A depth transect of deep-sea bamboo corals along the California margin provides evidence that coral strontium to calcium ratios (Sr/Cacoral) record seawater Sr/Ca ratios (Sr/Casw). A calibration was constructed utilizing Sr/Cacoral ratios and previously published Pacific Sr/Casw data (R2 = 0.53, n = 12, p < 0.01): Sr/Cacoral (mmol/mol) = 4.62*Sr/Casw (mmol/mol) − 36.64. Sr/Casw is ultimately governed by the remineralization of Sr-containing shells of surface water-derived marine organisms (e.g., Acantharia) at intermediate water depths. California margin Sr/Cacoral records from 792 and 1295 m document fluctuations in Sr/Casw that appear decadal-scale. These results suggest that Sr/Casw may not be as stable as previously assumed and may be influenced by surface productivity on short timescales.
All corals are from the California margin. Corals were collected via submersible in 2004 or 2007, with the exception of BB, collected via dredge. All coral specimens were utilized to sample outer edge calcite to compare to ambient seawater Sr/Ca composition via ICP-OES analyses. Four corals in bold are the focus of LA-ICP-MS analyses. Where possible, species identifications were made during shipboard sampling.
2. Materials and Methods
 Bamboo corals (genera Isidella, Keratoisis, and Lepidisis) were collected via remotely operated vehicle or dredge from 250 to 2136 m water depth (Table 1). In most cases, corals possessed living polyps at the time of sampling so were assumed to be actively calcifiying. We utilized established sampling methods for bamboo corals [Hill et al., 2011; LaVigne et al., 2011; Roark et al., 2005; Thresher et al., 2010]. Sr/Cacoral from outer margin calcite (most recent calcification) was determined for twelve corals using ICP-OES analyses, and compared to published Sr/Casw data. For these analyses, two replicate powdered samples (∼1 mg) were extracted from the outer 1–2 mm of each of coral. Powdered samples were dissolved in 1N HNO3 and analyzed for Sr/Ca at UC Davis Bodega Marine Laboratory via solution ICP-OES (JY Ultima 2 C). The analytical reproducibility for solution Sr/Ca measurements on an in-house matrix-matched bamboo coral consistency standard was ±1.3% RSD (n = 8).
 Four of the twelve corals (Table 1) were also thin sectioned (>100 um thickness) and polished for laser ablation-inductively coupled plasma mass spectrometry (LA-ICP- MS) analyses. Sr/Ca ratios via LA-ICP-MS were determined at UC Davis using a New Wave Research UP-213 laser ablation system equipped with a Nd: YAG deep UV-213 nm laser with Supercell, and an Agilent Technologies 7500a ICP-MS. The LA-ICP-MS analyses utilize a 15 Hz repetition rate, He sweep/carrier gas, and an individual circular 40 μm spot size (100 μm spacing). LA-ICP-MS data were acquired from outer margin to the center of the coral, following the growth axis (most recent to oldest material). Raw count rates were calibrated using a NIST-612 glass standard with periodic measurement of NIST 612 and USGS standard MACS-1 during the five day analytical period to determine accuracy, precision and instrumental drift. Off-line data analysis and blank correction were performed using Glitter data reduction software. Repeat analyses of MACS-1 indicated concentrations (206–242 ppm) consistent with the established range of Sr concentrations for this standard (200–240 ppm; USGS [Strnad et al., 2009]). Reproducibility of the Sr/Ca data was assessed by two mechanisms. First, repeated analyses of MACS-1 over multiple days (24 analyses, 2 days) yielded a SD of +/−0.04 mmol/mol. Additionally, parallel LA-ICP-MS tracks were collected from outer margin toward the center of coral sample T669 A1, yielding a reproducibility of +/−0.03 mmol/mol, based on the reproducibility of the mean of 30 points analyzed twice over 2 days. These data correspond to a ±0.85–1.05% precision on the LA-ICP-MS Sr/Cacoral measurements.
 Using a comparison of four corals in this study, the range of Sr/Cacoral ratios obtained via LA-ICP-MS for the outer 1 mm of the coral skeletons are found to be within +/−0.21 mmol/mol of solution ICP-OES analyses of the same coral (Table 2). We attribute the offsets between the two analytical methods to be a result of non-matrix match standardization by LA-ICP-MS (use of NIST glass standards) and small day to day variability in LA-ICP-MS. Therefore, we plot LA-ICP-MS data as anomalies from mean Sr/Cacoral; this method simultaneously removes variability associated with potential species-specific offsets given that results from three different species are reported here.
Table 2. Comparison of Coral Samples Analyzed Via ICP-OES and LA-ICP-MSa
ICP OES (mMol/mol)
LA-ICP MS (mMol/mol)
Samples for ICP-OES were drilled from the outer 1–2 mm and analyses reflect the mean of 2 discrete samples. Samples for LA-ICP-MS were ablated from outer 1 mm and reflect a mean of 10 discrete samples; 1σ error ranges shown.
3.246 +/− 0.025
3.460 +/− 0.035
3.156 +/− 0.012
3.029 +/− 0.025
3.052 +/− 0.042
3.086 +/− 0.028
3.163 +/− 0.013
3.068 +/− 0.031
 Two of the four corals sampled at depths of peak Acantharia remineralization were further investigated in time series (T661 A9, 792 m and T664 A17, 1295 m). These specimens were tied to independent chronologies based upon multiple calcite radiocarbon dates in each coral (Table 3), and a modern tiepoint dictated by the fact that both corals were sampled alive in 2004 (i.e., youngest age = 2004). These chronologies rely on the long-term average extension rate, which is preferred over relying on intermittent growth band counts [Hill et al., 2011; LaVigne et al., 2011; Noé et al., 2008]. The corals were sampled for radiocarbon analyses using a drill press (500 μm diameter bit). Radiocarbon analyses follow standard methods outlined in Hill et al. . Results are reported as conventional (uncorrected) radiocarbon age and per mil (‰) D14C as per international convention put forth in Stuiver and Polach  where:
The radiocarbon “age” of the calcite skeleton reflects that of the surrounding DIC, based upon previous results comparing bamboo corals and ambient seawater DIC from the California margin [Hill et al., 2011; Roark et al., 2005]. Corals T661 A9 and T664 A17 exhibit radiocarbon ages of 1635 +/−30 and 2010 +/−35 years, respectively, consistent with corals and seawater from intermediate depths along the California margin [Hill et al., 2011]. To construct a chronology, extension rates calculated from three radiocarbon ages for each coral are averaged to generate one extension rate for each coral (variability of extension rates from one sample to another fall within the margin of error; Table 3). We consider this method to be a ‘conservative’ approach to understanding coral growth and chronology, because it does not attempt to account for changes in growth that may have occurred throughout the lifetime of the coral.
Table 3. Radiocarbon Measurements and Calculated Growth Ratesa
Distance From Edge (mm)
Calculated Growth Rate (mm/yr)
One average extension rate is applied to each coral, because variability of extension rates from one sample to another fall within the margin of error. Previous research comparing growth band features with radiocarbon-based growth rates indicate that these two methods are typically consistent with one another but growth bands are not always identifiable in bamboo corals and can exhibit extremely variable extension rates instead of capturing a long-term average [Hill et al., 2011; LaVigne et al., 2011].
Sample T661 A9
0.048 +/− 0.015
0.063 +/− 0.044
Average growth rate
Sample T664 A17
0.144 +/− 0.094
0.160 +/− 0.118
Average growth rate
3. Coral Sr/Ca Proxy
 Although existing Sr/Casw data are sparse, we compare solution phase Sr/Cacoral data from twelve specimens (outer 1–2 mm margin samples; 250–2136 m water depth) with published water column Sr/Casw ratios collected from two transects in the western Pacific (10°S, 179°E; 45°N, 179°E; Figure 1) [de Villiers, 1999]. Powdered samples extracted from the outer 2 mm of coral calcite reflect an average of one decade of coral growth, based upon the range of bamboo coral growth rates reported here and in Hill et al. . Very few water column Sr/Casw data are available for comparison, so these sites, although geographically distant, provide the best opportunity for Sr/Ca calibration. The transect from 10°S exhibits a depth zone characterized by increased Sr/Casw from ∼250–1600 m, consistent with previous observations by Brass and Turekian  (Figure 1). The transect from 45°N exhibits a slightly deeper zone of increased Sr/Casw, extending to 2000 m. An analogous peak in Sr concentration is also apparent at 700 m in a very low resolution water column profile from the Northeast Pacific (28°N, 121°W) [Brass and Turekian, 1974]. Sites investigated in the Atlantic present a similar vertical gradient [de Villiers, 1999]. [de Villiers, 1999] propose that the higher surface Sr/Casw ratios at the 45°N site (relative to the 10°S site) reflect higher nutrient composition (thus, higher productivity) of these waters; this may also explain the broader depth zone of increased Sr/Casw at the 45°N site. Sr/Casw is also influenced by water mass history, with higher pre-formed ratios in the deep Pacific than in the North Atlantic [de Villiers, 1999].
 Sr/Cacoral ratios from within the zone of elevated Sr/Casw track the water column Sr/Casw trends, exhibiting elevated Sr/Cacoral ratios (3.20–3.30 mmol/mol) relative to those sampled from below this depth (3.05–3.18 mmol/mol; Figure 1, top). The observation that Sr/Cacoral changes with depth has been previously made in another bamboo coral study, based in the Southern Ocean and Western Pacific [Thresher et al., 2010]. It is also possible to develop a more direct comparison of Sr/Cacoral to Sr/Casw, by compiling a master water column profile of Sr/Casw ratios from across depths spanning those where coral samples were acquired (within ±100 m). This task is accomplished by merging the two Pacific Sr/Casw transects [de Villiers, 1999], averaging when possible their respective points at each discrete coral sample depth. In the case of the two corals that were acquired from depths not sampled by de Villiers , an estimated value of Sr/Casw was determined by linear interpolation of points from the master water column data set. This analysis yields a statistically significant relationship (R2 = 0.53, n = 12, p < 0.01) between Sr/Cacoral and Sr/Casw (Figure 1, bottom):
This correlation suggests that Sr/Casw exerts considerable control on Sr/Cacoral. Based upon these data, we calculate a partition coefficient of D = 0.35–0.38 (where D = Sr/Cacoral/Sr/Casw; Table 4). These values are comparable to those of other taxa. For example, D values for foraminifera, gastropods, coccolithophores are 0.15, 0.18, and 0.32 respectively, calculated utilizing published Sr/Cacalcite data and a surface ocean average Sr/Casw of 8.55 [de Villiers, 1999; Sosdian et al., 2006; Stoll et al., 1999, 2002]. Similarly, D values for fish otoliths, primarily controlled by Sr/Casw, range from 0.18 to 0.34 [Bath et al., 2000; Phillis et al., 2011]. The inclusion of multiple coral species does not bias the regression, as the slope of the line resulting from inclusion of only Isidella samples does not differ significantly from the full regression (t = 0.60, p > 0.5).
ICP-OES solution based analyses, reflecting an average of two samples for each coral. Temperature and salinity from Levitus . Calculated D-values for each coral are shown and described in the text.
 Variability in the twelve Sr/Cacoral samples is larger (2.35%) than the observed Sr/Casw variability across the same depth range (0.4%). This may in part be due to the temporal and spatial differences between the coral and water sampling – water sampling occurred at a single, discrete time, whereas the outer 2 mm of coral calcite reflects approximately a decade of growth, thus incorporating seasonal and annual variability in Sr/Casw. The total global Sr/Casw variability documented by de Villiers  was 2–3%, placing the Sr/Cacoral variability within the range of observed Sr/Casw variability.
 Sr/Casw exhibits a predictable relationship with other seawater nutrients, such as phosphate [Brass and Turekian, 1974; de Villiers, 1999]; based upon this, we can predict that corals along the California margin would experience a wider range of Sr/Casw values (with depth, and perhaps temporally) than those in the western Pacific (Figure S1 in the auxiliary material). Another possible driver of an increased Sr/Cacoral gradient with depth (relative to Sr/Casw plotted here) may be biases associated with the timing of calcification (e.g., if corals calcify faster during time periods of increased food/nutrients); however, very little is known about the temporal patterns of bamboo coral calcification rates. While no statistical relationship exists between Sr/Cacoral and parameters such as temperature and salinity (Table 4), there is an inverse relationship between depth (pressure) and Sr/Cacoral (r2 0.48, p < 0.02); however, Sr/Casw and depth (pressure) covary so cannot be distinguished as separate mechanisms influencing Sr/Cacoral. The relationship between Sr/Cacoral and depth may represent the effect of food quantity/quality to the corals, which could influence growth rate. However, we note that in four corals studied here, where it was possible to measure growth band width, there is no reliable statistical relationship between band width and Sr/Cacoral (Figure S2), a finding supported by another trace elemental investigation in bamboo corals [Sinclair et al., 2011].
4. A Centennial Scale Record of Sr/Ca Variability
 We plot Sr/Cacoral anomalies for four corals from margin to edge of the coral (e.g., most recent to oldest coral growth; Figure 2). Sr/Cacoral ratios from two Pacific corals located within PDW and below the celestite remineralization zone (2043 m and 2136 m) remain relatively stable (SD ± 1.15 and 0.98% of the mean, respectively; Table 5 and Figure 2), as one would expect for a depth region absent of strong Sr inputs. In contrast, the two corals from shallower sites near the deeper boundary of NPIW (792 m and 1295 m) exhibit higher Sr/Ca variability (SD ± 1.47 and 2.00% of mean, respectively; Table 5 and Figure 2). This dichotomy suggests that the Sr/Casw composition of NPIW is more variable than that of PDW. This trend is consistent with documented peaks in [Sr] and Sr/Casw in intermediate waters, due to the dissolution of Acantharia shells sinking from surface waters [Brass and Turekian, 1974; de Villiers, 1999; MacKenzie, 1964]. The Sr/Cacoral variation likely reflects changes in ambient Sr/CaSW due to variability in Acantharia flux from the surface ocean and subsequent remineralization of shell celestite at NPIW depths. Alternatively, the Sr/Cacoral record may indicate periodic shoaling of the upper boundary of PDW (lower in Sr/Casw).
Mean, standard deviation and variance expressed as a percentage of mean for Sr/Cacoral ratios. Shallower (NPIW) corals exhibit higher variability than deeper (PDW) corals. Reproducibility of LA-ICP-MS Sr/Cacoral data is 0.85–1.05%.
 We evaluate this further by analyzing the two shallower corals (792 and 1295 m; within NPIW) in time series. These corals exhibit notable oscillations in Sr/Cacoral over the past 100 years (Figure 3), suggesting a potential link between intermediate water Sr/Casw and decadal scale surface water variability. Development of more accurate chronological techniques in bamboo corals would further elucidate the timing relationships between these processes. The wide-ranging impacts of decadal scale climate change have been well documented in the surface ocean and terrestrial environments [Hare and Mantua, 2000; Mantua et al., 1997; Trenberth and Hurrell, 1994]. Although there is little doubt that dramatic surface productivity shifts will influence ecosystems beyond the photic zone, there is limited evidence to date for the penetration of short-term variations in environmental variability to the deep sea [Corno et al., 2007; Ruhl and Smith, 2004; Ruhl et al., 2008; Smith et al., 2009; Wu et al., 1999]. Bamboo corals may provide an opportunity to reconstruct such surface-to-deep water linkages.
 Here we provide evidence that Sr/Cacoral is a proxy for Sr/Casw, and that Sr/Cacoral records show significant variability on decadal timescales. These Sr/Cacoral records provide new insight into variations in deep sea Sr/Casw shifts, potentially driven by surface water productivity. These findings indicate broad scale shifts in deep ocean Sr/Casw content over the past 100 years. These results require a reconsideration of previous paleoceanographic data that assumed stable deep-sea Sr/Casw and provide evidence for direct linkages between surface water productivity and alterations to intermediate water geochemistry.
 California seamount sampling and geochemical analyses were supported by NOAA West Coast Polar Regions Research Program (NA030AR4300104 to T.M.H. and H.J.S.) and NSF (OCE 0647872 to T.M.H.). DAC and collection of samples using the R/V Western Flyer and ROV Tiburon were supported by a grant to MBARI from the David and Lucile Packard Foundation. A portion of this work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory (contract W-7405-Eng-48) including LLNL IGPP funding (H.J.S. and T.P.G.). We appreciate the technical support of M. Gras-Oqvist, UCD Interdisciplinary Center for Plasma Mass Spectrometry, and staff of LLNL CAMS. This manuscript benefited from suggestions by R. Thresher and two anonymous reviewers, as well as guidance from R. Zahn.