Food webs as proxies for ecological interaction networks

从历史上看，食物网一直是生态系统连通性的通用模型，也是复杂系统的标志性形象。然而，根据定义，食物网只描述了直接的营养相互作用（谁吃谁），这是物种相互作用的许多方式之一。本项目将发展并使用新兴的经验动态建模（EDM）方法来构建网络，显示几个生态系统中物种之间已实现的因果影响（相互作用网）。这些相互作用网是网络图，量化了一个物种在相互作用时对另一个物种的可测量影响。这种方法具有变革性，因为它衡量的是可能沿因果链间接传播的影响。它不同于当前的分析框架，因为它适应了这样一个事实，即生态相互作用不是恒定的，而是随着时间的推移而变化的，并取决于更广泛的生态环境（例如，消费者之间的竞争强度随着猎物的可用性而变化）。此外，该方法可用于从实时现场数据中生成复杂生态系统的动态数学模型。这项研究通过确定生态系统中的重要物种，揭示间接影响的隐藏因果链（例如揭示管理决策的意外后果），并通过为关键转变提供科学合理的早期预警信号，为生态管理提供可操作的建议，具有直接价值。这项工作的结果应该对资源管理人员和环境管理人员都极为有用。该项目将使用聚合交叉映射的EDM方法来确定因果关系并构建实现的影响网，使用来自四个生物群落的6个生态系统的数据：埃林湾港、北大西洋、加拿大实验湖区和奇瓦瓦沙漠。这些相互作用网络将与从文献综述中构建的经典食物网进行比较，以测试食物网描述最重要的生态系统相互作用的程度。将进行跨系统分析，以量化非营养性相互作用的类型和频率（例如，竞争与间接互惠），并在可能的情况下根据实验证据进行验证（例如，在两个研究地点进行的长期实验中已知的非营养性相互作用）。此外，还将测试交互网络识别重要节点和路径的能力，如关键捕食者和由传递因果链产生的间接影响。最后，该项目将超越静态生态网络的经典形象，承认互动是动态的（即随着时间的推移而变化）。

Historically, food webs have been the common model for ecosystem connectivity and the iconic image of a complex system. By definition, however, food webs depict only direct trophic interactions (who eats whom), one of many ways in which species may interact. This project will develop and use the emerging approach of empirical dynamic modeling (EDM) to construct networks showing the realized causal influences between species (interaction webs) for several ecosystems. These interaction webs are network diagrams that quantify the measurable impact of one species on another as they interact in time. The approach is transformative in that it measures impacts that may be propagated indirectly along causal chains. It is unlike current analytical frameworks, because it accommodates the fact that ecological interactions are not constant but change through time, and depend on the broader ecological context (e.g., where the intensity of competition among consumers changes with prey availability). Moreover, this approach can be used to generate dynamic mathematical models of complex ecosystems from real-time field data. This research has immediate value for producing actionable advice for ecological management by identifying the important species in ecosystems, exposing hidden causal chains of indirect effects (e.g. exposing unintended consequences of management decisions), and by providing scientifically sound early warning signs for critical transitions. The results of this work should be critically useful to resource managers and environmental stewards alike. This project will use the EDM method of Convergent Cross Mapping to identify causal interactions and construct realized influence webs, using data taken from 6 ecosystems spanning 4 biomes: Port Erin Bay, the North Atlantic Ocean, Canada's Experimental Lake Area, and the Chihuahua Desert. These interaction networks will be compared to classical food webs constructed from literature review to test the extent to which food webs depict the most important ecosystem interactions. A cross-system analysis will be done to quantify the type and frequency of non-trophic interactions (e.g. competition versus indirect mutualism), with validation against experimental evidence where possible (e.g. known non-trophic interactions from long-term experiments at two study sites). In addition, the interaction webs will be tested for the ability to identify important nodes and pathways, such as keystone predators and indirect effects resulting from transitive causal chains. Finally, this project will go beyond the classical image of a static ecological network, by acknowledging the reality that interactions are dynamic (i.e. change over time).