The reservoir of particulate organic carbon (POC) in the ocean, while considerably smaller than the dissolved organic carbon (DOC) pool (with an approximate POC : DOC ratio of 1:38, Denman et al., 2007), is an essential constituent of sequestered carbon, in addition to the dissolved and particulate inorganic carbon pools (Sarmiento and Gruber, 2006). POC sinks through the water column across isopycnals, scavenging other particles and transporting carbon and associated elements to deep waters, where some of them are remineralized or resuspended, and a minor fraction is eventually sequestered in the sediments (Eppley and Peterson, 1979; Stramska, 2009).

The POC in oceanic regions consists mainly of living organisms and marine snow, which is an aggregate of phytoplankton and small zooplankton fragments, fecal pellets, and other particles also known as organic aggregates, which includes non-living material and uncharacterized components from biological, mineral, and anthropogenic sources (Riley, 1963; Knauer et al., 1979; Dunbar and Berger, 1981; Hedges et al., 2000; Lutz et al., 2002; Stramski et al., 2008; Kharbush et al., 2020). Near active natural oil seeps or oil spills, POC can incorporate “dead carbon” or hydrocarbon-derived carbon, in a process that has been described as Marine Oil Snow Sedimentation and Flocculent Accumulation. This process has been shown to be an important pathway for the distribution and destination of spilled oil, which in the Gulf of Mexico accounts for up to 14% of the total oil released as a result of the Deepwater Horizon oil spill in 2010 (Daly et al., 2016).

Particulate organic carbon concentration in the euphotic layer is thought to reflect primary production, a process during which photosynthesis fuels the formation of organic carbon from dissolved inorganic carbon and nutrients, thereby decreasing the partial pressure of carbon dioxide in surface waters (Stramska, 2009). Many of these processes are sensitive to fluctuations in temperature, light availability, ocean circulation, vertical mixing, and nutrient inputs into the surface ocean’s euphotic layer (Dawson et al., 2018; Frenger et al., 2018; Dobashi et al., 2022).

In oligotrophic regions, the mixed layer is generally considered as a quasi-homogeneous with little to no variability in density and is usually depleted in nutrients due to uptake by primary producers (Falkowski et al., 1992; Velásquez-Aristizábal et al., 2022). The deepening of this layer allows for subsurface mixing of nutrients toward the surface, fueling biological productivity and consequently increasing POC concentrations (Gardner et al., 1999). These changes can be further driven by vertical mixing from the isopycnal displacement by anticyclonic and cyclonic eddies, especially under strongly stratified upper ocean conditions during summer (Gaube et al., 2014; Gaube et al., 2019).

Mesoscale processes are responsible for the mixing and transport of heat, salt, and biogeochemical tracers across the global oceans and may act as a moderating factor in global climate change (Faghmous et al., 2015). Thus, understanding ocean eddy dynamics and their role in influencing various oceanographic and biological phenomena is of keen scientific interest (Falkowski, 1998; Faghmous et al., 2015). The productivity enhancing effects of eddies is particularly important in low-nutrient environments, where mesoscale processes can regulate the net upward flux of limiting nutrients as a result of the undulation of nutrient isosurfaces: where the shoaling of isopycnal surfaces tends to bring nutrients into the euphotic zone thereby enhancing productivity, whereas their deepening leads to a lack of change or even a decrease in primary productivity (Gruber et al., 2011).

Remotely sensed sea surface heights (SSH) yield important information on the spatial distribution and intensity of ocean eddies and is strongly related to circulation patterns in the ocean, lending insight into nutrient transport crucial to the understanding of the spatial variability of biological production (Gruber et al., 2011). The measurements performed with various in situ and remote sensing platforms provide an effective tool for studying biogeochemical constituents of oceanic waters like chlorophyll-a, particulate inorganic carbon, and particulate organic carbon (Stramski et al., 2008; Ocean Biology Processing Group, G. M, 2022).

The GoM is a semi-enclosed sea linked to the Atlantic Ocean through the Florida Straits and to the Caribbean Sea through the Yucatan Channel (Oey et al., 2005). The deep-water region of the GoM is characteristic for its oligotrophic and nutrient-limited surface waters, which are relatively isolated from coastal eutrophic waters (Martínez-López and Zavala-Hidalgo, 2009; Pasqueron de Fommervault et al., 2017). The Loop Current (LC) brings warm water into the GoM, which leads to high stratification in its area of influence, especially during the summer. Higher intrusion of the LC into the GoM is observed more frequently during the summer, which along with the upper ocean warming, influences the circulation in the interior of the gulf (Sturges and Leben, 2000). In the GoM, positive SSH anomalies are indicative of anticyclonic eddies that are characterized by a deep, warm water layer (Müller-Karger et al., 2015; Dobashi et al., 2022).

The LC and detached LC eddies transport Caribbean Subtropical Underwater into the GoM, which is clearly distinguishable from the GoM common water (Vidal et al., 1992). This common water is formed by vertical convective mixing during the winter season, or by the mixing induced by the collision of detached LC eddies against the western GoM boundary (Schmitz et al., 2005). These mesoscale features drive hydrodynamic processes throughout the central GoM that fuel mixing processes and affect biological productivity patterns (Pérez-Brunius et al., 2012; Müller-Karger et al., 2015; Cervantes-Díaz et al., 2022).

The most energetic mesoscale events are the detachment of large anticyclonic eddies from the LC that drift to the west (Sturges and Leben, 2000), which are dynamically linked with cyclonic and anticyclonic eddies in the Gulf’s interior (Jouanno et al., 2016). The periods between detachments are highly variable (ranging between 0.5 and 18.5 months), implying that there are some years with no eddy detachments from the LC, while during other years these can happen up to three times per year (Sturges and Leben, 2000; Leben, 2005; Vukovich, 2007; Vukovich, 2012; Delgado et al., 2019).

In addition to the quasi periodic shedding of LC eddies, other processes and structures influence the productivity in the southern Bay of Campeche, like the semi-permanent cyclonic eddy in the southern Gulf (Pérez-Brunius et al., 2012), the confluence of seasonal along-shelf currents (Martínez-López and Zavala-Hidalgo, 2009), and upwelling processes along the eastern margin of the Bay of Campeche and off the northeast shelf of the Yucatan Peninsula (Zavala-Hidalgo et al., 2006).

The mixed layer depth’s (MLD) seasonal cycle is driven by the surface ocean cooling during winter months, fueled by northern winds that lead to the erosion of the stratification of its surface waters and a deepening of the mixed layer (Müller-Karger et al., 1991). This wintertime convection of the MLD due to strong winter winds or “nortes” ends during the late spring, when the warming cycle of the surface waters begins; the MLD becomes shallower throughout the summer into the early fall, when the stratification of the upper layers is strongest (Müller-Karger et al., 1991; Martínez-López and Zavala-Hidalgo, 2009). These cycles of winter mixing and summer stratification are thought to be the main controlling factors of the surface chlorophyll-a concentration in GoM (Müller-Karger et al., 2015).

Modeling efforts suggest the higher integrated content of chlorophyll-a (Chl-a) over 350 m occurs in winter whereas lower ocean concentrations happen in summer, while April-May and October-November are transitional periods (Damien et al., 2018). In addition, float data show that mesoscale activity is most likely the main source of variability for the Chl-a concentration in the deep-water region of the GoM, with higher concentrations observed in cyclones compared to anticyclones (Pasqueron de Fommervault et al., 2017; Linacre et al., 2019).

Additionally, cyclonic and anticyclonic eddies drive vertical mixing through the modulation of nutrient transport to the euphotic zone, thereby exerting influence on Chl-a concentrations in the euphotic layer (McGillicuddy, 2016; Honda et al., 2018) as well as POC concentrations (McGillicuddy, 2016). Still, there are limited field observations available to characterize how mesoscale physical processes may affect the POC of the deep-water region of the GoM (Müller-Karger et al., 2015).

POC in the GoM has been mostly studied over the continental shelf, particularly in the north and northeast. The transport of freshwater plumes and filaments from the Mississippi and the Atchafalaya Rivers advect higher nutrient concentration waters to the northern shelf region, especially to the continental shelves of Louisiana and Texas (Biggs, 1992; Trefry et al., 1994; Goñi et al., 1997; Bianchi et al., 1997; Wang et al., 2004). However, the role of mesoscale structures on POC concentrations in the GOM’s deep-water region, especially in the southern gulf and the Bay of Campeche, known for its relatively higher productivity within the gulf (Martínez-Lopez and Zavala-Hidalgo, 2009; Linacre et al., 2015), has yet to be examined. Hence, our study aims to understand the role of mesoscale eddies on POC concentrations in the upper layer of the southern GoM´s deep-water region.

The mechanisms that control POC concentrations in the ocean differ depending on the region and the season (Dobashi et al., 2022). Sources of organic carbon in the GoM can include terrestrial and marine sources (Bianchi et al., 1997). The contribution of terrestrial organic carbon from terrestrial sources, including estuaries, is particularly important in coastal and shelf waters and constitutes an important component of global organic carbon (Hedges, 1992). Specifically, river inputs play a significant role in the distribution of POC (Meybeck, 1982). The Mississippi-Atchafalaya River and the Grijalva-Usumacinta River Systems are the largest sources of freshwater inflow to the northern and southern GoM, respectively (Kemp et al., 2016). The Mississippi- Atchafalaya River System has been shown to influence the distribution of POC and plays an important role in the organic carbon transport to the shelf and northern oceanic region (Trefry et al., 1994; Rosenheim et al., 2013), while in the southern Gulf the Usumacinta-Grijalva River System is the main source of POC (Cuevas-Lara et al., 2021). In addition, POC fluxes are further influenced by mineral ballast, such as calcium carbonate, the concentrations of which are closely related to POC concentrations (Klaas and Archer, 2002).

The concentration profiles in the GoM’s deep-water region showed values that range between 2 and 7µM in the upper layer during the summer season. These concentrations are comparable with the values (4.48 ± 5.22 µM) previously reported for samples obtained under oligotrophic conditions in South and North Pacific Gyres in 2004 and 2012 (Stramski et al., 2022; Table 6 ). Also, POC concentrations in the Equatorial Pacific Ocean collected between 30°N-15°S and 180-160°W in 2012 range between 0.75 and 8.8 µM; the smallest values were from oligotrophic regions, and the highest for the high-nutrient Equatorial Pacific region (Graff et al., 2015).

Stramski et al. (2022) sampled different oceanographic conditions ranging from oligotrophic environments like the South Atlantic Gyre to higher productivity regions of the Atlantic Ocean. They report concentrations of 4.77 ± 1.68 µM along north-to-south transects covered in 2005 and 2010. In another study, POC concentrations ranging between 1.67 and 6.33 µM were reported from samples collected in the North Atlantic Drift, North Atlantic Subtropical Gyre, North Atlantic Tropical Gyre, Western Tropical Atlantic, the South Atlantic Gyre, and the South Subtropical Convergence during 2014 (Strubinger-Sandoval et al., 2022).

Reports from comparable oligotrophic regions like the Sargasso Sea showed POC concentrations between 1.18 and 2.96 µM from 1988 to 2004 (Duforêt-Gaurier et al., 2010; Table 6 ), while the Hawaii Ocean time-series showed a range between 2 and 5 µM (Świrgoń and Stramska, 2015). In the southern deep-water region of the GoM, we found POC concentrations that are comparable with those reported for the northern open waters of GoM by Liu and Xue (2020), who measured concentrations of 3 ± 1 µM in May between 2010 and 2013 along the coast of Louisiana in the north GoM. However, they also reported much higher concentrations in the surface waters influenced by the plumes near the Mississippi River Delta (values around 14 ± 6 µM). These high concentrations were attributed to the discharge of the Mississippi and Atchafalaya Rivers.

Hence, POC concentrations in the southern deep-water region of the GoM fall within the range observed in oligotrophic regions like the Sargasso Sea, and the subtropical gyres in the Atlantic and Pacific oceans during the summer season. An analysis of the global distribution of integrated POC₁₀₀ from 1995 to 2012 showed a range of values from ~20 to ~900 mmol m^-2 (Balch et al., 2018). The comparison of global estimates with our results is consistent with the oligotrophic nature of the southern deep-water region of GoM during the summer season.

POC concentrations in the water column showed the consistent well-known exponential decay with depth, with relatively high variability in the euphotic layer and decreasing concentrations below 700 m (<1 µM). To understand the mechanisms that could control this upper layer variability in the southern region of GoM, POC concentrations were integrated over different depths (100 m, MLD, MFD) and correlated with different variables. Integrated POC within the mixed layer depth and the fluorescence maximum depth did not show significant correlations with any of the oceanographic variables for any of the 3 years ( Supplementary Table 7 ).

However, when integrating over the euphotic layer (with depths of ~100m; Linacre et al., 2019; Kelly et al., 2021; Stukel et al., 2021; Stukel et al., 2022) our results showed a negative correlation with the MFD and the MLD and a positive one with the surface Chl-a concentration for the 2015 cruise. The POC₁₀₀ showed similar average integrated concentrations of ~276 mmol m^-2 for 2015 and 2016, although the variability was much higher in 2015 than in 2016. In contrast, during the 2017 cruise, POC₁₀₀ was lower, with a mean value of 207 mmol m^-2. Differences in POC₁₀₀ between years were significant due to the high variability between stations. These data suggest that MFD and MLD may occasionally play a role in modulating POC concentrations in the upper water column although it is not a consistent control over time and that there may be other mechanisms contributing to this intermittent relationship.

It is important to mention that although the three cruises took place during the summer season ( Table 1 ), the level of mesoscale activity differed ( Figure 3 ). During 2015 and 2016 mesoscale activity was higher in the central GoM than in 2017. The observed significant correlations between POC₁₀₀ and the MLD, MFD and surface Chl-a concentrations for the 2015 cruise may be related to the two large anticyclones that were present in the central deep-water region; these were the remnants of Nautilus eddy and the recently released Olympus eddy ( Figure 3 ). Thus, the higher variability in integrated POC concentrations found for 2015 ( Table 3 ), may be attributed to an intensification of mixing processes associated with the edges of the anticyclonic eddies and consequently with the vertical transport of nutrients fueling the POC production in the euphotic layer. According to the results from the algorithm of AVISO, the average number of anticyclones during the cruise in 2015 was 6.7, in 2016 it was 5.4 and in 2017 was 4.7. Hence, the number of mesoscale structures may be another contributing control to the variability of integrated POC_100, especially during the summer season characteristic for strongly stratified upper ocean conditions.

Large and highly energetic mesoscale eddies exert a strong influence on the biogeochemical variables of the upper water column (Gaube et al., 2019). Eddies are associated with vertical mixing, which affects nutrient transport and the Chl-a concentration in the euphotic layer (McGillicuddy, 2016), and our results in Figure 7 show that the POC upper layer concentrations may be modulated by their location within cyclonic and anticyclonic eddies. However, although we observed significant differences in the medians considering the first 5 m of the water column, we did not find significant differences in the integrated POC₁₀₀ medians.

Higher POC concentrations, associated with the shoaling of density profiles and the deep fluorescence maximum are consistently observed at stations near the CCE than at stations near the BCE. These features are consistent with a shallowing (deepening) of the pycnocline near the core (border) of cyclonic eddies. Cyclonic eddies show an anti-clockwise flow of currents with negative SSH_ns at its center and counterclockwise geostrophic velocities, a shallower thermocline and a conspicuous shoaling of isopycnals bringing colder, nutrient-rich subsurface waters closer to the euphotic zone at its center (Biggs et al., 1988; Seki et al., 2001), all of them enhancing phytoplankton productivity, higher chlorophyll and POC concentrations in the euphotic layer (Bidigare et al., 2003; McGillicuddy et al., 2003; Bakun, 2006).

The BCE shows consistently lower POC concentrations than stations in the CCE, with close to ~0.5 µmol L^-1 difference in their mean values ( Figure 7 ). The Chl-a concentrations at the center of a cyclonic eddy are consistently higher than at the borders, most likely as a result of the shoaling of the pycnocline allowing nutrients to reach the euphotic layer and thus fuel the phytoplankton productivity at depth (Gaube et al., 2014; McGillicuddy, 2016). Honda et al. (2018) have shown how cyclonic eddies may affect the POC flux in the oligotrophic region of the subtropical North Pacific Ocean through the shoaling of the pycnocline driven by the passage of cyclonic eddies over the observation area, resulting in increased Chl-a in the subsurface layer and enhanced POC flux to the deep-sea region. This flux is positively related to the concentration of POC in the water column (Roca- Martí et al., 2021).

In contrast, anticyclonic eddies with clockwise geostrophic velocities (arrows) are convergence zones favoring the piling up of relatively warmer and lower density water masses at their core ( Supplementary Figure 4 ), producing positive SSH_ns as seen in Figure 3 . Anticyclonic eddies are generally considered nutrient-depleted relative to cyclonic eddies (Biggs et al., 1988; Seki et al., 2001). However, the POC fluxes observed in anticyclones in the eastern tropical North Atlantic show the opposite behavior, with larger fluxes comparable to those observed in mesotrophic or eutrophic regions (Fiedler et al., 2016; Fischer et al., 2016). Here, stations at the CAE mostly showed lower POC average concentrations by ~0.4 µmol L^-1 ( Figure 7 ) than at those in the BAE. In addition, near the core of the anticyclones there was lower density gradient with depth, lower fluorescence and a deeper MFD. A characteristic transect through an anticyclonic eddy shows a deepening (shallowing) of the pycnocline and MFD near the core (border) of these mesoscale features.

Previous work reported differences in cyclonic and anticyclonic eddies within the GoM. Lee-Sánchez et al. (2022) analyzed vertical nutrient profiles for two cruises, one in the winter (February-March) of 2013 and the other in the beginning of summer (June) of 2016. Their results showed that mesoscale eddies play a strong modulating role in the vertical distribution of nitrate and nitrite, particularly in the first 250 m. In the cores of recently detached Loop Current eddies, the nitracline reached maximum depths, and there were lower nutrient concentrations in the euphotic layer. Data from ARGO measurements in the GoM showed that the depth and the magnitude of the DFM or deep chlorophyll maximum are also strongly controlled by the mesoscale variability, showing consistently lower chlorophyll concentration in anticyclones (Pasqueron de Fommervault et al., 2017 and Damien et al., 2021). Based on a biogeochemical model, primary productivity in the Loop Current Eddies is higher than the average rates in the surrounding open waters of the GoM. This anomalous behavior is explained by the mixed layer response during winter convective mixing, which reaches deeper into nutrient-rich waters due to the very low-density gradient with depth in the the anticyclonic eddy. Although we didn’t observe this process, since our data was always collected in the summer season, we coincide with their observation of the strong influence exerted by the mesoscale activity in the surface ocean chlorophyll concentration and more specifically within Loop Current Eddies (Damien et al., 2021).

All these observations show how mesoscale structures do exert some influence in the concentrations of POC in the upper layer. However, correlations of POC concentrations versus SSH_ns, which in principle is related to the mesoscale structures, didn’t show a significant relationship. To explore this apparent lack of correlation we calculated the Pearson correlation coefficient between satellite POC concentrations and SSH_ns in every coordinate of the data mesh of the GoM, considering the average of 8 days for the three summer months of 2015, 2016 and 2017. The relationship is presented in Figure 9 , and a clear negative relationship between the POC concentrations in the surface ocean and SSH_ns was observed over a broad region in the central abyssal region, especially during 2015 and 2016. This is most likely related to the east-west trajectories of large, detached eddies from the LC. These negative correlations show lower POC concentrations associated with higher SSH_ns, implying the importance of these mesoscale structures in controlling POC concentrations in the euphotic layer of the deep-water region of the GoM.

Although our observations show how the POC concentrations may be modulated by mesoscale dynamics in the southern deep-water region of the GoM, it is important to mention that other processes deserve further studies to explain the observed variability. For instance, primary production is associated with seasonal high freshwater inputs, including those from the Tamaulipas shelf in the west, the Mississippi-Atchafalaya River System in the north, and the outflows of the Grijalva-Usumacinta River System in the south ( Figure 1 ). In addition, further studies should be focused on the possible effects of increasing the stability of the upper water column due to rising sea surface temperatures in the interior of the GoM, and how this warming affects mixing in mesoscale features, and consequently primary production and upper layer POC concentrations to better understand the biological carbon pump in the GoM.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://zenodo.org/record/7302327.

YC: Formal analysis, methodology, visualization, data curation, investigation, conceptualization, writing – original draft preparation. SH: Project administration, funding acquisition, writing-review and editing, manuscript improvement. GV: Data analysis, manuscript improvement. JH: Conceptualization, supervision, investigation, methodology, data curation, resources, project administration, funding acquisition, writing-review and editing, manuscript improvement. All authors contributed to the article and approved the submitted version.