TPThelma Panaïotis

Thelma Panaïotis

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Thelma Panaïotis has a total of 115 co-authors. Here are the top co-authors by number of shared publications:

  • Jean-Olivier Irisson

    Sorbonne Université, CNRS, Laboratoire d'Océanographie de Villefranche, LOV, 06230 Villefranche-sur-mer, France

    9 shared publications
  • Barry B. Cael

    The University of Chicago Physical Sciences Division, Unknown, Chicago, IL, United States [US

    7 shared publications
  • Lars Stemmann

    Laboratoire d’Océanographie de Villefranche (Villefranche-sur-Mer, France)

    5 shared publications
  • Marc Picheral

    Sorbonne Université, CNRS, Laboratoire d'Océanographie de Villefranche, LOV, 06230 Villefranche-sur-mer, France

    5 shared publications
  • Marcel Babin

    Département de biologie, Université Laval, Québec, Canada

    5 shared publications
  • Amanda Elineau

    Sorbonne Université, CNRS, Laboratoire d'Océanographie de Villefranche, LOV, 06230 Villefranche-sur-mer, France

    4 shared publications

Latest Publications

  • Open Access

    A global consistent database of plankton and detritus from in situ imaging by the Underwater Vision Profiler 5

    by Ariadna C. Nocera, Lars Stemmann, Marcel Babin, Tristan Biard, Julie Coustenoble, François Carlotti, Laurent Coppola, Lucas Courchet, Laetitia Drago, Amanda Elineau, et al.

    Plankton and detritus are essential components of the Earth's oceans influencing biogeochemical cycles and carbon sequestration. Climate change impacts their composition and fluxes, and marine ecosystems as a whole. To improve our understanding of these changes, standardized observation methods and integrated global datasets are needed to enhance the accuracy of ecological and climate models. Here, we present a global dataset for plankton and detritus obtained by two versions (SD and HD) of the Underwater Vision Profiler 5 (UVP5). This release contains the images classified in 33 homogenized categories, as well as the metadata associated with them, reaching 3114 profiles and ca. 8 million objects acquired between 2008–2018 at global scale. The geographical distribution of the dataset is unbalanced, with the Equatorial region (30° S–30° N) being the most represented, followed by the high latitudes in the northern hemisphere and lastly the high latitudes in the Southern Hemisphere. Detritus is the most abundant category in terms of concentration (90 %) and biovolume (95 %), although its classification in different morphotypes is still not well established. Copepoda was the most abundant planktonic taxon, followed by Trichodesmium colonies. The two versions of UVP5 have different imagers, resulting in a different effective size range to analyze plankton and detritus from the images (HD objects > 600 µm, SD objects > 1 mm) and morphological properties (grey levels, etc.); however, both systems capture comparable qualitative patterns in object morphology and relative abundance across size classes, even though the absolute ranges differ. Therefore, recommendations are provided for the appropriate use of this data when conducting studies. A large number of images of plankton and detritus will be collected in the future by the UVP5, and the public availability of this dataset will help it being utilized as a training set for machine learning and being improved by the scientific community. This will reduce uncertainty by identifying previously unclassified objects and expand the classification categories, ultimately enhancing biodiversity quantification. The dataset that constitutes this first release is available on SEANOE at https://doi.org/10.17882/107583 (Nocera et al., 2025).

  • Open Access

    Assessing Zooplankton Communities' Influence on Particulate Organic Carbon Concentration

    by Thelma Panaïotis, B. B. Cael

    Zooplankton play a key role in setting the particulate organic carbon (POC) distribution in the ocean, but this role remains poorly quantified on large scales due to the complexity of zooplankton ecosystems and the sparsity and variability of observations. We address this by applying boosted regression trees to a global in situ zooplankton image database from the Underwater Vision Profiler 5 and an Argo float and satellite‐derived POC product. We compute taxonomic, morphological, and trophic community metrics, then identify three metrics—abundance, mean gray level, taxonomic evenness—that explain almost half of the spatial variability in the vertical concentration gradients of small POC (<100 μm). Partial dependence analysis and the consistency of the identified relationships with known zooplankton‐mediated particle‐processing pathways suggest mechanistically interpretable ecological linkages. These results advance quantification of zooplankton communities' influence on ocean carbon cycling and indicate carbon‐cycle‐relevant zooplankton community properties for improving ocean biogeochemical models.

  • Open Access

    Identifying the Environmental Drivers of Spatial Variability in Zooplankton Community Grazing Dynamics

    by S. A. Meyjes, B. B. Cael, C. M. Petrik, T. Panaiotis, T. Rohr, A. Mashayek

    Biogeochemical models used to estimate carbon export fluxes are highly sensitive to their parameterization of zooplankton grazing dynamics. Zooplankton grazing dynamics have been shown to vary spatially, however it is less clear how they are responding to underlying environmental drivers. In this study, we use machine learning regression models to quantify the relationship between spatial variability in observationally inferred zooplankton grazing dynamics and three environmental drivers. The majority ( = 0.80) of the variance in zooplankton grazing dynamics can be explained by sea surface temperature, mixed layer depth and chlorophyll. Globally, chlorophyll is the best predictor of zooplankton grazing dynamics, with slower grazing plankton communities associated with higher chlorophyll concentrations and deeper mixed layer depths. The findings show that zooplankton community grazing and composition is well constrained by biological drivers, which could provide a pathway to parameterize more diverse grazing dynamics in models and in turn simulate more realistic carbon export estimates.

  • Open Access

    A closer look at plankton: potential interactions inferred from centimeter-scale in situ observations

    by Thelma Panaiotis, Jean-Olivier Irisson, Mara Freilich, B.B. Cael

    Plankton are essential to marine ecosystems, supporting food webs and mediating biogeochemical processes such as carbon export to depth. Their spatial distribution influences ecosystem dynamics and serves as an indicator of environmental change. Although drifting plankton could be expected to exhibit random distribution, numerous studies have revealed significant heterogeneity in their spatial patterns. However, very few studies targeted plankton distribution at the centimeter scale in situ, despite its importance for understanding biological processes. We argue that centimeter-scale distances in plankton could reveal potential ecological interactions. Using an extensive in situ dataset of 18 million planktonic organisms collected by the In Situ Ichthyoplankton Imaging System (ISIIS), which images multiple organisms simultaneously and preserves their positions in the water column, we analyzed centimeter-scale distances in plankton. By comparing observed distances with those expected under a random distribution, we assessed potential interactions at three levels: among all organisms, within plankton groups and across groups. Our results show that planktonic organisms exhibit non-random distributions at the centimeter scale, with smaller distances than expected, suggesting potential ecological interactions. Notably, distances up to 11 cm were the most informative, which is much larger than typical interaction distances in plankton. Additionally, observed distances were compatible with a simple attraction model. Finally, we propose the non-randomness of distances as a novel metric of interaction strength in plankton ecological networks and compare it against classical empirical or co-occurrence networks. These results offer new insights into in situ interactions and how they shape plankton distribution at centimeter scale.

  • Plankton distribution across scales : contributions from artificial intelligence to plankton ecology

    by Thelma Panaïotis

    Distribution du plancton à diverses échelles : apport de l'intelligence artificielle pour l'écologie planctonique: En tant que base des réseaux trophiques océaniques et élément clé de la pompe à carbone biologique, les organismes planctoniques jouent un rôle majeur dans les océans. Cependant, leur distribution à petite échelle, régie par les interactions biotiques entre organismes et les interactions avec les propriétés physico-chimiques des masses d'eau de leur environnement immédiat, est mal décrite in situ, en raison du manque d'outils d'observation adaptés. De nouveaux instruments d'imagerie in situ à haute résolution, combinés à des algorithmes d'apprentissage automatique pour traiter la grande quantité de données collectées, nous permettent aujourd'hui d'aborder ces échelles. La première partie de ce travail se concentre sur le développement méthodologique de deux pipelines automatisés basés sur l'intelligence artificielle. Ces pipelines ont permis de détecter efficacement les organismes planctoniques au sein des images brutes, et de les classer en catégories taxonomiques ou morphologiques. Dans une deuxième partie, des outils d'écologie numérique ont été appliqués pour étudier la distribution du plancton à différentes échelles, en utilisant trois jeux de données d'imagerie in situ. Tout d'abord, nous avons mis en évidence un lien entre les communautés planctoniques et les conditions environnementales à l'échelle globale. Ensuite, nous avons décrit la distribution du plancton et des particules à travers un front de méso-échelle, et mis en évidence des périodes contrastées pendant le bloom de printemps. Enfin, grâce aux données d'imagerie in situ à haute fréquence, nous avons étudié la distribution à fine échelle et la position préférentielle d’organismes appartement au groupe des Rhizaria, des protistes fragiles et peu étudiés, dont certains sont mixotrophes. Dans l’ensemble, ce travail démontre l'efficacité de l'imagerie in situ combinée à des approches d’intelligence artificielle pour comprendre les interactions biophysiques dans le plancton et les conséquences sur sa distribution à petite échelle.

  • Open Access

    Assessing zooplankton communities' influence on particulate organic carbon concentration

    by Thelma Panaïotis, B. B. Cael

    Zooplankton play a key role in setting the particulate organic carbon (POC) distribution in the ocean, but this remains poorly quantified on large scales due to the complexity of zooplankton ecosystems and the sparsity and variability of observations. We address this by applying boosted regression trees to a global in situ zooplankton image database from the Underwater Vision Profiler 5 and an Argo float and satellite-derived POC product. We compute taxonomic, morphological, and trophic community metrics, then identify three metrics − abundance, mean grey level, and taxonomic evenness − that explain almost half of the spatial variability in the vertical concentration gradients of small suspended POC (< 100 µm). Partial dependence analysis, weak latitudinal gradients, and mechanistic plausibility suggest these relationships are not merely correlative. These results advance quantification of zooplankton communities’ influence on ocean carbon cycling and indicate carbon-cycle-relevant zooplankton community properties for improving ocean biogeochemical models.

  • Open Access

    Benchmark of plankton images classification: emphasizing features extraction over classifier complexity

    by Thelma Panaïotis, Emma Amblard, Guillaume Boniface-Chang, Gabriel Dulac-Arnold, Benjamin Woodward, Jean-Olivier Irisson

    Plankton imaging devices produce vast datasets, the processing of which can be largely accelerated through machine learning. This is a challenging task due to the diversity of plankton, the prevalence of non-biological classes, and the rarity of many classes. Most existing studies rely on small, unpublished datasets that often lack realism in size, class diversity and proportions. We therefore also lack a systematic, realistic benchmark of plankton image classification approaches. To address this gap, we leverage both existing and newly published, large, and realistic plankton imaging datasets from widely used instruments (see Data Availability section for the complete list of dataset DOIs). We evaluate different classification approaches: a classical Random Forest classifier applied to handcrafted features, various Convolutional Neural Networks (CNN), and a combination of both. This work aims to provide reference datasets, baseline results, and insights to guide future endeavors in plankton image classification. Overall, CNN outperformed the classical approach but only significantly for uncommon classes. Larger CNN, which should provide richer features, did not perform better than small ones; and features of small ones could even be further compressed without affecting classification performance. Finally, we highlight that the nature of the classifier is of little importance compared to the content of the features. Our findings suggest that compact CNN (i.e. modest number of convolutional layers and consequently relatively few total parameters) are sufficient to extract relevant information to classify small grayscale plankton images. This has consequences for operational classification models, which can afford to be small and quick. On the other hand, this opens the possibility for further development of the imaging systems to provide larger and richer images.


  • Open Access

    Identifying the environmental drivers of spatial variability in zooplankton community grazing dynamics

    by Sophie Alexandra Meyjes, B. B. Cael, Colleen M Petrik, Thelma Panaïotis, Tyler Weaver Rohr, Ali Mashayek

    Biogeochemical models used to estimate carbon export fluxes are highly sensitive to their parameterisation of zooplankton grazing dynamics. Zooplankton grazing dynamics have been shown to vary spatially, however, it is less clear how they are responding to underlying environmental drivers. In this study, we use machine learning regression models to quantify the relationship between spatial variability in observationally-inferred zooplankton grazing dynamics and three environmental drivers. The majority (R2=0.80) of the variance in zooplankton grazing dynamics can be explained by sea surface temperature, mixed layer depth and chlorophyll. Globally, chlorophyll is the best predictor of zooplankton grazing dynamics, with slower grazing plankton communities associated with higher chlorophyll concentrations and deeper mixed layer depths. The findings show that zooplankton community grazing and composition is particularly well constrained by biological drivers, which could provide a pathway to parameterise more diverse grazing dynamics in models and in turn simulate more realistic carbon export estimates.

  • Open Access

    A global consistent database of plankton and detritus from in situ imaging by the Underwater Vision Profiler 5

    by Ariadna Celina Nocera, Lars Stemmann, Marcel Babin, Tristan Biard, Julie Coustenoble, François Carlotti, Laurent Coppola, Lucas Courchet, Laetitia Drago, Amanda Elineau, et al.

    Plankton and detritus are essential components of the Earth’s oceans influencing biogeochemical cycles and carbon sequestration. Climate change impacts their composition and marine ecosystems as a whole. To improve our understanding of these changes, standardized observation methods and integrated global datasets are needed to enhance the accuracy of ecological and climate models. Here, we present a global dataset for plankton and detritus obtained by two versions of the Underwater Vision Profiler 5 (UVP5). This release contains the images classified in 33 homogenized categories, as well as the metadata associated with them, reaching 3,114 profiles and ca. 8 million objects acquired between 2008–2018 at global scale. The geographical distribution of the dataset is unbalanced, with the Equatorial region (30° S – 30° N) being the most represented, followed by the high latitudes in the northern hemisphere and lastly the high latitudes in the Southern Hemisphere. Detritus is the most abundant category in terms of concentration (90 %) and biovolume (95 %), although its classification in different morphotypes is still not well established. Copepoda was the most abundant taxa within the plankton, with Trichodesmium colonies being the second most abundant. The two versions of UVP5 (SD and HD) have different imagers, resulting in a different effective size range to analyse plankton and detritus from the images (HD objects >600 µm, SD objects >1 mm) and morphological properties (grey levels, etc.) presenting similar patterns, although the ranges may differ. Therefore, recommendations are provided for the appropriate use of this data when conducting studies. A large number of images of plankton and detritus will be collected in the future by the UVP5, and the public availability of this dataset will help it being utilized as a training set for machine learning and being improved by the scientific community. This will reduce uncertainty by classifying previously unclassified objects and expand the classification categories, ultimately enhancing biodiversity quantification. The dataset that constitutes this first release is available at SEANOE.


  • Open Access

    High throughput in situ imaging reveals complex ecological behaviour of giant marine mixotrophic protists

    by Thelma Panaïotis, Tristan Biard, Louis Caray--Counil, Robin Faillettaz, Jessica Y. Luo, Cedric M. Guigand, Robert K. Cowen, Jean-Olivier Irisson

    Mixotrophic organisms have the ability to combine photosynthetic and heterotrophic nutrition, and are ubiquitous in marine ecosystems. Here, we focus on Rhizaria, a diverse group of planktonic protists that included both mixotrophic and non-mixotrophic taxa, and that are particularly delicate and poorly sampled by standard plankton nets. Using high-frequency in situ imaging, we study the ecology of these organisms directly in their undisturbed environment. We examined the fine-scale distribution and orientation of ~230,000 organisms belonging to three groups of Rhizaria, including the mixotrophic taxa Acantharia and Collodaria, and the non-mixotrophic Phaeodaria. Overall, our results suggest that mixotrophic protists have better ability to control their position within the water column (depth, orientation) than non-mixotrophic ones. Our observations also point to several steps in the obscure life cycle of the mixotrophic Collodaria, during which de novo symbiont acquisition appears to involve active fine-scale buoyancy control. Taken together, these unprecedented results demonstrate that complex ecological behaviour can be achieved by "simple" organisms.