concentrations based on the BOUM dataset (Biogéochimie de l'Oligotrophie à l'Ultra-oligotrophie de la Méditerranée) collected in 2008.
This study compared two direct approaches, the TrOCA and the
MIX—optimum multiparameter MIXing analysis—to estimate the C
ant
accumulation at about 0.8–1.2 μmol.kg
−1
.yr
−1
. In 2013, the MedSea
dataset (Mediterranean Sea Acidification in A Changing Climate) allowed Hassoun et al. (2015a) to record C
ant
concentrations as high as
102 μmol/kg through the TrOCA approach. Such high values indicate
the acidification of the MS being between −0.15 and −0.05 pH units
(Hassoun et al., 2015a; Touratier and Goyet, 2011). This places the MS
at the top of the most acidified marine ecosystems (Touratier et al.,
2012). The C
ant
concentrations in the MS were also estimated using
either a high-resolution regional model (Palmiéri et al., 2015) or the
temporal variations of the stable carbon isotope ratio of dissolved inorganic carbon in the Levantine Basin (Sisma-Ventura et al., 2016).
The aforementioned studies highlighted the important role of the
western MS in trapping C
ant
, particularly, in the Algerian Basin (AB).
They promoted a better understanding of the basin's role in the regional
carbon cycle and C
ant
sequestration. The AB's complex mesoscale features (the Algerian Current, anticyclonic eddies, cyclonic gyres) are
essential for spreading the Modified Atlantic Waters (MAW) in the MS
and modulating the trophic regime of the AB. Nevertheless, the effects
of these physical structures on the carbonate system and its anthropogenic fraction have not been studied thoroughly (e.g., Moutin and
Prieur, 2012). This region has seldom been investigated by oceanographic cruises over the past few decades (e.g., the METEOR cruises in
October 2001 and April 2011, the BOUM in June 2008, and the MedSea
in May 2013). Additionally, these cruises have not covered important
areas of the AB.
This work benefits from a new dataset collected in the summer of
2014 that completely covers the AB. The SOMBA cruise (Système
d'Observations à la mer dans le Bassin Algérien) was conducted in the
context of MERMEX (Marine Ecosystems' Response in the
Mediterranean Experiment Program) that aims at studying the influence of global warming and anthropogenic activities on the marine
ecosystems of the MS (Durrieu de Madron et al., 2011). First, we addressed the spatial distribution of carbonate system parameters in the
AB by focusing on the east-west and north-south gradients. Second, the
C
ant
concentration was estimated using the TrOCA approach (Touratier
et al., 2007) in addition to a refitted Chen and Millero (1979) method
that was parametrized for preindustrial conditions using knowledge on
the MS carbonate system acquired over the last two decades. Finally,
the C
ant
concentrations obtained using the two methods were discussed,
emphasizing the role of mesoscale activity on the distribution of this
parameter. A first estimate of the amount of C
ant
sequestered by the AB
from the preindustrial period to the summer of 2014 was derived from
this new dataset.
2. Material and methods
2.1. SOMBA dataset
The SOMBA cruise, dedicated to studying the AB, was conducted
between August 14th and September 10th, 2014 on the French R/V
“Téthys II” (Mortier et al., 2014). This French-Algerian cruise had four
legs, including sampling at 70 hydrological stations along 7 sections
(Fig. 1). At each station, an underwater sampling system was lowered
from the surface to the bottom. The system included a SeaBird
SBE911+ CTD (Conductivity-Temperature-Depth) unit and a carousel
of 11 Niskin bottles (12 l each). The temperature and practical salinity
were measured with a precision of ± 0.002 °C and ± 0.003, respectively. The data were pretreated through manual check outs and spike
removal. The instrumentation errors were corrected by applying the
SBE Data Processing software. The sensors' drift was corrected based on
the manufacturer's recommendations, in addition to salinity measurements of discrete samples, taken directly from the Niskin bottles, using
an autosalinometer. The CTD was also interfaced with a dissolved
oxygen sensor (SBE43). Each station was sampled over eleven depth
levels. Discrete samples were collected for dissolved oxygen (12 stations), Dissolved Inorganic Carbon/Total Alkalinity (DIC/TA) (22 stations) and nutrients (70 stations).
Dissolved oxygen sensor responses were calibrated using daily
oxygen measurements performed by Winkler potentiometric titrations,
based on Langdon (2010)’s modified method. The sensor's calibration
coefficients were statistically adjusted by using multiple Winkler water
samples and sensor voltages over a wide range of oxygen calculations
(SBE, 2010). Thirty-two duplicates were sampled at different depths
with an estimated precision of 1.6 μmol/kg. Quality control checks
were performed for all the parameters, based on the recommendations
of the Global Ocean Ship-based Hydrographic Investigations Program
(Go-SHIP) (Swift, 2010). Subsequently, the data were flagged using
World Ocean Circulation Experiment (WOCE) standards. Only data
deemed “good” were selected for this study.
2.2. Geochemical parameters measurements
Carbonate system parameters were sampled in 500 ml borosilicate
glass vials (222 samples) and poisoned with 100 μl of a saturated
mercuric chloride solution (HgCl 2 ), based on Dickson et al. (2007)‘s
recommendations. Subsequently, the samples were analyzed for DIC
and TA at the SNAPO-CO 2 Laboratory (Service National d'Analyses des
Paramètres Océaniques du CO 2 , France) through closed-cell potentiometric titration, based on the procedure described by Edmond (1970).
The non-linear least squares procedure described in DOE (1994) was
used to determine equivalent points. Thirteen duplicates were homogeneously sampled over the cruise's time and space scales. Their repeatability was expressed using the short-term standard deviation—2 μmol/kg and 3.3 μmol/kg for TA and DIC, respectively. The
absolute differences (R) of the duplicate measurements did not exceed
the Upper Control Limit described by Dickson et al. (2007) –
UCL = 3.267 × R: 6.5 μmol/kg and 10.7 μmol/kg for TA and DIC,
respectively. Temporal drifts of the concentration of the diluted hydrochloric acid solution were corrected using Dickson Certified Reference Materials (CRM) provided by the University of California- San
Diego (batch 139: TA = 2250.8 ± 0.6 μmol/kg, DIC = 2023.2 ±
0.7 μmol/kg). Nutrient concentrations were determined by an automatic colorimetric procedure with a Technicon Auto Analyzer (Tréguer
and LeCorre, 1975) at the MIO Laboratory (Mediterranean Institute of
Oceanography- France). The precision of the nitrite, nitrate, soluble
reactive phosphate, and silicic acid measurements was 2%, 3–5%,
3–5%, and 5%, respectively while the detection limits were 0.03 μM,
0.05 μM, 0.02 μM, and 0.05 μM, respectively.
2.3. Carbonate system parameter calculations
Carbonate system properties were sampled at 22 stations of the
SOMBA cruise. We computed multiparametric linear regressions for TA
and DIC, versus potential temperature (θ), practical salinity (S P ), and
Apparent Oxygen Utilization (AOU) to extrapolate discrete TA and DIC
measurements over the entire basin. Researchers have already reported
a linear relationship between the TA and S p in the MS (Copin-Montégut
and Bégovic, 2002; Hassoun et al., 2015b). Nevertheless, this correlation depends on the region of the MS (e.g., Cossarini et al., 2015) and
seems stronger in the AB, characterized by fewer sources of TA variability than the eastern MS (the riverine and the Dardanelle inputs). Lee
et al. (2006) also showed that temperature can partly explain TA
variability.
DIC variability depends on both physical and biological processes
(e.g., Goyet and Davis, 1997) and was expressed against θ, S P , and AOU.
Several equations were tested for TA and DIC interpolations using the
SOMBA cruise data, either by accounting for the entire water column or
by dividing the dataset into three layers (surface, intermediate, and
M.A. Keraghel, et al.
Marine Chemistry 221 (2020) 103783
2
This study compared two direct approaches, the TrOCA and the
MIX—optimum multiparameter MIXing analysis—to estimate the C
ant
accumulation at about 0.8–1.2 μmol.kg
−1
.yr
−1
. In 2013, the MedSea
dataset (Mediterranean Sea Acidification in A Changing Climate) allowed Hassoun et al. (2015a) to record C
ant
concentrations as high as
102 μmol/kg through the TrOCA approach. Such high values indicate
the acidification of the MS being between −0.15 and −0.05 pH units
(Hassoun et al., 2015a; Touratier and Goyet, 2011). This places the MS
at the top of the most acidified marine ecosystems (Touratier et al.,
2012). The C
ant
concentrations in the MS were also estimated using
either a high-resolution regional model (Palmiéri et al., 2015) or the
temporal variations of the stable carbon isotope ratio of dissolved inorganic carbon in the Levantine Basin (Sisma-Ventura et al., 2016).
The aforementioned studies highlighted the important role of the
western MS in trapping C
ant
, particularly, in the Algerian Basin (AB).
They promoted a better understanding of the basin's role in the regional
carbon cycle and C
ant
sequestration. The AB's complex mesoscale features (the Algerian Current, anticyclonic eddies, cyclonic gyres) are
essential for spreading the Modified Atlantic Waters (MAW) in the MS
and modulating the trophic regime of the AB. Nevertheless, the effects
of these physical structures on the carbonate system and its anthropogenic fraction have not been studied thoroughly (e.g., Moutin and
Prieur, 2012). This region has seldom been investigated by oceanographic cruises over the past few decades (e.g., the METEOR cruises in
October 2001 and April 2011, the BOUM in June 2008, and the MedSea
in May 2013). Additionally, these cruises have not covered important
areas of the AB.
This work benefits from a new dataset collected in the summer of
2014 that completely covers the AB. The SOMBA cruise (Système
d'Observations à la mer dans le Bassin Algérien) was conducted in the
context of MERMEX (Marine Ecosystems' Response in the
Mediterranean Experiment Program) that aims at studying the influence of global warming and anthropogenic activities on the marine
ecosystems of the MS (Durrieu de Madron et al., 2011). First, we addressed the spatial distribution of carbonate system parameters in the
AB by focusing on the east-west and north-south gradients. Second, the
C
ant
concentration was estimated using the TrOCA approach (Touratier
et al., 2007) in addition to a refitted Chen and Millero (1979) method
that was parametrized for preindustrial conditions using knowledge on
the MS carbonate system acquired over the last two decades. Finally,
the C
ant
concentrations obtained using the two methods were discussed,
emphasizing the role of mesoscale activity on the distribution of this
parameter. A first estimate of the amount of C
ant
sequestered by the AB
from the preindustrial period to the summer of 2014 was derived from
this new dataset.
2. Material and methods
2.1. SOMBA dataset
The SOMBA cruise, dedicated to studying the AB, was conducted
between August 14th and September 10th, 2014 on the French R/V
“Téthys II” (Mortier et al., 2014). This French-Algerian cruise had four
legs, including sampling at 70 hydrological stations along 7 sections
(Fig. 1). At each station, an underwater sampling system was lowered
from the surface to the bottom. The system included a SeaBird
SBE911+ CTD (Conductivity-Temperature-Depth) unit and a carousel
of 11 Niskin bottles (12 l each). The temperature and practical salinity
were measured with a precision of ± 0.002 °C and ± 0.003, respectively. The data were pretreated through manual check outs and spike
removal. The instrumentation errors were corrected by applying the
SBE Data Processing software. The sensors' drift was corrected based on
the manufacturer's recommendations, in addition to salinity measurements of discrete samples, taken directly from the Niskin bottles, using
an autosalinometer. The CTD was also interfaced with a dissolved
oxygen sensor (SBE43). Each station was sampled over eleven depth
levels. Discrete samples were collected for dissolved oxygen (12 stations), Dissolved Inorganic Carbon/Total Alkalinity (DIC/TA) (22 stations) and nutrients (70 stations).
Dissolved oxygen sensor responses were calibrated using daily
oxygen measurements performed by Winkler potentiometric titrations,
based on Langdon (2010)’s modified method. The sensor's calibration
coefficients were statistically adjusted by using multiple Winkler water
samples and sensor voltages over a wide range of oxygen calculations
(SBE, 2010). Thirty-two duplicates were sampled at different depths
with an estimated precision of 1.6 μmol/kg. Quality control checks
were performed for all the parameters, based on the recommendations
of the Global Ocean Ship-based Hydrographic Investigations Program
(Go-SHIP) (Swift, 2010). Subsequently, the data were flagged using
World Ocean Circulation Experiment (WOCE) standards. Only data
deemed “good” were selected for this study.
2.2. Geochemical parameters measurements
Carbonate system parameters were sampled in 500 ml borosilicate
glass vials (222 samples) and poisoned with 100 μl of a saturated
mercuric chloride solution (HgCl 2 ), based on Dickson et al. (2007)‘s
recommendations. Subsequently, the samples were analyzed for DIC
and TA at the SNAPO-CO 2 Laboratory (Service National d'Analyses des
Paramètres Océaniques du CO 2 , France) through closed-cell potentiometric titration, based on the procedure described by Edmond (1970).
The non-linear least squares procedure described in DOE (1994) was
used to determine equivalent points. Thirteen duplicates were homogeneously sampled over the cruise's time and space scales. Their repeatability was expressed using the short-term standard deviation—2 μmol/kg and 3.3 μmol/kg for TA and DIC, respectively. The
absolute differences (R) of the duplicate measurements did not exceed
the Upper Control Limit described by Dickson et al. (2007) –
UCL = 3.267 × R: 6.5 μmol/kg and 10.7 μmol/kg for TA and DIC,
respectively. Temporal drifts of the concentration of the diluted hydrochloric acid solution were corrected using Dickson Certified Reference Materials (CRM) provided by the University of California- San
Diego (batch 139: TA = 2250.8 ± 0.6 μmol/kg, DIC = 2023.2 ±
0.7 μmol/kg). Nutrient concentrations were determined by an automatic colorimetric procedure with a Technicon Auto Analyzer (Tréguer
and LeCorre, 1975) at the MIO Laboratory (Mediterranean Institute of
Oceanography- France). The precision of the nitrite, nitrate, soluble
reactive phosphate, and silicic acid measurements was 2%, 3–5%,
3–5%, and 5%, respectively while the detection limits were 0.03 μM,
0.05 μM, 0.02 μM, and 0.05 μM, respectively.
2.3. Carbonate system parameter calculations
Carbonate system properties were sampled at 22 stations of the
SOMBA cruise. We computed multiparametric linear regressions for TA
and DIC, versus potential temperature (θ), practical salinity (S P ), and
Apparent Oxygen Utilization (AOU) to extrapolate discrete TA and DIC
measurements over the entire basin. Researchers have already reported
a linear relationship between the TA and S p in the MS (Copin-Montégut
and Bégovic, 2002; Hassoun et al., 2015b). Nevertheless, this correlation depends on the region of the MS (e.g., Cossarini et al., 2015) and
seems stronger in the AB, characterized by fewer sources of TA variability than the eastern MS (the riverine and the Dardanelle inputs). Lee
et al. (2006) also showed that temperature can partly explain TA
variability.
DIC variability depends on both physical and biological processes
(e.g., Goyet and Davis, 1997) and was expressed against θ, S P , and AOU.
Several equations were tested for TA and DIC interpolations using the
SOMBA cruise data, either by accounting for the entire water column or
by dividing the dataset into three layers (surface, intermediate, and
M.A. Keraghel, et al.
Marine Chemistry 221 (2020) 103783
2