Underwater Vision Profiler Research Articles

Beyond transfer learning: Leveraging ancillary images in automated classification of plankton

AbstractWe assess whether a supervised machine learning algorithm, specifically a convolutional neural network (CNN), achieves higher accuracy on planktonic image classification when including non‐plankton and ancillary plankton during the training procedure. We focus on the case of optimizing the CNN for a single planktonic image source, while considering ancillary images to be plankton images from other instruments. We conducted two sets of experiments with three different types of plankton images (from a Zooglider, Underwater Vision Profiler 5, and Zooscan), and our results held across all three image types. First, we considered whether single‐stage transfer learning using non‐plankton images was beneficial. For this assessment, we used ImageNet images and the 2015 ImageNet contest‐winning model, ResNet‐152. We found increased accuracy using a ResNet‐152 model pretrained on ImageNet, provided the entire network was retrained rather than retraining only the fully connected layers. Next, we combined all three plankton image types into a single dataset with 3.3 million images (despite their differences in contrast, resolution, and pixel pitch) and conducted a multistage transfer learning assessment. We executed a transfer learning stage from ImageNet to the merged ancillary plankton dataset, then a second transfer learning stage from that merged plankton model to a single instrument dataset. We found that multistage transfer learning resulted in additional accuracy gains. These results should have generality for other image classification tasks.

Rhizaria in the oligotrophic ocean exhibit clear temporal and vertical variability

Recently studies have shown that Rhizaria, a super-group of marine protists, have a large role in pelagic ecosystems. They are unique in that they construct mineral tests out of silica, calcium carbonate, or strontium sulfate. As a consequence, Rhizaria can have large impacts on the ocean’s cycling of carbon and other elements. However, less is known about Rhizaria ecology or their role in the pelagic food-web. Some taxa, like certain Radiolarians, are mixotrophic, hosting algal symbionts. While other taxa are flux-feeders or even predatory carnivores. Some prior research has suggested that Rhizaria will partition vertically in the water column, likely due to different trophic strategies. However, very few studies have investigated their populations over extended periods of time. In this study, we present data investigating Rhizaria abundance and vertical distribution from over a year of monthly cruises in the Sargasso Sea. This study represents the first quantification of Rhizaria throughout the mesopelagic zone in an oligotrophic system for an extended period of time. We use this data to investigate the hypothesis that Rhizaria taxonomic groups will partition due to trophic mode. We also investigate how their abundance varies in accordance with environmental parameters. Rhizaria abundance was quantified using an Underwater Vision Profiler (UVP5), an in-situ imaging device. Ultimately, we show that different Rhizaria taxa will have unique vertical distribution patterns. Models relating their abundance to environmental parameters have mixed results, yet particle concentration is a common predictive variable, supporting the importance of heterotrophy amongst many taxa.

First release of the Pelagic Size Structure database: global datasets of marine size spectra obtained from plankton imaging devices

Abstract. In marine ecosystems, most physiological, ecological, or physical processes are size dependent. These include metabolic rates, the uptake of carbon and other nutrients, swimming and sinking velocities, and trophic interactions, which eventually determine the stocks of commercial species, as well as biogeochemical cycles and carbon sequestration. As such, broad-scale observations of plankton size distribution are important indicators of the general functioning and state of pelagic ecosystems under anthropogenic pressures. Here, we present the first global datasets of the Pelagic Size Structure database (PSSdb), generated from plankton imaging devices. This release includes the bulk particle normalized biovolume size spectrum (NBSS) and the bulk particle size distribution (PSD), along with their related parameters (slope, intercept, and R2) measured within the epipelagic layer (0–200 m) by three imaging sensors: the Imaging FlowCytobot (IFCB), the Underwater Vision Profiler (UVP), and benchtop scanners. Collectively, these instruments effectively image organisms and detrital material in the 7–10 000 µm size range. A total of 92 472 IFCB samples, 3068 UVP profiles, and 2411 scans passed our quality control and were standardized to produce consistent instrument-specific size spectra averaged to 1° × 1° latitude and longitude and by year and month. Our instrument-specific datasets span most major ocean basins, except for the IFCB datasets we have ingested, which were exclusively collected in northern latitudes, and cover decadal time periods (2013–2022 for IFCB, 2008–2021 for UVP, and 1996–2022 for scanners), allowing for a further assessment of the pelagic size spectrum in space and time. The datasets that constitute PSSdb's first release are available at https://doi.org/10.5281/zenodo.11050013 (Dugenne et al., 2024b). In addition, future updates to these data products can be accessed at https://doi.org/10.5281/zenodo.7998799.

Global census of the significance of giant mesopelagic protists to the marine carbon and silicon cycles

Thriving in both epipelagic and mesopelagic layers, Rhizaria are biomineralizing protists, mixotrophs or flux-feeders, often reaching gigantic sizes. In situ imaging showed their contribution to oceanic carbon stock, but left their contribution to element cycling unquantified. Here, we compile a global dataset of 167,551 Underwater Vision Profiler 5 Rhizaria images, and apply machine learning models to predict their organic carbon and biogenic silica biomasses in the uppermost 1000 m. We estimate that Rhizaria represent up to 1.7% of mesozooplankton carbon biomass in the top 500 m. Rhizaria biomass, dominated by Phaeodaria, is more than twice as high in the mesopelagic than in the epipelagic layer. Globally, the carbon demand of mesopelagic, flux-feeding Phaeodaria reaches 0.46 Pg C y−1, representing 3.8 to 9.2% of gravitational carbon export. Furthermore, we show that Rhizaria are a unique source of biogenic silica production in the mesopelagic layer, where no other silicifiers are present. Our global census further highlights the importance of Rhizaria for ocean biogeochemistry.

Temporal evolution of plankton and particles distribution across a mesoscale front during the spring bloom

AbstractThe effect of mesoscale features on the distribution of planktonic organisms are well documented. Yet, the interaction between these spatial features and the temporal scale, which can result in sudden increases of the planktonic biomass, is less known and not described at high resolution. A permanent mesoscale front in the Ligurian Sea (north‐western Mediterranean) was repeatedly sampled between January and June 2021 using a SeaExplorer glider equipped with an Underwater Vision Profiler 6 (UVP6), a versatile in situ imager. Both plankton and particle distributions were resolved throughout the spring bloom to assess whether the front was a location of increased zooplankton concentration and whether it constrained particle distribution. Over the 5 months, the glider performed more than 5000 dives and the UVP6 collected 1.1 million images. We focused our analysis on shallow (300 m) transects, which gave a horizontal resolution of 900 m. About 13,000 images of planktonic organisms were retained. Ordination methods applied to particles and plankton concentrations revealed strong temporal variations during the bloom, with a succession of various zooplankton communities. Changes in particle abundance and size could be explained by changes in the plankton community. The front had a strong influence on particle distribution, while the signal was not as clear for plankton, probably because of the relatively small number of imaged organisms. This work confirms the need to sample both plankton and particles at fine scale to understand their interactions, a task for which automated in situ imaging is particularly adapted.

Size and transparency influence diel vertical migration patterns in copepods

AbstractDiel vertical migration (DVM) is a widespread phenomenon in aquatic environments. The primary hypothesis explaining DVM is the predation‐avoidance hypothesis, which suggests that zooplankton migrate to deeper waters to avoid detection during daylight. Copepods are the predominant mesozooplankton undergoing these migrations; however, they display massive morphological variation. Visual risk also depends on a copepod's morphology. In this study, we investigate hypotheses related to morphology and DVM: (H1) as size increases visual risk, increases in body size will increase DVM magnitude and (H2) if copepod transparency can reduce visual risk, increases in transparency will reduce DVM magnitude. In situ copepod images were collected across several cruises in the Sargasso Sea using an Underwater Vision Profiler 5. Copepod morphology was characterized from these images and a dimension reduction approach. Although in situ imaging offers challenges for quantifying mesozooplankton behavior, we introduce a robust method for quantifying DVM. The results show a clear relationship in which larger copepods have a larger DVM signal. Darker copepods also have a larger DVM signal, however, only among the largest group of copepods and not smaller ones. These findings highlight the complexity of copepod morphology and DVM behavior.

Structures of coexisting marine snow and zooplankton in coastal waters of Svalbard (European Arctic)

How plankton and particles are arranged spatially and the configurations of their co-occurrence shape the rates of organic matter production, utilization, and export within marine systems. The aim of this study was to examine whether the composition of marine snow (particles and aggregates >500 µm) and its coexistence with zooplankton change with depth layer, level of zooplankton dominance, chlorophyll fluorescence, and turbidity across the coastal–offshore gradients of hydrographically different Arctic fjords. The distribution and concentrations of zooplankton and marine snow were assessed in situ using an underwater vision profiler (UVP) in Svalbard waters during summer 2019. UVP counts of marine snow drastically outnumbered zooplankton at glacial stations, whereas zooplankton dominated offshore and in upper water layers, even in coastal waters. The most common compositional structure was dominance by an elongated morphotype of marine snow, often co-occurring with small dark (opaque) particles below 40 m depth, implying that these were the typical forms exported directly from surface layers. The other widespread type of structuring was dominance of UVP counts by copepods. They often coexisted with a flake morphotype of marine snow associated with high chlorophyll fluorescence. Structuring dominated by dark morphotypes was observed mainly near glaciers and in deep fjord basins. The highest amount of marine snow, represented by a high degree of dark morphotype, was observed in Hornsund, the most Arctic-type fjord. A Phaeocystis-associated agglomerated morphotype of marine snow occurred scarcely and only in more Atlantic-influenced fjords. A bimodal distribution pattern, with one abundance peak at the surface and another in deeper layers (>80 m) was observed offshore and in Kongsfjorden. This study emphasizes the high potential of UVPs for tracking links between plankton and detritus directly in their natural environment, and that variation in their co-occurrence may provide a proxy for the state of a pelagic ecosystem.

Nitrogen fixation rates in the Guinea Dome and the equatorial upwelling regions in the Atlantic Ocean

AbstractBiological nitrogen fixation is a key process balancing the loss of combined nitrogen in the marine nitrogen cycle. Its relevance in upwelling or high nutrient regions is still unclear, with the few available studies in these regions of the ocean reporting rates that vary widely from below detection limit to > 100 nmol N L−1 d−1. In the eastern tropical Atlantic Ocean, two open ocean upwelling systems are active in boreal summer. One is the seasonal equatorial upwelling, where the residual phosphorus associated with aged upwelled waters is suggested to enhance nitrogen fixation in this season. The other is the Guinea Dome, a thermal upwelling dome. We conducted two surveys along 23° W across the Guinea Dome and the Equator from 15° N to 5° S in September 2015 and August–September 2016 with high latitudinal resolution (20–60 nm between stations). The abundance of Trichodesmium colonies was characterized by an Underwater Vision Profiler 5 and the total biological nitrogen fixation in the euphotic layer was measured using the 15N2 technique. The highest abundances of Trichodesmium colonies were found in the area of the Guinea Dome (9°–15° N) with a maximum of 3 colonies L−1 near the surface. By contrast, colonies were almost absent in the Equatorial band between 2° N and 5° S. The highest nitrogen fixation rate was measured at the northern edge of the Guinea Dome in 2016 (ca. 31 nmol N L−1 d−1). In this region, where diazotrophs thrived on a sufficient supply of both phosphorus and iron, a patchy distribution was unveiled by our increased spatial resolution scheme. In the Equatorial band, rates were considerably lower, ranging from below detection limit to ca. 4 nmol N L−1 d−1, with a clear difference in magnitude between 2015 (rates close to zero) and 2016 (average rates around 2 nmol N L−1 d−1). This difference seemed triggered by a contrasting supply of phosphorus between years. Our study stresses the importance of surveys with sampling at fine-scale spatial resolution, and shows unexpected high variability in the rates of nitrogen fixation in the Guinea Dome, a region where diazotrophy is a significant process supplying new nitrogen into the euphotic layer.

Latitudinal gradient and influencing factors of deep-sea particle export along the Kyushu-Palau Ridge in the Philippine Sea

The export of particulate organic matter (POM) to deep-sea is crucial for deep-sea ecosystems. However, in situ measurements of large-scale POM export flux are scarce in the tropical and subtropical western Pacific, leading to reliance on biogeochemical models or sediment trap data from a few stations. To address this gap, the underwater vision profiler was used to measure particulate density and to calculate particulate organic carbon (POC) fluxes along the Kyushu–Palau Ridge (KPR) in the Philippine Sea. The results revealed a significant latitudinal gradient of POC fluxes: 37 % of the POC output from 200 m depth was preserved to 2000 m in the Western Pacific Warm Pool and up to 51 % was preserved in the North Pacific Subtropical Gyre. The near-bottom POC fluxes north of 25°N (1.64 ± 0.80 mg m−2 d−1) were significantly higher than the average near-bottom value of the entire transect (0.60 ± 0.43 mg m−2 d−1). Multiple linear regression analysis showed that the chlorophyll concentration had a significant positive effect on the POC fluxes at all depths, except near the bottom, while local factors such as mesoscale eddies and the interaction effect between the topography and current velocity only had significant effects on the POC fluxes at depths of >2000 m. Particle size spectrum analysis revealed that particles ranging from 64 to 323 μm in size exerted a dominant influence on the increase in the POC fluxes in the near-bottom layers situated north of 25°N. These findings indicated that the spatial heterogeneity of POC fluxes in the western Pacific was governed not only by upper ocean primary productivity but also by mesoscale processes, current velocity, and topography. These results provided crucial fundamental information for cartography of the distribution and simulation of the dynamics of deep-sea organisms along the KPR in the Philippine Sea.

Three major mesoplanktonic communities resolved by in situ imaging in the upper 500 m of the global ocean

AbstractAimThe distribution of mesoplankton communities has been poorly studied at global scale, especially from in situ instruments. This study aims to (1) describe the global distribution of mesoplankton communities in relation to their environment and (2) assess the ability of various environmental‐based ocean regionalizations to explain the distribution of these communities.LocationGlobal ocean, 0–500 m depth.Time Period2008–2019.Major Taxa StudiedTwenty‐eight groups of large mesoplanktonic and macroplanktonic organisms, covering Metazoa, Rhizaria and Cyanobacteria.MethodsFrom a global data set of 2500 vertical profiles making use of the Underwater Vision Profiler 5 (UVP5), an in situ imaging instrument, we studied the global distribution of large (>600 μm) mesoplanktonic organisms. Among the 6.8 million imaged objects, 330,000 were large zooplanktonic organisms and phytoplankton colonies, the rest consisting of marine snow particles. Multivariate ordination (PCA) and clustering were used to describe patterns in community composition, while comparison with existing regionalizations was performed with regression methods (RDA).ResultsWithin the observed size range, epipelagic plankton communities were Trichodesmium‐enriched in the intertropical Atlantic, Copepoda‐enriched at high latitudes and in upwelling areas, and Rhizaria‐enriched in oligotrophic areas. In the mesopelagic layer, Copepoda‐enriched communities were also found at high latitudes and in the Atlantic Ocean, while Rhizaria‐enriched communities prevailed in the Peruvian upwelling system and a few mixed communities were found elsewhere. The comparison between the distribution of these communities and a set of existing regionalizations of the ocean suggested that the structure of plankton communities described above is mostly driven by basin‐level environmental conditions.Main ConclusionsIn both layers, three types of plankton communities emerged and seemed to be mostly driven by regional environmental conditions. This work sheds light on the role not only of metazoans, but also of unexpected large protists and cyanobacteria in structuring large mesoplankton communities.

Particle size distribution at Ocean Station Papa from nanometers to millimeters constrained with intercomparison of seven methods

Particle size distribution (PSD) is a fundamental property that affects almost every aspect of the marine ecosystem, including ecological trophic interactions and transport of organic matter and trace elements. We measured PSDs using a suite of seven instruments in waters near Ocean Station Papa in the Northeast Pacific Ocean. These instruments and their sizing ranges are: Laser In-Situ Scattering and Transmissometer (LISST)-Volume Scattering Function meter (VSF) and Multispectral Volume Scattering Meter (MVSM), both sizing particles from 0.02 µm to 2000 µm; the LISST-100X, from 3 µm to 180 µm; the ViewSizer, from 0.3 µm to 2 µm; the Coulter Counter, from 2 µm to 40 µm; the Imaging Flow CytoBot (IFCB), from 5 µm to 100 μm; and the underwater vision profiler (UVP), from 100 µm to 2000 µm. Together, they cover an unprecedented size range spanning 5 orders of magnitude from 20 nm to 2 mm. The differences in size definition for the different instruments cause biases in comparing PSDs. The absolute differences in PSDs, after correcting for mean biases, were less than a factor of 3 among all the instruments, and within 50% among LISST-100X, LISST+MVSM, Coulter Counter and IFCB. We also found that particles of sizes <50 µm were not very porous; however, porosity must be considered for particles >50 µm. The merged PSDs, ranging from 0.02 µm to 2000 µm, showed little variation in the PSD slope in the upper 75 m of the water column even though the total number of particles decreased with depth. While submicrometer particles are numerically dominant, particles of sizes 1 µm to 100 µm account for 70–90% of the solid volume of particles. We expect that the results of this study will lead to improved estimates of mass and carbon flux in the study area.

New Estimate of Organic Carbon Export From Optical Measurements Reveals the Role of Particle Size Distribution and Export Horizon

AbstractExport of sinking particles from the surface ocean is critical for carbon sequestration and to provide energy to the deep biosphere. The magnitude and spatial patterns of this export have been estimated in the past by in situ particle flux observations, satellite‐based algorithms, and ocean biogeochemical models; however, these estimates remain uncertain. Here, we use a recent machine learning reconstruction of global ocean particle size distributions (PSDs) from Underwater Vision Profiler 5 measurements to estimate carbon fluxes by sinking particles (35 μm–5 mm equivalent spherical diameter) from the surface ocean. We combine global maps of PSD properties with empirical relationships constrained against in situ flux observations to calculate particulate carbon export from the euphotic zone (5.8 ± 0.1 Pg C y−1) and annual maximum mixed layer depths (6.1 ± 0.1 Pg C y−1). The new flux reconstructions suggest a less variable seasonal cycle in the tropical ocean and a more persistent export in the Southern Ocean than previously recognized. Smaller particles (less than 418 μm) contribute most of the flux globally, while larger particles become more important at high latitudes and in tropical upwelling regions. Export from the annual maximum mixed layer exceeds that from the euphotic zone over most of the low‐latitude ocean, suggesting shallow particle recycling and net heterotrophy in the deep euphotic zone. These estimates open the way to fully three‐dimensional global reconstructions of particle fluxes in the ocean, supported by the growing database of in situ optical observations.

Sampling uncertainties of particle size distributions and derived fluxes

AbstractIn this study, we provide a method to quantify the uncertainty associated with sampling particle size distributions (PSD), using a global compilation of Underwater Vision Profiler observations (UVP, version 5). The UVP provides abundant in situ data of the marine PSD on global scales and has been used for a diversity of applications, but the uncertainty associated with its measurements has not been quantified, including how this uncertainty propagates into derived products of interest. We model UVP sampling uncertainty using Bayesian Poisson statistics and provide formulae for the uncertainty associated with a given sampling volume and observed particle count. We also model PSD observations using a truncated power law to better match the low concentration associated with rare large particles as seen by the UVP. We use the two shape parameters from this statistical model to describe changes in the PSD shape across latitude band, season, and depth. The UVP sampling uncertainty propagates into an uncertainty for modeled carbon flux exceeding 50%. The statistical model is used to extend the size interval used in a PSD‐derived carbon flux model, revealing a high sensitivity of the PSD‐derived flux model to the inclusion of small particles (80–128 μm). We provide avenues to address additional uncertainties associated with UVP‐derived carbon flux calculations.

Siliceous Rhizaria abundances and diversity in the Mediterranean Sea assessed by combined imaging and metabarcoding approaches

Siliceous Rhizaria (polycystine radiolarians and phaeodarians) are significant contributors to carbon and silicon biogeochemical cycles. Considering their broad taxonomic diversity and their wide size range (from a few micrometres up to several millimetres), a comprehensive evaluation of the entire community to carbon and silicon cycles is challenging. Here, we assess the diversity and contribution of silicified Rhizaria to the global biogenic silica stocks in the upper 500 m of the oligotrophic North-Western Mediterranean Sea using both imaging (FlowCAM, Zooscan and Underwater Vision Profiler) and molecular tools and data. While imaging data (cells m-3) revealed that the most abundant organisms were the smallest, molecular results (number of reads) showed that the largest Rhizaria had the highest relative abundances. While this seems contradictory, relative abundance data obtained with molecular methods appear to be closer to the total biovolume data than to the total abundance data of the organisms. This result reflects a potential link between gene copies number and the volume of a given cell allowing reconciling molecular and imaging data. Using abundance data from imaging methods we estimate that siliceous Rhizaria accounted for up to 6% of the total biogenic silica biomass of the siliceous planktonic community in the upper 500m of the water column.

A global marine particle size distribution dataset obtained with the Underwater Vision Profiler 5

Abstract. Marine particles of different nature are found throughout the global ocean. The term “marine particles” describes detritus aggregates and fecal pellets as well as bacterioplankton, phytoplankton, zooplankton and nekton. Here, we present a global particle size distribution dataset obtained with several Underwater Vision Profiler 5 (UVP5) camera systems. Overall, within the 64 µm to about 50 mm size range covered by the UVP5, detrital particles are the most abundant component of all marine particles; thus, measurements of the particle size distribution with the UVP5 can yield important information on detrital particle dynamics. During deployment, which is possible down to 6000 m depth, the UVP5 images a volume of about 1 L at a frequency of 6 to 20 Hz. Each image is segmented in real time, and size measurements of particles are automatically stored. All UVP5 units used to generate the dataset presented here were inter-calibrated using a UVP5 high-definition unit as reference. Our consistent particle size distribution dataset contains 8805 vertical profiles collected between 19 June 2008 and 23 November 2020. All major ocean basins, as well as the Mediterranean Sea and the Baltic Sea, were sampled. A total of 19 % of all profiles had a maximum sampling depth shallower than 200 dbar, 38 % sampled at least the upper 1000 dbar depth range and 11 % went down to at least 3000 dbar depth. First analysis of the particle size distribution dataset shows that particle abundance is found to be high at high latitudes and in coastal areas where surface productivity or continental inputs are elevated. The lowest values are found in the deep ocean and in the oceanic gyres. Our dataset should be valuable for more in-depth studies that focus on the analysis of regional, temporal and global patterns of particle size distribution and flux as well as for the development and adjustment of regional and global biogeochemical models. The marine particle size distribution dataset (Kiko et al., 2021) is available at https://doi.org/10.1594/PANGAEA.924375.

Global Distribution of Zooplankton Biomass Estimated by In Situ Imaging and Machine Learning

Zooplankton plays a major role in ocean food webs and biogeochemical cycles, and provides major ecosystem services as a main driver of the biological carbon pump and in sustaining fish communities. Zooplankton is also sensitive to its environment and reacts to its changes. To better understand the importance of zooplankton, and to inform prognostic models that try to represent them, spatially-resolved biomass estimates of key plankton taxa are desirable. In this study we predict, for the first time, the global biomass distribution of 19 zooplankton taxa (1-50 mm Equivalent Spherical Diameter) using observations with the Underwater Vision Profiler 5, a quantitative in situ imaging instrument. After classification of 466,872 organisms from more than 3,549 profiles (0-500 m) obtained between 2008 and 2019 throughout the globe, we estimated their individual biovolumes and converted them to biomass using taxa-specific conversion factors. We then associated these biomass estimates with climatologies of environmental variables (temperature, salinity, oxygen, etc.), to build habitat models using boosted regression trees. The results reveal maximal zooplankton biomass values around 60°N and 55°S as well as minimal values around the oceanic gyres. An increased zooplankton biomass is also predicted for the equator. Global integrated biomass (0-500 m) was estimated at 0.403 PgC. It was largely dominated by Copepoda (35.7%, mostly in polar regions), followed by Eumalacostraca (26.6%) Rhizaria (16.4%, mostly in the intertropical convergence zone). The machine learning approach used here is sensitive to the size of the training set and generates reliable predictions for abundant groups such as Copepoda (R2 ≈ 20-66%) but not for rare ones (Ctenophora, Cnidaria, R2 < 5%). Still, this study offers a first protocol to estimate global, spatially resolved zooplankton biomass and community composition from in situ imaging observations of individual organisms. The underlying dataset covers a period of 10 years while approaches that rely on net samples utilized datasets gathered since the 1960s. Increased use of digital imaging approaches should enable us to obtain zooplankton biomass distribution estimates at basin to global scales in shorter time frames in the future.

Comparison of an In Situ Imaging Device and Net-Based Method to Study Mesozooplankton Communities in an Oligotrophic System

In the past several years, the capabilities of optical tools and in situ imaging devices have greatly expanded and are now revolutionizing the field of plankton research. These tools have facilitated the discovery of new plankton and enhanced the understanding of populations of fragile and gelatinous zooplankton. Imaging devices are becoming more accessible and regularly deployed on oceanographic studies and monitoring efforts. However, despite the increasing use of these tools, there are few studies which offer direct comparisons between in situ imaging devices and traditional-net based methods, especially in open-ocean, oligotrophic systems where plankton are sparser and less intensively sampled. This study compares estimates of mesozooplankton abundance calculated by net-tows and an Underwater Vision Profiler 5 (UVP5HD-DEEP) imaging system. Net tows were conducted with a Multiple Opening and Closing Nets with Environmental Sensing System (MOCNESS) device equipped with 153µm mesh. In total, four tows, each sampling eight distinct depth bins, were conducted aboard two cruises in the Sargasso Sea. Along each cruise, in situ images were collected using an Underwater Vison Profiler 5 (UVP5HD-DEEP). Using these methods, we estimated abundance of different mesozooplankton groups (>0.5 mm). Using established biovolume-biomass conversions, we also estimated the dry mass of certain zooplankton taxa. Furthermore, we address two methods for calculating density and biomass concentration from UVP data. Estimates of mesozooplankton abundance and biomass concentration were generally higher from MOCNESS methods than the UVP estimates across all taxa. It was found that there is not a reliable relationship between UVP estimates and MOCNESS estimates when directly comparing similar depth bins. Nonetheless, when integrating density and biomass concentrations throughout the water column, estimates are not significantly different between the methodology. This study addresses several important considerations for using in situ imaging tools and how to reconcile findings with traditional net-based methods.

Coupling Imaging and Omics in Plankton Surveys: State-of-the-Art, Challenges, and Future Directions

A major challenge in characterizing plankton communities is the collection, identification and quantification of samples in a time-efficient way. The classical manual microscopy counts are gradually being replaced by high throughput imaging and nucleic acid sequencing. DNA sequencing allows deep taxonomic resolution (including cryptic species) as well as high detection power (detecting rare species), while RNA provides insights on function and potential activity. However, these methods are affected by database limitations, PCR bias, and copy number variability across taxa. Recent developments in high-throughput imaging applied in situ or on collected samples (high-throughput microscopy, Underwater Vision Profiler, FlowCam, ZooScan, etc) has enabled a rapid enumeration of morphologically-distinguished plankton populations, estimates of biovolume/biomass, and provides additional valuable phenotypic information. Although machine learning classifiers generate encouraging results to classify marine plankton images in a time efficient way, there is still a need for large training datasets of manually annotated images. Here we provide workflow examples that couple nucleic acid sequencing with high-throughput imaging for a more complete and robust analysis of microbial communities. We also describe the publicly available and collaborative web application EcoTaxa, which offers tools for the rapid validation of plankton by specialists with the help of automatic recognition algorithms. Finally, we describe how the field is moving with citizen science programs, unmanned autonomous platforms with in situ sensors, and sequencing and digitalization of historical plankton samples.

Correcting estimations of copepod volume from two‐dimensional images

AbstractAccurate plankton biomass estimations are essential to study marine ecological processes and biogeochemical cycles. This is particularly true for copepods, which dominate mesozooplankton. Such estimations can efficiently be computed from organism volume estimated from images. However, imaging devices only provide two‐dimensional (2D) projections of three‐dimensional objects. The classical procedures to retrieve volumes, based on the equivalent spherical diameter (ESD) or the best‐fitting ellipse, are biased. Here, we present a method to correct these biases. First, a new method aims to measure body area and fit an ellipse. Then, the body of copepods is modeled as an ellipsoid whose 2D silhouette is mathematically derived. Samples of copepod bodies are simulated with realistic shapes/sizes and random orientations. Their total volume is estimated from their silhouettes using the two classical methods and a correction factor is computed, relative to the known, total, volume. On real data, individual orientations and volumes are unknown but the correction factor still holds for the total volume of a large number of organisms. The correction is around −20% for the ESD method and + 10% for the ellipse method. When applied to a database of 150,000 images of copepods captured by the Underwater Vision Profiler, the corrections decreased the gap between the two methods by a factor of 54. In addition, the same procedure is used to evaluate the consequence of the bias in the estimation of individual volumes on the slopes of normalized biovolume size spectra and show that they are, fortunately, not sensitive to the bias.

Constraining the Particle Size Distribution of Large Marine Particles in the Global Ocean With In Situ Optical Observations and Supervised Learning

AbstractThe abundance and size distribution of marine particles control a range of biogeochemical and ecological processes in the ocean, including carbon sequestration. These quantities are the result of complex physical‐biological interactions that are difficult to observe, and their spatial and temporal patterns remain uncertain. Here, we present a novel analysis of particle size distributions (PSDs) from a global compilation of in situ Underwater Vision Profiler 5 (UVP5) optical measurements. Using a machine learning algorithm, we extrapolate sparse UVP5 observations to the global ocean from well‐sampled oceanographic variables. We reconstruct global maps of PSD parameters (biovolume [BV] and slope) for particles at the base of the euphotic zone. These reconstructions reveal consistent global patterns, with high chlorophyll regions generally characterized by high particle BV and flatter PSD slope, that is, a high relative abundance of large versus small particles. The resulting negative correlations between particle BV and slope further suggests synergistic effects on size‐dependent processes such as sinking particle fluxes. Our approach and estimates provide a baseline for an improved understanding of particle cycles in the ocean, and pave the way to global, three‐dimensional reconstructions of PSD and sinking particle fluxes from the growing body of UVP5 observations.