Twentieth century δ 13 C variability in surface water dissolved inorganic carbon recorded by coralline algae in the northern North Pacific Ocean and the Bering

Introduction Conclusions References

The world's oceans have taken up approximately half of the anthropogenic CO 2 emitted into the atmosphere (Sabine et al., 2004a).Since anthropogenic emissions of CO 2 continue to increase, understanding and constraining CO 2 uptake by the ocean is of considerable importance.The ability of the ocean to take up anthropogenic CO 2 varies widely depending largely on the local/regional water temperature, ocean circulation, and biological productivity.This is particularly true in the subarctic Pacific Ocean, Figures where estimates of carbon uptake rates from direct measurements and proxy records vary substantially (Newsome et al., 2007;Quay et al., 2003;Tanaka et al., 2003), likely as a result of large seasonal and interannual variability in the physical and biological controls on surface water CO 2 (Sabine et al., 2004b).Therefore, robust, high-resolution records of seawater CO 2 in this region are needed to more accurately constrain the role of the northern North Pacific Ocean and Bering Sea in absorbing anthropogenic CO 2 .
The ratio 13 C/ 12 C of seawater dissolved inorganic carbon (DIC), expressed as δ 13 C DIC , is a useful method to measure the intrusion of anthropogenic CO 2 into surface waters (Quay et al., 1992).Specifically, δ 13 C DIC values record the δ 13 C-Suess effect (i.e., the burning of isotopically "light" fossil fuels, which decreases the δ 13 C value of carbon reservoirs), and δ 13 C in atmospheric CO 2 at Pt. Barrow, Alaska decreased 0.02‰ yr −1 from the early 1980s to the early 2000s (Keeling et al., 2001).
In contrast, repeated cruises in the subarctic Pacific recorded little to no decrease in δ 13 C values of surface water DIC between 1970 and 1990 (Quay et al., 2003) while δ 13 C DIC values decreased 0.012‰ yr −1 from 1997 to 2001 at the western North Pacific station KNOT (Tanaka et al., 2003).In these studies, the depressed influence of the δ 13 C-Suess effect in the surface waters of the subarctic Pacific water was attributed to rapid renewal of surface waters which prevents the subpolar gyres from equilibrating with the atmosphere (Quay et al., 2003).This reduces the penetration of atmospheric CO 2 into the surface waters, and drives the disequilibrium between atmospheric CO 2 and surface water δ 13 C DIC values.However, instrumental δ 13 C DIC measurements are limited both in temporal (i.e., repeat cruise data limited to two data points) and spatial (i.e., one location at station KNOT) distribution, and the rate of the δ 13 C decrease recorded in multi-decadal proxy records from different locations in the subarctic Pacific exceeds that of the instrumental records (e.g., 0.02 to 0.06‰ yr −1 (Newsome et al., 2007;Schell, 2001)).Although these proxy records are useful to supplement instrumental data, their source from marine mammals has inherent disadvantages.For example, it is difficult to differentiate changes in the baseline δ 13 C DIC relating to anthropogenic causes from temporal shifts in foraging zones of the marine mammals that Introduction

Conclusions References
Tables Figures

Back Close
Full have different isotopic values at the base of the food web (Newsome et al., 2007;Hirons et al., 2001).Variability within a marine mammal sample set, i.e., in size, gender, and age, may also obscure baseline changes (Hobson et al., 2004).In addition to the marine mammal-derived records, δ 13 C values from a gorgonian soft coral in the Alaskan stream declined at a rate of 0.015‰ yr −1 (Williams et al., 2007).Records from these corals are valuable as the corals are stationary throughout their life and thus not subject to migration or shift in foraging zones; however, the available records is limited and the factors that control the δ 13 C values of particulate organic matter, their primary food source, are still poorly understood.Therefore, a robust source of proxy δ 13 C DIC reconstructions from which to constrain the subarctic δ 13 C-Suess effect was not available prior to this study.

Sample collection
Two specimens of the coralline alga C. nereostratum (Attu 11-4 and AM4-1) were collected live from 10 m depth during a research cruise in August 2004 from the Aleutian Islands (Fig. 1).Attu 11-4 was collected offshore of the southern side of Attu Island (52 • 47.79 N, 173 • 10.80 E) and was primarily influenced by the Alaskan Stream.The Alaskan Stream originates in the Gulf of Alaska and turns into the Bering Sea through the Aleutian Islands passages where it is the dominant source of relatively warm, fresh and nutrient-rich water (Reed and Stabeno, 1994).AM4-1 was collected offshore of the northern side of Amchitka Island (51

Sample preparation and analyses
Untreated algal specimens AM4-1 and Attu 11-4 were sectioned in half and roughly polished onboard the research cruise.A 3 mm thick slice was removed from the museum specimen AM-KR-80 and mounted onto thin section glass.In the lab, the sections were polished with decreasing grit sizes on a Struers Labopol polishing disk.Polished sections were then photographed using an Olympus reflected light microscope attached to an automated sampling stage-imaging system equipped with geo.TS software (see (Hetzinger et al., 2009) for details).Annual increments in the calcified tissue were clearly visible in the resulting high quality images and were digitally mapped using the geo.TS software.This allowed for the location of each sample milled for stable isotope analysis (see below) to be carefully digitized on the photomosaic (Fig. 2).In addition, the widths of each annual growth increment were measured using the geo.TS software.Figures

Back Close
Full Polished and photographed sections were cleaned in an ultrasonic bath for three ten-minute intervals with deionized water and dried overnight.Working from the outside edge of the specimen which represents the most recent growth toward the oldest part of the sample, material for stable isotope analyses was removed using a highprecision, computer-driven micromill attached to an x, y, and z stage using digitized milling path positions.Sampling resolution for stable isotope analysis varied from one to twelve samples per growth increment depending on the width of the growth increment.Removed material was analyzed for δ 13
Chronologies developed in Halfar et al. (2007) and Hetzinger et al. (2009) (also see Sect.3.1) were applied to the specimens.Isotope data were interpolated to monthly intervals, and then subsequently resampled to annual resolution using AnalySeries software (Paillard et al., 1996).T-tests compared average δ 13 C values from 1937 to 1967 for AM-KR- 80 and Attu 11-4 and from 1968 to 2003 for AM4-1 and Attu 11-4.Attu 11-4 was combined with AM4-1 from 1967-2003 and the dataset was smoothed using a 7-year running average from 1937-2003.The line of best fit was calculated using SigmaPlot (Systat Software Inc., San Jose, CA) for this dataset to determine decadal rates of decrease from the 1960s to 2003.The specimen-specific offset for AM-KR-80 (see Sect. 3.3) prevented the inclusion of this sample in this dataset.The Aleutian Low Pressure Index (ALPI) is the most important index of climate-ocean dynamics in the North Pacific/Bering Sea (Benson and Trites, 2002;McFarlane et al., 2000).Annually-resolved ALPI data (Beamish et al., 1997; http://www.pac.dfo-mpo.gc.ca/ science/species-especes/climatology-ie/cori-irco/indices/alpi.txt) were also smoothed with a 7-year running mean.To create a robust dataset for time series analysis, each of the original annually-resolved algal δ

Growth increments and specimen chronologies
The three specimens displayed clear growth increments in the calcified tissue.The width of the annual growth increment varied from 140 to 680 µm with an average of 381 µm for all three specimens.There was no ontogenic or secular change in annual growth increment widths over time in any of the specimens.Seasonal variability in Mg/Ca ratios in all of the specimens and δ 18 O values in Attu 11-4 confirmed the annual periodicity of the growth increments (Halfar et al., 2007;Hetzinger et al., 2009).The outermost growth increment in all of the specimens was formed during the summer of the specimen collection and did not represent a whole year of growth.In specimens AM4-1 and Attu 11-4, the year prior to collection was assigned to the first full skeletal growth increment.In specimen AM-KR-80, the first full year of growth was not sampled and the year two-years prior to collection was assigned to the first full growth increment.Therefore, from Amchitka Island, 35 years were sampled from 1968 to 2003 in specimen AM4-1 and 31 years were sampled from 1937 to 1967 in AM-KR-80.From Attu Island, 116 years were sampled from 1887-2003 in specimen Attu 11-4.
Average δ 13 C values for Attu 11-4 and AM4-1 were depleted relative to predicted isotopic equilibrium by 3.5 and 3.0‰, respectively, for the period of 1992 and 1993 (Fig. 3).Average δ 18 O values for the same specimens were depleted relative to predicted isotopic equilibrium by 2.2 and 2.0‰, respectively, for the period of 1990, 1993, and 1995 (Fig. 3).Isotope values for AM-KR-80 were not compared to predicted Figures

Back Close
Full isotopic equilibrium as this specimen was collected prior to the early 1990s and before instrumental measurements of seawater isotopes were obtained close to the Aleutian Islands., with δ 13 C values of −0.4‰ at the start and −0.7‰ at the end of the record (Fig. 4).AM-KR-80 was collected prior to the beginning of growth of AM4-1, so direct comparisons between the two records are difficult, however δ 13 C values at the end of the AM-KR-80 record in 1967 were 0.6‰ lower than at the start of the AM4-1 record in 1968 (Fig. 4).In addition, the δ 13 C values for AM-KR-80 (average

Carbon and oxygen isotopes
C. nereostratum forms calcite depleted in both δ 13 C and δ 18 O relative to predicted isotopic equilibrium (Fig. 3).However, since δ 13 C and δ 18 O covary in only one specimen (AM4-1), and the range in δ 18 O values is small, we hypothesize that the correlation between δ 13 C and δ 18 O values in this specimen largely reflects anthropogenicallydriven changes in the isotopes relating to secular variability in δ 13 C DIC (δ 13 C-Suess effect; see below) and temperature/salinity (warmer/less saline; see Halfar et al., 2007), rather than a kinetic effect common to all C. nereostratum.Depleted δ 18 O values relative to predicted equilibrium are consistent with previous studies, and are expected to be species-specific and constant over time, thus facilitating the use of coralline algalderived oxygen isotope archives for paleoceanographic reconstructions (e.g., Halfar et al., 2000Halfar et al., , 2007Halfar et al., , 2008;;Lee and Carpenter, 2001;Wefer and Berger, 1991).Similar to δ 18 O, δ 13 C values in C. nereostratum are also depleted relative to predicted isotopic equilibrium by 3-3.5‰ (Fig. 3).Metabolic processes such as respiration and photosynthesis may alter δ 13 C values resulting in observations that deviate from predictions based on isotopic equilibrium (Swart, 1983).Respiration in the calcifying microenvironment may cause 13 C-depletion in CO 2 .On the other hand, photosynthesis may cause 13 C-enrichment in the same CO 2 microenvironment within the calcified tissue (i.e., photosynthesis removes 12 C).The δ 13 C values that we measured were depleted relative to equilibrium predictions.This result indicates that respiratory CO 2 contributed a significant amount of the carbon in the calcified tissue (Borowitzka, 1977).

δ 13 C records
Average δ 13 C values in AM-KR-80 were significantly lower than in Attu 11-4 (Fig. 5a).Although a specimen-specific effect cannot be ruled out, we hypothesize that the lower (25 m) plant was exposed resulting in decreased photosynthesis relative to Attu 11-4 and AM4-1 (10 m) plants.Light levels are known to influence growth rates in coralline algae (Adey, 1970) and presumably photosynthesis as well.Decreased photosynthesis from lower light levels will reduce discrimination against δ 13 C during growth, resulting in decreased δ 13 C values in the skeleton.Therefore, the collection depth of an algal specimen (or more specifically, a factor that accounts for differences in ambient light regimes) needs to be taken into consideration before comparing δ 13 C values among different specimens.In addition, since light levels may vary widely over very small spatial scales, for example as a result of local shading by macroalgae, averaging of more than one sample is recommended to reduce any influence of light on δ 13 C values.
δ 13 C values in marine calcium carbonate organisms are driven by changes in the δ 13 C DIC , as well as any metabolic and kinetic effects that are present (e.g., Mc-Connaughey, 2003).We hypothesize that kinetic effects are not a significant factor here (see Sect. 4.1), and assuming that offsets resulting from metabolic effects are constant over time and/or are accounted for by averaging multiple specimens, then δ 13 C values largely reflect δ 13 C DIC .Thus, the skeletal δ 13 C records can reconstruct the primary factors influencing oceanic δ 13 C DIC : anthropogenic δ 13 C-Suess effect (Andreasson and Schmitz, 1998), surface-ocean primary production (Krantz et al., 1987), upwelling events (Killingley and Berger, 1979), and water column stratification (Arthur et al., 1983).

Secular decrease in δ 13 C values
The decline in δ 13 C values recorded by the coralline algal specimens examined in this study clearly exceeds what can be attributed to the δ 13 C-Suess effect alone (Fig. 6).
This enhanced rate of decline is not consistent with instrumental or proxy records from outside of the northern North Pacific Ocean and Bering Sea (e.g., Swart et al., 2010).
This indicates that factors in addition to the δ 13 C-Suess effect drive the skeletal δ 13 C values.δ 13 C values in bowhead whale baleen and steller sea lions bone collagen (Schell, 2000(Schell, , 2001;;Hirons et al., 2001) also decline at an enhanced rate; therefore, Introduction

Conclusions References
Tables Figures

Back Close
Full an algal-specific effect or site-specific variability are unlikely to be the primary cause of the decline observed in our study.It would appear more likely that changes in the δ 13 C DIC pool to which both coralline algae and phytoplankton are exposed (the latter subsequently transmitted via the food web to whales and sea lions) drive the strongly decreasing δ 13 C values.Therefore, we explore the factors that influence δ 13 C DIC , in addition to the δ 13 C-Suess effect, that may drive the enhanced rate of decline.δ 13 C DIC in the northern North Pacific Ocean and Bering Sea regions is largely controlled by biological activity and upwelling/deep-water mixing (Tanaka et al., 2003).A decrease in primary productivity would increase the amount of CO 2 in surface waters, thus reducing carbon fractionation during the calcification process and lead to lower δ 13 C values in the resulting calcified tissues.In fact, a decline in phytoplankton cell growth rates, and hence primary productivity, has been hypothesized to drive the declining trends in the whale baleen and sea lion bone collagen δ 13 C (Schell, 2000(Schell, , 2001;;Hirons et al., 2001).However, available evidence does not support a decrease in new primary production (e.g., Hunt Jr. et al., 2002).In addition, there is no multi-year trend of decrease in growth-increment widths in the three C. nereostratum specimens over the past 50 years.An increase in upwelling and/or deepening of the mixed layer would also change CO 2 concentrations in the surface waters of northern North Pacific Ocean and Bering Sea (Patra et al., 2005), and thus δ 13 C DIC values and δ 13 C values of the algal calcified tissues.Authors of previous reports have suggested that upwelling and deep mixing may reduce the influence of the Suess effect in the subarctic Pacific as a consequence of dilution by deep water that has not been influenced by anthropogenic carbon emissions and the limited amount of time for CO 2 equilibration to take place (Gruber et al., 1999;Quay et al., 1992;Schell, 2001;Sonnerup et al., 1999).However, CO 2 concentrations increase and δ 13 C DIC decreases with depth (e.g, Koopnick et al., 1972; δ 13  reported by Bacastow et al. (1996).Increased CO 2 concentrations resulting from more upwelling and/or deepening of the mixed layer theoretically decreases fractionation during calcification resulting in lower δ 13 C values of the calcified tissue.In addition, the deeper water brought up to the surface will have low δ 13 C DIC values due to the remineralization of organic matter with depth, thus contributing to decreased δ 13 C DIC in surface waters.The mixed layer depth in the subarctic North Pacific has shoaled in recent decades, which has been attributed to warming and freshening that in turn has produced enhanced stratification of the surface waters (Freeland et al., 1997); therefore, upwelling is likely the driving factor of the enhanced rate of decline in the δ 13 C proxy records.Changes in the Aleutian Low pressure system may be driving the increase in upwelling.When the low pressure system is intense, oceanic upwelling under the centre of the Aleutian Low is increased (Reid, 1962;Thomson, 1981), bringing higher concentrations of CO 2 with low δ 13 C values to the surface.This upwelled CO 2 is moved to the edges of the Alaska gyre via horizontal divergence (similar to other nutrients and biomass; Brodeur and Ware 1992; Reid 1962) and is incorporated into the Alaskan Stream.Once incorporated into the Alaskan Stream, the CO 2 -rich water flows southwestward along the continental slope of the Alaska Peninsula and enters the Bering Sea through the Aleutian Island Archipelago.By this mechanism, CO 2 -rich water was transported to the environment in which the algal specimens grew.In contrast, when the Aleutian Low weakens, oceanic upwelling is reduced and water with less CO 2 and lower δ 13 C values is brought to the surface.The Aleutian Low Pressure Index (ALPI), which is a measure of the intensity of the Aleutian Low (Beamish et al., 1997), indicates that the Aleutian Low has generally been increasing in strength since the late-1940s (Fig. 6).In addition, the large scale variability in the ALPI is mirrored with an approximately three-year lag in the algal δ 13 C records from the start of the records up to 1989, which corresponds to the 1989 regime shift evident in the ALPI and in biological records (Beamish et al., 1999;Hare and Mantua, 2000).The lag likely represents the interval of time between the change in the pressure system, increase or decrease in upwelling, and transport of the water to the central Aleutian Islands where Figures

Back Close
Full There is large spatial and temporal variability in upwelling and ocean currents, and consequently CO 2 concentrations, throughout the subarctic North Pacific (Stabeno et al., 2004;Murphy et al., 2001).Therefore, the location of a specific δ 13 C DIC record relative to upwelling and/or currents transporting the upwelled water explains the large variability in δ 13 C instrumental and/or proxy records.The variable rate of decline recorded by the different instrumental and proxy records likely reflects local changes in δ 13 C DIC instead of basin-wide changes.As a consequence, extrapolation of a single record from one location to the entire basin provides an incomplete picture of carbon dynamics in the northern North Pacific Ocean and Bering Sea.Additional records are urgently needed to further constrain changes occurring and to calculate the role of subarctic oceans in taking up anthropogenic CO 2 .

Detrended δ 13 C variability
The significant spectral power on interannual timescales (4-7 years; Fig. 7) in the detrended and combined δ 13 C record corresponds to fluctuations of the El Ni ño Southern Oscillation (ENSO).Authors of previous studies suggested an atmospheric link between the tropical ENSO events and the northern Pacific/Bering Sea related to the winter position of the Aleutian Low such that it is intense and shifts eastward during El Ni ño conditions, driving warm, moist air northward along the northeast Pacific coast towards Alaska and into the Bering Sea (Niebauer, 1988;Emery and Hamilton, 1985).
The increase in intensity of the Aleutian Low associated with El Ni ño conditions will increase upwelling, bringing more CO 2 -rich water with low δ 13 C values into the surface waters.In contrast, a weakened Aleutian Low corresponds to cold events in the tropics (Emery and Hamilton, 1985).This supports a link between the tropical Pacific ENSO, and the subarctic Aleutian Low and algal δ 13 C records driving the 4-7 year variability.The significant spectral power on approximately 16-year timescales (Fig. 7) is generally Introduction

Conclusions References
Tables Figures

Back Close
Full consistent with the bidecadal (15-25 year) variations in the North Pacific Index (NPI) (Minobe, 1999).The NPI is another time series index representing the Aleutian Low, and is inversely correlated to the ALPI.Thus similar patterns between the NPI and the algal δ 13 C values are expected.These relationships support the influence of shifts in climate regimes (i.e., ENSO, PDO) on the surface water δ 13 C DIC values, in addition to the anthropogenic δ 13 C-Suess effect.

Summary
δ 13 C values from coralline algae clearly measure a decrease in δ 13 C values over the past 30 years exceeding that due solely to the anthropogenic δ 13 C-Suess effect.The enhanced rate of decline is consistent with records from some marine mammals.This indicates that the δ 13 C DIC which is taken up by primary producers, i.e., phytoplankton at the bottom of the marine food web and benthic coralline algae, is the cause of the enhanced rate of decline.We hypothesize that a combination of the δ 13 C-Suess effect and an increasingly intense Aleutian Low drives the strongly decreasing δ 13 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | C records from coralline algae theoretically are superior to existing subarctic Pacific δ 13 C animal-based proxy reconstructions because coralline algal skeletal δ 13 C values are not confounded by changes in the diet, foraging patterns, food web dynamics.Therefore, the goal of this paper is to demonstrate the potential of utilizing C. nereostratum as recorders of seawater δ 13 C DIC and extract robust, high-resolution records of subarctic δ 13 C DIC .
Discussion Paper | Discussion Paper | Discussion Paper | C and δ 18 O in inorganic high-Mg calcite were estimated from the equations of Romanek et al. (1992) and Kim and O'Neil (1997), respectively, with a correction of 0.17‰ for each mol% Mg for predicted δ 18 O (Jimenez-Lopez et al., 2006).The δ 13 C value of seawater was calculated from the average of six measurements from less than 15 m depth taken in June and August during two cruises in the northern Pacific during 1992 and 1993 (cruise ID: 31VIC92 0 and 3250210 1; data obtained from http://cdiac.ornl.gov/oceans/home.html;Key et al., 2004).The δ 18 O value of seawater was calculated from the average of 33 measurements from less than 15 m depth taken in June and August in 1990, 1993, and 1995 Discussion Paper | Discussion Paper | Discussion Paper |

13 C
records were normalized by subtracting the mean and dividing by the standard deviation.The resulting records were then combined to form a dataset comprised of Attu 11-4 and AM-KR-80 from 1937 to 1967 and Attu 11-4 and AM4-1 from 1968 to 2003.The line of best fit was calculated using SigmaPlot and the resulting residuals were used to produce a detrended δ13 C dataset.Wavelet analysis(Torrence and Compo, 1998) using IDL Wavelet Toolkit (available at: http://paos.colorado.edu/research/wavelets/)was applied to the combined and detrended dataset to determine the dominant modes of climate variability.
Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

δ 13 C
values more likely result from the lower light levels to which the deeper AM-KR-80 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | C DIC decrease by up to ∼2‰ within the top 200 m of the water column in measurements from WOCE P13N, http://cdiac3.ornl.gov/waves/discrete),and the ability of upwelling and/or mixing of deep water to drive surface water δ 13 C DIC values has previously been Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | the δ 13 C ultimately is taken up by the algae.The cause of the decoupling between the Aleutian Low and the algal δ 13 C records after the 1989 regime shift is unknown.
Discussion Paper | Discussion Paper | Discussion Paper | Figure 1.

Fig. 1 .Fig. 2 .Fig. 3 .Fig. 4 .
Fig. 1.Location of sample collection sites offshore of Attu and Amchitka Islands along the Aleutian Arc in the northern North Pacific Ocean and Bering Sea.Black circle indicates sample locations.(Map generated at http://www.planiglobe.com.) 615

Fig. 5 .Fig. 7 .
Fig. 5. Average (± SE) δ 13 C values for (A) AM-KR-80 and Attu 11-4 for the common period of 1937-1967 and (B) for AM4-1 and Attu 11-4 for the common period of 1968-2003.Averages were calculated from annual values as described in the methods section.* indicates significant differences between averages according to a t-test.