2.1 Study site and sampling strategy

The OUTPACE survey was performed aboard the RV L'Atalante during austral summer conditions between 18 February and 3 April 2015 in the WTSP Ocean, from New Caledonia (western part of the Melanesian archipelago) to French Polynesia, along a west–east transect covering ca. 4000 km between 17 and 22^∘ S (Fig. 1). This region is impacted by ENSO, known to be the most important mode of sea surface temperature (SST) variability on interannual to decadal timescales (Sarmiento and Gruber, 2006). The year 2015 was classified as an El Niño event, which was reflected in SST and Chl a satellite data (Moutin et al., 2017). Along this transect, two types of stations were sampled (Fig. 1): 15 short-duration stations (SD1 to SD15, 8 h) dedicated to a large-scale description and 3 long-duration stations for Lagrangian process studies. These three stations are as follows: station LD-A (19^∘12.8^′ S–164^∘41.3^′ E, 25 February–2 March) positioned in western Melanesian archipelago waters in the western part of the transect, offshore of New Caledonia; station LD-B (18^∘14.4^′ S–170^∘51.5^′ W, 15–20 March) in the eastern part of Melanesian archipelago waters, near Niue Island; and station LD-C (18^∘25.2^′ S–165^∘56.4^′ W, 23–28 March) in the eastern part of the transect, in the subtropical Pacific gyre, near the Cook Islands. All general characteristics of the stations are presented in Moutin et al. (2017, their Table 1).

Figure 1Transect of the OUTPACE cruise superimposed on quasi-Lagrangian-weighted mean Chl a of the WTSP during OUTPACE (see details in Moutin et al., 2017), with the two types of stations: short-duration stations 1 to 15 (×) and long-duration stations A, B and C (+). Along the transect, zooplankton samples were collected once at each short-duration station, whereas day–night sampling was performed each day at three strategic long-duration stations. Longitude is expressed as ^∘E. OUTPACE cruise (https://dx.doi.org/10.17600/15000900, Moutin et al., 2017).

Download

Real-time satellite images (altimetry, SST, ocean color) combined with drifter trajectories initiated during the first part of the cruise were used to define the best positions of these three stations on the basis of two criteria: sea surface chlorophyll levels to characterize the main sampled regions and minimum current intensity in each region to increase the chance of sampling a homogeneous water mass (Moutin et al., 2017; De Verneil et al., 2018). Stations LD-A and LD-B were characterized by local maxima of sea surface Chl a typical of the Melanesian archipelago zone, whereas Chl a minima characterized LD-C, representing typical waters of the subtropical gyre.

2.2 Meso-zooplankton sampling

Zooplankton collection was conducted at 14 of the SD stations (station SD-13 was not sampled for zooplankton) and at the 3 LD stations. SD stations were generally sampled during the day, except for SD-04 and SD-05, whereas LD stations were sampled once during the day and once during the night for each of the 5 days of station occupation. Sampling was done with a bongo net (70 cm mouth diameter) with 120 µm mesh nets mounted with filtering cod ends. The nets were equipped with HYDRO-BIOS flowmeters. Hauls were done from 200 m depth to the surface at a speed of 1 m s⁻¹. One of the cod ends was used for biomass measurements. The second one was preserved in 4 % buffered formaldehyde for later taxonomic identification, abundance, and size spectrum analyses. Volume filtered by the nets (V) was calculated using the formula $V=R\cdot S\cdot K$, combining the flowmeter counts (R; one count is a tenth of revolution), the mouth area of the net (S=0.38 m²), and the pitch of the impeller of the flowmeter (K) provided by the manufacturer, and it is equal to a 0.03 m count⁻¹.

2.3 Dry weight measurement

The biomass sample was processed onboard. Just after collection, each sample was filtered onto a pre-weighed GF/F filter (47 mm) and oven-dried at 60 ^∘C for 2 days. The average biomass concentration (in mg DW m⁻³) in the upper 200 m was calculated from the zooplankton dry weight (mg), obtained as the difference between the weight of the filter with and without the sample, taking into account the water-column-sampled volume. Biomass was also expressed in carbon, using a C∕DW ratio equal to 0.45 (Hansen et al., 1997).

2.4 Identification, abundance, and individual size and weight of the zooplankton taxa

The taxonomic composition was determined for each formalin sample. Samples were split using a Motoda box, and at least 100 individuals of the most abundant taxa were counted in each subsample under a dissecting microscope, a LEICA MZ6. Species/genus identification was done according to Rose (1933), Tregouboff and Rose (1957), and Razouls et al. (2005–2017). The abundance of the various taxa (groups, genera, or species) was divided by the sample volume to determine the concentration of individuals per cubic meter (ind. m⁻³). The diversity of the zooplankton was determined using the Shannon–Weaver index (Shannon and Weaver, 1949).

Approximates of the individual size (total length) and relative dimensions (length/width) of the different taxa were computed from literature values: summarized data for copepod species from Razouls et al. (2005–2017) and the mean size values of the other taxa from Tregouboff and Rose (1957) and Conway et al. (2003).

For comparison with ZooScan results (see below), we computed the body area of each taxon (A) from its dimensions to calculate its equivalent circular diameter (ECD):

We also estimated individual dry weight (DW) from the area (A) using the relationships obtained by Lehette and Hernández-León (2009) for subtropical copepods and meso-zooplankton.

2.5 Abundance, biomass, and size structure determined with the ZooScan

Samples were digitized with the ZooScan digital imaging system (Gorsky et al., 2010) to determine the size structure of the zooplankton communities, as detailed in Donoso et al. (2017). Each sample was divided into two fractions (<1000 and >1000 µm) and each fraction was then split using a Motoda box until it contained approximately 1000 objects. The resulting samples were poured onto the scanning cell, and zooplankton organisms were manually separated with a wooden spine in order to avoid overlapping organisms. After scanning, each image was processed using Zooprocess and the image analysis software Image-J (Grosjean et al., 2004; Gorsky et al., 2010). Only objects having an equivalent circular diameter (ECD) of >300 µm were detected and processed. Finally, the Plankton Identifier software (http://www.obs-vlfr.fr/~gaspari/Plankton_Identifier/index.php, last access: November 2018) was used for automatic classification of zooplankton into 12 categories. Among them, two categories of non-zooplankton organisms, aggregates, and fibers were grouped as detritus. A training set of about 1000 objects selected automatically from different scans was used to discriminate between and classify organisms, aggregates, and fibers. Afterwards, each scan was corrected using the automatic analysis of images.

Zooplankton abundance estimated from ZooScan (ind. m⁻³) was calculated from the number of validated vignettes in ZooScan samples, taking into account the scanned fraction and the sampled volume from the net tows. The zooplankton-estimated dry weight of each vignette was calculated from its area using the regression equation obtained for meso-zooplankton by Lehette and Hernández-León (2009).

Below, the terms “ZOOSCAN abundance” and “ZOOSCAN biomass” will indicate values derived from the laboratory ZooScan processing. The abundance and biomass of organisms were first calculated for four size fractions (<500, 500–1000, 1000–2000, and >2000 µm) based on their ECD and then summed to deliver the total average abundance and biomass per sample over the upper 200 m.

2.6 Stable isotope analyses

Nitrogen isotope ratios (δ¹⁵N) were measured for the zooplankton size fractions collected for biomass measurement, and for particulate organic matter (POM) samples collected at 5 m depth at each station. Zooplankton samples were first homogenized using a mortar and pestle and packaged into ∼1 mg subsamples. For POM analyses, water samples were collected in 4.4 L polycarbonate bottles at depths corresponding to 50 % and 1 % of light attenuation. The samples were immediately filtered on pre-burnt (450 ^∘C, 4 h) 25 mm GF/F filters. Stable isotope analysis was performed with an Integra CN, SerCon Ltd. EA-IRMS (elemental analyzer isotope-ratio mass spectrometry). δ¹⁵N values were determined in parts per thousand (‰) relative to the external standard of atmospheric N. Repeated measurements of an internal standard indicated measurement precision of ±0.13 ‰ for δ¹⁵N.

The mean δ¹⁵N value for each station was calculated as the mean of all size fractions, weighted by size fraction biomass. Subsequently, the contribution of DDN (%) to zooplankton δ¹⁵N (ZDDN) values at each station was calculated using a two-source mixing model as follows (Sommer et al., 2006):

where δ¹⁵N_zpl is the isotopic signature of the zooplankton collected; TEF is the trophic enrichment factor; δ¹⁵N_diazo is the isotopic signature of diazotrophs; δ¹⁵N_zplref is the isotopic signature of zooplankton assuming nitrate-based phytoplankton production. TEF and δ¹⁵N_diazo were set, respectively, at 2.2±0.3 ‰ (McCutchan et al., 2003; Vanderklift and Ponsard, 2003) and within a range of −1 ‰ to −2 ‰ (Montoya et al., 2002). δ¹⁵N_zplref was set at 6 ‰ for the Melanesian archipelago stations – a value calculated for the ocean west of New Caledonia where nitrogen fixation is reduced (Hunt et al., 2015) – and at 10.73 ‰ for the South Pacific gyre (GY) samples – the mean value of POM samples in the GY+2.2 ‰ trophic enrichment for the primary consumer level. Minimum, average, and maximum % ZDDN were estimated using the lower, mean, and upper limits of TEF and the δ¹⁵N_diazo values cited above.

2.7 Ancillary data from OUTPACE survey used for interpretations and comparisons

The acquisition of environmental data used in the present paper is presented in different companion papers. Briefly, temperature, salinity, and density were collected with a CTD (conductivity–temperature–depth) SeaBird SBE 9 and particle distribution with an underwater vessel profiler (UVP), both mounted on a rosette (de Verneil et al., 2018), whereas chlorophyll a and phaeophytin concentrations were estimated for different depths from Niskin bottle water samples using the fluorometric method (as described in Dupouy et al., 2018). The depth of the mixed layer (MLD) was calculated using a threshold density deviation of 0.03 kg m⁻³ from the value at a reference depth (de Verneil et al., 2018).

Integrated Chl a and POC (particulate organic carbon) were calculated from water samples collected at standard depths from the surface to 200 m, using Niskin bottles (see Spungin et al., 2018, for methodological details). Phytoplankton carbon biomass was estimated from Chl a using a C∕Chl a ratio of 50:1 (Wang et al., 2009). Abundance and distribution of unicellular (UCYN-A1, UCYN-A2, UCYN-B, and UCYN-C) and filamentous heterocystous (het-1) and non-heterocystous (Trichodesmium) diazotrophic microorganisms for all stations were taken from Stenegren et al. (2018, their Fig. 2). Primary productivity was determined using a ¹⁴C labeling method according to Van Wambeke et al. (2018).

Figure 2PCA on environmental variables: mixed-layer depth (MLD); total Chl a concentration (Chl a + Phae), % Chl a (ratio Chl a/Chl a + Phae), temperature, salinity averaged over the upper 0–200 m of the water column. Plots of the 15 stations (a) and variables (b) on the first factorial plan. The green circles delimit the clusters defined at a distance of 3.3: W-MA: western Melanesian archipelago; CE-MA: central and eastern Melanesian archipelago; BL: station B, blooming conditions; GY: subtropical gyre.

Download

A drifting array equipped with three PPS5 sediment traps and various captors was deployed at each LD station for 5 days at three depths (see Caffin et al., 2018a). Swimmers found in the trap were quantified and genera identified and weighed. Zooplankton C (Zoo-C), N (Zoo-N), and P (Zoo-P) mass was measured at each depth at each station (see methods in Caffin et al., 2018a). Only data from the sediment trap located at 150 m depth were used here.

2.8 Estimation of zooplankton carbon demand and grazing impact and of zooplankton excretion and respiration rates

The zooplankton carbon demand (ZCD in mgC m⁻³ d⁻¹) was computed based on estimates of biomass and of ration for each taxon:

where B_zoo is the biomass of zooplankton in mgC m⁻³, and Ration (d⁻¹) is the amount of food consumed per unit of biomass per day, calculated as

where g_z is the growth rate of zooplankton, r is the weight-specific respiration, with basal (r_b) and active (r_a) components, and A is assimilation efficiency.

Following Nival et al. (1975), we considered constant values of A=0.7 d⁻¹. For respiration, we applied a constant value for basal respiration (r_b=0.20 d⁻¹) derived from Hernández-León et al. (2008) for 20^∘ S zooplankton and assumed an activity-dependent respiration proportional to growth rate (r_a=0.25g_z) following Kiørboe et al. (1985).

g_z was calculated following Zhou et al. (2010):

as a function of sea water temperature (T, ^∘C), food availability (C_a, mgC m⁻³, estimated from Chl a), and weight of zooplankton individuals (w, mgC). C_a was used for herbivorous and omnivorous zooplankton taxa and replaced by POC for carnivorous zooplankton.

ZCD was thus estimated for each taxon and then summed to estimate the ZCD of total zooplankton. We considered two components: one for herbivorous and omnivorous zooplankton (ZCD_H) and one for carnivorous zooplankton (ZCD_C). To estimate the potential clearance of phytoplankton by zooplankton, we compared ZCD_H to the phytoplankton stock, converted to carbon assuming a classical C:Chl a ratio of 50:1, and to the phytoplankton primary production estimated by Van Wambeke et al. (2018). To estimate the grazing impact on phytoplankton size classes (pico-, nano-, and micro-phytoplankton), we applied the empirical relationship given by Wirtz (2012) to estimate the optimum prey size (D_opt, as µm equivalent spherical diameter) from the predator size (DZ, as µm ESD (equivalent spherical diameter)):

According to the root mean square deviations in log(D_opt) of the Wirtz regression model, we assumed a ±60 % food-size range around D_opt for each zooplankton taxon. When the calculated size range straddles the separation value between two phytoplankton size classes (e.g., 2 µm between pico- and nanoplankton), we assume that the grazing pressure on each phytoplanktonic class is proportional to the distance between the limit of the range and this separation value. Therefore, for each grazer, we implicitly assumed a constant clearance over the prey particle-size range.

Ammonium and phosphorus excretion rates were estimated for each taxon and station from the multivariate regression equations by Ikeda (1985), in which independent variables are animal body weight (carbon) and temperature. The daily ${\mathrm{NH}}_{\mathrm{4}}^{+}$ and ${\mathrm{PO}}_{\mathrm{4}}^{\mathrm{3}-}$ excretion values by total zooplankton equal the sum of values for all taxa. We estimated the potential contribution of zooplankton excretion to nitrogen and phosphorous requirements for phytoplankton from primary production using Redfields's ratios.

The contribution of migrating zooplankton to carbon export by respiration and excretion in deep water during the day was estimated at the long-duration stations by applying the respiration and excretion rates over 12 h to the biomass migrating at depth (difference of integrated 0–200 m zooplankton net biomass between night and day).

2.9 Statistical methods

Principal component analysis (PCA) was used to explore spatial patterns of the environmental variables data characterizing the zooplankton habitat: temperature, salinity, chlorophyll a, percentage of chlorophyll a (ratio $\mathrm{Chl}\text{-}a/\mathrm{Chl}\text{-}a+\mathrm{Phae}$) (average values of chlorophyll a and phaeophytin between 0 and 200 m depth, to be consistent with the net haul depth), and MLD. The data were normalized before the analyses were run using the Primer 6.0 software.

One-way or two-way analyses of variance were run, to explore the differences between day and night samples and between stations or zones for the environmental and zooplankton parameters. Post hoc Scheffé tests were performed to analyze paired differences. Spearman's rank correlations (R_s) were computed to test relationships between zooplankton variables and environmental parameters. Diversity was calculated for zooplankton and copepod taxa using the Shannon–Wiener diversity index.

Spatial variations in the zooplankton community composition were investigated using multivariate analysis, specifically nonmetric multidimensional scaling (NMDS). A Bray–Curtis matrix “species – stations” of square root transformed abundance data was used to estimate station similarity. The similarity matrix was then ordinated using NMDS. A SIMPER (percentage of similarity) analysis was performed to identify the species contributing most to similarity or dissimilarity between stations for the station groups identified by NMDS.

Finally, to select the environmental variables “best explaining” community patterns, we used the BEST procedure with the BIOENV algorithm, which maximizes a rank correlation between the environmental and zooplankton resemblance matrices. The environmental variables used are the same as those used in the PCA, and we also considered the abundance of Trichodesmium, derived from Stenegren et al. (2018). Analyses were run using Primer 6 for PCA and NMDS and with Statistica v.6 for ANOVA (analysis of variance), regression, and correlation.

Figure 3Zooplankton abundance and biomass along the OUTPACE west–east transect. (a) Abundance (ind. m⁻³) of small (ECD < 300 µm) and large (ECD > 300 µm) zooplankton determined by microscope counting (vertical bars) and of large zooplankton (ECD > 300 µm) determined by ZooScan (dark line). Averaged integrated Chl a concentrations (green line). (b) Cumulated zooplankton and detritus biomasses. Red line – values of total dry weight determined by weighing at each station. Black line total zooplankton biomass determined from microscopic counting. The zooplankton biomass fraction < 300 µm was determined from microscopic counting. Zooplankton biomass fractions > 300 µm (four fractions) and detritus biomass were estimated from ZooScan biovolumes. SD-01 to 15: short-duration stations. LD-A, LD-B, and LD-C: long-duration stations (average value and standard deviation over the 5-day sampling).

Download

3.1 Hydrology and trophic conditions along the transect

In the PCA of environmental data, the first two axes explained 70 % of the total variance, of which 50 % was accounted for by the first axis (Fig. 2). The first axis clearly separated the GY stations (stations LD-C, SD-14, SD-15), characterized by low Chl a but high temperature, salinity, and MLD values, from the stations of the Melanesian archipelago (MA). The second axis opposed two clusters of stations within this latter group: the first included the western stations close to Nouméa and Loyalty Islands (W-MA), and LD-B sampled in “blooming” conditions (called BL) characterized by a higher percentage of Chl a to total pigments (>67 %); the second cluster (57±0.09 % Chl a) grouped the stations referred to as central and eastern MA stations (CE-MA). Mean values of environmental data in each cloud are given in Table 1. Salinity was significantly lower in NA than in GY and BL (ANOVA; p<0.05), temperature was significantly lower in MA than in GY, and MLD was significantly deeper in GY than in W-MA and CE-MA (p<0.05). Chl a was significantly lower in GY than in the three other zones and % Chl a was significantly higher in W-MA and BL than in GY and CE-MA.

Table 2Mean values (± standard deviation) of zooplankton abundances from ZooScan and microscopic counts, percentage of taxonomic groups, and total copepod demographic parameters at the stations for the four clusters defined in the PCA on environmental variables (see Fig. 2) and for the three long-duration stations. W-MA: western Melanesian archipelago; CE-MA: central and eastern Melanesian archipelago; BL: station B (blooming conditions); GY: subtropical gyre. Letters below the mean values indicate homogeneous groups between zones (small letters) or LD stations (capital letters) according to post hoc Scheffé tests.

Download Print Version | Download XLSX

3.2 Spatial variations in zooplankton, abundance biomass, and size structure

The total zooplankton abundance estimated from microscope counting (Fig. 3a) and the total zooplankton biomass estimated from the cumulated biovolumes of organisms counted with the microscope (total or fraction < 300 µm) and ZooScan (fraction > 300 µm) (Fig. 3b) showed a general decreasing trend from west (SD-1) to east (SD-15), with local increases sometimes linked with Chl a increase (Tables 2 and 3). Detritus biomass (estimated with ZooScan) was also particularly high (40 %–50 %) in the Coral Sea region (SD-1 to SD-5 and LD-A), compared to other regions (17 % to 44 %) (Fig. 3b and Table 3). With the exception of stations SD-2, SD-3, and SD-9, total dry weights estimated from the biovolumes of counted organisms and particles (from binocular for ECD < 300 µm and from ZooScan for ECD > 300 µm) showed a good correspondence to measured total dry weight (R_s=0.721, p=0.001). In addition, total dry weights estimated from the ZooScan were well correlated with those estimated from binocular counting for the same size fraction (ECD > 300 µm): R_s=0.657, p=0.02. The total zooplankton abundance varied from 409 to 2017 ind. m⁻³ (Fig. 3a and Table 2). The highest values, but high variability as well, were observed in the New Caledonia region (SD1 to 4 and LD-A). There was a clear drop in abundance at GY stations (LD-C, SD-14, and SD-15) compared to all the other zones (W-MA, CE-MA, and BL; p<0.05). Microscope abundance showed relatively good agreement with ZOOSCAN abundance for the size fraction > 300 µm ECD (R_s=0.627, p=0.007). This fraction represented 49 % to 63 % of the total microscope counted zooplankton abundance (Fig. 3a and Table 2), whereas it represented 88 % to 98 % in terms of zooplankton biomass and was equally distributed in the different size classes, although with stronger variations for the >200 µm size class. The ratio of abundance of zooplankton size fractions above and below 300 µm ECD did not show any spatial trend.

Table 3Mean values (± standard deviation) among stations of each cluster defined in the PCA on environmental variables (see Fig. 2) of zooplankton biomass (top part of table, “Biomass (mg DW m⁻³)”) and percentage of total biomass for the different size fractions (bottom part of table, “Zooplankton biomass (%)”) and for the three long-duration stations. Zooplankton biomass estimated from weighing and from biovolume measurements from microscope and ZooScan observations. W-MA: western Melanesian archipelago; CE-MA: central and eastern Melanesian archipelago; BL: station B (blooming conditions); GY: subtropical gyre. Letters below the mean values indicate homogeneous groups between zones (small letters) or LD stations (capital letters) according to post hoc Scheffé tests.

Download Print Version | Download XLSX

Table 4Mean values (± standard deviation) per region of taxonomic diversity (H': Shannon index) and taxonomic richness (no. of taxa per sample) calculated for total zooplankton and copepod communities. W-MA: western Melanesian archipelago; CE-MA: central and eastern Melanesian archipelago; BL: station B, blooming conditions; GY: subtropical gyre. Letters below the mean values indicate homogeneous groups between zones (small letters) or LD stations (capital letters) according to post hoc Scheffé tests.

Download Print Version | Download XLSX

Zooplankton abundance was negatively correlated with water column temperature ($R_s=-\mathrm{0.511}$, p=0.028) and MLD ($R_s=-\mathrm{0.790}$, p=0.000). It was positively correlated with Chl a (R_s=0.498, p=0.042) when considering all of the transect stations, but the correlation was negative for stations in the New Caledonia region (SD1 to 4 and LD-A; $R_s=-\mathrm{0.900}$; p=0.037) and highly positive for other stations (SD-5 to SD-15, LD-B, and LD-C; R_S=0.804; p=0.002). Dry weight (weighed) as well as ZooScan zooplankton biomass (Fig. 3b and Tables 1 and 3) were both positively correlated with Chl a (R_S=0.588, p=0.013 and R_S=0.68, p=0.002, respectively). As for abundance, better correlations were found when considering stations SD-5 to SD-15 and LD-C (R_S=0.783, p=0.002 for both variables), whereas negative correlation was found with ZooScan zooplankton biomass for stations in the Coral Sea ($R_s=-\mathrm{0.900}$, p=0.035). Interestingly, detritus biomass was also well correlated with Chl a when considering the whole transect data (R_s=0.721, p=0.001) and data from stations outside the Coral Sea (R_s=0.755, p=0.004).

3.3 Taxonomic diversity in the different oceanic regions along the transect

From the 120 µm mesh size bongo net, 66 zooplankton taxa were identified (see Table S1), with 41 genera/species of copepods plus miscellaneous nauplii and copepodites). The total number of zooplankton taxa per sample varied from 25 to 40, and the Shannon index varied between 3.3 and 3.76 bit ind.⁻¹ (Table 4). These two variables displayed their minimum mean values in the GY zone. Copepods were the most abundant group (68 % to 86 % of total abundance), with a slight increase in their contribution from west to east (see Table 2), with a corresponding decrease in other contributors (gelatinous plankton and other holoplankton). Thus, copepod dominance was more prominent in the GY zone (79 % to 86 %) than at the other sites (<80 %). Among copepods, early life stages were dominant (69 %–88 % of copepod abundance) and included mostly copepodites (42 %–82 %). In the GY zone, the proportion of adults (mean = 15±2 %) was lower than in the three other zones (mean > 18 %), whereas the percentage of adult females was the lowest in W-MA. Clausocalanus/Paracalanus (25 % of copepod abundance), Oithona (19 %), Oncaea (18 %), Corycaeus (7.6 %), and Microsetella (4.6 %) were the most abundant copepod genera and were present at all stations sampled. All of these copepod taxa were listed in the top 10 species with respect to frequency of abundance for the four regions (Table 5), along with appendicularians, Thecosomata, and chaetognaths (except GY). Gelatinous zooplankton represented 8.3 % to 24.3 % of zooplankton abundance (see Table 2), with the lowest contributions at stations LD-C and SD-15 in the GY zone. They were dominated by appendicularians (8 %–17 %) and chaetognaths (0.8 %–3.3 %), whereas siphonophores, doliolids, salps, and hydrozoans represented <0.5 % of the total zooplankton abundance. Chaetognaths were rare in the GY zone (<1 %) and at SD-1 (0.2 %). Other holoplanktonic taxa (2.3 %–12.7 %) included Thecosomata (1.2 %–10.2 %), ostracods (1 %–4 %), and euphausiids (<1 %). Meroplankton was mostly polychaete larvae (0.2 %–0.5 %) and lamellibranch larvae (0.1 %–0.4 %), present in the four zones.

Table 5Top 10 taxa in frequency abundance for the four regions (W-MA: western Melanesian archipelago; CE-MA: central and eastern Melanesian archipelago; BL: station B, blooming conditions; GY: subtropical gyre).

Download Print Version | Download XLSX

The NDMS ordinations based on the relative abundance of the zooplankton taxa discriminated GY stations from the other stations (Fig. 4a; dissimilarity 20 %), mainly due to the contributions of Corycaeus (7 %) and Clausocalanus/Paracalanus (6.3 %) which were positively correlated to GY, and appendicularians (5.6 %), and chaetognaths (5 %), which were correlated to the other stations (Fig. 4b). However, the analysis did not discriminate between groups among W-MA, BL, and CE-MA stations, despite these groups being distinguishable on the basis of environmental data.

Figure 4NDMS of the main zooplankton taxa (>0.1 % abundance). (a) Plot of the stations with different colors between the regions identified with the environmental clustering (four regions: W-MA – western Melanesian archipelago; CE-MA – central and eastern Melanesian archipelago; BL – blooming conditions, station LD-B; GY – subtropical gyre). (b) Plot of the taxa.

Download

Figure 5Spatial variation in (a) log₁0 transformed mean nifH abundance values for three groups of diazotrophs across the transect: Trichodesmium, HET-1 , and UCYN-A and UCYN-B (adapted from Stenegren et al., 2018, their Fig. 2b). Abundance of (b) main copepod taxa and (c) main other zooplankton taxa.

Download

3.4 Relationships between zooplankton taxa and diazotrophic microorganisms

According to the BEST procedure, the environmental variables best explaining the zooplankton community pattern were Trichodesmium abundance, MLD, and Chl a (r=0.593, p=0.05), whereas temperature, salinity, and unicellular (UCYN) or heterocystous (het-1) cyanobacteria were not selected. The abundance of major zooplankton taxa along the transect showed a strong positive link with the abundance of diazotrophic microorganisms (Fig. 5). Positive correlations were found between heterocystous cyanobacteria HET-1 and Microsetella (R_s=0.52; p=0.032), Clauso-/Paracalanus (R_s=0.61; p=0.009), Oithona (R_s=0.66; p=0.004), and Appendicularia (R_s=0.53; p=0.030) and between UCYN-B and Ostracoda (R_s=0.61; p=0.009. Only Macrosetella gracilis (R_s=0.684, p=0.002) and Oncaea (R_s=0.484, p=0.049) showed a significant relationship with Trichodesmium. Among non-copepod taxa, only Thecosomata, showed a positive correlation with Trichodesmium (R_s=0.631, p=0.007) and displayed significant lower abundance in GY compared to W-MA, CE-MA, and BL.

3.5 Temporal dynamics of zooplankton at the three long-duration stations and comparison with sediment trap content

At LD-A and LD-B, the zooplankton biomass observed each day (Fig. 6a) showed a dome-shaped pattern, with an increase over the 3 first days followed by a decrease. At both stations, successive day–night samples showed a biomass increase during the night, mainly due to the size fraction >2000 µm (euphausids, large copepods, etc.). At station LD-C, the zooplankton biomass was rather stable over the 6 days, without day–night variations. At LD-A and LD-B, the proportion of detritus found in the sample was high and appeared to increase at LD-A. At LD-A, we observed a much higher abundance at day 5 that at day 1, which did not follow the biomass pattern, whereas at LD-B and LD-C, the abundance was rather stable (Fig. 6b). Interestingly, the taxonomic distribution at the three stations (Fig. 7a) showed a stable structure for LD-B and LD-C but a relative increase in small forms (nauplii, small copepods) parallel to a Chl a increase for LD-A. At LD-B, abundance and biomass of zooplankton did not respond to the strong decrease in the bloom within the last 2 days. At LD-C the stability of both abundance and biomass was parallel to the stability of Chl a.

In the sediment traps situated at 150 m, there was a greater relative contribution of copepods at LD-C compared to LD-A and LD-B, as observed in the water column (Fig. 7b). In contrast, appendicularians were a major contributor of the swimmers found in the LD-C trap compared to their frequency in the water column, and by comparison with their respective frequencies at LD-A and LD-B. At LD-B, there was a sharp decrease in swimmers over time in the traps mainly due to copepods but a relative increase in Ostracod. Pteropods had high relative contribution in the traps (20 %–30 % at LD-A, around 10 % at LD-B and LD-C), whereas their relative abundance in the water column was low (1 %–4 %).

Table 6Mean values (± standard deviation) of plankton stocks, zooplankton weight-specific ingestion, respiration and excretion rates, zooplankton grazing, excretion and respiration fluxes, and zooplankton vertical fluxes. W-MA: western Melanesian archipelago; CE-MA: central and eastern Melanesian archipelago; BL: station B, blooming conditions; GY: subtropical gyre. Letters below the mean values indicate homogeneous groups between zones (small letters) or LD stations (capital letters) according to post hoc Scheffé tests. Zooplankton stocks are estimated from the cumulated biovolume of binocular-counted organisms. Swimmer biomass in sediment trap from Caffin et al. (2018a).

Download XLSX

Figure 6Temporal variation in zooplankton abundance and biomass over 5 days at each of the three long-duration stations (LD-A, LD-B, and LD-C from left to right). (a) Abundance (ind. m⁻³) of small (ECD < 300 µm) and large (ECD > 300 µm) zooplankton determined by microscope counting (vertical bars – only for days 1 and 5) and of large zooplankton (ECD > 300 µm) determined by ZooScan (dark line). Chl a concentrations (green line). (b) Cumulated zooplankton biomass and detritus sampled. Red line – values of total dry weight determined by weighing at each station. The zooplankton biomass fraction < 300 µm was determined from microscopic counting. Zooplankton biomass fractions > 300 µm (four fractions) and detritus biomass were estimated from ZooScan biovolumes.

Download

Figure 7Comparison of zooplankton abundance (orange triangles) and percentage of main taxa (bars) of zooplanktonic organisms (a) in the water column (0–200 m) and (b) in the sediment trap (“swimmers”) at 150 m depth at each of the three long-duration stations (LD-A, LD-B, and LD-C) from left to right. Sediment trap data at day 5 cannot be considered for the analysis (see Caffin et al., 2018a).

Download

3.6 Estimation of fluxes related to meso-zooplankton

Biomass-weighted zooplankton δ¹⁵N values were lower in the regions W-MA and CE-MA, averaging 2.7 ‰ and 2 ‰, respectively, than in the GY, where zooplankton δ¹⁵N values averaged 8.5 ‰ (Fig. 8a). The δ¹⁵N values of the zooplankton corresponded with those of the POM, being lower west of the GY and increasing in the GY. We estimated that DDN contributed an average of 67 % and 75 % to zooplankton biomass in the W-MA and CE-MA regions, respectively (Fig. 8b). In the GY, the diazotroph contribution to zooplankton biomass decreased to an average of 22 % and showed a declining trend from west to east, with the lowest value of 7 % occurring at SD-15. The integrated phytoplankton standing stock derived from the water-column-integrated content of total chlorophyll a within the euphotic layer (Table 6) was highest at the LD-B and the lowest in the GY, although the stock generally decreased from west to east along the transect, as seen in Fig. 3a with the Chl a distribution pattern. Interestingly, regions with the lowest phytoplankton stocks (GY and CE-MA) presented the highest POC∕phytoplankton biomass ratio (3.73 and 3.67, respectively), whereas this ratio decreased to 2.46 in BL and to 2.60 in W-MA. The average zooplankton weight-specific rates of ingestion, ${\mathrm{NH}}_{\mathrm{4}}^{+}$ and ${\mathrm{PO}}_{\mathrm{4}}^{\mathrm{3}-}$ excretion, and respiration (Table 6) determined from allometric relationships for all zooplankton taxa (see “Material and methods”) were found to be rather stable over the different regions – although the test identified different ingestion in W-MA and GY, which reflected narrow variations in temperature and optimum available food supply (considering the contribution of phytoplankton and POC for the whole zooplankton community). The ingestion by herbivorous/omnivorous zooplankton (ZCD_H) represented between 19 % and 183 % of the estimated primary production over all stations, but this percentage was very heterogeneous in CE-MA (with an average of 72.6 %), more stable in other regions, and fell to 34.8 % in W-MA including LD-A. The grazing impact on the phytoplankton stock increased from east to west but was less in GY (9.4 %) than BL (17.6 %). The impact on picoplankton was low (<0.25 % stock d⁻¹) at all locations, and the grazing was distributed between nano- and microplankton in comparable proportions, with values particularly high in BL (mean was 73.45 % and 101.54 % of the stock per day for nano- and microplankton, respectively).

Figure 8(a) Biomass-weighted zooplankton and POM (5 m depth) nitrogen isotope ratios (δ¹⁵N). (b) Average percent contribution of diazotroph-derived nitrogen (DDN) to zooplankton biomass (ZDDN). W-MA: western Melanesian archipelago; CE-MA: central and eastern Melanesian archipelago; GY: subtropical gyre.

Download

Weight-specific excretion rates varied between 0.11 and 0.14 d⁻¹ for ${\mathrm{NH}}_{\mathrm{4}}^{+}$ and between 0.09 and 0.11 d⁻¹ for ${\mathrm{PO}}_{\mathrm{4}}^{\mathrm{3}-}$. Daily regeneration by zooplankton represented between 29.7 % and 77.2 % of phytoplankton needs for N and between 5.9 % and 165.6 % for P. The lowest and highest impacts of zooplankton on phytoplankton in terms of grazing and regeneration were found in the W-MA and CE-MA, respectively. Depth-integrated zooplankton respiration varied between 50.6 and 248.8 mgC m⁻² d⁻¹ and was significantly lower in GY than in W-MA, CE-MA, and BL (Table 6). The percentage of estimated zooplankton respiration rates to primary production was lowest in the W-MA region (7 % to 25 %), compared to the rates in the CE-MA (12 % to 112 %), BL (30 %), and GY regions (26 % to 52 %).

The biomass of migratory zooplankton to deep water during the 12 h daylight period was estimated from the difference of night and day biomass at the three long-duration stations (LD-A, LD-B, and LD-C), along with associated fluxes. The strongest impact of diel migration was observed at LD-A where half of the zooplankton biomass migrated, injecting 20 % of the surface zooplankton carbon biomass through respiration below 200 m. The biomass of migratory zooplankton and respiration below 200 m was reduced to half at LD-B, whereas no migration could be estimated at LD-C from our net tows. However, in terms of the percentage of primary production, the carbon released by zooplankton respiration below 200 m was comparable at the two stations LD-A and LD-B (3 % and 3.75 %, respectively). The daily biomass of zooplankton trapped in the sediment traps situated at 150 m at the LD stations was around 50 mgC m⁻² d⁻¹ , with no significant difference between stations (Table 6). At LD-A and LD-B, it represented, respectively, 12.7 % and 30.1 % of the migrating biomass.