胡桃属（Juglans L.）是世界范围内广泛分布的树种资源，具有重要的经济和生态价值。胡桃属内约22个物种，可分为黑核桃组（Sect. Rhysocaryon）、白核桃组（Sect. Cardiocaryon）和核桃组（Sect. Juglans）三个主要类群。但其中核桃组与其它两组的关系尚不明确，而美洲唯一的白核桃物种美洲白核桃（J. cinerea）的起源也同样不清楚。物种的全基因组中包含了很多进化过程中留下的信息，全基因组测序技术的发展也使得大量获取物种内多个个体的基因组信息成为可能。已有的研究已经提供了来自三个组中较为完整的参考基因组数据，本文在此基础上，针对胡桃属19个物种80个个体进行全基因组重测序，并对胡桃属的系统发育关系和进化历史进行了较为详细的分析。 首先，我们利用比对到参考基因组的方法获取了各样本的叶绿体基因组及核单拷贝同源基因，并构建了叶绿体树和基于多物种溯祖理论的核基因组的系统发育树。从中发现了核桃组与J. cinerea的位置具有核质谱系关系不一致的问题。四重频率检验拒绝了不完全谱系分选或多叉树的假设。据此，我们推断胡桃属物种间存在网状进化关系。 通过对物种分化后基因流情景的模拟，我们探究了物种分化后基因流对物种分化时间估计的影响。在物种分化后的有基因流存在的情景中，某些序列对分化自基因流时刻而非物种分化时刻，从而导致物种分化时间或分子钟推断被低估。通过对同源序列的成对距离的频率分布进行混合模型的拟合，我们发现该方法可以在一定程度上区分不同历史事件的同源序列组分，从而提高物种分化时间估计的准确性，避免分子钟速率的低估。通过对J. nigra， J. mandshurica和J. regia的直系同源基因的同义替代率（Ks）进行混合模型进行拟合，我们发现三个胡桃属物种任意两种间的成对距离均服从多个分布的混合模型，主峰在6-9百万年前（Ma）左右，与化石提示的J. mandshurica与J. nigra在45Ma分化的证据不吻合，提示胡桃属物种间可能在中新世晚期发生了次级接触，导致大量直系同源基因在此时分化。 为了进一步对胡桃属的网状进化关系进行刻画，我们利用种群结构分析、基于基因树推断的系统发育网络分析、全基因组的多态位点模式分析及近似贝叶斯计算的方法，揭示了核桃（J. regia）／铁核桃（J. sigillata）所在的核桃组是由来自白核桃组和黑核桃组的祖先支系古老杂交起源的，而美洲白核桃可能是由黑核桃祖先经历了来自白核桃祖先的大量核基因渐渗起源的。近似贝叶斯计算的方法也提示而杂交形成核桃组的时间可能在距今3.45 Ma的上新世晚期，而杂交的两个亲本在此后灭绝了。 最后，通过对生活在相似生态环境的多个北半球温带胡桃属物种近一百万年来的有效种群大小的变化动态的刻画，我们探讨了物种如何应对气候变化和内在环境的竞争。同一气候环境的胡桃属物种并没有对更新世的全球气温下降呈现出相同的种群大小变化趋势，相反，不同物种或不同种群的有效种群大小波动的幅度和时间都有所差异。这说明了生物的种群动态变化不仅受到外在气候因素的影响，更为关键的因素可能是内在生物因素，如物种间的竞争或者与特异病原体的协同进化。 本文为胡桃属植物积累了大量的种群基因组数据，同时也较为清晰地推断了胡桃属植物各组间的系统发育网络关系。此外，本文揭示了分化后基因流对系统发育分析和物种分化时间的影响。提醒后来的研究者在讨论物种进化关系时，不应忽视潜在的基因流的影响。最后，本文探索了如何利用基因组数据揭示物种进化历史，尤其是常被忽略的种间杂交和基因渐渗事件，为今后进行类似的系统发育基因组分析提供了参考的流程。

Juglans L. is a widely distributed tree genus with important economic and ecological value. There are about 22 species within genus Julgans, which can be divided into three sections, Sect. Rhysocaryon, Sect. Cardiocaryon and Sect. Juglans. But the origin of Sect. Juglans has remained mysterious because in phylogenies it occupies an unresolved position between Sect. Rhysocaryon and Sect. Cardiocaryon. Equally unclear is the origin of the only American butternut, J. cinerea. The genome of species contains a lot of information left over in the history of evolution, and the development of whole-genome sequencing has made it possible to to obtain a large number of genomic information from multiple individual samples. Previous research has provided one reference genome from each of the three sections. Here, we re-sequenced the whole genome of 80 individuals from 19 species of Juglans, and performed analyses in phylogenomics and population history. Firstly, we used reference-based mapping to assemble a chloroplast genome as well as to acquire nuclear single-copy genes for each individual. A chroloplast tree and a nuclear phylogenetic tree based on multispecies coalescent model was cunstructed. Cyto-nuclear discordance was observed in the place of Sect. Juglans and J. cinerea. A further quartet frequency test was performed with nuclear single-copy genes, and the result rejected null hypothesis of incomplete lineage sorting or polytomy. Hereby, we conclude the phylogenetic relationship within genus Juglans might be a network. We simulate and analyse data generated by post-divergence gene flow to characterize their impacts on inference of divergence time. In scenarios with gene flow, pair-wise orthologous could come from both gene flow and species divergence, which eventually leads to underestimation of divergence time or clock rate. By examining the distribution and using univariate mixture models to identify peaks in the distribution of simulated, we found this could help quantify the time of gene flow and improve the estimation of species divergence time. Then we conducted mixture models onto distributions of synonymous substitutions (Ks) from ortholog genes among J. nigra, J. mandshurica and J. regia. Results suggested multiple distributions between any pair of the three Juglans species, and a major distribution in 6-9 million years ago (Ma), while fossil evidence suggested the divergence of ancestor of J. mandshurica and J. nigra was ~45 Ma. This also imply massive gene flow has occurred among Juglans species. In order to further reveal the phylogenetic network of Juglans genus, we used phylogenetic-network analysis of single-copy nuclear genes, genome-wide site pattern probabilities, and Approximate Bayesian Computation (ABC). We discovered that J. regia (and its landrace J. sigillata) arose as a hybrid between the American and the Asian lineages and that J. cinerea arose from massive introgression from an immigrating Asian butternut into the genome of an American black walnut. ABC modelling placed the hybrid origin in the late Pliocene, ~3.45 Ma, with both parental lineages since having gone extinct in Europe. At last, in order to discuss whether species demography and diversification are driven primarily, by extrinsic environmental changes or intrinsic biological interactions, we assessed past fluctuations of population sizes over temperate Juglans species within a million years ago. Species or even among conspecific lineages from similar climatic environment did not react similarly to dramatic climatic oscillations following early Pleistocene cooling. So the population histories of temperate walnut species were not driven by extrinsic environmental changes alone, and a key role was probably played by species-specific factors such as coevolutionary interactions with specialized pathogens. In this study, a large number of population genomic data was accumulated for genus Julgans L. And we discovered a network relationship among sections in this genus. In addition, this work revealed the impacts of post-divergence gene flow on divergence time inference, and reminded following researchers to keep caution the influences of potential gene flow. Finally, this work explored and provided a phylogenomic analysis pipeline in the whole-genome sequencing era.