We describe RAPID: a Real-time Automated Plankton Identification Dashboard, deployed on the Plankton Imager, a high-speed line-scan camera that is connected to a ship water supply and captures images of particles in a flow-through system. This end-to-end pipeline for zooplankton data uses Edge AI equipped with a classification (ResNet) model that separates the images into three broad classes: Copepods, Non-Copepods zooplankton and Detritus. The results are transmitted and visualised on a terrestrial system in near real time. Over a 7-days survey, the Plankton Imager successfully imaged and saved 128 million particles of the mesozooplankton size range, 17 million of which were successfully processed in real-time via Edge AI. Data loss occurred along the real-time pipeline, mostly due to the processing limitation of the Edge AI system. Nevertheless, we found similar variability in the counts of the three classes in the output of the dashboard (after data loss) with that of the post-survey processing of the entire dataset. This concept offers a rapid and cost-effective method for the monitoring of trends and events at fine temporal and spatial scales, thus making the most of the continuous data collection in real time and allowing for adaptive sampling to be deployed. Given the rapid pace of improvement in AI tools, it is anticipated that it will soon be possible to deploy expanded classifiers on more performant computer processors. The use of imaging and AI tools is still in its infancy, with industrial and scientific applications of the concept presented therein being open-ended. Early results suggest that technological advances in this field have the potential to revolutionise how we monitor our seas.

The Pi-10 is the latest iteration of the Plankton Imager: a high-speed colour line-scan camera that images particles in a flow-through system. The Pi-10 is a cost-effective, easy to install and low maintenance automated instrument that can be used on any platform with access to water and power supply. We tested the Pi-10 on the research vessel Cefas Endeavour, connected to a continuous water supply pumping water at 34 L min-1. The instrument collected images of particles, within the size range of 180 um - 3.5 cm, automatically and continuously, alongside other vessel operations, in all weathers over a period of 18 days. The Pi-10 successfully captured and saved up 5000 images per minute, translating into a 46 GB of digital storage per day. When particle density exceeded 147 per litre, the instrument stopped saving all images, while still recording the number of particles that passed through the system. This is akin to subsampling, with more sub-sampling required in areas or times of high particle density (e.g at the time of spring plankton bloom or in turbid waters). The Pi-10 collects high volumes of data in a continuous manner, thus providing unprecedented fine spatial data. The high frequency nature of the instrument opens the door to new areas of research. These include, in particular, the observation of fine scale processes, the move towards real-time sampling, and the increased capability to build a digital twin of the oceans. As technologies continue evolving the Pi-10 performance will increase, being able to collect and save more and more images. ### Competing Interest Statement Authors Phil Culverhouse and Julian Tilbury are employed by Plankton Analytics Ltd which manufactor the Plankton Imager. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Marine species with a pelagic larval phase have the potential to disperse hundreds of kilometres via ocean currents, thus connecting geographically distinct populations. Connectivity between populations therefore plays a central role in population dynamics, genetic diversity and resilience to exploitation or decline and can be an important vector in the management of fisheries. The scallop, Pecten maximus, is a valuable benthic bivalve with a variety of management measures at both regional and national scales. A bio-physical numerical model was developed to simulate and characterise the larval transport and population connectivity of scallops across commercial fishing grounds within the Irish and Celtic Seas. The model incorporated realistic oceanographic currents and known behavioural traits of P. maximus larvae including spawning times, pelagic larval duration, and vertical migration during the various developmental stages i.e., passive, active swimming, vertical migrations, since growth rates change with temperature, which varies spatially and temporally, it was used in the model to determine when an individual larva changed its behaviour. Simulations showed a high degree of connectivity between most populations, with multiple connections allowing for substantial exchanges of larvae. The exception was a population off North Cornwall that was entirely reliant on self-recruitment. A sensitivity analysis of the biological parameters suggested that ocean current patterns primarily controlled the connectivity network, but the strength of the connections was sensitive to spawning date and the specific features of diel vertical migrations. The model identified weakly connected populations that could be vulnerable to overfishing, and populations that are ‘strong connectors’ and a vital source of larvae to maintain the metapopulation. Our approach highlights the benefits of characterising population connectivity as part of an effective management strategy for sustainable fisheries.

1. Evidence-based decisions relating to effective marine protected areas as a means of conserving biodiversity require a detailed understanding of the species present. The Caribbean island nation of St Lucia is expanding its current marine protected area network by designating additional no-take marine reserves on the west coast. However, information on the distribution of fish species is currently limited. 2. This study used baited remote underwater stereo-video to address this shortcom-ing by investigating the effects of depth and seabed habitat structure on demersal fish assemblages and comparing these assemblages between regions currently afforded different protection measures. 3. From the 87 stations visited a total of 5,921 fish were observed comprising 120 fish taxa across 22 families. Species richness and total abundance were higher within the highly managed region, which included no-take reserves. Redundancy analysis explained 17% of the total variance in fish distribution, driven predominantly by the seabed habitats. The redundancy analysis identified four main groups of demersal fishes each associated with specific seabed habitats. 4. The current no-take marine reserves protected two of these groups (i.e. fishes associated with the ‘ soft corals, hard corals or gorgonians ’ and ‘ seagrass ’ groups). Importantly, habitats dominated by sponges, bacterial mats, algal turfs or macroalgae, which also supported unique fish assemblages, are not currently afforded protection via the marine reserve network (based on the five reserves studied). These results imply that incorporation of the full breadth of benthic habitat types present would improve the efficacy of the marine reserve network by ensuring all fish assemblages are protected.

Being located between primary producers and fish, zooplankton are a key element of marine food webs, the trophic structure of which is dependent upon the size distribution of species. Changes in zooplankton community size structure have the potential to alter the food web structure and ultimately the quality of food for planktivorous fish. Zooplankton therefore play a key role in overall ecosystem health and are ideal indicators of environmental variability due to their short life cycle and sensitivity to environmental changes. In this study, we used a first approach to build and test the Copepod Mean Size and Total Abundance (CMSTA) ecological indicator in the Celtic Sea, using a similar methodology to that of the HELCOM Mean Size and Total Stock (MSTS) indicator. We explored relationships between zooplankton mean size, total abundance and biomass with hydrographic and biological variables representative of both lower and higher trophic levels (i.e. Chlorophyll-a levels and the biomass distribution of herring, sardine, anchovy, sprat and horse mackerel). Herring, the species with the strongest preference for larger prey, was the only fish that displayed a statistically significant positive correlation with copepod mean size. Displaying the stations spatially within the 2-dimentional (copepod mean size versus abundance) indicator plots, resulted in a strong pattern of herring distribution related to areas where copepods were the largest rather than the most abundant. Our preliminary results are in line with those obtained from previous studies, confirming that zooplankton mean size is able to reflect the state of the food web, and thus reinforce the importance as zooplankton size as a key trait to be routinely monitored. Our in-situ zooplankton size measurements were collected with the Plankton Imager (PI). The ability of this automated system to take measurements for every single organism in a sample meant that the mean size was calculated based on the actual size distribution of the sampled population. Additionally, because of the non-normal distribution of copepod size, the geometric mean was found to be a much more representative description of the community size than the arithmetic mean. We have demonstrated (1) the potential of the CMSTA as a potential indicator of ecosystem health status and climate effects, and (2) the value of the PI in collecting routine in-situ zooplankton size information for improving the power of the mean size indicator approach to describe the community structure.

The Plankton Imager (PI) is an underway semi-automated, high-speed imaging instrument, which takes images of all passing particles and classifies the mesozooplankton present. We used data (temperature, salinity and mesozooplankton abundance) collected in the Celtic Sea in spring and autumn from 2016 to 2019 to assess the ability of the PI to describe temporal changes in the mesozooplankton community and to capture the seasonality of individual taxa. The description obtained using the PI identified both seasonal and interannual changes in the mesozooplankton community. Variation was higher between years than seasons due to the large variation in the community between years in autumn, attributed to the breaking down of summer stratification. The spring community was consistent between years. The seasonality of taxa broadly adhered to those presented in the literature. This demonstrates the PI as a robust method to describe the mesozooplankton community. Finally, the potential future applications and how to make best use of the PI are discussed.

Three plankton collection methods were used to gather plankton samples in the Celtic Sea in October 2016. The Plankton Image Analysis (PIA) system is a high-speed color line scan-based imaging instrument, which continuously pumps water, takes images of the passing particles, and identifies the zooplankton organisms present. We compared and evaluated the performance of the PIA against the Continuous Automatic Litter and Plankton Sampler (CALPS) and the traditional ring net vertical haul. The PIA underestimated species abundance compared to the CALPS and ring net and gave an image of the zooplankton community structure that was different from the other two devices. There was, however, good agreement in the spatial distribution of abundances across the three systems. Our study suggests that the image capture and analysis step rather than the sampling method was responsible for the discrepancies noted between the PIA and the other two datasets. The two most important issues appeared to be differences in sub-sampling between the PIA system and the other two devices, and blurring of specimen features due to limited PIA optical depth of field. A particular advantage of the CALPS over more traditional vertical sampling methods is that it can be integrated within existing multidisciplinary surveys at little extra cost without requiring additional survey time. Additionally, PIA performs automatic image acquisition and it does remove the need to collect physical preserved samples for subsequent analysis in the laboratory. With the help of an expert taxonomist the system in its current form can also integrate the sampling and analysis steps, thus increasing the speed, and reducing the costs for zooplankton sampling in near real-time. Although the system shows some limitation we believe that a revised PIA system will have the potential to become an important element of an integrated zooplankton monitoring program.