QTLs for Tolerance of Drought and Breeding for Tolerance of Abiotic and Biotic Stress

15 Pages·2014·1.01 MB·English

QTLs for Tolerance of Drought and Breeding for
Tolerance of Abiotic and Biotic Stress: An Integrated
Approach
Shalabh Dixit1, B. Emma Huang2, Ma. Teresa Sta Cruz1, Paul T. Maturan1, Jhon Christian E. Ontoy1,
Arvind Kumar1*
1InternationalRiceResearchInstitute(IRRI),LosBan˜os,Laguna,Philippines,2ComputationalInformatics,CSIRO,DuttonPark,Queensland,Australia
Abstract
Background:Thecouplingofbioticandabioticstressesleadstohighyieldlossesinrainfedrice(OryzasativaL.)growing
areas. While several studies target these stresses independently, breeding strategies to combat multiple stresses seldom
exist.ThisstudyreportsanintegratedstrategythatcombinesQTLmappingandphenotypicselectiontodevelopricelines
with highgrainyield (GY)under droughtstressand non-stress conditions,and tolerance ofrice blast.
Methodology:A blast-tolerant BC F -derived population was developed from the cross of tropical japonica cultivar
2 3
Moroberekan(blast-anddrought-tolerant)andhigh-yieldingindicavarietySwarna(blast-anddrought-susceptible)through
phenotypic selection for blast tolerance at the BC F generation. The population was studied for segregation distortion
2 2
patternsand QTLsfor GYunderdrought were identifiedalongwith study ofepistaticinteractions forthe trait.
Results:Segregationdistortion,infavourofMoroberekan,wasobservedat50ofthe59loci.Majorityofthesemarkerloci
co-localizedwithknownQTLsforblasttoleranceorNBS-LRRdiseaseresistancegenes.Despitethepresenceofsegregation
distortion,highvariationforDTF,PHandGYwasobservedandseveralQTLswereidentifiedunderdroughtstressandnon-
stressconditionsforthethreetraits.EpistaticinteractionswerealsodetectedforGYwhichexplainedalargeproportionof
phenotypic varianceobserved in thepopulation.
Conclusions:This strategy allowed us to identify QTLs for GY along with rapid development of high-yielding purelines
toleranttoblastanddroughtwithconsiderablyreducedefforts.Apartfromthis,italsoallowedustostudytheeffectsofthe
selectioncycleforblasttolerance.ThedevelopedlineswerescreenedatIRRIandinthetargetenvironment,anddrought
andblasttolerantlineswithhighyieldwereidentified.Withtolerancetotwomajorstressesandhighyieldpotential,these
linesmayprovide yieldstability in rainfedrice areas.
Citation:DixitS,HuangBE,StaCruzMT,MaturanPT,OntoyJCE,etal.(2014)QTLsforToleranceofDroughtandBreedingforToleranceofAbioticandBiotic
Stress:AnIntegratedApproach.PLoSONE9(10):e109574.doi:10.1371/journal.pone.0109574
Editor:KeqiangWu,NationalTaiwanUniversity,Taiwan
ReceivedJune20,2014;AcceptedSeptember1,2014;PublishedOctober14,2014
Copyright:(cid:2)2014Dixitetal.Thisisanopen-accessarticledistributedunderthetermsoftheCreativeCommonsAttributionLicense,whichpermitsunrestricted
use,distribution,andreproductioninanymedium,providedtheoriginalauthorandsourcearecredited.
DataAvailability:Theauthorsconfirmthatalldataunderlyingthefindingsarefullyavailablewithoutrestriction.Allrelevantdataarewithinthepaperandits
SupportingInformationfiles.
Funding:TheworkwasfundedbyGlobalRiceSciencePartnership(GRiSP),GenerationChallengeProgram(GCP),Federalministryforeconomiccooperationand
development(BMZ),Germany,andtheBillandMelindaGatesFoundation(BMGF).Thefundershadnoroleinstudydesign,datacollectionandanalysis,decision
topublish,orpreparationofthemanuscript.
CompetingInterests:Theauthorshavedeclaredthatnocompetinginterestsexist.
*Email:[email protected]
Introduction systems and environments, the major focus of research in
increasingriceyieldpotentialhasbeenfortheirrigatedecosystem
Developing crop varieties with higher yield potential is a in the post-Green Revolution period. However, no major
challenging task in modern-day plant breeding. The Green breakthrough in breaking the yield potential barrier has been
Revolution led to a historic breakthrough in increasing the yield achieved until now.
potential of crops, such as rice and wheat, through the
The low-yielding rainfed lowland rice ecosystem has great
development of semi-dwarf varieties that has fed the increasing
potential for yield improvement. This ecosystem occupies about
populationtillnow.However,itispredictedthata60%increasein
38% of thetotal rice area, but contributes only21% tototal rice
agricultural production needsto be achieved by 2050 to feed the
production[2].InAsiaalone,34millionhaofrainfedlowlandrice
ever-growing population [1].
and 8 million ha of rainfed upland rice [3] experience drought
Unlike other cereals, rice is cultivated across a much wider
stressofvaryingintensityalmosteveryyear.Thesedroughtevents
range of ecosystems, including irrigated, rainfed upland, rainfed
areoftencoupledwithincreasedincidenceofotherbioticstresses,
lowland, flooded, and deepwater ecosystems across tropical and suchasblast,brownspot,andbacterialblight,leadingtoafurther
temperate climate conditions. Despite this diversity of cultivation
PLOSONE | www.plosone.org 1 October2014 | Volume 9 | Issue 10 | e109574
IntegratingQTLMappingandRiceBreedingforStressTolerance
A. e
S z
nIRRI- Plotsi 1.0 2.0 1.0 2.0 1.0 2.0 2.0 2.4 3.2
i
d
n
a
RI
R
I
at
n
o s
ulati ation
p c
po epli
g R 2 2 2 2 2 2 2 2 2
n
pi
p
a
m
esidentifiedfromthe Experimentaldesign 63810AL 63810AL 63810AL 63810AL 61210AL 61010AL 61010AL 6813AL 6813AL plandmoderatestress.
n u
li =
ought-tolerant Location IRRI IRRI IRRI IRRI IRRI IRRI IRRI IRRI-SA IRRI-SA dnon-stress,UMS
dr an
of owl
l
g =
n
ni ze NS
e si L
mudFstoingraida:pu1te0rpere.i1gnd31yg7r.o1Pup/ujoosogeupphdruutnllasaatlt.ttopriieooosnnndsee.daf0vone1er0vdl9eomn5lpo7oa4ppn.amg-ps0iet0nbrn1geltasssQatn-cTtdoLonslpedfhriotaeirnnotonhtsigyB.phCi2cgFrs3ac-dirneeeryniviienelddg TLsandscre Population 361 361 361 361 99 77 77 95 95 moderatestress,
Q
d
of an
d1.e0c–li2n.e5itnhraic2e1y[i4el]d..Riceyieldsintheseecosystemsremainlowat cation S=lowl
rdlaeerdvDgueeeclostppiporitnoehpiiogtnhrhte-riyoaiinimenlfdpoeiofdnrgttehacnedoscrseeoyusrotgaefhimndt-fsrte,oodulfeegrawharnteeatafsfscoaurartlstemivhcaaaorjvosv.ererAfbeadescetanowrimrthienasudhyleitig,ehtldoa- dforidentifi Ecosystem LSS LMS LNS LNS USS LSS LNS LMiS LNS everestress,LM
yielding varieties such as Swarna, Samba Mahsuri, IR64, and e s
ct nd
MTU1010 that were specifically developed for irrigated ecosys- u a
d wl
tems and suffer heavy yield losses under drought. On the other n o
o l
hand, several drought-tolerant cultivars, such as N22 and c =
Mngbyieuaenottludrelorsaabpcflkeooorrtfeeraktnroaatmlrineate,rilfarihscna’aincavaedlensodpeeflovecaoocolrtnwviosegiunddrmeaoiinenrvraestnrq’hguetpehasreeleoiftfryeyare.beiainInornfscte.iedcmTbaehaencnecodsyaseuyacsscbeatuiesolmoettisfivs,ctahtasrhetsrrisercoisnaulsrogegrlwshye, ofexperiments Season DS2011 DS2012 DS2011 DS2012 WS2012 DS2013 DS2013 WS2013 WS2013 =wetseason,LSSne.0109574.t001
csdagmtunreraalvedTtjiesionhvslbreoaeiysporr.ileteaisThalchcdsihkgoosithwn(srGo-sesyfpsYifsrteoee)oolfrsdflvu.eeiitnrncdhatdgeeinvsecvseraleaodsrwueoirelnofetpiiucqaergtsuoihocegwtonrmoiehtscphasbrpviititonenoerarltbetibuiareornaeeniannetnycdoedrifnfeoopglbrofoiwdorrmrittceiouhceduletgabriaphinrtslteade-ebtthdoiaalelebbietriiyrsootawttntoiioocctf DetailsTable1. Experiment 1 2 3 4 5 6 7 8 9 DS=dryseason,WSdoi:10.1371/journal.po
PLOSONE | www.plosone.org 2 October2014 | Volume 9 | Issue 10 | e109574
IntegratingQTLMappingandRiceBreedingforStressTolerance
varieties[5].Withincreasedprecisioninlarge-scalephenotyping, rainfed and irrigated ecologies across India, Nepal, and Bangla-
many studies have shown moderate to high heritability for GY desh andisregarded asa mega-variety ofrice.
underdrought[6–8].ThesestudiesuseddirectselectionforGYin Figure 1 presents the strategy used to develop the mapping
progeniesderivedfromhigh-yieldingdrought-susceptiblevarieties population.TheBC F plantsdevelopedthroughthecrossofthe
2 1
and drought-tolerant donors as a successful strategy to improve twoparentswereadvancedthroughthesingleseeddescent(SSD)
yield under drought. This strategy has also proven beneficial to methodtodevelopalargeBC F population.Thispopulationwas
2 2
combining high yield potential with drought tolerance and has screenedfortoleranceofriceblast.Blast-tolerantplantsidentified
resulted in the development and release of several high-yielding through this screening were advanced to the BC F generation
2 3
rice varieties with good yield under drought in countries such as through SSD. Seeds from BC F single plants were used to
2 3
India, Nepal, Bangladesh, Indonesia, and the Philippines [9]. develop a BC F mapping population to screen for GY under
2 3:4
Alongwiththereleaseofvarietiesthroughconventionalbreeding droughtstressandnon-stressconditions.Furthermore,singleplant
approaches,severallarge-effectQTLsforGYunderdroughthave selectionsweremadefromthemappingpopulationunderdrought
alsobeenidentified[2,10–15].Similarly,severalstudieshavealso stress. These selections were advanced, purified, and screened at
reportedindependentandepistaticQTLsforGYandothertraits IRRI, Philippines and IRRI-SA, India to identify lines that
of agronomic importance [16–19]. However, only a few studies performthebest inbothenvironments.
have reported QTLs leading to a yield advantage under both
drought stress and non-stress environments, taking grain yield as Stress ecosystems
theprimaryselectioncriterion.Occurrenceofdroughtisnormally FieldexperimentsformappingQTLswereconductedinupland
coupled with increased incidence of diseases such as rice blast, and lowland ecosystems. Throughout the study, the term
brown spot, andbacterial blight. However, very few studies have ‘‘upland’’ refers to field experiments conducted under dry direct-
been undertaken to understand the genetics of these abiotic and seeded, unpuddled, aerobic conditions, whereas ‘‘lowland’’ refers
biotic stresses simultaneously ina mapping population. to field experiments conducted under flooded, puddled, trans-
In our study, an advanced backcross (AB) QTL approach was planted, and anaerobic conditions. Experiments were conducted
used on a large backcross population developed from a cross underbothdroughtstressandnon-stressconditions.Inthisstudy,
between tropical japonica, drought- and rice blast-tolerant donor fully irrigated experiments with no drought stress imposed
Moroberakan with highly popular recipient indica rice variety throughout the season are referred to as non-stress experiments;
Swarna.Theobjectiveswereto:(1)developastrategytocombine whereas,thoseexperimentsonwhichdroughtstresswasimposed
breedingapproachesfortoleranceofblastanddroughtwithQTL during the reproductive stage (starting at 40 and 50 days after
mapping, (2) understand the effect of a selection cycle for blast seeding, respectively, under upland and lowland conditions) are
tolerance during population development on segregation distor- referred to as drought stress experiments. In order to further
tion and population structure, (3) identify QTLs for GY under classifythestressexperimentsintomild,moderate,orseverestress,
drought stress and non-stress conditions, and (4) develop high- the percentage reduction in yield as compared to non-stress
yielding blast- and drought-tolerant breeding lines from the conditions wasused asa criterion [6].
mappingpopulationandevaluatetheminthetargetenvironment.
Phenotypic evaluation and data collection
Materials and Methods Details of all the experiments, including population size,
experimental design, number of replications, and plot size are
Experiments for this study were conducted at the Experiment
presented in Table1. A population of 361 BC F -derived lines
Station of the International Rice Research Institute (IRRI), Los 2 3
was screened under lowland stress and non-stress conditions in
Ban˜os,Laguna, Philippines,duringthedryseasons(DS)of2011,
DS2011 and DS2012. A subset of 100 lines from this population
2012,and2013andthewetseason(WS)of2012.IRRIislocated
was screened under upland stress conditions in WS2012, along
at14u139Nand121u159E,atanelevationof21mabovemean
withparentsMoroberekanandSwarna.Breedinglinesdeveloped
sea level. The study also presents results from two experiments
throughplantselectionfromthispopulationwerescreenedunder
conducted at the IRRI South Asia breeding hub (IRRI-SA)
lowland stress and non-stress conditions in DS2013 at IRRI and
locatedattheInternationalCropResearchInstituteforSemi-Arid
WS2013 at IRRI-SA. All experiments were conducted as
Tropics (ICRISAT), Patancheru, Hyderabad (AP), India in
replicated field trials in a lattice design in continuous plots. The
WS2013. ICRISAT is located at 17u319 N and 78u159 E at an
experimental setup has proved effective for screening large rice
elevation of516m above sealevel.
mapping populations for grain yield under drought using smaller
plotsizes.ThemethodhasbeenusedinseveralQTLidentification
Plant material
studies,whichhaveledtotheidentificationoflargeandconsistent
A BC F -derived population developed from the cross of
2 3 QTLs now being used in marker-assisted breeding (MAB)
Moroberekan and Swarna was used in this study. Moroberekan,
programs [2,10–15]. Field and crop management were carried
the tolerant donor, is an upland adapted tropical japonica [20]
outusing theprotocolpresented inFile S1.
cultivar from New Guinea. It is a long-duration line with sturdy
In all experiments, data on days to flowering (DTF), plant
plant type, deep roots, and tolerance of drought and rice blast.
height(PH)atmaturity,andgrainyield(GY)wererecorded.DTF
However, this variety has poor yield potential due to its low
wasrecordedasthenumberofdaysfromseedingtothedaythat
tillering ability and lower number of grains per panicle. On the
50% of the plants had flowering tillers. The PH of three plants
otherhand,Swarna(MTU7029),thedrought-susceptiblerecipient
fromeachplotweremeasuredatmaturityfromthegroundtothe
parentusedinthisstudy,isalowland-adaptedhigh-yieldingindica
tipofthetallesttiller,andthenaveragedtogetthemeanPHfor
[2,21]varietyderivedfromthecrossVashishtha6Mahsuri.Itisa
analysis. GY from each plot was harvested at physiological
long-duration semi-dwarf variety with high tillering ability and
maturity, dried to a moisture content of 14%, and then weighed
grain yield but is highly susceptible to diseases such as rice blast
[11].Thisdatasetwasthenusedtocalculatethequantityusedin
and bacterial blight. This variety is grown on a large area in QTL analysis, namelyGY ofthegenotypes inkg ha21.
PLOSONE | www.plosone.org 3 October2014 | Volume 9 | Issue 10 | e109574
IntegratingQTLMappingandRiceBreedingforStressTolerance
Generation of genotypic data H acrossyears (2011and2012)was
Fresh leaves for all lineswere collected andfreeze-dried. DNA
wasextractedfromfreeze-driedleafsamplesbyamodifiedCTAB s2
mpuertihfioedd iannddeelipn-ewsewlleprleatgese.noTthyepeDdNuAsinwgasKtAhSenPaqruSanNtPifieadssaanysd. H~s2z(cid:2)s2GMG zs2e(cid:3)
G M rM
TheseSNPswereselectedassubsetsfromthesetof1,536and44K
SNP chips [22,23] converted to SNP assays and made available
[24],wheres2 isthegeneticvariance,s2 isthegenotype6year
through the integrated breeding platform (https://www. G GY
variance, s2 is the plot residual variance, and r and M are the
integratedbreeding.net/snp-marker-conversion). A total of 2,015 E
numberofreplicatesandyears,respectively.Variancecomponents
SNP markers were screened for polymorphism between the two
were calculated using the REML algorithm of PROC VAR-
parents. Out of the 2,015 SNPs screened for polymorphism, 591
COMP of SAS V.9.1 [25]. A class analysis was conducted to
polymorphicSNPlociwereidentified.Atotalof200polymorphic
understandtheeffectofDTFandPHonGY.Linesweredivided
markers spaced uniformly across the genome were used to
into five DTF and GY classes and heat maps were constructed
generate the genotypic data of the population. To ensure data
showingGYresponseinallpossiblecombinationsoftheseclasses
quality, lines having percentages higher than expected of
under drought stress and non-stress conditions. ‘Levelplot’ in R
heterozygote alleles (.40%), Moroberekan alleles (.40%), or
was usedtoconstruct theheatmaps.
missing data (.30%) were filtered before analysis. Furthermore,
markerswereremovedifthefrequencyofgenotypesderivingfrom
Moroberekan (homozygote and heterozygote) was ,2%, as they QTL mapping
were essentially monomorphic. Genetic analysis of the population included estimation of
segregation distortion, composite interval mapping (CIM), and
Statistical analysis estimation of epistatic effects. The level of segregation distortion
was assessed using the function ‘geno.table’ in the R package qtl
ANOVAandmeans. Data from all experiments for compu-
[26]. In order to understand the co-location of markers showing
tation of means and standard error of difference (SED) were
significant distortion to blast QTLs and known genes for disease
analyzedusingCROPSTATversion7.2.3.Mixedmodelanalysis
resistance in rice, a physical map of SNP markers (used in this
of datafromindividual years was carriedout usingthemodel
study), QTLs for blast resistance (reported in Moroberekan-
derivedmappingpopulations),andknowndiseaseresistancegenes
y ~mzgzrzb (r)ze
ijk i j k j ijk inricewasconstructed.Informationaboutdiseaseresistancegenes
aswellasthosewithintheidentifiedQTLsandtheirlocationswas
where y is the measurement recorded in a plot, mis the overall
ijk obtained from the MSU Rice Genome Annotation Project,
mean, g is the effect of the ith genotype, r is the effect of the jth
i j whereas QTL information was obtained from www.gramene.org
replicate,b(r)istheeffectofthekthblockwithinthejthreplicate,
k j andusedalongwiththeSNPmarkersfromthisstudytoconstruct
and e is the error. Genotypic effects were considered fixed and
ijk amapusingGGT2[27].TomapQTLsforGY,DTF,andPH,
thereplicates and blockeffects wererandom.
the genome was scanned for QTLs at a step size of 5 cM, using
Acombinedanalysiswasconductedforlowlandexperimentsto
compositeintervalmapping(‘cim’)inR/qtl[26]with5cofactors.
obtain line means across years (2011 and 2012) under stress and
The genome-wide significance level was determined based on
non-stress conditions.Themodel usedwas
1,000 permutations. A two-dimensional scan was also performed
to identify epistatic QTLs using ‘scantwo’ in R/qtl with 1,000
y ~mzgzmzr(m)zb ½r(m)(cid:2)zgmze permutations.
ijkl i l j l k j l i l ijkl
whereyijklisthemeasurementrecordedinaplot,mistheoverall Results
mean,g istheeffectoftheithgenotype,m istheeffectofthelth
i l
year, r(m) is the effect of the jth replicate within the lth year, Population development and segregation patterns
j l
b[r(m)]istheeffectofthekthblockwithinthejthreplicateofthe Our study reports a population development strategy using
k j l
lth year, gm is the effect of the interaction between the ith donors tolerant of more than one biotic and/or abiotic stress for
i l
genotypeandlthyear,ande istheerror.Theeffectsofreplicate simultaneous QTL mapping and pureline development in elite
ijkl
within year and block within replicate of year were considered genetic background (Fig. 1). In this study, tropical japonica
random, whereas the other effects were considered fixed. Due to cultivarMoroberekanwasusedasthedonorparentfortolerance
thehighvariabilityforDTFinthepopulation,adjustedmeanGY ofdroughtandriceblast,whereasindicavarietySwarnawasused
were also calculated by taking DTF as covariate using the above as the recipient. Our strategy involved developing a large BC F
2 2
models for QTL mapping. This was done to estimate the mean populationandscreeningthispopulationforblasttolerance.Only
GYoflinesfreefromthevariationcausedduetotheirdifference blast-tolerantplantswereadvancedfurthertodeveloptheBC F -
2 3
in flowering time(FileS3). derived mapping population, whichwas used to detect QTLs for
Heritability. Broad-sense heritability (H) of the traits for grain yield under drought and to develop purelines by plant
singleyears was calculated asshown below: selectioninthemappingpopulation.Sinceselectionpressurewas
applied for blast tolerance, segregation distortion was expected
s2 within the mapping population. The genotypic data generated
0
H~ G(cid:2) (cid:3) from BC2F3-derived lines were used to conduct the test for
s2 z s2e segregation distortion. A total of 50 and 9 marker loci showed
G0 r segregationdistortioninfavourofthedonorparentMoroberekan
andrecipient parent Swarna, respectively (TableS1). Thelargest
wheres2 isthegeneticvariance,s2 istheplotresidualvariance
G0 E segregation distortions were seen on chromosomes 3, 4, and 7
andristhenumberofreplications.Theformulausedtocalculate (TableS1).SegregationdistortioninfavourofSwarnawasseenat
PLOSONE | www.plosone.org 4 October2014 | Volume 9 | Issue 10 | e109574
IntegratingQTLMappingandRiceBreedingforStressTolerance
Figure2.Physicalmapofmarkersshowingsegregationdistortion,knowngenesfordiseaseresistanceinrice,andQTLsreported
forblastresistanceintheMoroberekan/CO39population.M = markersshowingdistortioninfavourofMoroberekan,S = markersshowing
distortion in favour of Swarna, N = genes coding for NBS-LRR proteins, and T = QTLs reported for blast resistance in the Moroberekan/CO39
population.
doi:10.1371/journal.pone.0109574.g002
chromosomes 1, 8, and 12. The physical map of SNP markers to the markers showing distortion may be expected as a result of
from this study, known disease-resistance genes, and meta QTLs phenotypicselection for blast tolerance.
forblastresistancealongthericegenome,highlightedNBS-LRR
disease resistance genes close to the markers showing significant Phenotypic variation
segregationdistortioninfavourofMoroberekanonchromosomes Table 2 presents the results of ANOVA, means, heritability,
3, 6, and 7 (Fig.2). Since the majority of blast resistance genes and percentage yield reduction (under drought stress) in the
codeforNBS-LRRtypeproteins,thepresenceofthesegenesclose experiments conducted at IRRI and IRRI-South Asia (IRRI-SA)
PLOSONE | www.plosone.org 5 October2014 | Volume 9 | Issue 10 | e109574
IntegratingQTLMappingandRiceBreedingforStressTolerance
ain II)
gr YNS
mean YR 93 45 66 80 83 40 GII)(L 1
S
htandpercentageyieldreduction(YR)oftrial Plantheight(cm) PP1P2MH 672566890.8**** 6102768280.87**** 687667590.70**** 61269110280.84**** 61189299100.69**** 61229010160.80**** 6107758970.85**** 684587460.86**** 61289110070.82**** 6NANA9760.59*** 6NANA11670.61**** droughtstressandnon-stressconditions. PH(LNSI)GY(LNSI)DTF(LNSII)PH(LN 1 0.317***1 220.166**0.226***1 0.894***0.123*0.0371 220.0580.861***0.368***0.095 respectively.
andplantheig P **** **** **** **** **** **** **** **** **** **** **** underlowland DTF(LNSI) 1 20.297*** 20.403*** 0.810*** 20.086 20.482*** 1%0.01%Plevels,
dPvaluesforgrainyield,daystoflowering, Daystoflowering PP1P2MH 6****1091109940.85 6****1001029250.83 6****1041069550.87 6****97989150.72 6****93938870.78 6****95959050.72 6****9511210050.88 6****1051058720.98 6****94958620.95 6****NANA9420.85 6****NANA9330.92 y,****=significantat0.01%Plevel. ystoflowering(DTF),andplantheight(PH) DTF(LMS)PH(LMS)GY(LMS) 1 20.0401 220.679***0.145**1 220.869***0.166***0.568*** 220.170***0.969***0.063 20.318***0.264***0.579*** 220.845***0.137*0.664*** 20.0950.801***0.238*** 20.457***0.0840.949**** andnon-stress.*,**,***,****=significantat5%,1%,0.
an bilit da owl
6MeansSED(M),broad-senseheritability(H)Table2.yieldunderstressandnon-stressconditions. 21Grainyield(kgha) ExperimentP1P2MH 6188876775630.60 62954232621156880.61 6Combined384121113906260.55 632234455243708680.78 6412484012385910490.54 6Combined1701428241186760.71 65691898273570.92 660012744330.93 6728936857738016280.67 68NANA47578560.83 69NANA789510270.31 aP1=Moroberekan,P2=Swarna,M=populationmean,H=heritadoi:10.1371/journal.pone.0109574.t002 Geneticcorrelationbetweengrainyield(GY),Table3. DTF(LSS)PH(LSS)GY(LSS) DTF(LSS)1 2PH(LSS)0.597***1 2GY(LSS)0.813***0.551***1 22DTF(LMS)0.936***0.495***0.716*** 22PH(LMS)0.0420.727***0.018 2GY(LMS)0.519***0.168***0.711*** 22DTF(LNSI)0.998***0.699***0.794*** 2PH(LNSI)0.238***0.831***0.185*** 2GY(LNSI)0.297***0.348***0.308*** 22DTF(LNSII)0.877***0.426***0.672*** 2PH(LNSII)0.114*0.668***0.204*** 2GY(LNSII)0.401***0.245***0.471*** LSS=lowlandseverestress,LMS=lowlandmoderatestress,LNS=ldoi:10.1371/journal.pone.0109574.t003
PLOSONE | www.plosone.org 6 October2014 | Volume 9 | Issue 10 | e109574
IntegratingQTLMappingandRiceBreedingforStressTolerance
Figure3.HeatmapsshowingtheeffectofDTFandPHonyieldperformanceoflinesunder(A)lowlandseverestress(LSS),(B)
lowlandmoderatestress(LMS),(C)lowlandnon-stress(LNSI)and(D)LNSIIconditions.Increasingintensityofcolorsfromyellowto
orangeindicatesincreasingyieldlevelswhilewhitecolorshowstheunavailablePHclass.
doi:10.1371/journal.pone.0109574.g003
PLOSONE | www.plosone.org 7 October2014 | Volume 9 | Issue 10 | e109574
IntegratingQTLMappingandRiceBreedingforStressTolerance
Figure4.CirculargenomeplotshowingthepositionsofQTLsidentifiedforgrainyield(adjustedforfloweringtime)underdrought
stressandnon-stressconditions.LSS = lowlandseverestress,LNS = lowlandnon-stress,LCS = lowlandcombinedstress,LCNS = lowland
combinednon-stress,UMS = uplandmoderatestress.
doi:10.1371/journal.pone.0109574.g004
under upland and lowland conditions. The full population was theWSexperiment(5)whenMoroberekanfloweredmuchearlier
screened in Experiments 1-4 under stress and non-stress condi- than Swarna. Similar observations were also recorded in seed
tions. Yield reduction in Experiments 1 and 2 varied with stress multiplicationtrialsunderlowlandconditionsinWS2013(datanot
levels in the respective seasons from severe for Experiment 1 to presented).
moderateinExperiment2.Experiments6and7followedsimilar
trendsasExperiments1and2hadintermsofyield.Bothparents Trait correlation and effect
hadsimilaryieldunderseverestress,however,Swarna(P2),being GeneticcorrelationsbetweenDTF,PH,andGYunderlowland
the high-yielding parent, had higher yield under moderate stress stress and non-stress conditions (DS2011 and DS2012) are
conditions in the lowland. The high yield of Swarna under presented in Table 3. These four experiments were conducted
moderate stress was also due to the poor adaptation of using the whole population, and hence, were used to understand
Moroberekan, an upland cultivar to lowland conditions. The traitcorrelations.ClearcorrelationpatternsbetweenDTFandGY
populationmeansremainedhigherthanthoseforparentsinboth were seen in all experiments. The two traits were negatively
stress experiments, suggesting the role of high yield and drought correlatedinbothstressandnon-stressexperiments.Regardlessof
tolerance indecidingtheyieldunder drought. stressseveritylevels,earlinessoflinesrelatedtohigheryieldunder
The advantage of Moroberekan (P1) over Swarna was seen stress environments. Also, under non-stress conditions, a similar
moreclearly underuplandconditions(Experiment 5),inwhicha pattern was observed, indicating that the reduced growth period
large difference was recorded in the yield of both parents. Mild didnotleadtoreducedyieldpotential.Thedegreeofcorrelation
stresslevelswererecordedinExperiment8asthisexperimentwas howeverwasmuchhigherinstressexperimentsthaninnon-stress
conducted in the wet season under naturally occurring drought experiments. In contrast to DTF, no clear correlation patterns
stress. Moderate to high heritability estimates were recorded for between PHandGY wereobservedinbothstressandnon-stress
yieldin allexperiments. conditions.
In all DS experiments, both parents had similar DTF under InordertofurtherunderstandtherelationshipsbetweenDTF,
stressandnon-stressconditions,buttheirprogenyshowedawide PH, and GY, a class analysis was conducted to measure GY for
rangeofDTFvalues.WhileSwarnaisawell-knownlong-duration varyingcombinationsofDTFandPHlevels(Fig.3).Atanylevel
variety,therewasapossibilityofMoroberekancontributingtothe of PH, greater differences were seen in GY for varying levels of
earliness of theprogeny. While, thisearliness wasnot seen in the DTF than vice versa. This agrees with the higher correlation
cultivar itself in DS experiments, this difference became clear in observed between DTF and GY than between PH and GY. In
PLOSONE | www.plosone.org 8 October2014 | Volume 9 | Issue 10 | e109574
IntegratingQTLMappingandRiceBreedingforStressTolerance
ht
g
u
o
dr
e
g
ctive-sta P ,0.001 ,0.001 0.009 0.009 0.047 0.012 0.025 0.050 0.014 0.001 0.003 ,0.001 ,0.001 ,0.001 ,0.001 ,0.001 0.008 0.002 0.008 0.031 ,0.001 ,0.001 0.024 ,0.001 ,0.001 0.040 0.028 ,0.001 ,0.001 ,0.001 ,0.001 0.031 ,0.001 ,0.001
u
d
o
pr
e
r
e
oderat A 281.6 211.3 552.5 1743.0 2350.0 2353.4 1223.0 620.7 417.6 1967.0 5.1 5.6 4.2 2.0 9.5 9.1 6.5 4.4 6.6 23.3 7.6 6.5 4.6 25.0 25.6 1.9 22.5 24.9 26.7 25.2 26.6 22.7 23.4 22.6
m
or
/
d
n
a 0 0 0 0 0 0 0 0 0 0 0 0 0
ere 2R 16. 14. 5.5 17. 4.9 4.7 10. 4.8 19. 25. 4.2 6.3 4.9 4.4 31. 9.1 4.5 3.6 9.1 4.9 9.1 7.4 2.8 30 8.4 3.4 4.9 29. 13. 34. 13. 7.9 17. 14.
v
e
s
er
d
n
u 0 0 0 0 0 0 0 0
ed cM 15.0 15.0 154. 13.5 69.3 54.1 36.7 212. 0.0 17.4 154. 70.9 95.0 43.0 150. 154. 75.0 43.0 74.3 58.2 154. 74.3 107. 20.0 91.0 69.3 5.0 20.0 60.4 20.0 60.4 130. 25.0 0.0
ntifi
e
d
i
F)
T
D
(
e
m
ti e
g m
n o
eri os
w m
dflo Chro 3 3 1 5 10 9 11 12 3 11 1 2 8 10 1 1 2 10 2 3 1 2 7 3 7 10 11 3 8 3 8 1 3 3
n
a
H),
P
(
ht
g
ei
h
me(GY),plant2. Peakmarker Id3000757 Id3000757 id1021344 wd5002636 id10006250 id9007259 id11009456 ud12000118 id3000019 id11002801 Id1021344 Id2006643 Id8007896 Id10004477 Id1021344 Id1021344 Id2007213 Id10004477 Id2007213 Id3002556 Id1021344 Id2007213 Id7005631 Id3000757 Id7003271 Id10006250 Id11000855 Id3000757 Id8004692 Id3000757 Id8004692 Id1014783 Id3000946 Id3000019
1
ti0
eringdDS2
wn
oa
fl1
or01 m
f2 e
djustedsinDS Ecosyst LSS LCS LNSI LNSI LNSI LCNS LCNS LCNS UMS UMS LSS LMS LMS LMS LCS LNSI LNSI LNSI LNSII LNSII LCNS LCNS LCNS LSS LSS LSS LSS LMS LMS LCS LCS LNSI LNSI LNSII
an
o
yieldnditi
no
aic
grss
e
QTLsfornon-str QTL qDTY3.2 qDTY3.2 qDTY1.1 qDTY5.1 qDTY10.2 qDTY9.2 qDTY11.2 qDTY12.2 qDTY3.2 qDTY11.1 qDTH1.1 qDTH2.1 qDTH8.1 qDTH10.1 qDTH1.1 qDTH1.1 qDTH2.1 qDTH10.1 qDTH2.1 qDTH3.1 qDTH1.1 qDTH2.1 qDTH7.1 qDTF3.1 qDTF7.1 qDTF10.1 qDTF11.1 qDTF3.1 qDTF8.1 qDTF3.1 qDTF8.1 qDTF1.1 qDTF3.1 qDTF3.1
d
4.n
a
Tablestress Trait GY PH DTF
PLOSONE | www.plosone.org 9 October2014 | Volume 9 | Issue 10 | e109574
IntegratingQTLMappingandRiceBreedingforStressTolerance
== general,lineswithmediummaturity(DTF=81–95days)werethe
P 0.023 ,0.001 0.044 ,0.001 ,0.001 2ofthepeakmarker,Rmbinednon-stress,USS mesmGhxeoeopnnswettreesitdmcaiocsetnhncvsetoiassm.trmeiHpanoattosirtoweadncetovtneoarsl,iltshltieolnensevetseplbwseeirytofhoofnrpmdplalatanhnntectseheehailgeecihvrgoethlssftsr.otahmcero1fos0su1r–a1el1lx0pfoecrumir-
ectco AnalysisofGYidentifiedseveralQTLsforeachtraitsignificant
effnd within and across environments (Fig 4.,Table4). The most
dditive=lowla ecoffnecsitsteunntdeQrTdLroiudgehnttifisetdresfsoracGrYosswbasotqhDTupYl3a.n2,dwahnicdhlhoawdlanand
a
1.5 3.3 4.0 7.0 4.2 =NS conditions. The QTL explained a phenotypic variance effect of
A 2 2 2 2 2 ALC 16.0% and 19.0% under lowland severe stress (LSS) and upland
detection.non-stress, mqQDoTTdLYer1ea1xt.1eplwastianrseesiddsea(nUtpiMfhieeSdn),owtryaepssipicdeecvntaitvriiefaileyndc(euTnoadfbel2re5U4.0)M.%AS.pcTaoorntdthfirteioombne.sTtthhiosef,
2R 3.9 21.0 7.8 11.0 19.0 oftheirowland GouYrkunnodwerleudpgela,nthdisdrisouthgehtfirssttretsims.eqDTY11.1isbeingreportedfor
der=l Inthisstudy,largeQTLsforDTFwereidentified,whichreflect
cM 125.0 25.0 118.0 90.0 0.0 oseafteraretheormbinedstress,LNS ictQbnhheeTrttohLnhmieesfgoosmastrtouioDmvdseytTe.cscFSoo2ern(,rqvse6Deils,ratTa7etilFn,oot83n.t,Q1sh1)ebT0wreLtaaQwsnaeTdacelLrn1soos1sDfsiaodTldretFhrnDootauiuTfngigeFdhhdtGwqasDeYttrrTeteuhsFsnae3ldas.q1eonDrrdoTedbmnYrsoo3eau.nri2nvg-sehelotddtrc.eutaAsosst
easthndco cidoenndtiitfiioednsac(Troasbsleth4e).geSnimomilaerlwy,ithforqDPTHH,1.1sevaesrathleQlaTrLgesstwaenrde
herwla most consistent QTL for the trait (Table4). Majority of the GY
me mber,wCS=lo QbeTreLksanu;nhdoewresvtreers,stwanodQnToLns-st(rqeDssTYw9e.2reancdonqtrDibTuYte10d.2)byforMyoierold-
Chromoso 1 3 6 2 3 omosomenueratestress,L uPdfonoHrndoeParrHnpndaoarnDen-ndTsttrFMeq,sDsomTrwoaFebj1ore0err.i1etayklsfaooonrf,cbotDhunTettroFiQbn)ueTwtQeLadssTbwLaylesfrooSerwcecaoaorncnnhtatrr.itibrbSauuiitmttee(didqlaDrbbTlyyyH,ttf3hho.ee1r
chrmod recipient Swarna.
thend While trangressive segregants were observed in the population
otewla for all three traits, the contribution of both parents to higher
malsdenMS=lo phvaearrvifeaonrbmceeacnopmcreeesoecnflteparforo.greIgnnryaoilnrindeeysriealtdnodufunthrdtehereprrdeersoxeupnglcaheintosftthreeepsisspthaatenincdoetnfyfopencic-t
Peakmarker Id1014783 Id3000946 Id6010576 Id2008112 Id3000019 bersbeforedecidseverestress,L sEbetQerpteTiwsstLsea1et0incc.o1nicsndeoitvnteeisroriasanctlset,inoltontlyecssistfhsouornwfodeGredrYeeppnuiiossnttnada-ttesiirtccreinnsisontentre-carsocatnrtciedtosiionstinsowswnesirt.wheeHoroetobhwsedererovvlnoeeecrd,.i
mn across the genome (Table 5). Other consistant loci include
Nuwla eQTL and eQTL at chromosomes 1 and 8 respectively
ely.=lo (Table1.51). The high8a.n1d consistent epistatic interactions of these
v
ctiSS locisuggesttheirimportanceinexplainingthegeneticbasisofGY
eL
espTL, in lines under non-stress conditions. In this study, lines with high
Ecosystem LCNS LCNS LCNS UMS UMS PH,andDTF,rainedbytheQ yirydeiieeecllniddptiiefpuineontdtdepe(nTratriaabeblnolettehqSSuw2da)ra.lorDuntogaeh,sopttrihtseetmrsbeeoserslieinnagetnhsmdawnonreoerSenwe-aaasrtbrrlnyleea-smstuoacntomudnreadirniintgintoaotnhinnsa-snwtthrteeehrsieers
QTLsforGY,varianceexpl 109574.t004 ctohocencLudinrittraeiiorsngngiedst.ednreotinufivgeihdrotfnrsotmrmeesnstthaen(mIdRanRpopIn-iS-nsAgtr)pesosfpocurolantGdioiYtniownuesnrtdeoeeruvnanlduaeatrtusetrdaanlilndy
Cont.Table4. TraitQTL qDTF1.1 qDTF3.1 qDTF6.1 qDTF2.2 qDTF3.1 qDTY,qDTH,andqDTFarepercentageofphenotypicuplandmoderatestress.doi:10.1371/journal.pone.0 tsSdhsMhhciroiAgageooln.thdurticeeefAgoisrrchinlmatstpsshi-neigltiosoltvonta,uteelirsegfrnbirihbycctuayaiyentetntlthwoodtneffleoipcenpssoantteenrrtrrseetertfehsgeossrssaleernamltiletsiGnevicaovaetYnnetethclsdcIsoeeoRfmiwrontRsraefeaerIrDllytg-eaehSectSeiAthtvo2eeal.n0edirvnn1sTyelev3ihwsnim.rebeaeoiFestrlRnedeiagImn2,RtuoetvrIRbhanaReseptlIeuR.5prhewaLIvpsireegiraenwrdhennees-edtsey(raFnieIfebuitiRglnslnnde.aRodtin5hlnetlIo)gyer--.
PLOSONE | www.plosone.org 10 October2014 | Volume 9 | Issue 10 | e109574