Assessinggeneticvariationingrowthanddevelopment

ofpotato

MuhammadSohailKhan

Thesiscommittee

Thesissupervisors

Prof.dr.ir.P.C.Struik

ProfessorofCropPhysiology,WageningenUniversity

Prof.dr.ir.F.A.vanEeuwijk

ProfessorofAppliedStatistics,WageningenUniversity

Thesisco‐supervisors

Dr.X.Yin

Assistantprofessor,CentreforCropSystemsAnalysis,WageningenUniversity

Dr.ir.H.J.vanEck

Assistantprofessor,LaboratoryofPlantBreeding,WageningenUniversity

Othermembers

Prof.dr.ir.M.K.vanIttersum,PlantProductionSystems,WageningenUniversity

Dr.J.‐P.Goffart,WalloonAgriculturalResearchCentre,Gembloux,Belgium

Prof.dr.ir.A.J.Haverkort,PlantResearchInternational(PRI),WageningenUR

Dr.ir.E.Heuvelink,HorticulturalSupplyChains,WageningenUniversity

Thisresearchwasconductedundertheauspicesof theC.T.deWitGraduateSchool

forProductionEcologyandResourceConservation.

Assessinggeneticvariationingrowthanddevelopment

ofpotato

MuhammadSohailKhan

Thesis

submittedinfulfilmentoftherequirementsforthedegreeofthedoctor

atWageningenUniversity

bytheauthorityoftheRectorMagnificus,

Prof.dr.M.J.Kropff,

inthepresenceofthe

ThesisCommitteeappointedbytheAcademicBoard

tobedefendedinpublic

on20September2012

at11a.m.intheAula.

MuhammadSohailKhan(2012)

Assessinggeneticvariationingrowthanddevelopmentofpotato

245pages

PhDThesis,WageningenUniversity,Wageningen,theNetherlands

Withreferences,withsummariesinEnglishandDutch

ISBN978–94–6173–359–7

Abstract

Khan,M.S., 2012.Assessinggeneticvariation ingrowthanddevelopmentofpotato.

PhDthesis,WageningenUniversity,Wageningen,theNetherlands,withsummariesin

EnglishandDutch,245pp.

Due to increasing food demand and changing diets potato (Solanum tuberosum L.) is

becoming a subsistence crop in many regions. However, the agronomy and whole crop

physiologyoftuberyieldproductionisextremelycomplex,duetogenotypeandenvironment

specific effects on crop physiological and morphological characteristics. Such intrinsic

complexities complicate the manipulation of yield determining traits and make their

prediction a challenging task. The genetic improvement of tuber yield can be understood

moremechanisticallyby investigatingand interpreting the relationshipsbetween themain

attributesofcropgrowth.

This thesisaimstodevelopanapproachtoquantify theyieldof individualgenotypes

and to estimate parameters which may reveal the effects of genetic and environmental

factorsontheimportantplantprocessescontrollingtuberyieldvariationamongalargesetof

F1 (SH83‐92‐488 RH89‐039‐16) genotypes of potato and a set of standard cultivars

coveringawiderangeofmaturitytypes.

Itfirstpresentsamodelapproachtoanalysethetimecourseofcanopycoverandtuber

bulkingduringtheentirecropcycleasafunctionofthermaltimeintermsoflargenumberof

physiological component traits and explain their inter‐relationships and impact on crop

maturity and tuber yield production across six contrasting field experiments. The results

indicatedthatthelengthofthecanopybuild‐upphase(DP1)wasconservativewithrespectto

genotype’smaturity type,but thedurationofmaximumcanopycover (DP2)and thedecline

phase(DP3)variedgreatly,withlatergenotypeshavinglongerDP2andDP3andthusahigher

areaunderwholegreencanopycurve(Asum).Valuesoftuberbulkingrate(cm)werehighest

for earlymaturing genotypes followed bymid‐late and then late genotypes. Latematuring

genotypeshadlongesteffectivedurationoftuberbulking(ED)followedbymid‐lateandearly

genotypes.Asaresulttuberyield(wmax)washigherinlategenotypesthaninearlygenotypes.

Theradiationuseefficiency(RUE)valueswerehighestforearlymaturinggenotypesfollowed

bymid‐lateandlategenotypeswhereasnitrogen(N)useefficiency(NUE)washighestinlate

maturinggenotypesfollowedbymid‐earlyandearlygenotypes.

High genetic variability and high heritability for most of these traits were found.

Resultsindicatedthatincreasedtuberyieldbyindirectselectionforoptimalcombinationof

important physiological traits can be achieved. While using these traits as a criterion for

selection, the causal physiological relationships and trade‐offs must be considered

simultaneously.

Our molecular dissection of traits determining the dynamics of canopy cover, tuber

bulking, and resource (radiation, nitrogen) use efficiencies identified several QTLs, the

mappingpositionofeach identifiedQTL, the interactionofQTLwithenvironment(QTLE)

and the magnitude of the QTL effect in explaining genetic variance in both SH and RH

parental genomes. The QTL results indicated that one particular chromosomal position at

18.2cMonpaternal(RH)linkagegroupVwastightlylinkedtothegenotype’searlinessand

controlling nearly all the traits and explaining thephenotypic variance byup to79%.This

suggestedthepleiotropicnatureoftheQTLformostofthetraitsdeterminingcropmaturity

andtuberyields.AnumberofQTLsfortraitswerenotdetectedwhentuberyieldpersewas

subjectedtoQTLanalysis.ThephenotypicvarianceexplainedbytheQTLsfortuberyieldper

sewasalsolowerthanforothertraits.

The physiological and quantitative knowledge gained was used to evaluate the

conventionalsystemofmaturitytypeandtoquantifyandre‐definetheconceptofmaturity

type on a physiological basis for a large set of genotypes. Four new physiological based

maturity criteria were developed based on four canopy cover and tuber bulking traits.

Physiologicalmaturity typecriteria tended todefinematurityclasses lessambiguouslyand

were easily and clearly interpretable compared to the conventionally used method of

definingmaturity.

Thecapabilityofanecophysiologicalmodel‘GECROS’wastestedtoanalysedifferences

in tuber yield of potato. Themodel yielded a reasonably good prediction of differences in

tuber yield across environments and across genotypes. Model analysis identified the

genotypickey‐parametersaffectingtuberyieldproductionandNmax(i.e.totalcropNuptake)

contributedmost to thedeterminationof tuberyield.Theresultsconcludedthatgenotypes

withhigherNmaxandlowertuberNcontentexhibitedhighertuberdrymatteryield.Further

analysis of the genotypic parameters should be performed in conjunction with molecular

markersinordertodeterminetheirgeneticcontrolandtoproceedtowardsQTL‐basedcrop

modellingapproach.

Thisthesisidentifiedthedominantcomponenttraitsmostlyinvolvedintheformation

of a tuber yield and gave insight into the possibilities of genetically and physiologically

manipulating the size or number of such traits. The information obtained should help in

marker‐assisted selection as well as in designing ideotypes for specific and/or diverse

environments.However, tomake significant contributions forbreeding, there is aneed for

furtherresearcheffortstoevaluatethecombinedphysiologicalandgeneticapproach.

Keywords:Potato(SolanumtuberosumL.),segregatingpopulation,canopycoverdynamics,

tuberbulkingdynamics,beta function, thermal time, componentsofvariance,genotype‐by‐

environment interaction (GE), heritability, QTL mapping, QTL‐by‐environment (QTLE)

interaction, complex traits, ecophysiological cropmodel, GECROS, cultivar choice,maturity

type,ideotypebreeding,tuberyield.

Dedicatedtomybelovedparents&

highlyesteemedsupervisors

Contents

Chapter1: Generalintroduction 1

Chapter2: Analysinggeneticvariationinpotato(SolanumtuberosumL.)

using standard cultivars and a segregating population. I.

Canopycoverdynamics

13

Chapter3: Analysisofgeneticvariationofpotato(SolanumtuberosumL.)

using standard cultivars and a segregating population. II.

Tuberbulkingandresourceuseefficiency

83

Chapter4: Model‐based evaluation of maturity type of potato using a

diverse set of standard cultivars and a segregating diploid

population

119

Chapter5: Anecophysiologicalmodelanalysisofyielddifferenceswithin

an F1 segregating population of potato (Solanum tuberosum

L.)grownunderdiverseenvironments

137

Chapter6: Generaldiscussion 159

References 191

Summary 219

Samenvatting(SummaryinDutch) 225

Acknowledgments 231

Curriculumvitae 241

PE&RCPhDEducationCertificate 243

Funding 245

Abbreviations Tb Basetemperature(˚C)

To Optimumtemperature(˚C)

Tc Ceilingtemperature(˚C)

tm1 Inflexionpointduringcanopybuild‐upphase(td)

t1,DP1 Period fromplant emergence tomaximumcanopy cover (moment and

durationsinceemergence,respectively)(td)

t2 Onsetofcanopycoversenescence(td)

te Timeofcompletesenescenceofcanopycoverorcropmaturity(td)

vmax Maximumlevelofcanopycover(%)

cm1 Maximumcanopygrowthrateduringbuild‐upphase(%td‐1)

c1 Averagecanopygrowthrateduringbuild‐upphase(%td‐1)

c3 Averagecanopysenescencerateduringdecline‐phase(%td‐1)

DP2 Totaldurationofmaximumcanopycoverphase(td)

DP3 Totaldurationofcanopysenescencephase(td)

A1 AreaundercanopycurveforDP1(td%)

A2 AreaundercanopycurveforDP2(td%)

A3 AreaundercanopycurveforDP3(td%)

Asum Areaunderwholegreencanopycurve(td%)

tB Onsetoflinearphaseoftuberbulking(td)

tE Endoflinearphaseoftuberbulking(td)

ED Totaldurationoftuberbulking(td)

cm Rateoftuberbulking(gtd‐1)

wmax Maximumtuberdrymatteratcropmaturity(gm‐2)

RUE Radiationuseefficiency(gDMMJ‐1)

NUE Nitrogenuseefficiency(gDMg‐1N)

mV Periodfromplantemergencetoonsetoftuberbulking(td)

mR Durationbetweenonsetoftuberbulkingandcropmaturity(td)

Hmax Maximumplantheight(m)

nSO MaximumtuberNconcentration(gNg‐1DM)

Nmax TotalcropNuptakeatcropmaturity(gNm‐2)

td Thermalday

cM centiMorgans

CHAPTER1

Generalintroduction IncreasingsustainableyieldAmajoragriculturalchallenge facing theworld today isprovidingsufficient foodto

meet thedemandsof a rapidly growingpopulation.The goal of anyplant breeding

programme is the development of new, improved cultivars or breeding lines for

particulartargetareasandtoincreasethesustainableyieldandqualityofcropplants

to meet projected increases in global food demand (FAO, 1996). This normally

involvesmanipulatingcomplextraitsassociatedwithplantgrowthanddevelopment,

usuallyinproductionenvironmentsthatarehighlyvariableandunpredictable.

Genetic improvementcanbeachievedbyselectingdirectly foraprimary trait

(suchasyield)inatargetenvironment(CeccarelliandGrando,1996).Thishasbeen

referredtoasempiricalor traditionalbreeding.Another‐ indirectway ‐ is toselect

forasecondarytrait(s)thatisputativelyrelatedwithahigheryieldpotentialand/or

totheimprovedbehaviorofthecropwhengrowninaparticularenvironment.Thisis

knownasanalyticalorphysiologicalbreeding.

Over thepastcentury,mostof theprogress inbreeding inmostcropsspecies

hasbeenderivedfromempiricalbreeding,whichhastakenyieldasthemaintraitfor

selection(Austinetal.,1989;Khush,1987;Russell,1993;Evans,1993;Slaferetal.,

1994).Forinstance,incereals,yieldincreaseshavebeenpossiblethroughthegradual

replacement of traditional tall cultivars by semi dwarf and fertiliser‐responsive

varietieswith superiorharvest indices.The term “GreenRevolution”was coined to

describethisprocess.

However, most agronomic traits are physiologically and genetically complex

(Larketal.,1995;Orfetal.,1999;DaniellandDhingra,2002;Stuberetal.,1999).For

instance, yield is the outcome ofmany underlyingmorphological and physiological

processes (Bezant et al., 1997). Furthermore, it is a quantitatively inherited trait

underpolygeniccontrol,andcharacterisedby lowheritabilityandahighgenotype‐

by‐environment(GE)interaction(AllardandBradshaw,1964;Jacksonetal.,1996;Tardieu, 2003). Such intrinsic complexities complicate the manipulation of

quantitativecroptraitsandmaketheirpredictionachallengingtask.Thereforenew

approache(s) which could complement traditional with analytical selection

methodologies to further improve crop yields and the overall efficiency of the

breedingprocessareofconsiderableinteresttoplantbreeders.

Chapter1

2

GenomicsbasedapproachesRecentadvancesinagriculturalgenomicsandmarker‐assistedselectionhavetotally

revolutionisedandenhancedtheplantbreedingprocess(SomervilleandSomerville,

1999;Yinetal.,1999a;Huangetal.,2002;Knight,2003).TheadventofDNA‐based

molecularmarkers(Botsteinetal.,1980;Patersonetal.,1988)andnewstateofthe

art statisticalmethods (Lander andBotstein, 1989; Jansen, 1993; Jansen and Stam,

1994) havemade it possible to localise polygenes controlling complex quantitative

traits and study the effect of individual genes on the trait, which was difficult in

classicalquantitativegenetics.

TheacronymQTLstandsforQuantitativeTraitLocus,withpluralQTLs,andhas

come to refer to a genomic region associated with a quantitative trait (Yin et al.,

1999a).Numerousstudieshavebeenreportedon identificationofQTLs forvarious

quantitativetraits inhumans,animals,andplants(Yinetal.,1999a).Theincreasing

knowledgeonQTLsforimportantagronomictraitscanprovidenewavenuesforthe

breederstospeeduptheprocessofincreasingcropyields(MorandiniandSalamini,

2003).

AlthoughQTLmappingtechnologieshaveallowedsignificantadvancesincrop

genetic improvement, there still exist challenges to increase the yield potential of

crops (Yin et al., 2003a). One major difficulty in the use of QTLs of traits is their

instability across environments (Reymond et al., 2003) due to complex QTL‐by‐

environment(QTLE)interactions.Forinstance,RibautandHoisington(1998)found13QTLsinastudyoffloweringdatesofmaize,butonlythreewerecommontothree

experimentswithdifferent levelsofwaterdeficit. Consequently,manyQTLs canbe

onlydetectedinanarrowrangeofenvironmentalconditions(Zhouetal.,2007)and

theclassicalgeneticmodelsbuiltfromQTLanalysishaveacorrectpredictiveability

onlyinalimitedrangeofconditions.

PopularQTL‐analysismethodswillhavedifficultieshandlingthesecomplexities.

Without increasing population sizes, limited reliability in QTL detection and

estimationcanbeachieved(KearseyandFarquhar,1998).Toovercomethesebottle

necks, support from other disciplines is required. There is an identified need to

separate factors influencing a given phenotypic trait and shifting from highly

integrated traits tomore gene‐related traits (Yin et al., 2002). Several studieshave

underlinedthepotentialinterestinbuildingsuchalink(Hammeretal.,2002,2006;

Tardieu, 2003; Yin et al., 2004). Functional genomics, systems biology, and crop

physiologycanjointlyimprovegeneticanalysisandbreedingefficiencies.

RoleofsystemsapproachesinmodelbasedbreedingThereisincreasingadvocacyfortheapplicationofcropphysiologicalknowledgeand

integrativemodellinginbreedingforcomplextraits(Camposetal.,2004;Edmeades

Generalintroduction

3

et al., 2004; Yin et al., 2004; Hammer et al., 2005). Crop modelling can play a

significant part in system approaches by providing a powerful capability for

phenotypepredictionandyieldscenarioanalysis(Hammer,1998;Cooperetal.,2002;

Hammeretal.,2004).

Crop growth models are simplified mathematical representations of the

interacting biological and environmental components of the dynamic plant system.

Ecophysiological crop growth models have been developed extensively during the

last decades by integrating knowledge across many disciplines such as crop

ecophysiology,micrometeorology,soilscience,andcomputingtechnologies(Loomis

et al., 1979; Ritchie, 1991; Boote et al., 1996; Bouman et al., 1996;McCown et al.,

1996). Thesemodels have been conventionally used to predict the performance of

given cultivars under various environmental conditions, and are now increasingly

beingusedinbreedingprogrammes(Aggarwaletal.,1997).

One of themain applications of ecophysiological crop growthmodels is their

explanationof differences in yieldpotential of genotypeson thebasis of individual

physiologicalparameters(Kropffetal.,1995;HaverkortandKooman,1997).Thisis

possible because their parameters can be used to mimic genetic characteristics of

plants when crop growth models become more mechanistic and comprehensive

(Boote et al., 1996). Values of these parameters are specific to each genotype and

constantunderawiderangeofenvironmentalconditionsthereforeareoftenreferred

toasgeneticcoefficients(Hoogenboometal.,1997;Booteetal.,2001;Tardieu,2003;

Bannayan,2007), reflecting theawareness that thevariationof theseparameters is

undergeneticcontrol(Stam,1996).Theseparameterscouldbethereforeconsidered

asquantitativetraitsandareamenabletofurtheranalysis(Yinetal.,2004;Quilotet

al.,2005),e.g.forQTLmapping,andevaluatinganddesigningideotypes(Loomisetal.,

1979;Kropffetal.,1995;Booteetal.,1996;HaverkortandKooman,1997;Yinetal.,

2004;YinandStruik,2008).

Ecophysiological models can make it possible to explore the impact of new

genotypes and the contribution of individual physiological traits on yield by

simulatinggenotypesunderanydefinedenvironmentalscenariosaswellasarange

of management options (Stam, 1996; Bindraban, 1997; Hoogenboom et al., 1997;

Asseng and Turner, 2007). Such models could help in understanding of GEinteraction andmay speedup crop improvement for specific environments (Kropff

andGoudriaan,1994;Robertsetal.,1996;Booteetal.,2001;Slafer,2003;Yinetal.,

2004; Mayes et al., 2005) and to assist plant breeding to better quantify crop

genotype‐phenotype relationships (Yin et al., 2000a, 2004; Reymond et al., 2003;

Hammeretal.,2006).

Improvedphenotypicpredictionsviacropmodellingresultfromtheirabilityto

Chapter1

4

dealwiththecomplexinteractionsamongplantgrowthanddevelopmentprocesses,

environmental effects and genetic controls. Thispredictive capacity of thedynamic

model could be used for favourably weighting more important QTLs during their

selection in marker‐assisted breeding programmes. Crop physiology using crop

models therefore can also reinforce the QTL analysis of complex traits, thereby

improving breeding efficiency and enhancing genetic design (Yin et al., 1999a,b,

2000a).

WhenQTL information is incorporated into cropmodels, the ‘QTL‐basedcrop

models’ could provide valuable insights into which combinations of alleles favour

adaptation to specific environments, therefore should narrow down genotype–

phenotypegaps.Theycanpotentiallybeapowerfultooltoresolvethephenomenon

ofGE interactionand thereforeovercome the limitation thatQTLmapping cannotextendanexperimentalresultofoneenvironmenttoanotherenvironmentcondition

(Messinaetal.,2006).Basedonrecentstudies(e.g.Reymondetal.,2003;Yinetal.,

2005a; Gu et al., 2012), there is now increasing advocacy forQTL‐basedmodelling

approaches,i.e.linkingcropmodellingwithQTLmappingforcomplextraits.

If cropphysiologyandgeneticsarecombined judiciously,cropphysiologyand

modellingresearchcanreinforce thegeneticanalysisof complex traitsand thereby

canofferthepossibilitiesofmodel‐basedbreeding.Promotingthisnovelroleofcrop

physiology,however,requiresanewgenerationofcropmodels.Suchmodelsshould

have robust model architecture, numerical consistency, stability, and enhanced

heuristics. Most importantly, crop models should include genotype‐specific

parameters for incorporating the latest findings offered by the rapid advances in

functionalgenomics(Yinetal.,2004).Suchmodelswouldgreatlyenhanceourability

to translate short‐time scale gene‐level information to the crop performance in

continuouslychangingfieldenvironments.

GECROS:AnimprovedstateoftheartecophysiologicalmodelRecently, a state of the art ecophysiological crop growth model called GECROS

(Genotype‐by‐Environment interaction on CROp growth Simulator), based on the

above lines of thinking for improved model structure and input parameters, was

developedbyYinandVanLaar(2005).GECROSusesnewalgorithmstosummarise

current knowledge of individual physiological processes and their interactions and

feedbackmechanisms.Itattemptstomodeleachsub‐processataconsistentlevelof

detail, so that no process is overemphasised or requires toomany parameters and

similarlynoprocessistreatedinatrivialmanner,unlessunavoidablebecauseoflack

of understanding. GECROS also tries to maintain a balance between robust model

structure,highcomputationalefficiency,andaccuratemodeloutput.Themodelcan

Generalintroduction

5

be used for examining responses of biomass and dry matter production in arable

cropstobothenvironmentalandgenotypiccharacteristics.Inthisstudyweexplored

the ability of GECROS to analyse differences in tuber yield in a set of varieties

covering awide rangeofmaturity types and awell‐adapteddiploidF1 segregating

populationofpotato(SolanumtuberosumL.).

Potato:OriginandimportancePotato(SolanumtuberosumL.)isanannualcropfromthegenusSolanum(family

Solanaceae) (Khurana et al., 2003). The potato originates from the mountains of

SouthAmericawhereithasbeenanimportantfoodcropforalongtime(Rowe,1993).

Inthe16thcenturythepotatowasintroducedtoEuropeasacuriosity,firsttoSpain

and then intoEngland (Hawkes,1990; Spooner et al., 2005).Gradually it becamea

majorfoodcrop,especiallywhenvarietieswereselectedwhichwereadaptedtolong

dayconditions.Inthe18thand19thcenturiesitwasalreadyanimportantfoodcrop,

especiallyforthepoorinvariouscountriesinEurope.Fromhere,potatospreadtoall

overtheworldasamajorfoodcrop(Harris,1992).

The genus Solanum comprises of many species of which over 200 are tuber

bearingaccording to the latest taxonomic interpretationbyHawkes (1990).Within

this genus, the section Tuberarium (Correll, 1962), also known as section Petota

(D'Arcy,1972),includesthetuber‐bearingmembers,ofwhichthecultivatedpotatois

best known. Solanaceous plants can be found throughout the world, though most

speciesdwellintropicalregionsofCentralandSouthAmerica.

Potatoes can be grown wherever it is neither too hot (ideally average daily

temperatures below21 oC) nor too cold (above 5 oC), and there is adequatewater

from rain or irrigation (Govindakrishan andHaverkort, 2006). The growing season

canbeasshortas75daysinthelowlandsubtropics,where90‐120daysisthenorm,

andas longas180daysinthehighAndes.Inthelowlandtemperateregionswhere

planting isdone inspringandharvesting inautumn,cropdurationis typically120‐

150 days, and yields are potentially high. Average fresh‐weight yields vary

tremendously by country from 2 to 45 t ha‐1 with a global average of 17.4 t ha‐1

(FAOSTAT,2010).Likemanyotherimportantcrops,potatoisapolyploid.Cultivated

potatovarietiesaretetraploid(4n=48)andmanywildspeciesarediploid(2n=24)

butmayrangeuptohexaploid(6n=72).Besides,potatoisoutcrossinginnatureand

phenotypic and genotypic variation is very common in potato. Therefore potato is

heterozygous, a characteristic that contributes to its extreme genetic diversity and

hasprobablybeenakeyfactorinitssurvival.

Potatoisamongtheworld’smostimportantagronomiccropsandisthehighest

Chapter1

6

rankednon‐cereal.Theroleofpotatoisnowwellrecognisedinhumannutrition,food

security, and national economy ofmany countries. Due to increasing food demand

and changing diets the traditionally high consumption of potatoes in Europe and

North America has recently combinedwith potato becoming a subsistence crop in

manyotherregionsparticularlyinareaswherepovertyishighlyconcentrated(Scott,

2002;http://www.cipotato.org).Currentlythecropisgrownatasignificantscalein

about 130 countries, and covers yearlyworldwide about 18million ha (Struik and

Wiersema,1999;FAOSTAT,2010).

Potatohighyieldpotential andhighnutritionalqualityhasmade themoneof

themostimportantplantfoodsources.Itisawholesomefoodwithalltheextremely

important and necessary dietary constituents, which are needed for health and

growth(Pushkarnath,1978;Li,1985;Burton,1989)(Figure1.1).

Comparedtootherrootsandtubersandalsomanycereals,potatotubershavea

high ratio protein to carbohydrates with a high nutritional value of the protein

(Shekhawatetal.,1994).Potatocanbeusednotonlyasahumanfoodoranimalfeed,

but also for seed tuber production, and industrial use. Various food processing

industries use potato to produce crisps, French fries, flakes, and canned potato,

whereas the non‐food industry uses potatoes to produce starch, alcohol, etc.

(Feustel,1987;StruikandWiersema,1999;Khuranaetal.,2003).

Figure1.1.Nutrientcontentofpotatoesper100g,afterboiling inskinandpeelingbefore

consumption.Source:UnitedStatesDepartmentofAgriculture,NationalNutrientDatabase.

In short, potato as a major food staple is contributing to the United Nations

MillenniumDevelopmentgoalsofprovidingfoodsecurityanderadicatingpoverty.In

recognitionoftheseimportantroles,theUNnamed2008astheInternationalYearof

Generalintroduction

7

Potato.With the growing global potato production requiresmodernisation of crop

improvement techniques and efficient utilisation of resources towards high yield

production.

YielddeterminingcomponentsinpotatoYield production can be expressed as the integrated response of distinct plant

processes to resources suchas radiation,nitrogenorwater.This response involves

the following twomain steps: theproductionofphotoassimilates, and their further

transformation into an economic (usually harvestable) component. In potato tuber

dry matter yield can be quantified by the summation of the daily incoming

photosynthetically active radiation (PAR) times the daily fraction of that radiation

interceptedbythecroptimestheradiationuseefficiencytimesthedailyfractionof

thetotaldrymatterpartitionedtothetubers(Struiketal.,1990).Thisisillustrated

bythefollowingequation:

1.1

wherePARinc=daily incomingphotosyntheticallyactiveradiation, fPARint= fraction

ofthePARintercepted,RUE=radiationuseefficiency,andHI=theharvestindex.

The seasonal rate of dry matter accumulation in potato is a function of

interceptionandutilisationofincidentphotosyntheticactiveradiation.Differencesin

the rate of dry matter accumulation can be attributed to differences in light

interception caused by variability in maximum LAI (leaf area index), in leaf

senescence(e.g., ‘canopycover’ inpotato)duringtuberbulking,andincanopy‐level

efficiency of utilisation of intercepted radiation as a result of changes in the

functionalityof leafphotosynthesisduringtuberbulking(Pashiardis,1988;Spitters,

1988;VanDeldenetal.,2001).Severalauthors(AllenandScott,1980;Burstalland

Harris,1983;VanderZaagandDoornbos,1987;Spitters,1988;Koomanetal.,1996)

haveanalysedthedifferencesinyieldofpotatocropsbasedoninterceptedradiation,

theefficiencyoftheuseofthisradiationfordrymatterproduction,andtheharvest

index.

Thegenetic improvementofyieldcanbeunderstoodmoremechanisticallyby

inspectingandinterpretingtherelationshipsbetweenthemainattributesofgrowth

andbypartitioningtheyieldcomponentsdescribedabove.Studiesonthemechanistic

basisbywhichbreedinghassuccessfullyincreasedyieldmayshedlightonpotential

futurealternatives.

HIRUEPARPARYield intinc f

Chapter1

8

NeedofthisstudyTheagronomyandphysiologyofyielddevelopment inpotato isextremelycomplex

due to genotype specific and environmental effects (i.e. GE) on crop physiologicaland morphological characteristics that are conditioned by the combination of

genotypeandenvironment,i.e.GE(Pashiardis,1988;AllenandScott,1992;JefferiesandMacKerron,1993;Tourneuxetal.,2003;Schittenhelmetal.,2006).Despite the

importanceofthisfact,fewstudiessoughttounderstandthephysiologicalprocesses

leadingtoyielddevelopmentandthefactorsaffectingtheseprocessesandtheseare

stillnotwellunderstood.Mostreportsareconcernedwithexperimentscarriedoutin

growthrooms,whichoftenuseveryshortstemsectionsasplantingmaterialandin

whichenvironmentalconditions,especiallylightintensity,differedgreatlyfromthose

normally prevailing during the period of initiation in the field. Insight into the

aboveground shoot development, particularlywith respect to canopy cover and its

relation with whole‐plant physiology are still underdeveloped (Almekinders and

Struik,1996).

Traditionally, crop growth simulation models have principally been used to

study and predict crop performance in response to environmental conditions and

managementpractices,whereasgenotypic impactsoncropperformance(especially

inthecontextofplantbreedingwherelargenumbersofgenotypesareinvolved)have

receivedlittleattention.Littleeffortshavefocusedonmodellingcanopydynamicsin

relationtogenotypewithenvironmentinteraction(GE),presumablyduetolackofsuitablemodellingapproachesanddatasets.Thereisaneedforamodelthatincludes

cropreactions to temperature,day lengthandsoilnitrogenavailabilityon theyield

dynamics of the crop (Allen and Scott, 1980;Kooman andRabbinge, 1996). Such a

modelmayalsoofferbreedersthescopetobreedforcultivarswhichmakethemost

effectiveuseofagivenenvironment.

MainobjectivesThe analysis conducted in this thesis is based primarily on data collected from six

contrasting field experiments carried out in Wageningen (52˚ N latitude), the

Netherlands, during 2002, 2004 and 2005. The plant material used in this study

consisted of 100 F1 diploid (2n= 2x= 24) potato genotypes derived from a cross

between two diploid heterozygous potato clones, SH83‐92‐488 RH89‐039‐16(RouppevanderVoortetal.,1997;VanOs,2006)referredtoasSHRH.Besides,fivestandard cultivars (i.e. Première, Bintje, Seresta, Astarte, and Karnico) were also

includedinourstudies.Thethesishasthefollowingobjectives:

To develop an approach to quantify and analyse the dynamics of important

Generalintroduction

9

growthanddevelopmentrelatedprocesses(i.e.dynamicsofgreencanopycover

and tuber bulking), where these processes are dissected into biologically

meaningful and genetically relevant component traits. We want to study the

impactoftheseprocessesontuberyieldproductionunderdiverseenvironmental

conditions;

To analyse how these processes are influenced quantitatively by genotype (G),

environment(E),andGEinteraction; Todevelopaphysiologybasedunambiguousmethodofassessingmaturitytype

inpotato;

ToidentifyfavourableQTLallelesforyielddeterminingphysiologicalcomponent

traits of potato for specific environment(s) to complement marker assisted

breedingprogrammes;

Toexamine theabilityandpotentialof theecophysiological cropgrowthmodel

‘GECROS’ in explaining yield differences in potato to help design strategies for

potatoideotypebreedingforspecificgenotypesand/orenvironments;

Toidentifyimportantphysiologicalandgenotypictraitswhichcouldbeusefulas

aselectioncriterionforimprovingcropyieldinpotato.

OutlineofthethesisThis thesis is composed of six chapters including this introduction (Chapter 1).

Chapters2 to5 present themain results of the study.Chapter2 andChapter3

describeamodellingapproachtoanalyseandquantifythedynamicsofcanopycover

andtuberbulking,respectively,duringtheentirecropcycleasafunctionofthermal

timeanditsvariabilityamongpotatogenotypesandtobreakthetimecourseandits

variationdownintobiologicallymeaningfulandgeneticallyrelevantcomponenttraits.

Wedissectandassess theroleofgenotype(G),environment (E),andGEon tuberyieldproduction;discusstheheritabilityofthesetraitsandtheirgeneticbackground

(i.e.QTLs)onthepaternalgenomes.Usingthisapproach,theaimistoquantitatively

analyse the sourcesof variation intouseful components of canopy cover and tuber

bulkingdynamicsandget insights into themostvitalprocesses thatcanbeused to

explore the possibilities of genetically manipulating potato tuber yield. Chapter4

discusseshowtoquantifyandassessmaturitytypeunambiguouslyforalargesetof

genotypes. It presents an improved approach to re‐define the concept of maturity

type in terms of important physiological traits thus relating the maturity to crop

phenology and physiology. Chapter 5 examines the performance of the

ecophysiologicalcropgrowthmodel‘GECROS’toanalysedifferencesintuberyieldin

asetofvarietiescoveringawiderangeofmaturitytypesandadiploid(SHRH)F1

Chapter1

10

segregatingpopulationofpotato.Finally,Chapter6discussesthemainfindingsand

overallcontributionofthisthesis.Itconcludeswithnewdirectionsandopportunities

of linking cropphysiologywith genetic systems and indicatesways to enhance the

power ofmolecular breeding strategies in potato. Figure 1.2 illustrates the overall

frameworkofthisstudy,whileFigure1.3furtherelaboratesthatplanandprovidesa

schematicoutlineofthisthesis.

Figure1.2.Overallframeworkofthisstudy.

Generalintroduction

11

Figure1.3.Schematicthesisoutline.

 

CHAPTER2

Analysinggeneticvariationinpotato(SolanumtuberosumL.)usingstandardcultivarsandasegregatingpopulation.

I.Canopycoverdynamics*

M.S.Khan,P.C.Struik,P.E.L.vanderPutten,H.J.Jansen,H.J.vanEck,F.A.vanEeuwijk,X.Yin

AbstractWe present a model based on the beta function to analyse the genotype‐by‐environment (GE) interactions for canopy cover dynamics in potato (Solanumtuberosum L.). The model describes the dynamics during the build‐up phase,maximum cover phase, and decline phase of canopy development through fiveparameters defining timing and duration of three phases and maximum canopycover (vmax). These five parameters were estimated for 100 individuals of an F1population, their parents, and five standard cultivars, using data from six fieldexperiments, and used to estimate secondary traits, related to rate of increase ordecreaseofcanopycoverandtheareaunderthecanopycovercurve for thethreephases.Thelengthofthecanopybuild‐upphasewasconserved,butthedurationofmaximum canopy cover (DP2)and the decline phase (DP3)varied greatly, with latematuringgenotypeshavinglongerDP2andDP3andthusahigherareaunderwholegreen canopy cover curve (Asum). Genetic variance of the onset (t2) or end (te) ofcanopysenescenceandAsumcontributedgreatly to thephenotypicvariance.StrongpositivephenotypicandgeneticcorrelationswereobservedbetweenDP2andvmaxorte,indicatingthatgenotypeswithlongerDP2couldbeindirectlyobtainedbyselectinggenotypes with high vmax or te. High broad‐sense heritability estimates across sixenvironmentswere recorded for t2, te,DP2, andAsum. Several quantitative trait loci(QTLs)weredetected for ourmodel parameters andderived traits explaining thevariancebyupto74%;manyoftheseQTLswerelocatedonpaternallinkagegroupV. Some of the QTLs were mapped to similar positions in the majority ofenvironments, but only few QTLs were stable across environments.We concludethatourapproachyieldedestimatesforagronomicallyrelevantcropcharacteristicsfor defining future breeding strategies in potato and the information obtainedshouldhelpinmarker‐assistedselection.Keywords:Potato(SolanumtuberosumL.),canopycoverdynamics,beta function,thermaltime,componentsofvariance,genotype‐by‐environment(GE)interaction,genetic variability, heritability, maturity type, QTL mapping, QTL‐by‐environment(QTLE)interaction._____________________________________*SubmittedtoFieldCropsResearch.

Chapter2

14

1.Introduction

Tuber drymatter yield in potato (Solanumtuberosum L.) can be quantified by the

summationofthedailyincomingphotosyntheticallyactiveradiation(PAR)timesthe

daily fraction of that radiation intercepted by the crop times the radiation use

efficiency times the daily fraction of the total drymatter partitioned to the tubers

(Struiketal.,1990).Thispaperisonthefirstpartoftheequation:thefractionofthe

incomingradiationintercepted,asinmanypotatogrowingregionstheabilityofthe

crop to produce and maintain a green canopy able to absorb and use incoming

radiation during the available growing season for biomass production determines

tuberyields(MollandKlemke,1990).

Thecropcanopycoverateachtimeduringthegrowingseasonistheresultant

oftherateanddurationofcanopygrowthandtherateofcanopysenescence(orthe

timing of the haulm killing) (Struik et al., 1990; Struik, 2007). This canopy cover

directly correlates with the fraction of incoming PAR absorbed (Khurana and

McLaren,1982;MacKerronandWaister,1985a;VanderZaagandDoornbos,1987;

VanDeldenet al., 2001).Under optimal conditions andwith long growing seasons,

earlyestablishmentoffullcanopycoveranditspersistenceoveralongperiodleadto

a higher yield due to better interception of incoming radiation (Martin, 1995),

although the biomass needs to be partitioned to tubers to obtain economic yield.

Thereisaclose,positivecorrelationbetweencanopycoverandtuberyield(Vander

Zaag, 1982; Vander Zaag and Demagante, 1987; Fahem and Haverkort, 1988). The

proper quantification of canopy dynamics is therefore one of the most important

elementsofmodellingpotatoproduction(Hodges,1991).

The variation in tuber yield among potato genotypes can – under conditions

without abiotic or biotic stress −be analysed in termsof differences in cumulative

light absorption, the efficiency with which the absorbed light energy is used for

biomass production, and partitioning of dry matter to the desired plant organ

(Pashiardis,1988;Spitters,1988;VanDeldenetal.,2001).Mostpotatogenotypesare

indeterminate (Allen and Scott, 1992; Struik and Ewing, 1995, Almekinders and

Struik, 1996) and (the continuation of) canopy growth largely depends on the

appearance of lateral (basal or sympodial) branches, and on the appearance,

expansion, and senescenceof leaveson thosebranches (Allen andScott, 1992;Vos

andBiemond,1992;Vos,1995a;AlmekindersandStruik,1996;Fleisheretal.,2006).

Thecanopycoverinpotatocropsmaythereforecompriseofleavesfrommainstems,

secondarystems,andaxillarybranches(Allen,1978).Potatogenotypesdifferinthe

degree towhich theybranchand theextentof sympodialbranching is indicativeof

theirmaturitytypeandthedegreeofdeterminacyintheirgrowthhabit(Firmanetal.,

1995).Once tuberbulkinghasbecomerapid, thebranch formationcomes toahalt

Geneticvariationinpotatocanopycoverdynamics

15

and the leaf senescence is no longer compensated by new leaf formation. The

durationofcanopycovercanbeconsidereda trait reflecting thematurity typeofa

specificcultivarorgenotype(Chapter4).

Genotypicdifferencesinpotatocanopycoverarelarge(Spitters,1988;Jefferies

andMacKerron, 1993; Tourneuxet al.,2003). Spitters (1988) for example showed

that differences in yields among five contrasting varieties could be explained by

variation in canopy cover over time (resulting in differences in cumulative light

absorption) and in lightuse efficiencyandharvest index. Likemanyother complex

quantitative traits, canopy cover may be controlled by many interacting genes of

which each only has a small effect (Lark et al., 1995; Orf et al., 1999; Daniell and

Dhingra, 2002; Stuber et al., 2003). The principal approach for the analysis of

quantitativetraitsistheuseofapopulationwhichsegregatesforthetraitsofinterest

(Tanksleyetal.,1989).Forpotato,individualsofsuchasegregatingpopulationcanbe

produced by crosses made between commercial cultivars or with individual

genotypesdevelopedfrompopulationimprovementproceduresandtheyarereadily

propagatedbyasexualreproduction(Bradshaw,1994).Commercialcultivarsutilised

as parents are heterozygous and segregation of characterswill be found in the F1

generationfollowinghybridisation.

Canopy cover dynamics also shows a high genotype‐by‐environment (GE)interaction (Pashiardis,1988;AllenandScott,1992;Schittenhelmet al.,2006).The

mainenvironmental factors influencing thecanopydynamicsunder fieldconditions

aretemperature,photoperiod,lightintensity,watersupply,andnitrogen(N)supply.

For our study the effects of temperature and N are most relevant and these are

thereforesummarisedinthisintroduction.

Temperaturestrongly influencesstemelongationandbranching(Marinusand

Bodlaender,1975;AllenandScott,1980;Struiketal.,1989a,AlmekindersandStruik,

1994, 1996) and leaf appearance, expansion, and senescence (Kirk and Marshall,

1992;Vos,1995a;Firmanetal.,1995;StruikandEwing,1995;VanDeldenetal.,2001;

Fleisher and Timlin, 2006; Fleischer et al., 2006; Struik, 2007), with optimal and

maximum temperatures for these processes contributing to canopy development

being 23‐25 ˚C and about 32 ˚C, respectively (Struik, 2007). N supply affects the

number of lateral (basal and sympodial) branches, the number of leaves on those

branches, the individual leaf size and leaf senescence (Fernando, 1958;Humphries

andFrench,1963;VosandBiemond,1992;AlmekindersandStruik,1996).Nitrogen

helpstoattaincompletecanopycoverearlyintheseason,especiallyunderrelatively

resource‐poorconditions(HaverkortandRutavisire,1986;Vos,2009),toextendthe

periodoffullcanopycover;andtoslowdownsenescence(SantelizandEwing,1981),

thusleadingtoincreasedlightinterception(Martin,1995).

Chapter2

16

Thereisaclearneedtoinvestigatethecomponentsofgeneticvariation(e.g.in

canopy dynamics) and to determine the GE interaction for a diverse populationundercontrastingenvironmentalconditions(Tarnetal.,1992;Bradshaw,1994).The

genetic variation and the GE interaction for canopy cover need to be defined bypreciselyassessingtherateanddurationofthebuild‐upphase,themaximumcanopy

coverandthedurationoftheperiodduringwhichthismaximumismaintainedand

therateanddurationofthecanopysenescence.Thesevariablesneedtobeoptimised

in such a way that the crop can complete its growth cycle within the period of

favourableweather conditions thus realising a high yield and an adequate harvest

index. There is a growing awareness that inorder to better analyse complex traits

using increasingly available genomic information, integration of quantitative crop

physiologywithgeneticsisrequired(Tardieu,2003;Yinetal.,1999a;Yinetal.,2004;

Hammeretal.,2006;YinandStruik,2008;Chenuetal.,2009;Messinaetal.,2009;

Hammeretal.,2010;TardieuandTuberosa,2010;YinandStruik,2010).

In this paper we first describe a quantitative approach to analyse the time

courseofcanopycoverduringtheentirecropcycleasafunctionofthermaltimeand

itsvariabilityamongpotatogenotypesandtobreakthetimecourseanditsvariation

down into biologicallymeaningful and genetically relevant component traits. Using

this approach, we aim to quantitatively analyse the sources of variation in canopy

cover dynamics among full‐sibs of an outcrossing segregating population, their

parents and a set of standard cultivars of potato. The heritability of and genetic

variation among different component traits of canopy cover are estimated (Van

Eeuwijk et al., 2005). Finally,we performQTLmapping of our traits to investigate

theirgeneticbasisanddiscusstheco‐locationsofQTLsforthesetraitsandtheirQTL‐

by‐environment (QTLE) interaction.Withsuch information,dominantcomponentsofcanopycovercanbe identifiedandtheirphenotypicandgeneticcorrelationscan

be assessed, with the ultimate goal of designing an effective breeding strategy of

potato.

2.Materialsandmethods

2.1.F1segregatingpopulationofSHRHandstandardcultivarsTheplantmaterialusedinthisstudyconsistedof100F1diploid(2n=2x=24)potato

genotypes derived from a cross between two diploid heterozygous potato clones,

SH83‐92‐488RH89‐039‐16(RouppevanderVoortetal.,1997;VanOsetal.,2006).Parentmaterialwasalso includedinthestudy.ThefemaleparentalcloneSH82‐93‐

Geneticvariationinpotatocanopycoverdynamics

17

488 is referred to as “SH” and themale parental clone RH89‐039‐16 as “RH”. Like

manyotherfull‐sibpopulations,theSHRHpopulationsegregatesformaturitytype,which is strongly associatedwith thedurationof canopy cover (VanderWal et al.,

1978;VanOijen,1991). It shouldbenoted thatdiploidpotatoesperfectlyresemble

tetraploidcultivarswithrespecttomanydevelopmentalprocesses,includingcanopy

development, but the most noticeable difference is a somewhat smaller plant size

(Hutten,1994).

Wealso included five standard (tetraploid) cultivars inour studies:Première,

Bintje, Seresta, Astarte and Karnico. These cultivars were chosen because of their

differences inmaturity typewhengrown in theNetherlands and they ranged from

early (Première) to very late (Karnico). This selection of standard cultivars would

allowabenchmarkingofconsequencesofmaturitytypeonthetimecourseofcanopy

cover.

We aimed at using disease‐free, uniform‐sized seed tubers of similar

physiologicalqualityandthereforeproducedseedtubersunderthesameconditions

and stored them together at 4 °C for approximately 6 months until planting. The

graded seed tubers were disinfected chemically to control Rhizoctonia and pre‐

sproutedbeforeplanting.

2.2.Fieldexperimentsandmeasurements

Six field experiments were carried out in Wageningen, the Netherlands (52˚ N

latitude),during2002,2004,and2005,withtwoexperimentsineachyear,torecord

canopycoverdynamics for100F1genotypes,2parents,and4 (Exps5and6)or5

(Exps 1‐4) standard cultivars. Karnicowas not included in the two experiments in

2005. Experiments differed in environmental conditions because theywere carried

out in different years, on different soils, and under different N fertiliser regimes,

thereby creating six contrasting environments (Table 2.1). Each experiment was

conductedusingarandomisedcompleteblockdesignwithtwoblocks.Thegenotypes

wererandomisedwithineachblock,consistingof106or107plots.Eachplothadsix

rows of 16‐18 plants. The seed bed was prepared following standard cultivation

practices. Seed tuberswereplanted in rows spaced0.75mapartwith0.27‐0.30m

betweenplantswithinthesamerows.Nitrogenfertiliserwasbroadcastedatplanting

in amounts depending on experiment (Table 2.1). Soil phosphorus and potassium

levels were kept sufficiently high to sustain normal crop growth. Crops were

managedwelltoensurethattheywerefreeofpests,diseases,andweeds.Irrigation

wasappliedwhennecessarytoavoiddroughtstress.

Chapter2

18

Table2.1.Descriptionoftheexperimentalsitesandexperimentalmethodsappliedin

Wageningen,theNetherlands.

Experimentno. 1 2 3 4 5 6

Year 2002 2002 2004 2004 2005 2005

Soiltype Clay Sand Clay Sand Sand Sand

Plantingpattern

(mm)0.750.30

0.750.30

0.750.27

0.750.27

0.750.27

0.750.27

Plantingdate 25April 25April 28April 28April 4May 26April

Plantdensity

(plantsm‐2)4.4 4.4 4.9 4.9 4.9 4.9

Netplotsize

(m2)23.0 23.0 20.7 20.7 20.7 20.7

SoilNat

planting

(kgha‐1)

n.d.1 n.d.2 12.5 23.0 81.0 49.5

FertiliserN

(kgha‐1)143 175 70 200 125 50

TuberNuptake

(kgha‐1)†− − 112.3 172.0 153.6 156.1

Weatherconditionsduringgrowingseason

2002 2004 2005

Airtemperaturemax(oC) 20.9 20.6 20.7

Airtemperaturemin(oC) 11.2 10.4 9.9

Rainfall(mm) 364.1 367.1 369.1

Relativehumidity(%) 78.2 75.6 76.5

Solarradiation(MJm‐2d‐1) 15.3 15.8 15.81n.d.=nodata.Guestimate:40kgNha‐1.2n.d.=nodata.Guestimate:60kgNha‐1.†DataforExperiments1and2werenotavailableandreplacedby“−”.

Plant emergencedatewas observed on thedatewhen50%of the plants had

emerged from the soil surface. Afterwards, green canopy cover (%) was visually

assessedonweeklybasisforeachplotduringthewholecropcyclebyusingagridas

Geneticvariationinpotatocanopycoverdynamics

19

describedbyBurstall andHarris (1983).Thegridconsistedofanaluminium frame

with thedimensions0.75m0.90m,adjusted to thecommonplantingpattern forpotato(rowwidth0.75m,plantingdistancewithintherow0.30m).Theframewas

dividedinto100equalrectanglesof0.075m0.090m.Thegridwasplacedabovethe potato canopy at 1 m from the ground and only those rectangles more than

halffilled with green leaves were counted by observing vertically above to avoid

parallaxerror(CadersaandGovinden,1999).Thegridwasplacedhalfwayoneach

side of the potato row to sample three plant positions. Observations were always

made at the same position in the field. The number of observations made on the

canopycoverduringthegrowingseasonperindividualplotwas17‐21.

Attheendofthegrowingseason,tubersofeachplotwereharvestedanddried

inanovenat70˚Ctoconstantweight.Insamplesofthegrowingseasonsof2004and

2005, tuber nitrogen content was determined by micro‐Kjeldahl digestion and

distillation (AOAC, 1984) in a fully accredited commercial laboratory. Average N

uptake was calculated and was used to characterise some of the environments in

termsoftotalNavailableforcropgrowth.

2.3.Modelapproach

2.3.1.Amodelforphasicdevelopmentofcanopycoverdynamics

Canopycoverdynamicsinpotatoasquantifiedbythegridmethodtypicallyfollowsa

patternthatcanbesubdividedintothreedistinctphases(Fig.2.1),i.e.build‐upphase

(P1),maximumcoverphase(P2),anddeclinephase(P3).

Build‐upphase(P1)

This first phase covers the period from emergence to full canopy cover, and is

dominated by appearance of stems, lateral branches and leaves, and expansion of

thoseorgans(AllenandScott,1992;VosandBiemond,1992;Vos,1995a;Fleisheret

al.,2006).Toobtainflexibleandasymmetricalcurves,canopycoverduringP1canbe

written,accordingtoasigmoidpartofthebetafunctionfordeterminategrowth(Yin

etal.,2003a),as:

11m11

1max with1 m11

1

tttt

tttt

vv ttt

0 (2.1)

Chapter2

20

Can

opy

cove

r (%

)

Thermal days (td)

vmax

0 m1t 2t1t et

P1 P2 P3

Figure2.1.Temporalcourseofpotatocanopydevelopment(fullcurve)fromtime0

to tedescribed by our three‐phase model, eqns (2.1‐2.3), respectively. P1 (canopy

build‐upphase),P2(maximumcanopycoverphase),andP3(canopydeclinephase).

Thedashedcurveisthemathematicalextensionofeqn(2.1).

where tm1 is the inflexion pointand vmax is themaximum value of canopy cover v,

whichisreachedattimet1(Fig.2.1).Equation(2.1)canbeappliedtocanopycover

withinthetimespanof0≤t≤t1;otherwise,vhastobesetat0ift<0oratvmaxift>

t1.

Maximumcoverphase(P2)

Duringthisphasethecanopycoverretainsitsmaximumlevelvmax.Thecanopycover

duringP2issimplygivenby:

21max with tttvv (2.2)

Geneticvariationinpotatocanopycoverdynamics

21

wheret2reflectsthelasttimepointwhencanopycoverisstillatitsmaximumand/or

onsetofcanopysenescence(Fig.2.1).

Declinephase(P3)

Thecanopycover starts todeclineafter time t2, and reaches zeroat theendof the

cropcyclei.e.te.Usuallythisdeclinefollowsareversedsigmoidpattern,whichcould

be formulatedasvmax[1‐f(t)] (cf.Yinetal.,2009),where f(t) is the function,suchas

eqn (2.1), describing a normal sigmoid pattern with time. Therefore, for P3, an

equationcouldbeformulated,usinganinflexionpointofthedeclinephase.However,

inmanycasestheestimatesforparametersshowedlargestandarderrors,indicating

over‐fittingwhenamodelhaving this inflexion isapplied incombinationwitheqns

(2.1)and(2.2)todescribethedataoftheentiretimecourseofcanopydynamics.To

reducethechanceofover‐fitting,wedevelopedanequationforthecanopycoverin

Phase3excludingthissecondinflexion,yetyieldingsatisfactoryresults:

e21

21

2e

emax with 2e

1

tttt

ttttttt

vv ttt

(2.3)

wheretereflectsthetimepointwhencanopycoveriszero.Thisequationisbasedon

adeclinepartofthebetaequation(Yinetal.,1995)andassumesthattediffersfrom

theextendedendpointofeqn(2.1)by(t2‐t1)(Fig.2.1).Equation(2.3)canbeapplied

withinthetimespanoft2≤t≤te;otherwise,vhastobesetat0ift>te.

Combiningeqns(2.1,2.2,and2.3)yieldsamodelwithfiveparameters:tm1,t1,t2,

te,andvmax.Anyfurtherreductionofparametersresultedinsignificantlossoffit(data

notshown).Themodeldescribescanopydynamicsforagivengenotype‐environment

combination. The model contains three segments but is smooth because the first

derivatives of vwith respect to tare zero at the joining points t1 and t2. Values of

model parameters tm1, t1, t2, te,and vmaxwere estimated for each individual plot of

every experiment with the iterative non‐linear least‐square regression using the

Gaussmethod,asimplementedinthePROCNLINoftheSASsoftware(SASInstitute

Inc.,2004).

2.3.2.Calculatingsecondarytraits

Once themodel parameterswere estimated, several secondary traitswere derived

reflecting the duration of the phases, the rates of canopy cover dynamics, and the

overallpotentialtointerceptlightduringthedifferentphases.

Chapter2

22

Asdefinedearlier,ourmodelsetstimezeroastheonsetofP1;sotheduration

ofP1 is numerically equal to t1, and theduration of the entire cycle is numerically

equaltote.ThedurationofP2(DP2)wascalculatedast2−t1.ThedurationofP3(DP3)

wascalculatedaste−t2.

Averagegrowthrate forP1(c1) isdefinedas:vmax/t1, andaveragesenescence

rate during P3 (c3) is vmax/Dp3. Maximum growth rate during P1 (cm1), which is

achieved at tm1, could be estimated according to the equation given by (Yin et al.,

2003a)as:

max1

m1

m111

m11m1

2 m11

m1

vtt

ttttt

c ttt

(2.4)

Theareaundereachsectionofthecurvewasestimatedbyintegratingeqns(2.1,

2.2,and2.3)overtimefortherespectivephases.Inanalogytothesolutiontothearea

forareversedsigmoidcurve(Yinetal.,2009),theintegralforP1canbeexpressedas:

232

m11

m111max1

ttttt

vA (2.5)

TheintegralforP2issimplyexpressedas:

12max2 ttvA (2.6)

TheintegralforP3ismorecomplicatedandcanbefoundas:

2‐22 1

1

12e12e

12e

2emax3

2e

1

tt

tttttt

tttttv

Att

t

(2.7)

ThesumofA1,A2,andA3results in theareaunderwholegreencanopycurve,Asum.

ThevalueofAsumreflectsthecapabilityofthecroptointerceptsolarradiationduring

thewholegrowingseason(cf.Vos1995b,2009).

2.4.Calculatingthermaltime

Ourexperimentswereconductedunder fieldconditions.To takeaccountofdiurnal

and seasonal variation in temperature under these conditions, the time axis in the

Geneticvariationinpotatocanopycoverdynamics

23

aboveequationsshouldbeinthermaltime,whichhaslongagobeenintroducedasa

scaleofreferenceforontogeny(Gallagher,1976,1979;MilfordandRiley,1980;Ong,

1983,Milfordetal.,1985;Bonhomme,2000).

WefollowedtheapproachofYinetal.(1995,2005),whodevelopedanequation

todescribethenon‐linearrelationshipbetweentemperature(T)andrateofgrowth

orofdevelopment(g(T))asafunctionofthreecardinaltemperatures, i.e.base(Tb),

optimum(To)andceilingtemperature(Tc):

t

oc

bo

bo

b

oc

c )(

c

TTTT

TTTT

TTTT

Tg

(2.8)

where ct is the temperature response curvature coefficient. When T≤ Tbor ≥Tc,

growthordevelopmentdoesnottakeplace.WhentemperatureisbetweenTbandTc,

g(T)hasabell‐shapedcurvewithmaximumvalue1atTo.

We used data sets of leaf appearance and leaf expansion rates from growth

chamber experiments of Fleisher et al. (2006) and Fleisher and Timlin (2006) to

estimatethevaluesofTb,To,Tc,andctofeqn(2.8).Theirdataonindividualleafarea

(cm2)andleafappearancerate(leavesplant‐1day‐1)ofpotatocultivarKennebecwas

usedtoestimateleafareaperplant(cm2plant‐1)atsix(14/10,17/12,20/15,23/18,

28/23,and34/29˚Cforday/night)temperatureswitha16/8hdiurnalcycleunder

ambient [CO2] of 370μmolmol‐1. From this data set the values ofTb,To,Tc, and ct

were determined using the iterative procedure of SAS‐NLIN, which follows a non‐

linear least squaresmethod forestimationofparameters (SAS Institute Inc.,2004).

Estimates (± standarderrors)were:Tb=5.5 (±5.2) ˚C,To=23.4 (±0.5) ˚C,Tc=34.6

(±1.3) ˚C,ct =1.6 (±0.5).Basedon thedataset,muchextrapolationwas involved in

estimatingTbwhich resulted in high standard error. Obviously,we have to assume

that the temperatureresponseswere thesame forallgenotypesand forallphases,

basedonassertionsthatwithinacropspecies,thegeneticdifferencesintemperature

sensitivity of phenology are relatively small (Ellis et al., 1990) and that cardinal

temperaturesinvariousdevelopmentalprocessesarequitecloseinpotato(e.g.Struik,

2007). A sensitivity analysis showed that uncertainties in cardinal‐temperature

valueshadlittleimpactonassessinggeneticdifferencesincanopydevelopment(data

notshown).

After the estimation of the temperature response parameters, eqn (2.8) was

used to convert days after emergence into thermal days after emergence (td).

Because the function g(T) is non‐linear and temperature fluctuates diurnally, g(T)

Chapter2

24

wasestimatedonanhourlybasisandhourlyg(T)valueswereaveragedtoobtainthe

daily value (Yin et al., 2005a). By definition, one thermal day is equivalent to one

actual day only if temperature at any moment of that day is at To (= 23.4 ˚C).

Obviously, every actual day was less than one unit thermal day. Required data of

hourlyairtemperatureswereobtainedfromaweatherstationinWageningenlocated

nearbytheexperimentalsites.

2.5.Statisticalandgeneticanalyses

All statistical analyses were performed in Genstat (Payne et al., 2009). A general

analysisofvarianceacrossenvironmentswasperformedtotestthesignificanceand

extent of differences between environments, all genotypes (including the F1

population, the parents, and standard cultivars) and GE interactions, where theeffectofblockwithinenvironmentwasincludedinthemodel.Meansofgenotype(G),

environment(E),andGEinteractiontermswerecomparedusingtheFisher’s leastsignificant difference (LSD) test. Further statistical analyseswere performed using

onlythe100genotypesoftheF1populationandthesearedescribedbelow.

2.5.1.Estimationofvariancecomponents

Weusedastatisticalmodel(VanEeuwijk,2003)toestimatethevariancecomponents

andtoassesswhatthecontributionwasofthegenotypicmaineffectsandtheGEtothetotalphenotypicvarianceforthemodelparameters(i.e.tm1,t1,t2,te,andvmax)and

secondarytraits(cm1,c1,c3,DP2,DP3,A1,A2,A3,andAsum):

ijkijijkjijk GEGEy (2.9)

where i= 1,…, 100; j= 1,…, 6; k= 1, 2; and yijk denotes the response variable of

genotype i, in environment j, blockk;µ is the grandmean;Ej is the environmental

maineffect;βjkstands forblockwithinenvironmenteffect;Gi is thegeneticeffectof

genotype i;GEij is the genotype‐by‐environment interaction effect for genotype i in

environment j,and ijk isaresidual term.Theunderlinedtermswereconsideredas

randomeffects,whichwereassumed tobenormallyand independentlydistributed

withzeromeanandapropervariance.

Therestrictedmaximumlikelihood(REML)procedurewasusedtoestimatethe

variancecomponents.Thesignificanceofthevariancecomponentswastestedusing

thelikelihoodratio(LR)test(Morrell,1998).Thistestdeterminesthecontributionof

asingle(random)factorbycomparingthefit(measuredasthedeviance,i.e.‐2times

Geneticvariationinpotatocanopycoverdynamics

25

the log likelihood ratio) for models with and without the factor. The phenotypic

variance,σ2Phwasestimatedfromthesevariancecomponentestimatesas(Bradshaw,

1994;FalconerandMackay,1996;LynchandWalsh,1998):

2ε

2GE

2G

2β

2E

2Ph (2.10)

whereσ2E=environmentalvariance,σ2B=blockvariance,σ2G=geneticvariance,σ2GE

=GEvariance,andσ2ε=experimentalerrorvariance.

2.5.2.Phenotypicandgeneticcoefficientofvariation

Scalingof the varianceby the traitmean, i.e. calculating the coefficient of variation

provides a more appropriate way to compare traits (Johnson et al., 1955; Houle,

1992). We therefore calculated the coefficient of variation (%) for each model

parameterandderivedtraitacrosssixexperiments,accordingto:

1002

CV (2.11)

where is thegrandmeanof thepopulation,and2

isavariancecomponent (i.e.

σ2phorσ2Gorσ2Eorσ2GE)fromeqn(2.9).

2.5.3.GGEbiplotanalysisStudyingquantitative traits is complicateddue toGE interactions (Uptmoor et al.,2008).GGEbiplotanalysisisaneffectivemethodtofullyexploremulti‐environment

trials for G and GE components of variation. It allows visual examination of therelationshipsamongthetestenvironments,genotypesandtheGE(Yanetal.,2000).For any particular environment, genotypes can be compared by projecting a

perpendicular fromthegenotypesymbols to theenvironmentvector, i.e.genotypes

thatare furtheralonginthepositivedirectionof theenvironmentvectorarebetter

performing and vice versa. The greater the distance from the origin to the

intersection of a genotype projection on an environment vector, the more this

genotype deviates from the average in this environment (Kroonenberg, 1997;

Chapman et al., 1997). GGE biplot analysis was performed to analyse the inter‐

relations among genotypes and environments. GGE biplots were constructed by

plotting the first principal component (PC1) scores of the genotypes and the

environments against their respective scores for the second principal component

Chapter2

26

(PC2).Theenvironment‐standardisedmethodofYan(2002)wasused.

2.5.4.Phenotypicandgeneticcorrelation

Product‐moment (Pearson) correlations were calculated, using genotypic means

acrossblockswithintrials,amongmodelparametersandsecondarytraits.

Thegeneticcorrelationswereestimatedusingthefollowingequation(Holland,

2006):

2G

2G

2G

G ji

ijijr

(2.12)

whereσ2Gijistheestimatedgeneticcovariancebetweentraits iand j;σ2Giandσ2Gjare

thegeneticvariancesoftraitsiandj,respectively.ThemultivariateREMLprocedure

was used to estimate the genetic variance and covariance estimates (Meyer, 1985;

Holland,2006).Thesignificanceofgeneticcorrelationswasdeterminedusingat‐test

after a z‐transformation of the correlation coefficients (Sokal and Rohlf, 1995;

Guttelingetal.,2007).

2.5.5.Heritability

We estimated the broad‐sense heritability (H2) (%) across all six environments by

using the estimatedvariance components of the linearmodel described asper eqn

(2.9),throughthefollowingequation(Bradshaw,1994;FalconerandMackay,1996):

100

t

2ε

e

2GE2

G

2G2

nn

H

(2.13)

wherene=6represents thenumberofenvironments,andnt=12 is theproductof

numberofblocksandenvironments.

The broad‐sense heritability (H2) was also estimated for each individual

experimentbyusingthefollowingformula(Bradshaw,1994):

100

b

2ε2

G

2G2

n

H

(2.14)

Geneticvariationinpotatocanopycoverdynamics

27

where σ2G = genetic variance; σ2ε = experimental error variance, and nb = 2

represents the number of blocks per individual environment. The variance

componentswereestimatedperindividualtrialbasis.

2.5.6.ExtensionofanAFLPmarkermapoftheSHRHpopulationTheAFLPprimer combinations used in this study have been previously applied to

createtheultra‐densegeneticmapof130SHRHgenotypesasdescribedinVanOsetal.(2006).Asour100F1lineswereonlypartlygenotypedforcreatingthatmap,

120 new SH RH genotypeswere fingerprintedwith AFLPTM (Vos et al., 1995) tomakeamapfor250individuals.GenomicDNAwasextractedfromfrozenleaftissue

accordingtoVanderBeeketal.(1992).AFLPmarkersweregeneratedaccordingto

standard protocols with radioactive labels, using four Eco‐Mse primer combinations

(Vosetal.,1995).TheAFLPprofilesof theparental cloneswerecomparedwithan

ultra‐densegeneticmapandAFLPproductsofequalelectrophoreticmobilitywhich

segregated inbothsetsof the lineswere identified.TheAFLP(only1:1segregating

markers)were firstmapped in theSHRHmappingpopulationusing JoinMap4.1(VanOoijen,2006)usingabinmappingapproach(VanOsetal.,2006).Themarker

dataweresplitintotwosetsonthebasisoftheirsegregationtype.Markersthatwere

heterozygous in thematernal parent (SH) and absent in the paternal parent (RH)

werescoredas<abaa>;“paternal”markersheterozygous inRHandabsentinSHwerescoredas<aaab>.

For map construction, recombination frequencies were converted into map

units (cM) by the use of the Kosambi function. Only markers with recombination

valueof<0.25wereconsideredasdescribedbyVanOsetal. (2006).Thematernal

andpaternal data setsweredivided into 12 and13 linkage groups, respectively. A

graphic representation of a map was made by the Map Chart software (Voorrips,

2002).

2.5.7.QTLdetection

Genstat version 14 (Payne et al., 2009) softwarewas used to identify QTLs for all

component traits. Eighty‐eight genotypes of our 100 F1 lines were covered in the

sample of the aforementioned 250 lines of SH RH population for the extendedmarkermap;dataofthese88lineswerethereforeusedfordetectionofQTLsforfive

model parameters and nine secondary traits. QTL analysis was performed

individuallyforallsixexperiments(environments).

QTL models were fitted for the two parents separately. Initially, the

Chapter2

28

conventionalSimple IntervalMapping (SIM)procedureasdescribedbyLanderand

Bostein(1989)andHackettetal.(2001)wasusedtoscanthegenomeforthemajor

QTLsperindividualenvironment.AQTLwasdeclaredtobesignificant(P<0.05)for

the threshold value (‐log10 (P) > 3.4). Secondly, amore sophisticatedQTLmapping

procedure, the Composite Interval Mapping (CIM) was performed to increase the

reliability of the QTL analysis (Zeng, 1994; Jansen, 1993; Jansen and Stam, 1994;

Jansen, 1995). In thismethod, backgroundmarkerswere selected to take over the

roleoftheputativeQTLsascofactorstoreducetheresidualvariance.Inouranalysis,

backgroundmarkersclosesttotheindicatedregionofputativeQTLswith–log10(P)

scoresexceedingthethresholdweregraduallyusedascofactors.Thisprocedurewas

repeateduntil no furtherQTLswere found.Thepercentageof the total phenotypic

variationexplainedbyQTLsidentifiedforeachtraitwasestimatedastheR2‐value.

3.Resultsanddiscussion

3.1.Modelperformanceindescribingcanopycoverdynamicsofgenotypes

Theuseofdatainthermaldaysresultedinamorestableparameterestimationthan

theuseofdata indays (datanot shown), as the confoundingeffectsofdiurnal and

seasonal temperature fluctuations during the experimental period could effectively

be removed using thermal days. The model for the canopy cover dynamics (i.e.

combinedeqns(2.1,2.2,and2.3))fittedwellforeachgenotypeofpotatosegregating

population,theirparentsandstandardcultivarsintheentiredataset,withR2values

rangingfrom0.94to1.00(n=17‐21;datanotshown).Figure2.2showsthecurve‐

fitting results for the SH and RH parents and for the standard cultivars that were

present in each experiment. In most cases, the standard error (SE) values for

observed points (i.e. average of two blocks) were much higher during the canopy

decline phase than during the first two phases, consistent with the fact that the

estimationof te involvedmuchextrapolationdue to lackofdatapoints in the third

phase of canopy growth. Overall, combined eqns (2.1‐2.3) could be very useful in

analysing the canopy cover dynamics of a diverse set of potato genotypes under

various environments and making inferences about the underlying process

controllingcanopygrowth.

3.2.Modelparametersandsecondarytraits

Usually,theestimatedvalueofeachmodelparameterdifferedverylittlebetweenthe

Geneticvariationinpotatocanopycoverdynamics

29

Figure2.2.Observed(obs)pointsandfitted(fit)curvesofstandardcultivarsandSH

andRHparentsforallsixexperiments(environments).Dataforcv.Karnicowerenot

availableinExps5and6(seethetext).

twoblocksandtherewasagoodpositivelinearcorrelationbetweenthetwoblocks

forallmodelparameters(R2valuesusuallyabove0.7;n=107).Table2.2givesthe

estimatedvaluesofthefivemodelparametersandtheninederivedsecondarytraits

for twoparents (SHandRH), five standardcultivars, and themeanvalueof theF1

0

20

40

60

80

100SH

0

20

40

60

80

100Première

0

20

40

60

80

100

0 20 40 60 80

Seresta

0

20

40

60

80

100Bintje

0

20

40

60

80

100

0 20 40 60 80

Astarte

0

20

40

60

80

100

RH

Thermal days (td)

Can

opy

cove

r (%

)

Chapter2

30

populationforallsixenvironments.

Environmenthadahighlysignificant(P<0.01)impactonallmodelparameters

and derived secondary traits, at least partly due to the purposeful variation in

availability ofN across trials (Table 2.1). Nitrogen has amain influence on canopy

development(PerumalandSahota,1986;Vos,1995a,b,2009),arisingfromeffectsof

Nonrateanddurationofappearanceofleavesandbranchesonthepotatoplant,and

on the active life span of individual leaves. Due to the lack of precise information

aboutamountofmineralNbecomingavailableduringthecourseofcanopygrowth,

weusedtheamountofNuptakebytubersasanindicatorofNavailability(Table2.1).

CropNuptakeisasite‐specificindicatorofNthatis“available”tothecrop(Sullivan

et al., 2008). Experiments with lower N uptake (especially Exp 3) also had lower

estimates for vmax, DP2, and canopy growth rates (cm1, c1, and c3). In potato, the

duration of P2 and whether or not full canopy cover is attained are very much

affectedbyNsupply.ThetotalgrowthperiodisprolongedforlargerrateofNsupply

becauseP2 extendswith betterN nutrition. Usually,within agronomically relevant

ranges,NhascomparativelylittleeffectonP1andaffectstherateofsenescenceonly

marginally (Vos, 2009). However, in ourmost extreme experiment (Exp 3)we did

observe such effects even to such an extent that the end of the crop cycle (i.e. te)

occurredlaterinthelowNenvironment(i.e.Exp3)incomparisontoenvironments

withhigherNavailability(hightuberNuptake)(Table2.2).

Themeanperformanceof standardcultivars,parentalgenotypes(SHandRH)

and the F1 population across the experimental sites (i.e. environments) varied

significantly (P<0.01) for all model parameters and secondary traits (Table 2.2).

Potatogenotypesvariedinthedurationoftheentirecropcyclewhichisreflectedin

variation in te. There were significant differences (P<0.05) among the standard

cultivars and parents SH and RH for t1, t2, and te, whereas tm1 did not differ

significantlyamonggenotypes(Table2.2).Latecultivarsgavehighervaluesfort2andte followedbymid‐lateandearlycultivars.Thereweresmalldifferencesamongthe

cultivars for vmax, but under certain environmental conditions late cultivars might

reach higher values than earlier ones, due to their comparatively later tuber set.

KoomanandRabbinge(1996)foundthat,comparedwithlatecultivars,earlypotato

cultivarsallocatea largerpartof theavailableassimilates to thetubersearly in the

growingseason,resultinginshortergrowingperiodsandalsoloweryields.Theinitial

phaseofcanopygrowth(i.e.t1orDP1)wasnotmuchaffectedbythematuritytypeof

thegenotypes,buttherewereverylargedifferencesamongstandardcultivarsforthe

durationofP2andP3(i.e.DP2andDP3,respectively).ThedurationofP2(DP2)tended

tobe longer for later cultivars.Estimatedvaluesof canopygrowth ratescm1andc1

variedsignificantly(P<0.05)amongthestandardcultivarsandtwoparents.However,

Geneticvariationinpotatocanopycoverdynamics

31

Table2.2. Estimatedmean values of fivemodel parameters (tm1,t1,t2,te, vmax)and

ninederived secondary traits (cm1, c1,c3,DP2,DP3,A1,A2,A3,Asum) as obtained from

across environment ANOVA for the two parents (SH and RH) and five standard

cultivars(listedinorderofincreasinglylongercropcycle).tdstandsforthermalday.

Dataforcv.KarnicowerenotavailableinExps5and6andarereplacedby“−”.

Parameter Exp.1 Exp.2 Exp.3 Exp.4 Exp.5 Exp.6 Mean¶

tm1(td)

SH 16.0 13.5 9.4 8.8 10.8 10.2 11.4a

RH 19.7 15.3 6.2 9.6 9.4 8.7 11.5a

Première 16.3 14.8 8.4 6.7 11.7 8.4 11.1a

Bintje 15.3 14.0 8.5 8.9 11.0 8.3 11.0a

Seresta 14.4 11.2 11.1 8.1 13.0 11.2 11.5a

Astarte 16.8 13.0 10.0 9.3 10.9 8.6 11.4a

Karnico 14.0 12.4 9.3 8.6 − − 11.1a

Mean† 16.1a 13.5b 9.0d 8.6e 11.1c 9.2d

F1mean‡ 18.0a 14.7b 9.7e 10.7d 11.6c 9.8e t1(td)

SH 30.1 25.0 21.5 19.9 24.1 17.4 23.0a

RH 27.6 26.1 23.6 16.5 14.7 14.4 20.5ab

Première 27.2 22.3 17.6 14.9 18.0 13.3 18.9b

Bintje 26.8 20.2 23.8 17.2 17.8 14.4 20.0ab

Seresta 26.3 19.1 22.6 15.9 20.1 17.7 20.3ab

Astarte 27.9 22.1 36.0 18.6 17.0 15.1 22.8a

Karnico 23.6 19.2 31.2 16.9 − − 22.7a

Mean† 27.1a 22.0c 25.2b 17.1e 18.6d 15.4f

F1mean‡ 31.0a 23.0d 25.8b 19.7e 23.8c 16.4f

Chapter2

32

Table2.2.(Continued)

Parameter Exp.1 Exp.2 Exp.3 Exp.4 Exp.5 Exp.6 Mean¶

t2(td)

SH 41.4 38.6 43.1 29.7 45.3 40.3 39.7cd

RH 36.1 35.9 29.0 31.1 47.6 28.3 34.7d

Première 42.2 28.6 31.6 28.2 40.6 38.9 35.0d

Bintje 40.5 36.5 42.2 45.9 52.5 41.1 43.1c

Seresta 52.1 47.0 56.3 62.4 69.8 55.7 57.2ab

Astarte 43.2 38.8 70.3 68.0 75.7 64.9 60.2a

Karnico 31.7 28.5 75.0 72.3 − − 51.9b

Mean† 41.0e 36.2f 49.6b 48.2c 55.3a 44.9d

F1mean‡ 41.8b 34.0e 37.9d 40.2c 49.0a 40.9bc te(td)

SH 57.8 51.8 63.7 57.8 58.2 54.4 57.3e

RH 51.3 51.1 52.9 53.8 57.1 47.4 52.3f

Première 50.8 45.7 54.0 41.3 57.3 51.3 50.1f

Bintje 67.7 60.6 76.9 65.3 65.5 57.2 65.5d

Seresta 69.7 71.7 85.2 73.3 78.3 72.1 75.0c

Astarte 80.6 80.2 79.4 78.3 84.5 78.0 80.2b

Karnico 82.9 82.5 87.5 90.4 − − 85.8a

Mean† 65.8c 63.4d 71.4a 65.7c 66.8b 60.1e

F1mean‡ 58.2c 53.4e 62.1b 58.1c 64.9a 57.2d vmax(%)

SH 100.0 100.0 64.7 100.0 98.0 100.0 93.8c

RH 93.6 64.2 57.8 100.0 100.0 100.0 85.9d

Première 100.0 100.0 68.9 99.8 100.0 100.0 94.8bc

Bintje 100.0 100.0 96.1 100.0 100.0 100.0 99.4ab

Seresta 100.0 99.9 88.2 100.0 100.0 100.0 98.0abc

Astarte 100.0 100.0 93.6 100.0 100.0 100.0 98.9ab

Karnico 100.0 100.0 100.0 100.0 − − 100.0a

Mean† 99.1b 94.9c 81.3d 100.0a 99.7ab 100.0a

F1mean‡ 95.3d 99.2ab 71.0e 98.8b 97.5c 99.9a

Geneticvariationinpotatocanopycoverdynamics

33

Table2.2.(Continued)

Parameter Exp.1 Exp.2 Exp.3 Exp.4 Exp.5 Exp.6 Mean¶

cm1(%d‐1)

SH 5.1 6.4 4.7 7.4 6.0 9.5 6.5b

RH 6.6 4.1 3.6 9.6 11.7 11.7 7.9a

Première 6.0 7.9 5.9 9.9 9.7 13.0 8.8a

Bintje 6.0 9.3 5.9 9.0 10.3 11.2 8.6a

Seresta 5.9 8.5 6.0 9.5 9.3 9.7 8.2a

Astarte 5.9 7.6 3.9 8.4 10.1 10.8 7.8ab

Karnico 7.0 9.0 4.8 8.9 − − 7.4ab

Mean† 6.1e 7.5d 5.0f 9.0c 9.5b 11.0a

F1mean‡ 5.1e 7.6c 4.3f 8.7b 7.0d 10.4a c1 (%td‐1)

SH 3.3 4.1 3.0 5.0 4.1 5.9 4.2b

RH 3.4 2.5 2.5 6.0 6.8 7.0 4.7b

Première 3.7 4.5 3.9 6.7 5.6 7.5 5.3a

Bintje 3.7 5.0 4.0 5.9 5.7 7.0 5.2ab

Seresta 3.8 5.3 3.9 6.3 5.1 5.7 5.0ab

Astarte 3.6 4.6 2.7 5.4 5.9 6.7 4.8ab

Karnico 4.3 5.2 3.2 5.9 − − 4.7b

Mean† 3.7e 4.4d 3.3f 5.9b 5.5c 6.6a

F1mean‡ 3.1d 4.4c 2.8e 5.2b 4.4c 6.2a c3 (%td‐1)

SH 6.1 7.6 3.2 3.9 8.1 8.2 6.2a

RH 6.2 4.3 2.5 4.6 11.1 5.2 5.7a

Première 11.6 5.9 3.1 7.4 6.7 8.1 7.1a

Bintje 3.8 5.0 2.8 5.4 8.6 6.3 5.3a

Seresta 5.7 4.1 7.3 11.2 11.9 7.2 7.9a

Astarte 2.7 2.5 11.0 9.7 12.7 8.5 7.9a

Karnico 2.0 1.9 8.1 8.0 − − 5.0a

Mean† 5.4c 4.5d 5.4c 7.2b 9.9a 7.2b

F1mean‡ 6.5b 6.1b 3.4c 6.4b 7.5a 7.1a

Chapter2

34

Table2.2.(Continued)

Parameter Exp.1 Exp.2 Exp.3 Exp.4 Exp.5 Exp.6 Mean¶

DP2(td)

SH 11.3 13.6 21.5 9.8 21.3 22.9 16.7c

RH 8.5 9.8 5.3 14.5 33.0 13.8 14.2c

Première 15.0 6.2 14.1 13.3 22.7 25.6 16.1c

Bintje 13.7 16.3 18.4 28.7 34.7 26.7 23.1b

Seresta 25.8 27.9 33.6 46.5 49.7 38.1 36.9a

Astarte 15.4 16.7 34.3 49.4 58.8 49.8 37.4a

Karnico 8.1 9.3 43.8 55.4 − − 29.1b

Mean† 14.0e 14.2e 24.4d 31.1b 36.7a 29.5c

F1mean‡ 10.7d 11.1cd 12.0c 20.5b 25.1a 24.5a

DP3(td)

SH 16.5 13.2 20.7 28.1 12.9 14.1 17.6bc

RH 15.1 15.3 23.9 22.7 9.5 19.2 17.6bc

Première 8.7 17.2 22.4 13.1 16.7 12.4 15.1c

Bintje 27.1 24.2 34.7 19.4 13.0 16.1 22.4b

Seresta 17.6 24.7 28.9 10.9 8.5 16.4 17.8bc

Astarte 37.4 41.5 9.1 10.3 8.8 13.1 20.0bc

Karnico 51.3 54.0 12.5 18.1 − − 34.0a

Mean† 24.8b 27.1a 21.7c 17.5d 11.6f 15.2e

F1mean‡ 16.5d 19.4b 24.3a 17.9c 16.0d 16.3d

A1(td%)

SH 1455.5 1192.3 723 1047.7 1239.1 785.4 1074a

RH 946.2 743.1 810 753.8 610 636 750b

Première 1201.1 900.3 612.3 777 738.7 561.2 798b

Bintje 1231.3 763.7 1286.5 844.4 756.8 658.7 924ab

Seresta 1247.5 862 1011.7 786.5 827.7 745.7 914ab

Astarte 1232.2 995.2 2023.2 919.2 708.6 691.6 1095a

Karnico 1045.7 795.8 1811.1 836.5 − − 1122a

Mean† 1194a 893b 1183a 852c 813d 680e

F1mean‡ 1331a 948d 1076c 904e 1159b 722f

Geneticvariationinpotatocanopycoverdynamics

35

Table2.2.(Continued)

Parameter Exp.1 Exp.2 Exp.3 Exp.4 Exp.5 Exp.6 Mean¶

A2(td%)

SH 1130 1361 1394 978 2091 2288 1540d

RH 797 610 326 1454 3296 1380 1311d

Première 1498 623 966 1323 2267 2558 1539d

Bintje 1373 1627 1752 2871 3472 2671 2294c

Seresta 2580 2789 3050 4648 4967 3806 3640a

Astarte 1537 1665 3171 4939 5875 4983 3695a

Karnico 807 926 4376 5543 − − 2913b

Mean† 1389e 1372e 2148d 3108b 3661a 2948c

F1mean‡ 1050c 1098c 906d 2065b 2472a 2442a A3(td%)

SH 1127.5 903.6 895.1 1826.9 859.9 941.9 1092bc

RH 971.4 679.9 910.6 1482 645.8 1257.8 991c

Première 605 1158.1 1010 878.1 1107.3 828.5 931c

Bintje 1807.5 1583.4 2185.9 1277.4 875.5 1065.5 1466b

Seresta 1196.8 1622 1545.7 732.8 588.5 1095.1 1130bc

Astarte 2445.7 2655.5 584.1 707.4 604.2 880 1313bc

Karnico 3257.6 3373.6 866.9 1188.4 − − 2172a

Mean† 1630b 1711a 1143c 1156c 780e 1011d

F1mean‡ 1074cd 1286a 1112bc 1169b 1039d 1073cd

Asum(td%)

SH 3713 3457 3012 3853 4190 4015 3707d

RH 2715 2033 2046 3690 4552 3273 3051e

Première 3304 2681 2589 2979 4113 3948 3269e

Bintje 4412 3974 5224 4993 5104 4396 4684c

Seresta 5024 5273 5607 6168 6383 5647 5684b

Astarte 5215 5316 5778 6566 7188 6555 6103a

Karnico 5110 5095 7054 7568 − − 6207a

Mean† 4213e 3976f 4473d 5116b 5255a 4639c

F1mean‡ 3455d 3332e 3094f 4138c 4685a 4246b

Means followed by different letters are significantly different according to Fisher’s

MultipleRangeTest(P<0.05).

Chapter2

36

¶Genotype(twoparentsandfourorfivecultivars)meanacrosssixexperiments(i.e.

environments).

†Meanofeachindividualenvironmentacrossgenotypes(twoparentsandfourorfive

cultivars).

‡F1meanofeachindividualenvironmentacrossF1population(100genotypes).

tm1:LSDforgenotype=1.6;LSDforenvironment=0.3;LSDforGE=2.8t1:LSDforgenotype=3.0;LSDforenvironment=0.5;LSDforGE=5.1t2:LSDforgenotype=5.5;LSDforenvironment=0.9;LSDforGE=9.6te:LSDforgenotype=4.0;LSDforenvironment=0.7;LSDforGE=6.9vmax:LSDforgenotype=4.9;LSDforenvironment=0.8;LSDforGE=8.5cm1:LSDforgenotype=1.3;LSDforenvironment=0.2;LSDforGE=2.3c1:LSDforgenotype=0.5;LSDforenvironment=0.1;LSDforGE=0.9c3:LSDforgenotype=2.7;LSDforenvironment=0.5;LSDforGE=4.8DP2:LSDforgenotype=6.1;LSDforenvironment=1.0;LSDforGE=10.6DP3:LSDforgenotype=6.6;LSDforenvironment=1.1;LSDforGE=11.5A1:LSDforgenotype=224.1;LSDforenvironment=37.2;LSDforGE=388.1A2:LSDforgenotype=568.2;LSDforenvironment=94.3;LSDforGE=984.2A3:LSDforgenotype=392.7;LSDforenvironment=65.2;LSDforGE=680.2Asum:LSDforgenotype=386.4;LSDforenvironment=64.1;LSDforGE=669.2the differences were not large. These rates were comparatively higher for early

maturingcultivarsthanthosematuringlate.AveragesenescencerateduringP3(i.e.

c3)wasnotsensitivetothematuritytypeofthegenotypes.

The total biomass production and accumulation of potato cultivars are

dependent on the intercepted PAR (Spitters, 1988; Vos and Groenwold, 1989; Van

Delden,2001).Theamountoflightinterceptedisinproportiontotheareaunderthe

wholecanopycurve(Vos,1995b).MeanestimatesforAsumrangedfrom3051to6207

td%forfivestandardcultivarsandtwoparentalgenotypes.Asumvalueswerehigher

for late cultivars likeKarnicoandAstarte than for early cultivars suchasPremière

(Table2.2).TheareaundercanopycoverduringP2,i.e.A2,significantlycontributed

to the higher values of Asumin late cultivars (Table 2.2). Mean values of Asum also

showedhighlysignificant(P<0.05)variationwithintheF1populationacrossthesix

environments(Table2.2).

The variation (the median, minimum and maximum values) of these model

parameters and secondary derived secondary traits for the F1 population per

individualexperimentisillustratedinFigs.2.3and2.4,respectively.Therangesofthe

Geneticvariationinpotatocanopycoverdynamics

37

five model parameters were consistently wider in Exp 3 than in the other

experiments(Fig.2.3).Incaseofgrowthrates(i.e.cm1,c1,andc3)widerrangeswere

observed in Exps 4 and 5 (Fig. 2.4). Wide ranges of variation were observed for

durationof thethreephasesofcanopydevelopment i.e.t1inExp3, forDP2inExp4,

andforDP3inExp2.Inthecaseoftheareasunderthecurves,therangeswithinthe

populationwerelargestinExp3forA1,inExps4and5forA2,andinExp2forA3.For

Asum,therangeswerehighestinExp3andvalueswerebetween1030and6101td%

(Fig.2.4).Theseresultswereinlinewiththeresultsforthelengthofthreephasest1,

DP2, and DP3. Within the F1 population, some genotypes particularly SHRH34‐H6,

SHRH83‐L9, and SHRH‐136) recorded the highest averageAsum with values (6146,

5836, and5739 td%, respectively). This shows that genotypes like thesemayhave

higher potential to intercept the PAR and tuber yield production. Most of the

parameterswerenearlynormallydistributed(Figs.2.5and2.6)whichmightindicate

theirtransgressivesegregation.

3.3.Phenotypic,geneticandenvironmentalvariances

Table 2.3 presents estimated values of phenotypic, genetic, and environmental

variancesforalltheparametersandthederivedsecondarytraitsintheF1population.

The results revealed considerable phenotypic and genetic variances for the traits.

Almostallthecomponentsofvariationweresignificant(P<0.01)(Table2.3).

Themajorportionofthephenotypicvariancewasaccountedforbythegenetic

variance component for t2, te, and Asum (Table 2.3). The contribution of the

environmentalvariancetothephenotypicvariancewasrelativelylargefortm1,t1,vmax,

andthederivedsecondarytraitscm1,c1,c3,DP2,andA2.Thecontributionof theGEinteractionvariancecomponenttothephenotypicvariancewaslargefortraitsDP3,A1,

andA3(Table2.3).

Estimatesofphenotypic(CVPh)andgenetic(CVG),environmental(CVE),andGEinteraction (CVGE) coefficients of variation for model parameters and derived

secondarytraitsacrossthesixexperimentsarepresentedinTable2.4.Estimatesof

CVPhrangedfrom15.4%to69.4%.Theseestimatesweresmallestforteandvmaxand

highest forDP2 andA2.Traitswith higherCVPh exhibitmore total variance and are

useful as selection criteria in breeding provided the trait is also heritable. Our F1

population reflected high CVG for almost all traits except for tm1, t1, vmax, and A1.Relatively higherCVG and lowerCVE estimateswere obtained for t2, te,A3, andAsum

suggesting that these traits, compared with the other ones, are under a greater

influenceofgeneticcontrol.TheCVEwascomparativelyhigherthanCVGfortraitstm1,

Chapter2

38

0

10

20

30

40

50

60[td] t1

0

5

10

15

20

25

[td] tm1

0

20

40

60

80

100

[td] te

0

20

40

60

80

1 2 3 4 5 6

[td] t2

0

20

40

60

80

100

1 2 3 4 5 6

[%] vmax

Experiment no.

Figure2.3.BoxplotsofgeneticmeansofanF1populationforfivemodelparameters

inall sixexperiments.Theboxesspan the interquartilerangeof the traitvalues, so

thatthemiddle50%ofthedatalaywithinthebox,withahorizontal lineindicating

themedian.Whiskersextendbeyondtheendsoftheboxasfarastheminimumand

maximumvalues.

Geneticvariationinpotatocanopycoverdynamics

39

Figure 2.4. Box plots of genetic means of an F1 population for seven secondaryderived traits in all six experiments. The boxes span the interquartile range of thetraitvalues,sothatthemiddle50%ofthedatalaywithinthebox,withahorizontallineindicatingthemedian.Whiskersextendbeyondtheendsoftheboxasfarastheminimumandmaximumvalues.

0

2

4

6

8

10c1

[% td-1]

0

5

10

15

20

1 2 3 4 5 6

[% td-1] c3

0

1000

2000

3000

4000[td %] A1

0

2000

4000

6000

8000A2

[td %]

0

1000

2000

3000

4000[td %] A3

0

2000

4000

6000

8000

1 2 3 4 5 6

[td %] Asum

0

5

10

15

20

25cm1

[% td-1]

Experiment no.

Chapter2

40

Figure2.5. Distribution of five model parameters among F1 genotypes across six

experiments.Thevaluesoftwoparents‘SH’and‘RH’areindicatedbyfullarrowand

dashedarrow,respectively.

100

30

25

70 90

20

15

80

10

5

85 95 75 0

80 50 75 45 65

30

25

55

20

15

60 70

10

5

0

Num

ber

of F

1 ge

noty

pes

14 15

25

20

15

8 12

10

5

10 0

11 13 9

tm1

20 32 28

25

20

18 24

15

10

5

0 26 22 30

t1

35 65 55 30 45

25

20

15

50 40

10

5

0 60

t2

te vmax

Geneticvariationinpotatocanopycoverdynamics

41

Figure 2.6. Distribution of seven secondary derived traits among F1 genotypes

acrosssixexperiments.Thevaluesoftwoparents‘SH’and‘RH’areindicatedbyfull

arrowanddashedarrow,respectively.

 

0800 1100 900

18

1200

15

13

10

1000

8

5

3

1800 800 1400

25

20

1000

15

10

1200

5

1600 0

20

2000 5000 0

10

3000 6000

5

4000

30

25

15

15

0 4000

5

2000 3000 0

1000

30

25

20

10

10

5.5 3.0 4.5 0

3.5 5.0 4.0

25

20

15

5

8

5

3

4 8 6

20

18

0 7 9

5

15

13

10

Num

ber

of F

1 ge

noty

pes

cm1 A2

c1 A3

A1

5 10 4

30

8

25

20

3 6

15

10

5

0 7 9

c3 Asum

Chapter2

42

Table2.3. Variance components for five model parameters (tm1, t1, t2, te, vmax)and

nine derived secondary traits (cm1, c1, c3, DP2,DP3, A1, A2, A3, Asum) within the F1

populationacrossallsixexperiments.

Parameter2

ph 2

E 2

β 2

G 2

GE 2

ε

tm1(td) 14.4 10.7** 0.1** 0.5** 1.2** 1.9

t1(td) 47.8 25.1** 0.5** 1.7** 13.6** 6.9

t2(td) 98.0 22.8** 2.1** 36.2** 13.7** 23.2

te(td) 88.2 15.4** 0.9** 54.1** 6.3** 11.5

vmax(%) 207.6 124.9** 0.2NS 16.9** 45.7** 19.9

cm1(%td‐1) 8.4 4.9** 0.4** 0.6** 1.0** 1.4

c1(%td‐1) 2.2 1.5** 0.1** 0.2** 0.2** 0.2

c3(%td‐1) 9.6 1.9** 0.1** 0.3** 1.5** 5.8

DP2(td) 128.5 44.3** 2.2** 33.5** 19.9** 28.6

DP3(td) 59.9 8.4* 2.5** 5.1** 11.6** 32.3

A1(td%) 166690 42225** 3096** 0.01 81689** 39680

A2(td%) 1348961 527517** 13971** 376931** 184725** 245817

A3(td%) 197626 4618NS 4557** 30015** 44409** 114027

Asum(td%) 1206327 372964** 9647** 611003** 99925** 112788**Significantat1%,*Significantat5%,NSNon‐significant.1Thevarianceestimatewasnegativesoassumedzero.

σ2Ph = phenotypic variance, σ2ß = block variance, σ2G = genetic variance, σ2 E =

environmentalvariance,σ2GE=genotypeenvironmentalinteractionvariance,σ2ε=residualvariance.

t1,vmax,cm1,c1,c3,DP2,DP3,andA2,thusreflectenvironmentalsensitivityofthesetraits.

TheCVGE ranged from4.3% to28.0%among the traits.The results further showed

thatCVGE exceeded theCVG for traitstm1, t1, vmax, cm1, c3,DP3, andA3. These results

showthemajorcontributionofGEtotheCVPhofthesetraitsinourexperiments.Itisnormally difficult to select for traits that are sensitive to environmental factors.

However, our results indicated that overall considerable genetic variability existed

for most traits which could be exploited for improvement of light interception

efficiency.

Geneticvariationinpotatocanopycoverdynamics

43

Table 2.4. The phenotypic coefficient of variation (CVPh%), genetic coefficient of

variation(CVG%),environmentalcoefficientofvariation(CVE%),andGEinteractioncoefficientofvariation(CVGE%)forfivemodelparameters(tm1,t1,t2,te,vmax)andnine

derivedsecondarytraits(cm1,c1,c3,DP2,DP3,A1,A2,A3,Asum)withintheF1population

acrossallsixexperiments.

Parameter Mean¶ CVPh CVG CVE CVGE

tm1(td) 12.4 30.5 5.7 26.3 8.8

t1(td) 23.3 29.7 5.6 21.5 15.8

t2(td) 40.7 24.3 14.8 11.7 9.1

te(td) 59.0 15.9 12.5 6.7 4.3

vmax(%) 93.6 15.4 4.4 11.9 7.2

cm1(%td‐1) 7.2 40.3 10.8 30.9 14.2

c1(%td‐1) 4.3 33.9 9.6 28.7 9.4

c3(%td‐1) 6.2 50.4 9.3 22.3 19.7

DP2(td) 17.4 65.1 33.3 38.3 25.6

DP3(td) 18.4 42.1 12.3 15.8 18.5

A1(td%) 1021 40.0 0.0 20.1 28.0

A2(td%) 1674 69.4 36.7 43.4 25.7

A3(td%) 1129 39.4 15.3 6.0 18.7

Asum(td%) 3801 28.9 20.6 16.1 8.3

¶GrandmeanoftheF1segregatingpopulationacrossallsixexperiments.

3.4.GGEbiplotanalysisHere we discuss only the results for trait t1, which had both a strong G and GE

component.TheGGEbiplotsrevealedthatthe1standthe2ndprincipalcomponents

accounted for 85.84% of the GE variation (Fig. 2.7). The environment vectorscovered a wide range of Euclidean space, indicating the existence of strong GEinteractionsamongthesixenvironmentsevaluated.

Theresultswerefurtheranalysedforuniquenessofenvironments.Asshownby

Fig.2.7,fourenvironments(Exps1,2,4,and6)weregroupedtogether.Thissuggests

thattheseenvironmentswerehighlycorrelatedandrelativelysimilarinthemanner

Chapter2

44

Figure 2.7. GGE biplot chart showing relationships among six experiments

(environments) for model parameter t1 in an F1 segregating population. AEC

indicates average environment coordinate (Yan, 2001; Yan andKang, 2003).A line

that passes through the biplot origin and average environment indicates themean

performanceofgenotypes.

theydiscriminateamonggenotypes.Foranyparticularenvironment,genotypescan

be compared by projecting a perpendicular from the genotype symbols to the

environmentvector, i.e.genotypesthatarefurtheralonginthepositivedirectionof

the environment vector are better performing and vice versa. The greater the

distance from the origin to the intersection of a genotype projection on an

environment vector, the more this genotype deviates from the average in this

environment (Kroonenberg, 1997; Chapman et al., 1997). The results showed that

environments (Exps 2 and 6) fell close to the origin. This could mean that these

Geneticvariationinpotatocanopycoverdynamics

45

environments may have little variability across genotypes (Kroonenberg, 1997).

However,markerofenvironment(Exp3)wasstandingfurthestapartfromtheorigin

which may suggest that this environment caused maximum variability across the

genotypes. These results are in line with our earlier described results that Exp 3

causedthemaximumvariabilityforsometraits.Thisenvironmentmightthereforebe

themaincontributortotheoverallGEonaccountofitslowestNavailabilitythantherestoftheenvironments.Itwouldnotbeexpectedthatthesamegenotypewouldbe

the most effective in most of the environments and therefore a focus on why

particulargenotypesperformexceptionallywellineitherloworhighinputsituations

couldenableselectionstrategiestobedevelopedforimprovedvarieties.Thisanalysis

therefore allowed a partial understanding of the environmental causes of the

observedGEinteraction.

3.5.Phenotypiccorrelationsofmodelparametersandthesecondarytraits

Table2.5illustratesthephenotypiccorrelationcoefficientsamongallthetraitswithin

theF1population.Theresultsshowedvariationintherelationshipacrossindividual

experiments.Forthecaseofthefivemodelparameters,therewereweakcorrelations

(r≤0.50)betweentm1andt1,andbetweent1andt2inmostexperiments,butnotin

Exps2and3,wheremoderatepositivecorrelationswerefound.Therewerestronger

positivecorrelations(r≥0.80)betweent2andteinExps3,4,5,and6.Exps1and2,

however,showedmoderate(r=0.63)tolow(r=0.44)t2‐tecorrelations,respectively

(Table 2.5). The results further showed little phenotypic correlation between the

thermaldaysrequiredforthedifferentphasesandthemaximumsoilcover:tm1‐vmax,

t1‐vmax, t2‐vmax, and te‐vmax. Only Exp 3 showedpositive strongcorrelationswith r =

0.51,0.79,0.69,0.69,respectively(Table2.5).

Therewerestrongnegativephenotypiccorrelationsbetweengrowthrates(cm1andc1)andt1andbetweenc3andDP3,whichsuggestsa trade‐offbetweenrateand

durationofthecanopybuild‐upandsenescence,respectively(Table2.5).Theresults

further showed strong negative correlations between t1 and DP2 in several

experiments. The correlations between DP2 and DP3were mostly weak and non‐

significant, only Exps 3 and 4 showed significant (P<0.05) and strong (r = ‐0.59)

negative correlations (Table 2.5). These results suggest that genotypes with slow

canopy build‐up had a relatively short period of maximum canopy cover (DP2).

Furthermore,t1andDP2alsodependeduponvmax,witht1generallybeinglongwithalowvmaxbutwithDP2generallybeingshortwhenvmaxwasbelow100%,whereasDP2

could be (much) longerwhen the canopy reached 100% cover. The results further

revealed strong positive correlations between DP2 and te, whereas in half of the

Chapter2

46

experimentsDP3andtewerepositivelycorrelated.ThissuggeststhatbothDP2andDP3

contribute positively to higher values of te. Later cultivars tended to senescemore

slowly (i.e. tended to have a longerDP3) than earlier cultivars. The results showed

weak correlations between A1, A2, and A3 (Table 2.5). The correlation coefficient

ranged from−0.03 to0.44.These results are in linewith those for t1,DP2, andDP3.There was a strong, positive correlation between A2and Asum in all experiments,

underliningtheimportantcontributionofvariationinA2tovariationinAsum.

Inshort,ourresultssuggestalargecomplexityofthegeneticandphysiological

inter‐relationsbetweenthevariousmodelparametersandderivedsecondarytraits.

3.6.Geneticcorrelationsofmodelparametersandthesecondarytraits

Table 2.5 illustrates the genetic correlation coefficients between the model

parametersandsecondarytraitswithintheF1population.Theresultsshowedweak

(r≤0.50)geneticcorrelationsbetweentm1andt1inExps4and5,andbetweent1and

t2inallexperimentsexceptExp2.Theotherexperimentsshowedpositiveandstrong

(r>0.50)tm1‐t1andt1‐t2geneticcorrelations.Positivegeneticcorrelationsbetweent2

andtewerestronginallexperiments(Table2.5).Therewaslittlegeneticcorrelation

between the thermal days required for the different phases and themaximum soil

cover: tm1‐vmax, t1‐vmax, t2‐vmax,and te‐vmax showedpoorcorrelations.However, these

correlations were stronger and positive in Exp 3 with r = 0.59, 0.87, 0.75, 0.75,

respectively (Table 2.5). There were strong negative genetic correlations between

growthrates(cm1andc1)andt1andbetweenc3andDP3.

The results further revealed weak genetic correlations between t1 andDP2in

Exps2,3,and6,whereasExps1,4,and5showedstrongnegativecorrelations(Table

2.5). The genetic correlations between DP2 and DP3were mostly weak and non‐

significant,onlyExp5showedsignificantanegativegeneticcorrelationwithr=‐0.79.

There were strong positive genetic correlations between DP2 and vmax in all the

experimentsexceptinExps3and6.ThetraitDP2alsoshowedstrongpositivegenetic

correlationswithteinallexperiments.Abouthalfofexperimentsalsoshowedstrong

positivegeneticcorrelationsbetweenDP3andte.Theseresultsindicatethatgenotypes

withlongerDP2andDP3couldbeindirectlyobtainedbyselectinggenotypeswithhigh

vmaxandte.

TherewereweakgeneticcorrelationsbetweenA1andA2,andbetweenA2andA3

in most experiments. However, there were strong positive genetic correlations

between A2 and Asum in all experiments and between A3 and Asum in half of the

experiments(Table2.5).Theseresultsare in linewiththose for t1,DP2,andDP3and

highlighttheimportanceofbothA2andA3tobeusedasanindirectselectionmeasure

Geneticvariationinpotatocanopycoverdynamics

47

**Significantat1%,*Significantat5%,N

S Non‐significant.

Table2.5.Phenotypic(low

ertriangle)andgenetic(uppertriangle)correlationcoefficientsam

ongallpairwisecomparisons

offivemodelparam

eters(tm1,t 1,t2,t e,v

max)andninederivedsecondarytraits(c m

1,c 1,c3,DP2,D

P3,A

1,A 2,A

3,A s

um)withintheF1

populationperindividualexperiment.Theunitforallmodelparam

etersistherm

alday(td)exceptforv

max(%

)andthe

derivedtraits A

1,A2,A3,Asum(td%)andc m

1,c 1,c3(%

td‐1).

t m

1

t 1

t 2

t e

v max

c m1

c 1

c 3

DP2

DP3

A1

A2

A3

Asum

Exp.1

t m1

0.56**

‐0.13N

S ‐0.12NS

‐0.11N

S ‐0.27**

‐0.40**

‐0.03NS

‐0.38**

‐0.05NS

0.21*

‐0.36**

‐0.10NS

‐0.30**

t 1

0.43**

‐0.44**

‐0.41**

‐0.60**

‐0.91**

‐0.92**

‐0.10NS

‐0.82**

‐0.20*

0.71**

‐0.82**

‐0.39**

‐0.70**

t 2

‐0.02NS

‐0.30**

0.78**

0.43**

0.53**

0.50**

0.07

NS

0.88**

0.21*

‐0.23*

0.87**

0.30**

0.82**

t e

‐0.07NS

‐0.38**

0.62**

0.37**

0.51**

0.48**

‐0.52**

0.72**

0.78**

‐0.27**

0.72**

0.80**

0.88**

v max

‐0.09NS

‐0.48**

0.35**

0.32**

0.85**

0.87**

0.26**

0.60**

0.15

NS

0.04

NS

0.62**

0.48**

0.73**

c m1

‐0.07NS

‐0.85**

0.42**

0.47**

0.75**

0.98**

0.15

NS

0.83**

0.27**

‐0.50**

0.84**

0.52**

0.82**

c 1

‐0.32**

‐0.88**

0.40**

0.44**

0.83**

0.94**

0.17

NS

0.81**

0.25*

‐0.41**

0.82**

0.51**

0.82**

c 3

0.02

NS

‐0.02NS

0.30**

‐0.47**

0.23*

0.10

NS

0.12

NS

0.10

NS

‐0.88**

0.12

NS

0.10

NS

‐0.69**

‐0.17NS

DP2

‐0.24*

‐0.75**

0.86**

0.64**

0.51**

0.75**

0.75**

0.22*

0.24*

‐0.52**

1.00**

0.40**

0.90**

DP3

‐0.08NS

‐0.21*

‐0.12NS

0.70**

0.09

NS

0.20

NS

0.20

*‐0.86**

0.03

NS

‐0.20NS

0.25**

0.95**

0.55**

A1

‐0.05NS

0.70**

‐0.14NS

‐0.24*

0.13

NS

‐0.53**

‐0.37**

0.13

NS

‐0.48**

‐0.18NS

‐0.52**

‐0.15NS

‐0.27**

A2

‐0.24*

‐0.75**

0.85**

0.65**

0.55**

0.77**

0.78**

0.21*

1.00**

0.04

NS

‐0.46**

0.42**

0.91**

A3

‐0.10NS

‐0.33**

‐0.01NS

0.74**

0.38**

0.41**

0.43**

‐0.74**

0.17

NS

0.95**

‐0.11NS

0.20*

0.72**

Asum

‐0.28**

‐0.62**

0.72**

0.85**

0.70**

0.73**

0.79**

‐0.11NS

0.84**

0.43**

‐0.17NS

0.86**

0.61**

Chapter2

48

t m

1

t 1

t 2

t e

v max

c m1

c 1

c 3

DP2

DP3

A1

A2

A3

Asum

Exp.2

t m1

0.82**

0.08

NS

‐0.10NS

‐0.45**

‐0.40**

‐0.76**

0.23*

‐0.21*

‐0.18NS

0.47**

‐0.21*

‐0.18NS

‐0.21*

t 1

0.69**

‐0.10NS

‐0.43**

‐0.78**

‐0.86**

‐0.99**

0.42**

‐0.42**

‐0.46**

0.89**

‐0.43**

‐0.46**

‐0.51**

t 2

0.05

NS

‐0.11NS

0.58**

0.48**

0.16

NS

0.13

NS

0.06

NS

0.94**

0.04

NS

‐0.14NS

0.94**

0.05

NS

0.72**

t e

‐0.07NS

‐0.31**

0.45**

0.46**

0.48**

0.45**

‐0.75**

0.67**

0.84**

‐0.53**

0.66**

0.85**

0.97**

v max

‐0.16NS

‐0.34**

0.21*

0.25*

0.76**

0.85**

‐0.19NS

0.69**

0.25**

‐0.75**

0.71**

0.27**

0.60**

c m1

‐0.25**

‐0.84**

0.15

NS

0.35**

0.41**

0.89**

‐0.39**

0.43**

0.49**

‐1.00**

0.44**

0.49**

0.52**

c 1

‐0.66**

‐0.97**

0.13

NS

0.33**

0.52**

0.87**

‐0.40**

0.45**

0.46**

‐0.89**

0.46**

0.47**

0.53**

c 3

0.17

NS

0.22*

0.26**

‐0.64**

‐0.05NS

‐0.18NS

‐0.20*

‐0.08NS

‐0.96**

0.47**

‐0.08NS

‐0.97**

‐0.62**

DP2

‐0.19NS

‐0.45**

0.94**

0.52**

0.30**

0.43**

0.45**

0.16

NS

0.19

NS

‐0.43**

1.00**

0.20*

0.82**

DP3

‐0.12NS

‐0.27**

‐0.20NS

0.78**

0.13

NS

0.28**

0.27**

‐0.89**

‐0.09NS

‐0.55**

0.19

NS

1.00**

0.71**

A1

0.26**

0.86**

‐0.14NS

‐0.33**

‐0.15NS

‐0.93**

‐0.80**

0.17

NS

‐0.43**

‐0.26**

‐0.44**

‐0.56**

‐0.55**

A2

‐0.20NS

‐0.45**

0.94**

0.52**

0.31**

0.43**

0.46**

0.16

NS

1.00**

‐0.09NS

‐0.43**

0.20*

0.82**

A3

‐0.12NS

‐0.27**

‐0.19NS

0.79**

0.18

NS

0.29**

0.29**

‐0.89**

‐0.08NS

1.00**

‐0.25*

‐0.07NS

0.72**

Asum

0.82**

0.08

NS

‐0.10NS

‐0.45**

‐0.40**

‐0.76**

0.23*

‐0.21*

‐0.18NS

0.47**

‐0.21*

‐0.18NS

‐0.21*

Table2.5.(Continued)

**Significantat1%,*Significantat5%,N

S Non‐significant.

Geneticvariationinpotatocanopycoverdynamics

49

Table2.5.(Continued)

**Significantat1%,*Significantat5%,N

S Non‐significant.

t m

1

t 1

t 2

t e

v max

c m1

c 1

c 3

DP2

DP3

A1

A2

A3

Asum

Exp.3

t m1

0.64**

0.69**

0.62**

0.59**

‐0.55**

‐0.32**

0.76**

0.35**

‐0.33**

0.57**

0.57**

0.26**

0.65**

t 1

0.40**

0.85**

0.77**

0.87**

‐0.65**

‐0.39**

0.94**

0.15

NS

‐0.42**

0.99**

0.45**

0.56**

0.88**

t 2

0.51**

0.70**

‐0.94**

0.75**

‐0.67**

‐0.39**

0.92**

0.67**

‐0.47**

0.82**

0.83**

0.29**

0.91**

t e

0.49**

0.69**

0.80**

0.75**

‐0.38**

‐0.23*

0.75**

0.66**

‐0.13NS

0.77**

0.73**

0.50**

0.91**

v max

0.51**

0.79**

0.69**

0.69**

‐0.12NS

0.13

NS

0.87**

0.16

NS

‐0.23*

0.92**

0.52**

0.72**

0.94**

c m1

‐0.02NS

‐0.53**

‐0.22*

‐0.22*

0.04

NS

1.00**

‐0.45**

‐0.33**

0.81**

‐0.49**

‐0.23*

0.22*

‐0.29**

c 1

0.02

NS

‐0.44**

‐0.16NS

‐0.16NS

0.17

NS

0.97**

‐0.16NS

‐0.19NS

0.46**

‐0.24**

‐0.04NS

0.30**

‐0.06NS

c 3

0.42**

0.60**

0.83**

0.48**

0.65**

‐0.09NS

‐0.02NS

0.41**

‐0.79**

0.95**

0.77**

0.28**

0.92**

DP2

0.31**

‐0.03NS

0.69**

0.42**

0.17

NS

0.22*

0.23*

0.56**

‐0.21*

0.12

NS

0.99**

‐0.22*

0.46**

DP3

‐0.21*

‐0.24*

‐0.60**

0.00

NS

‐0.22*

0.08

NS

0.05

NS

‐0.74**

‐0.59**

‐0.35**

‐0.54**

0.50**

‐0.31**

A1

0.37**

0.98**

0.71**

0.71**

0.87**

‐0.37**

‐0.27**

0.65**

0.02

NS

‐0.24*

0.45**

0.63**

0.90**

A2

0.45**

0.22

*0.83**

0.56**

0.47**

0.19

NS

0.24*

0.74**

0.93**

‐0.63**

0.29**

0.01

NS

0.73**

A3

0.18

NS

0.37**

‐0.05NS

0.44**

0.54**

0.07

NS

0.16

NS

‐0.21*

‐0.44**

0.67**

0.43**

‐0.26**

0.59**

Asum

0.53**

0.78**

0.87**

0.87**

0.93**

‐0.05NS

0.06

NS

0.72**

0.43**

‐0.29**

0.85**

0.68**

0.40**

Chapter2

50

**Significantat1%,*Significantat5%,N

S Non‐significant.

Table2.5.(Continued)

t m

1

t 1

t 2

t e

v max

c m1

c 1

c 3

DP2

DP3

A1

A2

A3

Asum

Exp.4

t m1

‐0.29**

0.57**

0.43**

0.25*

0.36**

0.24*

0.40**

0.55**

‐0.25**

‐0.39**

0.54**

‐0.25**

0.47**

t 1

‐0.39**

‐0.51**

‐0.54**

‐0.62**

‐0.86**

‐0.96**

0.11

NS

‐0.70**

‐0.19*

0.97**

‐0.71**

‐0.28**

‐0.64**

t 2

0.28**

‐0.31**

0.92**

0.50**

0.81**

0.65**

0.17

NS

0.97**

0.04

NS

‐0.62**

0.97**

0.10

NS

0.96**

t e

0.27**

‐0.37**

0.82**

0.32**

0.68**

0.60**

‐0.23*

0.91**

0.46**

‐0.64**

0.91**

0.52**

0.98**

v max

0.20*

‐0.34**

0.29**

0.21*

0.52**

0.71**

0.38**

0.59**

‐0.30**

‐0.54**

0.61**

‐0.15NS

0.51**

c m1

0.36**

‐0.81**

0.51**

0.50**

0.31**

0.89**

0.23*

0.90**

‐0.08NS

‐0.97**

0.90**

‐0.01NS

0.79**

c 1

0.31**

‐0.94**

0.43**

0.44**

0.50**

0.90**

0.08

NS

0.82**

0.04

NS

‐0.94**

0.82**

0.13

NS

0.72**

c 3

0.02

NS

0.03

NS

0.34**

‐0.21*

0.18

NS

0.15

NS

0.10

NS

0.11

NS

‐0.95**

0.08

NS

0.12

NS

‐0.95**

‐0.07NS

DP2

0.36**

‐0.59**

0.95**

0.82**

0.36**

0.70**

0.67**

0.28**

0.09

NS

‐0.79**

1.00**

0.17

NS

0.98**

DP3

0.01

NS

‐0.14NS

‐0.21*

0.39**

‐0.12NS

0.04

NS

0.05

NS

‐0.90**

‐0.13NS

‐0.18NS

0.07

NS

0.99**

0.29**

A1

‐0.56**

0.96**

‐0.35**

‐0.41**

‐0.22*

‐0.86**

‐0.90**

0.02

NS

‐0.61**

‐0.14NS

‐0.79**

‐0.25**

‐0.73**

A2

0.36**

‐0.59**

0.95**

0.82**

0.39**

0.70**

0.68**

0.29**

1.00**

‐0.14NS

‐0.61**

0.15

NS

0.97**

A3

0.03

NS

‐0.16NS

‐0.18NS

0.41**

0.03

NS

0.05

NS

0.10

NS

‐0.89**

‐0.10NS

0.99**

‐0.14NS

‐0.10NS

0.36**

Asum

0.26**

‐0.45**

0.91**

0.96**

0.40**

0.57**

0.56**

0.00

NS

0.92**

0.17

NS

‐0.46**

0.92**

0.22*

Geneticvariationinpotatocanopycoverdynamics

51

**Significantat1%,*Significantat5%,N

S Non‐significant.

Table2.5.(Continued)

t m

1

t 1

t 2

t e

v max

c m1

c 1

c 3

DP2

DP3

A1

A2

A3

Asum

Exp.5

t m1

0.12

NS

0.31**

0.40**

0.42**

‐0.18NS

‐0.16NS

0.18

NS

0.20*

0.05

NS

0.10

NS

0.21*

0.00

NS

0.32**

t 1

0.11

NS

‐0.05NS

0.05

NS

‐0.65**

‐0.99**

‐1.00**

‐0.56**

‐0.53**

0.30**

0.99

NS

‐0.52*

0.00

NS

‐0.26**

t 2

0.25*

‐0.04NS

0.97**

0.74**

0.11

NS

0.13

NS

0.79**

0.88**

‐0.75**

0.02

NS

0.88**

0.00

NS

0.99**

t e

0.28**

0.00

NS

0.83**

0.48**

‐0.06NS

‐0.03NS

0.72**

0.80**

‐0.60**

0.08

NS

0.80**

0.00

NS

0.93**

v max

0.30**

‐0.36**

0.47**

0.29**

0.51**

0.61**

1.00**

0.94**

‐1.00**

‐0.54**

0.93**

0.00

NS

0.77**

c m1

‐0.13NS

‐0.92**

0.07

NS

‐0.04NS

0.40**

0.99**

0.87**

0.57**

‐0.53**

‐1.00**

0.56**

0.00

NS

0.25**

c 1

‐0.13NS

‐0.96**

0.10

NS

‐0.01NS

0.49**

0.98**

0.83**

0.59**

‐0.52**

‐0.99**

0.59**

0.00

NS

0.29**

c 3

0.13

NS

‐0.09NS

0.54**

0.08

NS

0.37**

0.17

NS

0.18

NS

0.95**

‐0.85**

‐0.37**

0.97**

0.00

NS

0.98**

DP2

0.16

NS

‐0.52**

0.87**

0.71**

0.57**

0.51**

0.55**

0.50**

‐0.79**

‐0.46**

1.00**

0.00

NS

0.97**

DP3

‐0.06NS

0.07

NS

‐0.65**

‐0.11NS

‐0.44**

‐0.17NS

‐0.18NS

‐0.86**

‐0.59**

0.13

NS

‐0.80**

0.00

NS

‐0.85**

A1

0.07

NS

0.97**

0.02

NS

0.03

NS

‐0.18NS

‐0.91**

‐0.91**

‐0.03NS

‐0.46**

0.00

NS

‐0.46**

0.00

NS

‐0.19NS

A2

0.16

NS

‐0.52**

0.87**

0.71**

0.59**

0.51**

0.56**

0.50**

1.00**

‐0.58**

‐0.45**

0.00

NS

0.97

**

A3

0.00

NS

0.03

NS

‐0.58**

‐0.06NS

‐0.20*

‐0.12NS

‐0.10NS

‐0.86**

‐0.51**

0.96**

0.01

NS

‐0.50**

0.00

NS

Asum

0.23*

‐0.26**

0.92**

0.89**

0.62**

0.25*

0.31**

0.34**

0.91**

‐0.42**

‐0.17NS

0.92**

‐0.31**

Chapter2

52

**Significantat1%,*Significantat5%,N

S Non‐significant.

Table2.5.(Continued)

t m

1

t 1

t 2

t e

v max

c m1

c 1

c 3

DP2

DP3

A1

A2

A3

Asum

Exp.6

t m1

0.71**

0.29**

0.29**

0.00

NS

‐0.41**

‐0.72**

‐0.10NS

0.13

NS

0.11

NS

0.36**

0.13

NS

0.13

NS

0.20*

t 1

0.55

**

‐0.03NS

0.24*

0.00

NS

‐0.95**

‐1.00**

‐0.58**

‐0.24**

0.66**

0.91**

‐0.25**

0.67**

0.07

NS

t 2

0.24*

0.02

NS

0.92**

0.00

NS

0.16

NS

0.02

NS

‐0.13NS

0.98**

0.21*

‐0.18*

0.98**

0.21*

0.97**

t e

0.18

NS

0.16

NS

0.80**

0.00

NS

‐0.14NS

‐0.23*

‐0.53**

0.84**

0.58**

0.16

NS

0.84**

0.57**

0.98**

v max

‐0.18NS

‐0.25**

0.19

NS

0.15

NS

0.00

NS

0.00

NS

0.00

NS

0.00

NS

0.00

NS

0.00

NS

0.00

NS

0.00

NS

0.00

NS

c m1

‐0.20*

‐0.88**

0.11

NS

‐0.06NS

0.26**

0.93**

0.59**

0.35**

‐0.68**

‐1.00**

0.36**

‐0.69**

0.02

NS

c 1

‐0.57**

‐0.99**

‐0.02NS

‐0.15NS

0.30**

0.89**

0.55**

0.23*

‐0.63**

‐0.91**

0.23*

‐0.64**

‐0.07NS

c 3

0.05

NS

‐0.22*

0.14

NS

‐0.42**

0.04

NS

0.24*

0.20*

0.00

NS

‐1.00**

‐0.70**

0.00

NS

‐1.00**

‐0.38**

DP2

0.11

NS

‐0.20*

0.98**

0.75**

0.24*

0.30**

0.19*

0.19

NS

0.06

NS

‐0.38**

1.00**

0.06

NS

0.92**

DP3

‐0.04NS

0.23*

‐0.13NS

0.49**

‐0.03NS

‐0.26**

‐0.22*

‐0.91**

‐0.18NS

0.80**

0.06

NS

1.00**

0.43**

A1

0.15

NS

0.91**

‐0.09NS

0.10

NS

‐0.17NS

‐0.96**

‐0.88**

‐0.28**

‐0.28**

0.29**

‐0.38**

0.80**

‐0.01NS

A2

0.11

NS

‐0.20*

0.98**

0.75**

0.25**

0.30**

0.20*

0.18

NS

1.00**

‐0.18NS

‐0.28**

0.05

NS

0.92**

A3

‐0.03NS

0.25**

‐0.13NS

0.49**

‐0.01NS

‐0.27**

‐0.23*

‐0.91**

‐0.18NS

1.00**

0.30**

‐0.18NS

0.43**

Asum

0.12

NS

0.03

NS

0.91**

0.97**

0.22*

0.05

NS

‐0.03NS

‐0.24*

0.88**

0.28**

‐0.01NS

0.88**

0.29**

Geneticvariationinpotatocanopycoverdynamics

53

forplantswithhighAsum.

3.7.Estimatesofbroad‐senseheritability

Theestimatesofbroad‐senseheritability (H2)across sixenvironments (eqn (2.13))

varied greatly with traits under investigation (Table 2.6). The heritability values

rangedfrom31.1%to96.4%.Highestimates(H2>70%)wererecordedfort2,te,c1,

DP2,A2, andAsum.Moderateestimates (50%<H2<70%)wererecorded for tm1,vmax,

DP3, andA3 (Table2.6).On theotherhand t1andc3hadaweakheritability (37.4%

and31.1%,respectively).However,highlysignificantgeneticvariationinmodeltrait

among the F1 population may offset the relatively low heritability estimates, thus

makingthemresponsivetoselection.AlsoaccordingtoJonesetal.(1986),heritability

estimatesaslowas40%couldbeconsideredfavourableprovidedthattheselection

techniques have enough precision. Table 2.6 also presents the estimates of broad‐

Table2.6. Broad‐sense heritabilityH2 (%) estimates across six experiments (eqn

2.13)andperindividualexperiment(eqn2.14)forfivemodelparameters(tm1,t1,t2,

te,vmax)andninederivedsecondarytraits(cm1,c1,c3,DP2,DP3,A1,A2,A3,Asum)within

theF1populationperindividualexperiment.

Parameter H2† Exp.1 Exp.2 Exp.3 Exp.4 Exp.5 Exp.6

tm1(td) 58.3 59.2 52.5 68.7 40.8 74.3 69.2

t1(td) 37.4 79.9 55.7 89.2 69.5 77.6 75.6

t2(td) 89.6 70.6 64.6 86.4 85.6 79.0 79.8

te(td) 96.4 89.2 92.1 92.8 92.0 85.8 89.4

vmax(%) 64.6 88.2 22.9 92.6 56.5 46.1 0.0

cm1 (%td‐1) 67.6 82.5 57.8 30.4 66.5 76.4 68.6

c1 (%td‐1) 79.4 91.2 59.3 46.8 72.3 81.4 78.6

c3(%td‐1) 31.1 59.7 55.4 55.5 49.0 9.3 32.9

DP2(td) 85.5 82.6 64.4 54.5 88.6 80.8 81.2

DP3(td) 52.4 60.6 73.3 30.0 51.3 23.3 39.6

A1(td%) 0.0 57.6 41.3 93.2 57.6 70.1 65.5

A2(td%) 88.0 84.8 65.0 72.4 89.3 82.3 81.3

A3(td%) 64.0 69.2 73.0 69.1 48.7 0.0 41.0

Asum(td%) 95.9 92.3 90.6 94.9 94.0 88.1 92.6†Broad‐senseheritabilityacrosssixenvironments.

Chapter2

54

sense heritability (H2) per individual experiment (eqn (2.14)), which, in principle,

should be higher than the estimates across environments, because for individual

environmentsthegeneticandGEinteractioneffectsarelessconfounded. The high heritability estimates showed that most of the traits were little

influencedbytheenvironmentandgeneticdifferencesareexpectedtoremainstable

under varied environmental conditions. Furthermore, high estimates of heritability

indicated that these traits could be used readily in breeding for light interception

efficiencyinpotato.Atraitcanrespondtoselectiononlywhenithasheritablegenetic

variation(FalconerandMackay,1996).

3.8.GeneticmappingAmongatotalof566markers,325segregatedduetopolymorphisminthematernal

(SH)parent<abaa>,while241segregatedfromthepaternal(RH)parent<aaab>.Outof566markers,atotalof407markersweremappedandthetotaldatasetwas

split into maternal (SH) and paternal (RH) data sets. Lack of sufficient bridging

markerspreventedmakinganintegratedmap.

Thematernaldatasetcouldbesplitinto12linkagegroupsatarecombination

frequencythresholdof0.25.Twelveparentspecificlinkagegroupswereobtainedfor

both SH and RH (Figs. 2.8‐2.9). However, linkage group I was divided into two

subgroups(denotedasIAandIB,respectively)inthepaternal(RH)mapduetoalack

of a sufficient number of in‐between markers. Ninety‐five of the 325 AFLP SH

markerscouldnotbeassignedtotheSHlinkagegroups.IncaseofRHmarkers,64out

of241markerscouldnotbeassignedtotheRHlinkagegroups.ThenumberofAFLP

markers finally retained in thematernal andpaternalmapswas therefore230 and

177,respectively,coveringthegenomesizeof1902.9cM.

The length of the linkage groups in SHparentalmap ranged from35.7 cM to

129.3 cMwithamediandistanceof2.0 cMbetween the loci.TheRHparentalmap

rangedfrom28.0to101.1cMandthemediandistancebetween lociwas2.5cM. In

bothparentalmaps,thelargestgapbetweenlociwasonlinkagegroupXof10.5cM

and14.1cMinSHandRH,respectively.

Our linkagemapwasgenerallyconsistentwith theultra‐densemapdescribed

byVanOsetal.(2006).However,theSHlinkagegroupsVIIIandXwere37%andRH

linkagegroupVIIwas29%shorterthanintheultra‐densemapofVanOsetal.(2006).

Thisdiscrepancymay simplebedue todifferences in the sizeof SHRHmappingpopulations(cf.MaterialsandMethods).

Geneticvariationinpotatocanopycoverdynamics

55

EA

AC

MC

CA

_296

.5__

1_2

0.0

EA

CT

MC

TC

_240

.1__

1_2

2.0

PA

G/M

AG

C_1

26.5

__1_

23.

3P

AC

/MA

CT

_232

.4__

1_2

6.6

PA

C/M

AC

T_1

46.6

__1_

47.

8E

AG

TM

CA

G_1

98__

1_5

10.8

PC

A/M

AC

T_2

80.5

__1_

2024

.3(D

P2_

Exp

-4/A

2_E

xp-4

)---

----

EA

GT

MC

AG

_458

__1_

3033

.1E

AC

TM

CT

C_3

59.5

__1_

3135

.0(c

m1_

Exp

-4/c

1_E

xp-4

)---

----

EA

AC

MC

AG

_187

.9__

1_32

36.3

EA

TG

MC

AG

_147

.0__

1_32

37.1

PA

G/M

AC

T_1

29.5

__1_

3237

.5P

AC

/MA

GG

_107

.0__

1_32

37.9

EA

CA

MC

AG

_301

.9__

1_32

38.3

(vm

ax_E

xp-3

)---

----

EA

CA

MC

AG

_381

.9__

1_32

38.7

EA

GT

MC

AG

_110

__1_

3239

.1E

AG

AM

CT

C_4

03__

1_32

40.1

(t2_

Exp

-4,5

/DP

3_E

xp-5

/Asu

m_E

xp-4

)---

----

EA

AC

MC

CT

_217

.8__

1_32

41.6

EA

CA

MC

GT

_169

.6__

1_32

42.8

(A2_

Exp

-2)-

----

--P

AC

/MA

GC

_62.

6__1

_32

44.0

EA

GT

MC

AC

_119

__1_

3245

.2E

AA

CM

CC

T_3

31.6

__1_

3245

.8E

AC

TM

CT

G_1

15.1

__1_

3246

.4E

AA

CM

CA

G_2

04.5

__1_

3248

.1E

AA

CM

CA

G_8

1.5_

_1_3

249

.3E

AA

CM

CA

G_3

05.0

__1_

3250

.1E

AA

CM

CC

A_2

49.0

__1_

3251

.7E

AA

CM

CC

A_1

82.8

__1_

3252

.5E

AG

AM

CA

G_1

94.9

__1_

3254

.1P

CT

/MA

GG

_85.

9__1

_32

55.9

(A1_

Exp

-4)-

----

--P

CA

/MA

AC

_211

.6__

1_32

57.8

EA

TG

MC

AG

_510

.0__

1_32

59.4

PA

C/M

AA

C_2

88.4

__1_

3259

.8P

CT

/MA

GG

_106

.3__

1_32

60.2

EA

CA

MC

CT

_131

.7__

1_32

60.6

(DP

2_E

xp-4

/A2_

Exp

-4)-

----

--E

AG

AM

CA

G_2

28.3

__1_

3664

.0(t

1_E

xp-4

/vm

ax_E

xp-3

/cm

1_E

xp-4

/Asu

m_E

xp-3

)---

----

PA

G/M

AC

C_3

22.2

__1_

4269

.0(t

2_E

xp-4

)---

----

PA

T/M

AA

C_2

07.7

__1_

5280

.7(t

m1_

Exp

-5/A

sum

_Exp

-4)-

----

--E

AA

CM

CT

G_1

93.9

__1_

5281

.5P

AC

/MA

CT

_231

.3__

1_54

84.3

PA

C/M

AA

C_1

33.6

__1_

7094

.3E

AG

AM

CT

C_1

35__

1_75

100.

0P

AC

/MA

TG

_201

.9__

1_77

102.

1P

AC

/MA

GC

_270

.3__

1_82

107.

1P

AT

/MA

GG

_178

.6__

1_82

108.

8(D

P2_

Exp

-4/A

2_E

xp-4

)---

----

PA

T/M

AA

C_2

59.4

__1_

8411

0.0

PC

T/M

AG

G_2

36.9

__1_

8611

1.6

PA

C/M

AG

G_1

39.0

__1_

8611

4.5

PC

T/M

AG

G_1

51.2

__1_

9011

8.7

(t2_

Exp

-4)-

----

--P

AC

/MA

CT

_132

.6__

1_92

122.

0P

AG

/MA

CT

_190

.5__

1_95

124.

0P

AC

/MA

GG

_221

.8__

1_95

125.

6E

AC

TM

CT

G_3

19.0

__1_

9512

7.3

PC

A/M

AC

T_1

83.2

__1_

9512

9.3

SH

I

EA

CT

MC

TG

_13

3.9

__2

_40

.0

EA

GT

MC

AC

_31

7__

2_4

11.

5E

AA

CM

CC

T_1

77.

9__

2_4

11.

9E

AA

CM

CA

G_5

84.

6__

2_4

12.

4P

CT

/MA

GT

_50

3.9

__2

_21

3.5

PC

T/M

AG

G_1

39.

2__

2_2

15.

5

EA

AC

MC

AG

_23

6.3

__2

_12

24.

2

PA

G/M

AC

T_

133

.4_

_2_

203

2.0

PA

G/M

AG

T_

377

.9_

_2_

203

2.4

(A1_

Exp

-1)-

----

--E

AA

CM

CC

T_1

38.3

__2_

253

7.5

EA

AC

MC

AG

_22

6.4

__2

_43

53.

9

EA

CT

MC

AC

_21

1.2

__2

_52

60.

8

EA

GA

MC

AG

_14

2.7

__2

_56

64.

5

EA

CT

MC

TC

_18

4.1

__2

_88

95.

5

EA

GT

MC

AC

_397

__2

_95

108

.0

SH

II

PA

C/M

AG

T_1

13.6

__3_

140

.0

EA

GT

MC

AC

_212

__3_

12

2.5

EA

CT

MC

AG

_210

.1__

3_2

22.

9

EA

TG

MC

TC

_290

.5__

3_21

38.

8

PA

C/M

AG

G_1

58.8

__3_

425

7.5

EA

CT

MC

AC

_172

.0__

3_42

58.

3P

AT

/MA

GT

_397

.4__

3_4

25

9.5

PA

C/M

AG

T_2

61.8

__3_

637

5.2

EA

AC

MC

CA

_153

.6__

3_71

81.

8

PA

C/M

AG

C_3

07.7

__3_

758

6.4

SH

III

Figure2.8.LocationsofAFLPmarkersonmaternal(SH)map

(linkagegroupsI‐III).Thenum

beronrightsideisthegenetic

distanceincentiMorgans(cM

),leftsideismarkerdesignations.The

markerinboldshow

sthepositionofQTLassociatedwithtraits,

acrossenvironments.

Chapter2

56

EA

CT

MC

TC

_179

.3__

4_1

0.0

EA

CT

MC

TG

_207

.2__

4_13

10.7

EA

GA

MC

AG

_118

.4__

4_21

16.6

EA

GA

MC

AG

_178

.6__

4_21

17.0

(DP

2_E

xp2)

----

---E

AC

TM

CT

G_7

4.3_

_4_2

824

.8P

AC

/MA

CT

_198

.4__

4_28

25.6

EA

CT

MC

AC

_179

.3__

4_30

26.8

EA

GA

MC

AG

_137

.4__

4_30

28.0

EA

GT

MC

AG

_126

__4_

3229

.7E

AG

TM

CA

C_8

9__4

_31

30.5

EA

GA

MC

TC

_110

__4_

3131

.3

PA

C/M

AG

C_7

7.2_

_4_5

049

.4E

AA

CM

CT

G_3

06.1

__4_

5252

.3P

AG

/MA

GC

_432

.4__

4_52

52.7

EA

TG

MC

AG

_231

.0__

4_53

53.9

EA

TG

MC

AG

_187

.2__

4_63

61.2

PC

A/M

AG

G_9

3.4_

_4_7

374

.0

SH

IV

(c3_

Exp

-6/D

P3_

Exp

-6/A

3_E

xp-6

)---

----

SH

05B

012_

mat

uri

ty_l

ocu

s0

.0S

H05

B01

4_m

atu

rity_

locu

s0

.8S

H05

B01

5_m

atu

rity_

locu

s1

.2

EA

CT

MC

TG

_30

5__

5_28

13.

0E

AG

TM

CA

C_7

9__

5_28

14.

2

PA

G/M

AC

T_1

18.

5__

5_34

19.

7

PA

C/M

AG

T_2

81.

7__

5_38

22.

6

(vm

ax_E

xp-1

,4)-

----

--P

AC

/MA

AC

_190

.2__

5_44

26.

8E

AA

CM

CA

C_4

56.

2__

5_4

42

9.2

EA

AC

MC

AC

_25

0.1

__5_

44

32.

4E

AA

CM

CA

C_3

77.

6__

5_4

43

3.3

EA

CT

MC

TG

_36

5.9

__5_

44

35.

4E

AC

TM

CT

G_8

6.8

__5_

443

6.6

PA

C/M

AT

G_4

17.

8__

5_44

37.

8E

AC

TM

CT

C_1

92.

3__5

_43

39.

0P

AT

/MA

GG

_20

0.4

__5_

46

40.

6

EA

CT

MC

TG

_10

3.9

__5_

49

46.

6

PC

T/M

AG

G_3

94.

9__

5_5

35

7.1

PA

T/M

AG

G_2

06.

6__

5_5

76

1.4

PA

C/M

AG

C_3

58.

8__

5_7

07

1.6

(Asu

m_E

xp-3

)---

----

PA

C/M

AT

A_2

01.4

__5_

777

7.2

SH

V

PA

G/M

AG

C_3

41.7

__6

_60.

0

PA

G/M

AC

T_1

51.8

__6

_36.

9E

AG

AM

CA

G_2

93.9

__6

_38.

1P

AG

/MA

CT

_266

.8_

_6_1

9.9

PA

G/M

AC

T_1

45.2

__6

_512

.9E

AA

CM

CT

G_2

54.9

__6

_514

.3P

CT

/MA

GG

_114

.9_

_6_6

15.5

PA

C/M

AG

G_

81.2

__6

_615

.9P

AT

/MA

GG

_96

.5_

_6_8

18.8

EA

GT

MC

AG

_121

__6

_919

.6E

AT

GM

CT

C_1

91.

1__

6_11

21.2

EA

CA

MC

AC

_18

4.6_

_6_

1626

.2

PA

C/M

AT

A_3

30.

4__

6_21

35.5

PA

C/M

AC

T_

166.

3__

6_31

42.1

PA

C/M

AT

G_1

29.

6__

6_32

43.5

EA

AC

MC

TG

_15

4.7_

_6_

3344

.3E

AC

AM

CT

G_1

45.

9__

6_33

44.7

EA

AC

MC

AG

_14

1.8

__6_

3951

.1

(tm

1_E

xp-1

/Asu

m_E

xp-2

)---

----

EA

CA

MC

CT

_138

.9__

6_48

61.8

(t2_

Exp

-2/D

P2_

Exp

-2)-

----

--P

AT

/MA

AC

_155

.4__

6_56

73.5

PC

A/M

AA

C_2

73.

6__

6_63

82.0

PC

T/M

AG

G_

363.

2__

6_68

87.2

EA

GA

MC

TC

_21

2__

6_68

88.0

EA

CT

MC

AG

_29

5.3_

_6_

6892

.2

SH

VI

Figure2.8.(continued)LocationsofAFLPmarkersonmaternal(SH)map(linkagegroupsIV‐VI).Thenum

beronrightsideisthe

geneticdistanceincentiMorgans(cM

),leftsideismarkerdesignations.Themarkerinboldshow

sthepositionofQTLassociated

withtraits,acrossenvironm

ents.

Geneticvariationinpotatocanopycoverdynamics

57

EA

AC

MC

AC

_236

.0__

7_1

0.0

PC

T/M

AG

T_8

0.0

__7_

151

4.9

EA

GA

MC

TC

_461

__7_

252

2.9

PA

C/M

AG

T_

221.

5__

7_48

38.

0P

AG

/MA

GT

_18

5.5

__7_

493

9.7

PC

T/M

AG

G_

265.

3__

7_5

74

6.5

EA

CA

MC

GT

_98.

3__

7_63

52.

9

PA

C/M

AC

T_7

5.3

__7_

696

0.7

PA

C/M

AA

C_

297.

6__

7_70

62.

8E

AA

CM

CA

G_9

3.3

__7_

70

64.

0E

AG

AM

CT

C_3

13__

7_72

65.

2E

AG

TM

CA

C_

583.

1__

7_7

36

6.8

EA

CA

MC

TG

_66.

5__

7_71

67.

9E

AA

CM

CC

T_

136.

4__

7_71

68.

4E

AC

TM

CA

C_

440.

3__

7_71

69.

6E

AA

CM

CA

G_

390.

4__

7_72

73.

4P

CT

/MA

GG

_32

6.3

__7_

72

74.

6

EA

CA

MC

AC

_29

3.0

__7_

83

85.

8

EA

GA

MC

TC

_129

__7_

909

5.0

SH

VII

EA

AC

MC

TG

_11

7.2

__8_

10

.0

EA

CT

MC

TG

_23

8.4

__8_

66

.9

EA

CT

MC

AC

_474

.2__

8_13

14.

6P

AT

/MA

GG

_14

4.1_

_8_1

61

7.5

EA

CA

MC

AC

_87.

4__8

_16

18.

3E

AG

TM

CA

G_

90_

_8_1

72

0.4

EA

CT

MC

TC

_29

4.6_

_8_2

12

5.1

(DP

3_E

xp-6

/A3_

Exp

-6)-

----

--P

AT

/MA

AC

_283

.7__

8_29

33.

2E

AC

TM

CT

C_

126.

7__8

_31

35.

7

SH

VIII

PA

C/M

AT

G_9

4.2_

_9_3

0.0

PA

G/M

AA

G_3

37.6

__9_

1513

.5

EA

CT

MC

AG

_100

.9__

9_18

17.5

EA

AC

MC

CA

_208

.3__

9_19

20.4

PA

G/M

AC

C_2

28.0

__9_

2125

.0E

AC

AM

CC

T_1

27.0

__9_

2126

.6E

AG

TM

CA

C_14

1__

9_20

27.4

EA

AC

MC

CT

_431

.7__

9_21

27.8

EA

CA

MC

AG

_185

.8__

9_21

28.4

EA

CT

MC

AC

_430

.7__

9_21

29.5

EA

TG

MC

TC

_109

.7__

9_32

43.7

PA

G/M

AA

G_3

77.5

__9_

4954

.9P

AT

/MA

GG

_367

.0__

9_50

56.9

EA

CT

MC

TC

_160

.3__

9_74

80.5

(c3_

Exp

-2)-

----

--P

CA

/MA

GG

_294

.2__

9_74

80.9

EA

AC

MC

AC

_269

.9__

9_78

88.3

SH

IX

Figure2.8.(continued)LocationsofAFLPmarkersonmaternal(SH)map(linkagegroupsVII‐IX).Thenum

ber

onrightsideisthegeneticdistanceincentiMorgans(cM

),leftsideismarkerdesignations.Themarkerinbold

show

sthepositionofQTLassociatedwithtraits,acrossenvironm

ents.

Chapter2

58

E

AG

AM

CA

G_2

12.

3__1

1_1

0.0

EA

CA

MC

GT

_66.

3__1

1_6

5.5

PA

G/M

AG

T_1

47.

6__1

1_6

6.3

EA

GA

MC

TC

_28

9__1

1_6

7.1

(t1_

Exp

-6/c

1_E

xp-6

)---

----

EA

CA

MC

AC

_372

__11

_79

.2

(tm

1_E

xp5)

----

---P

AC

/MA

TA

_197

.8__

11_2

02

1.8

EA

CT

MC

TG

_138

.6__

11_4

53

6.7

PA

T/M

AG

G_1

32.7

__11

_45

37.

5

EA

TG

MC

AG

_84

.8__

11_4

84

2.2

PA

G/M

AC

T_1

94.4

__11

_58

57.

7

PA

C/M

AT

G_2

48.9

__11

_65

63.

0E

AC

AM

CA

G_1

52.6

__11

_66

64.

2P

AC

/MA

TA

_345

.5__

11_6

66

6.3

SH

XI

EA

CT

MC

AG

_195

.4__

12_4

90.

0

EA

CA

MC

AC

_187

.1__

12_5

616

.9E

AC

AM

CC

T_2

89.5

__12

_56

17.5

EA

CA

MC

AG

_85.

8__1

2_56

18.6

PA

C/M

AT

G_1

81.1

__12

_58

21.1

PA

T/M

AA

C_1

68.3

__12

_59

21.5

EA

TG

MC

AG

_236

.0__

12_6

021

.9E

AC

TM

CT

G_7

2.9_

_12_

6123

.1E

AA

CM

CC

A_1

65.1

__12

_63

26.4

PA

G/M

AC

T_2

09.8

__12

_72

36.7

SH

XII

PA

C/M

AG

T_1

29.5

__10

_1

0.0

PC

T/M

AG

G_1

27.

5__

10_

16

14.

6

PA

T/M

AG

G_2

38.

6__

10_

32

43.

4

EA

AC

MC

TG

_28

6.4

__1

0_4

85

7.7

PA

C/M

AT

A_2

55.

6__

10_

58

66.

2

EA

CA

MC

AC

_40

5.9

__1

0_6

67

1.4

PA

T/M

AG

G_3

79.

3__

10_

67

73.

5E

AC

TM

CA

G_1

49.

9__

10_

68

75.

1

PC

A/M

AG

G_3

75.

2__

10_

68

87.

5

SH

X

Figure2.8.(continued)LocationsofAFLPmarkersonmaternal(SH)map(linkagegroupsX‐XII).Thenum

ber

onrightsideisthegeneticdistanceincentiMorgans(cM

),leftsideismarkerdesignations.Themarkerinbold

show

sthepositionofQTLassociatedwithtraits,acrossenvironm

ents.

Geneticvariationinpotatocanopycoverdynamics

59

PC

A/M

AA

C_

153.

6__

1_1

0.0

PA

G/M

AG

T_

306.

4__

1_7

6.0

PA

C/M

AC

T_3

89.4

__1_

86

.8E

AG

TM

CA

C_

410

__1_

87

.6E

AA

CM

CC

A_1

27.

1__1

_91

0.0

PA

T/M

AA

C_2

79.8

__1

_10

12.

5E

AT

GM

CT

C_4

62.4

__1

_13

17.

5E

AT

GM

CT

C_1

92.9

__1

_13

18.

9E

AA

CM

CA

G_2

61.4

__1

_13

22.

0E

AA

CM

CA

G_1

39.9

__1

_13

23.

8E

AA

CM

CA

G_1

43.6

__1

_13

25.

1E

AA

CM

CA

G_1

96.4

__1

_13

26.

0E

AA

CM

CC

A_1

89.0

__1

_13

26.

9E

AC

AM

CA

C_2

26.7

__1_

132

8.5

EA

CT

MC

TG

_201

.4_

_1_1

32

9.0

PA

G/M

AG

T_3

09.9

__1

_13

29.

8E

AT

GM

CT

C_

96.7

__1

_13

31.

8E

AC

AM

CA

G_2

21.0

__1_

133

3.0

PA

T/M

AG

G_4

90.1

__1

_13

33.

4P

AC

/MA

GT

_70

.6_

_1_1

33

6.3

EA

CA

MC

TG

_150

__1

_13

37.

5E

AG

TM

CA

C_4

99.6

__1

_13

38.

7E

AC

TM

CT

C_3

77.7

__1

_13

39.

5E

AA

CM

CC

T_1

87.7

__1

_20

46.

4E

AC

AM

CT

G_1

52.6

__1_

204

8.0

(Asu

m_E

xp-6

)---

----

PA

T/M

AG

T_1

43.9

__1_

355

8.3

RH

IA

EA

CA

MC

AG

_58.

5__1

_60

0.0

EA

AC

MC

CT

_11

6.9_

_1_63

26.

1(t

2_E

xp-6

/DP

2_E

xp-6

/A2_

Exp

-6)-

----

--E

AC

AM

CC

T_1

44.

5__1

_63

27.

8

EA

CA

MC

GT

_296

.2__

1_71

33.

9

PA

T/M

AG

G_18

6.7_

_1_

8143.

4P

AT

/MA

GG

_19

4.5_

_1_

8144.

7E

AC

AM

CA

C_1

71.4

__1_

8447.

6

EA

AC

MC

TG

_237

.1__

1_95

55.

9

PC

A/M

AC

T_1

81.3

__1_

9860.

9P

AC

/MA

TA

_101

.3__

1_99

63.

0

RH

IB

EA

AC

MC

AC

_356

.7_

_2_1

0.0

PC

T/M

AG

G_2

12.4

__2

_1

4.2

EA

AC

MC

CT

_314

.9_

_2_1

4.6

EA

AC

MC

AC

_167

.7_

_2_6

10.9

EA

AC

MC

CA

_310

.7__

2_9

14.3

PA

G/M

AC

T_

132.

4__

2_20

23.0

(t2_

Exp

-5)-

----

--P

AG

/MA

CT

_230

.5__

2_21

23.4

EA

CT

MC

AC

_21

7.2_

_2_2

728

.1P

AT

/MA

AC

_57

0.4

__2_

3232

.2P

AC

/MA

CT

_21

1.1

__2_

33

33.5

PA

G/M

AG

C_

254.

4__

2_35

35.6

PA

G/M

AC

T_

146.

4__

2_37

38.2

PA

G/M

AG

C_

238.

8__

2_36

39.4

PC

T/M

AG

T_

225.

5__

2_5

149

.3E

AG

TM

CA

C_2

49__

2_5

150

.9E

AC

AM

CA

C_

266.

0__

2_51

52.1

(Asu

m_E

xp-5

)---

----

EA

CT

MC

TC

_444

.3__

2_72

73.8

PA

C/M

AT

A_9

6.8

__2_

7375

.5(A

sum

_Exp

-3)-

----

--P

AC

/MA

GG

_527

.2__

2_75

79.1

EA

GA

MC

TC

_130

__2_

80

84.7

RH

II

Figure2.9.LocationsofAFLPmarkersonpaternal(RH)map(linkagegroupsI‐II).Thenum

beronrightsideisthegenetic

distanceincentiMorgans(cM

),leftsideismarkerdesignations.Themarkerinboldshow

sthepositionofQTLassociatedwith

traits,acrossenvironm

ents.

Chapter2

60

E

AG

TM

CA

G_

129_

_4_

280.

0(t

1-E

xp-4

)---

----

EA

CA

MC

AC

_156

.4__

4_33

5.5

EA

AC

MC

TG

_23

3.9_

_4_

356.

8E

AG

AM

CT

C_

223_

_4_3

68.

4(t

e_E

xp-3

)---

----

EA

GA

MC

AG

_216

.2__

4_35

9.6

(tm

1_E

xp-2

)---

----

EA

CA

MC

TG

_69.

5__4

_35

11.2

(t1_

Exp

-2)-

----

--E

AC

AM

CT

G_1

38.9

__4_

3512

.8

EA

AC

MC

AC

_42

0.3_

_4_

4223

.5

EA

GA

MC

AG

_42

1.2_

_4_

6037

.0

(te_

Exp

-2)-

----

--E

AA

CM

CC

A_9

2.1_

_4_7

246

.9

RH

IV

EA

AC

MC

AG

_14

9.7_

_5_4

0.0

(c3-

Exp

-6/D

P3_

Exp

-6/A

3_E

xp-6

)---

----

PA

C/M

AG

T_1

90.2

__5_

41.

6E

AG

TM

CA

C_

290_

_5_4

2.8

(A1-

Exp

-1)-

----

--E

AC

AM

CC

T_1

09.4

__5_

44.

5

(tm

1_E

xp-3

/t1_

Exp

-1,3

)---

----

EA

GA

MC

TC

_470

__5_

1718

.2(t

2_E

xp-1

,3,4

,5,6

/te_

Exp

-1,2

,3,4

,5,6

/vm

ax_E

xp-3

)---

----

18.3

(cm

1_E

xp-1

,4/c

1_E

xp-1

,2,4

/c3_

Exp

-2,3

/DP

2_E

xp-1

,2,4

,5,6

/DP

3_E

xp-1

,2)-

----

--18

.4A

1_E

xp-3

/A2_

Exp

-1,4

,5,6

/A3_

Exp

-1,2

,3/A

sum

_Exp

-1,2

,3,4

,5,6

)---

----

18.5

EA

CA

MC

GT

_250

.1__

5_37

35.7

(A2_

Exp

-3)-

----

--E

AA

CM

CC

T_2

01.2

__5_

4243

.9P

CT

/MA

GT

_159

.2__

5_43

46.4

EA

TG

MC

AG

_304

.2__

5_46

48.9

EA

CT

MC

TC

_27

2.3_

_5_4

649

.3E

AC

TM

CT

C_3

44.

6__5

_46

50.5

EA

AC

MC

CA

_410

.3__

5_46

51.3

EA

AC

MC

AG

_135

.2__

5_46

52.5

EA

AC

MC

AG

_231

.8__

5_46

52.6

EA

CA

MC

AG

_102

.4__

5_46

53.3

EA

CA

MC

TG

_92.

9__5

_46

54.1

(c1_

Exp

-4)-

----

--P

AC

/MA

TA

_99.

4__5

_55

62.6

RH

V

PA

T/M

AA

C_2

98.3

__3_

30

.0P

AG

/MA

GT

_228

.5__

3_3

1.7

EA

GA

MC

TC

_13

2__

3_9

9.2

EA

CT

MC

AC

_13

3.1_

_3_1

61

5.8

PC

A/M

AA

C_2

14.

0__3

_35

35.

1

EA

GA

MC

AG

_41

0.1_

_3_3

73

9.5

EA

GA

MC

TC

_245

__3

_52

55.

1

PA

C/M

AG

T_2

67.

0__3

_72

74.

2

PC

A/M

AC

T_2

70.

0__3

_79

85.

5P

AT

/MA

GG

_15

7.1_

_3_8

08

7.2

RH

III

Figure2.9.(continued)LocationsofAFLPmarkersonpaternal(RH)map(linkagegroupsIII‐V).Thenum

beronrightsideis

thegeneticdistanceincentiMorgans(cM

),leftsideismarkerdesignations.Themarkerinboldshow

sthepositionofQTL

associatedwithtraits,acrossenvironm

ents.

Geneticvariationinpotatocanopycoverdynamics

61

EA

AC

MC

AG

_12

5.4b

__6_

10.

0

PA

T/M

AG

T_2

05.1

__6_

32

5.1

PA

G/M

AC

T_1

55.9

__6_

32

6.4

EA

AC

MC

CT

_37

7.0_

_6_1

53

3.2

EA

CA

MC

AG

_37

9.9_

_6_1

73

4.4

EA

CA

MC

TG

_269

.7__

6_18

35.

2E

AC

AM

CA

G_

193.

6__6

_17

35.

8E

AA

CM

CC

A_

231.

3__6

_16

37.

7E

AG

AM

CT

C_

539.

6__6

_21

40.

9E

AG

AM

CT

C_

466_

_6_2

14

1.8

EA

GA

MC

TC

_49

6.3_

_6_2

14

3.4

EA

GA

MC

TC

_52

6.0_

_6_2

14

3.8

EA

AC

MC

TG

_50

1.8_

_6_2

24

8.0

EA

CA

MC

AC

_390

.6__

6_23

48.

8E

AC

TM

CT

G_1

67.3

__6_

285

3.4

(te_

Exp

-3)-

----

--P

AT

/MA

AC

_272

.6__

6_28

53.

8(t

2_E

xp-3

)---

----

EA

CT

MC

AG

_39

3.7_

_6_3

45

9.7

PA

C/M

AT

G_1

09.4

__6_

436

6.0

EA

GT

MC

AG

_31

9__6

_45

66.

9E

AA

CM

CT

G_

152.

6__6

_46

69.

4E

AC

AM

CA

C_8

9.5_

_6_4

97

2.7

EA

GA

MC

AG

_50

7.5_

_6_4

97

4.1

EA

AC

MC

CT

_14

4.2_

_6_5

07

6.0

PA

T/M

AG

T_

251.

9__6

_52

78.

8

PA

C/M

AG

G_2

26.3

__6_

6910

1.1

RH

VI

PA

G/M

AG

T_1

80.1

__7_

440.

0P

AT

/MA

GT

_216

.8__

7_43

0.8

EA

GA

MC

AG

_555

.8__

7_66

19.5

EA

CT

MC

TG

_272

.0__

7_67

19.9

EA

CT

MC

TC

_337

.5__

7_68

20.3

(cm

1_E

xp-4

/c1_

Exp

-4)-

----

--P

AC

/MA

AC

_164

.8__

7_68

21.1

EA

CA

MC

AG

_90

.1__

7_73

28.0

RH

VII

PA

C/M

AT

A_2

66.2

__8_

50.

0

EA

AC

MC

AG

_107

.2__

8_2

210

.8E

AA

CM

CC

A_2

50.2

__8_

22

11.6

EA

AC

MC

CA

_437

.7__

8_2

213

.2E

AA

CM

CC

A_1

41.0

__8_

26

15.3

PC

T/M

AG

G_1

28.9

__8_

24

17.4

EA

CT

MC

AG

_274

.9__

8_3

628

.0

EA

GA

MC

TC

_19

8__8

_47

42.3

EA

CT

MC

TG

_140

.2__

8_8

273

.6P

AT

/MA

GG

_149

.8__

8_8

374

.5P

AG

/MA

GC

_298

.8__

8_8

376

.1

EA

AC

MC

AG

_290

a__8

_82

90.0

RH

VIII

EA

GT

MC

AG

_70_

_9_

10.

0E

AG

TM

CA

C_1

28__

9_1

2.1

EA

CA

MC

TG

_157

.0__

9_2

020

.5

PC

T/M

AG

G_1

15.9

__9

_33

38.4

PA

G/M

AA

G_1

14.5

__9

_55

60.1

EA

AC

MC

CT

_287

.5__

9_6

066

.8

PC

T/M

AG

G_4

86.3

__9

_67

76.6

PA

G/M

AG

C_

66.5

__9

_78

84.7

RH

IX

Figure2.9.(continued)LocationsofAFLPmarkersonpaternal(RH)map(linkagegroupsVI‐IX).Thenum

beronrightsideisthe

geneticdistanceincentiMorgans(cM

),leftsideismarkerdesignations.Themarkerinboldshow

sthepositionofQTLassociated

withtraits,acrossenvironm

ents.

Chapter2

62

P

AT

/MA

AC

_583

.1__

12_1

0.0

EA

GA

MC

TC

_372

__12

_54.

2

EA

CA

MC

AC

_243

.3__

12_2

123

.5

PC

T/M

AG

G_9

6.5_

_12_

2931

.7

PA

C/M

AC

T_2

29.7

__12

_43

43.8

PA

G/M

AC

C_2

29.9

__12

_46

47.2

EA

GA

MC

AG

_289

.0__

12_4

648

.4E

AG

TM

CA

G_2

44__

12_4

950

.5E

AG

TM

CA

C_1

11__

12_4

951

.3E

AA

CM

CA

G_2

86.3

__12

_49

51.7

EA

CT

MC

AG

_251

.5__

12_4

952

.1

EA

GA

MC

TC

_210

__12

_49

60.8

RH

XII

EA

AC

MC

AG

_409

.1__

10_1

0.0

PA

C/M

AG

G_2

71.

1__

10_1

110

.3

EA

AC

MC

CA

_21

6.8_

_10_

3431

.4

EA

TG

MC

AG

_14

0.3_

_10

_64

59.7

EA

AC

MC

TG

_11

2.7_

_10_

6460

.9

EA

CA

MC

AC

_51

3.7_

_10_

7975

.0

RH

X

EA

AC

MC

AC

_106

.6__

11_4

0.0

EA

AC

MC

AG

_291

.7__

11_

55.

3

EA

CT

MC

TG

_58

3.0_

_11_

22

20.8

PA

C/M

AG

C_

108.

2__1

1_27

28.7

EA

CA

MC

AC

_16

1.6_

_11_

3537

.9

EA

CT

MC

AG

_20

3.4_

_11_

4346

.4

PA

C/M

AG

C_

334.

1__1

1_61

57.3

EA

AC

MC

AG

_15

6.4_

_11_

6159

.0

EA

GT

MC

AC

_367

__1

1_8

278

.2P

AC

/MA

TG

_38

1.9_

_11_

84

81.5

PA

C/M

AT

A_

343.

3__1

1_84

82.4

EA

AC

MC

AC

_91.

1__1

1_84

84.0

RH

XI

Figure2.9.(continued)LocationsofAFLPmarkersonpaternal(RH)map(linkagegroupsX‐XII).The

numberonrightsideisthegeneticdistanceincentiMorgans(cM

),leftsideismarkerdesignations.

Themarkerinboldshow

sthepositionofQTLassociatedwithtraits,acrossenvironm

ents.

Geneticvariationinpotatocanopycoverdynamics

63

3.9.QTLdetectionQTL analysis for the fivemodel parameters and nine derived traitswas conducted

separately for the six environments (i.e. experiments). In total 41 QTLs were

identifiedonbothSHandRHparentalgenomesacrossallsixenvironments(Figs.2.8‐

2.9).IntheSHgenome,23QTLswereassociatedwitheightlinkagegroups(I,II,IV,V,

VI, VIII, IX, and XI). Eighteen QTLs were shared by the RH genome on six linkage

groups (I, II, IV, V, VI, and VII). The numbers of QTLs detected for each parental

chromosomearelistedin(Table2.7).

Table2.8summarisesthelistofQTLsdetected,theirparentalchromosomesand

mappositions and their characteristics (i.e. additive effects and variance explained

(R2) for each of the trait investigated for individual environment). All the QTLs

detected were significant at (P<0.05) with –log10(P) values ranging from 3.47 to

52.33.ThetotalfractionofphenotypicvarianceexplainedbyeffectsofeachQTLwere

moderate(rangingfrom<0.1%to74%),butahighpercentageofphenotypicvariance

wasaccountedforwhenconsideringtheglobalR2(rangingfrom28%to82%)(Table

2.8).

DetailedinvestigationoftheQTLsidentifiedforfivemodelparametersshowed

thatfiveQTLsweredetectedfortm1scatteredacrosslinkagegroups(SHI,SHVI,SH

XI,RHIV,andRHV)inalltheexperimentsexceptExps4and6,wherenoQTLwas

found(Table2.8).ThefractionofphenotypicvarianceexplainedbyindividualQTLs

ranged from 12% to 16%. The QTL (116_5_17) detected in Exp 3 shared the

maximum negative additive effect (‐4.1 td) and explained the maximum (16%)

phenotypic variance. However, two QTLs detected in Exp 5 (i.e. 66_1_52 and

Table2.7.DistributionofQTLsdetectedonparentalgenomes.

SH RH

Linkagegroup No.ofQTLs Linkagegroup No.ofQTLs

I 12 IA 1

II 1 IB 1

IV 1 II 3

V 3 IV 5

VI 2 V 5

VIII 1 VI 2

IX 1 VII 1

XI 2

Total 23 18

Chapter2

64

Table2.8.Maincharacteristicsofquantitativetraitloci(QTL)identifiedforfivemodelparam

eters(tm1,t 1,t2,t e,v

max)

andninederivedsecondarytraits(cm1,c 1,c3,DP2,D

P3,A1,A2,A3,Asum)withinthe‘SHRH’populationperindividual

experiment(i.e.environm

ent.).DatagivenintablearefromtheCIMm

appingm

ethod.QTLsm

arkedasboldare

detectedonlybytheCIMmethod,otherwisebyboththeCIMandSIMmethods.Exp.,experiment;position,positionof

maximum

–log 1

0(P);a,additiveeffectofthepresenceofparentalalleleatamarker;R

2 ,theindividualcontributionof

oneQTLtothevariationinatrait;globalR2 ,thefractionofthetotalvariationexplainedbyQTLsofthesametrait

withinsingleenvironment;td,thermalday.Sym

bol‘−’meansnoQTLwasdetected.

Parameter

Exp.

QTL

Linkage

group

Markername

Position

(cM)

‐log

10(P)

aR2

GlobalR

2

t m1(td)

1199_6_48

SHVI

EACAMCCT_138.9__6_48

61.8

3.47

2.902

0.13

299_4_35

RHIV

EACAMCTG_69.5__4_35

11.2

3.67

1.975

0.14

3116_5_17

RHV

EAGAMCTC_470__5_17

18.2

3.95

‐4.113

0.16

4−

−−

−−

−−

566_1_52

SHI

EAACM

CTG_193.9__1_52

81.5

4.61

2.427

0.16

0.29

293_11_20

SHXI

PAC/MATA_197.8__11_20

21.8

3.584

‐2.095

0.12

6−

−−

−−

−−

t 1(td)

1116_5_17

RHV

EAGAMCTC_470__5_17

18.2

6.23

7.404

0.23

298_4_35

RHIV

EACAMCTG_138.9__4_35

12.8

3.48

3.167

0.13

3116_5_17

RHV

EAGAMCTC_470__5_17

18.2

18.02

‐26.586

0.60

463_1_42

SHI

PAG/M

ACC_322.2__1_42

69

4.738

‐5.754

0.17

0.28

96_4_33

RHIV

EACAMCAC_156.4__4_33

5.5

3.423

4.691

0.11

5−

−−

−−

−−

6291_11_7

SHXI

EACAMCAC_372__11_7

9.2

3.371

‐2.591

0.14

Geneticvariationinpotatocanopycoverdynamics

65

Parameter

Exp.

QTL

Linkage

group

Markername

Position

(cM)

‐log

10(P)

aR2

GlobalR

2

t 2(td)

1116_5_17

RHV

EAGAMCTC_470__5_17

18.2

6.66

‐9.584

0.24

2200_6_56

SHVI

PAT/M

AAC_155.4__6_56

73.5

3.8

8.055

0.15

3116_5_17

RHV

EAGAMCTC_470__5_17

18.2

19.52

‐31.694

0.49

0.56

154_6_34RHVI

EACTMCAG_393.7__6_34

59.7

3.53

10.453

0.04

4116_5_17

RHV

EAGAMCTC_470__5_17

18.2

17.24

‐26.021

0.55

0.67

39_1_32

SHI

EAACM

CCT_217.8__1_32

41.6

1.445

6.521

0.10

67_1_52

SHI

PAT/M

AAC_207.7__1_52

80.7

0.248

1.915

0.01

81_1_92

SHI

PAC/MACT_132.6__1_92

122

1.555

5.433

0.01

5116_5_17

RHV

EAGAMCTC_470__5_17

18.2

16.008

‐27.999

0.52

0.64

39_1_32

SHI

EAACM

CCT_217.8__1_32

41.6

3.096

9.011

0.14

64_2_21

RHII

PAG/M

ACT_230.5__2_21

23.4

1.913

‐6.698

0.01

6116_5_17

RHV

EAGAMCTC_470__5_17

18.2

10.018

‐18.544

0.27

0.43

47_1_63

RHIB

EACAMCCT_144.5__1_63

27.8

5.463

11.577

0.08

Table2.8.(Continued)

Chapter2

66

Parameter

Exp.

QTL

Linkage

group

Markername

Position

(cM)

‐log

10(P)

aR2

GlobalR

2

t e(td)

1116_5_17

RHV

EAGAMCTC_470__5_17

18.2

19.5

‐19.351

0.51

2116_5_17

RHV

EAGAMCTC_470__5_17

18.2

24.25

‐27.232

0.55

0.62

108_4_72RHIV

EAACM

CCA_92.1__4_72

46.9

3.93

8.408

0.02

3116_5_17

RHV

EAGAMCTC_470__5_17

18.2

26.91

‐29.497

0.58

0.68

101_4_35RHIV

EAGAMCAG_216.2__4_35

9.6

3.95

7.58

0.03

153_6_28RHVI

PAT/M

AAC_272.6__6_28

53.8

3.62

7.112

0.02

4116_5_17

RHV

EAGAMCTC_470__5_17

18.2

20.612

‐32.356

0.65

5116_5_17

RHV

EAGAMCTC_470__5_17

18.2

19.527

‐24.133

0.63

6116_5_17

RHV

EAGAMCTC_470__5_17

18.2

52.33

‐49.912

0.74

v max(%)

1170_5_44

SHV

PAC/MAAC_190.2__5_44

26.8

5.66

22.66

0.21

2−

−−

−−

−−

3116_5_17

RHV

EAGAMCTC_470__5_17

18.2

26.28

‐54.862

0.56

0.65

21_1_32

SHI

EACAMCAG_381.9__1_32

38.7

3.8

4.243

0.02

63_1_42

SHI

PAG/M

ACC_322.2__1_42

69

4.78

16.533

0.12

4

170_5_44

SHV

PAC/MAAC_190.2__5_44

26.8

4.291

8.836

0.18

5

−−

−−

−−

−

6

−−

−−

−−

−

Table2.8.(Continued)

Geneticvariationinpotatocanopycoverdynamics

67

Parameter

Exp.

QTL

Linkage

group

Markername

Position

(cM)

‐log

10(P)

aR2

GlobalR

2

c m1(%td

‐1)

1116_5_17

RHV

EAGAMCTC_470__5_17

18.2

7.8

‐2.413

0.28

2

−−

−−

−−

−

3

−−

−−

−−

−

4

116_5_17

RHV

EAGAMCTC_470__5_17

18.2

5.21

‐3.987

0.16

0.41

18_1_32

SHI

EAACM

CAG_187.9__1_32

36.3

4.23

1.272

0.06

63_1_42

SHI

PAG/M

ACC_322.2__1_42

69

4.27

2.381

0.12

173_7_68RHVII

PAC/MAAC_164.8__7_68

21.1

3.55

2.997

0.08

5

−−

−−

−−

−

6

−−

−−

−−

−

c 1(%td

‐1)

1116_5_17

RHV

EAGAMCTC_470__5_17

18.2

7.34

‐1.165

0.26

2

116_5_17

RHV

EAGAMCTC_470__5_17

18.2

3.63

‐0.693

0.14

3

−−

−−

−−

−

4

116_5_17

RHV

EAGAMCTC_470__5_17

18.2

3.88

‐1.055

0.13

0.44

18_1_32

SHI

EAACM

CAG_187.9__1_32

36.3

5.82

1.236

0.18

131_5_55RHV

PAC/MATA_99.4__5_55

62.6

3.49

‐0.387

0.03

173_7_68

RHVII

PAC/MAAC_164.8__7_68

21.1

3.80

1.081

0.11

5

−−

−−

−−

−

6291_11_7

SHXI

EACAMCAC_372__11_7

9.2

3.477

0.899

0.14

Table2.8.(Continued)

Chapter2

68

ParameterExp.

QTL

Linkage

group

Markername

Position

(cM)

‐log

10(P)

aR2

GlobalR

2

c 3(%td

‐1)

1−

−−

−−

−−

2

116_5_17RHV

EAGAMCTC_470__5_17

18.2

8.76

4.649

0.29

0.39

265_9_74SHIX

PCA/M

AGG_294.2__9_74

80.9

3.64

2.619

0.13

3

116_5_17RHV

EAGAMCTC_470__5_17

18.2

6.905

‐5.338

0.28

4

−−

−−

−−

−

5

−−

−−

−−

−

6

1_5_12

SHV

SH05B012_maturity_locus

0

3.28

3.595

0.07

0.28

115_5_4

RHV

PAC/MAGT_190.2__5_4

1.6

5.641

4.523

0.18

DP2(td)

1116_5_17RHV

EAGAMCTC_470__5_17

18.2

11.48

‐16.987

0.37

2

133_4_28SHIV

EACTMCTG_74.3__4_28

24.8

4.09

6.655

0.09

0.35

200_6_56SHVI

PAT/M

AAC_155.4__6_56

73.5

4.30

6.996

0.13

116_5_17RHV

EAGAMCTC_470__5_17

18.2

4.08

‐10.208

0.14

3

−−

−−

−−

−

4

116_5_17RHV

EAGAMCTC_470__5_17

18.2

26.28

‐30.089

0.56

0.70

14_1_30

SHI

EAGTMCAG_458__1_30

33.1

6.82

2.83

0.03

62_1_36

SHI

EAGAMCAG_228.3__1_36

64

8.39

9.433

0.15

76_1_84

SHI

PAT/M

AAC_259.4__1_84

110

4.15

4.61

0.02

5

116_5_17RHV

EAGAMCTC_470__5_17

18.2

11.983

‐31.322

0.45

6

116_5_17RHV

EAGAMCTC_470__5_17

18.2

8.355

‐17.625

0.22

0.38

47_1_63

RHIB

EACAMCCT_144.5__1_63

27.8

4.686

11.283

0.04

Table2.8.(Continued)

Geneticvariationinpotatocanopycoverdynamics

69

ParameterExp.

QTL

Linkage

group

Markername

Position

(cM)

‐log

10(P)

aR2

GlobalR

2

DP3(td)

1116_5_17RHV

EAGAMCTC_470__5_17

18.2

5.26

‐9.768

0.20

2

116_5_17RHV

EAGAMCTC_470__5_17

18.2

14.39

‐19.569

0.42

3

−−

−−

−−

−

4

−−

−−

−−

−

5

39_1_32

SHI

EAACM

CCT_217.8__1_32

41.6

4.101

‐8.657

0.17

6

1_5_12

SHV

SH05B012_maturity_locus

0

4.573

‐8.988

0.08

0.46

242_8_29SHVIII

PAT/M

AAC_283.7__8_29

33.2

4.134

‐6.785

0.08

115_5_4

RHV

PAC/MAGT_190.2__5_4

1.6

9.326

‐12.803

0.24

A1(td%)

1102_2_25SHII

EAACM

CCT_138.3__2_25

37.5

3.48

275.871

0.13

0.33

114_5_4

RHV

EACAMCCT_109.4__5_4

4.5

6.23

412.691

0.22

2

−−

−−

−−

−

3116_5_17RHV

EAGAMCTC_470__5_17

18.2

20.042

‐2246.416

0.64

4

58_1_32

SHI

PCA/M

AAC_211.6__1_32

57.8

4.77

‐458.085

0.18

5

−−

−−

−−

−

6

−−

−−

−−

−

Table2.8.(Continued)

Chapter2

70

Parameter

Exp.

QTL

Linkage

group

Markername

Position

(cM)

‐log

10(P)

aR2

GlobalR

2

A 2(td%)

1116_5_17

RHV

EAGAMCTC_470__5_17

18.2

12.35

‐1764.776

0.39

2

53_1_32

SHI

PAC/MAGC_62.6__1_32

44

4.5

987.678

0.17

3

118_5_42

RHV

EAACM

CCT_201.2__5_42

43.9

4.398

‐1215.599

0.18

4

116_5_17

RHV

EAGAMCTC_470__5_17

18.2

25.9

‐3039.945

0.55

0.70

62_1_36

SHI

EAGAMCAG_228.3__1_36

64

8.48

973.111

0.15

76_1_84

SHI

PAT/M

AAC_259.4__1_84

110

4.17

469.087

0.02

14_1_30

SHI

EAGTMCAG_458__1_30

33.1

6.86

281.737

0.03

5116_5_17

RHV

EAGAMCTC_470__5_17

18.2

16.32

‐3198.355

0.45

6116_5_17

RHV

EAGAMCTC_470__5_17

18.2

8.35

‐1768.77

0.24

0.37

47_1_63

RHIB

EACAMCCT_144.5__1_63

27.8

4.634

1124.769

0.07

A 3(td%)

1116_5_17

RHV

EAGAMCTC_470__5_17

18.2

7.32

‐771.481

0.26

2

116_5_17

RHV

EAGAMCTC_470__5_17

18.2

14.76

‐1251.073

0.43

3

116_5_17

RHV

EAGAMCTC_470__5_17

18.2

4.473

‐787.222

0.18

4

−−

−−

−−

−

5

39_1_32

SHI

EAACM

CCT_217.8__1_32

41.6

3.68

‐436.12

0.14

6

1_5_12

SHV

SH05B012_maturity_locus

0

4.37

‐559.686

0.08

0.46

242_8_29SHVIII

PAT/M

AAC_283.7__8_29

33.2

3.85

‐422.482

0.08

115_5_4

RHV

PAC/MAGT_190.2__5_4

1.6

6.72

‐804.803

0.24

Table2.8.(Continued)

Geneticvariationinpotatocanopycoverdynamics

71

Parameter

Exp.

QTL

Linkage

group

Markername

Position

(cM)

‐log

10(P)

aR2

GlobalR

2

A sum(td%)

1116_5_17

RHV

EAGAMCTC_470__5_17

18.2

19.65

‐2359.408

0.50

0.57

43_1_35

RHIA

PAT/M

AGT_143.9__1_35

58.3

3.6

786.981

<0.1

2

116_5_17

RHV

EAGAMCTC_470__5_17

18.2

21.09

‐2188.154

0.52

0.59

199_6_48SHVI

EACAMCCT_138.9__6_48

61.8

3.91

719.654

0.03

3

116_5_17

RHV

EAGAMCTC_470__5_17

18.2

28.68

‐4112.081

0.69

0.82

63_1_42

SHI

PAG/M

ACC_322.2__1_42

69

3.953

878.855

0.07

179_5_77SHV

PAC/MATA_201.4__5_77

77.2

4.883

974.202

<0.1

81_2_75

RHII

PAC/MAGG_527.2__2_75

79.1

4.481

‐938.074

<0.1

4116_5_17

RHV

EAGAMCTC_470__5_17

18.2

40.06

‐2996.801

0.67

0.73

66_1_52

SHI

EAACM

CTG_193.9__1_52

81.5

4.11

419.656

0.06

39_1_32

SHI

EAACM

CCT_217.8__1_32

41.6

4.51

552.856

0.08

5116_5_17

RHV

EAGAMCTC_470__5_17

18.2

29.21

‐2991.844

0.59

0.65

79_2_72

RHII

EACTMCTC_444.3__2_72

73.8

3.7

‐815.825

<0.1

6116_5_17

RHV

EAGAMCTC_470__5_17

18.2

18.77

‐2185.245

0.49

Table2.8.(Continued)

Chapter2

72

293_11_20) together explained 29%of the phenotypic variancewith their additive

effectsrangingfrom(2.4tdto‐2.1td).

SixQTLswereassociatedwitht1onlinkagegroups(SHI,SHXI,RHIV,andRHV)

acrossenvironmentswiththeiradditiveeffectsrangingfrom(‐2.9tdto‐26.6td)with

phenotypicvarianceexplainedbyindividualQTLrangingfrom(11%to60%)(Table

2.8).MajorityofenvironmentsshowedoneQTLexceptExp4,wheretwoQTLswere

foundwith28%combinedexplainedphenotypicvariance.NoQTLcouldbefoundin

Exp5.AmongtheQTLsdetected, theQTL(116_5_17)on linkagegroupRHVwasa

majoronewithmaximumadditiveeffectof(‐26.6td).ThisQTLwasdetectedinboth

Exps1and3explaining23%and60%ofphenotypicvariance,respectively.

AtotalofeightdifferentQTLsweredetectedfort2acrosstheenvironmentswith

theiradditiveeffectsrangingfrom(1.9td to‐31.7td)(Table2.8).TheseQTLswere

associatedwith five parental linkage groups (SH I, SH VI, RH IB, RH II, and RH V)

explaining the phenotypic variance ranging from (1% to 55%). Two ormoreQTLs

were detected for the majority of environments. The QTL (116_5_17) on linkage

group RH V was a major one detected consistently throughout the environments

apartfromExp2.ThisQTLhadanegativeadditiveeffectthroughouttheexperiments

withmaximumeffectinExp3.ThemaximumphenotypicvariancebythisQTLwasin

Exp4(i.e.55%).

FourQTLswereassociatedwithteacrosstheenvironmentsexplainingthetotal

phenotypicvarianceranging from2%to74%(Table2.8).TheseQTLsweremainly

located on three paternal linkage groups (RH IV, RH V, RH VI) with their additive

effectsrangingfrom7.1tdto‐49.9td.ThemajorityofenvironmentsshowedoneQTL

except Exps 2 and 3, where two and three QTLs were found with 62% and 68%

combinedexplainedphenotypicvariance,respectively.AmongtheQTLsdetected,the

QTL(116_5_17)on linkagegroupRHVwasamajorone inall thesixenvironments

withmaximumadditiveeffectof(‐49.9td)inExp6.OnlythreeadditionalminorQTLs,

one inExp2(108_4_72)andtwoinExp3(101_4_35and153_6_28)weredetected,

withtheiradditiveeffectsrangingfrom7.1tdto8.4tdwithonly2‐3%ofassociated

phenotypicvariance.

FourindividualQTLswerefoundforvmaxlocatedonthethreeparentallinkage

groups (SH I, SHV, andRHV) (Table2.8). Their associated additive effects ranged

from 4.2% to ‐54.9% and the phenotypic variance ranged from 2% to 56%. QTLs

weredetected inonlyhalfof theenvironments (i.e.Exps1,3, and4).Among these

environments,atotalofthreeQTLsweredetectedinExp3withtheir65%combined

phenotypic explained variance. The QTL (116_5_17) on linkage group RHV was a

major one in this environment withmaximum additive effect of (‐54.9%). A QTL

(170_5_44)associatedwithlinkagegroupSHVwascommonlydetectedinbothExps

Geneticvariationinpotatocanopycoverdynamics

73

1and4with amaximumof 22.7%additive effect and21%associatedphenotypic

varianceinExp1.

Forthetraitsrelatedwithcanopygrowthrate,fourQTLsweredetectedforcm1,

fiveQTLsforc1,andfourQTLsforc3(Table2.8).TheseQTLswerescatteredonsix

parental linkage groups (SH I, SH V, SH IX, SH XI, RH V, RH VII). The phenotypic

varianceexplainedbyQTLs ranged from(3% to29%). In caseofcm1,oneand four

QTLs were found in Exps 1 and 4, respectively. No QTLs were found in other

environments.ThefourQTLsfoundinExp4jointlyexplained41%ofthephenotypic

variance. However, among these QTLs, (116_5_17) on linkage group RH V was a

major QTL with the maximum additive effect of ‐3.9 % td‐1. This QTL was also

commoninbothenvironments(Exps1and4)explaining28%and16%phenotypic

varianceforcm1.Incaseofc1,oneQTLwasdetectedinExps1,2,and6,whereasfour

QTLscouldbefoundinExp4.NoQTLswerefoundinExps3and5.ThethreeQTLs

found in Exp 4 jointly explained 44% of the phenotypic variance. The major QTL

(116_5_17)onlinkagegroupRHVwascommonlyfoundinExps1,2,and4withthe

additive effect ranging from ‐0.7 to ‐1.17 % td‐1 and 13% to 26% explained

phenotypicvariance.IntotalfourQTLsweredetectedforc3,twoeachinExps2and6,

oneinExp3,whereasnoQTLswerefoundintheotherenvironments.Thecombined

phenotypicvarianceinExps2and6was39%and28%,respectively.ThemajorQTL

(116_5_17) locatedon linkagegroupRHVwascommonlydetected inExps2and3

with positive (4.6% td‐1) and negative (‐5.3% td‐1) additive effectswith 29%and

28% explained phenotypic variance, respectively (Table 2.8). This switch from

positivetonegativeeffectisveryinterestingphenomenonandmightberelatedwith

theuniquenessofthesetwoenvironmentsasExp3wasclearlylowerinNavailability

thanExp2(Table2.1).Growthratecanbeseenastheintegrationofawiderangeof

processes and thus may depend on many factors. As we already mentioned in

previous sections, the canopy senescence ratewas slowed down to such an extent

thattheendofcropcycle(te)occurredverylateinthesaidenvironment(Table2.2).

For the traits related with total duration of canopy maximum and decline

phases(i.e.DP2andDP3),sixandfiveQTLsforeachweredetected,respectivelyacross

theenvironments(Table2.8).TheseQTLswerescatteredonsevenparental linkage

groups(SHI,SHIV,SHV,SHVI,SHVIII,RHIB,andRHV).Thephenotypicvariance

explainedbyQTLsrangedfrom2%to56%.ForDP2,oneQTLwasdetectedinExps2

and5,whilethree,fourandtwoQTLsweredetectedinExps2,4,and6,respectively.

NoQTLcouldbedetected inExp3.The fourQTLs found inExp4 jointlyexplained

70% of the phenotypic variance. Among the total QTLs detected for this trait,

116_5_17onlinkagegroupRHVwasamajorQTLdetectedinnearlyallenvironments

withitsadditiveeffectsrangingfrom‐10.2tdto‐31.3tdandexplaining14%to56%

Chapter2

74

phenotypicvarianceforDP2.IncaseofDP3,oneQTLwasdetectedinExps1,2,and5,

whereasthreeQTLswerefoundinExp6.NoQTLswerefoundinExps3and4.The

three QTLs found in Exp 6 jointly explained 46% of the phenotypic variance. The

majorQTLwas116_5_17onlinkagegroupRHVcommonlyfoundinExps1and2with

the additive effect ranging from ‐9.8 td to ‐19.6 td and 20% to 42% explained

phenotypicvariance,respectively.

Forthetraitsrelatedwiththeareaunderthecanopygrowthcurve(i.e.A1,A2,A3,

andAsum),QTLsdetectedwerescatteredonnineparentallinkagegroups(SHI,SHII,

SH V, SH VI, SH VIII, RH IA, RH IB, RH II, and RH V) (Table 2.8). The phenotypic

variance explained by QTLs ranged from <0.1% to 69%. A total four QTLs were

detectedforA1,i.e.twoinExp1andoneeachinExps3and4,whereasnoQTLswere

foundintheotherenvironments.ThephenotypicvarianceexplainedbyQTLsranged

from13%to64%.ThecombinedphenotypicvariancefortwoQTLsinExp1was33%.

AmongtheQTLsdetected116_5_17locatedonlinkagegroupRHVwasconsidereda

major QTL with its associated additive effect of ‐2246.4 td % and with 64%

phenotypic variance in Exp 3. Seven QTLs were found for A2 throughout

environments with their explained variance ranging from 2% to 55%. Among the

environments, oneQTLwas detected in Exps 1, 2, 3, and 5,whereas four and two

QTLs were estimated in Exps 4 and 6, respectively. The combined phenotypic

variance accounted byQTLs in Exps 4 and 6was 70% and 37%, respectively. The

majorQTL(116_5_17)on linkagegroupRHVwascommonly found inExps1,4,5,

and6withtheadditiveeffectrangingfrom‐1215.6td%to‐3198.4td%and18%to

55% explained phenotypic variance. In total five QTLs were detected for A3 with

phenotypicvariancerangingfrom8%to43%.OneQTLwasfoundeachinExps1,2,3

and 5; three in Exp 6,whereas noQTLswere found in Exp 4. The combinedQTLs

explainedphenotypicvariance inExp6was46%.QTL116_5_17 locatedon linkage

groupRHVwasamajoroneandcommonlydetectedinExps1,2,and3withadditive

effects ranging from ‐771.5 td% to ‐1251.1 td% with 18% to 43% range of

phenotypic explained variance. In case ofAsum,we estimated themaximumof nine

QTLsacrosstheenvironments.ThephenotypicvarianceexplainedbyindividualQTLs

ranged from<1% to69%.ThenumberofQTLs found in thevarious environments

was:oneQTLinExp6,twoinExps1,2,and5,threeinExp4,andfourQTLsinExp3

were found (Table 2.8). The combined phenotypic variance explained by the QTLs

rangedfrom57%to82%.TheQTL116_5_17onlinkagegroupRHVwasdetectedin

all the environments explaining the range of 49% to 69% phenotypic variance for

Asum.ThisQTLwasassociatedwithamajoradditiveeffectof‐4112.1td%inExp3.

In summary, QTLs with major effects were associated with paternal (RH)

linkagegroups,especiallyRHV(Table2.9),whereintotaloffiveQTLsweredetected

Geneticvariationinpotatocanopycoverdynamics

75

(Table2.7).QTLsonthis linkagegrouphadnegativeadditiveeffects, indicatingthat

RHallelesonthislinkagegroupshareanantagonisticeffectonthephysiologicaltraits

relatedwithcanopycover.OneparticularQTL(116_5_17)onthislinkagegroupwas

detected fornearlyall traitswithamajoradditiveeffectandexplainedmostof the

totalphenotypicvariance.Largenumberof additionalQTLswithminoreffectswas

mostlyassociatedwithmaternal(SH)linkagegroups(Tables2.8and2.9).Ourresults

areinlinewiththosebyVandenBergetal.(1996),whoalsoreportedthatmostof

thelocihadsmalleffects,butaQTLwithmaineffectwasfoundonchromosomeV.We

alsoobserved theco‐localisationofQTLs formany traits.For instanceclusteringof

manyQTLswere foundonposition18.2 cMonpaternal (RH) linkagegroup (Table

2.10). Here most of the traits (e.g. t2, DP2, A2, Asum) were tightly linked with QTL

(116_5_17) inmostoftheenvironments.ThiscouldmeanthatthisQTLisplayinga

pleiotropicroleindeterminingthesetraits.Thephenomenonofpleiotropycanhave

important implications for our understanding of the nature of genetic correlations

between different traits in certain regions of a genome and also for practical

applications in breeding because one of the major goals in breeding is to break

unfavourablelinkage(JiangandZeng,1995).Highgeneticcorrelationsbetweenthese

traitsconfirmtheserelations(Table2.5).

Table2.9.ListofparentallinkagegroupswithmajorandadditionalminorQTLs.

Parameter SHlinkagegroup RHlinkagegroup

MinorQTLs MajorQTLs MinorQTLs

tm1 I,VI,XI V IV

t1 I,XI V IVt2 I,VI V IB,II,VIte − V IV,VIvmax I,V V −cm1 I V VIIc1 I,XI V VIIc3 V,IX V −DP2 I,IV,VI V IBDP3 I,V,VIII V −A1 I,II V −A2 I V IBA3 I,V,VIII V −Asum I,V,VI V IA,II

Chapter2

76

QTL

Linkage

group

Markername

Position

(cM)

Exp.1

Exp.2

Exp.3

Exp.4

Exp.5

Exp.6

14_1_30

SHI

EAGTMCAG_458__1_30

33.1

DP2,A2

18_1_32

SHI

EAACM

CAG_187.9__1_32

36.3

c m

1,c1

39_1_32

SHI

EAACM

CCT_217.8__1_32

41.6

t 2,A s

um

t 2,D

P3

62_1_36

SHI

EAGAMCAG_228.3__1_36

63.9

DP2,A

2

63_1_42

SHI

PAG/M

ACC_322.2__1_42

68.9

v max,

Asum

t 1,c

m1

66_1_52

SHI

EAACM

CTG_193.9__1_52

81.5

Asum

t m1

76_1_84

SHI

PAT/M

AAC_259.4__1_84

109.9

DP2,A

2

1_5_12

SHIV

SH05B012_maturity_locus

0

c 3,D

P3,

A3

199_6_48

SHVI

EACAMCCT_138.9__6_48

61.8

t m1

Asum

200_6_56

SHVI

PAT/M

AAC_155.4__6_56

73.5

t 1,D

P2

242_8_29

SHVIII

PAT/M

AAC_283.7__8_29

33.2

DP3,A

3291_11_7

SHXI

EACAMCAC_372__11_7

9.2

t 1,c

147_1_63

RHIB

EACAMCCT_144.5__1_63

27.8

t 2,D

P2,

A2

115_5_4

RHV

PAC/MAGT_190.2__5_4

1.6

c 3,D

P3,

A3

116_5_17

RHV

EAGAMCTC_470__5_17

18.2

t 1,t2,t e,

c m1,c 1,

DP2,D

P3,

A2,A3,

Asum

t e,c

1,c 3,

DP2,D

P3,

A3,Asum

t m1,t 1,

t 2,te,

v max,c

3,A1,A3,

Asum

t 2,t e,

c m1,c 1,

DP2,A2,

Asum

t 2,te,

DP2,A

2,

Asum

t 2,te,

DP2,A

2,Asum

173_7_68

RHVII

PAC/MAAC_164.8__7_68

21.1

c m

1,c 3

Table2.10.Listofco‐localisedQTLsi.e.QTLsweresameformorethanonetrait.

Geneticvariationinpotatocanopycoverdynamics

77

This may indicate the difficulties of manipulating correlated traits

simultaneously. However, QTLs with similar behaviours could also be interesting

targetsforbreedingprogrammesastheyaremorelikelytobestableundervarious

environments.Moreover,itiswellknownthatlinkagegroupVharbourstheQTLfor

plantmaturityandvigourinpotato(Collinsetal.,1999;Oberhagemannetal.,1999;

Viskeretal.,2003;Bradshawetal.,2008).Ourresultsalsoconfirmthisfactasmostof

ourcanopygrowthanddevelopmenttraits(particularlyAsum)couldbeveryusefulin

definingthematuritytypeinpotato(seeChapter4).Ourresultsgiveaclearpicture

thatmaturityinpotatoismainlyexpressedfromthepaternal(RH)side.Besides,we

also found largenumberof independentQTLs(i.e. theydidnotcoincidewithother

traits)(Table2.11).TheseQTLshowever,withminoreffectscouldbeofgreatvalue

tobreedersforfurtherzoomingin.

AlthoughmanyoftheQTLsidentifiedweremappedtoasimilarpositioninmost

experiments,mostofthemwereexpressedinoneenvironmentbutnot intheother

ones(i.e.QTLE).Thiswasmostlyevidentfortraits(tm1,t1,vmax,cm1,c1,c3,DP3,A1,andA3)(Table2.12).ForallthesetraitstheGEcomponentofphenotypicvariancewasgreaterthantheGvariancecomponent(Table2.3).QTLscontrollingsuchtraitsoften

show low stability (Veldboom and Lee, 1996; Reymond et al., 2004). There were

markeddifferencesamongtheenvironmentsduetoNavailability(Table2.1).Besides,

the population was segregating for maturity. Variation in maturity type would be

expectedtocausevariationinanumberoftheparameterspresented(andfinalyield)

because of its relationship with the crop duration (e.g. te) and its effect on the

alignment between crop development processes and environment. These complex

interactions might have caused the lower repeatability of many QTLs over the

environmentsduetoQTLE.Severalresearchershaveidentifiedlocithat interactedwith the environment in different plant species e.g., photoperiod plasticity in

Arabidopsis(Ungereretal.,2003),growthandyieldinrice(Hittalmanietal.,2003),

floweringphenologyinbarley(Yinetal.,2005b)andyieldinbarley(Yinetal.,1999b;

Teulatetal.,2001;Voltasetal.,2001).

Only few QTLs were stable across the environments and therefore did not

show much of the QTLE (Table 2.13). For instance, the QTL 116_5_17 on RH VshowedupinalltheexperimentsforAsum(Fig.2.10). SuchQTLscouldbeusefulfor

markerassistedbreeding.

Chapter2

78

QTL

Linkage

group

Markername

Position

(cM)

Exp.1

Exp.2

Exp.3

Exp.4

Exp.5

Exp.6

21_1_32

SHI

EACAMCAG_381.9__1_32

38.7

v max

53_1_32

SHI

PAC/MAGC_62.6__1_32

44.0

A2

58_1_32

SHI

PCA/M

AAC_211.6__1_32

57.8

A1

67_1_52

SHI

PAT/M

AAC_207.7__1_52

80.7

t 2

81_1_92

SHI

PAC/MACT_132.6__1_92

121.9

t 2

102_2_25

SHII

EAACM

CCT_138.3__2_25

37.5

A1

133_4_28

SHIV

EACTMCTG_74.3__4_28

24.8

DP2

265_9_74

SHIX

PCA/M

AGG_294.2__9_74

80.9

c 3

293_11_20

SHXI

PAC/MATA_197.8__11_20

21.8

t m1

43_1_35

RHIA

PAT/M

AGT_143.9__1_35

58.3

Asum

64_2_21

RHII

PAG/M

ACT_230.5__2_21

23.5

t 2

79_2_72

RHII

EACTMCTC_444.3__2_72

73.8

Asum

96_4_33

RHIV

EACAMCAC_156.4__4_33

5.5

t 1

101_4_35

RHIV

EAGAMCAG_216.2__4_35

9.6

t e

98_4_35

RHIV

EACAMCTG_138.9__4_35

12.8

t 1

108_4_72

RHIV

EAACM

CCA_92.1__4_72

46.9

t e

114_5_4

RHV

EACAMCCT_109.4__5_4

4.5

A1

118_5_42

RHV

EAACM

CCT_201.2__5_42

43.9

A2

153_6_28

RHVI

PAT/M

AAC_272.6__6_28

53.8

t e

154_6_34

RHVI

EACTMCAG_393.7__6_34

59.7

t 2

Table2.11.ListofindependentQTLs(i.e.QTLsdetectedonlyonceforaparticulartraitinanyexperiment(environm

ent)).

Geneticvariationinpotatocanopycoverdynamics

79

Table 2.12. Number of QTLs detected across six experiments (environments).

Symbol‘−’meansnoQTLwasdetected.

Parameter Exp.1 Exp.2 Exp.3 Exp.4 Exp.5 Exp.6

tm1 1 1 1† − 2 −

t1 1 1 1† 2 − 1

t2 1 1 2† 4 3 2

te 1 2 3 1 1 1†

vmax 1 − 3† 1 − −

cm1 1 − − 4† − −

c1 1 1 − 4† − 1

c3 − 2 1† − − 2

DP2 1 3 − 4 1† 2

DP3 1 1† − − 1 3

A1 2 1 1† 1 − −

A2 1 1 1 4 1† 2

A3 1 1† 1 − 1 3

Asum 2 2 4† 3 2 1

TotalQTLs 15 17 18 28 12 18

†QTLswithmaximumadditiveeffects.

Chapter2

80

QTL

GroupMarkername

Position

(cM)

Exp.1Exp.2Exp.3Exp.4Exp.5Exp.6

39_1_32

SHI

EAACM

CCT_217.8__1_32

41.6

×

×

170_5_44SHV

PAC/MAAC_190.2__5_44

26.8

×

×

116_5_17RHV

EAGAMCTC_470__5_17

18.2

××

××

××

Table2.13.ListofstableQTLsacrosssixexperiments(i.e.SimilarQTLsdetectedinmajorityofenvironments).

Geneticvariationinpotatocanopycoverdynamics

81

Figure2.10.QTLEadditive effects for total areaunder canopy curve (Asum, td%)acrosspaternal(SHandRH)genomesforsixexperiments.Lettersaboveorbelowthe

barsindicatedifferentlinkagegroups.

4.Conclusions

In this study, we presented a quantitative model to describe canopy dynamics of

potato.Combinedwithgeneticanalysisweaimedtofurtherourknowledgeregarding

both the genetics and physiology of this trait of potato. The model successfully

describedthequantitativedifferencesincanopydynamicsofdiversegenotypesina

segregating F1 population of potato under varied environments and gave

physiological insightusingagronomicallymeaningfultraitsthatcharacterisecanopy

formation in potato. These traits are directly related to the ability of the adapted

genotypes to interceptphotosynthetically active radiation (PAR)and thus to create

high tuber yields, as we will show in Chapter 3. In this way, ourmodel approach

yieldedestimates foragronomically relevantcropcharacteristics thatareuseful for

definingfuturebreedingstrategiesinpotato.

The scope of crop improvement by breeding is determined by the amount of

heritable or genetic variation relative to that of non‐heritable or environmental

variation,andbythenatureandmagnitudeofGEinteractionwhichmaycomplicateselectionand testing inbreedingprogrammesandresult inreducedoverallgenetic

gain. Formost traits quantified in themodel, considerably high genetic variability

along with high heritability were recorded among the F1 population under study.

Thereareopportunities, therefore, toexploit thegeneticvariabilityavailable in the

-5000

-4000

-3000

-2000

-1000

0

1000

2000

1 2 3 4 5 6

Add

itive

eff

ects

Experiment no.

SH RH

V

V

II

IA

VV

II

V V

VI V II I

Chapter2

82

F1populationandtoselect forhighlyheritable traits inorder to improveradiation

interceptionefficiency.

In our second analysis, estimated physiological traits were subject to QTL

analysis.TheAFLPmarkerswere thereforegenerated foranextendedmarkermap

(Fig.2.9),towhichQTLweremappedforthemodelparametersandderivedtraits.In

total 42 QTLs were identified on both SH and RH parental genomes across all six

environments. QTLswithmajor effectswere associatedmainlywith paternal (RH)

linkagegroupV.OneparticularQTL(116_5_17)on this linkagegroupwasdetected

for nearly all traits with a major additive effect and explained most of the total

phenotypicvariance.SomeoftheQTLsweremappedtosimilarpositionsinmajority

oftheenvironments.OnlyfewQTLswerestableacrosstheenvironmentsandmostof

them showed QTLE. The stable QTLs across environments could be useful formarkerassistedbreeding.

ManyresearchersidentifiedQTLsfordifferenttraitsliketuberdormancy(Van

den Berg et al., 1996), tuber yield and starch content (Schäfer‐Pregl et al., 1998),

tubershape(VanEcketal.,1994),tuberfleshcolour(Bonierbaleetal.,1988),tuber

skincolour(Gebhardtetal.,1991),yield,agronomic,andqualitytraits(Bradshawet

al.,2008)inpotatoundernormalconditions.However,knowledgeaboutgeneticsand

physiologyoftraitsrelatedtocanopydynamicsandlightinterceptionisstill limited

andhardlyanyQTLshavebeenidentifiedforthesetraitsinpotato.

Ourquantitativeapproachincombinationwithmarkersofthewidelyavailable

and easy‐to‐use AFLP marker system identified QTLs that could be useful in the

developmentofmarkerassistedbreedingstrategiesinpotato.

Wehighlighted thepotentialofusingamodel toassist thegenetic analysisof

quantitative crop trait like canopy cover. Potato models intended for use inyield

predictions should first focus on improving genotype‐specific estimates of canopy

light interception under varied environmental conditions. Research investigations,

such as identifying potato canopy characteristics important for breeding trials,can

benefit by using our approach to provide an additional level of detail in canopy

development.

CHAPTER3

Analysisofgeneticvariationofpotato(SolanumtuberosumL.)usingstandardcultivarsandasegregatingpopulation.

II.Tuberbulkingandresourceuseefficiency*

M.S.Khan,X.Yin,P.E.L.vanderPutten,H.J.Jansen,H.J.vanEck,F.A.vanEeuwijk,P.C.Struik

AbstractQuantitative differences in the dynamics of tuber bulking of 100 genotypes in asegregatingF1population, theirparents (SH,RH)and five contrasting cultivarsofpotato(SolanumtuberosumL.)undervariousenvironmentswereanalysedusingapiece‐wise expolinear function. Tuber bulking was characterised by threeparameters: cm, ED and wmax, where cm and ED were growth rate and effectiveduration, respectively, of the linearphaseof tuberbulking, andwmaxwas the finaltuber dry weight.We also analysed radiation‐ and nitrogen‐use efficiencies (RUEandNUE,respectively),andtheirrelationshipswiththemodelparameters.Valuesofcmwere highest for earlymaturing genotypes followed bymid‐late and then lategenotypes.Latematuringgenotypeshadlongestperiodoftuberbulkingfollowedbymid‐lateandearlygenotypes.Asaresultwmaxwashigherinlategenotypesthaninearly genotypes. The RUE values were highest for early maturing genotypesfollowedbymid‐lateandlategenotypeswhereasNUEwashighestinlatematuringgenotypes followed by mid‐early and early genotypes. Genetic variability andheritabilitywerehighformosttraitsandphenotypicandgeneticcorrelationswerehigh(r>0.50)forthesetraitsaswell.PathcoefficientanalysisshowedthatRUE,cm,and a previously quantified parameter for total canopy cover Asum, had a majorinfluenceonwmax. Sixteen QTLs were detected for our model parameters and derived traitsexplainingthephenotypicvariancebyupto66%.QTLscontrollingmostofthetraitswereonpaternal(RH)linkagegroupV.OneparticularQTL(116_5_17)onpaternallinkage group V at position 18.2 cM was detected for all the traits with a majoradditiveeffectandexplainedmostofthetotalphenotypicvariance.AdditionalQTLsmostlyassociatedwithRHlinkagegroupsweredetectedfortraits(cm,tE,andED),whereasbothSHandRHlinkagegroupswereassociatedwithtraitsNUEandwmaxforminorQTLs.AnumberofQTLsfortraitswerenotdetectedwhentuberyieldpersewassubjectedtoQTLanalysis.ThephenotypicvarianceexplainedbytheQTLsfortuberyieldpersewasalsolowerthanforothertraits.Thegeneticparametersfoundin this study indicate that there are opportunities for improving tuberdrymatteryieldbyselectionforoptimalcombinationofimportantphysiologicaltraitsRUE,cm,andAsum.Keywords: Potato (Solanum tuberosum L.), tuber bulking dynamics, piece‐wiseexpolinear function, components of variance, genotype‐by‐environment (GE)interaction,geneticvariability,heritability,pathcoefficientanalysis,maturity type,QTLmapping,QTL‐by‐environment(QTLE)interaction.

___________________________________*SubmittedtoFieldCropsResearch.

Chapter3

84

1.Introduction

Tuberformationinpotato(SolanumtuberosumL.)consistsofacomplexanddynamic

sequenceofseveralindependentlyregulatedevents(EwingandStruik,1992;Jackson,

1999), including induction, initiation,set,bulking,andmaturation(Vreugdenhiland

Struik, 1989). These events are only possible when environment‐dependent steps

occurinanorchestratedway,includingthearrestofstolongrowth(Vreugdenhiland

Struik,1989),initiationofradialgrowth(CatchpoleandHillman,1969;Mingo‐Castel

etal.,1976;Struiketal.,1988;VreugdenhilandStruik,1989;EwingandStruik,1992)

andresourcestorage(Park,1990;Müller‐Röberetal.,1992).Thedifferentstepsand

events can occur independently of each other and are regulated by specific genes

(Struik et al., 1999; Kloosterman et al., 2005). The resulting, economically relevant

process of tuber bulking is, therefore, regulated by a large set of interacting genes

(Bachemetal.,2000).

The onset of tuber bulking greatly impacts subsequent growth, development,

andphysiologyoftheentirepotatocrop(Ewing,1990,EwingandStruik,1992;Van

Dam et al., 1996; Walworth and Carling, 2002), because the developing tubers

become thedominantsinkofbothcarbonandnitrogenassimilates (Oparka,1985).

Theonsetoftuberbulkingleadstoamoreorlessabruptpreferentialpartitioningof

assimilates to the tubers, thereby causing a reduction in the growth rate and

ultimatelyacompletehaltingofgrowthoffoliageandroots(MoorbyandMilthorpe,

1975; Ewing and Struik, 1992). However, the abruptness depends on thematurity

typeandotheraspectsofthegenotype‐specificphysiology(Struik,2007).Earlyonset

of tuber bulking may result in small plants with limited canopy cover and

consequently lowfinal tuberyields,whilst lateonsetof tuberbulking leadsto large

plantswithhighfinaltuberyields(BremnerandRadley,1966;Struik,2007).

Information on the different processes involved and factors affecting plant

developmentandtuberformationinpotatoareabundant(IvinsandBremner1965;

Ewing and Struik, 1992; Almekinders and Struik 1996; Kolbe and Stephan‐

Beckmann,1997;O’Brienetal.,1998;Jackson,1999;Struiketal.1999;Claassensand

Vreugdenhil,2000,andreferencestherein).However,moststudieshave focusedon

oneorveryfewofthesedevelopmentalprocesses,orononeoraverylimitednumber

ofgenotypes(cultivars).Moreover,thephysiologicalandgeneticbasesofvariability

amongsuchtraitshavenotbeenthoroughlyinvestigated,althoughsomeeffortshave

been made to study the temporal dynamics of important potato developmental

processesunderdiverseenvironmentalconditionsonasetofcontrastinggenotypes.

For instance,Spitters (1988)analysed thegenotypicdifferences in tuberbulkingof

potatoonalargesetofcommercialvarieties.Celis‐Gamboaetal.(2003)usedahighly

segregating diploid population of potato to study the temporal relationships

Geneticvariationintuberbulkingandresourceuseefficiencyofpotato

85

underlyingthedynamicsoftuberformationandotherdevelopmentalprocesses.

Difficultiesinmanipulatingyieldarerelatedtoitsgeneticcomplexity:polygenic

nature, interactions between genes (epistasis), and environment‐dependent

expressionofgenes(RibautandHoisington,1998).Manyenvironmentalandphysical

factors, such as temperature, day length, light intensity, water availability, and

nitrogen (N), have been demonstrated to influence potato tuberisation (Ewing and

Struik,1992; Jackson,1999).Forexample, temperatureexertsamajor influenceon

tuberisationanddrymatterpartitioningtotubers(Ewing,1981,1985),withcoolair

temperatures favouring induction to tuberise (Bodlaender, 1963; Gregory, 1965;

Epstein, 1966; Ewing, 1981; Manrique et al., 1984; Struik and Kerckhoffs, 1991),

whereas an increase in air temperaturemay reduce tuber drymatter content and

yield(Struiketal.,1989b).Nitrogenalsoplaysamajorroleintuberisation(Werner,

1934; Gregory, 1965). Nitrogen helps to attain complete canopy cover early in the

season, especially under relatively resource‐poor conditions (Haverkort and

Rutavisire, 1986; Vos, 2009) and to extend the period of full canopy cover thus

leading to increased light interception and tuber yield (Martin, 1995). Radiation is

important for dry matter accumulation (Monteith, 1977; Goudriaan and Monteith,

1990) and affects the early processes in tuber formation (Ewing and Struik, 1992)

andtherateoftuberbulking(BurstallandHarris,1983)anditsduration.Therefore,

resource(radiation,N)useefficiencymayhaveastrongbearingontuberbulkingand

finaltuberyields.

In our companion paper (Chapter 2), we have quantitatively analysed potato

canopy cover dynamics, using a set of varieties covering awide range ofmaturity

typesandawell‐adapteddiploidF1segregatingpopulation.Here,usingthesameset

ofplantmaterials,weaimtoanalysethedynamicsoftuberbulkinganditsvariability

by breaking them down into biologically meaningful and genetically relevant

component traits.We also analyse radiation use efficiency (RUE) and nitrogen use

efficiency (NUE) and study their relationships with the tuber bulking traits. We

quantify thegeneticparameters(variancecomponentsandheritability),phenotypic

andgeneticcorrelations,andpathcoefficientsofthesetraits.Finally,weperformQTL

mappingofthetraitsanddiscusstheirgeneticbasis.Thecombinedinformationfrom

Chapters2and3shouldgiveinsightsintothemostvitalprocessesthatcanbeusedto

explorethepossibilitiesofgeneticallymanipulatingpotatotuberyield.

Chapter3

86

2.Materialsandmethods

2.1.F1segregatingpopulationofSHRHandstandardcultivarsTheplantmaterialusedinthisstudyconsistedof100F1diploid(2n=2x=24)potato

genotypes derived from a cross between two diploid heterozygous potato clones,

SH83‐92‐488RH89‐039‐16(RouppevanderVoortetal.,1997;VanOsetal.,2006),orsimplythe‘theSHRHpopulation’.Thepopulationsegregatesformaturitytype.

BesidestheindividualF1genotypesandtheirtwoparents,wealsoincludedfive

standard cultivars in our studies: Première, Bintje, Seresta, Astarte, and Karnico.

These cultivars were chosen because of their differences in maturity type when

grownintheNetherlands,rangingfromearly(Première)toverylate(Karnico).This

selection of standard cultivars would allow a benchmarking of consequences of

maturitytypeonthetemporaldynamicsoftuberbulking.Furtherinformationabout

plantmaterialsisgiveninChapter2.

2.2.Fieldexperimentsandmeasurements

Six field experiments were carried out in Wageningen (52˚ N latitude), the

Netherlands,during2002,2004,and2005,withtwoexperimentsineachyear,using

the aforementioned plant materials. Details on the methodology and the

environmentalconditionshavebeendescribedinChapter2.Karnicowasnotpresent

inthetwoexperimentsin2005.

Tuberdrymatterwasmeasuredat threeharvestsduring thegrowingperiod.

Thefirstandsecondharvestswereplannedinsuchawaytomeasurethetuberdry

matterinthelinearbulkingphase,whilethelastharvestwasperformedatmaturity.

Tubersofeachplotwereharvestedanddriedinanovenat70˚Ctoconstantweight.

Forsamplesofthegrowingseasonsof2004and2005,nitrogen(N)concentrationin

tuberssampledattheendofthegrowingseasonwasdeterminedbymicro‐Kjeldahl

digestionanddistillation(AOAC,1984).TotalamountofN in tuberswascalculated

fromtheNconcentrationandtuberdryweight.

2.3.Amodelfortuberbulkingdynamics

Likethelifecycleofanyplantoritsorgan,potatotubergrowthasafunctionoftime

followsasigmoidpattern,includinganearlyacceleratingphase,alinearphaseanda

ripeningphase(Fig.3.1).WeusedtheexpolinearfunctionofGoudriaanandMonteith

(1990)todescribethetubergrowthduringtheexponentialphase:

Geneticvariationintuberbulkingandresourceuseefficiencyofpotato

87

Bm

m with1ln Bm tterc

w ttr (3.1)

wherewis tubermass,tis time,tBis themomentatwhichthe linearphaseof tuber

bulkingeffectivelybegins,rmistherelativegrowthrateinthe‘exponentialphase’,and

cmisthegrowthrateinthe‘linearphase’.

Eqn (3.1) can, inprinciple, beused todescribe the tuber growthof the linear

phase as well. However, eqn (3.1) tends to under‐estimate the true growth rate

becauseofitscurvilinearnature.Toobtainanobjectiveestimationofthegrowthrate

ofthelinearphase,weusedthefollowinglinearmodeltoquantifythesecondphase:

EBBmB with tttttcww (3.2)

wheretEis theend timeof the linearphase,wBis the tuberweightat time tB. Ifwe

knowrmandcm,thenwBcanbeestimatedfromeqn(3.1)as0.693cm/rm.

To represent a deflection in growth towards the third phase,Goudriaan and

Monteith(1990)suggesteda truncatedcurve that terminatesgrowthat the time tE,

when the maximum weight (wmax) is achieved. This is a brutal method, because

growthstopsgraduallyrather thanabruptly.However,given the limitednumberof

measurements in the time series (see above), we adopted the truncated curve

approach,with:

Emax with ttww (3.3)

wheretEiscalculatedastB+(wmax−wB)/cm.

Combiningeqns(3.1,3.2,and3.3)yieldsamodelwithfourparameters:rm,cm,

tB,andwmax,whilethetwootherparameterswBandtEarecalculatedas0.693cm/rm

andtB+(wmax−wB)/cm,respectively.Obviouslyanover‐fittingwouldbeobtainedifall

the fourparametersweretobedirectly fitted fromour limiteddatapoints foreach

genotype. IngramandMcCloud (1984) andVanDamet al. (1996) reported that rm

wasconservativeacrosspotatocultivarsatagiventemperature.Theirrmvalue0.34

d‐1 at the optimum temperature was used here for all genotypes. We also fixed

parameterwmaxas theaverageof twoblocksof the finalmeasuredweight.The two

remainingparameters (i.e.cm, tB) canbe estimated, butwith the valueof tB having

large standard error, probably due to insufficient data points for the early season.

Using these estimated values, we found that the initial weight (w0) at time zero,

calculatedusingeqn(3.1),didnotvarymuchacrossgenotypes.We,therefore,used

the value for w0 = 0.13 (g m‐2), the averaged w0 across all six

Chapter3

88

Figure 3.1. Tuber bulking dynamics in potato represented by the piece‐wiseexpolineargrowthfunction.

experiments and all genotypes, to further reduce the number of parameters to be

estimated.Whenw0isfixed,parametertBcanbecalculatedfromeqn(3.1)as:

1ln

1 m

m0

mB

crw

er

t (3.4)

This is inaccordancewithGoudriaan (1994),whosuggested that itmaybeamore

naturalsequencetoexpresstBasafunctionofinitialweight(w0)atemergence.

Eqn(3.4)andtheformulaeforcalculatingwBandtE,werecombinedwitheqns

(3.1‐3.3), for curve fitting. The fitting was performed for each genotype of every

experiment with the iterative non‐linear least‐square regression using the Gauss

method, as implemented in the PROCNLIN of the SAS software (SAS Institute Inc.,

2004). Obviously, our procedure estimated only cm, which together with wmax as

primary model parameters, characterises genotypic and environmental effects on

tubergrowth.ParameterstB,tE,andwBwerecalculatedfromtheequationsdescribed

earlier. The effective duration of tuber bulking (ED), an additional useful trait,was

calculatedastE−tB.

As in our first analysis for canopy cover (Chapter 2), all time variables and

Thermal days (td)

maxw  

Bt   Et  

mc  

mr  

Tub

er d

ry m

atte

r (g

m-2

)

Bw  

Geneticvariationintuberbulkingandresourceuseefficiencyofpotato

89

durationwereexpressedasthermaldays(td)toaccountfortheinfluenceofdailyand

seasonaltemperaturefluctuationsontubergrowth.Themethodforconversionofthe

actual days into tdwas givenbyYin et al. (2005) and its application to our potato

genotypes has been described in Chapter 2.Note that the td is equal to or smaller

thanthenumberofchronologicaldays.

2.4.CalculationofRadiationUseEfficiency(RUE)

Theradiationuseefficiency(RUE;gDMMJ‐1PARint)wasestimatedforeachindividual

plotbydividingthetotaltuberdrymatteratmaturity(gDMm−2)bythecumulative

interceptedphotosyntheticallyactiveradiation(MJPARintm−2)fortheentiregrowth

period.Dataofincidentglobalsolarradiationwereobtainedfromaweatherstation

in Wageningen located nearby the experimental sites. Daily incident PAR was

calculated as half of the global solar radiation (Spitters, 1988). To calculate

cumulativePARint,ourextensivedataonthepercentagegreencanopycover(Chapter

2)wereconvertedtoPAR‐interceptionpercentage,usinga linearrelationshipgiven

byBurstall andHarris (1983)asPARint(%)=0.956canopycover (%)–4.95.Dailyvalues of PARint were summed and the obtained cumulative PARint was used to

calculateseasonalaverageRUEonthetuber‐dryweightbasis.

2.5.CalculationofNitrogenUseEfficiency(NUE)

Aswe did notmeasure dryweight andN content in the organs other than tubers,

nitrogen use efficiency (NUE; g DM g‐1 N) was expressed on the tuber drymatter

basis, aswasRUE,bydividing the total tuberdrymatter (gDMm−2)by total tuber

nitrogen uptake (g N m−2). This way of presenting NUE means that NUE is

mathematicallyequivalenttotheinverseoftuberNconcentration(gNg‐1DM).

2.6.Statisticalandgeneticanalysis

All statistical analyses were performed in Genstat (Payne et al., 2009). Combined

analysisofvarianceacrossexperiments(i.e.environments)wasperformedtotestthe

significanceandextentofdifferencesamongenvironmentsandgenotypes(including

the F1 population, the parents, and standard cultivars). Means of genotype and

environmenttermswerecomparedusingtheFisher’sleastsignificantdifference(LSD)

test.FurtherstatisticalanalyseswereperformedonlyusingtheF1populationmeans

(100genotypes)acrosssixenvironmentsasdescribedbelow.

Chapter3

90

2.6.1.Estimationofvariancecomponents

The variance components for genetic, environmental and experimental error were

estimated through the REML procedure to assess their contribution to the total

phenotypic variance of the traits cm, tE,ED,RUE, NUE, andwmax. Significance levels

weredeterminedwithalikelihoodratiotest(Morrell,1998),whichteststhechange

indeviationafterremovingtherespectivevariancecomponentfromthemodel.The

changeindeviationisapproximatelychi‐squaredistributed(Littelletal.,1996).Once

these variance components were estimated, phenotypic variance (σ2Ph) was

calculatedasper followingequation (Bradshaw,1994;FalconerandMackay,1996;

LynchandWalsh,1998):

2ε

2E

2G

2Ph (3.5)

where σ2G is genetic variance, σ2E represents environmental variance, and σ2ε is

experimentalerrorvariance.

2.6.2.Phenotypicandgeneticcoefficientsofvariation

Coefficientsofvariation(%)werecalculatedaccordingtothefollowingequation:

1002

CV (3.6)

where is the grandmeanof thepopulation, andσ2X is a variance component (i.e.σ2Phorσ2Gorσ2E).

2.6.3.GGEbiplotanalysisGGEbiplot analysiswas performed to analyse the inter‐relations among genotypes

and environments. GGE biplots were constructed by plotting the first principal

component (PC1) scores of the genotypes and the environments against their

respective scores for the second principal component (PC2). The environment‐

standardisedmethodofYan(2002)wasused.

2.6.4.Heritability

Estimatesofbroad‐senseheritability(H2)(%)werecalculatedbyusingtheestimated

variancecomponents(FalconerandMackay,1996;Hollandetal.,2003):

Geneticvariationintuberbulkingandresourceuseefficiencyofpotato

91

100

t

2ε2

G

2G2

n

H

(3.7)

where(nt=6)istheproductofnumberofblocksandenvironments.

2.6.5.Phenotypicandgeneticcorrelation

Phenotypic correlations were calculated using the Pearson Correlation Coefficient.

The genetic correlations were calculated using the following equation (Holland,

2006):

2G

2G

2G

Gji

ijijr

(3.8)

whereσ2Gijistheestimatedgeneticcovariancebetweentraits iand j;σ2Giandσ2Gjare

the genetic variances of traits i and j, respectively. The variance and covariance

componentswereestimatedfrommultivariateREMLanalyses(Meyer,1985;Holland,

2006).Thesignificanceofgeneticcorrelationswasdeterminedusingat‐testafteraz‐

transformationofthecorrelationcoefficients(SokalandRohlf,1995;Guttelingetal.,

2007).

2.6.6.Pathcoefficientanalysis

The inter‐associations between the important yield determining components were

ascertained across all six experiments byworking out the path coefficient analysis

following the procedure of Dewey and Lu (1959). This was accomplished by

partitioning the direct and indirect effects of various physiological traits upon the

final tuberdrymatter. The final tuber drymatter (i.e.wmax)was considered as the

response variablewhile traits cm,ED,Asum, RUE, andNUEwere assumed to be the

predictorvariables,whereAsumistheareaunderthewholecanopy‐covercurveand

reflects the capability of the crop to intercept solar radiation during the whole

growingseason,asquantified inChapter2.Thedirecteffectsofpredictorvariables

werethepathcoefficientscomputedthroughmultipleregression.Apathcoefficientis

a standardized regression coefficient (Li, 1975). Indirect effectswere computed as

the product of the correlation coefficient between two variables and the path

Chapter3

92

coefficient from the secondvariable to the response variable. Let variables x1 to x5

refer to cm, ED, Asum, RUE, and NUE, respectively. The total effect of a predictor

variable x1 correlatedwith other predictor variables x2, x3, x4, and x5 on response

variablez,wouldbegivenby,forexample:

rzx1=Pzx1+(Pzx2×rx1x2)+(Pzx3×rx1x3)+(Pzx4×rx1x4)+(Pzx5×rx1x5) (3.9)

whererzx1istotalcorrelationbetweenzandx1,Pzx1representspathcoefficientfrom

x1tozrx1xi(i=2,3,4and5)iscorrelationcoefficientbetweenvariablesx1andxi,and

Pzxidenotespathcoefficientfromxitoz.Thesamelogicwasappliedtocomputerzx2,

rzx3,…,rzx5.

2.6.7.QTLdetection

Theparental(SH,RH)geneticmapdescribedinChapter2wasusedforQTLmapping

of traits. Eighty‐eightgenotypesof our100F1 lineswere covered in theextended

ultra‐densegeneticmapof250 linesofSHRHpopulation (cf.Chapter2);dataofthese 88 lines were therefore used for detection of QTLs for model parameters,

derivedtraitsand(N,radiation)useefficiencies.QTLanalysiswasdoneindividually

forallsixexperiments(environments)usingGenstatversion14(Payneetal.,2009)

software.Formoredetailsaboutthemappingprocedure,seeMaterialsandMethods

ofChapter2.

3.Resultsanddiscussion

3.1.Modelperformanceindescribingtuberbulkingdynamicsofgenotypes

Themodelfortuberbulkingdynamics(i.e.combinedeqns3.1‐3.3)fittedwellforeach

genotypeofthepotatosegregatingpopulation,theparentsandthestandardcultivars

intheentiredataset,withR2valuesrangingfrom0.90to1.00(n=6).Theestimated

tuberbulkingcurvesforthetwoparents(SHandRH)andfourstandardcultivarsare

showninFig.3.2.Thetransformationofcalendardaysintothermaltimeresultedina

more stable parameter estimation (data not shown), as thermal time effectively

removed the confounding effect of diurnal and seasonal temperature fluctuations

during theexperimentalperiod.Overall, combinedeqns (3.1‐3.3)areveryuseful in

analysing the tuber bulking dynamics of a diverse set of potato genotypes under

variousenvironments.

Geneticvariationintuberbulkingandresourceuseefficiencyofpotato

93

Figure3.2.Observed(obs)pointsandfitted(fit)curvesofparents(SHandRH)and

fourstandardcultivarsforallsixexperiments(environments).

3.2.Modelparametersandsecondarytraits

Resultsofcombinedanalysisofvarianceshowedhighlysignificant(P<0.01)effectsof

genotype(including100F1genotypes,theparents,andstandardcultivars)acrossthe

experimental sites (i.e. environments) on all model parameters and derived traits.

However,asexpected,differencesbetweencultivarsandacrossenvironmentsinthe

onset of tuber bulking (tB) were very small, because as mentioned earlier tB was

calculatedbyeqn(3.4)wherewoandrmwerefixedduetolimiteddatapointsforthe

earlygrowthphase.Similarly,thetuberweight(wB)achievedattBwascalculatedin

relation to rm and cm (see Materials and Methods). Therefore, we will no longer

analysethevariationoftBandwB.

 

0

500

1000

1500

2000

2500

0 20 40 60 80 100

SH

0

500

1000

1500

2000

2500

0 20 40 60 80 100

RH

0

500

1000

1500

2000

2500

0 20 40 60 80 100

Première

0

500

1000

1500

2000

2500

0 20 40 60 80 100

Bintje

0

500

1000

1500

2000

2500

0 20 40 60 80 100

Seresta

0

500

1000

1500

2000

2500

0 20 40 60 80 100

Astarte

Tub

er d

ry m

atte

r (g

m-2

)

Chapter3

94

Thereweresignificantdifferences(P<0.05)amongthestandardcultivarsforcm,

tE, ED, and wmax (Table 3.1). Values of cm were comparatively higher for early

maturingcultivarslikePremièreandmid‐earlycultivarslikeBintjethanforlateones

likeAstarteandKarnico.

Thevalues for tEandEDwerehigher for latematuringcultivars than formid‐

late and early cultivars. Themaximum tuber drymatter yield (wmax), on the other

handwashigherforlatematuringcultivarsthanformid‐lateandearlycultivars.The

highervaluesofwmaxin latematuringcultivarsmightbedue to theircomparatively

higher tE and longer ED. This is in line with the result of Kooman and Rabbinge

(1996),whofoundthatcomparedwithlatecultivars,earlypotatocultivarsallocatea

larger part of the available assimilates to the tubers early in the growing season,

resulting in shorter growing periods and also lower yields. Moreover, cultivars

differenceswithrespecttotuberyieldingpotentialcouldbeattributedtovariationin

efficiencyofassimilatepartitioningtothetubers(bulkingrate)andmaturityperiod.

HammesandDeJager(1990)andGawronskaetal.(1990)reportedtheexistenceof

varietaldifferenceswithrespecttotherateofnetphotosynthesisanddrymatter

production.

Table 3.2 shows trait values of the F1 segregating population in comparison

withtheirparents(SHandRH).ThemeansoftheF1segregatingpopulationwerenot

withintherangesoftheparentalclonesforparameterstEandED.Widerangeswere

observed forallmodelparametersandderived traits.Mostof theparameterswere

Table3.1.Estimatedmeanvaluesofdifferenttraitsforfivestandardcultivars(listed

inorderofincreasinglylongercropcycle),asobtainedfromcombinedANOVAacross

sixenvironments.tdstandsforthermalday.

Cultivarcm

(gDMtd‐1)

tE

(td)

ED

(td)

RUE

(gDMMJ‐1)

NUE

(gDMg‐1N)

wmax

(gDMm‐2)

Première 55.7a 37.3c 16.1c 2.7a 70.4c 1122b

Bintje 48.2ab 47.9b 27.4b 2.4a 72.1c 1375ab

Seresta 36.7bc 62.4a 42.8a 2.4a 88.5b 1579a

Astarte 35.4bc 63.0a 43.5a 2.2ab 90.9b 1533a

Karnico 31.2c 63.7a 44.4a 1.9b 101.8a 1487a

LSD

(P<0.05)13.2 10.1 10.7 0.5 7.7 263

Meanswithinacolumnfollowedbydifferentlettersaresignificantlydifferent

accordingtoFisher’sMultipleRangeTest(P<0.05).

Geneticvariationintuberbulkingandresourceuseefficiencyofpotato

95

nearlynormallydistributeddisplayingtransgressivesegregation(Fig.3.3).

Environment had highly significant (P<0.01) effects on all model parameters

and derived traits. There were significant differences (P<0.05) between the

experiments for cm, tE, ED, and wmax (Table 3.3), within five standard cultivars.

This was at least partly due to the purposeful variation in availability of N across

trials (Chapter 2). Figure 3.4 illustrates the variation (the median, minimum and

maximumvalues)ofthesemodelparametersandderivedtraitsfortheF1population

per individual experiment. The ranges of the parameters cm, and wmax were

consistentlywiderinExp3thanintheotherexperiments(Fig.3.4).IncaseoftEand

ED,widerrangeswereobservedinExps4and5,respectively(Fig.3.4).Thiscouldbe

attributedtovariedavailabilityofNperexperiment(Chapter2) in interactionwith

genotypespecificbehaviour, causingdifferent trade‐offsbetweenrateandduration

oftuberbulkingandtuberfinaldrymatterproduction.

3.3.RadiationUseEfficiency(RUE)

TheRUEdifferedsignificantly(P<0.01)amongthegenotypes(standardcultivarsand

F1segregatingpopulation).ThevaluesofRUErangedbetween1.9and2.7gDMMJ‐1

(Table 3.1). The calculated values of RUE were within the range reported in the

literature for Solanum tuberosumgenotypes under temperate conditions (Spitters,

1988; Scott and Wilcockson, 1978; Allen and Scott, 1980; Khurana and McLaren,

1982; MacKerron and Waister, 1985a; Stol et al., 1991; Kooman and Haverkort,

1995).

Table3.2. Estimatedmean values of different traits (across six environments) for

two parents (SH and RH) andmean, minimum (Min), maximum (Max), and range

withintheF1population.tdstandsforthermalday.

Parameter SH RH ‡Mean(±S.E.) Min Max Range

cm(gtd‐1) 39.8 29.0 35.5(±4.2) 19.9 47.4 27.5

tE(td) 50.8 46.9 51.0(±2.9) 42.5 62.9 20.4

ED(td) 30.9 27.8 31.8(±3.3) 22.4 45.1 22.7

RUE(gDMMJ‐1) 2.7 2.4 2.1(±0.2) 1.5 2.6 1.1

NUE(gDMg‐1N) 69.6 67.2 68.2(±5.6) 54.4 81.6 27.2

wmax(gm‐2) 1219 847 955.6(±50.0) 830.4 1115.7 285.3‡MeanofF1segregatingpopulation(100genotypes)acrosssixenvironments.

Chapter3

96

Table3.3.Estimatedmeanvaluesofdifferenttraitsforeachindividualenvironment

as obtained from combined ANOVA across five standard cultivars. td stands for

thermalday.

Parameter Exp.1 Exp.2 Exp.3 Exp.4 Exp.5 Exp.6 LSD

cm(gtd‐1) 30.6d 41.6c 46.1b 50.0a 30.9d 48.3a 2.2tE(td) 55.4d 42.9f 61.1b 59.4c 64.6a 53.0e 1.7

ED(td) 36.3d 22.8f 40.8b 38.8c 45.5a 32.5e 1.8

RUE(gDMMJ‐1) 2.1d 2.0e 2.8a 2.6b 2.0e 2.4c 0.09

NUE(gDMg‐1N) − − 102.5a 80.8c 69.4d 84.8b 1.5

wmax(gm‐2) 1086e 1003f 1834b 1954a 1341d 1555c 44

Meanswithinarowfollowedbydifferentlettersaresignificantlydifferentaccording

toFisher’sMultipleRangeTest(P<0.05).

Significant (P<0.05) differences were observed among the standard cultivars

withrespecttoRUE.TheRUEvalueswerecomparativelyhighforearlyandmid‐early

cultivars(PremièreandBintje,respectively) followedbymid‐late(Seresta)andlate

cultivars(Karnico).Theseresultswereexpectedbecausecumulativelightabsorption

tended to be greater for the later maturing cultivars, but on the other hand, they

exhibitedasmallerharvestindex(datanotshown).Owingtotheseoppositetrends

for cumulative light absorption andharvest index, tuber yield showed an optimum

relationshipwithmaturityclass.Earlymaturingcultivarsallocatealreadyinanearly

phase themajor part of their current assimilates to tuber growth, which is at the

expenseofcanopygrowth(Spitters,1988).Senescenceofleaves,withoutsignificance

formation of new leaves, causes an early senescence of foliage. Early cultivars had,

therefore,asmallercumulativelightabsorptionbutagreaterharvestindexandRUE.

On the other hand, late maturing cultivars maintain green, active foliage for an

extendedperiodoftime(SpittersandSchapendonk,1990).However,theinvestment

incanopygrowthisattheexpenseoftubergrowth.

Table3.2comparesthemean,andrangesofRUEbetweenthetwoparents(SH

andRH)andF1segregatingpopulation.RUEvaluesvariedfrom2.7gDMMJ‐1forSH

and 2.4 g DM MJ‐1 for RH parent, compared with mean (2.1 g DM MJ‐1) of F1

population.ThemeanRUEofF1populationwaslowerthanthevalueofeitherparent,

butthiswasassociatedwithawidevariationinthepopulationforRUEwithvalues

ranging from 1.5 g DM MJ‐1 to 2.6 g DM MJ‐1 (Table 3.2) and therefore displayed

transgressivesegregation(Fig3.3).Therearepossibilitiesthereforeforthebreeders

toexploitthisvariationforimprovingRUE.

Geneticvariationintuberbulkingandresourceuseefficiencyofpotato

97

Figure3.3. Distribution of five model parameters among F1 genotypes across six

experiments(environments).Thevaluesoftwoparents‘SH’and‘RH’areindicatedby

fullarrowanddashedarrow,respectively.

1.50

3.00

2.50

25

20

1.25

2.00

15

10

5

0

2.25

1.75

2.75

10

5

60 10 40 20 0

50 30

20

15

5

90 40 70 0

50

30

25

80 60

20

15

10

50 60

40

30

0 40

20

10

20 0

30 70 10

12.5

0.060 75

5.0

10.0

65 80

7.5

2.5

70

17.5

15.0

7.5

800

2.5

700 1100 900

20.0

17.5

0.01000 1200

5.0

15.0

12.5

10.0

tE NUE

wmax

Num

ber

of F

1 ge

noty

pes

ED

RUEcm

Chapter3

98

0

20

40

60

80

100

[g td-1] cm

0

20

40

60

80

100

120

[td] tE

0

20

40

60

80

100

1 2 3 4 5 6

[td] ED

0

1

2

3

4

5RUE[g DM MJ-1]

0

20

40

60

80

100

120

NUE[g DM g-1 N]

0

500

1000

1500

2000

1 2 3 4 5 6

[g m-2 ] wmax

Figure3.4.BoxplotsofgeneticmeansofanF1populationofdifferenttraitsinallsix

experiments. Theboxes span the interquartile rangeof the trait values, so that the

middle 50% of the data lay within the box, with a horizontal line indicating the

median. Whiskers extend beyond the ends of the box as far as the minimum and

maximumvalues.

Experiment no.

Geneticvariationintuberbulkingandresourceuseefficiencyofpotato

99

TheRUE showed significant (P<0.05)variationamongst environments for the

setofstandardcultivarsaswellasfortheF1segregatingpopulation.Forthestandard

cultivarsRUEmeanvalueswerehigherinExp3andlowerinExp5(Table3.3).For

the F1 population a similar trend was observed with wider and lower ranges of

variationinExp3andExp5,respectively(Fig.3.4)thanintheotherexperiments.We

surmise that such within and between experimental variations are the combined

result of differences in Asum associated with variation in maturity class and with

variedavailabilityofN(Chapter2).Plantnitrogenstatusandcropgrowthcycleboth

affectRUE,inadditiontotheeffectsofotherfactors(Green,1987;MuchowandDavis,

1988;SinclairandHorie,1989;Trapanietal.,1992).

TheRUEhasbeenusedtoexplaingenotypicdifferencesinpotatounderdiverse

environments (Sibma,1970,1977;VanderZaagandBurton,1978;MacKerronand

Waister, 1985a; Van der Zaag and Doornbos, 1987; Vander Zaag and Demagante,

1987;TrebejoandMidmore,1990;Komanetal.,1996).VanderZaagandDoornbos

(1987)concludedfromtheirtrialswith19cultivarsthatcultivardifferencesintuber

dry matter yield were mainly due to variation in RUE under varied growing

conditions.Itisconcludedthatanalysisofyieldvariationbetweenpotatogenotypes

can be obtained by interpreting that variation in terms of accumulated light

absorption,averageRUE,andharvestindex.3.4.NitrogenUseEfficiency(NUE)

Highlysignificant(P<0.01)effectsofgenotype(standardcultivarsandF1segregating

population) were observed for NUE (data not shown). Mean estimates for NUE

rangedbetween69.6to101.8gDMg‐1N(Table3.1).NUEvariedsignificantly(P<0.05)

among the standard cultivars. NUE values were comparatively higher for late

maturing cultivars like Karnico and Astarte than for mid‐early and early cultivars

suchasBintjeandPremière,respectively(seealsoZebarthetal.,2004).Theseeffects

weremainlyassociatedwithdifferencesinmaturitytype(VanKempenetal.,1996).

Latematuringcultivarscombinealongcanopycoverwithalongtuberbulkingperiod

(ED)andthereforeachievemoretuberdrymatteryield(wmax)perunitofNuptake

thanmidearlyandearlymaturingcultivars(Zebarthetal.,2008).Previousresearch

has also demonstrated that there is significant variation in crop uptake and use

efficiencyofNamongcommercialpotatocultivarsandadvancedclones(Kleinkopfet

al.1981;Lauer1986;Sattelmacheretal.,1990;PorterandSisson,1991a,b;Johnson

et al. 1995; Errebhi et al., 1998a,b, 1999; Sharifi et al., 2007; Zebarth et al., 2004,

2008).

MeancomparisonofNUEbetweentwoparents(SHandRH)andwithintheF1

populationisgivenby(Table3.2).NUEintheF1populationwas,onaverage,68.2g

Chapter3

100

DMg‐1Nwhichwasclosetothevaluesofthetwoparents(SH:69.6gDMg‐1N;RH:

67.2 g DM g‐1N). However, there was a very wide variation in NUEwithin the F1

population; it ranged from 54.4 g DM g‐1N to 81.6 g DM g‐1N. This wide range

suggested a transgressive segregation in F1 population for NUE (Fig. 3.3), which

could be further genetically manipulated for improving N use efficiency

characteristicsinpotato.

Therewassignificant(P<0.05)variationinNUEamongthesixenvironmentsfor

thestandardcultivars(Table3.3),aswellas fortheF1segregatingpopulation(Fig.

3.4); trendswereconsistent for the twogroupsofgenotypes.ThehighestNUEwas

recordedinExp3andthelowestinExp5,inlinewithlowertuberNuptake(11.3g

m‐2) observed in Exp 3 and higher tuber N uptake (15.3 gm‐2) observed in Exp 5

(Chapter2).ResponsestoNcanvarygreatlyfromsitetositeandfromyeartoyear.

Theydependonthecapacityof thesoil tosupplyNwhenthecropneeds it(Meyer,

1998)andonthecapacityofthecroptomakeefficientuseofthatN.

The NUE also varied within the experiments, with wide ranges of variation

observed in Exp 6 (Fig. 3.4). This might be due to the indirect effects of different

maturitygroupswithintheF1population.Thedifferentdurationsofthecropcycle,

asexpressedbymaturityclasses,isanobviousfactoronemightexpecttoaffecttheN

responsetoo(Vos,2009).

3.5.Phenotypic,geneticandenvironmentalvariances

Table 3.4 presents estimated values of phenotypic, genetic and environmental

variances for all the parameters and the derived traits in the F1 population. The

results revealed considerable phenotypic and genetic variances for all the traits

studied. All genetic and environmental components of variation were significant

(P<0.01)(Table3.4).

Thegeneticvariancecomponentcontributedamajorportiontothephenotypic

variance in traits cm, tE, andED (Table 3.4). The contribution of the environmental

variancetothephenotypicvariancewasrelativelylargeinwmax,RUE,andNUE(Table

3.4), probably because these traits were sensitive to nitrogen as we purposefully

appliedvarieddosesofnitrogenforcreatingcontrastingenvironments(seeMaterials

andMethods,Chapter2).Anumberofstudieshavealsoshownavaryingresponseto

ratesofNonthedrymatteryieldproductioninpotatocrops(e.g.PorterandSisson,

1991b;Maieretal.1994;Feibertetal.1998;Belangeretal.2000).

Estimates of phenotypic (CVPh) and genetic (CVG), and environmental (CVE)

coefficients of variation for traits across the six experiments are presented in

Table 3.5. Estimates of CVPh ranged from 19.5% to 54.7%. These estimates were

Geneticvariationintuberbulkingandresourceuseefficiencyofpotato

101

Table3.4.Variancecomponents fordifferenttraitswithintheF1populationacross

allsixexperiments.

Parameter σ2Ph σ2E σ2G σ2ε

cm(gtd‐1) 297.5 107.6** 120.6** 69.2tE(td) 253.9 51.1** 169.1** 33.7

ED(td) 301.0 65.1** 197.3** 38.6

RUE(gDMMJ‐1) 0.34 0.15** 0.09** 0.10

NUE(gDMg‐1N) 177.2 125.9** 20.7** 30.6

wmax(gm‐2) 47025 14706** 9902** 22417**Significantat1%.

σ2Ph=phenotypicvariance,σ2E=environmentalvariance,σ2G=geneticvariance,σ2ε=

residualvariance.

smallestforNUEandhighestforED.TraitswithhighCVPhexhibitlargetotalvariance

andareusefulasselectioncriteriainbreedingprovidedthetraitisalsoheritable.The

CVGestimateswerehigherthanCVEestimatesforalltraitsinvestigated.However,the

ratioofCVGoverCVEwasmuchgreaterintraitsNUE,te,andED,followedbyRUEand

wmax, whereas the lowest ratios were observed in cm. Our results indicated that

significant genetic variability existedwithin the F1 population formost traits. It is

therefore possible to utilise this wide genetic variability available for a breeding

programme aimed at improving tuber bulking dynamics, RUE,NUE, and ultimately

tuberdrymatterproduction.

Table 3.5. The phenotypic coefficient of variation (CVPh ), genetic coefficient of

variation (CVG), environmental coefficient of variation (CVE), and broad‐sense

heritability(H2)ofdifferenttraitswithintheF1populationacrossallsixexperiments.

Parameter ¶Mean CVPh(%) CVG(%) CVE(%) H2(%)

cm(gtd‐1) 35.5 48.6 30.9 29.2 91.3

tE(td) 51.0 31.3 25.5 14.0 96.8

ED(td) 31.7 54.7 44.3 25.5 96.8

RUE(gDMMJ‐1) 2.1 27.1 17.9 13.9 84.2

NUE(gDMg‐1N) 68.1 19.5 16.5 6.7 80.2

wmax(gm‐2) 957 22.7 12.7 10.4 72.6¶GrandmeanoftheF1segregatingpopulationacrossallsixexperiments.

Chapter3

102

3.6.GGEbiplotanalysisGGE biplot analysis allows visual examination of the relationships among the test

environments, genotypes, and the GE interaction (Yan, 2000). As N is the keyenvironmental variable affecting the important yield determining components of

potato(Honeycuttetal.,1996),herewediscussonlytheresultsfortraitNUEamong

the fourexperiments(Exps3‐4).TheGGEbiplotsrevealedthat the1standthe2nd

principalcomponentsaccountedfor84.04%oftheGEvariation(Fig.3.5).For any particular environment, genotypes can be compared by projecting a

perpendicular fromthegenotypesymbols to theenvironmentvector, i.e.genotypes

thatare furtheralonginthepositivedirectionof theenvironmentvectorarebetter

Figure3.5.GGEbiplot chart fornitrogenuseefficiency (NUE) inanF1 segregating

population.AECindicatesaverageenvironmentcoordinate(Yan,2001;YanandKang,

2003).Alinethatpassesthroughthebiplotoriginandaverageenvironmentindicates

themeanperformanceofgenotypes.

Geneticvariationintuberbulkingandresourceuseefficiencyofpotato

103

performing and vice versa. The biplot chart indicated that genotypes performed

exceptionallywell ineither loworhighinputenvironments(Fig.3.5).Thissuggests

that selection strategies can be developed for varieties with improved NUE. The

resultsfurthershowedthattwoenvironments(Exps4and5)weregroupedtogether.

Thissuggests that theseenvironmentswerehighlycorrelatedandrelativelysimilar

inthemannertheydiscriminateamonggenotypes.Theresultsfurtherrevealedthat

environments (Exps 4 and 5) fell close to the origin. This could mean that these

environments might have little variability across genotypes (Kroonenberg, 1997).

However,markersofenvironments(Exps3and6)werestandingfurthestapartfrom

the originwhichmight suggest that this environment causedmaximum variability

acrossthegenotypes.Theseenvironmentsmightthereforebethemaincontributorto

theoverallGEonaccountof their lowestandhighNavailability, respectively (seeTable2.1inChapter2)thantheotherenvironments.

3.7.Estimatesofbroad‐senseheritabilityAll estimates of broad‐sense heritability (H2) (eqn (3.7)) were very high and they

ranged from80.2 to96.8%(Table3.5). Inaprevioussectionweshowed that traits

likewmax, RUE, andNUEwere sensitive to environment.However, these traits also

had very high heritability estimates, which could mean that they can respond to

selection (Falconer and Mackay, 1996). High heritability estimates illustrate that

thesetraitshaveastronggeneticbasis,hencecouldbereliablyassessedandusedin

breeding for resource (radiation, N) use efficiency as well as tuber dry matter

productioninpotato.

3.8.Phenotypicandgeneticcorrelationsofmodelparametersandthesecondary

traits

Table 3.6 illustrates the phenotypic correlation coefficients among all the model

parametersandsecondary traitswithin theF1segregatingpopulationacrossall six

experiments.Allphenotypiccorrelationswerehighlysignificant(P<0.01)(Table3.6).

TheresultsshowedstrongnegativephenotypiccorrelationsbetweencmandtE(r=‐

0.83) and between cm andED (r= ‐0.84). These results suggest trade‐off between

tuberbulkingrateanddurationoftuberbulking.AsmentionedearliertBwasstable,

which means that ED was almost exclusively determined by tE; so, unsurprisingly

there was a strong positive correlation between tE and ED. The results further

revealed negative correlation (r = ‐0.37) between RUE and NUE. This suggested a

trade‐off between RUE and NUE, mainly caused by their negative and positive

Chapter3

104

Table 3.6. Phenotypic (lower triangle) and genetic (upper triangle) correlation

coefficientsamongallpairwisecomparisonsofdifferenttraitsacrosssixexperiments

ofanF1populationofpotato.tdstandsforthermalday.

Parameter cm tE ED RUE NUE wmax

cm(gtd‐1) − ‐0.91** ‐0.93** 0.95** ‐0.41** ‐0.20**

tE(td) ‐0.83** − 1.00** ‐0.91** 0.65** 0.66**

ED(td) ‐0.84** 1.00** − ‐0.93** 0.64** 0.64**

RUE(gDMMJ‐1) 0.69** ‐0.65** ‐0.65** − ‐0.75** ‐0.25**

NUE(gDMg‐1N) ‐0.28** 0.53** 0.52** ‐0.37** − 0.43**

wmax(gm‐2) ‐0.10** 0.55** 0.53** ‐0.02** 0.43** −**Significantat1%,NSNon‐significant.relationshipswithED,respectively(Table3.6).

Therewereweakbutnegative (r= ‐0.10)phenotypic correlationsbetweencm

andwmax(Table3.6).However, theresults indicatedstrongandpositivephenotypic

correlationsbetweentE,ED,andwmax(i.e.r=0.55and0.53,respectively).Fromthese

results, it seems that both rate and duration of tuber bulking are of importance in

determining final yield of potato crop. The positive role of NUEwas evident from

theseresultsduetoitspositivephenotypiccorrelationwithwmax(i.e.r=0.43).This

suggeststhatgenotypeswithhighNUEmayexhibithightuberyield.Theunderlying

relationships of wmax with important traits are elaborately discussed in the next

section.

Table 3.6 also illustrates the genetic correlation coefficients between the

model parameters and secondary traits within the F1 population. The results of

geneticcorrelationswereinlinewiththoseofphenotypiccorrelations.Asawholethe

coefficientvaluesforgeneticcorrelationswerecomparativelyhigherthanphenotypic

correlations.3.9.Pathcoefficientanalysis

Table 3.7 presents the results of path coefficient analysis describing thedirect and

indirect effects of different traits on tuber dry matter yield (wmax), while Fig. 3.6

presents the path coefficient structural model describing important relationships

among selected traits.The traitsAsum,RUEand cmhad thehighestdirect effects on

tuberdrymatteryield(wmax)(Table3.7).Ontheotherhand,very lowdirecteffects

onwmaxwereobservedfortraitsEDandNUE(i.e.0.06and‐0.04,respectively).The

results further illustrated that higher valueof direct effect ofAsum onwmaxwas the

Geneticvariationintuberbulkingandresourceuseefficiencyofpotato

105

Table3.7.Pathcoefficientanalysisofdirectandindirecteffectsofdifferenttraitson

thetuberdryyield(wmax)ofanF1population.tdstandsforthermalday.

Variable †Effect

Asum(td%)Directeffect 1.34

Indirecteffectvia cm ‐0.20ED 0.05

RUE ‐0.47NUE ‐0.02

Totalcorrelation 0.69cm(gtd‐1)

Directeffect 0.32Indirecteffectvia

Asum ‐0.85ED ‐0.05

RUE 0.47NUE 0.01

Totalcorrelation ‐0.10ED(td)

Directeffect 0.06Indirecteffectvia

Asum 1.21cm ‐0.27

RUE ‐0.46NUE ‐0.02

Totalcorrelation 0.53RUE(gDMMJ‐1)

Directeffect 0.69Indirecteffectvia

Asum ‐0.91cm 0.22ED ‐0.04

NUE 0.01Totalcorrelation ‐0.02

NUE(gDMg‐1N) Directeffect ‐0.04

Indirecteffectvia Asum 0.78cm ‐0.09ED 0.03

RUE ‐0.25Totalcorrelation 0.43

†Acrosssixenvironments.

Chapter3

106

AsumRUE

cm

Yield (wmax)

NUE

ED

1.34

0.59

-0.04

-0.37 -0.79

-0.720.52

0.90-0.28

-0.65 0.69

-0.84

0.320.06

0.69

Figure3.6.Pathcoefficientstructuralmodeldescribingdirectandindirecteffectsof

different traits on the tuber dry yield (wmax) across six experiments for an F1

segregatingpopulationofpotato.Thesolidlinerepresentsthecorrelationcoefficient

betweentwopredictorvariables;dashedlinerepresentsthepathcoefficientfromthe

predictorvariabletoresponsevariable(wmax).resultofsignificant,strongpositivecorrelationofAsumwithED(r=0.90)andNUE(r

= 0.59). On the other hand significant, strong, and positive correlations (r= 0.80)

between RUE and cmwere reflected in their higher values of direct effect onwmax.

Resultfurtherindicatedthattotalcorrelation(i.e.sumofdirectandindirecteffects)

betweenRUEandwmaxwasonly ‐0.02.Thiswasmainlydue to the strongnegative

indirecteffect(‐0.91)ofAsumonRUE.Thetotalcorrelationbetweencmandwmaxwas

alsolow(‐0.10). Inthiscasethestrongindirectnegativeeffect(‐0.85)ofAsumoncm

alsoplayeditsrole.

The above resultswere further supportedby the strongnegative correlations

between RUE and Asum (r = ‐0.79) and Asum and cm (r = ‐0.72) (Fig. 3.6). These

suggested that genotypes with higher Asumexhibit slow tuber bulking rate in the

Geneticvariationintuberbulkingandresourceuseefficiencyofpotato

107

linearphaseandmaybelessefficientinconvertingtheradiationinterceptedintodry

matteryieldoftubers.Thiscouldberelatedwiththeassimilationofdrymatterand

itsdistributionwithintheplant.Higherinvestmentintermsofbiomassallocationto

vegetative organsmay give highAsum and thereby higher total biomass, but on the

other hand a relatively low proportion may be used for the production of tubers,

especiallyifthemaintenancerequirementsarehigh.Excessivevegetativegrowthcan

be compensated to only a limited extent by redistribution of dry matter from

vegetativepartstotubers(VanHeemst,1986).

BasedontheseresultsitcanbestronglyconcludedthatAsum,RUE,andcmcould

be the traits having the strongest influence on the temporal dynamics of yield

formationinpotato.Fromabreedingperspective,anidealgenotypeshouldhavean

optimalAsum, without compromising RUE and cm. Our path analysis between traits

indicated the direction andmagnitude of correlated responses to selection and the

relative efficiency of indirect selection. In addition, our results suggest that while

using these traits as a criterion for selection, other causal relationships must be

consideredsimultaneously.

3.10.QTLdetection

Intotal16QTLswereidentifiedonbothSHandRHparentalgenomesacrossallsix

environments. In the SH genome, three QTLs were associated with two linkage

groups(IandV).ThirteenQTLslinkedtosevenlinkagegroups(IB,II, III,IV,V,VIII,

andVIII) on theRH genome.Table 3.8 describes thenumber ofQTLsdetectedper

parentalchromosome.

Table3.9summarisesthelistofQTLsdetected,theirparentalchromosomesand

mappositions and their characteristics (i.e. additive effects and variance explained

(R2))foreachofthetraitinvestigatedforindividualenvironment.AllQTLsdetected

were significant at (P<0.05)with –log10(P) values ranging from3.52 to 67.99. The

total fraction of phenotypic variance explained by effects of individual QTL ranged

from<0.1% to79%.Thepercentageofphenotypic variancewasevenhigherwhen

consideringtheirglobaleffects(rangingfrom27%to71%)(Table3.9).

ThreeQTLsweredetectedforcm(Table3.9).TheseQTLswereassociatedwith

linkage groups (RHV andRHVIII). TwoQTLsweredetected inExps1 and3with

63%and35%associatedcombinedQTLphenotypicvariance,respectively.Theother

environmentsshowedonlyoneQTL.ThemajorQTL116_5_17onlinkagegroupRHV

wascommonlydetected inmostofenvironments.ThisQTLhadmaximumnegative

additiveeffectthatrangedfrom21.488gtd‐1 to53.663gtd‐1andexplained24%to

57%phenotypicvariance.Theother twoadditionalQTLs(188_8_83and189_8_83)

Chapter3

108

Table3.8.DistributionofQTLsdetectedonparentalgenomes.

SH RH

Linkage

groupNo.ofQTLs

Linkage

groupNo.ofQTLs

I 2 IB 1

V 1 II 3

III 1

IV 1

V 4

VIII 2

X 1

Total 3 13onlinkagegroupRHVIIIhadpositiveadditiveeffects(i.e.23.719gtd‐1and12.058g

td‐1)with7%to13%associatedphenotypicvariance,respectively.

TwoQTLsweredetectedfortEandEDassociatedwithtwolinkagegroups(RH

IBandRHV)(Table3.9).OneQTLwasdetectedinalltheenvironmentsapartfrom

Exp 4 where two QTLs were detected. The phenotypic variance associated with

combinedQTLsdetected inExp4wasabout70%.QTL116_5_17on linkagegroup

RHVwasdetectedconsistently inall theenvironmentswithassociatedphenotypic

variancerangingfrom62%to79%.Here,thisQTLhadamaximumnegativeadditive

effect that ranged from ‐24.133 td to ‐66.019 td. The additional QTL 48_1_71 on

linkagegroupRHIBhadpositiveadditiveeffectsrangingfrom9.561tdto10.369td.

However, phenotypic variance explained by

theseadditionalQTLswaslessthan1%.

InthecaseofRUE,atleastoneQTLwasdetectedinmostofenvironmentsapart

fromExps2and6,wheretwoQTLsandnoQTLsweredetected,respectively(Table

3.9).ThetwoQTLsfoundinExp2jointlyexplained47%ofthephenotypicvariance.

The major QTL (116_5_17) on linkage group RH V was consistently found in

environmentswithadditiveeffectsrangingfrom0.641gDMMJ‐1to1.732gDMMJ‐1

with24%to55%associatedphenotypicvariance.

In total seven individual QTLs were detected for NUE (Table 3.9). The

distributionofQTLsweresuchasthreewerefoundinExp3,twoinExps5and6and

oneinExp4(Table3.9).ThephenotypicvarianceassociatedwithmultipleQTLsper

environmentrangedfrom27%to56%.AmongtheQTLs,116_5_17atposition18.2

Geneticvariationintuberbulkingandresourceuseefficiencyofpotato

109

Parameter

Exp.

QTL

Linkage

group

Markername

Position

(cM)

‐log

10(P)

aR2

GlobalR

2

c m(gtd

‐1)

1116_5_17

RHV

EAGAMCTC_470__5_17

18.2

24.97

38.833

0.57

0.63

189_8_83RHVIII

PAT/M

AGG_149.8__8_83

74.5

3.62

12.058

0.07

2116_5_17

RHV

EAGAMCTC_470__5_17

18.2

22.93

53.663

0.56

3116_5_17

RHV

EAGAMCTC_470__5_17

18.2

6.84

35.736

0.24

0.35

188_8_83RHVIII

PAG/M

AGC_298.8__8_83

76.1

4.01

23.719

0.13

4116_5_17

RHV

EAGAMCTC_470__5_17

18.2

24.54

51.958

0.57

5116_5_17

RHV

EAGAMCTC_470__5_17

18.2

17.74

21.488

0.48

6116_5_17

RHV

EAGAMCTC_470__5_17

18.2

18.8

34.06

0.50

t E(td)

1116_5_17

RHV

EAGAMCTC_470__5_17

18.2

67.62

‐55.624

0.79

2

116_5_17

RHV

EAGAMCTC_470__5_17

18.2

36.78

‐54.458

0.68

3

116_5_17

RHV

EAGAMCTC_470__5_17

18.2

19.376

‐38.459

0.63

4

116_5_17

RHV

EAGAMCTC_470__5_17

18.2

34.7

‐38.968

0.65

0.70

48_1_71

RHIB

EACAMCGT_296.2__1_71

33.9

3.67

9.561

<0.1

5

116_5_17

RHV

EAGAMCTC_470__5_17

18.2

51.11

‐24.133

0.63

6

116_5_17

RHV

EAGAMCTC_470__5_17

18.2

52.33

‐49.912

0.74

Table3.9.Maincharacteristicsofquantitativetraitloci(QTL)identifiedfordifferenttraitswithinthe‘SHRH’population

perindividualexperiment(i.e.environment.).DatagivenintablearefromtheCIMmappingmethod.QTLsmarkedasbold

aredetectedonlybytheCIMmethod,otherwisebyboththeCIMandSIMmethods.Exp.,experiment;position,positionof

maximum

–log 1

0(P);a,additiveeffectofthepresenceofparentalalleleatamarker;R

2 ,theindividualcontributionofone

QTLtothevariationinatrait;globalR

2 ,thefractionofthetotalvariationexplainedbyQTLsofthesametraitwithinsingle

environm

ent;td,thermalday.Sym

bols‘−’and‘*’m

eanlackofQTLanddata,respectively.

Chapter3

110

Parameter

Exp.

QTL

Linkage

group

Markername

Position

(cM)

‐log

10(P)

aR2

GlobalR

2

ED(td)

1116_5_17RHV

EAGAMCTC_470__5_17

18.2

67.99

‐60.873

0.79

2116_5_17RHV

EAGAMCTC_470__5_17

18.2

37.54

‐60.247

0.68

3116_5_17RHV

EAGAMCTC_470__5_17

18.2

18.756

‐41.077

0.62

4116_5_17RHV

EAGAMCTC_470__5_17

18.2

36.23

‐43.013

0.66

0.71

48_1_71

RHIB

EACAMCGT_296.2__1_71

33.9

3.7

10.369

<0.1

5116_5_17RHV

EAGAMCTC_470__5_17

18.2

52.34

‐66.019

0.74

6116_5_17RHV

EAGAMCTC_470__5_17

18.2

52.35

‐52.988

0.74

RUE

1116_5_17RHV

EAGAMCTC_470__5_17

18.2

22.72

1.732

0.55

(gDMMJ‐1)

2116_5_17RHV

EAGAMCTC_470__5_17

18.2

14.87

1.259

0.43

0.47

121_5_46RHV

EAACM

CAG_231.8__5_46

52.6

6.74

0.402

0.15

3116_5_17RHV

EAGAMCTC_470__5_17

18.2

9.862

1.329

0.39

4116_5_17RHV

EAGAMCTC_470__5_17

18.2

16.32

1.135

0.45

5116_5_17RHV

EAGAMCTC_470__5_17

18.2

6.67

0.641

0.24

6

−−

−−

−−

−

Table3.9.(Continued)

Geneticvariationintuberbulkingandresourceuseefficiencyofpotato

111

Parameter

Exp.

QTL

Linkage

group

Markername

Position

(cM)

‐log

10(P)

aR2

GlobalR

2

NUE

1*

* *

* *

* *

(gDMg‐1N)

2*

* *

* *

* *

3116_5_17

RHV

EAGAMCTC_470__5_17

18.2

3.6

‐8.79

0.12

0.39

84_3_3

RHIII

PAT/M

AAC_298.3__3_3

04.29

8.881

0.11

204_10_34

RHX

EAACM

CCA_216.8__10_34

31.4

4.99

‐10.835

0.17

475_2_51

RHII

EAGTMCAC_249__2_51

50.9

3.83

‐8.352

0.15

56_1_2

SHI

PAC/MACT_232.4__1_2

6.6

3.293

‐6.291

0.10

0.27

68_2_32

RHII

PAT/M

AAC_570.4__2_32

32.2

4.686

‐8.005

0.16

6116_5_17

RHV

EAGAMCTC_470__5_17

18.2

18.91

‐30.187

0.48

0.56

179_5_77

SHV

PAC/MATA_201.4__5_77

77.2

3.7

10.299

0.03

wmax(gm‐2)

1117_5_37

RHV

EACAMCGT_250.1__5_37

35.7

3.82

274.016

0.15

2

131_5_55

RHV

PAC/MATA_99.4__5_55

62.6

4.36

246.053

0.18

3

116_5_17

RHV

EAGAMCTC_470__5_17

18.2

15.84

‐630.531

0.44

0.53

81_2_75

RHII

PAC/MAGG_527.2__2_75

79.1

4.1

‐252.07

0.06

4

18_1_32

SHI

EAACM

CAG_187.9__1_32

36.3

3.52

244.584

0.13

5

−−

−−

−−

−

6

116_5_17

RHV

EAGAMCTC_470__5_17

18.2

8.94

‐458.44

0.29

0.41

99_4_35

RHIV

EACAMCTG_69.5__4_35

11.2

4.46

256.493

0.08

Table3.9.(Continued)

Chapter3

112

cMonthelinkagegroupRHVwasconsistentlydetectedinhalfoftheenvironments

explaining12%to48%phenotypicvarianceforNUE.ThiswasamajorQTLwithits

additiveeffectsrangingfrom‐8.79gDMg‐1Nto‐30.187gDMg‐1NinExps3and6,

respectively. These results indicated a threefold decrease in additive effects of this

QTLinthe lowNenvironment(i.e.Exp3;seeTable2.1inChapter2),whichmight

suggestthatallelesonpaternalchromosomeVatposition18.2cMmayreduceNUE

lessdrasticallyundersuchlowNconditions(Fig.3.4).Moreover,fouradditionalQTLs

were detected on linkage groups (RH II, RH III, SH I and SHV)with their additive

effects(‐8.005,8.881,‐6.291,and10.299gDMg‐1N,respectively).Interestingly,QTL

179_5_77atposition77.2 cMon thematernal (SH) linkagegroupVwasassociated

withapositiveadditiveeffectinExp6.Thiscouldsuggestthatallelesonthematernal

and paternal linkage groups V are associatedwith negative and positive effects on

NUE,respectively.

QTLmapping for tuberyield(wmax)perse showedsixQTLs(Table3.9).These

QTLswerescatteredoverfourparentallinkagegroups(SHI,RHII,RHIV,andRHV).

ThephenotypicvarianceexplainedbyQTLsrangedfrom(6%to44%).OneQTLwas

detectedinExps1,2,and4,whereastwoQTLsweredetectedinExps3and6.NoQTL

couldbedetectedinExp5.ThemultipleQTLsfoundinExps3and6jointlyexplained

53% and 41% of the phenotypic variance, respectively. Among the total QTLs

detectedforthiswmax,(116_5_17)onlinkagegroupRHVwasamajorQTLdetectedin

most of environments (i.e. Exps 3 and 6) with its additive effects ranging from

‐458.44 gm‐2 to ‐630.531 gm‐2with 29% to 44% associated phenotypic variance,

respectively(Table3.9).TheseresultsareinlinewithNUEandsuggestalsothatQTL

(116_5_17) at position 18.2 cM on paternal linkage group V is sensitive to the

environmentparticularlyNasnegativeeffectscausedbyallelesassociatedwiththis

QTLchangedinmagnitudewithrespecttoNavailability.Itisinterestingtonotethat

meanwmaxforExp3wasevenhigherthanExp6withhighNavailability(Table3.3).

As previouslymentioned, variability among the genotypeswas higher in Exp 3 for

most traits (Fig. 3.4). It would be expected that the some genotype were most

effective in thisenvironmentespecially forNUEas suggestedby thebiplotanalysis

(Fig.3.5).Thereforeafocusonwhyparticulargenotypesperformexceptionallywell

in loworhighinputsituationscouldenableselectionstrategiestobedevelopedfor

improvedvarieties.

InordertosummarisetheQTLmappingresults,QTLswithmajoreffectswere

associatedwithpaternal(RH)linkagegroupV,wheremaximumnumberoffourQTLs

wasdetected(Table3.8and3.9).OneparticularQTL(i.e.116_5_17)onpaternal(RH)

linkagegroupVatposition18.2cMwasdetectedforalltraitswithamajoradditive

effect and explainedmost of the total phenotypic variance (Table 3.10). Additional

Geneticvariationintuberbulkingandresourceuseefficiencyofpotato

113

QTLs with minor effects was mostly associated with only paternal (RH) linkage

groups for traits (cm, tE, and ED), whereas both maternal (SH) and paternal (RH)

linkagegroupswereassociatedwithtraitsNUEandwmaxforminorQTLs(Tables3.9

and3.10).OurresultsareinlinewiththosebyVandenBergetal.(1996),whoalso

reported thatmost of the QTLs had small effects, but a QTLwithmain effect was

found on chromosome V. It was further noted that paternal QTL (116_5_17) was

associatedwithitsnegativeadditiveeffectsformostofthetraitsincludingtuberyield

(wmax)perseexceptforcmandRUE,wherethisQTLshowedthepositiveeffects.This

indicatesthatRHallelesforthisQTLcausesynergisticeffectsduringtheearlyphase

ofplantgrowthwhenthetuberbulkingratesandRUEareattheirmaximum.

Itwasevident fromourresults thatvariation in tuberyield isassociatedwith

variationinbulkingrate,durationofcanopycover,andassociatedextentofradiation

interception (Fig. 3.6). However, these components are not physiologically

independent as genotypeswith a large tuber bulking ratemay effectively limit the

cropgrowthandduration(viaenhancedinternalplantcompetition)leadingtotheir

identified'earliness'.

We also observed co‐localisation of QTLs with many traits. For instance

clustering ofmanyQTLswere found on position18.2 cMon paternal (RH) linkage

group (Table 3.11).Heremost of the traits (e.g. cm, tE, andED)were tightly linked

withQTL116_5_17inmajorityoftheenvironments.ThiscouldmeanthatthisQTLis

playingapleiotropicroleindeterminingthesetraits.Thestronggeneticcorrelations

between these traits confirm these relations (Table 3.6). This may indicate the

difficulties of manipulating correlated traits simultaneously. However,

QTLs with similar behaviours could also be interesting targets for breeding

programmes as they aremore likely to be stable under various environments.Our

Table3.10.ListofparentallinkagegroupswithmajorandadditionalminorQTLs.

Parameter SHlinkagegroup RHlinkagegroup

MinorQTLs MajorQTLs MinorQTLs

cm − V VIII

tE − V IB

ED − V IB

RUE − V −

NUE I,V V II,III,X

wmax I V II,IV

Chapter3

114

resultsalsoindicatedanumberofindependentQTLsmainlyforNUE.TheseQTLsdid

notcoincidewithothertraits(Table3.12).TheseQTLshowever,withminoreffects

couldbeofgreatvalueforbreedingforNUE.

Several authors have indicated that some yield QTL coincide with those for

component traits, whereas other yield QTL map independently from component

traits (Xiao et al., 1995; Bezant et al., 1997). Few QTLs were expressed in one

environmentbutnotintheother(Table3.13).ThiswasmostlyevidentfortraitsRUE,

NUEandwmax.ThelowerrepeatabilityofsomeQTLsovertheenvironmentssuggests

QTLE interaction. Our results support this as for these traits the environmentalvariance component contributedmajorly to the total phenotypic variance for these

traits (Table 3.4). QTLs controlling such traits often show low stability (Veldboom

and Lee, 1996; Reymond et al., 2004). Several researchers have identified loci that

interactedwiththeenvironmentindifferentplantspeciese.g.yieldinbarley(Yinet

al.,1999a,b;Teulatetal.2001;Voltasetal.,2001).

OnlyoneQTL116_5_17onRHVshowedupinalltheexperiments(Table3.14).

ThisQTLwas thereforestableacross theenvironmentsandthereforedidnotshow

muchoftheQTLE.SuchQTLscouldbeusefulformarkerassistedbreeding.4.ConclusionsInthischapterwetriedtoelucidatetheonset,rate,anddurationoftuberbulkingas

yield‐determining, complex quantitative traits in a set of varieties covering awide

range ofmaturity types and a well‐adapted diploid F1 segregating population.We

presented a physiological and quantitative genetic analysis of the traits underlying

thedynamicsoftuberbulking.Ourmodeldescribedwellthetuberbulkingdynamics

andgaveinsightintothevitalunderlyingcomponenttraitsinfluencingthetuberdry

matter production. Our results showed that tuber bulking growth rate (cm) was

comparativelyhigherforearlymaturingcultivarsfollowedbylatecultivars.However,

theeffectiveperiodoftuberbulking(ED)waslongerforlatematuringgenotypes.Asa

result, tuberdrymatteryield (wmax)washigher in latematuringgenotypes than in

mid‐lateandearlygenotypes.

We also studied resource (radiation and N) use efficiencies and their

relationships with tuber dry matter yield production and with other important

physiological traits. The RUE values were higher for early maturing genotypes

followed by mid‐late and late genotypes. Mean NUE values were higher for late

maturing genotypes than formid early and early cultivars. Latematuring cultivars

have maximum canopy cover as well as a long tuber bulking period (ED) and

thereforethehighesttuberdrymatteryields(wmax)perunitofNuptake.

Geneticvariationintuberbulkingandresourceuseefficiencyofpotato

115

QTL

Linkage

group

Markername

Position

(cM)

Exp.1

Exp.2

Exp.3

Exp.4

Exp.5

Exp.6

48_1_71

RHIB

EACAMCGT_296.2__1_71

33.9

t E,ED

116_5_17RHV

EAGAMCTC_470__5_17

18.2

c m,tE,

ED,RUE

c m,tE,

ED,RUE

c m,tE,

RUE,

NUE,

wmax

c m,tE,

ED,RUE

c m.,t E,

ED,RUE

c m,tE,

ED,

NUE,

wmax

Table3.12.ListofindependentQTLs(i.e.QTLsdetectedonlyonceforaparticulartraitinanyenvironm

ent).

QTL

Linkage

group

Markername

Position

(cM)

Exp.1

Exp.2

Exp.3

Exp.4

Exp.5

Exp.6

6_1_2

SHI

PAC/MACT_232.4__1_2

6.6

NUE

18_1_32

SHI

EAACM

CAG_187.9__1_32

36.3

wmax

179_5_77

SHV

PAC/MATA_201.4__5_77

77.2

NUE

68_2_32

RHII

PAT/M

AAC_570.4__2_32

32.2

NUE

75_2_51

RHII

EAGTMCAC_249__2_51

50.9

NUE

84_3_3

RHIII

PAT/M

AAC_298.3__3_3

0

NUE

99_4_35

RHIV

EACAMCTG_69.5__4_35

11.2

wmax

117_5_37

RHV

EACAMCGT_250.1__5_37

35.7

wmax

121_5_46

RHV

EAACM

CAG_231.8__5_46

52.6

RUE

131_5_55

RHV

PAC/MATA_99.4__5_55

62.6

wmax

189_8_83

RHVIII

PAT/M

AGG_149.8__8_83

74.5

c m

188_8_83

RHVIII

PAG/M

AGC_298.8__8_83

76.1

c m

204_10_34

RHX

EAACM

CCA_216.8__10_34

31.4

NUE

Table3.11.Listofco‐localisedQTLs(i.e.QTLsweresameformorethanonetrait).

Chapter3

116

QTL Group MarkernamePosition(cM)

Exp.1 Exp.2 Exp.3 Exp.4 Exp.5 Exp.6

116_5_17 RHV EAGAMCTC_470__5_17 18.2 × × × × × ×

Table 3.13. Number of QTLs detected across six experiments (environments).

Symbols‘−’and‘*’meanlackofQTLanddata,respectively.

Parameter Exp.1 Exp.2 Exp.3 Exp.4 Exp.5 Exp.6

cm 2 1† 2 1 1 1

tE 1† 1 1 2 1 1

ED 1 1 1 2 1† 1

RUE 1† 2 1 1 1 −

NUE * * 3 1 2 2†

wmax 1 1 2† 1 − 2

TotalQTLs 6 6 10 8 6 7†QTLswithmaximumadditiveeffects.

Table3.14. List of stableQTLs across six experiments (i.e. SameQTLs detected in

majorityofenvironments).

Theprospectsofimprovingatargettraitbyselectingforcomponenttraitsare

determined by the genetic variation for the particular component traits and their

correlationwiththetarget trait.Highgeneticvariabilityalongwithhighheritability

wasrecordedformostofthetraits.Phenotypicandgeneticcorrelationswerehigh(r

>0.50)forthesetraits.Thephenotypicandgeneticcorrelationsamongmodeltraits

suggest a trade‐off between cm and ED and between RUE and NUE. Path analysis

showedthatcm,RUE,andAsumhadamajorinfluenceonwmax.

The genetic parameters found in this study indicated existence of significant

geneticvariabilitywithin theF1population formost traits studied. Itwaspossible,

therefore, to further analyse this wide genetic variability available, which could

potentiallybeexploited inbreedingprogrammesaimedat improving tuberbulking

dynamics,RUE,NUE,andultimatelytuberdrymatterproduction.

TheAFLPmarkerswerethereforeusedtoperformQTLmappingofthemodel

traitsandresource(radiation,N)useefficiencies.Intotal16QTLswereidentifiedon

both SH and RH parental genomes across all six environments. QTLs with major

Geneticvariationintuberbulkingandresourceuseefficiencyofpotato

117

effects were associatedmainlywith paternal (RH) linkage group V. One particular

QTL (116_5_17) on this linkage groupwas detected for nearly all the traitswith a

majoradditiveeffectandexplainedmostofthetotalphenotypicvariance.Additional

minorQTLsweremostlyassociatedwithonlypaternal(RH)linkagegroupsfortraits

(cm,tE,andED),whereasbothmaternal(SH)andpaternal(RH)linkagegroupswere

associatedwithtraitsNUEandwmaxforminorQTLs.

Ourquantitativeapproach in combinationwithAFLPmarkers identifiedQTLs

that could be useful in the development of marker assisted breeding strategies in

potato yield improvement. It was evident from our results that variation in tuber

yield is associated with variation in bulking rate, duration of canopy cover, and

associated extent of radiation interception. However, these components are not

physiologically independent as genotypes with a large tuber bulking rate may

effectively limit the crop growth and duration (via enhanced internal plant

competition)leadingtotheiridentified'earliness'.

 

CHAPTER4 Model‐basedevaluationofmaturitytypeofpotatousingadiversesetof

standardcultivarsandasegregatingdiploidpopulation*

M.S.Khan,H.J.vanEck,P.C.Struik

AbstractThe objective of this chapter is to evaluate the performance of the conventionalsystemofclassifyingmaturity type inpotatoand toprovideaconceptofmaturitytypebasedoncropphysiology.Wepresentanapproachinwhichphysiologicaltraitsareusedtoquantifyandassessmaturity typeunambiguously forasetofvarietiescoveringawiderangeofmaturityclassesandadiploidF1populationseparatingformaturity and well‐adapted to Dutch growing conditions, both grown in sixenvironments.Wedefinedphysiologicalmaturitybasedonfourtraits:thedurationofmaximumgreencanopy,theareaunderthegreencanopycoverprogresscurve,andtherateanddurationoftuberbulking.Theresultsindicatedthatphysiologicalmaturity typecriteria tended todefinematurity classes lessambiguously than theconventional criterion. Moreover, the conventional criterion was subject to morerandom noise and lacked stability and/or repeatability compared with thephysiological traits. The physiological maturity criteria also illustrated thephysiologicaltrade‐offsthatexistedbetweentheselectedtraitsandunderlinedthesubtle complexities in classifying maturity type. This study highlighted thecapabilities of differentmaturity type criteria in discriminating between differentmaturityclassesamongthelargesetofgenotypes.Ournewapproachinvolvingkeyphysiological traits could be beneficial in offering physiology‐based criteria to re‐definematuritytype.Animprovedcriterionbasedonimportantphysiologicaltraitswould allow relating the maturity to crop phenology and physiology. This newcriterionmaybeamenable to furthergeneticanalysisandcouldhelp indesigningstrategiesforpotatoideotypebreedingforgenotypeswithspecificmaturitytypes.Keywords:Potato(SolanumtuberosumL.),tuberbulking,earliness,maturitytype,breeding,segregatingpopulation,cultivarchoice,statisticaltools.

_____________________________________*AcceptedinPotatoResearch(2012).

Chapter4

120

1.IntroductionCropsundergosequentialdevelopmentalphasesfromemergencetosenescencethat

are characterised by their chronological age, but also by their phenology and

reproductivecapacity.Inmostannualcropplants,thetransitionfromthevegetative

tothereproductivephaseismarkedbytheonsetoffloweringandseedproduction;

hencefullyreproductiveplantsareconsideredmature(Bond,2000).However,inthe

case of potato (Solanum tuberosum L.), most genotypes maintain the capacity to

developnewleavesandcontinuetogrowthroughoutthemajorpartoftheirlifecycle

thankstothefactthatatleastsomevegetativemeristemsremainindeterminate(Vos

and Biemond, 1992; Struik and Ewing, 1995; Vos, 1995a; Almekinders and Struik,

1996; Fleisher et al., 2006). Progress tomaturity is therefore difficult to assess in

potato.

Cropdevelopmentinpotatoisacomplextraitthatmaybecomprisedofaseries

of phenological events such as completion of canopy growth, termination of

sympodialgrowth,theonsetoftuberformation,ortheonsetoftherapidincreasein

harvestindex,saggingofplants,senescenceofleaves,etc.(Struiketal.,2005;Struik,

2010). The sequential phases of growth determining the duration of phenological

stages of potato genotypes could be used to understand and define the concept of

maturitytypeandtoclearlyclassifygenotypes.

Theconventionallyuseddefinitionofmaturitytypeinpotato,however,israther

ambiguous (Haga et al., 2012). Many public and private institutions have tried to

define thematurity in potato and have come upwith their own definitions. Public

institutions such as the International Union for the Protection of New Varieties of

Plants (UPOV), and the authorities involved in granting plant breeders’ rights use

foliage maturity as descriptor among the many traits for DUS (Distinct, Uniform,

Stable) criteria.Maturity (Trait 36) in the UPOV document

(http://www.upov.int/edocs/tgdocs/en/tg023.pdf)isusingthecriterion“Thetimeof

maturity is reachedwhen80%of the leavesaredead.” Private institutions, such as

breeding companies and marketing boards release brochures to advertise the

specificitiesoftheircultivars.Maturitytypeindicationsinthesebrochuresareusually

based on the personal scale of the breeder. Furthermore, several databases on the

WorldWideWebareavailablesuchas“theEuropeancultivatedpotatodatabase”,

(http://europotato.org/display_character.php?char_no=18&character=Maturity).

Itisremarkable,thatmanyseemtoknowhowtoclassifyplantsintoaspecific

scale, and even to make conversion from one scale to another, without a proper

definition of this complex syndrome of developmental events. As a whole, the

conventional method or criterion of elucidating the maturity type used is mostly

basedonvisualobservationsmadeinthefieldatparticulartimeintervalsduringthe

Modelbasedanalysisofmaturitytypeofpotato

121

cropcycleandtransferringthatinformationtounit‐lessordinalscales.Thiscriterion

doesnothaveanybiologicalmeaningandinvolvesambiguityandmuchspeculation

in understanding the background of maturity, partly because of the different

viewpoints of growers and processors. For instance, the grower may consider

maturity as the onset of canopy senescence and/or when there is no further net

growth (Bodlaender and Reestman, 1968). The processor on the other hand may

consider the declining levels of reducing sugars in the tubers as an indication of

approaching maturity (Burton andWilson, 1970). Yet others think maturity to be

synonymouswithtuberdrymatterconcentration(BeukemaandvanderZaag,1979).

Onaccountofsuchdiscrepancies,theconventionalmaturitycriterionmayalso

notbe stable (changesof thematurity classof aparticular varietyhaveoftenbeen

reported), subject to personal bias and showing strong genotype‐by‐environment

(GE) interaction. Hence, there is a strong need for a clear and unambiguousdefinitionofthematuritytypeofpotatogenotypesbasedonaclearunderstandingof

potato physiology and of the GE. Crop physiologists, agronomists and breederswouldallprofitfromsuchacleardefinition.StudiesbyStruiketal.(2005)andStruik

(2010)haveshownthatredefiningthepotatomaturitybasedonphysiologicaltraits

ispromising.

Current state of the art in statistical knowledge and techniques has made it

possible to properly analyse large amounts of available physiological and genetic

information.Forinstance,clusteranalysisandpathcoefficientanalysisareuseful in

explorative data mining, information retrieval, and data summarisation. Cluster

analysis offers a meaningful grouping and/or sub‐grouping based on similarities

foundinthedata.Additionally,itofferswaystocharacteriseeachgroupintermsofa

clusterprototype, i.e.adataobject that is representativeof theotherobjects in the

group.Thismayhelpinpatternrecognition,targetingappropriatetreatment(s),and

studying typologies as well as finding the genetic relationships and/or diversity

amongthelargesetofgenotypes(Dunemannetal.,1994;Piluzzaetal.,2005;Arshad

et al., 2006;Ramet al., 2008). Inaddition,procedures likepath coefficient analysis

has proven to be useful in giving thorough understanding of relationships and

contributionof various characters to the target trait bypartitioning the correlation

coefficientintocomponentsofdirectandindirecteffects.Bhatt(1972)reportedthat

merelycorrelationstudiesdonotclearlyrevealsuchtypeofinformationandcanbe

misleadingifthehighcorrelationbetweentwotraitsisaconsequenceoftheindirect

effectofthetraits(DeweyandLu,1959).

Given the rich quantitative data set at hand from Chapters 2 and 3 and the

availabilityofstateoftheartstatisticaltools,thischapteraimstoevaluate,quantify,

andre‐definetheconceptofmaturitytypeofalargesetanddiversesetofgenotypes

Chapter4

122

ofpotato.Ourobjectiveistodevelopaphysiologicalapproachofassessingmaturity

typeduringthewholecourseofplantdevelopment.Thismethodshouldbereliable,

robust and should allow the phenotyping of large numbers of genotypes across

diverse environments. This study will also provide more insight into the genetic–

physiologicallinkageofmaturitytypeinpotato.

2.Materialsandmethods

2.1.Set‐upofexperiments

Datasetsforouranalysiscamefromsixfieldexperiments,performedinWageningen

(52˚Nlatitude),theNetherlands,during2002,2004,and2005withtwoexperiments

in eachyear.Theseexperimentsdiffered inenvironmental conditionsbecause they

werecarriedoutindifferentyears,ondifferentsoils,andunderdifferentNfertiliser

regimes(fordetails,seeChapter2).Theexperimentswereconductedunderstress‐

freegrowingconditionsandthestatisticaldesignsusedwererandomizedcomplete

block designs with two blocks. The planting material consisted of disease‐free,

uniform sized seed tubers of similar physiological quality from an F1 diploid

(2n=2x=24) potato population (100 genotypes) derived from a cross between two

diploid heterozygous potato clones, SH83‐92‐488 RH89‐039‐16, or simply the‘SHRH population’ (Rouppe van der Voort et al., 1997; Van Os et al., 2006). Thispopulationsegregatesformaturitytypeandiswell‐adaptedtoDutchconditions(Van

derWaletal.,1978;VanOijen,1991)thusmakingitidealforthisstudy.Asetoffive

standard cultivars (viz. Première, Bintje, Seresta, Astarte, and Karnico) covering a

widerangeofmaturitytypesunderDutchconditionswerealsoincludedinthisstudy

tobenchmarktheeffectsofmaturitytypeamongknowngenotypes.

2.2.Analyticalapproach

2.2.1.Selectionofphysiologicaltraits

Maturityisacomplexphenomenonaffectedbymanycomponentsofcropgrowthand

development. However, the variation in maturity between the genotypes could be

reflectedindifferingperiodsofattainingcriticalphysiologicalstagessuchascanopy

development,tuberinitiation,filling,andtheirtotalduration.InChapters2and3,we

developedphysiologicalapproachestoquantifythedynamicsofcanopydevelopment

Modelbasedanalysisofmaturitytypeofpotato

123

and of tuber bulking of potato during the entire crop cycle in the aforementioned

large setof genotypes.Thisworkyieldeda large setofbiologicallymeaningful and

geneticallyrelevantcomponenttraitsdirectlyrelatedtotheabilityof thegenotypes

to intercept photosynthetically active radiation and tuber dry matter production.

Most of these traits successfully explainedpart of the genetic variation inmaturity

typeandcouldbeusedtoenhancethebreedingeffortsaimingatelucidatingpotato

maturity.Inthisstudy,however,forthesakeofsimplicity,weselectedonlyfourtraits:

DP2,cm,ED,andAsumbasedontheirstronggeneticnatureandveryhighresponseto

direct selection (Chapter 6). DP2 is the duration of the phase with maximum and

constant canopy cover (in thermal days (td)). Asum is the area under whole green

canopycovercurve(intd%),cmistherateoftuberbulking(ingm‐2td‐1)andED is

thedurationoftuberbulking(intd).

2.2.2.Criteriaforre‐defininggenotypematuritytype

In the Netherlands, according to the conventional system, potato varieties are

officiallyclassifiedintofourmainmaturityclasses:veryearly,early/mid‐early,mid‐

late/late, and very late. However, for the sake of simplicity maturity classes

early/mid‐early and mid‐late/late were considered as early and late, respectively,

throughoutthetext,tables,andfigures.Genotypeswereallocatedtomaturityclasses

(veryearly,early,late,andverylate)basedoneachofthefourselectedphysiological

trait using the Ward’s minimum‐variance clustering method in SAS software (SAS

Institute Inc. 2004). A schematic representation of cluster analysis for identifying

differentmaturityclassesisdescribedbyFig.4.1.Theinputdataconsistedofmeans

of F1 genotypes, their two parents, and five standard cultivars across six

environments.

After the identification of the maturity classes, further sub‐clustering of

genotypes within each maturity class was done using the nearest neighbour

clusteringmethod.Eachsub‐clusterofgenotypeswithineachofthematurityclasses

was assigned different maturity scores according to the Netherlands Potato

ConsultativeFoundation(NIVAP,2007),forinstance,maturityscores8.0,8.5,9.0and

9.5intheclusterofveryearlygenotypes,maturityscores2.0,2.5,3.0,3.5forverylate,

etc.Table4.1describestherangesofscalesforeachmaturitytype.Finally,thiswhole

procedureresultedintore‐definedmaturitytypecriteriabasedonourfourselected

physiological traits. The scoring procedure was employed in order to make later

comparison possible between the conventional and newly re‐defined criteria of

genotype maturity type, now referred to as “physiological maturity type criteria”

throughout the remaining text. For the conventional criterion, we followed the

Chapter4

124

Figure4.1.AschematicrepresentationofclusteranalysisforsingletraitxwithinF1

population,wherex=DP2 orcmorED orAsum.Thedashed line showshypothetical

intersectionoffourmainclustersfortheselectionoffourmaturityclasses(veryearly,

early,late,andverylate).generalpracticeusedbybreeders, i.e.visualobservationsweremade inthe fieldat

particulartimeintervalsduringthecropcycleandtheinformationwastransferredto

unit‐lessordinalscales.FortheF1populationwehadthebreeders’assessments(i.e.

genotypematurityscales)basedontwoexperiments(environments)during2002.

Table4.1.Listofordinalscalevaluesfordifferentmaturityclasses.

Maturityclass Type Scalevalues

1 Veryearly 8.0,8.5,9.0,9.5

2 Early 6.0,6.5,7.0,7.5

3 Late 4.0,4.5,5.0,5.5

4 Verylate 2.0,2.5,3.0,3.5

Modelbasedanalysisofmaturitytypeofpotato

125

2.3.Statisticalanalysis

AllstatisticalanalyseswerecarriedoutusingGenstatcomputersoftware(Payneetal.,

2009). Path coefficient analysis was performed to ascertain the inter‐associations

between our selected physiological traits following the approach of Dewey and Lu

(1959).Thenon‐parametricMann‐WhitneyUtestwasusedtocheckthesignificance

ofdifferencesbetweentheconventionalandphysiologicalmaturitycriteriafortheF1

population. Descriptive statistics was performed and the relationships between

maturityscorespermaturitycriterion(i.e.conventionalandre‐definedcriteria)were

assessed using the Pearson correlation coefficient. Finally, regression analysis was

performedtocheckthecapabilityofconventionalandphysiologicalmaturitycriteria

to predict thematurity type of genotypes. As data consisted of ordinal scales, data

was log‐transformed prior performing the correlation and regression analysis to

satisfytestassumptions.

3.Resultsanddiscussion

3.1.Pathcoefficientanalysis

Wesurmisedbasedonourpreviousresults(seeChapters2and3)thatAsumcouldbe

seen as an expression of a series of underlying physiological, genetic, and

environmentalprocessesandinteractionsaffectingthematurityandyieldproduction

inpotato(cf.Vos,1995b;2009).Pathcoefficientanalysiswasthereforeaccomplished

bypartitioningthedirectandindirecteffectsofourselectedphysiologicaltraitsupon

Asum. Figure4.2presentsthepathcoefficientstructuralmodeldescribingimportant

relationshipsamongselectedphysiological traits.Amongtheselected traits,EDand

DP2hadthehighestdirecteffectsonAsum(0.69and0.51,respectively,seedashedlines

inFig.4.2).Ontheotherhand,cmshowedthelowest(0.22)directeffectonAsum.The

resultsfurtherindicatedastrongpositivecorrelationbetweenDP2andED(r=0.79)

andnegativecorrelationsbetweencmandDP2(r=‐0.63);cmandED(r=‐0.93)(Fig.

4.2).Theseresultssuggestthatgenotypeswithhighertuberbulkingrates(cm)exhibit

lower DP2 or ED and may mature early. Compared with late genotypes, early

genotypesallocatealargepartoftheiravailableassimilatestothetubersearlyinthe

growing season, leading to shorter crop cycle (Kooman and Rabbinge, 1996). In

conclusion,pathcoefficientanalysis indicated thatprogress tomaturity inpotato is

stronglyinfluencedbyourselectedphysiologicaltraits.Basedonthestrengthoftheir

directeffectsoncropmaturity,suchtraitscouldbecategorisedinfollowingorder:

ED>DP2>cm.

Chapter4

126

3.2.Descriptiveanalysisofdifferentmaturitytypedefiningcriteria

The first step in the analysis was to apply some standard descriptive statistical

methodstothedata.Table4.2showsthemeans,standarddeviations,andminimum,

maximumvaluesofgenotypesallocated inmaturityclassesunderconventionaland

ED

Asum(Maturity type)

DP2

0.79

0.51 0.69

- 0.93cm

0.22

-0.63

Figure4.2.Pathcoefficientstructuralmodeldescribingdirectandindirecteffectsof

different traits on the maturity type (defined by Asumi.e. area under whole green

canopy curve) across six experiments in an F1 population. DP2 represents the

durationofmaximumgreen canopyphase,cmandED aregrowth rateandeffective

durationofthe linearphaseoftuberbulking,respectively.Thesolid linerepresents

the correlation coefficient between twopredictor variables; dashed line represents

the path coefficient from the predictor variable to response variable (Asum). See

MaterialsandMethodsforfurtherdetails.

Modelbasedanalysisofmaturitytypeofpotato

127

Table4.2.Mean,minimum (Min),maximum (Max) values ofmaturity scores, and

percentage (%) of genotypes allocated to each maturity class for five different

maturitytypecriteriawithintheF1population(100genotypes).

Maturitytypecriterion1

andclass

Mean(±S.E.)2 Min Max%

Conventional

Veryearly 8.3(±0.1) 8.0 9.5 35.4

Early 7.0(±0.1) 6.0 7.5 53.8

Late 4.9(±0.1) 4.5 5.5 8.6

Verylate 3.0(±0.5) 2.5 3.5 2.2

DP2

Veryearly 8.3(±0.2) 8.0 9.0 7.3

Early 6.9(±0.1) 6.0 7.5 69.7

Late 4.8(±0.1) 4.0 5.5 16.5

Verylate 2.8(±0.2) 2.0 3.5 6.4

cm

Veryearly 8.6(±0.1) 8.0 9.5 18.5

Early 6.5(±0.1) 6.0 7.5 33.3

Late 4.9(±0.1) 4.0 5.5 34.3

Verylate 2.9(±0.1) 2.0 3.5 13.9

ED

Veryearly 8.7(±0.1) 8.0 9.5 52.8

Early 7.1(±0.1) 6.0 7.5 16.7

Late 4.6(±0.2) 4.0 5.5 10.2

Verylate 2.7(±0.1) 2.0 3.5 20.4

Asum

Veryearly 8.8(±0.2) 8.0 9.5 7.3

Early 7.0(±0.1) 6.0 7.5 62.4

Late 4.8(±0.1) 4.0 5.5 23.9

Verylate 2.9(±0.2) 2.0 3.5 6.41Conventional scale: based on breeders’ criteria; DP2 scale: based on duration of

maximumgreencanopycover;cmscale:basedontuberbulkingrate;EDscale:based

on effective duration of linear tuber bulking; Asumscale: based on the area under

wholegreenthecanopycurve.2MeanofF1segregatingpopulation(100genotypes)acrosssixenvironments.

Chapter4

128

physiological maturity type criteria. The mean values of maturity scaling varied

significantly (P<0.01) across fourmaturity classes (very early, early, late, and very

late) for eachmaturity type criterion. Examination of theminimum andmaximum

values illustrated that mean ranges of maturity classes were consistently similar

betweendifferentmaturitytypecriteria.Thiswasakindofcheckandprovedthatour

procedureofassigningthematurityscoreswassatisfactory.However,thepercentage

of genotypes allocated to eachmaturity classdifferedamongmaturity type criteria

(Table 4.2). Formaturity class ‘very early’, theDP2 andAsummaturity type criteria

allocatedminimum (7.3%) of total genotypes, while theED criterion allocated the

maximum(52.8%)numberofgenotypes.Incaseofmaturityclass‘early’,theEDand

DP2criteriaallocatedtheminimum(16.7%)andmaximum(69.7%)oftotalgenotypes,

respectively. For ‘late’ maturity class, the conventional criterion allocated the

minimum(8.6%)whereasthecriterionbasedoncmallocatedthemaximum(34.3%)

of total genotypes. On the other hand,minimum (2.2%) andmaximum (20.4%) of

totalgenotypeswereallocated tomaturity class ‘very late’by theconventionaland

EDmaturity–typecriteria,respectively.

These results illustrated the capabilities of these different maturity type

defining criteria in describing the extent of maturity type among the large set of

unknowngenotypes.

3.3.ComparisonofdifferentmaturitytypecriteriaThestatisticalcomparisonofdifferentmaturitytypecriteriafortheF1populationis

presented inTable4.3.Resultsof thenon‐parametricMann‐WhitneyUtestshowed

that there were significant (P<0.01) differences in maturity scores among the five

criteriaexceptbetweenconventionalandEDandbetweenDP2andAsum(Table4.3).

Figure 4.3 evaluates differences in maturity scales for standard cultivars. All

maturitycriteriagavedifferentscalestothestandardcultivars.Moreorlesssimilar

trend of maturity scaling was observed across the criteria following the order:

Première>Bintje>Seresta>Astarte>Karnico(Fig.4.3).

TheresultsindicatedfurthermorethatPremièrewasmarkedas‘veryearly’by

mostofthecriteria.However,accordingtoDP2andAsumcriteria,Premièrewasplaced

in the ‘early’ category. Bintje was marked as ‘early’ by all criteria, except the cm

criterionwhichputBintjeinthe‘veryearly’category.CultivarSerestawasmarkedas

‘late’ by the conventional,cmandED criteria, but the same cultivarwas considered

‘verylate’byDP2andAsumcriteria.AstarteandKarnicoweremarkedas‘verylate’by

all criteria except cm which considered these cultivars ‘late’. Apparently cm and

conventional criterion showed least and maximum resolution, respectively for

maturitytypeamongthestandardcultivars.

Modelbasedanalysisofmaturitytypeofpotato

129

Table 4.3. Mann‐Whitney U test for checking extent of variation among different

maturitytypecriteria.

Maturityclass1 Mann‐WhitneyUtest2

Conventionalscaleversus

DP2scale **

cmscale **

EDscale NS

Asumscale **

DP2scaleversus

cmscale **

EDscale **

Asumscale NS

cmscaleversus

EDscale **

Asumscale **

EDscaleversus

Asumscale **1Conventional scale: based on breeders’ criteria; DP2 scale: based on duration of

maximumgreencanopycover;cmscale:basedontuberbulkingrate;EDscale:based

oneffectivedurationoflineartuberbulking;Asumscale:basedontheareaunderthe

wholegreencanopyprogresscurve.2**Significantat1%,NS=non‐significant.

Theseresultshighlightedthecomplexitiesindefiningthematuritytypeinpotato.It

was notable that although conventionalmaturity criterionwas able to classify the

genotypes into different maturity classes, it could not offer a clear definition. In

contrast, results from physiological maturity type criteria were easily and clearly

interpretable. Our integrated approach involving different physiological traits

therefore could be very beneficial in offeringphysiology‐based criteria to re‐define

maturitytype.

Chapter4

130

Figure4.3.Comparisonofconventionalandphysiologicalmaturity typecriteria for

fivestandardcultivars.DP2representsthedurationofmaximumgreencanopyphase,

cmandEDaregrowthrateandeffectivedurationofthelinearphaseoftuberbulking,

respectively,Asumisareaunderwholegreencanopycurve.

3.4.Performanceandstabilityofconventionalmaturitytypecriterion

Our next objective was to check the performance of the conventional criterion in

termsofitscapacitytoclearlypickupgenotypesforeachphysiologicaltrait(i.e.DP2,

cm,ED,andAsum).Considerableoverlappingamongthematurityclasseswasobserved

intheconventionalcriterion(Fig.4.4).Theextentofstabilityand/orrepeatabilityof

maturity scores from the conventional criterionwere checked for the quantitative

physiologicaltraits.

Figure 4.5 compares the maturity scores from conventional method for one

physiological trait Asum. The results indicated that Asum values showed a lot of

variationbothwithinandthroughoutdifferentmaturityclasses.Forinstance,itwas

notedthattheconventionalcriterioncouldnotidentifygenotypeswithhighAsum(i.e.

very lategenotypes)asevident fromthe lackofmaturityscores intherangeof2‐4

(Fig.4.5).

These results indicated the weakness and lack of robustness of conventional

maturitycriterionindealingwithrathercomplexphysiologicalquantitativetraits.It

can be concluded that such criterion is not quite able to clearly interpret and

discriminate between maturity classes for the crucial growth related processes of

1

2

3

4

5

6

7

8

9

10

Conventional cm Asum

Mat

urity

sca

le

Maturity defining criteria

Premiére Bintje Seresta Astarte Karnico

cm ED AsumDP2

Modelbasedanalysisofmaturitytypeofpotato

131

Figure4.4.Boxplotsillustratingtherangesofgenotypesinfourmaturityclassesas

assessed by conventional criterion for physiological traits: DP2, cm, ED, and Asum,

withintheF1population(100genotypes).DP2representsthedurationofmaximum

green canopyphase,cmandED are growth rate andeffectivedurationof the linear

phaseoftuberbulking,respectivelyandAsumisareaunderwholegreencanopycurve.

Theboxesspantheinterquartilerangeofthetraitvalues,sothatthemiddle50%of

the data laywithin the box,with a horizontal line indicating themedian.Whiskers

extendbeyondtheendsoftheboxasfarastheminimumandmaximumvalues.VE=

veryearly;E=early;L=late;VL=verylate.

potato. An improved criterion based on important physiological traitswould allow

relating the maturity to crop phenology and physiology. Haga et al. (2012) used

flowerdevelopment, leafchlorophyllcontentand leafperoxidaseactivity toclassify

maturity of potato cultivars under temperate conditions. They observed that leaf

chlorophyllcontentwasthemostreliable indicatoramongthesethreevariables,an

indicator related toAsum as both reflect leaf senescence. However, leaf chlorophyll

content was mainly capable of discriminating between late and early‐medium or

medium‐late;itdidnotseparateearly‐mediumfrommedium‐latecultivarswell.

0

10

20

30

40

50D

P2

(td

)

0

10

20

30

40

50

60

c m (g

td

-1)

0

20

40

60

80

VE E L VL

ED

(td

)

0

2000

4000

6000

8000

VE E L VL

Asu

m (

td %

)

Maturity class

Chapter4

132

Figure 4.5. Extent of stability and/or repeatability of maturity scores from a

conventionalcriterionforquantitativetraits, forinstanceAsum(i.e.areaunderwhole

greencanopycovercurve)withintheF1population.

Our physiological based criteria of defining the maturity type could allow in

cleardiscriminatingbetweenallclasses.

3.5.Predictivecapabilityofdifferentmaturitytypecriteria

Differentmaturitycriteriawereanalysedfortheircapabilitiestopredictthematurity

in a large number of unknown genotypes across diverse environments. The

regression analysis was performed on the maturity scores from each maturity

criterion.Sofarourresultsindicatedthatnearlyallphysiologicaltraitssatisfactorily

definedthematuritytype.However,forthesakeofsimplicity,hereweconsideronly

onetraitAsumasasynonymtotheresponsevariable“maturity”.Figure4.6showsthe

results of regression between the observedAsum and theAsum (“maturity type”) as

predicted by regression analysis from the individual maturity criteria. The results

indicated thatalmostallmaturitycriteriapredicted thematuritywell, i.e.R2>0.50.

However, the conventional and cm based criteria predicted the maturity

comparatively less accurately with R2 values 0.68 and 0.51, respectively. Maturity

criterionbasedonEDgavethebest(R2=0.81)predictionsformaturitytypefollowed

0

2000

4000

6000

8000

1.0 2.5 4.0 5.5 7.0 8.5 10.0

Maturity score

Asu

m(t

d%

)

Modelbasedanalysisofmaturitytypeofpotato

133

Figure4.6.ComparisonofdifferentmaturitytypecriteriaforpredictingtheAsum(as

aproxyformaturitytype)withintheF1populationacrosssixenvironments.Panels

show regression analysis based on different maturity type criteria, i.e. (a)

conventional,(b)DP2(durationofmaximumgreencanopyphase),(c)cm(growthrate

during linear phase of tuber bulking), (d)ED (effective duration of tuber bulking).

NotethatvariableonY‐axis(i.e.Asum:areaunderwholegreencanopycovercurve)is

consideredequivalenttomaturity,seetext.

bycriterionbasedontraitDP2(i.e.R2=0.77).Theseresultswereinlinewithourpath

coefficientanalysisconclusions(Fig.4.2).Theseresultsconcludedthatoverall,most

of the physiological maturity type criteria were fully capable of reproducing the

maturitytypeofgenotypesacrossdiverseenvironments.

3.6.Direct/indirectselectionpossibilitiesformaturitytypeFigure 4.7 illustrates the relationships between maturity scores (Table 4.1) from

differentmaturity typecriteriawithin theF1populationacrossall sixexperiments.

Allcorrelationswerehighlysignificant(P<0.01)andpositive.

y = 0.6889x + 1198.7R² = 0.6889

1000

2500

4000

5500

7000 (a) y = 0.7692x + 881.46R² = 0.7691

1000

2500

4000

5500

7000 (b)

y = 0.511x + 1874.3R² = 0.5109

1000

2500

4000

5500

7000

1000 3000 5000 7000

(c) y = 0.8066x + 741.02R² = 0.8066

1000

2500

4000

5500

7000

1000 3000 5000 7000

(d)

Observed Asum (td %)

Pre

dict

ed A

sum (

td %

)

Chapter4

134

Figure4.7.Relationshipsbetweenmaturityscalevaluesfromdifferentmaturitytype

criteriawithinF1populationacrosssixenvironments.ValuesontheX‐axisandY‐axis

representlog‐transformedmaturityscales(cf.MaterialsandMethods).

The results indicated very strong phenotypic correlations between the Asum

criterionandDP2andED,andbetweencmandEDcriteria(r=0.89).Thisshowsthat

bothrateanddurationoftuberbulkingareimportantincontrollingthetotalduration

of crop growth and the capacity of the crop to intercept radiation, crucial in

determining final yield (Kooman andRabbinge, 1996). These results indicated that

maturity type of the genotypes based on Asum could be indirectly obtained by

selectinggenotypeswiththeproperDP2criterion.Ourresultsinpreviouschapters(2

and 3) showed that quantitative trait loci (QTL) for these physiological traits are

DP2 cm ED Asum

Con

vent

iona

l A

sum

E

D

c m

Maturity scale

Mat

urity

sca

le

Modelbasedanalysisofmaturitytypeofpotato

135

mostly co‐localised on chromosome V of the paternal (i.e. RH) genome in the F1

segregatingpopulationofpotatoandexplainmorethan50%ofthetotalphenotypic

variance.Moreover, our results in Chapter 2 indicated thatmaximumQTL additive

effectswereassociatedwithAsumthroughoutdifferentenvironments(seeFig.2.12),

illustrating thecloseassociationofmaturitywith thisphysiological trait.Therefore,

this locus could be used as an explanatory variable for further elucidating the

physiological as well as genetic aspects of maturity in potato. The conventional

criterion on account of its subjective nature does not explain maturity type any

further, inspiteof itshighcorrelationwithsomeof thephysiologicalmaturity type

criteria.

3.7.Applicationsofre‐definingmaturitytypebasedonphysiologicaltraits

Our study showed that re‐defining thematurity type on the basis of physiological

traits is useful. The new physiology‐based maturity definition could be applied in

plantbreedingandincropmanagement(Struik,2010).

Thephysiologicalmaturity criteria canplay a significant role indevising crop

escape strategies to cope with biotic stresses, such as late blight (Struik, 2010).

Strategieslikeadvancingthecropgrowthbeforethestressbecomestooharmfulmay

prove very useful. For instance, genotypes with early tuber set and faster tuber

bulking can better escape the disease severity. The physiological based maturity

criteriamayalso allow the selectionof thegenotypes formaturity type inorder to

breakthecorrelationbetweenresistanceandlatematurity.Characteristicsthatcould

beusefulindicatorsofcompetitiveabilityagainstweedsmayincludeagrowthhabit

thatallowsearlyattainmentoffullcanopycoveranditspresenceforlongerperiods

(i.e.genotypeswithhighDP2andlatematurity(e.g.highAsum)(JoenjeandKropff,1987;

Lotzetal.,1991;Grevsen,2003;Williamsetal.,2008).Thiscanhelppotatogrowersto

choosethosegenotypeswithahightolerancetoweedpressureandabletomaintain

highyieldsinthepresenceofweeds,especiallyinorganicagriculture.Inthiscontext,

inclusionofmaturitytypebasedonphysiologicalcharactersinbreedingprogrammes

couldallow the selectionof competitivegenotypes in field conditionswhereweeds

arealwayspresentaswellasenhancetheefficacyofoverallweedcontrol.

Thephysiologicalmaturity type criteriawouldalsogivemore insight into the

physiologicalnatureofgenotype’smaturitytypeandthereforewouldhelpinproper

planning of haulm killing. This would give the benefits of increased tuber quality

(Haldersonetal.,1985)andeconomicalsavings(costofdesiccant).

The physiological maturity type criteria take into account the quantitative

knowledge of the influence of environmental factors on the length of the growing

Chapter4

136

season and on drymatter partitioning, necessary for designing genotypes for such

environments(HaverkortandKooman,1997).Ourapproachthereforewouldhelpin

designingstrategiesforbreedingforgenotypeswithspecificmaturitytype.

4.Conclusions

Inthischapterwetriedtocreateacleardefinitionofmaturitytypeandtodevelopa

quantitative,reproduciblemethodtoassessthematuritytypeinalargesetofpotato

genotypes.Wepresentedanapproachwherebypotatophysiologicaltraitswereused

to explain thematurity type in a setof varieties coveringawide rangeofmaturity

typesandawell‐adapteddiploidF1segregatingpopulation.Thisprocedureresulted

in four re‐defined criteria of defining maturity type which we call “physiological

maturitytypecriteria”asopposedtotheconventionallyusedcriterion.

The physiological maturity type criteria were based on physiological and

quantitativeknowledgeandwereeasilyinterpretableintermsofphysiologicaltrade‐

offsthatexistedbetweentheselectedtraits.Theconventionalmaturitytypecriterion

could not offer this advantage on account of its subjective and ordinal nature. Our

results showed that the conventional maturity criterion showed considerable

overlappingamongthematurityclasseswithmaximumvariability.Besides, itcould

notofferacleardefinitionofmaturityonaccountofitssubjectivenature.Incontrast,

physiologicalmaturity type criteria tended toexplainmaturity type inpotatomore

clearlyandwerebettercapableofpredicting thematurity typeofgenotypesacross

diverseenvironmentsthantheconventionalcriterion.

This study gavemore understanding about the nature of genotype’smaturity

type,besides,italsohighlightedthecomplexityofdefiningmaturitytypeinapotato.

Our results however, showed that re‐defining the maturity type of a large set of

genotypesonthebasisofphysiologicaltraitsispossibleandmayproveuseful.Anew

physiology‐based maturity definition could offer wider applications in potato

breeding and crop management studies. This new criterion may be amenable to

further genetic analysis and could also help in designing strategies for potato

ideotype breeding for genotypes with specific maturity type for specific growing

conditionsand/orenvironments.

CHAPTER5 AnecophysiologicalmodelanalysisofyielddifferenceswithinasetofcontrastingcultivarsandanF1segregatingpopulationofpotato(SolanumtuberosumL.)grownunderdiverseenvironments*

M.S.Khan,X.Yin,P.E.L.vanderPutten,P.C.Struik

AbstractWetestedthecapabilityofthegenericecophysiologicalmodel‘GECROS’toanalysedifferences in tuber yield of potato in a set of cultivars covering awide range ofmaturitytypesandadiploidF1populationsegregatingformaturitytype.Themodelpredicts crop growth and development as affected by genetic characteristics andclimatic and edaphic environmental variables. The genotype‐specific model‐inputparametervalueswereestimatedandthemodelwasusedforpredictingthetuberyield of an F1 population, their parents and a set of cultivars in multiple fieldexperiments.Themodelyieldedareasonablygoodpredictionofdifferencesintuberyield across environments and across genotypes. Trends of growth and nitrogenuptake were adequately reproduced by the model. Model analysis identified thegenotypickey‐parametersaffectingtuberyieldproductionandNmax(i.e.totalcropNuptake)contributedmosttothedeterminationoftuberyield.GenotypeswithhigherNmaxandlowertuberNconcentrationexhibitedhighertuberdrymatteryield.Theseresultswerelargelyconfirmedbystatisticalpathcoefficientanalysis.GECROSmodelisausefultoolinanalysingthecontributionofindividualphysiologicaltraitstoyieldacross different environments. Such information can greatly facilitate thedevelopmentofpotatoideotypesforspecificenvironments.Keywords:Potato(SolanumtuberosumL.);ecophysiologicalcropmodel;GECROS;genotype‐by‐environment (GE) interaction; complex traits; yield; sensitivityanalysis;pathcoefficientanalysis;ideotype;plantbreeding._______________________*Tobesubmitted.

Chapter5

138

1.Introduction

Potato (Solanumtuberosum L.) is one of themost important andwidely cultivated

non‐cereal crops, inmany parts of theworld (Walker et al., 1999;Hijmans, 2001).

Duetoincreasingfooddemandandchangingdietspotatoisbecomingasubsistence

crop in many growing areas. There is a need to increase potato yield via genetic

improvement and/or altered cropmanagement. In order to efficiently improve the

targettraits,predictionofthephenotypiccharacteristicsofgenotypesundervarious

environmentalconditionsiscrucial(AssengandTurner,2007).

Mostagronomictraitsaregeneticallycomplex(DaniellandDhingra,2002;Lark

et al., 1995; Orf et al., 1999; Stuber et al., 2003); they have low heritability, are

strongly dependent on environmental changes and often show high genotype‐by‐

environment(GE)interactions(AllardandBradshaw,1964;Tardieu,2003;Cooperet al., 2002, 2005).There is a need todissect complex traits like yield into simpler

characters and to separate factors influencing a givenphenotypic trait and shifting

from highly integrated traits to genotype‐specific traits (Yin et al., 2002).

Ecophysiologicalcropgrowthmodelshavethepotentialtoassessacomplextraitata

higher organizational level, via integrating the information about processes at the

lowerlevel.Theirabilitytoincorporateknowledgeofphysiologicaltraitstosimulate

crop growth and yield as influenced by growing environment, and agronomic

practicessuggeststhepossibilityofusingmodelsasacropbreedingtool(Booteand

Tollenaar, 1994; Aggarwal et al., 1997; Bastiaans et al., 1997; Boote et al., 1998;

Uehara,1998;White,1998;Booteetal.,2001;Mavromatisetal.,2001,2002;Hammer

etal.,2002,2006;Tardieu,2003;Banterngetal.,2004;Hoogenboometal.,2004;Yin

etal.,2004;AssengandTurner,2007;Herndletal.,2007;Letortetal.,2007).

Oneof themain applicationsof thesemodels is todescribe thedifferences in

yieldpotentialofgenotypesbetweenorwithinabreedingpopulationonthebasisof

individual physiological parameters. These parameters could be considered as

quantitativetraitsandareamenabletofurtheranalysis(Yinetal.,2004;Quilotetal.,

2005),e.g. forevaluatinganddesigning ideotypes(Loomisetal.,1979;Kropffetal.,

1995; Boote et al., 1996; Cilas et al., 2006; Yin et al., 2003c, 2004; Yin and Struik,

2008). This is possible because these parameters, often regarded as ‘genetic

coefficients’,arespecifictoeachgenotypeandsupposedtobeconstantunderawide

range of environmental conditions (Boote et al., 2001; Tardieu, 2003; Bannayan,

2007). This model feature makes it possible to make predictions about the plant

developmentalprocessesofgenotypeinawiderangeofenvironments(Stam,1996;

Bindraban,1997;Hoogenboometal.,1997).Suchmodelscanquantifycropgenotype‐

phenotyperelationships(Yinetal.,2000b,2004;Reymondetal.,2003;Hammeretal.,

2006) and therefore are highly suitable for studying the GE (Shorter et al., 1991;

Anecophysiologicalmodelanalysisofyielddifferencesinpotato

139

Hunt,1993;KropffandGoudriaan,1994;Hammeretal.,1996;Chapmanetal.,2002;

Yinetal.,2003c;Banterngetal.,2004,2006;Yinetal.,2004;Suriharnetal.,2007)

andcouldassistwithmulti‐locationevaluationofcropbreedinglines(Liuetal.,1989;

Aggarwaletal.,1995;Palanisamyetal.,1995;Piperetal.,1998;Booteetal.,2001;

Slafer, 2003; Banterng et al., 2004; Mayes et al., 2005). For potato, Kooman and

Spitters (1995) showed that simulation models can be useful for predicting tuber

yieldandgaininginsightintocropgrowthprocessesandcanhelptoexploreoptions

forcropimprovement.

However, traditionally, crop models have principally been used to study and

predictcropperformanceinresponsetoenvironmentalconditionsandmanagement

practices,whereasgenotypicimpactsoncropperformance(especiallyinthecontext

ofplantbreedingwherelargenumbersofgenotypesareinvolved)havereceivedless

attention.Thisispartlyduetotheconstraintsimposedbytime,resources,andlarge

numberofgenotypesthatmakesitdifficulttomeasuredetailedgrowthdynamicsto

fullyderivethegenotype‐specificmodelcoefficients(Anothaietal.,2008),andpartly

due to the restricted capabilities ofmodels to represent genetic differences (White

andHoogenboom,1996;Hoogenboometal.,1997).

To be useful, the physiological frameworks used for trait dissection and

modelling atwhole‐crop levelmust realistically capture the functional basis of the

geneticvariationforcomplextraitsofinterest(Yinetal.,2000b).Mostexistingcrop

models,whichwereconstructed todealmostlywithagronomic issues,arenotwell

structuredinthisregardforinstanceforcaptureanduseofnitrogen(N)(Jeuffroyet

al.,2002)andforcarbon(C)andNpartitioning(Dingkuhn,1996).Theyalsolackthe

ability to describe subtle complexities associated with the differences between

genotypes(WhiteandHoogenboom,1996;Yinetal.,2004).

Inthisrespect,cropphysiologicalmodesofactionofthecomplextraitmustbe

understood and quantified and the crop model must be sufficiently detailed to

simulate the consequenceson growth anddevelopment generatedby the genotype

(G),environment(E)andultimatelyGE(Hammeretal.,1996).Tobecomeeffectivetools for addressing GE, existing models have to be improved, both in terms ofmodelstructureandinputparameters(Yinetal.,2004;YinandStruik,2008).Studies

are thus needed to develop and/or test the potential capabilities of current crop

models for plant breeding and to specify the necessary modifications for such

applications.

The objectives of this study are (i) to examine the ability of a recent

ecophysiologicalcropgrowthmodel‘GECROS’toexplainyielddifferencesamong100

genotypes from an F1 segregating population, their parents and a set of standard

cultivars of potato, and (ii) to analyse the relative importance of individual

Chapter5

140

physiological traits in determining yield differences. These analyses could have a

strongimplicationinexploringtheextenttowhichourmodeldissectstheroleofG,E,

andGEon tuber yield production, and indesigning strategies for potato ideotypebreedingforspecificenvironments.

2.Materialsandmethods

2.1.TheGECROSmodel

The model GECROS (Genotype‐by‐Environment interaction on CROp growth

Simulator) is a generic ecophysiological model that predicts crop growth and

development as affected by genetic characteristics and climatic and edaphic

environmental variables. For a detailed description of themodel, see Yin and Van

Laar(2005).Here,onlythekeyprocessesmodelled inGECROS(version2.0asused

byYinandStruik,2010)aresummarised.

Coupled modelling of CO2 diffusional (stomatal and mesophyll) conductance,

leaf photosynthesis and transpiration in dependence of leaf nitrogen content is

implemented according to Yin and Struik (2009). Subsequently, the results are

integrated for thewhole canopy,basedon the sun/shadeapproachofDePuryand

Farquhar(1997).TheGaussianintegrationisusedtoextendtheinstantaneousrates

intothedailytotalphotosynthesisandtranspiration.

Daily crop respiration is simulated based on the theoretical framework of

respiration components: (i) growth respiration, (ii) respiration for ammonium and

nitrateuptakeandnitratereduction;uptakeofother ions,phloem loading,and(iii)

residualmaintenancerespiration(CannellandThornley,2000).

Nitrogen demand ismodelled as themaximum value of the deficiency‐driven

and the growth activity‐driven demand. The first guarantees the actual plant N

concentration being above a critical value; the second is modelled based on the

optimum N‐C ratio for maximising the relative carbon gain, using the equation of

Hilbert (1990).TheactualNuptake is limitedby themaximumdailyNsupply rate

fromthesoil.

PartitioningofthenewlyproducedCandabsorbedNismodelledintwosteps:

first,betweentherootandtheshoot,andthenamongorganswithintheshoot.Root‐

shootpartitioning forC andN responds to environmental conditions, basedon the

root‐shoot functionalbalancetheory(e.g.Charles‐Edwards,1976)adjustedinorder

tomaximizerelativegrowthrate.

Intra‐shoot carbon partitioning to the structural stems and to the tubers are

Anecophysiologicalmodelanalysisofyielddifferencesinpotato

141

determinedaccording to theirexpecteddailycarbondemands,whicharedescribed

by the differential form of a sigmoid function for asymmetric determinate growth

(Yinetal.,2003a). In themodel, thepotential tubersinksize iscalculated fromthe

estimatedamountofNavailablefortubergrowth.Theremainingshoot‐carbongoes

eithertotheleavesortotheCreservepool,dependingonwhetherthegreen‐surface

areaindexbecomesC‐orN‐limited.

The intra‐shoot N partitioning is based on pre‐defined maximum tuber N

concentration(nSO,whichisagenotype‐specificparameter,Table5.1)andminimum

N concentration in the stems. If the N requirements for the tubers and structural

stemsaremetfromthecurrentNuptake,theremainingshootNgoestotheleaves,

which include the photosynthetically active green surfaces of other organs. If the

requirementsforthetubersarenotmet,remobilisationofNfirst fromthereserves

and then from the leaves and the roots takesplace, until the reserves aredepleted

and theNconcentrations in the leavesand roots reach theirminimumvalues.This

remobilisationresults in leafandrootsenescence,reflectingthephenomenonofso‐

called ‘self‐destruction’ (Sinclair andDeWit,1975). If the tuberN requirementsare

notmetbyshootNandremobilization,thetuberNconcentrationdeclines.

Green‐surfaceareaindexismodelledastheminimumvalueofC‐limitedandN‐

limited areas. The underlying principle is described in detail by Yin et al. (2000c,

2003b).Assuch,senescenceofleavesiscomputedasafunctionofNreductioninthe

canopy.Similarly,senescenceofrootsdependsonNreductionintheroots.

Dailyphenologicaldevelopmentrate(i),whichhasaunitofd‐1, iscalculated

both for the pre‐tuberisation period and for the tuber bulking period, respectively,

using genotype‐specific parameters mV and mR, representing the pre‐tuberisation

phase and the duration between onset of tuber bulking and crop maturity,

respectively(Table5.1).Phenologicalresponsetotemperatureisbasedonaflexible

bell‐shapednon‐linearbetafunction(Yinetal.,1995).Thisfunctionisimplemented

onanhourlybasis toaccount for thediurnal temperature fluctuation.Valuesof the

cardinaltemperaturesinthisfunctionweretakenfromChapter2forpotato.

Inthemodel,plantheightisrequiredmainlyforcalculatingthecarbondemand

forstructuralstems.Plantheight ismodelled to followasigmoidpattern,assuming

thatmaximumplantheight(Hmax,Table5.1) isreachedatamidwaybetweentuber

bulking and cropmaturity. The increment of plant height is decreased if available

carbonassimilatesdonotmeetthepotentialcarbondemandforstemgrowth.

GECROSrunswithatimestepofoneday.Requireddailyweatherdatainclude

minimumandmaximumtemperature,globalsolarradiation,vapourpressure,wind

speed,andrainfall.AirCO2concentrationshouldbeprovided.Otherrequiredinputs

aredailysupplyofwaterandNavailableforcropuptake.

Chapter5

142

2.2.Experimentaldata

Datasets for our model analysis came from six field experiments, performed in

Wageningen(52˚Nlatitude),theNetherlands,withtwoexperimentsineachyear,i.e.

2002(Exps1and2),2004(Exps3and4),and2005(Exps5and6)(Fordetailssee

Chapters2and3).Theseexperimentsdifferedinenvironmentalconditionsbecause

they were carried out in different years, on different soils and under different

nitrogen (N) fertiliser regimes, thereby creating six contrasting environments. The

plantmaterialfortheseexperimentsconsistedof100F1diploid(2n=2x=24)potato

genotypes derived from a cross between two diploid heterozygous potato clones,

SH83‐92‐488 RH89‐039‐16, hereafter referred to as the ‘SH RH population’(RouppevanderVoortetal.,1997;VanOsetal.,2006).Thispopulationsegregates

formaturitytype(VanderWaletal.,1978;VanOijen,1991).Besides, fivestandard

cultivars (viz.Première,Bintje, Seresta,AstarteandKarnico)werealso includedon

thebasisoftheirmaturitytypeunderDutchconditions.

To evaluate model performance in representing growth dynamics, field

measurementswereperformedonfewgenotypesduringyear2004(Exps3and4),

and2005(Exps5and6).Atotaloffiveharvestswerecarriedoutatintervalsof1‐2

weeksduringthecropperiod.Atleastfourrepresentativeplantspersamplingpoint

were takenandmeasurementsweremadeonbothabove‐ andbelow‐groundplant

parts for dry matter and N content in different plant parts. To determine the dry

weight, the plants were divided into their constituents of living and dead leaves,

stems,stolon,andtubers,andsub‐samplesofeachportionweredriedat70 ˚C toa

constant weight. Roots were not measured since they are difficult to harvest

(Kooman, 1995). Further sub‐sampleswereused to determinenitrogen content by

meansofmicro‐Kjeldahldigestionanddistillation.

Tomeasurethemodelpredictabilityforthefullsetof100genotypes,tuberdry

matterwasmeasuredatonlythreeharvestsduringthegrowingperiodandtuberN

content was measured at maturity (see Materials and Methods, Chapter 3). These

datasetscreatedthebasisfortheestimationofgenotype‐specificparametersandfor

model testing as described later. In addition, green canopy cover (%)was visually

assessedonweeklybasisduringthewholecropcycleforallgenotypes(Chapter2).

2.3.Estimationofmodelparameters

Table 5.1 gives the description of genotype‐specific parameters of GECROS, which

include total crop N uptake at maturity (Nmax). Nmax per se, as an accumulative

quantityinthecroplifecycle,shouldnothavebeenconsideredasaninputparameter.

However, therewas not sufficient information about the soil andmodelling of soil

Anecophysiologicalmodelanalysisofyielddifferencesinpotato

143

Table5.1.Listofgenotype‐specificparametersofGECROSmodel(SeeMaterialsand

Methods,section2.3).tdstandsforthermalday.

Trait Description Unit

mV Periodfromplantemergencetoonsetoftuberbulking td

mR Durationbetweenonsetoftuberbulkingandcrop td

Hmax Maximumplantheight m

nSO MaximumtuberNconcentration gNg‐1DM

Nmax TotalcropNuptakeatcropmaturity gNm‐2

processesisusuallyfullofuncertainties.Toreduceaninfluenceoflargeuncertainties

inpredictingedaphicvariablesforNsupply,wetookasimpleapproach,usingNmaxas

a model‐input parameter. The value of Nmax was estimated on the basis of the

assertion that the distribution of N over tubers and other components is probably

fairlyconservative(BiemondandVos,1992),assumingthatNaccumulationintubers

andhaulmaccountsfor77.5%and15%(ownmeasureddataonfewgenotypes),and

N accumulation in the roots for 7.5% (see data of Sharifi et al., 2005), of Nmax.

Procedures for estimation and correction for experiment‐specific parameter values

aredescribedbelow.

Of the five parameters, values of phenological parameters mV and mR were

derived from Chapters 2 and 3, where using the afore‐mentioned data sets, we

analysed the dynamics of potato canopy cover development and tuber bulking,

respectivelyasa functionof thermal time. Chapter2describes the threephasesof

canopycoverdevelopment(i.e.build‐upphase,maximumcoverphase,declinephase),

usingbiologicallymeaningfulcomponenttraits.Theendofthecanopybuild‐upphase

ismarkedbyt1atwhichthecanopycoverattainsitsmaximumlevel(seeFig.2.1in

Chapter 2). The end of canopy senescence (or crop maturity) is marked by te. In

Chapter3,wequantified the tubergrowthduringexponentialand linearphases, in

which tBcharacterises the onset of linear phase of tuber bulking (see Fig. 3.1 in

Chapter3).

Forfewgenotypeswithmoredataforgrowthdynamics,availabilityofsufficient

datapointsallowedtheestimationofmVastB.However,forothergenotypes,thepre‐

tuberisation phasemVis hard to estimate accurately. A sensitivity analysis showed

that this trait mattered little formodel prediction of tuber yield (see Results), we

assumedthatmV=t1.Becauset1issensitivetotheenvironmentalvariation(seeFig.

2.7),andExp6showedcomparativelylowervariationfort1(Chapter2),valuesoft1

Chapter5

144

basedonExp6wereusedforgenotype‐specificmV.ParametermRwascalculatedas

(te‐mv).Hereagain, inorder to reduce the randomnoise,mean teperF1genotypes

acrosssixexperimentswasusedforestimatingmR.

Notsurprisingly,valuesofplantheight,tuberNconcentration,andtotalcropN

uptake differed among the experiments (Fig. 5.1). To reduce the influence of

environmentalnoiseyetreflectthedifferencesamongtheexperiments,Hmax,nSO,and

Nmax were estimated as across‐experiment mean times a correction factor per

individual experiment. The correction factor was calculated as a slope of the

regression line between experiment‐specific parameter values versus its across‐

experiment mean. In the case of Exps 1 and 2, for which we lacked the data for

parameters nSO, Hmax and Nmax (Fig. 5.1), combined information from other

experimentsfortheseparameterswasused,withthecorrectionfactorobtainedfrom

theslopeoftheregressionbetweenobservedtuberyieldperExps1or2versusmean

observedtuberyieldacrosstheotherfourexperiments.

Cropswerenotsubjecttoanywaterstressduringthegrowth;so,ahighvalueof

dailywater supply from soil was given for simulation tomimic no drought stress.

DailyN supply from soilwas derived fromNmax, based on the fitted pattern of the

dynamicsoftheaccumulativecropNuptakeasobservedforthefewgenotypeswith

thedetailedmeasurements.

2.4.Modelanalysistoidentifymajoryield‐determiningtraits

Sensitivity analysis was performed to determine the importance of individual

genotype‐specific parameters in determining tuber yield potential. Data sets from

Exps3‐6wereusedforthispurpose.WefollowedtheapproachofYinetal.(2000b,

2005). First, the baseline simulation was conducted, where all genotype‐specific

parameter valueswere used as input for simulation. Then, parameterswere fixed,

oneatatime,attheiracross‐genotypemean.Theextent,towhichthepercentageof

varianceexplainedbythemodelwasdecreasedrelativetothepercentageexplained

by the baseline simulation, was used to rank the relative importance of the

parametersindeterminingtuberyields.

2.5.Statisticalpathcoefficientanalysis

Toconfirmtheresultofecophysiologicalmodelinganalysis,pathcoefficientanalysis

wasperformed,alsotocheckouttheimportantdirectandindirectinter‐associations

between the genotypic‐specific parameters and final tuber yield. This analysiswas

Anecophysiologicalmodelanalysisofyielddifferencesinpotato

145

Figure5.1.BoxplotsofmeansofanF1populationforthreemeasuredparametersin

fourexperiments.Theboxesspanthe interquartilerangeofthetraitvalues,sothat

themiddle50%of thedata laywithin thebox,withahorizontal line indicating the

median. Whiskers extend beyond the ends of the box as far as the minimum and

maximumvalues.PointsweremissingforExps1and2becausethesetraitswerenot

measuredtherein(seethetext).

0.0

5.0

10.0

15.0

20.0

25.0

30.0

35.0

3 4 5 6

0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.0

0.5

1.0

1.5

2.0

2.5

[m]

Plant N uptake

Plant height

[g N m-2]

Experiment no.

[g N g-1]

Tuber N conc.

Chapter5

146

performed on the mean of F1 genotypes across environments following the

procedureofChapter3.

3.Resultsanddiscussion

3.1.Modelparameters

The evaluation of genotype‐specific model parameters indicated that there were

strong genetic differences among the standard cultivars (Table 5.2). As expected,

valuesofmostoftheseparametersexceptnSOwerehigherforlatematuringcultivars

(Astarte and Karnico) than for mid‐late (Seresta), mid‐early (Bintje), and early

cultivars (Première). In contrast, values of nSO were higher for early maturing

cultivarslikePremièrefollowedbymid‐lateandlatecultivars(Table5.2).Ithasbeen

found in previous research that N uptake and use efficiency differ by cultivar

(Johnsonetal.,1995;Kleinkopfetal.,1981;Lauer,1986;PorterandSisson,1991a,b).

Early‐maturing cultivars stop leaf growth earlier than late‐maturing cultivars and

therefore exhibit less vegetative growth and a shorter tuber bulking period. This

pattern results in lower N use efficiencies and higher tuber N concentrations in

earliergenotypes(Vos,1997;Chapter3).

The results for theF1population indicated thatmost of theparameterswere

nearlynormallydistributed,apart frommRandHmax forwhich thedistributionwas

bimodal(Fig.5.2).TheF1populationdisplayedtransgressivesegregation,withmore

extremevaluesinthepopulationthantheirparents(SHandRH).

Table5.2.Estimatedmeanvaluesofgenotype‐specificparametersforfivestandard

cultivars(listedinorderofincreasinglylongercropcycle).tdstandsforthermalday.

(−)representslackofdata.SeeTable5.1forparameterdescriptions.

Cultivar mV

(td)

mR

(td)

Hmax

(m)

nSO

(gNg‐1DM)

Nmax

(gNm‐2)

Première 13.3 36.7 0.53 0.018 21.9

Bintje 14.4 51.1 0.82 0.017 25.8

Seresta 17.7 57.4 1.05 0.014 25.4

Astarte 15.1 65.1 1.14 0.013 24.6

Karnico − − 1.39 0.012 25.8

Anecophysiologicalmodelanalysisofyielddifferencesinpotato

147

Figure5.2.Distributionofgenotype‐specificparametersofGECROSmodelfor100F1

genotypes.The valuesof twoparents ‘SH’ and ‘RH’ are indicatedby full arrowand

dashedarrow,respectively.SeeTable5.1forparameterdescriptions.

mV

mR Hmax

Nmax n SO

Num

ber

of F

1 ge

noty

pes

Chapter5

148

3.2.Modelperformanceinrepresentinggrowthdynamics

The GECROS model efficiently simulated the dynamics of important potato

physiologicalgrowthprocessesunderarangeofenvironmentalconditions.Herewe

presenttheresultsoftwoexperimentsof2004(i.e.Exps3and4).Theseexperiments

were low and high in N availability, respectively (Fig. 5.1; Chapter 2). Figure 5.3

presentsmodelledandobserveddataofplantandtuberNuptakeanddrymatterfor

the two parental genotypes. Model evaluation showed good agreements between

observedandsimulatedgrowthpatternoftheseprocesses.

TheNaccumulationintubersshowedanS‐shapedgrowthpattern, incontrast

to the pattern of plant N uptake (Fig. 5.3). The results further showed that at the

beginning of the season the tuber drymatterwas underestimatedwhile plant dry

matterwas estimated ratherwell, whereas at the end of the season the simulated

tuberdrymatteryieldscomparedwellwiththemeasureddata.Thecurrentdata(Fig.

5.3) also indicated that the tuber aswell asplantdrymatter is reducedwhenN is

limiting (i.e. Exp 3), most likely via limited leaf expansion and interception of

radiation(VosandBiemond,1992).

3.3.Modelperformanceinexplainingyielddifferencesamonggenotypes

To evaluate the model’s ability to reproduce the observed yield‐based variation,

differences in the actual and modelled yield were compared across different

environments for the F1 population, their parents and a set of standard cultivars.

Overall, the model adequately simulated differences in tuber yields among these

genotypes inallexperiments(Fig.5.4).However, therewerediscrepanciesbetween

simulatedandmeasuredtuberyields,especiallyinExp3,wherethepredictedtuber

yield was under‐estimated in high‐yielding, and over‐predicted in low‐yielding

genotypes.WiththeconceptusedinGECROSfortuberNaccumulation,thefractionof

tuber N is a very sensitive input parameter, not only affecting the final N

concentration,butalsotheharvest indexandthepotentialtuberyield(YinandVan

Laar,2005).

Themodelexplainedvariationsinthetuberyield,withR2rangingfrom35%

to 65% for the F1 population, and 80% to 96% for the parents and five standard

cultivars,amongtheExps3‐6(Fig.5.4).However,incaseofExps1and2,themodel

could not satisfactorily explain the variation in the tuber yield (Fig. 5.4). This is

because there was a lack of information about N availability and therefore

assumptions had to be made for defining the related input parameters for those

experiments(seeMaterialsandMethods).

Anecophysiologicalmodelanalysisofyielddifferencesinpotato

149

Figure5.3.Modelled (lines) and observed (symbols) for genotypes (a) SHRH‐469

and(b)RH89‐039‐16, inExperiment3,and(c)SHRH‐469and(d)RH89‐039‐16, in

Experiment4,performedduring2004.

(a)

(b)

(c)

(d)

Thermal days (td)

0

2

4

6

8

10

12Plant Tuber

0

500

1000

1500Plant Tuber

0

5

10

15

20Plant Tuber

0

200

400

600

800

1000

1200Plant Tuber

0

5

10

15

20

25

30Plant Tuber

0

500

1000

1500Plant Tuber

0

10

20

30

40

0 20 40 60

Plant Tuber

0

500

1000

1500

2000

0 20 40 60

Plant Tuber

N u

ptak

e (g

m-2

)

Dry

mat

ter

(g m

-2)

Chapter5

150

Figure5.4.Comparisonbetweenpredictedandobservedtuberyieldfor(a)100F1

genotypes(smallcircles),(b)5standardcultivarsand2parentalgenotypes(SHand

RH), testedper each individual experiment (large circles).R2 valueswere obtained

from a linear regression; rRMSE is the relativeRMSE,calculated as the root‐mean‐

squareerrordividedbythemeanobservedtuberyield.

Observed yield (g m-2)

Pre

dict

ed y

ield

(g

m-2

)

0

500

1000

1500

2000

2500

Exp 1

(a) R2 = 0.02, rRMSE= 0.36(b) R2= 0.07, rRMSE= 0.37

0

500

1000

1500

2000

2500

Exp 2

(a) R2 = 0.08, rRMSE= 0.42(b) R2= 0.05, rRMSE= 0.46

0

500

1000

1500

2000

2500(a) R2 = 0.40, rRMSE= 0.23(b) R2= 0.88, rRMSE= 0.21

Exp 30

500

1000

1500

2000

2500(a) R2 = 0.50, rRMSE= 0.20(b) R2= 0.84, rRMSE= 0.09

Exp 4

0

500

1000

1500

2000

2500

0 1000 2000

(a) R2 = 0.35, rRMSE= 0.20(b) R2= 0.96, rRMSE= 0.09

Exp 50

500

1000

1500

2000

2500

0 1000 2000

(a) R2 = 0.652, rRMSE= 0.10(b) R2= 0.813, rRMSE= 0.11

Exp 6

Anecophysiologicalmodelanalysisofyielddifferencesinpotato

151

Themodelalsopredictedwell(i.e.R2=86%)themeantuberyieldacrossfour

environments (i.e. Exps 3‐6) (Fig. 5.5a). Furthermore, the model satisfactorily

predicted variation in the mean tuber yield across the F1 genotypes among the

environments (Fig.5.5b).Themodelpredicted57%of the totalobservedvariation,

when all observations for the F1 genotypes in the four environmentswere pooled

(Fig.5.5c).

The model performance was also evaluated by calculating the relative root

meansquareerror(rRMSE;Figs5.4‐5.5).TherRMSEvaluesrangedfrom0.10to0.42

in F1 population and 0.09 to 0.46 for the two parents and five standard cultivars,

amongtheExps3‐6(Fig.5.4).Asexpected,therRMSEvalueswerehigherforthefirst

twoexperiments(Exps1and2)mainlyduetolackofgoodfitasdescribedearlier.In

otherexperiments,rRMSEvalueswerereasonablygood.Theresultsfurtherindicated

lowrRMSEvaluesforthemeantuberyieldacross‐environments(i.e.0.09),acrossthe

F1genotypesamongtheenvironments(i.e.0.06),andwhenallobservationsforthe

F1genotypesinthefourenvironmentswerepooled(i.e.0.17)(Fig.5.5).

Overall, our results indicated that using as few as five measured genotype‐

specific parameters (Table 5.1), the GECROS model showed a good potential in

explainingtheobservedyielddifferencesamonggenotypes.

3.4.Model‐basedsensitivityanalysistoidentifyimportantyield‐definingtraits

Ecophysiological models can serve as an important tool to determinemodel input

parameters critical for yield potential (Yin et al., 2000b). Sensitivity analysis was

performedtoevaluatethe importanceof individualgenotype‐specificparameters in

determining tuber yield potential. The results of the baseline simulation, in which

genotype‐specificvalueswereapplied forallmodel‐inputparameters,werealready

presented in the previous section for individual experiments (Fig. 5.4). When the

parameters were fixed, one at a time at their across‐genotype mean, the model

explainedpercentagevariancedroppedfromthebaselinesimulationformostofthe

parameters throughout the environments (Table 5.3). This reduction in variance

explainedwasmoreevidentforNmaxfollowedbynSOthroughouttheexperiments.For

other parameters, in some cases, the percentage variance accounted for was even

higher than that obtained from the baseline simulation. Thiswas evident formV in

Exps3,4,and5andHmax inExps3and6.Almostsimilar trendswererecorded for

observationsinacross‐environmentmeansaswellaswhendatawerepooled.These

results conclude that the relative importance of parameters for tuber yield

determinationcanberankedas:Nmax>nSO>mR>mVandHmax.Obviously,Nmaxand

nSO were the most important parameters as they contributed most to the

Chapter5

152

Figure5.5.Comparisonbetweenpredictedandobservedtuberyield(gm‐2)testedin

across‐environment mean of 100 F1 genotypes (n = 100 points) (a), in across‐

genotypemean of four environments (i.e. Exps 3‐6) (b), and when data from 100

genotypes tested in the four experiments are pooled (n= 400) (c).R2 valueswere

obtainedfromalinearregression;rRMSEistherelativeRMSE,calculatedastheroot‐

mean‐squareerrordividedbythemeanobservedtuberyield.

Observed yield (g m-2)

Pre

dict

ed y

ield

(g

m-2

)

0

500

1000

1500

2000

2500

0 1000 2000

a

R2=0.86, rRMSE= 0.09

0

500

1000

1500

2000

2500

0 1000 2000

b

R2=0.89, rRMSE= 0.06

0

500

1000

1500

2000

2500

0 1000 2000

c

R2=0.57, rRMSE= 0.17

Anecophysiologicalmodelanalysisofyielddifferencesinpotato

153

Table5.3.Percentageoftheobservedtuberyieldvariationinfourexperiments(Exps3‐6)ofF1populationof

potatoaccountedforbytheGECROSmodelwhendifferentsetsofvaluesforfivegenotype‐specificparam

eterswere

used.SeeTable5.1forparameterdescriptions.

Set

of

para

met

ers

a

R2b

R

2c

R2d

mV

mR

Hm

ax

n SO

Nm

ax

E

xp. 3

E

xp. 4

E

xp. 5

E

xp. 6

+

+

+

+

+

0.

40

0.50

0.

35

0.65

0.

86

0.57

- +

+

+

+

0.44

0.

55

0.42

0.

59

0.90

0.

58

+

- +

+

+

0.35

0.

50

0.32

0.

56

0.81

0.

54

+

+

- +

+

0.48

0.

47

0.32

0.

68

0.90

0.

58

+

+

+

- +

0.21

0.

47

0.27

0.

52

0.62

0.

50

+

+

+

+

-

0.06

0.

0001

0.

02

0.24

0.

12

0.29

a The

‘+

’ sy

mbo

l in

dica

tes

the

para

met

er f

or w

hich

the

gen

otyp

e-sp

ecif

ic v

alue

was

use

d, a

nd ‘

-’ s

ymbo

l in

dica

tes

the

para

met

er f

or w

hich

the

gen

otyp

e m

ean

valu

e w

as u

sed

in t

he m

odel

sim

ulat

ion.

The

fir

st r

ow (

bold

sym

bols

and

tex

t)

indi

cate

s th

e re

sult

s of

the

base

line

sim

ulat

ion,

in w

hich

gen

otyp

e-sp

ecif

ic v

alue

s w

ere

appl

ied

to a

ll p

aram

eter

s.

b The

var

iati

on e

xpla

ined

per

eac

h ex

peri

men

t (nu

mbe

r of

poi

nts

n =

100

).

c The

var

iati

on e

xpla

ined

for

acr

oss-

envi

ronm

ent m

eans

in 1

00 F

1 ge

noty

pes

of p

otat

o (n

umbe

r of

poi

nts

n =

100

).

d The

var

iati

on ex

plai

ned

for

all

data

poi

nts

whe

n al

l th

e ob

serv

atio

ns f

or 1

00 F

1 ge

noty

pes

in t

he f

our

expe

rim

ents

wer

e

pool

ed (

n =

400

).

Chapter5

154

determinationoftuberyieldamongtheF1population(Table5.3).Lietal.(2006)also

performedasimilarsensitivityanalysisandreportedthatNavailabilitywasthekey

traitaffectingpotatotuberyield.CropNstatusaffectsphotosynthesiswhichinturnis

a functionof intercepted radiationby above‐groundvegetativeparts (Khurana and

McLaren,1982;Moll,1983;Marcelisetal.,1998).Thereisalinearrelationbetween

totalNuptakeandcanopygrowthanddevelopmentuntilamaximumlevelisreached

(Lemaire,1997).Thedurationoftheperiodwithfullcanopycoverislongerandthe

stageofdeclinestartslaterintimethehighertheNavailable(Chapter2).

The sensitivity analysis gave further insight into whether or not the relative

importanceofthefivemodel‐inputtraitsvarieswithenvironment.Suchinformation

about the relevance of different plant traits for yield under different growth

conditionscan,therefore,improvetheefficiencyofabreedingprogramme(Heuvelink

etal.,2007).However,ourresultsforexperiment‐specificsensitivityanalysisshowed

thattherelativeimportanceofthefivemodel‐inputtraitsvariedlittleamongthefour

experiments.

3.5.Pathcoefficientanalysis

Pathcoefficientanalysiswasperformed tounderstand theunderlying relationships

among the genotype‐specific parameters and with observed tuber yield (Fig. 5.6;

Table5.4).TheresultsrevealedthatparametersNmaxandnSOhadhighestsignificant

(P<0.01)positiveandnegativedirecteffects(i.e.0.91and‐0.47,respectively)onthe

finaltuberyield(Fig.5.6),whereasmV,mR,andHmaxhadleast(P>0.05)directeffects

onfinaltuberyield(i.e.‐0.05,‐0.10,0.00,respectively).Thesumofdirectandindirect

effects of parameters from path coefficient analysis revealed that there existed

significant (P < 0.01) positive effects of mR, Hmax, and Nmax on tuber yield (their

correlationcoefficientsas0.45,0.41,and0.81,respectively;Table5.4).

Path coefficient analysis has been very useful in partitioning the correlation

coefficient into components of direct and indirect effects and give thorough

understandingof actual causal relationships and contributionof various characters

on the target trait (Dewey and Lu, 1959; Samonte, 1998). Our results further

indicated thatrelationshipsofparameters(mR,Hmax)with tuberyieldwereactually

drivenbyindirecteffects,mainlythroughNmax(Table5.4).Theseresultsarelargelyin

line with our GECROS‐based sensitivity analysis, suggesting that genotypes with

higherNmaxmayexhibitmorefinaltuberyieldprovidednSOislow(Casaetal.,2005;

Chapter3).

Anecophysiologicalmodelanalysisofyielddifferencesinpotato

155

Figure5.6.Pathcoefficientstructuralmodeldescribingdirectandindirecteffectsof

differentgenotype‐specificparameters(mR,mV,Hmax,nSO,andNmax;seeTable5.1for

theirdescriptions)onthetuberyieldacrosssixexperimentsforanF1populationof

potato. The solid line represents the correlation coefficient between two predictor

variables;dashed linerepresents thepathcoefficient fromthepredictorvariable to

responsevariable(tuberyield).**SignificantatP<0.01,NSNon‐significant(P>0.05).

4.Conclusions

Yield variation in terms of growth and development of the crop is complex, for it

involves the effect of external factors on all the physiological processes, the inter‐

relationships between different processes and their dependence on the genetic

constituentoftheplant.Inthisstudyweexploredtheabilityoftheecophysiological

model‘GECROS’toanalysedifferencesintuberyieldinasetofcultivarsofmaturity

types and a diploid F1 segregating population. The model described well the

dynamics of important growth related processes in potato and gave important

insights into theunderlying component traits and factors influencing the tuberdry

matterproduction.Usingfivemeasuredgenotype‐specificparameters(mV,mR,Hmax,

Chapter5

156

Table5.4.Pathcoefficientanalysisofdirectandindirecteffectsofgenotype‐specific

parametersonthetuberyieldofanF1populationofpotato.tdstandsforthermalday.

SeeTable5.1forparameterdescriptions.

Variable Effect†mV(td) Directeffect ‐0.052Indirecteffectvia

mR 0.008Hmax 0.000nSO ‐0.028Nmax ‐0.090

Totalcorrelation ‐0.161mR(td) Directeffect ‐0.098Indirecteffectvia

mV 0.004Hmax 0.000nSO 0.226Nmax 0.318

Totalcorrelation 0.451Hmax(m) Directeffect 0.000Indirecteffectvia

mV 0.002mR ‐0.089nSO 0.213Nmax 0.287

Totalcorrelation 0.413nSO(gNg‐1) Directeffect ‐0.473Indirecteffectvia

mV ‐0.003mR 0.047Hmax 0.000Nmax 0.131

Totalcorrelation ‐0.298Nmax(gNm‐2) Directeffect 0.908Indirecteffectvia

mV 0.005mR ‐0.034Hmax 0.000nSO ‐0.068

Totalcorrelation 0.811†Acrossexperiments

Anecophysiologicalmodelanalysisofyielddifferencesinpotato

157

nSO, and Nmax), GECROS showed a good potential in explaining the observed yield

differencesamongthegenotypes.Simulatedtrendsofgrowthwereinagreementwith

themeasureddata.Themodelyieldedreasonablepredictionsofdifferencesintuber

yield across environments and across genotypes. The simulation results suggested

thatgenotype‐specificparameterNmaxcontributedmosttothedeterminationoftuber

yield.Theseresultswerefurtherprovedbypathcoefficientanalysissuggestingthat

genotypes with higher Nmaxand low nSO may exhibit more final tuber yield. Such

information would greatly facilitate the development of ideotypes for specific

environment(s).

AsawholethisstudyhighlightedtheusefulnessofGECROScropgrowthmodel

for exploring the impact of new genotypes and the contribution of individual

physiologicaltraitsonyieldbysimulatingtheresponsesofgenotypesacrossdifferent

environments.Inaddition,ourmodelapproachshedlightonsomeofthemechanisms

causing genotype differences in tuber yield and provided support for important

aspects of hypothesis regarding N accumulation. The model analysis identified N

uptake and use efficiency as possible areas for improvements in tuber drymatter

production. Opportunities for improving the modelling of and understanding of N

dynamics and source‐sink dynamics in potato have become apparent during the

study. Further analysis of the genotypic parameters should be performed in

conjunctionwithmolecularmarkersinordertodeterminetheirgeneticcontrol.Such

informationwould guide research towards key processes affecting crop yields and

mayhelporientatebreedingprogrammesassuggested,e.g.byYinetal.(2004);Yin

andStruik(2008).

 

CHAPTER6

Generaldiscussion Breedingforhigh‐yieldinggenotypes:AchallengeBreeding for high‐yielding cultivars for specific and/or varied environments is a

majorchallengetoachieveworldfoodsecurityinthenewcentury.Thepaceatwhich

progress has been made in achieving higher crop yields has become stagnant

comparedwiththatrequiredbythegrowingdemands(Cassman,1999).Thepaceand

efficiencyof geneticprogressmustnow increase tomeet theprojecteddemand for

agriculturalproducts(FAO,2009).

Molecularbiologyhasdeliveredcommercial successes inenablingcropplants

to better resist pests and tolerate herbicides and, more recently, in improving

product quality.However, the challenge remains as to how to manipulate more

complex growth anddevelopment traits associatedwith crop adaptation and yield.

The inability to connect information at gene level to the expressed phenotype in a

manner that is useful for selection and plant breeding has restricted adoption of

molecular approaches in plant breeding (Miflin, 2000). Enhanced capabilities in

genotyping have not been matched by development of enhanced capabilities in

phenotyping or by development of enhanced approaches to link genotype and

phenotype(Camposetal.,2004).Thesituationisparticularlychallengingforcomplex

traits associated with crop adaptation and yield (e.g. grain or tuber drymatter or

resource use efficiencies) as they are regulated by multiple interacting genes and

environmental conditions, so that gene‐to‐phenotype relationships are not

straightforward (Liet al., 2001; Luoet al., 2001).Hence, new approaches must be

developed to accelerate breeding through improving genotyping and phenotyping

methodsandbyincreasingtheavailablegeneticdiversityinbreedinggermplasm.

Potato:AcomplexcropInpotato,breedingforyieldpotentialandrangeofadaptation(ordegreeofstability

inperformanceoverawiderangeofenvironments)isparticularlydifficultduetothe

complexity of the crop’s whole‐plant physiology (cf. Van Dam et al., 1996; Celis‐

Gamboaetal.,2003).Growthandyieldaredeterminedbyadiversityofcomplexcrop

factorsincluding(a)therateofleafappearance,individualleafgrowth,finalleafsize,

and the life span of individual leaves, (b) the number of lower and sympodial

branches, (c) the overall rate of canopy development (e.g. increasing of N supply

Chapter6

160

levels accelerates crop development and the timewhenmaximum canopy cover is

reached),(d)lightinterceptionbythecropovertime,(e)therateofphotosynthesis,

(f)theonsetoftuberisation,and(g)finaltuberyieldandharvestindex.Furthermore,

likemanyothercomplexquantitativetraits,eachdevelopmentalstageandcropfactor

in potato is likely to be regulated and controlled by a large set of interacting

expressedgenesthroughouttheplantgrowthandthestronginfluenceofgenotype‐

by‐environment (GE) interactions (Allard and Bradshaw, 1964; Jefferies andMacKerron, 1993; Bachem et al., 2000; Schittenhelm et al., 2006). All these

complexitiesmakethemanipulationofyielddeterminingtraitsandtheirpredictiona

challengingtaskinpotato.

The traditional system of potato breeding is a laborious and time‐consuming

process, and the probability of finding superior cultivars is very low. Breeding

generallyinvolvesevaluationandselection,basedonmanydifferenttraitsincluding

yield, disease resistance and quality, of the clonally propagated progeny of a cross

between two clones. These parent clones can be existing cultivars or clones with

introgression from wild species. Normally, it takes more than 10 to 15 years to

produceanewcultivar(Caligari,1992;Hunt,1993;HaverkortandKooman,1997).

Roleofmulti‐environmenttrialsPlant breeders are mostly interested in characterising genotypes in diverse

environments to understand how a particular genotype performs in different

environments. Individuals in genetic populations are studied in replicated

experimentsanddistributedover ‘discrete’ environmentswhichareoften locations

and/oryears.Themaintaskofdataanalysisistoestimateparameterswhichreveal

theeffectsofgeneticandenvironmentalfactorscontrollingtheexpressionofspecific

traits. The parameters concerning effectiveness of selection strategies of individual

genotypes, forexample,areaimedtoexamineperformanceoftraitsandcorrelation

between them across environments, variability of entries due to genetic and

environmental components, heritability, and genetic gain over generations of

selectionfornewsuperiorvarietiesingrowingregions(Tai,2007).Informationlike

thesecouldhelptobetterunderstandthetypeandsizeofgenetic(G),environmental

(E), and GE interaction components of variation, and the reasons for theiroccurrence; and if necessary, a strategy to develop genotypes for particular target

environments.

The G or E components of variation for any trait normally show average

differencesbetweengenotypesorenvironments,respectively.Purelyenvironmental

effects, reflecting the different ecological potential of sites and management

Generaldiscussion

161

conditions, are not of direct concern for the breeding or recommendation of plant

varieties. However, differences between genotypes may vary widely among

environmentsinthepresenceofGEinteractioneffects.Therefore,thereisaneedtoquantifythe impactsofGEinordertoexplorethepotentialopportunitiestowardsyieldimprovementunderthespecificand/orvariedenvironments.Thiscouldhelpto

betterunderstandthetypeandsizeofG,E,andGEcomponentsofvariation,andthereasonsfortheiroccurrence.Itcouldalsohelptodevelop,ifnecessary,astrategyto

designanidealgenotype(i.e.ideotypebreeding)inpotato.

Inpotato,however,very littlesystematic,quantitative information isavailable

about genotypic differences in response to contrasting environmental conditions

and/or input levels, e.g. temperature, precipitation, and nitrogen (N) fertilisation

regimes. A better understanding of the physiological, genetic and environmental

regulation of relationships between the main attributes of growth and yield

determiningcomponentswillfacilitatethebreedingofvarietiesabletoperformwell

underspecificsetsofenvironmentalconditions.Inordertoachievethisgoalweneed

useful tools and more concerted efforts to understand GE, and the benefits ofintegratingaccumulatedandexpensivedatasets.RoleofcropphysiologyBesidesrecentdevelopmentsingenomics(suchasgenomesequencing)thatwill

provide useful tools andmassive amounts of information to speed up the plant

breedingprocess(Stuberetal.,1999;Miflin,2000),anoptionforimprovingbreeding

efficiency is to develop and utilise a thorough understanding of underlying

morphologicalandphysiologicalprocessesthatdetermineimportanttraitslikeyield

(Bindraban,1997).

Cropphysiology,asapowerful, rapidlydevelopingandquantitativediscipline,

hastheabilitytocomplementplantbreedingasphysiologicalstudiesprovideaway

todissectcomplextraitsintosimplercomponentsthatmightbeunderseparateand

probablysimplergeneticcontrol(Yinetal.,2002).Anyincreaseinyieldmusthavea

physiological basis that, if targeted, should permit a significant advance in the

identification of superior genotypes (Richards, 1996). So far, however, only in a

limitednumberof instancesphysiologicalapproacheshavebeenappliedtosupport

crop improvementprogrammes(Sinclairetal.,2004).Thissituationmaychange in

the future if the links between physiology and genetics are established or

strengthened (Miflin, 2000). There is a growing awareness that inorder to better

analysephenotypesofcomplextraitsforanycropgenotypeunderanyenvironmental

scenario using increasingly available genomic information, integration of genetics

Chapter6

162

withquantitativecropphysiologyisrequired(Tardieu,2003).

Systemsapproaches:TheirroleinassistingbreedingAshighlightedinprevioussections,progressincropimprovementandparticularlyin

molecular approaches to plant breeding is limited by our ability to predict plant

phenotypebasedonitsgenotype,especiallyforcomplextraitslikeyield(Hammeret

al.,2010).Newapproachescouldcomplementtraditionalapproacheswithanalytical

selectionmethodologies to further improvecropyieldsand theoverallefficiencyof

thebreedingprocess.Suchnewapproacheswillbeofconsiderableinteresttoplant

breeders(Chapter1).

An important advance in crop physiology during the last four decades is the

development of dynamic process‐based crop growth models for predicting yields

undervariedenvironments(Booteetal.,1996).Process‐basedecophysiologicalcrop

growth models have been developed by integrating knowledge across multi‐

disciplinessuchasplantphysiology,ecology,agronomy,soilscience,andmeteorology

(Loomisetal.,1979).

Asquantitativecropmodelsrepresentcausalitybetweencomponentprocesses

andyield,theycanpredictcropperformancebeyondtheenvironmentsforwhichthe

model parameters were estimated. In this respect, systems approaches based on

physiologicallysoundcropgrowthmodelscanquantifyandintegratecropresponses

togenetic,environmental,andmanagementfactors.Suchanintegratedapproachcan

guide research on questions such as bridging the gap between genotype and

phenotypeandtopotentiallyresolveGEinteractions(Chenuetal.,2009;Messinaetal., 2009;Hammeret al., 2010;TardieuandTuberosa,2010;YinandStruik,2010),

andinproposingideotypesforparticulartargetenvironments(Hammeretal.,1999;

Yinetal.,2000a,2004;YinandStruik,2008).Thisoffersamajorpotentialofapplying

modelling based approaches in plant breeding strategies (Jackson et al., 1996;

Chapter1).

YieldimprovementinpotatoTuber yield is still themost important trait to be selected for, since it remains the

basisonwhichthegrowerobtainsaneconomicreturn.Yieldproductioninpotatocan

be explained in terms of an integrated response of distinct plant physiological

processes(seeequation1.1,Chapter1).Physiologicaltraitsexplainingthedynamics

ofcanopycoverandtuberbulkingareveryusefulastheyareimportantdeterminants

ofyield(Hodges,1991).The totalbiomassproductionandaccumulationdependon

Generaldiscussion

163

the intercepted photosynthetically active radiation (PAR), which is directly

proportional to the crop green canopy cover (Burstall and Harris, 1983; Spitters,

1988; Tekalign and Hammes, 2005). Differences in the rate of dry matter

accumulation can be attributed to differences in light interception caused by

variability in growth and duration of maximum green canopy cover and its

senescence,andincanopy‐levelefficiencyofutilisationofinterceptedradiationasa

result of changes in the functionality of leaf photosynthesis during tuber bulking

(Pashiardis,1987;Spitters,1988;VanDeldenetal.,2001).Thetuberyieldisthenthe

productof interceptedradiationandradiationuseefficiency(RUE)anddetermined

bythefractionoftotalbiomassthatispartitionedtothetubers,influencedbytherate

anddurationoftuberbulking(VanderZaagandDoornbos,1987;Spittersetal.,1989;

Struik et al., 1990). All this shows that yield production in potato is the result of

interactions among several crop characters which are greatly influenced by both

geneticandenvironmentaland/ormanagementfactors.

Thegenetic improvementofyieldcanbeunderstoodmoremechanisticallyby

partitioning theyield into component traits (Chapters2 and3) andanalysing their

inter‐relationships(Bezantetal.,1997;Arausetal.,2008).Itisaveryusefulapproach

for the breeder to split a complex trait intomore simple ones, provided sufficient

geneticvariationexistsforthesetraits;thattheyhaveclearlyhigherheritabilitythan

thecomplextrait,andtheirassessmentwouldbeeasierthanthatoftuberyielditself.

Untilnow,limitedeffortshavebeenmadetostudythephysiologicalandgenetic

basisofvariabilityamongtraitsdeterminingcropperformance.Thekeyphysiological

and genetic processes leading to yield development and the factors affecting these

processesarestillnotwellunderstood.Furthermore,limitedinformationisavailable

on the temporal dynamics of important above‐ and below ground developmental

processes in potato under diverse environmental conditions on a large set of

contrasting genotypes. Majority of studies have focussed on one, or a very limited

number of genotypes and/or cultivars. This has made more difficult the

understandingofcomplexregulatorymechanismsunderlyingthedynamicsoftuber

formationandotherdevelopmentalprocesses.ThesisapproachKeeping in view the above discussion, this thesis aims to develop an approach to

quantify and predict the temporal dynamics of yield formation of individual

genotypesfromalargepopulationandtoestimateparameterswhichcouldrevealthe

effects of genetic and environmental factors on the important plant processes

controllingyieldformationinpotato.Besides,thisstudyalsoexploresthequestionof

Chapter6

164

whether using a crop growth and development modelling framework can link

phenotypiccomplexitytounderlyinggeneticsystemsinawaythatmaycontributeto

new insights into molecular breeding strategies as well as into possibilities to

indirectly select for higher yield. We approach this question by physiological

dissection and integrative modelling of complex traits in potato. Figures 1.2‐1.3

(Chapter1),presentaschematicrepresentationoftheoverallapproachemployedin

thisthesis.Likemostgeneticstudiesinpotato,wechoseadiploid(2n=2x=24)F1

segregatingpopulation(i.e.SHRH)inordertoavoidthecomplexityoftetrasomicinheritance(BradshawandMackay,1994;Meyeretal.,1998).Thispopulationiswell

adapted to the growing conditions of the Netherlands and segregates formaturity

(Van der Wal et al. 1978; Van Oijen 1991). Besides, we also used five standard

cultivarswithverydifferentmaturitytypes(i.e.Première,Bintje,Seresta,Astarte,and

Karnico)asstandardgenotypes.Thedatasetsusedinthesisareprimarilybasedon

six contrasting field experiments carried out in Wageningen (52˚ N latitude), the

Netherlands, during 2002, 2004, and 2005. These experiments differed in

environmental conditions because they were carried out in different years, on

different soils and under different N fertiliser regimes, thereby creating six

contrasting environments. For more details about the plant material used and

experimentaldetails(seeMaterialsandmethods,Chapter2).

Inthefollowingsections,Idiscussthemainfindingsandoverallcontributionof

thisthesis.

AnapproachforanalysingthedynamicsofcanopycoverandtuberbulkingAs mentioned in the previous section, intercepted radiation has been found to be

linearly correlatedwith the quantity of drymatter producedwhich, in the case of

potatoes, depends on the leaf canopy size, rate, and theduration over the growing

season to intercept radiant energy (Van der Zaag, 1984; MacKerron and Waister,

1985a,b;VanderZaagandDoornbos,1987;FahemandHaverkort,1988;VanDelden

etal.,2001).Tuberbulkinginpotato,ontheotherhandisthedeterminantprocessin

the formation of harvestable products of potato crop. It marks the onset of the

production phase of the potato crop after a dynamic sequence of several complex

physiological events (Ewing and Struik, 1992; Jackson, 1999; Struik et al., 2005).

Quantitativeunderstandingofbothcanopyandtuberbulkingdynamics is therefore

imperativetoexplainpotatoproductionunderparticularconditions.

Research has addressed many aspects of potato canopy growth and

development(e.g.FleisherandTimlin,2006;Fleisheretal.,2006;VosandBiemond,

1992);however,insightsintotherelationshipsamongenvironment,nutritional,and

Generaldiscussion

165

otherfactors,particularlywithrespecttocanopycoveranditsrelationwithwhole‐

plant physiology are still lacking (Almekinders and Struik, 1996). Despite the

importance of potato tubers as a source of food and a means of propagation, the

initiation,growthanddevelopmentoftubersandthefactorsaffectingtheseprocesses

are not well understood, especially in the context of large number of contrasting

genotypes and diverse multi‐environments under field conditions. Until now, few

studies have quantitatively and systematically explored the temporal dynamics of

tuber bulking for a wide range of genotypes, with the exception of the work by

Spitters(1988)andCelis‐Gamboaetal.(2003).

InChapter2,wethereforedevelopedaquantitativeapproachtobreakdownthe

timecourseofcanopycoverduringtheentirecropcycleasafunctionofthermaltime

into component traits. Canopy cover development pattern was subdivided into

distinctstagesusingabetafunctionbasedonYinetal.(2003,2009).Themodel(eqns

2.1‐2.3,Chapter2)describesthreephasesofcanopydevelopment(i.e.canopybuild‐

up phase, maximum cover phase, and canopy decline phase) through multiple

parameters defining the timing, rate, duration and area under green canopy curve

((seeFig.2.1inChapter2);Table6.1).Thesetraitswerebiologicallymeaningfuland

directlyrelatedtotheabilityoftheadaptedgenotypestointerceptPARandthusto

createhightuberyields.Manyofthesetraitsshowedsignificantgeneticvariation.

Inaddition,Chapter3presentsaquantitativeapproachtoanalysethedynamics

oftuberbulkingduringtheentirecropcycleanditsvariabilityamonga largesetof

potato genotypes and to break the time course and its variation down into

biologically meaningful and genetically relevant component traits. Tuber bulking

results from two basic processes, tuber initiation and tuber growth. Tuber growth

normally follows a sigmoid pattern as a function of time, including an early

acceleratingphase,alinearphase,andamaturationphase(seeFig.3.1inChapter3).

TheseprocessescanbedescribedbyusingtheexpolinearfunctionofGoudriaanand

Monteith (1990) as described by eqn 3.1 (Chapter 3). However, because of its

curvilinear nature the true growth rate in the linear phase is under‐estimated.We

thereforemodified the approach of Goudriaan andMonteith (1990) and quantified

thesephasestogetherwithapiece‐wiseexpolinearfunction(i.e.combinedeqns3.1‐

3.3;Chapter3).Severalparametersweregenerateddescribingtherateandduration

oflinearphaseoftuberbulking(Table6.1).

Toaccountfortheinfluenceofdailyandseasonaltemperaturefluctuationson

tuber growth, all time variables and durationwere expressed as thermal days (td)

followingour approach inChapter2.According to the results, ourmodel approach

successfully described the sigmoidal time‐course curve of canopy growth (Fig. 2.2,

Chapter 2) and the dynamics of tuber bulking (Fig. 3.2, Chapter 3) of individual

Chapter6

166

genotypes from a large set of population across varied environments via

agronomicallymeaningfultraits(asdiscussedlater).QuantificationofresourceuseefficienciesProductivity increase can also be accompanied by an increase in the genotype’s

resource use efficiency. Efficient utilisation requires both effective capture of

resources and efficient conversion into useable biomass. In potato, plant growth

includingcanopyproduction forefficient light interceptionandphotosynthesis (see

earliertext)dependsonadequatenutrition,andoptimisedfertiliserinputs,especially

N.BothradiationandNarethereforeessentialresourcesforefficientcropproduction.

Thevariationindrymattercontentamongcultivarscanalsopartlybeattributedto

the variation in efficiency of diverting of more dry matter to the tubers (Spitters,

1988; Struik et al., 1990). Therefore the interpretation of plant growth in terms of

accumulatedinterceptedPAR,Nuptakeandtheefficiencywithwhichtheseresources

are used for dry matter production should receive adequate attention in potato

breedingprogrammes.

The aim of this thesis was also to produce tools and materials suitable for

breeding new crop varieties with improved resource (i.e. radiation and N) use

efficiencies and to study how they are influenced by genetic and environmental

influences. Nonetheless, interventions in breeding based on understanding of the

geneticandphysiologicalbasisofcropperformancehavethepotentialtoaccelerate

geneticgainsforyieldandresourceuseefficiencies.Wethereforeanalysedradiation‐

andNuseefficiencies(RUEandNUE,respectively),andtheirrelationshipswithour

estimatedphysiologicalmodeltraits.

ThetermRUEreferstotheslopeoftherelationbetweentotaldrymatter(gm‐2)

and cumulative intercepted radiation (MJ m‐2) (Haverkort and Harris, 1987).

Similarly,NUE is generallydefined as the ratiobetween total plantdryweight and

plantNuptake.Inthisthesis,however,duetolackofdetailedmeasurementsoftotal

plant biomass for our large set of genotypes, radiation andN use efficiencieswere

estimatedonthebasisof tuber‐dryweight foreach individualgenotype/cultivarby

dividing the total tuber dry matter at maturity (g DM m−2) by the cumulative

intercepted PAR (MJ PARintm−2) and total tuber N uptake (g N m−2), respectively

(Chapter3).

Theresultsindicatedthatthereexistedasignificant(P<0.01)widevariationfor

bothRUEandNUEwithintheF1populationwithvaluesrangingfrom1.5gDMMJ‐1

to 2.6 g DM MJ‐1 and 54.4 g DM g‐1N to 81.6 g DM g‐1N, respectively (Table 3.2,

Chapter 3). Thesewide variations (see Fig. 3.3, Chapter 3) suggested transgressive

Generaldiscussion

167

Table6.1.DescriptionofparametersreproducedfromChapters2and3.tdstandsfor

thermalday.

Trait Description Unit

Canopycoverdynamics

tm1 Inflexionpointduringcanopybuild‐upphase td

t1,DP1 Periodfromplantemergencetomaximumcanopycover td

(momentanddurationsinceemergence,respectively)

t2 Onsetofcanopycoversenescence td

te Timeofcompletesenescenceofcanopycoverorcrop td

vmax Maximumlevelofcanopycover %

cm1 Maximumcanopygrowthrateduringbuild‐upphase %td‐1

c1 Averagecanopygrowthrateduringbuild‐upphase %td‐1

c3 Averagecanopysenescencerateduringdecline‐phase %td‐1

DP2 Totaldurationofmaximumcanopycoverphase td

DP3 Totaldurationofcanopysenescencephase td

A1 AreaundercanopycurveforDP1 td%

A2 AreaundercanopycurveforDP2 td%

A3 AreaundercanopycurveforDP3 td%

Asum Areaunderwholegreencanopycurve td%

Tubergrowth

tB Onsetoflinearphaseoftuberbulking td

tE Endoflinearphaseoftuberbulking td

ED Totaldurationoftuberbulking td

cm Rateoftuberbulking gtd‐1

wmax Maximumtuberdrymatteratcropmaturity gm‐2

segregation.

The continuous selection in most breeding programmes has very much

narrowed down the genetic base for important resource use efficiencies in most

moderndaypotatocultivars (Brown,1993;Love,1999).This thesis concluded that

there are enough possibilities to exploit the available variation in the SH RHbreeding population and may offer more efficient screening possibilities for

Chapter6

168

improvingresource(radiationandN)useefficiencies.Inthecomingsectionswewill

discussthebasisofthisvariation.

GeneticanalysisTheprospectsofimprovingatargettraitbyselectingforcomponenttraitsismostly

determinedbytheircorrelationwiththetargettrait(e.g.yield)aswellasamountof

heritableor genetic variation for theparticular component traits relative to thatof

non‐heritableor environmental variation, andby thenature andmagnitudeofGEinteraction. Therefore, an understanding of such relationships within potato

germplasmisimportanttoestablishabroadgeneticbaseforbreedingpurposes.

Results of the current thesis demonstrated that standard cultivars exhibited

significant (P<0.05) differences with respect to component traits controlling the

dynamicsofcanopycoverandtuberbulking,resource(radiation,N)useefficiencies,

andfinaltuberyield.Notsurprisingly,inmostcases,performanceofthecommercial

potatocultivarswassuperiortothatofthediploidF1population(Tables2.2;3.1‐3.2).

Tetraploid potatoes are typicallymore vigorous and high yielding (DeMaine, 1984;

Hutton,1994).Thesomewhatdecreasedvigourandyield indiploidsmaybedueto

theirploidyreductionandinbreedingdepression(Kotch,1987).

Estimates of different components of variance can be used to formulate the

mostefficientbreedingstrategyforimprovedgenotypes.Studyofsourcesofvariation

amongthetestedF1genotypesindicatedthattherewassignificant(P<0.01)genetic

variation for most traits determining the course of canopy development, tuber

bulking,and(radiation,N)resourceuseefficiencies(Chapters2and3).

The results further indicated that some genotypes performed best across the

environments. Table 6.2 presents themeans of top 10% of the out‐performing F1

genotypesfor fewselectedphysiological traits.Suchgenotypescouldbeveryuseful

forselectionandfurthertestinginbreedingprogrammes.

Heritabilityofa trait isakeycomponent indetermininggeneticadvance from

selection (Nyquist, 1991). The results indicated high broad‐sense heritability (H2)

estimates across environments for most traits. For instance, >80%H2values were

recorded for traits (t2, te, DP2, A2, Asum, cm, tE, ED, RUE, and NUE) (cf. Tables 2.6

(Chapter2)and3.5 (Chapter3)).Theseresultssuggested that thesearerepeatable

traits and strongly expressed across a range of environmental conditions. On the

otherhand lowbroad‐senseheritabilityaswellasminimumCVGestimates fort1or

DP1andc3indicatedthattotaldurationofcanopybuild‐upphaseandrateofcanopy

senescence, respectively is sensitive to the environment (mainly N) (Table 2.4,

Chapter2).

Generaldiscussion

169

No.

Genotype

t e(td)

Genotype

DP2(td)

Genotype

Asum(td%)

Genotype

c m(gtd

‐1)

1SHRH83‐L9*

79.5

SHRH83‐L9

32.6

SHRH34‐H6*

6146.4

SHRH11‐F10*

61.6

2SHRH34‐H6*

79.3

SHRH‐136

32.4

SHRH83‐L9*

5836

SHRH53‐J8*

58.1

3SHRH‐136*

75.4

SHRH‐150

28.3

SHRH‐136*

5739.3

SHRH118‐C7*

53.5

4SHRH29‐H2

73.1

SHRH‐148

27.5

SHRH‐148

5156.9

SHRH51‐J7*

53.5

5SHRH‐148

73.0

SHRH73‐B11

27.1

SHRH67‐K9*

5092.7

SHRH‐465

53.0

6SHRH9‐F8

71.2

SHRH‐428

26.1

SHRH29‐H2

5065.6

SHRH86‐L12*

52.0

7SHRH73‐B11

71.0

SHRH71‐K12

25.5

SHRH9‐F8

5061.0

SHRH17‐G5*

50.8

8SHRH67‐K9*

71.0

SHRH179‐E5

24.7

SHRH73‐B11

5022.4

SHRH138‐C10*

50.4

9SHRH111‐N12

70.6

SHRH9‐F8

24.5

SHRH‐150

4992.2

SHRH‐500*

50.3

10

SHRH179‐E5

70.1

SHRH89‐M3

24.3

SHRH179‐E5

4984.6

SHRH35‐H7

48.8

No.

Genotype

ED(td)

Genotype

RUE

(gDMMJ‐1)

Genotype

NUE

(gDMg‐1 N)

Genotype

wmax

(gm

‐2)

1SHRH83‐L9

68.0

SHRH‐477*

2.78

SHRH‐136*

78.7

SHRH42‐H12*

1215.9

2SHRH‐136

66.4

SHRH11‐F10

2.77

SHRH‐406*

78.2

SHRH‐406*

1177.9

3SHRH34‐H6

63.1

SHRH33‐H5*

2.77

SHRH71‐K12*

78.1

SHRH11‐F10

1166.2

4SHRH29‐H2

60.8

SHRH23‐G9*

2.64

SHRH114‐O3*

77.5

SHRH53‐J8*

1163.8

5SHRH67‐K9

60.3

SHRH‐500

2.61

SHRH34‐H6*

77.3

SHRH34‐H6

1158.0

6SHRH9‐F8

58.0

SHRH35‐H7*

2.59

SHRH‐148*

76.4

SHRH179‐E5

1122.0

7SHRH‐416

56.5

SHRH97‐M9*

2.56

SHRH17‐G5*

74.8

SHRH89‐M3*

1116.1

8SHRH87‐M1

54.2

SHRH95‐M7*

2.55

SHRH67‐K9*

74.2

SHRH6‐F5

1115.8

9SHRH71‐K12

53.2

SHRH‐465*

2.55

SHRH‐150*

73.8

SHRH97‐M9

1107.3

10

SHRH179‐E5

51.7

SHRH53‐J8

2.47

SHRH‐478*

73.3

SHRH83‐L9*

1094.0

Table6.2.ListoftoptenbestperformingF1genotypesforspecificphysiologicalmodeltraits.Traitvaluesindicategenotype

meanacrossenvironments.FordescriptionoftraitsseeTable6.1.*denotesparticulargenotypesthatperform

edwellinlowN

environm

ent(i.e.Exp3).

Chapter6

170

GE interaction is an extremely important issue in potato breeding as suchinteraction impedesplantbreedingprogress for complex traits suchasyieldand is

consideredtobeamongthemajorfactorslimitingresponsetoselection(Tarnetal.,

1992;Bradshaw,1994;Kang,1998).Thedevelopmentofshootsandtubersinpotato

isstronglyinfluencedbybothgeneticandenvironmentalfactors.Inthisstudy,itwas

noted that proportion of GE interaction variancewas greater than the G variancecomponentfortraitsdeterminingthecanopybuild‐upandsenescencephases,seeFig.

2.2(Chapter2).Thesetraitsmainlyincludedmaximumcanopycover(vmax),ratesof

canopygrowthandsenescence(cm1,c1,andc3),duration(DP1)andareaundercanopy

cover(A1)whencanopycoverreaches itsmaximum,duration(DP3)andareaunder

canopycover(A3)ofcanopysenescence(Table2.3,Chapter2).Thiscouldmeanthat

these traits may show a range of phenotypic expressions when subject to diverse

environmentsduetoGEinteraction.ThelargertheGEinteractioncomponent,thesmaller theheritabilityestimate; thus,selectionprogresswouldbereducedaswell.

However, themoderate (50% <H2< 70%) heritability estimates formost of these

traitsindicatedthatthesetraitsareresponsivetoselection(Table2.6,Chapter2).

Our results further indicated that some important physiological traits were

stable across the environments on account of their G component of variance being

higherthantheGEinteraction.Thesetraitsincludedonsetofcanopysenescence(t2),cropmaturity (te), duration (DP2) and area under canopy cover (A2), when canopy

coverisatmaximumlevel(or100%),andtotalareaundergreencanopycover(Asum)

(Table2.3,Chapter2).

Incaseoftraitsrelatedtothedynamicsoftuberbulkingandresource(radiation

andN)useefficiencies,thenatureofthedatasets(i.e.lackofindependentreplication

of estimations) did not allow estimating the GE component of variation for thesetraits. However, highCVG estimates for these traits alongwith high (>80%) broad‐

senseH2 (asdiscussedearlier) indicatedstronggeneticbasis for these traits(Table

3.5,Chapter3).

Studyofsourcesofvariationfortuberyieldperseontheotherhand,indicated

high contribution of environmental (E) variance in its total observed phenotypic

variance(Table3.4,Chapter3).Thisisnotsurprisingasyieldisacomplextraitand

thenetresultofmanyplantphysiologicalprocessesandtheirrelationshipswiththe

environmentparticularlyNasdiscussedinthelatersections.

For a successful breeding programme, genetic diversity and variability in a

populationplaysavitalrole(Omoiguietal.,2006).Thisthesisclearlyindicatedthat

thereexistsstronggeneticdiversityintheSHRHgermplasmfornearlyallthemodeltraits. There are opportunities to further analyse and exploit this wide genetic

variabilityavailable,whichcouldpotentiallybeusedinbreedingprogrammesaimed

Generaldiscussion

171

at improving canopy, as well as tuber bulking dynamics and resource (radiation, N) use

efficiencies and ultimately tuber dry matter production. RoleofenvironmentPlant breeders are mostly interested in characterising genotypes in diverse

environmentstounderstandhowaparticularsetofgenotypesperformdifferentlyin

different environments. A thorough understanding of the relationship between the

crop and the environment may help to increase potato productivity for specific

environments.

Inthisthesis,environmentrepresentsacombinationofyear,site,anddifferent

N availability regimes (Chapter 2). Typically, N is the key environmental variable

affecting the important yield determining growth and development processes of

potato(Honeycuttetal.,1996).Nmanagementisthereforeahighpriorityinpotato

cropping systems. In this study we lacked precise information about amount of

mineral N becoming available during the course of crop growth. However, crop N

uptake can be used as a site‐specific indicator of N that is “available” to the crop

(Sullivanetal.,2008).Therefore,toassesstheroleofN,totalamountofNuptakeby

tuberswas used as an indicator ofN availability per each environment (Table 2.1,

Chapter2).

Our results further indicated good correlations among the environments for

mostofphysiologicaltraits,e.g.Asum.However,thecorrelationsvariedmuchforsome

traits like t1 orvmax. It is obvious that traitsmore sensitive to environment and/or

GE interaction show more heterogeneity among the environments (Fig. 6.1).Rankingof genotypes changed across the environments (Table6.2)withmaximum

variabilityobservedwithin theF1populationunder lowNenvironment(i.e.Exp3)

(Table2.1,Figs.2.3‐2.4(Chapter2);Table3.4(Chapter3)).Suchchangesingenotype

rankinghavebeendefinedascross‐overGEinteractions(Baker,1988).Cross‐overtypes of interactions are important to breeders and production agronomists for

identifying adapted traits and may enable selection strategies for developing

improvedvarieties.

Plantbreedersandgeneticistshavealong‐standinginterestininvestigatingand

integrating G and GE interaction in selecting superior genotypes for specificapplicationsandforparticulartargetenvironmentswithhighestpossibleeconomical

yieldandresourceuseefficiencies(KangandPham,1991).Itwasremarkabletonote

thatsomegenotypes ‐besidesperformingbest inNsufficientenvironments–were

alsohighlyrankedunderresourcepoorconditions(i.e.Exp3withlowNavailability).

Chapter6

172

Figure6.1.H

eatmapsrevealingcorrelationsbetweensixenvironm

entsfortraits(a)t 1(b)vmax,and(c)A

sum

withinF1populationofpotato.FordescriptionoftraitsseeTable6.1.

(a)

(b)

(c)

Generaldiscussion

173

Table6.2 indicatessuchgenotypes for fewselectedphysiologicalmodel traits.

For instance genotypes SHRH34‐H6 and SHRH‐136 showedmaximumvalues for te

(timeofcropmaturity),Asum(areaunderwholegreencanopycurve)whereassome

other genotypes (e.g. SHRH11‐F10, SHRH53‐J8) indicated high cm (rates of tuber

bulking). It was further noted that the majority of best performing genotypes

indicatedhighresource(radiationandN)useefficienciesunderlowNenvironment

(Chapter 3; Table 6.2).Moreover, some of the best yielding genotypes also yielded

relativelyverywellunderlowNconditions(e.g.SHRH42‐H12,SHRH‐406,SHRH53‐J8,

and SHRH83‐L9; Table 6.2). Such genotypes therefore could be very useful for

breedingforwidelyadaptedpotatolines.

In addition, focus onwhy particular genotypes perform exceptionally well in

low or high input situations could enable selection strategies to be developed for

improvedvarieties.

Overall,theseresultsindicatethatNavailabilitymightbeoneofthekeydriving

factors for causing trade‐offs between the physiological traits in different

environments.Moreover,resultssuggestedthatNavailabilityanditsinteractionwith

genotype’smaturitytypemainlycontributetotheGEinteractionofthegrowthanddevelopment related processes in potato (see Fig. 3.5, Chapter 3). This thesis

therefore allowed a partial understanding of the environmental causes of the

observedGEinteractions.

UnderstandingmaturityinpotatoCropsundergosequentialdevelopmentalphasesfromemergencetosenescencethat

arenotonlycharacterisedbytheirchronologicalage,butalsobytheirphenologyand

reproductive capacity. Inmost annual crop plants, the transition to thematurity is

markedbytheonsetoffloweringandseedproduction(Bond,2000).However,inthe

caseofpotato,progresstomaturityisdifficulttoassess.Thisismainlybecausemost

potatogenotypespossess indeterminategrowthpatternandthereforemaintain the

capacity todevelopnew leaves and continue to grow throughout themajorpartof

theirlifecycle(StruikandEwing,1995;AlmekindersandStruik,1996;Fleisheretal.,

2006).

The sequential phases of growth determining the duration of phenological

stagesofpotatogenotypescouldbeusedtodefinethematuritytypeofgenotype(s).

In order to better understand the maturity type in potato and to study the

consequencesofmaturity typeon the canopycoverand tuberbulkingdynamicsas

wellasresource(radiation,N)useefficiencies,asetofcultivarscoveringawiderange

ofmaturitytypeswereincludedinthisthesis,seeMaterialandMethods(Chapter2).

Chapter6

174

Theresults indicated that lengthof thecanopybuild‐upphase (t1orDP1)was

conservative. Genotypic differences during early growth stages of potato are less

distinct (Spitters, 1988). The duration of maximum canopy cover (DP2) and the

canopy decline phase (DP3)varied greatly with maturity type, with late maturing

genotypeshavinglongerDP2andDP3andthusaccumulatedhigherareaunderwhole

greencanopycurve(Asum)(Table2.2,Chapter2).Theamountoflightinterceptedisin

proportiontotheareaunderthewholegreencanopycurve(Vos,1995a;Fig.6.2).

y = 0.1037x + 80.16R² = 0.9728

0

200

400

600

800

1000

0 2000 4000 6000 8000

Cum

ulat

ive

PAR

int (

MJ

m-2

)

Asum (td %)

Figure6.2.RelationshipbetweencumulatedPARinterceptionandareaunderwhole

greencanopycurve(Asum)inanF1populationofpotatoacrossenvironments.

Thetuberbulkingrate(cm)washighestforearlymaturinggenotypesfollowed

bymid‐lateandthenlategenotypes.Latematuringgenotypeshadlongestperiodof

tuber bulking (ED) followed by mid‐late and early genotypes. As a result crop

matured(tE) laterandtuberyield(wmax)washigher in lategenotypesthan inearly

genotypes(Table3.1,Chapter3).ThiswasinlinewithKoomanandRabbinge(1996)

and Spitters (1988), who reported that, comparedwith late cultivars, early potato

cultivarsallocatea largerpartof theavailableassimilates to thetubersearly in the

growing season, resulting in shorter growingperiods and also lower yields.On the

otherhand, latematuring cultivars combine a long canopy coverwith a long tuber

bulking period (ED) and therefore achievemore tuber drymatter yield (wmax) per

Generaldiscussion

175

unitofNuptake thanmid‐earlyandearlymaturingcultivars (Zebarthetal.,2008).

Figure6.3illustratestherelationshipbetweencropmaturityandfinaltuberyieldfor

asetofstandardcultivarsand(SHRH)segregatingpopulation.The results were further evaluated for relationship between maturity and

resource(radiation,N)useefficiencies.RUEvalueswerehighest forearlymaturing

genotypesfollowedbymid‐lateandlategenotypes(Table3.1,Chapter3).Thismight

berelatedwiththeleafage,whichislinkedwiththepartitioningofdrymatter.Ifthe

foliage of a crop remains green for a longer period, for example in late‐maturing

cultivars, andnonew leavesare formed in the latepartof thegrowing season, the

leavescanbecomesooldthattherateofphotosynthesisdeclinestowardstheendof

the season resulting into low RUE values (Van der Zaag and Doornbos, 1987). In

contrary,resultsindicatedhighNUEestimatesforpotatocultivarswithlatermaturity

(Zebarthetal.,2003,2004).Thereducedcanopysenescenceandgreaterpartitioning

ofN to above groundpartsmost likely contributed to thehighmeasuredvaluesof

NUEforlatecultivars(Sharifietal.,2005).

This thesis indicated that most of the physiological model traits (either

individually or in combination) elaborately explained the genetic variation in

maturitytype.Ourapproachcouldbeusefulinelucidatingtheroleofmaturitytypein

potato.

Thedefinitionofmaturitytypeasagenotypictraitinpotato,however,israther

ambiguousandcurrentlytherearesomanyunclearinterpretations.Manypublicand

privateinstitutionshavetriedtodefinethematuritytypeinpotatoandhavecomeup

with theirowndefinitions.The conventionalmethodor criterionof elucidating the

maturity type used is mostly based on visual observations made in the field at

particular time intervals during the crop cycle and transferring that information to

unit‐less ordinal scales. This criterion does not have any biological meaning and

involves ambiguity and much speculation in understanding the background of

maturity,partlybecauseofthedifferentviewpointsofgrowersandprocessors.The

added physiological and quantitative knowledge gained in Chapters 2 and 3 were

thereforeusedtoevaluatetheconventionalsystemofmaturitytypeandtoquantify

and re‐define the concept ofmaturity type on physiological basis for a large set of

genotypes. Four traits (DP2, cm,ED, andAsum,)were selected among the large set of

physiologicaltraitsbasedontheirhighdirectresponsetoselection(Table6.3),are‐

definedphysiologicalbasedmaturitycriteriawasdevelopedandcomparedwiththe

conventionally used criterion (Chapter 4). The results indicated that the

conventional‐criterion based maturity scores showed more chances of errors and

lack of repeatability in maturity scoring throughout the different maturity classes

(veryearly, early/mid‐early,mid‐late/late, andvery late).Moreover, themethodof

Chapter6

176

Figure 6.3. Final tuber yields of 100 F1 genotypes, 5 standard cultivars, and 2

parentalgenotypes(SHandRH),measuredineachindividualexperiment.

Thermal days (td)

Tub

er y

ield

(g

m-2)

0

500

1000

1500

2000

2500

Exp 1 Exp 2

0

500

1000

1500

2000

2500

Exp 3 Exp 4

0 25 50 75 100

Exp 60

500

1000

1500

2000

2500

0 25 50 75 100

Exp 5

F1 Première Bintje Seresta Astarte Karnico SH RH

Generaldiscussion

177

classification lacked the capability to clearly interpret and discriminate between

maturityclassesforthecrucialgrowthrelatedprocessesofpotato(e.g.Asum)(seeFig.

4.5,Chapter4).

The re‐defined criteria of maturity type based on physiological traits on the

other hand tended to define maturity classes less ambiguously and with a clear

conceptualbasis.Thisstudythereforeindicatedthatre‐definingthematuritytypeof

alargesetofgenotypesispossibleonthebasisofphysiologicaltraitsandmayprove

veryuseful.

Thephysiological traits related to the capability of genotype to interceptPAR

and its rate and duration of tuber bulkingwere indicated crucial formaturity and

therefore could be used to quantify variation inmaturity type among sets of new,

unknownpotatogenotypes(Chapter4).

Thenewphysiology‐basedmaturitydefinitionwouldnotonlyallowrelatingthe

maturitytocropphenologyandphysiologybutcouldalsoofferwiderapplicationsin

potatobreedingandcropmanagementstudies.Onesuchapplicationwouldbetohelp

indesigningbreedingstrategiesforpotatoideotypeswithspecificmaturitytypefor

specificgrowingconditionsand/orenvironments.Inter‐relationshipsamongthephysiologicaltraitsYield is a complex trait associated with many inter‐related components (Yan and

Wallace, 1995). The knowledge of quantitative relationships among physiological

charactersdeterminingyieldcanprovideuseful insightsintohowdifferencesofthe

complex trait and thesignificance in correlationsbetweencomponentandcomplex

traits(e.g.yield)comeabout(Chapter1).Inaddition,determiningthephysiological

traitsmostinvolvedintheformationofayieldcomponentcouldgiveinsightinto

thepossibilitiesofmanipulatingthesizeornumberofthecomponent.

Although information about the correlation of agronomic and morphological

characters with yields is helpful in the identification of the components of this

complexcharacter,yetthesedonotprovidepreciseinformationontherelative

importanceofdirectandindirectinfluencesofeachofthecomponentcharacters.

Manyresearchers(e.g.Bhatt,1972)havereportedthatmerelycorrelationstudiesdo

not clearly reveal such type of information and can be misleading if the high

correlation between two traits is a consequence of the indirect effect of the traits

(Dewey and Lu, 1959). Procedures like path coefficient analysis permit a thorough

understanding of relationships and contribution of various characters to the target

traitbypartitioningthecorrelationcoefficientintocomponentsofdirectandindirect

effects.

Chapter6

178

Path coefficient analysis was therefore performed to quantify the inter‐

relationshipsofourphysiologicalmodeltraitsandtoindicatewhethertheirinfluence

isdirectlyreflectedinthetuberyield(wmax)or iftheytakesomeotherpathwaysto

produceaneffect.Theresults revealed that traits likeAsum followedbyRUEandcm

exerted the highest direct influence on wmax (Table 3.7; Fig. 3.6,

Chapter3).

TheresultsindicatedthathighdirecteffectsofAsumonwmaxweremainlydueto

significant (P<0.05), strong positive phenotypic correlations betweenAsum,ED, and

NUE(Fig.3.6,Chapter3).Ontheotherhandsignificant,strong,positivephenotypic

correlations (r=0.69)betweenRUEandcmwerereflected in theirhighervaluesof

directeffectonwmax.However, results indicated that totaleffects (i.e. sumofdirect

and indirect effects) of traits cm andRUE onwmaxwere reducedmainly due to the

strong, negative, indirect effects ofAsum on these traits (Table 3.7, Chapter 3). The

resultsfurtherindicatedthateffectsoftraitsEDandNUEonthetuberyieldconsisted

mostly of positive indirect influencesmainly throughAsum, suggesting a correlated

responseinselection(Table3.7,Chapter3).

These findings thereforesuggested thatselection forAsum,RUE,andcmshould

be emphasised in breedingprogrammes for improving tuber yields.Donald (1968)

proposedthatanidealgenotypeisthatwithaspecificcombinationofcharacteristics

favourableforgrowthandyieldproductionbasedontheknowledgeofplantandcrop

physiology and morphology. Our results further indicated that an ideal potato

genotype from a breeding perspective is characterised by green canopy cover that

intercepts solar radiation for as long as possible (i.e. with higherAsum) during the

availablegrowingseasontoaccumulateasmuchdrymatteraspossiblemaintaininga

maximumcapacitytodivertdrymattertothetubers(i.e.withgreaterperiodoftuber

bulking(ED))withoutcompromisingtheoptimallevelsofgrowthrates(cm)andRUE

ensuringhighestpossibleeconomicaltuberyields.Figure6.4schematicallyhighlights

varioussituationsofpotatoideotypesunderbothoptimumandresourcepoor(lowN)

conditions.Researcherscouldthenstrivetoobtainanoptimumcombinationofyield

components thatwouldbestsuit therequirement forhighyieldingability inpotato

genotypesforanyparticularenvironment.IndirectselectionforyieldinpotatoInatraditionalcropbreedingprogramme,elite linesare inter‐crossedandthenthe

highest yielding linesare selected.However, yield selection is empiricaldue to low

heritabilityandahighgenotype–environment interaction (Reynoldsetal.,1999). It

alsorequires theevaluationofa largenumberofadvanced lines in fieldyieldtrials

Generaldiscussion

179

(a)

(b)

0

500

1000

1500

2000

0 250 500 750 1000

Cumulative PARint (MJ m-2)

Can

opy

cove

r (%

)

Tub

er d

ry m

atte

r (g

m-2)

0

500

1000

1500

2000

0 250 500 750 1000

0

20

40

60

80

100

0 25 50 75 100

0

20

40

60

80

100

0 25 50 75 100

Thermal days (td)

Figure6.4.Schematicrepresentationofthecourseofgreencanopycoverandtuber

drymatterproductionof5standardcultivars,and2parentalgenotypes(SHandRH)

for(a)lowNsituation(Exp3)and(b)optimumNsituation(Exp6).

over several years and locations (Ball and Konzak, 1993). Potato breeding

programmes therefore constantly need new strategies to improve efficiency by

increasingthefrequencyofselectedgenotypesandreducingtimeandcosts.

An alternative strategy that gives a high indirect response to selection for

averageyieldovertheproductionenvironmentshasthepotentialforscreeninglarge

numbers of genotypes in breeding programmes for identifying and selecting high‐

yielding lines (Shorter et al., 1991; Cooper et al., 1995; Marti et al., 2007). In this

context,thebestapproachwouldbetoselectforcomponenttraitsthatareputatively

relatedwith ahigheryieldpotential and/or to the improvedbehaviourof the crop

whengrowninaparticularenvironment.Thisisknownasanalyticalorphysiological

Chapter6

180

breeding(Chapter1).

Oneofthekeyobjectivesofthisthesiswastoidentifyphysiologicaltraitswhich

couldbeusefulasaselectioncriteriontoimprovecropyieldinpotato.Therefore,in

this section,ouraimwas toestimate theexpecteddirect selectionresponse (R) for

tuberyieldandalso forphysiological traits, correlatedselectionresponse(CR),and

efficiencyofindirectselection(CR/R)fortuberyieldfromphysiologicaltraits.Mean

values of F1 population were used to estimate the genetic correlations between

physiologicaltraitsandmeasuredtuberyield(Chapters2and3);estimatesforR,CR,

andCR/RwereestimatedfollowingthemethodsdescribedbyFalconer(1996)as:

x2xiHR (6.1)

yGyx rHiHCR (6.2)

xGy // HrHRCR (6.3)

where, i istheselectiondifferential(i=1.76at10%selectionintensity); 2xH and x

aretheheritabilityandphenotypicstandarddeviationvaluesfortraitx,respectively;

HxandHyarethesquarerootoftheheritabilityofthetraitxandy,respectively; Gr is

thegeneticcorrelationbetweentraitxandy, y isthephenotypicstandarddeviation

fortraity.

Theheritabilityandgeneticcorrelationswereestimatedaspereqns3.7and3.8,

respectively(cf.Chapter3).ValuesofR,CR,andCR/Rwereexpressedaspercentage

ofthegenotypesmeanforeachtraitinordertoallowcomparisonamongtraitswith

differentunits(Table6.3).

Theresults indicatedhigh(>50%)expectedordirectresponsetoselectionfor

physiologicalmodel traits:DP2,A2,Asum,cm,tE,andED.However, itwasnotablethat

direct selection response to tuber yieldperse (wmax)was very low (i.e. only 29%)

(Table 6.3). This was also supported by relatively lower proportion of genetic

componentofvarianceinthetotalphenotypicvarianceobservedfortuberyieldper

se(Chapter3).Theseresultsshowedtheaddedvalueandsuperiorityofphysiological

traitsovertuberyieldperse.

The results furthermore indicated that correlated response towmax based on

most physiological traits was higher than the direct selection for wmax. Using an

alternative indirect selection tool for yield is appropriate if the genetic correlation

between the target trait (e.g. yield) and component traits is very high, and the

correlatedresponseintheunselectedtraitbasedontheselectedtraitishigherthan

Generaldiscussion

181

Table 6.3. Genetic correlation coefficients (rG) between physiological traits and

measuredtuberyield(wmax),expectedselectionresponse(R)forphysiologicaltraits

and forwmax, correlated selection response (CR) and efficiency of indirect selection

(CR/R)forwmaxestimatedfromphysiologicaltraitsofpotato.Estimationswerebased

onF1populationmeanacrossexperiments.tdstandsforthermalday.Fordescription

aboutparameters,seeTable6.1.

Trait Unit rG R CR CR/R

tm1 td 0.24** 0.31 0.06 0.29

t1/DP1 td ‐0.05NS 0.20 ‐0.01 ‐0.09

t2 td 0.70** 0.22 0.22 0.57

te td 0.75** 0.27 0.25 0.57

vmax % 0.84** 0.18 0.23 0.94

cm1 %td‐1 0.53** 0.48 0.15 0.57

c1 %td‐1 0.46** 0.48 0.14 0.42

c3 %td‐1 0.17** 0.27 0.03 0.40

DP2 td 0.85** 0.98 0.27 0.73

DP3 td 0.15** 0.39 0.04 0.20

A1† td% − − − −

A2 td% 0.84** 1.07 0.27 0.69

A3 td% 0.41** 0.44 0.11 0.46

Asum td% 0.78** 0.49 0.26 0.59

cm gtd‐1 ‐0.20** 0.78 ‐0.06 ‐0.16

tE td 0.66** 0.53 0.22 0.49

ED td 0.64** 0.93 0.21 0.48

RUE gDMMJ‐1 ‐0.25** 0.41 ‐0.08 ‐0.22

Nuptake gm‐2 0.85** 0.29 0.24 0.91

NUE gDMg‐1N 0.43** 0.28 0.13 0.39

wmax gm‐2 0.29 **Significantat1%,NSNon‐significant.†Estimation was not possible due to zero genetic variance ( 2

G ), see Table 2.3,

Chapter2.

thedirectresponsetoselectionoftheunselectedtrait(Falconer,1996).Ourresults

furtherindicatedsignificant(P<0.01)andstronggeneticcorrelationsbetweentuber

Chapter6

182

yieldandmostofphysiologicalmodeltraits(Table6.3).Forinstance,traitssuchast2,

te,vmax,cm1,DP2,A2,Asum,tE,ED,Nuptake,anduseefficiencyshowedveryhigh(>0.80)

genetic correlations with wmax. The results therefore suggested that high indirect

responsecouldbeobtainedbyselectinggenotypesforthesetraitstoimprovetuber

yieldinpotato.However,ourresultsalsosuggestedthatwhileusingthesetraitsasa

criterion for selection, their causal physiological inter‐relationships and trade‐offs

must be considered simultaneously (see Fig. 3.6, Chapter 3). Finally, the results

concluded that indirect selection efficiency of most physiological traits was higher

thanthedirectselectionefficiency for tuberyield.Theselectionefficiency fromour

physiological traits ranged up to 94% (Table 6.3). However, it was noted that

efficiencyof indirect selectionwasnothigh for traits likecmandRUEdespite their

high direct physiological relationship with tuber yield determination (Fig. 3.6,

Chapter3).Thiswasmainlyduetotheweakphenotypic(Table3.6,Chapter3)aswell

asgeneticcorrelations(Table6.3)withtuberyieldobservedforthesetraits,mainly

influencedbythestrongnegativeindirecteffectsofAsum(asexplainedintheprevious

section).

Accordingtoourresults,therelativeimportanceoftraitsbasedontheir(>50%)

indirectselectionefficiencyfortuberyielddeterminationcanberankedas:vmax>N

uptake>DP2>A2>Asum>t2>te>cm1.Thesefindingsstronglyconcludedthatmost

physiological traits can be used successfully as indices of selection for yield

improvementinpotato.

Thisthesisthusindicatedthedirectionandmagnitudeofcorrelatedresponses

toselectionfortuberyieldandtherelativeefficiencyofindirectselection.Inaddition,

our results suggest that the development of such a selection indexmust integrate

several key physiological traits, their inter‐relationships and repeatability (high

heritability) for assessing yield in breeding programmes (Baker, 1986; Bouman,

1995).

QTLmappingofphysiologicaltraitsWhile environmental characterisation and physiological knowledge help to explain

andunravelgeneandenvironmentcontextdependencies,theanalysisofgeneeffects

on yield in diverse environments unmasks some of the key effects of genetic and

environmental variability on the target trait.The phenotype of most plant

characteristics varies quantitatively as it is under the influence both of the

environment and of genetic factors encoded at quantitative trait loci (QTLs)

(Gelderman,1975).Understandingthebasisoftheinteractionresponsesatthewhole

plant level (Reynolds et al., 2005; Raven et al., 2005; Trethowan et al., 2005) and

Generaldiscussion

183

identifyingQTLandmolecularmarkersassociatedwithdesirabletraits(Priceetal.,

2002;Nigametal.,2005;Habashetal.,2007)arelikelytohavealargeimpactonthe

researchaimedatimprovingyieldinpotato.

Many studies have identifiedQTLs for different traits in potato under normal

conditions for various agronomic, andquality traits (e.g.VandenBerget al., 1996;

Van Eck et al., 1994; Schäfer‐Pregl et al., 1998; Bradshaw et al., 2008). However,

knowledgeaboutgeneticbasisofimportantphysiologicaltraitscloselylinkedtothe

temporaldynamicsofyieldformationandresource(radiationandN)useefficiencies

isstilllimitedandhardlyanyQTLshavebeenidentifiedforthesetraitsinpotato.In

otherwords,physiologicalaspectsofcomplexquantitativetraitsinpotatohavesofar

received little attention from geneticists in QTL analysis although there are some

recentstudiesthatusestatisticalmodelstoanalysedatabasedonsemi‐quantitative

scalesdescribingcanopydevelopment(e.g.Hurtadoetal.,2012).

In this study, our molecular dissection of traits determining the dynamics of

canopy cover, tuber bulking, and resource (radiation, N) use efficiencies identified

severalQTLs,themappingpositionofeachidentifiedQTL,theinteractionofQTLwith

environment,andthemagnitudeofQTLeffectonexplaininggeneticvarianceinboth

SHandRHparentalgenomes(Chapters2and3).

The QTL results clearly showed that one particular chromosomal position at

18.2cMonpaternal(RH)linkagegroupVwascontrollingnearlyallthetraits(Tables

2.9and3.10),renderingthepopulationonwhichthisresearchwasdonelesssuitable,

albeititisagronomicallywelladapted.Forexample,itwasnotablethatQTLfortraits

likete,t2,DP2,A2,Asum,cm,tE,ED,andtuberyieldpersewereco‐localisedinmajorityof

theenvironments.ThisQTLwasassociatedwithmajoradditiveeffectsonmostofthe

traits and explained more than 50% of the phenotypic variance. Strong genetic

correlations among most of these traits (Chapters 2 and 3) and with tuber yield

(Table6.3) further supported these results and suggestedpleiotropicnatureof the

QTLformostofthetraits.

It iswellknownthatlinkagegroupVharbourstheQTLforplantmaturityand

vigourinpotato(Collinsetal.,1999;Oberhagemannetal.,1999;Viskeretal.,2003;

Bradshaw, et al., 2008). Our results also confirmed this fact as most of the traits

linkedtothischromosomewererelatedwithmaturity, for instanceDP2,cm,ED,and

Asum(Chapter3).Inotherwords,thisthesisgaveaclearpictureaboutmaturityina

SH RH segregating population of potato and indicated that maturity is mainlyexpressedfromthepaternal(RH)sideonlinkagegroupV.Itwasfurthernotablethat

this linkage group is mainly controlling earliness in genotypes on account of the

negative additive effects associatedwith amajorQTL foundhere formost of traits

includingtuberyieldperse(Chapters2and3).Thisinformationcouldbeveryuseful

Chapter6

184

in elucidating the genetic basis of yield determination in early and late maturing

genotypes.

TheQTL‐by‐environment(QTLE)interactionphenomenonwasevidentmainlyfortraitsdeterminingthecanopybuild‐upandsenescencephases,maximumcanopy

cover (vmax) ratesof canopygrowthand senescence (cm1,c1, andc3), duration (DP1)

andareaundercanopycover(A1)whencanopycoverreachesitsmaximum,duration

(DP3) and area under canopy cover (A3) of canopy senescence, resource (radiation

andN)useefficienciesandtuberyieldperse.Theseresultswereexpectedduetothe

presenceofrelativelylowergeneticvariancethantheirGEorEcomponentoftotalphenotypicvarianceinthesetraits,asdescribedinprevioussections(seeTables2.3

(Chapter 2) and 3.4 (Chapter 3)). Figure 6.5 illustrates such genetic complexity in

tuberyieldperse.SuchinconsistenciesinQTLbehaviourcancausemajorproblemsin

thedetectionofQTLs and their effectiveuse inmolecularbreeding.However, such

QTLs could be useful for marker assisted breeding for specific environments. Our

thesis as a whole provided a better understanding of the genetic background of

importantphysiologicalcomponentstraitsdeterminingmaturitytypeaswellastuber

yieldinpotato.

Figure 6.5. QTLE additive effects for tuber yield (g m‐2) parental (SH and RH)

genomes for six experiments. Letters above or below the bars indicate different

linkagegroups.

-700-600-500-400-300-200-100

0100200300400

1 2 3 4 5 6

Add

itive

eff

ects

Experiment no.

SH RH

V

V

V

V IVI

II

Generaldiscussion

185

AnecophysiologicalmodeltoanalysegeneticvariationintuberyieldAs already highlighted in previous sections, yield variation in terms of growth and

developmentofthecropiscomplex,foritinvolvestheeffectofexternalfactorsonall

thephysiologicalprocesses, the inter‐relationshipsbetweendifferentprocessesand

theirdependenceon thegenetic constituentof theplant.Oneof theapplicationsof

processed‐based ecophysiological crop growthmodels is their capability to explain

differencesintheyieldpotentialofcultivarsonthebasisof individualphysiological

parameters,andtheuseof thisknowledge forevaluatinganddesigningplant types

(Kropffetal.,1995;Yinetal.,2003c).Thisispossiblebecause,astheybecomemore

mechanistic and comprehensive, crop‐growth models can be used to mimic the

geneticcharacteristicsofplants(Chapter1).

However, a major limitation of current crop models in accounting for GEinteractions is the low resolution and accuracy of themodel in comparison to the

subtle differences between genotypes commonly observed inmanywell‐conducted

multi‐environmenttrials(WhiteandHoogenboom,1996).Acontrastingpointofview

is that crop physiological frameworks that are more readily aligned with plant

breeders’modesofactionarerequired(e.g.Shorteretal.,1991;Yinetal.,2004;Yin

andStruik,2008).Studieswheresimulationanalysesofvariationinatraithavebeen

confirmedinthefieldarerare.Certainly,itseemslogicalthatifcropmodelsaretobe

incorporatedintoacropimprovementprogramme,itisessentialthattheparameters

areeasilyandsimplyobtainedsothatbreederscanusethemandapplythemwithout

substantialinvestmentintimeanddatacollection.

YinandVanLaar(2005),alongtheaforementionedlinesofthinking,presented

a crop model, GECROS (Genotype‐by‐Environment interaction on CROp growth

Simulator), to overcome some of the weaknesses of earlier crop models. GECROS

capturestraitsofgenotype‐specificresponsestoenvironmentbasedonquantitative

descriptions of complex traits related to the phenology, root system development,

photosynthesis, stomatal conductance, and stay‐green traits. GECROS uses new

algorithms to summarise current knowledge of individual physiological processes

and their interactions and feedbackmechanisms (Chapter 5). It attempts tomodel

eachsub‐processataconsistentlevelofdetail,sothatnoprocessisoveremphasised

or requires too many parameters and similarly no process is treated in a trivial

manner,unlessunavoidablebecauseof lackofunderstanding.GECROSalso tries to

maintain a balance between robustmodel structure, high computational efficiency,

and accurate model output. The model can be used for examining responses of

biomass and dry matter production in arable crops to both environmental and

genotypiccharacteristics.

In this study (Chapter5),we thereforeexamined thepotentialof theGECROS

Chapter6

186

crop growth model in predicting yield differences of individual genotypes in a

segregatingpopulationofpotato,theirparentsandasetofcultivarscoveringawide

range of maturity types. We focused mainly on the genotype‐specific input

parameters of themodel. Table 5.1 (Chapter 5) gives the description of genotype‐

specific parameters of GECROS, which includemV andmR (i.e. period from plant

emergencetoonsetoftuberbulkinganddurationbetweenonsetoftuberbulkingand

cropmaturity,respectively);Hmax,plantheight;Nmax,totalcropNuptakeatmaturity;

nSO,maximumpotatoseedtuberNconcentration.Calibrationdataforthemodelinput

parameters were obtained from our extensive data sets of Chapter 2 and 3.

Procedures for estimation and correction for experiment‐specific parameter values

aredescribedinChapter5.

Themodeldescribedwellthedynamicsofimportantgrowthrelatedprocesses

in potato and gave important insights into the underlying component traits and

factors influencing the tuberdrymatterproduction.Using as fewas fivemeasured

genotype‐specific parameters (Table 5.1, Chapter 5), the model showed a good

potential in explaining the observed yield differences among genotypes. Simulated

trendsofgrowthwereincloseagreementwiththemeasureddata(Fig.5.3,Chapter

5). The model yielded reasonable predictions of differences in tuber yield across

environmentsandacrossgenotypes(Figs.5.4and5.5).

Simulationmodelscanbeveryusefultoevaluatecriticaltraitsneededforhigh

yield potential (Penning de Vries, 1991; Kooman and Haverkort, 1995; Yin et al.,

2000b). Themodel based sensitivity analysis indicated thatNmax andnSOwere the

most important parameters of GECROS model as they contributed most to the

determinationoftuberyield(seeChapter5,section3.4).Theresultssuggestedthat

genotypeswithhigherNmaxmayexhibitmorefinaltuberyield(Fig.6.6),providednSO

isloworinotherwordsgenotypeswithhighNUE(Casaetal.,2005;Chapter3).Our

path coefficient analysis also further indicated that relationships of most

physiological traits with tuber yield are actually driven mainly through plant’s

capabilitytouptakemaximumN(Table5.4,Chapter5).Theseresultsfurtherproved

ourearlierdescribedresultsthatplantNuptakecouldbeproposedasacriterionto

indirectlyselectfortuberyield.

This thesis therefore confirmed the usefulness of an ecophysiological model

‘GECROS’inexploringtheimpactofnewgenotypesandthecontributionofindividual

physiologicaltraitsonyieldbysimulatingtheresponsesofgenotypesacrossdifferent

environments. In addition, ourmodel analysis allowed us to identify the genotypic

key parameters and shed light on some of the vital physiological mechanisms

responsible for genotypic differences in tuber yield. The concepts captured in the

modeldrawmoreattentiontoNuptakeanduse

Generaldiscussion

187

Figure6.6.RelationshipsbetweenmeasuredplantNuptakeandtuberdrymatterin

F1populationofpotatoacrossenvironments.efficiency,aspossibleareasforimprovementsintuberdrymatterproduction.

The identification of both genotype‐specific key parameters and main

physiological processes involved in tuber yield formation should be useful from a

geneticpointof view. They couldguide research towards thesekeyprocesses and

orientatebreedingprogrammesassuggestedbyYinetal.(2004).Furtheranalysisof

the genotypic parameters should be performed in conjunction with molecular

markersinordertodeterminetheirgeneticcontrol.Suchinformationwouldgreatly

facilitatethedevelopmentofideotypesforspecificenvironments.

ConclusionsTheworkpresentedinthisthesisintendstoincreaseourphysiologicalknowledgeof

those factors applicable to plant breeding determining the yield of potato. We

presented a robust physiological framework for quantitatively dissecting the

phenotypic variation in potato growth and development to aid understanding of

underlying causes while simultaneously providing means to predict emergent

phenotypicconsequencesbyintegratingeffectsofvariationincomponentfactorsand

processesleadingtoyieldformationinpotato.

y = 45.777x + 106.04R² = 0.6513

0

500

1000

1500

2000

0 5 10 15 20 25 30

Tub

er

dry

ma

tter

(g m

-2)

N uptake (g m-2)

Chapter6

188

Thisthesisindicatedthatyieldpotentialinpotatocanbeenhancedbyselecting

formorphological, physiological, and phenological traits. Once this is accepted, the

challenge is in finding which traits to modify and the optimum phenotype or

expressionforthesetraits.Itwasconcludedthatparametersofthegrowthfunctions

determining thetemporaldynamicsofcanopycoverandtuberbulkinghaveaclear

meaningwithregardstotheprocessesofresourcecapturebytheplant,thusallowing

amore easy interpretation of the value andmagnitude of the growth components

associatedwithvariationintuberyieldamongcultivarsand/orgenotypes.However,

thesecomponentsarenotphysiologicallyindependentasgenotypeswithahightuber

bulking rate may effectively limit the crop growth and duration (via enhanced

internalplantcompetition)leadingtotheiridentified'earliness'.

In this context, this study also highlighted such trade‐offs among key

physiologicaltraitscausingsubtlecomplexitiesindefiningthematuritytypeinpotato.

Our results indicated that re‐defining the maturity type of a large set of new

genotypesbasedonphysiologicaltraitsispossibleandwouldproveuseful.Thenew

physiology‐based maturity definition could allow relating the maturity to crop

phenologyandphysiologyandcouldhelpindesigningpotatoideotypeswithspecific

maturitytypeforspecificgrowingconditionsand/orenvironments.

This study also highlighted the characterisation of the role of environments

mainlyN.Overall,resultsconcludedthatNavailabilityisthedrivingfactorforcausing

trade‐offs between the physiological traits and different environments. Moreover,

resultssuggestedthatNavailabilityanditsinteractionwithgenotype’smaturitytype

mainly contribute to the GE interaction of the growth and development relatedprocessesinpotatoinalargesetofF1segregating(SHRH)potatogenotypes.Thisthesis thereforeallowedapartialunderstandingof theenvironmentalcausesof the

observedGEinteractions.Ourquantitativeapproachincombinationwithmarkersofthewidelyavailable

and easy to use AFLP marker system identified QTLs that could be useful in the

development of marker assisted breeding strategies in potato yield improvement.

The QTL results showed that nearly all the physiological traits co‐localised at one

particular chromosomalpositionat18.2 cMonpaternal (RH) linkagegroupVwith

major effects. This QTL was associated withmajor additive effects onmost of the

traitsandexplainedmore than50%of thephenotypicvariance.Thissuggested the

pleiotropic nature of theQTL formost of the traits determining cropmaturity and

tuberyields.AnumberofQTLsfortraitswerenotdetectedwhentuberyieldperse

was subjected toQTL analysis. The phenotypic variance explained by theQTLs for

tuberyieldpersewasalsolowerthanforothertraits.

Our results also confirmed previous studies that most of the traits linked to

Generaldiscussion

189

linkagegroupVwererelatedwithmaturity. Itwas furthernotable that this linkage

groupismainlycontrollingearlinessingenotypesonaccountofthenegativeadditive

effectsassociatedwithamajorQTLfoundhereformostoftraitsincludingtuberyield

perse.

Thecapabilityofanecophysiologicalmodel‘GECROS’wastestedinthisthesisto

analysedifferencesintuberyieldofpotatoandanalysetheimportanceofgenotype‐

specificmodel‐inputparametertowardstuberyieldproduction.Themodelyieldeda

reasonably good prediction of differences in tuber yield across environments and

acrossgenotypes.Trendsofgrowthandnitrogenuptakewereadequatelyreproduced

bythemodel.TheGECROSmodelbasedsensitivityanalysisindicatedthattotalcrop

NuptakeandtuberNconcentrationwerethemostimportantparametersofGECROS

model as they contributed most to the determination of tuber yield. The results

concluded that GECROS model is a useful tool in analysing the contribution of

individualphysiologicaltraitstotuberyieldbysimulatingtheresponsesofgenotypes

acrossdifferentenvironments.Furtheranalysisof thegenotypicparameters should

be performed in conjunction with molecular markers. They could guide research

towardskeyprocessesandorientatebreedingprogrammesassuggestedbyYinetal.

(2004).

To conclude, breeding efforts aimed at meeting ongoing challenges, such as

breakingyieldbarriersandimprovingperformanceunderdiverseenvironments,are

more likely toachievesuccess ifphysiologicalunderstanding isusedtocompliment

traditional breeding approaches. This thesis yielded estimates for agronomically

relevant crop physiological and genetic characteristics and/or traits that are

promising fordefining futurebreeding strategies inpotato.High genetic variability

along with high heritability for most of these traits indicated that a more general

breeding goal, increased tuber drymatter yield by indirectly selection for optimal

combinationofimportantphysiologicaltraitscanbeachieved.Resultsalsoindicated

that while using these traits as a criterion for selection, the causal physiological

relationshipsandtrade‐offsmustbeconsideredsimultaneously.

Theapproachdescribedinthisthesisispromisingfordefiningfuturebreeding

strategiesinpotatoandtheinformationobtainedmayhelpindesigningideotypesfor

specificand/ordiverseenvironmentsaswellas inmarkerassistedselection(MAS).

AsmentionedinChapter1,currentapproachestoMASforcomplextraits lackgood

predictive power due to complex interactions among genes and/or environments.

One strategy to overcome these difficulties would be to combine QTL and

ecophysiological models, i.e. a step forward towards QTL‐based crop modelling.

However,smalldatasets forQTLmapping(i.e.only88genotypes,cf.Materialsand

Methods,Chapter2)andlowernumbersofindependentQTLsfoundforsingletraits

Chapter6

190

(seeTables2.8 (Chapter2)and3.9 (Chapter3))didnotallowachieving thisnovel

goal. In this context, future research should focus on this area and further steps

should be taken in linking cropmodellingwith QTLmapping for physiological and

genotypictraitsgeneratedbythisthesis.Wehopethatthestrongfoundationsetby

thisthesiswouldhelp inachievingbroadergoalofproceedingtowardsmodelbased

plantbreeding(Chapter1).Ontheselines,afollowupPhDresearchiscurrentlywell

underway including the same SH RH population as well as a very large set ofcultivars are grown in experiments in which nitrogen supply is included as an

experimentalfactor.

However,makingsignificantcontributionsintheseareasrequiresmuchcloser

collaborative research efforts between physiologists and plant breeders. Until such

research is conducted, and the benefits of shifting resources into the systems

approach can beweighed against reducing the resources put into the conventional

empiricalapproach,widespreadacceptanceofthenewmethodsisunlikely.

ReferencesAggarwal, P.K., Kropff, M.J., Cassman, K.G., Ten Berge, H.F.M., 1997. Simulating

genotypic strategies for increasing rice yield potential in irrigated, tropical

environments.FieldCropsRes.5,5–17.

Aggarwal,P.K.,Matthews,R.B.,Kropff,M.J.,1995.Opportunitiesfortheapplicationof

systemsapproaches inplantbreeding. In:Aggarwal,P.K.,Matthews,R.B.,Kropff

M.J., Van Laar, H.H. (Eds.), SARP Research Proceeding. International Rice

ResearchInstitute,LosBaños,Philippines,pp.135–144.

Allard, R.W., Bradshaw, A.D., 1964. Implications of genotype‐environmental

interactionsinappliedplantbreeding.CropSci.4,503–508.

Allen,E.J.,1978.Thepotato,thescientificbasisforimprovement.ChapmanandHall,

London,UK.

Allen,E.J.,Scott,R.K.,1980.Ananalysisofgrowthofthepotatocrop.J.Agric.Sci.94,

583–606.

Allen,E,J.,Scott,R.K.,1992.Principlesofagronomyandtheirapplicationinthepotato

industry. In: Harris, P. (Ed.), The potato crop: The scientific basis for

improvement.ChapmanandHall,London,UK,pp.816–881.

Almekinders, C,J.M., Struik, P.C., 1994. Photothermal response of sympodium

development and flowering in potato (Solanumtuberosum L.) under controlled

conditions.Neth.J.Agric.42,311–329.

Almekinders, C.J.M., Struik, P.C., 1996. Shoot development and flowering in potato

(SolanumtuberosumL.).PotatoRes.39,581–607.

Anothai, J., Patanothai, A., Pannangpetch, K., Jogloy, S., Boote, K.J., Hoogenboom, G.,

2008.Reduction in data collection for determination of cultivar coefficients for

breedingapplication.Agric.Syst.96,195–206.

Araus, J.L., Slafer,G.A.,Royo, C., Serret,M.D., 2008.Breeding for yieldpotential and

stressadaptationincereals.Crit.Rev.PlantSci.27,377–412.

Arshad, M., Ali, N., Ghafoor, A., 2006. Character correlation and path coefficient in

soybeanGlycinemax(L.)Merrill.PakistanJ.Bot.38,121–130.

Asseng, S., Turner, N.C., 2007. Modelling genotype × environment × management

interactionstoimproveyield,wateruseefficiencyandgrainproteininwheat.In:

Spiertz, J.H.J., Struik, P.C., Van Laar, H.H. (Eds.), Scale and complexity in plant

systemsresearch:Gene‐plant‐croprelations.Springer,pp.93–104.

AssociationofOfficialAnalyticalChemistry(AOAC),1984.Officialmethodsofanalysis

of the association of official analytical chemists. 14th ed. Washington D.C., pp.

1141.

Austin, R.B., Ford, M.A., Morgana, C.L., 1989. Genetic improvement in the yield of

References

192

winterwheat:afurtherevaluation.J.Agric.Sci.112,295–301.

Bachem,C.,VanderHoeven,R.,Lucker,J.,Oomen,R.,Casarini,E.,Jacobsen,E.,Visser,

R., 2000. Functional genomic analysis of potato tuber life‐cycle. Potato Res. 43,

297–312.

Baker, R.J., 1986. Selection indices in plant breeding. CRC Press Inc., Boca Raton,

Florida.

Baker, R.J.,1988. Tests for crossover genotype‐environmental interactions.Can. J.

PlantSci.68,405–410.

Ball,S.T.,Konzak,C.F.,1993.Relationshipbetweengrainyieldandremotely sensed

datainwheatbreedingexperiments.PlantBreed.110,277–282.

Bannayan, M., Kobayashi, K., Marashi, H., Hoogenboom, G., 2007. Gene‐based

modellingforrice:Anopportunitytoenhancethesimulationofricegrowthand

development?J.Theor.Biol.249,593–605.

Banterng, P., Patanothai, A., Pannangpetch, K., Jogloy, S., Hoogenboom, G., 2004.

Determination of genetic coefficients of peanut lines for breeding applications.

Eur.J.Agron.21,297–310.

Banterng,P.,Patanothai,A.,Pannangpetch,K.,Jogloy,S.,Hoogenboom,G.,2006.Yield

stability evaluation of peanut lines: A comparison of an experimental versus a

simulationapproach.FieldCropsRes.96,168–175.

Bastiaans,L.,Kropff,M.J.,Kempuchetty,N.,Rajan,A.,Migo,T.R., 1997. Can simulation

modelshelpdesignricecultivarsthataremorecompetitiveagainstweed?Field

CropsRes.51,101–111.

Belanger, G.,Walsh, J.R., Richards, J.E.,Milburn, P.H., Ziadi,N., 2000. Comparison of

three statistical models describing potato yield response to nitrogen fertilizer.

Agron.J.92,902–908.

Beukema,H.P.,VanderZaag,D.E.,1979.Potatoimprovement.Somefactorsandfacts.

InternationalAgricultureCentre,Wageningen,theNetherlands.

Bezant, J., Laurie, D., Pratchett, N., Chojecki, J., Kearsey, M., 1997. Mapping QTL

controllingyield andyield components in a springbarley (Hordeumvulgare L.)

crossusingmarkerregression.Mol.Breed.3,29–38.

Bhatt,G.M.,1972.Significanceofpathcoefficientanalysisindeterminingthenatureof

characterassociation.Euphytica22,338–343.

Biemond,H.,Vos,J.,1992.Effectsofnitrogenonthedevelopmentandgrowthofthe

potatoplant.2.Thepartitioningofdrymatter,nitrogenandnitrate.Ann.Bot.70,

37–45.

Bindraban, P.S., 1997. Bridging the gap between plant physiology and breeding.

Identifying traits to increase wheat yield potential using system approaches.

WageningenUniversity,theNetherlands,pp.145.

References

193

Bodlaender, K.B.A., 1963. Influence of temperature, radiation and photoperiod on

development and yield. In: Ivins, J.D., Milthorpe, E.L. (Eds.), The growth of the

potato.Proceedingsofthe10thEasterSchoolofAgriculturalScience,Universityof

Nottingham,pp.199–210.

Bodlaender,K.B.A.,Reestman,A.J.,1968.Theinteractionofnitrogensupplyandplant

densityinpotatoes.Neth.J.Agric.Sci.16,165–176.

Bond,B.J.,2000.Age‐relatedchangesinphotosynthesisofwoodyplants.TrendsPlant

Sci.5,349–353.

Bonhomme,R.,2000.Basesandlimitstousing‘degreeday’units.Eur.J.Agron.13,1–

10.

Bonierbale,M.,Plaisted,R.L.,Tanksley,S.D.,1988.RFLPmapsbasedonacommonset

ofclonesrevealmodesofchromosomalevolutioninpotatoandtomato.Genetics

120,1095–1103.

Boote,K.J., Jones, J.W.,Hoogenboom,G., Pickering,N.B., 1998.TheCROPGROmodel

for grain legumes. In: Tsuji G.Y. et al., (Eds.), Understanding options for

agriculturalproduction.KluwerAcademicPubl.,Dordrecht,theNetherlands,pp.

99–128.

Boote, K.J., Jones, J.W., Pickering, N.B., 1996. Potential uses and limitation of crop

models.Agron.J.88,704–716.

Boote, K.J., Kropff,M.J., Bindraban, P.S., 2001. Physiology andmodelling of traits in

cropplants:implicationsforgeneticimprovement.Agric.Syst.70,395–420.

Boote,K.J.,Tollenaar,M.,1994.Modelinggeneticyieldpotential. In:Boote,K.J.etal.

(Eds.),Physiologyanddeterminationofcropyield.ASA,CSSA,andSSSA,Madison,

pp.533–565.

Botstein, D., White, R.L., Skolnick, M., Davis, R.W., 1980. Construction of a genetic

linkage map in man using restriction fragment length polymorphisms. Am. J.

Hum.Genet.32,314–331.

Bouman,B.A.M.,1995.Cropmodellingandremotesensingforyieldprediction.Neth.J.

Agric.Sci.43,143–161.

Bouman,B.A.M.,VanKeulen,H.,VanLaar,H.H.,Rabbinge,R.,1996.The‘Schoolofde

Wit’ cropgrowth simulationmodels:Apedigreeandhistorical overview.Agric.

Syst.52,171–198.

Bradshaw, J.E., 1994. Quantitative genetics theory for tetrasomic inheritance. In:

Bradshaw, J.E., Mackay, G.R. (Eds.), Potato Genetics. CAB International,

Wallingford,UK,pp.71–99.

Bradshaw, J.E., Mackay, G.R., 1994. Potato Genetics. Wallingford, UK: CAB

International,pp.552.

Bradshaw,J.,Hackett.C.,Pande,B.,Waugh,R.,Bryan,G.,2008.QTLmappingofyield,

References

194

agronomic and quality traits in tetraploid potato (Solanum tuberosum subsp.

tuberosum).Theor.Appl.Genet.116,193–211.

Bremner, P.M., Radley, R.W., 1966. Studies in potato agronomy. II. The effects of

variety and timeof plantingon growth, development andyield. J.Agric. Sci.66,

253–262.

Brown,C.R.,1993.Originandhistoryofthepotato,Am.PotatoJ.,70,363–374.

Burstall,L.,Harris,P.M.,1983.Theestimationofpercentage light interception from

leafareaindexandpercentagegroundcoverinpotatoes.J.Agric.Sci.100,241–

244.

Burton,W.G.,1989.Thepotato.PotatoRes.32,483–484.

Burton,W.G.,Wilson, A.R., 1970. The apparent effect of the latitude of the place of

cultivationuponthesugarcontentofpotatoesgrowninGreatBritain.PotatoRes.

13,269–283.

Cadersa, Y., Govinde, N., 1999. Relationship between canopy cover and light

interception in potato in a tropical climate. Food and Agricultural Research

Council,Réduit,Mauritius,pp.137–144.

Caligari,P.D.S.,1992.Breedingnewvarieties. In:Harris,P.M.(Ed.),ThePotatoCrop.

Thescientificbasisforimprovement.Chapman&Hall,London,pp.334–372.

Campos,H.,Cooper,M.,Habben,J.E.,Edmeades,G.O.,Schussler,J.R.,2004.Improving

droughttoleranceinmaize:aviewfromindustry.FieldCropsRes.90,19–34.

Cannell, M.G.R., Thornley, J.H.M., 2000. Modelling the components of plant

respiration:someguidingprinciples.Ann.Bot.85,45–54.

Casa, R., Pieruccetti, F., Sgueglia, G., Lo Cascio, B., 2005. Potato tuber quality

improvement through nitrogen management optimisation: Review of

methodologies.ActaHort.684,65–72.

Cassman, K.G., 1999. Ecological intensification of cereal production systems: yield

potential, soil quality, and precision agriculture. Proc. Nat. Acad. Sci., USA96,

5952–5959.

Catchpole, A.H., Hillman, J., 1969. Effect of ethylene on tuber initiation in Solanum

tuberosumL.Nature223,1387.

Ceccarelli, S., Grando, S., 1996. Drought as a challenge for the plant breeder. Plant

GrowthRegul.20,149–155.

Celis‐Gamboa,C.,Struik,P.C.,Jacobsen,E.,Visser,R.G.F.,2003.Temporaldynamicsof

tuber formation and related processes in a crossing population of potato

(Solanumtuberosum).Ann.Appl.Biol.143,175–186.

Changjian, J., Zhao‐Bang, Z., 1995. Multiple trait analysis of genetic mapping for

quantitativetraitloci.Genetics140,1111–1127.

Chapman, S.C., Cooper, M., Hammer, G.L., 2002. Using crop simulation to generate

References

195

genotype by environment interaction effects for sorghum in water‐limited

environments.Aust.J.Agric.Res.53,379–389.

Chapman,S.C.,Crossa,J.,Edmeades,G.O.,1997.Genotypebyenvironmenteffectsand

selectionfordroughttoleranceintropicalmaize.I.Twomodelpatternanalysisof

yield.Euphytica95,1–9.

Charles‐Edwards, D.A., 1976. Shoot and root activities during steady‐state plant

growth.Ann.Bot.40,767–772.

Chenu, K., Chapman, S.C., Tardieu, F., McLean, G., Welcker, C., Hammer, G.L., 2009.

Simulatingtheyieldimpactsoforgan‐levelquantitativetraitlociassociatedwith

droughtresponseinmaize‐a“gene‐to‐phenotype”modelingapproach.Genetics

183,1507–1523.

Cilas, C., Bar‐Hen, A., Montagnon, C., Godin, C., 2006. Definition of architectural

ideotypesforgoodyieldcapacityinCoffeacanephora.Ann.Bot.97,405–411.

Claassens,M.M.J., Vreugdenhil, D., 2000. Is dormancybreaking of potato tubers the

reverseoftuberinitiation?PotatoRes.43,347–369.

Collins,A.,Milbourne,D.,Ramsay,L.,Meyer,R.,Chatot‐Balandras,C.,Oberhagemann,

P.,DeJong,W.,Gebhardt,C.,Bonnel,E.,Waugh,R.,1999.QTLforfieldresistance

to late blight in potato are strongly correlated with earliness and vigour. Mol.

Breed.5,387–398.

Cooper, M., Chapman, S.C., Podlich, D.W., Hammer, G.L., 2002. The GP problem:

Quantifyinggene‐to‐phenotyperelationships.InSilicoBiol.2,151–164.

Cooper,M.,Podlich,D.W.,Smith,O.S.,2005.Gene‐to‐phenotypemodelsandcomplex

traitgenetics.Aust.J.Agric.Res.56,895–918.

Cooper, M., Woodruff, D.R., Eisemann, R.L., Brennan, P.S., DeLacy, I.H., 1995. A

selection strategy to accommodate genotype by environment interaction for

grain yield of wheat: Managed‐environments for selection among genotypes.

Theor.Appl.Genet.90,492–502.

Correll, D.S., 1962. The potato and its wild relatives. Contributions from Texas

ResearchFoundation,BotanicalStudies4,1–606.

Daniell,H.,Dhingra,A.,2002.Multigeneengineering:dawnofanexcitingnewerain

biotechnology.Curr.Opin.Biotech.3,136–141.

D'Arcy,W.G., 1972. Solanaceae Studies II: Typification of Subdivisions of Solanum.

Ann.MOBot.Gard.59,262–278.

DePury,D.G.G.,Farquhar,G.D.,1997.Simplescalingofphotosynthesisfromleavesto

canopieswithouttheerrorsofbig‐leafmodels.PlantCellEnviron.20,537–557.

DeMaine,M.J.,1984.Patternsofvariationinpotatodihaploidfamilies.PotatoRes.27,

1–11.

Dewey,D.R.,Lu,K.H.,1959.Acorrelationandpath‐coefficientanalysisofcomponents

References

196

crestedwheatgrassseedproduction.Agron.J.51,515–518.

Dingkuhn,M., 1996.Modelling concepts for the phenotypic plasticity of drymatter

andnitrogenpartitioninginrice.Agric.Syst.52,383–397.

Donald,C.M.,1968.Thebreedingofcropideotypes.Euphytica17,385−403.

Dunemann,F.,Kahnau,R.,Schmidt,H.,1994.GeneticrelationshipsinMalusevaluated

byRAPD'fingerprinting'ofcultivarsandwildspecies.PlantBreed.113,150–159.

Edmeades, G.O., McMaster, G.S., White, J.W., Campos, H., 2004. Genomics and the

physiologist:bridgingthegapbetweengenesandcropresponse.FieldCropsRes.

90,5–18.

Ellis, R.H., Hadley, P., Roberts, E.H., Summerfield, R.J., 1990. Quantitative relations

betweentemperatureandcropdevelopmentandgrowth.In:Jackson,M.T.,Ford‐

Lloyd, B.V., Parry, M.L. (Eds.), Climatic change and plant genetic resources.

BelhavenPress,London,pp.85–115.

Epstein,E.,1966.Effectofsoiltemperatureatdifferentgrowthstagesongrowthand

developmentofpotatoplants.Agron.J.58,169–171.

Errebhi, M., Rosen, C.J., Gupta, S.C., Birong, D.E., 1998a. Potato yield response and

nitrateleachingasinfluencedbynitrogenmanagement.Agron.J.90,10–15.

Errebhi,M., Rosen, C.J., Lauer, F.I., Martin,M.W., Bamberg, J.B., 1999. Evaluation of

tuber‐bearingSolanumspecies for nitrogen use efficiency and biomass

partitioning.Amer.J.PotatoRes.76,143–151.

Errebhi,M., Rosen, C.J., Lauer, F.I.,Martin,M.W., Bamberg, J.B., Birong, D.E., 1998b.

Screening of exotic potato germplasm for nitrogen uptake and biomass

production.Amer.J.PotatoRes.75,93–100.

Evans,L.T.,1993.Cropevolution,adaptationandyield.CambridgeUniversityPress.

Ewing,E.E.,1981.Heatstressandthetuberizationstimulus.Amer.J.PotatoRes.58,

31–49.

Ewing,E.E.,1985.Cuttingsassimplifiedmodelsofthepotatoplant.In:Li,P.H.(Ed.),

Potatophysiology.AcademicPress,Orlando,Florida,pp.153–207.

Ewing,E.E.,1990.Inductionoftuberizationinpotato.In:Vayda,E.,Park,W.D.(Eds.),

Themolecularandcellularbiologyofthepotato.Wallingford:CABInternational,

pp.25–41.

Ewing, E.E., Struik, P.C., 1992. Tuber formation in potato: induction, initiation and

growth.Hort.Rev.14,89–198.

Fahem,M.,Haverkort,A.J.,1988.Comparisonofthegrowthofpotatocropsgrownin

autumnandspringinNorthAfrica.PotatoRes.31,557–568.

Falconer, D.S., Mackay, T.F.C., 1996. Introduction to Quantitative Genetics. Fourth

edition.AddisonWesleyLongman,Harlow,Essex,UK.

FAO,1996.Food requirements and population growth. Technical Background

References

197

DocumentNo.4.Rome.

FAO, 2009. Proceedings of the expertmeeting on “How to feed theworld in 2050”

24‐26June2009,FAOHeadquarters,Rome.

FAOStatisticsDivision(FAOSTAT),2010.http://faostat.fao.org.

Feibert, E.B.G., Shock, C.C., Saunders, L.D., 1998.Nitrogen fertilizer requirements of

potatoesusingcarefullyscheduledsprinklerirrigation.Hort.Sci.32,262–265.

Fernando,L.H.,1958.Studiesonthe leafgrowth,effectofmineralnutrientsandthe

interdependenceoftheleavesofaplant.PhDthesis,UniversityofLondon.

Feustel,I.C.,1987.Miscellaneousproductsfrompotatoes.In:Talburt,W.F.,Smith,O.

(Eds.),PotatoProceedings.AVIPub.Co,Westport,Conn,pp.727–746.

Firman, D.M., O’Brien, P.J., Allen, E.J., 1995. Appearance and growth of individual

leavesinthecanopiesofseveralpotatocultivars.J.Agric.Sci.125,379–394.

Fleisher,D.H., Shillito, R.M., Timlin,D.J., Kim, S‐H, Reddy, V.R., 2006. Approaches to

modelingpotatoleafappearancerate.Agron.J.98,522–528.

Fleisher,D.H.,Timlin,D.J.,2006.Modelingexpansionofindividualleavesinthepotato

canopy.Agric.ForestMeteorol.139,84–93.

Gallagher, J.N., 1976. The Growth of Cereals in Relation to Weather. PhD thesis,

UniversityofNottingham.

Gallagher,J.N.,1979.Fieldstudiesofcerealleafgrowth:I.Initiationandexpansionin

relationtotemperatureandontogeny.J.Exp.Bot.30,625–636.

Gawronska, H., Dwelle, R.B., Pavek, J.J., 1990. Partitioning of photoassimilates by

potatoplants(SolanumtuberosumL.)asinfluencedbyirradiance.II.Partitioning

patternsby four clones grownunderhighand low irradiance.Am.Potato J. 67,

163–176.

Gebhardt, C., Ritter, E., Barone, A., Debener, T., Walkemeier, B., Schachtschabel, U.,

Kaufmann, H., Thompson, R.D., Bonierbale, M.W., Ganal, M.W., Tanksley, S.D.,

Salamini, F., 1991. RFLP maps of potato and their alignment with the ho‐

meologoustomatogenome.Theor.Appl.Genet.83,49–57.

Geldermann, H., 1975. Investigations on inheritance of quantitative characters in

animalsbygenemarkers.I.Methods.Theor.Appl.Genet.46,319–330.

Goudriaan,J.,1994.Usingtheexpolineargrowthequationtoanalyseresourcecapture.

In:Monteith, J.L., Scott, R.K., Unsworth,M.H. (Eds.), Resource capture by crops.

Nottingham:NottinghamUniversityPress,pp.99–110.

Goudriaan,J.,Monteith,J.L.,1990.Amathematicalfunctionforcropgrowthbasedon

lightinterceptionandleafareaexpansion.Ann.Bot.66,695–701.

Govindakrishan, P.M., Haverkort, A.J., 2006. Ecophysiology and academic

management.In:Gopal,J.,Khurana,S.M.P.(Eds.),Handbookofpotatoproduction,

improvement,andpostharvestmanagement.FoodProductsPress,NewYork,pp.

References

198

179–229.

Green,C.F.,1987.Nitrogennutritionandwheatgrowthinrelationtoabsorbedsolar

radiation.Agric.ForestMeteorol.41,207–248.

Gregory,L.E.,1965.Somefactorsfortuberizationinthepotatoplant.Amer.J.Bot.43,

281–288.

Grevsen,K.,2003.Weedcompetitiveabilityofgreenpeas(PisumsativumL.)affected

byseedingrateandgenotypecharacteristics.Biol.Agric.Hort.21,247–261.

Gu, J., Yin, X., Struik, P.C., Stomph, T.J., Wang, H., 2012. Using chromosome

introgressionlinestomapquantitativetraitlociforphotosynthesisparametersin

rice (Oryzasativa L.) leavesunderdroughtandwell‐watered field conditions. J.

Exp.Bot.63,455–469.

Gutteling, E.W., Doroszuk, A., Riksen, J.A.G., Prokop, Z., Reszka, J., Kammenga, J.E.,

2007. Environmental influence on the genetic correlations between life‐history

traitsinCaenorhabditiselegans.Heredity98,206–213.

Habash,D.Z.,Bernard,S.,Schondelmaier,J.,Weyen,J.,Quarrie,S.A.,2007.Thegenetics

ofnitrogenuseinhexaploidwheat:Nutilisation,developmentandyield.Theor.

Appl.Genet.114,403–419.

Hackett, C.A., Bradshaw, J.E., McNicol, J.W., 2001. Interval mapping of QTLs in

autotetraploidspecies.Genetics159,1819–832.

Haga, E., Weber, B, Jansky, S., 2012. Examination of potential measures of vine

maturityinpotato.Am.J.PlantSci.3,495–505.

Halderson, J.L., Haderlie, L.C., Corsini, D.L., 1985. Potato vine kill: Pulling, chemical

killingandrollingeffectsonyieldandqualityofRussetBurbank.Am.PotatoJ.62,

281–288.

Hammer, G.L., 1998. Crop modelling: current status and opportunities to advance.

ActaHort.456,27–36.

Hammer, G.L., Butler, D., Muchow, R.C., Meinke, H., 1996. Integrating physiological

understanding and plant breeding via crop modelling and optimization. In:

Cooper,M.,Hammer,G.L. (Eds.), Plant adaptation and crop improvement. CAB

International,ICRISAT&IRRI,Wallingford,UK,pp.419–441.

Hammer,G.L.,Chapman,S.C.,VanOosterom,E.,Podlich,D.W.,2005.Traitphysiology

andcropmodellingasa frameworkto linkphenotypiccompexity tounderlying

geneticsystems.Aust.J.Agric.Res.56,947–960.

Hammer,G.L., Chapman, S.C., Snell, P., 1999. Crop simulationmodelling to improve

selection efficiency in plant breeding programs. In: Williamson et al., (Eds.),

proceedingsof9thAssemblyWheatBreedingSocietyofAustralia,pp.79–85.

Hammer,G,L.,Cooper,M.,Tardieu,F.,Welch,S.,Walsh,B.,VanEeuwijk,F,Chapman,S.,

Podlich, D., 2006. Models for navigating biological complexity in breeding

References

199

improvedcropplants.TrendsPlantSci.11,587–593.

Hammer, G.L., Kropff, M.J., Sinclair, T.R., Porter, J.R., 2002. Future contributions of

crop modelling‐‐from heuristics and supporting decision making to

understandinggeneticregulationandaidingcropimprovement.Eur.J.Agron.18,

15–31.

Hammer, G.L., Sinclair, T.R., Chapman, S.C., Van Oosterom, E., 2004. On systems

thinking,systemsbiology,andtheinsilicoplant.PlantPhysiol.134,909–911.

Hammer, G.L., Van Oosterom, E., McLean, G., Chapman, S.C., Broad, I., Harland, P.,

Muchow, R.C., 2010. Adapting APSIM to model the physiology and genetics of

complexadaptivetraitsinfieldcrops.J.Exp.Bot.61,2185–2202.

Hammes, P.S., De Jager, J.A., 1990. Net photosynthetic rate of potato at high

temperature.PotatoRes.33,515–520.

Harris, P.M., 1992. The Potato Crop. In: Harris, P.M. (Ed.), The Scientific Basis for

Improvement.ChapmanandHall,London,UK,pp.1–12.

Haverkort,A.J.,Harris,P.M.,1987.Amodelforpotatogrowthandyieldundertropical

highlandconditions.Agric.ForestMeteorol.39,271–282.

Haverkort, A.J., Kooman, P.L., 1997. The use of systems analysis and modelling of

growthanddevelopment inpotato ideotypingunderconditionsaffectingyields.

Euphytica94,191–200.

Haverkort,A.J.,Rutavisire,C.,1986.Fertilizerutilizationundertropicalconditions.2.

Effect of applying nitrogen, phosphorus and potassium on the relationship

betweeninterceptedradiationandyieldofpotatoesincentralAfrica.PotatoRes.

29,357–365.

Hawkes,J.G.,1990.Thepotato:evolution,biodiversityandgeneticresources.Oxford:

BelhavenPress.

Herndl,M.,Shan,C.,Wang,P.,Graeff,S.,Claupein,W.,2007.Amodelbasedideotyping

approach for wheat under different environmental conditions in North China

plain.Agric.Sci.China6,1426–1436.

Heuvelink, E., Marcelis, L.F.M., Bakker, M.J., Van der Ploeg, A., 2007. Use of crop

growthmodelstoevaluatephysiologicaltraitsingenotypesofhorticulturalcrops.

In:Spiertz,J.H.J.,Struik,P.C.,VanLaar,H.H.(Eds.),ScaleandComplexityinPlant

SystemsResearch:Gene‐Plant‐CropRelations.Springer,pp.223–233.

Hijmans,R.J.,2001.Globaldistributionofthepotatocrop.Am.J.PotatoRes.78,403–

412.

Hilbert, D.W., 1990. Optimization of plant root: shoot ratios and internal nitrogen

concentration.Ann.Bot.66,91–99.

Hittalmani, S.,Huang,N.,Courtois,B.,Venuprasad,R., Shashidhar,H.E., Zhuang, J‐Y.,

Zheng, K‐L., Liu, G‐F., Wang, G‐C., Sidhu, J.S., Srivantaneeyakul, S., Singh, V.P.,

References

200

Bagali,P.G.,Prasanna,H.C.,McLaren,G.,Khush,G.S.,2003.IdentificationofQTL

forgrowth‐andgrainyield‐relatedtraitsinriceacrossninelocationsofAsia.

Theor.Appl.Genet.107,679–690.

Hodges, T., 1991. Crop growth simulation and the role of phenologicalmodels. In:

Hodges,T.(Ed.),Predictingcropphenology.CRCPress,Boston,pp.3–5.

Holland,J.B.,2006.Estimatinggenotypiccorrelationsandtheirstandarderrorsusing

multivariate restricted maximum likelihood estimation with SAS proc mixed.

CropSci.46,642–654.

Holland, J.B., Nyquist, W.E., Cervartes‐Martinez, C.T., 2003. Estimating and

interpreting heritability for plant breeding: an update. Plant Breed. Rev. 22, 2–

112.

Honeycutt, C.W., Clapham,W.M., Leach, S.S., 1996. Crop rotation andN fertilization

effectsongrowth,yield,anddiseaseincidenceinpotato.Amer.J.PotatoRes.73,

45–61.

Hoogenboom,G.,White,J.W.,Acosta‐Gallegos,J.,Gaudiel,R.G.,Myers,J.R.,Silbernagel,

M.J.,1997.EvaluationofaCropSimulationModelthatIncorporatesGeneAction.

Agron.J.89,613–620.

Hoogenboom,G.,White, J.W.,Messina,C.D.,2004.Fromgenometocrop: integration

throughsimulationmodeling.FieldCropsRes.90,145–163.

Houle,D.,1992.Comparingevolvabilityandvariabilityofquantitativetraits.Genetics

130,195–204.

Huang,J.,Pray,C.,Rozelle,S.,2002.Enhancingthecropstofeedthepoor.Nature418,

678–684.

Humphries,E.C.,French,S.A.W.,1963.Theeffectsofnitrogen,phosphorus,potassium

andgibberellic acidon leaf areaandcelldivision inMajesticpotato.Ann.Appl.

Biol.52,149–162.

Hunt,L.A.,1993.Designingimprovedplanttypes:abreedersview‐point.In:Penning

de Vries, F.W.T., Teng, P.S., Metselaar, K. (Eds.), Systems approaches for

agriculturaldevelopment.KluweracademicPublishers,theNetherlands,pp.3–17.

Hunt, L.A., Pararajasingham, S., Jones, J.W., Hoogenboom, G., Imamura, D.T., Ogoshi,

R.M.,1993.GENCALC–softwaretofacilitatetheuseofcropmodelsforanalyzing

fieldexperiments.Agron.J.87,1090–1094.

Hurtado,P.X.,Schnabel,S.K.,Zaban,A.,Veteläinen,M.,Virtanen,E.,Eilers,P.H.C.,Van

Eeuwijk,F.A.,Visser,R.G.F.,Maliepaard,C.,2012.Dynamicsofsenescence‐related

QTLsinpotato.Euphytica183,189–302.

Hutten,R.C.B.,1994.Basicaspectsofpotatobreedingviathediploidlevel.PhDthesis,

WageningenUniversity,ISBN9054852925.

Ingram,K.T.,McCloud,D.E.,1984.Simulationofpotatocropgrowthanddevelopment.

References

201

CropSci.24,21–27.

Ivins,J.D.,Bremner,P.M.,1965.Growth,developmentandyieldinthepotato.Outlook

onAgric.4,211–217.

Jackson, S.D., 1999.Multiple signalling pathways control tuber induction in potato.

PlantPhysiol.119,1−8.

Jackson, P., Robertson,M., Cooper,M., Hammer, G., 1996. The role of physiological

understanding inplant breeding; fromabreedingperspective. FieldCropsRes.

49,11–37.

Jansen,R.C.,1993.IntervalMappingofMultipleQuantitativeTraitLoci.Genetics135,

205–211.

Jansen,R.C.,1995.Geneticmappingquantitativetraitlociinplants:anovelstatistical

approach.PhDdissertation,WageningenUniversity,theNetherlands.

Jansen,R.C.,Stam,P.,1994.HighResolutionofQuantitativeTraitsIntoMultipleLoci

viaIntervalMapping.Genetics136,1447–1455.

Jefferies,R.A.,MacKerron,D.K.L.,1993.Responsesofpotatogenotypestodrought.II.

Leafareaindex,growthandyield.AnnalsofAppliedBiology122,105–112.

Jeuffroy, M.H., Ney, B., Ourry, A., 2002. Integrated physiological and agronomic

modellingofNcaptureandusewithintheplant.J.Exp.Bot.53,809–823.

Jiang,C.,Zeng,Z.‐B.,1995.Multipletraitanalysisofgeneticmappingforquantitative

traitloci.Genetics140,1111–1127.

Joenje,W.,Kropff,M.J.,1987.Relativetimeofemergence,leafareadevelopmentand

plantheightasmajorfactorsincrop–weedcompetition.BritishCropProtection

Conference,BritishCropProtectionCouncil,Brighton,UK,pp.971–978.

Johnson, H.W., Robinson, H.F., Comstock, R.E., 1955. Genotype and phenotype

correlationsinsoybeansandtheirimplicationinselection.Agron.J.47,477–483.

Johnson,C.L.,Tindall,T.A.,Thornton,M.,Brooks,R.A.,1995.PetioleNO3‐Nsufficiency

curvesinnewlydevelopedpotatocultivars.Proc.Univ.IdahoWinterCommodity

Schools27,209–216.

Jones,A.,Dukes,P.D.,Schalk,J.M.,1986.Sweetpotatobreeding.In:Basset,M.J.,(Ed.)

Breedingvegetablecrops.AVIPublishingCompany,Inc,Connecticut,pp.1–35.

Kang, M.S., 1998. Using genotype‐by‐environment interaction for crop cultivar

development,Adv.Agron.62,199–252.

Kang,M.S.,Pham,H.N.,1991.Simultaneousselectionforhighyieldingandstablecrop

genotypes.Agron.J.83,161–165.

Kearsey, M.J., Farquhar, A.G.L., 1998. QTL analysis in plants; where are we now?

Heredity80,137–142.

Khurana, S.C.,McLaren, J.S., 1982. The influence of leaf area, light interception and

seasononpotatogrowthandyield.PotatoRes.25,329–342.

References

202

Khurana, S.M.P., Minhas, J.S., Pandey, S.K., 2003. The potato: production and

utilizationinsubtropics.MehtaPublishers,NewDelhi,India.

Khush,G.S.,1987.Ricebreeding:Past,presentandfuture.J.Genet.66,195–216.

Kirk,W.W.,Marshall,B.,1992.Theinfluenceoftemperatureonleafdevelopmentand

growthinpotatoesincontrolledenvironments.Ann.Appl.Biol.120,511–525.

Kleinkopf, G.E., Westeman, D.T., Dwelle, R.B., 1981. Dry matter production and

nitrogenutilizationbysixpotatocultivars.Agron.J.73,799–802.

Kloosterman, B., Vorst, O., Hall, R.D., Visser, R.G.F., Bachem, C.W., 2005. Tuber on a

chip: differential gene expression during potato tuber development. Plant

Biotechnol.J.3,505–519.

Knight,J.,2003.Cropimprovement:Adyingbreed.Nature421,568–570.

Kolbe, H, Stephan‐Beckmann, S., 1997. Development, growth and chemical

compositionofthepotatocrop(SolanumtuberosumL.).II.Tuberandwholeplant.

PotatoRes.40,135–153.

Kooman,P.L.,1995.Yieldingabilityofpotatocropsasinfluencedbytemperatureand

daylength. Ph.D. Thesis. Wageningen Agricultural University, Wageningen, the

Netherlands.

Kooman, P.L., Fahem, M., Tegera, P., Haverkort, A.J., 1996. Effects of climate on

differentpotatogenotypes.1.Radiationinterception,totalandtuberdrymatter

production.Eur.J.Agron.5,193–205.

Kooman,P.L.,Haverkort,A.J.,1995.Modellingdevelopmentandgrowthofthepotato

crop influencedbytemperatureanddaylength:LINTUL‐POTATO. In:Haverkort,

A.J., MacKerron, D.K.L. (Eds.), Potato Ecology and Modelling of Crops Under

ConditionsLimitingGrowth.KluwerAcademicPublishers,Dordrecht,pp.41–60.

Kooman, P.L., Rabbinge, R., 1996. An analysis of the relation between dry matter

allocationtothetuberandearlinessofapotatocrop.Ann.Bot.,77,235–242.

Kooman,P.L.,Spitters,C.J.T.,1995.Coherentsetofmodelstosimulatepotatogrowth.

In: Kabat, P., Marshall, B., Van den Broek, B.J., Vos, J., Van Keulen, H. (Eds.),

Modelling and parameterization of the soil‐plant‐atmosphere system. A

comparisonofpotatogrowthmodels,WageningenPress,Wageningen,pp.253–

274.

Kotch, G.P., 1987. The production of haploids and their use in genetic studies in

potatoes.UniversityofWisconsin‐Madison,pp.342.

Kroonberg,P.M.,1997.IntroductiontobiplotsforG×Etables.ResearchreportNo.51.

CentreforStatistics,theUniversityofQueensland,Brisbane,Queensland.

Kropff, M.J., Goudriaan, J., 1994. Competition for resources capture in agricultural

crops.In:Monteith,R.K.Scott.,Unsworth,M.H.(Eds.),Resourcecapturebycrops.

NottinghamUniversityPress,Loughborough,Leicestershire,pp.233–253.

References

203

Kropff, M.J., Haverkort, A.J., Aggarwal, P.K., Kooman, P.L., 1995. Using systems

approaches to design and evaluate ideotypes for specific environments. In:

Bouma,J.,Kuyvenhoven,A.,Bouman,B.A.M.,Luyten,J.C.,Zandstra,H.G.(Eds.),

Eco‐regional approaches for sustainable land use and food production, Kluwer,

Dordrecht,pp.417–435.

Lander, E.S., Botstein, D., 1989.MappingMendelian factors underlying quantitative

traitsusingRFLPlinkagemaps.Genetics121,185–199.

Lark, K.G., Chase, K., Adler, F., Mansur, L.M., Orf, J.H., 1995. Interactions between

quantitative trait loci in soybean in which trait variation at one locus is

conditionaluponspecificalleleatanother.Proc.Natl.Acad. Sci.USA.92,4656–

4660.

Lauer,D.A.,1986.ResponseofNooksackpotatoestonitrogenfertilizer.Am.PotatoJ.

63,251–262.

Lemaire,G.,Gastal,F.,1997.Nuptakeanddistributioninplantcanopies.In:Lemaire,

G.(Ed.),Diagnosisofnitrogenstatusincrops.Springer‐Verlag,Berlin,pp.3–41.

Letort, V., Mahe, P., Cournede, P.H., De Reffye, P., Courtois, B., 2007. Quantitative

genetics and functional‐structural plant growth models: Simulation of

quantitativetraitlocidetectionformodelparametersandapplicationtopotential

yieldoptimization.Ann.Bot.101,1243–1254.

Li,C.C.,1975.Pathanalysis‐aprimer.TheBoxwoodPress.PacificGrove,CA.

Li,P.H.,1985.PotatoPhysiology.AcademicPress,NewYork.

Li,H.,Parent,L.E.,Karam,A.,2006.Simulationmodelingofsoilandplantnitrogenuse

inapotatocroppingsysteminthehumidandcoolenvironment.Agric.Ecosyst.

Environ.115,248–260.

Li, Z.K., Luo, L.J.,Mei, H.W.,Wang, D.L., Shu, Q.Y., Tabien, R., Zhong, D.B., Ying, C.S.,

Stansel,J.W.,Khush,G.S.,Paterson,A.H.,2001.Overdominantepistaticlociarethe

primarygeneticeffectsofinbreedingdepressionandheterosisinrice.I.Biomass

andgrainyield.Genetics158,1737–1753.

Littell,R.C.,Milliken,G.A.,Stroup,W.W.,Wolfinger,R.D.,1996.SASsystemformixed

models.SASInstitute,Cary,N.C.

Liu,W.T.H.,Botner,D.M.,Sakamoto,C.M.,1989.ApplicationofCERES‐maizetoyield

predictionofaBrazilianmaizehybrid.Agric.ForestMeteorol.45,299–312.

Loomis,R.S.,Rabbinge,R.,Ng,E.,1979.Explanatorymodelsincropphysiology.Annu.

Rev.PlantPhysiol.30,339–367.

Lotz,L.,Groeneveld,R.M.V.,DeGroot,N.,1991.Potentialforreducingherbicideinputs

in sugar beet by selecting early closing cultivars. Proceedings Brighton Crop

ProtectionConference,Weeds,BritishCropProtectionCouncil,Brighton,UK,pp.

1241–1248.

References

204

Love, S.L., 1999. Founding clones, major contributing ancestors, and exotic

progenitorsofprominentNorthAmericancultivars,Am. J.PotatoRes.,76,263–

280.

Luo,L.J., Li,Z.K.,Mei,H.W., Shu,Q.Y.,Tabien,R.,Zhong,D.B.,Ying,C.S., Stansel, J.W.,

Khush, G.S., Paterson, A.H., 2001. Overdominant epistatic loci are the primary

genetic effects of inbreeding depression and heterosis in rice. II. Grain yield

components.Genetics158,1755–1771.

Lynch,M.,Walsh,B.,1998.GeneticsandAnalysisofQuantitativeTraits.Sunderland,

MA,SinauerAssociates,Inc.,pp.980.

MacKerron,D.K.L.,Waister,P.D.,1985a.Asimplemodelofpotatogrowthandyield.

Part 1.Model development and sensitivity analysis. Agric. ForestMeteorol. 34,

241–252.

MacKerron,D.K.L.,Waister,P.D.A.,1985b.Asimplemodelofpotatogrowthandyield.

PartII.Validationandexternalsensitivity.Agric.ForestMeteorol.34,285–300.

Maier,N.A.,Dahlenburg,A.P.,Williams,C.M.J., 1994.Effectsofnitrogen,phosphorus

and potassium on yield, specific gravity, crisp colour and tuber‐chemical

compositionofpotato(SolanumtuberosumL.)cv.Kennebec.Aust.J.Exp.Agric.34,

813–824.

Manrique, L.A., Tsuji, G.Y., Uehara, G., Fox, R.L., 1984. Winter and summer

performance of potato (Solanum tubersosum L.) in isohyperthermic regimes.

Amer.J.PotatoRes.61,41–56.

Marcelis,L.F.M.,Heuvelink,E.,Goudriaan,J.,1998.Modellingbiomassproductionand

yieldofhorticulturalcrops:areview.ScientiaHort.74,83–111.

Marinus, J., Bodlaender, K.B.A., 1975. Response of some potato varieties to

temperature.PotatoRes.18,189–204.

Marti, J.,Bort, J., Slafer,G.A.,Araus, J.L., 2007.Canwheatyieldbeassessedbyearly

measurements of normalized difference vegetation index? Ann. Appl. Biol. 150,

253–257.

Martin, R.J., 1995. The effect of nitrogen fertilizer on the recovery of nitrogenby a

potatocrop.In:Proc.Ann.Conf.Agron.Soc.NewZealand25,97–104.

Mavromatis, T., Boote,K.J., Jones, J.W., Irmak,A., Shinde,D.,Hoogenboom,G., 2001.

Developinggenetic coefficients for cropsimulationmodelswithdata fromcrop

performancetrials.CropSci.4,41–51.

Mavromatis, T., Boote, K.J., Jones, J.W., Wilkerson, G.G., Hoogenboom, G., 2002.

Repeatability of model genetic coefficients derived from soybean performance

trailsacrossdifferentstates.CropSci.42,76–89.

Mayes, S., Parsley, K., Sylvester‐Bradley, R., May, S., Foulkes, J., 2005. Integrating

genetic information into plant breeding programmes: how will we produce

References

205

varieties from molecular variation, using bioinformatics? Ann. Appl. Biol. 146,

223–237.

McCown,R.L.,Hammer,G.L.,Hargreaves,J.N.G.,Holzworth,D.P.,Freebairn,D.M.,1996.

APSIM: a novel software system for model development, model testing and

simulationinagriculturalsystemsresearch.Agric.Syst.50,255–271.

Messina, C., Hammer, G., Dong, Z., Podlich, D., Cooper, M., 2009. Modelling crop

improvementinaGEMframeworkviagene‐traitphenotyperelationships. In:Sadras, V.O., Calderini, D. (Eds.), Crop physiology: applications for genetic

improvementandagronomy.Elsevier,theNetherlands,pp.235–265.

Messina, C.D., Jones, J.W., Boote, K.J., Vallejos, C.E., 2006. A gene‐based model to

simulatesoybeandevelopmentandyieldresponsestoenvironment.CropSci.46,

456–466.

Meyer,K.,1985.Maximumlikelihoodestimationofvariancecomponentsofvariance

components for a multivariate mixed model with equal design matrices.

Biometrics41,153–165.

Meyer, R.D., Marcum, D.B., 1998. Potato yield, petiole nitrogen, and soil nitrogen

responsetowaterandnitrogen.Agron.J.90,420–429.

Meyer, R.C., Milbourne, D., Hackett, C.A., Bradshaw, J.E., McNichol, J.W., Waugh, R.,

1998. Linkage analysis in tetraploid potato and association of markers with

quantitative resistance to late blight (Phytophthora infestans). Mol. Gen. Genet.

259,150–160.

Miflin,B.,2000.Cropimprovementinthe21stCentury.J.Exp.Bot.51,1–8.

Milford,G.F.J.,Pocock,T.O.,Riley,J.J.,Messem,A.B.,1985.Ananalysisofleafgrowthin

sugarbeet.III.Leafexpansioninfieldcrops.Ann.Appl.Biol.106,187–203.

Milford,G.F.J.,Riley, J.,1980.Theeffectof temperatureonleafgrowthofsugarbeet

varieties.Ann.Appl.Biol.94,431–443.

Mingo‐Castel, A.M.,Negm, F.B.Young, R.E., Smith, O.E.,1976.Kinetin induced

tuberization of potato in vitro: on the mode of action of kinetin.Plant Cell

Physiol.17,557–570.

Moll,A.,1983.PhotosyntheserateandEtragsleistungvonKartoffelklonen.PotatoRes.

26,191–202.

Moll,A.,Klemke,T.,1990.Asimplemodel for theevaluationofhaulmcharacters in

potatobreeding.ArchivfürZüchtungsforschung20,151–158.

Monteith, J.L.,1977.Climateand theefficiencyofcropproduction inBritain.Philos.

TransR.Soc.Lond.281,277–294.

Moorby, J.,Milthorpe,F.L.,1975.Potato.In:Evans,L.T.(Ed.),CropPhysiology:Some

casehistories.CambridgeUniversityPress,London,pp.225–257.

Morandini,P.,Salamini,F.,2003.Plantbiotechnologyandbreeding:alliedforyearsto

References

206

come.TrendsPlantSci.8,70–75.

Morrell, C.H., 1998. Likelihood ratio testing of variance components in the linear

mixedeffectsmodelusingrestrictedmaximumlikelihood.Biometrics54,1560–

1568.

Muchow, R.C., Davis, R., 1988. Effect of nitrogen supply on the comparative

productivity of maize and sorghum in a semi‐arid tropical environment. II.

Radiationinterceptionandbiomassaccumulation.FieldCropsRes.18,17–30.

Müller‐Röber, B., Sonnewald, U., Willmitzer, L., 1992. Inhibition of ADP‐glucose

pyrophosphorylase in transgenic potatoes leads to sugar‐storing tubers and

influencestuberformationandexpressionoftuberstorageproteingenes.EMBOJ.

11,1229–1238.

NetherlandsPotatoConsultativeFoundation(NIVAP),2007.NetherlandsCatalogueof

PotatoVarieties.TheHague,pp.288.

Nigam S.N., Chandra, S., Sridevi, K.R., Bhukta, M., Reddy, A.G.S., Rachaputi, N.R.,

Wright,G.C.,Reddy,P.V.,Deshmukh,M.P.,Mathur,R.K.,Basu,M.S.,Vasundhara,S.,

Varman, P.V., Nagda, A.K., 2005. Efficiency of physiological trait‐based and

empirical selection approaches for drought tolerance in groundnut. Ann. Appl.

Biol.146,433–439.

Nyquist,W.E.,1991.Estimatesofheritabilityandpredictionofselectionresponsein

plantpopulations.Crit.Rev.PlantSci.,10,235–322.

Oberhagemann, P., Chatot‐Balandras, C., Bonnel, E., Schäfer‐Pregl, R., Wegener, D.,

Palomino, C., Salamini, F., Gebhardt, C., 1999. A genetic analysis of quantitative

resistancetolateblightinpotato:towardsmarkerassistedselection.Mol.Breed.

5,399–415.

O’Brien, P.J., Allen, E.J., Firman, D.M., 1998. A review of some studies into tuber

initiationinpotato(Solanumtuberosum)crops.J.Agric.Sci.130,251–270.

Omoigui,L.O.,Ishiyaku,M.F.,Kamara,A.Y.,Alabi,S.O.,Mohammed,S.G.,2006.Genetic

variabilityandheritabilitystudiesofsomereproductivetraits incowpea(Vigna

unguiculata(L.)Walp.).Afr.J.Biotechnol.5,1191–1195.

Ong, C.K., 1983. Response to temperature in a stand of pearl millet (Pennisetum

typhoides).I.Vegetativedevelopment.J.Exp.Bot.34,322–336.

Oparka, K.J., Davies, H.V., 1985. Translocation of assimilates within and between

potatostems.Ann.Bot.56,45−54.

Orf, J.H., Chase, K., Adler, F.R., Mansur, L.M., Lark, K.G., 1999. Genetics of soybean

agronomictraits:II.Interactionsbetweenyieldquantitativetraitlociinsoybean.

CropSci.39,1652–1657.

Palanisamy,S.,Aggarwal,P.K.,Thiyagarajan,T.M.,Ranganathan,T.B.,1995.Simlating

yields and ranking of rice genotype in multi‐location trials. In: Aggarwal, P.K.,

References

207

Matthews, R.B., Kropff, M.J., Van Laar, H.H. (Eds.), SARP Research Proceeding.

InternationalRiceResearchInstitute,LosBaños,Philippines,pp.91–95.

Park, W.D., 1990. Molecular approaches to tuberization in potato. In: Vayda, M.E.,

Park, W.D. (Eds.), The molecular and cellular biology of the potato. Redwood

Press,Melksham,UK,pp.44–55.

Pashiardis,S.M.,1988.Improvementsofpotatoyields.ActaHort.214,27–45.

Paterson,A.H.,Lander,E.S.,Hewitt,J.D.,Peterson,S.,Lincoln,S.E.,1988.Resolutionof

quantitativetraitsintoMendelianfactorsbyusingacompleteRFLPlinkagemap.

Nature355,721–728.

Payne, R.W.,Murray, D.A., Harding, S.A., Baird, D.B., Soutar, D.M., 2009. GenStat for

Windows(12thEdition)Introduction.VSNInternational,Hemel,Hempstead.

PenningdeVries, F.W.T., 1991. Improving yields: designing and testingVHYVs. In:

SystemsSimulationsatIRRI.IRRIRes.Pap.Ser.No.151,pp.13–19.

Perumal,N.K.,Sahota,T.S.,1986.Investigationsongrowthandtuberizationofpotato

atdifferentplantingdatesandnitrogenlevels.Int.J.Trop.Agric.4,63–72.

Piluzza,G.,Pecetti,L.,Bullitta,S.,Piano,E.,2005.Discriminationamongsubterranean

clover (Trifolium subterraneum L. complex) genotypes using RAPD markers.

Genet.Resour.Crop.Evol.52,193–199.

Piper,E.L.,Boote,K.J.,Jones,J.W.,1998.Evaluationandimprovementsofcropmodels

usingregionalcultivartrialdata.Trans.ASAE.14,435–446.

Porter, G.A., Sisson, J.A., 1991a. Petiole nitrate content of Maine‐grown 'Russet

Burbank'and 'Shepody'potatoes in response tovaryingnitrogenrates.Amer. J.

PotatoRes.68,493–505.

Porter,G.A.,Sisson,J.A.,1991b.Responseof'RussetBurbank'and'Shepody'potatoes

tonitrogenfertilizerintwocroppingsystems.Amer.J.PotatoRes.68,425–443.

Price, A.H., Cairns, J.E., Horton, P., Jones, H.G., Griffiths, H., 2002. Linking drought‐

resistance mechanisms to drought avoidance in upland rice using a QTL

approach:progressandnewopportunities to integrate stomatal andmesophyll

responses.J.Exp.Bot.53,989–1004.

Pushkarnath,1978.Potatoinsub‐tropics.OrientLongmanLimited,NewDelhi.

Quilot,B.,Kervella,J.,Génard,M.,Lescourret,F.,2005.Analysingthegeneticcontrolof

peach fruit quality through an ecophysiological model combined with a QTL

approach.J.Exp.Bot.56,3083–3092.

Ram, S.G., Parthiban, K.T., Kumar, R.S., Thiruvengadam, V., Paramathma, M., 2008.

Genetic diversity among Jatropha species as revealedbyRAPDmarkers. Genet.

Resour.CropEvol.55,803–809.

Raven, J.A.,Andrews,M.,Quigg,A., 2005.Theevolutionofoligotrophy: implications

forthebreedingofcropplantsforlowinputagriculturalsystems.Ann.Appl.Biol.

References

208

146,261–280.

Reymond, M., Muller, B., Leonardi, A., Charcosset, A., Tardieu, F., 2003. Combining

quantitative trait loci analysis and an ecophysiological model to analyze the

genetic variability of the responses of maize leaf growth to temperature and

waterdeficit.PlantPhysiol.131,664–675.

Reymond, M., Muller, B., Tardieu, F., 2004. Dealing with genotype×environment

interactionviaamodelingapproach:acomparisonofQTLsofmaizeleaflengthor

widthwithQTLsofmodelparameters.J.Exp.Bot.55,2461–2472.

Reynolds, M.P., Pellegrineschi, A., Skovmand, B., 2005. Sink‐limitation to yield and

biomass:asummaryofsomeinvestigationsinspringwheat.Ann.Appl.Biol.146,

39–49.

Reynolds,M.P., Rajaram, S., Sayre, K.D., 1999. Physiological and genetic changes of

irrigatedwheatinthepost‐greenrevolutionperiodandapproachesformeeting

projectedglobaldemand.CropSci.39,1611–1621.

Ribaut,J.‐M.,Hoisington,D.,1998.Marker‐assistedselection:newtoolsandstrategies.

TrendsPlantSci.3,236–239.

Richards, R.A., 1996. Increasing the yield potential of wheat: Manipulating sources

and sinks. In: Renolds, M.P., Rajaram, S., McNab, A. (Eds.), Increasing yield

potentialinwheat:Breakigthebarriers.CIMMYT.MexicoD.F.,pp.134–149.

Ritchie, J.T., 1991. Specifications of the ideal model for predicting crop yield. In:

Muchow,R.C., Bellamy, J.A. (Eds.), ClimaticRisk inCropProduction:Model and

Management for the Semi‐arid Tropic and Subtropic. CAB International,

Wallingford,UK,Intl.Symp.S.Lucia,Brisbane,Queensland,Australiapp.97–122.

Roberts,E.H.,Qi,A., Ellis,R.H., Summerfield,R.J., Lawn,R.J., Shanmugasundaram, S.,

1996.Useoffieldobservationstocharacterizegenotypicfloweringresponsesto

photoperiodandtemperature:asoybeanexemplar.Theor.Appl.Genet.93,519–

533.

Rouppe van der Voort, J.,Wolters, P., Folkertsma, R., Hutten, R., Van Zandvoort, P.,

Vinke,H.,Kanyuka,K.,Bendahmane,A.,Jacobsen,E.,Janssen,R.,Bakker,J.,1997.

Mappingof the cystnematode resistance locusGpa2 inpotatousing a strategy

basedoncomigratingAFLPmarkers.Theor.Appl.Genet.95,874–880.

Rowe, R.C., 1993. PotatoHealthManagement. Department of Plant Pathology, Ohio

StateUniversity,Wooster,Ohio,USA.

Russell,W.A.,1993.AchievementsofmaizebreedersinNorthAmerica.In:Baenziger,

P.S. (Ed.), International Crop Science. I. Crop Science Society of America, Inc.,

Madison,WI,pp.225–233.

Samonte,S.O.,Wilson,L.T.,McClung,A.,1998.Pathcoefficientanalysisofyieldand

yield‐relatedtraitsoffifteendiversericegenotypes.CropSci.38,1130–1136.

References

209

Santeliz, G., Ewing, E.E., 1981. Effects of nitrogen fertilization on growth and

developmentofpotatoes.Amer.J.PotatoRes.58,517–518.

SAS Institute Inc.,2004.SAS9.1.3HelpandDocumentation,SAS Institute Inc.,Cary,

NC.

Sattelmacher, B., Kuene, R.,Malagamba, P.,Moreno, U.B., 1990. Evaluation of tuber

bearing Solanum species belonging to different ploidy levels for its yielding

potentialatlowsoilfertility.PlantandSoil129,227–233.

Schäfer‐Pregl, R., Ritter, E., Concilio, L., Hesselbach, J., Lovatti, L., Walkemeier, B.,

Thelen,H.,Salamini,F.,Gebhardt,C.,1998.Analysisofquantitativetraitloci(QTLs)

andquantitativetraitalleles(QTAs)forpotatotuberyieldandstarchcontent.

Theor.Appl.Genet.97,834–846.

Schittenhelm, S., Sourell, H., Lopmeier, F.J., 2006. Drought resistance of potato

cultivarswithcontrastingcanopyarchitecture.Eur.J.Agron.24,193–202.

Scott,G.J.,2002.Maps,modelsandmuddles:World trendsandpatterns inpotatoes

revisited.PotatoRes.45,45–77.

Scott, R.K., Wilcockson, S.J., 1978. Application of physiological and agronomic

principles to the development of the potato industry. In:Harris, P.M. (Ed.), The

potatocrop:thescientificbasisforimprovement.Chapman,London,pp.678–704.

Sharifi,M.,Zebarth,B.J., Coleman,W.,2007. Screening fornitrogen‐useefficiency in

potatowitharecirculatinghydroponicsystem.Commun.SoilSci.PlantAnal.38,

359–370.

Sharifi,M.,Zebarth,B.J.,Hajabbasi,M.A.,Kalbasi,M.,2005.Drymatterandnitrogen

accumulation and rootmorphological characteristics of two clonal selection of

Russet Norkotah potato as affected by nitrogen fertilization.J. Plant Nutri. 28,

2243–2253.

Shekhawat,G.S.,Khurana,P.S.M.,Pandey,S.K.,Chandla,V.K.,1994.Potato:Present&

Future.Proceedingsof theNationalSymposiumheldatModipuramduring1–3,

March, 1993). Indian Potato Association, Central Potato Research Institute,

Shimla‐171001,HP,India,pp.347.

Shorter,R.,Lawn,R.J.,Hammer,G.L.,1991.Improvinggenotypicadaptationincrops:a

roleforbreeders,physiologistsandmodellers.Exp.Agric.27,155–175.

Sibma, L., 1970. Relation between total radiation and yield of some field crops in

Netherlands.Neth.J.Agric.Sci.18,125–131.

Sibma,L.,1977.MaximizationofarablecropyieldsintheNetherlands.Neth.J.Agric.

Sci.25,278–287.

Sinclair,T.R.,DeWit, C.T., 1975.Photosynthateandnitrogen requirements for seed

productionbyvariouscrops.Science189,565–567.

Sinclair, T.R., Horie, T., 1989. Leaf nitrogen, photosynthesis, and crop radiation use

References

210

efficiency:Areview.CropSci.29,90–98.

Sinclair,T.R.,Purcell,L.C.,Sneller,C.H.,2004.Croptransformationandthechallenge

toincreaseyieldpotential.TrendsPlantSci.9,70–75.

Slafer, G.A., 2003. Genetic basis of yield as viewed from a crop physiologist's

perspective.Ann.Appl.Biol.142,117–128.

Slafer,G.A.,Satorre,E.H.,Andrade,F.H.,1994.Increasesingrainyieldinbreadwheat

frombreedingandassociatedphysiologicalchanges.In:Slafer,G.A.(Ed.),Genetic

improvementoffieldcrops.MarcelDekker,NewYork,pp.1–68.

Sokal,R.R.,Rohlf,F.J.,1995.Biometry.Freeman,SanFrancisco.

Somerville,C.,Somerville,S.,1999.Plantfunctionalgenomics.Science285,380–383.

Spitters,C.J.T.,1988.Ananalysisofvariationinyieldamongpotatocultivarsinterms

of lightabsorption, lightutilizationanddrymatterpartitioning.ActaHort.214,

71–84.

Spitters, C.J.T., Schapendonk, A.H.C.M., 1990. Evaluation of breeding strategies for

droughttoleranceinpotatobymeansofcropgrowthsimulation.PlantSoil,123,

193–203.

Spitters,C.J.T.,VanKeulen,H.,VanKraalingen,D.W.G.,1989.Asimpleanduniversal

crop growth simulator: SUCROS87. In: Rabbinge, R., Ward, S.A., Van Laar, H.H.

(Eds.), Simulation and systems management in crop protection. Simulation

Monographs,Pudoc,Wageningen,theNetherlands.

Spooner, D.M., McLean, K., Ramsay, G., Waugh, R., Bryan, G.J., 2005. A single

domestication for potato based on multilocus amplified fragment length

polymorphismgenotyping.Proc.NatlAcad.Sci.102,14694–14699.

Stam, P., 1996. Quantitative genetics: a moving frontier to the benefit of crop

production. In: Annual Report for 1995, The C. T. deWit Graduate School for

ProductionEcology,WageningenAgriculturalUniversity,theNetherlands,pp.6–

9.

Stam,P.,VanOoijen,J.W.,1995.JoinMap™version2.0:softwareforthecalculationof

geneticlinkagemaps.Wageningen:CPRO‐DLO.

Stol,W.,DeKoning,G.H.L.,Kooman,P.L.,Haverkort,A,J.,VanKeulen,H.,Penningde

Vries,F.W.T.,1991.Agroecologicalcharacterizationforpotatoproduction,CABO

report,155,CABO,Wageningen,pp.53.

Struik,P.C.,2007.ResponsesofthePotatoPlanttoTemperature.In:Vreugdenhil,D.,

Bradshaw, J., Gebhardt, C., Govers, F., Taylor, M., MacKerron, D. (Eds.), Potato

BiologyandBiotechnology:AdvancesandPerspectives,Elsevier,pp.367–393.

Struik,P.C.,2010.Canphysiologyhelpustocombatlateblightinpotato?PotatoRes.

53,277–287.

Struik, P.C., Ewing, E.E., 1995. Crop physiology of potato (Solanum tuberosum):

References

211

responses to photoperiod and temperature relevant to crop modelling. In:

Haverkort,A.J.,MacKerron,D.K.L. (Eds.),Ecologyandmodellingofpotatocrops

under conditions limiting growth. Kluwer Academic Publishers, Dordrecht, pp.

19–40.

Struik,P.C.,Geertsema, J., Custers, C.H.M.G., 1989a.Effectsof shoot, root and stolon

temperatureon thedevelopmentof thepotato (SolanumtuberosumL.) plant: I.

Developmentofthehaulm.PotatoRes.32,133–141.

Struik,P.C.,Geertsema, J., Custers,C.H.M.G.,1989b.Effectsof shoot, rootandstolon

temperatureonthedevelopmentofthepotato(SolanumtuberosumL.)plant.III.

Developmentoftubers.PotatoRes.32,151–158.

Struik,P.C.,Haverkort,A.J.,Vreugdenhil,D.,Bus,C.B.,Dankert,R.,1990.Manipulation

oftuber‐sizedistributionofapotatocrop.PotatoRes.33,417–432.

Struik, P.C., Kerckhoffs, L.H.J., 1991. Temperature effects on the development of

shoots, stolons, tubers and roots in Solanumtuberosum L. Proc. Golden Jubilee

Symp.Gen.Res.Educ.:Currenttrendsandthenextfiftyyears.

Struik,P.C.,VanHeusden,E.,Burger‐Meijer,K.,1988.Effectsofshortperiodsoflong

days on the development, yield and size distribution of tubers of Solanum

tuberosumL.cv.Bintje.Neth.J.Agric.Sci.36,11–22.

Struik,P.C.,Visker,M.H.P.W.,Pauwels,J.B.,Colon,L.T.,2005.Arecuttingssuitablefor

assessingmaturitytypeinpotato(Solanumtuberosum)?Ann.Appl.Biol.147,27–

34.

Struik, P.C., Vreugdenhil, D., Van Eck, H.J., Bachem, C.W., Visser, R.G.F., 1999.

Physiologicalandgeneticcontroloftuberformation.PotatoRes.42,313–331.

Struik, P.C., Wiersema, S.G., 1999. Seed Potato Technology. Wageningen Pers,

Wageningen,theNetherlands.

Stuber,C.W.,Polacco,M.,Senior,M.L.,1999.Synergyofempiricalbreeding,marker‐

assisted selection, and genomics to increase crop yield potential. Crop Sci. 39,

1571–1583.

Sullivan, D.M., McQueen, J.P.G., Horneck, D.A., 2008. Estimating nitrogen

mineralization in organic potatoproduction.Oregon StateUniversity Extension

Service.EM8949‐E.

Suriharn, B., Patanothai, A., Pannangpetch, K., Jogloy, S., Hoogenboom, G., 2007.

Derivationofcultivarcoefficientofpeanutbreedinglinesforcropimprovement.

CropSci.47,605–619.

Tai, G.C.C., 2007. The Canon of Potato Science: 7. Genotype‐by‐Environment

Interaction.PotatoRes.50,231–234.

Tanksley,S.D.,Young,N.D.,Paterson,A.D.,Bonierbale,M.W.,1989.RFLPmappingin

plantbreeding:newtoolsforanoldscience.Biotechnol.7,257–269.

References

212

Tardieu,F.,2003.Virtualplants:modellingasatoolforthegenomicsoftoleranceto

waterdeficit.TrendsPlantSci.8,9–14.

Tardieu,F.,Tuberosa,R.,2010.Dissectionandmodellingofabioticstresstolerancein

plants.Curr.Opin.PlantBiol.13,206–212.

Tarn, T.R., Tai, G.C.C., De Jong,H., 1992. Breeding potatoes for long day, temperate

climates.PlantBreed.Rev.9,217–332.

Tekalign, T., Hammes, P.S., 2005. Growth responses of potato (Solanumtuberosum)

grown in a hot tropical lowland to applied paclobutrazol: 1. Shoot attributes,

assimilateproductionandallocation.N.Z.J.CropHort.Sci.33,35–42.

Teulat, B., Merah, O., Souyris, I., This, D., 2001. QTLs for agronomic traits from a

Mediterraneanbarleyprogenygrowninseveralenvironments.Theor.Appl.Genet.

103,774–787.

TheEuropeancultivatedpotatodatabase.(http://www.europotato.org).

Tourneux, C., Devaux, A., Camacho, M.R., Mamani, P., Ledent, J.F., 2003. Effects of

water shortage on six potato genotypes in the highlands of Bolivia (I):

morphologicalparameters,growthandyield.Agronomie23,169–179.

Trapani,N.,Hall, A.J., Sadras, V.O., Vilella, F., 1992.Ontogenetic change in radiation

useefficiencyofsunflower(HelianthusannuusL.)crops.FieldCropsRes.29,301–

316.

Trebejo, I., Midmore, D.J., 1990. Effect of water stress on potato growth, yield and

wateruseinahotandacooltropicalclimate.J.Agric.Sci.114,321–334.

Trethowan,R.M.,Reynolds,M., Sayre,K.,Ortiz‐Monasterio, I.,2005.Adaptingwheat

cultivarstoresourceconservingfarmingpracticesandhumannutritionalneeds.

Ann.Appl.Biol.146,405–413.

Uehara, G., 1998. Synthesis. In: Tsuji, G.Y., Hoogenboom, G., Thornton, P.K. (Eds.),

UnderstandingOptionsforagriculturalproduction.KluwerAcademicPublishers,

Dordrecht,theNetherlands,pp.389–392.

Ungerer,M.C.,Halldorsdottir,S.S.,Purugganan,M.D.,Mackay,T.F.C.,2003.Genotype‐

environment interactions at quantitative trait loci affecting inflorescence

developmentinArabidopsisthaliana.Genetics165,353–365.

UnionfortheProtectionofNewVarietiesofPlants(UPOV).(http://www.upov.int)

Uptmoor, R., Schrag, T., Stützel, H., Esch, E., 2008. Cropmodel based QTL analysis

across environments and QTL based estimation of time to floral induction and

floweringinBrassicaoleracea.Mol.Breed.21,205–216.

VanDamJ.,KoomanP.L.,StruikP.C.,1996.Effectsoftemperatureandphotoperiodon

earlygrowthandfinalnumberoftubersinpotato(SolanumtuberosumL.).Potato

Res.39,51–62.

VanDamJ.,Kooman,P.L.,Struik,P.C.,1996.Effectsof temperatureandphotoperiod

References

213

on early growth and finalnumber of tubers in potato (Solanum tuberosum L.).

PotatoRes.39,51–62.

Van Delden, A., Kropff, M.J., Haverkort, A.J., 2001. Modeling temperature‐ and

radiation–drivenleafareaexpansioninthecontrastingcropspotatoandwheat.

FieldCropsRes.72,119–142.

Van den Berg, J. H., Ewing, E.E., Plaisted, R.L.,McMurry, S., Bonierbale,M.W., 1996.

QTLanalysisofpotatotuberization.Theor.Appl.Genet.93,307–316.

Van der Beek, J.G., Verkerk, R., Zabel, P., Lindhout, P., 1992. Mapping strategy for

resistancegenesintomatobasedonRFLPsbetweencultivars:Cf‐9(resistanceto

Cladosporiumfulvum)onchromosome1.Theor.Appl.Genet.84,106–112.

Van der Wal, A,F., Bouma, W.F., Huijsman, C.A., Maris, B., Van Suchtelen, N.J.,

Wiersema, H.T., 1978. Breeding to maximize the physiological potential of

potatoes for yield. Survey papers 7th Triennial Conf. Eur. Assoc. Potato Res.

(EAPR),Warsaw,Poland,pp.23–33.

Van der Zaag, D.E., 1982. 3. Water supply to potato crops. Netherlands Potato

ConsultativeInstituteandMinistryofAgricultureandFisheries,pp.20–21.

VanderZaag,D.E.,1984.Reliabilityandsignificanceofsimplemethodofestimating

thepotentialyieldofpotatocrop.PotatoRes.27,51–73.

Van der Zaag, D.E., Burton, W.G., 1978. Potential yield of the potato crop and its

limitation. Survey papers 7th Triennial Conf. Eur. Assoc. Potato Res. (EAPR),

Warsaw,Poland,pp.7–23.

Van der Zaag, D.E., Doornbos, J.H., 1987. An atempt to explain differences in the

yielding ability of potato cultivars based on differences in cumulative light

interception, utilization efficiency of foliage and harvest index. Potato Res. 30,

551–568.

VanderZaag,P.,Demagante,A.L.,1987.Potato(Solanumspp.)inanisohyperthermic

environment.I.Agronomicmanagement.FieldCropsRes.17,199–217.

VanEck,H.J., Jacobs, J., Stam, P., Ton, J., Stiekema,W.J., Jacobsen, E., 1994.Multiple

allelesfortubershapeindiploidpotatodetectedbyqualitativeandquantitative

geneticanalysisusingRFLPs.Genetics137,303–309.

VanEeuwijk,F.A.,2003.GenotypebyEnvironment Interaction‐BasicsandBeyond.

PlantBreeding:TheArnelR.HallauerInternationalSymposium.

VanEeuwijk,F.A.,Malosetti,M.,Yin,X.,Struik,P.C.,Stam,P.,2005.Statisticalmodels

for genotype by environment data: from conventional AVOVA models to eco‐

physiologicalQTLmodels.Austr.J.Agric.Res.56,883–894.

VanHeemst,H.D.J., 1986. Thedistribution of drymatter during growthof a potato

crop.PotatoRes.29,55–66.

VanKempen,P.,LeCorre,P.,Bedin,P.,1996.Phytotechnic.In:Rousselle,P.,Crosnier,

References

214

R.Y.(Eds.),Lapommedeterre,INRA,Paris,pp.363–414.

Van Oijen, M., 1991. Light use efficiencies of potato cultivars with late blight

(Phytophthorainfestans).PotatoRes.34,123–132.

Van Ooijen, J.W., 2006. JoinMap 4.1, software for the calculation of genetic linkage

mapsinexperimentalpopulations.PlantResearchInternational,Wageningen.

VanOs,H.,Andrzejewski,S.,Bakker,E.,Barrena,I.,Bryan,G.J.,Caromel,B.,Ghareeb,B.,

Isidore,E.,DeJong,W.,VanKoert,P.,Lefebvre,V.,Milbourne,D.,Ritter,E.,Rouppe

VanderVoort, J., Rousselle‐Bourgeois, F., VanVliet, J.,Waugh,R., Visser, R.G.F.,

Bakker,J.,VanEck,H.J.,2006.Constructionofa10,000‐markerultradensegenetic

recombination map of potato: Providing a framework for accelerated gene

isolationandagenomewidephysicalmap.Genetics173,1075–1087.

Veldboom,L.R.,Lee,M.,1996.Geneticmappingofquantitative trait loci inmaize in

stressandnon‐stressenvironments.I.Grainyieldandyieldcomponents.CropSci.

36,1310–1319.

Visker,M.H.P.W.,Keizer,L.C.P.,VanEck,H.J.,Jacobsen,E.,Colon,L.T.,Struik,P.C.,2003.

CantheQTL for lateblight resistanceonpotatochromosome5beattributedto

foliagematuritytype?Theor.Appl.Genet.106,317–325.

Voltas,J.,Romagosa,I.,Ullrich,S.E.,VanEeuwijk,F.A.,2001.Identificationofadaptive

patternsinthe‘Steptoe’Morex’barleymappingpopulationintegratinggenetic,

phenotypic and environmental information. Presented at the 7th Quantitative

TraitLocusMappingandMarker‐AssistedSelectionWorkshop,Valencia,Spain.

Voorrips,R.E., 2002. MapChart: software for thegraphicalpresentationof linkage

mapsandQTLs.J.Hered.93,77–78.

Vos, J., 1995a. Foliar development of the potato plant and modulations by

environmentalfactors.In:Kabat,P.,VandenBroek,B.J.,Marshall,B.,Vos,J.(Eds.),

Modelling and parameterization of the soil‐plant‐atmosphere system. A

comparisonofpotatogrowthmodels.WageningenPers,Wageningen,pp.21–38.

Vos, J., 1995b. Potato ecology and modelling of crops under conditions limiting

growth. (Current Issues in Production Ecology). In: Haverkort, A.J.,MacKerron,

D.K.L.(Eds.),KluwerAcademicPublishers,Dordrecht,theNetherlands.

Vos, J., 1997. The nitrogen response of potato (Solanumtuberosum L.) in the field:

nitrogenuptakeandyield,harvestindexandnitrogenconcentration.PotatoRes.

40,237–248.

Vos,J.,2009.Nitrogenresponsesandnitrogenmanagementinpotato.PotatoRes.52,

305–317.

Vos,J.,Biemond,H.,1992.Effectsofnitrogenonthedevelopmentandgrowthofthe

potatoplant:I.Leafappearance,expansiongrowth,lifespansofleavesandstem

branching.Ann.Bot.70,27–35.

References

215

Vos, J., Groenwold, J., 1989. Genetic differences in water‐use efficiency, stomatal

conductanceandcarbonisotopefractionationinpotato.PotatoRes.32,113–121.

Vos, P., Hogers, R., Bleeker,M., Reijans,M., Lee, Tvd., Hornes,M., Friters, A., Pot, J.,

Paleman, J., Kuiper, M., Zabeau, M., 1995. AFLP: a new technique for DNA

fingerprinting.Nucl.AcidsRes.23,4407–4414.

Vreugdenhil,D.,Struik,P.C.,1989.Anintegratedviewofthehormonalregulationof

tuberformationinpotato(Solanumtuberosum).Physiol.Plant.75,525–531.

Walker,T.S.,Schmiediche,P.E.,Hijmans,R.J.,1999.Worldtrendsandpatternsinthe

potatocrop:aneconomicandgeographicsurvey.PotatoRes.42,241–264.

Walworth,J.L.,Carling,D.E.,2002.Tuberinitiationanddevelopmentinirrigatedand

non‐irrigatedpotatoes.Amer.J.PotatoRes.19,1–9.

Werner, H.O., 1934.The effect of a controlled nitrogen supply with different

temperatures andphotoperiods upon the development of the potato plant.Res.

Bull.Neb.Agric.Exp.Stn,No.75.

White, J.W., 1998.Modelingand crop improvement. In:Tsuji,G.Y.,Hoogenboom,G.,

Thornton,P.K.(Eds.),Understandingoptionsforagriculturalproduction.Kluwer

AcademicPublishers,Dordrecht,theNetherlands,pp.179–188.

White,J.W.,Hoogenboom,G.,1996.Integratingeffectsofgenesforphysiologicaltraits

intocropgrowthmodels.Agron.J.88,416–422.

Williams, M.M. II., Boydston, R.A., Davis, A.S., 2008. Crop competitive ability

contributestoherbicideperformanceinsweetcorn.WeedRes.48,58–67.

Xiao, J.,Li, J.,Yuan,L.,Tanksley,S.D.,1995.Dominance is themajorgeneticbasisof

heterosis inriceasrevealedbyQTLanalysisusingmolecularmarkers.Genetics

140,745–754.

Yan, W., 2001. GGEbiplot ‐ a Windows application for graphical analysis of

multienvironmenttrialdataandothertypesoftwo‐waydata.Agron.J.93,1111–

1118.

Yan,W.,2002. Singular‐valuepartition forbiplot analysisofmultienvironment trial

data.Agron.J.94,990–996.

Yan, W, Hunt, L.A., Sheng, Q., Szlavnics, Z., 2000. Cultivar evaluation and mega‐

environmentinvestigationbasedonGGEbiplot.CropSci.40,597–605.

Yan, W., Kang, M.S., 2003. GGE biplot analysis: A graphical tool for breeders,

geneticists,andagronomists.CRCPress,BocaRaton,FL.

Yan,W.,Wallace, D.H., 1995. Breeding for negatively associated traits, Plant Breed.

Rev.13,141–177.

Yin,X.,Chasalow,S.D.,Dourleijn,C.J.,Stam,P.,Kropff,M.J.,2000a.Couplingestimated

effectsofQTLs forphysiological traits toa cropgrowthmodel:predictingyield

variationamongrecombinantinbredlinesinbarley.Heredity85,539–549.

References

216

Yin,X.,Chasalow,S.D.,Stam,P.,Kropff,M.J.,Dourleijn,C.J.,Bos,I.,Bindraban,P.S.,2002.

Useofcomponentanalysis inQTLmappingofcomplexcroptraits:acasestudy

onyieldinbarley.PlantBreed.121,314–319.

Yin, X., Goudriaan, J., Latinga, E.A., Vos, J., Spiertz, H.J., 2003a. A flexible sigmoid

functionofdeterminategrowth.Ann.Bot.91,361–371.

Yin, X., Guo, W., Spiertz, J.H.J., 2009. A quantitative approach to characterize sink‐

source relationships during grain filling in contrasting wheat genotypes. Field

CropsRes.114,119–126.

Yin,X.,Kropff,M.J.,Goudriaan,J.,Stam,P.,2000b.Amodelanalysisofyielddifferences

amongrecombinantinbredlinesinbarley.Agron.J.92,114–120.

Yin, X., Kropff, M.J., McLaren, G., Visperas, R.M., 1995. A nonlinear model for crop

developmentasafunctionoftemperature.Agric.ForestMeteorol.77,1–16.

Yin, X., Kropff, M.J., Stam, P., 1999a. The role of ecophysiological models in QTL

analysis:theexampleofspecificleafareainbarley.Heredity82,415–421.

Yin, X., Lantinga, E.A., Schapendonk, A.H.C.M., Zhong, X., 2003b. Some quantitative

relationships between leaf area index and canopy nitrogen content and

distribution.Ann.Bot.91,893–903.

Yin, X., Schapendonk, A.H.C.M., Kropff,M.J., VanOijen,M., Bindraban, P.S., 2000c. A

genericequation fornitrogen‐limited leafarea indexand itsapplication incrop

growthmodelsforpredictingleafsenescence.Ann.Bot.85,579–585.

Yin,X.,Stam,P.,Dourleijn,C.J.,Kropff,M.J.,1999b.AFLPmappingofquantitativetrait

loci foryield‐determiningphysiologicalcharactersinspringbarley.Theor.Appl.

Genet.99,244–253.

Yin, X., Stam, P., Kropff, M.J., Schapendonk, A.H.C.M., 2003c. Crop Modeling, QTL

Mapping,andTheirComplementaryRoleinPlantBreeding.Agron.J.95,90–98.

Yin,X.,Struik,P.C.,2008.Applyingmodellingexperiencesfromthepasttoshapecrop

systemsbiology:theneedtoconvergecropphysiologyandfunctionalgenomics.

NewPhytologist179,629–642.

Yin, X., Struik, P.C., 2009. C3 andC4 photosynthesismodels: An overview from the

perspectiveofcropmodelling.NJAS‐WageningenJ.LifeSci.57,27–38.

Yin, X., Struik, P.C., 2010. Modelling the crop: from system dynamics to systems

biology.J.Exp.Bot.61,2171–2183.

Yin,X.,Struik,P.C.,Kropff,M.J.,2004.Roleofcropphysiology inpredictinggene‐to‐

phenotyperelationships.TrendsPlantSci.9,426–432.

Yin, X., Struik, P.C., Tang, J., Qi, C., Liu, T., 2005a. Model analysis of flowering

phenologyinrecombinantinbredlinesofbarley.J.Exp.Bot.56,959–965.

Yin,X.,Struik,P.C.,VanEeuwijk,F.A.,Stam,P.,Tang,J.,2005b.QTLanalysisandQTL‐

basedpredictionoffloweringphenologyinrecombinantinbredlinesofbarley.J.

References

217

Exp.Bot.56,967–976.

Yin,X.,VanLaar,H.H.,2005.Cropsystemsdynamics:anecophysiologicalsimulation

model for genotype by environment interactions. Wageningen Academic

Publishers,theNetherlands,pp.155.

Yuan, F.M., Bland, W., 2005. Comparison of light‐ and temperature based index

modelsforpotato(SolanumtuberosumL.)growthanddevelopment.Am.J.Potato

Res.82,345–352.

Zebarth,B.J.,Leclerc,Y.,Moreau,G.,2004.Rateandtimingofnitrogenfertilizationof

RussetBurbankpotato:Nitrogenuseefficiency.Can.J.PlantSci.84,845–854.

Zebarth, B.J., Leclerc, Y., Moreau, G., Gareau, R., Milburn, P.H., 2003. Soil inorganic

nitrogencontentincommercialpotatofieldsinNewBrunswick.Can.J.SoilSci.83,

183–195.

Zebarth,B.J.,Tai,G.,Tarn,R.,DeJong,H.,Milburn,P.H.,2004.Nitrogenuseefficiency

characteristicsofcommercialpotatocultivars.Can.J.PlantSci.84,589–598.

Zebarth, B.J., Tarn, T.R., De Jong, H., Murphy, A., 2008. Nitrogen use efficiency

characteristicsofAndigenaanddiploidpotatoselections.Am.J.Pot.Res.85,210–

218.

Zeng, Z. B., 1994. Precisionmapping of quantitative trait loci. Genetics 136, 1457–

1468.

Zhou,L.,Wang, J.,Yi,Q.,Wang,Y.,Zhu,Y.,Zhang,Z.,2007.Quantitative trait loci for

seedlingvigorinriceunderfieldconditions.FieldCropsRes.100,294–301.

 

SummaryThesustainedincreaseinglobalfoodproductionisunprecedented.Amajorchallenge

forplantandcropscienceisdiscoveringnewwaystocontinuetoincreasecropyield

in a sustainablemanner. Potato (Solanumtuberosum L.) is among theworld’smost

important cropswith awell‐recognised role in humannutrition, food security, and

national economyofmany countries.Due to increasing fooddemand and changing

dietspotatoisbecomingasubsistencecropinmanyregions.However,theagronomy

and whole crop physiology of tuber yield production are difficult as each

developmentalstageinpotatoislikelytoberegulatedandcontrolledbyalargesetof

interacting expressed genes and because the genotype‐by‐environment (GE)interactions are strong. Breeding for highly‐yielding crop cultivars for specific

environments and the prediction of their behaviour are major challenges. New

technologies must be developed to accelerate breeding through improving

genotyping and phenotyping methods and by increasing the available genetic

diversityinbreedinggermplasm.

Interventions in breeding based on understanding of the genetic and

physiologicalbasisofcropperformancehavethepotentialtoaccelerategeneticgains

for yield and resource use efficiency. The genetic improvement of yield can be

understoodmoremechanisticallybyinvestigatingandinterpretingtherelationships

between the main attributes of growth. Such studies may shed light on potential

futurealternativesforsustainableyieldimprovement.

Cropmodellingcanplayasignificantpartinsystemapproachesbyprovidinga

powerful tool for phenotype prediction and yield scenario analysis; to explore the

impact of new genotypes and the contribution of individual physiological traits on

yield by simulating genotypes under various environmental scenarios and

management options.Modellingmay also prove useful in understanding GE, thushelpingtospeedupthecurrentlystagnantpaceofcropimprovement.Nevertheless,

insightsintothephysiologicalprocessesleadingtoyielddevelopmentandthefactors

affectingtheseprocessesarestillunderdeveloped.

This thesis aims to develop an approach to quantify and predict the yield of

individual genotypes and to estimate parameters which may reveal the effects of

genetic and environmental factors on important plant processes controlling tuber

yield variation. The analysis conducted in this thesis is based primarily on data

collected from six contrasting field experiments carried out inWageningen (52˚N

latitude), theNetherlands,during2002,2004and2005.Theplantmaterialusedin

thisstudyconsistedof100F1diploid(2n=2x=24)potatogenotypesderivedfroma

Summary

220

crossbetweentwodiploidheterozygouspotatoclones,SH83‐92‐488RH89‐039‐16.Five standard cultivars (Première, Bintje, Seresta, Astarte, and Karnico) and the

parentalcloneswerealsoincludedinthisstudy.

The thesis is composed of six chapters. Chapter 1 gives an overview of

complications indealingwithquantitativetraits likeyieldandhighlights theroleof

systems approaches via incorporating crop modelling and crop physiological

knowledge in potato breeding programmes. It presents the possible opportunities

fromsuchcombinedeffortstowardsreinforcingthegeneticanalysisofcomplextraits.

Chapter2presentsaquantitativemodelapproachbasedonthebetafunctionto

analysethetimecourseofcanopycoverduringtheentirecropcycleasafunctionof

thermaltime.Themodeldescribesthecanopydevelopmentdynamicsinthreephases

(build‐up phase, maximum cover phase, canopy decline phase) through five

parameters defining the timing and duration of the phases and maximum canopy

cover (vmax). These five parameters were further used to derive secondary traits,

related to rate of increase or decrease of canopy cover, and the area underwhole

green canopy cover curve (Asum). The latter trait Asum was directly related to the

ability of the genotypes to intercept photosynthetically active radiation (PAR) and

thustorealizetuberyield.Themodelsuccessfullydescribedquantitativedifferences

in canopydynamicsof adiversesetof genotypes ina segregatingF1populationof

potatoundervariedenvironments.

Theresults indicated that lengthof thecanopybuild‐upphase (t1orDP1)was

conservative,butthedurationofmaximumcanopycover(DP2)andthedeclinephase

(DP3)varied greatly,with later genotypeshaving similarDP1, butlongerDP2 andDP3

and thus generate higher values of Asum. Strong positive phenotypic and genetic

correlations were observed between DP2 and vmax, indicating that genotypes with

longerDP2couldbeindirectlyobtainedbyselectinggenotypeswithmaximumvalue

ofcanopycover(vmax).Formosttraitsquantifiedinthemodel,highgeneticvariability

andhighheritabilitywererecorded.Geneticvariancecontributedgreatlytothetotal

phenotypicvarianceformosttraits,forinstance,timeofonset(t2)orendofcanopy

senescence(te),DP2,andAsum,indicatingtheirstronggeneticbackgroundandstability

acrossenvironments.

SeveralQTLsweredetectedacross theexperiments for themodelparameters

and derived traits determining the canopy dynamics. One particular chromosomal

positionat18.2cMonpaternal (RH) linkagegroupVwascontrollingnearlyall the

traits and showed its pleiotropic nature. This QTL had major additive effects and

explained 74%of the phenotypic variance.Most of the traits linked to this linkage

group were related with earliness. Such information could be very useful in

identifyingthegeneticbasisofmaturitytypeinpotato.Thischapteryieldedestimates

Summary

221

foragronomically relevantcropcharacteristics thatcouldbepromising fordefining

future breeding strategies for traits influencing the radiation interception and

therebyfinalyield.

Chapter3describesaquantitativemodelapproachtoanalysethedynamicsof

tuberbulkingduringtheentirecropcycleasafunctionofthermaltimeusingapiece‐

wise expolinear function. The model breaks down tuber growth into parameters

related with the growth rate and effective duration of the linear phase of tuber

bulking.Furthermore, radiation‐ and nitrogen use efficiencies (RUE and NUE,

respectively), and their relationships with the model parameters are also studied.

Rateoftuberbulking(cm)washighestforearly‐maturinggenotypesfollowedbymid‐

lateandthenlategenotypes.Ontheotherhand,late‐maturinggenotypeshadlongest

effectivedurationoftuberbulking(ED)followedbymid‐lateandearlygenotypes.As

aresult,finaltuberyield(wmax)washigherinlategenotypesthaninearlygenotypes.

The results further illustrated that RUE values were highest for early‐maturing

genotypesfollowedbymid‐lateandlategenotypeswhereasNUEwashighestinlate‐

maturinggenotypesfollowedbymid‐earlyandearlygenotypes.Thephenotypicand

geneticcorrelationssuggestedphysiologicaltrade‐offsbetweencmandEDaswellas

between RUE and NUE. High genetic variability along with high heritability was

recorded for most traits which illustrated their strong genetic background. Path

coefficientanalysisshowedthatRUE,cm,andAsum(previouslydefinedinChapter2),

hadamajor influenceonwmax.SixteenQTLsweredetected fortraitsstudiedinthis

chapter, explaining the phenotypic variance by up to 79%. In accordance with

Chapter2,QTLscontrollingmostof thetraitsweredetectedatposition18.2cMon

paternal(RH)linkagegroupVwithmajoradditiveeffectsandexplainedmostofthe

totalphenotypicvariance.OurresultsindicatedthatanumberofQTLsfortraitswere

notdetectedwhentuberyieldpersewassubjectedtoQTLanalysis.Thephenotypic

varianceexplainedby theQTLs for tuberyieldpersewasalso lower than forother

traits.Thissuggests thatcomplextraits likethesearecontrolledbyQTLsthatoften

show low stability across environments. Overall, parameters found in this chapter

indicatedthatthereareopportunitiesforimprovingtuberdrymatteryieldinpotato

by selection for optimal combination of important physiological traits likeRUE, cm,

andAsum.

Chapter4addressestheissueofmaturitytypeinpotato,ascurrentlythereare

somanyunclear interpretations.This chapter tries to create a clear and consistent

definition of the concept of maturity type and presents an approach to assess the

maturitytypeinalargesetofpotatogenotypesinarathersimpleandreproducible

way. The added physiological and quantitative knowledge gained from Chapters 2

and3wereusedtoquantifymaturitytypeunambiguouslyforasetofvarietiesanda

Summary

222

segregating population across diverse environments. The re‐defined physiological

basedmaturitycriteriaarebasedonfourtraits:DP2,cm,ED,andAsum.There‐defined,

four physiological maturity criteria were compared with the conventionally used

criterion. The results indicated that physiological maturity type criteria tended to

definematurityclasseslessambiguouslyincomparisontotheconventionalcriterion.

Moreover, theconventionalcriterionwassubject tomorerandomnoiseand lacked

stabilityand/orrepeatabilitywithrespecttophysiologicalquantitativetraits.There‐

defined criteria also clearly illustrated the physiological trade‐offs that existed

between the selected traits and underlined the subtle complexities in defining the

maturity type. The new definition would not only allow relating maturity to crop

phenologyandphysiologybutcouldalsoofferwiderapplicationsinpotatobreeding

and cropmanagement studies. One such applicationwould be to help in designing

strategiesforpotatoideotypebreedingforgenotypeswithspecificmaturitytypes.

Chapter 5 sheds light on some of the key mechanisms causing genotype

differences in tuberyieldandprovidessupport for importantaspectsofhypothesis

regarding nitrogen accumulation. It tests the capability of the generic

ecophysiological model ‘GECROS’ to analyse differences in tuber yield of potato

withinanF1population,theparents,andasetofcultivars.Themodelpredictscrop

growth and development as affected by genetic characteristics and climatic and

edaphic environmental variables. The genotype‐specific model‐input parameter

valueswereestimatedandthemodelwasusedforpredictingthetuberyieldofafore‐

mentionedplantmaterial inmultiple field experiments. Themodelwas reasonably

good in predicting the differences in tuber yield across environments and across

genotypes.Trendsofgrowthandnitrogenuptakewereadequatelyreproducedbythe

model.Modelanalysis identified thegenotypickey‐parametersaffecting tuberyield

productionandNmax(i.e.totalcropNuptake)contributedmosttothedetermination

oftuberyield.TheresultsshowedthatgenotypeswithhigherNmaxandlowertuberN

content exhibited higher tuber dry matter yield. Such information can greatly

facilitatethedevelopmentofpotatoideotypesforspecificenvironments.Theresults

confirmedthattheGECROSmodelisusefulforexploringtheimpactofnewgenotypes

and the contribution of individual physiological traits on yield by simulating the

responsesofgenotypesacrossdifferentenvironments.

Chapter6 finally discusses themain findings and overall contribution of this

thesis. It indicatesways to enhance the power ofmolecular breeding strategies as

wellasideotypebreedinginpotato.Itstressesupontheuseofcombinedknowledge

from the fields of crop physiology, modelling, and genetics in elucidating and

understanding the complex traits. Such an integrated approach can reinforce the

geneticanalysisofcomplextraitslikeyield,therebyimprovingbreedingefficiency.In

Summary

223

this context, future research should focus on this area and further steps should be

taken in linking cropmodellingwith QTLmapping for physiological and genotypic

traits generated by this thesis. The strong base set by this thesis could help in

achieving this broader goal ofmodelbased plant breeding. However, tomake such

significantstrides,muchclosercollaborative researcheffortsbetweenphysiologists

andplantbreedersaredirelyneeded

 

SamenvattingWereldwijdblijftdevraagnaarvoedseltoenemen.Eengroteuitdagingvoorplant‐en

gewaswetenschappen ishetontdekkenvannieuwemanierenomopeenduurzame

manierdegewasopbrengstenteblijvenverhogen.Deaardappel(Solanumtuberosum

L.) is één van de belangrijkste gewassen in de wereld. Het gewas speelt een

belangrijke rol in de humane voeding, voedselzekerheid, en de nationale economie

van vele landen. Door de toenemende vraag naar voedsel en veranderende

voedingspatronenwordtdeaardappelinveleregio’seenbelang‐rijkerbasisvoedsel.

In het algemeen zijn agronomie en gewasfysiologie van opbrengstvorming lastig

aangezien elk ontwikkelingsstadium waarschijnlijk wordt gereguleerd en

gecontroleerddooreengrotereekstotexpressiegebrachteenopelkaarinwerkende

genen en omdat de genotypemilieu (GE) interacties sterk zijn. Veredeling voorhoog‐opbrengenderassenvoorspecifiekemilieusenhetvoorspellenvanhungedrag

zijn daarbij belangrijke uitdagingen. Nieuwe technologieën moeten worden

ontwikkeldomhetveredelingsprocesteversnellen.Denkaanverbeterdemethoden

voorgenotyperingen fenotypering, alsmedemodellendiehet fenotypevoorspellen

uit het genotype. Daarmee kan bestaande en nieuwe genetische diversiteit beter

benutwordeninhetveredelingsproces.

Inzicht in de genetische en fysiologische basis van opbrengstvorming kan

bijdragenaanmeerefficiënteselectiemethodenvoordegewenstegenetischeaanleg,

in het bijzonder voor eigenschappen die bijdragen aan betere benutting van de

hulpbronnen lichtenstikstof (RUE,NUE).Degenetischeopbrengst‐verbeteringkan

meermechanistischwordenbegrependoorhetonderzoekeneninterpreterenvande

relaties tussen de belangrijkste groeikarakteristieken. Door het onderzoeken en

interpreteren van de relaties tussen de belangrijkste processen tijdens de groei en

ontwikkelingvandeplantingenetischdiversmateriaalkanmeerspecifiekgestuurd

wordenindeveredelingvanopbrengst‐componenten.Dergelijkestudieskunnenlicht

werpen op mogelijke toekomstige alternatieven voor duurzame

opbrengstverbetering.

Gewasmodellering kan een belangrijke rol spelen in systeembenaderingen

omdatmodellenkrachtige instrumentenzijnvoorhetvoorspellenvanhet fenotype

en voor het analyseren van opbrengstscenarios.Modellen kunnen ook behulpzaam

zijn bij verkennen van het opbrengstpotentieel van nieuwe genotypen en van de

bijdrage van individuele fysiologische kenmerken aan deze opbrengst, omdat ze

gebruikt kunnen worden voor het simuleren van de groei van genotypen onder

verschillendemilieu‐scenario's en teelttechniek.Modellering kanooknuttig zijn bij

Samenvatting

226

hetbegrijpenvanGE.Tochzijndeinzichtenindefysiologischeprocessendieleidentotdeopbrengstvormingende factorendievaninvloedzijnopdezeprocessennog

onderontwikkeld.

Dit proefschrift beoogt een aanpak te ontwikkelen voor het kwantificeren en

voorspellenvandeopbrengstvoor individuelegenotypenenvoorhet schattenvan

relevante parameters. Het gaat daarbij om parameters die de effecten kunnen

blootleggen van genetische factoren en omgevingsfactoren op belangrijke

plantprocessendiegezamenlijkdevariatie inknolopbrengstbepalen.Deanalyse in

dit proefschrift is hoofdzakelijk gebaseerd op gegevens die zijn verzameld in zes

contrasterende veldexperimenten uitgevoerd nabij Wageningen (52˚

Noorderbreedte),Nederland, inde jaren2002,2004en2005.Hetplantmateriaal in

deze studie bestond uit een diploïde F1 populatie (2n = 2x = 24) van 100

nakomelingen verkregen uit een kruising tussen twee heterozygote diploïde

aardappelklonen, SH83‐92‐488 en RH89‐039‐16. Vijf standaardrassen (Première,

Bintje, Seresta,AstarteenKarnico)endeoorspronkelijkeouderklonenwerdenook

opgenomenindezestudie.

Hetproefschriftbestaatuitzeshoofdstukken.Hoofdstuk1geefteenoverzicht

van complicaties in het omgaan met kwantitatieve kenmerken zoals opbrengst en

benadrukt de rol van de systeembenaderingen via het inpassen van

gewasmodellering en gewasfysiologische kennis in aardappelveredelings‐

programma's. Het presenteert de mogelijkheden van dergelijke gecombineerde

inspanningenomtekomentoteengenetischeanalysevancomplexeeigenschappen.

In Hoofdstuk 2 wordt een kwantitatieve modelbenadering op basis van de

bètafunctie gepresenteerd om het verloop van de bodembedekkingsgraad te

analyseren tijdens de gehele gewascyclus als functie van de thermotijd. Hetmodel

beschrijft de dynamiek van de ontwikkeling van de bodembedekkingsgraad in drie

fasen (opbouwfase, fase vanmaximale bodembedekking, afstervingsfase). Het doet

dat doormiddel van vijf parameters die samen aanvang, eind en duur van de drie

fasen bepalen alsmede de maximale bedekkingsgraad (vmax). Deze vijf parameters

werden verder gebruikt om secundaire kenmerken te schatten, gerelateerd aan

snelheid van toe‐ of afname van de bedekkingsgraad, en de oppervlakte onder de

totale curve van de bodembedekking tegen de thermotijd (Asum). Deze laatste

eigenschap, Asum, was direct gerelateerd aan het vermogen van de genotypen om

fotosynthetischactievestraling(PAR)teonderscheppenendusomknolopbrengstte

realiseren.Hetmodelbeschreefmetsucceskwantitatieveverschillenindynamiekin

bodembedekkingvandezegenetischdiversesetvangenotypenineenuitsplitsende

F1populatievanaardappelgeteeldonderuiteenlopendeomstandigheden.

De resultaten gaven aan dat de lengte van de opbouwfase van de bodem‐

Samenvatting

227

bedekking (t1 of DP1) conservatief was, maar de duur van de maximale

bodembedekkingsgraad (DP2) en de duur van de fase van afsterving (DP3) liepen

uiteen,waarbijlateregenotypenvergelijkbareDP1,maarlangereDP2enDP3haddenen

dushogerewaardenvoorAsumgenereerden.Erwerdensterke,positievefenotypische

en genetische correlaties gevonden tussen DP2 en vmax, hetgeen aangeeft dat

genotypenmeteenlangereDP2 indirectzoudenkunnenwordenverkregendoorhet

selecteren van genotypen met een hoge waarde voor de maximale

bodembedekkingsgraad (vmax). Voor de meeste met het model gekwantificeerde

eigenschappen was de F1 populatie uitermate variabel in vergelijking met de

standaardrassen. Voor demeeste eigenschappenwerd een hoge erfelijkheidsgraad

waargenomen; met andere woorden de genetische variantie droeg in belangrijke

mate bij aan de totale fenotypische variantie. Dit was bijvoorbeeld het geval voor

tijdstipvanaanvang(t2)ofheteindevanloofafsterving(te),DP2,enAsum,duidendop

eensterkegenetischebasisenstabiliteitovermilieus.

Over de verschillendemilieuswerden voor demodelparameters en afgeleide

kenmerken die de bodembedekkingsdynamiek bepalen verschillende QTLs

gedetecteerd. De ouder RH89‐039‐16 bleek heterozygoot voor een bijzonder

genetische locus op 18.2 cM op chromosoom 5. De uitsplitsende allelen van deze

ouderbepaalden indenakomelingenbijnaallegemeteneigenschappenen toonden

zo een pleiotroop karakter. De andere ouder SH83‐92‐488 splitste voor deze locus

nietuit.Deallelenvandezelocus(QTL)hadgroteadditieveeffectenenverklaardetot

74%vande fenotypischevariantie.Erwas sprakevangenetischekoppeling tussen

QTLsvoorvroegheidenmodelparameters.Dergelijkeinformatiekanzeernuttigzijn

bij het identificeren en karakteriseren van de genetische factoren betrokken bij

vroegrijpheid in aardappel en daaraan gerelateerde kenmerken. Dit hoofdstuk

leverde schattingen op voor landbouwkundig relevante gewaskenmerken die

veelbelovendkunnenzijn voorhetbepalenvan toekomstigeveredelingsstrategieën

voor eigenschappen die de lichtonderschepping, en daarmee uiteindelijk de

opbrengst,beïnvloeden.

Hoofdstuk 3 beschrijft een kwantitatieve modelbenadering waarmee de

dynamiekvandeknolgroeigedurendedegehelegewascycluswerdgeanalyseerdals

functie vande thermotijdmetbehulp van een in trajectenopgedeelde expolineaire

functie te analyseren. Het model ontleedt knolgroei tot parameters die betrekking

hebbenopdegroeisnelheidendefeitelijkeduurvandelineairefasevanknolvorming.

Bovendienwerdendegebruiksefficiëntievanstralingenstikstof(respectievelijkRUE

en NUE), en hun relaties met de modelparameters bestudeerd. Snelheid van

knolvorming (cm) was het hoogst voor de vroegrijpe genotypen gevolgd door

middellateendanlategenotypen.Anderzijds,laatrijpegenotypenhaddendelangste

Samenvatting

228

knolvormingsduur (ED) gevolgd door middellate en vroege genotypen.

Dientengevolge was de uiteindelijke knolopbrengst (wmax) hoger voor de late

genotypendanvoordevroegegenotypen.Erwerdtevensgevondendatdewaarden

voorRUEhethoogstwarenvoorvroegrijpegenotypen,gevolgddoormiddellateen

late genotypen, terwijl NUE juist het hoogst was in late genotypen, gevolgd door

middelvroege en vroege genotypen. De waargenomen fenotypische en genetische

correlatiessuggererendateromfysiologischeredeneneenafruilbestaattussencmen

ED.Hetzelfdegeldt tussenRUEenNUE.Voordemeestekenmerkenwerdeenhoge

genetischevariatie eneenhogeerfelijkheidsgraadgevonden, suggererenddat ze in

sterkemategenetischbepaaldzijn.PadcoëfficiëntanalysetoondeaandatRUE,cm,en

Asum(eerderbeschreveninHoofdstuk2)eengroteinvloedopwmaxhadden.Voorde

eigenschappendieindithoofdstukwerdenbestudeerdwerden16QTLsgedetecteerd,

diemaximaal 79%vande fenotypische variantie verklaarden.Overeenkomstigmet

wat in Hoofdstuk 2werd gevonden, beheerste de locus op positie 18.2 cM van de

vaderlijke(RH)koppelingsgroepVhetgrootstedeelvandeeigenschappen.Vooralle

eigenschappen vertoonde deze locus grote additieve effecten en verklaarde het

grootstedeelvandetotalefenotypischevariantie.Onzeresultatenlatenziendateen

aantal QTLs die voor de deeleigenschappen werden gedetecteerd niet werden

gevondenwanneer knoldrogestofopbrengstpersewerd onderworpen aan eenQTL

analyse. De fenotypische variantie verklaard door de QTLs voor

knoldrogestofopbrengst per se was ook lager dan voor andere eigenschappen. Dit

suggereert dat complexe eigenschappen als knoldrogestofopbrengst worden

gecontroleerd door QTLs die vaak slechts in beperkte mate stabiel zijn over de

milieus.Overhet geheel genomengavendeparameters indithoofdstukaandat er

kansenzijnomdeknoldrogestofopbrengst inaardappel teverbeterendoorselectie

opeenoptimalecombinatievanbelangrijkefysiologischeeigenschappen,zoalsRUE,

cm,enAsum.

Hoofdstuk 4 gaat in op de kwestie van vroegheid in aardappel, omdat

momenteel veel onduidelijk is over dit concept. Dit hoofdstuk probeert een

eenduidige en consistente definitie van het begrip vroegheid te geven en geeft een

benaderingomvroegheidtebepalenineengrotereeksaardappelgenotypen,opeen

vrij eenvoudige en reproduceerbare manier. De nieuwe fysiologische en

kwantitatieve kennis opgedaan in de Hoofdstukken 2 en 3 werd gebruikt om

vroegheid eenduidig te kwantificeren voor een set van rassen en een uitsplitsende

populatie geteeld in diverse milieus. De nieuwe definitie van vroegheid werd

gebaseerdopvierverschillendefysiologischekenmerken:DP2,cm,EDenAsum.Devier

nieuwe, fysiologische vroegheidscriteria werden vergeleken met het gebruikelijke

criterium.Deresultatengavenaandatdefysiologischevroegheidscriteriadoorgaans

Samenvatting

229

de vroegheidsklassen op een minder dubbelzinnige manier definieerden dan het

conventionelecriterium.Bovendiengafhetconventionelecriteriummeerruisenwas

het minder stabiel en minder herhaalbaar dan de fysiologische kenmerken. De

nieuwe criteria blijken overigens ook de fysiologische wisselwerkingen te

onderstrepen die bestonden tussen de geselecteerde eigenschappen en

onderstreepten tevens de complexiteit van vroegheid bij aardappel. De nieuwe

definitie maakt het niet alleen mogelijk vroegheid te relateren aan fenologie en

fysiologie, maar kan ook breed worden toegepast in de aardappelveredeling en

onderzoeknaardeteeltechniekvanaardappel.Eénvandietoepassingenzoukunnen

zijn het ontwerpen van strategieën voor veredeling van aardappelideotype voor

specifiekevroegheids‐klassen.

Hoofdstuk5werptlichtopenkelevandebelangrijkstemechanismendieertoe

leiden dat genotypen verschillen in knoldrogestofopbrengst en biedt houvast voor

belangrijke veronderstellingen aangaande de ophoping van stikstof. Het test de

mogelijkheid om met het generieke gewasgroeimodel GECROS ecofysiologische

verschillen in knoldrogestofopbrengst van nakomelingen uit een F1 populatie, de

ouders, en een reeks cultivars te analyseren. Het model voorspelt de groei en

ontwikkelingvangewassenonder invloedvangenetischekenmerkenenbodem‐en

klimaatfactoren. De genotype‐specifieke model‐input parameterwaarden werden

geschatenhetmodelwerdgebruiktvoorhetvoorspellenvandeknolopbrengstvan

voornoemdplantmateriaalinverschillendeveldexperimenten.Hetmodelvoorspelde

de verschillen in knoldrogestofopbrengst over omgevingen en tussen genotypen

tamelijk goed. Het model reproduceerde op adequate wijze de trends in groei en

stikstofopname. Met behulp van de modelanalyse konden de belangrijkste

genotypische parameters worden geïdentificeerd die van invloed zijn op de

knoldrogestofopbrengst.DeNmax(demaximaletotaleNopname)droeghetmeestbij

aan de uiteindelijke knoldrogestofproductie. De resultaten toonden aan dat

genotypen met een hogere Nmax en een lager N‐gehalte in de knol een hogere

knoldrogestofopbrengsthadden.Dergelijkeinformatiekanaanzienlijkbijdragenaan

deontwikkelingvanaardappelideotypenvoorspecifiekeomgevingen.Deresultaten

bevestigdendathetmodelGECROShandigisvoorhetverkennenvanhetbelangvan

nieuwe genotypen en van de bijdrage van individuele fysiologische eigenschappen

aandeopbrengstdoorhetsimulerenvandereactiesvangenotypeninverschillende

milieus.

Hoofdstuk6 behandelt tot slot de belangrijkste bevindingen en de algemene

bijdrage van dit proefschrift aan de wetenschap. Het geeft manieren aan om

moleculaire veredelingsstrategieën en ideotypeveredeling in de aardappel te

versterken. Het benadrukt de noodzaak om kennis op de terreinen van de

Samenvatting

230

gewasfysiologie, modellering, en de genetica te combineren teneinde complexe

eigenschappen beter te begrijpen. Een dergelijke geïntegreerde aanpak kan de

genetische analyse van complexe eigenschappen zoals opbrengst versterken en

verhoogt aldus de efficiëntie van het veredelen. In dit verband dient toekomstig

onderzoek zich te richten op dit gebied en dienen verdere stappen te worden

genomenomgewasmodellerentekoppelenaanQTLmappingvandefysiologischeen

genotypische eigenschappendie in dit proefschrift zijn beschreven.Dit proefschrift

legteenstevigebasisvoorhetbrederedoelvandoorgewasmodellenondersteunde

plantenveredeling. Echter, dergelijke stappen kunnen alleen gezet worden als de

samenwerkingtussenfysiologenengeneticigeïntensiveerdwordt.

Acknowledgements

Thisthesisisasteptowardsunderstandingofgeneticvariationamongalargesetof

potatogenotypes.Thisstudy isapartofanextensiveresearcheffort toexplorethe

possibilities to link crop physiology, modelling, and genetic analysis for a better

understandingofpotatogrowthanddevelopmentthatmayhelpinunderstandingthe

yield dynamics as well as in designing ideotypes for specific and/or diverse

environments. This thesis has resulted into four years and more of field and

laboratory experimentation, literature review, data collection and modelling and

geneticanalysis. I sincerelyhope that this thesiswill, albeitmodestly, contribute to

thescientificknowledgeofpotatoscience.

Pursuing thisPhDprojectwithsomany interesting facetswas indeeda lovely

andmemorableexperienceforme.Itwasjustlikeclimbingahighpeak,stepbystep,

accompaniedwithchallenges,hardships,encouragement,andhope.NowwhenIfind

myselfatthetopenjoyingthebeautifulscenery,Irealisethatitwasinfactpossible

duetoexcellentteamworkthatgotmeherefinally.

ThecompletionofmydissertationandsubsequentPhDhasbeenalongrichand

learning journey. This journey actually started in 2007, when my dream to attain

highereducationfromawellrenownedworldleadinguniversitybecamearealitydue

to a scholarship award from Higher Education Commission (HEC), Government of

Pakistan under “Overseas PhD Scholarship Program Phase‐II”. I was very lucky to

cometotheWageningenUniversity(WUR),theworld’sleadingacademicinstitution.I

couldnothavesucceededwithoutthe invaluablesupportofseveral.Hence, Iwould

hereliketotakethisopportunitytoextendmymany,manysincerethankstoallthe

personsandinstitutionswhohavecontributedinthedifferentaspectsofthisstudy.

Myappreciationismuchdeeperthanafewwordscansay.

Firstandforemost,Iwouldliketoexpressmydeepandsinceregratitudetomy

multidisciplinaryPhDsupervisoryteammadeupofProf.dr.ir.PaulC.Struik(Centre

forCropSystemsAnalysis,WUR(CSA)),Prof.dr. ir.FredA.vanEeuwijk(Biometris,

WUR), Dr. Xinyou Yin (CSA), and Dr. ir. Herman J. van Eck (Laboratory of Plant

Breeding,WUR).Theteamwasapillarofsupportall theway fromthebeginning. I

couldnothavewishedforbettercollaboratorsandcoaches.Eachofyouhadaunique

support and efforts towards my success, which will always be remembered and

appreciated.Thankyouverymuchforembarkingwithmeonthislongjourney.Your

relentlesscontributionsandinsighthavebeenofgreatvaluetome.Youhaveallmade

itpossibleformetocommenceandcompletethisenormoustask.This is inspiteof

thedifficultchallengesandpredicamentsthatwewereallfacedwith.Toallthefour

supervisors, I have learned a lot under your tutelage and grown not just as a

Acknowledgements

232

researcher,butaspersontooandforthisIamindebtedtoyouall.

Aboveall,myspecialandsinceredeeply felt thanks tomyhonorificpromotor

Prof.dr.ir.PaulC.Struik,whoacceptedmeashisPhDstudentwithoutanyhesitation

when I requested him for an acceptance letter. Thereafter, he offeredme somuch

advice,patiently supervisingme,andalwaysguidingme in the rightdirection.This

thesiswould not have been possiblewithout the help, support and patience ofmy

principlesupervisorProf.PaulC.Struik,nottomentionhisadviceandunsurpassed

knowledgeofpotatophysiology.Heguidedmethroughoutmystudyperiod.Itisnot

sufficienttoexpressmygratitudewithonlyafewwords.Hewasinvolvedinalmost

allaspectsofplanningandoperationalaspectsoftheresearchcarriedout,resulting

inmanyfruitfuloutcomes.HewastheheadoftheCropPhysiologychairgroupatmy

departmentCSA.Nevertheless,hewasalwaysavailablewheneverIneededhisadvice.

Healwaysguidedmewithhisquickresponses,sharpanalyses,andnolessimportant,

great mentorship. Whenever necessary, he kept me going with a few stimulating

words.

DearPaul,Iamverygratefulforyourunreservedsupport,concern,day‐to‐day

guidance, enthusiasm, and keen interest in the progress of the research and

numerous constructive discussions throughout my study, which were always

encouragingand forsparingmuchofyourprecious timeeditingand improving the

manuscripts and chapters of my thesis. The door of your office was always open,

encouragingmetoenterwithoutanxiety.Yourswiftnessingoingthroughdrafts,even

during your journeys to a faraway destination is somuch appreciated. I was very

luckyhavingyouasmysupervisor.Paul, thankyouverymuchforhelpingmeinall

my queries, for editing themanuscripts, for you always helpful suggestions during

ourprojectmeetings, for yourworthy letters on gettingmyextensions and solving

administrativeissues.Thanksalsoforyourpatienceandfornevergivingupwithme.

You will remain my best role model for a scientist, mentor, and teacher. I am

privilegedtohavehadtheopportunitytobeyourstudentasItrulybelievethatyou

are the best supervisor I ever could have had to pursue my doctoral studies. My

special thanks for introducing me to the fascinating world of “potato science”, for

sharing with me your incredible knowledge and ideas about the potato crop and

physiology. Your enthusiasmand love for science is contagious.Honestly, Iwish to

workwithyouagaininthefuture! Ialsowishtoextendmysincereheartfeltthanks

to yourwife Prof. dr. ir. EdithT. Lammerts vanBueren for her kindness and great

hospitality.

Iamdeeplygratefultomyco‐promotorProf.dr.ir.FredA.Eeuwijkforsharing

hisknowledgeandexperienceontheappliedstatisticalgenetics,forhisdetailedand

constructivecomments,andforhisoverall inputandimportantsupportthroughout

Acknowledgements

233

this work. Prof. Fred provided countless valuable comments and suggestions to

improvethequalityofthisthesis.

DearFred,thanksalotforyourcriticalviewsonmyresearchworkduringour

teammeetingsandpositiveandveryhelpfulcommentsonmydraftmanuscripts,for

ourdiscussionswhichwerealwaysacontinuoussourceofinspirationformeandalso

foryourmotivation.Iamdeeplygratefulforyourpositivefeedbacks,expertguidance

on statistical genetics, insightful comments and suggestions throughout the study

whichwere essential to strengthening the finalwork. Thank you for being patient

withme.Itwasanhonourworkingwithyou!

Iwillforeverbethankfultomydailysupervisordr.XinyouYin,forhisexcellent

guidance duringmy PhD study.Workingwith Yin, I learned that Imust look after

every detail; to be precise and concrete. Throughout my thesis‐writing period, he

providedencouragement,soundadvice,goodteaching,andlotsofgoodideas.Iwould

have been lost without him. His detailed comments on my draft manuscripts

improvedconsiderablymyability towritescientificpaperswhilehisoverallefforts

guaranteed high quality of this thesis. Xinyou is a great teacher and mentor. He

encouragesandexpectedmetothinkmoreindependentlyaboutmyexperimentsand

results.

Dear Yin, I greatly admire your high level of expertise in the field of crop

modellingandyetmodesty,yourcapacityofunderstandingandanalysingasituation

in the breadth of your knowledge. Your door was always open for me to discuss

whateverdoubtsIhadinmymind.Youwerealwaysnexttome,anytimeoftheday,

providing ample scientific knowledge and vital advices for my life. You were

instrumentalinthemodellingexercises,thanksforintroducingmetothewonderful

galaxyofcropmodelling.Youconstantlychallengedmetogetthebestoutofmeand

thiswasalwaysanencouragingexperienceforme.Ourconversationsenlightenedmy

wayofthinking,andsoIwouldliketogivemysincerethanksforyourgeneroushelp.

Youremainedasupporterandprovidedinsightanddirectionrightuptotheend.For

this,Icannotthankyouenough.Iamforevergrateful.Ialsothankyourfamilyforall

the hospitality and nice time duringNewYear dinners. Thank you Yin; it’s a great

pleasureworkingwithyou!

Iwouldalsoliketogiveaheartfelt,specialthankstomyseconddailysupervisor

Dr. ir.HermanJ.vanEckforhisknowledgeableadvice,encouragementandpositive

perspectivesatalltimes.Hehasbeenmotivating,encouraging,andenlightening.He

remainedagreatsupport.Hehasbeenhelpfulinprovidingadvicemanytimesduring

my PhD period. Dr. Herman has provided very insightful discussions about the

research.IamalsoverygratefultoHermanforhisscientificadviceandknowledgeon

potatobreedingandgenetics,andformanyinsightfuldiscussionsandsuggestions.

Acknowledgements

234

Dear Herman, thank you very much for your valuable comments and

suggestions.Yourwideknowledgeaboutthepotatogeneticsandyourlogicalwayof

thinking has been a good basis for the present thesis. I wish I could have learned

moreaboutfascinatingscienceofpotatogeneticsfromyou.

Performingatopqualityresearchisnotpossibleintheabsenceofqualitydata

sets.IwasreallyfortunateenoughtohaveIng.PeterE.L.vanderPutteninmyteam.

Peterneedsaspecialwordofthanks.Hehasbeenindispensableinmyexperimental

work.Heisaverydedicatedworkerandhehadalwaysthepatiencetoexplaintome

whateverneeded.

DearPeter, I reallyadmireall yourhardworkanddedication in settingupof

fieldexperiments,collectionandcompilingdata.Therichdatayoucollectedisindeed

astrongfoundationofthisthesisandhelpedmetofocusmoreonmymodellinggoals.

Iamverygrateful toyouforprovidingmethenecessarydatafilesandteachingme

thepracticalaspectsofpotatoexperiments. Iwouldhavebeenunabletomake it to

whereIamtoday!

Ialsowishtoextendmydeepfeltthanksandappreciationtotheotherexternal

membersofmythesiscommitteeProf.dr.ir.M.K.vanIttersum,Dr.J.‐P.Goffart,Prof.

dr. ir. A.J. Haverkort, Dr. ir. E. Heuvelink whose helpful suggestions increased the

readabilityandqualityofthisthesis.Thankyouverymuchfortakingyourprecious

timeinreadingmybook!

IwanttoextendmymanythankstopeoplefromBiometrisfortheiradviceand

helpinsolvingstatisticalproblemsonacoupleoftimeswhendirelyneeded.Special

thanks to Dr. Marcos Malosseti, Mr. Paul Keizer, and Dr. Martin Boer for their

excellentco‐operationandassistancewiththeGenstatprocedures.Specialthanksto

Dr.HansJansenforintroducingmetoJoinMapandforgenerouslyhelpingwiththe

QTLanalysis inGenstat. IwouldalsoliketoacknowledgethenicesupportofICTof

Wageningen UR for their prompt assistance and for providing the necessary

softwares.TheWURlibrarystaffofalsodeservesspecialthanksfortheirunreserved

assistanceandco‐operation.

I am also grateful to all the staffmembers of theWageningen Plant Sciences

Experimental Centre (Unifarm), without the continuous support and assistance of

whommyresearchworkwouldhavebeenpracticallyimpossible.

IspentallmytimeatthedepartmentofCentreforCropSystemsAnalysis(CSA).

IconsidermyyearsspentatCSAtobeacruciallearningperiodandmorethanthata

great privilege. Tome CSA is also synonymouswith “the Centre of Excellence”, an

entity inWageningenURwith itsownflavouranduniqueness inscience;providing

topqualityresearchsupportandtrainingonmultifacetedaspectsofplantproduction,

fromgenotypestocropsystemsanalysisandproductionchains.Iamdeeplygrateful

Acknowledgements

235

toallthestaffmembersofCSAfortheirencouragement,support,andexpertadvice

whenever required. Thank you very much for offering me such an excellent

atmosphere, for making me feel at home, and for being so helpful and friendly.

Thanks for all the nice social activities, PV events, for the Dutch dinners, and

particularlyfunwiththetypicalDutchgames.IkzalmijnverblijfbijCSAaltijdblijven

herinnerenalseenvandebestetijdendieikooitgehadheb.

My life was easier in the Netherlands thanks to the professionalism and

dedicationoftheCSAstaffmembers.MyspecialthanksgotoMrs.JennyElwood,Mr.

EugeneGoosens,Mr.SjaakTijnagel,Mrs.WampievanSchowenburgandMrs.Sjanie

vanRoekel,fortheirhelpintheadministrativematters,Mrs.GijsbertjeBerkhoutand

Mr.Alex‐JandeLeeuwfortheirassistanceinfinancerelatedmatters,Dr.Ing.Aadvan

Ast,Mrs.Ans (J.E.)Hofman, and Ing.HenrietteDrenth for theirkindassistanceand

help on the technical issues.Many thanks for taking over all theworries frommy

head so that I could just concentrate onmy studieswith total dedication. Bedankt

allemaalvoorjullieadviesenhulp!

DearJenny,thankyouverymuchforyourwarmthandkindnessandhelpingme

withmyinitialdaysofPhDstudies.DearSjanieandWampie,thankyouverymuchfor

alwayslisteningtomyproblemsandforshowingalotofcare,help,andsupport,this

meanalottomeandIamverygrateful forthat.Sjanie, ikspreekhierbijmijngrote

waarderinguitvoorjeaanmoeding,supportenhetaltijdbereidzijnommijmeteen

groteglimlachtehelpen.Bedanktvoorallehulpenbedanktommijtestimulerenom

meerNederlandstepraten!Wampie, ikdankjouhartelijkvoordesupportdejeme

altijdgaf,vooraljehulpvoorde‘IND’en‘Zorgtoeslagzaken’.

Iamalsograteful todr.ClaudiusvandeVijver(coordinatorof theC.T.deWit

Graduate School for ProductionEcology andResource Conservation) for his advice

and assistance during the course of my PhD studies and for organising excellent

informative PhD related programmes, PE&RC weekends, PE&RC days, etc. I really

enjoyedmyselfbeingparticipantoftheseprogrammesandlearnedalotabouthowto

organisemyPhDstudieswell.

IwouldliketothankallthecolleaguesandfriendsoftheCSAanddepartment

PlantProductionSystems (PPS) for thegood chatswehadduring coffeeand lunch

breaks,andseminarsandgatheringsandcolleaguesforprovidingastimulatingand

funenvironments. Ihavereallyenjoyedsharingexperienceswithyouallduringthe

coffeebreaksandPVevents.

IamverygratefultotheformerandcurrentstaffmembersoftheCSA:Prof.dr.

ir.NielsP.R.Anten,Prof.dr.HolgerMeinke,Prof.dr.Huub(J.H.J.)Spiertz,Dr. ir. Jan

Vos, dr. ir. Wopke van derWerf, Dr. ir. Willemien J.M. Lommen, Dr. ir. Tjeerd‐Jan

Stomph,Dr.ir.LammertBastiaans,Dr.JochemB.Evers,Dr.GerhardBuck‐Sorlin,Dr.

Acknowledgements

236

CorA.Langeveld,Dr.PepijnvanOort,Dr.ir.Felix(J.J.A.)Bianchi,andtomycolleagues

andfriendsfromPPS:Prof.dr.ir.HermanvanKeulen,Prof.dr.KenGiller,Prof. dr. ir.

Martin van Ittersum, Dr. ir. Bert Janssen, Dr. ir. P.A. Leffelaar, Dr. Nico de Ridder, Ir.

JoostWolf, Ing. Elco Meuter, Ing. Bert Rijk, Dr. Ir. Gerrie van de Ven, Dr. ir. Maya

Slingerland,Dr.MarkvanWijk,Dr.PytrikReidsma,Mr.MarcelLubbers,Mrs.Riavan

Dijk,andMrs.CharlotteSchilt.

Iamgratefultoallfellow(formerandcurrent)PhDsandfriendsfrombothCSA

andPPS,forbeingthesurrogatefamilyduringthemanyyears.Thanksaredueto(in

alphabetical order): Aart van der Linden, Abdoul Aziz Niane, Aisha Abdulkadir,

AmbechaGemechis,AndréMonteiro,Dr.ArgyrisKanellopoulos,AshwiniNarasimhan,

BenjaminKibor,BennoBurema,CharlesBucagu,ChristiaanBiemond,ClaraVenegas,

DechassaLemessa,Dr.DonaldGaydon,EssegbemonAkpo,EstherRonner,Dr.Gisella

Cruz García, Junfei Gu, Herman Berghuijs, Inga Kurnosova, Joyce Challe, Julienne

Fanwoua, JunqiZhu, LennyvanBussel,Dr.MadeleineFlorin,MarciaThaisdeMelo,

Marieke Sassen, Maryia Mandryk, Masood Awan, Mazhar Ali Khan, Meike van

Opeusden, Dr. Michiel de Vries, Murilo de Arruda, Dr. Myriam Adam, Nicodème

Fassinou,NiteenKadam,NoméSakané, Dr.Rik Schuiling,RobertOkello, Sanderde

Vries, Dr. Sander van Delden, Sheida Sattari, Dr. Sinead O’ Keeffe, Sjaak Aben, Dr.

Sotiris Archontoulis, Dr. Susana Akrofi, Tommie Ponsioen,Wenfeng Cong,Wenjing

Ouyang,WietskevanderStarre,WillemWesterhuis,andYangYu.Thankyouall for

makingourgroupsmorevibrantandenjoyabletowork.

Iwishtoextendmysincerethanksandappreciationtomydearcolleagueand

best friend Junfei Gu for the great friendship and for helping me get through the

difficult times, and for all the emotional support, camaraderie, and caring he

provided.IwishyoulotsofsuccessandgoodluckwithyourPhDstudies.Thanksto

Dr.Gerhardforourendlessscientificdiscussionsrangingfromcropmodellingtovast

scientific matters; thanks also to your wife Evelyne and cute Coquinette for the

excellenthospitality.DearDr.JanVosandProf.dr.ir.HuubSpiertz,Iamverygrateful

to the great hospitality offered by you and your families. Dear Joost, I will always

remember our interesting talks and discussions during the coffee breaks and

sometimes in the late hours. Thanks for your encouragement, support, and expert

guidanceondiversescientificmatters.

Iwould liketothankmyoffice‐matesDr.GisellaCruzGarcía,MazharAliKhan

andMasoodAwan,forbeingveryhelpfulandsupportive.DearGisella,thankstoyou

andyourhusbandPaul,forthegreathospitalityandcare.DearAliandAwan,thanks

for always standing bymy side in times of need. I really felt like at homewith your

company.IwishyougreatsuccesswithyourrespectivePhDprojects.

IwarmlythankMarjoleinTiemens‐HulscherandCesarAndresOspinaNietofor

Acknowledgements

237

becomingmyparanymphs. Your presence and support onmy special day is highly

appreciated.Thegreatthingisthatweallsharethesameinterestsandambitions,i.e.

toexplorethefascinatingworldofpotatoscience.Inthisrespect,Iamreallyproudof

youbeingmyparanymphs!DearCesar,thanksforbeingagreatcolleagueandfriend.

IalsowishyoulotsofsuccessandgoodluckwithyourPhDproject.

Iwouldalsoliketothankmycolleaguesandfriendsfromotherdepartmentsof

WUR, particularly (in alphabetical order): AhmedAbd‐El‐Haliem,Dr. Anna Finkers,

Dr. Anitha Kumari, Dr. Arwa Shahin, Bas Allema, Dr. Bas Engel, ChristosKissoudis,

DennisOonincx, Dr. Fradin Emilie, Dr.Gerard vanderLinden,Dr.HenkSchouten, Jet

Vervoort, Jimmy Ting Hieng Ming and wife, Mansoor Karimi, Mohamad Arif,

Mohamed El Soda, Dr. Nasim Irani, Nasr Ahmed, Partha Sarathy, Paula Ximena

Hurtado,PaulinaKerbiriov,QiiWeicong, Dr.SameerJoshi,SuxianZhu,TarekSoliman,

Thijs van Dijk and Yusuf Sen. Thank you very much for your support and great

interaction.

Imetwithmanyinterestingpeoplefromaroundtheworldandmademanynew

friends during my studies in Wageningen. I consider Wageningen “the city of life

sciences” truly my second home now. I really enjoyed and learned about a wide

spectrumofcultureshere. Iwould liketo thankmyfriendsandcolleagues fortheir

lovely friendship and support and making my time in Wageningen the most

memorable in my life. Many thanks to you my good friends, particularly (in

alphabetical order): Alba Tros, Alberto Franco Solís, Aleksander Banasik, Alisa

Shlyuykova,AntonAgarev,AtivanderHoning,AzizPatwary,BenoitCarréres,Daniel

Kinpara, Diego Chacon, Dimitris Mitsopoulos, Dmitry Zolotarev, Elena Sokolova,

Eliana de Souza, Dr. Fareeduddin Noori, Habibullah Asad, Hamed Rashidi, Inga

Kurnosova,JanneDenolf,KhanWaliKamran,KunHan,LiyakatMujawar,LolaRomera,

Manuel Jesùs, Mariya Tarazanova, Maya Lievegoed, Rahatullah Mohmand, Romi

Gaspric,SandraDelgado,SatyaSriram,SergeyKleymenov,Dr.SouravBhattacharjee,

Tanya Huizer, Tooryalay Nasery, Victoria Naipal,Wasil Akhter,Wei Qin andmany

more.Icouldnotincludeeverybody’snameasthelistisverylong.Yoursalutations,

smiles, jokes, encouragements, social, and academic discussions were simply

infectious.

DearAlberto,thanksforyourwarmthandunselfishfriendship,forallthegood

time we had together, for you great sense of humour. I immensely enjoyed your

company.IwishyougreatsuccessandgoodluckwithyourPhDstudies.

Thanks tomyTurkish friends for yourwarmwelcome and hospitality onmy

visittoTurkey,especially:Prof.Dr.MehmetEminÇalışkan,Dr.FundaArslanoğlu,Dr.

ErdoğanÖztürk,andDr.YasinBedrettinKaran.Manythanksforyourbestwishesand

encouragementevenwhenfaraway,çokteşekkürederim!

Acknowledgements

238

IwouldalsoliketothankmyPakistanifriendsandfamiliesinWageningenand

elsewhereintheNetherlands.Myheartiestgratitudeto(inalphabeticalorder):Abbas

Gul, Abid Maan, Akmal Nazir, Dr. Farrakh Mehboob, Farooq Ahmad, Ghulam Abbas,

GhulamMustafa, Hafiz Sultan, Haider Ali, Hamid Shah, HashimDurrani, Hazrat Ali,

Imran Bhatti, Imran Haider, Kashif Khan, Dr. Luqman Aslam, Mateen Ahmad, Dr.

MazharAli,Dr.MazharIqbalZafar,MubarakAli,MuhammadImtiaz,Dr.Muhammad

Irshad, Dr. Muhammad Jamil and family, Muhammad Rashid, Muhammad Sohail,

MunawarHussain and family, Dr. NadeemKhan, Dr.Nazir AhmadKhan,Noorullah

Khan,SaadGilani,Dr.SabazAliKhan,SadeeqUrRahman,SaifMalik,Dr.SajidRehman

and family,ShafqatQaisrani,SuhailYousaf,ZakirDahri,andZeshanHassan. I spent

mostjoyousandpleasanttimewithyouinWageningen.Iwishyoumuchsuccessand

goodluckwithyourcurrentandfutureendeavours.

Isharedaspecial friendshipwithDr.NazirAhmadKhan,NoorullahKhan,and

Dr.SabazAliKhan.Iwillrememberourinterestingculturaldinnerevenings,ourtalks

overpracticallyanythingonthisearth,especiallyoverhistory,politics,Pashtomusic,

younameit,althoughImustadmitourfavouritetopicwashowtoimproveanduplift

theeducationsystemofPakistan.

Iwould like to conveymyheartfelt thanks tomyhome institute, Agricultural

Research Institute (ARI), D.I.Khan for their companionship and encouragement. I

thankthestaffofARI for theirgeneroussupport. Iconsidermy initial jobperiodat

ARIasacruciallearningperiodofmylife.

My sincere thanks go to all the institutions andorganisations,which gaveme

theirindispensablegeneroussponsoring,withouttheirsupportandfinancialhelp,it

would not have been possible for me to pursue and to complete this PhD project

successfully.MysincerethanksareduetotheHECPakistan,forprovidingmeafully

fundedscholarshiptofulfilmydreamofhighereducationintheNetherlands.Iwould

liketoacknowledgethefinancial,academic,andtechnicalsupportofentireHECteam

during the course ofmy studies. Iwant to thank theNetherlands Organization for

InternationalCooperationinHigherEducation(NUFFIC)fortheirunreservedsupport

and excellent assistance and co‐operation. I am thankful to the European Union

SeventhFrameworkProgrammeunderproject (NUE‐CROPSFP7‐CP‐IP222645) for

supportingmystudies.IamverygratefultomyhostdepartmentintheNetherlands

i.e. CSA for offering me necessary financial support during the end phase of my

studies.

Ofcoursenoacknowledgmentswouldbecompletewithoutgivingthankstomy

parents.ThecontinuoussupportandencouragementIreceivedfrommyfamilyback

in Pakistan was a great source of strength. Without their encouragement and

understandingitwouldhavebeenimpossibleformetocompletelydevotemyselfto

Acknowledgements

239

myPhDstudies. Iwish toexpressmyeverlastingsenseofgratitudeand lovetomy

affectionate parents, for their endless patience, relentless support, understanding,

andencouragementall thewaythroughoutmystay intheNetherlands.Yourcaring

wordsofencouragementandconstantassurancehavehelpedmegetthroughsomeof

themostdifficulttimes.Icouldnothavedoneitwithoutyou!Mostimportantofallis

myfatherDr.RahmatUllahKhan,whostimulatedmecontinuouslyduringmystudies.

Hehasbeena constant inspiration.Mymother, especially,playedavery important

roleduringmyPhD,providingmewithherample love,moral supportandshowed

greatpatienceandunderstandingduringalltheseyears.Shewasagreatrolemodel

ofresilienceandstrength.Iowemyeveryachievementtobothofthem.Thisthesisis

alsodedicated toyoubothasa small tokenof appreciation foreverything thatyou

havedoneformealltheseyears.Myspecialgratitudeisduetomygrandmotherfor

herblessings,tomybelovedyoungerbrotherYasirKhan,mysistersMariyaKhanand

UmaimaKhan,mybrother‐in‐lawsDr.KashifWaseemKhan,AdnanKhanand their

families, for their loving unequivocal support throughout, as always, forwhichmy

mereexpressionofthankslikewisedoesnotsuffice.

MystudytooklongtimeandImayhaveneglectedtoacknowledgeotherpeople

whohavehelpedmealongtheway.Ifso,pleaseacceptmyapologies,andknowthat

youareappreciated.

ToallthoseI’vemetalongthewayinsomewayaffectedme,mysincerethanks!

MuhammadSohailKhan

Wageningen,September2012

241

CurriculumvitaeMuhammadSohailKhanwasbornon6thJune1981

inKhyberPakhtunkhwa,Pakistan.Hecompletedhis

basic edu cation in 1998 and a fewmonths later he

entered as a student at the Faculty of Agriculture,

Gomal University, Dera Ismail Khan, Khyber

Pakhtunkhwa,Pakistan. In2002,heobtainedhis4‐

yrB.Sc.(Hons)Agriculturediplomawithdistinction.

HecontinuedhisstudiesinthesameuniversityforhisMastersandobtainedhis2‐yr

M.Sc.(Hons)Agriculturediplomawithdistinction(GoldMedal)in2004.Hechoseto

specialize in the field of Horticulture during his Masters studies and worked on

evaluatingtheeffectsofdifferentauxinsonthegrowthandestablishmentofDamask

Rose(RosadamascenaMill.)cuttingsindifferentgrowingmedia.

SoonaftercompletinghisMasters,hebrieflyworkedasaResearchAssistantat

the Arid Zone Research Institute Dera Ismail Khan, Pakistan Agricultural Research

Council (PARC). His work mostly involved developing improved production and

propagationtechnologiesforaridhorticulturalfruits.In2005,hewasappointedasa

ResearchOfficerattheHorticultureSectionofAgriculturalResearchInstitute(ARI),

DeraIsmailKhanundertheauspicesofAgriculturalResearchGovernmentofKhyber

Pakhtunkhwa, Pakistan. He was mainly involved in the conservation of elite

indigenous fruit germplasm as well as multi‐environment evaluation of different

exotic cultivars of important regional fruits: Datepalm (Phoenixdactylifera),Mango

(MangiferaindicaL.),Ber(ZizyphusjujubaL.),Falsa(Grewiaasiatica),etc. It is there

thathebecamefascinatedwithstudyingtheroleofG×Einteractiononplantgrowth

andphysiology.

Inearly2007,hewasawardedan“OverseasPhDScholarship”fromtheHigher

Education Commission of Pakistan (HEC) in collaboration with the Netherlands

Organization for International Cooperation in Higher Education (NUFFIC). A few

monthslater,hecametoWageningenUniversity,theNetherlandstopursuehisPhD

study under the supervision of Prof. dr. ir. Paul C. Struik (Centre for Crop Systems

Analysis, (CSA)),Prof.dr. ir.FredA.vanEeuwijk (Biometris),Dr.XinyouYin (CSA),

andDr.ir.HermanJ.vanEck(LaboratoryofPlantBreeding).HisPhDresearchwason

“Assessing genetic variation in growth and development of potato (Solanum

tuberosumL.)”whichresultedintotheoutcomeofthisthesis.AfterdefendinghisPhD

thesis,hewilljoinhisparentinstituteinPakistan.Hehasdevelopedastronginterest

incropphysiology,quantitativegenetics,andmodel‐basedplantbreeding.Hewould

liketoinvesthisfutureresearcheffortsexcellingintheseareas.

PE&RCPhDEducationCertificate

With the educational activities listed below the PhD

candidate has complied with the educational

requirementssetbytheC.T.deWitGraduateSchoolfor

ProductionEcologyandResourceConservation(PE&RC)

whichcomprisesofaminimumtotalof 32ECTS(=22

weeksofactivities)

Writingofprojectproposal(6ECTS)

- Genotype‐by‐environment (G×E) interaction on the dynamics of yield

formationinpotato(SolanumtuberosumL.)(2012)

Writingofprojectproposal(4.5ECTS)

- Assessing genetic variation in growth and development of potato (Solanum

tuberosumL.)(2008)

Post‐graduatecourses(6ECTS)

- Artofmodelling;PE&RC(2008)

- Modernstatisticsforthelifesciences;Biometris(2009)

- Multivariateanalysis;PE&RC(2010)

Deficiency,refresh,brush‐upcourses(3ECTS)

- Cropecology(2008)

- Systemsanalysis,simulationandsystemsmanagement(2008)

Competencestrengthening/skillscourses(6.6ECTS)

- Informationliteracy,includingintroductiontoEndnote;WGS(2008)

- Projectandtimemanagement;WGS(2008)

- Scientificwriting;CENTA(2009)

- Techniquesforwritingandpresentingscientificpapers;WGS(2009)

- Presentationskills;CENTA(2010)

- Howtowriteaworld‐classpaper;LibraryWageningenUR(2011)

PE&RCAnnualmeetings,seminarsandthePE&RCweekend(1.2ECTS)

- PE&RCWeekend(2008)

- PE&RCDay(2010)

PE&RCPhDEducationCertificate

244

Discussiongroups/localseminars/otherscientificmeetings(5.8ECTS)

- SENSEContextsymposium;WageningenUR,theNetherlands(2007)

- Ontheevolutionofplant‐pathogeninteractions: fromprinciplestopractices;

WageningenUR,theNetherlands(2008)

- 40YearstheoryandmodelatWageningenUR,theNetherlands(2008)

- Plant roots from gene to ecosystem; Radboud University, the Netherlands

(2008)

- Ecologyandexperimentalplantsciences2;WageningenUR, theNetherlands

(2009)

- Biometrisstatisticalgeneticsmeeting(2009‐2012)

- The potato root system; Plant Breeding, Wageningen UR, the Netherlands

(2010)

- YoungPSGmeetings(2010)

- Plantsciencesseminars(2010‐2012)

- PlantBreedinginthegenomicsera;WageningenUR,theNetherlands(2011)

Internationalsymposia,workshopsandconferences(5.6ECTS)

- InternationalEucarpiaCongress,“PotatobreedingaftercompletionoftheDNA

sequenceofthepotatogenome”;WageningenUR,theNetherlands(2010)

- International symposium on agronomy and physiology of potato, “Potato

Agro‐physiology‐2010”;Nevşehir,Turkey(2010)

Lecturing/supervisionofpractical’s/tutorials(0.6ECTS)

- Researchmethodsincropscience;2days(2010)

Supervisionof4MScstudents;7days

- Model based estimation of parameters describing the dynamics of canopy

covergrowthinalargesetofpotato(SolanumtuberosumL.)genotypes

- Estimationofbetathermaltime

- Non‐linearregressioninSAS

- MixedmodelanalysisinGenstat

245

Funding

The research work described in this PhD thesis was co‐funded by the European

Community financial participation under the Seventh Framework Programme for

Research, Technological Development and Demonstration Activities, for the

Integrated Project NUE‐CROPS FP7‐CP‐IP 222645. The views expressed in this

publicationarethesoleresponsibilityoftheauthoranddonotnecessarilyreflectthe

views of the European Commission. Neither the European Commission nor any

personactingonbehalfoftheCommissionisresponsiblefortheusewhichmightbe

made of the information contained herein. Scholarship grant to the author was

grantedbytheHigherEducationCommission(HEC),GovernmentofPakistanunder

“OverseasPhDScholarshipProgramPhase‐II”andCentreforCropSystemsAnalysis

(CSA).