<%BANNER%>

Breeding for Improved Growth, Wood Quality, and Chemistry for Southern Pines by Combining Quantitative Genetics and Asso...

Permanent Link: http://ufdc.ufl.edu/UFE0024692/00001

Material Information

Title: Breeding for Improved Growth, Wood Quality, and Chemistry for Southern Pines by Combining Quantitative Genetics and Association Mapping
Physical Description: 1 online resource (171 p.)
Language: english
Creator: Li, Xiaobo
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: asreml, association, bayesian, correlation, functional, genetics, heritability, marker, pca, pine, pymbms, snp, statistics, stiffness, trait, wood
Forest Resources and Conservation -- Dissertations, Academic -- UF
Genre: Forest Resources and Conservation thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: In the economically important southern pine species, loblolly and slash, the genetic architecture of wood chemistry and stiffness were determined and association genetics was used to identify genes that potentially regulate these economically important traits in loblolly pine. The in-tree stiffness of juvenile corewood in a progeny trial of 139 families of slash pine had moderate level of genetic control with a heritability (h^2=0.42) which is generally higher than growth traits, i.e., height and diameter at breast height. No significant genetic correlations were observed between velocity stiffness, DBH and volume growth. Wood chemistry in the juvenile corewood of a clonally propagated progeny trial of 61 families of loblolly pine was measured by pyrolysis molecular beam mass spectrometry (pyMBMS). Genetic analysis of all 421 peaks in the spectra identified 32 with significant genetic control. Of these peaks, about half are chemically identified. The chemically identified mass to charge peaks alone or summed together for five and six carbon sugars, lignin and C20 extractives all had low levels of genetic control with a clonal repeatabilities of 0.11 - 0.16, which are lower than growth traits (height and DBH). Pairwise genetic correlations between peaks showed that known lignin and sugar peak intensities are strongly negatively, genetically correlated; while known carbohydrate and lignin peaks are strongly positively, genetically correlated with others from the same class. In addition, 9 chemically unidentified peaks are strongly correlated with other chemically identified peaks, suggesting the possible chemical class of these peaks. Principal component and wavelet analyses of the spectra show that the genetic variation is better segmented by wavelet than by principal components. Finally, a Metropolis Hastings algorithm was developed and used to search for combinations of 2, 3, 4 and 5 peaks that maximized the heritability. This search algorithm successfully identified many combinations of known peaks within the same class, and also identified combinations of unknown with chemically identified peaks, supporting pairwise genetic correlations. The performance of a new Bayesian association method that uses a Gibbs sampler to impute missing data (BAMD) was tested using simulated data for structured and unstructured populations and real data from a structured population to identify SNPs known to significantly affect phenotypes that have varying degrees of genetic control. Association genetic analyses were completed with BAMD to identify significant SNPs for wood property phenotypes. From a set of 2182 genotyped SNPs in a richly structured clonally propagated CCLONES population, 87 significant SNPs were identified for wood chemistry and in-tree velocity stiffness. In-tree wood velocity stiffness is a moderately heritable trait, and 6 SNPs were significant at the 95% level, but an additional 9 were identified at the 90% level.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Xiaobo Li.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Peter, Gary F.
Local: Co-adviser: Casella, George.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024692:00001

Permanent Link: http://ufdc.ufl.edu/UFE0024692/00001

Material Information

Title: Breeding for Improved Growth, Wood Quality, and Chemistry for Southern Pines by Combining Quantitative Genetics and Association Mapping
Physical Description: 1 online resource (171 p.)
Language: english
Creator: Li, Xiaobo
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: asreml, association, bayesian, correlation, functional, genetics, heritability, marker, pca, pine, pymbms, snp, statistics, stiffness, trait, wood
Forest Resources and Conservation -- Dissertations, Academic -- UF
Genre: Forest Resources and Conservation thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: In the economically important southern pine species, loblolly and slash, the genetic architecture of wood chemistry and stiffness were determined and association genetics was used to identify genes that potentially regulate these economically important traits in loblolly pine. The in-tree stiffness of juvenile corewood in a progeny trial of 139 families of slash pine had moderate level of genetic control with a heritability (h^2=0.42) which is generally higher than growth traits, i.e., height and diameter at breast height. No significant genetic correlations were observed between velocity stiffness, DBH and volume growth. Wood chemistry in the juvenile corewood of a clonally propagated progeny trial of 61 families of loblolly pine was measured by pyrolysis molecular beam mass spectrometry (pyMBMS). Genetic analysis of all 421 peaks in the spectra identified 32 with significant genetic control. Of these peaks, about half are chemically identified. The chemically identified mass to charge peaks alone or summed together for five and six carbon sugars, lignin and C20 extractives all had low levels of genetic control with a clonal repeatabilities of 0.11 - 0.16, which are lower than growth traits (height and DBH). Pairwise genetic correlations between peaks showed that known lignin and sugar peak intensities are strongly negatively, genetically correlated; while known carbohydrate and lignin peaks are strongly positively, genetically correlated with others from the same class. In addition, 9 chemically unidentified peaks are strongly correlated with other chemically identified peaks, suggesting the possible chemical class of these peaks. Principal component and wavelet analyses of the spectra show that the genetic variation is better segmented by wavelet than by principal components. Finally, a Metropolis Hastings algorithm was developed and used to search for combinations of 2, 3, 4 and 5 peaks that maximized the heritability. This search algorithm successfully identified many combinations of known peaks within the same class, and also identified combinations of unknown with chemically identified peaks, supporting pairwise genetic correlations. The performance of a new Bayesian association method that uses a Gibbs sampler to impute missing data (BAMD) was tested using simulated data for structured and unstructured populations and real data from a structured population to identify SNPs known to significantly affect phenotypes that have varying degrees of genetic control. Association genetic analyses were completed with BAMD to identify significant SNPs for wood property phenotypes. From a set of 2182 genotyped SNPs in a richly structured clonally propagated CCLONES population, 87 significant SNPs were identified for wood chemistry and in-tree velocity stiffness. In-tree wood velocity stiffness is a moderately heritable trait, and 6 SNPs were significant at the 95% level, but an additional 9 were identified at the 90% level.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Xiaobo Li.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Peter, Gary F.
Local: Co-adviser: Casella, George.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024692:00001


This item has the following downloads:


Full Text

PAGE 1

1

PAGE 2

2

PAGE 3

3

PAGE 4

IthankDr.GaryPeterforhisinvaluableguidanceallthewaythroughmyPh.D.Hisvisiontocombinegeneticsandstatisticsasatooltounderstandassociationbetweengenotypeandimportantphenotypictraitsmotivatedmetojoinandcontributetotheeld.Hehasbeenextremelysupportiveinallaspectsintheseveyears.ThroughhimIhavelearnedagreatdealaboutscience,criticalthinkingandwriting,aswellasgeneralapproachestoproblems.IthankDr.GeorgeCasellaforactingasmyco-advisor.HisteachingandadviceinExperimentalDesignandMonteCarloMethodsbroadenedmyknowledgebaseinmystatisticallearningandindependentthinking.Ireallyappreciatehispatientadvisementandconsiderablementoring.Dr.DudleyHuberhelpedmeunderstandquantitativegeneticsandplantbreedingwithonetooneteachinganddailydiscussions.Dr.MatiasKirsthelpedmeunderstandpopulationgenetics.Dr.JohnDavishelpedmeunderstandplantgeneticsandassociationmapping.Dr.LaurenMcIntyrehelpedmetoclearmymindonthestructureofthedissertation.Dr.AutherBergprovidedvaluablesuggestionstofunctionaldataanalysis.I'dliketothankallofthemfortheirvaluablesuggestionsandtime,withoutwhichprogressinmyPhDwouldhavebeenextremelydicult.EventhoughDr.MaryChristmanwasnotamemberofmycommittee,sheprovidedgeneroushelponstatisticalanalysisandconsulting.FromDr.TimMartinIlearnedagreatdealoftreebiologyandwaterrelationsinplants.Dr.GuyNasonprovidedvaluablesuggestionsandcommentsonhowtousetheWavesleshRpackage.IalsowanttothankourlabmanagerChrisDervinisforhiskindhelpinpreparingexperimentalmaterialsaswellassoftwareassistance.IthankmembersoftheForestBiotechnologyandGenomicslabandSFRCcolleaguesTaniaQuesada,LilianaParisi,GustavoRamirez,BriannaMiles,QibinYu,LuisOsorio,GregPowell,SalvadorGezan,BrianRoth,PatricioMunoz,EvandroNovaes,KathySmith,PhilipBocock,TimilsinaNilesh,DerekDrost,AlejandroWalker,BarbaraKahn,AnnaMwanikiandSTAT 4

PAGE 5

5

PAGE 6

page ACKNOWLEDGMENTS ................................. 4 LISTOFTABLES ..................................... 9 LISTOFFIGURES .................................... 12 LISTOFSYMBOLS .................................... 13 ABSTRACT ........................................ 14 CHAPTER 1LITERATUREREVIEWANDPROJECTOVERVIEW ............. 16 1.1Introduction ................................... 16 1.1.1ForestsandSociety ........................... 16 1.1.2WhyLoblollyPine ........................... 17 1.1.3WoodProperties ............................ 18 1.1.4ConiferWoodMechanicalProperties ................. 20 1.1.4.1ModulusofElasticity ..................... 20 1.1.4.2WoodSpecicGravity .................... 21 1.1.5WoodChemicalProperties ....................... 22 1.2QuantitativeGeneticsofWoodPropertiesinConifers ............ 23 1.2.1StatusofTreeImprovementinSouthernPines ............ 25 1.2.2ApproachestoDiscoverGenesControlWoodPropertiesinConifers 27 1.2.3AssociationGenetics .......................... 29 1.2.4AssociationMappingApproaches ................... 31 1.3ResearchObjectives ............................... 34 2BREEDINGFORIMPROVEDGROWTHANDCOREWOODSTIFFNESSINSLASHPINE ................................... 36 2.1Introduction ................................... 36 2.2MaterialsandMethods ............................. 40 2.2.1MaterialandExperimentalDesign ................... 40 2.2.2VelocityStinessMeasurement ..................... 41 2.2.3SamplingStrategyforVelocityMeasurements ............ 42 2.2.4StatisticalAnalysis ........................... 42 2.3Results ...................................... 44 2.3.1GeneticControlofGrowth,RustResistanceandWoodStiness .. 45 2.3.2EnvironmentalInuencesonGrowthandWoodStiness ...... 45 2.3.3BreedingValuesofHeight,VolumeandV2 46 2.3.4SelectionIndicestoMaximizeGeneticGaininVolumeandVelocityStiness ................................. 46 2.4Discussion .................................... 48 6

PAGE 7

............................... 51 3GENETICARCHITECTUREANDCORRELATIONSOFJUVENILECOREWOODCHEMICALCOMPOSITION ................... 53 3.1Introduction ................................... 53 3.1.1ResearchObjectives ........................... 57 3.2MaterialsandMethods ............................. 57 3.2.1SampleCollection ............................ 58 3.2.2PyrolysisMolecularBeamSpectrometryDataCollection ...... 58 3.2.3StatisticalAnalysis ........................... 59 3.2.4GeneticModel .............................. 59 3.2.4.1Metropolis-HastingsAlgorithm ............... 62 3.2.4.2FunctionalDataAnalysis .................. 62 3.3Results ...................................... 63 3.3.1Heritabilityandtype-Bgeneticcorrelationofidentiedpeaksandmajorcomponents ............................ 63 3.3.2GeneticCorrelationsbetweenHeritablePeaks ............ 69 3.3.3PeakClusterHeritability ........................ 70 3.3.4ExploratoryFunctionalDataAnalysis ................. 74 3.3.5LigninContentandItsCorrelationswithGrowth .......... 76 3.3.6GeneticCorrelationsbetweenWoodChemistryandOtherImportantTraits ................................... 77 3.4Discussion .................................... 79 3.5Summary .................................... 82 4ASSOCIATIONDESIGNANDSIMULATION .................. 84 4.1Introduction ................................... 84 4.2BayesianHierarchicalModel .......................... 86 4.2.1FamilyBasedLinearModel ....................... 86 4.2.2BayesianApproach ........................... 88 4.3SimulationandValidationStudy ....................... 90 4.3.1SimulationofPopulationandSNPdesign ............... 90 4.3.1.1StructuredPopulation .................... 90 4.3.1.2UnstructuredPopulation ................... 91 4.3.1.3StructuredPopulationwithMissingGenotype ....... 91 4.3.1.4ValidationoftheModelwithCarbonData ......... 92 4.3.1.5PowerCalculation ...................... 92 4.3.2SimulationResults ........................... 92 4.3.2.1SNPDetectionforClonalRepeatabilityof0.2and0.4 .. 92 4.3.3MinorGenotypeFrequencyandPowerCalculation .......... 95 4.3.3.1CarbonRevisit ........................ 97 4.4ChapterSummary ............................... 100 7

PAGE 8

................... 102 5.1Introduction ................................... 102 5.1.1AssociationGenetics .......................... 102 5.1.2ApproachestoDiscoverGenesControlWoodPropertiesinConifers 103 5.1.2.1AssociationMappingApproaches .............. 106 5.2ResearchObjectives ............................... 109 5.3MaterialsandMethods ............................. 110 5.3.1SNPGenotyping ............................ 110 5.3.2StatisticalAnalysis ........................... 111 5.4Results ...................................... 112 5.4.1LigninContent ............................. 112 5.4.2Carbohydrates .............................. 113 5.4.2.1C5SugarandRelatedPeaks ................. 114 5.4.2.2C6SugarsandRelatedPeaks ................ 115 5.4.2.3CombinationofMostHeritableC5andC6Peaks ..... 117 5.4.3ComparingSignicantAssociationsforCarbohydratePeaks ..... 117 5.4.4C20ResinAcidExtractives ....................... 118 5.4.5VelocityStiness ............................ 119 5.5Discussion .................................... 120 5.6ChapterSummary ............................... 123 6CONCLUSION .................................... 124 6.1ContributiontoQuantitativeGeneticsandBreedingforSouthernPine .. 125 6.2ContributiontoAssociationGeneticsinLoblollyPine ............ 127 6.2.1SimulationStudies ........................... 128 6.2.2AssociationMappingofWoodPropertyTraits ............ 129 6.3FutureResearchandConsiderations ...................... 130 APPENDIX AMETROPOLISHASTLINGALGORITHMTOFINDHERITABLEPEAKCLUSTERS ...................................... 132 BWaveletAnalysisCodeinR ............................. 140 CASREMLCodetoEstimateVarianceComponentsofSinglePeak ........ 142 DSNPPrescreeningANOVACodeinR ....................... 143 EBLASTResultsfortheGeneAnnotation ...................... 145 REFERENCES ....................................... 154 BIOGRAPHICALSKETCH ................................ 171 8

PAGE 9

Table page 1-1Primaryeectofwoodpropertiesonnalproductquality ............ 20 2-1Leastsquaremeansforthetraits .......................... 40 2-2Overallnarrow-senseheritabilityestimates ..................... 46 3-1IndividualandacrosssiteclonalrepeatabilityestimatesandtypeBgeneticcorrelationsofsignicantsinglepeakintensities ......................... 67 3-2Acrosssiteheritabilityestimatesandstandarddeviations()andtypeBgeneticcorrelationsfortheC5andC6carbohydratesandtheircorrespondingindividualm/zpeaks ....................................... 68 3-3Acrosssiteheritabilityestimatesandstandarddeviations()andtypeBgeneticcorrelationsforligninanditscorrespondingindividualm/zpeaks ........ 68 3-4Acrosssiteheritabilityestimatesandstandarddeviations()fortheC20diterpenoidextractiveandtheircorrespondingpeaksinpine .................. 69 3-5PhenotypicandtypeAclonalgeneticcorrelationof32signicantsinglepeaks 71 3-6SomesetofpeakclustersfromMetropolis-Hastingsroutine ............ 74 3-7PC1andPC2loadingofmajorpeaksfromPrinciplecomponentanalysis .... 75 3-8Across-siteclonalrepeatabilityandtypeBgeneticcorrelationsofwaveletandPCAcoecients ................................... 77 3-9LSmeansandstandarddeviation()ofseedlingandrooted-cuttinglignincontentofinterandintraprovenancecrossesgrownatCuthbert,GAandNassau,FL .. 77 3-10Correlationsbetweenlignincontentandothertraits ................ 79 3-11CorrelationsbetweenC6andothertraits ...................... 79 3-12Correlationsbetweenresinextractivecontentandothertraits .......... 79 4-1Priorforthemissinggenotype ............................ 92 4-2SummaryofdetectedSNPsunder0.2ofclonalrepeatability ........... 93 4-3PercentageofdetectedSNPsamongallSNPsfordierentadditivevalueintervalsat0.2CR ....................................... 93 4-4PercentageofdetectedSNPsamongallSNPsatdierentadditivevalueintervalsat0.4CR ....................................... 94 9

PAGE 10

....................................... 94 4-6SignicantSNPsfromBAMassociationmodelat95%condenceinterval .... 99 4-7SignicantSNPsfromBAMDassociationmodelat90%condenceinterval ... 101 5-1NumberofSNPsindierentcategoriesdependingonthenumberofminorhomozygotesfoundintheSNPcall ................................. 111 5-2SignicantSNPsforlignincontentat95%Bayesiancondenceinterval;pstdevstandsforphenotypicstandarddeviation ..................... 113 5-3SignicantSNPsforthesumofallassignedC5andC6sugarsat95%Bayesiancondenceinterval;pstdevstandsforphenotypicstandarddeviation ...... 114 5-4SignicantSNPsforsumofallassignedC5sugarpeaks(m/z57+73+85+96+114)at95%Bayesiancondenceinterval;pstdevstandsforphenotypicstandarddeviation ............................................. 114 5-5SignicantSNPsform/z114at95%Bayesiancondenceinterval;pstdevstandsforphenotypicstandarddeviation .......................... 115 5-6SignicantSNPsforthesumofthechemicallyassignedC6sugar ........ 115 5-7SignicantSNPsforheritableC6sugarpeaks ................... 116 5-8SignicantSNPsform/z144at95%Bayesiancondenceinterval;pstdevstandsforphenotypicstandarddeviation .......................... 116 5-9SignicantSNPsforthesumofthemostheritableC5sugar(m/z114)andC6sugar(m/z144)at95%Bayesiancondenceinterval;pstdevstandsforphenotypicstandarddeviation .................................. 117 5-10SummaryofSNPsfoundintwoormoreAssociationTestsforCarbohydrates .. 118 5-11SignicantSNPsfortheresinacidextractivepeakm/z285at95%Bayesiancondenceinterval .................................. 118 5-12SignicantSNPsfortheresinacidextractivepeakm/z300at95%Bayesiancondenceinterval .................................. 119 5-13SignicantSNPsforthesumofC20resinacidextractivepeaksm/z285+300at95%Bayesiancondenceinterval ......................... 119 5-14SignicantSNPsforvelocitystinessat95and90%Bayesiancondenceinterval.Boldindicatessignicantatthe95%level ..................... 120 5-15SignicantSNPsthathavenohitorhaveunknownfunctionfromtheBLAST 121 A-1Clustersof2,3,4,and5peaksthathasthetop30clonalrepeatabilities .... 134 10

PAGE 11

...................................... 137 E-1GenefunctionsforthediscoveredsignicantSNPsfromtheBLAST ....... 146 11

PAGE 12

Figure page 2-1Breedingvaluefamilyrankcorrelations ....................... 47 2-2Simulationofpercentgaininvolumeandstinessfordierentselectionindices 48 3-1Somerepresentativemassspectrometrysamples .................. 64 3-2AcrosssiteclonalrepeatabilityofthewholespectrumofCCLONE ....... 65 3-3Histogramsofclonalrepeatability(CR)ofall421peaks ............. 66 3-4PC1andPC2ofthesamplesfromtwotestsites .................. 76 3-5Histogramofthelignincontentbysiteandseedlingsource ............ 78 4-1DetectedSNPsandtheircorrespondingadditivevalue .............. 95 4-2Minorgenotypefrequencyandadditivevalue ................... 96 4-3Additivevalueanddetectingpower ......................... 97 4-4AutocorrelationoffoursignicantSNPsforcarbonbeforethinning) ....... 99 4-5AutocorrelationoffoursignicantSNPsforcarbonafterthinning) ........ 100 12

PAGE 13

BAMD BayesianAssociationModelwithMissingDataBLUE BestLinearUnbiasedEstimatorBLUP BestLinearUnbiasedPredictorCR DenotesRametClonalRepeatabilityCRmean DiameteratBreastHeightFDR FalseDiscoveryRateLD LinkageDisequilibriumMCMC MarkovChainMonteCarloMFA MicrobrilAngleM-Halgorithm Metropolis-HastingsalgorithmMOE ModulusofElasticityQTDT QuantitativeTransmissionDisequilibriumTestQTL QuantitativeTraitLocipyMBMS PyrolysisMolecularBeamMassSpectrometryPCA PrincipalComponentAanalysisRb SingleNucleotidePolymorphismwco WaveletCoecient 13

PAGE 14

Intheeconomicallyimportantsouthernpinespecies,loblollyandslash,thegeneticarchitectureofwoodchemistryandstinessweredeterminedandassociationgeneticswasusedtoidentifygenesthatpotentiallyregulatetheseeconomicallyimportanttraitsinloblollypine. Thein-treestinessofjuvenilecorewoodinaprogenytrialof139familiesofslashpinePinuselliottiihadmoderatelevelofgeneticcontrolwithaheritability(h2=0:42)whichisgenerallyhigherthangrowthtraits,i.e.,heightanddiameteratbreastheight.Nosignicantgeneticcorrelationswereobservedbetweenvelocitystiness,DBHandvolumegrowth. Woodchemistryinthejuvenilecorewoodofaclonallypropagatedprogenytrialof61familiesofloblollypinePinustaedawasmeasuredbypyrolysismolecularbeammassspectrometry(pyMBMS).Geneticanalysisofall421peaksinthespectraidentied32withsignicantgeneticcontrol.Ofthesepeaks,abouthalfarechemicallyidentied.Thechemicallyidentiedmasstochargepeaksaloneorsummedtogetherforveandsixcarbonsugars,ligninandC20extractivesallhadlowlevelsofgeneticcontrolwithaclonalrepeatabilitiesof0.11-0.16,whicharelowerthangrowthtraits(heightandDBH).Pairwisegeneticcorrelationsbetweenpeaksshowedthatknownligninandsugar 14

PAGE 15

TheperformanceofanewBayesianassociationmethodthatusesaGibbssamplertoimputemissingdata(BAMD)wastestedusingsimulateddataforstructuredandunstructuredpopulationsandrealdatafromastructuredpopulationtoidentifySNPsknowntosignicantlyaectphenotypesthathavevaryingdegreesofgeneticcontrol. AssociationgeneticanalyseswerecompletedwithBAMDtoidentifysignicantSNPsforwoodpropertyphenotypes.Fromasetof2182genotypedSNPsinarichlystructuredclonallypropagatedCCLONESpopulation,87signicantSNPswereidentiedforwoodchemistryandin-treevelocitystiness.In-treewoodvelocitystinessisamoderatelyheritabletrait,and6SNPsweresignicantatthe95%level,butanadditional9wereidentiedatthe90%level. 15

PAGE 16

1.1.1ForestsandSociety 2009 ).Worldwide,theeconomicvalueofallindustrialtimberproductsisestimatedtobe>400billionUSdollars.AlthoughtheUnitedStatesistheworld'smajorproducerofforestproducts,italsoconsumesnearly30%oftheworld'sproductionofindustrialtimberandoverallisanetimporterofwoodandforestproducts( 1993 ). Asthehumanpopulationexpands,forestlandcontinuestobeturnedintoagriculturalelds,townsandcities.Thisdecreaseinforestlandchallengesforesterstosupplymorewoodtokeepupwiththealsoincreasingdemandforforestproducts.Bytheendof2020,itispredictedthatglobaldemandforwoodwillexceeditsavailability,600-800millionm3peryear,( 1996 ; 1996 ).Howcantheforestindustryremaincompetitivewithotherresourcessuchasconcrete,steel,aluminum,plastics,orotherbers;howcanthedemandtosupplytimberbedonesustainably;andhowcanweconserveournaturalforestsforourfuturegenerations?Therearenosimpleanswerstothesecomplexquestions.However,oneapproachtokeepupwiththeincreaseddemandfor 16

PAGE 17

2002 ).WidespreadimplementationofplantationforestryalsoenablesthedevelopmentandapplicationofforestbiotechnologiessuchasrecombinantDNAmethods,planttissueculture,andmolecularmarkersthathavethepotentialtofurtherboostproductivity.Inthepast15years,commercialsuccessofgeneticallyengineeredcropplantssuchassoybean,maizeandcottonhasprovidedexamplesfordevelopmentanddeploymentofgeneticallyengineeredforesttrees( 2007 ).Thus,forestbiotechnologycanhelpmeetthechallengeofdevelopinganddeployinghighyieldinggenotypeswhichinturncouldsupplymorewoodfromlesslandandtherebyreducetheareaofnaturalforestthatneedstobecuttosupplyourcurrentandfutureneedswithrenewable,environmentallyacceptableproducts. 2005 ).However,duringthistimetheyieldsofwoodhaveincreased3-4fold( 2004 ).Thisincreaseinproductivityisduetothedevelopmentandimplementationofplantationforestrypracticesstartingin1952on2millionacresandexpandingtodayto30millionacres,withanadditional25millionacrespredictedinthenearfuture( 2002 ).In1990,southernpineplantationsaccountedforabout75%oftheplantationestateintheUnitedStates.Ontheseplantations,eighthundredmilliontoonebillionloblollyandslashpineseedlingsareplantedannuallyand80%oftheseseedlingscomefromgeneticallyimprovedseed( 2002 ).Becauseofthiswidespread 17

PAGE 18

1996 ; 2000 ). 1984 ; 1985 ; 2007 ).Forexample,itiswellknownthatthewoodandberpropertiesaredramaticallydierentbetweenangiospermandgymnospermspecies.Gymnosperm,orsoftwoods,havehigherproportionsofligninandlowerproportionsofcellulosethanthatofangiosperms,orhardwoods( 1984 ; 1999 ; 2004 ).Inaddition,angiospermshaveberandvesselswhereasgymnospermsonlyhavetracheids.Gymnospermspeciesareespeciallyimportantbecausethewoodisusedprincipallyinconstructionandfortheproductionofpackagingpaperssuchascardboard.Ofthegymnosperms,itistheconiferousspeciesthatarethebeststudied.Forconiferspecies,thewoodandberpropertiesofthestemchangedramaticallyfromthecenter,orpith,totheouterportion,orbark,andsomewhatfrombasetotip.Woodpropertiescanbecharacterizedintwodimensions,juvenilevs.maturewoodfromthebasetothetopofthestemandcorewoodvs.outerwoodfromthepithtothebark( 2004 ).Thebestcharacterizeddierencesinwoodpropertiesoccurat1.4meterabovethesoil,wheretheinnercorewoodformedduringthejuvenilephaseofgrowthandtheouterwoodformedduringthematurephaseofgrowth.Ingeneral,thematureouterwoodmakesbettersolidwoodproductsbecauseithaslesscompressionwoodandknots,highermodulusofelasticityduetolowercellulosemicrobrilangles(MFA),andhighermodulusofruptureduetohigherspecicwooddensity.Inaddition,matureouterwoodmakesbetterpaperproductsduetogreatercellulosecontents,longertracheidsandlowerMFA 18

PAGE 19

2004 ; 2005 ).Therefore,matureouterwoodispreferredforprocessingintosolidwoodandpaperproducts. Withinanormalseasonofsecondarygrowth,conifertreesformearlywoodandlatewood.Earlywood,alsoreferredtoasspringwood,isformedduringthebeginningofthegrowingseasonwhilelatewood,alsoreferredtoassummerwood,isformedduringthelaterpartofthegrowingseason.Earlywoodandlatewooddiermostintheirdensity,withearlywoodtracheidsbeinglargerindiameterandtheirsecondarycellwallsbeinghalfthethicknessoflatewoodtracheids.Inaddition,latewoodhasslightlylongertracheidsandcanhaveslightlylowerMFA.ThepropertiesoflatewoodareimportantforsolidwoodpropertiesbecauselatewoodhasgreaterMOEthenearlywood.Visualgradingoflumberqualityisinpartbasedontheproportionoflatewood. Conifersalsomakeamorespecializedcompressionwoodthatformsinthemainstembeneathwherethebranchesconnect,andalsoontheundersideofbenttrees( 1986 ).Compressionwooddierssubstantiallyfromearlywoodandlatewood,havinghigherdensitybecauseofitsthickercellwalls,roundcells,andthemiddlelamellaeisdegraded.Thechemicalcompositionofcompressionwoodisalsoverydierent,being50%ligninand<35%celluloseandauniquehemicelluloseispresentincompressionwood.TheMFAofcompressionwoodtracheidsishigh,45degrees.Compressionwoodisundesirableforbothsolidwoodandpaperproductionyetdesirableforenergyproduction. Tomeettheincreasingdemandforforestproductsfromfeweracres,forestersareusingmoreintensivemanagementpractices,suchassitepreparation,nutrientmanagement,fertilization,( 1988 )andherbicidetreatment( 1996 ),aswellasfastergrowinggenotypes( 2004 ; 2004 ; 2006 ; 2007 ).Loblollyandslashpinegrownwithintensivemanagementpracticesontreeplantationsreachharvestablesizesatyoungeragesandhencehaveshorterrotationtimes.Currentpracticesreducetheageofthetreesatharvestfrom40+yearsto15-30years,dependingonthesitequality.Amajoreectofgrowingtreesmuch 19

PAGE 20

2002 ; 2007a ).Table 1-1 showstheprimaryeectofwoodpropertiesinrelationtotheirnalproducts( 2007 ).Itneedstobeemphasizedthatwoodpropertiesareactuallycorrelatedalongthetransitionfromjuvenilecorewoodtomatureouterwood. Table1-1. Primaryeectofwoodpropertiesonnalproductquality PropertySolidwoodFiberboardPaperBioenergy WoodDimensionDimensionConversion,ConversionChemistrystabilitystabilityLightstability StinessMOEMOETensileStrength{ WoodMORBonding,Bonding,HydrolysisDensityTensileStrengthTensileStrength MFAMOEDimensionTensileStrength{StabilityDimensionstability 1.1.4.1ModulusofElasticity 2008 ).MOEistheratioofappliedstresstothechangeinshapeofanelasticbody,andisameasureoftheresistanceofamaterialtodeformation.ItisoneofthemostimportantpropertiesinsolidwoodproductassessmentbecausewoodwithahigherMOEinthelongitudinaldirectionisstierandthereforebearsgreaterloadswithoutbuckling,makingitmoredesirableforbuildingconstruction.Individualberstinessisalsoanimportantdeterminantofpaperstrength.Thus,theimportanceofMOEforwoodandpaperproductshasstimulatedresearchintothegenetic,physiologicalandenvironmentalmechanismsthatcontrolthiswoodproperty.MOEincreasesmorethan2foldbetweencorewoodandouterwoodbecausecelluloseMFAdeclines.BothMFAandMOEchangerapidlyduringjuvenileandtransitionphasesbut 20

PAGE 21

2004 ).Anotherindexforwoodstrengthisthemodulusofrupture(MOR).Itisthemaximumberstressatfailure.MORisofgreaterimportanceforcompositematerialsratherthansolidtimbertesting( 1997 ; 2008 ). 1993 ).For20-25yearoldplantationgrownloblollypine,thecross-sectionalareaweightedspecicgravityofearlywoodwasreportedatabout0.33andlatewoodat0.72( 2002 ).Regionalvariationinwoodspecicgravityofplantedloblollypineissignicant.LargevariationinringSGexistswithinthenaturalrangeofloblollypinewithringSGbeinghigherintheSouthAtlanticandGulfregionscomparedwithalltheinlandPiedmontandhillyregions( 2008 ).SGisthemostimportantpredictorofMOR.SGisalsoimportantforMOEbecausetogetherwithMFAtheyaccountfor93%ofthevariationinloblollypine( 1999 ).WoodSGincreasesfromthecorewoodtoouterwood.ThisincreaseinwoodSGisduetotherelativeincreaseintheproportionoflatewoodtoearlywoodformedduringoneseasonofgrowthandtoaincreaseinthedensityofthelatewooditselfintheouterwood( 2006 ).Inconiferspecies,itiswellestablishedthatwoodSGisundermoderatetohighgeneticcontrol( 1977 ; 1984 ; 1991 ; 1994 ; 1998 ; 1999 ; 2001 ; 2003 ; 2007 ).Althoughmydissertationresearchdidnotinvestigatethisimportantwoodproperty,thegeneticarchitectureof 21

PAGE 22

1999 ).Celluloseisthemostabundantcompoundinwoodandbecauseofitshighdegreeofpolymerization(upto20,000),linear1,4-glucosestructureandsemicrystallinestateitisadominantdeterminantofwoodtensilestrength.Hemicellulosesareshortchained(sugarbackbonesofupto250),branchedpolysaccharidesthathavestronganitiesforcelluloseandarecovalentlyboundwithlignin;henceithasbeenproposedthattheymediateimportantcross-linkingfunctionsbetweenligninandcellulose( 1976 ).Thecompositionofhemicellulosesvariesdramaticallybetweenconiferandangiospermtrees( 1999 ).Ligninisacomplexthree-dimensionalphenolicpolymerthatactsasagluetoholdcellulosemicrobrilstogetherandalsowoodbersthemselves( 1999 ; 2005 ). Theproportionofcellulose,hemicelluloseandligninaecttheprocessingeciency,yieldandthequalityofpulp,paperandbioenergyproductsthatcanbeobtainedfromwood.Thehigherlignincontent,shortertracheidlength,andlowerspecicgravityofcorewoodaregenerallyassociatedwithhigherpulpingcostsandlowerpulpyields( 1984 ; 1996 ; 2003 2006 ).Cellulosecontentofawoodcoreextractedat1.4mwasshowntobeaverygoodpredictorofkraftpulpyieldofthewholetreeinEucalyptusnitens,withpotentially86%ofvariationinpulpyieldatKappa18accountedforbythecoresample( 2002 ).ThecorrelationofpulpyieldwithcelluloseandglucosewasgoodforbothEucalyptusglobuluswoodsamplesranginginagesfrom6to32years,and21sevenyearoldE.nitenswoodsamples 22

PAGE 23

1996 ).Paperstrengthdependsontheberlength,cellwallthickness,berchemicalcontent,anddegreeofdamagetocellulose( 2004 ; 2005 ).PulpmechanicalstrengthandespeciallytensilestrengthisshowntobedirectlyproportionaltocellulosecontentandMFA( 1999 ). 1996 ).Mosttraitsarequantitativetraitsthathavecomplexpatternsofinheritance;theyarecontrolledbymanygeneswithvaryingeects-butingeneral,severalgenescanhaverelativelylargeeectbutmostgeneshavesmalleectsonthephenotypeandtheenvironmentcanalsohavelargeeectsonthephenotypesthustherearegenotypebyenvironmentinteractions.Thevastmajorityofecologicallyandeconomicallyimportanttraitsinfoodcropsandforesttrees,suchasgrowth,grainyield,woodmechanicalandchemicalproperties,areinheritedasquantitativetraits. Quantitativegeneticanalysesofwoodchemistryareimportantforbreedingprogramsbecausethewoodchemicalcompositiondramaticallyaectstheprocessingofwoodintopulp,paper,andbioenergy.Currentlyinconifers,onlyafewreportsofthegeneticanalysisofwoodchemistryvariation( 1980 ; 1997 ; 2001 ; 2006 )areavailable.Theliteratureshowsthatwoodchemicalcompositionisunderlowtohighgeneticcontroldependingonthestudy. 1997 )determinedtheclonalrepeatabilityofPinusradiata(radiata)ligninconcentrationinthemiddlelamellaestimatedbyinterferencemicroscopyandfoundthatforthe10clonesstudiedtheclonalrepeatabilitywas0.70,andtheclonalrepeatabilityforglucoseandxylosecontentwas0.38and0.70respectively. 2000 )studiedthewoodextractivecontentofthenorthernScotspinePinussylvestrisinSwedenandfoundthattheheritabilityofpinosylvin,pimaricresinacid,abieticresinacid, 23

PAGE 24

2006 )studiedthewoodchemicalcompositionvariationandgeneticparametersoffourteenfull-sibfamiliesgeneratedbya6-parenthalf-diallelmatingdesignplantedon4sitesandfoundthatacrosssitenarrow-senseheritabilityofindividualtreelignincontentwash2=0.12forjuvenilewoodandh2=0.23fortransitionwood;andfor-cellulosenarrow-senseheritabilityatindividualtreelevelwas0.15forjuvenilewoodand0.09fortransitionwood.Asignicantlimitationofthesestudiesisthatthemagnitudeofthestandarddeviationinheritabilityistypicallysimilartotheheritabilityestimateitself.Thishigherrorinestimationismostlikelyduetothetechnicalchallengesassociatedwithwoodchemistryphenotypingbytraditionalwetchemicalmethodswhichareslowandexpensive.Consequently,thereisnoliteraturedescribingthequantitativegeneticsofwoodchemistrywitharelativelylargenumberoffamiliesandmultiplereplications. Alternativehighthroughputmethodsfordeterminingwoodchemistrysuchasnearinfraredspectroscopy(NIR)andpyrolysismolecularbeammassspectrometry(pyMBMS)arebeingdevelopedforbothsoftwoodandhardwoodspecies( 1999 ; 2002 ; 2004 ),buttheyhaveyettobeappliedtolargegeneticteststodissectthegeneticarchitectureoftheseimportanttraits.pyMBMScanprovidefastanddetailedinformationabouttheplantcellwallchemistryandhasbeenusedtorapidlyidentify,classifyandpredictcellwallchemistryinmanytreespeciesandherbaceousbiomass( 1982 ; 1999 ; 2002 ; 2006 ; 2009 ).ForexampleusingpyMBMS,thewoodchemicalcompositionofgeneticallymodiedPopuluslinesexpressingaBttoxingenedidnotdiersignicantlyfromcontrollines( 2006 ),whileincorn,lignincontentis33-97%higherinBttoxingenemodiedplantsthaninthecontrolplants( 2001 ),indicatinggenesfunctiondierentiallyinwoodyplants. 24

PAGE 25

2009 ). Incontrasttowoodchemistry,thegeneticcontrolofMOEorstinessismuchbetterunderstoodandresultsshowthistraitisundermoderatetohighgeneticcontrolinconiferspecies( 2002 ; 2006 ; 2006 ; 2007a ).MOEandin-treeacousticstinessshowedhighnarrow-senseheritabilityfrom0.46-0.53inanopen-pollinatedradiatapinePinusradiataprogenytestinNewZealand( 2002 ).In-treeacousticstinessforafull-sibslashpineprogenytrialpopulationwith139familiesgrownonsixsiteshadanarrow-senseheritabilityof0.41( 2007a ).Forclearwood,MFAandstinessarehighlycorrelated( 1992 ; 1994 ).Inonetrialcomposedof50open-pollinatedfamiliesandasecondtrialcomprising20control-pollinatedfamiliesinNewSouthWales,Australia,MFA,wooddensityandMOEmeasuredfromwoodcoresusingx-raydiractionanddensitometrywerefoundtohavehighlevelsofgeneticcontrolinthecorewoodandmoderatelevelsinouterwood( 2006 ).MOEforhybridlarchshowedanarrow-senseheritabilityof0.44in19full-siblarchfamilies( 2006 ).Incontrast,individualtreenarrow-senseheritabilityof4thyearearlywoodMFAwasshowntobelowat0.17inloblollypinepopulationgrownattwosites( 2004b ). 1999b ; 2004 ).The 25

PAGE 26

Inadditiontotheuniversity-industrycooperativesdevelopingimprovedmaterialthroughtraditionalbreedingandselection,afewcompaniesareusingclonalselectionandpropagationmethodstoidentifyimprovedvarietiesofloblollypine.Theuseofclonalselectionandmassvegetative/clonalpropagationhasthepotentialtosignicantlyincreasethegainfromagivencycleduetotheapplicationofmorestringentselectioncriteria.Moreoverthemultiplicationandplantingofmanygeneticallyidenticalcopiesofthebestgenotypeshasthepotentialtonotonlyincreasegrowthratesbutalsotoimprovestanduniformityinsizeandgradeleadingtogreaternancialgain( 2009 ). Marker-assistedbreedingandselectionarealsoviablealternativesforsouthernpinebreeding.Inmarker-assistedselection(MAS)thebestprogeniescanbeselectedwithappropriatemarkersorhaplotypesusinghighthroughputgenotypingtechnology.ImplementingMAShasthepossibilityofenhancingselectionspeedandeciency,aswellasdrasticallyreducingthecostsofconventionalbreedingwithmultiplelocationprogenyeldtests.Forexample,inmaize,byapplyingacceleratedmarkerassistedrecurrentselection(MARS),breederscanmakeselectiondecisionsthreetofourtimesperyearinsteadofthetypicalonetotwotimesperyear( 2007 ).Inrice,MAScanbeusedtoeliminateabout40%ofthematerialonaveragefromthebreedingpopulationsandthusextensivelyreduceeldandgreenhouselaborcosts( 2006 ).Inpea,MASinearlygenerationssignicantlyincreasedtheeciencyofthebreedingprocesscomparedwithconventionalselectionintheF3andlatergenerations( 2006 ).TheseexamplesshowthatbycombiningMASandphenotypicinformation,theeciencyoftheselectionissignicantlyimprovedandthecostsofa 26

PAGE 27

2002a ; 2003 ; 2006 ).Inloblollypine,QTLmappingforwoodpropertieshasbeencompletedinathree-generationoutbredpedigreewithfourgrandparentsandtwoparents( 2002a );Inthispedigree,172progenygrownon6siteswerephenotypedandQTLmappingmethodsusedwithanRFLPbasedmaptoidentifynineQTLsforwoodspecicgravity,veforMFAandeightuniqueQTLsassociatedwithcellwallchemistry.Inanotherloblollypinepedigree,twopopulationswereusedforvericationpurpose,onepopulationof457progenyfromarematingoftheparentsoftheQTLpedigreeandanotherunrelatedpopulationconsistingof445progenyfromthebasepedigree;halfoftheQTLweredetectedinmultipleseasonsindicatingmoderatetemporalstability;however,fewQTLwerefoundtobeincommonatthedierentpopulations( 2003 ).Inhybridlarch,68AFLPmarkerswereusedtodetectQTLforwooddensitytraitwitheighthalf-sibfamiliesfromacrossofEuropeanlarch(LarixdeciduaMill.)asfemaleparentand8Japaneselarches(Larixkaempferi)asmaleparents,9markerswereshowntobelinkedtotheringspecicgravity( 2002 ),8markerswereshowntobelinkedwithearlywoodspecicgravityand7markerswereshowntobelinkedwithlatewoodspecicgravity.Fourmarkersco-segregatedwithringSGandearlywoodSG,1marker 27

PAGE 28

2008a b ).Inmaritimepine(Pinuspinaster),186individualsfromanoutbredpedigreeofthecrossoftwoparentswereusedtoidentifyQTLsthatcontrolwoodchemistryandwoodpropertytraits.Inaddition,10candidategenesinvolvedinsecondarycellwallbiosynthesisanddepositionchosenbasedonthefunctionalinformationgatheredfromtreespeciesandmodelplants,-5genes(Korrigan,CesA01,CesA3,PFK,Susy)involvedinpolysaccharidesynthesis,4genes(PAL,C4H,CAD,CCoA-OMT)involvedinlignin,andonecellwallprotein(AGP)weremapped.Ofthese9candidategenes,KorriganmappedinaQTLintervalthataectshemicellulosecontentandbercharacteristics( 2006 ).TheabovestudiesshowthatwoodpropertiesareundersucientgeneticcontroltodetectQTLinawidenumberofconiferspeciesusingtraditionalmethods.Ingeneral,similarnumbersofQTLaredetectedbetweenfamilieswithinaspeciesandevenbetweenspecies(i.e.,loblollyandDouglasr),andthatthesemultiplelocihaverelativelysmalleectsonthephenotypes.TheyalsoshowthattheQTLidentiedarenotconsistentlyfoundwithdierentfamilies.Inaddition,becausethenumberofprogenytestedwasoftensmalltheQTLidentiedarelikelytobeoverestimatesoftheirrealeectonthephenotypes( 1998 ; 2003 ).Finally,asignicantdisadvantageoflinkageanalysisisthatitislowresolutionandinconiferswithlargegenomesizesmakeitnearlyimpossibletoidentifythecausativegene(s)withinthelargeintervals. 28

PAGE 29

1991 ; 2000 ).Inloblollypine,collectionsofESTsenablecomparisonswithmodelplants,providetheabilitytoconductdetailedexpressionstudiesusingmicroarrayexperimentsandprovideasequenceresourceforthediscoveryofpolymorphismsthatcanbeusedinassociationgeneticstudies.Inloblollypine,threelargeESTsequenceeortshavebeenreported,onewithdierentiatingsecondaryxylem( 2003 ),onewithrootsaftervarioustreatments( 2006 ),andonewithsomaticembryos( 2006 ).Intotaltheseeortshavecontributed>300,000ESTsequencesandunigenesetof16,000-33,000dependingonthecontigingalgorithmused.Asmallnumberofmicroarrayprojectshavebeenreported( 2001 ). Athirdapproachtodiscoveringgenesthatcontrolwoodpropertiesinconifersisassociationgeneticmapping.Associationgeneticsisapopulationbasedsurveythatuseshistoricalselectionandrecombinationeventswhichoccurredthroughoutallgenerationsofthepopulationtoidentifysignicantassociationsofamolecularmarkerwithaphenotypictrait( 2003 ; 2004 ; 2006 ).Associationgeneticanalysesarewellsuitedtoconifersingeneralandloblollypineinparticularbecauseoftheirlongevolutionaryhistories,theirlargenaturalpopulations,out-crossingmatingsystem,substantialESTsequenceresources,highnucleotidediversity,andrelativelyshortaverageLDs. 1994 ; 1996 ).Astrengthofthisapproachover 29

PAGE 30

2003 ; 2004 ; 2006 ).Aweaknessofthisapproachisthatitalonecannotestablishcausation.Recently,associationgeneticshasbeenusedsuccessfullyinmaizetoidentifycandidategenesunderlyingsimplyinheritedtraits(endospermcolorandsweetnesstaste),andcandidategeneswithrelativelylargeeectonquantitativecomplextraitswithhigh(oweringtime)andmoderateheritability(starchconcentration)( 2001 ; 2002 ; 2004 ; 2005 ; 2005 ; 2006 ).Inaddition,ithasbeenusedsuccessfullytoidentifycandidategenescontrollingwoodproperties( 2007 ),andwateruseeciency( 2008 ). LinkageDisequilibrium(LD)isthenon-randomassociationofSNPallelesfromdierentlociwithaphenotypictraitandplaysaveryimportantroleingeneticassociationstudies.StudiesofLDandassociationmappinghavebeenreportedinhumans( 1994 ; 2000 ; 2000 ; 2001 ; 2006 )aswellasinplants( 1999 ; 2001 ; 2002 ; 2004 ; 2006 ; 2007 ; 2007 ).ThedetectionofLDmightimplythattherecombinationfractionbetweentwomarkersisrelativelysmallandthustheSNPmarkerislikelytobe,ortobecloselylinked,withthecausativegene.UnlikeanimalsandArabidopsis,whereLDaveragesover150kb,LDinconifersappearsshort,averaging2-3kb,andSNPsshowinggeneticassociationwiththetraitofinterestarelikelytobelocatedincloseproximitytothecausalloci( 2004 ; 2007 ).LDdecaydependsonthematingsystem,eectivepopulationsize,mutation,populationdriftandpopulationbottleneck,etc( 2001 ; 2006 ).InDouglasFir,thenucleotidediversityishighandtheLDdecreasessteadilywithingenes 30

PAGE 31

2005 ).InMaritimepine,rapiddeclinesinLDbetweensiteswithinmostgeneswasobserved( 2003 ).InScotspine,lowLDwasobservedevenbetweencloselylinkedloci( 2002 ).Inloblollypine,nucleotidediversityishigh( 2007 )andLDvariesamonggenesbutonaverager2decaysto0.20within1500bps( 2004 ).ItshouldbenotedthatLDcanexistbetweenunlinkedlociinthestructuredpopulationduetoselection,geneticdrift,migration,mutation( 2005 ). 1993 ; 2003 ).Forquantitativetraits,LDassociationbetweenamarkerandtraitofinterestcanbedoneusingregressionanalysis( 2001 ),structuredassociationtests( 2000 ; 2001 ),andmixedlinearmodels( 2006 )andQTDTtest( 2008 ).LDassociationmappingcanpreciselypositiontheQTLthatcontrolsthetraitandidentifythecausativemutationwhenLDisshort( 2003 ). Numerousmarkersystems(i.e.AFLP,SSR,etc.)havebeenusedinQTLandLDmapping;however,themarkersofchoiceformostassociationgeneticexperimentsaresinglenucleotidepolymorphisms(SNP).SNPsprovidemuchhigherresolutionthanothermarkertypesbecausetheyareeasilyidentiedfromhighthroughputsequencedataandtheyarecloselyspaced.Forexample,inloblollypineaSNPwithincodingregionsoccursonaverageevery50basepairs( 2004 ).Onelimitationofassociationanalyseswithout-crossingplantshavinghighlevelsofnucleotidediversityisthatSNPdiscoveryandgenotypingcostsaretoohightopermitwholegenomeassociations.Thus,mostassociationstudiesintheseorganisms,likeloblollypine,use 31

PAGE 32

2003 ).Inapopulationofloblollypine,signicantassociationsbetweenasmallnumberofSNPlociandwoodpropertyphenotypesweredetectedinaninitialanalysis( 2007 ). Oneconstraintonassociationmappingishiddenpopulationstructure.Populationstructurecanconfoundassociationtestingbycausingspuriouscorrelations,leadingtoanelevatedfalse-positiverate( 1994 ).Thus,itisimportanttodeterminethecorrectpopulationstructureoronlyconductwithinfamilytests,suchasthetransmissiondisequilibriumtest(TDT),whichusesparentswithaectedospring( 1998 ).TDTdetermineswhetheralleleAisassociatedwithdiseasebycomparingthenumberofparentswhotransmitAtotheiraectedospringwiththenumberwhotransmittheotherallele( 1999 ).QuantitativeTransmissionDisequilibriumTest(QTDT)isanapproachbasedonregressionmethodsforassociationtestingofquantitativetraitsoffamily-basedsamples( 1998 ).Whenpopulationstraticationisnotpresent,quantitativetraitlinkagedisequilibriumtestcanbemorepowerfulbyaddingfoundergenotypeinformation( 2005 ).StructuredassociationwasrstproposedbyPritchard.Unlinkedmarkerlocicanbeusedtodetectpopulationstraticationandtoinferpopulationstructureaswellasancestryestimatesofthesampledindividualsusingamodel-basedBayesianclusteringalgorithmSTRUCTURE( 32

PAGE 33

).Ancestryestimatesareassignedtoindividualsinthesubpopulationandtestedforassociationwithinsub-populations( 1999 ; 2001 ).Mixed-modelapproachesaccountingforeithermultiplelevelsofrelatedness( 2006 )orgenome-widedierences( 2007 )inrelatednesswereshowntoperformwellincontrollingTypeIandIIerrorrate,reducingfalsepositiverate,aswellasincreasingstatisticalpower.In 2006 ),anonymousSSRmarkerswereusedtoestimatepopulationstructure(Q)andtherelativekinshipmatrix(K);andthenQandKweretintoamixedmodeltotestformarker-traitassociation.Thisapproachaccountsforrelatednessatboththefamilyandpopulationlevel.In 2007 ),theyappliedtheKandQmatricestotestforassociationbetweenmarkerandoweringtimeofmaizeandsuccessfullyreducedthefalse-positiverateclosetotheexpectedlevels,indicatingthatconfoundingbypopulationstructurewassignicantlyreduced. InaPinusradiatafull-sibfamilystudy,amixedmodelwasusedtocalculatethevariancecomponents,estimateandpredicttheparentalbreedingvalues( 2004 ).Thebreedingvaluesfordiameteratbreastheight(DBH),stemstraightness,andwooddensitywereusedforthemarker-traitassociationswith34SSRmarkers.Nosignicantassociationswerefound,butitispossiblethattheover-representationofthefemaleparentsinthese200full-sibfamiliescouldhavebiasedtheprocessofdetectingassociation,i.e.,thefrequencyoffavorablealleleswasgenerallyhigherinthefemalepopulation( 2004 ).Inaloblollypineassociationstudy( 2007 ),amixedlinearmodelwithapairwisekinshipmatrixamongindividualstoaccountforrelatednesswastfor42singlemarkersandwoodpropertytraits.22nuclearmicrosatelliteswerealsousedtotestforpopulationstructure.Themodel-basedclustering(STRUCTUREsoftware, 2000 ))showedapatterntypicalofunstructuredpopulations.Morerecently,anassociationstudyusingQTDTinvestigatedwaterrelationsmeasuredbythecarbonisotopediscriminationof 33

PAGE 34

2008 ). 2005 ).Thealpha-latticedesignmoreecientlyreducestheerrorvariancethantherandomizedcompleteblockdesignbyincorporatingrowandcolumneectsintothedesignusingacyclicmethodofvarietyconstruction( 1997 ; 2005 ).Twopreviousstudiessuccessfullydissectedthegeneticarchitectureoftherootingofthecuttingsandresistancetofusiformrustandpitchcankerinthesamepopulation( 2005 ; 2005 ). Theobjectivesforthisprojectare: Toaccomplishtheseobjectives,woodpropertytraitsweremeasuredfor3800treesrepresenting999genotypesgrownontwooftheCCLONESsites.Woodchemistry(lignin,C5andC6sugarcontent)wasdeterminedbypyrolysismolecularbeammassspectrum 34

PAGE 35

Themarkerinformationwasobtainedfor2182SNPsfromgenotypingthe999dierentclonallypropagatedlines.Atotalof7216SNPswereselectedforgenotyping.SNPgenotypedatawereobtainedthroughtheIluminaInniumplatform.Geneticanalysiswasthenperformedforassociationtesting.Growth,woodstinessandwoodchemistrytraitswerethentestedforassociationusingtheseSNPmarkers.BayesianMarkovChainMonteCarlomethodswereusedtoperformtheassociationtesting.AGibbssamplerwasusedtoruntheMCMCchainormissingdataimputation.Chapter4focusedontheassociationgeneticsmodelingandsimulation.InChapter5,wetestforgenesthatmaycontrolthesewoodpropertytraitsusingassociationgenetics. 35

PAGE 36

1999a ; 2004 ).Theemphasisoftreeimprovementprogramsonthesetraitsratherthanonwoodpropertieswasbasednotonlyontheireconomicimportance,butalsoontherealitythatmeasuringwoodpropertiesusingtraditionalmethodswastooexpensiveforthelargepopulationsoftreestestedinbreedingprograms(e.g., 2002 )).Becausesoftwoodtreesgrownonintensivelymanagedplantationsareharvestedyounger,theyhavegreaterproportionsofcorewood,producedduringthejuvenilephaseofgrowth,whichhasadramaticallylowermodulusofelasticity(MOE)thantheouterwood,producedduringthematurephaseofgrowth( 1994 ; 1995 ; 2001 ; 2004 ; 2004 ).Thus,astheprimarygoaloftreeimprovementistoimprovewoodyieldstheimportanceofintegratingmeasuresofcorewoodMOEwithgrowthanddiseaseresistanceintobreedingprogramshasincreased( 2000a ). MOE,theratioofappliedstresstothechangeinshapeofanelasticbody,isameasureoftheresistanceofamaterialtodeformation.ItisoneofthemostimportantwoodpropertiesbecausewoodwithahigherMOEinthelongitudinaldirectionisstierandthereforebearsgreaterloadswithoutbuckling,makingitmoredesirableforbuildingconstruction.ThisimportanceofMOEforwoodandpaperproductshasstimulatedresearchintothegenetic,physiologicalandenvironmentalmechanismsthatcontrolthis 2007a ) 36

PAGE 37

1980 )),acoustics(e.g., 1983 ); 2002 )),andx-raydiraction( 2001 ).Theselaboratorymethodsshow,forexample,inEucalyptusspeciesthat>92%ofthevariationinMOEisexplainedbyvariationinwooddensityandcellulosemicrobrilangle(MFA),andthatvariationinMFAaloneaccountsfor>86%ofthevariationinMOE(EvansandIlic2001,YangandEvans2003).ThecorewoodMOEofsoftwoodtreesislowcomparedwithouterwood,becauseofreducedwooddensityandhigherMFA(reviewedby 2004 )).Overallincreasesinwooddensitycanonlyaccountforupto20%oftheincreaseinMOEbetweenthecoreandouterwood( 1994 ; 2003 ; 2004 ).VariationinouterwoodMOEisprimarilyduetovariationinwooddensityasMFAdoesnotvarysubstantially. UsinglaboratorybasedmethodstodirectlymeasureMFAandMOEisstilltoocostlytointegratefullyintomostsoftwoodbreedingprograms.Thus,lessexpensiveandmorerapidmethodsthatmeasureMFAandMOEaredesired,preferablyonesthatcanbeusedwithstandingtreesintheeld 2003 ).Acousticsisonepotentialmethodthathasreceivedsignicantinterest.AcousticmethodshavebeenextensivelytestedandarewellacceptedformeasuringwoodstinessandMOEinthelaboratory(e.g., 2002 ); 2003 ).Severalstudieshaveshownthatindriedwood,MOEmeasuredbytraditionalstaticbendingtestsishighlycorrelatedwithacousticmethods( 1991 ; 1991 ; 1993 ; 1993 ; 2001 ; 2002 ).AcousticshavebeenusedregularlytoevaluateMOEindriedandnishedsolidwoodandveneerproducts 1965 ); 1977 ); 1994 ); 1997 ). Similartostaticbendingtests,severalfactors,includingknotsand(or)compressionwood,woodmoisturecontent,spiralgrain,externalstress,andtreeform,haveimportanteectsonvelocityasameasureofMOE.Moisturecontentaectsthespeedofacoustic-wave 37

PAGE 38

1990 ; 1996 ; 2001 ; 2002 ; 2003 ; 2005 ).Theeectsofwooddensityandmoisturecontentonattenuationaresmallcomparedwithgrainangle( 2002 ).Severalstudiesfoundthatacousticvelocityisaectedbyknotsbecausetheyarerichincompressionwood( 1979 ; 1982 ; 2001 ).Whentheacousticwavestartedfromthevicinityofaknot,thewavecontourwasdramaticallyalteredandvelocitywasreducedaroundtheknot.However,whenthesignalstartedfurtherfromaknot,thewavesmoothlycircumventedtheknot;thus,theaverageeectoftheknotwasaccountedforbutnotexaggerated. 2002 )foundthatroundandstraightlogsproducebettercorrelationsbetweenresonanceMOEandstatic-bendingMOE.Seasonaleectsonthemeasurementofstandingtreesusingultrasonicequipmentwerenotstatisticallysignicant( 1994 ; 2001 ). AsignicantamountofresearchhasalsotestedtheabilityofacousticmethodstomeasurethelongitudinalMOEorstinessinwetlogs( 1997 ; 2000 ; 2000b ; 2002 )andinstandingtrees( 1993 ; 1994 ; 2001 ; 2001 ; 2001 ; 2002 ; 2002 ; 2004 ).Acousticmethodsnowhavebeenusedsuccessfullytosegregatewetlogsintodierentstinessclassespriortoprocessingintosolidwoodandpaperproducts( 2002 ; 2003 ). 2002 )comparedfouracousticinstrumentswithstatic-bendingtestsformeasuringstinessinstandingfour-year-oldradiatapinetreesandfelledlogs.Theyfoundthatacousticmethodsconductedwithgreenwoodandstatic-bendingtestswerehighlycorrelated,andtheFAKKOPacousticinstrumentusedonstandingtreescomparedwellwithotheracousticinstrumentssuchastheHITMANusedwithfelledlogs.Sincemeasurementsinlive 38

PAGE 39

2002 ; 2004 ; 2005 ).Thus,becauseacousticinstrumentsarereasonablylowincostandstandingtreescanbemeasuredrapidlyintheeld,theyholdpromiseforbreederstofullyintegratewoodstinessintotheirtreeimprovementprogramsandselectforfamiliesthatgrowfasterwithbetterstemform,greaterdiseaseresistance,andhigherMOE,particularlyinthecorewood( 2003 ; 2002 ; 2004 ). Currently,onlytwopapershavereportedtheinvestigationoftheutilityofacousticinstrumentsfordeterminingthegeneticcontrolofwoodstiness.InPinusradiateD.Don, 2002 )showedthatthevelocitystinessof12-year-oldprogenyfrom72rst-generation,open-pollinatedfamiliesgrownononesitehadanarrow-senseheritabilityof0.47usingtheHITMANwithfelledlogsand0.46usingtheFAKOPPwithstandingtrees.Theseresultscomparedwellwiththenarrow-senseheritabilitydeterminedwithclearwoodbendingstinessfromthesametrees.Morerecently, 2004 ),usingtheHITMANwithfelledradiatapinelogs,reportedatypeBgeneticcorrelationof0.54acrossonesiteinAustraliaandoneinNewZealandand0.94betweentwositesinNewZealand.Thesetestswithradiatapineshowpromise,butwhethermeasurementofvelocitystinesswithin-treeacousticinstrumentswillbebroadlyusefulfortree-breedingprogramsdependsonitslevelofgeneticcontrolestimatedinlargertrials,thedegreeofgenotypebyenvironmentinteractions,geneticcorrelationswithothertraitsthatareimportantfortreeimprovement,andconrmationinotherspecies. Theobjectiveofthisresearchwastoinvestigatetheutilityofusingin-treeacousticvelocitymeasurementsforestimating: 39

PAGE 40

Thisresearchcanprovidevaluableinformationforthedesignofanddecisionmakinginpinetree-breedingprograms. 2.2.1MaterialandExperimentalDesign 1993 )detailedthebreedingstrategyforthesetrials.Inthesetrialsthenumberofpollenparentsusedinanygivenyearwas20unrelatedslashpinerst-generationselections.Thesamepollenmixwasusedwithallofthefemales;soweassumethefamiliesapproximatehalf-sibfamilies.ThesixsitesmeasuredinthisstudywereasubsetofthePolymixItestingprogramandlocatedintheatwoodsofFloridaandSouthGeorgia(Table 2-1 ).Thesiteswerechosentorepresentenvironmentswhereslashpineisplantedoperationally.Thesesiteswereuniformandwell-managedreceivinggoodsoilpreparation:abroadcastweedcontrolbeforeplantingandspotherbicidesprayinginyearstwoandthree.Fertilizer,561kg/haof10-10-10plusmicroelements,wasappliedinthespringbeforethethirdgrowingseason. Table2-1. LeastsquaremeansandstandarderrorsforV2,height,DBHandvolumefromeachsiteat8-yearsoldforV2and5-yearsoldforgrowthvariables SiteLocationV2(km2=s2)Height(m)DBH(cm)Volume(dm3) 760Yulee,FL6.56(1.72)5.35(0.69)8.76(1.45)17.20(6.49)762Fargo,GA7.53(1.61)6.28(0.58)9.94(1.13)23.97(6.20)763WhiteSprings,FL6.26(1.55)5.28(0.64)8.85(1.39)17.31(6.08)765Munson,FL7.19(1.70)6.43(0.52)10.81(1.01)28.66(6.26)766Perry,FL7.25(1.63)6.31(0.56)11.11(1.33)29.94(7.22)767Lyons,GA5.41(1.37)3.62(0.74)7.05(1.78)9.31(5.06)AVERAGE6.705.559.4221.01 40

PAGE 41

2003 ).Thevolumeofthe5-year-oldtreeswasestimatedindm3usingtheequationof 1983 ). 2004 ; 2005 ),weuseV2torepresentthedynamiclongitudinalMOE. InpreliminarytrialstotestwhetherV2variedsignicantlywithintrees,thefourcardinalpositionsof30treesweremeasured.Whilehighlysignicantdierenceswere 41

PAGE 42

2002 )wasusedtoanalyzethedata.Fortheinitialanalysis,V2ofeight-year-oldtreeswasregressedagainstalltheparametersincludingsite,theve-year-oldheightandDBH,andtheirinteractions.Parametersnotsignicantintheregressionmodelweredeleted.WeanalyzedthedatatodenethephenotypicassociationsandtoidentifytheregressioncoecientsonDBHandheightbyusingDBHandheightascovariates.Thereducedmodelisasfollows: 42

PAGE 43

Weassumedthatthepolymixprogenieswerehalf-sibfamiliesandnarrow-senseheritabilitywascalculatedusingthefollowingequation(FalconerandMackay1996): ^h2=42f TypeBgeneticcorrelationswerecalculatedas: ^rB=2f Itestimatesthestabilityofthegeneticentriesacrosssites,withhighertypeBvaluesofatraitindicatingthatfamilymeansforagiventraitweremorerepeatableacrossthesesixsites. TypeAgeneticcorrelationsbetweentwophenotypictraitsexplainthegeneticcorrelationbetweentwotraitsmeasuredonthesameindividual.BivariateanalysisinASREMLwasusedtocalculatetypeAgeneticcorrelations.Themodelforbivariateanalysis,basedontheunivariatemodel(3),is: whereYdenotesthephenotypefortraitst1,t2,andt3,bdenotesthexedeectsofthemeans,sitesandblockswithinsitesfortraitst1andt2withincidencematrixX,udenotestherandomeectsofparent(MVN(0;GA))(whereGisthetraitgeneticrelationshipmatrixandAisthenumeratorrelationshipmatrix(additivecoecientmatrix)andtest-by-parentinteraction)(uMVN(0;SA)),sinceDBHandvolumeare 43

PAGE 44

G=0BBBB@2f(t1)f(t2;t1)f(t3;t1)f(t1;t2)2f(t2)f(t3;t2)f(t1;t3)f(t2;t3)2f(t3)1CCCCA S=0BBBB@2sf(t1)sf(t2;t1)sf(t3;t1)sf(t1;t2)2sf(t2)sf(t3;t2)sf(t1;t3)sf(t2;t3)2sf(t3)1CCCCA ThetypeAgeneticandenvironmentalcorrelationsbetweentwotraitswascalculatedusingtheequation. ^rA=t1;t2 where2f1isthevariancefortrait1and2f2isthevariancefortrait2.Thenumeratoristhegeneticfamilycovariancebetweenthesetwotraits. 2-1 ).ThesmallesttreeswereattheLyons,Georgiasite,andthelargesttreeswereatthePerryandMunson,Floridasites(Table 2-1 ).ThemeanV2rangedfromalowof5.41inLyons,Georgiatohighsof7.53and7.25km2=s2inPerry,FloridaandFargo,Georgia,respectively.NosignicantdierencewasobservedbetweenthemeangrowthandwoodstinessofthesitesinFloridaversusGeorgia. 44

PAGE 45

2-2 ).Thehighheritabilityofvelocitystinessincorewoodissimilartothatfoundforotherwoodproperties,suchaswooddensity,inanumberofconiferousspecies( 1994 ),andmicrobrilangleandMOEinjuvenileradiatapine( 2006 ).Asexpected,heightandDBHhadstrongfavorablegeneticcorrelationsbetweenthemselvesandwithvolume.V2showedveryweakbutunfavorablegeneticcorrelationswithDBH(-0.12)andvolume(-0.08)(Table 2-2 ).IncontrasttoDBHandvolume,asignicant,favorable,butsmall,geneticcorrelationwasestimatedbetweenV2andheight(0.19)(Table 2-2 ). 2-2 ).TypeAenvironmentalcorrelationswerecalculatedastheresidualvariance.Althoughmanyfactorscontributetotheresidualvariance,suchasmeasurementinaccuraciesorpoormodelselection,thesefactorsareusuallysmallandthuswe,aswellasothers,interpretthatthemajorcomponentofthisresidualvariancecomesfromvariationinthegrowthenvironment( 1971 ; 1982 ).TypeAenvironmentalcorrelationsindicatethatmicrositeswhichpromotedgrowthdidsobystimulatingbothheightandDBH(re0:74)(Table 2-2 ).Incontrast,environmentalcorrelationsbetweengrowthtraitsandV2wereweak.MicrositesthatstimulatedheightgrowthonlyweaklypromotedincreasesinV2;whereas,micrositesthatstimulateddiametergrowthweaklyreducedV2(Table 2-2 ).Geneticbyenvironmentalinteractionsforvelocitystinesscouldnotbedeterminedreliablybecausechangesinrankacrosssitescouldbeinuencedbytherelativelysmallnumberoftreessampledperfamiliesfamilyatthesixsites. 45

PAGE 46

Overallnarrow-senseheritabilityestimates(diagonal,boldprint),typeAenvironmentalcorrelations,(belowdiagonal)andtypeAgeneticcorrelations(abovediagonal)ofV2,DBH,height,andvolumeacrossall6sitesandtheirstandarderrors(parentheses) TraitV2DBHHeightVolume 2-1 ).EventhoughthenegativecorrelationbetweenvolumegainandV2isnotstatisticallysignicant,atradeobetweenthebreedingvalueforvolumeandV2wasobserved(Figure 2-1 ).ThetwofamilieswiththelargestgainsinvolumehavenegativebreedingvaluesforV2,whilethetwofamilieswiththesmallestgainsinvolumeshowpositivebreedingvaluesforV2(Figure 2-1 ).Ifthetraitsaresegregatingindependentlythenone-quarterofthefamilieswouldbeexpectedtohavepositivebreedingvaluesforbothtraits.Aboutone-fthoftheslashpinefamiliestestedhadpositivebreedingvaluesforV2andheight(30of139)andforV2andvolume(29of139).Twenty-onefamilieshadpositivebreedingvaluesforheight,volumeandV2stronglysuggestingindependentsegregationofV2withthesegrowthtraitsinthisslashpinebreedingpopulation. 46

PAGE 47

BreedingvaluefamilyrankcorrelationsofheightandV2(A)withra=0:23andvolumegainandV2(B)withra=0:02 47

PAGE 48

2-2 ).Whentheweightkis1,wherestinesswastotallyneglectedintheselection,thegainforvolumeis22.4%andthegainforstinessisabout0%.Whentheweightkwas>0.7,thepercentgaininvolumereachedaplateau,becausethesamefamilieswereselected.Whentheweightkwas0,indicatingvolumegainwastotallyneglectedintheselection;thegainforstinesswas24%andthegainforvolumewas1.2%.Theseresultsagreedwiththendingsthattherewasnosignicantcorrelationbetweenvolumegainandstinessinourstudy.Whenthevalueofkisbetween0.2and0.5,thepercentagegainforbothtraitsexceeded10%. Simulationofpercentgaininvolumeandstinessfordierentselectionindices.Thesoliddotisforstinessgainandtheopendotisforvolumegain 48

PAGE 49

2-2 )andwassimilartowooddensitymeasuredfromextractedcores( 1994 ; 2002 ).Weobtainedapreciseestimateofheritabilityforin-treevelocitystinessofcorewoodbytestingalargenumberoffamiliesfromaslashpinebreedingpopulation.Theh2estimatesfoundherewithalargeslashpinebreedingpopulationwereverysimilartothosereportedby 2002 )usingtheFAKOPPstress-waveinstrumentwithfewerradiatapinefamiliesandgrownononlytwosites.Thesedataindicatethatthelevelofgeneticcontroloverin-treevelocitystinessissimilarin8-year-oldslashand12-year-oldradiatapinepopulationsmeasured. Interestingly,thenarrow-senseheritabilityofin-treevelocitystinessforcorewoodobservedherewasatthehigherendoftherangeforbroadandnarrow-senseheritabilityofMFAreportedinanumberofsoftwoodspeciesincludingloblollypine( 2004 ; 2004a ; 2006 ).Themostlikelyexplanationforthehighheritabilityobservedforin-treevelocitystinessinslashpineisthechoiceofcambialageformeasurement.Forradiatapine, 2006 )reportedthattheheritabilityofMOEandMFAatbreastheightwashighestforrings4-8andwasloweratotherages.Inloblollypinetheindividualtreenarrownarrow-senseheritabilityofMFAforrings4and5ontwositeswas0.3-0.4( 2004a ).OursandKumar's2003( 2004 )ndingsshowthattheacross-siteheritabilityforin-treestinessissimilartoMFAinclearwoodcoresampleswhencomparedatsimilarcambialages.ThistogetherwiththeknowledgethatMFAisacriticaldeterminantofMOE( 2001 ; 2003 ; 2003 )supportsthehypothesisthatthelevelofgeneticcontroloverin-treevelocitystinessissimilartoMFAmeasuredinwoodcoresfrombreastheight.AnimportantquestionremainsregardingthestrengthofthegeneticcorrelationbetweenMFAandin-treeacousticvelocity. Genesforgrowthandin-treevelocitystinesstendtosegregateindependently.About20%ofthefamilieshavepositivebreedingvaluesforbothgrowthandvelocitystiness, 49

PAGE 50

1985 ; 1995 ).Thus,ourdatasuggestthatmorelatewoodand(or)longertracheidsareformedintallerslashpinetrees. Theacross-sitetypeBgeneticcorrelationsforvelocitystinesswerelower(0.68vs.0.94)than( 2004 )reportedforradiatapinefortwoNewZealandsites.Thissuggeststhatslashpineshowedmoregeneticbysiteinteractionforvelocitystinessthanradiatapine.However,thiscomparisonislimitedbythesmallnumberoftreessampledperfamilypersiteinourstudyandthefactthatonlytwositesweretestedwithradiatapine( 2004 ).Interestingly,theacross-siteTypeBgeneticcorrelationbetweensitesinNewZealandandAustraliawassignicantlylower( 2004 ).Inloblollypine,genotypebyenvironmentinteractionswerereportedforMFA( 2004a ),suggestingthatstinessandMFAofdierentgenotypesvarydependingontheenvironment.Overall,themagnitudeofthegenotypebyenvironmentinteractionsfoundherewassmall,suggestingthateveniftheyarerepeatableandpredictableforspecicenvironments,theadditionalgainsinstinesswhichcouldarisefrommatchingspecicfamiliestocertainsiteswouldalsobesmall. Thehigherh2estimateofin-treevelocitystinessindicateslessenvironmentalinuencethanobservedwithgrowthtraits.Thelackofasignicantgeneticcorrelationbetweenstemdiameterandcorewoodstinessindicatesthattheslightunfavorable 50

PAGE 51

2-1 )butratheraectedbytheenvironment.Thespecicvariable(s)intheenvironmentthatcausedthisweaknegativecorrelationareunknown.Slowgrowingtreeshavesmallerannualringsandbecausethein-treevelocitystinessismeasuredoveraxeddistancefromthebark,thestinessmeasurementincludesringsthatareyoungerincambialagecomparedwithfastergrowingtrees.Because,insoftwoods,ringsclosertothepiththataroseatayoungercambialageareknowntobelesssti,itisexpectedthatsiteswithslowergrowingtreesofthesameagewouldonaveragehavelowerstiness( 1994 ).Forthisexplanationtoholdtheearlywoodtolatewoodratiomustalsobesimilar.Alternatively,comparisonofMFAandtracheidlengthatthesamecambialagesshowedthatfast-growingNorwaysprucetreesfromdierentsitestendtohavehigherMFAs( 2004 )andshortertracheids( 1998 ).WoodwithhigherMFAinthetracheidshaslowerstiness. Becauseofitssizeandclearwoodcontent,thebottomlogofthetreeisconsideredthemostvaluableeventhoughthelesssticorewoodoccupiesagreatervolumethantheotherlogsharvestedfromsoftwoodtrees( 1985 ; 2004 ; 2004 ).TheresultspresentedhereandbyKumarshowthatin-treemeasurementsofcorewoodstinesscanbeusedtoselectforimprovedcorewoodstinessinthislogtherebyenhancingitsvalue.Itisunclearhowselectingforimprovedcorewoodstinessinthebottomlogwillaectcorewoodstinessinlogsfromhigherupinthetree.However,limitedresearchsuggestscorewoodstinessinlogshigherupthetreewillalsoincrease,althoughthishasnotbeenaddressedsuciently( 2004 )andPeter,unpublisheddata). 51

PAGE 52

52

PAGE 53

1984 ; 1999 ; 2008 ).Celluloseisthemostabundantcompoundinwoodandbecauseofitshighdegreeofpolymerization(upto20,000),linearbeta-D-1,4linkedglucosechainandsemi-crystallinestateitisadominantdeterminantofwoodtensilestrength( 1964 ).Hemicelluloseshaverelativelyshortsugarbackbonesofupto250andarebranchedpolysaccharideswithstronganitiesforcelluloseandarecovalentlyboundwithlignin;henceithasbeenproposedthattheymediateimportantcross-linkingfunctionsbetweenligninandcellulose( 1976 ).Thehemicellulosecompositionofwoodfromangiospermtreesispredominantly4-o-methyl-glucuronoxylan,whilegymnospermwoodcontainsbothgalactoglucomannanandarabinoglucuronoxylan( 1999 ).Ligninisacomplexthree-dimensionalphenolicpolymerthatholdscellulosemicrobrilstogetherandalsowoodbersthemselves( 1999 ; 2005 ).Gymnospermscontainprincipallyguaiacylligninwhileangiospermssynthesizebothguaiacylandsyringyllignin( 2005 ; 2004 ). Theproportionofcellulose,hemicelluloseandligninaecttheprocessingeciency,yieldandthequalityofpulp,paperandbioenergyproductsthatcanbeobtainedfromwood.Woodwithhigherligninandlowercellulosecontentsincreasechemicalpulpingcostsandlowerpulpyields( 1984 ; 1996 ; 2003 2006 ; 2007 ),andwillnegativelyaecttheeciencyandeconomicsofpretreatmentaswellassaccharicationduringbioconversiontoethanol( 53

PAGE 54

; 2008 ; 2008 ; 2008 ; 2009 ).Incontrast,forthermalconversionbygasication,pyrolysisorcombustion,woodwithhigherligninandlowercarbohydratecontentswouldbepreferred.Thus,aneconomicopportunityexiststogeneticallyalterwoodchemicalcompositiontoenhancetheyieldandeciencyofconversiontopulp,paperandbiofuels. Inordertoincorporatewoodchemistrytraitsintobreedingprograms,itisrstnecessarytounderstandtheirgeneticarchitectureandgeneticcorrelationswithotherimportanttraits,suchasgrowth,modulusofelasticity,anddiseaseresistance.Currentlyinconifers,onlyafewreportsofthegeneticanalysisofwoodchemistryvariation( 1980 ; 1997 ; 2001 ; 2006 )areavailableandtheheritabilityestimatesrangefromlowtohighdependingonthechemicalcompoundandthepopulationstudied.Forexample, 1997 )determinedtheclonalrepeatabilityofPinusradiata(radiata)ligninconcentrationinthemiddlelamellaestimatedbyinterferencemicroscopyandfoundthatforthe10clonesstudiedtheclonalrepeatabilitywas0.70,andtheclonalrepeatabilityforglucoseandxylosecontentwas0.38and0.70respectively. 2000 )studiedtheheartwoodextractivecontentofthenorthernScotspinePinussylvestrisandfoundthattheheritabilityofpinosylvin,pimaricresinacid,abieticresinacid,fattyacidandsterolcontentswere0.59,0.55,0.59,0.33-0.37,0.62respectivelywith160treesinafull-sibprogenytest. 2006 )studiedvariationinwoodchemicalcompositionandgeneticparametersoffourteenfull-sibfamiliesgeneratedbya6-parenthalf-diallelmatingdesignplantedon4sitesandfoundthatacrosssitenarrowsenseheritabilityofindividualtreelignincontentwash2=0.12forjuvenilewoodandh2=0.23fortransitionwood;andfor-cellulosenarrowsenseheritabilityatindividualtreelevelwas0.15forjuvenilewoodand0.09fortransitionwood.Asignicantlimitationofallofthesepublishedstudiesisthatthemagnitudeofthestandarddeviationinheritabilityistypicallysimilartotheheritabilityestimateitself.Thishigherrorintheestimateismostlikelyduetothetechnicalchallengesassociated 54

PAGE 55

Alternativehighthroughputmethodsfordeterminingwoodchemistrysuchasnearinfraredspectroscopy(NIR)( 2004 ; 2002 ; 2005 ),Fouriertransforminfraredspectroscopy( 1998 )andpyrolysismolecularbeammassspectrometry(pyMBMS)( 1999 ; 2008 )arebeingdevelopedforbothgymnospermandangiospermspecies,buttheyareonlynowbeingappliedtolargegeneticteststodissectthegeneticarchitectureoftheseimportanttraits.pyMBMSprovidesdirectanddetailedinformationabouttheplantcellwallchemistryandhasbeenusedtorapidlyidentify,classifyandpredictcellwallchemistryinanumberoftreeandherbaceousspecies( 1982 ; 1999 ; 2002 ; 2006 ; 2009 ).ForexampleusingpyMBMS,thewoodchemicalcompositionofgeneticallymodiedPopuluslinesexpressingaBttoxingenedidnotdiersignicantlyfromcontrollines( 2006 ),whileinmaize,lignincontentis33-97%higherinBttoxingenemodiedplantsthaninthecontrolplants( 2001 ),indicatingthisgenesactsdierentinmaizethanPopulusplants.Morerecently,pyMBMSwasusedsuccessfullytodeterminethegeneticcontrolofwoodchemistryinalargenumberofprogenyfromonefamilyofPopulus( 2009 ). InpyMBMS,asmallamountofgroundwood(20mgdrymass)ispyrolyzedandthegaseousmoleculesarerapidlydilutedtominimizesecondary,gasphaseinteractions,andaredirectedinamolecularbeamintoamassspectrometer( 1999 2006 ).Themassspectrometerseparatesandquantiesthesmallmolecularweightionizedmoleculeswithmasstocharge(m/z)ratiosfrom30to450beingusedforanalyses.Currently,only32ofthe421peaks(8%)havebeenchemicallyidentied;fourfor5carbonsugars(C5),vefor6carbonsugars(C6),twofrombothC5andC6sugars,eightforguaiacyllignin, 55

PAGE 56

1999 ; 2001 ; 2006 ; 2006 ).Areliableestimateofthepercentlignincontentinwoodcanbeobtainedfromsummingallligninderivedpeaksandnormalizingtothesumoftheligninpeaksofwoodwithknownlignincontent( 1999 ; 2008 ).AlthoughpyMBMSprovideshigh-throughputwoodchemistrydata,ourunderstandingofthespectraandhowtoutilizethemarestilllimited.SofaranalyseswithpyMBMShaveonlyusedthesumoftheassignedpeaksderivedfromlignin,C5orC6sugarsforgeneticandphenotypicanalyses( 2002b ; 2009 ).Nothoroughanalysisofanalyticalapproachestoecientlyusemoreofthespectrumorcapturemostofthevariationinthespectrahasbeenreportedinpines.TowardsthisgoalofusingthewholepyMBMSspectrainquantitativegeneticanalyses,wehavetestedthreedierentapproaches,includingsinglepeakheritabilityandpairwisegeneticcorrelations,heritabilityofparameterestimatesfromfunctionaldataanalysis,andMetropolis-Hastingsalgorithmtosearchcombinationsof2,3,4and5peaksthatincreasetheheritabilityofthecluster. Functionaldataanalysissummarizespatternsinthepeaksasafunctionratherthanregardingeachindividualpeakindependently,therebytransformingorreducingthedimensionalityofthedatatoidentifyimportantpatternsandvariation( 1998 ).Waveletdecompositionandprincipalcomponentanalysis(PCA)aretwomethodsthatcanbeusedtoreducehighdimensionalmassspectra( 2009 ).Waveletanalysisdecomposeshighlydimensionaldata.Therequirementisthattheoriginalfunctionaldataneedstohave2mdatapointsandtherearemlevelsofcoecientthatyoucanextract(level=0,1,2,...,m1)( 2009 ).Asofthewritingofthisthesisnoliteratureisfoundthatestimatesthegeneticheritabilityofcoecientsfromthewaveletdecomposition.Alternatively,PCAisanexploratorymultivariatestatisticalmethodtosimplifycomplexmulti-dimensionaldatasets( 1994 ).Ithasbeenusedinanimalbreedingtoinvestigatepre-breedingandreproductivetraits( 1977 56

PAGE 57

).Inapple,PCAwasusedtodeterminethegeneticheritabilityoffruitshapetraits;therstprinciplecomponentshapetraitaccountedfor76.8%ofthetotalphenotypicvariationandshowedanacross-siteheritabilityof0.79( 2000 ).ReportedheretheresultsofwaveletandPCAanalysesarecomparedandusedtoimproveourunderstandingofthegeneticcontrolofwoodchemicalcompositionandpyMBMSasamethod. 57

PAGE 58

2005 ).ThisstudyiscalledComparingClonalLinesonExperimentalSites(CCLONES). 1999 2006 ; 2008 ).Briey,4-5mgofgroundsamplewereloadedintoapyrolysistubeandplacedintotraysfortheautosampler.Sampleswerepyrolyzedinareactorconsistingofaquartztubewithheliumowingthroughat5lmin1(atSTP).Thesampleswereheatedinaheliumatmosphereat550?Candthepyrolysisgasintroducedintoamolecularbeammassspectrometer.Massspectrawerecollectedandthepeakintensitiesofthemasstochargeratiosfrom30to450werenormalizedtototalioncurrentpriortoanalysis( 2006 ; 2008 ).Atypicalrunhad48samples,inwhich4samplesarestandardsfromknownspecies.Inourexperiment,thestandardbiomasssamplesareradiatapine(8493),poplar(8492),wheatstraw(8494),andsugarcanebagasse(8491)fromNationalInstituteofStandardsandTechnology(http://www.nist.gov).These4samplesareput 58

PAGE 59

2006 ).IncontrasttopastpyMBMSexperiments,notechnicalreplicationwasusedsincetechnicalreplicationof40randomlychosenpinesamplesmatchedverywell,andthisagreedwithapreviouslargegeneticexperimentwithPopuluswoodwhichshowednosignicantdierencebetweenthetechnicalreplications( 2009 ). 3.2.4GeneticModel whereyijklmnorepresentsthepeakfortheithtest,jthrep,kthincompleteblock,lthpropaguletype,mthandnthparentofothgenotype Thexedeectsareshownbelow 59

PAGE 60

Theclonalrepeatabilityestimatesforthecuttingpropaguletypel=1iscalculatedasfollows: Thenarrowsenseheritabilityforthecuttingiscalculatedasfollow: Thenarrowsenseheritabilityfortheseedlingsourcepropaguletypel=2iscalculatedasfollow: Theonlydierencebetweenthecomponentstobeincludedtocalculatethenarrowsenseheritabilityoftheseedlingsandcuttingsisthatnorandomeectcomponentsfor'clone'anditshigherterminteractionswerenotincluded.ThetypeBgeneticcorrelation 60

PAGE 61

Inordertounderstandthegeneticcorrelationbetweenpercentlignincontentandotherimportanttraitssuchasgrowth,crowndimension,andwoodproperties,abivariateanalysiswasperformedbetweenlignincontentandallothertraitswiththesametermslistedinthemodelabove.Geneticcorrelationsbetweentwophenotypictraitsexplainthegeneticcorrelationbetweentwotraitsmeasuredonthesameramet.BivariateanalysisinASREMLwasusedtoestimatetypeAgeneticcorrelations.Themodelforbivariateanalysis,basedontheunivariatemodel(3.2.1),is: whereYdenotesthephenotypefortraitst1andt2,bdenotesthexedeectsofthemeans,testsites,blockswithinsitesandreplicationwithinblocksfortraitst1andt2withincidencematrixX,udenotestherandomeectsofparent(MVN(0;GA))(whereGisthetraitgeneticrelationshipmatrixandAisthenumeratorrelationshipmatrix(additivecoecientmatrix)andtest-by-parentinteraction)(uMVN(0;SA))dened,G=0B@2f(t1)f(t1;t2)f(t1;t2)2f(t2)1CAand S=0B@2sf(t1)sf(t1;t2)sf(t1;t2)2sf(t2)1CAandwithincidencematrixZ,andeistheresidualfortraitst1andt2,eMVN(0;EIn)withE=0B@2e(t1)e(t1;t2)e(t1;t2)2e(t2)1CA. ThetypeAgeneticbetweentwotraitswascalculatedusingtheequation. 61

PAGE 62

Letbeadistributiononthatwewishtoestimate.Weassumethathasadensity.LetQ(:;:)beatransitionprobabilityfunctiononwithrespectto,suchthatforeveryx,Q(x;:)hasadensityq(x;:)withrespectto.ThisQ(:;:)willnotnecessarilyhaveasastationarydistribution.Q(:;:)wasusedtogetacandidatepoint,butweacceptthecandidateonlywithacertainprobability.Thatprobabilityissetupsothattheresultingchaindoeshaveasastationarydistribution.Qisarbitraryandq(x;y)>0ifandonlyifq(y;x)>0. q(x,y)canbesymmetricornot.theacceptancefunction(x;y)forx,y2,theacceptancefunction(x;y)is: Ifthechainiscurrentlyinxwithpeakcluster(a,b,c,d),acandidateywillbechosenwithpeakcluster(a,b,c,e)withprobability(x;y)thisnewclusterywillbeaccepted;andtheremainingprobabilitythatthechainwillstayintheoldclusterx. ForWaveletanalysis,peaks50-305wereselectedbecausetheymettherequirementofthedatadimensionof2m,inwhichm=8.Therst3levels(level=0,1and2)of 62

PAGE 63

2009 ).ThepyMBMSdatasetisveryspikyandthustheDaubechiesorthonormalcompactlysupportedwaveletwasusedbecauseitissuitableforextremalphasedataset( 1988 ).Boundariesweretreatedassymmetricsuchthatafunctionbeyonditsboundariesisassumedtobeasymmetricreectionofthefunctionintheboundary. 3.3.1Heritabilityandtype-Bgeneticcorrelationofidentiedpeaksandmajorcomponents 1987 ):C5sugarshave5characteristicmasstochargeratio(m/z)peaksandC6sugarshave6characteristicpeaksofwhichtwo(m/z57,73)areincommon.Twentyoftheotheridentiedpeaksarederivedfromligninand3areC20componentsoftheextractive. Figure 3-1 showstypicalmassspectraoffourtreesinourpopulation.Interestingly,tree40254hasarelativelyhighintensityatpeak300,whichcorrespondstodihydroabeiticacid,aC20extractivecomponent. PreviousgeneticanalyseswithpyMBMSdatahavesummedtheintensitiesofthechemicallyidentiedpeaksderivedfromC5orC6sugarsand/orligninpriortoanalysis( 2002a ; 2009 ).However,focusingonthe8%ofknownpeaksdoesnotutilize92%ofthespectrumwhichmaycontainpeaksthatcontributesignicantlytovariationinwoodchemicalcomposition.Thus,ageneticanalysiswiththefullmodelwasdoneindependentlyforall421singlepeaksforeachsiteandacrossbothsites(Figure 3-2 andTable 3-1 ).Forthesinglesites,128peaksatCuthbert,GAand102peaksatNassau,FLweresignicantlyheritablewiththemajorityofhavingCRs<0:1.ThehighestsinglesiteCRatCuthbert,GAandNassau,FL,were0.23 63

PAGE 64

SomerepresentativemassspectrafromtheCCLONEpopulation (m/z182)and0.24(m/z285),respectively.Only32peakswerefoundtohavesignicantacross-siteheritability(Table 3-1 ).Asexpected,theacrosssiteclonalrepeatabilitiesaretypicallybetweenorbelowthosefromthesinglesites.Thesubstantiallyfewerpeakswithsignicantheritabilityacrosssitesisexpectedbecauseoftheoveralllowgeneticcontrolcoupledwithlowsamplesize,consequently,apparentlylargeenvironmentaleectonwoodchemicaltraits.Thenarrow-senseheritabilityestimatesfortheclonesandseedlingsweresimilar,althoughthestandarddeviationissubstantiallygreaterfortheseedlingsduetothesmallernumbersampled.Interestingly,ofthe32peakswithsignicantacrosssiteheritability,14havebeenchemicallyidentied:5-6derivedfromlignin,2fromC5andC6sugars,4fromC6sugars,1fromC5sugars,and2fromC20extractive.Forthe 64

PAGE 65

3-1 ),and10othershavem/z>200,whichmakesthemlargerthanalloftheknownpeaksderivedfromlignocellulosiccompounds. Figure 3-2 showsthefrequencyoftheclonalrepeatabilityforallpeaks.Mostofthepeaksarenotgeneticallycontrolled. AcrosssiteclonalrepeatabilityofthewholespectrumoftheCCLONEloblollypinepopulation Table 3-2 showstheclonalrepeatability,narrow-senseheritabilityoftherootedcuttingsandseedlings,andtypeBgeneticcorrelationsofthenineidentiedC5andC6sugarpeaksandtheirsums.FortheC5sugarpeaks,onlym/z73and114areheritableandthereforethesepeakintensitiesmustdeterminethesignicantclonalrepeatability 65

PAGE 66

Histogramsofclonalrepeatabilityofall421peaks (0.11)forthesum(57+73+85+96+114)ofC5sugarpeaks.Interestingly,m/z114istheonlyuniquepeaktotheC5sugarsas57and73comefrombothC5andC6sugarsandderivesfromxylaninPopulus( 2008 ; 2009 ).Thenarrow-senseheritabilityofm/z114isgreaterthanthesumoftheC5peaksbecausetheotherknownC5peaksarenotheritable.IncontrasttotheC5sugars,5ofthe6C6peakshavesignicantgeneticcontrol(Table 3-2 ).Theclonalrepeatabilityforthesumof(57+60+73+98+126+144)C6sugarpeaksis0.14andthenarrow-senseheritabilityfortheclonesis0.17,whichisgreaterthananyofthesinglepeaks.Overall,summingallheritableknownC5andC6sugarpeaksgivesthehighestheritability0.16;however,thisisnotstatisticallygreaterthanthesumoftheknownC6peaks(Table 3-2 ). 66

PAGE 67

IndividualandacrosssiteclonalrepeatabilityestimatesandtypeBgeneticcorrelationsofsignicantsinglepeakintensities m/zCRCutCRNasCRbothRbassigned 600.15(0.05)0.065(0.04)0.074(0.04)0.77(0.14)C5,C6sugar730.134(0.04)0.086(0.04)0.068(0.04)0.80(0.15)C5,C6sugar980.077(0.04)0.056(0.03)0.065(0.02)0.95(0.12)C6sugar1140.18(0.04)0.14(0.04)0.108(0.02)0.77(0.14)C5sugar1260.082(0.04)0.121(0.04)0.1(0.02)0.88(0.20)C6sugar1440.086(0.05)0.073(0.04)0.116(0.02)0.94(0.07)C6sugar 1240.093(0.04)0.073(0.04)0.074(0.03)0.67(0.20)Guaiacol1380.116(0.04)0.132(0.04)0.101(0.02)0.99(0.01)Methylguaiacol1500.085(0.04)0.12(0.04)0.055(0.02)0.69(0.37)Vinylguaiacol1640.102(0.04)0.062(0.04)0.05(0.02)0.99(0.01)Guaiacol1800.116(0.05)0.083(0.04)0.113(0.03)0.84(0.15)Syringylethene1820.237(0.04)0.118(0.04)0.149(0.02)0.85(0.13)Syringaldehyde2850.068(0.04)0.199(0.04)0.134(0.02)0.91(0.10)Dehydroabieticacid3000.093(0.03)0.247(0.04)0.116(0.02)0.96(0.09)Dehydroabieticacid 310.098(0.03)0.062(0.03)0.078(0.02)0.99(0.01)-430.109(0.03)0.11(0.03)0.095(0.02)0.99(0.01)-440.085(0.03)0.109(0.03)0.053(0.02)0.99(0.01)-1160.075(0.04)0.037(0.04)0.051(0.02)0.99(0.01)-1280.112(0.04)0.105(0.04)0.106(0.02)0.83(0.14)-1450.073(0.04)0.053(0.04)0.067(0.02)0.99(0.18)-1810.088(0.04)0.054(0.04)0.074(0.02)0.55(0.24)-2390.062(0.04)0.153(0.04)0.054(0.02)0.92(0.13)-2560.125(0.04)0.016(0.04)0.060(0.02)0.01(0.14)-2640.072(0.04)0.075(0.04)0.075(0.02)0.93(0.19)-2960.088(0.04)0.06(0.04)0.072(0.02)0.92(0.33)-2980.112(0.03)0.161(0.04)0.111(0.02)0.8(0.25)-3010.102(0.04)0.108(0.04)0.081(0.02)0.64(0.25)-3160.063(0.04)0.107(0.04)0.078(0.02)0.99(0.01)-3180.07(0.04)0.028(0.04)0.05(0.02)0.99(0.01)-3300.042(0.04)0.086(0.04)0.055(0.02)0.68(0.27)-3420.042(0.04)0.043(0.04)0.052(0.02)0.99(0.01)-3440.102(0.04)0.142(0.04)0.105(0.03)0.83(0.20)Ofthe20peaksidentiedascomponentsoflignin6aresignicantlyheritable(Table 3-3 ).Ofthe5heritableG-ligninpeaks,onlym/z138hadaclonalrepeatability>0.1.Unexpectedly,m/z182hasthehighestheritabilityofallpeaksinpine(Table 3-1 ).InPopulus,m/z182hasbeenidentiedassyringaldehyde 2008 ); 2009 );however,becausegymnospermscannotsynthesizeS-ligninthisisnotlikely 67

PAGE 68

Acrosssiteheritabilityestimatesandstandarddeviations()andtypeBgeneticcorrelationsfortheC5andC6carbohydratesandtheircorrespondingindividualm/zpeaks 570.03(0.02)0.03(0.01)0.00(0.00)-850.02(0.02)0.01(0.01)0.00(0.00)-960.01(0.02)0.00(0.00)0.00(0.00)-1140.11(0.03)0.17(0.06)0.14(0.14)0.77(0.14)C50.11(0.03)0.11(0.04)0.09(0.09)0.71(0.17)57+73+85+96+114600.07(0.02)0.12(0.05)0.22(0.11)0.77(0.14)730.07(0.02)0.12(0.04)0.17(0.10)0.80(0.15)980.07(0.02)0.10(0.03)0.03(0.07)0.95(0.12)1260.10(0.02)0.07(0.03)0.09(0.07)0.88(0.20)1440.12(0.03)0.15(0.05)0.15(0.09)0.94(0.07)C60.14(0.03)0.17(0.05)0.16(0.10)0.87(0.10)57+60+73+98+126+144C5C60.15(0.03)0.16(0.05)0.20(0.12)0.83(0.12)57+60+73+85+96+98+114+126+144C5C6heri0.16(0.03)0.18(0.05)0.25(0.13)0.87(0.11)60+73+98+114+126+1441141440.16(0.03)0.19(0.06)0.20(0.15)0.98(0.38)114+144 Table3-3. Acrosssiteheritabilityestimatesandstandarddeviations()andtypeBgeneticcorrelationsforligninanditscorrespondingindividualm/zpeaks 1240.09(0.02)0.07(0.03)0.11(0.06)0.67(0.20)guaiacol1370.04(0.02)0.04(0.02)0.08(0.05)0.76(0.32)ethylguaiacol,homovanillin,coniferylalcohol1380.12(0.02)0.06(0.03)0.09(0.05)0.86(0.38)methylguaiacol1500.09(0.02)0.03(0.02)0.05(0.07)0.69(0.37)vinylguaiacol1520.03(0.02)0.01(0.02)0.06(0.06)-4-ethylguaiacol,vanillin1640.05(0.02)0.04(0.02)0.07(0.07)-allyl-+propenylguaiacol1780.01(0.01)0.02(0.01)0.03(0.05)-coniferylaldehyde1820.15(0.02)0.14(0.05)0.05(0.07)0.85(0.13)syringaldehydes-lignin0.10(0.02)0.11(0.04)0.01(0.04)0.99(0.01)154+167+168+182+194+208+210g-lignin0.09(0.02)0.05(0.03)0.17(0.10)0.55(0.25)124+137+138+150+164+178lignin0.13(0.02)0.08(0.08)0.16(0.09)0.64(0.18)S+G+120+180+181 3-4 showstheheritabilityestimatesfortheindividualandsumsofC20diterpenoidpeaks.Thedihydroabeiticacid(m/z285,300)peaksandtheirsumsallhavesignicantheritability.Interestingly,theseC20diterpeneextractivecompoundshavesimilarlevelsofgeneticcontrolasthemainlignocellulosiccomponents.ThisistherstreportofusingpyMBMStoshowtheheritabilityofextractive.However,itshouldbenotedthatbecausethewoodcoreswerestoredat4oCforuptoaweekpriorto 68

PAGE 69

Table3-4. Acrosssiteheritabilityestimatesandstandarddeviations()fortheC20diterpenoidextractiveandtheircorrespondingpeaksinpine 2850.13(0.02)0.12(0.04)0.09(0.06)0.91(0.09)Dihydroabeiticacid3000.12(0.02)0.09(0.03)0.11(0.06)0.96(0.09)Dihydroabeiticacid3020.01(0.01)0.01(0.01)0.02(0.02)-AbeiticacidDihydroabeiticacid0.14(0.02)0.11(0.04)0.11(0.06)0.95(0.08)285+300Diterpenoids0.08(0.01)0.06(0.02)0.08(0.05)-285+300+302 3-1 ).Interestingly,typeBcorrelationsare>0.9forC6sugars,m/z98,126,and144;whereas,theyare<0.75form/z73and60,whichhavecomponentsfrombothC5andC6sugars.ThetypeBgeneticcorrelationsforC20extractivepeakswereverystrong,suggestinglimitedgenotypebyenvironmentinteractions.IncontrasttothehightypeBgeneticcorrelationsfoundforC6carbohydrateandC20extractive,theligninpeakshaverelativelylowtypeBgeneticcorrelationstypically<0.7forthesinglepeaksand0.64fortotalnormalizedlignincontent.Thisindicatessubstantialgenotypebyenvironmentinteractionoccursforlignin. 3-5 ).Insomecasesthebivariateanalysisdidnotconvergebecausethevariancecomponentatthefamilylevelwasverysmallandinthesecasesthefamilyvariancewasnotincludedinthemodeltogetconvergence.Becauseonlyhalfoftheheritablepeakshavebeenchemicallyidentied,stronggeneticcorrelationsbetweenknownandunknownpeakshavethepotentialtoassistinthechemicalidenticationoftheunknownpeaks.Asexpected,strongnegativegenotypicandphenotypiccorrelationsbetweenmostofthecarbohydrateandligninpeaksaswellasstrongpositivegenotypic 69

PAGE 70

70

PAGE 71

PhenotypicandtypeAclonalgeneticcorrelationof32signicantsinglepeaks

PAGE 72

3-5 .Continued m/z23926428529629830030131631833034234460-0.02.-0.2-0.57-0.43-0.27-0.34-0.38..-0.2-0.473-0.54-0.35-0.24-0.52-0.48-0.45-0.5-0.42..-0.44-0.398-0.16-0.63-0.51-0.77-0.67-0.41-0.73-0.62-0.47-0.990.32-0.51140.03-0.52-0.4-0.18-0.47-0.54-0.58-0.31-0.54-0.860.02-0.02116..-0.14.-0.48-0.35.....-0.42124-0.080.170.1.0.230.250.190.35..0.110.79126..-0.26-0.69-0.39-0.27-0.4-0.24..-0.14-0.55128-0.24-0.51-0.35-0.62-0.59-0.47-0.44-0.37-0.57-0.55-0.2-0.171380.2.-0.070.340.11-0.16-0.15..-0.1-0.210.51440.06-0.54-0.18-0.56-0.55-0.37-0.48-0.16..-0.09-0.39145-0.2-0.7-0.31-0.04-0.45-0.45-0.63-0.27..-0.580.29150-0.140.130.010.810.090.010.160.080.02-0.07-0.040.51640.4-0.040.290.650.310.130.44.-0.02.0.120.81800.01.0.130.60.420.220.280.39.0.450.530.431810.59.0.340.610.190.420.510.58....182-0.48.-0.09-0.63-0.18-0.11-0.33..0.71..239{0.62.0.990.99....0.11-0.540.422640.18{.0.640.350.670.550.630.76.00.22850.700.18{0.830.9.....-0.450.172960.340.230.39{.0.8......2980.620.210.690.57{0.85...-0.04-0.230.353000.680.320.790.530.79{..-0.420.183010.570.220.660.410.640.72{..-0.330.253160.580.220.650.520.720.770.62{.-0.06-0.240.253180.340.240.370.430.500.560.420.52{...3300.130.250.130.380.350.340.200.310.35{-0.420.18342-0.020.20-0.030.210.150.140.060.150.200.41{.344-0.040.13-0.020.250.110.090.040.140.210.200.14{

PAGE 73

Afterthechainwasrun50,000times,thetop30peakpairsandclustersfromtheclusterofkoutofrunareretained.SincetheRprogramcannotutilizethefamilypedigreeinformationforthemixedmodel,thepeakclusterswereexportedandrunagaininASRemltogetamorepreciseestimateofthevariancecomponentsandcorrespondingclonalrepeatabilityforeachpeakpairandcluster.Thetop30resultsfor2,3,4,and5groupsarereportedinappendixA.Table A-1 showstheresultthatincludesallthepeaksintheMetropolis-Hastingsroutine.Initially,peaks43and44dominatedtheclusterswiththehighestclonalrepeatability.Onereasonispresumablythattheyhavearatherhighintensitycomparedwithmanyotherpeaks.Afterexcludingpeaks43and44,theMetropolis-Hastingsalgorithmwasrunagainandtheresultsareshowninappendixtable A-2 .Itisnoteworthythattheclonalrepeatabilityofthetop30peaksissimilar,0.1-0.18,astheindividualpeaks. Someinterestingpeakclustersaresummarizedintable 3-6 .Forpairsofpeaks,theheritabilityofpeak60increaseswhencombinedwitheitherm/z73,114,126,or144.Similarly,form/z114and126thatclusterwithpeak73,botharecarbohydratepeaks.Itisstrikingthatalltheseknowncarbohydratepeaksrankedinthetop30pairsafter50,000permutations.Theheritabilityofm/z180increaseswhencombinedwithanotherpeakfromG-lignin,m/z150,orwhencombinedwith300and302,C20extractivepeaks.However,itismorediculttoexplainwhytheC5peakm/z114clusterswithaligninpeakm/z138.Forclustersof3peaks,theMetropolis-Hastingsalgorithmfoundmanygroupsofpeaksfromthesamechemicallyidentiedclass.Forexample,theclusterofpeaks73,114,144withaclonalrepeatabilityof0.16areallidentiedascarbohydrates.Similarly,forclustersof4peaks,cluster(73,98,114,126)and(60,114,126,144)have 73

PAGE 74

Table3-6. SomesetofpeakclustersfromMetropolis-Hastingsroutine VarianceestimateHeritabilityRbfamfemalecloneres CRh2cuttingh2seedling 114,138 0.0010760.00270.00580.0643 0.160.140.41 0.8483114,144 0.000840.004670.006310.082 0.160.190.45 0.866673,114 00.006550.009880.1238 0.150.170.39 0.790060,126 0.001160.01030.018660.2445 0.140.140.41 0.843560,114 0.003250.012250.019820.2815 0.140.150.39 0.772373,126 00.004330.007020.1019 0.130.140.34 0.85360,144 0.003250.011810.015590.2727 0.130.150.32 0.8070124,180 0.003960.007910.014370.2394 0.120.110.33 0.802160,73 0.002670.014980.020460.3625 0.120.140.29 0.7944150,180 0.002030.006560.01510.2096 0.120.110.39 0.8737180,302 0.001770.004350.011730.2183 0.090.070.37 0.9013 73,114,144 0.000090.005680.007210.0955 0.160.190.39 0.8502124,180,300 0.002250.005270.009590.1269 0.150.140.38 0.880860,114,138 0.002010.006680.007820.1391 0.140.160.26 0.827360,98,144 0.001160.008250.010.1728 0.140.160.31 0.838660,98,114 0.000990.008440.012140.1768 0.140.160.38 0.813160,73,144 0.001940.010370.013490.2223 0.140.160.32 0.820560,73,126 0.000630.009350.014830.202 0.140.150.38 0.841060,73,114 0.001690.010650.016020.2258 0.140.160.37 0.79360,73,98 0.000530.009890.01280.2161 0.130.150.31 0.8281 60,114,126,144 0.001070.006750.010670.1306 0.160.170.46 0.854073,98,114,126 00.003830.006010.0745 0.150.170.43 0.827960,98,126,128 00.005230.0090.1082 0.150.160.45 0.860260,73,128,144 0.001270.007060.009980.1468 0.140.160.37 0.8157 73,114,126,128,144 00.003550.005220.0621 0.160.190.48 0.850960,73,116,126,182 0.000280.003410.006620.0781 0.150.140.44 0.835860,73,117,126,144 0.000460.004650.006790.0942 0.150.160.38 0.856560,73,98,126,139 00.004790.007060.095 0.140.170.37 0.866460,73,98,144,211 0.000390.005150.006490.1026 0.140.170.33 0.8473 74

PAGE 75

3-7 showsthePC1andPC2loadingsofthemajorpeakslistedinorderofimportance.Forexample,peak180hasthehighestnegativePC1loadingandtheimportanceofthepeaksdecreasesfromlefttoright;andpeak60hasthehighestpositivePC1loadingandtheimportanceofthepeaksincreasesfromrighttoleft.InthepositivePC1loadings,all5peaksarefromcarbohydrates.InthenegativePC1,5areknownG-ligninpeaks,2areunknownandoneisaC20extractive. Table3-7. PC1andPC2loadingofmajorpeaksfromPrinciplecomponentanalysis PCloadingLoadingsignContributingPeaks PC1Negative180124150164137138110300Positive144114735760 PC2Negative571373025512415010969107Positive1281271631451389868180115126114144 Figure 3-4 showsPC1andPC2ofthesamplesfromthetwotestsites.GenerallysamplesfromthelowligninNassau,FLsitehavehigherPC1valuesandclusterontheright. WaveletanalysisisanalternativedatareductiontechniquethatisbettersuitedtomassspectrathanPCA,becausethisapproachcanbettertspikydatawithlargedierencesinintensitiesinneighboringpeaks,suchasistypicalofmassspectra.ThepyMBMSspectraweremodeledusingthewaveletmethodof( 2009 )andthecoecientofthewaveletisthevectorofsetsofwaveletcoecientsatdierentresolutionlevels(level=0,1,2,...,m-1).Weusedlevel0,1,and2inouranalysis. Table 3-8 showstheacrosssiteclonalrepeatabilityandtypeBgeneticcorrelationestimatesforwaveletandPCAcoecients.Forthewaveletcoecients,rstlevel(wco0),secondlevel(wco11andwco12)andpartofthirdlevel(wco21)showrelativelyhighclonalrepeatabilitywhicharecomparabletothesingleandsumsoftheheritablepeaks.ForPCAcomponents,onlyPCA1hasasignicantclonalrepeatabilityof0.15andmoderately 75

PAGE 76

PC1andPC2ofthesamplesfromtwotestingsites hightypeBgeneticcorrelation;therestoftheprinciplecomponentsarenotheritable,indicatingdicultyinusingthemforselectioninabreedingprogram. 2006 ).Lignincontentseemstobenormallydistributed(Figure 3-5 );nosignicantdierencewasobservedfortheAtlanticCoastalandFloridaprovenances,orbetweenrootedcuttingsandseedlingsTable 3-9 .However,thelignincontentintheCCLONESpopulationvariedsignicantlybetweenthedierentsites,withanaverageatCuthbert,GAof30%and28%atNassau,FL. 76

PAGE 77

Across-siteclonalrepeatabilityandtypeBgeneticcorrelationsandtheirstandarddeviation()ofwaveletandPCAcoecients; CoecientcrtypeB wco00.10(0.02)0.73(0.14)wco110.08(0.02)0.72(0.14)wco120.09(0.02)0.99(0.01)wco210.10(0.02)0.78(0.12)wco220.04(0.02)0.53(0.61)wco230.07(0.02)0.60(0.28)wco240.03(0.02)0.99(0.01) pca10.15(0.03)0.78(0.12)pca20.02(0.01)0.35(0.26)pca30.00(0.00)0.00(0.00)pca40.04(0.02)0.69(0.29)pca50.01(0.02)0.00(0.00) Table3-9. LSmeansandstandarddeviation()ofseedlingandrooted-cuttinglignincontentofinterandintraprovenancecrossesgrownatCuthbert,GAandNassau,FL CuthbertCuttings30.25(0.21)30.62(0.19)31.06(0.22)CuthbertSeedlings30.58(0.24)30.34(0.21)30.91(0.25)NassauCuttings27.87(0.21)28.21(0.19)28.14(0.22)NassauSeedlings28.06(0.26)27.87(0.22)27.99(0.26) 3-10 ,Table 3-11 andTable 3-12 showgeneticcorrelationsbetweenlignincontent,C6sugarandC20extractivepeakintensitiesandgrowthatage4(DBH04,HT04),crownwidthatage2(CW02),chemicalcomposition(C5,C6),andage5in-treewoodvelocitystinesstraits.Asignicant,stronglynegativegeneticcorrelationbetweenC5,C6sugarsandlignincontentwasobservedatalllevelsasexpected,becauseascarbohydratecontentsincrease,lignincontentmustdecrease(Table 3-10 ).Thisnegativegeneticcorrelationwasalsoobservedpreviously 2006 ).Additionally,asexpectedC6andC5sugarsarestrongly,positivelycorrelatedatalllevels(Table 3-11 ).Interestingly,therewasaweak,positivegeneticcorrelationbetweenlignincontentwithDBH04andHT04andtherewasaweak,negativegeneticcorrelationbetweenlignin 77

PAGE 78

Histogramofthelignincontentbysiteandseedlingsource contentandC6sugarrespectivelywithDBH04andHT04.Interestingly,lignincontentandcrownwidtharemoderately,positivelygeneticallycorrelatedandC6sugarsmorestrongly,negativelycorrelatedwithcrownwidth.Althoughitisgenerallytruethatwidercrownsindicatelargerbranches,andlargerbranchesproducemorecompressionwoodinthestemwhichhasgreaterlignincontentasaresultofthehighlyligniedsecondarywall( 1971 ; 2001 ),itisnotclearthatthisexplainsthecorrelation,asthecoreswerenotcollectednearnodes.Aweak,positivecorrelationbetweenC6sugarandvelocitystinesswasobserved,butstinesswasuncorrelatedwithlignincontent.C20extractivecontentisnotwellcorrelatedwithgrowth,crownwidthorC5,C6andlignincontent.However,itisinterestingthathigherC20extractiveare 78

PAGE 79

Table3-10. Variouslevelsofthegeneticandphenotypiccorrelationsandtheirstandarddeviation()betweenlignincontentandgrowthtraits;height(HT),diameteratbreastheight(DBH),andwoodchemistrytraits;5carbonsugars(C5),6carbonsugars(C6),crownwidthacrossbed(CWac)atyear2(02);andin-treevelocitystinessatyear5 Parental0.57(0.30)0.67(0.29)0.66(0.24)0:87(0.10)0:83(0.10)0:15(0.31)Clone(fam)0.24(0.12)0.37(0.11)0.22(0.11)0:59(0.12)0:65(0.10)0:08(0.09)Family0.56(0.25)0.53(0.24)0.57(0.19)0:81(0.11)0:78(0.10)0:07(0.26)Clone0.37(0.12)0.43(0.11)0.36(0.11)0:69(0.08)0:72(0.07)0:08(0.11)Phenotype0.01(0.03)0.08(0.03)0:04(0.03)-0.59(0.01)0:66(0.01)0:08(0.03) VariouslevelsofthegeneticandphenotypiccorrelationsbetweenC6andgrowthtraits;height(HT)anddiameteratbreastheight(DBH)atyear4;and5carbonsugars(C5),lignincontent(lignin),crownwidthacrossbed(CWac)atyear2(02);andin-treevelocitystinessatyear5 Parental-0.43(0.23)-0.54(0.22)-0.77(0.17)0:98(0.03)0:83(0.10)0:23(0.25)Clone(fam)-0.14(0.14)-0.50(0.13)-0.39(0.12)0:83(0.06)0:65(0.10)0:22(0.10)Family-0.48(0.21)-0.52(0.20)-0.67(0.15)0:98(0.03)0:78(0.10)0:21(0.23)Clone-0.31(0.13)-0.50(0.11)-0.52(0.10)0:91(0.03)0:72(0.07)0:21(0.11)Phenotype-0.02(0.03)-0.13(0.03)0:13(0.03)0.84(0.01)0:66(0.01)0:11(0.03) VariouslevelsofthegeneticandphenotypiccorrelationsbetweenC20extractiveandgrowthtraits;height(HT),diameteratbreastheight(DBH),6carbonsugars(C6),lignincontent(lignin),crownwidthacrossbed(CWac)atyear2(02);andin-treevelocitystinessatyear5 Parental0.50(0.24)0.45(0.25)0.45(0.24)0:43(0.21)0:16(0.30)0:25(0.27)Clone(fam)-0.01(0.12)0.32(0.11)0.28(0.11)0:44(0.15)0:07(0.16)0:35(0.10)Family0.35(0.21)0.36(0.21)0.39(0.19)0:40(0.19)0:12(0.25)0:22(0.23)Clone0.15(0.12)0.33(0.11)0.33(0.10)0:42(0.12)0:09(0.14)0:30(0.10)Phenotype-0.01(0.03)-0.10(0.03)0:11(0.03)-0.35(0.02)0:04(0.02)0:09(0.03) 79

PAGE 80

2006 ).TheseresultsalsocomparewellwiththoserecentlyreportedinPopulususingpyMBMSwheretotalligninhadaclonalrepeatabilityof0.150.04( 2009 )comparedwith0.130.02forpine,andclonalrepeatabilitiesofthesummedC5andsummedC6sugarpeakswere0.11and0.14,respectivelyinpine;whileinPopulustheywere0.16and0.17,respectively.Overalltheheritabilityestimatesofthelignocellulosiccomponentswerequitesimilarinthetwospecies,evenconsideringthePopulusstudywasfromasinglefamilyderivedfromapseudobackcrossbetweenaP.trichocarpaXP.deltoidesF1hybridandaP.deltoidesgrowninthegreenhouse,whileourstudyused61familiesofeldgrownloblollypinetestedontwositeslocatedindierentregionsofitsnaturalrange.TypeBgeneticcorrelationsforlignin(0.64)andC5sugar(0.71)indicatesubstantialdierencesinrankbetweenthetwositessuggestingsignicantgenotypebyenvironmentinteractions.ThisagreeswithpreviousresultsinPinusradiataD.Donwhichshowedthatlignicationissensitivetoenvironmentalconditionssuchastemperature( 1991 )andwithresultsfromwetchemicalanalysesofloblollypine( 2006 ).Incontrast,typeBgeneticcorrelationsforC6sugars(0.87)andC20extractive(0.95)werestronger,indicatinglimitedgenotypebyenvironmentinteractions. Thebestmethodforanalyzingthepy-MBMSdataforgeneticanalysisofwoodchemicalcompositionhasnotbeendetermined.Becausetheindividualpeakscorrespondtochemicalsofuniqueidentity,achallengeistoidentifysingleorcombinationsofm/zintensitiesthatbestrepresenttheactualamountofthepolymericcellulose,hemicelluloses,andligninaswellasextractiveinthewood.Additionally,forgeneticvariationtobedetected,adierenceinpeakintensitybetweengenotypesneedstobegreaterthanvariationamongstclonallyreplicatedcopiesofeachgenotypeformostalloftheprogenyinafamily.Thus,threedierentapproaches:1)singlepeakgeneticanalysisandpairwise 80

PAGE 81

Singlepeakanalysesshowedthatonly32(8%)ofthepeaksweresignicantlyheritable.Thus,for92%ofthepeaksthevariationwasverysmallorthevariationamongclonalcopiesofthegenotypeswasmuchgreaterthanbetweendierentgeneticentriessuchthatsignicantdierencesinprogenywerenotdetected.Theheritabilityofthesinglepeakswasverysimilartothoseobtainedaftersummingthechemicallyidentiedpeakswithinthesameclass.Thesumofonlytheheritablepeakswithinaclassgavethehighestestimatesofgeneticcontrolalthoughthissumwasnotstatisticallydierentthanthemostheritablepeak.Interestingly,only14ofthe32heritablepeakshavebeenchemicallyidentied,suggestingtheneedtochemicallyidentifytheadditional18heritablepeaks.Forthese18unknownpeaks,strongpairwisegeneticcorrelationsbetweencarbohydrateandligninpeaksorextractivepeakspredicttheclassidentityforabouthalfofthem:2carbohydrate(m/z116,128),3lignin(m/z181,330,344)and3extractive(m/z239,296,298).Thesinglemostheritablepeakwasm/z182butitschemicalidentityisunknown.Unlikethe8peakswithstrongcorrelationswithmultipleotherpeaks,m/z182wasonlymoderatelygeneticallycorrelatedonlywithm/z330andhadapatternofweak,positivecorrelationsconsistentwithitbeingacarbohydratepeak. AMCMCsamplingsearchalgorithmtondclustersofpeaksthathaverelativelyhighergeneticheritabilitywasdevelopedandusedtoconrmmeaningfulpeakclusterswithknownfunctionalgroups.Thismethodwasparticularlysuccessfulatidentifyingclustersofknowncarbohydratepeaks.Inaddition,theclusteringofheritablepeaksofunknownchemicalidentitywasconsistentwiththepairwisegeneticcorrelations.Interestingly,onlyinclustersof5peaksdidthesearchidentifypeakswithnoacrosssiteheritabilitythatwhenaddedtooneswithsignicantheritabilityincreasedtheheritabilityofthepeaktotal. 81

PAGE 82

Thewoodchemistryofloblollypineshowedarelativelylowlevelofgeneticcontrol,indicatingalargeportionofthevariationinthenaturalpopulationiscausedbyenvironmentandthatalteringwoodchemicalcontentthroughtraditionalbreedingandselectionwillbedicult.Strongpairwisegeneticcorrelationswerefoundbetweeneightchemicallyunidentiedpeaksandpeakswithknownchemicalidentityprovidingselectpeaksforfuturechemicalidentication.ItwasalsoshownthatwaveletandPCAcoecientswereheritableatsimilarlevelsaspeakintensities,althoughthewaveletmethodseparatesmoreofthevariancethanthePCA.Finally,anovelM-Halgorithmwasdevelopedandimplementedtoidentifypairsandclustersofuptovepeaksthatwhencombinedincreasedtheheritability.Theresultsofthisrandomsearchwerehighly 82

PAGE 83

83

PAGE 84

1994 ; 1996 ).AstrengthofthisapproachovertypicalQTLstudiesisthatittakesadvantageofthehistoricalselection,mutationandrecombinationeventswhichoccurredthroughoutthenaturalhistoryofthepopulationtodramaticallyimprovetheresolutionofthemapping( 2003 ; 2004 ; 2006 ).Aweaknessofthisapproachisthatitalonecannotestablishcausation.Recently,associationgeneticshasbeenusedsuccessfullyinmaizetoidentifycandidategenesunderlyingMendeliantraits(endospermcolorandsweetnesstaste),andcandidategeneswithrelativelylargeeectonquantitativecomplextraitswithhigh(oweringtime)andmoderateheritability(starchconcentration)( 2001 ; 2002 ; 2004 ; 2005 ; 2005 ; 2006 ).Inadditioninloblollypine,ithasbeenusedsuccessfullytoidentifycandidategenescontrollingwoodproperties( 2007 ),andwateruseeciency( 2008 ). Associationmappingtestssignicantdierencesinphenotypebetweengenotypesinthepopulationthatcarrythedominantandrecessivehomozygousallelesinthepopulation.Animportantassumptionofthisapproachisthatthepopulationisunstructured.Populationstructurecanseriouslyconfoundassociationtestingbycausingspuriouscorrelations,leadingtoanelevatedfalse-positiverate( 1994 ).Thus,itisimportanttodeterminethecorrectpopulationstructureoronlyconductwithinfamilytestssuchasthetransmissiondisequilibriumtest(TDT),whichusesparentswithaectedospring( 1998 ).TDTdetermines 84

PAGE 85

1999 ).QuantitativeTransmissionDisequilibriumTest(QTDT)isanapproachbasedonregressionmethodsforassociationtestingofquantitativetraitsoffamily-basedsamples( 1998 ).Whenpopulationstraticationisnotpresent,thequantitativetraitlinkagedisequilibriumtestcanbemademorepowerfulbyaddingfoundergenotypeinformation( 2005 ).StructuredassociationwasrstproposedbyPritchardetal.,whereunlinkedmarkerlocicanbeusedtodetectpopulationstraticationandtoinferpopulationstructureaswellasancestryestimatesofthesampledindividualsusingamodel-basedBayesianclusteringalgorithmSTRUCTURE( 2000 ).Ancestryestimatesareassignedtoindividualsinthesubpopulationandtestedforassociationwithinsub-populations( 1999 ; 2001 ).Mixed-modelapproachesaccountingforeithermultiplelevelsofrelatedness( 2006 )orgenome-widedierences( 2007 )inrelatednesswereshowntoperformwellincontrollingTypeIandIIerrorrate,reducingfalsepositiverate,aswellasincreasingstatisticalpower.In 2006 ),anonymousSSRmarkerswereusedtoestimatepopulationstructure(Q)andtherelativekinshipmatrix(K);andthenQandKweretintoamixedmodeltotestformarker-traitassociation.Thisapproachaccountsforrelatednessatboththefamilyandpopulationlevel. 2007 ),appliedtheKandQmatricesinassociationtestsbetweenmarkerandoweringtimeofmaizetosuccessfullyreducethefalse-positiverateclosetoexpectedlevels,indicatingthatconfoundingbypopulationstructurewassignicantlyreduced. Despitetheseadvances,twomajorstatisticalproblemsstillexistwhenperformingtrait/SNPassociations.Therstproblemiscreatedbyhavingmanythousandsofmarkersandmanyfewerobservationalunits,genotypes,thiscausesamulti-colinearityproblemwheretheSNPeectsarenotmathematicallyindependent(thecontrastof 85

PAGE 86

2008 )developedanewmethodcalledBayesianAssociationwithMissingData(BAMD)inherPhDThesisandVikneswaranGopalprogrammeditintheRpackage.Weareapplyingitforassociationgeneticsinstructuredandunstructuredpopulationsinsouthernpines. 4.2.1FamilyBasedLinearModel Thefollowingmodelaccountsforthecontributionsofthefamilystructureandpotentialmarkersforacertaintrait. Y=X+Z+ whereY=(Y1;Y2;:::;Yn)isan1randomvectorsofclonalleastsquaremeansforthephenotypictrait, 86

PAGE 87

(4{2) LetG11,G22,andG12bethegenotypicvalueforthethreegenotypes(twohomozygotesandtheheterzygote).Theregressioncoecientisthusdenedasbj=G11-G22fortheadditiveeect.NowdeneiastheithSNPandjasthejthgenotypeclass,thenthefollowingdeneshowtocalculatetheadditiveeectoftheithSNP:xij=1;foronehomozygote0;forheterzygote1;foranotherhomozygote Althoughtheentiremarkersetcanbeincorporatedintoamixedlinearmodelcontext,thisintroducesasignicantcomputationproblem.First,theincidentmatrixwillhavemissingvaluesduetothemissingmarkergenotypes.Second,nearbymarkersmightbehighlycorrelatedandcauseamulticolinearityproblemforthelinearmodel.Insuchasituation,eitheramodelcanbeselectedthatdoesnottthedatawellbutthatgivesanswersthatareeasytoobtain,e.g.areinclosedform,oramodelcanbeselectedthatisappropriateforthedatabutiscomputationallyproblematic.Inpractice,thelargenumbersofSNPsarerstscreenedforeectusingAnalysisofVariance.TheSNPsarethenrankedinverselybythep-valuesfromtheANOVAwithinclusionoftherst400SNPsintheassociationtests. 87

PAGE 88

(4{3) Bayesianapproachescreateposteriordistributionsofandasafunctionoftheobserveddata,whichiscalculatedastheproductofthelikelihoodf(yj)andpriorinformationforeachparameterf().Itrequirespriorinformationforeachparameterbasedonpreviousknowledge.Ifinsucientinformationisavailablefortheprior,anuninformativepriorcouldbeused.TheBayesianapproachintegratesoverallvaluesofeachparameterbyusingMarkovChainMonteCarlo(MCMC)searchmethods. Bayesianmodelstreatalltheparametersasiftheyareallrandomlysampledfromacertaindistribution.Inourmodel,fortheadditiveeect,eachbjisassumedtobesampledfromanormaldistributionwithzeromeanand2j;theirdistributionarecalledthepriordistributions.GiventheSNPmarkerandthephenotypicdata,Bayesiananalysisinfersthedistributionoftheseparametersfromtheobserveddatabycalculatingtheconditionaldistributionoftheparametersgivenbytheobserveddata,calledtheposteriordistribution.TheBayesianapproachoftenimplementsthisprocessbyaniterativeMarkovChainMonteCarlo(MCMC). Weproposedtousethefollowingpriordistribution:p()/1,p(20)/1=20,p(bj)/N(0;2j),p(2j)/1=2j,forjin1;2;:::;s.Thejointpriordistributionoftheseparameterstakestheproductofthepriorsofeachindividualparameter.Letb=bjandu=2j,thenthelikelihoodfunctionis: 88

PAGE 89

220nXi=1(yisXj=1(xijbj))2g TherearetwokindsofMarkovchainsthataretypicallyused,onebeingtheGibbssampler,theotherbeingtheMetropolisalgorithm.AGibbssamplerisaspecialcaseoftheMetropolisalgorithm,anditisapopularMarkovchainMonteCarloroutinetogeneraterandomvariatesfromdistributionsthatarediculttosample.Randomvariablesaregeneratedfromthemarginaldistributionindirectlyforcomplexmodels( 1992 ; 2008 ). Supposeisadistributiononaspace=1234,andsupposeweknowtheconditionaldistributionX(i)jfX(j);j6=ig,oratleastweareabletogenerateobservationsfromtheseconditionaldistributions.WechooseastartingpointX0=(X(1)0;X(2)0;X(3)0;X(4)0)anditerateaccordingtothefollowingscheme. IfXm=(m;m;2m;2m)iscurrentstate,nextstateXm+1=(m+1;m+1;2m+1;2m+1)oftheMCMCisformedasfollows: step1.Generatem+1fromj(X(j)jj6=1)(;m;2m;2m), step2.Generatem+1fromj(X(j)jj6=2)(m+1;;2m;2m), step3.Generate2m+1fromj(X(j)jj6=2)(m+1;2m+1;;2m), step4.Generate2m+1fromj(X(j)jj6=2)(m+1;2m+1;2m);) IntheMCMC-implementedBayesiananalysis,theparameterissampledfromtheabovejointposteriordistribution.Itisdesirabletotesthowwelltheapproachdetects 89

PAGE 90

4.3.1SimulationofPopulationandSNPdesign 2009 )andtotestthepowerofthemodelunderdierentheritabilitylevelswithdierentpopulationstructure.Theexperimentaldesignforthesimulationstudyiscomposedofthefollowing. Four-hundredindependentSNPlociweresimulatedforeachrun.Two-hundredSNPshadanadditiveeectdrawnfromauniformdistributionontheintervalof0to0:3andtheremaining200hadnoeect.The200SNPshaveaneectaccountingfor>90%ofthegeneticvariability,andallSNPshadallelefrequency'sbetween0:2and0:8inthepopulationfromwhichtheparentsweredrawn.AlllociwerethenrunsimultaneouslywiththeBAMDprogram.A95%BayesiancondenceintervalwasusedtodeclaresignicanceoftheSNPeect,i.e.ifthe(2:5%;97:5%)intervaldoesnotincludezero,thentheSNPwasdetectedassignicant. 90

PAGE 91

FormissingSNPs,simulationdataweregeneratedwithonly10SNPs.BasedonmixedmodelcontrastswithnomissingdataSNPs1,3,5,and7weresignicantatthe=0.05level,SNPs2,4,6,8,10hadnoeect(generatedasnulls)andSNP9wasmarginallysignicant.LevelsofmissingSNPgenotypeswerecomparedbyrandomlydeleting0,2,5,and10%oftheSNPgenotypesacrossallSNPs.InformativepriorswereusedforthemissingSNPgenotypesbasedonParentalinformation.Table 4-1 showedthepriorforthemissinggenotypesunderMendelianrandomization.Forexample,ifboththeparentsareheterozygous,Aa,thentheprogenygenotypewillhavea25%chancetobeAA,50%chancetobeAa,and25%tobeaa.Insomecases,neitherparentgenotypeisknown;eachgenotypegroupisassignedaprobabilityof0.33.ThemissingdataisimputedduringeachiterationintheGibbsSamplerwiththeirparentalgenotypesasprior.Table 4-1 showedthepriorforthemissinggenotype. 91

PAGE 92

Priorforthemissinggenotype Parent1 Parent2 AAABBBUnknown AA AA1.0 AB AA0.5AB0.5AA0.25AB0.5BB0.25 BB AB1.0AB0.5BB0.5BB1.0 Unknown 0.33each0.33each0.33each0.33each 4.3.2.1SNPDetectionforClonalRepeatabilityof0.2and0.4 4-2 summarizesthedetectedSNPsunderCR=0.2 92

PAGE 93

Table4-2. SummaryofdetectedSNPsunder0.2ofclonalrepeatability Parameter StructuredUnstructured Replication 5rep20rep50rep 5rep20rep50repdiscovered 395464 264156falsepositive 1176 799 FDR 28.2%12.9%9.4% 26.9%22.0%16.1% Table 4-3 showsthenumberofsignicantSNPsdetectedineachofthreecategoriesandshowsthedistributionofthedetectedSNPpercentageattheseadditivevalueintervalswhenCRwas0.2.Ingeneral,whentheadditivevalueincreases,thepowertodetectassociatedSNPsalsoincreases. Table4-3. TotalnumberofSNPsinadditivevalueintervals(0.3maximumvalue)withnumberdetectedassignicantandpercentagedetectedforstructuredandunstructuredpopulationswiththreelevelsofreplicationoftheclonalentriesatarametlevelrepeatabilityof0.2;sigabbreviatesforthedetectedtruesignicantSNPsandtotalabbreviatestotalnumberofsignicantSNPs Popu.Add.5rep20rep50repval.sigtotal% sigtotal% sigtotal% Structured>0.20115719.3% 255644.6% 387451.4%0.1-0.275911.9% 176028.3% 176327.0%<0.1108312.0% 5846.0% 3634.8% Unstructured>0.20146521.5% 176028.3% 365862.1%0.1-0.22613.3% 147319.2% 136619.7%<0.13744.1% 1671.5% 7769.2% Table 4-4 summarizesthedetectedSNPsandthepercentageofthetotalsimulatedsignicantSNPsatCR=0.4.Atthislevelofclonalrepeatability,thetraitisgenerally 93

PAGE 94

Table4-4. TotalnumberofSNPsinadditivevalueintervals(0.3maximumvalue)withnumberdetectedassignicantandpercentagedetectedforstructuredandunstructuredpopulationswiththreelevelsofreplicationoftheclonalentriesatarametlevelrepeatabilityof0.4 Parameter StructuredUnstructured Replication 5rep20rep50rep 5rep20rep50repdiscovered 434746 345058falsepositive 323 8710FDR 7.0%4.3%6.5% 23.5%14.0%17.2% Table4-5. TotalnumberofSNPsinadditivevalueintervals(0.3maximumvalue)withnumberdetectedassignicantandpercentagedetectedforstructuredandunstructuredpopulationswiththreelevelsofreplicationoftheclonalentriesatarametlevelrepeatabilityof0.4;sigabbreviatesforthedetectedtruesignicantSNPsandtotalabbreviatestotalnumberofsignicantSNPs Popu.Add.5rep20rep50repval.sigtotal%sigtotal%sigtotal% Structured>0.20217129.6%297737.7%317143.7%0.1-0.2176825.0%216731.3%105418.5%<0.12613.3%4567.1%3754.0% Unstructured>0.20145724.6%245940.7%245841.4%0.1-0.297811.5%126817.6%136719.4%<0.13654.6%7739.6%117514.7% Figure 4-1 showstheregressionofthedetectedSNPscorrespondingtotheiradditivevaluesatclonalrepeatabilityof0.2(left)and0.4(right)respectively.Asexpected,astrongtrendisobservedthatastheSNPeectsbecomelarger,thepercentageofdetectionincreasescorrespondingly. 94

PAGE 95

DetectedSNPsandtheircorrespondingadditivevalue 4-3 summarizesthedetectedSNPswiththeircorrespondingadditivevaluesandminorgenotypefrequencytoinferthepowerofthetest. 95

PAGE 96

4-2 showedtheregressionofthesignicantSNPadditivevalueanditscorrespondingminorgenotypefrequencyatclonalrepeatabilityof0.2and0.4respectivelyfortwopopulations. MinorgenotypefrequencyandthedetectedtruesignicantSNPadditivevalueunderRRof0.2and0.4 ThepowercurvesforSNPdetectionwithBAMDcontinuedthebasictenantthatextremelyprecisephenotypesareoptimalforSNPdetection(Figure 4-3 ).Thisgureshowsthatwhenclonalmeanrepeatabilityincreasesfrom0.8to1.0,foradditiveeectsof0.4,anincreaseof15%inthepowertodetectsignicantSNPswasobserved.ThecurvesalsoillustratethedicultiesindetectingSNPsoflowadditiveeectinthatwhenthecontrastofthetwohomozygotesislessthan9%ofaclonalmeanstandarddeviationdetectionwaslessthan20%. 96

PAGE 97

AdditivevalueanddetectingpoweroftruesignicantSNPsforthreeclonalmeanrepeatabilitylevels 2009 ) 97

PAGE 98

1996 ).Uninformativepriorsmeanthatdataaremissingatrandomandhaveequalprobabilityforeachgenotypegroup. CCLONESisastructuredpopulationwith61full-sibfamilies;eachfamilyhasabout15progeny;theprogenywereclonallypropagatedasrootedcuttingstogeneratetherametsthatwereplantedontwotestsitesCuthbert,GA,andPalatka,FL.Allofthe46availableSNPsweresimultaneouslytestedfor50,000iterationsusingBAMD.Therst10,000wereconsideredburn-in,andaftertheburn-in,thechainwasthinnedevery4iterationstoensure'independentsampling'fromthechain.Figure 4-4 showstheautocorrelationofthechainaftertheburn-in.Figure 4-5 showstheautocorrelationofthechainaftertheburn-inandthinning.Theautocorrelationisreducedsignicantlyafterthethinning,andthusatotalof10,000samplesforeachchainforourstatisticswereused. Table 4-6 s10andein2 s1)werealsofoundwithQTDT( 2008 ).TheCaf1 s1SNPwassignicantwiththeuninformativepriorbutnotsignicantwithinformativeprior.Whenusingtheinformativeprior,twoSNPs(dhn1 s2andcpk3 s5)weredetectedassignicant;whiletheuninformativepriorfailedtodetectthem.Theseresultsindicatethatinformativepriorsimprovethepowerofdetection.WiththeinformativepriorsBAMDdetected3ofthesamesignicantSNPsastheQTDTtest. 4-6 and 4-7 indicatessignicantonGonzlez-Martnez2008paper. 98

PAGE 99

AutocorrelationoffoursignicantSNPsforcarbonbeforethinning Table4-6. SignicantSNPsfromBAMassociationmodelat95%condenceinterval SNPInformativePriorUninformativePrior low2.5%Meanupper97.5% low2.5%Meanupper97.5%caf1 s1*-0.0080.0510.110 0.0130.0720.129ccoaomt s10*-0.103-0.058-0.012 -0.097-0.052-0.005cpk3 s5-0.101-0.052-0.004 -0.096-0.0480.001dhn1 s2*0.0160.0650.113 -0.0030.0440.092ein2 s1*0.0100.0770.142 0.0070.0670.126 AsMartinez-Gonzalezhasshown,therearesomeothermarginallysignicantSNPsatalpha=0.1,wealsosummarizedourSNPsthataresignicantatalpha=0.1level(Table 4-7 ).Inthiscase,caf1 s1wassignicantusingQTDTanditalsoappearedtobesignicantat0.1levelwithinformativepriorusingBAMD. 99

PAGE 100

AutocorrelationoffoursignicantSNPsforcarbonafterthinning 100

PAGE 101

SignicantSNPsfromBAMDassociationmodelat90%condenceinterval SNPInformativePriorUninformativePrior low5%Meanupper95% low5%Meanupper95%caf1 s1*0.0010.0510.101 0.0230.0720.120comt4 s4-0.0020.0540.109 0.0030.0580.113ccoaomt s10*-0.095-0.058-0.019 -0.090-0.052-0.013cpk3 s5-0.093-0.052-0.012 -0.089-0.048-0.007setc s5-0.087-0.045-0.003 -0.087-0.044-0.001dhn1 s2*0.0240.0650.105 0.0040.0440.085ein2 s1*0.0210.0770.132 0.0170.0670.116 101

PAGE 102

5.1.1AssociationGenetics 1994 ; 1996 ).Astrengthofthisapproachovertypicalquantitativetraitloci(QTL)studiesisthatittakesadvantageofthehistoricalselection,mutationandrecombinationeventswhichoccurredthroughoutthenaturalhistoryofthepopulationtodramaticallyimprovetheresolutionofthemapping( 2003 ; 2004 ; 2006 ).Recently,associationgeneticswasusedsuccessfullyinmaizetoidentifycandidategenesunderlyingMendeliantraits(endospermcolorandsweetnesstaste),andcandidategeneswithrelativelylargeeectonquantitativelyinheritedtraitswithhigh(oweringtime)andmoderateheritability(starchconcentration)( 2001 ; 2002 ; 2004 ; 2005 ; 2005 ; 2006 ).Inaddition,ithasbeenusedsuccessfullytoidentifycandidategenescontrollingwoodproperties( 2007 ),andwateruseeciency( 2008 ). LinkageDisequilibrium(LD)isthenon-randomassociationofsinglenucleotidepolymorphism(SNP)allelesfromdierentlociwithaphenotypictraitandplaysaveryimportantroleingeneticassociationstudies.StudiesofLDandassociationmappinghavebeenreportedinhumans( 1994 ; 2000 ; 2000 ; 2001 ; 2006 )aswellasinplants( 1999 ; 2001 ; 2002 ; 2004 ; 2006 ; 2007 ; 2007 ).LDdecaydependsonthematingsystem,eectivepopulationsize,mutation,populationdriftandpopulationbottleneck 102

PAGE 103

2001 ; 2006 ).ThedetectionofLDmightimplythattherecombinationfractionbetweentwomarkersisrelativelysmallandthustheSNPmarkerislikelytobecloselylinkedwiththecausativegene.UnlikeanimalsandArabidopsis,whereLDaveragesover150kb,LDinconifersappearsshort,averaging2-3kb,andSNPsshowinggeneticassociationswiththetraitofinterestarelikelytobelocatedincloseproximitytothecausalloci( 2004 ; 2007 ).InDouglasFir,thenucleotidediversityishighandtheLDdecreasessteadilywithingenes( 2005 ).InMaritimepine,rapiddeclinesinLDbetweensiteswithinmostgeneswasobserved( 2003 ).InScotspine,lowLDwasobservedevenbetweencloselylinkedloci( 2002 ).Inloblollypine,nucleotidediversityishigh( 2007 )andLDvariesamonggenesbutonaverager2decaysto0.20within1500bps( 2004 ).ItshouldbenotedthatLDcanexistbetweenunlinkedlociinthestructuredpopulationduetoselection,geneticdrift,migration,andmutation( 2005 ). 2002a ; 2003 ; 2006 ).Inloblollypine,QTLmappingforwoodpropertieshasbeencompletedinathree-generationoutbredpedigreewithfourgrandparentsandtwoparents.Inthispedigree,172progenygrownon6siteswerephenotypedandQTLmappingmethodswereusedwithrestrictionfragmentlengthpolymorphism(RFLP)basedmaptoidentifynineQTLsforwoodspecicgravity,veforMFAandeightuniqueQTLsassociatedwithcellwallchemistry( 2002a ).Inanotherloblollypinepedigree,twopopulationswereusedfor 103

PAGE 104

2003 ).Inhybridlarch,68ampliedfragmentlengthpolymorphism(AFLP)markerswereusedtodetectQTLforwooddensitywitheighthalf-sibfamiliesfromacrossofEuropeanlarch(Larixdecidua)asthefemaleparentand8Japaneselarch(Larixkaempferi)asthemaleparents,9markerswereshowntobelinkedtoringspecicgravity(SG),8markerslinkedwithearlywoodspecicgravityand7markerswithlatewoodspecicgravity.Fourmarkersco-segregatedwithringSGandearlywoodSG,1markerco-segregatedwithringSGandlatewoodSG,and1markerco-segregatedwithlatewoodSGandearlywoodSG.TheproportionofvariationofringSGexplainedbymarkersrangedfrom3.6%to5.8%.( 2002 ).IncoastalDouglasr(Pseudotsugamenziesii),anampliedfragmentlengthpolymorphism(AFLP)mapof19linkagegroupswithanaveragedistanceof9.3cMbetweenmarkerswasconstructedfromeightfull-sibfamilieseachconsistingof40progeny.Usingthismap,3QTLsshowedsignicanteectsforwoodspecicgravityand7QTLswereidentiedforwoodchemistry:1forgalactosecontent;2forglucosecontent;1formannosecontent;1forarabinosecontent;1forlignincontent)( 2008a b ).Inmaritimepine(Pinuspinaster),186individualsfromanoutbredpedigreeofthecrossoftwoparentswereusedtoidentifyQTLsthatcontrolwoodchemistryandwoodpropertytraits.Inaddition,10candidategenesinvolvedinsecondarycellwallbiosynthesischosenbasedonthefunctionalinformationgatheredfromtreespeciesandmodelplants,-5genes(Korrigan,CesA01,CesA3,PFK,Susy)involvedinpolysaccharidesynthesis,4genes(PAL,C4H,CAD,CCoA-OMT)involvedinlignin,andonecellwallprotein(AGP)weremapped.Ofthese9candidategenes,KorriganmappedwithinaQTLintervalthataectshemicellulosecontentandbercharacteristics( 2006 ).Theabovestudiesshowthatwood 104

PAGE 105

1998 ; 2003 ).Finally,asignicantdisadvantageoflinkageanalysisinconifersistheirlargegenomesizesmakeitnearlyimpossibletoidentifythecausativegene(s)withinthephysicallylargeintervals. Expressedsequencetags(ESTs)arerelativelyshortsequences(usually200to800nucleotideslong)thataregeneratedbysequencingeitheroneorbothendsofthemRNAsfromexpressedgenes.ESTsareapowerfulandcosteectivewaytoidentifyandcataloggeneswithinaspecictissueanddevelopmentalstagethatcouldcontrolspecictraits( 1991 ; 2000 ).Inloblollypine,collectionsofESTsenablecomparisonswithmodelplants,providetheabilitytoconductdetailedexpressionstudiesusingmicroarrayexperimentsandprovideasequenceresourceforthediscoveryofpolymorphismsthatcanbeusedinmappingstudies.Inloblollypine,threelargeESTsequencingeortshavebeenreported,onewithdierentiatingsecondaryxylem( 2003 ),onewithrootsaftervarioustreatments( 2006 ),andonewithzygoticandsomaticembryos( 2006 ).Intotal,theseeortshavecontributed>300,000ESTsequencesandunigenesetof16,000-33,000dependingonthecontigingalgorithmused.Asmallnumberofmicroarrayprojectsinconifershavebeenreported( 2001 ; 2008 ). Athirdapproachtodiscoveringgenesthatcontrolwoodpropertiesinconifersisassociationgeneticmapping.Associationgeneticsisapopulationbasedsurveythatuseshistoricalselectionandrecombinationeventswhichoccurredthroughoutall 105

PAGE 106

2003 ; 2004 ; 2006 ).Associationgeneticanalysesarewellsuitedtoconifersingeneralandloblollypineinparticularbecauseofitslongevolutionaryhistory,largenaturalpopulations,out-crossingmatingsystem,substantialESTsequenceresources,highnucleotidediversity,andrelativelyshortaverageLDs. 1993 ; 2003 ).Forquantitativetraits,LDassociationbetweenamarkerandtraitofinterestcanbedoneusingregressionanalysis( 2001 ),structuredassociationtests( 2000 ; 2001 ),andmixedlinearmodels( 2006 )andQTDTtest( 2008 ).LDassociationmappingcanpreciselypositiontheQTLthatcontrolsthetraitandidentifythecausativemutationwhenLDisshort( 2003 ). Numerousmarkersystems(i.e.AFLP,SSR)havebeenusedinQTLandLDmapping;however,themarkersofchoiceformostassociationgeneticexperimentsaresinglenucleotidepolymorphisms(SNP).ThehighfrequencyandclosespacingofSNPsprovidemuchhigherresolutionthanothermarkertypesandtheyareeasilyidentiedfromhighthroughputsequencedata.Forexample,inloblollypineaSNPwithincodingregionsoccursonaverageevery50basepairs( 2004 ).Onelimitationofassociationanalyseswithout-crossingplantshavinghighlevelsofnucleotidediversityisthatSNPdiscoveryandgenotypingcostsaretoohightopermitwholegenomeassociations.Thus,mostassociationstudiesintheseorganisms,likeloblollypine,usecandidategenes.Candidategenesareoftenselectedbasedonhomologousgenesidentied 106

PAGE 107

2003 ).InapopulationofloblollypineaninitialanalysisdetectedsignicantassociationsbetweenasmallnumberofSNPlociandwoodpropertyphenotypes( 2007 ). Oneconstraintonassociationmappingishiddenpopulationstructure.Populationstructurecanseriouslyconfoundassociationtestingbycausingspuriouscorrelations,leadingtoanelevatedfalse-positiverate( 1994 ).Thus,itisimportanttodeterminethecorrectpopulationstructureoronlyconductwithinfamilytestssuchasthetransmissiondisequilibriumtest(TDT)whichusesparentswithaectedospring( 1998 ).TDTdetermineswhetheralleleAisassociatedwithdiseasebycomparingthenumberofparentswhotransmitAtotheiraectedospringwiththenumberwhotransmittheotherallele( 1999 ).QuantitativeTransmissionDisequilibriumTest(QTDT)isanapproachbasedonregressionmethodsforassociationtestingofquantitativetraitsoffamily-basedsamples( 1998 ).Whenpopulationstraticationisnotpresent,quantitativetraitlinkagedisequilibriumtestcanbemorepowerfulbyaddingfoundergenotypeinformation( 2005 ).StructuredassociationwasrstproposedbyPritchardetal.,unlinkedmarkerlocicanbeusedtodetectpopulationstraticationandtoinferpopulationstructureaswellasancestryestimatesofthesampledindividualsusingamodel-basedBayesianclusteringalgorithmSTRUCTURE( 2000 ).Ancestryestimatesareassignedtoindividualsinthesubpopulationandtestedforassociationwithinsub-populations( 1999 ;

PAGE 108

2001 ).Mixed-modelapproachesaccountingforeithermultiplelevelsofrelatedness( 2006 )orgenome-widedierences( 2007 )inrelatednesswereshowntoperformwellincontrollingTypeIandIIerrorrate,reducingfalsepositiverate,aswellasincreasingstatisticalpower.In 2006 ),anonymousSSRmarkerswereusedtoestimatepopulationstructure(Q)andtherelativekinshipmatrix(K);andthenQandKweretintoamixedmodeltotestformarker-traitassociation.Thisapproachaccountsforrelatednessatboththefamilyandpopulationlevel.In 2007 ),theyappliedtheKandQmatricestotestforassociationbetweenmarkerandoweringtimeofmaizeandsuccessfullyreducedthefalse-positiverateclosetotheexpectedlevels,indicatingthatconfoundingbypopulationstructurewassignicantlyreduced. InaPinusradiatafull-sibfamilystudy,amixedmodelwasusedtocalculatethevariancecomponents,estimateandpredicttheparentalbreedingvalues( 2004 ).Thebreedingvaluesfordiameteratbreastheight(DBH),stemstraightness,andwooddensitywereusedforthemarker-traitassociationswith34SSRmarkers.Nosignicantassociationswerefound,butitispossiblethattheover-representationofthefemaleparentsinthese200full-sibfamiliescouldhavebiasedtheprocessofdetectingassociation,i.e.,thefrequencyoffavorablealleleswasgenerallyhigherinthefemalepopulation( 2004 ).Inaloblollypineassociationstudy( 2007 ),amixedlinearmodelwithapairwisekinshipmatrixamongindividualstoaccountforrelatednesswastfor42singlemarkersandwoodpropertytraits.22nuclearmicrosatelliteswerealsousedtotestforpopulationstructure.Themodel-basedclustering(STRUCTUREsoftware, 2000 ))resultshowedapatterntypicalofunstructuredpopulations.Morerecently,anassociationstudyusingQTDTinvestigatedwaterrelationsmeasuredbythecarbonisotopediscriminationofneedlesand46SNPmarkersintheCCLONESpopulation.AlthoughnoSNPswerefoundtobesignicantafterapplyingaBonferronicorrection,Dhn-1andein2-s1showed 108

PAGE 109

2008 ). 2005 ).Thealpha-latticedesignmoreecientlyreducestheerrorvariancethantherandomizedcompleteblockdesignbyincorporatingrowandcolumnintothedesignusingacyclicmethodofvarietyconstruction( 1997 ; 2005 ). Toaccomplishtheseobjectives,woodpropertytraitsweremeasuredforover3800treesrepresenting999genotypesgrownontwooftheCCLONESsites.Woodchemistry(lignin,C5andC6sugar,andC20extractivecontent)wasdeterminedbypyrolysismolecularbeammassspectroscopyofrings3and4( 2008 )(seechapter3).WoodstinesswasmeasuredwiththeDirectorST300in-treeacousticvelocityinstrument.Heightandstemdiameterat1.4m,andcrowndimensionswereavailablefromtheForestBiologyResearchCooperative'sbaselinemeasurements. Themarkerinformationwasobtainedfor2182SNPsfromgenotypingthe999dierentclonallypropagatedlines.Atotalof7600SNPswereselectedforgenotyping.SNPgenotypedatawereobtainedthroughIluminaInniumplatform.WoodstinessandwoodchemistrytraitswerethentestedforassociationusingtheseSNPmarkerswithaBayesianMarkovChainMonteCarlomethodusingaGibbssamplertoruntheMCMCchainormissingdataimputation( 2009 ).Chapter4focusedonsensitivity 109

PAGE 110

2005 ).Woodcoresfromthestemwerecollectedfromthehighintensitytreatmentat0.8-1.0mabovethegroundfrom4yearoldtrees,fromtworeplicationsattheCuthbert,GAandNassau,FLsites.Thephenotypictraits,includinggrowth,chemistry(seechapter3)andintreewoodstinesswiththeDirectorST300fromFibreGen( 2007b )werephenotypedforthetreesgrownonthesesites. Needlesamplesof999genotypesfrom61full-sibfamiliesfortheCCLONEpopulationrepresentedatthesetwositeswerecollectedintheeldandstoredonicepriorstorageat-200C.SampleswerethengroundandDNAwasextractedusingQaigenDNeasyminikit.DNAsampleswerethendividedintotwoportions,onefortheSNPgenotypingatUCDavisandonefortheSSRmarker. 2005 ).SampleswithGenTrainscoreslessthan0.7wereexcludedfromtheanalysis,becausetheyshowdierentlevelsofclustering.Forexample,somehaveclearly 110

PAGE 111

5-1 ). Table5-1. NumberofSNPsindierentcategoriesdependingonthenumberofminorhomozygotesfoundintheSNPcall Lessthan103804Between11and15198Morethan152182 Onepracticalissueforassociationtestingisthehighcomputationalburdenwhenallthefactors(i.e.,site,replication,aswellasincompleteblock)arerunintheassociationmodelwithhundredsofSNPeects.BecausegenotypesgrownondierentsitesactuallyhavetheidenticalSNPgenotypeinformation,priortoassociationtesting,asimplemodel 5{1 wasusedtoaccountforenvironmentaleectsonthephenotypicvariation.Thus,theleastsquaremeansforeachgenotype(CLONE)wereusedasphenotypicvaluesforassociationtestingwhichkeepsallofthegeneticvariationpriortoassociationtests. Although,2182SNPswereselected,notalltheSNPsaretestedwiththeassociationmodel.Thisisbecause(1)manySNPshavenoeectswiththetraitofinterest;(2)toomanySNPsinthemodelcancausenoiseandthusmakeitdiculttoidentifysignicantSNPs;and(3)forpracticalpurpose,moreSNPsmeanslongercomputationtimes.Thus, 111

PAGE 112

5{2 )thatincludedonlythefamilyandSNPeectinthemodeltoidentifytheSNPswiththelargesteects. EachSNPwasrunwiththeaboveANOVAmodelandthereducedmodelwithouttheSNPeect.Achi-squaretestwasthenusedtotesttheeectofthatSNPforthetraitandthep-valueofthechi-squaretestwasstoredforthatSNP.Aloopforall2182SNPswasappliedinRtocalculatethep-valueforallthese2182SNPsandthep-valuesweresortedtoobtainthetop400SNPsforeachtrait.These400SNPswerethentestedinthenalassociationmodel.TheRcodetoruntheaboveprocessisshowninAppendix D E 5-2 ).Thep valuesforthetop400SNPsforlignincontentrangedfrom1.48E-11to0.02.Tenoutofthese13signicantSNPshavep valueslessthan0.005.SNP0 14949hadthelargesteect 112

PAGE 113

Table5-2. SignicantSNPsforlignincontentat95%Bayesiancondenceinterval;pstdevstandsforphenotypicstandarddeviation 0 149497.3unknown0 183224.2Os08g05181000 188874.1aminoacidpermease0 54703.5planttype,putativeserinethreoninekinase0 8863.5NoHitsFound2 102073.3hypotheticalprotein2 14474.8unknown2 34633.7Activatingsignalcointegrator,putative2 44554.2GTPbindingprotein,putative2 49055.2mitochondrialAcetylornithineaminotransferase2 64344.0unnamedproteinproduct2 94573.8predictedproteinUMN 59424.3conservedhypotheticalproteinUMN 8014.1hypotheticalprotein 2008 ),3)theC6peak,m/z144,wasthemostheritableandislikelytocomefromcellulose,4)m/z114and144areonlymoderatelygeneticallycorrelated. First,anassociationtestwasdonewiththesumofallchemicallyidentiedcarbohydratepeaks(m/z57+60+73+85+98+114+126+144),whethertheywereheritableornot.ElevensignicantSNPswithp valueslessthan0.006(Table 5-3 )wereidentied.SNP2 4905hasthelargesteectwithanadditivevalueofabout7.1%ofthetotalphenotypicstandarddeviation. 113

PAGE 114

SignicantSNPsforthesumofallassignedC5andC6sugarsat95%Bayesiancondenceinterval;pstdevstandsforphenotypicstandarddeviation 0 107895.6glycosidehydrolasefamily28protein0 139587.0hypotheticalproteinOsJ 364450 150035.8NoHitsFound0 175576.0aminoacidtransporter,putative0 3664.1heatshockfactorprotein70 91484.7short-chaindehydrogenase,putative0 91696.5putativeheatshockprotein70interactingprotein2 14474.7unknown2 49057.1mitoch.AcetylornithineaminotransferaseCL1287Contig15.8unknownUMN 54324.8predictedprotein valuesforthetop400SNPsforthesumofC5sugarpeaks(m/z57+73+85+96+114)is8E-10to0.021,andthe10signicantSNPshavep valueslessthan0.0195.Table 5-4 summarizesoftheinformationaboutthe10SNPsthatweresignicantlyassociatedwithC5sugars.Thetop2SNPswere0 14949and0 1974butneitherhassignicantsimilaritytosequencesinGenBank.CL148ctg1isaputativeauxin-responsiveproteinand2 5064isaputativemitogen-activatedproteinkinase. Table5-4. SignicantSNPsforsumofallassignedC5sugarpeaks(m/z57+73+85+96+114)at95%Bayesiancondenceinterval;pstdevstandsforphenotypicstandarddeviation 0 140397.8hypotheticalproteinOsJ 364450 1494910.5unknown0 164587.3unknown0 18968.5predictedprotein0 197410.2unknown2 50649.2putativemitogen-activatedproteinkinase2 86116.2hypotheticalproteinOsJ 09149CL1287ctg19.4unknownCL148ctg19.0auxin-responsiveprotein,putativeCL1595ctg17.7Lhca4protein 5-5 )with95%condence.SNP0 10789hasthelargesteectonm/z114intensityanditsadditivevalueisabout11.7%ofthetotalphenotypicstandard 114

PAGE 115

Table5-5. SignicantSNPsform/z114at95%Bayesiancondenceinterval;pstdevstandsforphenotypicstandarddeviation 0 1021010.0Anthocyanin5-aromaticacyltransferase,putative0 103739.7GTP-bindingprotein-like;roothairdefective3protein-like0 1078911.7glycosidehydrolasefamily28protein0 139587.4hypotheticalproteinOsJ 364450 150037.9NoHitsFound0 154669.7aminoacidtransporter,putative0 175579.2aminoacidtransporter,putative0 26898.4NoHitsFound0 8349.8NoHitsFound2 14479.5unknown2 15637.5hypotheticalproteinOsJ 087912 54389.2Os03g0194900CL1568Contig19.2nohitfoundCL4420Contig17.5nohitfoundUMN 54329.4predictedproteinUMN 689110.0unknownprotein 5-6 ).SNP0 10789explainsthegreatestproportionofthephenotypicstandarddeviation(10.5%)andsimilarinsequencetomembersoftheglycosidehydrolasefamily28enzymesandwasalsoidentiedwithm/z114. Table5-6. SignicantSNPsforthesumofthechemicallyassignedC6sugar(m/z57+60+73+98+126+144)peaksat95%Bayesiancondenceinterval;pstdevstandsforphenotypicstandarddeviation 0 1078910.5glycosidehydrolasefamily28protein0 3669.6hypotheticalproteinOsI 321080 89637.5NoHitsFound0 102109.2putativeanthocyanin5-aromaticacyltransferase2 14478.9unknown2 35429.2unknown2 48617.0unknown2 49058.0mitochondrialacetylornithineaminotransferase2 98517.5F-boxfamilyprotein/WD-40repeatfamilyproteinCL1287Contig110.0unknown2 48619.7unknown

PAGE 116

valuesforthetop400SNPsrangefrom4.38E-11to0.01,andthe11signicantSNPshavep valueslessthan0.002(Table 5-7 ).SNP0 15003explainsthehighestpercentageofthephenotypicstandarddeviation(10.2%)butisunrelatedtoanyothersequenceinthedatabase. Table5-7. SignicantSNPsforheritableC6sugarpeaks(m/z60+73+98+126+144)at95%Bayesiancondenceinterval;;pstdevstandsforphenotypicstandarddeviation 0 102199.7NoHitsFound0 107898.8glycosidehydrolasefamily28protein0 1500310.2NoHitsFound0 3668.4putativeheatshockfactor0 89638.4NoHitsFound2 14478.8unknown2 49058.1mitoch.Acetylornithineaminotransferase2 98518.3F-boxandwd40domainprotein,putativeCL1052ctg18.8Nitric-oxidesynthase,putativeCL1287ctg110.0unknownUMN 54328.6predictedprotein 5-8 ).SNP0 6817hasthelargesteectonm/z144intensityanditsadditivevalueisabout12.3%ofthetotalphenotypicstandarddeviation. Table5-8. SignicantSNPsform/z144at95%Bayesiancondenceinterval;pstdevstandsforphenotypicstandarddeviation 0 137228.8dirigent-likeprotein0 139587.9hypotheticalproteinOsJ 364450 1500310.7NoHitsFound0 681712.3similartofungalresistanceprotein;ASC1-likeprotein20 91488.4short-chaindehydrogenase,putative0 91699.2putativeheatshockprotein70-interactingprotein2 490510.0Acetylornithineaminotransferase,mitochondrial2 50268.4hypotheticalproteinOsJ 281522 96839.4PREDICTED:hypotheticalproteinCL3061Contig18.4calmodulinbindingprotein,putativeUMN 68919.4unknown

PAGE 117

5-9 ).SNP0 6817explainsthegreatestpercentageofthephenotypicstandarddeviationof10.6%. Table5-9. SignicantSNPsforthesumofthemostheritableC5sugar(m/z114)andC6sugar(m/z144)at95%Bayesiancondenceinterval;pstdevstandsforphenotypicstandarddeviation 0 107899.3glycosidehydrolasefamily28protein0 139588.2hypotheticalproteinOsJ364450 150037.4NoHitsFound0 2156.1NoHitsFound0 681710.6ASC1-likeprotein/putativeresistanceprotein0 91486.5short-chaindehydrogenase,putative0 91697.4putativeheatshockprotein70(HSP70)-interactingprotein2 15637.2hypotheticalproteinOsJ 087912 49058.6Acetylornithineaminotransferase,mitochondrialCL1052Contig18.1hypotheticalproteinOsJ 05018 5-10 ).ForC6sugars,asimilarnumberofsignicantSNPswasfoundwiththesumofallchemicallyknownpeaksandtheheritablepeaks,andasexpected11ofthemareincommonbecause5ofthe6peakshavesignicantheritability.Incontrast,nocommonSNPswerefoundbetweenthesumofallC5sugarpeaksandtheheritableC5m/z114peak(Table 5-10 ).Thisisprobablyduetothenoiseintroducedbyincludinginthesum3peaksthatlackgeneticcontrol(seeChapter3).Interestingly,theassociationtestwiththesumofthemostheritableC5(m/z114)andC6sugar(m/z144)peaks,identied10signicantSNPsofwhich7werealsosignicantinotherassociationtestsand4ofthese7SNPs-0 10789,0 15003,2 4905,andCL1052Ctg1-werefoundin4ormoreofthedierentassociationtests(Table 5-10 ). 117

PAGE 118

SummaryofSNPsfoundintwoormoreAssociationTestsforCarbohydrates SNPc5t114c6c6ht144t114144c5c6 0 10210**0 17557**0 6817**0 8963**2 1563**2 9851**CL1052ctg1**UMN 6891**0 366***0 9148***0 9169***UMN 5432***0 13958****2 1447****CL1287ctg1****0 10789*****0 15003*****2 4905*****GrandTotal10161111111011 5-11 ).SNP0 18301explainedthegreatestamountofvariationforpeakm/z285with12.4%ofphenotypicstandarddeviation. Table5-11. SignicantSNPsfortheresinacidextractivepeakm/z285at95%Bayesiancondenceinterval 0 114756.7proteinbindingprotein,putative0 120539.6NoHitsFound0 136738.7NoHitsFound0 1830112.4NoHitsFound2 10237.6540Sribosomalptn.SA/P40-likeptn.2 103996.5Indole-3-aceticacid-inducedproteinARG7,putative2 134311.3LAC3(laccase3);copperionbinding/oxidoreductase2 27187.4Os08g01071002 90177.0CytochromeP450716B22 91026.57hypo.ptn.OsJ 01438CL1376ctg18.7NoHitsFound

PAGE 119

5-12 ). Table5-12. SignicantSNPsfortheresinacidextractivepeakm/z300at95%Bayesiancondenceinterval 0 114756.38proteinbindingprotein,putative0 120537.34NoHitsFound0 136738.18NoHitsFound0 183018.56NoHitsFound2 103995.94put.Indole3-aceticproteinARG72 13438.76LAC3/copperionbinding/oxidoreductase2 36026.32NoHitsFound2 25038.14Os08g0107100UMN 58306.32NoHitsFound0 114707.84hypo.ptn.MtrDRAFT AC148819g4v22 10236.3240Sribosomalptn.SA/P40-likeptn.CL3375Contig15.95hypo.ptn.OsI 02082CL4776Contig16.46put.carboncataboliterepressorptn.UMN 34887.11NoHitsFound0 144357.37put.integralmembraneptn. 5-13 ).These7SNPsdonotoverlapwithotherwoodchemistryorwoodstinesstraits.Allofthese7SNPshavep valueslessthan0.01.OnlySNP2 10399hasapositiveeect.SNP2 1343'sadditivevalueisabout8.8%ofthetotalphenotypicstandarddeviation. Table5-13. SignicantSNPsforthesumofC20resinacidextractivepeaksm/z285+300at95%Bayesiancondenceinterval 1205314.9NoHitsFound0 1367316.6NoHitsFound0 1830118.9NoHitsFound2 1039912.7Indole-3-aceticARG7/SAURfamilyptn.2 134317.0LAC3/copperionbinding/oxidoreductaseCL1376Contig114.8CytochromeP450716B2UMN 583013.2put.alanyl-tRNAsynthetase 119

PAGE 120

5-14 ).Velocitystinessistheonlytraitforwhichthe90%condenceintervalisreported,becauseithasstrongergeneticcontrolthanthewoodchemistrytraits.Thep valuesforthetop400SNPsrangedfrom1.55E-12to0.0022.SNP2 10183hasthelargesteectonwoodvelocityanditsadditivevalueisabout12.1%ofthetotalphenotypicstandarddeviation. Table5-14. SignicantSNPsforvelocitystinessat95and90%Bayesiancondenceinterval.Boldindicatessignicantatthe95%level 11269.247caeicacidO-methyltransferase0 150759.995GrxS16-glutaredoxinsubgroupII0 369910.138NoHitsFound2 1018312.246put.diseaseresistanceprotein2 156310.323hypotheticalproteinOsJ 087912 346310.947put.Activatingsignalcointegrator0 128627.532unknown0 132409.721L-aspartateoxidase0 135658.503unknown0 31929.397ProteincbxX,chromosomal,putative0 70467.610Beta-ureidopropionase,putative0 971311.245hypotheticalprotein2 102078.992hypotheticalprotein2 111710.090NoHitsFound2 39899.609WRKYtranscriptionfactor1CL2076Ctg1 048.375NoHitsFoundUMN CL298Ctg1 049.547NoHitsFound 5-15 ).66showedsignicantsequencesimilaritywithgenesofknownorputativefunction. Table 5-10 showedthatcarbohydratetraitsarehighlycorrelatedandthusSNPsassociatedwiththemwereidentiedusingindividualaswellassumsofknownC5andC6sugarpeaks.Forexample,SNPs0 10789,0 15003,and2 4905weresignicantinveofthesevenassociationtestsforcarbohydrates.SNP0 10789hassequencesimilaritytotheglycosidehydrolasefamily28proteins.SNP0 15003isinanESTthathasno 120

PAGE 121

SignicantSNPsthathavenohitorhaveunknownfunctionfromtheBLAST 10 102190 1286220 120530 1356530 136730 1494940 150030 197450 183012 144760 215CL1287Contig170 268980 369990 834100 886110 8963122 1117132 486114CL1376Contig115CL1568Contig116CL4420Contig117UMN 3488 4905hassequencesimilaritywithamitochondrialacetylornithineaminotransferase.Interestingly,SNP0 4905wassignicantforlignincontentaswell.SNP0 17557hassequencesimilaritywithputativeaminoacidtransportersandwasassociatedwithC5sugarsbutnotC6sugars,includingm/z114(abreakdowncompoundofxylan),andSNP0 9148isinagenewithsequencesimilaritytoshort-chaindehydrogenasesandwassignicantform/z114,indicatingthatthesegenesmayberelatedtoarabinoglucuronxylanbiosynthesisoneofthemainhemicellulosesinpine.Inaddition,SNPs0 6817and0 13958werealsosignicantforbothm/z114+144andm/z114;suggestingtheirinvolvementinarabinoglucuronxylanbiosynthesis.ItisalsointerestingthatseveralSNPswereassociatedwithbothcarbohydrateandlignincontent(Table 5-10 )andthedirectionoftheSNPeectfordierenttraits(notshowninthetable).SNP0 14949wassignicantforbothC5sugar(negative)andlignincontent(positive);0 9169wassignicantforbothm/z114+144(positive)andlignincontent(negative);andSNP2 4905wassignicantforbothC5C6(negative)andlignincontent(positive)withoppositeeects.Itwasshowninchapter3thatligninandcarbohydratewerestrongly,negativelygeneticallycorrelated.Signicantassociationsforin-treevelocitystinessidentify6strongcandidatesatthe 121

PAGE 122

10183hadthelargesteectandhassequencesimilaritytoleucinerichrepeat(LRR)familyproteins.LRRproteinsareofteninvolvedinplantdefenseresponses,butotherfunctionshavebeenidentied( 2005 ).SNP2 3463issimilarinsequencetoanactivatingsignalcointegratorproteinfromcastorbeanandapredictedproteinfromPopulustrichocarpa. SNP2 1563wassignicantforvelocitystinessandthesumofC5andC6aswellasthesumofpeaks114and144.BecauseSNP2 1563contributedpositivelytobothm/z114+144andvelocitystiness,onepossibilityisthatthisgeneaectsthecellulosecontentorcrystallinity,bothofwhichwouldaectvelocitystiness.Inchapter3,itwasshownthatcarbohydratecontentandvelocitystinesswereweaklypositivelygeneticallycorrelated.SNP2 10207contributednegativelytolignincontentandpositivelytothevelocitystiness. Recentgenome-wideassociationstudieshaveshedlightonthegeneticarchitectureofcomplexorquantitativetraitsinDrosophila,miceandhumans( 2009 ).Thesestudiesshowthatcomplextraitsintheseorganismsarecontrolledbylargenumbersofloci,eachwithquitesmalleectonthephenotypicvariation.AnotherndingfromthesestudiesisthatmanysignicantSNPsareinintragenicregions,andthoseincodingregionsareingenesthatoftenwouldnotbepredictedonthebasisofsequenceannotationtobedirectlyinvolvedinthetrait( 2006 ; 2009 ; 2009 ).OurresultswiththerelativelysmallnumberofSNPstestedhereforwoodpropertiesareconsistentwiththewholegenomestudiesintheseotherorganisms,suggestingthatthetrendforcomplextraitsbeingcontrolledbymanylociwithsmalleectandlimitedapriorirelationshipwiththephenotypemayalsoholdtrueinloblollypine.Inaddition,thesesmalleectsareconsistentwithpreviousQTLstudiesinloblollypineshowingthatwoodchemistryandmicrobrilangleareinheritedascomplextraitswithmanygenesinuencingthephenotype( 2000 ; 2003 ). 122

PAGE 123

123

PAGE 124

Theoverallgoalofthisdissertationwastoimproveourunderstandingofthegeneticcontrolandarchitectureofwoodpropertytraitsinconiferspecies.Thestudywasdividedintotwomajorparts: Fortherststudy,woodpropertytraitswerecollectedinmultipleenvironmentsforbothspeciestoevaluatecrosssiteheritability,genotypebyenvironmentinteractions,andtrait-traitcorrelations.Thisstudywasdividedintotwomajorparts: Forloblollypineastructuredpopulationfromadiallelmatingdesignof31parentsfrom3seedsourcesandclonallypropagatedprogenymaterialwastestedinmultipleenvironmentstogiveusbetterphenotypicinformation.SNPinformationfor999genotypeswasusedinassociationtests.Thisstudycanbeseparatedintothreeparts: Basedontheresultsofthechapters1to5,Isummarizemymainndingsandimplicationsasfollow: 124

PAGE 125

WeusedpyMBMStodeterminevariationinwoodchemicalcomponentsintheloblollypineCCLONESpopulation.WefoundthatpyMBMSisafast,ecient,androbustmethodtodeterminethechemicalcompositionofwood.Standardquantitativeanalysesofthechemicalcompositionofjuvenilewoodshowedcomplexinheritancewithsignicant,butlow,levelsofgeneticcontrolforC5andC6sugars,ligninanddihydroabeiticacid.pyMBMSprovidesarichsetofchemicalinformationthatneedstobeminedusingnovelapproachestomoreeectivelyextracttheinformationandcompareitwiththetraditionalwetchemistrytoanalyzethechemicalcomposition.Wediscoveredthatthemajorityofpeakswerenotheritable.Inall32peakswereheritablealbeitatalowlevel.Fourteenofthe32chemicallyidentiedpeakswereheritableand18additionalpeaksofunknownchemicalidentitywerealsosignicantlyheritable. BivariatepairwisegeneticcorrelationsbetweenheritablepeaksshowedthatC5andC6sugarpeaksarestrongly,negativelycorrelatedwithligninpeaks.TheC5andC6sugarpeaksarestronglypositivelycorrelatedwitheachother.Interestingly,20m/zpeaksare 125

PAGE 126

TypeBgeneticcorrelationsforlignincontentwere0.64indicatingsignicantdierencesinrankbetweenthetwositessuggestinggenotypebyenvironmentalinteractions.Incontrast,typeBgeneticcorrelationsfortotalC5(0.71)andtotalC6(0.87)peaksweregreaterthantotallignin,indicatingthatcarbohydratecontent,possiblycellulose,hadlittlegenotypebyenvironmentalinteractions. Threedierentapproaches,singlepeakgeneticanalysis,Metropolis-Hastingsalgorithm(M-H),anddatareductionmethods,wereusedtodeterminewhethermoreofthepeaksthanthosethathavebeenchemicallyassignedcanbeusedinouranalyses.First,ageneticanalysisofsinglepeaks,ratherthansums,identied32ofthe421assignicant.Ofthe32geneticallycontrolledpeaks,halfhaveanassignedchemicalidentity.Theheritabilityofthesinglepeakswasverysimilartothoseobtainedfromthesumofchemicallyidentiedpeakswithinaclass,andpeak182wasthemostheritable.Importantly,allfouroftheuniquelyassignedC6sugarpeaksandfouroutofthesixG-ligninassignedpeakswereheritable.However,onlytwooftheC5sugarpeakswasheritable. AMetropolis-Hastingsalgorithmwasdevelopedtosearchforclustersofpeaksthathaverelativelyhighergeneticheritabilitywhencombined.Withthisapproach,weidentiedclustersofmeaningfulpeakclusterswithknownfunctionalgroups.Forpairsofpeaks,theheritabilityofpeak60increaseswhencombinedwitheitherm/z73,114,126,or144.Similarly,form/z114and126thatclusterwithpeak73,botharecarbohydratepeaks.Itisstrikingthatalltheseknowncarbohydratepeaksrankedinthetop30pairsafter50,000permutations.Theheritabilityofm/z180increaseswhencombinedwithanotherpeakfromG-lignin,m/z150,orwhencombinedwith300and302,C20extractivepeaks.Forclustersof3peaks,thesearchalgorithmfoundmanygroupsofpeaksfromthesamechemicallyidentiedclass.Forexample,theclusterofpeaks73,114,144witha 126

PAGE 127

PCAandWaveletanalysisgiveusthedatareductionofsuchhighdimensionaldataandhelpusidentifymajorvariationswithinallpeaks.ThelevelsofgeneticheritabilityfromPCAanalysisshowedthatonlyPC1hassignicantclonalrepeatabilityof0.15;PC1explainthemajorityofthevariationinthedataandthusithasgoodpotentialinbeingappliedasacompositetraitinthebreedingprogram;whiletherestofthePCcomponentscouldnotprovidedirectinformationforselectionduetolackofheritance(allare<0:05).Thewaveletcoecientsshowedverysimilarpatternsofheritabilityasthoseofsingleheritablepeaks.WaveletanalysishasanotheradvantageisthatitcandealwithspikyfunctionaldatacomparedwithPCAanalysis.Ourresultsalsodemonstratedthatlevel0,level1andlevel2waveletcoecientsareallshowntobeheritableandthusitshouldhavegoodpotentialinapplicationinthebreedingprogramassomecompositetraits. 127

PAGE 128

2009 ). Afamily-basedmixedlinearmodelwasappliedtoaccountforboththefamilystructure(relatedfull-sibfamiliesandcloneswithinfamilies)oftheCCLONEpopulationinordertoincreasethepowertodetectassociationsandthemarkerinformationtominimizetheerrorvariance. 128

PAGE 129

Weidentiedatotalof87signicantSNPsthatwereunique.Seventeen(20%)ofthesewerenotsimilarinsequencetoanysequenceinthedatabase.Sixweresimilarinsequencetogenesthathavenoknownfunction.66SNPsshowedsignicantsequencesimilaritywithgenesofknownorputativefunction. Sincecarbohydratetraitsarehighlycorrelated,wendseveralSNPsareassociatedwithmultiplecarbohydratetraits.Forexample,SNPs0 10789,0 15003,and2 4905weresignicantinveofthesevenassociationtestsforcarbohydrates.SNP0 10789isinanESTthathassequencesimilaritytotheglycosidehydrolasefamily28proteins.SNP0 15003isinanESTthathasnosimilaritytoanyothersequenceinthedatabase,andSNP0 4905hassequencesimilaritywithamitochondrialacetylornithineaminotransferase.SNP0 17557hassequencesimilaritywithputativeaminoacidtransportersandwasassociatedwithC5sugarsbutnotC6sugars,includingm/z114indicatingthatthisgenemayberelatedmoretocelluloseorgalactoglucomannanthanarabinoglucuronxylanbiosynthesis.Incontrast,SNP0 9148wassignicantforC5+C6,P114144,andm/z114 129

PAGE 130

6817and0 13958weresignicantforbothP114144andm/z114;suggestingtheirinvolvementinarabinoglucuronxylanbiosynthesis.ItisalsointerestingthatseveralSNPswereassociatedwithbothcarbohydrateandlignincontent(Table 5-10 ).SNP0 14949wassignicantforbothC5sugarandlignincontent;0 9169wassignicantforbothP114144andlignincontent;andSNP2 4905wassignicantforbothC5C6(negative)andlignincontent(positive).Itwasshowninchapter3thatligninandcarbohydratewerestrongly,negativelygeneticallycorrelated. Wealsoidentied3SNPsthatareassociatedwithligninandcarbohydratetraits,i.e,SNP0 14949wassignicantforbothC5sugarandlignincontent;0 9169wassignicantforbothP114+144andlignincontent;andSNP2 4905wassignicantforbothC5C6andlignincontent. Lignincontentandvelocitystinesswereshowntobeweaklynegativelycorrelated;SNP2 10207wasnegativelycontributingtolignincontentandpositivelycontributingtothevelocitystiness.Carbohydrateandvelocitystinessareweaklybutsignicantlygeneticallycorrelated;SNP2 1563waspositivelyassociatedwithbothP114+144andvelocitystiness. 130

PAGE 131

ForAssociationMapping,ourmethodisverygeneralcanbeappliedtootherorganisms.Oursimulationresultsshowedthatthepowerofthetestcanbeimprovedwithmoreprecisephenotyping.Futureresearchshouldemphasize: Inconclusion,thisresearchistherstlargescalequantitativeandassociationgeneticstudyofthegeneticarchitectureofeconomicallyimportantquantitativetraitsinloblollypine.Thisinformationisofvaluetothebreedersbecausethegeneticcorrelationbetweentraitssuggestshowdirectselectionforgrowthaectswoodpropertytraits.Usingmarkerassistedbreedingandselectiontocombineidentiedsignicantfavorableallelesforthetargetedspecictrait,breederscandeveloplinesthatcontainthesefavorableallelesandpotentiallyhavebettereconomicvalueinarelativelyshorterbreedingcycle. 131

PAGE 134

Clustersof2,3,4,and5peaksthathasthetop30clonalrepeatabilities Peakgroup VarianceestimateHeritabilitytypeBfamfemalecloneres crnhcnhs 114,138 0.0010760.00270.00580.0643 0.160.140.41 0.8483660,114 0.0032480.01230.01980.2815 0.140.150.39 0.7723860,126 0.0011560.01030.01870.2445 0.140.140.41 0.8435660,144 0.0032540.01180.01560.2727 0.130.150.32 0.8070160,73 0.0026720.0150.02050.3625 0.120.140.29 0.79445150,180 0.0020340.00660.01510.2096 0.120.110.39 0.8737443,44 0.0010540.01850.04280.6073 0.110.110.28 142,43 1.01E-070.010.04070.4701 0.110.070.35 0.9999943,45 0.0018770.01020.03520.4505 0.110.080.35 0.9999943,117 1.01E-070.00930.02690.376 0.110.090.32 0.9999943,141 1.01E-070.00830.02730.3707 0.110.080.33 0.9999943,84 1.01E-070.00860.03110.4113 0.10.070.34 0.9999943,111 1.01E-070.00860.02680.3761 0.10.080.32 0.9999943,112 1.01E-070.00950.0270.3904 0.10.080.31 0.9999943,53 1.01E-070.00870.02780.3899 0.10.080.32 0.9999943,230 1.01E-070.00810.02570.3635 0.10.080.31 0.9999943,246 1.01E-070.00840.02480.3664 0.10.080.3 0.9999943,70 1.01E-070.0070.03120.3977 0.10.060.36 0.9999943,153 1.01E-070.00830.02470.3658 0.10.080.3 0.9999943,182 1.01E-070.00910.02720.4042 0.10.080.3 0.9999931,43 5.15E-080.01410.05680.7666 0.10.070.29 0.9999443,79 1.01E-070.00910.02380.3766 0.10.090.3 0.9999943,110 1.01E-070.010.02530.4013 0.10.090.3 0.9999943,109 2.63E-070.00890.02410.3889 0.10.080.29 0.9999943,60 1.01E-070.00890.0390.5324 0.10.060.39 0.8457643,77 1.60E-060.0080.02310.3617 0.10.080.3 0.9999939,43 1.01E-070.0080.02720.4043 0.090.070.29 0.999843,96 1.01E-070.00820.02490.391 0.090.070.27 0.9999943,58 1.01E-070.010.02660.4371 0.090.080.28 0.9999943,57, 1.01E-070.00770.03380.5055 0.090.050.34 0.9999773,114,144 9.19E-050.00570.00720.0955 0.160.190.39 0.85028760,114,126 0.0010170.00790.01390.1665 0.150.160.46 0.83044360,114,128 0.0016950.00720.01230.1557 0.150.150.45 0.7762960,116,126 0.0005280.00540.01010.1213 0.150.150.45 0.87408460,114,138 0.0020120.00670.00780.1391 0.140.160.26 0.82732460,73,144 0.0019430.01040.01350.2223 0.140.160.32 0.82059260,73,145 0.0013120.00780.01050.1764 0.130.150.31 0.81286857,60,114 0.0016040.00760.01240.1972 0.130.130.34 0.762984 134

PAGE 135

A-1 .Continued Peakgroup VarianceestimateHeritabilitytypeBfamfemalecloneres crnhcnhs 60,69,73 0.0025210.00850.01330.2228 0.130.130.3 0.75690942,43,126 1.01E-070.00480.02270.2281 0.120.070.4 0.99997943,44,162 0.0003350.00860.01980.2658 0.120.110.29 0.99998843,44,116 0.0004540.0080.02020.2685 0.120.10.29 0.99998743,44,111 0.0007020.00850.01990.2752 0.120.110.29 0.99998843,44,63 0.0004980.00830.01950.2704 0.120.110.29 0.99998843,44,172 0.0005360.00820.0190.2688 0.120.110.28 0.99998843,44,182 0.0001510.00910.02010.2899 0.110.110.28 0.99998943,45,98 0.0006560.00470.01890.2228 0.110.070.38 0.99997843,44,301 0.0004820.0080.0180.2611 0.110.110.27 0.99998733,43,44 0.0004040.00830.01920.2743 0.110.10.27 0.99998843,44,78 0.0007230.00820.01850.2703 0.110.10.27 0.99998843,44,142 0.0003450.0080.0190.2684 0.110.10.28 0.99998743,44,107 0.0005760.00810.0170.2561 0.110.110.28 0.99998743,44,274 0.0002740.00830.01750.2662 0.110.110.26 0.99998843,45,206 0.0007180.00460.01530.1996 0.110.080.34 0.99997843,44,163 0.0011050.00840.01620.2667 0.110.110.24 0.99998843,45,110 0.0009320.00540.01560.2156 0.110.090.34 0.99998131,43,45 1.60E-060.00760.03280.3871 0.110.070.34 0.99978942,43,272 1.01E-070.0040.01680.2053 0.110.070.33 0.99997443,44,180 0.0029950.01220.01140.317 0.110.130.14 0.99999243,110,112 1.01E-070.00510.01210.1898 0.10.090.3 0.9999842,43,44,271 1.01E-070.0052050.0150890.180867 0.120.10.3 0.9999842,43,44,145 1.01E-070.0049210.0152550.178109 0.120.090.31 0.9999843,44,46,294 0.0003810.0046250.01120.151438 0.120.110.28 0.9999843,44,112,148 0.000190.0050780.0113080.156643 0.120.110.29 0.9996943,44,205,264 0.0001890.0043760.0113270.149703 0.120.10.29 0.9999840,43,44,128 0.00030.0039190.0126580.153959 0.120.090.31 0.9999743,44,140,271 0.0001350.0050640.0108120.154484 0.120.110.28 0.9999843,44,112,299 0.0003730.0050260.0108570.154941 0.120.110.28 0.9999843,44,133,184 0.0003980.0045060.0106160.147931 0.120.10.29 0.9996542,43,44,124 0.0003270.0066410.0137310.203351 0.120.110.27 0.9999843,44,84,182 1.01E-070.0053050.0131050.178209 0.120.10.3 0.9999843,44,140,275 0.0001750.0050840.0106560.155288 0.120.110.27 0.9999843,44,84,301 7.70E-050.0047460.0119010.161141 0.120.10.29 0.9999843,44,174,244 0.0003040.0045980.0107110.151416 0.110.10.28 0.9999843,44,182,294 1.39E-050.005060.0111790.161248 0.110.110.27 0.9999843,44,182,299 0.0002140.0050430.0106890.158443 0.110.110.26 0.9999843,44,84,303 1.60E-060.0048890.0114710.161069 0.110.110.28 0.99998 135

PAGE 136

A-1 .Continued Peakgroup VarianceestimateHeritabilitytypeBfamfemalecloneres crnhcnhs 43,44,140,164 0.0002330.005510.0095540.15616 0.110.120.27 0.9997143,44,140,181 1.60E-060.0054150.010.156969 0.110.120.25 0.9999840,43,44,261 0.0001910.004380.0114070.15581 0.110.10.27 0.9996343,44,164,294 0.000260.0049950.0093770.150079 0.110.120.27 0.9996843,44,108,142 0.0003130.0043070.0101760.146627 0.110.10.28 0.9999843,44,214,298 0.0003750.0043870.0091840.141838 0.110.110.25 143,44,253,294 0.000190.0045040.010050.149062 0.110.10.26 0.9999843,44,184,298 0.0003380.0043530.0092120.141855 0.110.110.25 0.9986643,44,182,298 9.46E-050.0048190.0096320.152026 0.110.110.25 0.9980443,44,130,298 0.0003760.0042930.0090250.140551 0.110.110.25 0.9996340,43,44,62 0.0001760.0044030.0107940.155754 0.110.10.25 0.9996439,40,43,44 0.0002080.0043650.012150.168411 0.110.090.26 0.9799343,44,57,294 0.0002170.0038910.0143320.185943 0.10.070.35 0.9999760,98,114,126,130 1.29E-040.00420.00660.0821 0.150.170.45 0.8480943,44,45,205,264 6.84E-040.00320.00920.1109 0.130.10.32 0.999543,44,111,128,229 2.81E-040.00270.0080.0966 0.120.10.33 0.9999643,44,45,182,264 5.74E-040.00350.0090.1197 0.120.10.3 0.9930443,44,111,214,229 2.66E-040.00310.00720.0988 0.120.110.29 0.9999743,44,111,145,154 2.29E-040.00290.00730.0969 0.120.10.3 0.9999743,44,84,87,279 2.11E-080.00340.00810.1099 0.120.110.3 0.9995343,44,84,128,296 5.30E-050.00270.00810.1012 0.120.090.32 0.9999643,44,84,145,154 1.02E-070.0030.00820.1043 0.120.10.31 0.9999743,44,86,182,224 3.89E-050.00340.00770.1073 0.120.110.29 0.9999743,44,82,111,182 3.29E-040.00340.00780.1105 0.120.110.29 0.9999743,44,82,182,224 2.57E-040.00330.00770.1082 0.120.110.29 0.9999743,44,94,111,229 2.55E-040.00290.00690.0956 0.120.10.29 0.9999637,43,44,116,154 2.38E-040.00280.00720.0977 0.120.10.28 0.9999639,43,44,127,244 1.98E-040.00270.00820.1039 0.120.090.3 0.9994243,44,45,60,116 3.15E-040.00220.01240.1304 0.120.060.45 0.9579539,43,44,127,301 1.82E-040.00270.00790.1005 0.120.090.29 0.9977437,43,44,154,228 1.98E-040.0030.00690.0982 0.110.110.27 0.9999743,44,83,111,238 2.56E-040.00320.00720.1032 0.110.110.28 0.9999743,44,83,111,279 2.60E-040.00320.0070.1037 0.110.10.28 0.9999743,44,77,82,182 2.05E-040.00330.00710.1072 0.110.110.28 143,44,81,150,180 1.56E-030.00550.00480.1347 0.110.140.17 0.9639843,44,81,195,234 2.40E-040.00320.00670.1046 0.110.10.27 0.999543,44,81,167,234 2.45E-040.00320.00660.1038 0.110.10.27 0.9994931,43,44,109,228 1.30E-070.00440.01280.1746 0.110.090.27 0.9996439,43,44,148,296 8.02E-050.00290.00660.0994 0.110.10.25 0.9994543,44,81,142,234 1.71E-040.00310.00680.1041 0.110.10.27 0.9999743,44,81,150,234 4.21E-040.00350.00580.1085 0.110.110.24 0.9845643,44,70,91,131 1.44E-040.00230.00680.0957 0.110.080.31 0.9999643,44,57,154,239 1.74E-040.00250.00880.1186 0.10.070.34 0.99996 136

PAGE 137

Clustersof2,3,4,and5peaksthathasthetop30clonalrepeatabilitieswithoutpeak43,44 Peakgroup VarianceestimateHeritabilitytypeBfamfemalecloneres crnhcnhs 60,72 0.002430.006880.011330.2084 0.110.110.31 0.7129373,144 0.000360.005710.00630.1154 0.130.170.27 0.8694114,144 0.000840.004670.006310.082 0.160.190.45 0.8666173,126 00.004330.007020.1019 0.130.140.34 0.85360,99 0.001810.006460.011060.1922 0.120.120.32 0.7363260,129 0.002160.007310.00990.1944 0.120.130.28 0.7630160,96 0.001590.006450.010780.2051 0.110.110.3 0.7281960,123 0.00160.007530.00910.1898 0.120.140.25 0.79514114,138 0.001080.002670.005810.0643 0.160.140.41 0.8483660,145 0.002130.007810.010850.1993 0.120.130.3 0.7894960,127 0.002140.007410.011220.1977 0.120.130.31 0.7619960,85 0.001890.006790.011340.2099 0.110.110.28 0.74575150,180 0.002030.006560.01510.2096 0.120.110.39 0.8737460,280 0.001950.006780.008520.1724 0.120.130.28 0.7572560,116 0.001840.007560.012080.1972 0.130.130.34 0.8040257,60 0.001680.009860.01270.3206 0.090.110.21 0.76737180,302 0.001770.004350.011730.2183 0.090.070.37 0.9013873,114 00.006550.009880.1238 0.150.170.39 0.7900260,115 0.002860.008630.013760.2231 0.130.130.34 0.78059124,180 0.003960.007910.014370.2394 0.120.110.33 0.80218180,300 0.003180.006590.012940.1838 0.140.120.35 0.9168560,69 0.003940.008830.016130.2834 0.110.110.31 0.7105660,128 0.002130.008790.014130.2255 0.130.130.35 0.7685760,97 0.003010.00930.013460.2521 0.120.130.3 0.7641360,98 0.001170.011230.014720.2693 0.120.140.3 0.8109760,70 0.002880.010690.015010.2888 0.120.130.29 0.7604660,144 0.003250.011810.015590.2727 0.130.150.32 0.8070160,126 0.001160.01030.018660.2445 0.140.140.41 0.8435660,114 0.003250.012250.019820.2815 0.140.150.39 0.7723860,73 0.002670.014980.020460.3625 0.120.140.29 0.7944560,73,264 0.001270.006910.008750.1585 0.130.150.29 0.8004260,73,89 0.001040.007080.009640.1691 0.130.140.3 0.8031460,73,111 0.000950.006560.009410.1585 0.130.140.31 0.78991124,180,300 0.002250.005270.009590.1269 0.150.140.38 0.8808760,116,126 0.000530.005350.010090.1213 0.150.150.45 0.8740873,114,144 0.000090.005680.007210.0955 0.160.190.39 0.8502960,114,138 0.002010.006680.007820.1391 0.140.160.26 0.8273260,115,144 0.001960.006780.009840.1496 0.140.150.37 0.8239260,128,144 0.001540.00690.009970.1506 0.140.150.37 0.81312180,300,302 0.001880.003550.009760.1912 0.090.070.38 0.972856,60,73 0.001190.008270.011790.2053 0.120.140.31 0.8303160,73,123 0.001070.007550.009290.1695 0.130.150.28 0.8169960,70,98 0.001050.007530.010390.1804 0.130.140.32 0.79442 137

PAGE 138

A-2 .Continued Peakgroup VarianceestimateHeritabilitytypeBfamfemalecloneres crnhcnhs 60,115,126 0.000910.006020.011110.1359 0.150.150.44 0.8516560,97,144 0.002160.007120.009640.1628 0.140.150.33 0.8063360,98,144 0.001160.008250.010.1728 0.140.160.31 0.8386160,69,114 0.003030.006860.01280.181 0.140.130.38 0.7229160,126,128 0.000660.006020.011490.1364 0.150.150.46 0.8326660,73,116 0.001140.00760.011280.1749 0.130.150.34 0.8234960,69,126 0.002010.006080.012790.1629 0.140.130.41 0.7807160,97,126 0.001220.006280.01090.1489 0.140.140.4 0.8219160,114,128 0.00170.007150.012340.1557 0.150.150.45 0.7762960,98,114 0.000990.008440.012140.1768 0.140.160.38 0.8131660,73,128 0.001530.008340.012770.1928 0.140.150.35 0.7877760,70,114 0.002170.008190.013190.1866 0.140.150.4 0.7724160,126,144 0.001290.007680.012210.1616 0.150.160.42 0.8631360,73,98 0.000530.009890.01280.2161 0.130.150.31 0.8281960,73,144 0.001940.010370.013490.2223 0.140.160.32 0.8205960,73,126 0.000630.009350.014830.202 0.140.150.38 0.8410760,73,114 0.001690.010650.016020.2258 0.140.160.37 0.793273,98,114,126 00.003830.006010.0745 0.150.170.43 0.8279160,114,126,283 0.000510.004410.007650.0922 0.150.160.46 0.8291760,114,126,294 0.000640.004470.007580.093 0.150.160.45 0.8367860,114,134,144 0.001560.004920.006740.0981 0.150.170.37 0.8229560,73,144,270 0.001030.005880.007460.1235 0.140.160.32 0.8279960,114,126,153 0.000830.004460.007740.0931 0.150.160.46 0.8325860,61,126,144 0.000780.004750.007270.0977 0.150.160.4 0.8622260,114,126,156 0.000550.004540.008110.0946 0.150.160.48 0.8407360,114,144,178 0.001340.005190.007430.1036 0.150.170.4 0.8248760,72,73,98 0.000460.005930.00820.1333 0.130.150.32 0.8057960,73,98,129 0.00030.006130.007610.1297 0.130.160.3 0.8309660,73,126,275 0.000350.005160.008320.1124 0.140.150.38 0.8313460,73,126,184 0.00030.005280.008210.1129 0.140.160.38 0.8397460,73,84,144 0.001040.005780.00860.128 0.140.150.36 0.8017560,73,116,128 0.000850.00530.008520.1173 0.140.150.38 0.8142360,73,114,224 0.000940.00590.008720.1259 0.140.160.36 0.7885160,72,73,144 0.001270.006190.008560.1374 0.140.150.34 0.7992160,73,126,138 0.000570.006070.006940.1217 0.140.170.26 0.8748160,69,114,128 0.001890.004960.009490.1213 0.150.140.42 0.7377360,73,114,239 0.000990.005760.008670.1243 0.140.150.36 0.7839160,114,138,144 0.001620.005890.006730.1107 0.150.180.3 0.8561560,98,126,128 00.005230.0090.1082 0.150.160.45 0.86027 138

PAGE 139

A-2 .Continued Peakgroup VarianceestimateHeritabilitytypeBfamfemalecloneres crnhcnhs 60,73,115,128 0.001180.005770.009340.1276 0.140.150.39 0.7953960,72,73,114 0.001110.006320.010330.1391 0.140.150.4 0.7743860,69,114,126 0.001690.005450.010730.1284 0.150.140.45 0.7805860,73,128,144 0.001270.007060.009980.1468 0.140.160.37 0.8157760,97,114,126 0.0010.00570.009660.1209 0.150.160.45 0.8115560,73,97,144 0.001690.007250.00960.1554 0.140.160.33 0.8127360,73,114,128 0.001140.007270.011680.1494 0.150.160.42 0.7904260,114,126,144 0.001070.006750.010670.1306 0.160.170.46 0.8540160,114,126,128,164 0.000540.002830.005240.0657 0.140.140.39 0.7939360,73,138,144,163 0.000740.004530.004340.0895 0.130.170.23 0.8946560,73,126,239,264 0.00030.003310.005030.0697 0.140.160.37 0.8346860,70,72,114,145 0.000990.003620.006170.0815 0.150.150.44 0.7700460,73,84,138,144 0.000750.004310.004720.0877 0.130.170.25 0.8464460,97,98,144,172 0.000770.003780.004690.079 0.140.160.33 0.8348860,73,126,139,156 0.000180.003480.005460.0728 0.140.160.38 0.8550360,73,126,129,203 0.00020.003720.005720.0779 0.140.160.38 0.8478460,73,84,126,139 0.000160.003410.0060.0748 0.140.150.41 0.8259160,69,126,128,142 0.0010.002890.006290.0734 0.150.130.45 0.7809260,69,73,137,144 0.000730.004120.006590.101 0.130.140.32 0.8170960,69,73,126,150 0.000750.003230.005440.0792 0.130.140.31 0.7975560,70,73,100,123 0.000540.004010.005460.0876 0.140.150.32 0.8218260,73,116,126,182 0.000280.003410.006620.0781 0.150.140.44 0.8358160,73,126,134,145 0.000490.003780.005390.077 0.140.160.35 0.8611660,70,114,120,128 0.000940.003640.006070.0779 0.150.150.43 0.785560,73,116,126,283 0.000220.003730.006090.0775 0.150.160.41 0.8601560,69,98,126,142 0.00070.003520.006650.0828 0.140.140.41 0.8272860,73,98,128,281 0.000350.004290.006160.0906 0.140.160.36 0.8195360,97,126,143,144 0.000730.003710.005630.0775 0.150.160.41 0.8488160,62,114,126,128 0.00040.003560.006580.0735 0.160.160.51 0.8226460,73,126,128,287 0.000240.004050.006690.0849 0.150.160.41 0.8317373,114,126,128,144 00.003550.005220.0621 0.160.190.48 0.8509860,73,98,144,211 0.000390.005150.006490.1026 0.140.170.33 0.8473760,73,117,126,144 0.000460.004650.006790.0942 0.150.160.38 0.8565860,73,114,144,150 0.000950.00450.005940.0889 0.150.170.32 0.8226960,66,73,126,144 0.000490.004850.007080.0974 0.150.170.39 0.8600460,73,98,126,139 00.004790.007060.095 0.140.170.37 0.8664360,73,126,144,309 0.000520.004940.00720.0983 0.150.170.39 0.8564660,98,114,126,145 0.000140.00470.007240.0895 0.150.170.45 0.86661 139

PAGE 146

GenefunctionsforthediscoveredsignicantSNPsfromtheBLAST 0 10210unknownABR184708E-38160Piceasitchensis0 10210unknownABK246382E-37159Piceasitchensis0 10210hypo.ptn.XP 0022788016E-35151Vitisvinifera0 10210unnamedptn.productCAO157886E-35151Vitisvinifera0 10210hypo.ptn.CAN621196E-35151Vitisvinifera0 10210Antho.5aromaticacyltransferaseEEF354919E-34147Ricinuscommunis0 10373hypo.ptn.XP 0022702133E-61146Vitisvinifera0 10373hypo.ptn.CAN818693E-61146Vitisvinifera0 10373roothairdefect.GTPbindingptn.NP 1993293E-59143Arabidopsisthaliana0 10373ptn.SEY1,put.EEF465661E-58132Ricinuscommunis0 10373hypo.ptn.XP 0022796733E-57131Vitisvinifera0 10373unnamedptn.productCAO456993E-57131Vitisvinifera0 10789polygalacturonasefam.ptn.NP 1941136E-84314Arabidopsisthaliana0 10789glycosidehydrolasefam.28ptn.ACC912501E-83313Arabidopsishalleri0 10789unknownACF864271E-80303Zeamays0 10789hypo.ptn.OsI 02888EAY749901E-80303OryzasativaIndica0 10789unnamedptn.productCAO658172E-80302Vitisvinifera0 10789Polygalacturonaseprecursor,put.EEF513363E-80301Ricinuscommunis0 1126unknownABR179303E-59172Piceasitchensis0 1126unknownABR175707E-58176Piceasitchensis0 1126unknownABK241465E-48153Piceasitchensis0 1126caeicacidO-methyltransferaseCAC216015E-46145Pinuspinaster0 1126unknownABK224909E-24103Piceasitchensis0 1126unknownABK214171E-23101Piceasitchensis0 11470unknownABK226898E-1996.3Piceasitchensis0 11470predictedptn.XP 0023249522E-1378.6Populustrichocarpa0 11470MtrDRAFT AC148819g4v2ABD284673E-1171.2Medicagotruncatula0 11470hypo.ptn.XP 0022720016E-1170.1Vitisvinifera0 11470unknownptn.NP 5682346E-1170.1Arabidopsisthaliana0 11470conservedhypo.ptn.EEF302618E-1169.7Ricinuscommunis0 11475unknownABK248441E-58179Piceasitchensis0 11475unknownABK236905E-55168Piceasitchensis0 11475unknownABK264754E-54164Piceasitchensis0 11475hypo.ptn.OsI 04887EAY769293E-963.2OryzasativaIndica0 11475ptn.bindingptn.,put.EEF315234E-959.3Ricinuscommunis0 11475hypo.ptn.BAD824401E-554.7OryzasativaJaponica0 12862unknownABR182441E-22110Piceasitchensis0 12862unknownABK219041E-22110Piceasitchensis0 12862unknownABK253644E-1481.6Piceasitchensis0 12862unknownABK266762E-1069.3Piceasitchensis0 13240L-aspartateoxidaseXP 0022743616E-66254Vitisvinifera0 13240unnamedptn.productCAO147696E-66254Vitisvinifera0 13240l-aspartateoxidase,put.EEF448662E-62243Ricinuscommunis0 13240predictedptn.XP 0017814271E-61240Physcomitrellapatens0 13240predictedptn.XP 0022985253E-59232Populustrichocarpa0 13240L-aspartateoxidaseNP 5683046E-58228Arabidopsisthaliana0 13565unknownABK255671E-63246Piceasitchensis

PAGE 147

E-1 .Continued QueryHitAccessionevalueScoretaxa 0 13565unknownABK234982E-63245Piceasitchensis0 13565unknownABK257343E-63244Piceasitchensis0 13565unknownABK224745E-61237Piceasitchensis0 13565unknownABR179246E-60234Piceasitchensis0 13565unknownABK241241E-59233Piceasitchensis0 13722unknownABR172325E-50116Piceasitchensis0 13722unknownABK212311E-49116Piceasitchensis0 13722dirigent-likeptn.ABR277213E-2177Piceasitchensis0 13722hypo.ptn.XP 0022693971E-2078.2Vitisvinifera0 13722unknownABK261262E-2072.8Piceasitchensis0 13722hypo.ptn.CAN613154E-2079Vitisvinifera0 13958unknownABR180871E-107392Piceasitchensis0 13958predictedptn.XP 0017845984E-55218Physcomitrellapatens0 13958predictedptn.XP 0017660019E-45184Physcomitrellapatens0 13958predictedptn.XP 0017796225E-1275.9Physcomitrellapatens0 13958unknownACN255673E-1069.7Zeamays0 13958hypo.ptn.OsJ 36445EEE533966E-1068.9OryzasativaJaponica0 14039predictedptn.XP 0017588651E-863.5Physcomitrellapatens0 14039conservedhypo.ptn.EEF327512E-862.4Ricinuscommunis0 14039hypo.ptn.XP 0022782893E-962Vitisvinifera0 14039unknownAAM624981E-860.1Arabidopsisthaliana0 14039unknownptn.NP 5679931E-860.1Arabidopsisthaliana0 14039unnamedptn.productBAD437001E-860.1Arabidopsisthaliana2 1343unknownACN403476E-49197Piceasitchensis2 1343predictedptn.XP 0023199551E-42176Populustrichocarpa2 1343put.laccaseBAD938582E-42176Arabidopsisthaliana2 1343LAC3/oxidoreductaseNP 1805802E-42176Arabidopsisthaliana2 1343laccase90dXP 0023151302E-42176Populustrichocarpa2 1343laccase90bXP 0023121871E-41173Populustrichocarpa0 14435unknownABR165382E-42114Piceasitchensis0 14435unknownABK243562E-42119Piceasitchensis0 14435hypo.ptn.XP 0022667043E-38112Vitisvinifera0 14435hypo.ptn.CAN729833E-38112Vitisvinifera0 14435put.integralmembraneptn.NP 1776589E-38117Arabidopsisthaliana0 14435F18O14.22AAF794451E-37120Arabidopsisthaliana0 14949unknownABR184737E-1894.7Piceasitchensis0 14949unknownABR180969E-861.2Piceasitchensis0 14949unknownABR164446E-858.5Piceasitchensis0 15075hypo.ptn.LOC100193022NP 0011316621E-55220Zeamays0 15075glutaredoxin,grx,put.EEF302714E-55218Ricinuscommunis0 15075hypo.ptn.XP 0022638235E-55218Vitisvinifera0 15075Grx S16-glutaredoxinsubgroupIIACG472252E-54216Zeamays0 15075predictedptn.XP 0017853202E-54216Physcomitrellapatens0 15075Os12g0175500NP 0010662913E-54216Oryzasativa0 15466Os05g0424000NP 0010555921E-859.3Oryzasativa0 15466hypo.ptn.XP 0022780861E-858.9Vitisvinifera0 15466hypo.ptn.isoform2XP 0022858794E-957.4Vitisvinifera0 15466hypo.ptn.isoform1XP 0022858784E-957.4Vitisvinifera

PAGE 148

E-1 .Continued QueryHitAccessionevalueScoretaxa 0 15466aminoacidtransporter,put.EEF281305E-857Ricinuscommunis0 15466aminoacidpermeaseCAH594255E-857Plantagomajor0 16458unknownABR179563E-33145Piceasitchensis0 17557aminoacidtransporter,put.EEF424835E-44182Ricinuscommunis0 17557hypo.ptn.XP 0022638172E-43180Vitisvinifera0 17557unnamedptn.productCAO667092E-43180Vitisvinifera0 17557unknownACN276714E-43179Zeamays0 17557Os02g0700500NP 0010478407E-43178Oryzasativa0 17557neutralaminoacidtransportptn.XP 0023043849E-43177Populustrichocarpa0 18322hypo.ptn.OsI 29896EEC838892E-1690.1OryzasativaIndica0 18322Os08g0518100NP 0010622492E-1690.1Oryzasativa0 18322conservedhypo.ptn.EEF345791E-1277.4Ricinuscommunis0 18322predictedptn.XP 0023090121E-1277.4Populustrichocarpa0 18322put.microtubule-associatedptn.CAE542733E-1276.3Triticumaestivum0 18322hypo.ptn.XP 0022694944E-1275.9Vitisvinifera0 18887unknownABR180021E-108394Piceasitchensis0 18887hypo.ptn.isoform1XP 0022858781E-99366Vitisvinifera0 18887aminoacidtransporterAAD160141E-99366Nepenthesalata0 18887hypo.ptn.isoform2XP 0022858793E-99365Vitisvinifera0 18887hypo.ptn.XP 0022758813E-99365Vitisvinifera0 18887aminoacidpermeaseXP 0023070532E-98362Populustrichocarpa0 1896unknownABK258814E-20102Piceasitchensis0 1896predictedptn.XP 0017537211E-864.3Physcomitrellapatens0 1896predictedptn.XP 0017583953E-859.3Physcomitrellapatens0 1896predictedptn.XP 0017555992E-756.6Physcomitrellapatens0 1974unknownACN398071E-21106Piceasitchensis0 1974unknownABK226531E-21106Piceasitchensis0 3192unknownABK251363E-97245Piceasitchensis0 3192ptn.cbxX,chromosomal,put.EEF482532E-82229Ricinuscommunis0 3192unknownACN336543E-81230Zeamays0 3192hypo.ptn.LOC100193785NP 0011323423E-81230Zeamays0 3192Os02g0266300NP 0010465053E-80228Oryzasativa0 3192hypo.ptn.OsJ 06177EEE567023E-80228OryzasativaJaponica0 366hypo.ptn.OsI 32108EAZ098206E-48194OryzasativaIndica0 366put.heatshockfactorBAD463586E-48194OryzasativaJaponica0 366Os09g0526600NP 0010637266E-48194Oryzasativa0 366heatshockfactorABG734433E-47192Oryzabrachyantha0 366heatshockfactorptn.7ACG371514E-46188Zeamays0 366heatshockfactorptn.7ACG473385E-46188Zeamays0 366hypo.ptn.OsI 32108EAZ098206E-48194OryzasativaIndica0 366put.heatshockfactorBAD463586E-48194OryzasativaJaponica0 366Os09g0526600NP 0010637266E-48194Oryzasativa0 366heatshockfactorABG734433E-47192Oryzabrachyantha0 366heatshockfactorptn.7ACG371514E-46188Zeamays0 366heatshockfactorptn.7ACG473385E-46188Zeamays0 5470unknownABR180563E-47192Piceasitchensis0 5470unknownACN410415E-46188Piceasitchensis

PAGE 149

E-1 .Continued QueryHitAccessionevalueScoretaxa 0 5470unknownABR164381E-39167Piceasitchensis0 5470put.serine-threonineptn.kinaseEEF355203E-35152Ricinuscommunis0 5470hypo.ptn.XP 0022714052E-34149Vitisvinifera0 5470unnamedptn.productCAO431835E-34148Vitisvinifera0 6817unknownABK244391E-27126Piceasitchensis0 6817longevityassurancefactor,put.EEF271327E-24114Ricinuscommunis0 6817hypo.ptn.XP 0022708001E-23113Vitisvinifera0 6817Longevityassurancehomolog3NP 0010310372E-23112Arabidopsisthaliana0 6817ASC1-likeptn.Q8W4Y54E-23111Solanumlycopersicum0 6817ASC1-likeptn.2ACG462044E-22108Zeamays0 7046unknownABR170233E-76288Piceasitchensis0 7046unknownACF852414E-70267Zeamays0 7046hypo.ptn.OsI 26017EAZ038814E-69264OryzasativaIndica0 7046Os07g0485100NP 0010596564E-69264Oryzasativa0 7046Beta-ureidopropionase,put.EEF296214E-68261Ricinuscommunis0 7046predictedptn.XP 0023104856E-68260Populustrichocarpa0 9713predictedptn.XP 0023107822E-33145Populustrichocarpa0 9713unknownptn.NP 1799595E-33144Arabidopsisthaliana0 9713hypo.ptn.AAN599555E-33144Arabidopsisthaliana0 9713hypo.ptn.XP 0022818051E-32143Vitisvinifera0 9713hypo.ptn.CAN614971E-32143Vitisvinifera0 9713unknownABK286702E-31139Arabidopsisthaliana0 9148unknownABK240783E-87325Piceasitchensis0 9148hypo.ptn.XP 0022815572E-80302Vitisvinifera0 9148short-chaindehydrogenase,put.EEF519542E-76289Ricinuscommunis0 9148predictedptn.XP 0022994444E-71271Populustrichocarpa0 9148hypo.ptn.XP 0022629814E-71271Vitisvinifera0 9148predictedptn.XP 0023221831E-70270Populustrichocarpa0 9169predictedptn.XP 0023157551E-77293Populustrichocarpa0 9169predictedptn.XP 0023116361E-74283Populustrichocarpa0 9169hypo.ptn.XP 0022831553E-74282Vitisvinifera0 9169unnamedptn.productCAO712153E-74282Vitisvinifera0 9169heatshockptn.70-interactingptn.EEF452441E-73280Ricinuscommunis0 9169hypo.ptn.CAN614002E-73280Vitisvinifera2 10183unknownACN400327E-45184Piceasitchensis2 10183diseaseresistanceptn.,put.AAG600981E-44183Arabidopsisthaliana2 10183leucine-richrepeatfam.ptn.NP 6834861E-44183Arabidopsisthaliana2 10183predictedptn.XP 0017676171E-43180Physcomitrellapatens2 10183predictedptn.XP 0017677152E-43179Physcomitrellapatens2 10183NBS-LRRdiseaseresist.ptn.homologCAD450293E-43178Hordeumvulgare2 10207unknownABK227564E-75285Piceasitchensis2 10207unknownABK254563E-58229Piceasitchensis2 10207hypo.ptn.XP 0022775313E-46189Vitisvinifera2 10207hypo.ptn.XP 0022758417E-46187Vitisvinifera2 10207unnamedptn.productCAO457017E-46187Vitisvinifera2 10207hypo.ptn.CAN786447E-46187Vitisvinifera2 1023unknownABK263401E-42177Piceasitchensis2 1023unknownABK245171E-42177Piceasitchensis2 1023predictedptn.XP 0023223284E-37159Populustrichocarpa2 102340Sribosomalptn.SA/P40-likeptn.O803777E-37158Daucuscarota2 1023predictedptn.XP 0023033279E-37158Populustrichocarpa2 1023unknownABK928599E-37158Populustrichocarpa

PAGE 150

E-1 .Continued QueryHitAccessionevalueScoretaxa 2 10399unknownABR170953E-68262Piceasitchensis2 10399hypo.ptn.XP 0022756444E-30135Vitisvinifera2 10399unnamedptn.productCAO651914E-30135Vitisvinifera2 10399Indole-3-aceticacid-inducedARG7EEF462174E-29132Ricinuscommunis2 10399SAURfam.ptn.XP 0023261434E-28129Populustrichocarpa2 10399SAURfam.ptn.XP 0023247242E-27126Populustrichocarpa2 1447unknownABR168764E-69265Piceasitchensis2 1447unknownABK235613E-29132Piceasitchensis2 1563hypo.ptn.CAN613812E-34150Vitisvinifera2 1563unknownACN285116E-34148Zeamays2 1563hypo.ptn.OsJ 08791EEE580121E-33147OryzasativaJaponica2 1563avonWillebrandfactorAdomainAAK986951E-33147Oryzasativa2 1563Os02g0806700NP 0010484501E-33147Oryzasativa2 1563hypo.ptn.isoform2XP 0022852712E-33146Vitisvinifera2 2503unknownABK269632E-64248Piceasitchensis2 2503unknownABR163624E-35151Piceasitchensis2 2503Os08g0107100NP 0010607955E-31137Oryzasativa2 2503hypo.ptn.OsI 27526EEC827817E-31137OryzasativaIndica2 2503hypo.ptn.ACG367574E-29131Zeamays2 2503predictedptn.XP 0023099161E-28129Populustrichocarpa2 2718unknownABK242424E-49198Piceasitchensis2 2718cellwallbeta-glucosidaseAAS979603E-41172Secalecereale2 2718Os03g0749300NP 0010512753E-40168Oryzasativa2 2718beta-D-glucosidaseCAA070704E-40168Tropaeolummajus2 2718beta-D-glucanexohydrolaseAAC491707E-401672 2718unnamedptn.productCAO674371E-39166Vitisvinifera2 3463Activatingsignalcointegrator,put.EEF521735E-37158Ricinuscommunis2 3463hypo.ptn.LOC100193257NP 0011318784E-36155Zeamays2 3463predictedptn.XP 0023179801E-35154Populustrichocarpa2 3463hypo.ptn.XP 0022759441E-34150Vitisvinifera2 3463unnamedptn.productCAO625441E-34150Vitisvinifera2 3463hypo.ptn.CAN675441E-34150Vitisvinifera2 3542hypo.ptn.XP 0022758292E-64249Vitisvinifera2 3542Stachyosesynthaseprecursor,put.EEF317663E-64249Ricinuscommunis2 3542predictedptn.XP 0023305893E-64249Populustrichocarpa2 3542unknownABK957343E-64249Populustrichocarpa2 3542unknownABK947753E-64249Populustrichocarpa2 3542alkalinealphagalactosidaseIAAZ814241E-60237Cucumissativus2 3989WRKYtranscriptionfactor1ACA048881E-860.8Piceaabies2 4455unknownABK262008E-91337Piceasitchensis2 4455hypo.ptn.XP 0022660773E-30136Vitisvinifera2 4455unknownABK259391E-29134Piceasitchensis2 4455GTPbindingptn.,put.EEF423001E-29134Ricinuscommunis2 4455hypo.ptn.XP 0022727597E-29131Vitisvinifera2 4455hypo.ptn.XP 0022691887E-29131Vitisvinifera2 4905predictedptn.XP 0017839336E-66254Physcomitrellapatens2 4905predictedptn.XP 0017524344E-61238Physcomitrellapatens

PAGE 151

E-1 .Continued QueryHitAccessionevalueScoretaxa 2 4905mitoch.Acetylo.aminotransferaseO048667E-61237Alnusglutinosa2 4905hypo.ptn.XP 0022721581E-59233Vitisvinifera2 4905predictedptn.XP 0023347511E-58229Populustrichocarpa2 4905predictedptn.XP 0023237662E-58229Populustrichocarpa2 5026unknownABR177485E-24114Piceasitchensis2 5026hypo.ptn.OsJ 09254EAZ254383E-1068.6OryzasativaJaponica2 5026hypo.ptn.OsI 09817EAY883623E-1068.6OryzasativaIndica2 5026Os03g0127500NP 0010488323E-1068.6Oryzasativa2 5026hypo.ptn.XP 0022693634E-964.7Vitisvinifera2 5026hypo.ptn.OsJ 28152EEE690914E-964.7OryzasativaJaponica2 5064put.mitogen-activatedptn.kinaseACA048569E-73277Piceaabies2 5064put.mitogen-activatedptn.kinaseACI877961E-40170Cupressussempervirens2 5064unknownABK244882E-40169Piceasitchensis2 5064unnamedptn.productCAO710821E-1484.3Vitisvinifera2 5064bigmapkinase/bmk,put.EEF454107E-1481.6Ricinuscommunis2 5064hypo.ptn.XP 0022797195E-1379Vitisvinifera2 5438unknownABK245514E-94348Piceasitchensis2 5438hypo.ptn.XP 0022732708E-68261Vitisvinifera2 5438predictedptn.XP 0023087967E-65251Populustrichocarpa2 5438Os03g0194900/membraneptn.NP 0010492547E-65251Oryzasativa2 5438hypo.ptn.OsI 10369EAY888902E-64249OryzasativaIndica2 5438hypo.ptn.XP 0022642447E-64248Vitisvinifera2 6434unknownABK259889E-31137Piceasitchensis2 6434hypo.ptn.XP 0022714732E-1587Vitisvinifera2 6434unnamedptn.productCAO488692E-1587Vitisvinifera2 6434predictedptn.XP 0023230732E-1586.7Populustrichocarpa2 6434unnamedptn.productCAO430872E-1586.7Vitisvinifera2 6434unnamedptn.productCAO427492E-1586.7Vitisvinifera2 8611unknownABK232022E-32143Piceasitchensis2 8611predictedptn.XP 0017577886E-30134Physcomitrellapatens2 8611predictedptn.XP 0017776702E-27126Physcomitrellapatens2 8611hypo.ptn.OsJ 09149EEE581984E-1172.4OryzasativaJaponica2 8611Os03g0113500NP 0010487364E-1172.4Oryzasativa2 8611hypo.ptn.AAO170164E-1172.4Oryzasativa2 9017CytochromeP450716B2Q50EK01E-85320Piceasitchensis2 9017CytochromeP450716B1Q50EK12E-85319Piceasitchensis2 9017unknownABK269681E-79300Piceasitchensis2 9017unknownABR167079E-76287Piceasitchensis2 9017hypo.ptn.XP 0022646434E-68261Vitisvinifera2 9017unnamedptn.productCAO665214E-68261Vitisvinifera2 9102hypo.ptn.XP 0022711476E-965.5Vitisvinifera2 9102predictedptn.XP 0023015181E-860.8Populustrichocarpa2 9102hypo.ptn.OsJ 01438EAZ115704E-959.3OryzasativaJaponica2 9102hypo.ptn.OsI 01543EAY736554E-959.3OryzasativaIndica2 9102Os01g0300600NP 0010428184E-959.3Oryzasativa2 9102pM5ptn.-likeBAD878234E-959.3OryzasativaJaponica2 9457unnamedptn.productCAO398785E-63244Vitisvinifera2 9457hypo.ptn.XP 0022672751E-62243Vitisvinifera2 9457predictedptn.XP 0023233582E-62243Populustrichocarpa2 9457predictedptn.XP 0023092332E-61239Populustrichocarpa2 9457hypo.ptn.XP 0022835654E-61238Vitisvinifera2 9457predictedptn.XP 0017749666E-61238Physcomitrellapatens

PAGE 152

E-1 .Continued QueryHitAccessionevalueScoretaxa 2 9683unknownABK268682E-91339Piceasitchensis2 9683hypo.ptn.XP 0022652517E-40168Vitisvinifera2 9683hypo.ptn.XP 0022639002E-38163Vitisvinifera2 9683conservedhypo.ptn.EEF350906E-38161Ricinuscommunis2 9683predictedptn.XP 0023018211E-37160Populustrichocarpa2 9683predictedptn.XP 0017810612E-37159Physcomitrellapatens2 9851F-boxandwd40domainptn.,put.EEF407374E-29132Ricinuscommunis2 9851predictedptn.XP 0023128501E-28130Populustrichocarpa2 9851hypo.ptn.XP 0022699191E-26124Vitisvinifera2 9851F-boxptn./WD-40repeatptn.NP 5684085E-26122Arabidopsisthaliana2 9851F-boxdomaincontainingptn.ACG356271E-21107Zeamays2 9851F-boxdomaincontainingptn.ACG334451E-21107ZeamaysUMN 5432unknownABK252453E-71271PiceasitchensisUMN 5432unknownABK246665E-71271PiceasitchensisUMN 5432predictedptn.XP 0017717267E-58227PhyscomitrellapatensUMN 5432predictedptn.XP 0017690971E-57226PhyscomitrellapatensUMN 5432predictedptn.XP 0017840151E-55219PhyscomitrellapatensUMN 5432predictedptn.XP 0017674275E-53211PhyscomitrellapatensUMN 5830predictedptn.XP 0023136834E-44181PopulustrichocarpaUMN 5830predictedptn.XP 0017531254E-41171PhyscomitrellapatensUMN 5830hypo.ptn.ACG270535E-41170ZeamaysUMN 5830hypo.ptn.XP 0022789519E-41169VitisviniferaUMN 5830unnamedptn.productCAO474829E-41169VitisviniferaUMN 5830alanyl-tRNAsynthetase,put.EEF282891E-39166RicinuscommunisUMN 5942unknownABK226843E-1379.3PiceasitchensisUMN 5942predictedptn.XP 0023021616E-1275.1PopulustrichocarpaUMN 5942predictedptn.XP 0023067171E-1174.3PopulustrichocarpaUMN 5942conservedhypo.ptn.EEF380655E-1068.6RicinuscommunisUMN 5942hypo.ptn.XP 0022634545E-1068.6VitisviniferaUMN 5942unknownptn.NP 1927272E-967ArabidopsisthalianaUMN 6891unknownABR183241E-552.8PiceasitchensisUMN 801predictedptn.XP 0017700157E-36153PhyscomitrellapatensUMN 801hypo.ptn.XP 0022632409E-33143VitisviniferaUMN 801unnamedptn.productCAO177409E-33143VitisviniferaUMN 801predictedptn.XP 0017838493E-32141PhyscomitrellapatensUMN 801predictedptn.XP 0017555555E-31137PhyscomitrellapatensUMN 801predictedptn.XP 0023103221E-28129PopulustrichocarpaCL1052ctg1unknownABR169511E-108395PiceasitchensisCL1052ctg1hypo.ptn.OsJ 05018EEE561386E-53211OryzasativaJaponicaCL1052ctg1unknownACN269172E-52209ZeamaysCL1052ctg1hypo.ptn.OsI 05484EAY841017E-52207OryzasativaIndicaCL1052ctg1Nitric-oxidesynthase,put.EEF515644E-51205RicinuscommunisCL1287ctg1unknownABK226471E-72277PiceasitchensisCL1287ctg1unknownABK247253E-58229PiceasitchensisCL1287ctg1unknownABK270445E-57225PiceasitchensisCL1287ctg1unknownABK210491E-43181PiceasitchensisCL1287ctg1unknownABK269446E-42175PiceasitchensisCL1287ctg1unknownABK269532E-41173Piceasitchensis

PAGE 153

E-1 .Continued QueryHitAccessionevalueScoretaxa CL148ctg1unknownABK245513E-1792.8PiceasitchensisCL148ctg1hypo.ptn.XP 0022801752E-1483.6VitisviniferaCL148ctg1hypo.ptn.CAN673342E-1483.6VitisviniferaCL148ctg1auxin-responsiveptn.,put.NP 1914662E-1483.2ArabidopsisthalianaCL148ctg1hypo.ptn.CAN665256E-1481.6VitisviniferaCL148ctg1Auxin-inducedroot12precursorEEF303788E-1481.3RicinuscommunisCL1532ctg1unknownptn.NP 1872661E-144515ArabidopsisthalianaCL1532ctg1conservedhypo.ptn.EEF453161E-142508RicinuscommunisCL1532ctg1hypo.ptn.LOC100279591NP 0011460601E-139497ZeamaysCL1532ctg1hypo.ptn.ACG287021E-139497ZeamaysCL1532ctg1predictedptn.XP 0023144091E-138497PopulustrichocarpaCL1532ctg1predictedptn.XP 0023118291E-138497PopulustrichocarpaCL1595ctg1Lhca4ptn.CAA789321E-143510PinussylvestrisCL1595ctg1Lhca4ptn.CAA789011E-140500PinussylvestrisCL1595ctg1put.chloro.a/b-bindingptn4CAC844911E-139498PinuspinasterCL1595ctg1unknownABK242361E-138495PiceasitchensisCL1595ctg1unknownACN405981E-138494PiceasitchensisCL1595ctg1put.chlorophyllA/Bbindingptn.EEF478471E-114415RicinuscommunisCL2076ctg1unknownACN399651E-134482PiceasitchensisCL2076ctg1predictedptn.XP 0023329951E-124447PopulustrichocarpaCL2076ctg1put.Celldivisionptn.EEF419891E-122443RicinuscommunisCL2076ctg1lament.temp.-sensitiveH2BNP 0011207211E-122443ZeamaysCL2076ctg1lament.temp.-sensitiveH2ANP 0011207201E-122442ZeamaysCL2076ctg1FtsH-likeptn.PftfprecursorAAD172301E-122442NicotianatabacumCL3061ctg1unknownABR184341E-118427PiceasitchensisCL3061ctg1unknownABK246425E-33145PiceasitchensisCL3061ctg1unnamedptn.productCAO706685E-31139VitisviniferaCL3061ctg1hypo.ptn.XP 0022657458E-31138VitisviniferaCL3061ctg1predictedptn.XP 0023142602E-30137PopulustrichocarpaCL3061ctg1calmodulinbindingptn.,put.EEF486725E-30135RicinuscommunisCL3375ctg1unknownABR180111E-72277PiceasitchensisCL3375ctg1hypo.ptn.XP 0022817824E-42175VitisviniferaCL3375ctg1unknownptn.NP 1979412E-40170ArabidopsisthalianaCL3375ctg1predictedptn.XP 0023242521E-38164PopulustrichocarpaCL3375ctg1conservedhypo.ptn.EEF459252E-38163RicinuscommunisCL3375ctg1hypo.ptn.OsI 02082EEC707097E-37158OryzasativaIndicaCL4776ctg1hypo.ptn.XP 0022809906E-74281VitisviniferaCL4776ctg1unnamedptn.productCAO234426E-74281VitisviniferaCL4776ctg1hypo.ptn.CAN734966E-74281VitisviniferaCL4776ctg1hypo.ptn.OsJ 31500EEE509515E-73278OryzasativaJaponicaCL4776ctg1put.cataboliterepressorptn.EEF312133E-72276RicinuscommunisUMN CL298c140Sribosomalptn.S25,put.EEF385455E-30135RicinuscommunisUMN CL298c1unknownABK238397E-30134PiceasitchensisUMN CL298c1unknownABK933341E-29134PopulustrichocarpaUMN CL298c140Sribosomalptn.S25AAQ227263E-29132GlycinemaxUMN CL298c1unknownACJ839063E-29132MedicagotruncatulaUMN CL298c1unknownABK211384E-31139PiceasitchensisCL1052ctg1Os02g0104700NP 0010456146E-53211OryzasativaCL4776ctg1Os10g0412100NP 0010645875E-73278Oryzasativa

PAGE 154

154

PAGE 155

155

PAGE 156

156

PAGE 157

157

PAGE 158

158

PAGE 159

159

PAGE 160

160

PAGE 161

161

PAGE 162

162

PAGE 163

163

PAGE 164

164

PAGE 165

165

PAGE 166

166

PAGE 167

167

PAGE 168

168

PAGE 169

169

PAGE 170

170

PAGE 171

XiaoboLiwasbornSeptember16thinXiangxiang,Hunan,China,whichhasbeencreditedasoneofthehistoricalculturecentersofChina.XiaobogothisBachelordegreeinWoodSciencefromBeijingForestryUniversityin1999.HegothisMasterdegreeinWoodSciencefromChineseAcademyofForestryin2002.HethenwenttoLouisianaStateUniversityandgraduatedwithaMasterdegreeinForestryin2004.HestartedhisstudyatUFfromMay2004andHegothisMasterdegreeinStatisticsfromUniversityofFloridain2007.HegraduatedwithhisPhDdegreeinstatisticalgeneticsfromSchoolofForestResourcesandConservationinDecember2009.Xiaobohasbeenveryinterestedinapplyingstatisticalgeneticstoimprovedesiredtraitsineconomicallyimportantcrops.Hisresearchinterestinthefuturewillbeombiningquantitativegeneticsandmolecularmarkerassociationmappingforcropspeciestobreedbettercropgenotypestoghtpovertyandhunger,andtoimprovethelivesofpeople. 171