The study is part of the CSIC-SUMMER cruise carried out on board the RV “Sarmiento de Gamboa” around the Iberian Peninsula from September 28^th to October 25^th, 2020 (Olivar et al., 2022). Five zones were sampled repeatedly throughout the day and night: south of the Balearic Islands (Z1; 38.5°N, 2.5°E), western Alborán Sea (Z2; 36°N, 4°W), Gulf of Cádiz (Z3; 36°N, 8°W), off Lisbon (Z4; 38.1°N, 9.4°W), and off Galicia (Z5; 42°N, 9.5°W). Briefly, Z1 is an oligotrophic zone, Z2 and Z3 are transition zones between the Atlantic and Mediterranean waters, but Z2 is more productive than Z1 and Z3. Z4 and Z5 are more productive zones due to the influence of the upwelling (for detailed area descriptions see Supplementary Material SM1). We spent between 48 and 60 hours at each zone conducting repeated stations sampling zooplankton and micronekton for a minimum of two consecutive days during day- and nighttime. However, Z5 was limited to just one day and one-night stations due to rough sea conditions (Supplementary Table S1).

The vertical profiles of temperature, conductivity, and fluorescence were recorded from the surface to 1000 m depth using a SeaBird SBE 911plus CTD equipped with a Seabird-43 Dissolved Oxygen sensor and a Seapoint Fluorometer mounted on a rosette sampler equipped with 24 Niskin bottles of 12 L each. Fluorescence obtained in vertical profiles in the upper 200 m depth were converted to chlorophyll a (Chl a) according to Yentsch and Menzel (1963). The vertical profiles of temperature, salinity, and fluorescence were averaged every 1 dBar. Monthly average values (October 2020) of sea surface temperature (SST) were downloaded from the NASA’s OceanColorWeb site with a spatial resolution of 4x4 km and processed using the proto-algorithm from MODIS Ocean Team Computing Facility (MOTCF) based on satellite infrared retrievals of ocean temperature. Net primary production (NPP) was obtained from remote sensing data following Behrenfeld and Falkowski (1997) through the Ocean Productivity website (https://www.science.oregonstate.edu/ocean.productivity/index.php) for the specific dates of the cruise and using the Vertical Generalized Production Model (VGPM) as the standard algorithm.

Sediment trap-derived passive flux to estimate gravitational export was measured at 150 m depth using a free-drifting multi-trap array with eight cylinders, as the model described by Knauer et al. (1979) and using the procedure described in Hernández-León et al. (2019a). The traps were deployed for approximately 24 h, each equipped with cylinders containing filtered seawater enriched with NaCl (∼45 g·L^-1 analytical grade) to increase density. No poisons were added to retard bacterial decomposition. After recovering, samples were filtered onto pre-combusted (450 °C for 12 h) 25 mm Whatman GF/F filters. Then, they were frozen at -20 °C until analysis in a Carlo Erba CHNSO 1108 elemental analyzer (UNESCO, 1994).

Zooplankton samples were obtained using a MOCNESS-1 net with a 1 m² mouth opening area fitted with 200 μm mesh size (Wiebe et al., 1976). Oblique hauls were made from 700 m depth to the surface in eight strata: 700-600, 600-500, 500-400, 400-300, 300-200, 200-100, 100-50, 50–0 m, at about 1.5-2.5 knots (Supplementary Table S1). Filtered volume was measured using an electronic flowmeter. After sampling, representative organisms - based on visual inspection of the most abundant species (mainly copepods, euphausiids, and chaetognaths) - were gently picked for enzymatic measurements, frozen in liquid nitrogen (-196 °C) and preserved at -80 °C. The rest of the sample was preserved in 4% buffered formalin and seawater. In the laboratory, a subsample was selected for taxonomic analysis. Samples were digitized using an EPSON scan version 4990 at 2400 dpi, processed in ZooProcess (Gorsky et al., 2010; Vandromme et al., 2012), and uploaded to EcoTaxa (Picheral et al., 2017) to AI-supported manual classification. The body area (in pixel)-to-dry weight (DW) conversion was done using the equations provided by Lehette and Hernández-León (2009), assuming an uncertainty based on the standard error of the regression slope for general mesozooplankton (slope = 1.54 ± 0.03) and to carbon weight (CW) using a conversion factor of 0.40 ± 0.08 based on a literature review (Banse, 1996; Dam and Peterson, 1993; Andersen and Hessen 1991; Andersen et al., 2016. See SM4. Sensitivity analysis). Biomass estimates need to be corrected for potential net avoidance, which is commonly referred to as ‘capture efficiency’ (CE). For zooplankton, we assumed no net avoidance (Skjoldal et al., 2013). A sensitivity analysis for the conversion factors was carried out and is described below.

Details of the overall micronekton sampling operations during the cruise have been published by Olivar et al. (2022). Briefly, day and night samples were obtained using a Mesopelagos midwater trawl (Meillat, 2012) with a total length of 58 m and a graded mesh netting of 30 mm near the mouth and 4 mm in its lower part. At the end of the net, a VERDA multi-sampler (Castellón and Olivar, 2023) was fitted to enable stratified sampling (Supplementary Table S1). The ship course was kept constant during the hauls, and ship speed was maintained at ca. 2 knots using the Speed Over Ground system provided by the onboard GPS. The volume of water filtered was calculated as the product of mouth opening of the net and the distance travelled by the vessel. Mean net opening area was 30 m². The distance travelled during each haul (d), was calculated by spherical trigonometry applying the Haversine formula, which takes into account the earth radius (R) and the differences of the latitude (Δφ=φ₁−φ₀) and longitude (Δλ=λ₁−λ₀) at the beginning (0) and end (1) of the hauls.

On board, fish and decapods were sorted and identified. Wet weight was measured using a marine precision balance POLS S-182 P-15 (precision 2 g). Selected species of mesopelagic fishes and decapods (based on a visual assessment of the most abundant types) were frozen in liquid nitrogen and stored at -80°C for later metabolic analysis. Biomass was estimated converting wet weight (WW) to DW using conversion factors of 0.18 ± 0.01 for decapods (Pakhomov et al., 2019) and 0.23 ± 0.04 for fishes (López-Pérez et al., 2020)s, and then to CW using the above-mentioned ratio. All micronekton CW estimates need to be corrected for potential net avoidance (CE). For the net we used for micronekton (Mesopelagos), there are, to our knowledge, no published data on CE. Hence, we assumed a CE of 20 ± 13% based on a literature review (Gjøsaeter, 1984; May and Blaber, 1989; Koslow et al., 1997; Davison, 2011; Pakhomov et al., 2019) of other midwater trawls. The sensitivity analysis of these conversions, using a range of conversion factors, is described in Supplementary Material (SM4. Sensitivity analysis.).

To compare the relative contribution of zooplankton and micronekton to the carbon flux, active fluxes (AF) were estimated as the sum of the respiratory (RF), mortality (MF), gut flux (GF), and excretion (EF) (Equation 1).

Briefly, RF was estimated using electron transfer system (ETS) activity, measured following the method of Packard (1971) modified by Owens and King (1975); Kenner and Ahmed (1975), and Gómez et al. (1996). Respiration in carbon units was estimated applying a R/ETS ratio of 0.5 ± 0.1 (Hernández-León and Gómez, 1996; Hernández-León et al., 2019c; Couret et al., 2024) and a respiratory quotient of 0.97 ± 0.4 (Omori and Ikeda, 1984; Hernández-León and Ikeda, 2005). MF was estimated from growth assuming steady-state conditions (growth = mortality) in the mesopelagic zone, using the equation of Ikeda and Motoda (1978) relating respiration and growth and applying a conversion factor of 0.75 ± 0.19 for zooplankton (Omori and Ikeda, 1984) and 0.66 ± 0.17 for micronekton (Brett and Groves, 1979). EF was estimated using the values given by Steinberg et al. (2000), assuming that the excretion of dissolved organic carbon makes up 24 ± 26% (range = 5-42%) of the total carbon metabolized. GF was estimated from respiration by assuming that zooplankton feeding is 2.5 ± 0.5 times respiration (Ikeda and Motoda, 1978), and migrant zooplankton egested 50% of the gut content at depth (Ariza et al., 2015). For micronektonic migrants, we assumed they egest an amount equivalent to the 40% of the respired carbon (Brett and Groves, 1979) and that they transport feces to the mesopelagic because of their density and the long gut passage time of large animals. Assuming that micronektonic migrants egest after the downward migration, the egestion should be double in relation to respiration during 24 h. Therefore, we used an egestion equivalent to 80 ± 16% of the respired carbon (Ariza et al., 2015). Detailed information is given in Supplementary Material (SM2, 3).

To account for the uncertainties and potential error propagation, we carried out a sensitive analysis for the biomass and metabolic rate estimates of both zooplankton and micronekton (mesopelagic fish and decapods) using the Monte Carlo method with a range of conversion factors (CF) (Supplementary Table S2). Briefly, we incorporated uncertainties in measurements and conversion factors at each calculation step and conducted 100,000 simulations. The mean and standard deviation of these simulations were then calculated and presented as the final estimates. Detailed information is given in Supplementary Material (SM4).

To identify environmental patterns that control fluxes and assess the correlation between biomass, migrant biomass and active flux of zooplankton, decapods, and mesopelagic fish with environmental parameters, we first calculated the average values of temperature, salinity, chlorophyll a concentration, and oxygen concentration for the regions in the approximate mixed layer (0–50 m), epipelagic (0–200 m) and mesopelagic (200–700 m) layers. We also included satellite-derived NPP and POC flux measured by the sediment traps. The correlation between all parameters was calculated using R’s co function and visualized using the ComplexHeatmap package (Gu, 2022). We checked whether correlations are significant (p< 0.05) using simple linear regression. Note that a discrepancy between a strong correlation and the lack of significance in the linear regression (or vice versa) could occur due to the small sample size (n = 5) and, in some cases, narrow range of parameter values. All regressions were visually checked, but care should be taken to not overinterpret individual results. All analyses were performed in the programming language R (R Core Team, 2022). The sampling map was generated using the geographic information system QGIS (V.3.22.3) (QGIS Development Team, 2021).

The vertical temperature profiles (Figure 1A) showed lower average values in the northern Atlantic, both in the epipelagic layer (Z4 = 13.7°C and Z5 = 14.3°C) and mesopelagic layer (Z4 = 11.9°C and Z5 = 10.9°C), compared to the Mediterranean stations (Z1 = 17.1°C; Z2 = 16.9°C and Z1 = 13.5°C; Z2 = 13.3°C, respectively). The Gulf of Cádiz showed epipelagic values similar to the Mediterranean ones (17.2°C), but lower mean temperature at depth (11.5°C). Salinity profiles (Figure 1B) showed high mean values in the epipelagic at Z1 and Z2 (37.8 and 36.9, respectively), compared to the Atlantic stations, increasing at depth due to the high salinity of the Mediterranean Sea (38.5 for both). Salinity at Z3 decreased with depth, with a mean value of 36.3 in the epipelagic zone and 35.7 in the mesopelagic zone, likely influenced by the presence of less saline Atlantic waters. Z4 and Z5 exhibited a consistent salinity pattern in the upper 400 m, with mean values of approximately 35.6-35.7. Beyond this depth, salinity increased slightly, reaching 36.3 at Z4 and 36.0 at Z5 by 800 m. Surface oxygen levels were similar across all stations, with values of 4.7, 5.0, 4.9, 5.1, and 5.2 ml·L^-1 at Z1 through Z5, respectively (Figure 1C). Higher oxygen concentrations were observed at depth at Z4 and Z5, coinciding with the layers of lowest temperature and salinity. At Z1 and Z3, an oxygen peak occurred at 50–75 m depth, corresponding to the onset of the thermocline and a decline in temperature.

Epipelagic Chl a values (Figure 1D) were generally low, displaying maximum values lower than 0.4 mg·m^-3 at Z1, from 0.4-0.9 mg·m^-3 at Z2,<0.3 mg·m^-3 at Z3, and 0.7 mg·m^-3 at Z5, with the exception of Z4 in the upwelling zone off Portugal, where surface Chl a concentrations reached values close to 3 mg·m^-3. Higher net primary production values occurred at the Atlantic stations (1006 ± 449 mg C·m^-2·d^-1) compared to the Mediterranean stations (451 ± 141 mg C·m^-2·d^-1), showing the highest values in Z4 (Table 1).

Due to strong currents at Z2 and rough sea conditions at Z5, it was not possible to measure trap-derived POC flux at these stations. Across the remaining stations, POC flux values were relatively consistent, ranging from 4.2 ± 0.2 mg C·m^-2·d^-1 at Z3 to 7.9 ± 3.9 mg C·m^-2·d^-1 at Z1 (Table 1). No significant relationship (assessed by linear regression) was found between POC flux, net primary production, or active flux values (Supplementary Table S3).

Vertical profiles of zooplankton biomass showed the expected patterns of higher biomass in the upper 100 m layer and decreasing with depth (Figure 2A), except at Z2 where a biomass peak was observed in the mesopelagic layer between 400 and 500 m depth. At Z1-3, mean zooplankton biomass in the epipelagic layer was higher during nighttime, whereas at the northern Atlantic stations (Z4-5) daytime mean biomass was greater. Micronekton biomass vertical profiles revealed higher values in the upper 300 m during nighttime compared to deeper layers, where greater variability in biomass distribution between day and night was found (Figure 2B). This pattern was consistent for both mesopelagic fish (Figure 2C) and decapods (Figure 2D).

Zooplankton specific ETS activity profile represents the average specific ETS activity of the zooplankton community (Figure 3A). Except for Z3, zooplankton specific ETS activities were higher in the upper 200 m depth during nighttime, gradually decreasing with depth. Copepods (Figure 3B) and euphausiids (Figure 3C) exhibited higher specific ETS activities, with more pronounced day-night and spatial variability compared to chaetognaths (Figure 3D). Micronekton (mesopelagic fish and decapods average specific activity) ETS values were not measured at all depths across the stations (Figure 3E). Generally, micronekton specific ETS activity stations (Figures 3E–G) showed higher values at depth during the daytime, except at Z3.

Zooplankton respiratory flux was almost 6-fold higher in the Mediterranean (26.9 ± 20.2 mg C·m^-2·d^-1) than in the Atlantic (4.6 ± 4.5 mg C·m^-2·d^-1) (Supplementary Table S4, Figure 4). In contrast, total micronekton (mesopelagic fish and decapods) respiratory flux was lower in the Mediterranean by a factor of 2 compared to the Atlantic (5.7 ± 2.5 and 10.8 ± 5.3 mg C·m^-2·d^-1, respectively). This difference was driven by changes in decapods migration: Like for zooplankton, mesopelagic fish respiratory fluxes were higher at Mediterranean stations (2.9 ± 2.8 mg C·m^-2·d^-1) compared to the Atlantic stations (1.1 ± 0.8 mg C·m^-2·d^-1), while decapods respiratory fluxes were higher by a factor of 3 at Atlantic stations (9.7 ± 6.6 mg C·m^-2·d^-1) compared to the Mediterranean stations (2.8 ± 1.7 mg C·m^-2·d^-1) (Figure 4). Mortality, gut and excretion fluxes followed the same patterns as these are estimated based on respiration fluxes.

Considerable variability in zooplankton migrant biomass was observed across zones, with higher values at the Mediterranean stations (497 and 953 mg C·m^-2 at Z1 and Z2, respectively), accounting for 70% of the total migrant biomass, compared to Z3 (173 mg C·m^-2), Z4 (174 mg C·m^-2), and Z5 (29 mg C·m^-2) (Table 1, Figure 5). In contrast, micronekton migrant biomass was higher in the Atlantic (Z3-5) compared to the Mediterranean (Z1-2) where migrant biomass exceeded 50% (Table 1, Figure 5). Fish migrant biomass ranged from 6 at Z5 to 108 mg C·m^-2 at Z1, while decapods biomass exhibited larger variability (ranging from 56 at Z1 to 450 mg C·m^-2 at Z4) with an average biomass value 3-fold higher than mesopelagic fish migrants. At Z1, fish accounted for a larger proportion of migrant biomass than decapods, but the proportion of decapods increased progressively across the stations, reaching over 80% of the total migrant biomass at Z5 (Figure 5).

Total active flux (zooplankton and micronekton) displayed the highest value at station Z2 (133.8 mg C·m^-2·d^-1) and the lowest at station Z5 (24.6 mg C·m^-2·d^-1), while the remaining stations exhibited similar values: Z1 = 51.7, Z3 = 50.2, and Z4 = 58.4 mg C·m^-2·d^-1 (Table 1, Figure 5). In the Mediterranean (Z1-2), micronekton active flux contributed to the total flux by 13% at Z1 and 18% at Z2 to the carbon flux, while zooplankton active flux was the dominant contributor (Figure 5). In contrast, at Z4 and Z5, zooplankton active flux accounted for only 7% and 19% of the total flux, respectively, playing a relatively minor role. Instead, micronekton activity drove the majority of the flux in these zones due to decapods flux. Notably, POC flux was substantially lower, contributing just 13%, 8%, and 9% of the total carbon flux at Z1, Z3, and Z4, respectively.

Zooplankton migrant biomass was significantly correlated with salinity (positive) and oxygen (negative) in the mesopelagic layer (Figure 6). Similarly, zooplankton active flux was negatively correlated with oxygen in the mesopelagic. Fish active flux exhibited significant positive correlations with temperature in the mesopelagic layer and salinity in the epipelagic layer but was negatively correlated with oxygen in the first 50 m depth. Decapods biomass was significantly positively correlated with NPP and Chl a in the upper 50 m depth, and migrant biomass negatively correlated with salinity and temperature in the upper 50 m depth. Additionally, decapods active flux was significantly positively correlated to NPP, Chl a in the upper 50 m depth, and Chl a in the epipelagic layer. Strikingly, the matrix correlation revealed that decapods exhibited an opposite pattern compared to zooplankton and mesopelagic fish, showing positive correlation with productivity (both NPP and Chl a) but a negative correlation to temperature and salinity. Finally, POC flux exhibited a negative correlation with oxygen in the mesopelagic layer.

We only found a significant relationship between decapods and NPP (R² = 0.845, p<0.05, n=5), but no significant relationship between zooplankton nor fish and NPP (not shown). Fish migrant biomass was the only variable that exhibited a significant positive relationship with biomass, while no significant relationship was observed between biomass and active flux in any of the other groups. Noteworthy here is that for both fish and decapods, total biomass and migrant biomass was positively correlated, whereas for zooplankton there appeared to be a negative correlation. Finally, in all three communities, migrant biomass was -as expected due to autocorrelation- significant positive correlation with active flux (Figure 7).

The relative contribution of each group (zooplankton, decapods, and mesopelagic fish) to the carbon flux shifted from the Mediterranean Sea to the Atlantic Ocean. While zooplankton dominated the carbon flux in the Mediterranean Sea, micronekton -specifically decapods- played a particularly prominent role in the North Atlantic Ocean (Figure 4).

While mesopelagic fish have recently become an intense subject of study (see Aksnes et al., 2023), our data highlights that the role of decapods in the active flux is substantial and may exceed active flux by both zooplankton and fishes. Despite the high abundance of decapods in many parts of the world’s oceans (Flock and Hopkins, 1992), their active flux has received little attention (Angel and Pugh, 2000; Schukat et al., 2013; Pakhomov et al., 2019). Estimates of decapod active flux vary widely between studies with no clear agreement on the values. In the North Atlantic Ocean, previous estimates of decapod active flux varied by an order of magnitude [2 versus 12.1 mg C·m^-2·d^-1; Angel and Pugh (2000); Schukat et al. (2013)], while our estimates were up to 20 times higher (ranging from 15.1 to 42.1 mg C·m^-2·d^-1) compared to Angel and Pugh (2000). On the other hand, in the central North Pacific Subtropical Gyre, decapod active flux estimates were up to two orders of magnitude lower than those in our study (0.4 to 0.6 mg C·m^-2·d^-1; Pakhomov et al., 2019). The large discrepancies between studies and regions, in addition to the limited availability of data, emphasize the urgent need to increase research efforts to better understand the contribution of decapods to the carbon flux.

We observed a strong, although not statistically significant, positive correlation between decapod active flux and total biomass. While total biomass itself does not directly determine active flux, our results suggest that higher total biomass is associated with higher migrant biomass (Díaz-Pérez et al., 2024), which is directly used in flux calculations. Therefore, total decapod biomass can serve as a rough indicator of potential active flux. Enhanced decapod biomass has been associated with oxygen minimum zones (Vereshchaka et al., 2016; Hernández-León et al., 2019a), lower mesopelagic fish biomass (Ariza et al., 2015), and high productivity (Schukat et al., 2013; Hernández-León et al., 2019a). The latter scenario may explain our findings, as decapod biomass (and active flux) were significantly correlated with NPP and Chl a (Figure 5). Similarly, in the Costa Rica Dome (Stukel et al., 2018) and in the open-ocean upwelling of the Guinea Dome (Hernández-León et al., 2019a), active transport by these pelagic micronektonic fauna was the dominant vertical transport mechanism in areas of high productivity. Although no direct explanation has been provided, it seems that the continuous fueling of primary production, the role of protists as an important intermediate trophic level in these upwelling systems (Ward and Follows, 2016; Armengol et al., 2019), or both, in these areas of persistent productivity influence the contribution of the decapods to the biological pump. Despite the limited research on how environmental variables influence decapods distribution, high primary production appears to be a common factor driving increased decapods biomass.

In terms of both biomass and total active flux, the Alboran Sea (Z2) stands out in our dataset. The relatively high zooplankton migrant biomass observed here, compared to other stations, may be linked to the influence of the anticyclonic gyre. This gyre enhances both migration from deeper waters and active flux by deepening the thermocline, which promotes the sinking of bacteria and phytoplankton due to inward motion (Arístegui et al., 1994, 1997), thereby providing a significant source of carbon (Yebra et al., 2005). Previous research on zooplankton active flux in that area reported a similar value of zooplankton migrant biomass to ours (993 and 953 mg C·m^-2, respectively), but a zooplankton respiratory flux 2.4 times lower than ours (17.2 and here 41.1 ± 5.4 mg C·m^-2·d^-1, respectively) (Yebra et al., 2018). Even though the values of migrant biomass were rather similar, the higher estimates in zooplankton respiratory flux in our study imply higher zooplankton active flux. The enhanced NPP observed in this study compared to Yebra et al. (2018) (84 and 611 mg C·m^-2·d^-1, respectively) could explain the discrepancy in both respiration and active flux due to increased metabolic activity. This aligns with the expectation that productive areas, characterized by a large and consistent food supply, tend to exhibit higher carbon flux (Hernández-León et al., 2019a). On the other hand, discrepancies in the respiratory fluxes might be related to variations in zooplankton body size (Hernández-León et al., 2024), as smaller organisms tend to exhibit higher respiration rates (Ikeda, 1985). Yebra et al. (2018) sampled the core of western anticyclonic gyre, that was found to host larger organisms compared to nearby areas (Valcárcel-Pérez et al., 2019). Thus, larger organisms at the core of the eddy likely exhibit lower respiration rates compared to the smaller organisms found in the surrounding areas, such as our sampling station.

The relatively low zooplankton migrant biomass values and active flux observed at the northern Atlantic stations could possibly be attributed to zooplankton biomass patchiness, which is common in high-productivity areas and particularly pronounced in upwelling zones (Barton et al., 1998). These sites (Z4 and Z5) also had higher zooplankton biomass at the surface during the day, contrary to expected diel vertical migration patterns. Mesoscale structures, such as eddies and upwelling filaments, promote large differences in phytoplankton and zooplankton populations over short periods and distances in coastal transition zones (Hernández-León et al., 2024). In this context, we suggest that the higher zooplankton biomass found during daytime off Portugal and Galicia was related to important advection due to Ekman offshore transport in upwelling systems, which could induce large mesoscale variability (Batchelder et al., 2002).

While decapods biomass was influenced by the productivity of the area, zooplankton and mesopelagic fish biomass in the water column showed no significant relationship with any of the analyzed environmental variables (Figure 6). This lack of correlation hampers the direct explanation for the distinct vertical profiles observed between the Mediterranean and the Atlantic Ocean. These results suggest that zooplankton and mesopelagic fish biomass vertical distribution in the water column should be also affected by other factors beyond the traditional environmental metrics of Chl a, NPP, temperature, salinity, and oxygen. Drivers may include nutrient stress impacting phytoplankton diversity (Tian et al., 2017), the depth of the deep chlorophyll maximum affecting the trophic coupling between phytoplankton and heterotrophic prokaryotic production (Marañón et al., 2021), as well as predation pressure and competition among zooplankton species (Gage and Tyler, 1992). Furthermore, shifts in the zooplankton feeding ecology -such as a transition from omnivorous to carnivorous or gelatinous filter-feeding zooplankton- could reduce food quality for fish, making it less nutritious (Heneghan et al., 2023).

Finally, it is worth mentioning that only fish biomass was significantly related to fish migrant biomass (Figure 7). Thus, higher zooplankton and decapods biomass in the water column does not imply higher migratory biomass or higher fluxes as one might expect. These findings suggest that factors beyond biomass, such as species composition (Hays et al., 2001), behavior (Forward, 1988), trophic interactions (Pinti et al., 2019) or food supply, play a critical role in governing vertical migration and carbon transport. It becomes increasingly evident that a comprehensive understanding of the influence of environmental conditions on biomass vertical distribution, requires looking beyond traditional metrics (e.g., temperature, salinity, oxygen, and productivity) to uncover the mechanisms shaping biomass distribution.

We found no clear relationship between NPP and total active flux, with active flux estimates up to 10 times higher than previously reported under similar NPP (Ariza et al., 2015; Hernández-León et al., 2019a). Instead, the total active flux varied depending on the composition of the migrating community, which, similarly to biomass, appeared to be influenced by different environmental variables. According to the correlation matrix, active flux by zooplankton and mesopelagic fish was negatively correlated with oxygen (although, not significantly) and positively correlated with temperature and salinity, whereas decapods displayed the inverse pattern (Figure 6). Our loose interpretation of this pattern is that the active flux by these groups is on a broader scale influenced by water masses, with active flux by zooplankton and mesopelagic fish being, relatively, more prevalent in the warmer and saltier waters of the Mediterranean, whereas decapods dominate active flux in the colder fresher Atlantic waters.

Of particular interest is the significantly negative relationship between oxygen concentrations (in the mesopelagic) and zooplankton active flux. This relationship might be related to the influence of oxygen availability on the community structure, vertical distribution (Ekau et al., 2010), metabolic activity (Kiko and Hauss, 2019), as well as feeding and excretion rates (Robinson et al., 2010). However, care should be taken with zooplankton correlation as it is expected the avoidance of the mesopelagic zone if oxygen levels decline below approximately 20 µmol O₂·kg^-1 (Hauss et al., 2016). Thus, enhanced active flux by low oxygen might be related to other factors such as food availability while minimizing predation risk from larger predators that are less tolerant of low oxygen levels (Gilly et al., 2013).

Finally, our findings indicated that carbon flux was mainly driven by diel vertical migrants rather than by the passive flux (see Table 1, Figure 4). Recent global models suggested that active flux of carbon due to zooplankton diel vertical migration accounts for 10-18% of the passive flux (Aumont et al., 2018; Archibald et al., 2019; Nowicki et al., 2022) in areas of low productivity (Koppelmann and Weikert, 2007). In contrast, in high-productivity areas, carbon flux has been suggested to be primarily carried out by diel migrants, due to high biomass of low-turnover organisms such as zooplankton (Yebra et al., 2005, 2018; Hernández-León et al., 2019a). Thus, active flux is expected to be low on a global scale due to widespread oligotrophic conditions, as the transfer of organic carbon to the deep ocean should be enhanced along with net primary production (Davison et al., 2013; Hernández-León et al., 2020). However, our findings suggest that the relationship between the passive and active flux is not directly dependent on productivity.

This study examined different pathways of the biological carbon pump from the Mediterranean Sea to the Atlantic Ocean, revealing strong regional differences in the active flux. We found no clear relationship between NPP and total active flux; instead, flux varied according to the composition of the migrating community. In the Mediterranean Sea, zooplankton were the dominant contributors, with enhanced migrant biomass driving a substantial zooplankton-mediated carbon export. By contrast, in the North Atlantic Ocean, micronekton, particularly decapods, played a prominent role, while mesopelagic fishes contributed more modestly but in patterns linked to the distribution of water masses. Decapod active flux appeared influenced by regional productivity, whereas zooplankton and mesopelagic fish fluxes were more closely associated with the properties of the water column. Finally, the relatively low contribution of passive flux compared to total active flux underscores the central role of the mesopelagic-migrant pump, with distinct taxonomic groups dominating in different oceanic regions.