URoL:EN: Emergence of function and dynamics from ecological interaction networks

URoL:EN：生态相互作用网络中功能和动态的出现

• 批准号: 2222478
• 负责人: Michael Gil
• 金额: $ 300万
• 依托单位: University of Colorado at Boulder
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2022
• 资助国家: 美国
• 起止时间: 2022-10-01 至 2027-09-30
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2222478&HistoricalAwards=false
• 关键词: URoL; EN; Emergence; function; dynamics

Biological systems at virtually all levels of organization are defined by diversity – diversity not only of their constituent units, but also of the interactions among those units. Understanding how the core functions of these systems emerge from such diverse interactions is a fundamental challenge that cuts across fields ranging from physiology and medicine to ecology and environmental engineering. Scientists have long sought to simplify biological interactions by classifying them as either negative (e.g., competitors vying for resources) or positive (e.g., cells cooperating within tissue). Yet, upon close inspection, biological interactions regularly defy this simple dichotomy, and are instead “multivalent”: they involve many different types of interaction occurring simultaneously. Examples include bacteria that exchange genes conferring antibiotic resistance while competing for limiting nutrients, or trees that share carbon through mycorrhizal networks even as they compete for light and water. Multivalent interactions such as these appear to be ubiquitous in biology, but they are not easily incorporated into the conventional network models that are widely used to study system-scale dynamics. This project aims to reveal the Rule of Life that governs how function and dynamics emerge in systems of multivalent interactors. The researchers will decipher this Rule using a model ecosystem: coral reefs. In coral reefs, consumption of algae by fish drives dynamics of diverse fish populations and promotes a coral-dominated ecosystem state. The project team will rigorously measure how multivalent interactions among fish species scale up to influence ecosystem dynamics. In doing so, the team will develop data collection technologies and mathematical tools that will provide a generalizable methodology for measuring and understanding one of the most elusive, yet fundamental aspects of complex biological systems: biological interactions themselves. The project’s findings will inform management of coral reef ecosystems vital to over 500 million people worldwide. Moreover, the project will create a technical undergraduate internship program for students from historically underrepresented backgrounds and an international reef monitoring program that empowers local citizen scientists.Through three aims, the project team will develop methods that discover hidden structure in complex networks of ecological interactions and exploit that structure to understand ecosystem-level function and responses to environmental change. The first phase of the project will use field experiments, new camera technologies, and computer vision to directly measure ecological interactions among species, along with a novel algorithmic modeling framework to describe interaction behaviors based on quantitative behavioral traits. The second phase will use manifold learning methods to search behavioral trait data for “functional clusters” of species and exploit this structure to derive coarse-grained dynamical models of the system. These models will be used to evaluate how the structure of interactions drives emergent ecosystem function. In the third phase, the team will use data-driven dynamical models to project how system function and state will respond to future environmental change. Coarse-grained ecosystem models will be developed to understand long-term, system-scale responses to environmental perturbations. These models will incorporate empirically derived evolutionary rates to evaluate the capacity for evolutionary rescue to influence how ecosystems will respond to change. The high-throughput data-collection technologies, dimension reduction tools, and dynamical modeling methods developed in this research will help provide a novel toolkit for data-driven discovery of the dynamical structure of interactions and their consequences in complex biological systems.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.

几乎所有层次的生物系统都是由多样性来定义的——不仅是组成单位的多样性，而且是这些单位之间相互作用的多样性。了解这些系统的核心功能是如何从这些不同的相互作用中产生的，是一个跨越生理学、医学、生态学和环境工程等领域的基本挑战。长期以来，科学家们一直试图通过将生物相互作用分类为消极（例如，竞争资源）或积极（例如，组织内细胞合作）来简化它们。然而，仔细观察就会发现，生物相互作用经常违背这种简单的二分法，而是“多价”的：它们涉及同时发生的许多不同类型的相互作用。例如，细菌在竞争有限的营养物质的同时交换赋予抗生素耐药性的基因，或者树木在竞争光和水的同时通过菌根网络共享碳。诸如此类的多价相互作用似乎在生物学中无处不在，但它们不容易被纳入广泛用于研究系统尺度动力学的传统网络模型中。该项目旨在揭示控制多价相互作用者系统中功能和动态如何出现的生命规则。研究人员将使用一个生态系统模型来解读这一规则：珊瑚礁。在珊瑚礁中，鱼类对藻类的消耗驱动了不同鱼类种群的动态变化，并促进了珊瑚主导的生态系统状态。项目团队将严格测量鱼类之间的多价相互作用如何扩大到影响生态系统动态。在此过程中，该团队将开发数据收集技术和数学工具，为测量和理解复杂生物系统中最难以捉摸但最基本的方面之一——生物相互作用本身——提供一种通用的方法。该项目的研究结果将为珊瑚礁生态系统的管理提供信息，珊瑚礁生态系统对全球5亿多人至关重要。此外，该项目将为来自历史上代表性不足的背景的学生创建一个技术本科实习计划，并为当地公民科学家提供一个国际珊瑚礁监测计划。通过三个目标，项目团队将开发方法，发现生态相互作用复杂网络中的隐藏结构，并利用该结构来理解生态系统层面的功能和对环境变化的反应。该项目的第一阶段将使用实地实验、新的摄像技术和计算机视觉来直接测量物种之间的生态相互作用，以及基于定量行为特征的新型算法建模框架来描述相互作用行为。第二阶段将使用多种学习方法搜索物种“功能集群”的行为特征数据，并利用这种结构推导出系统的粗粒度动态模型。这些模型将用于评估相互作用的结构如何驱动新兴的生态系统功能。在第三阶段，团队将使用数据驱动的动态模型来预测系统功能和状态将如何响应未来的环境变化。将开发粗粒度的生态系统模型，以了解对环境扰动的长期系统级反应。这些模型将纳入经验推导的进化速率，以评估进化拯救影响生态系统如何应对变化的能力。本研究开发的高通量数据收集技术、降维工具和动态建模方法将有助于为复杂生物系统中相互作用的动态结构及其后果的数据驱动发现提供一个新的工具包。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。

项目成果

Demystifying image-based machine learning: a practical guide to automated analysis of field imagery using modern machine learning tools

揭秘基于图像的机器学习：使用现代机器学习工具自动分析现场图像的实用指南

• DOI: 10.3389/fmars.2023.1157370
• 发表时间: 2023
• 期刊: Frontiers in Marine Science
• 影响因子: 3.7
• 作者: Belcher, Byron T.;Bower, Eliana H.;Burford, Benjamin;Celis, Maria Rosa;Fahimipour, Ashkaan K.;Guevara, Isabela L.;Katija, Kakani;Khokhar, Zulekha;Manjunath, Anjana;Nelson, Samuel
• 通讯作者: Nelson, Samuel