指导老师：李德铢  研究员

学位论文题目：《喜马拉雅红豆杉的谱系地理学研究》

论文摘要：

本研究中，我们选用父性遗传的叶绿体trnL-F片段和双亲遗传的SSR标记，以采集自喜马拉雅红豆杉整个分布区的43个居群共815个个体为研究材料，对其开展了谱系地理学和居群遗传学研究，以探讨其遗传多样性、遗传结构和谱系分化历史及其成因等问题，并在此基础上提出了该物种的保护策略。主要研究结果和结论如下：

利用四个叶绿体和一个核基因片段对欧亚红豆杉属植物进行了DNA条形码研究。结果表明，分布于欧亚地区的红豆杉属植物包括11个种（谱系）。须弥红豆杉复合群形成了7个明显不同的谱系，各自均具有不同的地理分布区。其中，三个谱系与传统分类学的三个分类群相对应，其余四个谱系可能为新分类群或隐存种。分布于东喜马拉雅和横断山南部地区的喜马拉雅红豆杉包含两个谱系。

叶绿体trnL-F片段共检测到29种单倍型。喜马拉雅红豆杉具有较高的遗传多样性（HT = 0.809, π = 0.00117）和明显的谱系地理学结构（NST = 0.768 > GST = 0.469, P < 0.01）。AMOVA的结果表明较多的遗传变异（61.8%）存在于组间，仅有30.16%的遗传变异存在于居群内。SAMOVA分析结果把43个居群分为了两组：HM组（谱系）包括16个居群，主要分布于横断山南部地区（云南西北部澜沧江和金沙江上游地区）；EH组（谱系）包括了剩余的27个居群，主要分布于东喜马拉雅至云南高原的西部地区。单倍型系统发育及网状分析的结果表明，所有单倍型聚为两个不同的单系分支，单倍型的分布与SAMOVA分组的结果吻合。基于叶绿体trnL-F片段的平均进化速率估算的两个谱系间分歧时间大约在7.6-3.4 Ma BP。失配分析和中性检测的结果均表明这两个谱系历史上都发生过居群快速扩张，但具有不同的居群扩张历史。我们推测，这两个谱系的分化可能由青藏高原隆升导致的地理隔离所致，东亚季风和印度洋季风的加强加剧了这两个谱系间的生境异质化，最终导致它们发生了异域物种形成。　  　　

通过跨种筛选和新开发的策略得到了20对有多态性的SSR引物。从中选择了11个微卫星位点对喜马拉雅红豆杉进行了STR分型检测，选取了9个微卫星位点用于居群遗传学分析。结果表明，喜马拉雅红豆杉具有较低的遗传多样性（HO = 0.39, HE = 0.48）及明显的遗传分化（GST = 0.21，P < 0.001）。AMOVA分析的结果表明有68.27%的遗传变异存在于居群内，23.97%的遗传变异存在于组间，仅有7.76%的遗传变异位于组内居群间。Barrier分析表明，怒山山脉成为两个谱系间的天然地理隔离障碍，阻碍了谱系间的基因流。STRUCTURE、PCO以及NJ分析的结果表明，喜马拉雅红豆杉分化为HM和EH两个明显不同的谱系，这与cpDNA结果一致。结合cpDNA的结果，根据遗传多样性的分布进一步确认两个谱系各自拥有多个不同的冰期避难所。在EH谱系中，东部的居群具有较高的遗传多样性，而西部的居群遗传多样性则较低，我们推测，东喜马拉雅地区西部的居群可能由西藏东南部的居群沿喜马拉雅山脉南坡由东向西扩散形成的。进一步分析发现，两个谱系的生态位存在明显分化，因此我们认为这两个谱系可能随着青藏高原的快速隆升和季风气候的影响经历了异域物种形成，它们已分化为完全不同的遗传实体。　　　 

综合本研究的结果，喜马拉雅红豆杉包含两个进化显著单元（ESU），在制定保护策略时需要区别对待。保护策略上，就地保护应该是有效保护喜马拉雅红豆杉野生资源的优先保护策略，迁地保护和近地保护可作为就地保护的有效补充。

Abstract：In the present study, we sampled 815 individuals from 43 populations of T. wallichiana across its entire distribution range to investigate the distribution of genetic diversity, levels of genetic differentiation, and phylogeographic history. To this end we acquired sequences of the chloroplast trnL-F intron-spacer region and population genetic data from nine nuclear microsatellite loci. The main results and conclusions can be summarized as follows:
We analyzed 47 accessions, representing all taxa treated in current floristic work and covering most of the distribution range of Taxus in Eurasia. We used a DNA barcoding approach with four chloroplast regions (rbcL, matK, trnH-psbA, trnL-F) and the nuclear ribosomal DNA internal transcribed spacer (ITS).The results of the DNA barcoding indicated the existence of eleven yew lineages (= barcode taxa) in Eurasia. Seven lineages were identified among the T. wallichiana complex, of which three lineages corresponded to T. wallichiana s.s., T. chinensis and T. mairei, while the remaining four lineages potentially represent new or cryptic species. Some of these show a distinct geographical distribution and identifiable morphological characters. The recognized species T. wallichiana was divided into two barcode taxa (i.e. T. wallichiana s.s. and Hengduan type), which mainly distributed in the East Himalayas and the Hengduan Mountian regions, respectively.

A total of 29 cpDNA haplotypes were identified based on trnL-F sequence data. The species showed a high level of genetic diversity at the species level (HT = 0.809, π = 0.00117) with a strong phylogeographic structure (NST = 0.768 > GST = 0.469, P < 0.01). The AMOVA results revealed that 61.8% of the genetic variation was partitioned among groups, and 30.2% within populations, suggesting a significant genetic differentiation among populations. The 43 populations analyzed were divided into two groups by a SAMOVA analysis; the HM group (equal to HM lineage) included 16 populations from the south Hengduan Mountian region (Lancang and Jinsha River area in northwest Yunnan), and the EH group (equal to EH lineage) contained the remaining 27 populations mainly from the East Himalayas and West Yunnan Plateau. Neighbor-joining (NJ) and network analyses of the cpDNA haplotypes suggested that the 29 haplotypes formed two distinct monophyletic clades, which were consistent with the two groups defined by SAMOVA. The divergence time of the two lineages was estimated at 7.6-3.4 Ma before present, based on average mutation rates of trnL-F sequences. Both mismatch distribution analysis and neutrality test implied that the two lineages rapidly expanded, but have a different evolutionary history. The evolutionary history of T. wallichiana may be as follows: the ancestor of the two lineages was geographically split and isolated due to the rapid uplift of the QTP, which led to an intensified East Asian and Indian monsoon. This resulted in wetter Eastern Himalaya and a drier Hengduan Mountain region, to which each two lineage adapted. This resulted in ecological differences of the two lineages, and finally in an allopatric speciation event.

Forty-one microsatellite loci were tested for cross-species amplification, of which 10 loci amplified successfully and showed polymorphisms. A total of 65 newly designed microsatellite loci were evaluated for T. wallichiana, of which 10 exhibited polymorphisms. The 20 suitable markers were characterized on 58 individuals from two T. wallichiana populations from China. The polymorphic microsatellite markers screenedin this study will be useful for further investigations inconservation genetics of the endangered T. wallichiana and related species.
We chose 11 microsatellite loci for genotyping of 815 individuals from 43 populations of T. wallichiana. Nine loci (two SSR loci were discard due to high frequency of null alleles detected in many populations) were at last used for population genetics analysis. A low level of genetic diversity (HO = 0.39, HE = 0.48) but a significant genetic differentiation (GST = 0.21) among populations was estimated at the species level. The AMOVA analysis revealed that 68.27% of the genetic variation was partitioned within populations, 23.97% among groups, and only 7.76% among populations within the groups. The barrier analysis indicated Nushan Mountains as a possible natural geographic barrier for gene flow between the HM and EH lineages. STRUCTURE, PCO and NJ analyses indicated two distinct lineages in T. wallichiana, which corresponded with the cpDNA results. Thus, the microsatellite results as well as the cpDNA data confirmed the existence of multiple refugia for T. wallichiana. Within the EH lineage, a higher genetic diversity was found in the eastern populations and a lower in the western populations with a cline in the East Himalayas. This indicated population expansion from refugia in Southeast Xizang westward alone the East Himalayas to central Nepal. The two lineages occupy different ecological niches with differentiation of habitat factors (e.g. mean annual precipitation and mean annual temperature) based on ecological information analysis. We suggest that T. wallichiana has likely experienced an allopatric speciation event that followed the rapid uplift of the QTP and the formation and intensifying of the Asian monsoon in the Miocene. Based on molecular and morphological differences, and their contrasting ecological requirements, the two lineages can be regarded as two distinct taxonomic entities.

Two evolutionarily significant units (ESU) were confirmed in this study, namely the HM lineage and the EH lineage, respectively. Because of their different evolutionary history and ecological differences, they may require different conservation managements. To devise proper conservation strategies, in situ conservation should take priority for this species, complemented by ex situ and near situ conservation approaches.

 

周  伟  博士　 　 专业：植物学　 

指导老师：李德铢  研究员、王红 研究员

学位论文题目：《二型花柱植物滇丁香单态和偏态居群起源》

论文摘要：

本研究在滇丁香分布区内进行广泛而密集的采样，总共获得25个有代表性的野生居群，这些居群包括19个二态居群（Dimorphic population）、4个长花柱型单态居群（Long-styled monomorphic population）和2个长花柱型偏态居群（Long-styled morph biased population）。由于叶绿体（cp）DNA相对核基因微卫星（SSR）变异速率更慢，我们首先通过两个叶绿体片段（trnL-trnF和rpl20-rps12）联合分析初步查明滇丁香居群在较大时间尺度上的谱系分化格局及其动因。在此基础上，通过7对自行开发的微卫星标记验证上述分化模式，并借助其灵敏的遗传多样性检测能力，进一步推测该物种在近期内所经历的遗传分化历史及其驱动因素。此外，我们选取6个花部形态特征，包括柱头高度、花药高、柱头长度、花药长度、花冠筒长度和雌雄异位距离，对滇丁香二态和单态居群花部综合征变异模式进行比较分析，探讨滇丁香不同花型构成的居群间繁育系统的转变对花部形态分化的潜在影响。基于分子和形态两方面的研究，本研究的主要结果总结为以下三方面：

（一）滇丁香谱系分化模式

叶绿体片段联合分析表明，二型花柱植物滇丁香约早在更新世中晚期（0.351Mya）分化成两支，即东部分支（Eastern isoplethic lineage）和中西部分支（Central anisoplethic & Western isoplethic lineage）。这两个分支的地理分异位置大致位于中国－喜马拉雅森林植物亚区（Sino-Himalayan Forest Floristic Subkingdom）和中国－日本森林植物亚区（Sino-Japanese Forest Floristic Subkingdom）分界线南端，即田中线（Tanaka Line）位置。我们推测，滇丁香两个分支的形成是由于连续分布区中部（即田中线以西至云南双江一带）出现生境片段化所导致的隔离分化（vicariance）。

微卫星标记数据不仅支持根据叶绿体片段推导的上述结论，同时，它进一步揭示中西部分支可以细分为中部分支（Central anisoplethic lineage）和西部分支（Western isoplethic lineage），其中由单态和偏态居群构成的中部分支其遗传多样性显著低于由二态居群构成的西部分支。结合中西部分支经历快速扩张事件和种群遗传多样性差异，我们认为当片段化区域植被和气候条件恢复之后，西部分支（主要指长花柱个体）快速向中部区域扩散迁移，形成目前的中部分支。

可见，滇丁香物种分布区经历了一个早期的片段化和一个近期的修复过程。但是，从本研究结果来看在后期分布区重新修复连接过程中，主要贡献来自于片段化区域西侧居群的扩张运动，而东侧居群很少有向西迁移的迹象。从目前的分布情况来看，早期分化的两支在地理分布上已经重新融合，而且相互之间也已经出现了一定程度的基因交流，如居群SP和居群MG。

（二）滇丁香单态和偏态居群起源

首先，研究结果表明滇丁香长花柱型单态居群起源于西部邻近的二态居群。这些二态居群以种子散布的形式借助风力传播向其东部开拓新的生境，在此过程中，长花柱个体由于具备高自交亲和性特点而得以在新的生境中建立起稳定的初始居群（Founder population）。就目前各居群的花型组成来看，最早的初始居群应该是现在的偏态居群SJ，它的前身应该是长花柱型单态居群，其近期转化为偏态居群是由于西部短花柱个体的侵入。在SJ这个早期的单态居群建立之后，随后依次形成的单态居群分别是LC、JP、PB、LJS和SP。可见，长花柱单态居群的形成最主要的特点是其自交亲和性决定的建立者效应。

其次，研究发现滇丁香长花柱型偏态居群是有单态居群转化而来的，它并非由二态居群中短花柱个体频率下降所致。根据我们的研究结果，本研究中所涉及的两个偏态居群SJ和SP都是由单态居群转化而来，其前身都应该是单态居群。但是，它们转化的方式可能并不相同，对于SJ而言，它更可能是由于短花柱个体通过来自西部二态居群的种子流形式输入；相反，对于SP而言，尽管它的长花柱个体从西部二态居群通过逐步扩散而来，但是它的短花柱个体则应该来自于东部的二态居群，然而其输入方式则更可能是花粉流。

以动态的观点来看，滇丁香分布区的扩张是通过自交亲和的长花柱个体先行，短花柱自交不亲和个体后行的方式向前逐步推进来实现。由于这种特殊的扩张方式，使得分布区的边缘地带呈现为长花柱型单态居群，次边缘地带呈现为长花柱型偏态居群，而在核心地带呈现为比率平衡的二态居群。

（三）滇丁香繁育系统转变与花部形态分化

微卫星标记结果表明，滇丁香单态居群相对二态居群异交率显著降低，这个结果并不令人感到意外，因为单态居群由于缺乏配对花型而无法实现非选型交配，有鉴于此其异交率必将有所下降。然而，值得期待的是，既然单态居群面临繁育方式的转变，那么构成此类居群的长花柱型花相对二态居群中同类型花有没有在花部形态特征上发生新的调整呢？形态学分析结果表明：首先，在单花水平上，它们的雌雄异位距离显著缩小了；其次，在多花水平上，它们的雌雄器官空间非法重叠程度显著提高了。单态居群长花柱型花在形态上所发生的这两方面变异将导致两个截然相反的结果：前者有利于促进自花授粉提高自交率；后者在传粉昆虫介导下有利于促进同株异花授粉或者异株授粉，能同时提高自交和异交。单态居群长花柱型花花部形态所发生的这一系列转变可能有两种原因，第一，由于原有的二型花柱传粉系统的崩解，长花柱型个体摆脱先前传粉适应特征的束缚而呈现随机变异所致；另外，也有可能是出与繁育系统转变需求所引起的定向选择结果。

Abstract：To reveal the population history, wenty-five populations (19 dimorphic and isoplethic populations, four long-styled monomorphic populations and two long-styled morph biased populations) of distylous Luculia pinceana were sampled and studied. In general, the divergence rate was significantly lower in plastid DNA than nuclear microsatellite (SSR) variation. Therefore, a combination of two chloroplast (cpDNA) sequences variations (trnL-trnF and rpl20-rps12) of L. pinceana was used to evaluating the species phylogeographic pattern and its demographic history. We further developed seven primer pairs: LP58, LP45, LP4, LP18, LP154, LP107 and LP198 to assess population genetic structure and differentiation at microsatellite loci with its substantially high level of polymorphicsms. Morphologically, six floral morphological characters (stigma height, anther height, stigma length, anther length, floral tube length and stigma-anther separation) were analysed to approach the potential connetion between transition of mating system and floral morphological differentiation patterns. The analyses of genetic data from cpDNA and SSR, coupled with the floral morphology results, provided three major findings.

1. Phylogeographic pattern of L. pinceana

According to rate-calibrated sequence divergences (dA), the separation between the west-central and eastern lineages detected in cpDNA most likely occurred at the beginning of the Mid-Pleistocene, about 0.315 Mya, coinciding with ‘Stage I’ of China’s ‘Penultimate Second Glacial Period’ (c. 0.333 - 0.316 Mya). This major split in the species co-localizes with a biogeographic boundary between Sino-Himalayan Forest Floristic Subkingdom and Sino-Japanese Forest Floristic Subkingdom, i.e., the Tanaka Line. We presume that the species was initially split into the west-central and the eastern lineages by vicariance.

The results of SSR genotyping analysis further indicated that the central lineage was derived from the west-central lineage by relatively recent dispersals, since a significant loss of genetic diversity and a decrease of crossing rate were detected in this lineage. Combined with the facts of sudden population expansion in the west-central lineage and a significant loss of genetic diversity in the central lineage, we suggested that the central anisoplethic lineage was derived from the western isoplethic lineage (the long-styled morph) by founder dispersal events.

Therefore, it is clear that the distribution of L. pinceana have experienced a process of fragmentation early and a subsequent recovery more recently. However, our results indicated that the most contribution was come from the recolonization of western populations but not the westward movement of eastern population. In the present, the splited two lineages have connected again, and few gene flows have occured between lineages, e.g. populations SP and MG.

2. Origin of monomorphic and morph biased populations in L. pinceana

Firstly, our results indicated that the monomorphic populations of L. pinceana were originated from adjacent western dimorphic populations. These dimorphic populations expanded their new habit by seed dispersal with assistance of wind. In this process, the long-styled individual established the stable founder population because of its self-compatibility.

As far as floral morph composition, the first established monomorphic population appeared to be SJ as it presented morph biased individuals in precent. The transition from monomorphic to morph biased was resulted from the invasion of short-styled individual from the western lineage. The monomorphic population established at the marginal habit actted as stepping stone to facilitate the emengerce of following monomorphic populations, e.g. populations LC, JP, PB, LJS and SP. Therefore，the most significant feature for origin of monomorphic population is the founder events of self-compatibility of long-styled individuals.

Secondly, our results indicated that the morph biased populations of L. pinceana was originated from monomorphic population rather than dimorphic population. In our study, two sampled morph biased populations SJ and SP were all came from monomorphic, but it may be in a different way in transfer of short-styled gene. The short-styled gene in population SJ was more likely to come from western dimorphic populations by seed flow, however, it was more likely to come from eastern dimorphic populations by pollen flow with respect to population SP.

Overall, with a dynamic view, the expansion of distribution was dispersed by self-compatible long-styled individuals and following with self-incompatible short-styled ones. This special expansion mode leads to the significant distribution patterns, with long-styled monomorphic populations always presented in the front edge, the dimorphic populations in the core region and the long-styled morph biased populations occurred in-between.

3. Mating system transition and floral morphological variation in L. pinceana

The results of codominant SSR markers indicated that the crossing rate was significantly lower in dimorphic populations. It is in expection, because the disassortative mating was impossible for absence of mating partner (short-styled morph). Since the mating system composition was changed between dimorphic and monomorphic populations, is there any floral morphological adjustment occurred? Our morphmetric analysis revealed two important findings. Firstly, at the single flower level, the stigma-anther separation in monomorphic population was significantly reduced compared with dimorphic population. Secondly, at the multi-flower level, the illegitimate spatial matching of sexual organs was increased in monomorphic populations. The modification of floral morphology would have lead to contrary influence on mating systems, the former will promote selfing and the latter will facilitate crossing within or between individuals. There are two possible expalanations for morphological variation patterns revealed in monomorphic population. Firstly, the former pollination system was disaggegated and the long-styled morph escaped the constraint of floral syndrome to random variation; an alternative explanation is that the transition of mating system from disassorsitive to selfing or intra-morph crossing lead to the directional selection on floral morphology. Further study is needed to give a more certain answer to this problem.