Patterns of mesozooplankton community composition and vertical fluxes in the global ocean

Get Complete Project Material File(s) Now! »

Zooplankton sampling, digitization, biomass estimates

A Hydrobios Multinet (with a 300μm mesh and an aperture of 0.25m2) was used to sample zooplankton (Roullier et al., 2014; Pesant et al., 2015) in five distinct water layers ranging from the surface to occasionally 1300 m depth. The five depth layers were locally defined as a function of the vertical structure of the water column according to the profiles of temperature, salinity, fluorescence, nutrients, oxygen, and particulate matter (Pesant et al., 2015). The Multinet was equipped with a flowmeter to measure the volume of seawater filtered by each net tow (Pesant et al., 2015). Day and night net tows were conducted at ten stations. Sampling at the other stations occurred at day or night, depending on cruise schedule and operational constraints. Once collected, the samples were preserved in a solution of buffered formaldehyde-Borax solution (4%). In the laboratory, the samples were rinsed and split with a Motoda box (MOTODA, 1959). The final split was analysed with the Zooscan imaging system (Gorsky et al., 2010) which allowed a rapid and semi-automatic analysis of zooplankton samples. In total, the samples comprised nearly 400,000 images of living zooplankton and detritus. These images were imported into Ecotaxa, an online platform which allows an automatic prediction of the taxonomic classification of each single image followed by a manual validation/correction. The organisms were then identified manually down to the order, sometimes to the family and rarely down to the genus level. All copepods were sorted at the family level apart from the smallest copepods that cannot be recognised at the family level from the image. They were all grouped into one category called Other-copepoda or other-Calanoida if their morphological features allowed classifying them as Calanoida. This initial sorting allowed classifying zooplankton into 119 taxa. As many taxa showed a very small contribution to total zooplankton abundance, the 119 taxa were grouped into 19 taxonomic groups (Table 1). Those include all the major zooplankton groups that are frequently observed in the oceans. To investigate vertical patterns in mesozooplankton abundance, these 19 groups were further aggregated into eight groups representing a combination of taxonomic and functional classification (Table 1). Once the zooplankton images were sorted, Ecotaxa enabled us to extract the concentration and the biovolume of each mesozooplankton group at every station and for every net tow, while accounting for the Motoda fraction and the volume sampled. The biovolume was computed for each individual zooplankton using the minor and major length axes assuming a prolate ellipsoidal shape (Gorsky et al., 2010). The biomass was calculated for the 8 large groups using the equations for the different taxa : 𝐵𝑜𝑑𝑦 𝑊𝑒𝑖𝑔ℎ𝑡(μ𝐶)=𝑎𝑆𝑏.

Analyzing zooplankton and particles vertical distributions

Vertical attenuation rates of zooplankton (abundance and biomass) and estimated particle fluxes were estimated, from the five sampled layers for zooplankton and from the 5 meter resolution profile of estimated vertical flux using a linear regression of the log-log (i.e. natural logarithm) with the following equation : 𝑋=𝑋0(𝑍/𝑍0)𝑏 (3).
where 𝑋 represents the zooplankton abundance, the zooplankton biomass or the particle vertical flux at the depth level 𝑍, 𝑋0the zooplankton biomass or abundance and vertical particle flux at the depth 𝑍0 (chosen as median depth of the surface net) and 𝑏 the slope taken as a proxy of the attenuation rate of zooplankton biomass zooplankton abundance or particle flux. In the rest of the manuscript, A_zoo represents the slope b of vertical profile for zooplankton abundance, B_zoo the slope b of vertical profile for zooplankton biomass, A_flux the slope b of vertical profile for particle flux, and P_flux1, P_flux2 and P_flux3 the particle flux in respectively the epipelagic, upper and lower mesopelagic. To analyse latitudinal patterns in attenuation rates, the slope values were separated into three latitudinal bands based on the latitudinal position of their corresponding sampling stations: intertropical (0°-30°), temperate (30-60°) and polar (60°-90°). The intertropical stations gathered both OMZ and non-OMZ stations, which allowed us to analyse the effect of oxygen depletion on zooplankton and particles. Non-parametric variance analyses (Kruskal and Wallis tests) were performed to test for differences in slope values (i.e. zooplankton and particles attenuation rates) between latitudinal bands.

Multivariate analysis of community composition

To explore the response of zooplankton community composition to environmental drivers across depth layers, the non-interpolated abundances of the 19 taxonomic groups mentioned above were aggregated into three layers: the epipelagic layer (0-200m), the upper mesopelagic layer (200-500m) and the lower mesopelagic layer (500-1000m). To analyse separately the three depth layers, the samples collected in overlapping layers (18.59% of the total samples) were not included in the statistical analysis (Table S1). To characterise the environmental conditions of each layer at each sampling station the median values of the following contextual environmental variables were used: temperature (T), salinity (S), oxygen (O2), nitrate concentration (NO3), chlorophyll a concentration (Chl_a), phytoplankton size fractions (𝑓_𝑚𝑖𝑐𝑟𝑜, 𝑓_𝑛𝑎𝑛𝑜, and 𝑓_𝑝𝑖𝑐𝑜), concentration of suspended particles (SPM) and particle flux (P_Flux). The measurements of all these environmental variables are available on PANGAEA ( To estimate the strength of the Diel Vertical Migration (DVM) at 10 stations, pairwise Wilcoxon tests were performed to compare in each layers the abundance and biomass of each taxa between day and night. For those 10 same pairs of stations, we used an analysis of similarities (ANOSIM) to test for significant variations in community composition between day and night samples. The ANOSIM tested whether the inter-groups difference (day and night groups) was higher than the intra-groups difference, by providing an R coefficient. An R coefficient close to one suggests dissimilarity between groups, while R value close to zero suggests an even distribution of high and low ranks within and between groups. An R value below zero suggests that dissimilarities are greater within groups than between groups (Clarke and Gorley, 2001). ANOSIM tests were performed within each layer using log-transformed (where log is the natural logarithm) abundances and Bray-Curtis distance among stations.
For each depth layer, a canonical redundancy analysis (RDA) was performed based on the abundances of the 19 mesozooplankton groups and the above-mentioned environmental variables to explore the explanatory power of these variables in structuring the mesozooplankton community. The RDA is an extension of the multiple regression analysis applied to multivariate data (P. Legendre and Legendre, 1998). It allows representing the response variables (abundances of the 19 mesozooplankton groups) in a “constrained” reduced space, i.e., constructed from the explanatory variables (the environmental variables). For each RDA, the following variables were used as “supplementary variables” of the analysis in order to visualize their correlation with the environmental structuring of the mesozooplankton community (i.e., to visualise their position in the RDA space): attenuation of particle flux (A_flux), attenuation of zooplankton abundance (A_zoo), attenuation of zooplankton biomass (B_zoo) and the Shannon index (H’). Beforehand, a Hellinger transformation was performed on the mesozooplankton abundances. A preliminary RDA based on all samples together showed a very strong effect of depth on mesozooplankton community composition (Fig. S1). Therefore, to avoid such a strong effect of depth on the community composition analysis, three separate RDAs were performed on each of the three layers defined above. Significant axes were identified using the Kaiser-Guttman criterion (Legendre and Legendre, 1998).
Data manipulation and statistical analyses were performed with Matlab 2018b (MATLAB 9.5) for the vertical profiles of abundance and biomass and statistical test (Wilcoxon test, Kruskal-wallis test), R environment v3.5.1 (using the following packages: vegan version 2.5-5, ggplot2 version 3.1.1, ggrepel version 0.8.0 and ggfortify version 0.4.7) for the redundancy analysis and PRIMER6 (Version 6.1.12) and PERMANOVA+ (Version 1.0.2) for the ANOSIM test.

Important environmental factors for the mesozooplankton community composition in the epipelagic layer

High latitude marine ecosystems are characterised by a combination of lower species diversity and shorter food webs (Laws, 1985; Stempniewicz et al., 2007) sustained by higher concentrations of large microphytoplankton cells (diatoms). On the contrary, low latitude ecosystems are featured by more complex and diverse food webs (Saporiti et al., 2015; Uitz et al., 2006) adapted to lower production rates ensured by smaller cells (i.e. pico- and nanoplankton). How the dynamics of zooplankton community composition and vertical particle flux follow this scheme remains more elusive. Previous field-based studies reported peaks in zooplankton species richness in the tropics (Rombouts et al., 2010; Rutherford et al., 1999; Yasuhara et al., 2012), which is in line with the above-mentioned theory that the tropical food-depleted regions promote more complex food-webs with higher species richness. Our RDAs results result supports the view that, on a global scale, temperature and the production regime of surface ecosystems are the main drivers of zooplankton community structure in the epipelagic layer. Therefore, our observations fall in line with the theory described above: in the epipelagic layers, less abundant and more diverse zooplankton thrive in warm, pico- and nanophytoplankton-dominated waters contrary to the polar waters where zooplankton is much more abundant but less diverse. Polar waters are characterised by a higher contribution of microphytoplankton to total phytoplankton biomass and higher concentrations of particles. Our results also indicate that the two polar communities are not completely alike, as the Arctic is dominated by calanoid copepods while euphausiids and small undetermined copepods dominate in the few stations sampled in the Southern Ocean. Such differentiation has been previously shown by several authors who found that Arctic zooplankton were dominated by Calanus spp. (Balazy et al., 2018a; H-J Hirche and Mumm, 1992), whereas it was shown that Antarctic zooplankton were dominated by euphausiids, small calanoids, cyclopoids (i.e. Oithona spp., Oncaea spp.) and salps (Park and Wormuth, 1993; Quetin et al., 1996; Ross et al., 2008).

READ  Communication Significance among Cross-Functional Team

Important environmental factors for the mesozooplankton community structure in the mesopelgic layer

The RDA displayed similar general patterns in both the upper and lower mesopelagic layers with the stations in the Indian Ocean (with low zooplankton concentration in low oxygen) and Arctic Ocean (with high zooplankton concentration in oxygenated conditions) being in all layers well separated from the other stations in the remaining ocean. The highest mesopelagic zooplankton concentrations were found at stations associated with high microphytoplankton concentrations in the epipelagic and high particle flux, suggesting a strong impact of surface production regime on the mesozooplankton in the mesopelagic (Hernández-León et al., 2020). The inter-basin differences in zooplankton concentration and community composition was slightly lower in the lower mesopelagic compared to the upper mesopelagic, probably due to the more homogeneous habitat (Fernández de Puelles et al., 2019).
In general, stations associated with anoxic or hypoxic conditions at midwater depth displayed lower zooplankton concentrations in the mesopelagic and different community composition from oxygenated layers. The stations that sampled the tropical OMZ showed higher proportions of gelatinous carnivorous zooplankton (Cnidaria), gelatinous filter feeders (Tunicata), Mollusca and small omnivorous grazers (Cladocera). Anoxic or strongly hypoxic conditions in the mesopelagic may have selected for those taxa adapted to low oxygen (Vaquer-Sunyer and Duarte, 2008a). In mesopelagic oxygenated waters, the stations of the Tropical Pacific Ocean displayed a higher diversity stemming from the higher and relatively even abundances of large protists (i.e. Rhizaria and Foraminifera), chaetognaths, crustaceans (Ostracods, Euphausiacea, Eumalacostraca, Amphipoda), and various copepod families (Corycaeidae, Oithonidae, Oncaeidae, small Calanoida). These zooplankton communities were associated with oligotrophic conditions at surface (Fig. 4), lower zooplankton abundances, lower concentrations of suspended matter, weaker particle fluxes and weaker attenuation rates (Fig. 3B) and with stronger attenuation rates of organisms’ abundances (Fig. 3A). Therefore, we evidence oligotrophic regimes where the zooplankton community constitutes a network of diverse taxa that is not as efficient at using the low amount of material fluxing. Our results are consistent with previous studies suggesting that those oligotrophic regimes can be relatively efficient at exporting the slow-sinking fraction of the little carbon that is produced in the surface layers due to a low impact of zooplankton grazing on the sinking particles (Guidi et al., 2009; Henson et al., 2015). This could help explain why we found the lowest particle flux attenuation rates in the tropics.

Zooplankton sampling and digitization

Zooplankton samples were collected by Hydrobios Multinet composing of five sequential plankton nets equipped with a 300μm mesh and an aperture of 0.25m2, equipped with a flow meter to calculate the volume of filtered water that ranged from 5 to 502 m3 (median value of 113 m3) (Stéphane Pesant et al., 2015). Plankton sampling was carried out at 57 stations and at five different depth levels between the surface and 1300 m. Only 10 stations were sampled both at day and night. Once collected, the samples were preserved in a solution of buffered formaldehyde (4%). In the laboratory, the samples were rinsed and fractionated with a Motoda box. The final fraction was analyzed with the ZOOSCAN which allowed a rapid and time-efficient analysis of the plankton samples and stored the data in digital form (Gorsky et al., 2010). The resolution of the ZOOSCAN (2400 dpi) allowed for morphometric measurements and a taxonomic classification of the images that was adequate for the goal of our study. The observed volume by the UVP5 for each sample was much lower ranging from 0.033 to 3.77 m3 (median value of 0.69 m3).
The multinet samples comprised nearly 400.000 images of organisms. Images were imported into Ecotaxa (, an online platform connected to the database which allowed to visually validate the taxonomic classification of each particle/organism previously semi-automatic predicted by the Ecotaxa web application. The images were sorted into the different groups for each instrument. For the Zooscan, the high definition images allowed to identify down to the family (and sometimes down to genus). Those taxa were further aggregated into 19 taxonomic groups (Table SM1) for which abundances were high enough for the following statistical analysis including the NBSS estimates.
Once the zooplankton images were sorted, Ecotaxa enabled to extract the concentration of each taxa and the biovolume of each living organism for every station and net (depth interval). The concentrations were used to describe the community composition. The biovolume was computed for each individual zooplankton using the minor and major length provided by the software and computed assuming an ellipsoidal shape (Gorsky et al., 2010). Because the sampling depths were not constant during the cruise, most further statistical analysis were performed in three depth layers, 0-200. 200-500 and 500-1000m depth by calculating mean concentrations in each layers. Few nets intersecting the layers were removed in that case.

Table of contents :

Chapitre I: Introduction générale
I.1. Le plancton et le cycle du carbone dans l’océan
I.2. La zone mésopélagique:
I.3. Plancton et changement climatique
I.4. Les zones à minimum d’oxygène
I.5 La structure en taille du plancton marin
I.6 Taux physiologiques
I.7. Objectifs de cette thèse
Chapitre II. Patterns of mesozooplankton community composition and vertical fluxes in the global ocean
II.1. Résumé
II.2. Article soumis et accepté
Materials and methods
Supplementary material
II.3. Comparaison des communautés entre les biogéographies spatiales existantes dans l’océan global
Meilleur prédicteur de la distribution spatiale du zooplancton
Chapitre III.1: Diversity of zooplankton NBSS in the global ocean
III.1.1. Résumé
III.1.2. Article en préparation
Material and methods
Supporting Material
Chapitre III.2: Contribution of plankton to the Normalized Biovolume Size Distribution in the upper 1000m of open ocean
III.2.1. Résumé
III.2.2. Article en préparation
Materials and methods
Chapitre IV : Distribution en taille, budget de la respiration et de l’excrétion des crustacés dans la zone inter-tropicale de l’océan mondial : TARA Océans- GEOMAR-MALASPINA
IV.1. Introduction
IV.2. Matériels et Méthodes
IV.2.1. Les données d’hydrologie, d’optique, d’imagerie et satellites
IV.2.2. Echantillons du Zooplancton et digitalisation
IV.2.3. Identification des images
IV.2.4. Construction du spectre et distribution en taille du plancton
IV.2.5. Analyse statistique
IV.2.6. Estimation des Biomasses et des taux de respiration et d’excrétion
IV.3. Résultats
IV.3.1. Variabilité des NBSS
IV.3.2. Sources de variance
IV.3.3. Pentes spectrales, tailles et biovolumes des crustacés
IV.3.4. Respiration et excrétion des crustacés
IV.4. Discussion
IV.4.1. Qualité des données et les limites de notre travail
IV.4.2. Les sources de variance dans la distribution spatiale des crustacés
IV.4.3. Les grands types de NBSS des crustacés
IV.4.4. Biomasse, Respiration, excrétion
IV.4.5. Contribution de la respiration des crustacés au Budget de carbone.
IV.5. Conclusion
IV.6. Références
Chapitre V: Conclusion générale et perspectives
V.1. Résumé des résultats principaux
V.1.1. Structure verticale de la communauté du mésozooplancton dans l’océan mondial
V.1.2. Distribution 3D des NBSS du zooplancton
V.1.3. Etudes comparatives des NBSS collectés et in situ
V.1.4. Budget carbone des crustacés dans les zones inter-tropicales
VI.2. Perspectives générales


Related Posts