Evidence for enhanced primary production driving significant CO2 drawdown associated with the Atlantic ITCZ

Evidence for enhanced primary production driving significant CO 2 drawdown associated with the Atlantic ITCZ J. Severino P. Ibánhez a,b, ⁎,Manuel Flores Montes c , Nathalie Lefèvre d a Instituto de Investigacións Mariñas, Consejo Superior de Investigaciones Científicas (IIM-CSIC), Eduardo Cabello 6, 36208 Vigo, Spain b Biogeochemistry Research Group, School of Natural Sciences, Trinity College Dublin, College Green, Dublin 2, Ireland c Department of Oceanography –DOCEAN, Federal University of Pernambuco –UFPE, Av. Arquitetura, s/n, Cidade Universitária, 50740-550 Recife, PE, Brazil d IRD-LOCEAN, Sorbonne Université, 4 place Jussieu, 75252 Paris Cedex 05, France HIGHLIGHTS •Rainfall associated with the Atlantic ITCZ promotes a large reduction of oceanic CO 2 emissions to the atmosphere. •Thermodynamics and physics only explain 41% of the observed CO 2 drawdown associated with rainfall in the tropical Atlantic. •The freshening caused by the ITCZ seems to enhance primary production in the otherwise oligotrophic tropical Atlantic. GRAPHICAL ABSTRACT ABSTRACTARTICLE INFO Editor: Julian Blasco The intense rainfall associated with the Intertropical Convergence Zone (ITCZ), a narrow zone of confluence of the northeast and southeast trades, can significantly alter sea surface salinity, the chemistry of inorganic C and the resulting sea-air CO 2 exchange in the tropics. We have analyzed extensive underway data collected from 2008 until 2014 and recorded by an autonomous CO 2 system installed on a commercial ship that crosses the central tropical Atlantic (5°S to 15°N, 18°W to 36°W) to disentangle the effects of the ITCZ over the carbonate system there. Based on statistically significant linear co-variance of sea surface fugacity of CO 2 (fCO 2sw ) and sea surface salinity in the areas affected by the ITCZ, we calculated CO 2 drawdown rates associated with the impact of the ITCZ in the central tropical Atlantic ranging from 0.11 ± 0.02 to 2.35 ± 0.08 mmol m −2 d −1 . These were calculated by comparing the observed fCO 2sw with that expected without surface seawater carbonate system dilution and increase in gas transfer caused by the ITCZ. The observed decrease in fCO 2sw associated with the freshening caused by the ITCZ is much larger than expected from thermodynamics alone. 59.1 ± 4.1 % of the total observed CO 2 drawdown associated with the ITCZ cannot be explained by abiotic processes. Instead, we found significant negative correlations between underway sea surface salinity and remote-sensed chlorophyll ain the areas affected by the ITCZ. Different to other tropical oceanic basins, the tropical Atlantic receives large amounts of continental dust originated from Africa. Wet dust deposition driven by the ITCZ appears associated with the interannual variability of the CO 2 drawdown associated with the ITCZ. Fertilization driven by the ITCZ seems to enhance primary production in the otherwise oligotrophic tropical Atlantic, thus significantly lowering CO 2 emissions to the atmosphere. Keywords: Tropical Atlantic Oceanic carbonate system Sea-air CO 2 flux Rainfall Diazotrophy Science of the Total Environment 838 (2022) 156592 ⁎Corresponding author at: Instituto de Investigacións Mariñas, Consejo Superior de Investigaciones Científicas (IIM-CSIC), Eduardo Cabello 6, 36208 Vigo, Spain. E-mail address: [email protected] (J.S.P. Ibánhez). http://dx.doi.org/10.1016/j.scitotenv.2022.156592 Received 11 January 2022; Received in revised form 14 May 2022; Accepted 6 June 2022 Availableonline08June2022 0048-9697/© 2022 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/). Contents lists available at ScienceDirect Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv 1. Introduction The global ocean is a net sink of atmospheric CO 2 ,absorbingabout23% of the anthropogenic CO 2 emissions to the atmosphere during the last decade (Friedlingstein et al., 2020). Nevertheless, sea-air CO 2 fluxes are not uniformly distributed throughout the ocean's surface. Since atmospheric CO 2 has a relatively homogeneous global distribution, the role of ocean's surface as a source or a sink of atmospheric CO 2 depends on the distribution of the fugacity of CO 2 (fCO 2 ) in surface seawaters. Thus, the tropical band is generally a source of CO 2 to the atmosphere due to high sea surface temperatures (SST) and the equatorial upwelling of CO 2 -rich waters (e.g. Landschützer et al., 2016;Takahashi et al., 2014). The tropical Atlantic is the second largest oceanic CO 2 source to the atmosphere after the tropical Pacific, largely fuelled by the zonal spread of CO 2 -rich waters originated in the equatorial upwelling (Andrié et al., 1986). Nevertheless, different to the tropical Pacific, the tropical Atlantic receives the discharge of some of the largest rivers in the world, such as the Amazon, Niger and Congo rivers, that spread through vast areas beyond the coastal zone driven by the prevailing winds and complex surface currents (Bianchi and Allison, 2009). These large outer river plumes are significant atmospheric CO 2 sinks due to enhanced primary production in the otherwise oligotrophic tropical Atlantic (Cooley et al., 2007;Ibánhez et al., 2015;Körtzinger, 2003, 2010;Lefèvre, 2009;Lefèvre et al., 2017a;Ternon et al., 2000). The zonal spread of these continental freshwater sources further overlaps with the freshening caused by the Intertropical Convergence Zone (ITCZ), significantly affecting the fCO 2 of surface waters across the basin (Ibánhez et al., 2016;Lefèvre et al., 2010;Oudot et al., 1995). The ITCZ is a narrow zonal band of intense rainfall located in the zone of confluence of the northeast and southeast trades, where they converge to form the rising branch of the Hadley circulation, accounting for 32 % of the global rainfall (Kang et al., 2018). The ITCZ seasonally migrates following the hemispheric asymmetry of SST, being at its northernmost location during boreal summer-early autumn (Donohoe et al., 2013;Kang et al., 2018). The intensity of the rainfall associated with the ITCZ in the tropical ocean is capable of significantly changing sea surface salinity (SSS) and SST across the basin (e.g. Dessier and Donguy, 1994;Grodsky et al., 2020). This freshening caused by the ITCZ dilutes both sea surface total alkalinity (TAlk) and dissolved inorganic C (DIC), enhanced by density stratification, thus lowering sea surface fCO 2 (fCO 2sw ) and increasing pH (Ashton et al., 2016;Ho and Schanze, 2020;Turk et al., 2010;Woolf et al., 2019). Rainfall is also responsible for the transport of CO 2 to surface waters, termed wet deposition, although a general drop of fCO 2sw caused by rainfall prevails (Sarmiento and Gruber, 2006;Turk et al., 2010). Rainfall further enhances sea-air CO 2 exchange rates by increasing gas transfer at the seawater surface through turbulence caused by rain drops (Harrison et al., 2012;Ho et al., 2000). This effect of rain over the gas transfer at the sea surface decreases at increasing wind speeds (Harrison et al., 2012). Combined, the physical and thermodynamic effects of rainfall over the sea-air exchange of CO 2 can either reduce CO 2 emissions in areas of CO 2 outgassing or enhance sea surface CO 2 uptake in oceanic CO 2 sink areas (Ho and Schanze, 2020), thus significantly impacting sea-air CO 2 exchange globally (Ashton et al., 2016). A lowering of the fCO 2sw proportional to SSS changes caused by the rainfall associated with the ITCZ has been observed in the tropical Pacific (Ho and Schanze, 2020;Turk et al., 2010), as well as in the tropical Atlantic (Lefèvre et al., 2010;Oudot et al., 1995). In the tropical Pacific, the physical and thermodynamic effects of rainfall over the carbonate system successfully explained the observed fCO 2sw in the areas affected by the ITCZ (Ho and Schanze, 2020;Turk et al., 2010). Different to the tropical Pacificand adding to the spatial overlapping of the freshening caused by the ITCZ and the zonal spread of large river plumes, the tropical Atlantic receives the largest fluxes of continental dust of all oceans originating from northwest Africa (Mahowald et al., 2009) that crosses the basin reaching the Caribbean Sea and America (Prospero et al., 2014). Wet deposition associated with the ITCZ is one of the main transport pathways of these particles (van der Does et al., 2020), thus fertilizing vast oligotrophic oceanic areas (Schlosser et al., 2014). This raises high uncertainties about the expected role of the ITCZ in the sea-air CO 2 exchange in the tropical Atlantic. Nevertheless, the drivers of surface seawater carbonate system variability in the areas affected by the ITCZ in the tropical Atlantic have not been addressed to date. In this study, we use routine underway measurements performed along a commercial line between France and Brazil that crosses the central tropical Atlantic. The spatial and temporal coverage of the underway measurements performed permit to obtain a complete picture of the seasonal influence of the ITCZ over fCO 2sw in the area. The main objective of this study is therefore to evaluate the impact of the intense rainfall associated with the ITCZ over the fCO 2sw and sea-air CO 2 fluxes in the tropical Atlantic. The recorded underway fCO 2sw is compared to that expected from thermodynamics and the main drivers of the observed changes in the local carbonate system are disentangled. 2. Materials and methods 2.1. Underway measurements and sea-air CO 2 exchange An automated CO 2 equipment based on infrared detection (LI-COR Inc. model 7000, Lincoln, USA; Pierrot et al., 2009) was installed on board commercial vessels since 2008 sailing the commercial route Le Havre (mainland France) –Santos (Brazil), consecutively on the Motor Vessel (MV) Monte Olivia (2008–2009), MV Rio Blanco (2009–2012), MV Santa Cruz (2013–2014) and MV Cap San Lorenzo (2014 onwards; Table 1). During the ships' voyages, underway fCO 2sw and atmospheric fCO 2 (fCO 2atm ) determinations were made. A thermosalinograph (SeaBird Scientific model SBE21, Bellevue, USA) and a barometer (Druck model RPT350, Baker Hughes Company, Leicester, UK) were also installed in the vessels, thus measuring underway SST, SSS and atmospheric pressure (Pres). In this study, we use ships' voyages from 2008 until 2014 that recorded sufficient data in the area of influence of the ITCZ in the tropical Atlantic (Fig. 1). Data used were collected between 15°N and 5°S (Lefèvre and Diverrès, 2021a,b) from 35 voyages (Table 1). During March and April 2011, the atmospheric molar fraction of CO 2 (xCO 2atm ) could not be recorded onboard the MV Monte Olivia due to a problem with the atmospheric pumping. For this period, the ship atmospheric measurements are replaced by the monthly xCO 2atm recorded at the atmospheric stations of the NOAA/ESRL Global Monitoring Division (http://www.esrl.noaa.gov/gmd/ccgg/iadv/) of Ascension Island (7.97°S, 14.40°W), Farol De Mae Luiza Lighthouse (5.80°S, 35.19°W; data since 2010), Ragged Point (13.17°N, 59.43°W) and Tenerife (28.31°N, 16.50°W) are used. Monthly xCO 2atm measurements at these stations are linearly interpolated at the position of the underway measurements. fCO 2atm is then calculated as fCO2atm ¼xCO2atm Pres pH2O ðÞ C(1) where pH 2 O is the water vapor pressure at 100 % humidity calculated from underway SST and SSS, and C is the fugacity coefficient calculated from Weiss (1974). The same procedure is used to calculate fCO 2atm along the tracks of the voyages where underway XCO 2atm was measured. Comparison of both calculated and measured fCO 2atm gives very good results, with deviations (average 1.70 μatm, n = 40,925; see Supplementary materials) within the precision range of the equipment (<2μatm). Sea-air CO 2 fluxes (F) are calculated according to F¼ktotal SofCO2sw fCO2atm ðÞ (2) where S o is the solubility of CO 2 (Weiss, 1974)andk total is the gas transfer velocity. Both wind speed and rainfall enhance sea-air gas exchange by affecting k total in a nonlinear manner (Harrison et al., 2012): ktotal ¼kwind þ1−exp −αβðÞ½krainfall ð3Þ where k wind is the gas transfer velocity associated with the wind solely, k rainfall is that promoted by rainfall solely, αis a non-dimensional fitting J.S.P. Ibánhez et al. Science of the Total Environment 838 (2022) 156592 2 parameter of the empirical Eq. (3) and βis the ratio between the kinetic energy flux caused by rainfall (KEF rain ) and that caused by wind (KEF wind ). KEF rain can be simplified to KEF rain = 0.0112 Rn (Harrison et al., 2012), where Rn is the rainfall rate, while KEF wind can be calculated as: KEFwind ¼ρCdU2 10 (4) where ρis the air density and C d is the drag coefficient. k rainfall and k wind are calculated following the parametrizations made by Harrison et al. (2012) and Ho et al. (2006), respectively: k600ðÞ rainfall ¼63:02 KEFrain ðÞ 0:6242 (5) k600ðÞ wind ¼0:266 U2 10 (6) where U 10 is the wind speed at 10 m height above sea surface. 2.2. Remote-sensed and reanalysis data U 10 is obtained from the European Centre for Medium-Range Weather Forecasts (ECMWF) daily reanalysis data (ERA-interim, 0.125° resolution). Daily rainfall data are obtained from the Tropical Rainfall Measuring Mission (TRMM) (Huffman et al., 2007)(http://precip.gsfc.nasa.gov/; 0.25° resolution). In both cases, the regular grid of daily data was linearly interpolated at the position of each underway measurement and using the date of passage (Table 1). U 10 is used to calculate wind stress (τ) and thus, Ekman pumping (We) from the equation (Pickett and Paduan, 2003): We¼Curl τ ρf  (7) where ρis the density of seawater (assumed 1025 Kg m −3 )andfis the Coriolis parameter. Rainfall events can promote stratification in the open ocean that can last for several hours depending on factors such as wind speed or surface current velocity (Boutin et al., 2015). The gridded daily rainfall TRMM data are additionally used to identify the areas affected by recent rainfall along the tracks of the VOS used here. The accumulated rainfall corresponding to a four days period prior the passage of the ships was linearly interpolated at the position of the underway measurements. This arbitrary integration time-period was varied (±2 days) offering highly similar results (data not shown). An accumulated rainfall threshold of 1 mm is used to discern among areas affected and not affected by recent rainfall. Sea surface current velocities obtained from the Ocean Surface Current Analyses –Real time (OSCAR) data (1/3° resolution; JPL Physical Oceanography DAAC; developed by ESR) are used to separate the different surface waters and thus avoid fCO 2sw changes caused by the mixture of oceanic water masses with different origin and carbonate system properties. In this case, we opt for using the monthly OSCAR product to avoid surface water masses misidentification due to short-term reversal of surface water direction. Sea surface circulation in the tropical Atlantic is dominated by the zonal component, with the westward North Equatorial Current (NEC) and South Equatorial Current (SEC) separated by the eastward North Equatorial Countercurrent (NECC; Fig. 1). The SEC further presents three branches, although only two appear in the study area (Fig. 1), separated by the South Equatorial Undercurrent (Stramma, 1991). At the surface, the divide between the two branches of the SEC in the study area varies seasonally from westward to eastward propagation (Stramma, 1991). The gridded zonal component of the OSCAR dataset was linearly interpolated at the position of each underway measurement and used to identify the limits of the three main surface currents in the study area. From North to South, the limit of the NEC is stablished where the zonal component of the velocity changes to eastward propagation when the NECC is present, Table 1 VOS voyages used in this study. The limits of the ship tracks used here and the dates the ship performed each track are also shown. ID Dates of the voyages Name Minimum latitude Maximum latitude Representative month 115–18 July 2008 MV Monte Olivia 5°S 15°N July 28–10 October 2008 MV Monte Olivia 5°S 15°N October 319–21 November 2008 MV Monte Olivia 4°S 15°N November 49–11 December 2008 MV Monte Olivia 5°S 15°N December 51–3 January 2009 MV Monte Olivia 5°S 14°N January 620–23 January 2009 MV Monte Olivia 5°S 15°N January 711–13 February 2009 MV Monte Olivia 5°S 15°N February 83–6 March 2009 MV Monte Olivia 5°S 15°N March 914–17 April 2009 MV Monte Olivia 5°S 15°N April 10 15–18 December 2009 MV Rio Blanco 5°S 15°N December 11 20–23 January 2010 MV Rio Blanco 5°S 15°N January 12 10–12 February 2010 MV Rio Blanco 5°S 15°N February 13 3–6 March 2010 MV Rio Blanco 5°S 15°N March 14 23–26 March 2010 MV Rio Blanco 5°S 15°N March 15 9–12 May 2010 MV Rio Blanco 5°S 15°N May 16 4–7 June 2010 MV Rio Blanco 5°S 15°N June 17 28 June-1 July 2010 MV Rio Blanco 5°S 15°N June 18 22–26 July 2010 MV Rio Blanco 5°S 15°N July 19 15–18 August 2010 MV Rio Blanco 5°S 15°N August 20 9–13 September 2010 MV Rio Blanco 5°S 15°N September 21 28–30 October 2010 MV Rio Blanco 5°S 15°N October 22 24–28 March 2011 MV Rio Blanco 5°S 15°N March 23 18–20 April 2011 MV Rio Blanco 5°S 15°N April 24 12–16 May 2011 MV Rio Blanco 5°S 15°N May 25 4–7 June 2011 MV Rio Blanco 5°S 15°N June 26 30 June–4 July 2011 MV Rio Blanco 5°S 15°N July 27 23–26 July 2011 MV Rio Blanco 5°S 15°N July 28 6–9 June 2012 MV Santa Cruz 5°S 15°N June 29 1–4 July 2012 MV Santa Cruz 5°S 15°N July 30 20–22 August 2012 MV Santa Cruz 5°S 15°N August 31 13–15 September 2012 MV Santa Cruz 5°S 10°N September 32 4–7 April 2013 MV Santa Cruz 5°S 15°N April 33 28 April–1 May 2013 MV Santa Cruz 5°S 15°N April 34 23–26 January 2014 MV Santa Cruz 5°S 15°N January 35 11–15 December 2014 MV Cap San Lorenzo 5°S 15°N December J.S.P. Ibánhez et al. Science of the Total Environment 838 (2022) 156592 3 or until an intensification of the westward propagation caused by the SEC when the NECC is not present. The NECC is identified as the area with eastward surface water propagation in the 3–10°N latitudinal band, between the SEC and the NEC. Finally, the two branches observed in the SEC are assumed to have similar carbonate system characteristics and included in a single SEC discrimination. The reliability of remote-sensed wind speed, rainfall and the zonal component of surface currents obtained from reanalysis was tested by comparing them with those properties measured in the Prediction and Research Moored Array in the Tropical Atlantic (PIRATA; Bourlès et al., 2019; Fig. 1). Remote-sensed and reanalysis properties were linearly interpolated at the location of the five (two with available surface current data) PIRATA moorings closer to the study area framed in Fig. 1 (5°S to 15°N, 18°W to 36°W). Both sets of data showed very good agreement for the 2008–2014 period (see Supplementary materials). Additionally, remote-sensed, monthly-averaged Chlorophyll a(Chl a) concentration, reanalysis wet and dry dust deposition rates and remotesensed SSS are used in this study. Monthly-averaged Chl aconcentration is obtained from the Moderate Resolution Imaging Spectroradiometer (MODIS)/Aqua satellite with a 4 km resolution and processed with the OC3M algorithm and standard NASA global coefficients (O'Reilly et al., 1998). The monthly data are linearly interpolated at the position of each underway measurement. Monthly wet and dry dust deposition rates are obtained from the NASA's Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2), produced by the Goddard Earth Observing System-Data Assimilation System (GEOS-DAS, version 5.12.4; Gelaro et al., 2017). MERRA-2 provides monthly dust deposition reanalysis data from 1980 with a spatial resolution of 0.5° latitude and 0.66° longitude and discriminated into 5 dust size bins. The total wet and dry dust deposition rates are calculated by summing up the 5 dust size classes. Finally, monthly SSS data (0.25° resolution) are obtained from the Soil Moisture and Ocean Salinity (SMOS) mission, available from the Ocean Salinity Expertise Center (CECOS), IFREMER (France). Rainfall and wet and dry dust deposition climatologies in the area of study are computed as the long-term (1998–2014), monthly average in the area framed within 5°S-15°N and 18°W-36°W (Fig. 1). The period used to calculate these climatologies is arbitrarily chosen to obtain a longterm mean of these properties beyond the period of study, thus limiting the impact of anomalous years over the long-term mean when using short periods. Wet and dry dust deposition and rainfall anomalies are then computed for the studied period (2008–2014). Furthermore, the position of the ITCZ and the maximum wet dust deposition are calculated as the monthly maximum precipitation/wet dust deposition latitudinal band within the same framed area and averaged for the 18°W-36°W longitudinal band. 2.3. Thermodynamic calculations CO 2 solubility, vapor pressure, fCO 2sw and the equilibrium constants of the carbonate system are all affected by SST and SSS changes (Dickson et al., 2007). The overall effect of SST changes over fCO 2sw is determined through the thermodynamic coefficient presented by Takahashi et al. (2009): ∂ln fCO2sw ∂SST ¼0:0433−8:710−5SST ð8Þ Fig. 1. a. Tracks of the 35 ships' voyages of the voluntary observing ships (VOSs) crossing the tropical Atlantic and used in this study. The long-term mean surface velocity for the month of August (calculated for the 1998–2014 period) is shown, with the indication of the main surface currents present in the studied area (marked square): the North Equatorial Current (NEC), the North Equatorial Countercurrent (NECC) and the South Equatorial Current (SEC). The location of the Prediction and Research Moored Array in the Tropical Atlantic (PIRATA; Bourlès et al., 2019) is also shown (circles), together with the moorings used in this study (black circles). b. Monthly latitudinal ITCZ position calculated as the monthly maximum precipitation latitudinal band within 5°S–15°N and averaged for the 18–36°W longitudinal band. J.S.P. Ibánhez et al. Science of the Total Environment 838 (2022) 156592 4 Similarly, Sarmiento and Gruber (2006) describe the effect of SSS variations over fCO 2sw when resulting from the admixture of waters with highly different DIC and TAlk content, such as the influence of rainfall on seawater, as: ∂ln fCO2sw ∂ln SSS ¼γsþγDIC þγTAlk ¼γrain (9) where γ s refers to the effect of SSS changes over the dissociation constants of the carbonate system and is close to 1, and γ DIC and γ TAlk are the sensitivities of fCO 2sw to changes in DIC (i.e. Revelle factor) and Talk, respectively. Reported globally averaged values of γ rain range from 1.6 at low latitudes to 1.7 at high latitudes (Sarmiento and Gruber, 2006). To evaluate the magnitude of each of the thermodynamic terms of Eq. (9) in the tropical Atlantic, we first use the highly significant linear relationship of TAlk with SSS found by Lefèvre et al. (2010) to estimate TAlk Fig. 2. Latitudinal distribution of sea surface temperature (SST), sea surface salinity (SSS), interpolated rainfall accumulated during the 4 days prior the passage of the VOS, sea surface fCO 2 and the interpolated zonal component of the monthly-averaged surface velocity (u; positive values indicate eastward movement) for the 35 voyages used in this study. Voyages are organized by representative month and year is denoted by the color code. Note that during January 2009, March and June 2010, July 2011 and April 2013 two voyages were performed. J.S.P. Ibánhez et al. Science of the Total Environment 838 (2022) 156592 5 along the tracks. DIC is then estimated from the calculated TAlk, underway fCO 2sw , SST and SSS using the CO2SYS program (van Heuven et al., 2011) and the dissociation constants of carbonic acid in seawater found by Mehrbach et al. (1973) refitbyDickson and Millero (1987). Calculated TAlk and DIC are then used to estimate γ DIC and γ TAlk in the areas affected by rainfall associated with the ITCZ as follows: γDIC ¼ΔfCO2=fCO2 ðÞ=ΔDIC=DICðÞ (10) γTAlk ¼ΔfCO2=fCO2 ðÞ=ΔTAlk=TAlkðÞ (11) To calculate γ DIC, we impose a DIC change of 1 μmol Kg −1 keeping unvaried TAlk and calculate the resulting fCO 2 using the CO2SYS program. γ TAlk is calculated similarly by imposing a change of 1 μmol Kg −1 in TAlk. To evaluate the effect of the freshening caused by the ITCZ over underway fCO 2sw in the tropical Atlantic, we look for significant linear relationships of ln fCO 2sw vs ln SSS (Eq. (9)) in the areas directly affected by recent rainfall, i.e. those where the accumulated rainfall in the four days prior the passage of the VOS was higher than 1 mm. The average SST from all the underway determinations used here is first used in Eq. (8) as reference temperature to remove the thermodynamic effect of temperature over fCO 2sw. To evaluate the total contribution of the ITCZ to the sea-air CO 2 exchange in the region, sea-air CO 2 exchange in the regions where significant linear relationships of ln SSS vs ln fCO 2sw are found is decomposed as follows: FCO2sea−air ¼FCO2oceanic þFCO2itcz ð12Þ where FCO 2sea−air is the sea-air CO 2 exchange calculated with underway fCO 2sw measurements, FCO 2oceanic is that expected without the effects of the ITCZ and FCO 2itcz is that associated with the overall impact of the rainfall associated with the ITCZ. To estimate FCO 2oceanic ,wefirst need to estimate an fCO 2sw representative of the different surface oceanic waters transported by the surface current system. The broader spatial extent of the freshening caused by the ITCZ often causes a drop in SSS in an entire surface current, particularly in the NECC, and adds to mixing of waters with different origin towards the limits of the different surface currents, thus hindering the association of an fCO 2sw value to each of the surface water masses. As a conservative estimate, for each of the areas where significant linear relationships of ln fCO 2sw vs ln SSS are found, the highest measured SSS within these relationships is assumed as representative of each of the currents analyzed without the influence of the ITCZ. This is used together with the ln SSS vs ln fCO 2sw linear relationships found to calculate an fCO 2sw representative of the oceanic waters of each current and thus, to compute FCO 2oceanic .FCO 2itcz is then calculated from the difference between FCO 2sea−air and FCO 2oceanic . Further to this analysis, FCO 2itcz is decomposed into: FCO2itcz ¼FCO2thermodynamics þFCO2unknown (13) where FCO 2thermodynamics accounts for the expected impact of the ITCZ over the carbonate system in the area and the enhanced sea-air transfer coefficient caused by rainfall and FCO 2unknown denotes the unexplained sea-air CO 2 exchange. We use underway SSS changes and the theoretical γ rain (1.6; Sarmiento and Gruber, 2006)tocalculatefCO 2sw and thus FCO 2thermodynamics . 3. Results 3.1. Underway SST, SSS and fCO 2sw variability in the central tropical Atlantic The monthly distribution of the 35 voyages analyzed here show the seasonal variability of the measured properties in the study area (Table 1; Fig. 2). The latitudinal distribution of underway SST in the tropical Atlantic shows a seasonal pattern in the north of the studied region (15°N), with lower temperatures during boreal winter (down to 22 °C; Fig. 2 January to April) and higher during boreal summer (up to 30 °C; Fig. 2 July to October). In the equatorial area, SST shows lower temporal variability, remaining all year-round in between 26 and 30 °C. SSS measured underway shows Fig. 3. Latitudinal distribution of ΔfCO 2 calculated for the 35 ships' voyages used here. Year is denoted by the color code. Note that during January 2009, March and June 2010, July 2011 and April 2013 two voyages were performed. J.S.P. Ibánhez et al. Science of the Total Environment 838 (2022) 156592 6 two distinctive areas: one characterized by constant SSS characteristic of oceanic waters, and another one affected by the freshening caused by the intense rainfall associated with the ITCZ and potentially by the Amazon river plume during boreal summer in the NECC (Lefèvre et al., 2020). The position of the ITCZ migrates seasonally and is placed at its northernmost latitude during August/September, and at the southern-most latitude during March to May (Fig. 1;e.g.Lefèvre et al., 2010). This seasonal migration of the position of the ITCZ is observable in the short-term (4 days) rainfall interpolated at the positions of underway measurements used here, with rainfall concentrated in the 5°S-5°N latitudinal band from March to May and in the 5–13°N latitudinal band during August–September (Fig. 2). The location of measured underway SSS minima almost perfectly matches the areas affected by short-term rainfall, thus following the seasonal displacement of the ITCZ (Fig. 2). Underway fCO 2sw shows significant latitudinal gradients associated with the freshening caused by the ITCZ (Lefèvre et al., 2010)andthe main system of surface currents in the area (Fig. 2;Lefèvre et al., 2014). As rainfall typically contains no total alkalinity (TAlk) and small amounts of Dissolved Inorganic C (DIC), intense rainfall dilutes these parameters in the surface ocean and promotes a lowering of fCO 2sw . This effect can be seen underway in the zones with lower SSS (Fig. 2), and acts together with SST, the surface circulation and the surface transport of different water masses in shaping surface fCO 2sw levels. The three zonal sea surface currents that dominate the circulation in the study area (i.e. the westward NEC and SEC separated by the eastward NECC; Fig. 1) are clearly identified in the interpolated zonal component of the surface circulation obtained from the OSCAR data (Fig. 2). In the southern limit of the studied area, the two branches of the SEC (~5°S5°N) are identified as two different westward (negative u) maximum zonal velocities, sometimes separated by eastward transport as during March to May (Fig. 2). The SEC transports CO 2 -rich waters originated from the equatorial upwelling system and the Benguela Current (e.g. Andrié et al., 1986;Stramma, 1991), which explains the high underway fCO 2sw characteristic of the ~5°S-5°N latitudinal band throughout the year outside the areas affected by the ITCZ (Fig. 2). In the northern limit of the studied area, the NEC, mainly fed by the Canary Current that transports waters originated in the Northern hemisphere (e.g. HernándezGuerra et al., 2005), is characterized by weak westward velocities along the tracks of the vessels used here (Fig. 2;~5–15°N). There, fCO 2sw shows a clear seasonal pattern; lower fCO 2sw occurs during boreal winter while the higher fCO 2sw in the NEC is verified from June to October. SST is the primary driver of the fCO 2sw variability in the NEC (Ibánhez et al., Table 2 Impact of the ITCZ over the carbonate system in the tropical Atlantic. ID corresponds to the identification of each VOS voyage used in this study as in Table 1. Data with shortterm interpolated rainfall higher than 1 mm was used to obtain the thermodynamic effect of salinity changes caused by rainfall (γ rain )overfCO 2sw through linear regression (r) of ln fCO 2swnorm vs ln SSS. Estimated fCO 2sw sensitivities to DIC (Revelle factor) and TAlk (TAlk sensitivity) changes were used to calculate the expected thermodynamic effect of rainfall over fCO 2swnorm (Calc. γ rain ) for each voyage used. ID Month Year Current SSS range γ rain r Revelle factor TAlk sensitivity Calc. γ rain n 1 7 2008 SEC 34.7–35.7 4.61 ± 0.06 0.95 8.93 ± 0.00 −8.34 ± 0.00 1.59 ± 0.01 558 NECC 33.7–35.8 2.85 ± 0.08 0.88 8.81 ± 0.00 −8.22 ± 0.00 1.59 ± 0.01 412 2 10 2008 SEC 35.1–35.5 4.98 ± 0.05 0.98 9.06 ± 0.00 −8.46 ± 0.00 1.59 ± 0.00 289 3 11 2008 SEC 34.5–36.2 3.50 ± 0.03 0.98 8.88 ± 0.00 −8.29 ± 0.00 1.59 ± 0.00 360 4 12 2008 SEC 33.7–36.2 2.89 ± 0.04 0.96 8.98 ± 0.00 −8.39 ± 0.01 1.59 ± 0.01 427 5 1 2009 SEC 34.5–36.2 3.45 ± 0.03 0.99 8.93 ± 0.00 −8.34 ± 0.00 1.59 ± 0.01 353 6 1 2009 SEC 35.0–36.1 4.26 ± 0.09 0.88 9.07 ± 0.00 −8.47 ± 0.00 1.60 ± 0.01 687 7 2 2009 SEC 34.1–36.2 1.45 ± 0.02 0.95 9.08 ± 0.00 −8.48 ± 0.00 1.60 ± 0.00 494 8 3 2009 SEC 34.6–36.0 1.97 ± 0.04 0.93 9.06 ± 0.00 −8.46 ± 0.00 1.60 ± 0.00 427 9 4 2009 SEC 35.0–36.0 1.25 ± 0.04 0.84 8.93 ± 0.00 −8.34 ± 0.00 1.59 ± 0.00 368 10 12 2009 SEC 35.6–36.4 4.52 ± 0.12 0.85 9.09 ± 0.00 −8.50 ± 0.00 1.60 ± 0.01 572 NEC 34.8–35.8 2.66 ± 0.05 0.95 8.83 ± 0.00 −8.25 ± 0.00 1.59 ± 0.00 283 11 1 2010 SEC 35.2–36.3 3.47 ± 0.05 0.95 9.04 ± 0.00 −8.45 ± 0.00 1.60 ± 0.01 440 12 2 2010 SEC 35.4–36.2 3.75 ± 0.04 0.98 9.10 ± 0.00 −8.46 ± 0.00 1.60 ± 0.01 318 15 5 2010 SEC 35.0–36.2 2.92 ± 0.06 0.89 8.90 ± 0.00 −8.31 ± 0.00 1.59 ± 0.00 712 16 6 2010 SEC 35.1–36.2 2.71 ± 0.04 0.92 8.83 ± 0.00 −8.24 ± 0.00 1.59 ± 0.00 806 17 6 2010 SEC 34.8–36.1 3.29 ± 0.03 0.97 8.87 ± 0.00 −8.28 ± 0.00 1.59 ± 0.01 619 NECC 34.6–35.8 2.15 ± 0.05 0.90 8.77 ± 0.00 −8.19 ± 0.00 1.59 ± 0.00 360 18 7 2010 SEC 35.2–36.1 3.19 ± 0.09 0.84 8.89 ± 0.00 −8.30 ± 0.00 1.59 ± 0.01 594 NECC 34.8–36.1 2.60 ± 0.04 0.94 8.73 ± 0.00 −8.14 ± 0.00 1.59 ± 0.00 697 19 8 2010 NECC 34.7–36.2 2.77 ± 0.05 0.90 8.77 ± 0.00 −8.19 ± 0.00 1.59 ± 0.00 670 20 9 2010 NECC 34.6–36.3 3.02 ± 0.05 0.92 8.70 ± 0.00 −8.12 ± 0.00 1.58 ± 0.00 636 21 10 2010 SEC 34.1–36.2 4.24 ± 0.07 0.96 8.77 ± 0.00 −8.19 ± 0.00 1.59 ± 0.00 305 NECC 33.7–36.2 2.56 ± 0.04 0.94 8.67 ± 0.00 −8.09 ± 0.00 1.58 ± 0.01 500 22 3 2011 SEC 34.9–35.9 2.29 ± 0.06 0.79 8.89 ± 0.00 −8.30 ± 0.00 1.59 ± 0.00 830 23 4 2011 SEC 35.3–36.0 2.30 ± 0.07 0.85 8.88 ± 0.00 −8.29 ± 0.00 1.59 ± 0.00 464 24 5 2011 SEC 35.1–36.2 2.89 ± 0.03 0.96 8.81 ± 0.00 −8.22 ± 0.00 1.59 ± 0.00 837 25 6 2011 SEC 35.2–36.3 2.86 ± 0.07 0.88 8.88 ± 0.00 −8.29 ± 0.00 1.59 ± 0.01 445 NECC 34.5–35.6 2.25 ± 0.06 0.92 8.75 ± 0.00 −8.16 ± 0.00 1.58 ± 0.00 242 26 7 2011 NECC 34.4–36.1 2.45 ± 0.07 0.84 8.75 ± 0.00 −8.16 ± 0.00 1.58 ± 0.01 473 27 7 2011 SEC 35.2–36.3 2.54 ± 0.07 0.91 8.96 ± 0.00 −8.37 ± 0.00 1.59 ± 0.01 303 NECC 34.0–36.1 3.33 ± 0.07 0.91 8.77 ± 0.00 −8.18 ± 0.00 1.58 ± 0.00 481 28 6 2012 SEC 35.1–36.2 3.52 ± 0.03 0.98 8.97 ± 0.00 −8.38 ± 0.00 1.59 ± 0.01 482 NECC 34.2–35.5 4.55 ± 0.09 0.96 8.75 ± 0.00 −8.17 ± 0.00 1.58 ± 0.01 217 29 7 2012 SEC 35.4–36.3 4.68 ± 0.06 0.96 9.00 ± 0.00 −8.41 ± 0.00 1.59 ± 0.00 383 NECC 34.2–35.9 2.53 ± 0.04 0.96 8.76 ± 0.00 −8.18 ± 0.00 1.58 ± 0.01 372 30 8 2012 NECC 34.6–35.9 4.41 ± 0.13 0.89 8.86 ± 0.00 −8.27 ± 0.00 1.59 ± 0.01 293 31 9 2012 NECC 34.4–35.9 2.89 ± 0.07 0.91 8.80 ± 0.00 −8.21 ± 0.00 1.59 ± 0.01 386 32 4 2013 SEC 34.7–36.2 2.35 ± 0.03 0.96 8.99 ± 0.00 −8.39 ± 0.00 1.59 ± 0.00 515 33 4 2013 SEC 35.0–36.2 3.00 ± 0.02 0.98 8.92 ± 0.00 −8.33 ± 0.00 1.60 ± 0.00 585 34 1 2014 SEC 35.2–36.5 2.03 ± 0.03 0.94 9.10 ± 0.00 −8.50 ± 0.00 1.59 ± 0.00 498 NEC 35.5–36.0 4.38 ± 0.11 0.92 8.93 ± 0.00 −8.34 ± 0.00 1.59 ± 0.00 285 35 12 2014 SEC 35.8–36.4 3.89 ± 0.09 0.91 9.19 ± 0.00 −8.59 ± 0.00 1.60 ± 0.01 389 NEC 34.7–35.9 3.12 ± 0.02 0.99 8.90 ± 0.00 −8.31 ± 0.00 1.59 ± 0.00 401 J.S.P. Ibánhez et al. Science of the Total Environment 838 (2022) 156592 7 2017;Lefèvre et al., 2019) and strongly determines the seasonality observed. In between the SEC and the NEC and from June to November, the northward shift of the trade winds promotes the intensification of the NECC (Fonseca et al., 2004), which becomes visible in the eastward component of the surface velocity centered at about 5°-8°N along the tracks of the vessels used here (Fig. 2,JUN-NOV).BoththeCO 2 - oversaturated North Brazil Current (Lefèvre et al., 2017b), fed by the SEC, and the NEC can contribute to the water (and carbonate system properties) transported eastward by the NECC (Zhang et al., 2003). Due to the different origin of the waters transported by the surface current system in the study area, the NEC, NECC and SEC present different carbonate system properties reflected in significantly different fCO 2sw levels (see Supplementary materials). The high fCO 2sw observed in the 5°S-5°N latitudinal band characteristic of the SEC determines that the waters transported by this current are a permanent source of CO 2 to the atmosphere (i.e. positive ΔfCO 2 ,Fig. 3). The only exceptions are the areas affected by recent rainfall (Fig. 2), that lowers fCO 2sw becoming sporadically a sink of atmospheric CO 2 (e.g. Fig. 3 JUL). Further North, ΔfCO 2 in the NECC shows sea surface CO 2 undersaturation (Fig. 3, JUN-NOV). This is accompanied by a drop in SSS in the latitudinal band of the NECC (Fig. 2, JUN-NOV). During boreal summer, the ITCZ is at its northernmost location and the associated rainfall directly impacts surface waters transported by the NECC. During this period of the year, the North Brazil Current retroflection transports the Amazon River plume waters into the NECC and towards the central tropical Atlantic, spatially overlapping with the zonal freshening caused by the ITCZ (Ibánhez et al., 2016). Lefèvre et al. (2020) recently identified the presence of the Amazon plume as far from the river mouth as at a mooring placed at 8°N 38°W. Our underway determinations were performed further East, at about 29°W along the 8°N latitudinal band. Although the distance between the mooring and the underway determinations presented here is still considerable, the significant CO 2 drawdown created by the spread of the Amazon plume could be spread further East contributing together with the ITCZ to the observed CO 2 undersaturation within the NECC. Finally, in the North of the study area (from ~9–10°N; Fig. 3), ΔfCO 2 reflects a marked seasonal pattern in the NEC. From November to March, the NEC is a net sink of atmospheric CO 2 (i.e. negative ΔfCO 2 ;Fig. 3, NOV-MAR) coinciding with the lower annual SST levels (Fig. 2, NOV-MAR), while it acts as a net source of CO 2 to the atmosphere from June to October (i.e. positive ΔfCO 2 ; Fig. 3, JUN-OCT). 3.2. ITCZ impact on the carbonate system in the tropical Atlantic In the areas affected by recent rainfall and after removing the thermodynamic effect of SST over underway fCO 2sw (Eq. (8); hereafter termed fCO 2swnorm ), we found robust, highly significant linear relationships of ln fCO 2swnorm vs ln SSS (r >0.79, n from 217 to 837) in 33 of the 35 ships' voyages analyzed here. These correspond to SSS gradients varying from 0.4 to Fig. 4. Ln fCO 2swnorm vs ln SSS constructed with the underway data with interpolated, short-term rainfall higher than 1 mm, discriminated by surface current (SEC, NECC, NEC) and month. Color code shown in the figure legend is used to indicate year. During January 2009, June 2010, July 2011 and April 2013, significant linear relationships of ln fCO 2swnorm vs ln SSS were found in the northbound and southbound voyages. J.S.P. Ibánhez et al. Science of the Total Environment 838 (2022) 156592 8 2.5 salinity units and are found in the three main surface currents present in the study area (Table 2). Following the seasonal migration of the ITCZ, these significant linear relationships of ln fCO 2swnorm vs ln SSS are found mainly in the SEC from December to May and in the NECC from June to November (Fig. 4). The overall thermodynamic effect of salinity changes caused by rainfall over fCO 2sw ,γ rain , calculated from these linear regressions (Eq. (9)), vary from 1.25 ± 0.04 to 4.98 ± 0.05, with an average value of 3.12. With the exception of γ rain values obtained during the ships' voyages performed in February and April 2009, all the remaining γ rain are well above the γ rain reported from global means of the carbonate system sensitivity to rainfall at low latitudes (γ rain = 1.6; Sarmiento and Gruber, 2006). To evaluate the contribution of the different thermodynamic terms included in the overall γ rain , we calculate the sensitivity of underway fCO 2sw to changes in TAlk (γ TAlk )andDIC(γ DIC )(Eqs.(10),(11)). Calculated γ DIC and γ TAlk are highly consistent among the different ships' voyages, ranging from 8.67 ± 0.00 to 9.19 ± 0.00 (γ DIC ) and −8.59 ± 0.00 to −8.09 ± 0.00 (γ TAlk ;Table 2). The resulting theoretical γ rain calculated from γ s ,γ DIC and γ TAlk (Eq. (9)) remain systematically in the range of 1.58–1.60 (Table 2), i.e. the value proposed by Sarmiento and Gruber (2006) for low latitudes (1.6). These results suggest that the impact of the rainfall associated with the ITCZ over the surface water carbonate system in the tropical Atlantic cannot be explained solely by thermodynamics and chemical dilution of surface waters. 3.3. Sea-air CO 2 exchange in the tropical Atlantic affected by the ITCZ Despite the lowering in fCO 2sw caused by rainfall, the areas affected by the ITCZ in the SEC were permanent sources of CO 2 to the atmosphere, ranging from 0.30 ± 0.01 to 3.31 ± 0.04 mmol m −2 d −1 (Table 3). This contrasts with the resulting sea-air CO 2 fluxes found in the areas affected by the ITCZ in the NECC and the NEC, which commonly acted as sinks of atmospheric CO 2 . Nevertheless, the overall impact of the ITCZ over the sea-air CO 2 exchange in the tropical Atlantic is highly significant. CO 2 fluxes attributed to the impact of rainfall associated with the ITCZ range from −0.11 ± 0.01 mmol m −2 d −1 to −2.35 ± 0.08 mmol m −2 d −1 found in the SEC during May 2013 and in the NECC during August 2012, Table 3 Sea-air CO 2 exchange affected by the ITCZ along the tracks of the vessels used in this study, discriminated by surface current in the tropical Atlantic. Sea-air CO 2 flux corresponds to that determined from underway measurements (FCO 2sea-air ), ITCZ CO 2 drawdown (FCO 2ITCZ ) corresponds to the calculated overall impact of the ITCZ over the seaair CO 2 exchange, while unexplained CO 2 flux corresponds to that not explained by thermodynamics (FCO 2unknown ). The significant negative correlations between MODIS/ aqua, monthly Chl aand measured SSS and fCO 2sw normalized to temperature (fCO 2swnorm ) are also shown, together with the number of interpolated Chl avalues used (n Chl. a). ID corresponds to the identification of each VOS voyage used in Table 1. p-value is indicated with the following code: *** p <0.0001, ** p <0.001, * p <0.01, while absence of code indicates p <0.05. No significant (p >0.05) Chl. avs SSS and Chl. avs fCO 2swnorm relationships are denoted by empty fields. ID Month Year Current FCO 2sea-air FCO 2ITCZ FCO 2unknown n FCO 2 Chl. avs SSS (r) Chl. avs fCO 2swnorm (r) n Chl. a mmol m −2 d −1 mmol m −2 d −1 mmol m −2 d −1 1 7 2008 SEC 0.81 ± 0.03 −1.30 ± 0.02 −0.84 ± 0.01 558 −0.60*** −0.63*** 512 NECC 0.14 ± 0.06 −1.06 ± 0.06 −0.60 ± 0.04 412 2 10 2008 SEC 1.54 ± 0.09 −1.52 ± 0.07 −1.17 ± 0.05 289 −0.42*** −0.33*** 108 3 11 2008 SEC 0.78 ± 0.05 −1.02 ± 0.04 −0.56 ± 0.03 360 −0.26*** −0.26*** 224 4 12 2008 SEC 1.05 ± 0.05 −0.43 ± 0.02 −0.24 ± 0.01 427 −0.59*** −0.47*** 202 5 1 2009 SEC 0.96 ± 0.07 −1.64 ± 0.04 −0.99 ± 0.03 353 −0.70*** −0.72*** 138 6 1 2009 SEC 2.11 ± 0.06 −1.60 ± 0.09 −1.51 ± 0.05 687 −0.62*** 336 7 2 2009 SEC 0.99 ± 0.05 −0.26 ± 0.01 0.00 ± 0.00 494 −0.82*** −0.73*** 180 8 3 2009 SEC 1.98 ± 0.03 −0.44 ± 0.02 −0.08 ± 0.01 427 −0.15 225 9 4 2009 SEC 0.75 ± 0.01 −0.16 ± 0.01 0.00 ± 0.01 368 10 12 2009 SEC 3.31 ± 0.04 −1.32 ± 0.04 −0.89 ± 0.03 572 −0.85*** −0.81*** 145 NEC −0.44 ± 0.03 −0.67 ± 0.03 −0.30 ± 0.01 283 11 1 2010 SEC 1.60 ± 0.05 −0.63 ± 0.02 −0.34 ± 0.01 440 −0.77*** −0.81*** 163 12 2 2010 SEC 0.86 ± 0.01 −0.30 ± 0.02 −0.17 ± 0.01 318 0.21* 191 15 5 2010 SEC 1.57 ± 0.06 −0.63 ± 0.04 −0.25 ± 0.02 712 −0.66*** −0.72*** 566 16 6 2010 SEC 0.84 ± 0.03 −0.28 ± 0.01 −0.11 ± 0.01 806 −0.38*** −0.27*** 616 17 6 2010 SEC 2.23 ± 0.06 −0.60 ± 0.03 −0.30 ± 0.02 619 −0.65*** −0.60*** 503 NECC 0.05 ± 0.01 −0.42 ± 0.02 −0.12 ± 0.01 360 −0.54*** −0.22* 188 18 7 2010 SEC 1.84 ± 0.04 −1.16 ± 0.03 −0.56 ± 0.03 594 −0.66*** −0.62*** 560 NECC 0.12 ± 0.01 −0.30 ± 0.01 −0.11 ± 0.00 697 19 8 2010 NECC 0.21 ± 0.05 −1.98 ± 0.05 −0.80 ± 0.03 670 20 9 2010 NECC −0.20 ± 0.05 −1.95 ± 0.04 −0.90 ± 0.02 636 −0.14 200 21 10 2010 SEC 0.51 ± 0.08 −2.28 ± 0.08 −1.38 ± 0.06 305 −0.67*** −0.55*** 129 NECC −0.38 ± 0.06 −1.97 ± 0.06 −0.69 ± 0.03 500 −0.23*** 417 22 3 2011 SEC 1.71 ± 0.04 −0.56 ± 0.02 −0.17 ± 0.01 830 −0.77*** −0.53*** 379 23 4 2011 SEC 0.81 ± 0.02 −0.31 ± 0.01 −0.10 ± 0.01 464 −0.14 −0.43*** 282 24 5 2011 SEC 0.30 ± 0.01 −0.88 ± 0.03 −0.39 ± 0.01 837 −0.57*** −0.54*** 704 25 6 2011 SEC 1.60 ± 0.02 −1.07 ± 0.04 −0.49 ± 0.03 445 NECC 0.00 ± 0.02 −0.45 ± 0.01 −0.11 ± 0.01 242 −0.36*** −0.49*** 119 26 7 2011 NECC 0.00 ± 0.01 −0.27 ± 0.01 −0.09 ± 0.01 473 −0.62*** −0.30*** 226 27 7 2011 SEC 1.39 ± 0.03 −0.92 ± 0.06 −0.33 ± 0.03 303 −0.41*** 258 NECC −0.31 ± 0.03 −1.49 ± 0.04 −0.75 ± 0.03 481 28 6 2012 SEC 3.19 ± 0.09 −1.54 ± 0.06 −0.84 ± 0.04 482 −0.55*** −0.56*** 438 NECC −0.51 ± 0.04 −0.93 ± 0.04 −0.59 ± 0.02 217 29 7 2012 SEC 2.11 ± 0.08 −1.07 ± 0.04 −0.68 ± 0.03 383 NECC −0.42 ± 0.01 −0.37 ± 0.02 −0.13 ± 0.01 372 30 8 2012 NECC −0.92 ± 0.07 −2.35 ± 0.08 −1.45 ± 0.06 293 31 9 2012 NECC −0.16 ± 0.02 −0.56 ± 0.02 −0.23 ± 0.01 386 −0.51*** 233 32 4 2013 SEC 1.24 ± 0.04 −0.30 ± 0.04 −0.31 ± 0.02 515 33 4 2013 SEC 1.62 ± 0.04 −0.11 ± 0.02 −0.28 ± 0.01 585 −0.46*** −0.49*** 452 34 1 2014 SEC 0.91 ± 0.03 −0.29 ± 0.01 −0.06 ± 0.06 498 −0.76*** −0.78*** 372 NEC −0.72 ± 0.07 −1.02 ± 0.04 −0.65 ± 0.03 285 35 12 2014 SEC 2.75 ± 0.08 −0.77 ± 0.02 −0.47 ± 0.02 389 NEC −0.16 ± 0.02 −0.29 ± 0.02 −0.22 ± 0.01 401 J.S.P. Ibánhez et al. Science of the Total Environment 838 (2022) 156592 9