Fear of predation may influence food webs more than actual predation. However, the mechanisms and magnitude of nonconsumptive predator effects are largely unknown in unicellular-dominated food webs such as marine plankton. We report a general mechanism of chemically induced predator effects in marine plankton. Copepods, the most abundant zooplankton in the oceans, imprint seawater with unique polar lipids—copepodamides—which trigger toxin production and bioluminescence in harmful dinoflagellates. We show that copepodamides also elicit defensive traits in other phytoplankton, inducing the harmful algal bloom-forming diatom Pseudo-nitzschia seriata to produce 10 times more toxins, and colony-forming diatoms to decrease colony size by half. A 1-year study in the northeast Atlantic revealed that natural copepodamide concentrations are high enough to induce harmful algal toxins and size reduction in dominant primary producers when copepods are abundant. We conclude that copepodamides will structure marine plankton toward smaller, more defended life forms on basin-wide scales.

Object detection is a fundamental task in digital microscopy, where machine learning has made great strides in overcoming the limitations of classical approaches. The training of state-of-the-art machine-learning methods almost universally relies on vast amounts of labeled experimental data or the ability to numerically simulate realistic datasets. However, experimental data are often challenging to label and cannot be easily reproduced numerically. Here, we propose a deep-learning method, named LodeSTAR (Localization and detection from Symmetries, Translations And Rotations), that learns to detect microscopic objects with sub-pixel accuracy from a single unlabeled experimental image by exploiting the inherent roto-translational symmetries of this task. We demonstrate that LodeSTAR outperforms traditional methods in terms of accuracy, also when analyzing challenging experimental data containing densely packed cells or noisy backgrounds. Furthermore, by exploiting additional symmetries we show that LodeSTAR can measure other properties, e.g., vertical position and polarizability in holographic microscopy.

Marine zooplankton release chemical cues, which trigger defenses in unicellular phytoplankton, such as increased toxin production and changes of colony sizes. Here, we identify the structure of two novel alarm cues belonging to the group of copepodamides. Similar to the known copepodamides, one of the compounds described is shown to trigger both amnesic and paralytic shellfish toxin production and chain length shortening in Skeletonema marinoi. In addition, we report the putative structures of another 21 copepodamides, which constitute 28% of the total copepodamides extractable from whole animals, suggesting that the copepodamide concentrations have been underestimated in earlier studies. We introduce a structure‐based nomenclature to handle the increasing number of copepodamides. Analysis of 12 copepod species showed that marine calanoid and freshwater cyclopoid copepods contain copepodamides. The only harpacticoid included in the analysis, Tigriopus californicus, did not appear to produce detectable amounts of copepodamides. Feeding experiments revealed that copepodamide compositions depend on both diet and species‐specific properties. Copepodamides induce both morphological and biochemical defensive traits in phytoplankton and may drive large‐scale trait–mediated effects in marine food webs. The more comprehensive list of copepodamides reported here makes it possible to explore the role of the copepodamide signaling system in the pelagic ecosystem in greater detail.

The implementation of deep learning algorithms has brought new perspectives to plankton ecology. Emerging as an alternative approach to established methods, deep learning offers objective schemes to investigate plankton organisms in diverse environments. We provide an overview of deep‐learning‐based methods including detection and classification of phytoplankton and zooplankton images, foraging and swimming behavior analysis, and finally ecological modeling. Deep learning has the potential to speed up the analysis and reduce the human experimental bias, thus enabling data acquisition at relevant temporal and spatial scales with improved reproducibility. We also discuss shortcomings and show how deep learning architectures have evolved to mitigate imprecise readouts. Finally, we suggest opportunities where deep learning is particularly likely to catalyze plankton research. The examples are accompanied by detailed tutorials and code samples that allow readers to apply the methods described in this review to their own data.