Search for: All records

As the basis of oceanic food webs and a key component of the biological carbon pump, planktonic organisms play major roles in the oceans. Their study benefited from the development of in situ imaging instruments, which provide higher spatio-temporal resolution than previous tools. But these instruments collect huge quantities of images, the vast majority of which are of marine snow particles or imaging artifacts. Among them, the In Situ Ichthyoplankton Imaging System (ISIIS) samples the largest water volumes (> 100 L s -1 ) and thus produces particularly large datasets. To extract manageable amounts of ecological information from in situ images, we propose to focus on planktonic organisms early in the data processing pipeline: at the segmentation stage. We compared three segmentation methods, particularly for smaller targets, in which plankton represents less than 1% of the objects: (i) a traditional thresholding over the background, (ii) an object detector based on maximally stable extremal regions (MSER), and (iii) a content-aware object detector, based on a Convolutional Neural Network (CNN). These methods were assessed on a subset of ISIIS data collected in the Mediterranean Sea, from which a ground truth dataset of > 3,000 manually delineated organisms is extracted. The naive thresholding method captured 97.3% of those but produced ~340,000 segments, 99.1% of which were therefore not plankton (i.e. recall = 97.3%, precision = 0.9%). Combining thresholding with a CNN missed a few more planktonic organisms (recall = 91.8%) but the number of segments decreased 18-fold (precision increased to 16.3%). The MSER detector produced four times fewer segments than thresholding (precision = 3.5%), missed more organisms (recall = 85.4%), but was considerably faster. Because naive thresholding produces ~525,000 objects from 1 minute of ISIIS deployment, the more advanced segmentation methods significantly improve ISIIS data handling and ease the subsequent taxonomic classification of segmented objects. The cost in terms of recall is limited, particularly for the CNN object detector. These approaches are now standard in computer vision and could be applicable to other plankton imaging devices, the majority of which pose a data management problem. 

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. 

Abstract. Marine particles of different nature are found throughout the globalocean. The term “marine particles” describes detritus aggregates andfecal pellets as well as bacterioplankton, phytoplankton, zooplankton andnekton. Here, we present a global particle size distribution datasetobtained with several Underwater Vision Profiler 5 (UVP5) camerasystems. Overall, within the 64 µm to about 50 mm size range coveredby the UVP5, detrital particles are the most abundant component of allmarine particles; thus, measurements of theparticle size distribution with the UVP5 can yield importantinformation on detrital particle dynamics. During deployment, which ispossible down to 6000 m depth, the UVP5 images a volume of about 1 Lat a frequency of 6 to 20 Hz. Each image is segmented in real time, andsize measurements of particles are automatically stored. All UVP5units used to generate the dataset presented here wereinter-calibrated using a UVP5 high-definition unit as reference. Ourconsistent particle size distribution dataset contains 8805 verticalprofiles 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. 

The rise of in situ plankton imaging systems, particularly high‐volume imagers such as the In Situ Ichthyoplankton Imaging System, has increased the need for fast processing and accurate classification tools that can identify a high diversity of organisms and nonliving particles of biological origin. Previous methods for automated classification have yielded moderate results that either can resolve few groups at high accuracy or many groups at relatively low accuracy. However, with the advent of new deep learning tools such as convolutional neural networks (CNNs), the automated identification of plankton images can be vastly improved. Here, we describe an image processing procedure that includes preprocessing, segmentation, classification, and postprocessing for the accurate identification of 108 classes of plankton using spatially sparse CNNs. Following a filtering process to remove images with low classification scores, a fully random evaluation of the classification showed that average precision was 84% and recall was 40% for all groups. Reliably classifying rare biological classes was difficult, so after excluding the 12 rarest taxa, classification accuracy for the remaining biological groups became > 90%. This method provides proof of concept for the effectiveness of an automated classification scheme using deep‐learning methods, which can be applied to a range of plankton or biological imaging systems, with the eventual application in a variety of ecological monitoring and fisheries management contexts.

 

Plankton imaging systems supported by automated classification and analysis have improved ecologists' ability to observe aquatic ecosystems. Today, we are on the cusp of reliably tracking plankton populations with a suite of lab‐based and in situ tools, collecting imaging data at unprecedentedly fine spatial and temporal scales. But these data have potential well beyond examining the abundances of different taxa; the individual images themselves contain a wealth of information on functional traits. Here, we outline traits that could be measured from image data, suggest machine learning and computer vision approaches to extract functional trait information from the images, and discuss promising avenues for novel studies. The approaches we discuss are data agnostic and are broadly applicable to imagery of other aquatic or terrestrial organisms.