【作者】 李博；

【导师】 李火根；

【作者基本信息】 南京林业大学 ， 林木遗传育种， 2013， 博士

【摘要】 种群遗传结构及其影响因子的研究是保护遗传学的重要内容，它不仅是探讨物种形成以及植物适应和进化机制的基础，也是针对物种制定有效保护策略和措施的依据。影响植物居群内遗传结构的生态和进化因素很多，而遗传变异的空间分布是种群遗传结构的重要特征之一。空间遗传结构是影响群体短期进化过程的重要因素，开展种群空间遗传结构研究有助于了解种群动态，对于濒危植物的保护策略的制订有重要的指导意义。鹅掌楸(Liriodendron chinense (Hemsl.) Sarg.)为木兰科鹅掌楸属植物，星散分布于我国华东、中南及西南地区以及越南北部。由于鹅掌楸种群规模小，生境片段化，自然结实率低，天然更新不良，从而被列为我国二级保护树种。本文以1个残存的天然种群和2个遗传结构清楚的人工群体为研究对象，采用SSR分子标记进行亲本分析与谱系重建，进而揭示鹅掌楸空间遗传结构，期望为鹅掌楸的保护策略制定提供理论依据。主要研究结果如下：三种常用亲本分析软件效率的比较。在亲本分析中，常用基于最大似然法的软件(如CERVUS, COLONY和PAPA)进行亲本推定，但不同软件各有其特点，其分析结果与亲本推定效率有时相差较大。为比较三种常用亲本分析软件(CERVUS3.0, COLONY2.0和PAPA2.0)的效率及特点，本文以包含278个鹅掌楸潜在亲本及90个鹅掌楸已知亲本的子代群体为材料，利用13对EST-SSR位点信息，分析子代亲本并与其真实亲本进行对照，进而评价三种亲本分析软件的效率，期望为林木亲本分析软件选择提供参考。研究结果表明，13个SSR位点在实验群体中检测出的等位基因数(A)、观测杂合度(Ho)及Shannon多样性指数(I)平均值分别为5.23，0.4766和1.4636，表明该批引物多态性高，适合用于亲本分析。就亲本分析效率而言，总体上看，谱系清楚的子代远高于谱系不清的子代，单亲分析远高于双亲分析。三种软件的分析效率及特点各有差别。对于谱系不清的林下更新子代苗木，双亲分析时，三种软件的分析效率差别不大。但单亲分析时，CERVUS与COLONY软件分析效率较高，其亲本推定的准确率分别为78.87%和69.90%，高于PAPA软件的分析效率。而对于谱系清楚的控制授粉子代，双亲分析时，采用PAPA软件对异交子代亲本推定准确率为34.72%，高于CERVUS及COLONY软件；而CERVUS与COLONY软件则适合于自交子代的亲本分析，其亲本推定的准确率分别可达83.33%及81.2%，远高于PAPA软件。广西猫儿山鹅掌楸天然群体谱系重构。利用17对SSR分子标记结合叶绿体PsbA-TrnH片段序列信息，对广西猫儿山鹅掌楸天然群体进行谱系重构。研究结果表明：该群体不同个体间在单碱基（A/T）的重复序列中其重复次数不同，猫儿山鹅掌楸天然群体具有4种不同的单倍型。对49个未成年子代进行亲本分析，共有47个子代推断获得了亲本。其中，有24个子代的单倍型与候选亲本的单倍型一致，有17个子代可推断出双亲。综合上述信息对该鹅掌楸天然群体谱系关系进行了重构。共构建了12对同胞组合，其中全同胞1对，同父异母半同胞4对，同母异父半同胞4对，以及共同亲本不能确定是母本还是父本的半同胞3对。鹅掌楸群体空间遗传结构。利用SSR标记分析了3个鹅掌楸群体的空间遗传结构。遗传背景清楚的人工群体的空间遗传结构强度值及其子代的空间遗传结构强度值分别为0.01341，0.02493和0.01762，0.00989。天然群体及其子代群体的空间遗传结构强度值为0.02700，0.01423。鹅掌楸的空间遗传结构属较高水平，且与已报道的具有类似生活史特征物种的空间遗传结构相吻合。鹅掌楸种群基因流格局。基因流是影响植物种群遗传结构的重要因子。分别采用亲本分析方法以及空间遗传结构(SGS)分析方法对鹅掌楸基因流进行直接与间接估计。结果表明，鹅掌楸种子流及花粉流大小在不同群体间差别较大。基因流间接估计结果显示，鹅掌楸3个不同群体的基因流及花粉流随居群的成年个体密度增大而减小。在种群密度较大的两个人工群体中，鹅掌楸种子的传播距离大于花粉的传播距离，种子流对基因流的贡献程度较大；而在种群密度较小的天然群体中，其结果则正好相反。广西猫儿山鹅掌楸天然种群空间分布格局。为进一步探讨鹅掌楸濒危的机制，从而为鹅掌楸天然种群保护提供参考依据，本文还分析了广西桂林猫儿山鹅掌楸天然种群结构。以种群生命表及生存分析理论为基础，以林木径级结构代表龄级结构，编制鹅掌楸种群静态生命表，分析种群结构的动态变化特点。总体上，鹅掌楸种群龄级结构为金字塔型：幼龄个体数量较多，中龄个体数居中，而老龄个体则相对数量较少，表现为增长型种群。在第Ⅳ龄级时该种群出现死亡高峰，其种群存活曲线属Deevey-Ⅱ型。表明该天然种群具有前期种群数量快速减少，中后期稳定，末期衰退的特点。同时，应用O-ring函数对该鹅掌楸天然种群不同龄级的空间分布格局及龄级间的关系进行分析。分析发现，从小尺度空间（小于24内）角度看，该群体的空间分布格局表现为聚集分布；在中等尺度空间（24～48m）范围内，该群体则表现为随机分布；从较大尺度空间（49～88m）范围看，该群体的空间分布格局趋于均匀分布；而当空间尺度大于88m后，又表现为随机分布。鹅掌楸不同龄级立木的空间分布格局存在差别。成年个体及幼龄个体在小尺度空间范围内表现为聚集分布，在中尺度及大尺度空间范围内表现为随机分布。老龄个体在不同的空间尺度下均表现为随机分布。在小尺度上，不同龄级立木之间表现为正关联；而在较大尺度上，不同龄级立木之间表现为不关联。鹅掌楸亲、子群体遗传多样性比较。采用14个SSR位点检测了3个鹅掌楸群体亲本及子代群体的遗传多样性。结果表明，与木兰科其他物种相比，鹅掌楸群体具有较高的遗传多样性，且亲、子代间的遗传多样性差别不大。更多还原

【Abstract】 Population genetic structure is a major aspect in conservation genetics. It is not only thebasis of the evolution of plants, but also the foundation for the protection of endangered species.Although there are many factors that affect the evolution of plant population, the spatial geneticstructure (SGS) of plants is one of the important factors, for SGS plays key roles in theshort-term evolution. Therefore, studies on SGS may help us to know the population dynamics,as well as to develop effective strategies for ex-situ conservation of endangered species.Liriodendron chinense (Hemsl.) Sarg, a species from the genus of Liriodendron inMagnoliaceae family, is now recognized as an endangered species due to its low ratio of fullseeds, hardness in natural regeneration, small population size and fragmentated habitats. In thispaper, the fine-scale spatical genetic structure of L. chinense was analyzed for three populations(one natural population and two artificial populations) using microsatellite markers. This studyaimed to answer the three following quentions:(1) Is there a strong spatial genetic structurewithin the isolated natural population?(2) Whether the sub-populations grouped by age classcould also exhibit obvious spatial genetic structure?(3) What about the gene flow in the isolatednatural population? In one word, the final goal of this paper is to explore the mechanism of thespatial genetic structure in L. chinense. The main results are as follows:At present, softwares in the basis of maximum likelihood assignment, such as CERVUS,COLONY and PAPA, were often used to analyze parentage. However, the outputs generatedfrom different softwares were much different though applied to the same experimental population.These were mostly attributed to the different characteristics of softwares. In order to compare thepower of three softwares mentioned above on parentage assignment,13microsatellites had beenused to detect an experimental population of Liriodendron which includes278potential parentsand90progenies whose real parents were unknown. Three softwares (CERVUS3.0,COLONY2.0, PAPA2.0) were used to assign parentage for each progeny, the power of threesoftwares were evaluated by checking the assigned parentage to their real parents for eachprogeny. The results are as follows: the average values of number of alleles (A), effective numberof alleles (Ne) and Shannon diversity index (I) in this experimental population were5.23,0.4766,and1.4636respectively, indicating that these13SSR loci were high polymorphic and suitablefor parentage analysis. As a whole, the power of parentage assignment for control pollinationprogenies was higher than that for open pollination progenies, and paternity analysis was betterthan parentage analysis. Each of these three softwares had its specific characteristics. When theywere applied to analyze progenies whose pedigree were not clear, no significant differenceamong three softwares was found when appling to parentage analysis. However, when appling topaternity analysis, the differences among three softwares in terms of power were significant. Theparentage assignment accuracy of CERVUS and COLONY were78.87%and69.90% respectively, much higher than that of PAPA. As for control pollination progeny whose pedigreewas clear, the parentage assignment accuracy of PAPA was34.72%, higher than other twosoftwares when applied to outcrossing progenies. Nevertheless, if applied to selfing progenies,the parentage assignment accuracy of CERVUS and COLONY were83.33%and81.2%respectively, much higher than that of PAPA. Our findings might be helpful in software choicefor parentage analysis in plants.To investigate the spatial genetic structure of Liriodendron chinense, three populations weretaken as target populations, one natural population from Maoershan Reserve, Guilin, Guangxiprovince and two artificial populations from Nanjing and Zhengjiang, Jiangsu provincerespectively.The natural population in Maoershan Reserve is an isolated population and has formedcomplex pedigree relationship in its long-term breeding process. To reconstruct the pedigree ofthis population, all the individuals within this population were detected with17SSR locicombined with chloroplast DNA sequences of PsbA-TrnH gene. Single nucleotide mutation (A/T)among individuals was detected checked. Altogether, there were4types of haplotype in thispopulation. Among49offsprings,47offsprings were successfully assigned to their parentage.The haploid of24offsprings were consistent with their candidate parents.17offsprings coulddistinguish their paternity from maternity. The pedigree of this natural population wasreconstructed based on the results of parentage assignment.The spatial genetic structure for three populations of L.chinense was investigated using17microsatellite markers. The parameters of spatial genetic structure (Sp) of all individuals andtheir offsprings in2artificial populations were0.01341，0.02493and0.01762，0.00989respectively, while in the natural population, the parameters of spatial genetic structure (Sp)were0.02700,0.01423respectively. Alike other species with similar biological characteristics,L.chinense displays strong fine-scale spatial structure in natural population.Gene flow plays key parts in the genetic structure of plant populations. The quantification ofgene flow can be achieved either by parentage analysis directly or by SGS analysis indirectly. Inthis paper, two methods above were used simultaneously to estimate gene flow of L.chinense.The differences among three populations were obvious for both seed flow and pollen flow. Atrend was found that the gene flow and the average pollen dispersal distance decreased withincreased population density in three populations investigated. In two artificial populations withhigher population density, the seed dispersal distance was greater than the pollen dispersaldistance, and the seed flow accounted for the majority of gene flow, while in the naturalpopulation with low population density, the trend was just the opposite.The population structure dynamic of L. chinensis was also analyzed for the naturalpopulation in Maoershan Reserve. Based on the theory of population life table and survivalanalysis, the life table in the natural population of L. chinensis was constructed by replacing ageclasses with size-classes，and the population structure dynamic was also predicted. The censusdata showed that the number of young seedling was larger than that of middle-aged and oldindividuals, indicating that the population was in an incremental status. The survival curve in this population appeared to be a Deevey-Ⅱ type, with a peak of mortality in size class Ⅳ. It can beconcluded that the population size of L. chinensis in Maoer Reserve declined sharply in earlystage，stablized in middle stage and fell rapidly in last stage.The spatial distribution pattern in this natual population of L.chinense was investigated inthe application of O-ring function. The distribution pattern of this population was closely relatedto the spatial scale, with an aggregation distribution in less than24m, a random distribution in24-48m spatial scale, a hypodispersion distribution in49--88m spatial scale and a randomdistribution in larger than88m. The fine-scale spatial distribution pattern was obviousdifferences among growth stages in this natural population. Both the adulted and youngindividuals showed aggregation distribution at smaller spatial scale, and ramdon distribution atother spatial scale. Whereas, for those aging individuals,only one pattern of random distributionwas identified at all spatial scales. The spatial correlation among development stages werepositive at smaller scale，while at larger scale, no close association was identified．The genetic diversity was also compared between parent population and its offspringspopulation based on14SSR loci. Overall, L. chinense contains rather high polymorphismcompared with other species in magnoliaceae. And there was no difference on the geneticdiversity between parent population and offspring population.更多还原