Interannual variability of pteropod shell weights in the high-CO 2 Southern Ocean

Interannual variability of pteropod shell weights in the high-CO2 Southern Ocean D. Roberts, W. R. Howard, A. D. Moy, J. L. Roberts, T. W. Trull, S. G. Bray, and R. R. Hopcroft Antarctic Climate and Ecosystems Cooperative Research Centre, Hobart, 7001, Tasmania, Australia Australian Antarctic Division, Department of the Environment, Water, Heritage and the Arts, Channel Highway, Kingston, 7050, Tasmania, Australia Marine and Atmospheric Research, Commonwealth Scientific and Industrial Research Organisation (CSIRO), Castray Esplanade, Hobart, 7000, Tasmania, Australia Institute of Antarctic & Southern Ocean Studies, University of Tasmania, Hobart, 7001, Tasmania, Australia Institute of Marine Science, Univ. of Alaska Fairbanks, Fairbanks, Alaska 99775-7220, USA Received: 1 October 2008 – Accepted: 1 October 2008 – Published: 26 November 2008 Correspondence to: D. Roberts (d.roberts@utas.edu.au) Published by Copernicus Publications on behalf of the European Geosciences Union.


Introduction
The oceans have absorbed about one-third of the total anthropogenic CO 2 released into the atmosphere (Sabine et al., 2004).The addition of CO 2 to seawater lowers both Figures
The Southern Ocean is a substantial sink for atmospheric CO 2 , taking up a large proportion of the anthropogenic carbon in the global oceans (Sabine et al., 2004).We deployed the first of a series of sediment traps at 2000m below the sea surface at 47 • S, 142 • E in the Southern Ocean (Fig. 1) in 1997/1998 to capture a multiyear series of marine calcifiers falling through the water column (Bray et al., 2000;Trull et al., 2001).The fate of aragonite calcifiers in the Southern Ocean is of particular concern, especially as global carbon cycle models suggest that polar-regions will be the first to experience [CO 2− 3 ] below aragonite saturation (Orr et al., 2005;Royal Society, 2005;McNeil and Matear, 2008).As aragonite is less stable than calcite, the other main form of biogenic calcium carbonate (CaCO 3 ) (Mucci, 1983), aragonite producers are likely to respond to ocean acidification more markedly than calcite producers.
One major group of aragonite producers is the thecosomatous (shelled) pteropods: planktonic gastropods (Lalli and Gilmer, 1989).As the only group of pelagic aragonite producers, pteropod production and dissolution play important roles in the upper-ocean alkalinity cycle (Betzer et al., 1984;Byrne et al., 1984;Gangstø et al., 2008).One of the most common pteropods of the Southern Ocean, Limacina helicina antarctica (Fig. 2a), has a geographic range that extends from the Subtropical Front to the Antarctic coast (van der Spoel and Dadon, 1999).Two intra-specific morphotypes of this common Southern Ocean pteropod have been identified: Limacina helicina antarctica forma Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version Interactive Discussion antarctica (Fig. 2b), dominant in Antarctic waters between Antarctica and the Polar Front, and Limacina helicina antarctica forma rangi (Fig. 2c), found more commonly in Subantarctic waters north of the Antarctic Circumpolar Current (van der Spoel and Dadon, 1999).Both morphotypes were collected in our sediment trap series.We use mean whole pteropod shell weight as a measure of calcification in these aragonite producers.This approach has been used in studies of planktonic foraminifera, where clear correlations between shell weights and [CO 2− 3 ] in seawater have been observed (Bijma et al., 2002).Our investigation represents the first attempt to estimate interannual variations in in situ pteropod calcification and establish a benchmark against which future impacts of ocean acidification in the Southern Ocean may be detected.

Sediment trap deployments and sample retrieval
A McLane 21-cup sediment trap was first deployed to a depth of 2000 m at 47 • S, 142 • E in 1997/1998 (Fig. 1), and sediment traps have been recovered from the same site in 1999/2000, 2000/2001, 2003/2004 and 2005/2006 to date.The conical sediment trap collects particles sinking through the ocean to a 0.5 m 2 surface area into individual cups prepared with buffered and poisoned solutions that are open for periods varying between 5 and 60 days (Supplementary • S, 153 • E), which was treated with sodium chloride (5 g l −1 ) to increase solution density, sodium tetraborate (1 g l −1 ) as a pH buffer and mercuric chloride (3 g l −1 ) as a preservative (Bray et al., 2000;Trull et al., 2001) and Howard, 2003).The resulting 150 µm-1 mm size fraction was used to determine mean pteropod shell weights for each sediment trap cup recovered.We focus on pteropods within the 150 µm-1 mm size fraction herein, as pteropods in this size fraction tend to dominate Limacina helicina populations (Fabry, 1989;Collier et al., 2000) and are likely to be most sensitive to changes in carbonate chemistry (Byrne et al., 1984;Honjo et al., 2008).Pteropods may be digested or dissolved before they reach the sediment traps so we assume our samples represent a minimum estimate of pteropod flux at the depth of the sediment trap.Whole pteropods (pitted or fragmented shells were not included in our analyses) were picked from dried aliquots, identified and enumerated.Limacina helicina antarctica (Fig. 2a) was the most common pteropod retrieved in the sediment trap series.These shells were separated into morphotypes according to van der Spoel and Dadon's (1999) separation of taxa (Fig. 2b, c), weighed in batches per cup for each season retrieved using a Mettler Toledo microbalance (preci-sion=0.1 µg) and mean shell weights per morphotype per cup per trap calculated (Fig. 3a, c; Supplementary Table 1 http://www.biogeosciences-discuss.net/5/4453/ 2008/bgd-5-4453-2008-supplement.pdf).Flux weighted mean shell weights per morphotype (Fig. 3b, d; Table 1) were calculated from sediment trap pteropod shell (Fig. 4a; Table 2) and mass (Fig. 4b; Table 3) flux to remove bias generated by collection time differences and allow interannual comparisons.

Statistical analyses
As a result of non-uniform cup sampling periods (varying between 5 and 60 days), limited sample sizes and our batch-weighing approach we are unable to quantify within-sample error distributions or characterize seasonal variability.Accordingly, nonparametric statistical tests were employed to detect interannual trends in shell weight for either morphotype as these tests make no implicit assumptions about underlying distributions.Using a two-tailed Spearman rank-correlation coefficient test (Anderson et al., 1981) we estimate an interannual shell weight trend (P =0.14) for forma antarctica only (Table 4a).The Spearman rank-correlation coefficient test does not take into account possible within-sample correlations or allow quantification of the magnitude of trends.To quantify the trend in forma antarctica shell weight we assumed normality and applied a weighted least-squares regression (Davies and Goldsmith, 1972) to estimate a slope of -1.17±0.47 µg yr −1 (P =0.02) (Table 4b).Note that the shell weight for 2005/2006 is based on a single shell.However, our estimates of trends in the data are insensitive to removal of this data point.Without this final data point, the estimated linear trend is -1.19±0.49(P =0.02).Similarly, without the final data point the Spearman rank-correlation coefficient is only slightly changed (P =0.12).Spearman rank-correlation coefficient tests were also used to detect interannual trends in sediment trap pteropod shell and mass flux (Table 5) and to investigate correlations between forma antarctica shell weights and sea surface temperature (SST) and chlorophyll-a concentrations in the Subantarctic (Table 6).

Biogeochemical trends
One objective of this study was to seek to identify environmental variables responsible for the observed changes in mean Limacina helicina antarctica forma antarctica shell weights.We examined decadal-scale changes in both physical (temperature) and ecological (primary production) parameters at 47 • S, 142 Reynolds, 1997) and chlorophyll-a concentrations (Chl-a) (SeaWiFS, 2008) (Fig. 5a,b).
In addition, we sought to estimate changes in carbonate chemistry of the water masses the pteropods inhabit from discrete carbonate chemistry measurements made in 1995 and 2001 (Fig. 5c).Average mixed-layer (top 100 m) [CO 2− 3 ] was calculated form World Climate Research Programme Climate Variability and Predictability Program (CLIVAR) and World Ocean Circulation Experiment (WOCE) SR3 transects between Tasmania and Antarctica on voyages AA9404 (Tilbrook and Rintoul, 1995) and AA0301 (Tilbrook et al., 2001).Average [CO 2− 3 ] was calculated from total alkalinity, total carbon dioxide, temperature, pressure, salinity, phosphate and orthosilicate using the CO2Sys Excel macro (version 1.02) (Lewis and Wallace, 1998) and constants from Mehrbach et al. (1973) as modified by Dickson and Millero (1987).Aragonite solubility was calculated from the equation of Mucci (1983) that includes pressure adjustments K sp following Ingle (1975).Pre-industrial mixed-layer [CO 3 ], based on estimates of anthropogenic CO 2 inventories in the Subantarctic Southern Ocean in this sector (Sabine et al., 2004;Feely et al., 2004).

Results and discussion
The two morphotypes of Limacina helicina antarctica show different shell weight trends in our nine-year sediment trap series.There is an interannual linear shell weight trend for forma antarctica of -1.17±0.47 µg yr −1 (P =0.02) (Fig. 3a; Table 4).For a constant shell size distribution and shell flux, this change in shell weight is equivalent to a reduction in calcification of ∼35% (Fig. 3b).If this rate of shell weight decrease is valid and is sustained, Limacina helicina antarctica forma antarctica may be unable to sustain a shell by 2020.We do not find the same trend in shell weight for Limacina helicina antarctica forma rangi (Fig. 3c, d 3).Maximum flux was recorded in the first year of sediment trap collection (0.77 mg m −2 day −1 ) and the contribution of pteropod aragonite in that year is between 0.02 and 0.92% of the total PIC flux (Trull et al., 2001).Estimates of CaCO 3 flux in Ross Sea sediment traps (a large proportion of which is Limacina helicina in southern Ross Sea traps) are much greater than in this study (Collier et al., 2000) (Table 7).However, our Subantarctic pteropod fluxes are similar to those from sediment traps deployed in the deep Arctic Ocean (Fram Strait -2000 m and Bear Island -1700 m) in 1984/1985 (Meinecke and Wefer, 1990) where maximum flux estimates were 0.51 and 0.37 mg m −2 day −1 respectively and the pteropods Limacina helicina and Limacina retroversa collectively account for 1.7% and 0.4% of the total CaCO 3 flux to the Arctic Ocean traps (Table 7)(though these fluxes include the >1 mm size fraction).
When looking for a causal mechanism for the rate of shell weight loss in Limacina helicina antarctica forma antarctica shell weights we found no correlation (Table 6) between this morphotype and either SST (Fig. 5a) or Chl-a (Fig. 5b) over the past decade.We determined average mixed-layer (upper 100 m) [CO 2− 3 ] has changed from 155 µmol kg −1 in 1995 to 146 µmol kg −1 in 2001 (Fig. 5c), implying a reduction in the saturation state for the mineral aragonite (expressed by the index Ω aragonite ) of ∼−0.2/decade.This is consistent with other estimates of decadal-scale carbonate saturation decrease due to anthropogenic CO 2 uptake in this region (McNeil et al., 2001;Matear and Lenton, 2008).We suggest the most likely explanation for the shell weight loss observed in Limacina helicina antarctica forma antarctica is a reduction in [CO 2− 3 ] in the Subantarctic Southern Ocean over the past decade.The intra-specific differences in shell weight trends between Limacina helicina antarctica forma antarctica and Limacina helicina antarctica forma rangi may be due to differences in their calcification response to changes in carbonate chemistry.The two forms of Limacina helicina antarctica are morphologically and ecologically distinct so distinct physiological responses in calcification are plausible.Introduction

Conclusions References
Tables Figures

Back Close
Full Alternatively, the differences in shell-weight trends may be explained by differential aragonite loss due to dissolution in the water column.The sediment trap in this study was deployed at a depth below the aragonite saturation horizon (ASH) so any pteropods falling into the trap would have been sinking through water corrosive to aragonite (Feely et al., 2002).As aragonite is more soluble in seawater than calcite (Mucci, 1983;Gehlen et al., 2005), aragonite producers will be first affected by changes in the saturation horizon.Our sediment trap samples were collected at 2000 m below the sea surface, 800 m below the Sub-Antarctic Southern Ocean's present average ASH depth of ∼1200 m (Fig. 1c), and some pteropods would be likely to experience dissolution.Hence, an alternative hypothesis may be that reduced aragonite saturation is not yet inhibiting shell growth but accelerating dissolution during export to the sediment traps.
To address this, we apply a semi-quantitative index of pteropod dissolution, based on shell opacity, and originally developed for the congeneric taxon Limacina inflata in the Atlantic (Gerhardt and Henrich, 2001), to the Limacina helicina antarctica shells in this study.We see signs of dissolution in 30-40% of the forma antarctica shells and in 50-70% of the forma rangi shells but there is no apparent trend in opacity for either morphotype over the last decade.As we see typically ∼25% more dissolution in forma rangi yet no systematic shell weight change in this morphotype collected in the sediment trap series we suggest that dissolution is unlikely to be the main causal mechanism for the shell weight trend observed in forma antarctica.
Our observations of pteropod shell weights in the Southern Ocean indicate that there is still much to learn about aragonite calcifiers' responses to the acidifying conditions in ocean ecosystems.Recent experiments on the coccolithophore Emiliania huxleyi (Iglesias-Rodriguez et al., 2008) reveal potential differences in calcification response to high-CO 2 environments among morphotypes of that species.Such intra-specific differences may explain the differences in trends for the two morphotypes of Limacina helicina antarctica.Differential responses to elevated CO 2 conditions may result in competitive advantages that could drive ecosystem shifts (Fabry, 2008)

Conclusions
Our observations represent the first attempt to estimate interannual variations in pteropod calcification and establish a benchmark against which future impacts of ocean acidification may be detected.The changes in high-latitude seawater chemistry anticipated by the end of the century could alter the structure and biodiversity of high-latitude ecosystems, with impacts on multiple trophic levels (Orr et al., 2005).Though, as yet, the implications of these likely impacts are not clear, we know that pelagic molluscs can be important calcifiers in the Southern Ocean (Royal Society, 2005) and act as food sources and grazers in the Antarctic food web (Lalli and Gilmer, 1989;Seibel and Dierssen, 2003;Hunt et al., 2008).South of the Antarctic Polar Front pteropods dominate the export flux of CaCO 3 (Collier et al., 2000) and Limacina helicina antarctica replaces krill at times as the dominant Southern Ocean zooplankton group (Lalli and Gilmer, 1989).The production of carbonate by pteropods may also be a modulator of the efficiency of biological particulate carbon transfer to the interior of the ocean (Byrne et al., 1984).Finally, the production and dissolution of aragonite is a constraint on models of the impact of ocean acidification on pelagic ecosystems and on marine alkalinity-cycle feedbacks (Gangstø et al., 2008).As the small but discernable decadal variability we document here may represent an emerging trend we recommend closer monitoring of the Southern Ocean ecosystem for further effects on these pelagic snails, and other, marine calcifiers.

Fig. 3 .
Fig. 3. (a) Limacina helicina antarctica forma antarctica: a linear decrease in mean shell weight of 1.17±0.47µg yr −1 (P =0.02) is seen since 1997/1998.Note that the shell weight for 2005/2006 is based on a single shell.However, our estimates of trends in the data are insensitive to removal of this data point.Without this final data point, the estimated linear trend is -1.19±0.49(P =0.02).(b) Limacina helicina antarctica forma antarctica: flux weighted mean shell weight±standard error per sediment trap.Note there is no standard error estimate for the 2005/2006 shell weight data as it represents a single shell.(c) Limacina helicina antarctica forma rangi: no significant trend is seen for this morphotype's mean shell weight since 1997/1998.(d) Limacina helicina antarctica forma rangi: flux weighted mean shell weight±standard error per sediment trap.
. Identifying in situ biological responses of Southern Ocean pteropods and other organisms will Introduction

Table 2 .
Limacina helicina antarctica morphotypes and total pteropod shell flux per sediment trap (shells m −2 d −1 ) from 1997/1998 to 2005/2006 at 47 • S, 142 • E. Limacina helicina antarctica morphotype calculations are based on the number (n) of sediment trap cups containing those species.Total flux is based on the number (n) of sediment trap cups containing any pteropod species.

Table 3 .
Limacina helicina antarctica morphotypes and total pteropod mass flux per sediment trap (mg m −2 d −1 ) from 1997/1998 to 2005/2006 at 47 • S, 142 • E. Limacina helicina antarctica morphotype calculations are based on the number (n) of sediment trap cups containing those species.Total flux is based on the number (n) of sediment trap cups containing any pteropod species.

Table 4b .
Weighted least squares regression of mean shell weight trend with time for Limacina helicina antarctica forma antarctica from 1997/1998 to 2005/2006 at 47 • S, 142 • E.

Table 7 .
Total pteropod flux (mg m −2 d −1 ) comparisons between this study and others.Calculations are made on the number of sediment trap cups (n) that successfully captured material.