The Solomon Sea: its circulation, chemistry, geochemistry and biology explored during two oceanographic cruises

The Solomon Sea is a semi-enclosed sea located at the western boundary of the equatorial Southwest Pacific Ocean, between Papua New Guinea and the Solomon Islands (Figure 1). Suspecting the importance of the Solomon Sea circulation to the general Pacific circulation, Lindstrom et al. (1987) made two surveys in 1985 and 1986 that revealed large and vigorous transports toward the equator. This region appeared as an essential pathway for waters originating from mid-to high latitudes and joining the western equatorial Pacific and the equatorial current system. Water properties indicated that the Solomon Sea was in particular the major source of the Equatorial Undercurrent (Tsuchiya et al., 1989), and thereby of the downstream surface waters of the cold tongue in the eastern equatorial Pacific. Since then, the area had essentially only been surveyed by underway shipboard current profilers during research vessel transits or geophysical cruises (Cravatte et al., 2011).

Recently, the CLIVAR Southwest Pacific Ocean Circulation and Climate Experiment (SPICE) pinpointed the region as a key passageway for the low-latitude western boundary currents, providing important components for the oceanic and climatic variability of the equatorial Pacific at interannual and decadal time scales (Ganachaud et al., 2014). Water mass mixing, and geochemical and biogeochemical enrichments in the Solomon Sea may indeed impact the downstream surface waters and the productivity of the equatorial cold tongue.

The oceanic conditions in the Solomon Sea are distinctive in several ways. First, strong currents interacting with numerous islands and straits impact water properties (Melet et al., 2011). Then, water masses coming in contact with the coasts undergo not only dynamical transformations but also enrichments of macro-and micro-nutrients and trace elements through land/sea exchange, or boundary exchanges (BE) (Grenier et al., 2011), including river discharge, shelf inputs and submarine groundwater discharges. Trace elements and their isotopes that are injected from BE provide information on transformation and fate of particles and on water masses as well as mixing. Quantifying these BE is a priority of the international GEOTRACES program (e.g., Grenier et al., 2014).

In the context of the SPICE program, two major multidisciplinary cruises were organized inside the Solomon Sea: the Pandora cruise in July–August 2012, and the MoorSPICE cruise in March 2014. Pandora documented the physics, geochemistry, biology and chemistry of the area (Figure 2) for full cruise track from New Caledonia. MoorSPICE partly repeated the hydrographic measurements, then explored the Bismarck Sea to the North (Figure 2). The main objectives of these two cruises were to document the circulation, mixing, water mass transformations and pathways; to document the chemistry, geochemistry, and the enrichment of waters in the area. In addition, it has been recently shown that the Solomon Sea is a place of large, but essentially undocumented nitrogen fixation (Bonnet et al., 2009): these cruises were an opportunity to better document N₂ fixation. They also allowed the first evaluation of both anthropogenic carbon and ocean acidification in the region.

The objectives of this paper are to present the general context of these cruises and to provide an initial overview for subsequent more detailed analyses by physical oceanographers, geochemists and biologists. The scientific rationale for these cruises and the measurements made are presented. The climatic context is provided, and key results inferred from observations made during these cruises are also summarized.

Recent targeted efforts in the context of the SPICE program have helped to understand aspects of the mean oceanic circulation within the Solomon Sea from a diverse source of observations (vessel-mounted current profiler compilations, gliders, drifters and hydrological data) and modeling studies (Ganachaud et al., 2014). All previous works agree on the following mean circulation pattern: the bulk of thermocline transport into the Solomon Sea enters through the southern opening of the Solomon Sea where the Gulf of Papua Current joins the North Vanuatu Jet (NVJ) in forming the low latitude western boundary current called the New Guinea Coastal Undercurrent (NGCU) (Figure 1); (e.g., Melet et al., 2010; Gasparin et al., 2012; Kessler and Cravatte, 2013). The NGCU travels equatorward, splitting around Woodlark Island, then rejoins before it bifurcates just south of New Britain. The westward branch exits the Solomon Sea through Vitiaz Strait while the eastward branch travels along the coast of New Britain as the New Britain Coastal Undercurrent (NBCU) (Cravatte et al., 2011). The majority of the NBCU exits through Solomon Strait and the remainder passes through St. Georges Channel. In the near surface, the mean circulation is slightly different, with water entering the Solomon Sea through Solomon Strait and exiting southward through the southern entrance (Melet et al., 2010; Hristova and Kessler, 2012). Surface velocities can also be weaker than subsurface, with strong seasonal variations (Cravatte et al., 2011). At deeper levels, the circulation in the Solomon Sea and straits is quasi unknown.

Mean total transport through the Solomon Sea is estimated to be ~30 Sv above 1300 m (Gasparin et al., 2012). The strength of the currents varies on a seasonal timescale with the strongest equatorward flows from June to September during the southeasterly monsoon (Cravatte et al., 2011; Hristova and Kessler, 2012). Currents also vary on El Niño Southern Oscillation (ENSO) timescale in the upper 250 m, with transports increasing during El Niño events and decreasing – or even reversing – during La Nina events (Davis et al., 2012; Melet et al., 2013). Additionally, surface and subsurface mesoscale eddies and submesoscale features dominate instantaneous current fields (Djath et al., 2014; Gourdeau et al., 2014; Hristova et al., 2014).

Water masses, currents, their variation and the partition of their transports between the different straits have therefore been inferred from numerical simulations and scattered measurements mainly confined to the upper 300 m. However, we have not as yet had simultaneous measurements of physics and geochemistry, and our understanding of the internal and deep Solomon Sea has been extremely limited by the lack of measurements.

The Solomon Sea is an interesting site for its specific biology and chemistry, in particular, atmospheric dinitrogen (N₂) fixation. N₂ fixation is the process by which N₂ is reduced to ammonia by specialized microbes termed diazotrophs. This process is the major external source of nitrogen for the ocean, even more important than riverine and atmospheric inputs (Gruber, 2004). N₂ fixation thus provides nitrogenous nutrients for phytoplankton communities and contributes to carbon dioxide (CO₂) uptake and climate regulation (Karl et al., 2002). The South Pacific Ocean has been ranked as the second major basin contributing to global N₂ fixation after the Indian Ocean (Luo et al., 2012), although studies performed in the South Pacific are much less numerous than those performed, for example, in the Indian (Luo et al., 2012), in the North Atlantic (Benavides and Voss, 2015) and the North Pacific Oceans (Karl et al., 1997; Dore et al., 2002).

The tropical Pacific is characterized by a strong zonal gradient in nutrient distribution, with the eastern basin affected by the cold equatorial tongue where nutrient availability is higher but trace metals limit primary production (Aufdenkampe et al., 2002; Le Bouteiller, 2003), and the western basin dominated by a warm pool where metals are available through land runoff, hydrothermalism, and volcanic activity (Labatut et al., 2014). In the western Pacific, the permanent water column stratification inhibits nutrient upwelling (Pena et al., 1994) and N₂ fixation appears as the main source of nitrogenous nutrients for primary production (Yoshikawa et al., 2005; Bonnet et al., 2009). This zonal pattern is also reflected in N₂ fixation activity and in the diversity of diazotrophs, with low N₂ fixation rates in the eastern side of the South Pacific (Raimbault and Garcia, 2008) dominated by heterotrophic diazotrophs (Bonnet et al., 2008; Halm et al., 2012), and potentially high rates provided by autotrophic diazotrophs on the western side (Bonnet et al., 2009; Moisander et al., 2010). However, very few studies have quantified N₂ fixation in the Southwest Pacific, in particular in the Solomon Sea, which presents ideal temperature conditions for diazotrophs to bloom extensively (Bonnet et al., 2017).

Numerous processes that could affect water properties are occurring at the land-ocean interface and within the water column. These include advective, diffusive and turbulent transports; biological uptake; remineralization; river runoff; dissolution of lithogenic minerals and precipitation of authigenic ones (i.e., precipitated locally within the water column); adsorption on particles followed by scavenging or desorption from particles followed by the release of the chemical species. Understanding the sources of these trace elements is fundamental for their integration in any model coupling circulation and biogeochemistry. Chemical elements and isotopes identified that help to constrain the nature and rate of these processes are considered as “oceanic” tracers and part of the GEOTRACES core parameters (GEOTRACES science plan, online at http://www.geotraces.org). The Coral and Solomon Seas are of particular interest to provide indications of these processes because of their energetic dynamics, where boundary exchange is suspected to be of importance (Lacan and Jeandel, 2005; Grenier et al., 2013; Jeandel and Oelkers, 2015).

The surface deposition from volcanic plumes is also a potential source of various species. Papua New Guinea (PNG), the Solomon Islands and Vanuatu form the Melanesian volcanic arc with large emitters of volcanic species and sulfate precursors (Bani et al., 2012; McCormick et al., 2012). Air chemistry of the marine boundary layer and free troposphere of the Solomon Sea is largely impacted by volcanic plumes from PNG volcanoes (i.e., Rabaul and Bagana, located in New Britain and Bougainville Islands, respectively; Figure 1) but also from remote volcanic plumes. The Vanuatu Ambrym volcano is ranked as one of the top emitters, producing between 8% and 17% of global volcanic fluxes of SO₂, HCl, HF, HBr, Hg and ²¹⁰Po (Allard et al., 2015).

Data sets on cloud chemistry and pH in wet deposition are very limited in the region. Nagamoto et al. (1990) provided the first survey of cloud chemistry across the Western Pacific and recorded rainwater pH values in the range of 5 to 6 in the equatorial and remote island region north of Solomon Islands. They suggested a role of the biogenic source (dimethyl sulfide) in rain composition. But over Vanuatu, Vetch and Haefeli (1997) measured a pH of 2 in rainwater 4 km away from Ambrym’s crater (Figure 1). Lefevre et al. (2015) showed that Ambrym is the main source of sulfur species in the Coral and Solomon Seas, with a similar contribution to the regional sulfur budget as dimethyl sulfide (DMS) emitted from marine organisms.

The methylmercury (MeHg) cycle is also unclear; in particular, we do not know whether MeHg is produced primarily in marine sediments and subsequently advected into the open ocean or whether it is produced in situ in the oceanic water column, during the remineralization of sinking organic matter (Fitzgerald et al., 2007; Mason et al., 2012).

The global ocean absorbs about 30% of the annually-emitted anthropogenic carbon slowing the rapid accumulation of CO₂ in the atmosphere that has recently reached 400 ppm (global CO₂ mean in March 2015, Le Quéré et al., 2015). But the ocean carbon uptake leads to a decrease in pH, the so-called “other” CO₂ problem of ocean acidification, which will have a harmful impact on marine organisms and ecosystems. Since pre-industrial times, the mean pH in the surface ocean has dropped by about 0.1 units. In order to follow both the anthropogenic CO₂ penetration in the ocean and changes in acidification, observations are needed to be conducted regularly in open oceans, marginal seas and coastal Zones. Very few observations of the inorganic carbon system have been conducted previously in the Coral and Solomon Seas (Suzuki et al., 2013; Lauvset et al., 2016; Olsen et al., 2016). It is thus interesting to explore the CO₂ system in this area (also referred as southeastern part of the Coral Triangle), including the anthropogenic CO₂ penetration and associated ocean acidification and compare our new observations with previous analysis conducted in open ocean tropical and subtropical zones (e.g., Alvarez et al., 2009). Interestingly, this region, part of the so-called North Australia region in the project called MARCATS (MARgins and CATchments Segmentation), presents large uncertainty in estimates of air-sea CO₂ fluxes whether depending on data-based method (CO₂ source) or ocean biogeochemical models (CO₂ sink) (Bourgeois et al., 2016). A recent study using data-based reconstruction (Lenton et al., 2016) also suggests low pH in this region (8.03 – 8.05) but authors also concluded that at present the Coral Sea is not experiencing marginal conditions with respect to ocean acidification (values of the aragonite saturation state <3.5).

Most data analyzed in this paper originate from the two research cruises: Pandora carried out aboard the French research vessel l’Atalante (also recorded as GEOTRACES cruise GP12), and the MoorSPICE cruise aboard US research vessel Thomas G. Thompson. As part of the Tropical Ocean Climate Study (TOCS) project, a cruise took place aboard the R/V Mirai (MR14–02) in February 2014, around the New Ireland (Figure 2b). This cruise was conducted by the Japan Agency for Marine-Earth Science and Technology (JAMSTEC) in order to explore the oceanic leakage between the Solomon Sea and the western equatorial Pacific through the NICU pathways (Figure 2b). Data acquired during this cruise are not discussed here.

The R/V l’Atalante departed Nouméa, New Caledonia, on 28 June 2012, and returned to port on 6 August (Figure 2a; see Figure S1a for a larger-scale map). Scientific equipment included two SBE-911 + CTD/Rosette systems. The primary standard rosette was used at all stations outfitted with duplicated temperature-conductivity-oxygen SBE sensors, a Chelsea Aqua3 fluorometer, a Wetlab C-star transmissometer and a pair of RDI built Lowered-Acoustic Doppler Current Profilers (L-ADCP) attached to the CTD frame (150 kHz and 300 kHz, down- and up-looking, respectively) to measure currents along the rosette profile. The rosette was fitted with 22 × 12L Niskin bottles to obtain water samples for the standard parameters (see below).

The secondary clean rosette was designed for obtaining uncontaminated samples for trace metals, with a titanium CTD body and epoxy coated frame, and included one set of SBE temperature-conductivity-oxygen sensors, a Wetlab Westar fluorometer and 12L Go-Flo bottles. A set of 7 Maclane or Challenger in situ pumps for sampling of particulate matter was deployed at 5 stations on a hydrography wire. Underway continuous recordings are summarized in Table S1 as well as specific discrete sampling.

The Pandora cruise track included a meridional section along 163°E, from 18°S to 9°S, to sample incoming waters from the SEC (Figure 2a) followed by a section across the southern entrance of the Solomon Sea that closely followed a 2007 cruise track (Gasparin et al., 2012). Three more sections across the Solomon Sea were conducted to follow the paths and transformations of water masses (Figure 2a). Other sections included one on the eastern side of the Solomon Islands and two across Solomon and Vitiaz Straits. Details of the parameters sampled during stations are given in Table S2. Along the cruise track, eleven long duration stations with repeated casts (3–39 h, Table S2, last row) allowed the necessary large-volume water samplings required for geochemical analyses. Moreover, the repeated CTD casts during these stations provided additional information on the high frequency variability of properties and currents. Overall, measurements were made at 83 stations, for a total of 170 casts, including 30 clean sample casts, and 6 on situ pump deployments (Figure 2a).

A total of 9 moorings were deployed over the northern exit channels of Vitiaz and Solomon strait, and in St. Georges Channel from 2012 to March 2014 (MoorSPICE cruise), with two CTD moorings on either side of the southern mouth of the Solomon Sea since 2012. These latter two moorings were combined with bottom pressure inverted echosounders to capture the geostrophic transport between these two endpoint moorings. They will remain deployed (with some intermediate servicing) until summer 2017, and are part of a boundary current transport monitoring system measuring the inflow into the Solomon Sea using gliders (Davis et al., 2012).

The MoorSPICE cruise aboard the R/V Thomas G. Thompson was carried out from 28 February 2014 to 31 March 2014. Scientific equipment included a dual set of temperature-conductivity-oxygen SBE-911+ CTD, a Wetlab ECO fluorometer, a Wetlab C-star transmissometer and a pair of RDI L-ADCPs attached to the CTD frame (300 kHz down- and up-looking). The rosette was fitted with 24 12–L Niskin bottles.

The ship directly sailed from Nouméa to the Solomon Strait and St. Georges Channel to recover the moorings previously deployed during the Pandora cruise; and a number of CTD stations were reoccupied. A saw-tooth pattern across the Bismarck Sea was then performed, in an area with poor in situ data coverage. The cruise continued through Vitiaz Strait, where mooring work was done. Then the track closely followed that of Pandora across the northern section of the Solomon Sea (in a reversed pattern). At the entrance to the Solomon Sea, most of the Pandora stations were reoccupied. Upon reaching Guadalcanal the ship sailed back to Nouméa. Overall, 57 stations were surveyed, some of them repeated, for a total of 78 casts (Figure 2b). Details of underway and on-station sampled parameters are given in Tables S1 to S3 (Section Supplemental Material). There were 6 Argo deployments, 9 moorings that were deployed during Pandora were recovered, and 4 additional moorings were re-deployed.

For both cruises, data processing followed the GO-SHIP guidelines (online at http://www.go-ship.org/HydroMan.html). Temperature and conductivity sensors were calibrated by the manufacturer soon after the cruises. Calibration parameters were applied after checking agreement with salinity bottle samples. Every CTD cast was checked for spikes that might be due to particulate matter in the sensor circuits or electrical problems, and flagged accordingly. Dissolved oxygen sensor data were adjusted by comparison with a Winkler titration determination of water samples (Langdon, 2010; Uchida et al., 2010; Saout-Grit et al., 2015).

Shipboard ADCP velocity data were processed and calibrated using the CODAS software (online at http://currents.soest.hawaii.edu). LADCP velocity data was processed using the GEOMAR/LDEO software v10.8 (Visbeck, 2002). Continuous surface temperature and salinity as well as basic meteorological data were recorded by both ships using the underway thermosalinograph and meteorological systems, respectively.

During the July 2012 cruise, nine moorings were deployed in Vitiaz and Solomon straits, and St. Georges Channel (Figure 2b). Moorings in the straits were recovered and four of them were re-deployed in Vitiaz Strait and Solomon Strait during the March 2014 cruise, and recovered in August 2015. They provided continuous measurements of near full-depth temperature, salinity and velocity. Two subsurface ADCP moorings were also deployed off the eastern coast of New Ireland Island in August 2012 during TOCS MR12–03 cruise; they were recovered in February 2014 during TOCS MR14–02 cruise. They provided continuous current velocity data from near surface to subsurface. Two moorings were also deployed on either side of the southern entrance of the Solomon Sea in 2012 (Figure 2b). These moorings were part of a boundary current heat transport monitoring system also using gliders and pressure inverted echosounders (Davis et al., 2012). These data will be presented in future papers.

Dissolved nutrients (N, P, Si) were sampled and analyzed at high precision following the GO-SHIP guidelines (Hydes et al., 2010) but based on pasteurization (Aminot and Kerouel, 1995; Daniel et al., 2012). Pasteurized samples were analyzed on a SEAL-AA3 automated Segmented Flow Analysis, following (Aminot and Kerouel, 2007; Hydes et al., 2010).

Discrete samples for total dissolved inorganic carbon (DIC) and total alkalinity (TA) were collected at selected stations, poisoned with HgCl₂ after sampling and measured back at LOCEAN laboratory in Paris-France using a potentiometric titration (e.g., Metzl et al., 2010). For calibration, we used certified reference material (CRM, Batch #119) provided by Prof A. Dickson (Scripps Institution of Oceanography, San Diego, USA). The accuracy of DIC and TA are estimated to be ±2.6 μmol kg^–1 and ±2.1 μmol kg^–1, respectively. Using DIC/TA pairs, other inorganic carbon properties were calculated (Pierrot et al., 2011), including the fugacity of CO₂ (fCO₂), carbonate ions, calcium carbonate solubility (for aragonite and calcite), and pH. For all derived calculations we used recommended constants K1 and K2 from Mehrbach et al. (1973) refit by Dickson and Millero (1987; 1989), and pH is seawater scale.

The presence of volcanogenic acidic sulfur was monitored using 30 rainwater samples during Pandora. pH and temperature in rainwater were measured using a rain gauge fixed on the fore deck, which provided a simple and affordable measurement. A glass electrode was used to determine pH values in rain samples immediately after collection (WTW 3110 pH meter). The pH readings were standardized with commercial buffer solutions.

The abundance of different diazotroph phylotypes was assessed through quantitative polymerase chain reaction (qPCR) counts of the nifH gene (encoding for a subunit of the nitrogenase enzyme, responsible for the N₂ fixation activity) as previously described (Church et al., 2005; Moisander et al., 2008). For the purpose of this study, only the main phylotypes are reported (i.e., Trichodesmium and the sum of unicellular diazotrophic cyanobacteria, UCYN). N₂ fixation and primary production were measured at four depths: 5, 50, 100, 150 m during Pandora (Table S2); and 5, 15, 30, 70 m during MoorSPICE (Table S3) using ¹⁵N/¹³C dual isotopic labeling and in situ-simulated 24–h incubation as described in Bonnet et al. (2015); Berthelot et al. (2017).

During Pandora (also GEOTRACES GP12), a comprehensive group of trace element samples and some isotopes (TEIs) were for subsequent measurements. They are listed in the appendix together with their role as geochemical tracers. Al, Mn, Hg speciation and dissolved Fe were available for this study.

Dissolved iron (dFe) samples were collected using a trace metal clean rosette and Go-Flo bottles with a slight N₂ overpressure to force filtration through 0.2 μm filter cartridges (Sartrobran-300, Sartorius). They were acidified to pH 2.0 with ultrapurQR hydrochloric acid (HCl, Merck) and analyzed on board after at least 24 h by flow injection with chemiluminescence detection (Sarthou et al., 2007). Samples for total and dissolved aluminum (Al) and manganese (Mn) were collected into 100 mL LDPE bottles. Total samples were collected directly into the sample bottle, while dissolved samples were collected after their passage through 0.2 μm cellulose acetate Sartobran-300 capsule filters. Samples were acidified to pH < 7 with 12 N Optima HCl at least one week prior to analysis. Al and Mn were determined by flow injection analysis using fluorimetric (Resing and Measures, 1994), and spectrophotometric (Resing and Mottl, 1992) detection, respectively. The manganese chemistry was modified by adding 4 g of nitrilo tri-acetic acid per liter of reaction buffer.

Total Hg (tHg) and methylmercury (MeHg) were measured according to Heimburger et al. (2015). The 9 profiles sampled during the Pandora cruise consisted of 89 unfiltered acidified (HCl 0.4%, v/v) samples.

The Solomon Sea is a region subject to significant annual variability (influenced by the Australian monsoon), interannual variability (linked to El Niño Southern Oscillation cycles), and intraseasonal variability (linked essentially to Madden-Julian Oscillation events, cyclones and synoptic westerly wind events). The Pandora and MoorSPICE cruises sampled two contrasted seasons, associated with significant changes in winds and precipitation. Pandora took place during the dry strong-trade winds season (June—November), in July—August 2012, and MoorSPICE during the wet weak-trade winds season (December—May), in March 2014. At this time of year, the SPCZ is fully developed over the Solomon Sea and the winds weaken or even reverse direction to become northwesterly.

The Pandora cruise occurred following the 2011 La Niña event and coincided with a return to neutral state, also present during the MoorSPICE cruise (Figure 3). The MoorSPICE cruise followed a series of strong equatorial westerly wind bursts and corresponding downwelling eastward Kelvin waves, associated with the onset of the “aborted 2014 El Niño” (Menkes et al., 2014; McPhaden et al., 2015).

In general, El Niño periods in the Coral Sea—Solomon Sea region are characterized by stronger southeasterly wind stress anomalies, with clockwise (upwelling) curl northeast of the Solomon Islands where the wind stress anomalies turn westerly (Kessler and Cravatte, 2013). During Pandora, winds were strongly southeasterly (Figure 4a), with July corresponding to the height of the southeast trade season. The warmest recordings of sea surface temperature (SST) during Pandora were largely confined equatorward and east of the Solomon Islands, with cooler waters within the Solomon Sea and south toward the winter hemisphere (Figure 4a). In contrast, regional winds were extremely weak during MoorSPICE (Figure 4b), in keeping with the expected change in the seasonal cycle, but also the coinciding transition from a cold phase to a warm phase of ENSO. During MoorSPICE, SST was largely above 29°C in most of the Solomon Sea. The warmest waters were found in the tropics east of the Solomon Islands, and likely reflected the eastward shift of the western Pacific warm pool in response to westerly wind bursts earlier in the year associated with the El Niño conditions that ultimately failed to materialize (Menkes et al., 2014; McPhaden et al., 2015). Thus, Solomon Sea SST and wind patterns are tied to the regional scale seasonal cycle as can be visualized in Figure 4.

On intraseasonal time scales, the air-sea coupling of the western Pacific is influenced by the Madden-Julian Oscillation (MJO), a system of large-scale coupled patterns of atmospheric convection and winds that propagate eastward along the equator (Madden and Julian, 1972). An active MJO is characterized by deep convective anomalies that can block the incoming solar radiation and produce anomalously cool SST. The Wheeler and Hendon (2004) index breaks the MJO cycle into eight phases that correspond to the shifting location of the maximum intraseasonal winds and convection (derived from satellite measurements of outgoing longwave radiation (OLR, W m^–2). During Pandora, MJO anomalies based on the Wheeler and Hendon Index suggest an extremely weak active phase of the MJO, while for MoorSPICE, MJO activity was strong in late February, just before the cruise (not shown).

These stable conditions suggest that at least during the time of the cruises, MJO and intraseasonal wind forcing did not likely play a strong role in upper ocean variability or circulation. However, the apparent weak variability during both cruises was fortuitous since four anticyclones migrated south of 10°S during Pandora, nevertheless leaving near stationary conditions over the Solomon Sea. Atmospheric conditions were more unstable during MoorSPICE, promoting the formation of tropical cyclones (TC): weak TC HADI (Category 1, March 9–11 south of 10°S) and severe TC LUSI (Category 3, March 9–16, over the north Vanuatu Islands). A third TC (GILLIAN, March 21–26) formed in the Southwest Pacific too, but just outside the Solomon Sea (http://www.bom.gov.au). In the end, none of these TCs directly impacted oceanic conditions in the Solomon Sea during the cruise time of MoorSPICE.

Similar OLR values of around 220–230 W m^–2 were found during both Pandora (Figure 4c) and MoorSPICE (Figure 4d). Low OLR (cloudier) was mostly confined north of the equator during Pandora, probably indicating the location of the western Pacific Intertropical Convergence Zone. During MoorSPICE, the minimum in OLR to the east of the Solomon Sea indicates the equatorward axis of the South Pacific Convergence Zone (SPCZ), which is more active during the Southern Hemisphere summer.

The July–August 2012 cruise conditions were therefore mainly representative of the strong-trade wind season, while the March 2014 cruise conditions were representative of the opposite weak-trade wind season. Interannual and intraseasonal variability are not expected to be of major influence.

The sea surface conditions during both cruises were assessed using the along-track direct measurements of in situ winds, sea surface salinity (SSS) and SST (Table S1; Figures 5 and S3 for complete cruise track). During Pandora, we observed strong (~20 m s^–1) and unidirectional winds oriented equatorward along the prevalent trade wind axis (Figure 5a). During MoorSPICE winds were much weaker (~5 m s^–1) along an identical axis and were poleward to the north of 6°S and equatorward to the south of this latitude (Figure 5b). In situ winds (Figure 5) agree well with the monthly mean satellite winds (Figure 4) for both cruises.

Sea surface waters in the Solomon Sea appeared to be saltier and colder during Pandora compared to MoorSPICE (Figure 5) except northeast of Bougainville Island. The SSS was dominated by small scale gradients during Pandora, while during MoorSPICE salinity features were much more cohesive on a large scale, especially within the northern exit straits of the Solomon Sea. The SST exhibited a stronger zonal cross-basin gradient during Pandora compared to MoorSPICE, with lower temperature on the western part of the Solomon Sea (Figure 5c, d). Similar patterns were found in the monthly mean satellite SST for both cruises (Figure 4a, b), consistent in SST and SSS seasonal variability described in Delcroix et al. (2014). These colder and saltier waters are due to northwestward oceanic advection (Figures 5, S3).

Results of the pH survey are displayed in Figure 6a. The atmospheric transport and dispersion of acidic species released from Rabaul, Bagana and Ambrym volcanoes were forecast onboard using the atmospheric circulation model WRF-Flexpart (Mari et al., 2014; Lefevre et al., 2015) (Figure 6b). Cloud waters in the pristine marine boundary layer are characterized by pH values ranging from 6 to 7, but large pH decreases of 1 to 2 units were recorded in cloud waters contaminated by the volcanic plumes. A very acidic rainfall (pH = 4.1) was found >500 km downwind of Ambrym volcano (3 July 2012). The volcanic source of acidifiers was confirmed by air trajectory results (Figure 6b). Six days later (9 July 2012), the volcanic plume had crossed over the Louisiades archipelago (Figure 1), more than 1000 km away from Ambrym volcano acid rain with a collected pH of 5.4.

The presence of volcanogenic acidifiers is explained by the long range transport of sulfate precursors from Ambrym discussed in Lefevre et al. (2015): from July 5–15. A strong anticyclone (1030 hPa) in the northern Tasman Sea (http://www.bom.gov.au) drove a rapid and steady southeasterly trade wind flux, promoting the transport of volcanogenic acid species from Ambrym to the sampling site within a couple of days.

It is worth noting that close to Rabaul volcano, cloud chemistry is also influenced by local magma degassing (Figure 6b). Finally, the pH distribution (Figure 6a) is in good agreement with the mean distribution of surface sulfate concentrations and total sulfate (dry + wet) flux deposition displayed in Lefevre et al. (2015, their Figure 13a, b).

The observed currents were generally strong with high energy at mesoscale (~100 km) and submesoscale (~10 km) so that snapshots from each cruise contain variability on many time scales. Currents derived from satellite altimetry and drifters suggest that the mean surface circulation during Pandora was dominated by a westward flow in most of the Solomon Sea (Figure 7a). The shipboard ADCP shows a strong and narrow NGCU entering the Solomon Sea (Figure 7c). The NGCU core is located between 500 and 1000 m depth with a surface reversal above 100 m (Germineaud et al., 2016, Figure S4a), as also observed by repeated glider transects (Davis et al., 2012).

As the NGCU splits around Woodlark Island, a strong, shallower stream traveled to the west through the Milne Bay, while the other stream was eastward, and stronger below 100 m (153.5°E, 9°S; Figure 7c). This asymmetry is likely due to the sill depth of the passage between Woodlark Island and Trobriand (151.5°E, 8.5°S) which is about 350 m (Figure S4a). A moderate (~0.5 m s^–1) surface-enhanced flow was observed entering the Solomon Sea through the very narrow Indispensable Strait (Figure 7c). The portion of the NGCU that exited the Solomon Sea through Vitiaz Strait was strong (>1 m s^–1) over the full depth (Germineaud et al., 2016).

While the bifurcation of the NGCU south of New Britain was not sampled, the resulting NBCU was captured at 152°E with flow along the coast to the northeast at all depths and a portion exiting through St. Georges Channel with the current’s core between 100 and 400 m depth (Germineaud et al., 2016, Figure 7c). The circulation at Solomon Strait was stronger near the coast of New Ireland and northward below 100 m as the southward flow was confined to the upper 100 m and to the western half of the Solomon Strait (Figure 7c).

East of Solomon Islands and south of 6°S (Figure 7a), there is a branch of the westward SEC with an eastward current to the north (2°–6°S). Cyclonic circulations occur South of Papua New Guinea (11°S, 148°E) and northeast of Bougainville Island (5°S, 156°E) (Figure 7a). This last feature is related to a strong northwestward surface flow along the east coast of Bougainville Island (155°E, 5°S; Figure 7c).

During MoorSPICE the surface circulation was dominated by cyclonic eddies at the southern entrance of the Solomon Sea (Figure 7b). The current was southward through Solomon Strait, contributing to a large anti-cyclonic eddy inside the Solomon Sea (8°S, 153°E). East of Solomon Islands, the flow was eastward at 8°S, that is, against the time-mean flow in the SEC with large eddies centered at 6°S and 10°S.

Compared to during the Pandora cruise, the NGCU was shallower and wider during MoorSPICE, with no surface reversal as it entered the Solomon Sea (Germineaud et al., 2016, Figures 7d, S4b). No NGCU bifurcation around Woodlark Island was observed during MoorSPICE, with eastward flow east of Milne Bay, that is, opposite to the direction found during Pandora.

This contrast was also observed in drifter climatology during this season (Hristova and Kessler, 2012). However the merging of a very strong and deep NGCU was observed north of the Woodlark Island chain. The outflow through Vitiaz Strait was unexpectedly strong (>1 m s^–1), and so not consistent with the expected seasonal decrease. Nevertheless, the flow decreases with depth, and the total transport through Vitiaz Strait was much lower (Germineaud et al., 2016). The NBCU had a similar speed and direction to that found during Pandora. Surface flow (0–100 m) in both St. Georges Channel and Solomon Strait was entirely southward into the Solomon Sea, then to the southwest along the coast of New Britain.

In addition, strong variations in spatial mixing patterns were also detected between both cruises. Observations from the Pandora and MoorSPICE cruises and the Argo database were used to estimate the dissipation of kinetic energy and mixing in the Solomon Sea using the Thorpe scale method and finescale parametrization (Alberty et al., 2017). The mean profile of dissipation in the Solomon Sea provided in that work shows that the maximum is near the surface and decreases with depth. Depth-mean dissipation in the Solomon Sea was elevated by a factor of 8 relative to the rest of the equatorial Pacific and spatial patterns of mixing vary by up to two orders of magnitude along isopycnal layers. The results suggest seasonal variability of mixing with enhancement during strong monsoonal winds, though the full seasonal cycle remains unresolved.

The two cruises therefore show contrasted circulations, which we attribute to the expected seasonal variability as both the intraseasonal and interannual contrasts are low between the two cruises (Section on Climatic Context). Pandora was characterized by strong and well defined currents that are typical of austral winter (e.g., Cravatte et al., 2011; Hristova and Kessler, 2012), while MoorSPICE was characterized by weaker currents and a greater activity of cyclonic eddies, typical of early austral summer (Gourdeau et al., 2014; Hristova et al., 2014).

Germineaud et al. (2016) provide a detailed analysis of thermocline and intermediate water properties and we summarize here the main features. Near the surface, water properties have an essentially homogeneous salinity during Pandora (34.4–35.4; Figure 8a) and a large range during MoorSPICE (33.2–35.6; Figure 8b).

Below the surface, water masses are identified by marked temperature and salinity extremes. During Pandora, salinities reach a maximum around 200 m depth (>35.8) at σ_θ = 24.5 (Figure 8a) characterizing the South Pacific Tropical Water (SPTW, Wyrtki, 1962), also referred as the Upper Salinity Maximum with south/east Pacific origins (Donguy 1994; Qu and Lindstrom, 2002). This upper salinity maximum was found at σ_θ ~ 25 during MoorSPICE with salinities (<36) lower than during Pandora (Figure 8b).

Below the SPTW is the Subtropical Mode Water (STMW; σ_θ = 25 – 26.5 kg m^–3, 12°–20°C) coming from the northern part of the Tasman Sea (Donguy and Henin, 1977; Holbrook and Maharaj, 2008). Deeper, waters in the density range close to σ_θ = 27.0 kg m^–3 (5°–8°C, Figure 8a) indicate the Subantarctic Mode Water (SAMW; McCartney, 1977) formed at the northern edge of the Antarctic Circumpolar Current south of Tasmania and then advected northward into the South Pacific (e.g., Grenier et al., 2013; Kessler and Cravatte, 2013).

Near 800 m depth at σ_θ = 27.2 kg m^–3, a salinity minimum (<34.6) associated with a maximum in oxygen content (~150–200 μmol kg^–1; Figure 8a) identifies the Antarctic Intermediate Water (AAIW) that originates in the southeast Pacific (Qu and Lindstrom, 2004). AAIW is also associated with marked increases in silicate (Figure 8c, d) and nitrate (Figure 8e, f) observed below σ_θ ~ 27. Northeast of Bougainville Island (~5°S–155°E; Figure S9), the highest salinities of the intermediate waters indicate the presence of the South Equatorial Pacific Intermediate water (SEqPIW that emanates from the equatorial region, Bostock et al., 2010; Germineaud et al., 2016).

An increase in salinity at σ_θ = 27.4 (>1000 m) associated with high nutrient content characterizes the Upper Circumpolar Deep Water (UCDW), which is consistent with Sokolov and Rintoul (2000, their Figure 12b). We observed quite low oxygen content associated with high concentrations in nitrate (>30 μmol kg^–1) and silicate (>100 μmol kg^–1), with high salinity (>34.5) during both cruises. The UCDW emanates southeast of New Zealand, before reaching the anticyclonic gyre in the South Pacific (Tsimplis et al., 1998; Kawabe and Fujio, 2010).

Below, the major deep and bottom waters are characterized by the Lower Circumpolar Deep Water (LCDW), which corresponds to an increase in salinity and a decrease in silicate near 2000 m in the Southern Ocean (Orsi et al., 1999). This deep water flows northward across the Southwest Pacific via complex pathways (Kawabe and Fujio, 2010, their Figure 2c), although its presence within the Coral and Solomon Seas remains poorly documented.

Bonnet et al. (2015) and Berthelot et al. (2017) have provided a detailed quantification of N₂ fixation and diazotrophs in the Solomon Sea; here we summarize here the main results and trends. In general, Trichodesmium dominated the diazotroph community in the Solomon Sea with up to 8.2 × 10¹⁰nifH copies m^–2 during Pandora and up to 4.8 × 10⁹nifH copies m^–2 during MoorSPICE (Figure 9a) (Bonnet et al., 2015). These abundances are in the upper range of nifH-derived Trichodesmium abundances in the world ocean (Luo et al., 2012). UCYN were the second most abundant phylotypes during Pandora (average of 4.0 × 10⁹nifH copies m^–2; Figure 9). Maxima were observed at the centre of the Solomon Sea (8.0 × 10⁹nifH copies m^–2) during MoorSPICE, with much lower maximum abundances in the Bismarck Sea (<1.3 × 10⁸nifH copies m^–2) (Figure 9b). Trichodesmium dominates over UCYN phylotypes in the Solomon Sea in contrast with other South Pacific areas such as the Coral Sea. This dominance seems to be associated with the meridional temperature gradient between the Solomon and the Coral Sea, as well as the higher trace metal availability in the Solomon Sea (Section on geochemistry, Moisander et al., 2010; Bonnet et al., 2015).

In the Solomon Sea, photic N₂ fixation rates ranged from 30 to 5449 (average 624) μmol N m^–2 d^–1 during Pandora and 0.2 to 4364 (average 1396) μmol N m^–2 d^–1 during MoorSPICE (Figure 9c). They were significantly higher during austral summer conditions (MoorSPICE) than during austral winter conditions (Pandora), suggesting a possible seasonal effect. Nonetheless, rates during both seasons were in the upper range of those reported in the global N₂ fixation MAREDAT database and even surpassed its upper rates (100–1000 μmol N m^–2 d^–1). This comparison suggests that the Solomon Sea is a hot spot of N₂ fixation in the global ocean, although the reasons behind these high rates need to be further investigated. We hypothesize that having SST values >27°C all year round (i.e., optimal for diazotrophic growth and nitrogenase activity, Breitbarth et al., 2007) associated with significant micro- and macronutrient inputs provide ideal conditions for N₂ fixation. In particular, phosphate turnover times have been reported to be >2 days (i.e., non-limiting for phytoplankton communities) further south around Melanesian archipelagoes (Moutin et al., 2005), which is likely the case in the Solomon Sea as surface phosphate concentrations were never exhausted during both cruises (always >0.08 μmol L^–1).

In addition, iron, an important component of the nitrogenase enzyme which catalyzes N₂ fixation (Raven, 1988), is continuously provided to the Solomon Sea through land runoff, hydrothermal, and volcanic activity (see section Dissolved Iron below and Labatut et al., 2014) and possibly through energetic currents interacting with the islands (Grenier et al., 2013; 2014; Djath et al., 2014). The dissolved Fe measurements performed along a west–east zonal section in the equatorial Pacific (Slemons et al., 2010) confirm that the Solomon Sea is a primary source of iron for the equatorial Pacific.

Altogether, these conditions provide optimal environmental conditions for diazotrophs to bloom extensively and likely explain why this regional sea is a hot spot of N₂ fixation. The basin-scale South Pacific N₂ fixation rate was estimated at 46 Tg N y^–1 (Luo et al., 2012), equivalent to a daily water column integrated rate of 125 μmol N m^–2 d^–1 (considering an area of 72 × 10⁶ km² for the South Pacific). The N₂ fixation rates measured during the Pandora and MoorSPICE cruises are larger by a factor of 7 (529 μmol N m^–2 d^–1 on average and up to >5000 μmol N m^–2 d^–1; Figure 9c), confirming that the Solomon Sea contributes importantly to fixed nitrogen inputs in the South Pacific as a whole.

In an attempt to assess the importance of N₂ fixation in this region, we extended the depth-integrated N₂ fixation rates obtained during both cruises to the Solomon Sea surface (7.2 × 1011 m²). Assuming that diazotrophs fix on average 600 μmol N m^–2 d^–1 during the austral winter conditions (Pandora) and 1300 μmol N m^–2 d^–1 during austral summer conditions (MoorSPICE), N₂ fixation would introduce 3.5 Tg N y^–1 to the Solomon Sea. This amount corresponds to ~10% of the total nitrogen fixed for the entire South Pacific Ocean as reported in Luo et al. (2012), while the Solomon Sea represents only 1% of its total area. Consequently, the N₂ fixation budget for the South Pacific needs to be updated by including these recent studies performed in the Solomon Sea.

The nitrogen demand derived from primary production data (Figure 9) (assuming a C:N ratio of 6.6) indicates that N₂ fixation fueled on average 5.5 and 15.1% of total primary production in the Solomon Sea during the Pandora and MoorSPICE cruises, respectively. This contribution is remarkably high compared to other oceanic regions such as the Northwestern Pacific (Shiozaki et al., 2013), Southeastern Pacific (Raimbault and Garcia, 2008), Northeast Atlantic (Benavides et al., 2013) or the Mediterranean Sea (Bonnet et al., 2011; Ridame et al., 2014), where the contribution of N₂ fixation to primary production is generally <5%. These higher contributions confirm the importance of N₂ fixation in supplying new nitrogen to the surface waters of the Solomon Sea.

By using high-resolution nanometer scale secondary ion mass spectrometry (nanoSIMS) coupled with ¹⁵N₂ isotopic labelling and flow cytometry cell sorting, Bonnet et al. (2016) showed during MoorSPICE that 10% of diazotroph-derived N was quickly (48 h) transferred to non-diazotrophic plankton, in particular diatoms, whose abundance increased 3-fold in that time frame. Diatoms are efficient carbon exporters to depth (Nelson et al., 1995), suggesting that N₂ fixation may sustain an indirect export pathway of organic matter in the Solomon Sea. It would be interesting in further studies to quantify the part of exported material sustained by diazotrophy in such a hot spot of N₂ fixation.

Although during Pandora samples were taken at contrasted locations, depth profiles for both TA and DIC were relatively depth-independent below 2000 m, as compared to surface gradients (Figure 10a, b). Below the surface, a pronounced TA subsurface maximum around 150 m was clearly recognized in the SPTW (Figure 8). In intermediate waters (500–1000 m), TA and DIC regularly increased with depth.

However, at station 44 (4°S–155°E, i.e. outside the Solomon Sea in the western tropical Pacific) we observed localized DIC maxima around 500–700 m (highlighted in Figure 10b) in the core of the low oxygen layer characterizing old SEqPIW of northern origins (Section on water properties). In deep waters (>1000 m) DIC and TA concentrations presented, like nutrients, a maximum around 2000–3000 m. However, lower DIC and TA concentrations (highlighted in Figure 10a, b) were found around 3000 m at station 29 (11°S, 156°E) associated with lower nutrient concentrations, high salinity and high oxygen (11.5°S, 155.7°E).

In surface waters, the range for TA (2270–2290 μmol kg^–1), for DIC (1920–1955 μmol kg^–1), for pH (8.06–8.09) and for Aragonite saturation state (3.8–4) are typical values observed in the eastern Indian and western Pacific tropical zones (Takahashi et al., 2014; Lenton et al., 2016). Sea surface fCO₂ (calculated) was around 350 μatm at 14°S–162°E and higher at 370 μatm around 4°S–152°E, a difference that could be explained by warming in the north. The fCO₂ range derived from surface DIC/TA data is coherent with underway fCO₂ measurements conducted in this region during July–August 2012 (Trans–Future 5 cruise reported on Figure 11, available on SOCAT-V3 data-base, online at www.socat.info, Bakker et al., 2016).

In July 2012 the global mean atmospheric concentration was about 392 ppm (or 378 μatm when converting to fCO₂) (Le Quéré et al., 2015) Therefore, during the cruise the region was a CO₂ sink (oceanic fCO₂ was lower than in the air). To understand how the oceanic carbon uptake propagates in the water column, we evaluated the anthropogenic CO₂ concentrations (C_ant) using the method based on the TrOCA tracer (a combination of TA/DIC pairs, temperature, salinity and oxygen data) previously applied in the Indian Ocean along several WOCE sections (Touratier et al., 2007; Alvarez et al., 2009). The new TrOCA method proposed by Touratier et al. (2007) was considered the most suitable for the investigated region because no transient tracers (e.g., CFCs) were measured during the Pandora cruise. The C_ant concentrations are presented only below 200 m (Figure 10b) as the method is not suitable to evaluate C_ant in surface layers.

The C_ant vertical gradient suggests a significant penetration of anthropogenic CO₂ down to about 1500 m. Between 200 and 1000 m C_ant concentrations decreased sharply from about 50 μmol kg^–1 to 10 μmol kg^–1. Below 1500 m, the C_ant concentrations are low and within the error of the method (±7μmol kg^–1).

We also estimated the anthropogenic pH_ant by calculating the difference between modern pH, (based on DIC and TA concentrations in 2012) and pre-industrial pH values (based on DIC-C_ant and TA concentrations). As expected, the pH_ant profiles mirror the C_ant penetration (Figure 10b). A large pH_ant value of about –0.1 units is estimated in upper layers (200 m) and is also significant at depth (–0.05 units at 500 m).

The DIC and TA observations obtained during the Pandora cruise enable the first evaluation of both anthropogenic carbon and ocean acidification in the Solomon Sea. Added to historical observations conducted in the Western Tropical Pacific ocean (Suzuki et al., 2013), these data can be used to investigate in more detail the oceanic inorganic carbon system and its variability in a broader context.

We present below an initial examination of the concentration of Fe, Al, Hg and Mn in the Solomon Sea.

Depth profiles of dissolved Fe (dFe) resemble nutrient-like profiles, with low concentrations at the surface and an increase with depth (Figure 12a). The surface dFe concentrations ranged from 0.11 to 0.44 nmol L^–1 (Figure 12b) and showed significantly (p < 0.05) higher dFe concentrations near the islands (0.35 ± 0.06 nmol L^–1) than in the Coral Sea (0.23 ± 0.07 nmol L^–1) in the first 150 m. Similar concentrations have been observed by Labatut et al. (2014) off the Papua New Guinea coast due to river runoff. Volcanic inputs from the Rabaul volcano (close to station 42) could also be involved. This enrichment is confined near islands and offshore, dFe concentrations (0.19 ± 0.02 nmol L^–1) are not significantly (p > 0.05) different from those in the Coral Sea. High primary production and Trichodesmium abundance could explain the similar dFe stocks (Bonnet et al., 2015). The deep water dFe concentrations ranged from 0.62 to 1.29 nmol L^–1 (Figure 12c) and suggest a sedimentary and/or hydrothermal source of dFe in the Solomon Strait (Labatut et al., 2014).

Enrichments in Al and Mn concentrations in the region relative to the open ocean indicate that coastal and sedimentary input is important in this region and that this boundary exchange is likely driven by strong currents scouring the coastal margin. However, quantification of boundary exchange requires the coupling of concentrations of rare earth elements (REE) and measurements of Nd isotopes (Lacan and Jeandel 2005; Grenier et al., 2013), currently underway The evolution of Al concentration along the NGCU pathways near the shallow shelf (<500 m) of the Rossel Island (station 34) suggests a noticeable source at about 400 m depth (Figure 13), at the core of the NGCU (Germineaud et al., 2016). There was a strong Mn gradient in the upper water column, also suggestive of sediment flux from the shelf (Figure 13). These stations also show evidence of enrichment at depth from deep sedimentary input.

The NGCU flows near the shallow sill along the Woodlark and Trobriand Islands (station 71) before it bifurcates into the NBCU. Here, there is an increase in aluminum and manganese at depth, as well as a peak in aluminum (but not manganese) around 500 m. This pattern could either be the signal from the Louisiades, or a lateral input from the nearby shallow sill. The absence of Mn enrichment at this depth suggests that the aluminum is from a more distant source like the Louisiades and that the Mn has been scavenged during transport. There is also deep water enrichment off of New Britain (station 60). In the Solomon Strait (station 42), the enrichment from the Louisiades is overwhelmed by the strong continental inputs through the strait.

Based on the Al profile, as the NBCU flows through the Solomon Strait it entrains both deep sedimentary inputs and lateral input from coastal margins. Continuing along the NGCU through the Vitiaz Strait (station 77), there are lateral inputs from the shelves, as well as enrichment at depth. Sites located adjacent to each other in the Solomon Strait (station 42) and northeast of Bougainville (station 43) are remarkably different. The Solomon Strait is enriched in aluminum and manganese relative to the Bougainville coast, suggesting that as water exits the Solomon Strait it flows more northwards than eastwards (Figure 7). This interpretation is consistent with ADCP findings, which show no flow between these neighboring sites. Rather, the Solomon Islands Coastal Undercurrent flows north, and there is some concurrence between this station and the south side of the Solomon Islands (station 13): they have similar Mn distributions, and similar Al distributions deep in the water column.

In the Coral Sea, upstream of the Solomon Sea, station 4 located within the North Caledonian Jet (NCJ) showed similar distributions of aluminum and manganese to those of station 82, further north. Lateral inputs from the shelf off New Caledonia enrich station 4 in both aluminum and manganese, relative to station 82. Based on these data, the predominant inputs of trace metals to the Solomon Sea are from coastal margin fluxes at depth, as well as lateral inputs in straits, where water is in contact with sediment at both the seafloor and along the coasts. There are high surface concentrations at many stations that reach thermocline depths (~300 m). In general, average aluminum and manganese concentrations (~12 nM and ~0.5 nM, respectively) are elevated relative to open ocean values. The most elevated concentrations in aluminum are found at station 77, along the outflow of the Vitiaz Strait, suggesting high terrestrial inputs in the surface waters.

The Solomon Sea is, in many locations, enriched in Al and Mn relative to open ocean values, suggesting that the interaction between strong currents and coastal boundaries is an important mechanism by which Al and Mn concentrations are enriched in this area. Such hypothesis will be confirmed, or not, when REE and Nd will be analyzed.

We present 9 vertical profiles (to 1300 m–depth) for tHg and MeHg that were sampled during the Pandora cruise. Total Hg concentrations varied only little, both vertically and horizontally (mean = 0.90 ± 0.18 pM, range = 0.55 – 1.69 pM, n = 87). The vertical tHg profiles showed neither a clear surface enrichment nor indications for sedimentary sources (Figure 14), and correspond to concentrations generally observed in the open ocean (Lamborg et al., 2014). The MeHg vertical distributions showed a similar pattern at all stations (mean = 0.17 ± 0.14 pM, range = 0.01 – 0.42 pM, n = 89).

MeHg concentrations were generally low (0.043 ± 0.021 pM, n = 39) at the surface (0–200 m), increase gradually from 200 to 400 m-depth (0.195 ± 0.103 pM, n = 21) to reach their maximum below 400 m-depth, (0.337 ± 0.034 pM, n = 29). The fact that MeHg and tHg distributions are relatively homogeneous in this dynamic zone and comparable to what has been observed eastward in the open Equatorial Pacific (Mason and Fitzgerald, 1990; Munson et al., 2015) suggests that MeHg is formed in situ in the oceanic water column and that boundary exchange has little or no influence. Concurrent near-margin Hg scavenging (Cossa et al., 2004) and Hg release from margin sediments could indeed result in zero net effect on tHg concentrations. It is very unlikely, however, that those boundary exchange processes would not affect MeHg distributions.

The Pandora and MoorSPICE multidisciplinary cruises documented the Solomon Sea and part of the Coral and Bismarck Seas during two contrasted seasons, in austral winter (July–August 2012) and austral summer (March 2014). This paper presents the climatic context of the cruises, explains the scientific rationales behind all measurements made and describes the data acquisition. It presents an ensemble view of available observations and some initial results on physics, biogeochemistry and geochemistry so that those can be used as a reference for further, refined physical, geochemical and biogeochemical studies and stimulate multidisciplinary analyses.

Both cruises took place during neutral or weak ENSO conditions, and during weak active phases of MJO, precluding a strong influence of these phenomena on observed conditions. It is more likely that seasonal variability is the main driver of observed differences between the two cruises. Pandora took place during austral winter, and was associated with colder sea surface temperature, strong regional monsoonal wind forcing, and well-defined currents, whose transports were in the upper range of what had been previously estimated. In contrast, MoorSPICE took place in austral summer, and was associated with warm SSTs, much weaker winds, weaker currents and a greater activity of cyclonic eddies.

Measurements of temperature, salinity, oxygen, nutrients and currents allowed an unprecedented description of the fine-scale structure of water mass pathways, properties and mixing at different densities. Special attention was paid to the entrance and exits of the Solomon Sea, and these sections were sampled from top to bottom. Waters enter the Solomon Sea through the southern entrance in the narrow and deep NGCU confined to the western boundary, joining a weaker branch in mid-basin. This strong current further splits around islands and bifurcates on the New Britain coast, ultimately exiting the Solomon Sea through three narrow exits. The main branch exits northwestward through Vitiaz Strait, where the flow is found to extend to the bottom during both cruises. Other branches exit through St. Georges Channel and Solomon Strait. In the surface, waters also enter the Sea through Solomon Strait. Unexpected features include the deep extension of the currents, and strong contrasts between the interior of the Solomon Sea and conditions east of the Solomon Strait (Germineaud et al., 2016).

These energetic currents and the complex topography make this area potentially key for water mixing. Spatial patterns of mixing have been analyzed using CTD and LADCP profiles and reveal high seasonal dependence dissipation, which is much higher in the Solomon Sea than in other equatorial Pacific regions.

Biogeochemical measurements revealed new and important results. The Solomon Sea appears as a hot spot of N₂ fixation in the global ocean. Trichodesmium was quite abundant during both cruises and dominated the diazotroph community. N₂ fixation rates measured were among the highest of the global ocean, especially in the narrow exits of the Solomon Sea where diazotrophs accumulated. The Pandora cruise carbon measurements also enabled the first evaluation of anthropogenic carbon and oceanic acidification in the Solomon Sea. Anthropogenic carbon signal penetrates deep in the water column, down to 1500 m, and acidification, large in the surface layer, was still found to be significant at 500 m depth.

Trace elements and isotopes were collected at selected locations along the water pathways; though many tracers are still currently being analyzed, the first results clearly reveal land to ocean inputs. A surface dissolved iron enrichment, probably due to river runoff or volcanic inputs, is confined close to the coasts (Labatut et al., 2014). Dissolved iron concentrations in deep waters suggest the existence of a deep sedimentary and/or hydrothermal source. Aluminum and manganese concentrations also suggest significant sediment flux from the shelf, in shallow and deep waters, and lateral inputs in the straits. Surprisingly, total mercury and methylmercury vertical distributions were similar to those in open ocean, and did not reveal any substantial surface enrichment, nor sedimentary sources. These results suggest that boundary exchange has little or no influence on MeHg formation. Further analyses of other tracers (more specifically REE concentrations and Nd isotopes) will help to constrain the nature and rate of the boundary exchanges at the land-ocean interface. The question of how these micronutrient elements will be transported downstream toward the equator remains open.

The data collected during these two cruises provide snapshots of the Solomon Sea conditions. They give unprecedented detailed information, but have inherent limitations, since a snapshot view of the currents can be strongly impacted by variability on many timescales, in particular by mesoscale eddies, intraseasonal or interannual variability. To complement these cruises, moorings have been deployed at the entrance and exits of the Solomon Sea (in Vitiaz Strait, St. Georges Channel and Solomon Strait) for a period of 18 months to 3 years. They sampled velocity, temperature and salinity at high frequency from top to bottom. Together with the repeated glider sections at the entrance of the Solomon Sea, with the PIES endpoint moorings, they will provide estimates of mass and heat transport variability from the tropics to the equator via the low-latitude western boundary current.

We hope to learn much from these observations about the scales of variability of this transport, which is a key element of the meridional subtropical cells supplying waters of mid to high latitude origin into the western equatorial Pacific. It is recognized that sustained boundary current observations are primary missing elements of the ocean observing system, and that their transport of mass, heat and freshwater should be monitored through dedicated observations (TPOS2020, report available online at http://tpos2020.org/first-report/). We also hope to learn from these measurements how to effectively built a sustained observing system for this region.

During Pandora (also GEOTRACES GP12), a comprehensive group of trace element samples were collected and are currently being measured. They are listed below, along with their scientific application:

1. Dissolved and particulate trace metal concentrations (Fe, Co, Mn, Cu, Zn, Pb, Cd) and some of their isotope distributions (Fe, Zn, Cu, Pb, Cd) will be used to assess the flux of trace elements provided to the water masses by their contact with the adjacent continent to the Solomon Sea. In addition, some of them as for example Fe, Co, Cu, or Zn are essential to the phytoplankton development and considered as micro-nutrients.

2. Concentrations and isotopic composition of these micronutrients help to constrain their origin, fate and cycling in seawater. For example, hydrothermal sources can provide iron that is isotopically light compared to the simple dissolution of erosion products, allowing the differentiation of the element sources (Radic et al., 2011; Conway and John, 2014; Labatut et al., 2014; Klar et al., 2017). Lead isotopes are good tracers of the lead sources in the different water masses.

3. Being conservative lithogenic tracers, particulate Al and ²³²Th allow discrimination of the terrestrial from the biogenic fraction of the suspended material pool.

4. Radium and actinium isotopes (²²⁶Ra, T_1/2 = 1602 years; ²²⁸Ra, T_1/2 = 5.75 years; ²²⁴Ra, T_1/2 = 3.66 day; ²²³Ra, T_1/2 = 11.4 day) are all decay products of insoluble elements. Soluble, they often show high activity in shallow water overlying continental shelves via epibenthic outflux, and decrease in abundance by natural radioactive decay and mixing. Because of their chronometer properties, Ra isotopes provide information on water transit time and mixing processes (van Beek et al., 2008; Sanial et al., 2015).²²⁷Ac (half-life of 21.8 years) is produced in the bottom sediments. ²²⁷Ac will be analyzed on the same samples as for Ra and will help in constraining the vertical mixing on a longer time-scale (Geibert et al., 2002).

5. Rare Earth (REE) concentrations and Nd isotopes: The continental margins surrounding the world ocean display different isotopic signatures, reflecting their different geological origin. Old granitic sediments display low ¹⁴³Nd/¹⁴⁴Nd ratios while the volcanic sediment have high ¹⁴³Nd/¹⁴⁴Nd ratios. This property allows the quantification of the fraction of weathered material that dissolves at the continent/ocean interface (Arsouze et al., 2009; Jeandel and Oelkers, 2015). In addition, the distributions of Nd isotopic signatures and REE concentrations between the dissolved and particulate fractions of the collected samples are complementary tracers that allow quantifying exchange processes between these two phases.

6. Thorium and protactinium isotopes: Thorium is insoluble in seawater. ²²⁸Th and ²³⁰Th are used as tracers of particle scavenging rates, settling rates and dissolved/particulate exchange (Henderson, 2003; Venchiarutti et al., 2008; Hayes et al., 2013; 2014). ²³¹Pa is also insoluble (but slightly more soluble than Th) and also produced in the water column by the decay of a soluble parent. Although more soluble than Th, ²³¹Pa is still particle reactive. Coupled with ²³⁰Th isotope data, its distribution provides a powerful means to estimate dissolved/particle exchange rate, boundary scavenging and ventilation rates (Henderson, 2003; Venchiarutti et al., 2008; Hayes et al., 2013; 2014).

7. Stable Li and Pb isotopes: hydrothermal fluids have a characteristic stable Pb isotope signatures (as do Nd) that is altered near a vent. Li concentrations and Li isotopic signatures are also good specific tracers of hydrothermalism because seawater has an isotopic signature of about δ 7Li = +33‰ whereas hydrothermal fluids are close to δ 7 Li = +5–10‰.

8. Total Hg (tHg) and methylmercury (MeHg) are contaminants, the last one being accumulated and toxic for the trophic chain (Heimburger et al., 2015).

1. Pandora data are freely available and can be accessed from SISMER and CYBER web server at http://www.obs-vlfr.fr/proof/php/pandora/pandora_overall.php.

2. MoorSPICE data are freely available and can be accessed from SISMER and CYBER web server at http://www.obs-vlfr.fr/proof/php/MoorSPICE/moorspice_overall.php.

3. R/V Mirai MR14–02 are freely available and can be accessed from the JAMSTEC data website http://www.godac.jamstec.go.jp/darwin/datatree/e.

This work is a contribution to the CLIVAR/SPICE and GEOTRACES International programmes. We are grateful to the ship crews (RV l’Atalante, RV Thompson and RV Mirai) who made possible the various in situ measurements, thereby requiring skills and care. The collaboration with SOPAC/SPC, PI-GOOS, and University of Papua New Guinea was greatly appreciated. We thank the technical division of INSU (DT-INSU, O. Desprez de Gesincourt, L. Fichen, F. Perault, E. de Saint-Leger, C. Brachet and L. Souarnec); US-IMAGO (D. Lopez, L. Jamet), IRD Center Noumea (J. Carso, V. Perrin, C. Hartmann, J.-L. Fuda, D. Varillon), LEGOS support staff (B. Cournou, N. Lacroux, A. Baritaud, M. Mena) for their efforts to ensure a timely delivery of instruments/air tickets or participation at different stages of the cruise organization. We thank the Hydraulics Laboratory at SIO for their mooring expertise (P. Harvey, S. Kawamoto). We thank Patrick Raimbault (MIO) who did the mass spectrometric analyses related to biology. The Pandora data processing followed the US-IMAGO (IRD) information system (online at http://www.ird.fr/us191/spip.php?article66). Dissolved Fe analyses were funded by Labex Mer. Most geochemical graphics were made under ODV (Schlitzer, R., Ocean Data View, http://odv.awi.de, 2015). MoorSPICE data can be found through Cruise https://doi.org/10.7284/903044. Also thank C. Schmechtig who ensured the Pandora and MoorSPICE data delivery through the LEFE-CYBER Database (online at http://www.obs-vlfr.fr/proof/).

The accomplishments were made possible through concurrent contributions of national funding agencies. The MoorSPICE and Pandora cruises have been co-funded by NSF grant OCE1029487, and by ANR project ANR- 09-BLAN-0233-01 and INSU/LEFE project Solwara (IDAO and CYBER). Carbonate measurements were supported by European program FP7/CARBOCHANGE (grant 264879). NM thanks the Service National dAnalyses des Paramètres Océaniques du CO₂ (SNAPO-CO₂) supported by CNRS/INSU and UPMC (OSU ECCE-Terra). CG was supported by a PhD scholarship from the French Ministry of Research and Education. South Solomon Sea endpoint moorings (US NOAA funded CORC project grant NA10OAR432015) Mercury analyses were funded by ERC-2010-StG_20091028 grant from the European Research Council to JES.

The authors have no competing interests to declare.

AG, SC, JS, CJ, and GE coordinated the projects, cruises and article writing. CG and MA contributed to the physical analysis and produced corresponding figures; MB, SB, HB, MR lead the biological analyses; JL the volcanic aspects. The geochemistry aspects were analyzed by NM (carbonates); L-EH (Hg); SM, JR (Al/Mn); FQ, GS, FL (Fe). Cruise data acquisition and treatment was ensured by JG and FB. TH lead the JAMSTEC cruise; WK, MK, EP, US, PV-B, MS and JES were involved at different stages of the results presented here.