Study site
Located on the coast of Arraial do Cabo (Rio de Janeiro, Brazil, Fig. 1), the study site occurred between latitudes 22° 58' − 23° 06' South and longitudes 42 ° 10 '- 42 ° West. The Brazilian coast presents a NE-SW orientation in this region, causing the NE wind to blow parallel to the coast and the shallow coastal waters to move toward the open sea due to Ekman's transport19. Cabo Frio coastal upwelling phenomenon brings the deepest, coldest, and richest nutrient waters20.
Campos Basin is directly influenced by the South Atlantic Subtropical Anticyclone (SASA) 21, the main feature of the atmospheric circulation over the South Atlantic Ocean that affects the Brazilian weather and climate22. This feature is responsible for thermodynamic stability and low-intensity northeast winds that predominate in the southeast region of Brazil21. Teleconnection patterns (wave propagation in the upper atmosphere) can change the intensity and location of the South Atlantic subtropical high at any time of year, and changes in the positioning of the SASA generates a significant change in the wind pattern23.
Sun et al.24 analyzed data from 1979 to 2015 to compare the austral summer mean location of the SASA with the Southern Annular Mode (SAM)25 and extended multivariate El Niño–Southern Oscillation (ENSO) index (MEI)26. The authors found that when the SASA shifts poleward, the SAM is in a positive phase of La Niña, and when it shifts equatorward, the SAM is in a negative phase during El Niño. The wind speeds observed across Brazil can also describe this shift in the SASA pattern.
Three water masses occur at Arraial do Cabo from the shore to 500 m deep. The Tropical Water (TW) mass is found on the outer side of the continental shelf in the first 200 m and is characterized by a temperature range between 27.37ºC and 28.26ºC and a salinity range between 36.44 psu and 37.55 psu1,27,28. The South Atlantic Central Water (SACW) mass, ranging from 142 to 567 m in depth, is characterized by temperatures between 13.33ºC and 15.59°C and salinity between 35.41–35.78 psu1,27,29,30,31,20. The SACW is a water mass rich in nutrients that reach the surface through the coastal Cabo Frio upwelling phenomenon20,32. The Coastal Water (CW) mass is found in the inner part of the continental shelf33, resulting from the mixture of continental waters, TW, and SACW. The CW is characterized by temperatures above 23ºC and salinity below 34 psu34. The primary water flux in the region is the Brazil Current System (BC). This current acts until approximately 500 m deep, carrying the TW and the SACW towards the south31.
The climate in the study area is classified as hot-semiarid35, characterized by intense regional evaporation and reduced rainfall if compared to adjacent areas, yielding an arid climate36.
Wind velocity and surface temperature of the sea
Wind velocity and surface sea temperature (SST) data from the Cabo Frio (RJ) region, organized in a temporal series, were used to characterize the marine upwelling phenomenon. The daily averages of the wind velocity represented the NE (0° – 45° direction) and SW (180° – 270° direction) quadrants from September 2006 to December 2016. The data series of SST daily averages referred to the period between 1994 and 2016 (National Institute of Meteorology – www.inmet.gov.br). The data series of SST weekly averages referred to the period between 1994 and 2016 (data from the Admiral Paulo Moreira Marine Research Institute – IEAPM).
The daily average wind velocity from the NE and SW quadrants was tabulated with the respective SST (Sea Surface Temperature) averages from 2006 to 2016. In cases where the data of average wind velocity did not have corresponding SST data in the temporal series, the SST was estimated using the Kriging interpolation method. The estimate was obtained from the interpolation of the SST data in a regular grid, where: (X) corresponded to the time in days, (Y) to the time in years, and (Z) to the daily and annual SST37. A univariate statistic of residual values of SST resulting from the interpolation model was used to evaluate the standard error of the data and to indicate the function that best fits the input data. The Q-Q (quantile-quantile) relation was used to analyze the wind velocity, NE, and SW, compared to its respective SST38. The data were compared to those of the El Niño and La Niña periods checking the influence of these events on wind velocity and SST39. Trend curves were also obtained relating the average wind velocity with the SST. Pearson's R2 coefficient was used to assess data correlation.
Sampling design
The Diadorim R/V from the Brazilian Navy collected the water samples on January 20, 2016, using a Northeast/Southwest oriented transect. The first station (S1) was located between Cabo Frio Island and the continent, and the furthermost station (S4) was located 14 km from the coast, at the 100 meters isobath. (Table 1, Fig. 1).
Table 1
Geographic coordinates (datum WGS84), distance from the coast, and depth of the stations sampled in this study (S1, S2, S3, and S4).
Stations
|
Coordinates
(dd°mm'ss.ss")
|
Distance (km)
|
Depth (m)
|
S1
|
22° 59' 54.75"S
42°00'53.27"W
|
0.130
|
20
|
S2
|
23° 01' 43.05"S
42° 02' 49.22"W
|
4.84
|
75
|
S3
|
23° 03' 28.40"S
42° 04' 44.11"W
|
9.46
|
85
|
S4
|
23° 05' 20.56"S
42° 06' 42.17"W
|
14.29
|
100
|
The sampling campaign involved a two-scale analysis. The spatial scale was held perpendicular to the coastline (S1 to S4), Fig. 1. The ship was anchored in station 12 for 12 hours for the temporal scale. Both campaigns were performed on the same day. In the spatial approach, water was sampled from the surface (~ 3 m) between 6:30 h and 10:00 h (UTC) using a pump adapted to a hose without forming bubbles. The other water samples below the surface were collected through a Niskin bottle of 10 L for the middle (half of the total depth) and bottom (~ 5 m above the seafloor).
In the temporal approach, in station 2, the surface water samples were collected hourly (12 samples total), between 13:00 h to 01:00 h (UTC), while middle and bottom water samples were collected in alternate hours: 13, 15, 17, 21, 23, and 1 h.
Analytical methods
Water measurements, such as temperature, depth, and dissolved oxygen, were performed with a CTD vessel (Midas Valeport).
Thermodynamic modeling
The modeling and the procedural calibrations described in this manuscript were performed with the Marine Chemical Analysis (AQM) program1,2,16,17. The AQM is a package of thermodynamic equations executed via MS Excel, which can predict the complex composition of the marine carbonate system. This package is based on measurements that can be relatively inexpensive (pH, temperature, and alkalinity), reducing the overall costs of ocean acidification monitoring programs. The AQM program is available upon request to the corresponding author’s email.
Statistical procedure
Non-parametric Kruskal Wallis test was chosen for comparisons between groups. All statistical tests were performed using the Statistica 7.0 software (TIBCO) with a significance level set at p < 0.05.
Analytical procedure
The analytical method was based on international procedures for studies involving the chemistry of inorganic carbon dioxide in marine waters15,40,41 with the necessary adaptations.
Total alkalinity (TA)
For the determination of TA, water samples were collected and filtered in a Nalgene filtration system through GF/ F filters before being transferred to BOD-type flasks (300 mL, Kimble brand) and immediately analyzed40.
The potentiometric determination was conducted with duplicate samples in an open thermostated glass cell, where 3 mL (to obtain v1) and 10 mL (to obtain v2) of HCl 0.1 M were added to each 100 mL sample42. The method consists in determining the slope of the line by obtaining two points for the function of Gran (F): F (1), defined by v1, and F (2), defined by v242. A Thermo Scientific Orion Star potentiometer coupled to the Orion glass reference electrode cell, model 8102BNUWP was used for potentiometric determinations. The pH electrode was calibrated daily with "Tris" buffer (0.04 m) for sample readings (maximum 12 samples per day). Due to the reduced number of samples per day, the short period of the oceanographic cruise, and the constant working conditions (electricity source, solutions, and equipment), we chose to verify the electrode performance at the beginning and the end of the oceanographic cruise. The electrode's percent efficiency ranged between 99.49% and 99.54% concerning the theoretical Nernst value (59 mV). More details are available in hydrogen potential (pH).
The analytical precision and accuracy were calculated from five replicates of the reference material (Dickson–CRM, for oceanic CO₂ measurements, batch 104)43, which obtained a 95% recovery rate from the expected value (Table 2). The calculated TA was obtained by the AQM program through the equation: TA (µmol/kg) = 660 + 47.6S, defined by Hunter44 for waters of the Atlantic and Pacific oceans by the GEOSECS Program. The normalized total alkalinity (NTA) was obtained by the AQM program using the equation: NTA (µmol/kg) = TA (µmol/kg) x 35/Salinity (g/kg), where 35 was assumed to be the representative salinity of the water masses.
Table 2
Total alkalinity measured from five replicates of the certified reference material (Dickson, oceanic CO₂ measurements, batch 134).
|
Total alkalinity
|
Expected value
|
2222.61 µmol/kg
|
Measured (mean)
|
2108.0 µmol/kg
|
Absolute error
|
−115.0
|
Relative error
|
−5.0
|
Variation coefficient
|
1.66%
|
Sample volume
|
50 L
|
Hydrogen potential (pH)
The total pH of the water samples collected during the cruise was determined in the "wet laboratory" as follows: pHT (= -log([H+] + [(HSO4−]/co), where co is the thermodynamic concentration (1 mol/kg-soln).
The internal solution of the combined pH electrode was filled up with 0.7 m NaCl to reduce the potential liquid junction. The electrode's electromotive force (emf) was related to the molar concentration of the proton [H+], as shown in Eq. 2.
\(E={E}^{o}-\left(\frac{RT}{F}\right)lnln \left[{H}^{+}\right]\)
|
Equation 2
|
where: E° is the standard electrode potential, determined by titrating a 0.7 m NaCl solution with 0.179 M HCl45. The pHT (total scale) values were measured immediately after each collection at a constant temperature of 25 ℃ in a thermostatic cell connected to a microprocessed thermostatic bath with external circulation (Qimis) to avoid temperature bias46. The determinations were made by the Thermo Scientific Orion Star potentiometer coupled to the Orion glass reference electrode with a 0.7 m NaCl outer chamber filling solution, model 8102BNUWP. The analytical slope for the electrode was within ± 0.13 mV (theoretical Nernst value at 25 oC). The electrode was calibrated with a "Tris" buffer (0.04 m) prepared in the laboratory47, where pH values were assigned by spectrophotometry (m-cresol method) 16,40,48. The "Tris" buffer allows the accuracy of 0.001 pH units units47,49. Subsequently, using the AQM program, the pH results were corrected for the temperature recorded at the sampling moment (pHt = pH25 + A + Bt + Ct2)50.
Calcium (Ca) and total boron (TB)
The determination of Ca and TB was conducted using a MIP OES (microwave-induced plasma optical emission spectrometer, 4200 MP-AES, Agilent brand). The external analytical curves were made with monoelementary standards (1000 mg/L, VHG®) with concentrations ranging from 0.1 to 10 mg/L in an ultrapure water matrix. A matrix influence test was conducted in which it was found that both the boron and calcium signals did not show any significant difference between the ultrapure water matrices and the 500 mg/L NaCl solution. The calculated calcium and boric acid were also obtained using the equations: [Ca2+]T = 2.938x10− 4xS15, and [B]T = 0.000416x(S/35)51. The analyzed versus calculated values of Ca and TB in water samples showed a relative error (RE%) from 0.2–8% (Table 3).
Table 3
Calcium and boron concentrations (µmol/L) in seawater at three depths: surface (s), middle (m), and bottom (b) at the four sampling stations (S1, S2, S3, and S4). The relative error (RE%) assumes the analyzed value as the expected result.
Station
|
Ba µmol/L
|
Bc µmol/L
|
RE (%)
µmol/L
|
Caa µmol/L
|
Cac µmol/L
|
RE %
µmol/L
|
S1s
|
412
|
411
|
-0.24
|
11073
|
10618
|
-4.11
|
S1m
|
419
|
412
|
-1.68
|
11091
|
10584
|
-4.58
|
S1b
|
425
|
411
|
-3.29
|
11126
|
10777
|
-3.14
|
S2s
|
403
|
411
|
1.99
|
11073
|
10990
|
-0.75
|
S2m
|
410
|
411
|
0.24
|
11231
|
10857
|
-3.33
|
S2b
|
425
|
412
|
-3.06
|
11423
|
10614
|
-7.08
|
S3s
|
408
|
410
|
0.49
|
11248
|
10954
|
-2.61
|
S3m
|
412
|
411
|
-0.24
|
11300
|
10772
|
-4.67
|
S3b
|
418
|
412
|
-1.44
|
11493
|
10545
|
-8.25
|
S4s
|
406
|
410
|
0.99
|
11266
|
10842
|
-3.76
|
S4m
|
413
|
411
|
-0.48
|
11353
|
10938
|
-3.66
|
S4b
|
422
|
412
|
-2.37
|
11580
|
10792
|
-6.80
|
Ba = total analyzed boron; Bc = total calculated boron (AQM Program); Caa = total analyzed calcium; Cac = total calculated calcium (AQM Program).
Speciation and quantification of the carbonate system
All parameters from the inorganic CO2 system (CO2, \({CO}_{3}^{2-}\), \({HCO}_{3}^{-}\), DIC, ρCO2, ΩCalc, and ΩArag) were calculated using the carbonate system dissociation constant K 52 defined as follows:
1. \({lnk}_{B}^{*}\)53
2. \({lnk}_{Si}^{*}\)54
3. \({lnk}_{1}^{*}\)(H3PO4)55
4. \({lnk}_{2}^{*}\) (\({H}_{2}{PO}_{4}^{-}\))55
5. \({lnk}_{3}^{*}\) (\({HPO}_{4}^{2-}\))55
6. \({lnk}_{2}^{*}\) (\({CO}_{3}^{2-}\))56
The aqueous concentrations (CO2(aq)) and the partial pressure (ρCO2) were calculated from the variables of temperature, salinity, pH, and TA and by using the thermodynamic and stoichiometric constant K (\({pk}_{1}^{o}\), \({pk}_{2}^{o}\), \({pk}_{1}^{*}\), and \({pk}_{2}^{*}\)) 57,58. The AQM was also used in this phase, aiding the calculations.
Air-sea CO2 fluxes
The CO2 flux equation between oceans and the atmosphere is defined between aqueous CO2 and saturated CO2, defined as follows (Eq. 3):
\({F}_{{CO}_{2}}={k}_{T}\left[{CO}_{2water}-{CO}_{2saturated}\right]\)
|
Equation 3
|
Aqueous CO2 and saturated CO2 are components that characterize the "balance" of the flux equation (Eq. 3) determined in this study using the AQM program. Wind velocity has a significant effect on the gas transfer equation. The relationship between gas exchange and wind velocity can have non-linear effects on calculating the gas transfer velocities for particular wind velocity measurements, depending on the sampling design and the wind velocity1. The most critical parameter in the gases transfer velocity equation (Eq. 4) in terms of the function with wind velocity is based on the gaseous exchange coefficient (kT in cm/h):
\({k}_{T}=0.31\times {u}^{2}\times \left[\frac{{S}_{c}}{660}\right]-0.5\)
|
Equation 4
|
where: u is the wind velocity module at 10 m from the surface in m/s, Sc is the Schmidt number of CO2 in seawater59,60, and 660 is the Sc value in seawater at 20oC. The Schmidt number is defined as the water kinematic viscosity divided by a gas diffusion coefficient defined as follows (Eq. 5) by a polynomial:
\({S}_{C}=A-BT+C{T}^{2}-D{T}^{3}\)
|
Equation 5
|
The kT calculation (Eq. 4) considers that the wind velocity (u) has a fundamental quadratic dependence on the CO2 flux calculation. Typically, the wind velocity sampling for the kT calculation takes the climatological average into account at 10 m above the water surface61. However, such a procedure can include an error related to the average wind incompatibility and the CO2 measured in-situ. In addition, reading the wind speed on the sea surface (10 meters) can produce an inaccurate result due to the orographic average of the wind induced by the vegetation. Thus, the closer the wind measurement is performed to the water surface, the more significant the effect of wind on Sc will be2.