Early-warning signals of a regime shift (EWS) indicate, for a wide range of systems, if a tipping-point is being approached. In ecology, EWS are well established from a theoretical perspective but are far from unequivocal when applied to field data. The gap between theory and application is caused by a set of limitations, like the lack of coherence between different EWS, data acquisition issues, and false results. Experiments assessing EWS may provide an empirical mechanistic understanding of why an EWS was observed (or failed to be observed), which often cannot be elucidated by simple computational modeling or pure environmental data. Here we focused on aquatic experiments to explore to what extent the existing EWS experiments can bridge the gap between the theory and real-world application. For that, we used the Thomson-ISI Web of Science© database to retrieve EWS experiments executed before early-2020, detailing their experimental designs and each EWS assessed. Success rates - correct anticipation of tipping points – were around 70% for the most used EWS (assessment of autocorrelation, variance, recovery, and shape of the distribution using abundance, Chlorophyll-a, Phycocyanin, and dissolved oxygen data). Yet, no EWS showed to be 100% reliable, and their use demands cautious interpretation. As a rule, we observed that experiments were not designed to tackle issues encountered in real-world situations. They lack a deep mechanistic understanding of why, when, and how an EWS was observed or not. When experiments did aim to assess issues encountered in the real world, the experimental designs were often of low ecological significance. We also investigated the relationship between sampling and the success rate of EWS, observing that the sampling regime might have to be tailor-made towards specific monitoring objectives. Moreover, experiments have taught us that the use of EWS can be more versatile than expected, going from monitoring the extinction of single populations to the anticipation of transient regime shifts. Most of the experiments presented here supported empirical proof of the existence of EWS in aquatic systems. Still, to bridge the gap between theory and application, experiments will have to move closer to real-world conditions and better support a mechanistic understanding of why EWS may succeed or fail to anticipate a regime shift. For that, we provide six elements to take into account when designing experiments that could enhance the capabilities of EWS to go beyond the stage of proof-of-concept.

政权转移 (EWS) 的早期预警信号表明，对于各种系统而言，是否正在接近临界点。在生态学中，从理论角度来看，EWS 已经很成熟，但在应用于现场数据时却远非明确。理论与应用之间的差距是由一系列限制造成的，例如不同 EWS 之间缺乏一致性、数据获取问题和错误结果。评估 EWS 的实验可以提供对 EWS 被观察到（或未能被观察到）的经验机制理解，这通常无法通过简单的计算模型或纯环境数据来阐明。在这里，我们专注于水生实验，以探索现有的 EWS 实验在多大程度上可以弥合理论与实际应用之间的差距。为了那个原因，我们使用 Thomson-ISI Web of Science© 数据库检索了 2020 年初之前执行的 EWS 实验，详细说明了他们的实验设计和评估的每个 EWS。对于最常用的 EWS（使用丰度、叶绿素-a、藻蓝蛋白和溶解氧数据评估自相关、方差、恢复和分布形状），成功率 - 临界点的正确预测 - 约为 70%。然而，没有任何 EWS 显示为 100% 可靠，它们的使用需要谨慎解释。通常，我们观察到实验并非旨在解决现实世界中遇到的问题。他们对为什么、何时以及如何观察到 EWS 缺乏深入的机械理解。当实验确实旨在评估现实世界中遇到的问题时，实验设计通常具有较低的生态意义。我们还调查了采样与 EWS 成功率之间的关系，观察到采样制度可能必须针对特定的监测目标量身定制。此外，实验告诉我们，EWS 的使用可以比预期的更通用，从监测单个种群的灭绝到预测短暂的政权转变。这里介绍的大多数实验都支持水生系统中存在 EWS 的经验证据。尽管如此，为了弥合理论和应用之间的差距，实验必须更接近现实世界的条件，并更好地支持对为什么 EWS 可能成功或未能预期政权转变的机制理解。为了那个原因，