Introduction
Antarctic continental shelves are viewed as strong anthropogenic CO2
sinks and therefore play an important role in global biogeochemical cycles
(Arrigo et al., 2008). These highly productive regions also support ecosystems
that are exposed to rapid environmental change
(Ducklow et al., 2007, 2012). Conditions along the
western shelf of the Antarctic Peninsula (WAP, Fig. 1) are characterized
by rapid ocean–atmosphere warming, sea-ice retreat and melting of glaciers
(Ducklow et al., 2012; Stammerjohn et al., 2012; Meredith et al., 2013), impacting
phytoplankton concentrations (Montes-Hugo et al., 2009) and higher
trophic level organisms such as krill, fish, and Adèlie penguins
(Ducklow et al., 2007, 2012; Schofield et al., 2010).
Climate and oceanographic trends are also mirrored in the inorganic carbon
dynamics, which could feed back to global carbon cycling and/or enhance the
projected fast progression of Southern Ocean acidification
(McNeil and Matear,
2008; Steinacher et al., 2009; Bopp et al., 2013), thereby imposing additional environmental stressors
on the ecosystem.
Map of the West Antarctic Peninsula (WAP) and study area of the
Palmer Antarctica Long Term Ecological Research (PAL-LTER) project. The red
box shows the main study grid that has been sampled for inorganic carbon
chemistry since 1993, and is defined in this study as the central
subregion. The black squares indicate the stations (20 km apart) arranged
in onshore to offshore lines spaced 100 km apart along the peninsula. The
inorganic carbon measurements from stations south of the central subregion
were only added in 2009. The central subregion also contains surface
underway pCO2 observations that were used in the trend analysis
(Sect. 3.5). P: Palmer Station on Anvers Island; A: Adelaide Island; and
MB: Marguerite Bay.
In the WAP, carbon biogeochemistry is controlled by an interplay of physical
and biological mechanisms, which include photosynthesis, respiration,
freshwater input, gas exchange, sea-ice cover, winds, and horizontal
advection (Carrillo and Karl, 1999; Carrillo et al., 2004; Wang et al., 2009;
Montes-Hugo et al., 2010). The physical oceanography of the region is strongly influenced by
equatorward flow at the continental shelf/slope break associated with the
eastward flowing Antarctic Circumpolar Current that abuts the continental
slope along the WAP region. On the shelf, there are indications of one or
more cyclonic circulation cells with poleward flow inshore (Hofmann et al., 1996; Dinniman and Klinck, 2004;
Martinson et al., 2008). Water mass properties are strongly influenced by subsurface
intrusions onto the continental shelf of warm, nutrient- and dissolved inorganic carbon (DIC)-rich Upper
Circumpolar Deep Water (UCDW), which appears to be modulated by topographic
depressions and canyons (Martinson et al., 2008; Dinniman et al., 2011; Martinson and McKee, 2012). In winter, respiration
processes and the entrained deep CO2-rich water increase the DIC
concentration in surface waters to supersaturated levels of CO2 with
respect to the atmosphere (Carrillo et al., 2004; Wang et al., 2009;
Tortell et al., 2014; Legge et al., 2015). From austral spring through summer, sea ice retreats from
north to south and from offshore to inshore (Smith and Stammerjohn, 2001). If not
counteracted by strong winds, freshwater from melting sea ice, glaciers, and
snow (Meredith et al., 2013) stabilizes the water column in close proximity
to the inshore and southward moving sea-ice edge. Stratification and
presumably iron availability provide favorable conditions for phytoplankton
blooms (Garibotti et al., 2003, 2005; Vernet et al., 2008), resulting in
a strong drawdown of DIC and flux of CO2
from the atmosphere into the ocean
(Carrillo et al., 2004; Montes-Hugo et al., 2009; Wang et al., 2009).
Subsequent iron depletion results in a decreasing trend of chlorophyll a (Chl a) from onshore to offshore, with interannual differences in the gradient
strength, depending on the onset of the sea-ice retreat
(Garibotti, 2005; Garibotti et al., 2005), but possibly also the timing of
sampling in relation to the timing of sea-ice retreat and phytoplankton
blooms.
The inorganic carbon dynamics are further complicated by large-scale
atmospheric patterns. The El Niño–Southern Oscillation (ENSO) and
Southern Annular Mode (SAM) drive the WAP climate and oceanography on
interannual to multidecadal timescales (Yuan and Martinson, 2001; Stammerjohn
et al., 2008a). During La Niña years, storms become longer and more
intense, temperatures increase, and sea-ice extent decreases in the WAP
region as a result of a strong low-pressure system driven by the poleward
displacement of the polar jet (Yuan, 2004). Positive SAM phases are also
associated with positive temperature anomalies over the Antarctic Peninsula
and decreased sea-ice extent (Kwok, 2002; Stammerjohn et al., 2008b).
Furthermore, the SAM brings the Southern Hemisphere westerly winds closer to
Antarctica, which amplifies the typical features of La Niña. During these
periods, nutrient and CO2-rich Circumpolar Deep Water intrudes more
frequently on to the shelf (Martinson et al., 2008), potentially increasing
[CO2] on the shelf. On the other hand, weaker and fewer storms and
spatial and temporal extension of sea-ice coverage are observed in negative
phases of SAM, with associated stronger stratification of the water column
and enhanced biological productivity (Saba et al., 2014). These features are
further intensified when a negative SAM coincides with El Niño
(Stammerjohn et al., 2008b).
The WAP oceanography and ecosystem have been intensely observed as part of
the PAL-LTER (Palmer Long Term Ecological Research) program
(http://pal.lternet.edu/) over the past two decades
(Ducklow et al., 2007, 2012). Since 1993, this
multifaceted data set has also contained seawater inorganic carbon measurements
taken each January along transects shown in Fig. 1. We complement the
summertime inorganic carbon measurements from PAL-LTER with surface underway
pCO2 measurements that cover all four seasons
(Takahashi et al., 2015). Here, we describe the spatial, seasonal, and
interannual variability in the inorganic carbon system over the past two
decades with the intention to improve our understanding of the main physical
and biological controls. Furthermore, such a uniquely long data set allows
us to gain first insights into the impacts of ocean acidification on the
region.
Data and methods
In situ data and calculation of carbonate system variables
We used discrete measurements of seawater DIC, total alkalinity (TA), and
nutrients collected during ship-based cruises as part of the PAL-LTER
program, along with temperature and salinity from CTD casts. The data were
gathered along the PAL-LTER sampling grid (Fig. 1), which runs 500 km
along the coast and 250 km across the shelf. The along-shelf transects were
spaced every 100 km, with 20 km between the stations. The data were
collected on an annual summertime cruise each January–February from 1993
through 2012. Carbon system sample collection and analysis were performed by
David Karl and Chris Carrillo for data prior to 2003, and by Hugh Ducklow
and Matthew Erickson for data from 2003 onward, with the exception that DIC
analysis was done by Taro Takahashi in 2003 and 2004. No TA data were
collected during 2003–2004.
Following the WOCE-JGOFS protocols, discrete samples of DIC and TA (300 mL)
from Niskin bottle casts were preserved with 200 µL of saturated
HgCl2 solution and sealed (Dickson and Goyet, 1994). DIC was analyzed by coulometric
determination of extracted CO2
(Johnson et al., 1987). TA was measured
with the potentiometric titration method. Certified reference materials
(provided by A. G. Dickson, Scripps Institution of Oceanography) were used to
assure internal consistency of data with a precision of
±2 µmol kg-1 for DIC and ±5 µmol kg-1 for TA. Water for
inorganic nutrient analysis was subsampled from Niskin bottles into acid-washed 50 mL Falcon tubes and frozen at -70 ∘C. The samples were
first analyzed using a Lachat Quickchem 8000 autoanalyzer at the University
of California at Santa Barbara Marine Science Institute Analytical Lab
(1993–2007) and later at the Marine Biological Laboratory (Woods Hole MA,
2008–2012). Inorganic nutrient data reach a precision of ±1 %.
All PAL-LTER data and a detailed description of the sampling methodology are
publicly available at http://pal.lternet.edu/ (dissolved
inorganic nutrients, PAL-LTER data set 27).
Calculated pH and saturation state for aragonite (Ωarag) were
determined from DIC, TA, temperature, salinity, phosphate, silicate, and
pressure using the CO2SYS MATLAB version (van Heuven et al., 2011). To
determine the carbonate variables we applied the dissociation constants for
carbonic acid by Dickson and Millero (1987; refit from
Mehrbach et al., 1973). The CO2 solubility equations of Weiss (1974) and dissociation constants for boric acid by
Dickson (1990) were also used to determine pH and Ωarag. pH is reported on the total H+ ion concentration scale
(pHT).
Comparison of deep-water (off-shelf) dissolved inorganic carbon
(DIC, µmol kg-1) and total alkalinity (TA,
µmol kg-1) data from Palmer Station Long Term Ecological
Research (PAL-LTER) with other available cruise data. (a) Station
locations and (b) DIC and (c) TA depth profiles from
PAL-LTER cruises (1998–2012), World Ocean Circulation Experiment (WOCE) and
Climate and Ocean – Variability, Predictability, and Change (CLIVAR) cruises
along parts of sections A21 (2006, 2009) and S4P (1992, 2011). The directly
measured parameters are listed in the parentheses and were used to calculate
TA if not directly measured.
The Lamont-Doherty Earth Observatory (LDEO) measured surface underway
pCO2 with a precision of ±0.5 %, together with salinity and
temperature in various seasons between 1999 and 2013, using a shower-type
water–gas equilibrator and infrared CO2 gas analyzer (see
www.ldeo.columbia.edu/pi/CO2 for the operational and engineering
details; Takahashi et al., 2015). A range of five standard gas mixtures
spanning between 100 and 700 ppm mole fraction CO2 certified by the
Earth System Research Laboratory of the National Oceanic and Atmospheric
Administration (NOAA) was used to calibrate the system every 4 h.
Comparison with deep-water WOCE/CLIVAR inorganic carbon system
data
We checked the consistency of the PAL-LTER DIC and TA data by comparing
PAL-LTER deep-water (> 2000 m), offshore TA, and DIC measurements
to deep-water data collected during the World Ocean Circulation Experiment
(WOCE) and Climate and Ocean – Variability, Predictability, and Change
(CLIVAR) cruises along parts of sections A21 and S4P that were overlapping
with the PAL-LTER grid (data available at
http://www.nodc.noaa.gov/woce/wdiu/). The WOCE and CLIVAR shipboard
measurements were calibrated using seawater certified reference materials
(prepared by A. G. Dickson, Scripps Institute of Oceanography), leading to
an estimated precision of ±2 µmol kg-1. DIC was measured on
all cruises. When necessary, TA was calculated from DIC and either fCO2
or pCO2 following the same procedure as described in Sect. 2.1.
Figure 2a shows the stations along the WAP where deep-water samples were
taken during PAL-LTER and WOCE cruises. PAL-LTER DIC and TA measurements
were within the range of sampled/calculated DIC and TA from the WOCE and
CLIVAR cruises (Fig. 2b and c). After removing five outliers, mean
deep-water DIC (DICmean= 2260.6 ± 3.8 µmol kg-1) and
TA (TAmean= 2365.4 ± 7.0 µmol kg-1) from PAL-LTER
cruises corresponded well with the data measured/calculated from WOCE
cruises (DICmean= 2261.8 ± 3.0 µmol kg-1;
TAmean= 2365.9 ± 9.3 µmol kg-1).
Comparison with surface underway pCO2 data
We also undertook a quality check of the PAL-LTER discrete surface DIC and
TA data (depth < 5 m) by comparing PAL-LTER pCO2, which was
calculated using observed DIC and TA values, to LDEO pCO2. LDEO
pCO2 samples that were collected during the PAL-LTER cruises were
spatially matched with the PAL-LTER-derived pCO2 values by choosing the
nearest latitude and longitude pair within a 1 km distance. Four PAL-LTER
pCO2 outliers that underestimate/overestimate pCO2 relative to the
underway observations by more than 150 µatm were removed. Analysis of
the corrected data set with a linear regression type II model suggests a
correlation of r= 0.82 (Fig. A1 in the Appendix, Table 1). Some of the observed
discrepancies may be attributed to errors in matching the times of bottle
samples with those of underway pCO2 measurements. Seawater inorganic
carbon chemistry is highly variable along the WAP due to the influence of
productivity, respiration, freshwater and upwelling of CO2-rich
subsurface water (Carrillo et al., 2004). Small matching errors may
therefore introduce small DIC and TA offsets, which would translate into
larger fractional differences in pCO2 due to the large Revelle factor
(∂ ln pCO2/ ∂ ln DIC) common in the region
(Sarmiento and Gruber, 2006).
Comparison of Lamont-Doherty Earth Observatory of Columbia
University (LDEO) underway pCO2 (µatm) data (Takahashi et al., 2015) with the
pCO2 (µatm) derived from PAL-LTER discrete surface samples over
the Palmer-Long Term Ecological Research (PAL-LTER) sampling grid. The
PAL-LTER discrete pCO2 sample values were computed using the dissolved
inorganic carbon (DIC, µmol kg-1) and total alkalinity (TA, µmol kg-1). The analysis is based on the data after removing outliers
as explained in the text.
Mean (SD)
r
Slope
Intercept
n
2005
LDEO
293 (79)
0.94
1.05 (±0.06)
-45.7 (±17.0)
49
PAL-LTER
322 (75)
2006
LDEO
248 (46)
0.90
0.95 (±0.06)
13.2 (±15)
55
PAL-LTER
248 (48)
2007
LDEO
261 (61)
0.87
1.04 (±0.08)
14.7 (±18.5)
60
PAL-LTER
237 (59)
2008
LDEO
340 (28)
0.53
0.61 (±0.14)
158 (±42.5)
48
PAL-LTER
299 (37)
2009
LDEO
318 (24)
0.58
0.47 (±0.13)
179 (±37.9)
27
PAL-LTER
292 (37)
2010
LDEO
327 (35)
0.54
1.62 (±0.57)
-167 (±174)
20
PAL-LTER
305 (27)
2011
LDEO
226 (98)
0.93
0.97 (±0.9)
0.60 (±21.4)
21
PAL-LTER
233 (101)
2012
LDEO
354 (36)
0.46
1.44 (±0.63)
-47.7 (±172)
21
PAL-LTER
279 (30)
All
LDEO
290 (69)
0.82
1.08 (±0.04)
-5.57 (±12.2)
300
PAL-LTER
275 (65)
Results
Here, we examine the observed spatial summer patterns of DIC, TA, pHT, and
Ωarag along the WAP and explore the underlying biological and
physical drivers. We then discuss regional carbon–nutrient drawdown
ratios and present our seasonal Ωarag predictions that give
initial insights into the chemical environment in the more poorly sampled
spring, fall, and winter months. Finally, using the LTER and LDEO data sets,
we investigate temporal trends over the past two decades.
Spatial summertime patterns of the inorganic carbon system
Surface waters in the PAL-LTER region exhibited high spatial and interannual
variability in DIC (min = 1850 and max = 2173 µmol kg-1), TA
(min = 2087 and max = 2396 µmol kg-1), and salinity (min = 30.3 and max = 33.9) across the
shelf. As a result, surface Ωarag reached levels as low as 0.98
in 1996, while maximum Ωarag values were > 3 in
several years (Fig. 3). Offshore, DIC (min = 2072 and max = 2255 µmol kg-1), TA (2265 and
2355 µmol kg-1), and salinity (min = 33.4 and max = 34) were
less variable, resulting in a smaller Ωarag range (min = 1.14
and max = 2.41). Additional aragonite undersaturation was detected between
100 and 200 m depth in 2005 and 2007 (Fig. 3). At depths > 70 m, which is below the mixed layer depth, Ωarag was < 1.5 in all years.
Depth profiles of aragonite saturation state (Ωarag)
for the years 1993 through 2012. The aragonite saturation horizon for each
year is located where the profile crosses the red line
(Ωarag= 1.0).
Maps of summertime averages of surface (a) pCO2, (b) pHT,
(c) aragonite saturation state (Ωarag), (d) total alkalinity (TA,
µmol kg-1), (e) salinity, (f) dissolved inorganic carbon (DIC, µmol kg-1), and (g) salinity-normalized
DIC (sDIC, µmol kg-1)
across years with available DIC and TA measurements (1993–1999, 2001–2002,
and 2005–2012). Salinity-normalized PO43- (sPO43- µmol kg-1) and salinity-normalized NO3-
(sNO3-, µmol kg-1) were averaged across 1993–1996, 1999,
and 2001–2012. Averages are only shown for regions where samples were taken
in five or more years. Occupied stations are shown by black dots.
To gain a spatial overview of the general summertime surface features (upper
5 m), we linearly interpolated the observations in space and averaged across
years with available DIC and TA (or nutrient) measurements. Averages are only
shown for regions where samples were taken in more than 5 years (Fig. 4). The
resulting pCO2, pHT, Ωarag, TA, salinity, DIC, and
nutrient fields exhibited clear onshore–offshore gradients. With the
exception of DIC, all variables also followed a north–south gradient. Mean
summertime surface pCO2 was lowest (< 200 µatm) in
the southern coastal region and was about 60 to 70 µatm lower than
in the northern near-shore regions (Fig. 4a). The highest mean summertime
pCO2 values were found in the northern slope region
(300–325 µatm). The opposite pattern was reflected in
Ωarag and pHT, with highest values
(Ωaragmax= 2.6 and pHTmax 8.3) close to the
coast and south of 66.5∘ S (Fig. 4b and c), decreasing along the
coast towards the north to pHT ∼ 8.2 and Ωarag
∼ 1.9, and reaching the lowest levels in northern offshore waters
(pHTmin= 8.1; Ωaragmin= 1.7). TA also
exhibited north–south and onshore–offshore gradients, with values as low as
2185 µmol kg-1 in the northern near-shore regions and as high
as > 2300 µmol kg-1 offshore. The low TA values
along the northern part of the coast coincided with the lowest salinity
values of 31.8, suggesting dilution of TA due to freshwater input (Fig. 4d
and e). Higher TA values offshore were also reflected in increased DIC and
salinity concentrations, with temperatures between 1.3 and 1.5 ∘C.
DIC also exhibited an onshore–offshore gradient with values about 80 to
100 µmol kg-1 lower in the near-shore region compared to
offshore, but there was no significant north–south gradient despite the
presence of freshwater in the north (Fig. 4f). Salinity-normalized DIC (sDIC,
normalized with UCDW salinity = 34.7) was lowest in the southern region,
thereby indicating that biological processes likely counteracted the expected
north–south DIC gradient due to the pronounced freshwater influence on DIC
in the north (Fig. 4g).
Physical and biological drivers of the inorganic carbon system
In this section we examine the physical and biological mechanisms that
control the observed variability in DIC and TA. DIC can decrease (increase)
through dilution with freshwater (evaporation), organic matter production
(remineralization), CO2 outgassing to the atmosphere (CO2 uptake),
and/or precipitation of CaCO3 (dissolution). While positive net
community production decreases DIC, the biological effect of organic matter
production on TA depends on the source of nitrogen, where nitrate consumption
increases TA and ammonium consumption decreases TA (Goldman and Brewer,
1980). Since nitrate is more abundant than ammonium in WAP surface waters
(Serebrennikova and Fanning, 2004), nitrate was assumed as the nitrogen
source. With a Redfield stoichiometry of 6.6 mol C mol N-1, TA
should increase by 1/6.6=+0.15 µmol TA per µmol DIC
consumed. Precipitation of biological CaCO3 material reduces both DIC
and TA, with the effect on TA twice as large as that on DIC
(2 µmol µmol-1). TA is not affected by gas exchange
but does vary as a result of dilution and evaporation.
Indications of surface reductions in TA and DIC due to freshwater input are
evident along the WAP, and therefore freshwater processes (sea-ice and
glacial melt, precipitation; Meredith et al., 2013) appear to be important
factors influencing the summertime carbon dynamics along the WAP. Figure 5
shows TA (circles) and DIC (diamonds) as a function of salinity. The black
lines represent the dilution lines for TA and DIC, which were calculated
following Yamamoto-Kawai et al. (2009). UCDW end members are based on
average TA and DIC concentrations in the water mass identified as UCDW
(black frames; Martinson et al., 2008). Upper-ocean TA follows its dilution
line closely, with stronger positive deviations of about 35 µmol kg-1 on average. In contrast, DIC values fall considerably below the
dilution line. A DIC drawdown of about 60 µmol kg-1 is visible
in the winter water (grey diamonds), which increased to more than 200 µmol kg-1 in the mixed layer, leading to Ωarag as low as
1.5 and as high as 3.9.
Scatterplots of dissolved inorganic carbon (DIC,
µmol kg-1) illustrated as diamonds and total alkalinity (TA,
µmol kg-1) illustrated as dots as a function of salinity. The
data points are color-coded by the aragonite saturation state
(Ωarag). The solid lines illustrate the dilution lines using
S= 34.7, TA = 2350 µmol kg-1, and
DIC = 2253 µmol kg-1 as end members for UCDW, and S= 0, TA = 300 µmol kg-1, and
DIC = 300 µmol kg-1 as end members for meltwater
(Yamamoto-Kawai et al., 2009). WW: winter water
(T ≤ -1.2 ∘C; 33.85 ≤ S ≤ 34.13); UCDW:
Upper Circumpolar Deep Water (1.7 ∘C ≥ T ≤ 2.13 ∘C; 34.54 ≤ S ≤ 34.75) following
(Martinson et al., 2008).
The DIC drawdown relative to the salinity mixing–dilution line is most
likely due to biological production of organic matter. Figure 6 shows sDIC
as a function of salinity-normalized TA (sTA) for waters shallower than UCDW
(orange dots). The regression line (solid black line, sTA = -0.11 × sDIC
+ 2601, RMSE = 18.6) ± 2σ (dashed lines) for estimated
measurement precision (σ=±5 µmol kg-1) is similar
to the nitrate-based photosynthesis line (blue line), indicating that the
large decrease in DIC with the concomitant smaller increase in TA was mainly
due to net biological production of organic matter. The photosynthesis line
is based on winter water (WW) DIC and TA end members (blue dots) and a slope
of -1/6.2. According to the Redfield ratios
(C / N / P = 106:16:1;
Redfield, 1958), photosynthetic utilization of 1 mol of NO3 increases TA
by 1 µmol kg-1
(Wolf-Gladrow et al., 2007) and decreases DIC by
106/16 (6.6). However, since the TA titration was performed to a pHT of
about 3, the TA values include residual PO4-3, which leads to this
slightly shallower slope of 6.2.
Salinity-normalized total alkalinity (sTA, µmol kg-1)
as a function of salinity-normalized dissolved inorganic carbon (sDIC,
µmol kg-1) for waters shallower than the Upper Circumpolar
Deep Water (UCDW, orange circles). A linear fit between sTA and sDIC is shown
by the black solid line. The dotted black lines indicate 2σ for
estimated measurement precision of σ=±5 µmol kg-1.
The blue line illustrates the trend if sTA and sDIC of the winter water (WW)
were only influenced by photosynthesis (1:-6.2). Grey dots represent sTA as
a function of sDIC corrected for gas exchange in the waters above the WW, and
the linear fits with the estimated measurement precision are the grey solid
and dashed lines respectively. WW: T ≤ -1.2 ∘C,
33.85 ≤ S ≤ 34.13; UCDW:
1.7 ∘C ≥ T ≤ 2.13 ∘C, 34.54 ≤ S ≤ 34.75, following (Martinson et al., 2008).
The intense, biologically driven DIC drawdown and resulting pCO2
undersaturation in the mixed layer may have led to some CO2 uptake from
the atmosphere that tends to reduce the apparent DIC deficit; thus the
estimated biological drawdown from observed DIC values in Fig. 6 may be
underestimated and needs to be corrected for air–sea CO2 gas exchange
from the period of biological drawdown to the sampling time. To account for
DIC concentration changes due to gas exchange with the atmosphere, we
assumed a constant atmospheric concentration of 390 µatm between 1993
and 2012, and a gas transfer rate (k) of 5 (±1) mmol CO2 m-2 µatm-1 month-1, which is the estimated mean rate for
the Southern Ocean area south of 62∘ S (Takahashi et al., 2009).
The change in DIC (µmol kg-1 month-1) due to gas transfer
into the mixed layer (ML) of d meters depth is
ΔDIC=k×Δt×ΔpCO2/d.
ΔpCO2 (pCO2atm–pCO2ML) was between -143 and 312 µatm, as pCO2ML ranged from 533
to 78 µatm, indicating that there was potential for both oceanic
CO2 uptake and outgassing. Assuming that d= 50 m
(Ducklow et al., 2013), we estimate that the monthly ΔDIC due
to air–sea CO2 gas exchange was in the range of -14 to 31 µmol kg-1 month-1. Since the first large phytoplankton blooms generally
occur after the sea ice retreats in November (Δt∼ 3 months), we assume that by the time of sampling at the end of January,
ΔDIC would fall in the range -43 to 94 µmol kg-1. The DIC
corrected for gas exchange is illustrated as grey dots in Fig. 6. While
applying the gas exchange correction flattens the regression line (grey
line) somewhat, the photosynthesis line (blue) still remains within the
estimated error bounds of the gas-exchange-corrected regression line (grey
dotted lines), further emphasizing that photosynthesis is the key biological
driver of the summertime carbonate system west of the Antarctic Peninsula.
Nutrient vs. carbon drawdown
Ocean carbon, nitrogen and phosphorus cycles are governed by organic matter
production and subsequent remineralization and are strongly correlated on a
global average with the proportions C / N / P = 106:16:1
(Redfield, 1958). Our findings suggest that the carbon–nutrient cycles along
the WAP depart from the standard Redfield values (Fig. 7). In a few samples,
the standing stock of PO43- became depleted before NO3-, and
overall the regression indicates a low N : P ratio of 9.8 ± 0.4 in
the mixed layer (Fig. 7a, black) and N : P = 11.7 ± 0.3 for all
data (dark grey) relative to the standard Redfield value of
16 mol N mol P-1. The mole / mole C : P ratio was also
considerably smaller than the Redfield ratio (Fig. 7b). C : P yielded
43.1 ± 2.3 in the mixed layer and 55.0 ± 1.7 for all data.
However, after applying the gas exchange correction on DIC (see Sect. 2), the
C : P ratio shifted closer to the Redfield ratio and resulted in a value of
80.5 ± 2.5 (light-grey dots and lines). Correcting the DIC for gas
exchange shifted the C : N molar ratio from 4.5 ± 0.2 (mixed layer
depth) and 4.7 ± 0.1 (all data) to 6.7 ± 0.2 and resulted in a
Redfield-like C : N ratio.
Plot of salinity-normalized nutrients and dissolved inorganic carbon
(sDIC, µmol kg-1), (a) sPO43-
(µmol kg-1) vs. sNO3- (µmol kg-1),
(b) sPO43- vs. sDIC, and (c) sNO3- vs.
sDIC. Observations within the mixed layer (depth < 50 m) are
illustrated by black circles. The light-grey dots in (b) and
(c) show sDIC corrected for gas exchange as a function of
sPO43- and sNO3-, respectively. A linear fit is
represented by the solid black line for the mixed layer, by the solid grey
line for all data, and by the light-grey line for the gas-exchange-corrected
sDIC in (b) and (c). The dashed black lines are the
nutrient drawdown lines using the corresponding Redfield ratio and data from
the Upper Circumpolar Deep Water (UCDW) as end members.
Seasonal variability
To get insights into the carbon dynamics during winter, spring, and fall,
when direct measurements of DIC, TA, and nutrients are either scarce or not
available, we developed a regional TA algorithm (based on PAL-LTER summertime
data). In combination with seasonal LDEO pCO2, salinity, and
temperature data, we calculated Ωarag for the missing seasons.
Due to the weak correlation between PAL-LTER temperature and TA (r= 0.50), we based the TA algorithm on salinity only (Fig. A2, r= 0.88).
Applying the Akaike information criterion (Burnham and Anderson, 2002), we
determined that TA along the WAP will be best represented by a first-order
linear model. We then randomly divided the PAL-LTER surface measurements
(depth < 5 m) into 10 data subsets using the 10-fold
cross-validation method (Stone, 1974; Breiman, 1996). Using 9 of the 10 data
sets we derived a regression model, predicted the TA with the model, and
calculated the model coefficients and root mean square errors (RMSEs). We
repeated these steps so that every data subset was left out once. The
coefficients for the final model were calculated from the mean of the ten
regression coefficients. We found the best fit to predict TA
(TApred, µmol kg-1) in the following equation:
TApred=57.01(±0.88)×S+373.86(±35.26),
which resulted in a linear correlation coefficient of r= 0.88 and a RMSE
of 15.2 µmol kg-1 (Fig. A2). In combination with the
pCO2 measurement precision of 3 µatm, the RMSE of TA prediction
resulted in a mean error in calculated Ωarag of 0.0219 units
and pHT of 0.0043 (Glover et al., 2011). Note that the calculated Ωarag and pHT estimates implicitly require that the approximately
linear summertime TA–salinity relationship hold for the other seasons, a
reasonable assumption if dilution and mixing substantially affect TA
patterns.
Summertime LDEO underway pCO2 values were, on average, lower than
during the rest of the year (Fig. 8a). While only a small percentage of
these summertime values reached levels higher than the atmospheric CO2
concentration, 70 % of the water samples taken in winter were
supersaturated with regard to atmospheric CO2 (> 390 µatm). Spring and fall pCO2 values were also generally higher than
summertime measurements and ranged from 207 to 506 µatm and 90 to 414 µatm.
Seasonal variability in the inorganic carbon system. Relative
frequency distribution of (a) measured underway surface partial
pressure pCO2 (µatm), (b) predicted surface total
alkalinity (TA, µmol kg-1) from underway salinity, and
(c) predicted surface aragonite saturation state
(Ωarag) in summer (red), fall (orange), winter (blue), and
spring (yellow). The x axis represents the range of Ωarag,
TA, and pCO2 with a relative frequency distribution ≥0.0001.
Our salinity-based algorithm predicted the majority of all TA ranging
between 2200 and 2300 µmol kg-1 in all seasons, with the most
frequent occurrence of highest TA in winter and spring (Fig. 8b). Some
summertime TA was predicted to be as low as 2056 µmol kg-1.
Prediction of seasonal Ωarag revealed that surface waters of
the WAP were supersaturated with regard to aragonite throughout the years
(Fig. 8c). The most frequent occurrence of low Ωarag was in
winter and spring, when most of the predicted values resulted in
Ωarag < 1.4. 20 % of spring and winter values
were Ωarag < 1.2, with the lowest predicted surface
Ωarag reaching near aragonite undersaturation in winter.
Similar to the LTER observations, predicted summertime Ωarag
displayed a large range, spanning from 1.1 to 4.1, with the majority of
predictions between 1.3 and 1.8. Biological production in summer is
sufficiently intense to prevent low Ωarag values during the
active growing season, when its effects might
be most pronounced.
Temporal trends
Trend analysis of the PAL-LTER data showed no statistically significant
annual trends (at the 95 % confidence level) in the measured carbon
parameters, temperature, or salinity in surface waters in summer between 1993
and 2012 (Table 2). As a comparison, we conducted a trend analysis for the
LDEO surface underway pCO2 data set (1999–2013) in the same region.
LDEO observations show an increasing, but not statistically significant,
trend in surface pCO2, supporting our results above (Table 3). The
largest increasing trend was found in fall
(1.9 ± 0.95 µatm yr-1), but this trend was also slightly outside the confidence interval
and therefore statistically not significant.
Mean annual trend (1993–2012) of Palmer-Long Term Ecological
Research (PAL-LTER) surface (depth < 5 m) carbonate chemistry and
hydrography from the West Antarctic Peninsula (central subregion). Regression
statistics include the mean annual rate (yr-1), standard error (SE),
number of measurements (NM), number of years (NY), r-squared, and p value
for aragonite saturation state (Ωarag), pHT, dissolved
inorganic carbon (DIC, µmol kg-1), total alkalinity (TA,
µmol kg-1), temperature (∘C), and salinity. Trends
with a p value < 0.05 would be considered statistically significant at the
95 % confidence level. Points
that were outliers at 95 % probability level were excluded (o).
Parameter
Rate (yr-1) ± SE
NM(o)
NY
r2
p value
Surface (< 5 m depth)
Ωarag
0.001 ± 0.01
892 (17)
18
0.04
0.9127
pHT
0.002 ± 0.002
892 (8)
18
0.03
0.2784
DIC (µmol kg-1)
-0.18 ± 1.03
907 (0)
18
0.00
0.8677
TA (µmol kg-1)
0.58 ± 0.63
907 (0)
18
0.05
0.3681
Temperature (∘C)
-0.01 ± 0.02
1076 (8)
20
0.01
0.4629
Salinity
0.01 ± 0.01
1060 (8)
20
0.12
0.1349
Discussion
The 20-year-long PAL-LTER seawater inorganic carbon time series showed a
distinct upper-ocean spatial pattern of onshore–offshore and north–south
gradients and suggests that the summertime carbon dynamics are primarily
controlled by biological productivity and freshwater input in near-shore
areas.
Surface Ωarag was distributed across a wide range (< 1 to values > 3) in freshwater-influenced areas with salinities
S < 32 (Fig. 5). To better understand how such a wide range of
Ωarag at relatively low salinities was possible, we quantified
the effect of freshwater and biological production. Mixing of seawater with
sea-ice or glacial meltwater leads to a “dilution” of CO32- ions
and a decrease in Ωarag because TA and DIC in glacial and
sea-ice meltwater are much lower than in seawater
(Anderson et al., 2000; Yamamoto-Kawai et al., 2009). Calculations of
salinity-normalized Ωarag using sDIC and sTA showed that freshwater
input decreased Ωarag by up to 0.2 units along the coast.
Despite the negative effect of freshwater on Ωarag, the water
in the south was nonetheless highly supersaturated with CaCO3. The
salinity-normalized DIC in the near-shore southern region of the PAL-LTER
sampling grid was up to 177 µmol kg-1 lower than elsewhere,
suggesting that near-shore phytoplankton blooms balanced out the negative
effect of freshwater on Ωarag and even increased Ωarag by up to 2 units. In 2005, when the above-described pattern was
particularly conspicuous, high Chl a (up to 20 µg L-1) in the southern
coastal area of the sampling grid provides further evidence that high
primary productivity led to the observed high Ωarag despite the
presence of freshwater. Similar results were found after the calving event
of the Mertz Glacier tongue in eastern Antarctica, where enhanced primary
productivity increased the Ωarag and thereby counteracted the
effect of dilution by meltwater input (Shadwick et al., 2013).
Trend analysis (1999–2013) of Lamont-Doherty Earth Observatory of
Columbia University (LDEO) surface continuous underway pCO2 (µatm), salinity, and temperature (∘C) measurements from within the
central subregion of the Palmer-Long Term Ecological Research (PAL-LTER)
sampling grid (Fig. 1, red box). Regression statistics include mean rate,
standard error (SE), number of measurements (NM), number of years (NY),
r-squared, and p value. Trends with a p value < 0.05 would be
considered statistically significant at the 95 % confidence level.
Parameter
Season
Rate ± SE
NM
NY
r2
p value
pCO2 (µatm yr-1)
Summer
1.45 ± 2.97
94 774
12
0.01
0.6361
Fall
1.90 ± 0.95
42 655
14
0.26
0.0685
Winter
0.43 ± 0.77
26 314
11
0.04
0.6304
Spring
1.22 ± 2.72
14 813
9
0.03
0.6678
Temperature (∘C yr-1)
Summer
0.03 ± 0.05
94 774
13
0.03
0.5515
Fall
0.00 ± 0.05
42 655
14
0.01
0.9279
Winter
0.00 ± 0.04
26 314
13
0.00
0.9262
Spring
0.01 ± 0.03
14 813
9
0.04
0.8598
Salinity (yr-1)
Summer
-0.02 ± 0.02
53 713
12
0.10
0.3294
Fall
0.02 ± 0.01
55 823
13
0.14
0.0988
Winter
-0.01 ± 0.01
28 063
10
0.01
0.6631
Spring
-0.01 ± 0.01
53 713
11
0.05
0.1422
Our findings of onshore–offshore and latitudinal gradients of carbon
parameters are supported by previous results that suggest similar patterns
for several physical and biogeochemical parameters. Summertime surface
temperature, salinity, and NO3-+ NO2- are generally lower
close to the coast, while Chl a, primary production, Si(OH)2, and
water column stability decrease from the coast toward the open ocean (Smith,
2001; Garibotti et al., 2003; Vernet et al., 2008). The freshwater along the
coast may originate, to a large part, from melting of glacial ice and snow
(Meredith et al., 2013). Such glacial and snow-melt plumes have been
correlated with increased primary production due to a stabilization of the
mixed layer, which creates favorable conditions for phytoplankton blooms
(Dierssen et al., 2002). This in turn is thought to be the dominant control
of the onshore–offshore gradient of phytoplankton variability and associated
biologically impacted parameters. The north–south gradients possibly reflect
the timing of phytoplankton blooms in the north and south. As such, blooms in
the north occur sooner than blooms in the south (Smith et al., 2008) – thus
on average the PAL-LTER January cruise takes place after the bloom in the
north, and during the blooms in the south. This may also be the reason for
the nutrient depletion along the coast, despite low biological productivity
at the time of sampling in the north (Fig. 4h and i). However, it is
important to note that, as a result of changes in ice cover, cloud formation,
and wind over the past 30 years, biological productivity has increased in the
southern part of the WAP and significantly decreased north of 63∘ S
(Montes-Hugo et al., 2009). The observed DIC drawdown in the winter water
(Figs. 5 and A3) may be a result of biological productivity, which is
supported by previous observations of Chl a maxima in the euphotic part of
the winter water, likely due to increased iron concentrations there
(Garibotti et al., 2003; Garibotti, 2005). However, it is also
possible that lateral advection or
vertical mixing of low-DIC water into the winter water has
contributed to this signal.
Low Ωarag values (< 1.35) observed offshore coincided
with surface waters supersaturated with regard to atmospheric CO2,
salinities > 33.5, and temperatures between 1.3 and 1.5 ∘C (not shown). These physical properties are associated with
modified UCDW, a mixture between UCDW and Antarctic Surface Water
(Smith et al., 1999), and indicate that upwelling of DIC- and TA-rich water
into the mixed layer may lead to lower Ωarag conditions
offshore (Carrillo et al., 2004).
The PAL-LTER data indicate N : P uptake ratios lower than the Redfield ratio
of 16:1, and uptake ratios similar to our findings (14:1) are common for the
polar region of the Southern Ocean (Weber and Deutsch, 2010;
Martiny et al., 2013). Our observed low ratio may be the result of a high abundance of
diatoms with low N / P ratios in this cold and nutrient-rich environment
(Arrigo, 1999; Arrigo, 2002; Green and Sambrotto, 2006; Martiny et al.,
2013). Rubin et al. (1998) observed a similar N / P ratio of
13.0 ± 1.2 in the mixed layer south of the polar front, and an even
lower N / P ratio of 11.3 ± 0.3 was observed in the iron-spiked
mixed layer during the iron fertilization experiment in the Subantarctic
South Pacific (Hales and Takahashi, 2012). Consistent with the low N / P
ratio, the observed C : P ratio (80.5 ± 2.5, corrected for gas
exchange) was also lower than the classic Redfield ratio. This indicates that
the regional phosphate cycle shows non-Redfield behavior, which is in
agreement with the observed C : P ratio of 91.4 ± 7.9 in the mixed
layer south of the polar front (Rubin et al., 1998). For the same region,
Rubin et al. (1998) describe Redfield behavior of C / N nutrient
utilization, which corresponds with our gas-exchange-corrected C / N
nutrient utilization ratio of 6.7 ± 0.2. Recently published work
suggests that C / N / P ratios exhibit a latitudinal pattern, with a
range of 66:11:1 to 74:13:1 at higher latitudes in the Southern Ocean
(Martiny et al., 2013) and can therefore be significantly lower than what we
found in this study.
TA variability was largely driven by dilution through freshwater input and
mixing (Fig. 5), which is well characterized by the salinity-derived TA
relationship presented in Sect. 3.4. However, biological mechanisms such
as photosynthesis, respiration, CaCO3 precipitation, and dissolution
also played an important role in controlling TA concentrations in the water
column and at the surface (Fig. 6). Neglecting these important drivers may
be responsible for the large RMSE of our predicted TA (Fig. A2) relative
to other studies that either had additional parameters at hand (i.e., O2
or nutrients) to derive inorganic carbon system parameters in coastal
environments
(Juranek et al.,
2009; Kim et al., 2010; Evans et al., 2013) or used salinity algorithms to predict TA in
open-ocean regions (Takahashi et al., 2014). Furthermore, TA varied by
more than 70 µmol kg-1 at salinities > 33.7, which led
to an unbalanced distribution of residuals (Fig. A2c). Increasing TA at
higher salinities and nearly constant DIC concentrations has been observed
before in Arctic and Antarctic regions
(Dieckmann et al., 2008; Fransson et al., 2011; Rysgaard et al., 2012; Shadwick et al., 2014; Legge et al., 2015) and may be due to
formation of ikaite crystals (CaCO3 6H2O; Suess et al., 1982) that store TA in sea ice and, upon melting,
release the excess TA into the surface water
(Rysgaard et al., 2012, 2013). However,
reasons for the observed increasing TA at higher salinities along the WAP
remain speculative, since direct evidence of ikaite formation/dissolution
such as an increase in DIC associated with TA increase is missing (Fig. 6). A combination of other mechanisms, such as upwelling of high-salinity, high-TA waters concomitant with biological DIC drawdown, could have
increased TA : DIC ratios at high salinities. Finally, the WAP region is very
dynamic, with large seasonal changes that may affect the carbon system in
ways not representable by one algorithm and may therefore require seasonally
adjusted algorithms.
Despite the above-described shortcomings in our salinity-derived TA
algorithm, the estimated Ωarag values give a useful overview of
the seasonal distribution and variability in Ωarag (Fig. 8). Error propagation of pCO2 measurement precision and TA prediction
accuracy suggests that the predicted error for Ωarag may be
as little as 0.02 (Glover et al., 2011). The seasonal estimations of Ωarag suggest that some winter and springtime Ωarag values were
near Ωarag= 1 and 20 % were between 1.0 and 1.2 (Fig. 8). Short-term exposure to low levels of Ωarag may cause severe
dissolution of live pteropod shells and has already been observed in the
Scotia Sea (Bednaršek et al., 2012). Surface aragonite
undersaturation along the WAP may be a result of ocean acidification and may
not have been common during preindustrial times (Hauri et al., 2015).
The large uncertainties in our estimated temporal trends are caused
inherently by the large spatial and temporal variability in our data.
Nevertheless, our mean rates of 1.45 ± 2.97 for summer and 0.43 ± 0.77 µatm yr-1 for winter suggest that the surface water pCO2
has been increasing at a slower rate than the atmospheric pCO2 rate of
about 1.9 µatm yr-1, and that the air–sea CO2 driving
potential has been increasing. Our results may be compared with the recent
analysis of the 2002–2015 time-series data obtained across the Drake Passage
by Munro et al. (2015). In the waters south of the polar front (their Zone 4,
closest to the LTER area), they observed that the surface water pCO2
increased at a rate of 1.30 ± 0.85 µatm yr-1 in summer and
0.67 ± 0.39 µatm yr-1 in winter, which are comparable with
ours along the WAP. We observed the strongest but still statistically
insignificant increase in surface pCO2 in fall (1.9 µatm yr-1, p= 0.0685). This increase corresponds with the mean
atmospheric pCO2 increase of 1.9 µatm per year, which causes a pHT
decrease of about 0.02 per decade (Takahashi et al., 2014).
Interestingly, Stammerjohn et al. (2008a, b) found that sea-ice extent and wind are also changing most rapidly in spring and fall, which
may enhance sea–air gas exchange and therefore facilitate positive pCO2
trends. Furthermore, it is likely that the strong counter-effect of
biological productivity successfully masks the pCO2 trend in summer,
and decreased gas exchange due to sea ice weakens the trend in winter.
However, the WAP climate and oceanography are regulated by large-scale
atmospheric patterns, such as El Niño–Southern Oscillation and Southern
Annular Mode (Stammerjohn et al., 2008a), which may also influence the
region's inorganic carbon chemistry on an interannual scale. A longer
measurement period may be needed in order to be able to distinguish with
certainty between natural variability and secular trends
(Henson et al., 2010).