meanabsoluteerror(MAE).
2.2.2. EuropeanGrapevinemoth
2.2.2.1. Modeldescription. Tosimulatethephenologyofthe Euro-pean grapevinemoth,we adoptedtheapproach of Cossuet al.
(1999),whojoinedtwosub-models,onesub-modelforthe
sim-ulationofadultflights(Arcaetal.,1993)andoneforthesimulation oflarvalstages(Cossuetal.,1999).Thesub-modelforadultflightsis basedontheaccumulationofgrowingdegree-days(GDDs),which arecalculatedusingthedoublesinemethod(Coop,2009),relying onthepreviousday’smaximumandthecurrentday’sminimum temperatureforthefirsthalfoftheday,andthecurrentday’s max-imumandminimumtemperatureofthedayforthesecondhalf oftheday.GDDswerecalculatedadoptingalowerthreshold tem-peratureof8◦Candupperthresholdof28◦C(Cossuetal.,1999), andcalculatedfromthe1stofJanuarytotheattainmentofeach flight-specificheatrequirements,estimatedbyArcaetal.(1993).
Themodelforlarvalemergenceisbasedonanempirical rela-tionshiplinkingtherateofdevelopmentofeggswithtemperature.
Thisrelationshipis expressedbytheLogancurve (Logan etal.,
1976):
v
(T)=a(eb(T−Tinf)−eb(Tsup−Tinf)−c(Tsup−T)) (2) where:v
(T)=rateofeggdevelopment;T=meandailytemperature; a,b,c,TsupandTinf=parameterstobecalibratedusingobservationsoflarvalstages.Weusedtheparameterscalibratedandvalidated
byCossuetal.(1999)usingtherateoflarvalemergenceat
differ-enttemperatures(Rapagnanietal.,1988):a=0.29737;b=0.18337; c=0.18798;Tinf=10◦C;Tsup=35◦C.
Cossuetal.(1999)linkedthemodelforlarvalstageswiththe
modelforadultflightsassumingthattherearethreedaysbetween thebeginning oftheflightandthestart ofoviposition.Because wewereinterestedinthenumberofgenerationsoccurringbefore diapause,westoppedthesimulationoflarval emergenceatthe timeoftheyearwhenphotoperiod≤12h(approximatedas21st September)asdifferentstudiesreporta criticalphotoperiodfor diapauseinductionbetween13and12h(Roehrich,1969;Deseo
etal.,1981;RoditakisandKarandinos,2001).
2.2.2.2. Modelvalidation. Weadoptedthemodelwiththe parame-tersoriginallycalibratedbyCossuetal.(1999).Inordertovalidateit forthestudyarea,weranthemodelusingtemperatureseriesfrom localmeteorologicalstationsandcomparedmodeloutputto Euro-peangrapevinemothcatchesatsixsites(381observationsintotal) locatedinthesouthernandnorthernareasoftheregion(Fig.1). Mostobservationscamefromthesouthernareasofthestudyregion (267southern observations against114northernobservations). Thetemperatureseriescamefromthemeteorologicalstationsof CaldaroforthenorthernsitesandPreaboccoandAvioforthe south-ernsites.Ofthesesixsites,fivehadobservationsduringtheperiod from2006to2009,andattheremainingsite(Caldaro)duringtwo periods:2001–2005and2008–2009.Atthefivesites,trapswere monitoredwithaweeklyfrequencyfromthefirstweekofMay tothelastweekofAugusteachyear.InCaldaro,trapswere mon-itoredwitha dailyfrequencyfromthebeginningof Junetothe endofJuly,thuscoveringonlytheperiodofthesecondgeneration flight.Inaddition,atthisstation,themonitoringsitewasmoved fromitsinitialposition(Caldaro-site1)toaslightlymoreelevated position(Caldaro-site2)after2005(Menke,2007).Thenameand thealtitudeofeachsitearegiveninFig.1.
The model was operated for the six monitoring sites; we simulatedup tofouryearly flight periods,not overlappingand takingplacefromMaytoSeptember.Thesimulatedweekly pres-ence/absenceofflightwascomparedwiththeweeklyobservations (presence/absenceofinsectsinthetraps)usingcontingencytables. Predictedflightperiods(P)werecomparedwithobservedflight periods(O);allthepossiblecombinationsofPvs.Owereorganized ina2×2contingencytablewiththefollowingcategories:(i)true positives(TP),whenthemodelcorrectlypredictstheoccurrenceof flight;(ii)truenegatives(TN),whenthemodelcorrectlypredicts theabsenceofflight;(iii)falsepositives(FP),whenthemodel pre-dictsflightoccurrencebutnoadultsweretrappedinthatperiod; (iv)falsenegatives(FN),whenthemodelpredictsflightabsencebut aflightisobserved.Weadoptedthe“overallaccuracy”(ameasure ofthepercentageofcorrectlyclassifiedinstances)asameasureof modelperformanceasfollows:
Overallaccuracy=(TP+TN)/(TP+TN+FP+FN) (3) Sensitivity,specificityandlikelihoodratiosofthemodel pre-dictionswereevaluatedbymeansofBayesiananalysis(Yuenand
Hughes,2002).
2.2.3. Powderymildew
2.2.3.1. Modeldescription. Amodelforpowderymildewepidemics inthestudyareawasdevelopedbasedontherelationshipbetween:
(i) theseverity of powdery mildewepidemics on bunchesand (ii) the number of infection cycles occurring during the pow-derymildew-susceptiblegrowthstagesofthegrapevineplants,i.e. betweenbudburstand8◦Brix(Delp,1954;ChellemiandMarois,
1991).
DiseaseseveritywasassessedatweeklyintervalsbetweenMay andAugust,in2002–2008,intwountreatedChardonnayvineyards inthestudyarea,atPressano,onahill-slope,andZambana,inthe valleyfloor(Fig.1).Theobservationsreportedthepercentageof dis-easedarea(i.e.,diseaseseverity),whichwasvisuallyestimatedon 100randomleavesfromMaytoharvest(BBCH89),and100random bunchesfromflowering(BBCH65)untilharvest(BBCH89).Disease assessmentsonbuncheswereaveragedtoprovideweeklydisease severityforthetwosites.
Thenumberofcyclesbetweenbudburst(simulatedusingthe FENOVITISmodel)andAugustwascalculatedbasedonthelength ofthelatencyperiodofE.necator,usingtheequationdevelopedby
Caffietal.(2011)asapartofamechanisticmodelforthe
occur-renceofprimaryinfections;themodelofCaffietal.(2011)was validatedinItaly,innorthern(i.e.PiedmontandLombardy), cen-tral(i.e.MarcheandEmiliaRomagna),andsouthernregions(i.e. Basilicata).Thisequationwasdevelopedbyfittingexperimental dataontherelationshipbetweentemperatureanddurationofthe latencyperiod(Delp,1954;Analytis,1980;ChellemiandMarois,
1991;Calonnecetal.,2008)byapolynomialinsuchawaythat
latencydurationisminimalattheoptimaltemperatureof22–27◦C
(PearsonandGoheen,1988;Fernandez-Gonzalezetal.,2009;Caffi
etal.,2011)andincreasesatsub-optimaltemperaturesaboveor belowthisrange.Indetail,thedailyprogressoflatency(LATP)is calculatedasfollows:
LATPi= LAT1
i;LATi=+ Ti−Ti
2 (4)
with=44.7; =0.067;=3.244;T=meandailytemperature(◦C); i=dayoftheyear.
Starting frombudburst, LATPi wasaccumulateddaily;when
LATPi≥1,sporulationstartsonthepowderymildewcoloniesand
anewdiseasecyclestartsover.
Thediseasemodelisbasedonalogisticequation,inthe follow-ingform:
DS= c
1+ea−bNC (5)
whereDS=diseaseseverityonbunches;NC=numberofinfection cycles;a,b,c=modelparameters.
2.2.3.2. Modelcalibration. ParametersofEq.(4)wereestimatedfor theepidemicswhichdevelopundertwoscenariosofconduciveness forthedisease:(i)highconducivenessand(ii)low-intermediate conduciveness.The“diseaseconduciveness”conceptaccountsfor allthevariablesaffectingthediseasedevelopment,otherthanthe weather(e.g., level of overwintered inoculum, training system, vinevigour),whichwerenotspecificallyconsideredinthiswork. Thediseaseseveritydataoftheobservedvineyardswithafinal diseaseseverity≥95%wereusedforthefirstscenario,while vine-yardswithafinaldiseaseseveritybetween26.7and64.3%were usedforthesecondscenario.Onlydiseaseseverityobservations takenongrapevinebuncheswasusedtodevelopthemodel.The model parametersfor thetwo scenarios wereestimatedbythe non-linearregressionprocedureofSPSS(ver.15,SPSSInc.),which minimizestheresidualsumsofsquaresusingtheMarquardt algo-rithm.Thefollowing wereusedasindicatorsofgoodness-of-fit: themagnitudeof thestandarderrorsof themodelparameters, thecoefficientofdeterminationadjustedforthedegreesof free-dom,thenumberofiterationsrequiredbytheMarquardtalgorithm to converge on parameter estimates, and the magnitude and
Table1
Parameterestimates(heatunits)andperformanceoftheextendedFENOVITISmodel.Mean,Max,MinandSD:mean,maximum,minimumvaluesandstandarddeviations oftheobservedandsimulatedphenophasedates(dayoftheyear).N(cal):numberofobservationsinthecalibrationdataset.N(val)numberofobservationsinthevalidation dataset.MAE(cal):meanabsoluteerrorofthemodelpredictionsonthecalibrationdataset(internalMAE);MAE(val):meanabsoluteerrorofthemodelpredictionsonthe validationdataset(externalMAE);BBCH61:beginningofflowering;BBCH71:endofflowering/fruitset;8◦Brix:attainmentof8◦Brix;21◦Brix:attainmentof21◦Brix.
Submodel Observations Simulations Heatunits N(cal) N(val) MAE(cal)(days) MAE(val)(days)
Mean Max Min SD Mean Max Min SD
BBCH61 152.8 161 147 4.9 153.4 164 148 5.5 21.59 4 5 0.9 2.1
BBCH71 165 176 160 5.2 167.1 182 161 6.7 32.03 4 5 1.0 2.5
8◦Brix 211.9 230 200 8.7 210.5 233 198 9.6 67.61 9 9 5.1 1.9
21◦Brix 245.8 269 235 10.0 248.4 292 230 17.6 95.71 9 9 2.7 3.9
distributionofresiduals(datanotshown).Theestimated parame-tersforhighandlow-intermediateconducivenessscenarioswere: (i)a=9.119±1.861,b=0.998±0.207, c=0.98±0.015(R2=0.93); (ii)a=14.632±2.067,b=1.214±0.257,c=0.49±0.118(R2=0.95). Theindicatedasymptoticstandarderrors(±)accountforthe vari-ability in estimating the model parameters with a non-linear regressionprocedure.
2.3. Meteorologicaldataandtemperatureprojections
Meteorologicaldata(dailyminimumandmaximum tempera-ture)touseasinputsinthegrapevineandpest/pathogenmodel were obtained from four different weather station networks belonging to: theAgenzia Regionale per la Prevenzione e Pro-tezioneAmbientaleoftheVenetoregion(Veronaprovince);the ServizioMeteorologicoof theautonomousprovinceof Bolzano, the Südtiroler Beratungsring für Obst- und Weinbau (Bolzano province);andtheFondazioneE.Mach(Trentoprovince).Weather stationswerelocatedbetween0and5kmfromthemonitoringsites (Fig.1).
Forthesimulationoffutureclimate,weusedstatistically down-scaled temperature series based on the output of the Hadley Centre’sAtmosphere-OceanGeneralCirculationModel(AOGCM) HADCM3(Popeetal.,2000).Thecomparisonofthemodel out-puts from the CMIP – Coupled Model Intercomparison Project
(http://cmip-pcmdi.llnl.gov/)shows that HADCM3stemperature
projectionisintheaverageoffivebenchmarkmodelsfrom2000 totheyears2040sforA2scenario,andslightlyincreasesabovethe groupaverageafterthefirsthalfofthecentury.Thedownscaling procedureusedinthisworkwaspreviouslydescribedbyCaffarra
andEccel(2010).Themodeloutputhadaresolutionof2.5◦
lati-tudeby3.75◦longitude.Thestatisticaldownscalingalgorithmwas adaily-resolution“transferfunctionmethod”(Ecceletal.,2009), appliedseriesbyseries.
Weconsidered theA2and B2scenarios, downscaledfor the stationsof SanMichele(representativeofthevalley floor envi-ronment) and Cembra (representative of the mountain slope environmentathigherelevation)fortheperiod1990–2080.These twoscenarios arefromIPCC’s “SpecialReport onEmission Sce-narios”–SRES(NakicenovicandSwart,2000)andrepresenttwo intermediatehypothesesontheincreaseinconcentrationof green-house gases (GHG). Whereas both scenarios show a moderate temperatureincreasefromtodayto2050,afterwards,theA2 sce-nariowarmsupabruptlywhiletheB2scenariomaintainsaconstant butmoderatetemperatureincrease.TheA2storylineandscenario familydescribesaveryheterogeneousworld,whoseemissionrates ingeneraldonotstaywithintheboundsofanyself-regulation.The underlyingthemeisself-relianceandpreservationoflocal iden-tities.Economicdevelopmentisprimarilyregionallyorientedand percapitaeconomic growthand technologicalchangearemore fragmented.The B2 storyline and scenario family assumes the adoptionofself-imposedstandardsofreductionintheemissions. Itdescribesaworldinwhichtheemphasisisonlocalsolutionsto
economic, social,andenvironmentalsustainability.Itis aworld with continuouslyincreasing global populationat a lower rate thaninA2,intermediatelevelsofeconomicdevelopment,andless rapidandmorediversetechnologicalchangethanintheB1andA1 storylines.Attheendofthesimulationperiod,meanannual tem-peraturesinthestudyregionareprojectedtoexceedpresentvalues byabout2and4.5◦C,respectivelyintheB2andA2scenarios.
2.4. Analysis
First,weevaluatedtheprojectionsforEuropeangrapevinemoth andpowderymildew;second,wecombinedthemwithgrapevine phenologyprojections,andfinallyweconsideredtheinteraction amongthepest,thediseaseandthehost.Themodelprojections werevisuallyevaluatedusingscatterplots.Ten-yearmoving aver-ageswereappliedtotheprojectedseriestovisuallyassesstemporal trends.Todeterminetheoverallsignificanceoftrends,wefitted linearmodelstothesimulateddataovertheprojectionperiod. Cen-soredregressionwasappliedwhereverthevariableofinterestwas observableonlywithinadefinedintervalsuchasforpercentage
data(Schnedler,2005).
Inordertodefinethepercentageoftemporaloverlapbetween the Europeangrapevinemothlarvae-resistant growthstages of grapevine(floweringandfruitset,BBCH61to71)andthemoth larval stage, we considered the number of days during which both larvae and the larvae-resistant growth stages occurred (co-occurrence days) and calculated the percentage of overlap as: ((number of co-occurrence days)/(duration of the resistant phenophase))×100.
Thenumberoflarvalgenerationsbeforeharvestwascalculated adopting21◦Brixasanindicatorof harvest.When,accordingto thesimulations,thisstageoccurredinthemiddleofalarval gen-eration (i.e.afterthe onsetof larvae,but beforetheirend), we calculatedthepercentportionoflarvalgenerationoccurringbefore harvestas:((numberofdaysfromthebeginningoflarval emer-genceto21◦Brix)/(totaldurationoflarvalgeneration))×100.This ratioexpressedthetime lapseduring whichlarvaeco-occurred withripeberriesasafractionofthetotaldurationofthelarval generation.
Forpowderymildew,weproducedprojectionsofdisease sever-ity for both low-intermediate and high disease conduciveness scenarios and evaluatedtheimpact oftemperature increaseon diseaseepidemicsinrelationwithgrapevinephenology.
3. Results
3.1. Validationofpestandhostmodels
Inthestudyareafloweringusuallyoccursatthebeginningof Juneandlastsaround10–14days(periodbetweenbeginningof floweringandfruitset)dependingontheelevation.Ripeningoccurs betweentheendofAugust(valleyfloor,onwarmyears)andthe endofSeptember(mountainslopes,oncoolyears)(Table1).
Table2
OverallaccuracyandlikelihoodratiosoftheEuropeangrapevinemothmodelpredictions(adultflights)usingobservationsfromsixvineyardsinthestudyarea.The observationperiodrangesbetween2001and2009.
Observed Predicted Total Overallaccuracya Likelihoodratio(LR) Priorprobability(P) Posteriorprobability(P)
Yes(P+) No(P−)
Yes(O+) 222(0.9)b 33(0.1)c 255 0.80 LR(O+)2.6 P(O+)0.67 P(O+,P+)0.84 P(O+,P−)0.28
No(O−) 43(0.3)d 83(0.7)e 126 LR(O−)0.2 P(O−)0.33 P(O−,P+)0.16 P(O-,P−)0.72
Total 265 116 381
a(TP+TN)/(TP+TN+FP+FN).
bTruepositiveproportion,TPP(sensitivity). cFalsenegativeproportion,FNP.
dFalsepositiveproportion,FPP(minusthespecificity). eTruenegativeproportion,TNP(specificity).
Table3
Ratesofchangeinthetimingofphenophasesoverthesimulationperiod(dayyr−1)(slope),R2ofthelinearregressionmodelfittedtothesimulatedphenologicalseriesand meanphenophasedates±standarddeviation,simulatedoverthedecades1991–2000and2071–2080,foreachsiteandscenario.
Site Scenario Phenophase Slope R2 Meanphenophasedate
1991–2000 2071–2080 San Michele A2 BBCH08 −0.091 0.102* 101.0±8.7 94.2±7.8 BBCH61 −0.181 0.343** 152.8±7.6 135.2±5.8 BBCH71 −0.207 0.399** 168.5±6.4 148.8±6.0 8◦Brix −0.268 0.565** 213.8±7.2 188.4±6.2 21◦Brix −0.374 0.623** 250.8±9.5 220.2±6.0 B2 BBCH08 −0.026 0.010 100.9±5.9 96.6±7.3 BBCH61 −0.153 0.268** 153.9±3.1 137.5±4.9 BBCH71 −0.158 0.289** 168.3±3.1 152.3±4.4 8◦Brix −0.290 0.489** 92.0±3.9 72.7±4.4 21◦Brix −0.290 0.499** 252.2±6.0 226.1±4.8 Cembra A2 BBCH08 −0.096 0.127** 100.9±6.6 92.7±7.7 BBCH61 −0.186 0.368** 149.4±8.1 130.6±6.6 BBCH71 −0.237 0.453** 166.7±6.7 143.4±6.6 8◦Brix −0.335 0.644** 213.8±7.2 186.8±6.6 21◦Brix −0.658 0.709** 272.3±15.5 218.1±7.4 B2 BBCH08 −0.032 0.017 102.0±5.4 95.4±4.3 BBCH61 −0.182 0.294** 156.3±4.4 134.8±4.1 BBCH71 −0.193 0.318** 171.9±5.0 151.8±4.5 8◦Brix −0.245 0.477** 219.2±5.3 169.0±4.8 21◦Brix −0.468 0.528** 270.6±2.5 228.2±4.6 *p<0.01. **p<0.001.
The predictingperformanceof thesub-modelsforgrapevine phenologicalsusceptibility was considered satisfactory as they yieldedMeanAbsolute Errors(MAEs)that werelowerthanthe monitoringresolution,whichwas7daysforphenophase8◦Brix
and21◦Brixand 3–4daysfor BBCH61 and71 (Table 1).There
was a reasonable agreement between simulated and observed numberoftotalgenerationsfortheperiod1991–2010,asinthe studyareausually 2–3generationsof mothsareobserved each
year (Anfora et al., 2007) similarto the mean 2.2–2.8
genera-tionssimulatedinCembraandSanMicheleoverthesameperiod.
TheEuropeangrapevinemothmodelshowedhighoverall accu-racy(80%),sensitivity(0.9),andspecificity(0.7).Thisimpliesthat itwasabletocorrectlypredictflightsandabsenceofflightsmost ofthetime,asconfirmedbythelargepositivelikelihoodratio(2.6) andthesmallnegativelikelihoodratio(0.2).Themodelshowed a slightly better performance when predicting flight presence [P(O+,P+)=0.84]thanflightabsence[P(O−,P−)=0.72](Table2).The probabilityofmispredictingtheoccurrenceofflightsisalso rea-sonablylow(P(O+,P−)= 0.28).Whereasmostobservationscame fromthesouthernareasofthestudyregion(267southern obser-vationsagainst114northernobservations),theoverallaccuracy ofthemodelwassimilarwhencalculatedseparatelyforthetwo regions (88% vs. 74%, for southern and northern observations, respectively).
3.2. Projections
3.2.1. Grapevine
Allgrowthstagesshowedatrendtowardsearlieroccurrence, butweremorepronouncedinsummer(i.e.,attainmentof8and 21◦Brix)andforthemountainsiteofCembra.Ingeneral,the pro-jectedratesofadvancewerestrongerfortheA2scenario,dueto itslargerprojectedtemperatureincrease(Table3).Thesetrends wereallhighlysignificantexceptforbudburst(BBCH08)fortheB2 scenarioatbothsites.Whencomparingthe1990swiththe2070s, simulatedharvestdatesadvancedby25days(SanMichele,B2)up to32days(Cembra,A2).Themeandurationofflowering(BBCH61 to71)atthebeginningofthesimulatedperiodwasabout15–16and 15.5–17daysinSanMicheleandCembra,respectively(B2andA2 scenarios,respectively),butdecreasedonlyslightlyintheB2 sce-nario,to13.5and16daysinSanMicheleandCembra,respectively (meandurationforthe2070s).Duringthisperiod,thebeginningof floweringadvancedbybetween14and23days(SanMichele,B2 andCembra,A2),correspondingtoashiftfromthefirstweekof Junetothesecond–thirdweekofMay.
3.2.2. Europeangrapevinemoth
Whereasboththetotalnumberoflarvalgenerationsandthe numberoflarvalgenerationsbeforeharvest(pre-harvest genera-tions)increasedoverthesimulationperiod,thistrendwasclearly
Fig.2.Modelsimulationsoftotal(blackpoints)andpre-harvest(greypoints)larvalgenerationsaccordingtoscenariosA2andB2forthetwositesunderstudy:SanMichele (lowerelevation),Cembra(higherelevation).10-yearmovingaveragelinesaresuperimposedontheseries.Thelegendforallplotsisdisplayedatthebottomleftoftheright plot.
more pronounced in the first than in the second case(Fig. 2). Regressionanalysisconfirmedthesignificanceoftheseincreasing trendsforbothsitesand scenarios,withratesofincrease rang-ing from 0.05 generations/decade in San Michele– B2to 0.18 generations/decadeinCembra–A2(totalgenerations)andfrom 0.03generations/decadeinSanMicheleB2andCembraB2to0.05 generations/decadeinSanMichele–A2(pre-harvestgenerations)
(Table4).Atthebeginningoftheprojectedperiodthetotal
num-beroflarvalgenerationsandthenumberofpre-harvestgenerations wereoftenthesame,duetoarelativelylateharvestdateoccurring arounddiapauseinduction(simulatedon21stSeptember). Nev-ertheless,astemperatureincreasedoverthecentury,harvestdate waspredictedprogressivelyearlierwhereasthepredictedtimingof thephotoperiod-drivendiapauseinductiondidnotchange, result-inginfewerpre-harvestthantotalgenerations.Thesiteshowing thehighestincreaseinpre-harvestgenerationswasSanMichelein theA2scenario(from2.5inthe1990sto2.9inthe2070s),whereas
Table4
Ratesofchangeinthenumber oftotalandpre-harvest larvalgenerationsof Europeangrapevinemothoverthesimulationperiod(1991–2080)(generations decade−1),slope,R2andp-valueofthelinearregressionmodelfittedtothesimulated seriesforeachsiteandscenario(allwithp<0.001).
Scenario Totalgenerations Pre-harvestgenerations
Slope R2 Slope R2
SanMichele A2 0.15 0.694 0.05 0.584
B2 0.05 0.508 0.03 0.343
Cembra A2 0.18 0.714 0.04 0.228
B2 0.11 0.564 0.03 0.273
Cembrashowedonlysmallincreasesinbothscenarios(from2.3 and2.2inthe1990sto2.4and2.4inthe2070s,forA2andB2, respectively)(Table5).
The temporaloverlap betweenresistantplantgrowthstages andfirst-generationlarvaeoccurrenceshowedahighinterannual
Table5
Europeangrapevinemoth:meannumberoftotalandpre-harvestgenerations±standarddeviation,inthreesampledecadesatthebeginning,middleandendoftheprojection period(1991–2080).
1991–2000 2021–2030 2071–2080
Totalgenerations Pre-harvestgenerations Totalgenerations Pre-harvestgenerations Totalgenerations Pre-harvestgenerations
SanMichele A2 2.79±0.28 2.54±0.17 3.1±0.19 2.69±0.07 3.95±0.10 2.94±0.07
B2 2.69±0.14 2.68±0.11 2.79±0.12 2.73±0.08 3.63±0.28 2.88±0.09
Cembra A2 2.33±0.33 2.26±0.21 2.70±0.25 2.48±0.17 3.66±0.28 2.44±0.08
Fig.3.AnnualpercentagesoftemporaloverlapbetweenEuropeangrapevinemothlarvaeandlarvae-resistantphenophasesingrapevinesaccordingtoscenariosA2andB2 forthetwositesunderstudy:SanMichele(lowerelevation)andCembra(higherelevation).10-yearmovingaveragelinesaresuperimposedontheseries.
variability at both sites and for both scenarios, ranging from total(100%overlap) tonone (0%overlap).However, while San Micheleshoweda clear overalltrend of decreasingoverlap for both scenarios, Cembra did not, especially in the B2 scenario (Fig. 3).In fact, censoredregression analysisconfirmed signifi-cantnegativetrendsforthis variableonlyatSanMichele(both scenarios)(Table6).Thispatternisalsoshownbythedecrease inthemeanpercentagesoftemporal overlapbetweenresistant growth stages and first-generation larvae, for San Michele in both scenarios (from 51 to13% and from 39 to25% fromthe 1990s to 2070s, for A2 and B2, respectively) contrasting with the lack of a clear trend in Cembra in the B2 scenario (non-significantincreasefrom71to86%fromthe1990stothe2070s).
Table6
Ratesofchangeinthepercentageofoverlapbetweenlarvaeandthelarvae-resistant phenophaseoverthesimulationperiod(1991–2080)(%overlapyr−1)(slope)andR2 ofthecensoredlinearregressionmodelfittedtothesimulatedseriesforeachsite andscenario. Scenario Slope R2 SanMichele A2 −0.543 0.393* B2 −0.232 0.090** Cembra A2 −0.111 0.031 B2 0.075 0.017 *p<0.001. **p<0.05. Table7
Powderymildew:meannumberofcyclesanddiseaseseverity±standarddeviation,inthreesampledecadesatthebeginning,middleandendoftheprojectionperiodfor thehighandlow-intermediatediseaseconduciveness(cond.)scenarios.
1991–2000 2021–2030 2071–2080
N◦cycles Highcond. (%)
Low-intermediate cond.(%)
N◦cycles Highpressure (%)
Low-intermediate cond.(%)
N◦cycles Highcond. (%) Low-intermediate cond.(%) San Michele A2 14.78±0.27 97.64±0.09 47.21±0.52 14.66±0.28 97.59±0.13 46.94±0.75 13.72±0.52 96.84±0.84 42.60±4.51 B2 14.97±0.23 97.70±0.07 47.58±0.41 14.73±0.20 97.62±0.08 47.14±0.44 14.14±0.29 97.31±0.25 45.24±1.50 Cembra A2 13.40±0.19 96.59±0.26 40.93±1.53 13.53±0.28 96.74±0.29 41.82±1.72 12.98±0.37 95.79±0.85 36.67±4.23 B2 14.12±0.21 97.31±0.14 45.28±0.85 14.02±0.22 97.24±0.17 44.82±1.03 13.57±0.28 96.79±0.35 42.11±20.8
Fig.4. ModelsimulationsofthenumberofpowderymildewinfectioncyclesaccordingtoscenariosA2andB2forthetwositesunderstudy:SanMichele(lowerelevation) (blacksymbols),Cembra(higherelevation)(greysymbols).10-yearmovingaveragelinesaresuperimposedontheseries.Thelegendforbothplotsisdisplayedatthebottom leftoftherightplot.
3.2.3. Powderymildew
Atbothsites,thedurationofthesusceptibilitywindowto pow-derymildew(i.e.,BBCH08to8◦Brix)decreasedbetween20and25 daysbetweenthe1990sandthe2070s(fortheB2andA2scenarios, respectively),mainlyasaneffectofanearlier8◦Brixphenophase, asbudburstadvancedsignificantlybyaboutoneweekonlyinthe A2scenario,forbothSanMicheleandCembra.Projectionsshowed adecreaseinthenumberofpowderymildewcyclesastemperature increasedoverthesimulationperiod.Thistrendwasclearand sig-nificant(p<0.001) at both sitesand for both scenarios, though morepronounced for theA2 scenario(R2 rangingfrom0.25in CembraB2to0.45SanMicheleA2)(Fig.4,Table8).Thesite show-ingthelargestdecreasewasSanMichelefortheA2scenario(from 14.8inthe1990sto13.7inthe2070s).InCembratheeffectof cli-matechangeonpowderymildewwaslessnoticeable,forboththe A2(from13.4inthe1990sto13inthe2070s)andB2scenarios (from14.1inthe1990sto13.6inthe2070s)(Table7).
Simulationsofdiseaseseverityshowedsignificantdecreasing trendsinpowderymildewepidemicsoverthecentury(Fig.5and
Table8).However,whereasinahighdiseaseconduciveness
sce-nario,thedecreaseindiseaseseveritywasminimal(forexample inSanMichelefortheA2scenarioitdecreasedfrom97.6%to96.8% fromthe1990stothe2070s),inthelow-intermediate conducive-nessscenario,changesweremorenoticeable(inSanMichelefor theA2scenarioitdecreasedfrom47.2%to42.6%fromthe1990s tothe2070s).Cembrashowedasimilartrend,withapronounced decreaseindiseaseseverity,especiallyinthelow-intermediate dis-easeconducivenessscenario,andintheA2scenario(from40.9%in the1990sto36.7%inthe2070s).
4. Discussion
4.1. Europeangrapevinemoth
Thereiswidespreadconcernthatthepredictedfuture warm-ingwillincreasethepressureofinsectpestsanddiseasesoncrops
(Porteretal.,1991;Estayetal.,2009;Olesenetal.,2011).Increased
temperaturesandearlieronsetofthegrowingseasonmayreduce wintermortality,increasetherateofinsectmetabolismand devel-opment(Baleetal.,2002),andincreasethenumberofgenerations ofmultivoltinespecies(Laˇst ˚uvka,2009).Recentworksshowthat thesechangesarealreadytakingplace(Battistietal.,2005;Raffa etal.,2008),andimportantinsect-pestslikethegrapevinemoth areproducingadditionalbroodscomparedtopastdecades(Pavan
etal.,2006;Martin-Vertedoretal.,2010).Forexample,in
south-ernSpaintheextraordinarilyhightemperaturesrecordedinthe spring-summerof2006resultedinEuropeangrapevinemoth com-pletingfourcompletegenerationsasopposedtotheusualthree
(Martin-Vertedoretal.,2010).Inagreementwiththese
observa-tions, inourmodellingprojections risingtemperaturesresulted in a markedincrease inthemeannumber of generations com-pletedbythismoth.However,whenweconsideredthenumberof pre-harvestlarvalgenerationsasopposedtothetotalnumberof larvalgenerations, theincreasingtrend wasnotsopronounced. Morespecifically,thediscrepancybetweenthesetwovariableswas initiallyminimal,butbecamelargerovertheprojectedperiod,most likelyduetotheirdifferentenvironmentalcontrols.Infact,while thesimulationofthetotalnumberofgenerationswithinayearwas drivenonlybytemperatureandphotoperiod,simulationof pre-harvestgenerationsalsodependedonharvesttime,whichwasin turndependentontemperatureduringripening.Increasing tem-peraturesresultedinamarkedadvanceinsimulatedharvestdates, butdidnotalterthetimingofdiapauseinduction,whichis con-trolledbyphotoperiod.TheseresultssuggestthatforChardonnay, apotentialincreaseinthepressurebyEuropeangrapevinemoth maybebalancedbyanearlierharvestandlesstimeforlarvaeto producedamage.Inagreementwiththiseffect,Martin-Vertedor
etal.(2010)notedthatthefourcompletegenerationsrecordedin
thewarm2006inSpainwerefollowedbyafifthincompleteadult generationafterharvest,whichpresumablydiedwithout produc-ingadditionaldamagenorviableprogeny.Ontheotherhand,in viticulturalareaswherebothlateandearlyripeningvarietiesare grown,thismightleadtohigherpressureonlaterripeningrather thanearlyripeningvarieties.Toassessthispossibility,amodelling approachincludingdifferentgrapevinevarietieswithinaspecific areawouldbeuseful.
Theprojectionofthetimingofresistantplantgrowthstagesin conjunctionwiththeoccurrenceoffirstgenerationlarvaemadeit possibletoassesscurrentandfuturepatternsofplant–pest inter-actions.Accordingtothesimulations,currently,atallsitesandfor allscenariosthereisaconsiderabletemporaloverlapbetween lar-vaeandtheflowering–fruitsetwindowin Chardonnay.These resultsaresupportedbyfieldobservationsfromthestudyarea. However,thisoverlapmightdecreaseinthefutureatwarmer,low elevationsitessuchasSanMichele.Infact,whileincreased temper-atureresultedinasignificantlyearliermothphenology,itadvanced grapevinephenologytoalesserextent,resultinginadecreaseinthe synchronybetweenlarvaeandresistantphenologicalstages.The
Fig.5.Modelsimulationsofpowderymildewseverity(%)inascenariooflow-intermediateconduciveness(blacksymbols)andhighconduciveness(greysymbols)for thediseaseaccordingtoscenariosA2andB2forthetwositesunderstudy:SanMichele(lowerelevation),Cembra(higherelevation).10-yearmovingaveragelinesare superimposedontheseries.Thelegendforallplotsisdisplayedinthebottomleftoftherightplot.
sametrendwaspredictedforCembra,thoughitwasnotstatistically significantinthesimulationperiodandwasonlydetectedinthe warmerA2scenario.Thissimulationsuggeststhatinthewarmer, more profitable viticultural areas of the study region, increas-ingtemperaturemighthavea detrimentalimpactoncropyield duetoincreasedasynchronybetweenthelarvae-resistantgrowth stagesofgrapevineandlarvaeoccurrence.Today,inwarmerareas suchassouthernItalyorSpain,Europeangrapevinemothadults arereportedtoemergeonemonthearlierthaninnorthernItaly, inlateMarch–early April,whilelarvaeappear1–2weeks
after-ward(Moleas, 2005). However, Chardonnay flowering in these
areasoccursonly2–3weeksearliercomparedtonorthernItaly, sothereisashortertemporaloverlapbetweenlarvaeoccurrence andresistantgrowthstagesofgrapevine.Thissituationresembles
theonedepictedbythemodelsimulations.However,itisdifficult toextrapolatecurrentpatternsofEuropeangrapevinemoth infes-tationsinareaslikeSicilytothefuturesituationofnorthernItaly becauseofthedifferenceinlocalfloras.Infact,thesemothsare generalistsintheirareaofprovenance(theMediterraneanbasin), and laytheireggsand feedonavariety ofhost-plants, includ-ingCompositae,Convulvolaceae,Oleaceae,Rhamnaceae,Rosaceae, andVitaceae(StavridisandSavopoulou-Soultani,1998).Thus,in Mediterraneanareasthevariouslocalplantspeciesmightdivert the first generation larvae from grapevines. A study by Pavan
etal.(2009)reports thatintheVenetoregion(anareacloseto
thestudy area, but slightly warmer),levels of infestationwere higher on very earlyvarieties with hairless inflorescences, like ChardonnayandPinotgris.However,innoneofthevarietiestested
Table8
Ratesofchangeinpowderymildewdiseaseseverityoverthesimulationperiod(1991–2080)(%overlapyr−1),slopeandR2ofthecensoredlinearregressionmodelfittedto thesimulatedseriesforeachsiteandscenario(p<0.001inallcases).
Scenario Highcond. Low-intermediatecond. Numberofcycles
Slope R2 Slope R2 Slope R2
San Michele A2 −0.009 0.002 −0.053 0.008 −0.013 0.447 B2 −0.004 0.001 −0.026 0.004 −0.009 0.379 Cembra A2 −0.014 0.002 −0.075 0.012 −0.008 0.304 B2 −0.006 0.001 −0.034 0.007 −0.006 0.254
didlarvaeclustersexceedtheeconomicinjury level(100larval nestsper100clusters).Climatewarmingmightexacerbate this vulnerabilityinsomeareas,likethealpineregion,wherethereare fewnaturalhostsofthispestandgrapevinesarethemainfood source.
4.2. Powderymildew
Thedurationofthelatencyperiodisparticularlyimportantfor powderymildew,whoseinfectiondoesnotstrictlydependon par-ticularenvironmentalconditions,likerainfallofthepresenceoffree wateronplantorgans.Infectionbythispathogenisvirtually con-tinuousandthereforetheepidemicprogressionmostlydepends onthedailydoseofinoculumabletoinfectthehost,which in turndependsonhowmanyinfectioncycleshavebeencompleted, leadingtosporulatingpowderymildewcolonies(Xu,1999).
Themorepronounceddecreaseinthemeannumberof infec-tion cycles at the low elevation site can be related to the temperatureresponseoflatencyduration,ratherthantoa differ-enceinhostsusceptibilityduration(showinga similardecrease acrosssites). Latency duration is minimal at optimal tempera-turesbetween20and28◦C,andincreasesattemperaturesbelow or above this range. Therefore, as the climate of San Michele (lowerelevation)is currentlyintheoptimaltemperaturerange forpowderymildewdevelopmentduringthesusceptibleperiod of grapevine (spring–summer), a temperature increase would inevitably increase thelatency duration. On theother hand,at the higher elevations of Cembra, a temperature increase from itscurrentlycooler conditionsdoesnotimply suchalargeshift fromoptimaltemperatures.Predictedseverityofpowderymildew decreasedmoreinthelow-intermediatethaninthehighdisease conducivenessscenario,duetothedifferenteffectthatthedecrease ofmeannumberofannualcycleshasondiseaseseverityinthetwo scenarios(from13.5–14.8to13.0–13.7intheA2scenario,for Cem-braandSanMichele,respectively,andfrom14.1and15to13.6 and14.1intheB2scenarioforCembraandSanMichele, respec-tively):whileinthehighlyconducivescenarioanearlystartofthe epidemicleadstohighestdiseaseseverityfromthe12thto13th cycleonwards,inthelow-intermediatescenariothemaximum pre-dictedseverityoccursfromthe14thto15thcycleonwards.The factthatthedecreaseinsimulateddiseaseseverityishigherfor theA2thantheB2scenario,butissimilaracrosssitesis counter-intuitive,giventheabovementioneddifferenceincycledecreases acrosssites(SanMicheleshowingalargerdecreasethanCembra). Thiseffectistheresultofthenon-linearityandinteractionbetween thedifferentmodelsused:inthiscasethehigherdamagedecrease inA2isrelated,atbothsites,tothedecreaseindiseasecycles occur-ringincorrespondencewithafasterdecreaseindiseaseseverity (aroundtheinflectionpointofthediseaseseverityfunction).
Theseresultssuggestthatclimatechangemightdecreasethe severityofpowderymildew,especiallyduringextremeyearswith particularlyhightemperatures, asthose predictedby themore pessimistic(“business-as-usual”)A2scenario,andinthescenario withlow-intermediateconducivenessfor thedisease.Similarly,
Calonnecetal.(2008)modelledtheinteractionbetweengrapevine
andpowderymildewandfoundalowerimpactofpowderymildew ongrapevine ona particularlywarm year (2003)compared to an average year (1998). This situation might occur more fre-quentlyiftemperaturesincreasefurther,decreasingoverallmildew severity.
4.3. Conclusions
The multiple interactions among pests, diseases and plants make it necessary toconsiderthem jointly as a system,rather thanseparateelements(Grulke,2011).Thisviewfindssupportin
thepresentsimulations,whichshowastrongeffectofhost-plant developmentonpestanddiseasepressureandsuggestthatcurrent interactions betweenhost-pestandhost-diseasemaybealtered by climate change. Simulations suggested that in the warmer, more profitable viticultural areas of the study area, increasing temperature might have a detrimental impact on susceptibil-ity to European grapevine moth due to increased asynchrony betweenthelarvae-resistantgrowthstagesofgrapevinesand lar-vae. On the other hand, the increase in pest pressure due to theincreasednumber ofgenerations mightnot beassevereas expectedonthebasisofthepest-modelonly,duetotheadvance inharvestdateswhichwouldlimitdamagesoflatergenerations, especiallyforearlyripeninggrapevinevarietieslikeChardonnay. Simulations for powdery mildew showed a decrease in simu-lated disease severity especially in years with a later onset of theepidemics(assumingalow-intermediatedisease conducive-ness) and in the climate scenario with a higher temperature increase.
Whilethepresent simulationsofferinformationonpotential impactsofclimatechangeonviticulture,theyshouldbetakenwith cautionduetothechallengeofnotbeingabletoaccountforallof thefactorsaffectingthehost-pestandhost-diseasedynamicsthat mightchangeinthefuture.Plantpathogensaregenerallyhighly adaptableandlikelytoexploitanycompromiseinplantdefence caused byclimatechange.Instabilityofvariationis commonin manypathogens,althoughthemechanismsareoftennotknown
(Gregoryetal.,2009).Manyreportsstressthatpathogensshowan
increasedabilitytomutateandgeneratevariantsunderstressed conditions(Hastingsetal.,2000;Twissetal.,2005).Inparticular, powderymildewisabletoreproducebothsexuallyandasexually producingahighnumberofspores,whichincreasesthechances toproducestrainsadaptedtochangedclimaticconditions.In par-ticular,E.necatorcouldadaptbydevelopingashorterdurationof latencyathighertemperatures,orbyincreasingasexual sporula-tion,whichwould inbothcasesresultinanincreaseindisease severity.Alsoanearlierascosporereleaseadaptedtoanearlierbud breakofvinescouldledtoagreaternumberofinfectioncyclesthan thosepredictedinthiswork.
Similartopathogens,insectsaregenerallyadaptableorganisms
(Bradshawand Holzapfel, 2008;Laˇst ˚uvka, 2009).Whileclimate
changemightresultinadisruptionoftheirsynchronywiththe host,adaptiveprocessesarelikelytoquicklyrestorethissynchrony
(Robinet and Roques,2010).For example, in Japan,Gomiet al.
(2007)reportedadecreaseinthelengthofthecritical
photope-riodfordiapauseinductioninthefallwebworm(Hyphantriacunea) whichenabledthisspeciestoproduceoneadditionalbroodperyear andexploitwarmertemperatures.
Indeed, predictingpotentialconsequencesof climatechange ontheinteractionbetweentrophiclevelsisacomplicatedmatter becauseofthenumberoffactorsinvolved.Arangeof environmen-talfactorsnotdirectlyconsideredbythemodelsemployedinthis studymight affecttrophicinteractions inthefuture. For exam-ple,elevatedCO2concentrations,suchasthosepredictedinthe
future,causeadecreaseinleafnitrogenandincreasecarbohydrates andphenolics,withpotentialeffectsontheinteractionbetween insectherbivoreandplants(BezemerandJones,1998).Changes in precipitationand humiditymightalsoimpact thephenology andpopulationdynamicsofbothpest/pathogenandhost.Inorder tomodeltheseinteractions,weneedtoimproveour understand-ingofthewholesystem,throughexperimentsandobservations takingintoaccountdifferenttrophiclevels.Similarly,modelling studiesaimedatpredictingtheimpactsofclimatechangeon agri-culturalecosystemsshouldintegratealltheseresponses(Hoover
andNewman,2004;Pontietal.,2009)toobtainamorerealistic
pictureofallpotentialeffectsandbetterassistthedevelopmentof mitigationpolicies.
Acknowledgements
The authorswish tothankGianfranco Anfora, FabioZottele, LenCoop,FriedrichMenke,Francesco Penner,MaurizioBottura, FrancescoFellinfordatasupply,adviceandproofreading.
References
Analytis,S.,1980.Obtainingofsub-modelsformodelingtheentirelifecycleofa pathogen.J.PlantDis.Prot.87,371–382.
Anfora,G.,Schmidt,S.,Ioratti,C.,2007.Morfologiaebiologia.In:Anfora,G.,Angeli, G.,Baldessari,M.,Ioriatti,C.,Marchesini,E.,Mattedi,L.,Menke,F.,Mescalchin, E.,Schmidt,S.,Tasin,M.,Varner,M.(Eds.),Letignoledellavite.,pp.22–31. Arca, B., Cossu, A., Delrio, G., Locci, L., 1993. Individuazione dei gradi
giorno relativi allo sviluppo della Lobesia botrana (Den. Et Schiff.) in Sardegna. Atti Convegno Nazion. Protezione delle colture: osser-vazioni, previsioni, decisioni, Pescara 7–8 Ottobre, pp. 325–334 (http://www.sar.sardegna.it/pubblicazioni/notetecniche/nota3/pag018.asp). Bale,J.,Masters,G.,Hodkinson,I.,Awmack,C.,Bezemer,T.,Brown,V.,Butterfield,
J.,Buse,A.,Coulson,J.,Farrar,J.,Good,J.E.G.,Harrington,R.,Hartley,S.,Jones, T.H.,Lindroth,R.L.,Press,M.C.,Symrnioudis,I.,Watt,A.D.,Whittaker,J.B.,2002. Herbivoryinglobalclimatechangeresearch:directeffectsofrisingtemperature oninsectherbivores.Glob.Chang.Biol.8,1–16.
Baker,R.H.A.,Sansford,C.E.,Gioli,B.,Migletta,F.,Porter,J.R.,Ewert,F.,2005. Combin-ingadiseasemodelwithacropphenologymodeltoassessandmappestrisk: Karnalbuntdisease(Tilletiaindica)ofwheatinEurope.In:PlantProtectionand PlantHealthinEurope:IntroductionandSpreadofInvasiveSpecies,Symposium ProceedingsNo.81(pp.89–94).BCPCAlton,Hampshire,UK.
Battisti,A.,Stastny,M.,Netherer,S.,Robinet,C.,Schopf,A.,Roques,A.,Larsson,S., 2005.Expansionofgeographicrangeinthepineprocessionarymothcausedby increasedwintertemperatures.Ecol.Appl.15,2084–2096.
Bradshaw,W.E.,Holzapfel,C.M.,2008.Geneticresponsetorapidclimatechange: it’sseasonaltimingthatmatters.Mol.Ecol.Jan.17,157–166.
Bendek,C.,Campbell,P.,Torres,R.,Donoso,A.,Latorre,B.A.,2007.Therisk assess-ment indexingrapepowderymildewcontroldecisions andtheeffectof temperatureandhumidityonconidialgerminationofErysiphenecator.SJAR 5,522–532.
Bezemer,T.M.,Jones,T.H.,1998.Plant-insectherbivoreinteractionsinelevated atmosphericCO2:quantitativeanalysesandguildeffects.Oikos82,212–222. Caffarra,A.,Eccel,E.,2010.IncreasingtherobustnessofphenologicalmodelsforVitis
viniferacvChardonnay.Int.J.Biometeorol.54,255–267.
Caffarra,A.,Eccel,E.,2011.Projectingtheimpactsofclimatechangeonthe phenol-ogyofgrapevineinamountainarea.Aust.J.GrapeWineRes.17,52–61. Caffi,T., Rossi,V.,Leger,S.E.,Bugiani,R.,2011. Amechanistic model
simulat-ingascosporicinfectionsbyErysiphenecator,thepowderymildewfungusof grapevine.PlantPathol.60,522–531.
Campbell,P.,Bendek,C.,Latorre,B.A.,2006.Riskofpowderymildew(Erysiphe neca-tor)outbreaksongrapevineinrelationtoclusterdevelopment.Cien.Inv.Agr. 34,1–6.
Calonnec,A.,Cartolaro,P.,Naulin,J.M.,Bailey,D.,Langlais,M.,2008.Ahost–pathogen simulationmodel:powderymildewofgrapevine.PlantPathol.57,493–508. Carisse,O.,Bacon,R.,Lefebvre,A.,Lessard,K.,2009.Adegree-daymodeltoinitiate
fungicidesprayprogramsformanagementofgrapepowderymildew(Erysiphe necator).Can.J.PlantPathol.31,186–194.
Carroll,J.E.,Wilcox,W.F.,2003.Effectsofhumidityonthedevelopmentofgrapevine powderymildew.Phytopathology93,1137–1144.
Chakraborty,S.,Tiedermann,A.V.,Teng,P.S.,2000.Climatechange:potentialimpact onplantdiseases.Env.Poll.108,317–326.
Chellemi,D.O.,Marois,J.J.,1991.Developmentofademographicgrowthmodelfor Uncinulanecatorbyusingamicrocomputersreadsheetprogram. Phytopathol-ogy81,250–254.
Chuine,I.,Cour,P.,Rousseau,D.D.,1998.Fittingmodelspredictingthedatesof floweringoftemperate-zonetreespeciesusingsimulatingannealing.PlantCell Environ.21,455–466.
Coop,L.,2009.IntegratedPlantProtectionCenteratOregonStateUniversityandthe WRIPMCenters.http://uspest.org/cgi-bin/ddmodel.pl(accessed2009). Cortesi,P.,Gadoury,D.M.,Seem,R.C.,Pearson,R.C.,1995.Distributionand
reten-tionofcleistotheciaofUncinulanecatoronthebarkofgrapevines.PlantDis.79, 15–19.
Cossu,Q.A.,Delrio,G.,DiCola,G.,Gilioli,G.,1999.Modellimatematicinella pro-tezioneintegratadellecoltureinSardegna.CollanadiAgrometeorologiaperla Sardegna,NotaTecnica3.
Delbac,L.,Lecharpentier,P.,Thiery,D.,2010.Larvalinstarsdeterminationforthe EuropeanGrapevineMoth(Lepidoptera:Tortricidae)basedonthefrequency distributionofhead-capsulewidths.CropProt.29,623–630.
Delp,C.J.,1954.Effectoftemperatureandhumidityonthegrapepowderymildew fungus.Phytopathology44,615–626.
Deseo,K.V.,Brunelli,F.M.A.,Bertaccini,A.,1981.Observationsonthebiologyand diseasesofLobesiabotranaDenandSchiff.(Lepidoptera:Tortricidae)in central-northItaly.ActaPhytopathol.Hung.16,405–431.
Dukes,J.S.,Pontius,J.,Orwig,D.,etal.,2009.Responsesofinsectpests,pathogens, andinvasiveplantspeciestoclimatechangeintheforestsofnortheasternNorth America:whatcanwepredict?Can.J.For.Res.39,231–248.
Easterling,W.E.,Aggarwal,P.K.,Batima,P.,Brander,K.M.,Erda,L.,Howden,S.M., Kirilenko,A.,Morton,J.,Soussana,J.-F.,Schmidhuber,J.,Tubiello,F.N.,2007.Food, fibreandforestproducts.In:Parry,M.L.,Canziani,O.F.,Palutikof,J.P.,vander Linden,P.J.,Hanson,C.E.(Eds.),ClimateChange2007:Impacts,Adaptationand Vulnerability.ContributionofWorkingGroupIItotheFourthAssessmentReport oftheIntergovernmentalPanelonClimateChange.CambridgeUniversityPress, Cambridge,UK,pp.273–313.
Eccel,E.,Rea,R.,Caffarra,A.,eCrisci,A.,2009.Riskofspringfrosttoapple produc-tionunderfutureclimatescenarios:theroleofphenologicalacclimation.Int.J. Biometeorol.53,273–286.
Estay,S.A.,Lima,M.,Labra,F.A.,2009.Predictinginsectpeststatusunderclimate changescenarios:combiningexperimentaldataandpopulationdynamics mod-elling.J.Appl.Entomol.133,491–499.
Fernandez-Gonzalez,M.,Rodriguez-Rajo,J.,Jato,V.,Aira,M.J.,2009.Incidenceof fungalsinavineyardofthedenominationoforiginRibeiro(Ourense– North-WesternSpain).Ann.Agric.Environ.Med.16,263–271.
Fuhrer,J.,2003.Agro-ecosystemresponsetocombinationsofelevatedCO2,ozone, andglobalclimatechange.Agric.Ecosyst.Environ.97,1–20.
Gabel,B.,Roehrich,R.,1995.Sensitivityofgrapevinephenologicalstagestolarvae ofEuropeangrapevinemoth,LobesiabotranaDen.etSchiff.(Lep.,Tortricidae). J.Appl.Entomol.119,127–130.
Gadoury,D.M.,Pearson,R.C.,1988.Initiation,development,dispersalandsurvival ofcleistotheciaofUncinulanecatorinNewYorkvineyards.Phytopathology78, 1413–1421.
Gadoury,D.M.,Seem,R.C.,Pearson,R.C.,Wilcox,W.F.,Dunst,R.M.,2001.Effectsof powderymildewonvinegrowth,yield,andqualityofconcordgrapes.PlantDis. 85,137–140.
Gadoury,D.M.,Seem,R.C.,Ficke,A.,Wilcox,W.F.,2003.Ontogenicresistanceto powderymildewingrapesberries.APS93,5.
Gomi,T.,Nagasaka,M.,Fukuda,T.,Hagihara,H.,2007.Shiftingofthelifecycleand life-historytraitsofthefallwebworminrelationtoclimatechange.Entomol. Exp.Appl.125,179–184.
Gregory,P.J.,Johnson,S.N.,Newton,A.C.,Ingram,J.S.I.,2009.Integratingpests andpathogensintotheclimatechange/foodsecuritydebate.J.Exp.Bot.60, 2827–2838.
Garrett,K.A.,Nita,M.,DeWolf,E.D.,Gomez,L.,Sparks,A.H.,2009.Plantpathogens asindicatorsofclimatechange.In:ClimateChange:ObservedImpactsonPlanet Earth.Elsevier,pp.425–437.
Grulke,N.E.,2011.Thenexusofhostandpathogenphenology:understandingthe diseasetrianglewithclimatechange.New.Phytol.189,8.
Gutierrez,A.P.,Ponti,L.,Cossu,Q.A.,2009.EffectsofclimatewarmingonOliveand olivefly(Bactroceraoleae(Gmelin))inCaliforniaandItaly.ClimaticChange95, 195–217.
Hall,A.,Jones,G.,2008.Effectofpotentialatmosphericwarmingon temperature-baedindicesdescribingAustralianwinegrapegrowingconditions.Aust.J.Grape WineRes.14,1–23.
Hastings,P.J.,Bull,H.J.,Klump,J.R.,Rosenberg,S.M.,2000.Adaptiveamplification: aninduciblechromosomalinstabilitymechanism.Cell103,723–731. Hoover,J.K.,Newman,J.A.,2004.Tritrophicinteractionsinthecontextofclimate
change:amodelofgrasses,cerealAphidsandtheirparasitoids.Glob.Change Biol.10,1197–1208.
Kang, Y.H.,Khan, S., Ma, X.Y., 2009. Climate changeimpacts oncrop yield, crop waterproductivityand foodsecurity– A review.Prog.Nat.Sci. 19, 1665–1674.
Laˇst ˚uvka,Z.,2009.Climatechangeanditspossibleinfluenceontheoccurrenceand importanceofinsectpests.Plant.Prot.Sci.45,53–62.
Logan,J.A.,Wollkind,D.J.,Hoyt,S.C.,Tanigoshi,L.K.,1976.Ananalyticalmodel fordescriptionoftemperaturedependentratephenomenainarthropods.Env. Entomol.5,1133–1140.
Maracchi,G.,Sirotenko,O.,Bindi,M.,2005.Impactsofpresentandfutureclimate variabilityonagricultureandforestryinthetemperateregions:Europe.Climatic Change70,117–135.
Martin-Vertedor,D.,Ferrero-Garcia,J.J.,Torres-Vilas,L.M.,2010.Globalwarming affectsphenologyandvoltinismofLobesiabotranainSpain.Agric.For.Entomol. 12,169–176.
Menke,F.,2007.Unesempiopratico:lalottaalletignoledellaviteinAltoAdige.In: Anfora,G.,Angeli,G.,Baldessari,M.,Ioriatti,C.,Marchesini,E.,Mattedi,L.,Menke, F.,Mescalchin,E.,Schmidt,S.,Tasin,M.,Varner,M.(Eds.),Letignoledellavite., pp.32–39.
Metropolis,N.,Rosenbluth,A.V.,Rosenbluth,M.N.,Teller,A.H.,Teller,E.,1953. Equationofstatecalculationsbyfastcomputingmachines.J.Chem.Phys.21, 1087–1092.
Moleas,T.,2005.Letignoledellavite:notiziebioetologicheetecnichedicontrollo. In:Ragusa,S.,Tsolakis,H.(Eds.),Ladifesadellavitedagliartropodidannosi.,pp. 97–147.
Moschos,T.,Souliotis,C.,Broumas,T.,Kapothanassi,V.,2004.ControloftheEuropean grapevinemothLobesiabotranainGreecebythematingdisruptiontechnique: athree-yearsurvey.Phytoparasitica32,83–89.
Nakicenovic,N.,Swart,R.,2000.SpecialReportonEmissionScenarios.Cambridge UniversityPress,UK,pp.570–579.
Olesen, J.E., Bindi, M., 2002. Consequences of climate change for Euro-pean agricultural productivity, land use and policy. Eur. J. Agron. 16, 239–262.
Olesen,J.E.,Trnka,M.,Kersebaum,K.C.,Skjelvåg,A.O.,Seguin,B.,Peltonen-Sainio,P., Rossi,F.,Kozyra,J.,Micale,F.,2011.ImpactsandadaptationofEuropeancrop productionsystemstoclimatechange.Eur.J.Agron.34,96–112.
Panter,S.N.,Jones,D.A.,2002.Age-relatedresistancetoplantpathogens.Adv.Bot. Res.38,251–280.
Pavan,F.,Floreani,C.,Barro,P.,Zandigiacomo,P.,DallaMontà,L.,2010.Influence ofgenerationandphotoperiodonlarvaldevelopmentofLobesiabotrana.Env. Entomol.39,1652–1658.
Pavan,F.,Stefanelli,G.,Cargnus,E.,Villani,A.,2009.Assessingtheinfluenceof inflo-rescencetraitsonthesusceptibilityofgrapetovinemoths.J.Appl.Entomol. 133,394–401.
Pavan,F.,Zandigiacomo,P.,DallaMontà,L.,2006.Influenceofthegrape-growing areaonthephenologyofLobesiabotranasecondgeneration.Bull.Insectol.59, 105–109.
Pearson,R.C.,Goheen,A.C.,1988.CompendiumofGrapeDiseases.A.P.S.
Ponti,L.,Cossu,Q.A.,Gutierrez,A.P.,2009.ClimatewarmingeffectsontheOlea europea–BactroceraoleaesysteminMediterraneanislands:Sardiniaasan exam-ple.Glob.ChangeBiol.15,2874–2884.
Pope,V.,Gallani,M.L.,Rowntree,P.R.,Stratton,R.A.,2000.Theimpactofnewphysical parameterizationsintheHadleyCentreclimatemodel:HadAM3.Clim.Dyn.16, 123–146.
Porter,J.H.,Parry,M.L.,Carter,T.R.,1991.Thepotentialeffectsofclimaticchangeon agriculturalinsectpest.Agric.For.Meteorol.57,221–240.
Raffa,K.F.,Aukema,B.H.,Bentz,B.J.,Carroll,A.L.,Hicke,J.A.,Turner,M.G.,Romme, W.H.,2008.Cross-scaledriversofnaturaldisturbancespronetoanthropogenic amplification:thedynamicsofbarkbeetleeruptions.Bioscience58,501–517. Rapagnani,M.R.,Caffarelli,V.,Barlattini,M.,1988.Lobesiabotrana(Den.etSchiff.):
studioinlaboratoriodelciclodisviluppoinfunzionedellatemperatura.In:Atti XVCongr.Naz.Ital.Entomol.L’Aquila,pp.973–980.
Robinet,C.,Roques,A.,2010.Directimpactsofrecentclimatewarmingoninsect populations.Integr.Zool.5,132–142.
Roditakis,N.E.,Karandinos,M.G.,2001.Effectsofphotoperiodandtemperatureon pupaldiapauseinductionofgrapeberrymothLobesiabotrana.Physiol.Entomol. 26,329–340.
Roehrich,R.,1969.Ladiapausedel’EudémisdelaVigneLobesiabotranaSchiff.(Lép.: Tortricidae):inductionetélimination.Ann.Zool.Ecol.Anim.1,419–431.
Salinari,F.,Giosue,S.,Tubiello,F.N.,Rettori,A.,Rossi,V.,Spanna,F.,Rosenzweig,C., Gullino,M.L.,2006.Downymildew(Plasmoparaviticola)epidemicsongrapevine underclimatechange.GlobalChangeBiol.12,1299–1307.
Sall,M.A.,1980.Epidemiologyofgrapepowderymildew:amodel.Phytopathology 70,338–342.
Schnedler,W.,2005.Likelihoodestimationforcensoredrandomvectors.Econ.Rev. 24,195–217.
Stavridis,D.G.,Savopoulou-Soultani,M.,1998.Larvalperformanceonand oviposi-tionpreferenceforknownandpotentialhostsbyLobesiabotrana(Lepidoptera: Tortricidae).Eur.J.Entomol.95,55–63.
Thomson,S.V.,2000.Epidemiologyoffireblight.In:Vanneste,J.L.(Ed.),FireBlight theDiseaseanditsCausativeagent,Erwiniaamylovora.CABIPublishing, Walling-ford,UK,pp.9–36.
Torres-Vila,L.M.,Rodriguez-Molina,M.C.,McMinn,M.,Rodriguez-Molina,A.,2004. Larvalfoodsourcepromotescyclicseasonalvariationinpolyandryinthemoth Lobesiabotrana.OxfordJ.16,114–122.
Twiss,E.,Coros,A.M.,Tavakoli,N.P.,Derbyshire,K.M.,2005.Transpositionis mod-ulatedbyadiversesetofhostfactorsinEscherechiacoliandisstimulatedby nutritionalstress.Mol.Microbiol.57,1593–1607.
Webb,L.,Whetton,P.,Barlow,E.W.R.,2008.Climatechangeandwinegrapequality inAustralia.Clin.Res.36,99–111.
Xu,1999.Effectsoftemperatureonthelatentperiodoftherosepowderymildew pathogen,Sphaerothecapannosa.PlantPathol.48,662–667.
Yamamura,K.,Yokozawa,M.,2002.Predictionofageographicalshiftinthe preva-lenceofricestripevirusdiseasetransmittedbythesmallbrownplanthopper, Laodelphaxstriatellus(Fallen)(Hemiptera:Delphacidae),underglobalwarming. App.Entomol.Zool.37,181–190.
Yuen,J.,Hughes,G.,2002.Bayesiananalysisofplantdiseaseprediction.PlantPathol. 51,407–412.
Woiwod,I.,1997.DetectingtheeffectsofclimatechangeonLepidoptera.J.Insect. Conserv.1,149–158.
