Development of a pearl millet Striga-resistant genepool: Response to five cycles of recurrent selection under Striga-infested field conditions in West Africa

(1)

ContentslistsavailableatScienceDirect

Field

Crops

Research

j o ur na l h o me pa g e :w w w . e l s e v i e r . c o m / l o c a t e / f c r

Development

of

a

pearl

millet

Striga-resistant

genepool:

Response

to

ﬁve

cycles

of

recurrent

selection

under

Striga-infested

ﬁeld

conditions

in

West

Africa

Boubacar

A. Kountche

a,b,d,e

_,

_C.

_Tom

_Hash

b

_,

_Harouna

_Dodo

b

_,

_Oumarou

_Laoualy

a,b

_,

Moussa

D. Sanogo

c

_,

_Amadou

_Timbeli

c

_,

_Yves

_Vigouroux

d,e

_,

_Dominique

_This

e

_,

Randy

Nijkamp

f

_,

_Bettina

_I.G.

_Haussmann

b,∗

a_University_Abdou_Moumouni,_Faculty_of_Agronomy,_BP_10960,_Niamey,_Niger

b_{International}_Crops_Research_Institute_for_the_Semi-Arid_Tropics_(ICRISAT),_BP_12404,_Niamey,_Niger c_Institut_d’Economie_Rural_(IER),_Station_de_Cinzana,_BP_214,_Segou,_Mali

d_Institut_de_Recherche_pour_le_{Développement}_(IRD),_BP_64501,₃₄₃₉₄_Montpellier_cedex_5,_France e_Montpellier_SupAgro,₂_Place_Viala,₃₄₀₆₀_Montpellier_Cedex_1,_France

f_Crop_and_Weed_Ecology_Group,_Wageningen_University,_PO_Box_430,₆₇₀₀_AK,_Wageningen,_The_Netherlands

a

r

t

i

c

l

e

i

n

f

o

Articlehistory: Received17April2013

Receivedinrevisedform10July2013 Accepted11July2013

Keywords: Strigahermonthica Downymildew Cenchrusamericanus Gainfromselection Resistancebreeding

a

b

s

t

r

a

c

t

Strigahermonthica(Del.)Benth.isapersistentthreattopearlmillet[Cenchrusamericanus(L.) Mor-rone,comb.nov.]production,especiallyinWestAfrica.Thisstudyaimedatevaluatingtheresponse ofadiversifiedpearlmilletgenepooltofivecyclesofrecurrentselectiontargetingStrigaresistanceand panicleyield,andtoalesserextentdownymildew[Sclerosporagraminicola(Sacc.)J.Schroet.]resistance. Two-hundredfull-sibfamilies(FS)representingtheC5selectioncyclewereevaluatedtogetherwiththe genepoolparentallandraces,experimentalvarietiesderivedfrompreviouscyclesandlocalchecksin Striga-infestedfieldsatSadoré(Niger)andCinzana(Mali).Substantialandmostlysignificantselection progresscouldbedocumented.Theaccumulatedpercentagegainfromselectionamountedto51%/1% lowerStrigainfestation(measuredbyareaunderStriganumberprogresscurve,ASNPC),46%/62%lower downymildewincidence,and49%/31%higherpanicleyieldoftheC5-FScomparedtothemeanofthe genepoolparentsatSadoré/Cinzana,respectively.Experimentalvarietiesselectedfrompreviouscycles alsorevealedlowerASNPCandmostlyhigheryieldcomparedtogenepoolparentsattheirselectionsites. SignificantgeneticvariationamongtheC5-FSandoperativeheritabilitiesof76%(Cinzana),84%(Sadoré) and34%(combinedacrosslocations)forASNPCwillenablecontinuedselectiongainforStrigaresistance. Highgenotype×environmentinteractionvariancesforalltargettraitssuggestthatdifferent experimen-talvarietiesneedtobeextractedfromthegenepoolfordifferentsites.Thegenepool-derivedvarieties willbefurthervalidatedon-farmandareexpectedtocontributetointegratedStrigacontrolinpearl milletinWestAfrica.

Abbreviations:ASNPC,areaunderStriganumberprogresscurve;das,daysaftersowing;DM,downymildewincidence;E,environment;FLO,timeto50%ﬂowering;G, genotype;G×E,genotypebyenvironment;HYD,panicleyield;ICRISAT,InternationalCropsResearchInstitutefortheSemi-AridTropics;IRD,InstitutdeRecherchepourle Développement.

∗ Correspondingauthor.Presentaddress:UniversityofHohenheim,InstituteofPlantBreeding,SeedScienceandPopulationGenetics(350a),70599Stuttgart,Germany. E-mail addresses: [email protected] (B.A.Kountche), [email protected] (C.T. Hash), [email protected] (H. Dodo), [email protected] (O. Laoualy),

[email protected](M.D.Sanogo),[email protected](Y.Vigouroux),[email protected](D.This),[email protected](R.Nijkamp),

http://dx.doi.org/10.1016/j.fcr.2013.07.008

Open access under CC BY license.

(2)

1. Introduction

Pearl millet [Cenchrus americanus (L.) Morrone, comb. nov.] (Chemisquy et al., 2010)is thestaple food and fodder cropof millionsof poor ruralfamilies in Africaand India (Hashet al., 2000;Senthilvel etal.,2008).Inthearidand semi-aridregions wherethiscerealiscultivated,pearlmilletplaysanimportantrole in foodsecurity.However, grainyieldsare generallylow (circa 350–600kgha−1).Grainyield is limited by abiotic stresseslike droughtandlowsoilfertility,andbioticstressessuchasdowny mildewdiseaseandtheparasiticweedStriga.

Strigahermonthica (Del.) Benth.is themostwidespread and destructivewitchweedaffectingcereals.IntheStriga-proneregions ofsub-SaharanWesternandCentralAfrica,S.hermonthicaremains themajorbioticandpersistentthreattopearlmilletproductionand productivity(Wilsonetal.,2004).About40%ofthecerealproducing areasisseverelyinfestedwithStrigainsub-SaharanAfrica(Gurney etal.,2006).Grainyieldlossesupto100%havebeenreportedin susceptiblecerealcultivarsunderhighinfestationlevels, particu-larlyunderdroughtconditions(Gurneyetal.,2006;Ejeta,2007; Amusanet al.,2008).Severaloptions have beenrecommended forStrigacontrol:handweeding,herbicides,rotationsofcashor trapcropswithcereals,improvedsoilfertility,intercroppingand biologicalcontrol(Haussmannetal.,2000).However,the diver-sityofthefarmingsystemsandoftheparasitehaverenderedthe useofa singlecontrolmethodineffective,leadingtothe devel-opmentofintegratedStrigacontrolpackagesthatcombineseveral controlmeasures(e.g.,Badu-Apraku,2010).Breedingcultivarsthat canwithstandparasiteinfectionisconsideredtobecentralto inte-gratedapproachestoStrigacontrolinpearlmillet.Employinghost resistancetoS.hermonthicaissimpleandeconomicalfor subsis-tencefarmerstoadopt(Ejeta,2007;MenkirandKling,2007;Wilson etal.,2000).

Inheritanceof traitsassociated withresistancetoStrigaspp. hasbeenreportedforsorghum,maizeandrice(Haussmannetal., 2001a,2004;Amusanetal.,2008;Gurneyetal.,2006).Thelow stimulationofS.asiaticaseedgerminationhasbeenreportedto beunderthecontrolof a singlerecessivegene inthesorghum cultivars Framida,555, SRN 6496,and SRN 39 (Hessand Ejeta, 1992;EzeakuandGupta,2004).Onemajorgeneandseveralminor genesareinvolvedinstimulationofS.hermonthicaseed germina-tioninsorghum(Haussmannetal.,2001a),andthemajorgenehas recentlybeenmapped(Satishetal.,2011).FieldresistancetoStriga inmaizeisconsideredaquantitativelyinheritedtrait(Kim,1994; GethiandSmith,2004;Amusanetal.,2008).Theheredityof host-plantresistancetoStrigainpearlmilletisnotwelldocumented. Fewpearl milletcultivars havebeenreportedaspartially resis-tantortoleranttoStriga,andresistance(oratleastalowerlevel ofsusceptibility)incertainpearlmilletmaterialswasshowntobe dominant(Ramaiah,1987).However,theveryexistenceofStriga resistanceincultivatedpearlmillethasbeenquestionedbyothers (ChisiandEsele,1997).Lackofpreciseandvalidatedinformation aboutStrigaresistanceinpearlmilletmaybepartiallyduetothe factthatthepearlmillet/S.hermonthicapathosystemisparticularly complex.Bothhostandparasitearehighlyout-crossingspecies, whichresultsineach plantinapearlmilletlandrace,improved open-pollinatedvarietyor genepoolpopulationhavinga differ-entgenotypeandthereforecarrying potentiallydifferentalleles forStrigaresistanceorsusceptibility.Similarly,eachS.hermonthica plantinanaturalpopulationcarriespotentiallydifferentallelesfor virulence.

PartialquantitativeresistancetoS.hermonthicawasreportedin wildpearlmilletrelativesfromAfrica(Wilsonetal.,2000,2004). However,fromabreeder’spointofview,itappearspreferableto workwithinadynamicanddiversiﬁedgenepoolofcultivatedpearl milletsusingrecurrentselectionmethodsaswildmilletrelatives

carrymanyundesirableallelesthatneedtobeeliminated,thereby slowingdownselectionprogress.

Downymildew[causedbytheoomyceteSclerospora gramini-cola(Sacc.)J.Schroet.]isapseudo-fungaldiseaseofpearlmillet, causing up to 60% croploss (Singh, 1990; Thakur et al., 2011) and needs tobeconsidered in pearl millet improvement. Both Strigaanddownymildewpathogensarehighlyvariable,and dif-ferentecotypescanbepresentatdifferentlocations(Haussmann etal.,2001b;Thakuretal.,2006;Yoshidaetal.,2010),underlining theneedtoperformselectionatcontrastinglocationstoexploit genotype×locationinteractions and/or identifystablevarieties. Accordingtoourknowledge,thereisnoStriga-resistantpearlmillet varietyreleasedinWestAfrica.

Todate,increasedresistancetoStrigathroughrecurrent selec-tion has not been reported for pearl millet. Since 2006, the InternationalCropsResearchInstitutefortheSemi-AridTropics (ICRISAT)inNiger, inpartnershipwiththenationalagricultural researchprograminMali(Institutd’EconomieRurale,IER),has con-ductedrecurrentselectionina diversiﬁedpearlmilletgenepool toincreasefrequenciesofdesirableallelesforStrigaanddowny mildewresistanceandpanicleyieldusingtwotestsitesinWest Africa(SadoréinNigerandCinzanainMali).Intotal,thisgenepool wassubjectedtoﬁvecyclesofrecurrentselection,andprovidesa uniqueopportunitytoexaminetheeffectivenessofrecurrent selec-tionforimprovingStrigaanddownymildewresistance,andyield performanceincultivatedpearlmillet.

Theobjectivesofthisstudywere(i)todescribethegenepool developmentandselectionprocess,(ii)toexaminetheresponse toﬁvecyclesofrecurrentselectionforpanicleyieldandresistance toS.hermonthicaanddownymildew,(iii)toestimate quantitative-geneticparametersandpredictpossiblefutureselectiongainsfrom thegenepool’sC5-Full-sibpopulation.

2. Materialsandmethods

2.1. Striga-resistantgenepooldevelopment

TheinitialstepfordevelopingthepearlmilletStriga-resistant genepoolconsistedoffieldevaluationof64pearlmilletlandraces that had been collected in Striga-infested fields jointly by IRD (theFrenchResearch InstituteforDevelopment)and ICRISATin 2003(Bezançon etal., 2009).The 64 landraceswereevaluated ina four-replicatefieldtrialunderartificialStrigainfestationat the ICRISAT-Sadoré research station (13◦15 N, 2◦18 E) in the 2005rainy season.Six landraceswere identified asless Striga-sensitiveandrelativelyhigheryielding(M141,M239,M029,M197, M017and KBH).During the2005/2006off-season,both full-sib (FS)progenies(intra-andinter-population)andS1(self-pollinated) progenieswerecreatedfromthesesixselectedgenepoolparents (Table1).Theseprogeniesweresubjectedtothefirstcycleof eval-uationin the2006rainyseasonina four-replicatefield trialat ICRISAT-Sadoré. Separate,artificially infested fields (S1 and S2, sownwithintwodaysinterval)wereusedforevaluatingFSand S1progeniesinthisandthefollowingyears.Evaluationof Striga-resistantgenepoolprogeniesatCinzana,Mali(13◦17N,5◦56W) wasstartedlaterinthe2008rainyseasoninanaturallyinfested farmer’sfieldnearthestation.Thisfieldhadbeencarefullyselected forstronganduniformStrigainfestationinthepreviousyear.An overviewofthediversifiedStriga-resistantgenepooldevelopment processispresentedinFig.1.

Intheﬁrstandsubsequentcycles,selectionwasbasedonthe fol-lowingcriteria:above-averageStrigaresistance[i.e.below-average areaunder Striga number progress curve (ASNPC)(Haussmann etal.,2000)orStrigaemergencesupportedbytheprogeny], above-averagepanicleyield,andbelow-averagedownymildewincidence

(3)

Table1

Selectionprotocolsandphenotypicdataforeachcycleofrecurrentselection(2006–2010).

Selection cycle(year)

Selection method

Location No.ofprogeny Proportionof

selectedunits(%) Traits†

Evaluated Recombined ASNPC HYD DM FLO h2 _TM _h2 _TM _h2 _TM _h2 _TM C0(2006) Full-sib Sadoré 169 53 31 45.0 1308.5 53.9 126.7 61.5 37.5 44.7 85.7 S1 116 25 22 30.6 325.1 30.6 160.6 51.1 18.3 25.8 107.5 C1(2007) Full-sib Sadoré 164 33 20 29.1 354.9 47.8 190.0 43.8 7.5 48.2 72.5 S1 91 20 22 37.9 10.5 49.1 102.5 71.0 7.0 44.2 81.7 C2(2008) Full-sib Cinzana 70 – – 59.1 46.3 32.3 390.2 51.1 14.9 35.3 71.5 Sadoré 196 59 30 64.7 1263.0 37.0 291.9 61.3 19.9 47.9 74.8 S1 Sadoré 91 12 13 19.7 904.2 55.9 193.1 60.4 20.4 55.6 80.0 C3(2009) Full-sib Cinzana 183 65 25 57.6 2074.9 16.4 280.5 32.8 9.4 27.5 67.8 Sadoré 183 55.9 4265.9 34.1 483.5 60.5 7.2 72.4 42.6 C4(2010) Full-sib Cinzana 103 56 54 55.5 196.9 22.7 201.5 66.9 7.0 34.3 76.4 Sadoré 162 61.8 39.6 27.9 140.8 52.7 11.6 23.9 85.9 S1 Sadoré 55 16 20 29.9 244.2 54.5 207.8 28.9 22.5 55.2 67.4

† _ASNPC:_area_under_Striga_number_progress_curve;_HYD:_panicle_yield_(g_m−2_);_DM:_downy_mildew_incidence_(%);_FLO:_time_to_50%_ﬂowering_(days_after_sowing)._The_trial

means(TM)andbroad-senseheritability(h2₎_on_a_plot_basis_are_reported_for_each_trait.

(DM) (number of infested hills recorded during plant tillering andupdatedafterﬂowering,dividedbythetotalnumberofhills afteremergence,multipliedby100).Inlatercyclesalsoan over-allagronomicsuitabilityscorewastakenintoaccount,andonly entrieswithabove-averagescoreswereselected.Thepercentage ofselectedentriesrangedfrom13to54%,anddependedonthe numberofentriestested,thenumberofsuperiorentriesidentiﬁed forthedesirabletraitcombinations,andthedesiretohaveatleast 50progeniescontributingtothenextgeneration,inordertoavoid inbreedingdepressioninsubsequentgenerationsofthegenepool. In the following selection cycles, S1 and FS selection were alternated,withnewFSprogeniesbeingcreatedfromselectedS1

progeniesandbothnewFSprogeniesandS1progeniesbeing cre-atedfromtheselectedFSprogenies.Duringeachrecombination step,crossingplanswereestablishedtoassurethateachselected progenywascrossedtoseveralotherselectedprogeniesandalso thateachselectedprogenycontributed aboutthesamenumber of progenytothe nextgeneration.Thiswasdoneto maximize theeffectivepopulationsizeinthefollowinggenerationsofthe genepool,andtherebytoavoidinbreedingdepression.

Thegenepoolwaskeptopen,sothatpotentiallyinteresting lan-draces(reportedbyfarmerstobeStrigaresistant)wereincluded astestentriesduringtheevaluationcycles(allfour-replicateﬁeld trials),andifabove-averageStrigaresistanceandacceptableyield

Eva lu a ti on Test of 64 landraces for

selecting the

genepool parents C₂Evaluation using 196 FS & 91 S1 at Sadoré & 70 FS at

Cinzana (Mali)

& checks† C₃Evaluation at Sadoré & Cinzana using 183 FS & checks† C₄Evaluation using 162 FS & 55 S1at Sadoré & 103 FS at Cinzana & checks† Full-sib (FS) & selfed progeny (S₁) creation using 6 landraces representing the genepool parents

25 selected S1

used for new FS creation & 53 selected FS

used for both new FS/S₁ creation Creation of two experimental varieties C₅Evaluation &

Test of selection

progress at Sadoré

& Cinzanausing

200 FS & checks†

Introduction of M017 & PE5461 as new germplasm

Introduction of PE00702, PE05606

& Serkloc as new

germplasm Creatio_experimentaln of three

varieties

Rainy season 2005 Rainy season 2006 Rainy season 2007 Rainy season 2008 Rainy season 2009 Rainy season 2010 Rainy season 2011

Off-season 2005/2006 Off-season 2006/2007 Off-season 2007/2008 Off-season 2008/2009 Off-season 2009/2010 Off-season 2010/2011 Off-season 2011/2012

Re com b inat ion & Creat ion of new te st uni ts 20 selected S1

used for both new FS/S₁

creation

12 selected S1

used for both new FS/S₁

creation

65 selected FS used for both

new FS/S₁ creation

16 selected S₁ & 56 selected FS used for

new FS creation

28 selected FS used for new FS creation C₀Evaluation at Sadoré(Niger) using 169 FS & 116 S1 & checks† M017 & PE5461 C1Evaluation at Sadoré using 164 FS & 91 S1 & checks†

PE00702, Serkloc & PE05606

Creation of two experimental

varieties

Legend:

FS = Full-sib families; S1= Self-pollinated progenies; = New germplasm included in field evaluations; † Checks included genepool parents and other controls.

Fig.1. SchematicillustrationofthegenepooldevelopmentprocessfollowinganopenrecurrentselectionschemewithcyclicFSandS1evaluation,selectionandrecombination combinedwithintrogressionofnewgermplasmandextractionofnewexperimentalvarieties.

(4)

performancewasproven,thelandracewasincludedintheFSandS1 creationforthenextselectioncycle.Atotalofﬁvenewgermplasm wereincludedattworecombinationcycles(inC1 andC2,Fig.1). Basedonselectionsmadefrom2008and2009rainyseasondata, theﬁrstexperimentalvarietieswereextractedfromthegenepool (afterthe3rdand4thselectioncycle).Fortheexperimental vari-etiesderivedin2008,15bestprogeniesatSadoréwithlongpanicles (EV08-Sad-LP),and15bestprogeniesatSadoréwithshorter pan-icles(EV08-Sad-SP)wererecombinedduringthe2008/2009dry season.Fortheexperimentalvarietiesderivedin2009,the15best progeniesfromeachsite,Sadoré(EV09-Sad)andCinzana (EV09-Cin),andfromacombinedanalysisacrosstwosites(EV09-Comb) wererecombined.Withtherecombinationof15bestprogenies,a higherselectionintensitywasappliedfortheexperimental vari-etiescomparedtotheoneappliedtothegenepool.

Only one improvedpopulation/genepoolresulted from each selectioncycle.ThereasonsforusingS1recurrentselectionin com-binationwithFSrecurrentselectionwerethefollowing:selection amongS1 progeniesgiveshigherchancestoselectfordesirable recessive resistancealleles and has theoretically theadvantage oflargergeneticvarianceamongthetestedentries.Thevariable numbersofS1 andFSprogeniesincludedineachselectioncycle (Table1)wereduetovariablesuccessinFS/S1creationinthe off-seasonnurseries–duetoirrigationproblemsandconcomitantheat stress,grasshopperdamageorpoorseedsetting.A summaryof eachselectioncycleincludingevaluatedunits,recombinationunits, heritabilityandtrialmeanisgiveninTable1.

2.2. Evaluationofselectionprogressafterﬁveselectioncycles 2.2.1. Geneticmaterials

Thetestofselectionprogressafterﬁveselectioncycleswas inte-gratedintotherecurrentselectionprogenyevaluationprocess.A setof200C5-FSprogenieswereproducedinthe2010/2011dry sea-sonbyrecombining56FSprogeniesand16S1progeniesselected fromtheC4evaluationcycle,and3experimentalvarieties (EV09-Sad,EV09-CinandEV09-Comb).Remnantseedsofeachselected progenyweresowninaone-rowplotofsevenhillsandcrosses wereperformed accordingtoa deﬁned crossing planto assure thateachprogenywascrossedtoatleastthreedifferentselected progeniesandthateachprogenycontributedthesamenumberof progenies(about three)tothenextgeneration.For thepresent study,the200C5 derivedFSfamilies togetherwith20 controls [including genepoolparental varieties,5 experimentalvarieties (seeabove)andasetof8agronomicallyeliteopen-pollinated vari-eties(OPVs)]wereevaluatedforStrigaresistanceduringthe2011 rainyseason.Theseelitevarietieshaveacceptablelevelsof resis-tancetodownymildewandwideadaptabilitybutaregenerally susceptibletoStriga.

2.2.2. Fieldexperimentprocedures

Theexperiment wasconducted attwo sites: inNiger under artiﬁcialStriga infestationatICRISAT-Sadoré(13◦15 N,2◦18 E) andinMaliundernaturallyhighlyStriga-infestedﬁeldconditions nearIER-Cinzana(13◦17N,5◦56W).Theseenvironmentsare con-trastingintermsofecology(SahelianvsSudanian),andsoiltype. Randomizedincompleteblockdesignswithfourreplicationswere usedineachtestenvironment.Eachfull-sibfamilywasgrownina single-rowplotof3mlengthwithinter-rowspacingof1.60mof distancetoprovidemorelightforStrigaemergence,enoughspace forStrigacountingandtoreducebordereffects(Haussmannetal., 2000).Withinplot-rows,distancebetweenhillswas0.50m, pro-vidingsevenhillsperrow.Initiallyabout15seedsweresownper hill.Thehillswerethenthinnedtotwoplantsperhillthreeweeks afteremergence.Fertilizationofthetrialswasperformedduring soilpreparationwhen50kgha−1of15–15–15NPKfertilizerwas

applied.WeedsotherthanStrigawereeliminatedbyhandona regularbasistokeepthefieldcleanfor Strigaobservations.The StrigaseedsusedforartificialinfestationatICRISAT-Sadoréwere collectedfrompearlmilletfieldtrialsthereandfarmers’fieldsnear Sadoré attheend ofpreviousseasons and hadbeenstored for atleastoneyeartobreakseeddormancy.TheStrigaseedswere mixedwithfinelysievedsandthatservedascarrierandensured uniforminfestation(Kim,1991).About50,000viableStrigaseeds wereappliedperm2_to_ensure_a_{sufficiently}_high_Striga_pressure (Haussmannetal.,2000).

2.2.3. Measuredvariables

Datawerecollectedatbothsitesonaplotbasisfortraitsrelated topearlmilletphenologicaland agronomicalcharacters,downy mildew(DM)incidence,andStrigaresistance.Thefloweringtime (FLO)indaysaftersowing(das)wasrecordedwhen50% ofthe plantshadexertedstigmas.Duringharvest,thenumberof produc-tivepanicleswascountedandweighedtodeterminethepanicle yield (HYD) (gm−2).The number of hills per plotwith downy mildewwasrecordedattillering,verified/adjusted after flower-ingandusedforcalculatingthedownymildew(DM)incidence(in percentage):numberofinfestedhills,dividedbythetotalnumber ofhillsafteremergence,multipliedby100.Inaddition,thenumber ofemergedStrigaplantsforeachplotwasrecordedat67,101and 121das.SuccessiveStrigacountswerethenusedtocalculatethe “AreaunderStrigaNumberProgressCurve”(ASNPC)(Haussmann etal.,2000): ASNPC= n−1

i=0

Yi+Y(i+1) 2

(t(i+1)−ti)

wherenisthenumberofStrigaassessmentdates,YitheStrigacount attheithassessmentdate,tithedaysaftersowingattheith assess-mentdate,t0is0,andY0is0.LowASNPCmeansvaluesindicate resistance,andhighvaluessusceptibilitytoStriga.

2.2.4. Statisticalanalysis

Analysisofvariance(ANOVA)wasperformedforallmeasured traitsfollowingtheincompleteblockdesign.Datafromthetwo locations(SadoréandCinzana)wereanalyzedseparatelyandthen theadjustedmeansofeachenvironmentwereusedforcombined analysis.ThestatisticalsoftwarePLABSTATv3A(Utz,2005)was usedfordataanalysis.Thediscretedownymildewdatawere sub-jectedtoalog-transformationtomeetANOVAassumptions.The binomialdistributionwiththeLogitfunctionwasusedfor analyz-ingthetransformeddownymildewdata.Thegeneralizedlinear model(GLM)commandimplementedintherestrictedmaximum likelihood (REML) package from theGenStat 14thEd. software (Payneetal.,2011)wereusedfortheanalysis.Back-transformed meanswereestimated.Allsourcesof variation,entries, replica-tionsandincompleteblockswithinreplications,wereconsideredas randomeffects.Forfurtheranalysis,theC5full-sibfamilies (C5-Full-sib),genepoolparent(Parent),agronomiccontrols(Control)and experimentalvarieties(EVs)wereseparatedintodifferentgroups. t-testswereperformedforpairedgroupmeanscomparisonsby dividingthemeansdifferencewiththeirstandarderrorof differ-ence.Theselectiondifferential(S)isthedifferencebetweenthe meanofselectedunitsandtheoverallmeanofthebase popula-tion.Estimatesofbroad-senseheritability(h2₎_on_a_plot_basis_in areplicated trialwerecalculatedforalltraitsfollowing the for-muladescribed byHaussmannetal.(2000).Predictedgain was calculatedbymultiplyingheritabilitybytheselectiondifferential

(5)

(Falconer,1981).Thepercentagegainpercycle(G%)wascalculated followingtheformula:

G%=

( ¯Xn− ¯X0)/ ¯X0

×100

n ,

whereG%istheaveragepercentagegainpercycle, ¯Xnisthemean atlatestselectioncycle, ¯X0isthegenepoolparentsmean,andnis thenumberofselectioncycles.

3. Results

3.1. Assessmentofthegenepooldevelopment:S1vsFSselection Due to the genetic load of the pearl millet landraces used (which were never self-pollinated before), the comparatively poorerseedlingestablishmentoftheS1 materialsturnedoutto bea disadvantageintheS1 selection process,leadingtohigher experimentalerrors(lowerheritabilities)intheS1intheC0,C2and C4trials,especiallyforASNPCandDM(Table1).IntheC0 evalua-tion,S1progeniesfloweredthreeweekslaterthantheFS.Thiswas duetopoorerS1 plantestablishmentafteraseveredroughtthat occurredafterplanting.Relativelyhighpanicleyield(484gm−2) wasrecordedinC3intheFS,whichwasduetothefavorable2009 rainyseasonconditions.SincebothS1andFSprogenieswere eval-uatedinseparatefieldtrials,andsinceboth S1 andFSselection schemeswerecombined,arealcomparisonofmeansandselection efficienciesisnotpossibleinthisstudy.

3.2. Entrymeans,geneticdifferentiationandcorrelationsamong traits

Meanperformance of the200C5-FSfor panicle yield (HYD) amountedto151gm−2atSadoré,whichwashigherthanthemean panicleyieldrecordedatCinzana(130gm−2)andacrossthetwo sites(141gm−2)(Table2).MeanvaluesfortheareaunderStriga numberprogresscurve(ASNPC)wereloweratCinzana(246.9)than atSadoré(1068.1)(Table2).Highlysigniﬁcantdifferences(P<0.01) weredetectedamongthegenotypesforﬂoweringtime(FLO), pan-icleyield(HYD),ASNPCanddownymildewincidence(DM)from thesingle-siteanalysesofvariance(Table2).Excellentplot-based heritabilitiesof84and76%wereestimatedforASNPCatSadoré andCinzana,respectively.

Resultsfromthecombinedanalysesofvarianceshowedhighly signiﬁcantdifferences (P<0.01) among genotypes (G), environ-ments(E)(datanotshown)andgenotype-by-environment(G×E) interactionsforallthemeasuredtraits.G×Einteractionvariances (VG_×E)werelargerthangeneticvariances(VG)fortheStrigaand downymildewresistance-relatedtraits. Incontrast,thegenetic variancecomponentofpearlmilletﬂoweringtimewashigherand theoneforpanicleyieldsimilartotherespectiveG×Evariance component(Table2).

Forthepresenttrial,estimatesofbroad-senseheritability(h2₎ fromthecombinedanalysiswerehighforﬂoweringtime(h2₌_71%) andpanicleyield(h2₌_50%)._Moderate_heritability_values_(34%_and 38%)wereobservedforStrigaanddownymildewresistancetraits (Table2),duetothelargerG×Evariancecomponentsforthese traits.

Highpanicleyieldwassignificantlybutonlyweaklyassociated withresistancetoStriga(coefficientofcorrelationr=_−0.38**and −0.15*)anddownymildew(r=−0.34**and−0.22**)forSadoréand Cinzana,respectively.Combinedacrossthetwosites,highpanicle yieldwasalsoweaklyandsignificantlyassociatedwithlater flow-ering.NeitherStrigaresistancenordownymildewresistancewere relatedtofloweringtime.

3.3. Testofselectionprogress

UnderSadoréconditions,themeanofthefull-sibfamilies(FS) createdfromtheC5selectioncyclewassignificantlylower(P<0.01) thanthemeanofthegenepoolparents(Parent)forbothStriga resis-tance(ASNPC:−51%)anddownymildewincidence(DM:−46%) (Table3).Inotherwords,theadvancedC5-FSpopulationpresents significantlybetterresistancetoStrigaanddownymildewthanthe genepoolparentalvarietiesfromwhichitwasderived.Panicleyield wassignificantlyhigher(+49%)intheFSprogeniescomparedto thegenepoolparentalpopulation.Similarly,significantselection gainswereobservedatCinzanaforpanicleyield(+31%)anddowny mildewincidence(−62%),butnotforStrigaresistance,wherethe C5-FSmeanforareaunderStriganumberprogresscurvewasonly 1%lowerthantheParentgroup,andtheobserveddifferencewas statisticallynon-significant.Atbothsites,nosignificantdifference wasdetectedforfloweringtime betweenFSandParent groups (Table3),indicatingthattheselectionfortheothertraitsdidnot resultinfloweringtimechangesinthegenepool.

TheaveragerealizedpercentagegainfromselectionforStriga resistance(ASNPC)was−10%cycle−1atSadoréand−0.3%cycle−1 atCinzana,downymildewsusceptibilitydecreasedby9%cycle−1 atSadoréand21%cycle−1atCinzana.Theselectiongainforpanicle yieldwasequalatbothSadoréandCinzana,where10%cycle−1was quantiﬁed(Table3).

Under Sadoré conditions, most experimental varieties (EVs) derivedfromthegenepoolin2008(3rdselectioncycle)or2009 (4thselectioncycle)presentbetterresistancetoStrigaanddowny mildewthanthegenepoolparentalpopulation,exceptEV09-Cin (forASNPC),EV08-Sad-SPandEV09-Comb(forDM)(Table3).All EVswerehigheryieldingcomparedtothegenepoolparents,except forEV08-Sad-SP.AtCinzana,thederivedvarietiesEV08-Sad-SP, EV09-CinandEV09-CombshowedlowermeanforASNPC,while allEVsweremoreresistanttodownymildewandhigheryielding comparedtotheParentgroup,exceptforEV09-Cin(forHYD).At bothsites,onlynon-signiﬁcantchangeswereobservedbetween EVsandgenepoolparentalmeansforﬂoweringtime.Underboth sites,theControlgroupsupportedthehighestStrigadamageand

Table2

SinglesiteandcombinedanalysisofvarianceresultsfromtheC5-Full-sibpopulationevaluatedatSadoréandCinzanaduringthe2011rainyseason.

Trait Sadoré Cinzana Combinedacrosstwosites

Mean VG h2(%) Mean VG h2(%) Mean VG VG×E h2(%)

ASNPC 1068.1 231.7** _83.5 _246.9 _11.1** _76.2 _656.4 _27.5** _100.4** _34.1

HYD(gm−2) 151.2 971.6** _36.1 _130.3 _310.3** _21.5 _140.7 _334.4** _317.5** _49.8

DM(%) 20.1 213.4** _42.9 _6.7 _90.8** _53.7 _13.4 _51.1** _119.1** _38.3

FLO(das) 71.1 9.5** _48.6 _51.6 _6.0** _38.6 _61.3 _5.9** _2.5** _70.6

TheanalysisofvarianceisperformedonareaunderStriganumberprogresscurve(ASNPC),panicleyield(HYD,gm−2_),_downy_mildew_(DM)_incidence_(%)_and_time_to_50%

ﬂowering(FLO)indaysaftersowing(das).

TheanalysiswasperformedusingPLABSTAT.Foralltraits,genetic(VG)andG×Einteraction(VG×E)variancecomponentsarehighlysigniﬁcant.Basicdatadescription(trial meansandheritability)aregivenforeachselectedtrait.

(6)

Table3

Realizedresponsetoselection:estimatedmeansofgenepoolparents,C5-Full-sib(FS)population,newexperimentalvarieties(EVs)extractedfromthegenepoolandthe

controlgroup,evaluatedduringthe2011rainyseasonatSadoréandCinzana.

Trait† Location Meanperformance‡

Genepool parents

C5-Full-sibpopulation Realized

gain(%)

Gainper cycle(G%)

EV08-Sad-LP EV08-Sad-SP EV09-Sad EV09-Cin EV09-Comb Control

ASNPC Sadoré 1969.4 965.2** −51.0 −10.2 611.3 687.3 262.1 1019.0 601.3 3447.3 Cinzana 224.5 222.5ns _−0.9 _−0.3 _301.6 _202.6 _242.6 _86.2 _186.7 _993.1 HYD(gm−2) Sadoré 105.0 156.0** 48.6 9.7 152.4 105.5 125.9 131.5 162.2 80.1 Cinzana 101.8 133.3** 30.9 10.3 118.2 108.2 122.7 93.8 111.6 89.8 DM(%) Sadoré 36.2 19.5** −46.1 −9.2 8.4 33.5 19.6 15.5 30.8 21.1 Cinzana 15.6 5.9** −62.2 −20.7 0.0 7.8 3.7 3.2 7.5 19.0

†_ASNPC₌_area_under_Striga_number_progress_curve;_HYD₌_panicle_yield_(g_m-2_);_DM₌_downy_mildew_incidence_(%);_EV₌_Striga_resistant_experimental_variety;_08/09₌_year

ofextractionfromthegenepool;Sad/Cin/Comb=selectionbasedonperformanceatSadoré,Cinzanaandcombinedacrossthetwolocations.

‡_**_andns_:_Difference_to_genepool_parents_signiﬁcant_by_t-test_at_1%_and_{non-signiﬁcance}_for_the_C

5-Full-sibpopulationvsgenepoolparentsmeanscomparison. lowestpanicleyield,whileitsperformancewasbetteratSadoré

fordownymildewresistancecomparedtothegenepoolparental means(Table3).

3.4. Predictedresponsefromfutureselection

Considering the200C5-FS asthebasepopulationfor future selection,thesigniﬁcantgeneticvariationandmoderatetohigh heritabilitiesforthetargettraitsatbothsitesindividuallyandin thecombinedanalysisshouldenablefurtherpositiveresponseto selection.

Predictedgainscycle−1 forStrigaresistance(recordedonthe areaunderStriganumber progress curve)were −572,−81and −122forSadoré,Cinzanaandacrossthetwosites,respectively, andrangedfrom−3%to−6%fordownymildewincidence(Table4). Forbothtraits,theexpectedgainscycle−1atSadoréwere remark-ablyhighercomparedtothetwootherconditions.Predictedgains cycle−1forpanicleyieldweresimilarbetweenSadoréandacross thetwosites(12gm−2cycle−1).Comparativelysmallerbut posi-tivegeneticgainsforresistancetoStrigaanddownymildewandfor increasesinpanicleyieldmaythereforebeexpectedunderCinzana conditions(Table4).

Resultsfromthecombinedanalysesofvariancerevealedahigh amountofG×Einteractionvariance especiallyforresistanceto Strigaanddownymildew,butalsoforpanicleyieldandlower her-itabilitiesforASNPCandDMcomparedtothesingle-siteanalysis (Table2).Thesesuggestthatselectionforspeciﬁcadaptationto eachofthetwotestsitesmaybemorepromisingthanselection forwideadaptationacrossthesetwosites.Thismeansthat dif-ferentFSfamiliesneedtobeselectedandrecombinedforthetwo

contrastingsitesSadoréandCinzanatomaximizegainsfrom selec-tionandtoderivethebestvarietiesforeachlocation.

4. Discussion

4.1. Recurrentselection:aneffectivetoolforenhancingStriga resistanceinpearlmillet

Recurrentselectioninbroad-basedpopulationscanbeusedto enhancequantitativeresistancetoatargetpestordisease.Such quantitativeresistance,whichisduetomanygenelociwithsmaller effects,isusuallymoredurablethanmonogenic,qualitative resis-tance(Berneretal.,1995;MenkirandKling,2007).Accordingtoour knowledge,thisisthefirstreportdescribingsignificantprogress fromrecurrentselectionforquantitativeStrigaresistanceinpearl millet.Simultaneously,Strigaresistance,downymildewresistance andpanicleyieldwereimprovedwhilemaintainingfloweringtime whichdeterminesgeneraladaptationtothetargetsites.The rig-orousevaluationunderfieldconditionsusingfourreplicatesand ASNPC as combined measure for Striga resistance (resultingin lowerexperimentalerrorandhigherbroad-senseheritabilities), and thelargenumber of progenyevaluated in each generation seemtobeeffectiveforimprovingthegenepool.Self-pollinated progenyevaluationisexpectedtoleadtobettergenetic differen-tiation,buttheS1 evaluation partiallysufferedfrompoorplant establishment(seedlinginbreedingdepression),resultinginlower heritabilitiesinsomeseasons.TheFSselection,ontheotherhand, haslessprobabilitytoidentifyandmoveforwarddesirable reces-sivealleles compared toS1 selection, but hastheadvantageof thetestentriesbeingintheirnatural,highlyheterozygousstate.

Table4

Predictedfuturegainfromselection.

Traits† _Location _C₅_-Full-sib

population mean

SelectedFS familiesmean

Selectiondifferential Heritability Predictedgain

ASNPC Sadoré 965.2 284.1 −681.1 0.84 −572.1 Cinzana 222.5 115.4 −107.1 0.76 −81.4 Across-sites 593.9 235.1 −358.8 0.34 −122.0 HYD(gm−2) Sadoré 156.0 188.2 32.2 0.36 11.6 Cinzana 133.3 148.5 15.2 0.22 3.3 Across-sites 144.7 168.1 23.4 0.50 11.7 DM(%) Sadoré 19.5 5.2 −14.3 0.43 −6.1 Cinzana 5.9 1.3 −4.6 0.54 −2.5 Across-sites 12.7 4.4 −8.3 0.38 −3.2

FLO(das) Sadoré 71.3 71.0 −0.3 0.49 −0.1

Cinzana 51.6 51.0 −0.6 0.39 −0.2

Across-sites 61.4 61.2 −0.2 0.71 −0.1

(7)

Sincebothselectionschemeswerecombinedwhichsavedtime(no additionalcycleneededforrecombinationofS1s),amoredetailed comparisonofselectionefficienciesinS1vsFSselectionisnot pos-siblein thisstudy.Based onthepresent findings,therecurrent selectionobjective–improvingtheStriga-resistantgenepoolmean whilemaintainingthegeneticvariability–islargelyattainedby thefifthselectioncycle,especiallyatSadoré.Differentrecurrent selectionmethodshavebeenreportedbyRattundeandWitcombe (1993)toimprovesubstantiallygrainyieldandresistancetodowny mildewinfivepearlmilletcomposites.Theseresultsareinlinewith ourfindingswhichequallyrevealedlargegainsoverfivecyclesof selection.Also,recurrentselectionhasbeenusedsuccessfullyin pearlmillettoimprovedownymildewresistance,grainyieldand otheragronomictraits(e.g.,WeltzienandKing,1995;Goviletal., 1982;BidingerandRaju,2000;Baskaranetal.,2009).InNigeria, recurrentselectionofmaizeforgrainyieldandStrigaresistance hadbeenreportedtobesuccessful(Badu-Apraku,2010;Menkir andKling,2007).

Thelowermeasuredresponsefromselectionfor Striga resis-tanceatCinzanamaybeattributedtothefollowingfactors: -predominanceofselectionatSadoré,i.e.,lowernumberandlater

startofthegenepoolevaluationsatCinzana; -selectionofthegenepoolparentsmainlyfromNiger;

-signiﬁcantG×EinteractionforASNPC acrossSadoré and Cin-zana,combinedwithpredominanceofselectionatSadorémight haveresultedinspeciﬁcadaptationofthegenepooltotheStriga populationatSadoré;

-fluctuations inselection pressure, resulting fromyear-to-year variation in the level of parasiteinfestation, especially under theconditionsof naturalStriga infestationaswasthecase at Cinzana(artificial Strigainfestationmight thereforebe gener-allypreferable,butrequiresstrictisolationtopreventanyStriga seedtomigrate;butsuchisolationwasnotpossibleatCinzana; obviously,artificialinfestationcannotberecommendedif farm-ers’fieldsareusedduringevaluation–herecarefulfieldselection inthepreviousyeariskeytosuccess);

-thelowStrigainfestationlevelatCinzanaduringtheevaluation ofselectionprogress,whichmighthavebeensub-optimalfor dif-ferentiationamongthetestedentries–butthisislessprobable sincetheestimatedplot-basedheritabilitywassufﬁcientlyhigh atCinzanaduringthetestyear.

4.2. FutureusesoftheStriga-resistantpearlmilletgenepool Inthepresentstudy,markedpredictedresponseswere esti-matedfor alltarget traits,suggesting that enhanced Striga and downymildewresistance,aswellasincreasedpanicleyieldare expectedalsointhenextcycleofselection.Thefactthatuseful geneticvariationstillexistsinthepopulationmaybeduetothe introgressionoffurtherresistantgermplasminoneselectioncycle andmaintenance of alargeenougheffective populationsize in thegenepoolineach selectioncycle.Thebroad-basedgenepool can therefore serve as base population for furtherselection in collaborativebreedingprogramsandforextractionofnew exper-imentalvarietieswithspeciﬁcadaptationtodifferenttargetsites. Experimentalvarietieshave beenand are beingextractedfrom thegenepool for Sadoré and Cinzana, and are being also vali-dated inon-farm trials.The genepoolmayalsobesharedwith otherpearlmilletbreedersinWestAfrica,especiallyBurkinaFaso, Nigeria,andSenegal,sothattheresistanceisvalidatedandnew experimentalvarietiesareextractedfornewtargetsitesinthose countries.Suchcultivarswithquantitativeresistancemayplayan importantroleinintegratedStrigacontrolinpearlmilletinWest Africa.

4.3. Needforselectionforspeciﬁcadaptation

Thepresence of significantG×E interactionfor Striga resis-tance and other traits in our study may be attributed to the differencesinenvironmentalfactorssuchassoiltype,amountof rainfall(SahelianvsSudanian),andputativedifferencesfor viru-lencebetweentheparasitepopulationsatSadoréandCinzana.The totalobservedG×Einteractionconsistsofinteractionsofthepearl milletgenepoolwithenvironmentalfactorsofthetestlocations, interactions betweenpearl millet genepool and S. hermonthica populations at the two test sites and the threefoldinteraction betweenpearlmilletgenepool,S.hermonthicapopulationand envi-ronmentalfactors,butthesedifferenttypesofinteractioneffects cannotbeseparatedinfieldtrialsduetoquarantineregulations. Onealternativewouldbetomovebeyondthetwosites(Cinzana andSadoré)andrathertestatnumerousenvironmentsinseveral countriesinordertoperformanAMMIanalysis,apowerfultool forunderstandingcomplexG×Einteractions.However,thefact thattheG×Einteractionvarianceswereconsiderablyhigherthan thegeneticvariancesforresistancetoStrigaanddownymildew suggeststhatbestFSfamiliesdifferbetweenthetwoevaluation environments,andthatthebestFSidentifiedateachsiteshouldbe recombinedseparatelytocreateexperimentalvarietieswith spe-cificadaptationtoeachsiteanditsbroaderenvironmentalcontext. AlsoBadu-Apraku (2010) foundsignificant G×E interaction forgrainyieldandStrigaresistanceparametersinanextra-early maturingmaizepopulationinthreetestlocationsinNigeria,and Haussmannetal.(2001b)reportedhighlysignificantG×E inter-actionforgrainyieldandareaunderStrigaseverityprogresscurve (ASVPC,ameasureofStrigaemergenceandvigorthroughoutthe season)inAfricansorghumtrials.Bothstudiesthereforealso sup-porttheneedtoselectforspecificadaptationinStrigaresistance breeding,especiallyifthetargetsitesdiffersubstantiallyfromeach otherandrepresentputativedifferentStrigaecotypes.

4.4. FutureprospectsforStrigaresistancebreedingand managementinpearlmillet

Intra-speciesgeneticdiversityhasbeenreportedinS. hermon-thica(Yoshidaetal.,2010), indicatingthattheparasiteishighly variable,andthereforerespondstoselectionpressurerapidly.The complexnatureofthepathosystemandhighadaptabilityofStriga toachanginghostpopulationlikelyrequiresthatStriga-resistant varietiesbemanagedand combinedwithintegrated Striga con-trolmeasures,tohindertheparasitetoovercometheresistance. Genetically heterogeneous open-pollinated varieties where dif-ferentplantscarrydifferentresistanceallelesmaycontributeto stabilityofresistanceovertime,becausetheyresultinnon-uniform selectionpressureontheparasite.Maintenancebreedingofsuch Striga-resistant open-pollinated varieties may need to be done underStrigainfestedconditions,inordertomaintainandimprove theirresistance.Thisisespeciallytrueincasethereareany“costsof resistance”thatleadtolowerperformanceofresistantplantsinthe absenceofStriga.Toourknowledge,anycostsofStrigaresistance –ifexistentatall–arenotyetwellunderstood.

Severalmechanismshave beenimplicatedinloweringStriga parasitism,includinglowgerminationstimulantproduction(Hess etal.,1992;EjetaandButler,1993;Haussmannetal.,2001a;Jamil etal.,2011),reducedcapacitytoelicithaustorialinductionofStriga (Gurneyetal.,2003;Richetal.,2004),mechanicalbarriers(Maiti et al.,1984; Gurney et al.,2006; Amusan etal., 2008; Yoshida andShirasu,2009),andhypersensitiveresponse(Mohamedetal., 2003;Cissokoetal.,2011).Ejetaetal.(2000)hadalsosummarized otherpotentialpost-germinationmechanismsofresistancethat impedeattachmentandemergenceofStrigaincrops.Whetherthe observedgainsinStrigaresistancewithselectioninthepearlmillet

(8)

Striga-resistantgenepoolareduetotheseorothermechanismsis yettobedetermined.AbetterunderstandingofStrigaresistance mechanismsinpearlmilletanddevelopmentofappropriate, cost-efﬁcientscreeningproceduresforcomponenttraitscouldrender futurerecurrentselectionmoreefﬁcient.

Thedevelopmentofamarker-assistedpopulationimprovement schemeforenhancingquantitativeStrigaresistanceinpearlmillet couldhelprenderselectionevenmoreeffective.Duetodecreasing genotypingcosts(Elshireetal.,2011),largernumbersofentries couldbe screened for markers linked to resistancealleles, fol-lowed byﬁeld phenotyping of a selectedsubset ofthe entries, andthiswouldresultinanincreasedeffectiveselectionintensity comparedtophenotypicselectionalone.Theso-called“non-Striga years”(yearswithoveralllimitedStrigainfestation)thatoftenlead topoordifferentiationamongthetestedentrieswouldnolonger hinderselectionprogressasmuchastheydoatpresent,because selection wouldbe doneusingthemarkers. In years whenthe ﬁeldphenotypingissuccessfulindifferentiatingthetestedentries, theresultscanbeusedtore-calibratethemarker-basedselection index.Effortsarepresentlyunderwayinourgrouptodevelopsuch amarker-assistedFSpopulationimprovementschemeforStriga resistanceinpearlmillet.

5. Conclusions

Fivecyclesof phenotypicrecurrentselectionin a diversified pearl millet genepool resulted in significant improvement for Strigaanddownymildewresistance,andincreasedpanicleyield. Significantgeneticvarianceandsufficientlylargebroad-sense her-itabilitiesintheC5populationindicategoodprospectsforfurther selection within and extraction of new Striga-resistant experi-mental varieties out of this genepool. Large G×E interactions observedinthepresentStrigaresistancetrialsunderlinetheneed formulti-locationtestingtodevelopnewvarietaloptionswith spe-cificadaptationtocontrastingtargetenvironments.Becauseofthe Strigavariability,resistantvarietiesneedtobecarefullymanaged andcombinedwithotherintegratedStrigacontrolmeasures,in ordertohindertheparasitetoovercomethedeployedresistance.

Acknowledgements

ThisworkhasbeenundertakenaspartoftheCGIARResearch Program onDrylandCereals. The financial support of the McK-nightFoundationisgratefullyacknowledged(ICRISATgrantsno. 06-14and10-134).ThisresearchwasconductedatICRISAT-Sadoré andIER-Cinzana;theauthorsexpresstheirappreciationtoBoubé Guida,HamadouBoubé,AdamaTraoréandallotherfieldassistants atSadoréandCinzanafortheirgreathelpwiththeexperiments. OurgratefulthanksareextendedtoDr.RogerSternandZobinou Mawusi(ReadingUniversity,UK)fortheirvaluablestatistical assis-tance,andtotheresourcepersonsofthee-writeshoporganizedby theCollaborativeCropResearchProgramoftheMcKnight Founda-tion.Finally,thispublicationwasmadepossiblethroughsupport providedbytheIRD-DSFtothefirstauthor.

References

Amusan,I.O.,Rich,P.J.,Menkir,A.,Housley,T.,Ejeta,G.,2008.ResistancetoStriga her-monthicainamaizeinbredlinederivedfromZeadiploperennis.NewPhytologist 178,157–166.

Badu-Apraku,B.,2010.EffectsofrecurrentselectionforgrainyieldandStriga resis-tanceinanextra-earlymaizepopulation.CropSci.50,1735–1743.

Baskaran,K.,Hash,C.T.,Senthilvel,S.,BhaskerRaj,A.G.,Muthiah,A.R.,2009.SSRallele frequencychangesinresponsetorecurrentselectionforpearlmilletgrainyield andotheragronomictraits.J.SATAgric.Res.7.

Berner,D.K.,Kling,J.G.,Singh,B.B.,1995.Strigaresearchandcontrol:aperspective fromAfrica.PlantDis.79,652–660.

Bezanc¸on,G.,Pham,J.-L.,Deu,M.,Vigouroux,Y.,Sagnard,F.,Mariac,C.,Kapran, I.,Mamadou,A.,Gérard,B.,Ndjeunga,J.,Chantereau,J.,2009.Changesinthe diversityandgeographicdistributionofcultivatedmillet(Pennisetumglaucum [L.]R.Br.)andsorghum(Sorghumbicolor(L.)Moench)varietiesinNigerbetween 1976and2003.PlantGenet.Resour.CropEvol.56,223–236.

Bidinger,F.R.,Raju,D.S.,2000.Responsetoselectionforincreasedindividualgrain massinpearlmillet.CropSci.40,68–71.

Chemisquy,M.A.,Giussani,L.M.,Scataglini,M.A.,Kellogg,E.A.,Morrone,O.,2010.

PhylogeneticstudiesfavourtheuniﬁcationofPennisetum,Cenchrusand Odonte-lytrum(Poaceae):acombinednuclear,plastidandmorphologicalanalysis,and nomenclaturalcombinationsinCenchrus.Ann.Bot.(Oxf.)106,107–130.

Chisi,M.,Esele,P.,1997.DiscussionforSessionVII-Breedingforresistancetoother abioticstressesandStriga.In:ProceedingInternationalConference:Genetic ImprovementofSorghumandPearlMillet.INTSORMIL,Lincoln,NE,pp.525–528.

Cissoko,M.,Boisnard,A.,Rodenburg,J.,Press,M.C.,Scholes,J.D.,2011.NewRice forAfrica(NERICA)cultivarsexhibitdifferentlevelsofpost-attachment resis-tanceagainsttheparasiticweedsStrigahermonthicaandStrigaasiatica.New Phytologist192,952–963.

Ejeta,G.,2007.BreedingforStrigaresistanceinSorghum:exploitationofanintricate host–parasitebiology.CropSci.47,S-216–S-217.

Ejeta,G.,Mohammed,A.,Rich,P.,Melake-Berhan,A.,Housley,T.L.,Hess,D.E.,2000.

SelectionforspeciﬁcmechanismsofresistancetoStrigainsorghum.In: Hauss-mann,B.I.G.,etal.(Eds.),BreedingforStrigaresistanceincereals,Proceedingsof aWorkshop.IITA,Ibadan,Nigeria,18–20August1999,Margraf,Weikersheim, Germany,pp.29–37.

Ejeta,G.,Butler,L.G.,1993.Host–parasiteinteractionsthroughouttheStrigalife cycle,andtheircontributionstoStrigaresistance.Afr.CropSci.J.1,75–80.

Elshire, R.J., Glaubitz, J.C., Sun, Q., Poland, J.A., Kawamoto, K., Buckler, E.S., Mitchell, S.E., 2011. A robust, simple genotyping-by-sequencing (GBS) approach for high diversity species. PLoS ONE 6, e19379,

http://dx.doi.org/10.1371/journal.pone.0019379.

Ezeaku,I.E.,Gupta,S.C.,2004.Developmentofsorghumpopulationsforresistanceto StrigahermonthicaintheNigerianSudanSavanna.Afr.J.Biotechnol.3,324–329, Availableonlineathttp://www.academicjournals.org/AJB

Falconer,D.S.,1981.IntroductiontoQuantitativeGenetics.Longman,NewYork,USA, 340pp.

Gethi,J.G.,Smith,M.E.,2004.GeneticresponsesofsinglecrossesofmaizetoStriga hermonthica(Del.)BenthandStrigaasiatica(L.)Kuntze.CropSci.44,2068–2077.

Govil,J.N.,Pokhriyal,S.C.,Murty,B.R.,1982.Full-sibandreciprocalrecurrent relec-tioninrelationtopearlmilletimprovement.Theor.App.Genet.62,25–30.

Gurney,A.L.,Slate,J.,Press,M.C.,Scholes,J.D.,2006.Anovelformofresistanceinrice totheangiospermparasiteStrigahermonthica.NewPhytologist169,199–208.

Gurney,A.L.,Grimanelli,D.,Kanampiu,F.,Hoisington,D.,Scholes,J.D.,Press,M.C., 2003.NovelsourcesofresistancetoStrigahermonthicainTripsacumdactyloides, awildrelativeofmaize.NewPhytologist160,557–568.

Hash, C.T., Yadav, R.S., Cavan, G.P., Howarth, C.J., Liu, H., Qi, X., Sharma, A., Kolesnikova-Allen, M.A., Bidinger, F.R., Witcombe, J.R., 2000. Marker-assistedbackcrossingtoimproveterminaldroughttoleranceinpearlmillet. In:Ribaut, J.-M., Poland, D. (Eds.), MolecularApproaches for theGenetic Improvement ofCereals for Stable Production in Water-Limited Environ-ments,AStrategicPlanningWorkshopheldatCIMMYT.ElBatan, Mexico, 21–25June1999.InternationalMaizeandWheatImprovementCenter (CIM-MYT),Mexico,D.F.,Mexico,pp.114–119http://www.cimmyt.org/english/docs/ proceedings/molecApproaches/pdfs/markerAssisted.pdf

Haussmann,B.I.G.,Hess,D.E.,Welz,H.G.,Geiger,H.H.,2000.Improved methodolo-giesforbreedingStriga-resistantsorghums.FieldCropsRes.66,195–211.

Haussmann,B.I.G.,Hess,D.E.,Omanya,G.O.,Reddy,B.V.S.,Welz,H.G.,Geiger,H.H., 2001a.MajorandminorgenesforstimulationofStrigahermonthicaseed ger-minationinsorghum,interactionwithdifferentStrigapopulations.CropSci.41, 1507–1512.

Haussmann,B.I.G.,Hess,D.E.,Reddy,B.V.S.,Mukuru,S.Z.,Kayentao,M.,Welz,H.G., Geiger,H.H.,2001b.Patternanalysisofgenotype×environmentinteraction forStrigaresistanceandgrainyieldinAfricansorghumtrials.Euphytica122, 297–308.

Haussmann,B.I.G.,Hess,D.E.,Omanya,G.O.,Folkertsma,R.T.,Reddy,B.V.S.,Kayentao, M.,Welz,H.G.,Geiger,H.H.,2004.Genomicregionsinﬂuencingresistancetothe parasiticweedStrigahermonthicaintworecombinantinbredpopulationsof sorghum.Theor.App.Genet.109,1005–1016.

Hess,D.E.,Ejeta,G.,1992.InheritanceofresistancetoStrigainsorghumgenotype SRN39.PlantBreeding109,233–241.

Hess,D.E.,Ejeta,G.,Butler,L.G.,1992.Selectingsorghumgenotypesexpressinga quantitativebiosynthetictraitthatconfersresistancetoStriga.Phytochemistry 31,493–497.

Jamil,M.,Rodenburg,J.,Charnikhova,T.,Bouwmeester,H.J.,2011.Pre-attachment StrigahermonthicaresistanceofNewRiceforAfrica(NERICA)cultivarsbasedon lowstrigolactoneproduction.NewPhytologist192,964–975.

Kim, S.-K., 1991. Breeding for tolerance and general resistance in maize: a novel approachto combatingStrigain Africa. In:Improving Striga Man-agement in Africa. Proceedings of the Second GeneralWorkshop of the Pan-AfricanStrigaControlNetwork(PASCON),Nairobi,Kenya,23–29June1991, pp.168–176.

Kim,S.-K.,1994.GeneticsofmaizetoleranceofStrigahermonthica.CropSci.34, 900–907.

Maiti,R.K.,Ramaiah,K.V.,Bisen,S.S.,Chidley,V.L.,1984.Acomparativestudyofthe haustorialdevelopmentofStrigaasiatica(L.)Kuntzeonsorghumcultivars.Ann. Bot.(Lond.)54,447–457.

(9)

Menkir,A., Kling,J.G.,2007. Responseto recurrentselectionforresistance to Strigahermonthica(Del.)Benthinatropicalmaizepopulation.CropSci.47, 674–684.

Mohamed,A.,Ellicott,A.,Housley,T.L.,Ejeta,G.,2003.Hypersensitiveresponseto StrigainfectioninSorghum.CropSci.43,1320–1324.

Payne,R.W.,Harding,S.A.,Murray,D.A.,Soutar,D.M.,Baird,D.B.,Glaser,A.I.,Welham, S.J.,Gilmour,A.R.,Thompson,R.,Webster,R.,2011.IntroductiontoGenStatfor Windows14thEdition.VSNInternational,5theWaterhouse,WaterhouseStreet, HemelHempstead,HertfordshireHP11ES,UK.

Ramaiah,K.V.,1987.Breedingcerealgrainsforresistancetowitchweed.In: Mus-selman,L.J.(Ed.),ParasiticWeedsinAgriculture,vol.I,Striga.CRCPress,Boca Raton,FL,pp.227–242.

Rattunde,H.F.W.,Witcombe,J.R.,1993.Recurrentselectionforincreasedgrain yieldandresistanceto downymildewinpearlmillet. PlantBreeding110, 63–72.

Rich,P.J.,Grenier,C.,Ejeta,G.,2004.Strigaresistanceinthewildrelativesofsorghum. CropSci.44,2221–2229.

Satish,K.,Gutema,Z.,Grenier,C.,Rich,P.J.,Ejeta,G.,2011.Taggingandvalidation ofmicrosatellitemarkerslinkedtothelowgerminationstimulantgene(lgs)for Strigaresistanceinsorghum[Sorghumbicolor(L.)Moench].Theor.App.Genet,

http://dx.doi.org/10.1007/s00122-011-1763-9.

Senthilvel,S.,Jayashree,B.,Mahalakshmi,V.,Sathish,K.P.,Nakka,S.,Nepolean,T., Hash,C.T.,2008.Developmentandmappingofsimplesequencerepeatmarkers forpearlmilletfromdataminingofexpressedsequencetags.BMCPlantBiol.8, 119,http://dx.doi.org/10.1186/1471-2229-8-119.

Singh,S.D.,1990.Sourcesofresistancetodownymildewandrustinpearlmillet. PlantDis.74,871–874.

Thakur,R.P.,Rao,V.P.,Sharma,R.,2011.Inﬂuenceofdosage,storagetimeand tem-peratureonefﬁcacyofmetalaxyl-treatedseedforthecontrolofpearlmillet downymildew.Eur.J.PlantPathol.129,353–359.

Thakur, R.P., Shetty, H.S., Khairwal, I.S., 2006. Pearl millet downy mildew researchinIndia:progressandperspectives.SATeJournal2,1–6http://www. icrisat.org/journal/cropimprovement/v2i1/v2i1pearlmillet.pdf

Utz,H.F.,2005.PLABSTATVersion3A:Acomputerprogramforstatisticalanalysis ofplantbreedingexperiments.InstituteofPlantBreeding,SeedScienceand PopulationGenetics,UniversityofHohenheim,D-70593Stuttgart,Germany.

Weltzien,E.R.,King,S.B.,1995.Recurrentselectionfordowny-mildewresistancein pearlmillet.PlantBreeding114,308–312.

Wilson,J.P.,Hess,D.E.,Hanna,W.W.,2000.ResistancetoStrigahermonthicainwild accessionsoftheprimarygenepoolofPennisetumglaucum.Phytopathology90, 1169–1172.

Wilson,J.P.,Hess,D.E.,Hanna,W.W.,Kumar,K.A.,Gupta,S.C.,2004.Pennisetum glau-cumsubsp.monodiiaccessionswithStrigaresistanceinWestAfrica.CropProtect. 23,865–870.

Yoshida, S.,Ishida, J.K., Kamal, N.M., Ali, A.M., Namba, S.,Shirasu, K., 2010. A full-length enriched cDNA library and expressed sequence tag anal-ysis of the parasitic weed, Striga hermonthica. BMC Plant Biol. 10, 55,

http://dx.doi.org/10.1186/1471-2229-10-55.

Yoshida,S.,Shirasu,K.,2009.Multiplelayersofincompatibilitytotheparasitic witchweed,Strigahermonthica.NewPhytologist183,180–189.