Sea-surface temperature reconstruction from trace elements variations of tropical coralline red algae

Quaternary Science Reviews 93 (2014) 34e46

Contents lists available at ScienceDirect

Quaternary Science Reviews journal homepage: www.elsevier.com/locate/quascirev

Sea-surface temperature reconstruction from trace elements variations of tropical coralline red algae Nicolas Darrenougue a, *, Patrick De Deckker a, Stephen Eggins a, Claude Payri b a b

Research School of Earth Sciences, The Australian National University, Canberra 0200, Australia Institut de Recherche pour le Développement, UR227, Nouméa, New Caledonia

a r t i c l e i n f o

a b s t r a c t

Article history: Received 26 August 2013 Received in revised form 22 February 2014 Accepted 9 March 2014 Available online 22 April 2014

We used laser ablation inductively coupled plasma mass spectrometry (LA-ICPMS) to obtain highresolution variations of the Mg/Ca, Sr/Ca and Li/Ca composition of free-living forms (i.e. rhodoliths) of the coralline red algal species Sporolithon durum in order to test their potential to archive seawater temperature information. A monitoring experiment was conducted based on alizarin red S (ARS) staining of rhodoliths specimens collected in various locations across a w1 km2 rhodolith bed in the vicinity of Nouméa, New Caledonia, where in situ temperature (IST) variations were recorded for 22 months between November 2009 and August 2011. A >45-year comparison of Mg and trace elements with seasurface temperature (SST) was established from the analysis of 5 different branches belonging to three of the largest (7.4e8.5 cm in diameter) rhodolith specimens observed at the site. Consistent mean Mg/Ca, Sr/Ca and Li/Ca concentrations and seasonal patterns are found for the rhodoliths’ last living years (2009 e2011) across 43 branches and for the full 1963e2008 period across the 5 branches. Average elemental concentrations (Mg/Ca: 0.31
0.04 mol/mol; Sr/Ca: 3.5
0.4 mmol/mol and Li/Ca: 0.08
0.02 mmol/ mol) fall within range of those found in the literature. Individual element variations show good reproducibility between records and Mg/Ca, Sr/Ca and Li/Ca co-vary systematically. Combined records of Mg/ Ca, Sr/Ca and Li/Ca are highly correlated with the IST monthly pattern for the 2009e2011 period (0.82 < r < 0.91; p < 0.001) and with local variations of monthly SST for the 1963e2008 period (0.65 < r < 0.85; p < 0.001), with Mg/Ca systematically being the best fit to monthly seawater temperature variations. Inter-annual Mg/Ca anomalies show significant correlation with the Oceanic Nino Index (ONI), indicating that S. durum rhodoliths also have the capacity to record the regional climate pattern in the tropical Pacific. Finally, consistent variations between the combined Mg/Ca record in S. durum rhodoliths and one Sr/Ca record of a Porites sp. coral from the same site, as well as a similar relationship with local SST at both monthly and interannual scales, suggest that S. durum rhodoliths have the potential to compare favourably with corals in terms of SST reconstruction. Ó 2014 Elsevier Ltd. All rights reserved.

Keywords: Rhodoliths CCA Sporolithon durum Laser ablation Mg/Ca Sr/Ca Li/Ca Alizarin red S ENSO

1. Introduction Coralline red algae are globally-distributed calcareous marine organisms that deposit a high-magnesium calcite skeleton during growth (e.g. Adey and MacIntyre, 1973; Bosence, 1983b). They occur either as attached forms on hard bottoms or as free-living nodules (i.e. rhodoliths) on unstable substrata (Steneck, 1986) where they can develop into thick crusts or nodules up to 20 cm in diameter (Adey and MacIntyre, 1973; Littler et al., 1991; Frantz et al., 2005).

* Corresponding author. Research School of Earth Sciences, Building 142, Mills Road, The Australian National University, Canberra ACT 0200, Australia. Tel.: þ61 33 6 9518 1663. E-mail address: [email protected] (N. Darrenougue). http://dx.doi.org/10.1016/j.quascirev.2014.03.005 0277-3791/Ó 2014 Elsevier Ltd. All rights reserved.

Coralline algae generally exhibit slow growth rates (0.015e 2.17 mm/yr e Foster, 2001; Blake and Maggs, 2003; Böhm et al., 1978) and individual organisms can live for more than 800 years (Halfar et al., 2007). This places coralline red algae among the oldest living shallow-water calcifiers (Frantz et al., 2005). As they grow, environmental information may be recorded in their calcite skeletons, including their annual growth pattern, which comprises alternating small cells and heavily calcified cell walls generally produced in winter, and longer cells with less calcified cell walls typically produced in summer (e.g. Halfar et al., 2008; Kamenos and Law, 2010). This commonly appears as paired clear and dark bands on visual examination of sectioned coralline red algae skeletons (Kamenos et al., 2008; Burdett et al., 2010; Kamenos and Law, 2010).

N. Darrenougue et al. / Quaternary Science Reviews 93 (2014) 34e46

Providing high-resolution archives is crucial to improve our understanding of climate variability at the seasonal and interannual timescales worldwide. In this endeavour, a number a recent studies have focussed on the climate information that is contained in coralline red algae, both within their growth increment variations and in the geochemical composition of their skeleton (e.g. Halfar et al., 2000). These endeavours have provided critical decadal to century-length reconstructions of environmental parameters, mainly for the northern hemisphere high latitudes (Halfar et al., 2007, 2008, 2011; Hetzinger et al., 2009, 2011; Kamenos, 2010; Chan et al., 2011; Williams et al., 2011). Particularly, the variation in Mg composition of coralline red algae with seawater temperature has proved the most promising tool for palaeo-environmental reconstructions (Halfar et al., 2000; Kamenos et al., 2008; Hetzinger et al., 2009, 2011; Kamenos, 2010). Sr/Ca records in coralline red algae are scarce but have also been shown to present significant, positive correlations with temperature for various highlatitude species (Kamenos et al., 2008; Hetzinger et al., 2011). Sporolithon durum is widely distributed in oceanic shallow waters, from tropical to temperate-cold environments (e.g. Townsend et al., 1995; Womersley, 1996; Goldberg and Heine, 2008; Basso et al., 2009). It has been reported to live for several decades (Goldberg and Heine, 2008; Darrenougue et al., 2013) and can display annual increments in its growth pattern (Darrenougue et al., 2013), however environmental reconstructions are yet to be generated for this species. Here, we assess the potential for Mg and trace elements variations in S. durum rhodoliths to be used for sea-surface temperature (SST) reconstruction in a tropical environment. This is aimed at confirming the potentiality of coralline red algae as reliable palaeoclimate archives where other proxies (e.g. corals) are absent or unusable. We have used laser ablation inductively coupled plasma mass spectrometry (LA-ICPMS) to measure Mg/Ca, Sr/Ca and Li/Ca variations at a monthly to sub-monthly resolution. The results were compared to solution ICPMS and ICP atomic emission spectrometry (ICP-AES) measurements on the same samples to assess the accuracy of the LA-ICPMS calibration. Variations of the studied elements were compared to instrumental records of both in situ temperature (IST) and local SST. The Oceanic Nino Index (ONI) was used to investigate the potential of interannual Mg/Ca anomalies in S. durum to record regional scale climate pattern. In addition, an evaluation of Mg/Ca in coralline algae as a tool for SST reconstruction was tested against the Sr/Ca record of a Porites sp. coral from the same site. 2. Material and methods 2.1. Study site and rhodoliths collection New Caledonia is a group of islands located in the western tropical Pacific Ocean, w1000 km offshore Australia. The 8000-km2 barrier reef on the west side of the main island defines one of the largest lagoons in the world, with an area of 23,400 km2 (Dandonneau et al., 1981; Labrosse et al., 2000). The typical annual pattern of climate in New Caledonia alternates between a warm and wet austral summer season (JaneMar) and a cooler and humid winter season (JuleSep). Dry conditions are generally observed in spring (OcteDec) and autumn (ApreJun). The interannual climatic variations are closely related to the El Niño Southern Oscillation (ENSO) pattern, with cooler and drier conditions during periods of El Niño whereas La Niña periods are generally responsible for warmer temperatures and heavier rainfall (Nicet and Delcroix, 2000). The Ricaudy Reef (22180 5700 S; 166 270 2600 E) is one of the fringing coral reefs that borders Nouméa, the most populated city

35

in New Caledonia (Fig. 1). Ricaudy Reef is located at the southern end of the Sainte Marie Bay in the southwest region of New Caledonia, where seawater within the lagoon is subject to episodic freshwater inputs from the nearby Coulée River (e.g. Fernandez et al., 2006). All samples studied here were collected at 4e5 m depth by hand using SCUBA from a rhodolith bed that covers w1 km2 on the edge of the Ricaudy Reef. The rhodolith samples are monospecific, of ellipsoidal shape and present a degree IV branching structure (according to the classification in Bosence, 1983a). A small number of rhodolith samples, representative of the size range observed within the bed (w4e8 cm in diameter), were selected for marking and monitoring in February 2011. These specimens were stained onshore for a period of 48 h using alizarin red S (ARS), at a concentration of 7.5 mg/l in seawater taken from the Ricaudy Reef (adapted from Payri, 1997). They were then returned to Ricaudy reef and placed in a w1  1 m2 mesh enclosure within their natural environment, where they were left to grow for 28 weeks before being retrieved at the end of August 2011. The IST was recorded hourly for the entire period using a TinyTag TG-4100 Aquatic 2 data logger, that had been attached to the enclosure since November 2009. For longer-term environmental reconstruction, three rhodoliths were collected in October 2009 (for the BSA specimen) and February 2011 (for the MSA and SSA specimens; note that BSA, MSA and SSA are catalogue numbers for the studied rhodolith branches). These specimens were selected for being some of the largest nodules within the bed (7.4e8.5 cm in diameter for SSA and MSA, respectively). 2.2. Mg and trace elements analyses Rhodoliths were dried and impregnated into araldite resin before being cut into w5-mm thick sections, parallel to their long axis (Fig. 2A). After polishing and ultrasonic cleaning, the thick sections were oven-dried (40  C) prior to LA-ICPMS analyses and/or sampling for solution ICPMS and ICP-AES analyses. 43 different rhodolith branches in which the alizarin stain was clearly marked (Fig. 2B), were analysed by LA-ICPMS along with five different branches belonging to the samples BSA (3 branches), SSA and MSA (1 branch each) which were also analysed by solution ICPMS and ICP-AES. LA-ICPMS analyses were conducted using an ArF excimer laser (193 nm wavelength) and in-house built ANU HelEx laser ablation cell in combination with a Varian 820 ICPMS, at the Research School of Earth Sciences (RSES) of the Australian National University (ANU). The laser ablation sampling was performed using a spot diameter of 42 mm, laser fluence of w5 J/cm2 and a pulse rate of 10 Hz. The laser was scanned across pre-defined tracks following the main growth axis along the branches at a speed of 5 mm/s. The ICPMS was set to time-resolved analysis mode using 1 point-perpeak on each of the isotopes 7Li, 24Mg, 25Mg, 43Ca and 88Sr, and different dwell times that resulted in a total integration time of w1 s per measurement cycle. The same analysis protocol was applied to each run and comprised a pre-analysis ablation to remove surface contamination, and 60 s bracketing measurements of standard materials and background levels before and after the sample acquisition, which typically lasted about 7 min for analyses of the last living years and about 2 h for the longer-term analyses. The data reduction procedure followed that described in Longerich et al. (1996), with 43Ca used as an internal standard to correct for variation in ablation yield. The bracketing background levels were extrapolated to correct for drift in instrumental background noise. The standard deviation of the measured background was used to determine detection limits (defined here as 3s of the background

36

N. Darrenougue et al. / Quaternary Science Reviews 93 (2014) 34e46

Fig. 1. Location map showing the Ricaudy Reef study site, at the southern end of the Sainte Marie Bay in the Southwest region of New Caledonia. The site where instrumental SST is recorded daily since 1958 (source: IRD e Institut de Recherche pour le Développement, Nouméa) is also shown.

standard deviation), which were typically: <0.19 ppm Li, <24 ppm Mg (at 24 amu), <9 ppm Mg (at 25 amu), <0.2& Ca and <0.12 ppm Sr. The bracketing standard measurements were used to make a linear correction for instrument drift during the run. A dolomite marble of known Mg/Ca composition, from the Carrara region, Italy (Herz and Dean, 1986), was used as an external, matrix-matched standard for Mg/Ca calibration. The glass reference NIST SRM 612, was used for Li/Ca and Sr/Ca calibration. Each LA-ICPMS cycle measurements correspond to w5 mm distance along the branches which were averaged to produce an effective w30-mm resolution so as to minimise instrument measurement noise (except for the “full resolution data”, see later). This is equivalent to a monthly to submonthly sampling resolution. All results are presented as element/ Ca molar ratios, and are plotted as distance from the outer growing edge of the rhodolith branches. Both solution ICPMS and ICP-AES analyses were performed on samples consisting of w1 mg of carbonate powder that were obtained by hand-drill (2-3 mm-diameter  w1 mm-deep holes) along branches of the BSA, SSA and MSA rhodoliths. With this strategy every sample encompasses 3 to 5 annual growth bands. Each sample (4 from both BSA and SSA and 5 from MSA) was dissolved into 10-ml of 2% HNO3 prior to analyses. Solution ICPMS analyses were performed on a Varian 820 quadrupole ICPMS (at RSES, ANU). Calibration curves were generated using four standard solutions with differing trace metal concentrations (prepared from AccuTrace solutions e National Institute of Standard and Technology) that were measured before, during and at the end of the analysis run. An ultrapure 2% HNO3 solution was used to measure instrument background levels and to

determine detection limits (3s of the background) for Li (0.026 ppm), Mg (2.5 ppm), Ca (59 ppm) and Sr (0.03 ppm). Instrument reproducibility, determined by measuring the same sample repeatedly (n ¼ 6) through the run was 2.8% for Li, 2.1% for Mg, 1.8% for Ca and 1.1% for Sr (1s). Mg, Ca and Sr were also analysed using a Varian Vista Pro Axial ICP-AES at the RSES, ANU, following the method of de Villiers et al. (2002). A matrix-matched standard solution made from AccuTrace solutions (National Institute of Standard and Technology) was employed for instrument calibration and also drift correction through the analytical run. Detection limits for Mg, Ca and Sr were respectively 0.01, 0.10 and 0.005 mg/l, and the standard deviation of the measured elemental concentrations were <0.74% (Mg), <0.75% (Ca) and <0.70% (Sr). The detection limit was too high to enable the determination of Li. 2.3. Age model determination Age models for each rhodolith were established using multiple approaches (see Darrenougue et al., 2013 for details). In brief, the collection date and the position of the ARS stain layer were combined with the seasonal cycle of major Mg/Ca variations to determine the chronology for the last living year of the monitored specimens. For rhodoliths BSA, MSA and SSA, concordant results were obtained between radiocarbon dating and the combined Mg/ Ca seasonal cycles and growth band counting method. Note that for the MSA and SSA rhodoliths, different branches with overlapping Mg/Ca cycles and visually recognisable continuous growth bands were used and for the chronology. Elements records for

N. Darrenougue et al. / Quaternary Science Reviews 93 (2014) 34e46

37

Fig. 2. A: Whole BSA rhodolith specimen (top) and a section parallel to the long axis of growth (bottom) showing the path of three laser tracks (dashed lines). Scale bar: 2 cm. B: Close up on the top part of a rhodolith branch where the laser track appears as the vertical trough and the alizarin red S stained layer is pointed by the pink arrow. Also note the growth increments figured by the alternation of clear and darker bands at different scales (see Darrenougue et al., 2013 for details). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

these overlapping parts of branches were averaged. The AnalySeries data analysis software (Paillard et al., 1996) was used to tie high and low Mg/Ca seasonal peaks to months with the highest and lowest seawater temperatures (both IST or SST) for any given year. Note that, although rhodoliths appear not to grow consistently throughout the year (see Darrenougue et al., 2013), the sampling resolution was such that we could attribute at the least, one data point to any given month throughout the studied period. The same anchor points were used for the Li/Ca and Sr/Ca records (Fig. 3). Data between these anchors were interpolated linearly and resampled to produce a monthly resolution. Given the previously determined, average annual growth rate for the S. durum branches studied here (w0.6 mm/y, Darrenougue et al., 2013) and the spatial resolution obtained from LA-ICPMS analyses (w5 mm per measurement, 42 mm spot diameter), full resolution records would correspond to be approximately fortnightly. However, due to the limitations of the assumptions of our age model, we consider monthly-resolution data to be appropriate for comparison with instrumental environmental records. Interannual (i.e. year-to-year) variations of Mg and trace elements were obtained by applying a 25-point Hanning filter (i.e. weighted average) to the monthly data to filter out seasonal variations from the primary data (e.g. Corrège et al., 2000; Le Bec et al., 2000; Nicet and Delcroix, 2000). 2.4. Statistical analyses We used the AnalystSoft Inc., StatPlus:mac LE 2009 statistical analysis program (www.analystsoft.com) to perform linear regressions providing statistical comparisons between the Mg/Ca, Sr/ Ca and Li/Ca records from the LA-ICPMS analyses (Table 1). This was also used to compare the records obtained from those elemental ratios with the instrumental temperature dataset available for the study site (Table 2). Throughout the manuscript, for each

regression, the Pearson’s correlation coefficients (r) as well as the associated level of significance (p) are indicated. When monthlyresolved data were resampled to interannual resolution, the loss of degrees of freedom was taken into account and the p values recalculated accordingly.

3. Results 3.1. LA-ICPMS All analysed rhodolith branches, including the ARS stained rhodoliths and those used for long-term reconstruction, produce comparable average elemental concentrations, standard deviations and concentration range (Table 1). The average Mg/Ca value and standard deviations for all branches vary from 0.29 to 0.31
0.04e 0.03 mol/mol, with an overall population mean value of 0.30
0.04 mol/mol. Individual Mg/Ca values range from 0.19 to 0.49 mol/mol (Table 1). The average Sr/Ca concentration for all the records is 3.5
0.4 mmol/mol, with average values ranging from 3.4 to 3.6 mmol/mol, depending on the analysed branch. The minimum Sr/Ca value is 2.2 mmol/mol and maximum is 5.2 mmol/ mol. Li/Ca concentrations average 0.08
0.02 mmol/mol for all branches, with a large range of individual Li/Ca values from 0.03 to 0.17 mmol/mol (Table 1). The high cross-correlation coefficient between each elemental ratio, at every level of comparison (Table 1) indicates strong covariation of Mg/Ca, Sr/Ca and Li/Ca in the S. durum rhodoliths (see also Fig. 3). Sr/Ca and Li/Ca systematically display the strongest relationship with correlation coefficients varying from r ¼ 0.77 to r ¼ 0.90, and an r-value of 0.87 for the entire longer-term reconstruction dataset (p < 0.001). The pairs Mg/CaeSr/Ca and Mg/Cae Li/Ca show very similar correlation coefficients for any particular branch. Mg/Ca and Sr/Ca correlation coefficients ranging from

38

N. Darrenougue et al. / Quaternary Science Reviews 93 (2014) 34e46

Calendar years 2009

2007

2005

2003

2001

1999

0.40 0.35 0.30 4.4 0.25 4.0 0.20

3.6 3.2

Li/Ca (mmol/mol)

2.8 0.10

Sr/Ca (mmol/mol)

Mg/Ca (mol/mol)

2011

2.4

0.08 0.06

MSA1

0.04 0.02 0

2

4

6

8

10

Distance from top (mm) Fig. 3. Variation of Mg/Ca, Sr/Ca and Li/Ca along the top 10 cm of the MSA1 branch as recorded by LA-ICPMS. The top axis shows the corresponding chronology with and anchor points attributed to winter (blue triangles) and summer (red triangles) months of every year. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

r ¼ 0.62 to r ¼ 0.84 and the ones of Mg/Ca and Li/Ca vary from r ¼ 0.76 and r ¼ 0.85 (p < 0.001; Table 1). 3.2. Mg and trace elements calibration Mg/Ca and Sr/Ca results determined by LA-ICPMS, solution ICPMS and ICP-AES, and Li/Ca results determined by LA-ICPMS and solution ICPMS agree well, both in terms of absolute ratio values and in the variation trends measured along the rhodolith branches (Fig. 4). This indicates the validity of using the LA-ICPMS technique to obtain accurate Mg/Ca, Sr/Ca and Li/Ca ratio values in S. durum rhodoliths. Nonetheless, the results of the three methods do not match perfectly, even for the solution ICPMS and ICP-AES for which splits of the same samples were analysed. 3.3. Reproducibility of elements variations The elemental correlation matrices presented in Fig. 5A for the period November 2009eAugust 2011 suggest that monthly resolved data are slightly more reproducible than the fullresolution dataset. However, the degree of correlation is highly variable across the 43 analysed branches, with correlations ranging from insignificant (r < 0.3; p > 0.05) to highly significant (up to r ¼ 0.9; p < 0.001). Overall, Mg/Ca variations are the most reproducible across the different branches, with 95% of significantly correlated pairs of records (899 out of 946; r > 0.3; p < 0.05) at monthly resolution. Sr/Ca and Li/Ca are also very reproducible, with, respectively, 78% (738 out of 946) and 83% (785 out of 946) of significantly correlated pairs at monthly resolution (Fig. 5A). The reproducibility of Mg and trace elemental ratio across the five rhodoliths branches that span the last >45 years was assessed for both monthly and interannual variations (Fig. 6). With the exception of one Sr/Ca records comparison, all compared records (n ¼ 60) are significantly correlated (p < 0.05). At monthly resolution, Mg/Ca and Li/Ca records across branches are the most reproducible with most correlation coefficients between r ¼ 0.4

and r ¼ 0.6. Sr/Ca is less well reproduced at this resolution, with half of the r-values falling between 0.4 and 0.6 and the other half ranging from 0.2 to 0.4. All correlations are, however, statistically significant (p < 0.001; n ¼ 590) At the interannual scale, higher correlations are generally observed for all studied elements, with rvalues often >0.6 (Fig. 6). At interannual resolution, Mg/Ca variations produce the highest and most consistent correlation coefficients, whereas the correlation coefficients for Sr/Ca are inconsistent, ranging from insignificant at the 95% level (for the pair BSA1-BSA2; p ¼ 0.2) to up to r ¼ 0.8; p < 0.05 (Fig. 6).

3.4. Element variations and seawater temperature Between November 2009 and August 2011, well-defined seasonal variations are observed in the monthly average branch Mg/Ca, Sr/Ca and Li/Ca ratio values. These average branch values closely match the IST pattern (Fig. 5B), and this translates into strong correlations between each elemental ratio and instrumental IST (r > 0.82; p < 0.001; Table 2). Mg/Ca displays the best fit to the IST dataset over this 22-month period (r ¼ 0.91; p < 0.001 e Fig. 5C; Table 2). For the period 1963e2008, all monthly variations of Mg/Ca, Sr/ Ca and Li/Ca display significant correlations with local SST at the 99.9% confidence interval (0.4 < r < 0.8; p < 0.001 e Fig. 6; Table 2). Again, Mg/Ca ratio values show the best fit to SST, with correlation coefficients for most branches being between r ¼ 0.6 and r ¼ 0.8 (p < 0.001). Sr/Ca and Li/Ca show slightly lower correlations (0.4 < r < 0.6; p < 0.001). The relationship between the studied elements variations and SST is generally even stronger at the interannual level. All five Mg/Ca records are highly correlated with SST variations (0.6 < r < 0.8; p < 0.05) as well as the majority of the Li/Ca records. Sr/CaeSST correlation coefficients are also greater compared to the monthly resolution (Fig. 6). Monthly elemental variations over the 1963e2008 period for the 5 analysed branches are displayed in Fig. 7, along with the combined record obtained by averaging the 5 individual records, and the instrumental local SST

N. Darrenougue et al. / Quaternary Science Reviews 93 (2014) 34e46

39

Table 1 Average values, standard deviation and range of LA-ICPMS monthly records of Mg/Ca, Sr/Ca and Li/Ca for the 43 branches corresponding to the November 2009eAugust 2011 period (ARS stained), as well as along each of the five entire rhodolith branches studied here and the whole dataset (i.e. results of the entire LA-ICPMS analysis) corresponding to the 1958e2010 period (Whole). Pearson’s correlation coefficients (r) obtained for the cross-correlation between every element for every branch and for the whole dataset are also displayed (p < 0.05) along with the number of data points (n) used to establish these correlations. Rhodolith branches

Average
1s (range)

ARS stained BSA1 BSA2 BSA3 MSA1 SSA1 Whole

0.31 0.31 0.29 0.30 0.31 0.30 0.30

Correlation coefficient (r)

Mg/Ca (mol/mol)








0.04 0.04 0.04 0.04 0.04 0.03 0.04

Sr/Ca (mmol/mol)

(0.18e0.47) (0.21e0.47) (0.20e0.42) (0.21e0.44) (0.22e0.43) (0.21e0.40) (0.20e0.47)

3.6 3.6 3.5 3.6 3.4 3.5 3.5










0.4 0.4 0.3 0.3 0.3 0.4 0.4

(2.8e5,2) (2.5e4.7) (2.7e4.3) (2.6e4.5) (2.3e4.4) (2.2e4.5) (2.2e4.7)

Li/Ca (mmol/mol) 0.09 0.09 0.08 0.09 0.07 0.08 0.08

Table 2 Regression equations for Mg/Ca, Sr/Ca and Li/Ca vs. temperature relationships using monthly-resolved data of both the November 2009eAugust 2011 period and the 1963e2008 period. Equations are presented under the form: Temperature ¼ a*Proxy þ b, with standard errors (SE) associated with a and b. IST: in situ temperature; SST: sea-surface temperature. Pearson’s correlation coefficients (r) associated with these regressions are also displayed as well as the significance level (p-level) and the number of data points used for the calculation of each regression. Studied period

Proxy

Temperature relationship

a SE

Nov 2009e Mg/Ca IST ¼ 65 Mg/Ca þ 5
7 Aug 2011 Sr/Ca IST ¼ 8 Sr/Ca  4
1 Li/Ca IST ¼ 140 Li/Ca þ 13
21 1963e2008 Mg/Ca SST ¼ 68 Mg/Ca þ 3.1
2 Sr/Ca SST ¼ 5.9 Sr/Ca þ 3
0.3 Li/Ca SST ¼ 103 Li/Ca þ 15.3
5

b SE r

p-Level n


2
5
2
0.6
1
0.4

<0.001 22 <0.001 22 <0.001 22 <0.001 530 <0.001 530 <0.001 530

0.91 0.82 0.84 0.85 0.65 0.66










0.02 0.02 0.02 0.02 0.02 0.02 0.02

(0.04e0.17) (0.05e0.17) (0.04e0.14) (0.05e0.17) (0.03e0.13) (0.03e0.16) (0.03e0.17)

n

Mg/CaeSr/Ca

Mg/CaeLi/Ca

Sr/CaeLi/Ca

0.76 0.79 0.75 0.84 0.62 0.74 0.68

0.73 0.77 0.77 0.85 0.79 0.76 0.67

0.77 0.90 0.85 0.87 0.80 0.89 0.87

946 1143 683 913 1033 974 4746

variation over this period. The relationship between the combined elemental ratio records for Mg/Ca, Sr/Ca and Li/Ca, and SST at monthly resolution are quantified in Fig. 8 and Table 2. Fig. 9 shows the combined record average and standard deviation field ranges plotted against the local SST, at interannual resolution. Averaging the individual branch values to produce a combined record results in significantly higher correlations with local SST. Mg/Ca again maintains the strongest relationship with SST at monthly and interannual resolutions (r ¼ 0.85; p < 0.001 and r ¼ 0.83; p < 0.05, respectively in Figs. 8 and 9). The correlation coefficients for the average Sr/Ca and Li/Ca ratio values versus SST are similar at monthly resolution (r ¼ 0.65 and r ¼ 0.66, respectively; p < 0.001 e Fig. 8) and are higher when the interannual variations are considered (r ¼ 0.71 and r ¼ 0.75, respectively; p < 0.05 e Fig. 9).

Mg/Ca (mol/mol)

4. Discussion 4.1. Elements concentrations in S. durum rhodoliths

0.34 0.32 0.30 0.28 4.0 3.6 3.2

0.12 0.10 0.08

28-30

23-25

0-1

13-15

23-25

29-31

17-19

2-5

8-10

20-22

30-33

0.04

1-3

0.06 11-13

Li/Ca (mmol/mol)

2.8

Sr/Ca (mmol/mol)

0.26

Distance (mm)

Distance (mm)

Distance (mm)

BSA

MSA

SSA

Fig. 4. Mg and trace elements concentrations along the BSA, MSA and SSA rhodoliths (distances are from the outside toward the center of the nodule) measured by LAICPMS (black lines), solution ICPMS (dark grey lines) and ICPM-AES (light grey lines). Error bars correspond to the standard error on each measurement for solution ICPMS and ICPAES data point, and to the standard error of the mean of each element values recorded between the corresponding distance, for the LA-ICPMS data points.

The average Mg/Ca value recorded for both the November 2009eAugust 2011 period from the ARS-stained branches and the 1963e2008 period from the five different rhodolith branches are similar, indicating a consistent Mg incorporation in S. durum both across the Ricaudy Reef rhodolith bed and over time. Our Mg/Ca results for S. durum are consistent with the 0.08e0.40 Mg/Ca (molar) range reported for coralline red algae in general (Chave, 1954), albeit near the higher end for most of the other species studied (Table 2). This likely reflects the colder temperature ranges of most coralline red algae species compared to S. durum (see Table 2), and the incorporation of higher Mg concentrations at higher temperatures (e.g. Oomori et al., 1987). Calcification rates could also determine the level of incorporation of Mg into calcite, with higher Mg occurring with faster calcification (Rimstidt et al., 1998). This could account for the difference between S. durum and Lithothamnium crassiusculum as the extension rates of S. durum (Darrenougue et al., 2013; see also Table 2) are significantly higher than for L. crassiusculum (Halfar et al., 2000). However for different cold-water coralline red algal species exhibiting much lower growth rates, Kamenos et al. (2008) reported Mg/Ca values that are of a similar range to the ones of L. crassiusculum (Halfar et al., 2000; Table 2). Given this, and as previously suggested (e.g. Halfar et al., 2000; Kamenos et al., 2008), our results seem to indicate a most probable role of inter-species variability in the differences in Mg composition of coralline red algae. The range of Sr/Ca values presented here in S. durum is within values reported for cold-water species of coralline red algae (2.2e 5.5 mmol/mol; Kamenos et al., 2008; Hetzinger et al., 2011 e Table 2), which suggests that temperature might not play a major role in the observed Sr content of coralline red algae. Similarly,

40

N. Darrenougue et al. / Quaternary Science Reviews 93 (2014) 34e46

Fig. 5. A: Cross-correlation matrix diagrams for the records of Mg/Ca, Sr/Ca and Li/Ca along the top part (w2 mm) of the 43 stained rhodolith branches studied here. Both full (Full) and monthly (Monthly) resolution data are considered. The percentages correspond to the proportion of significant pairwise (i.e. between two profiles) correlations (r > 0.3; p < 0.05) over the entire dataset (n ¼ 903). The scale is such that the clearer the colour, the higher the correlation coefficient. B: Monthly-resolved variations of Mg/Ca, Sr/Ca and Li/ Ca recorded along the 43 stained rhodolith branches (pale lines) over the November 2009eAugust 2011 period. Average elements records as well as monthly variations of the in situ temperature (IST) (thick coloured lines) are also displayed for the same period. C: Scatter plots of the relationships between monthly data of average Mg/Ca, Sr/Ca and Li/Ca against IST for the November 2009eAugust 2011 period. Linear regression line (black line) and 95% confidence intervals (shaded areas) are displayed as well as the corresponding Pearson’s correlation coefficients (r); p < 0.05. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

extension rate is also to be excluded as a primary factor of Sr average concentration among coralline algae. Neither Sr/Ca concentrations nor Sr/Ca variations showed a significant difference after H2O2 treatment of a S. durum rhodolith branch (our unpublished results), suggesting Sr content in rhodolith’s calcite is not significantly affected by the presence of organic matter. Our results therefore seem to indicate that the Sr content in coralline red algae is not species dependant and rather reflect the relative uniformity of the Sr/Ca ratio in the global ocean (de Villiers, 1999). Li/Ca concentrations have not been reported previously for coralline red algae. The values presented in this study are globally of the same order than those in the shells and skeletons of other calcareous organisms, although Li/Ca concentrations for S. durum rhodoliths (0.03e0.17 mmol/mol) are higher than the range reported in corals (0.005e0.009 mmol/mol; Marriott et al., 2004a; Rollion-Bard et al., 2009) and various species of foraminifera (0.003e0.024 mmol/mol; Bryan and Marchitto, 2008; see also Delaney and Boyle, 1986; Hall and Chan, 2004; Marriott et al., 2004b; Yu et al., 2005). They are also slightly higher than the ones reported in brachiopods (0.018e0.05 mmol/mol; Delaney et al., 1989) but lower than lacustrine ostracods Li/Ca concentrations (0.04e0.20 mmol/mol; Zhu et al., 2012). More reports of Li concentration in coralline red algae will be crucial to better understand the values presented here.

4.2. Elements relationship with SST All studied element vs. time records show highly significant, positive relationships with IST and local SST at both monthly and interannual resolutions (Figs. 5B, C and 6). This suggests temperature is the principal factor controlling variations of Mg/Ca, Sr/Ca and Li/Ca in S. durum rhodoliths. These elements-SST relationships are enhanced when the records are averaged to create a combined record for Ricaudy Reef rhodoliths (Figs. 7 and 8). Clearly, averaging the records from different branches of the same rhodolith and different rhodoliths eliminates a significant fraction of the noncoherent elemental variability that occurs between branches and individual rhodoliths, which is primarily due to infra-seasonal growth rate variations between branches and the resulting uncertainties in the corresponding branch’s age models (see Darrenougue et al., 2013). 4.2.1. Mg/Ca Out of the three analysed elements, Mg/Ca displays the strongest correlation with IST and with local SST over both monthly to interannual timescales, and therefore is likely to provide the most accurate temperature reconstructions for S. durum rhodoliths at both monthly and interannual resolution. The highest correlation over the 2009e 2011 period is obtained between the combined Mg/Ca record and the

N. Darrenougue et al. / Quaternary Science Reviews 93 (2014) 34e46

41

Fig. 6. Cross-correlation matrix diagrams for the records of Mg/Ca, Sr/Ca and Li/Ca along the five rhodoliths branches analysed with LA-ICPMS, considering both the monthly and interannual data. The correlation of each record with the local SST is also displayed. All correlations are statistically significant at the 99.9% and 95% levels for, respectively, monthly and interannual datasets, except for *. The scale is such that the clearer the colour, the higher the correlation coefficient.

IST at monthly resolution (r ¼ 0.91; p < 0.05; Table 2). Kamenos et al. (2008) reported similar results when comparing monthly-resolved Mg/Ca data in Lithothamnium glaciale and Phymatolithon calcareum to IST over a comparable period of time. The Mg/Ca relationship with local SST over the last w50 years (average: r ¼ 0.85; individual records 1965

1970

1975

1980

1985

0.6 < r < 0.8; p < 0.001) is also very high. This is consistent with the high correlations reported for cold-water coralline red algae over records of several decades (Hetzinger et al., 2009, 2011). The Mg composition of coralline red algae typically increase by about 1 mol % MgCO3 (0.013 Mg/Ca mol/mol) per degree Celsius 1990

1995

2000

2005

2010

SST ( C)

30

25

20

0.35 0.3 0.25

Sr/Ca (mmol/mol)

4.5

Mg/Ca (mol/mol)

0.4

0.2 4

3.5

0.12 0.1 0.08 0.06

Li/Ca (mmol/mol)

3

0.04 1965

1970

1975

1980

1985

1990

1995

2000

2005

2010

Fig. 7. Monthly-resolved variations of Mg/Ca, Sr/Ca and Li/Ca recorded along the five individual rhodolith branches (thin pale lines) over the 1963e2008 period. Average elements records as well as monthly variations of local SST (thick coloured lines) are also displayed for the same period. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

42

N. Darrenougue et al. / Quaternary Science Reviews 93 (2014) 34e46

SST ( C)

Mg/Ca (mol/mol)

20

22

24

26

28

0.36

0.32

0.28

0.24

Li/Ca (mmol/mol)

Sr/Ca (mmol/mol)

4.0

3.6

3.2

0.10

0.08

0.06

0.04

20

22

24

26

28

SST ( C) Fig. 8. Scatter plots of the relationships between monthly data of average Mg/Ca, Sr/Ca and Li/Ca against local SST. Linear regression line (black line) and 95% confidence intervals (shaded areas) are displayed as well as the corresponding Pearson’s correlation coefficients (r); p < 0.001.

(e.g. Halfar et al., 2000; Table 2). The regression slopes derived from the S. durum combined record for Mg/Ca versus IST variation over the period November 2009eAugust 2011 period is 0.013, and for Mg/Ca versus local SST over the period 1963e2008 is 0.011. Despite the likely occurrence of species-specific effects on the absolute Mg concentration in coralline red algae, the sensitivity of Mg variation to seawater temperature is very consistent between species and in different environments (Table 2). For S. durum the rate of Mg increase with temperature is þ3.5%  C1, which compares closely to the value of 3.1
0.4% reported for abiotic calcite (Oomori et al., 1987). This indicates there may be little or no significant vital effect control over Mg/Ca variation in coralline red algae skeletons, as has been suggested previously (Ries, 2006; Kamenos et al., 2008). However, this is in contradiction with the findings of Hetzinger

et al. (2011), and implies that the presence of a vital effect on the Mg/Ca variations might be genera specific. 4.2.2. Sr/Ca and Li/Ca Sr/Ca variations with IST or local SST in S. durum are also highly significant but are generally weaker than for Mg/Ca (Figs. 5C and 8), consistent with observations made in previous studies (Kamenos et al., 2008; Hetzinger et al., 2011). The correlations between Sr/ Ca and both IST and local SST, as well as the amplitude of Sr/Ca variation with temperature agree well with previously reported values (Kamenos et al., 2008; Hetzinger et al., 2009). The tight range of Sr/Ca (mol/mol) variation in S. durum (i.e. 0.7  104 to 0.8  104/ C) falls well within the range (0.6  104 to 2  104/ C) that has been obtained by Kamenos et al. (2008) and Hetzinger et al. (2011) for four different coralline red algal species from cold water environments. Li/Ca variations show a positive correlation with seawater temperature that is of the same order as the Sr/Ca variations. The strong correlation observed between Li/Ca and Sr/Ca variations in the individual records suggests that Li/Ca and Sr/Ca share similar influential factors upon their variations. Abiotic calcite precipitation shows negative correlations with temperature for both Sr/Ca (Kinsman and Holland, 1969) and Li/Ca (e.g. Marriott et al., 2004a). Even though the mode of incorporation of these elements may be different according to the organism considered, this negative correlation is observed in various biogenic carbonates (Sr/Ca in corals: see Corrège, 2006 for a review; Li/Ca in foraminifera: e.g. Hall and Chan, 2004; Marriott et al., 2004b; Bryan and Marchitto, 2008; in brachiopods: Delaney et al., 1989; in corals: Marriott et al., 2004a; Montagna et al., 2006; Rollion-Bard et al., 2009; Hathorne et al., 2009). A strong vital effect was suggested by Kamenos et al. (2008) to explain the positive Sr/Ca-SST relationship. To try to understand the positive relationship observed in coralline red algae between Sr/Ca, Li/Ca and SST, one might need to consider the high Mg concentration of the calcite deposited by these organisms. Indeed, it has been shown that an increase in Mg2þ concentration in the parent solution would result in an increase of the distribution coefficients of both Sr2þ (Carpenter and Lohmann, 1992) and Liþ (Okumura and Kitano, 1986), which will, therefore, be more easily incorporated into calcite. Okumura and Kitano (1986) also showed a positive linear relationship between the MgCO3 concentration and the Li content in calcite. Consequently, we could speculate that the variations of Sr/Ca and Li/Ca in coralline red algae may be primarily controlled by the variations in Mg incorporation into the calcite skeleton. This could explain (1) the fact that Sr/Ca and Li/Ca variations are positively correlated to SST, despite the fact that abiotic calcite precipitation and observations in other calcareous organisms predict negative correlations; and (2) the lower extent of the Sr/Ca-SST and Li/Ca-SST relationships compared to the one of Mg/Ca-SST. Indeed, for Sr/Ca and Li/ Ca, SST variations may only be an indirect control, the major influential factor being the Mg/Ca variations, itself being directly controlled by SST variations. It is recommended that further investigations be designed to isolate the factors (temperature, calcification rate, and Mg content) that control Sr and Li incorporation into the coralline red algal calcite. 4.3. Regional climate record Mg/Ca variations in S. durum from the Ricaudy Reef provide a reliable record of local SST variations at monthly to interannual timescales. When Mg/Ca variations in S. durum are compared to the Oceanic Nino Index (ONI), a measure of ENSO climate variability in the tropical Pacific (http://ggweather.com/enso/oni.htm), the average interannual Mg/Ca anomaly is found to correlate

N. Darrenougue et al. / Quaternary Science Reviews 93 (2014) 34e46

1965

1970

1975

1980

1985

1990

1995

2000

43

2005

25 0.32 24 0.28

23

Mg/Ca (mol/mol)

0.36

26

SST ( C)

2010

22 3.9

25 3.6 24 23

3.3

Sr/Ca (mmol/mol)

SST ( C)

26

22

0.10

25 24

0.08 23 22

Li/Ca (mmol/mol)

SST ( C)

26

0.06 1965

1970

1975

1980

1985

1990

1995

2000

2005

2010

Fig. 9. Interannual variations of average Mg/Ca (orange), Sr/Ca (purple) and Li/Ca (yellow) records, super imposed over interannual local SST variations (blue lines) for the period 1965e2006. Standard deviations (shaded areas) of the elements averages as well as the corresponding Pearson’s correlation coefficients (r); p < 0.05, are also displayed. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

significantly (p < 0.05) with ONI variations over the period 1965e 2008, with the best fit (r < 0.5; p < 0.05) obtained using a 4month lag (Fig. 10A). A 4-month lag also gives the best correlation (r < 0.6; p < 0.05) between ONI and the local SST anomalies (Fig. 10A). This is consistent with the lag time found by Delcroix and Lenormand (1997), who obtained the closest response of SST in the lagoon of New Caledonia to observed ENSO variability with a 3month lag. The longer lag obtained for the Ricaudy Reef and instrumental SST measurement site could reflect their near shore locations within the lagoon, where water residence times are longer than the lagoon more generally by approximately 15 days (Jouon et al., 2006; Ouillon et al., 2010). Mg/Ca anomalies in S. durum from the Ricaudy Reef appear to faithfully record the regional climatic variability and ENSO anomalies in the Pacific (Fig. 10B). In particular, most La Niña periods (i.e. negative ONI values that are associated with higher SST in New Caledonia) appear as positive Mg/Ca anomalies. The El Niño periods (i.e. positive ONI and lower SST) are less obvious in the S. durum Mg/Ca record. Specifically, the strong 1982e83 and 1997e98 El Niño events (Caviedes, 1984; McPhaden, 1999) are missing from the Mg/Ca anomalies (Fig. 10B). However, these particular events resulted in only a slight decrease in the local SST near the Ricaudy Reef (Fig. 10B). 4.4. Mg/Ca in rhodoliths vs. Sr/Ca in corals It is only recently that the potential for coralline red algae to provide high-resolution archives of environmental changes has

been assessed (e.g. Kamenos et al., 2008; Hetzinger et al., 2009). These particular studies focused on high-latitudes species where environmental reconstruction has been largely restricted to bivalve molluscs that have well-known limitations for this purpose due to an ontogenic effect as they grow (Wanamaker et al., 2011). One of the most used and accurate, high-resolution palaeo-environmental archives are tropical scleractinian corals (e.g. Corrège, 2006). A comparison of local temperature reconstructions from both scleractinian corals and coralline red algae will enable further assessment of the reliability and potential of coralline red algae for palaeo-temperature reconstructions. A high-resolution Sr/Ca record from a Porites sp. coral has been generated from the Ricaudy Reef and covers 17 years of SST variations from 1975 to 1992 (Montaggioni et al., 2006). The correlation between Sr/Ca from this coral and SST is negative at monthly resolution (r ¼ 0.72; p < 0.001). The SST relationships obtained from the Sr/Ca variations in the Porites sp. coral and the Mg/Ca variations in the S. durum combined record agree well at monthly resolution and produce significant correlation coefficients (r ¼ 0.79 and r ¼ 0.87 respectively for the coral and rhodolith records; p < 0.001 e Fig. 11). This suggests the high-resolution Mg/Ca reconstruction of SST from S. durum rhodolith combined record may be more accurate than the Sr/Ca reconstruction from a Porites sp. coral at this site. However, if any single rhodolith branch record is considered, each observed Mg/Ca-SST correlation is slightly lower than the single coral Sr/Ca-SST correlation, although, still significant at the 95% level. This indicates that Mg/Ca variations in a single rhodolith branch are likely to be more influenced by additional factors to SST

44

N. Darrenougue et al. / Quaternary Science Reviews 93 (2014) 34e46

Fig. 10. A: Lagged cross-correlation diagram between local SST, the combined Mg/Ca record anomalies and the Oceanic Nino Index (ONI) interannual variations. The scale is such that lighter colours correspond to stronger anti-correlations. A ‘þ’ symbol stands for a positive correlation coefficient and the stars highlights the þ4 months lag, for which the highest correlation was obtained. Correlations are significant for jrj>0.3; p < 0.05. B: Variations of interannual local SST anomaly, average Mg/Ca anomaly and ONI for the 1965e 2007 period. Positive anomalies in the ONI record (red areas) represent El Nino periods and negative anomalies (blue areas) are characteristic of La Nina periods. Note: the ONI representation takes into account the 4-month lag discussed in the text. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

SST ( C)

Calendar years 1980

1985

8.9

26

9.0

24 9.1 22 9.2

28

0.36

26 0.32

24 22

0.28

22

24

26

28

r = -0.79

9.0 9.1 9.2

Mg/Ca (mol/mol)

20

8.9

0.36

Mg/Ca (mol/mol)

SST ( C)

28

SST ( C)

20

1990

Sr/Ca ( mol/mol)

1975

0.36

r = 0.87

0.32

0.28

Sr/Ca ( mol/mol)

20

8.9

0.36

9.0

0.32

9.1 0.28 9.2

0.32

0.28 r = -0.72

1975

1980

1985

Calendar years

1990

8.9

9

9.1

9.2

Sr/Ca ( mol/mol)

Fig. 11. Left, top 2 plots: Sr/Ca from Porites sp. and Mg/Ca from S. durum (black lines) compared to the monthly and interannual variations of local SST (grey lines) over the same period. Note the reversed y axis for the Sr/Ca records. Bottom plot: Monthly variations of the Sr/Ca record from a Porites sp. coral (grey line e Montaggioni et al., 2006) compared against the average Mg/Ca record from the S. durum rhodoliths (black line e this study) over the 1975e1992 period. Right: Scattered plots of the relationships between monthly data of the above-mentioned parameters. Corresponding linear regression lines (black lines) and 95% confidence intervals (shaded areas) are displayed as well as Pearson’s correlation coefficients (r); p < 0.001.

N. Darrenougue et al. / Quaternary Science Reviews 93 (2014) 34e46 Table 3 Cross correlation matrix between the records of Sr/Ca in Porites sp. coral (Montaggioni et al., 2006), average Mg/Ca in S. durum (this study) and the local SST at the monthly (p < 0.001) and interannual (p < 0.05) resolutions.

than the Sr/Ca variations of a single coral colony. Similar conclusions can be drawn from a comparison of the interannual variations of Mg/Ca in S. durum rhodoliths and Sr/Ca in Porites sp. coral with local SST (Table 3), which again results in a slightly better SST reconstruction using the rhodolith Mg/Ca combined record. 5. Conclusion High-resolution LA-ICPMS records of Mg/Ca, Sr/Ca and Li/Ca from various S. durum rhodolith branches all show, respectively, consistent average values and a similar range of variation over both the rhodoliths’ last living years and a w50-year period. Concentration values obtained for these elements ratios mostly fall within the range of previously published values for other coralline red algal species (Mg/Ca and Sr/Ca) and other calcareous organisms (Li/Ca). For the two studied time periods, a comparison with the other recorded elemental ratios shows that Mg/Ca variations are the most reproducible between different rhodolith branches. However, all Mg/Ca, Sr/Ca and Li/Ca average records are significantly correlated to the IST and local SST at both monthly and interannual timescales. Mg/Ca variations show the best fit with local SST, indicating that Mg/Ca variations in S. durum are likely to be the least affected by non-environmental factors. This also confirms previously published suggestions on the preferential use of Mg/Ca over other trace elements for SST reconstructions. S. durum rhodoliths are able to record the regional climate pattern as shown by the significant correlation between interannual Mg/Ca anomalies and the ONI over the entire studied period and a comparison of Mg/Ca variations in S. durum with Sr/Ca variations in a Porites sp. coral from the same study site suggests that coralline red algae have the potential to be a proxy as reliable as corals in terms of SST reconstruction. Acknowledgements The authors are very grateful to J. Butscher (IRD) for the invaluable help provided for the rhodoliths collection and throughout the monitoring experiment. The late G. Cabioch (IRD) was instrumental in organising this research project. Technical help by L. Kinsley, C. Alibert, T. Phimphisane and J. Vickers (RSES) is gratefully acknowledged. Comments from M. Davies (RSES) greatly improved an early version of this manuscript. Funding for this research was obtained through an ANU postgraduate scholarship and a RSES field allowance awarded to ND, and an ARC-DP grant awarded to PDD. Fieldwork was also supported by a 2011 AFAS (AustralianeFrench Association for Science and technology) grant awarded to ND. This manuscript was improved as a result of pertinent comments received by two anonymous reviewers to whom we are grateful.

45

References Adey, W.H., MacIntyre, I.G., 1973. Crustose coralline algae: a re-evaluation in the geological sciences. Geol. Soc. Am. Bull. 84, 883e904. Basso, D., Nalin, R., Nelson, C.S., 2009. Shallow-water Sporolithon rhodoliths from North Island (New Zealand). Palaios 24, 92e103. Blake, C., Maggs, C.A., 2003. Comparative growth rates and internal banding periodicity of maerl species (Corallinales, Rhodophyta) from northern Europe. Phycologia 42 (6), 606e612. Böhm, L., Schramm, W., Rabsch, U., 1978. Ecological and physiological aspects of some coralline algae from the western Baltic. Calcium uptake and skeleton formation in Phymatolithon calcareum. Kiel. Meeresforsch. 4, 282e288. Bosence, D.W.J., 1983a. Description and classification of Rhodoliths (Rhodoids, Rhodolites). In: Peryt, T.M. (Ed.), Coated Grains. Springer-Verlag, Berlin, pp. 217e224. Bosence, D.W.J., 1983b. The occurence and ecology of recent rhodoliths e a review. In: Peryt, T.M. (Ed.), Coated Grains. Springer-Verlag, Berlin, pp. 225e242. Bryan, S.P., Marchitto, T.M., 2008. Mg/Ca-temperature proxy in benthic foraminifera: new calibrations from the Florida Straits and a hypothesis regarding Mg/ Li. Paleoceanography 23 (2), PA2220. Burdett, H., Kamenos, N.A., Law, A., 2010. Using coralline algae to understand historic marine cloud cover. Palaeogeogr. Palaeoclimatol. Palaeoecol. 302, 65e70. Carpenter, S.J., Lohmann, K.C., 1992. Sr/Mg ratios of modern marine calcite: empirical indicators of ocean chemistry and precipitation rate. Geochim. Cosmochim. Acta 56 (5), 1837e1849. Caviedes, C.N., 1984. El Nino 1982e83. Geogr. Rev. 74 (3), 267e290. Chan, P., Halfar, J., Williams, B., Hetzinger, S., Steneck, R., Zack, T., Jacob, D.E., 2011. Freshening of the Alaska Coastal Current recorded by coralline algal Ba/Ca ratios. J. Geophys. Res. 116 (G1), G01032. Chave, K.E., 1954. Aspects of the biogeochemistry of magnesium 1. Calcareous marine organisms. J. Geol. 62 (3), 266e283. Corrège, T., 2006. Sea surface temperature and salinity reconstruction from coral geochemical tracers. Palaeogeogr. Palaeoclimatol. Palaeoecol. 232 (2), 408e428. Corrège, T., Delcroix, T., Récy, J., Beck, W., Cabioch, G., 2000. Evidence for stronger El Nino-Southern Oscillation (ENSO) events. Paleoceanography 15 (4), 465e470. Dandonneau, Y., Debenay, J., Dugas, F., Fourmanoir, P., Magnier, Y., Rougerie, F., 1981. Le lagon de la Grande Terre, Présentation d’ensemble, Sédimentologie et hydrologie du sud-ouest. Atlas de Nouvelle-Calédonie et dépendances. ORSTOM, Paris, pp. 21e24. de Villiers, S., 1999. Seawater strontium and Sr/Ca variability in the Atlantic and Pacific oceans. Earth Planet. Sci. Lett. 171 (4), 623e634. de Villiers, S., Greaves, M., Elderfield, H., 2002. An intensity ratio calibration method for the accurate determination of Mg/Ca and Sr/Ca of marine carbonates by ICPAES. Geochem. Geophys. Geosyst. 3 (1), 1001. Delaney, M., Popp, B., Lepzelter, C., Anderson, T., 1989. Lithium-to-calcium ratios in modern, cenozoic, and paleozoic articulate brachiopod shells. Paleoceanography 4 (6), 681e691. Delaney, M.L., Boyle, E.A., 1986. Lithium in foraminiferal shells: implications for high-temperature hydrothermal circulation fluxes and oceanic crustal generation rates. Earth Planet. Sci. Lett. 80 (1e2), 91e105. Delcroix, T., Lenormand, O., 1997. ENSO signals in the vicinity of New Caledonia, South Western Pacific. Oceanol. acta 20 (3), 481e491. Fernandez, J.M., Ouillon, S., Chevillon, C., Douillet, P., Fichez, R., Gendre, R.L., 2006. A combined modeling and geochemical study of the fate of terrigenous inputs from mixed natural and mining sources in a coral reef lagoon (New Caledonia). Mar. Pollut. Bull. 52 (3), 320e331. Foster, M.S., 2001. Rhodoliths: between rocks and soft places. J. Phycol. 37, 659e667. Frantz, B.R., Foster, M.S., Riosmena-RodrÌguez, R., 2005. Clathromorphum nereostratum (Corallinales, Rhodophyta): the oldest alga? J. Phycol. 41 (4), 770e773. Goldberg, N., Heine, J.N., 2008. Age estimates of Sporolithon durum (Corallinales, Rhodophyta) from Rottnest Island, Western Australia, based on radiocarbondating methods. J. R. Soc. West. Aust. 91, 27e30. Halfar, J., Steneck, R.S., Joachimski, M., Kronz, A., Wanamaker Jr., A.D., 2008. Coralline red algae as high-resolution climate recorders. Geology 36 (6), 463e 466. Halfar, J., Steneck, R.S., Schöne, B.R., Moore, G.W.K., Joachimski, M., Kronz, A., Fietzke, J., Estes, J., 2007. Coralline alga reveals first marine record of subarctic North Pacific climate change. Geophys. Res. Lett. 34, L07702. Halfar, J., Williams, B., Hetzinger, S., Steneck, R.S., Lebednik, P., Winsborough, C., Omar, A., Chan, P., Wanamaker, A.D., 2011. 225 years of Bering Sea climate and ecosystem dynamics revealed by coralline algal growth-increment widths. Geology 39 (6), 579e582. Halfar, J., Zack, T., Kronz, A., Zachos, J.C., 2000. Growth and high-resolution paleoenvironmental signals of rhodoliths (coralline red algae): a new biogenic archive. J. Geophys. Res. 105 (C9), 22,107e22,116. Hall, J.M., Chan, L.H., 2004. Li/Ca in multiple species of benthic and planktonic foraminifera: thermocline, latitudinal, and glacialeinterglacial variation. Geochim. Cosmochim. Acta 68 (3), 529e545. Hathorne, E., Felis, T., Suzuki, A., Kawahata, H., 2009. Lithium Content of the Aragonitic Skeletons of Massive Porites Corals: a New Tool to Reconstruct Tropical Sea Surface Temperatures?. American Geophysical Union, Fall Meeting 2009, abstract #PP34B-05. Herz, N., Dean, N.E., 1986. Stable isotopes and archaeological geology: the Carrara marble, northern Italy. Appl. Geochem. 1 (1), 139e151.

46

N. Darrenougue et al. / Quaternary Science Reviews 93 (2014) 34e46

Hetzinger, S., Halfar, J., Kronz, A., Steneck, R., Adey, W.H., Lebednik, P.A., Schöne, B.R., 2009. High-resolution Mg/Ca ratios in a coralline red alga as a proxy for Bering Sea temperature variations from 1902 to 1967. Palaios 24, 406e412. Hetzinger, S., Halfar, J., Zack, T., Gamboa, G., Jacob, D.E., Kunz, B.E., Kronz, A., Adey, W., Lebednik, P.A., Steneck, R.S., 2011. High-resolution analysis of trace elements in crustose coralline algae from the North Atlantic and North Pacific by laser ablation ICP-MS. Palaeogeogr. Palaeoclimatol. Palaeoecol. 302 (1e2), 81e94. Jouon, A., Douillet, P., Ouillon, S., Fraunié, P., 2006. Calculations of hydrodynamic time parameters in a semi-opened coastal zone using a 3D hydrodynamic model. Cont. Shelf Res. 26 (12e13), 1395e1415. Kamenos, N.A., 2010. North Atlantic summers have warmed more than winters since 1353, and the response of marine zooplankton. Proc. Natl. Acad. Sci. 107 (52), 22442e22447. Kamenos, N.A., Cusack, M., Moore, P.G., 2008. Coralline algae are global palaeothermometers with bi-weekly resolution. Geochim. Cosmochim. Acta 72 (3), 771e779. Kamenos, N.A., Law, A., 2010. Temperature controls on coralline algal skeletal growth. J. Phycol. 46 (2), 331e335. Kinsman, D.J.J., Holland, H.D., 1969. The co-precipitation of cations with CaCO3 e IV: the co-precipitation of Sr2þ with aragonite between 16 and 96 C. Geochim. Cosmochim. Acta 33 (1), 1e17. Labrosse, P., Fichez, R., Farman, R., Adams, T., 2000. New Caledonia. In: Sheppard, C.R.C.E. (Ed.), Seas at the Millennium: an Environmental Evaluation, pp. 723e736. Le Bec, N., Juillet-Leclerc, A., Corrège, T., Blamart, D., Delcroix, T., 2000. A coral d180 record of ENSO driven sea surface salinity variability in Fiji (south-western tropical Pacific). Geophys. Res. Lett. 27 (23), 3897e3900. Littler, M.M., Littler, D.S., Dennis Hanisak, M., 1991. Deep-water rhodolith distribution, productivity, and growth history at sites of formation and subsequent degradation. J. Exp. Mar. Biol. Ecol. 150 (2), 163e182. Longerich, H.P., Jackson, S.E., Günther, D., 1996. Laser ablation inductively coupled plasma mass spectrometric transient signal data acquisition and analyte concentration calculation. J. Anal. Atomic Spectrom. 11 (9), 899e904. Marriott, C.S., Henderson, G.M., Belshaw, N.S., Tudhope, A.W., 2004a. Temperature dependence of 7Li, 44Ca and Li/Ca during growth of calcium carbonate. Earth Planet. Sci. Lett. 222 (2), 615e624. Marriott, C.S., Henderson, G.M., Crompton, R., Staubwasser, M., Shaw, S., 2004b. Effect of mineralogy, salinity, and temperature on Li/Ca and Li isotope composition of calcium carbonate. Chem. geol. 212 (1), 5e15. McPhaden, M.J., 1999. Genesis and evolution of the 1997e98 El Nino. Science 283, 950. Montaggioni, L.F., Le Cornec, F., Corrège, T., Cabioch, G., 2006. Coral barium/calcium record of mid-Holocene upwelling activity in New Caledonia, South-West Pacific. Palaeogeogr. Palaeoclimatol. Palaeoecol. 237 (2), 436e455. Montagna, P., McCulloch, M., Mazzoli, C., Silenzi, S., Schiaparelli, S., 2006. Li/Ca ratios in the Mediterranean non-tropical coral Cladocora caespitosa as a potential paleothermometer. Geophys. Res. Abstr. 8, 03695.

Nicet, J.B., Delcroix, T., 2000. ENSO-related precipitation changes in New Caledonia, southwestern tropical Pacific: 1969e98. Mon. Weather Rev. 128 (8), 3001e 3006. Okumura, M., Kitano, Y., 1986. Coprecipitation of alkali metal ions with calcium carbonate. Geochim. Cosmochim. Acta 50 (1), 49e58. Oomori, T., Kaneshima, H., Maezato, Y., Kitano, Y., 1987. Distribution coefficient of Mg2þ ions between calcite and solution at 10e50  C. Mar. Chem. 20 (4), 327e336. Ouillon, S., Douillet, P., Lefebvre, J.P., Le Gendre, R., Jouon, A., Bonneton, P., Fernandez, J.M., Chevillon, C., 2010. Circulation and suspended sediment transport in a coral reef lagoon: the south-west lagoon of New Caledonia. Mar. Pollut. Bull. 61 (7e12), 309e322. Paillard, D., Labeyrie, L., Yiou, P., 1996. Macintosh program performs time-series analysis. EOS Trans. AGU 77 (39), 379. Payri, C., 1997. Hydrolithon reinboldii rhodolith distribution, growth and carbon production of a French Polynesian reef. In: Proc. 8th International Coral Reef. Symposium Panama, pp. 755e760. Ries, J.B., 2006. Mg fractionation in crustose coralline algae: geochemical, biological, and sedimentological implications of secular variation in the Mg/Ca ratio of seawater. Geochim. Cosmochim. Acta 70, 891e900. Rimstidt, J.D., Balog, A., Webb, J., 1998. Distribution of trace elements between carbonate minerals and aqueous solutions. Geochim. Cosmochim. Acta 62 (11), 1851e1863. Rollion-Bard, C., Vigier, N., Meibom, A., Blamart, D., Reynaud, S., Rodolfo-Metalpa, R., Martin, S., Gattuso, J.P., 2009. Effect of environmental conditions and skeletal ultrastructure on the Li isotopic composition of scleractinian corals. Earth Planet. Sci. Lett. 286 (1), 63e70. Steneck, R.S., 1986. The ecology of coralline algal crusts: convergent patterns and adaptative strategies. Annu. Rev. Ecol. Syst. 17, 273e303. Townsend, R.A., Woelkerling, W.J., Harvey, A.S., Borowitzka, M., 1995. An account of the red algal genus Sporolithon (Sporolithaceae, Corallinales) in southern Australia. Aust. Syst. Bot. 8 (1), 85e121. Wanamaker Jr., A.D., Hetzinger, S., Halfar, J., 2011. Reconstructing mid-to highlatitude marine climate and ocean variability using bivalves, coralline algae, and marine sediment cores from the Northern Hemisphere. Palaeogeogr. Palaeoclimatol. Palaeoecol. 302 (1e2), 1e9. Williams, B., Halfar, J., Steneck, R.S., Wortmann, U.G., Hetzinger, S., Adey, W., Lebednik, P., Joachimski, M., 2011. Twentieth century 13C variability in surface water dissolved inorganic carbon recorded by coralline algae in the northern North Pacific Ocean and the Bering Sea. Biogeosciences 8 (1), 165e174. Womersley, H.B.S., 1996. The marine benthic flora of southern Australia, Part III B: Gracilariales, Rhodymeniales, Corallinales and Bonnemaisoniales. Australian Biological Resources Study, Canberra. Yu, J., Day, J., Greaves, M., Elderfield, H., 2005. Determination of multiple element/ calcium ratios in foraminiferal calcite by quadrupole ICP-MS. Geochem. Geophys. Geosyst. 6 (10), 1029. Zhu, Z., Chen, J., Li, D., Ren, S., Liu, F., 2012. Li/Ca ratios of ostracod shells at Lake Qinghai, NE Tibetan Plateau, China: a potential temperature indicator. Environ. Earth Sci. 67 (6), 1735e1742.