Article pubs.acs.org/ac
Strontium Randomly Substituting for Calcium in Fish Otolith Aragonite Zoë A. Doubleday,*,† Hugh H. Harris,*,‡ Christopher Izzo,† and Bronwyn M. Gillanders†,§ †
Southern Seas Ecology Laboratories, DX 650 418, School of Earth and Environmental Sciences, The University of Adelaide, Adelaide, South Australia 5005, Australia ‡ School of Chemistry and Physics, The University of Adelaide, Adelaide, South Australia 5005, Australia § Environment Institute, The University of Adelaide, Adelaide, South Australia 5005, Australia S Supporting Information *
ABSTRACT: The chemistry of fish ear bones (otoliths) is used to address fundamental questions in fish ecology and fisheries science. It is assumed that strontium (Sr), the most important element used in otolith chemistry research, is bound within the aragonitic calcium carbonate lattice of otoliths via random chemical replacement of calcium; however, this has never been tested and three other alternatives exist with regard to how Sr may be incorporated. If any variation in the mode of incorporation occurs, otolith chemistry data may be misinterpreted, impacting how fish and fisheries are understood and managed. Using X-ray absorption spectroscopy (specifically, analysis of extended X-ray absorption fine structure or EXAFS), we investigated how Sr is incorporated within fish otoliths from seven species collected from a range of aquatic environments. For comparison, aragonitic structures from other aquatic taxa (cephalopods and coral) were also analyzed. The results consistently indicated for all samples that Sr randomly replaces Ca within the aragonite lattice. This research explicitly shows how Sr is bound within otoliths and validates a fundamental and longheld assumption in aquatic research.
F
otolith. Due to the chemical characteristics of Sr, it has been assumed until now that it, along with many other elements, randomly substitutes for Ca within the aragonite crystal. This is seen as the ideal scenario for otolith chemistry studies, as elements in the otolith are more likely to reflect the physical and chemical properties of the water in a predictable and consistent manner. Yet, this fundamental assumption, on which a plethora of important studies have been based, has never been tested in otolith aragonite for any element. A small number of X-ray spectroscopic studies has investigated how Sr is incorporated in the aragonitic skeletons and shells of several coral and bivalve species and has verified random Sr−Ca substitution in the matrices of both taxa.7−10 However, one study observed, in two coral species, that 40% of Sr resided as discrete clusters of strontianite, rather than being randomly distributed.11 This suggests that variation in Sr concentration within the coral skeletons may be, in part, caused by variation in the occurrence of aragonite and strontianite rather than variation in the environment. Such an observation may have significant implications for how Sr coral chemistry is
ish ear bones, or otoliths, are paired calcified structures found in all scalefish (teleost) species.1 They are used for hearing and balance and are typically composed of aragonite, a form of calcium carbonate. Otoliths grow incrementally throughout the lifetime of a fish and can provide excellent time-calibrated archives of chemical information.2 Otolith chemistry research, which seeks to reconstruct the environmental histories of fish based on this chemical information, has increased exponentially around the world since its development in the 1980s.1−3 It has been applied to a large number of species representing a diverse range of aquatic environments, providing critical insights into the dispersal and migration patterns, stock structure, recruitment dynamics, and habitat utilization of fish.2−6 Strontium (Sr) is considered the most informative and extensively used element in otolith chemistry studies due to its high abundance, ease of detection, and heterogeneous distribution throughout a wide range of aquatic environments. The seminal otolith chemistry review by Campana1 identified four possible ways Sr may be incorporated into the otolith: (1) random substitution of Ca within the aragonite crystal lattice; (2) nonrandom substitution of Ca within the crystal lattice, forming Sr-rich clusters of the mineral strontianite (Sr carbonate); (3) within the interstitial spaces of the crystal lattice; or (4) in association with the organic component of the © 2013 American Chemical Society
Received: October 22, 2013 Accepted: December 3, 2013 Published: December 3, 2013 865
dx.doi.org/10.1021/ac4034278 | Anal. Chem. 2014, 86, 865−869
Analytical Chemistry
Article
Table 1. Taxonomic Details, Habitat and Salinity (ppt), and Sr Concentration (ppm) of Samples Analyzed Using EXAFS common name golden perch Murray cod black bream mulloway snapper ocean perch small-mouthed hardyhead
southern calamary squid Australian giant cuttlefish scleractinian coral Port Jackson shark (vertebra)a broadnose sevengill shark (tooth)a a
scientific name Macquaria ambigua Maccullochella peelii Acanthopagrus butcheri Argyrosomus japonicus Chrysophrys auratus Helicolenus percoides Atherinosoma microstoma
Sepioteuthis australis Sepia apama Plesiastrea versipora Heterodontus portusjacksoni Notorynchus cepedianus
family Fish Species Percichthyidae Percichthyidae Sparidae Sciaenidae Sparidae Sebastidae Atherinidae
Other Taxa Loliginidae Sepiidae Faviidae Heterodontidae Hexanchidae
n 1 1 1 1 1 1 4
1 1 1 1 1
habitat and salinity (ppt)
Sr concentration (ppm)
freshwater, 0 freshwater, 0 estuarine (brackish), 30 estuarine (brackish), 28 coastal marine, 35 offshore marine, 35 estuarine (hypersaline) 1. 35 2. 60 3. 88 4. 132
972 1975 1652 843 1077 1046
marine, marine, marine, marine, marine,
3573 1042 4946 n/a n/a
35 35 35 35 35
1116 2027 1605 1875
Represents nonaragonitic out-group; note Sr was not analyzed for these samples.
reference standard (NIST612) was analyzed every 90 min to account for instrument drift. X-ray absorption spectroscopy (XAS), specifically extended X-ray absorption fine structure (EXAFS) analysis, was used to characterize the solid phase structure of Sr in each sample. Sr K-edge EXAFS data were collected on the XAS beamline at the Australian Synchrotron (AS), Victoria, Australia. EXAFS is an analytical technique that provides information regarding the average local environment surrounding a specific element within a chemically complex sample, including accurate interatomic distances and an estimate of the number and atomic size of neighboring atoms (see ref 13 for more details). XAS can thus be used to determine how Sr is bound within otolith aragonite and to distinguish between the four possible scenarios described by Campana.1 No sample preparation was required for the XAS reported herein; however, for some of the larger samples, a subsample of material was analyzed due to chamber-size restrictions of the cryostat. The 3 GeV electron beam at the AS was maintained at a current of 200 mA in top-up during the sample analysis. The X-ray beam was tuned with a Si (111) monochromator in the energy ranges of 15905−16085 eV for pre-edge (10 eV steps), 16085−16155 eV for the XANES region (0.5 eV steps), and then 0.035 Å−1 steps in k-space to 14 Å−1 for the EXAFS region. A metallic Pb foil, recorded in transmission mode downstream of the sample, was used as an external standard to calibrate the energy scale to the first peak of the first derivative of the Pb LI edge (15861.0 eV). The data were recorded in fluorescence mode on a 100-pixel Ge detector array at 90° to the incident beam (Canberra/UniSys), and the samples were maintained at ∼10 K during data collection in a Cryo Industries (Manchester, New Hampshire, USA) closed-cycle He cryostat. Calibration, averaging and background subtraction of all XAS data, and single-scattering EXAFS curve-fitting were performed using the EXAFSPAK software package (G.N. George, SSRL). The curve-fitting presented here followed the approach developed in earlier work where three distinct single-scattering shells were used to model the data.10 The first and second shells were Sr−O and Sr−C with coordination numbers fixed at nine and six, respectively, while the third had a coordination
interpreted and subsequently applied as a proxy of past environmental conditions.9 In light of this, it is crucially important to determine how Sr is bound within otolith aragonite. If Sr resides in the interstitial spaces, rather than the crystal lattice, for instance, it would more likely leach out of the structure postdeposition and not reflect the environment in which the fish has lived. Furthermore, if Sr forms discrete clusters of strontianite, rather than randomly replacing Ca, changes in Sr concentration throughout the otolith structure are less likely, again, to reflect variation in the environment. Either of these two scenarios could critically influence how chemistry data are interpreted and thus the way fish and fisheries are understood. Using an Xray spectroscopic approach, our study explicitly addressed this knowledge gap by examining how Sr is bound in the otolith aragonite lattice.
■
EXPERIMENTAL SECTION
Otoliths were analyzed from a range of native Australian fish species collected from freshwater, estuarine, marine, and hypersaline environments (Table 1). For structural comparison with other biogenic aragonites, a coral skeleton, cuttlefish cuttlebone, and squid ear bone (statolith) were also analyzed (Table 1). A shark tooth and vertebra were analyzed as a nonaragonitic “out-group”. As Sr concentration may influence the mode of incorporation, Sr concentration in each sample (see Table 1) was determined using laser ablation inductively coupled plasmamass spectrometry (LA ICPMS). Each sample (i.e., 1 otolith or statolith from each individual or a subsample of material from each of the other structures) was embedded in epoxy resin, sectioned, and mounted onto slides using standard procedures (see ref 12 for more details). Samples were analyzed for 88Sr on a New Wave Nd:Yag 213 nm UV laser connected to an Agilent 7500cs ICPMS. The laser was operated in Q-switch mode with a pulse rate of 5 Hz and power of 75%. Laser transects, 30 μm in width, were run at 5 μm s−1 from core to edge in the otoliths and statolith and at a randomly selected location in the other samples. Calcium was used as an internal standard, and a 866
dx.doi.org/10.1021/ac4034278 | Anal. Chem. 2014, 86, 865−869
Analytical Chemistry
Article
number fixed at six, but with a fully variable ratio of Sr−Ca and Sr−Sr contributions. Because the Sr−Sr component was always identified as minor in this work, the Sr−Sr distance was fixed to the value determined previously for pure strontianite10 and the Sr−Ca and Sr−Sr Debye−Waller factors (or disorder factors) were constrained to be equal. However, all other bond distances and Debye−Waller factors were allowed to vary during the curve-fitting analysis.
a distinct peak at approximately 2.5 Å. This represents the summation of the scattering interactions from the nearestneighbor oxygen and carbon atoms surrounding the central metal in aragonite, in this case Sr. The peak was significantly modified in the FT spectra of the nonaragonitic samples, suggesting a larger range of nearest-neighbor distances, but with a shorter average bond length. Curve-fitting analysis of the EXAFS data (see Table 2 and Supporting Information, Figures S1−S13, for more details) yielded structural parameters that were similar to those that have been reported previously for Sr in aragonitic coral matrices.9,10 The Sr−O and Sr−C bond distances fitted to the data for the aragonitic samples ranged from 2.57 to 2.61 Å and from 2.93 to 2.97 Å, respectively, consistent with the values determined for coral. The second distinct FT peak observed in all aragonitic samples, at approximately 4 Å, corresponds to the scattering interaction from the nearest-neighbor metal atoms surrounding the average photon-absorbing metal in aragonite (Figure 1). Data relating to this peak is particularly informative in relation to the structural state of Sr in aragonite: if the nearest metal to Sr is Ca, for instance, it can be assumed that Sr randomly substitutes for Ca, and if the nearest metal to Sr is another Sr atom, it can be assumed that Sr clusters together to form strontianite.7,9,11 The curve-fitting analysis indicated that the FT peak at 4 Å in our aragonitic samples was well modeled by a Sr−Ca interaction, which ranged from 3.97 to 4.00 Å in length, again consistent with the coral data. The curve fitting process allowed the inclusion of a partial Sr−Sr interaction for this FT peak at the ideal distance of 4.166 Å determined previously (see ref 10). However, the Sr−Sr interaction was never found to be a major component, with a maximum contribution of 12% (for coral) and an average contribution of 6%, significantly less than the estimated standard deviations for %Ca (Table 2). We noted that the inclusion of the extra Sr−Sr interaction in the model did not improve the fit error for any of the samples, despite the addition of an extra variable. As such, we found no evidence for the presence of this interaction in the aragonite samples. The distribution of Sr, however, is typically variable within otoliths due to variation in Sr exposure throughout the lifetime of the fish. It is feasible, therefore, that inhomogeneous incorporation of Sr within the otolith may lead to different regions having increased Sr−Sr interactions to those reported herein. Such heterogeneity in Sr distributions may explain the subtle variations in EXAFS profiles among taxa (Figure 1) and could contribute to the %Ca values in Table 2 being less than 100%. It is unlikely, however, given the significant uncertainties in EXAFS determined coordination numbers (i.e., ± 25%) and the calculated estimated standard deviation for %Ca discussed above, that such variation could be explicated by a bulk analysis of this type. Our EXAFS analysis produced larger Debye−Waller (disorder) factors that might otherwise have been expected, due to treating a range of distinct crystallographic distances as equal (see Table 2) and from a possible minor self-absorption effect of Sr at the concentrations reported. We noted, however, that the Debye−Waller factors reported here were significantly lower than those reported by Finch et al.10 This is a reflection of the fact that our data were recorded at cryogenic temperatures, while the previous data were presumably recorded at room temperature (the data collection temperature was not specifically reported). Our data also showed a significant FT peak at approximately 5 Å that is absent from the earlier work. This is presumably a complex multiple
■
RESULTS AND DISCUSSION Sr K-edge EXAFS Fourier Transform (FT) spectra for all aragonitic samples are shown in Figure 1. These spectra display
Figure 1. Phase-corrected Fourier transforms of Sr K-edge EXAFS spectra for all samples analyzed. The radius (Å) relates to the radial distribution of atoms around Sr. The peak at approximately 4 Å, highlighted by the red line, represents the Sr−Ca shell, which is characteristic of random Sr−Ca replacement in aragonite crystal. Samples in italics represent nonaragonitic out-group, and numbers after hardyhead samples represent salinity that sample was collected from. 867
dx.doi.org/10.1021/ac4034278 | Anal. Chem. 2014, 86, 865−869
Analytical Chemistry
Article
Table 2. EXAFS Curve Fitting Results for Aragonitic Samplesa species
R1
R2
R3
σ12
σ22
σ32
ΔE0b
%Cac
fit errord
Murray cod golden perch mulloway black bream snapper ocean perch hardyhead (35) hardyhead (60) hardyhead (88) hardyhead (132) squid cuttlefish coral
2.603(6) 2.588(5) 2.571(4) 2.597(5) 2.583(6) 2.606(6) 2.594(5) 2.588(5) 2.606(5) 2.603(5) 2.579(4) 2.591(5) 2.575(5)
2.984(11) 2.952(13) 2.93(2) 2.966(10 2.968(13) 2.973(10) 2.953(14) 2.946(12) 2.968(10) 2.964(8) 2.941(13) 2.962(12) 2.95(2)
3.996(12) 3.984(9) 3.971(7) 3.992(3) 3.966(10) 4.003(13) 3.986(10) 3.982(9) 4.006(12) 4.000(10) 3.978(8) 3.989(9) 3.969(10)
0.0076(4) 0.0081(4) 0.0070(3) 0.0085(4) 0.0085(5) 0.0071(4) 0.0072(4) 0.0073(3) 0.0077(4) 0.0078(4) 0.0083(3) 0.0074(3) 0.0087(4)
0.0084(14) 0.011(2) 0.0174(3) 0.0095(10) 0.010(2) 0.0080(13) 0.0117(14) 0.011(2) 0.0083(12) 0.0076(1) 0.013(2) 0.012(2) 0.018(3)
0.012(2) 0.011(2) 0.0093(11) 0.013(2) 0.011(2) 0.013(2) 0.0114(10) 0.011(2) 0.013(2) 0.012(2) 0.0107(12) 0.012(2) 0.012(1)
−10.7(9) −10.4(8) −11.4(7) −10.3(8) −11.5(9) −10.2(9) −10.5(8) −10.6(7) −9.5(8) −9.8(7) −10.6(7) −10.9(7) −11.6(7)
95(23) 100(17) 97(14) 91(20) 100(18) 94(25) 93(19) 93(18) 95(23) 89(18) 91(15) 100(17) 88(18)
0.598 0.579 0.478 0.581 0.610 0.610 0.550 0.526 0.598 0.560 0.508 0.505 0.535
Interatomic distances (R, Å); Debye−Waller factors (σ2, Å2). The subscript labels refer to Sr−O (1, coordination number fixed at nine), Sr−C (2, coordination number fixed at six), and Sr−Ca/Sr (3, total coordination number fixed at six) scattering interactions included in the model. bΔE0 = E0 − 16120 (eV) where E0 is the threshold energy. cSee data analysis section in Supporting Information for description of the %Ca parameter. All fits used starting parameters that resulted from the best fit to the black bream data, except that the %Ca was lowered to ∼15% prior to the start of each optimization. The Debye−Waller factors for the Sr−Ca and Sr−Sr interactions were constrained to be equal in all fits. The Sr−Sr bond distance was fixed in all cases to 4.166 Å, as previously determined for pure strontianite.10 Three component fits with a Sr−Ca coordination number set at six and no Sr−Sr component yielded fit errors within ±0.002 of those in the table indicating that the inclusion of the Sr−Sr component did not improve the fit. The k-range was 1−14.0 Å−1, and a scale factor (S02) of 0.9 was used for all fits. Values in parentheses are the estimated standard deviation derived from the diagonal elements of the covariance matrix and are a measure of precision. dThe fit-error is defined as [Σk6(χexp − χcalc)2/Σk6χexp2]1/2 and, as this depends on the noise level in each data set, cannot be directly compared across samples. a
■
scattering interaction that is enhanced at low temperature and by the longer data range in our study, but we have not attempted to model it here. Our results verify that Sr randomly substitutes for Ca in otolith aragonite and aligns with other studies that have examined the structural state of Sr in bivalve7,11 and coral9,10 aragonite. In contrast, Greegor et al.11 observed a 60/40% mixture of Sr in aragonite and strontianite in two scleractinian coral species, respectively, confirmed by a distinct double peak at approximately 4 Å in the FT of the Sr EXAFS. In Greegor’s study, Sr concentration in the coral skeletons ranged from 7000 to 7500 ppm, approximately 30% greater than the scleractinian coral analyzed in this study and 70% greater than the highest Sr level observed in the otoliths. The higher levels of Sr typically found in corals, relative to other biogenic aragonites, may influence how Sr is bound within the aragonitic lattice;1,8 however, the more recent EXAFS coral studies suggest that strontianite is not as widespread in coral aragonite as previously believed.10
ASSOCIATED CONTENT
* Supporting Information S
Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Authors
*Phone: +61 8 83131485. E-mail: zoe.doubleday@adelaide. edu.au (Z. A. Doubleday). *Phone: +61 8 83135060. E-mail:
[email protected] (H. H. Harris). Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We would like to thank Chris Glover for XAS technical support and advice and the South Australian Museum (sample code: H382) and Gretchen Grammer and Brenton Zampatti for donating samples. Parts of this research were undertaken on the XAS beamline at the Australian Synchrotron, Victoria, Australia (Reference No: AS131/XAS/5823). Funding was also provided through ARC grants awarded to B.M.G. (DP110100716, FT100100767) and H.H.H. (DP0985807).
■
CONCLUSIONS Our study validates the long-held assumption that Sr randomly replaces Ca within otolith aragonite and further corroborates the use of Sr as a key environmental tracer in otolith chemistry studies. The consistent results observed across species, families, and aquatic environments suggest the assumption likely holds true for other fish taxa. However, it should be noted that the median concentration of Sr in fish otoliths is approximately 2000 to 2500 ppm,1 and Sr may behave differently in otoliths with much higher levels of Sr than those analyzed in this study. Our study also confirms the assumption relating to Sr−Ca replacement in cephalopod statoliths, which have analogous research applications to otoliths. Further investigations into how other elements, commonly used in otolith chemistry (e.g., barium and magnesium), are incorporated within otoliths will further validate the approach and its utility in addressing fundamental questions in aquatic ecology.
■
REFERENCES
(1) Campana, S. Mar. Ecol.: Prog. Ser. 1999, 188, 263−297. (2) Elsdon, T. S.; Wells, B. K.; Campana, S. E.; Gillanders, B. M.; Jones, C. M.; Limburg, K. E.; Secor, D. H.; Thorrold, S. R.; Walther, B. D. Oceanogr. Mar. Biol. Annu. Rev. 2008, 46, 297−330. (3) Campana, S. E.; Thorrold, S. R. Can. J. Fish. Aquat. Sci. 2001, 58, 30−38. (4) Gillanders, B. M. Estuarine, Coastal Shelf Sci. 2005, 64, 47−57. (5) Walther, B. D.; Limburg, K. E. J. Fish Biol. 2012, 81, 796−825. (6) Pangle, K. L.; Ludsin, S. A.; Fryer, B. J. Can. J. Fish Aquat. Sci. 2010, 67, 1475−1489. (7) Foster, L. C.; Allison, N.; Finch, A. A.; Andersson, C. Geochem., Geophys., Geosyst. 2009, 10, Q03003.
868
dx.doi.org/10.1021/ac4034278 | Anal. Chem. 2014, 86, 865−869
Analytical Chemistry
Article
(8) Allison, N.; Finch, A. A.; Sutton, S. R.; Newville, M. Geochim. Cosmochim. Acta 2001, 65, 2669−2676. (9) Finch, A. A.; Allison, N. Geochim. Cosmochim. Acta 2003, 67, 4519−4527. (10) Finch, A. A.; Allison, N.; Sutton, S. R.; Newville, M. Geochim. Cosmochim. Acta 2003, 67, 1197−1202. (11) Greegor, R. B.; Pingitore, N. E. J.; Lytle, F. W. Science 1997, 275, 1452−1454. (12) Doubleday, Z. A.; Izzo, C.; Woodcock, S. H.; Gillanders, B. M. Aquat. Biol. 2013, 18, 271−280. (13) Levina, A.; Armstrong, R. S.; Lay, P. A. Coord. Chem. Rev. 2005, 249, 141−160.
869
dx.doi.org/10.1021/ac4034278 | Anal. Chem. 2014, 86, 865−869
