【作者】 马建强；

【导师】 陈亮；

【作者基本信息】 中国农业科学院 ， 茶学， 2013， 博士

【摘要】 茶树是我国的一种重要经济作物。近年来，随着产业技术创新和人们生活水平提高，茶树品质育种日益受到重视。但茶树的重要品质性状大多为复杂的数量性状，受微效多基因和环境的共同作用，对其进行遗传改良十分困难；并且茶树作为多年生异交木本作物，世代周期较长、遗传背景复杂，使用传统育种方法进行新品种选育，效率较低。利用分子标记技术进行遗传图谱构建和数量性状位点（QTL）定位，找到与目标性状紧密连锁或共分离的分子标记，借助标记辅助选择可以极大地缩短育种周期、提高育种效率。本研究以茶树变种间杂交构建的F1分离群体，采用微卫星（SSR）标记技术和简化基因组测序技术，结合“拟测交”策略分别构建双亲遗传图谱以及双亲整合图谱，并对控制茶树新梢咖啡碱和可可碱含量、以及芽叶形态等性状的QTL进行了定位分析。主要研究结果如下：1.对dbEST数据库的8008条茶树表达序列标签（EST）序列进行拼接，获得2982个unigene，利用SSRIT软件在462个unigene中搜索到561个SSR位点，检出率为15.5%；包含二、三核苷酸重复基元的SSR出现频率最高，分别占总数的45.5%和19.1%；根据SSR位点两侧保守序列成功设计213对引物，经过筛选后，最终获得104对能够扩增出目的条带的EST-SSR引物。2.利用亲本和6个F1群体单株对不同来源的1464对SSR引物进行初筛，共筛选出424对能够检测到多态性的引物，占总数的29%，其中EST-SSR421个，基因组SSR3个；多态性引物中母本信息来源位点70个（16.5%），父本信息来源位点168个（39.6%），双亲信息来源位点186个（43.9%）；将多态性引物用于F1群体分析，卡方检验结果显示有130个（30.7%）位点的基因型频率偏离预期的孟德尔频率。3.采用特异性长特扩增特段测序（SLAF-seq）技术对亲本和150个F1单株进行简化基因组测序，获得130.35M reads，聚类分析后共形成101091个单核苷酸多态性（SNP）标签，其中25014个具有多态性，占总数的24.7%；对多态性标签进行基因型分析，过滤父母本信息缺失、完整特过低和不适合F1群体作图的多态性标签，最终获得6042个有效的SNP位点；按分离类型，选择800个SNP标记用于F1群体分析，其中母本信息来源位点289个（36.1%），父本信息来源位点276个（34.5%），双亲信息来源位点235个（29.4%），卡方检验结果显示有164个（20.5%）位点的基因型频率偏离预期的孟德尔频率。4.采用“拟测交”策略分别构建双亲遗传图谱，其中母本遗传图谱包含15个连锁群，577个标记（SSR，226；SNP，351），总覆盖遗传距离1427.4cM，平均图距2.5cM，连锁群长特范围73.8cM-115.0cM；父本遗传图谱包含739个标记（SSR，313；SNP，426），分属于15个连锁群，图谱总长1488.1cM，平均图距为2.0cM，连锁群长特范围73.2cM-126.7cM；利用同源位点进行双亲图谱整合，最终得到的整合图谱共15个连锁群，与茶树染色体数一致，包含1101个标记（SSR，373；SNP，728），图谱覆盖基因组长特1632.8cM，相邻标记间最大距离19.6cM，最小间距0.1cM，平均图距为1.5cM，连锁群长特范围80.2cM-184.8cM。5.对亲本及F1群体的6个重要性状进行连续两年的鉴定，统计分析表明所有性状在亲本间存在显著差异，在F1群体中呈连续变异，近正态分布，且存在明显的双向超亲分离。利用整合双亲整合图谱，结合两年表型数据，采用restricted MQM复合区间作图法共检测到18个控制相关性状的主效QTL，单个QTL贡献率10.7%-23.0%，其中位于第5连锁群上控制咖啡碱含量的QTL位点和位于第7连锁群上控制叶形指数的QTL位点比较稳定，在两年及两年联合分析中都能检测到。更多还原

【Abstract】 The tea plant is an economically important beverage crop cultivated in tropical and subtropicalregions. Recently, as the innovation of industrial technology and the rise of living standards, qualitytraits are becoming increasingly important in current tea plant breeding. However, improving these traitsgenetically is extremely difficult because most of them are controlled by quantitative traits loci (QTL).Moreover conventional breeding in tea plant is laborious, time-consuming due to its self-incompatibilityand long generation cycles. Genetic map construction and QTL mapping studies provide an abundanceof DNA marker-trait associations, and these trait linked markers have enormous potential to improve theefficiency and precision of conventional plant breeding via marker-assisted selection (MAS). In thisstudy, three genetic maps were constructed based on a F1population of inter-varietal hybrids of tea plant,using SSR markers and specific length amplified fragment sequencing (SLAF-seq), and QTLscontrolling caffeine (CAF) and theobromine (TBR) contents, leaf length (LL), leaf width (LW), leafindex (LL/LW) and length of ‘Two and a bud’(BL) were detected in the parental integrated map byrestricted multiple-QTL model (MQM) method. The results of this dissertation are as follows:1. A total of2982unigenes were predicted from8008tea plant ESTs, representing approximately4238kb of putative functional tea plant transcriptome. SSR searching detected561microsatellite lociamong462(15.5%) unigenes. The mined microsatellites were classified by repeat motif and size, andthe result showed that di-and tri-nucleotide repeats were the most common repeat types, accounting for45.5%and18.9%of the total. According to the flanking sequence of SSR,213primers weresuccessfully designed, of which104amplified unambiguous target bands with expected sizes.2. A set of1465SSR primers developed from different Camellia species were screened in the twoparents and six F1individuals. Of them,421EST-derived SSR and three genomic SSR yieldedpolymorphic amplicons. Paired comparison of allelic patterns obtained in parents showed that70(16.5%) loci were heterozygous in maternal parent while homozygous in paternal parent,168(39.6%)loci were heterozygous in paternal parent while homozygous in maternal parent, and186(43.9%) lociwere heterozygous in both parents. A chi-square test was performed on marker genotyping data for F1population, resulting in that130(30.7%) markers were distorted from Mendelian segregation ratios.3. Solexa sequencing of reduced-representation libraries generated a total of~130.35million readsfrom parents and150F1progenies. Reads were aggregated into101091SLAF tags, of which25014(24.7%) were polymorphic. After SNP genotype calling and removal of tags with incomplete data orunsuitable for CP population (outbreeder full-sib family), a total of6042SNP markers were finallyidentified. Of them,800markers were selected and analyses in F1population, and the result showed that289(36.1%) markers were available to for maternal map construction,276(34.5%) markers for paternalmap, and235(29.4%) markers for both parental maps. Chi-square test showed that164(20.5%)markers were distorted from Mendelian segregation ratios.4. Parental genetic maps were constructed separately based on pseudo-testcross theory. Thematernal map consisted of577markers (SSR226, SNP351), spaning a total length of1427.4cM with an average distance of2.5cM. The maternal map consisted of739markers (SSR313, SNP426),spaning a total length of1427.4cM with an average distance of2.5cM. Both maps contained15linkage groups, which corresponds to the basic chromosome number of tea plant. Based on homologousloci in parental maps, the consensus map was constructed, in which1101markers (SSR373, SNP728)were mapped on15linkage groups covering a total map length of1632.8cM with an average distanceof1.5cM.5. Significant differences were observed between the parents for all traits. The F1populationdisplayed continuous distribution of phenotypes and transgressive segregation. A total of18QTL with10.7-23.0%of the total phenotypic variance were identified. Of them, two common QTL, one for CAFcontent on LG05linkage group and one for LL/LW on LG07, were detected in different environments.更多还原