Article pubs.acs.org/est
Uranium in Larval Shells As a Barometer of Molluscan Ocean Acidification Exposure Christina A. Frieder,*,† Jennifer P. Gonzalez, and Lisa A. Levin Integrative Oceanography Division, Scripps Institution of Oceanography, University of California San Diego, 9500 Gilman Drive, La Jolla, California 92093-0218, United States S Supporting Information *
ABSTRACT: As the ocean undergoes acidification, marine organisms will become increasingly exposed to reduced pH, yet variability in many coastal settings complicates our ability to accurately estimate pH exposure for those organisms that are difficult to track. Here we present shell-based geochemical proxies that reflect pH exposure from laboratory and field settings in larvae of the mussels Mytilus californianus and M. galloprovincialis. Laboratory-based proxies were generated from shells precipitated at pH 7.51 to 8.04. U/Ca, Sr/Ca, and multielemental signatures represented as principal components varied with pH for both species. Of these, U/Ca was the best predictor of pH and did not vary with larval size, with semidiurnal pH fluctuations, or with oxygen concentration. Field applications of U/Ca were tested with mussel larvae reared in situ at both known and unknown pH conditions. Larval shells precipitated in a region of greater upwelling had higher U/Ca, and these U/Ca values corresponded well with the laboratory-derived U/Ca-pH proxy. Retention of the larval shell after settlement in molluscs allows use of this geochemical proxy to assess ocean acidification effects on marine populations.
■
INTRODUCTION Most calcifying marine invertebrate larvae are sensitive to reduced seawater pH associated with increased partial pressure of carbon dioxide, decreased carbonate ion concentration, and decreased calcium carbonate saturation state (Ω).1,2 Changes in larval performance in the plankton can influence success after settlement and thus have profound ecological impacts on populations.3 Yet the realized potential for decreased larval performance due to pH conditions is unknown for two reasons: (1) There is a lack of information regarding larval distributions in space and time, and (2) there is limited knowledge of corresponding environmental pH conditions. Natural geochemical proxies in larval carbonate structures are one means by which the larval pH-exposure history could be estimated, since variations in the environment can be recorded in calcified structures (e.g., shells, otoliths, and statoliths).4 Given that the larval shell component is retained during the early juvenile phase in molluscs, pH proxies could also be utilized to evaluate carry-over effects of these parameters on postlarval growth and survival, and as a means to determine the range of pH conditions experienced by larvae that have successfully recruited to adult populations. The search for a geochemical proxy that reflects carbonate chemistry has intensified with growing interest in modern and past changes in oceanic pCO2 and pH. δ11B and various trace and minor element-Ca ratios (e.g., Mg/Ca, Sr/Ca, B/Ca, and U/Ca) have been studied as possible proxies for low pH in planktic and benthic foraminifera, and invertebrates.5−9 Uranium incorporation into carbonates has gained increasing © 2014 American Chemical Society
scrutiny as a potential carbonate system proxy because uranium exists in seawater in the form of multiple uranyl carbonate complexes whose relative abundances depend on seawater pH and [CO2− 3 ]. As pH decreases, the proportions of bicarbonate and monocarbonate uranyl complexes increase and are matched with an increase in the free forms uranyl and uranyl hydroxide.10 Uranyl tricarbonate, the most abundant complex, is not incorporated into the calcite structure as easily as the bicarbonate and monocarbonate uranyl complexes; the relative abundance of these latter two complexes increases with decreasing pH.8,9 Accordingly, U/Ca in biogenically precipitated carbonates of corals and foraminifera increases with decreasing seawater pH.6,8,9,11 pH proxies have rarely been explored as an ecological tool. A valuable application would be to determine pH exposures of organisms that are difficult to track such as tiny invertebrate larvae. Other environmental parameters are known to affect the elemental composition of the larval carbonate including salinity, temperature, proximity to land, exposure to hypoxia, pollution, and upwelling experienced during the larval stage.12 Endogenous factors such as growth and ontogenetic effects also influence element signatures.13,14 To probe for a geochemical proxy that reflects pH exposure, two mytilid mussel species were cultured from embryos to Received: Revised: Accepted: Published: 6401
January 30, 2014 April 28, 2014 May 5, 2014 May 5, 2014 dx.doi.org/10.1021/es500514j | Environ. Sci. Technol. 2014, 48, 6401−6408
Environmental Science & Technology
Article
Table 1. Experimental Conditions for Experiments A, B, and C Used to Rear Mussel Larvae of Mytilus californianus and M. galloprovincialis species M. californianus
M. galloprovincialis
experiment (reps)
pHT
A (3)a A (3) A (3)b B (3) B (3) C (3) C (3) fieldc B (2) B (2) C (2) C (2)
8.04 7.51 7.51 ± 0.15 7.90 7.68 7.64 8.00 8.05 7.91 7.61 7.59 7.95
[O2] (μmol kg−1)
temperature (°C)
salinity
AT (μmol kg−1)
[CO2− 3 ] (μmol kg−1)
223 104 230 101 232 231 86 234 87
16.5 16.5 16.5 15.9 15.8 16.3 16.1 16.5 17.2 17.2 16.9 17.1
33.53 33.53 33.53 33.49 33.49 33.55 33.54 33.23 33.64 33.65 33.60 33.62
2235 2241 2240 2228 2227 2233 2232 2225 2250 2252 2240 2241
155 52 52 ± 17 115 73 68 142 156 124 67 63 134
U/Casw (μmol mol−1)
1.09 1.07 1.17 1.36 1.20 0.98 1.17 1.36
a
[O2] was not controlled or measured during Expt. A. bThird treatment during Expt. A cycled pH by 0.3 units on a semidiurnal basis with a mean of 7.51. cEnvironmental data for field-cultured larvae were collected with seapHOx instrumentation that recorded pHT, [O2], temperature and salinity every 15 min. [CO2− 3 ] was calculated from pH along with AT determined from a discrete sample taken at the beginning of the outplant.
method based on Dickson et al.,17 using 1-cm cuvette cells and commercially available m-cresol dye with dye corrections based on Clayton and Byrne.18 This method was calibrated with certified reference material from the Marine Physical Laboratory at Scripps Institution of Oceanography, and uncertainty of measurements was ±0.03 pH units. pH values are reported at the in situ temperature and on the total pH scale. [O2] measurements were made with a modified Winkler method with an accuracy of ±2 μmol kg−1. AT measurements were determined using an open-cell, potentiometric titration with an accuracy of ±2 μmol kg−1. pCO2, aragonite saturation state (Ωa), and [CO2− 3 ] were calculated from average AT and pH using CO2SYS with dissociation constants from Mehrbach et al.,19 as refit by Dickson and Millero.20 The average propagated uncertainties based on uncertainty in pH and AT for pCO2, Ωarag, and [CO2− 3 ] were ±72 μatm, ± 0.09, and ±5.7 μmol kg−1, respectively. On the basis of discrete samples taken during Expt. A on M. californianus, the average offset in pH among replicates from the set condition was 0.02 units. For Expt. B, the average offset among replicates from the set condition was 0.01 pH units and 3 μmol O2 kg−1 for both species. For Expt. C, the average offset of replicates from the set condition was 0.02 pH units and 3 μmol O2 kg−1 for both species. For laboratory experiments, larvae were cultured in 4-L containers nested within a 7.5-L bucket that received flowthrough treatment seawater. Ten larvae mL−1 in Expt. A and 50 larvae mL−1 in Expts. B and C were transferred to the experimental culturing vessels within 2-h post fertilization. Culturing buckets were treated as replicates, and there were 2 or 3 per treatment for each experiment (Table 1). Larvae were cultured for 8 days, fed ad libitum daily starting on day 2 a premixed diet of 50 000 cells mL−1 during Expt. A and 100 000 cells mL−1 during Expts. B and C of Isochrysis sp., Pavlova sp., Thalassiosira weissf logii, and Tetraselmis sp., and maintained on a 12h:12h light/dark cycle. On day eight postfertilization, larvae were frozen at −20 °C for later analysis. Temperatures ranged between 15.8 and 16.5 °C in M. californianus experiments conducted between November 2012 and January 2013, and 16.9 and 17.2 °C in M. galloprovincialis experiments conducted during May 2012 (Table 1).
aragonitic shell-bearing veligers across a range of pHT levels from 7.51 to 8.04. This corresponded with a range of [CO2− 3 ] from 52 to 156 μmol kg−1. The shells were analyzed for multiple trace and minor elements via laser-ablation inductively coupled plasma mass spectrometry. This method consumed the entire larval shell, and so elemental signatures likely reflect the entire growth period under study. Ensuing proxies were tested in the field under known and unknown pH conditions. This was accomplished by examining larvae reared in situ in parallel with a pH sensor, and at multiple locations along a gradient of upwelling.
■
MATERIALS AND METHODS Larval Culturing Methods. Five separate larval culturing experiments were carried out, three with Mytilus californianus and two with M. galloprovincialis, to produce shells precipitated from differing combinations of pH and dissolved oxygen ([O2]) (Table 1). Adult M. californianus were collected from the Scripps Institution of Oceanography pier in southern California, U.S.A. (32.867° N 117.257° W), and adult M. galloprovincialis were collected from San Diego Bay, California, U.S.A. (32.725° N 117.203° W). Gametes from at least four females and four males were used per experiment. Separate methods were utilized to manipulate seawater conditions for Experiment A and Experiments B and C. To maintain the desired pH conditions and incorporate fluctuations in Expt. A, we designed a feedback system in which target values were designated by the user, and a Honeywell Durafet III pH sensor continuously monitored the pH of one replicate per treatment. To study [O2] and pH effects in Expts. B and C, the desired chemistry was achieved by equilibrating seawater with a gas mixture of N2, O2 and CO2 using a Liqui-Cel membrane contactor. The desired gas composition was mixed from individual gas cylinders using mass flow controllers that provided the user with complete control over both the pCO2 and [O2] concentrations.15 Detailed experimental seawater manipulations, sampling, and measurement procedures for temperature, salinity, [O2], pH, and total alkalinity (AT) have been described.15,16 Briefly, discrete samples were taken daily for determination of pH and [O2], and AT and salinity were determined at the beginning and end of each experiment. pH samples were determined spectrophotometrically via a modified 6402
dx.doi.org/10.1021/es500514j | Environ. Sci. Technol. 2014, 48, 6401−6408
Environmental Science & Technology
Article
known importance from previous geochemical tagging studies in the region.21,23,24 The isotopes measured (63Cu, 66Zn, 88Sr, and 238U) were then ratioed to 48Ca. Data Treatment. Element ratios were transformed using log transformations to meet the normality and homogeneity of variance assumptions of regression analysis. All tests were conducted with averaged element-Ca ratios per experimental replicate container. In order to determine whether there were relationships between pH and individual element-Ca ratios in larval shells, simple linear regressions were performed. We then used simple linear regression to test whether there were ontogenetic effects (indicated by larval size) on individual element-Ca ratios. Larval size was measured as length of the veliger shell, anterior to posterior dimension parallel to the hinge line, on 30 veligers from within each replicate with ImageJ from photographs taken with a microscope-camera setup. In order to examine the utility of a multielemental proxy, we then performed a principal component analysis (PCA) to produce a linear combination of the ratios Cu/Ca, Zn/Ca, Sr/ Ca, and U/Ca. Linear regression analyses were used to investigate the degree of association between the principal components and pH.
To compare shell-element signatures of laboratory-cultured larvae with those of field-developed larvae, M. californianus embryos were cultured in the field from 12-h post fertilization to prodissoconchs. The field experiment was carried out from 19−27 November, 2011. Embryos were loaded into an acidwashed larval home.21 The home was secured to a mooring line at 7-m water depth immediately offshore of a mussel aggregation at Bird Rock, San Diego, California, U.S.A. (32.81° N 117.29° W). The larval home floated within 0.5 m of a SeapHOx instrument package that continuously logged pH, [O2], temperature, and salinity. Details of instrumentation are provided in Frieder et al.22 Upon retrieval, the contents of the home were frozen at −20 °C for later sorting and larval shell analysis. Environmental conditions corresponding with the larval outplant are provided in Table 1. To test for spatial variation in U/Ca of larval shells from regions of differing upwelling intensity, shell elemental data were obtained from a southern California, U.S.A. trace-element fingerprinting study of realized population connectivity for Mytilus californianus and M. galloprovincialis.23 The outplant conducted during May 2007 was chosen because elemental concentrations in shells were analyzed on the same instrument as in this study. Only data from M. galloprovincialis were available. Sites from this outplant were split into three regions, and only open coast sites were included to avoid salinity influences on uranium within estuaries.24 The North County region included Agua Hedionda, Oceanside Harbor, and Agua Lagoon, the Central County region included Dike Rock, La Jolla, and Ocean Beach, and the South County region included Imperial Beach and Cabrillo. Larval shell U/Ca (see next section) was averaged from 20 individual shells for each site, except for Oceanside Harbor and Dike Rock (n = 6 and 13 larval shells, respectively). Temperature data were recorded with the use of a HOBO Pendant temperature data logger located next to larval homes at each site. Sample Preparation and Analysis. All materials and containers coming into contact with larvae were seawaterleached or acid-leached for at least 2 weeks using 10% nitric acid and then rinsed thoroughly with ultrapure water. Sample preparation was carried out in a clean room environment using trace-metal-clean techniques. Larvae from each replicate were sorted into 1.5-mL acid-clean sample cups and rinsed with three aliquots of Milli-Q water to remove seawater and foodassociated debris. Shells were soaked in a clean solution of 100−300 μL 15% Optima grade hydrogen peroxide (Fisher Chemical) buffered in 0.05 M Suprapur NaOH (EMD Chemicals) for 4.5−7 h to remove all organic material, rinsed in ultrapure water three times, transferred to a depression slide, and allowed to dry under a Class-100 laminar-flow hood. Dry shells were individually transferred to a petrographic slide covered in double-sided tape (3 M Scotch Brand). Between 29 and 47 shells from each replicate of Experiments A, B, and C were mounted for analysis. Shells were analyzed using a Thermo Element 2 singlecollector ICP-MS operating in low-resolution mode with a New Wave Research UP-213 laser ablation unit (at the University of California Santa Barbara). Detailed calibration and analytical procedures are provided in the Supporting Information. Four element isotopes were selected for analysis. Uranium was selected based on its known association with carbonate ions in seawater, strontium was chosen based on previous observations that Sr/Ca varies with pH in planktic foraminifera,11 and the additional elements copper and zinc were chosen based on their
■
RESULTS Laboratory-Derived pH Proxies from Larval Carbonates. Two element-Ca ratios, U/Ca and Sr/Ca, were negatively correlated with pH in both species (Figure 1; Table 2). Element ratios on average were higher in Mytilus galloprovincialis larval shells than M. californianus larval shells; 35% higher for U/Ca and 7% higher for Sr/Ca (Figure 1). Since Keul et al.11 revealed that incorporation of uranium in benthic foraminiferal calcite indicated [CO32−], we also performed regressions of U/Ca as a function of [CO2− 3 ]. These yielded a similar coefficient of determination, r2, as when pH was used as the predictor variable, since these experiments varied pH by changing total dissolved inorganic carbon with constant total alkalinity (Table 2). Using measured seawater U/ Ca ratios (Table 1), the partition coefficient for uranium in larval shell (DU) was calculated according to DU = (U/Ca)shell/ (U/Ca)seawater. DU was negatively correlated with pH in both species (Table 2; Figure 1). Multielemental signatures using U/Ca, Sr/Ca, Cu/Ca, and Zn/Ca were generated with principal components analysis (PCA) to explore whether a linear combination of multiple elements provided a more reliable model for pH. This PCA resolved three principal components that together explained 97.7% and 97.4% of the total variance in elemental composition for M. californianus and M. galloprovincialis, respectively (Table 3). For M. californianus, the first principal component had high loadings from all four elements: Cu/Ca, Zn/Ca, Sr/Ca, and U/ Ca; for M. galloprovincialis, the first principal component had high loadings from three of the elements: Zn/Ca, Sr/Ca, and U/Ca. Regression analysis of PC1 against pH revealed a significant negative correlation for both species (Figure 1; Table 2). Of the remaining principal components, only PC3 varied with pH in M. californianus and had high loadings of Cu/ Ca and U/Ca (Figure 1; Tables 2 and 3). These regressions did not perform as well as the single elemental regressions with pH. While U/Ca varied as a function of pH, it is important to ensure that other environmental parameters do not influence the incorporation of uranium into the larval shell. For example, in many coastal settings pH and dissolved oxygen ([O2]) covary with strong fluctuations in time and space.22,25−27 To 6403
dx.doi.org/10.1021/es500514j | Environ. Sci. Technol. 2014, 48, 6401−6408
Environmental Science & Technology
Article
Table 3. Principal Component Analysis Based on Cu/Ca, Zn/Ca, Sr/Ca, and U/Ca for (a) Mytilus californianus and (b) M. galloprovincialis Larval Shell Compositiona (a) Mytilus californianus PC1 eigenvalues percent (%) cum. percent (%) element log(Cu/Ca) log(Zn/Ca) log(Sr/Ca) log(U/Ca)
PC1 eigenvalues percent (%) cum. percent (%) element log(Cu/Ca) log(Zn/Ca) log(Sr/Ca) log(U/Ca)
PC2
PC3
2.41 0.86 60.3 21.4 60.3 81.7 eigenvectors −0.462 0.578 0.416 0.796 0.624 0.033 0.473 −0.177 (b) Mytilus galloprovincialis 2.25 56.2 56.2 eigenvectors 0.162 −0.496 0.634 0.571
0.64 16.0 97.7 0.534 −0.203 −0.088 0.816
PC2
PC3
1.14 28.5 84.6
0.51 12.8 97.4
−0.840 0.364 0.172 0.364
0.517 0.755 0.109 0.388
a
All experiments and treatments are included in principal components analysis. Variables with high loadings on each of the principal components are indicated in boldface.
test for interactive effects of low [O2] and variable pH on the incorporation of elements, a low oxygen treatment and a variable pH treatment were incorporated in the culturing experiments (Table 1). U/Ca of larval shells precipitated at pH values between 7.9 and 8.0 was not significantly altered by exposure to high or low [O2] (e.g., 220−240 μmol kg−1 versus 86−104 μmol kg−1) (M. californianus t test: p-value = 0.898; M. galloprovincialis t test: p-value = 0.084). Similarly, U/Ca of larval shells precipitated in low pH treatments were not significantly different between the high and low [O2] treatments (M. californianus t test: p-value = 0.111; M. galloprovincialis t test: p-value = 0.209). When M. californianus larvae were exposed to the same mean pH (7.51), incorporating
Figure 1. U/Ca, Sr/Ca, partition coefficient for uranium (DU) and principal components (U/Ca, Sr/Ca, Cu/Ca, and Zn/Ca) in larval shells as a function of pH and [CO2− 3 ] for the mussel species Mytilus californianus (brown circles) and M. galloprovincialis (blue triangles). Element-Ca of field-cultured M. californianus larval shells (open brown circle) are added to the plot for comparison but not included in statistical analysis. Error bars are ±1 SE.
Table 2. Summary of Element-Ca, Partition Coefficient of Uranium (DU), and Principal Component Linear Relationships with a −1 pH and [CO2− 3 ] (μmol kg ) in the Mussel Species Mytilus californianus and M. galloprovincialis response Mytilus californianus log(U/Ca) log(Sr/Ca) PC1 PC3 DU log(U/Ca) Mytilus galloprovincialis log(U/Ca) log(Sr/Ca) PC1 DU log(U/Ca)
relationship
r2
p
1.093(±0.15) 0.994(±0.184) 35.0(±10.9) 17.7(±5.7) 5.35(±0.69) 0.247(±0.010)
− − − − − −
0.116(±0.019) pH 0.066(±.024) pH 4.52(±1.41) pH 2.29(±0.73) pH 0.625(±0.089) pH 0.0006(±0.0001) [CO32−]
0.67 0.30 0.37 0.35 0.83 0.65
