Genomic approaches to inference of population history and multispecies community assembly

长期以来，物种之间以及物种内种群之间的关系一直被比作一棵树上的树枝——达尔文自己的笔记中就有一个著名的例子。自从DNA测序技术发展以来，人们已经开发出了一些方法，可以根据序列重建种群之间的历史关系。了解种群之间的历史关系（包括祖先种群分裂为后代种群，以及个体在种群之间的分散）在生物学的许多领域都很重要，包括我们自己的物种是在何时何地进化的。在保护生物学中，种群历史的重建使我们能够确定哪些物种可能拥有最大的遗传多样性（这样我们就知道在哪里集中保护工作），哪些种群通过迁徙联系在一起（因此可能相互支持），哪些种群是孤立的（因此灭绝的风险更高）。更一般地说，如果我们想要正确理解种群对自然选择的反应，推断种群关系也是必不可少的。种群历史通常是通过在大量个体中取样的少量基因（通常是5个或更少）的数据推断出来的。造成这种情况的一个主要原因是很难获得更多基因的序列数据，而且人们相信，要了解发生了什么，必须对许多个体进行采样。最近的重大进展改变了这一观点。首先，通过新的“下一代”测序技术的发展，序列数据的可用性有了质的飞跃，能够生成任何物种基因组中数千个基因的数据。其次，凝聚理论的进步表明，在少数个体中取样许多基因要比反之（通常的做法）好得多。这是令人兴奋的，因为这意味着我们甚至可以研究那些不受欢迎或不可能对许多个体进行采样的稀有动物。然而，基因组数据集的庞大规模使得用现有方法分析它们变得困难或不可能。该项目的一个主要目标是开发更好的工具，从基因组数据集推断人口历史，这些数据集将在网络上提供给所有人使用。然后，我们将把我们的新工具应用于两个天然昆虫群落（欧洲橡树瘿和东澳大利亚无花果）的真实数据，每个群落都包括食草动物和它们的天敌寄生蜂。通过比较每个群落中不同物种的种群历史，我们将测试食草动物和寄生性动物是通过空间和时间一起传播，还是在一定的时间尺度上加入它们的群落。这是当前生态学研究的一个重要领域，因为物种之间的长期联系通常会导致强烈的生态依赖性，而这种相互作用的破坏（例如通过人类强加的栖息地变化）可能很难恢复。我们还将测试更具体的假设，即食草动物可以暂时逃离它们的天敌，从而享受一定程度的“无敌人时间”。我们之所以选择这些，部分是因为它们在多物种相互作用研究中作为模型系统的重要性，部分是因为所有涉及的昆虫物种的遗传学方面（雄性中存在一组染色体）使得为它们生成和分析基因组数据集特别容易。虽然这个例子侧重于与生物多样性有关的问题，但我们将开发的方法可以同样适用于更广泛的关联，例如人类与其寄生虫之间的关联。从少数个体中提取信息的能力为更好地利用现有样本或最大限度地减少对稀有物种的影响提供了巨大的潜力。这些机会将在项目支持的3个研讨会上与利益相关者讨论，并在SSERC支持的暑期学校与学校教师交流。

Relationships between species, and between populations within species, have long been likened to branches in a tree - Darwin's own notes include a well-known example. Since the development of DNA sequencing, methods have been developed that allow reconstruction of historical relationships among populations from sequence. Understanding of historical relationships among populations (which includes both the splitting of ancestral populations into daughter populations, and dispersal of individuals between populations) is important in many areas of biology, including where and when our own species evolved. In conservation biology, reconstruction of population history allows us to identify where species are likely to have their greatest genetic diversity (so we know where to concentrate conservation efforts), which populations are connected by migration (and so may support each other) and which are isolated (and so at higher risk of extinction). More generally, inferring population relationships is also essential if we want to correctly understand how populations are responding to natural selection.Population history is usually inferred using data for only a small number of genes (usually 5 or less), sampled in lots of individuals. A major reason for this has been the difficulty in getting sequence data for more genes, and a belief that sampling of many individuals is necessary to understand what is going on. Recent major advances have changed this view. First, there has been a quantum leap in availability of sequence data through development of new "nextgen" sequencing technologies, able to generate data for thousands of genes across the genome of any species. Second, advances in coalescent theory show that it is much better to sample many genes in a small number of individuals than vice versa (the common practice). This is exciting because it means we can work even with rare animals for which sampling of many individuals is unwelcome or impossible. However, the sheer size of genomic datasets makes it difficult or impossible to analyse them with available methods. A major aim of this project is the development of better tools for inference of population history from genomic datasets, which will be made available on the web for all to use.We will then apply our new tools to real data for two natural insect communities (European oak galls and eastern Australian figs), each of which comprises herbivores and their parasitoid wasp natural enemies. By comparing population histories across species in each community, we will test whether herbivores and parasitoids spread together through space and time, or joined their communities over a range of timescales. This is a major area of current research in ecology that matters because long associations between species commonly result in strong ecological dependence, and disruptions of such interactions (for example through human-imposed habitat change) can be very hard to restore. We will also test the more specific hypothesis that herbivores can escape their natural enemies for a while, and so enjoy a measure of 'enemy-free time'. We choose these in part because of their importance as model systems in the study of multispecies interactions, and in part because an aspect of the genetics of all the insect species involved (presence of a single set of chromosomes in males) makes it particularly easy to generate and analyse genomic datasets for them. And while this example focuses on a biodiversity-related issue, the methods we will develop can be applied equally to more applied associations, such as those between humankind and their parasites.Ability to extract information from small numbers of individuals provides enormous potential to make better use of existing samples, or minimise impact on rare species. These opportunities will be discussed with stakeholders at 3 supported workshops through the project, and communicated to school teachers in a supported SSERC summer school.