MIM: Discovering in reverse – using isotopic translation of omics to reveal ecological interactions in microbiomes.

淡水河流中的微生物群包括数百种藻类、细菌、真菌和微观动物。这些微生物群是如何相互作用的？这种相互作用在很大程度上是竞争性的吗，就像生物群体争夺稀缺资源一样？当一个生物体合成另一个生物体需要的资源时，有多少是有益的？这些相互作用对食物网、淡水藻类的生长和环境中营养物质的循环有多重要？这项研究的重点是Cladophora glomerata，一种生长在全球温带湖泊和河流中的绿色淡水藻类，特别是它的微生物群，即生活在Cladophora表面的数百种微生物。研究人员将使用化学示踪剂来揭示碳和氮等元素如何在微生物群中移动，因为这些元素对支持生产力很重要。研究人员还将测试微生物组如何响应营养和光，这些因素驱动淡水生态。这项研究将提高对微生物群如何支持健康生态系统以及可能将系统转变为有害藻华的因素的理解。新的机器学习工具将用于开发预测框架，以了解微生物物种如何相互作用并影响宏观食物网（昆虫，鱼类，鸟类）。此外，将培训十多名学生，并制作面向普通观众的教育材料，包括视频、书面故事和艺术作品。该培训计划强调在所有职业阶段教授研究人员如何通过视觉、书面和口头讲故事来交流他们的科学。该项目将开发绿色巨藻，Cladophora，作为理解微生物组控制过程的模型系统。这项研究将通过结合假设驱动的生态理论和数据驱动的机器学习能力来推进微生物组科学。研究人员将把“组学”数据与最先进的稳定同位素技术相结合，例如定量稳定同位素探测，qSIP，它使用18O标记水的稳定同位素测量单个微生物物种的生长和死亡率；qSIP和Chip-SIP采用13C和15N，通过测量核酸同位素组成来量化分类群特有的C和N同化；和NanoSIMS，分析C和N在微生物群及其宿主中移动的时空模式。将进行实验来研究(1)微生物组在一个有充分记录的连续序列中的变化，(2)对营养添加的反应的生产力变化，以及(3)对光的反应。跨学科的方法将通过超越常用物种共现矩阵的数据矩阵，在生物学和机器学习方面发展新的科学。将添加的其他特征包括分类群特定的生长和死亡率、碳和氮的吸收、代谢和系统发育。添加这些将提高我们理解复杂微生物组中微生物相互作用的能力。使用机器学习，研究人员将建立和训练吸收这些多个同时的数据矩阵的模型（合在一起，一个张量）。这种方法将促进产生相互作用术语的努力，这些术语反映了构成生态群落的关键相互作用（竞争、互惠、促进、寄生或捕食）的重要性。该项目由数学和物理科学部（MPS）的化学部（CHE）和生物科学部（BIO）的环境生物学部（DEB）共同资助。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Microbiomes in freshwater rivers comprise hundreds of species of algae, bacteria, fungi and microscopic animals. How do these groups of micro-organisms interact? Are the interactions largely competitive, as groups of organisms vie for scarce resources? How many are helpful, where one organism synthesizes a resource that another needs? How important are these interactions for food webs, for the growth of algae in freshwater, and for the cycling of nutrients in the environment? This research focuses on Cladophora glomerata, a green freshwater alga that occurs in temperate lakes and rivers worldwide, and specifically on its microbiome, the hundreds of micro-organisms that live right on the surface of Cladophora. Researchers will use chemical tracers that reveal how elements like carbon and nitrogen move through the microbiome, because these elements are important for supporting productivity. Researchers will also test how the microbiome changes in response to nutrient and light, factors that drive the ecology of freshwaters. This research will improve understanding of how microbiomes support healthy ecosystems and the factors that can shift systems to harmful algal blooms. Novel machine learning tools will be used to develop a predictive framework for understanding how microbial species interact with one another and affect macroscopic food webs (insects, fish, birds). In addition, over ten students will be trained and educational materials including videos, written stories and artwork geared towards general audiences will be created. The training program emphasizes teaching researchers, at all career stages, and how to communicate their science through visual, written and oral storytelling.This project will develop the green macro-alga, Cladophora, as a model system for understanding processes governing microbiomes. This research will advance microbiome science by combining hypothesis-driven ecological theory with the data-driven power of machine learning. Researchers will integrate ‘omics’ data with state-of-the-art stable isotope techniques, such as quantitative stable isotope probing, qSIP, which measures growth and mortality rates of individual microbial species using stable isotopes of 18O labeled water; qSIP and Chip-SIP using 13C and 15N, which quantifies taxon-specific C and N assimilation by measuring the isotopic composition of nucleic acids; and NanoSIMS, which analyzes spatial and temporal patterns of C and N movement through the microbiome and its host. Experiments will be performed to study (1) shifts in the microbiome over a well-documented successional sequence, (2) changes in productivity in response to nutrient addition, and (3) responses to light. Interdisciplinary approaches will develop new science in biology and machine learning through data matrices that span beyond the commonly used species co-occurrence matrices. Additional features that will be added include taxon-specific growth and mortality rates, carbon and nitrogen uptake, metabolism, and phylogeny. Adding these will improve our ability to understand microbial interactions in complex microbiomes. Using machine learning researchers will build and train models assimilating these multiple, simultaneous matrices of data (together, a tensor). This approach will advance efforts to generate interaction terms that reflect the importance of the key interactions (competition, mutualism, facilitation, parasitism, or predation) that structure ecological communities.This project is co-funded by the Division of Chemistry (CHE) in the Mathematics and Physical Sciences (MPS) Directorate and by the Division of Environmental Biology (DEB) in the Biological Sciences (BIO) Directorate.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.