Context and scientific objectives Climate change is one of the biggest challenges of our times, with impact already measurable on our planet and its oceans. To anticipate potential changes in oceanic ecosystems and their consequences on associated services to humans, we need to develop efficient monitoring and modelling methods for physics, chemistry, and biology1. While progress has been made in the first two areas, the biological component of the ocean remains particularly difficult to study at temporal and spatial scales that are relevant for society2. Although Digital Twins of the Ocean (DTOs) place data and numerical technologies at the forefront of environmental science3, their broad scopes are not necessarily meant to resolve processes at regional scales, which are pertinent for stakeholders. Regional DTOs, which collect extensive and diverse observations in a given area, are crucial for enhancing our understanding of the environment at more relevant scales. Among the biological components of the ocean, plankton is extremely diverse4 and plays several critical roles. Phytoplankton is responsible for about half of the global primary production5. The particles resulting from the death and excretion of plankton sequester large amounts of carbon at depth, through biological carbon pumps6, hence contributing to the regulation of climate. Plankton is also critical for marine food webs, directly supporting some of the largest fisheries on earth7. Finally, planktonic organisms are very sensitive to environmental change because they spend their usually short lifespans in the conditions of the watermass they are embedded in. Plankton biomass and diversity are therefore recognized as Essential Ocean and Biodiversity Variables8 (EOVs and EBVs) and used as indicators9 for ecosystem assessment for the Marine Strategy Framework Directive (MSFD). Because of this importance, collecting and integrating plankton data into the EDITO is the target of several ongoing efforts.
While the interannual distribution of plankton over large latitudinal gradients is relatively well known10, their concentrations may vary by orders of magnitude over kilometres or days. Indeed, fronts, eddies, meanders, and filaments generate significant vertical fluxes and redistribute heat, salt, nutrients, oxygen, and carbon above and below the surface mixed layer11. Consequently, phyto- and zooplankton communities exhibit finescale spatial heterogeneity, responding dynamically to the complex interplay of mesoscale (10-100 km) and submesoscale (1-10 km) oceanic features12. When sampling the same point in time, the signature of these features are sudden spikes. Furthermore, the intrinsic dynamics of plankton communities combine with these local phenomena to generate extraordinary bursts of diversity and concentration, over very short time scales13. These hotspots of biological activity can propagate all the way to top predators14. Although these small spatio-temporal scales are vital for understanding plankton dynamics, they are also very challenging to capture. As a consequence, most plankton data currently being integrated into international databases come from relatively sparse ship-based samples and low resolution (e.g., monthly) time series, which is not enough to describe sub-basin processes accurately. The only technology capable of capturing the biological variability of plankton at the same, fine, spatio-temporal scales as its biogeochemical and physical environment is quantitative imaging15: cameras that take large quantities of images in a controlled manner. Among them, the Underwater Vision Profiler 616 is a particularly successful instrument (170 units sold since 2021) that can be deployed from ships, on moorings and on autonomous instruments (gliders, floats). It counts and sizes marine particles from 80µm and images plankton from 0.6mm and up. To collate the massive amount of data generated, specialised, collaborative, online databases were developed17 and leverage machine learning to accelerate the classification of images18. They follow best practices to make data FAIR and harvestable by EU-level and international data aggregators19. The association of high-resolution plankton imaging and physico-biogeochemical datasets, processed with efficient data pipelines, can resolve intense local carbon export events through the subduction of surface water masses20, detect transient anoxic events or harmful algal blooms21, and capture trends in plankton communities by capturing and eliminating the dayto-day variability18, among other processes. All these endeavours require tight data integration into regional DTOs that can then deliver fit-to-purpose products to stakeholders.

The objectives of AQUASPECT are to:
1. Improve plankton imaging technology, through the commercialisation of a new version of the UVP6 targeting smaller sizes, the UVP6m (count from 10µm, image from 100µm);
2. Augment existing monitoring efforts with high spatial or temporal resolution sampling of plankton using imaging, in three contrasting environments (the oligotrophic Mediterranean Sea, the eutrophic North Sea and the low salinity Baltic Sea);
3. Integrate large datasets of EBVs and concomitant environmental data, into regional DTOs, as well as the European DTO (EDITO);
4. Provide more relevant information to stakeholders relative to biodiversity, carbon export, ecosystem state indicators, and effects of anoxia, through these regional DTOs.