Locations for this study have been selected to represent four different ocean sectors (Figure 1; Table 1) where productivity is relatively high and hence barite is abundant in the upper water column however these sites represent different saturation states within the mesopelagic zone.

North Pacific: OC1608A, C-SNOW cruise, California current system, coastal North Pacific Station 1 (1,929 m water depth), and OC1608A, C-SNOW cruise, California current system, Santa Barbara basin, Station 2 (4,880 m water depth). These stations are at the edge of the North Pacific oligotrophic gyre. The North Pacific Subtropical Gyre harbors one of the largest biomas on Earth, it is a relatively stable oligotrophic environment, with low surface concentrations of nitrogen and phosphorus. Nutrients derived from advective transport from depth into the surface ocean stimulates primary production in this region (Karl and Church, 2017; Robidart et al., 2019). The two stations in the Santa Barbara basin are on the edge of this gyre and are characterized by relatively high productivity and high phytoplankton biomass that supports a productive pelagic ecosystem (Letelier, et al., 2019). The productivity is fueled by intensive coastal upwelling induced by northerly winds along the California margin (e.g., Brzezinski and Washburn, 2011; Abella-Gutiérrez and Herguera, 2016).

Atlantic: MV1101, Great Calcite Belt (GCB) 1, South Atlantic, Station 117 (Rosengard et al., 2015; Balch et al., 2016). During the GCB1 cruise the R/V Melville crossed the Atlantic sector from Punta Arenas, Chile, to Cape Town, South Africa, sampling between 39^° S and 59^° S. Station 92 from this cruise has been previously analyzed for marine barite (Martinez-Ruiz et al., 2018) and station 117 has been selected for this study (5,048 m water depth). Station GCB1-117 is located in the subtropical region of the South Atlantic Ocean and is dominated by seasonal coccolithophores and diatoms blooms (Smith et al., 2017).

Indian ocean: RR1202, Great Calcite Belt (GCB) 2, South Indian Ocean, Station 63 (Rosengard et al., 2015; Balch et al., 2016; Smith et al., 2017). During the GCB2 cruise, the R/V Revelle crossed the Indian sector from Durban, South Africa, to Perth, Australia, sampling between 37^° S and 60^° S. Station 63 has been selected for this study (1,310 m water depth). The region is characterized by elevated surface reflectance that is thought to result from high seasonal concentrations of coccolithophores. Data for multiple parameters sampled during the GCB cruises including chlorophyll, particulate inorganic carbon (PIC), POC, biogenic silica (BSi), coccolithophore concentration, calcification, photosynthesis, dissolved inorganic carbon (DIC), total alkalinity, iron limitation of phytoplankton and ²³⁴Th-based vertical flux rates are available for these stations (Rosengard et al., 2015; Balch et al., 2016). Shipboard scientists reported dense coccolithophore populations that exported small, highly degraded, and compact particles out of the euphotic zone. Coccolithophore blooms are considered very efficient in transferring POC to the base of the mesopelagic zone, although the magnitude of exported POC is not as high as in diatom-rich regions (e.g., Henson et al., 2012).

South Pacific: NBP1101, Seafarers cruise, Ross Sea, Station 14. This station is located off the Ross Sea Shelf in the Pacific (1,887 m). Samples were collected between January 17, and February 13, 2011 aboard the R/V Nathaniel B (Hatta et al., 2017). The Ross Sea continental shelf is one of the most productive areas in the Southern Ocean (e.g., Smith, Jr. et al., 2014). Here a significant supply of dissolved Fe to surface waters is required to sustain high productivity (Sedwick et al., 2011; Hatta et al., 2017), and include dust, sea-ice, icebergs and upwelling of deeper waters as some of the main inputs (e.g., Measures et al., 2012; Marsay et al., 2014).

Size-fractionated particulate material has been collected using multiple MUL-VFS (Bishop et al., 1985) and battery-operated McLane in-situ pumps (LV-WTS) (Rosengard et al., 2015). Analyzed samples and corresponding depths are indicated in Table 1. Particulate Ba concentrations (pBa) have been determined at the South Atlantic (Great Calcite Belt) and South Pacific Ocean (Antarctic sector) stations, but these data are not available at other stations. In situ deployed filters were processed using the protocol described in Bishop and Wood (2008). Samples for pBa were collected on PES filters to ensure low blank and Ba concentrations in the particulate leachate was analyzed using an iCAP RQ inductively-coupled plasma mass spectrometer. Quantification was achieved via comparison of blank- and indium-normalized ion beam intensities in samples against those measured in a serially diluted multi-element standard that was prepared in house. Precision is generally better than ±3% relative standard deviation. POC samples were collected on pre-combusted QMA filters and concentrations were measured using a CHN elemental analyzer immediately on the ship during the cruises as described in Rosengard et al. (2015). Sampling details and complication associated with particles collected by MUL-VFS as well as retention efficiency are discussed in detail in Bishop et al. (2012). In the South Indian ocean station, both large (>51 μm) and small (1–51 μm) size particulates were analyzed. For the rest of stations, only filters retaining the 1–51 μm fraction were analyzed. The MUL-VFS sampling was found to be highly suitable for barite particles retention and QMA filters were ideal for barite microscopic detection and observation.

To place the p[Ba] data in context, we calculated the barite saturation state of seawater with respect to barite (Ω_barite) at the depths of sample collection Ω_barite is the ratio between the barium and sulfate ion activity product and the barite solubility product. Values of Ω <1, =1, and >1 indicate under-, perfect-, and super-saturation, respectively. For consistency with the literature, we consider water samples with Ω between 0.9 and 1.1 as being in saturation equilibrium (e.g., Monnin et al., 1999). Since co-located samples for analysis of dissolved Ba and sulfate were generally not available for our study, we estimated Ω_barite from nearby stations with reliable published Ba concentration depth profiles (see Table 2 for details and Figure 1 for locations). Calculations were performed using PHREEQC version 3 (Parkhurst and Appelo, 2013). Values of Ω_barite were computed for each sample based on input parameters of in situ temperature, d[Ba], pressure (estimated from depth). Both pH and salinity were prescribed in all calculations at 8.1 and 35, respectively. The major ion composition of seawater in the calculations was based on that reported by Kester et al. (1967). We believe that using a fixed salinity in our calculations is a reasonable assumption given the relatively minor effect this property has on Ω_barite over the range of salinities encountered in open ocean seawater.

Quartz fiber filters (Whatman QMA) have been used for scanning electron microscopy (SEM) observation and analyses. Representative filter pieces were coated with carbon for observation under the SEM using an AURIGA FIB-FESEM Carl Zeiss SMT microscope equipped with an energy dispersive X-ray (EDX) detector system (Centre for Scientific Instrumentation, University of Granada). Filter pieces were grounded in an agate mortar and then dispersed in ethanol by sonication for approximately 3 min. Particulate matter released from the filter was deposited on carbon-film-coated copper grids for high-resolution transmission electron microscopy (HRTEM) observation by using a FEI TITAN G2 60–300 microscope with a high brightness electron gun (X-FEG) operated at 300 kV and equipped with a Cs image corrector CEOS (Centre for Scientific Instrumentation, University of Granada). For analytical electron microscopy (AEM), a SUPER-X silicon-drift windowless EDX detector was used. EDX maps and selected area electron diffraction (SAED) patterns were also collected on barite particles for crystallographic characterization and for determining major constituents composition.

To date, a large body of work from multiple oceanographic expeditions and sampling location has provided ample datasets of dBa and non-lithogenic pBa distribution in the ocean water column. In general, the GEOTRACES-era datasets exhibit similar depth-dependent patterns in pBa to those shown here; however, this study also adds novel results regarding the mechanisms behind these distributions. It has been demonstrated that barite abundance shows significant spatial differences mostly related to productivity. Also, significant differences are recognized with depth in the water column since processes involved in precipitation are occurring at certain depths and the barite may dissolve deeper in the water column. Since pBa is closely correlated with the flux of organic carbon, it is enriched in the mesopelagic zone and typically shows a maximum abundance at intermediate depths (200–600 m) (e.g., Dehairs et al., 1980; Bishop, 1988; Dehairs et al., 1991; Dymond and Collier, 1996; Dehairs et al., 1997; Dehairs et al., 2008; Stenberg et al., 2008; Jacquet et al., 2011; Planchon et al., 2013; Lemaitre et al., 2018; Conte et al., 2019). Overall, vertical pBa profiles are similar to those of calculated oxygen consumption rates, which supports the link between organic matter degradation and barite formation (e.g., Dehairs et al., 1997). Importantly, pBa has been correlated with rates of microbial degradation of organic matter, which further support the link to oxygen consumption and carbon respiration. Barium isotopes also support barite formation at mesopelagic depths as demonstrated by enrichment in the isotopically-heavy Ba in seawater (¹³⁸Ba) and depletion of the lighter Ba (¹³⁴Ba) due to the preferential incorporation of the lighter Ba isotopes in barite (Horner et al., 2015; Hsieh and Henderson, 2017; Bates et al., 2017; Bridgestock et al., 2018). Indeed a local maximum in δ¹³⁸Ba at depths between 200 and 600 m in diverse ocean basins indicates that barite precipitation mostly occurs at these depths (Horner et al., 2015; Bates et al., 2017). Our pBa data from the two analyzed stations at the South Atlantic (Great Calcite Belt) and South Pacific (Antarctic sector) stations are also consistent with the idea of enhanced barite formation at this depth the mesopelagic zone. Profiles show higher pBa concentrations below 200 m and down to 600 m with a maximum in at about 200–400 m. Though qualitative, SEM observations from all the analyzed stations similarly show higher barite abundance at these intermediate depths.

As discussed above, barite formation is linked to organic carbon mineralization and export production, however a notable spatial variability in the Ba:Corg ratio is found over ocean regions. For instance, sediment trap samples from the Western Atlantic have significantly lower Ba/Corg values than samples from the Pacific (e.g., Dymond and Collier, 1996). The causes for this spatial variability are still poorly understood. Differences in the Ba:Corg ratio have been related to the efficiency of mineralization of POC in the mesopelagic zone relative to the exported amount (e.g., Francois et al., 1995). Thus, using algorithms that relate Ba to carbon export may not be appropriate in regions of highly variable carbon flux. Moreover, barite formation may be affected by the rate at which particles sink, given that particles that are quickly removed from the water column by rapid sinking may reduce the likelihood for precipitation of particulate Ba phases (McManus et al., 2002). This has also added uncertainty to sedimentary Ba interpretations and paleoproductivity reconstructions, particularly to the use of Ba content as a quantitative proxy for reconstructing productivity. In general, the poor understanding of barite distribution in the water column stems from our limited knowledge of the processes leading to barite saturation. A better knowledge of such processes may improve our ability to assess changes in past productivity. Results from experimental work and from the analyses of pBa phases in the water column have recently shed light into potential mechanisms leading to saturation and precipitation of barite in microenvironments within sinking particulate matter in the mesopelagic zone (Martinez-Ruiz et al., 2018; Martinez-Ruiz et al., 2019). Collectively this work emphasizes the role of biofilms as Ba-concentrating agents in a process that could be termed organo-mineralization. According to the Encyclopedia of Geobiology, organo-mineralization (Défarge, 2011) is a process of mineral formation mediated by organic matter (OM), independent of the living organisms which the OM derives from. The organic compounds may be excretion products or detached parts of living organisms, or relics and by-products of dead organisms that have been released into waters or incorporated into soils, sediments, or rocks. Our observations from suspended marine particulate matter agree with this process and further support previous findings. Our SEM and TEM observations at multiple stations also demonstrate that barite forms through a P-rich amorphous precursor phase, seen in the pBa composition, ranging from Ba-phosphate to Ba-sulfate consistent with previous studies from two Atlantic Ocean stations (Martinez-Ruiz et al., 2019). The present study adds further evidence in support of these findings. Moreover, although the Ω_barite (Table 2) in the epipelagic and upper mesopelagic zone at the newly studied ocean sectors differs from site to site, this does not seem to be a major control in barite precipitation as no relation between Ω_barite and pBa or barite abundance is evident. At all the studied stations, barite crystals show similar characteristics in terms of composition, size, distribution and association to organic material, prominently showing association with EPS-like morphologies, which further supports EPS production is a major factor in promoting barite formation ( Figures 2–5 ) as well as organo-mineralization as a common process for barite formation throughout the ocean. Considering that the relation of pBa with export production depends on microbial processes related to organic matter degradation, barite formation is therefore linked to the suite of complex processes involved in the ocean biological carbon pump. Specifically, the fraction of primary production that leaves the upper ocean and is exported to depth, defined as export production is the fraction that “fuels” barite formation. Export production depends on diverse factors such as phytoplankton and zooplankton community structures, the formation of aggregates, sinking by ballasting, and bacterial mineralization rates (e.g., Francois et al., 2002; Cavan et al., 2015; Belcher et al., 2016; Le Moigne et al., 2016) and these factors change in space and time. For example, Henson et al. (2019) demonstrated that low primary production and high export efficiency regimes tend to occur when macro-zooplankton and bacterial abundances are low in the surface ocean. Thus, a large fraction of primary production is exported, likely as intact cells or phytoplankton-based aggregates. In contrast, when macro-zooplankton and bacterial abundances in the surface ocean are high, the export efficiency decreases. These results support that the whole ecosystem structure, rather than just the phytoplankton community, play a major role in export efficiency (Dehairs et al., 1992). All these factors not only depend on seasonality but can also be very different at regional and global scales. Hence, appropriate knowledge of the processes involved in carbon export fluxes, the formation of organic aggregates, and particle sinking is required for assessing the relationship between productivity and Ba proxies, and proper interpretative care and caution are required for using Ba as a proxy for export production. As the relation between export production and barite largely depends on microbial processes and EPS production, temperature, oxygen abundance, and the type of organisms involved may impact this relationship. Although the reason for Ba accumulation in bacteria and EPS or in other living organisms is not yet well understood, it is known that bioaccumulation of Ba occurs throughout the ocean. The nucleation and crystallization of barite results in the formation of a highly stable mineral that is hard to dissolve under oxygenated conditions, consequently a relatively large fraction of the particulate barite that forms reaches the sediments (Paytan and Kastner, 1996). Barite accumulation in the sediment would therefore represent a record of the combination of diverse processes including export productivity, organic matter degradation, bacterial activity, and EPS production. This complexity should be considered when interpreting temporal and spatial variability in the Ba:Corg ratios and in barite accumulation in marine sediments.

Understanding the microbial processes leading to the formation of the mineral barite in the oceanic water column is crucial to determining the utility of Ba proxies for paleo-productivity and paleo-chemistry reconstructions. How and why Ba associates with organic matter in microenvironments and how the rates of organic matter decomposition affect barite production are key questions that link primary productivity or export productivity to barite abundance in marine sediments. The important role that EPS and microbial cells may play in nucleation and crystallization in the ocean is still far from being well understood at the molecular scale. Even though the EPS in the ocean have been widely investigated, their role in mediating mineral precipitation remains mostly unknown, in particular as it pertains to barite formation. In seawater, these secretions facilitate attachment to surfaces leading to the formation of biofilms, organic colloids, and larger aggregations of cells (marine snow). Though difficult to measure accurately, EPS represent a significant portion of the bioavailable carbon pool in the ocean. These substances occur in a range of molecular sizes, with diverse physical and chemical properties, and their composition includes polysaccharides, proteins, lipids, and nucleic acids (Decho and Gutierrez, 2017). In general, the attachment of microbes to surfaces, or to each other, provides higher environmental stability than being a free-living cell and may be favorable in the ocean (e.g., Flemming et al., 2016). The EPS matrix of biofilms provides a three-dimensional architecture framework (Decho, 2000) that is the building block of the aggregates suspended in the water column. These organic aggregates are known to be very rich in microbial communities with abundances up to two orders of magnitude higher than in the surrounding seawater environment (e.g., Alldredge et al., 1986; Herndl, 1988).

While both experimental work and observations in diverse natural environments have demonstrated that functional groups associated with EPS are able to bind different metal ions (e.g., Braissant et al., 2007; Tourney and Ngwenya, 2014), their precise role for the binding, trapping and concentrating metals in the open ocean has not been sufficiently investigated. For instance, in bioremediation, it has been demonstrated that the polyanionic nature of the EPS promotes the binding of heavy and toxic metal ions, and EPS use for detoxification of heavy metals is well known. Many examples have been described in the literature such as in the remediation of Cd, Cr, Pb, Ni, Cu, Al, and U (e.g., Beech and Cheung, 1995; Iyer et al., 2005; Bhaskar and Bhosle, 2006; Gerber et al., 2018). However, as many of these studies had commercial purposes, very few have addressed the ecological implications of marine EPS in metal biogeochemical cycles. Loaec et al. (1997, 1998) reported the heavy metal-binding capacity of EPS produced by hydrothermal vent bacteria and suggested that this could represent a survival strategy for the bacteria by reducing their exposure to toxic metals released from the hydrothermal vents. Major elements such as Na, Mg, Ca, K, Sr, and Si, have also been shown to be adsorbed by marine bacterial EPS (Gutierrez et al., 2008). Likewise, Fe uptake by EPS in eukaryotic phytoplankton has been investigated (Hassler et al., 2011; Gutierrez et al., 2012), and binding of Th to carboxylate, phosphate and sulfate groups in marine EPS has also been shown (Alvarado Quiroz et al., 2006). Nevertheless, the ecological implications of these binding processes are not well understood, and Ba has not been investigated yet in this regard. Thus, our data open an unexplored field and support the crucial role that EPS play not only for Ba bioaccumulation but also for mineral precipitation in the ocean, with important implications for paleo-oceanographic reconstructions. Although further investigations are required to elucidate precise nucleation and crystal growth processes, the available data strongly support that biofilm matrix is crucial for metal precipitation in the ocean. In this dynamic environment, with abundant microbial cells, polysaccharides and water, together with excreted cellular products (e.g., Sutherland, 2001), functional groups including sulfate and phosphate would contribute to the overall negative charge of the EPS, and these functional groups would interact with metals promoting precipitation.

The use of barite as a proxy to gain insights into past microbial processes is a promising tool in paleoceanographic research. It is broadly known that microbial communities play a major role in biogeochemical cycles since they play a role both at the base of the oceanic food web and as decomposers (e.g., Falkowski et al., 2008; Robidart et al., 2019). Further knowledge of microbial productivity and structure of communities is required for predicting future marine ecosystem functions, and the impact of increasing environmental effects on ocean ecosystems. This is challenging at present because biogeochemical processes and microbial communities are very complex, but it is far more complex for the past because a record of microbial processes is not usually well preserved in marine sediments beyond the preservation of biomarkers and some minerals that form through direct or indirect association with microorganisms. Accordingly, although barite accumulation rates are closely correlated with carbon export to the deep ocean (Carter et al., 2020 and references therein), the occurrence of barite may also reflect Ba utilization in the surface ocean through microbial processes, EPS production and organic matter mineralization. Except in sulfate reducing environments, barite is well preserved in marine sediments, thus the presence of pelagic barite particles is an indication of past bacterial respiration processes. In fact, barite has been proposed as a good proxy reflecting average mineralization processes (e.g., Cardinal et al., 2005), which are in turn a major control in the global carbon cycle and atmospheric carbon sequestration (e.g., Cavan et al., 2017). Our observations at diverse ocean sectors and depths showing formation within organic aggregates commonly rich in EPS, further support that mineralization due to microbial respiration is responsible for barite formation and consequently barite is a bioindicator for such processes.

The occurrence of barite in marine particles may have additional effects that have so far not been thoroughly investigated. For example, the potential role of barite particles within marine snow in the ballasting and mineralization controls of carbon sedimentation. It has been demonstrated that particulate minerals, for instance eolian dust, can be incorporated into organic aggregates and act as ballast enhancing the marine carbon export hence the significant increase in the sinking velocities of aggregates (Van der Jagt et al., 2018). Although biological processes affecting the fragmentation and mineralization of large particles are the most important factors determining the POC profiles (e.g., Lam and Bishop, 2007), barite is a high-density mineral that could also affect export processes. Another important aspect that still requires further investigation is the distribution of barite at greater depths than the mesopelagic zone. To date most of the studies on barite distribution have focused on mesopelagic depths and little is known about distribution and potential precipitation or dissolution with depth since only very few data from deeper samples are available (Conte et al., 2019). Furthermore, in the bathypelagic ocean (depth >1,000 m), Archaea and Viruses are particularly important in the microbial loop, but they remain largely unexplored in deep waters, and interactions between microbes and minerals beyond bacterial precipitation is almost unknown, except some recent work on the role of viruses in carbonate precipitation (e.g., Lan et al., 2020; White III et al., 2020).