It is necessary to separate B completely from the potential Ca matrix for isotopic determination. A microsublimation technique, modified from Gaillardet et al. (2001), was used to extract B from calcium carbonates to circumvent any column separation or complicate a chemical reagent. It is more effective for the purification and also for the elimination of potential contamination (Wang et al., 2020). Furthermore, this technique reduces the influence of organic matter from the ion exchange resin or the sample itself. A solution, ~40 µl, was placed on the cover of a 5-ml perfluoroalkoxy (PFA) conical beaker (Savillex, Eden Prairie, Minnesota USA), and heated on a dry bath system (Firefox Dry Bath 6100, Taipei City, Taiwan) for 12 h at 98°C. In order to eliminate matrix effects, a series of recovery experiments were conducted to assess our B separation under various matrices of Ca- and Na-added solutions (Wang et al., 2010).

Precise analytical methods for boron isotopic measurements in natural materials have been discussed in detail (Spivack and Edmond, 1986; Guerrot et al., 2011). Despite the importance of marine biogenic carbonates for paleoceanographic studies, B isotopic applications are scarce owing to analytical difficulties, which require an accurate and high-precision technique for a small amount of sample (a few nanograms of B). Recent advances in multicollector inductively coupled plasma mass spectrometry (MC-ICP-MS) techniques are capable of precise isotopic analyses for a wide range of samples (Stewart et al., 2021).

In this study, a precise analytical technique for B isotopic determination in coral skeletons was developed using MC-ICP-MS (Thermo Fisher Neptune, Bremen, Germany) at the Earth Dynamic System Research Center (EDSRC), NCKU. The MC-ICP-MS has more advantages than the traditional thermal ionization mass spectrometry (TIMS). For instance, 1) the required sample amount is relatively small, 2) both the sample throughput and analytical precision are significantly higher than TIMS, and 3) the mass discrimination inherited from the instrument itself can be corrected by the sample-standard bracketing procedure.

A quartz spray dual chamber [Stable Sample Introduction (SSI), ESI] and a 50-μl PFA-ST nebulizer (ESI) were used to minimize boron blank and enhance signal stability. The Ni sampler and skimmer cones were used and the RF power was set at 1,200 W. Sample gas was adjusted to 1.05–1.10 L min⁻¹, and the flow rates of the cooling gas and auxiliary gas were set at 15.0 and 0.7 L min⁻¹, respectively. The typical sensitivity of B is ~0.7–1 V for 100 ppb at an uptake rate of 60 μl min⁻¹. Ammonia solution was used to reduce the memory effect. Boron isotopes, ¹¹B and ¹⁰B, were measured simultaneously on H3 and L3 Faraday cups, respectively, and 60 ratios with 6 blocks were collected with 2 s integration. Each analysis takes approximately 2 min. A quartz spray dual chamber coupled to a 50-μl PFA-ST nebulizer was used for sample introduction.

For most elements, memory effects can be resolved by washing with diluted HNO₃ in the introduction system. In the case of boron, however, the interference cannot be reduced effectively using HNO₃. As described by Al-Ammar et al. (1999), the addition of ammonia solution (NH₄OH) to the spray chamber of ICP-MS can reduce the absorbability of B-bearing species, thus reducing memory effects more effectively.

Unlike the common internal or external correction for precise isotope determinations, B only has two isotopes and cannot be corrected for the mass bias by traditional normalization methods. The instrumental mass bias correction for B can be carried out by the sample-standard bracketing technique (Aggarwal et al., 2003), and δ¹¹B values can be directly calculated by comparing the measured boron isotope ratios to standards. To establish a reliable isotope standard for boron using MC-ICP-MS, δ¹¹B values of different standard materials were also analyzed by TIMS (Thermo Fisher, Triton T1), in positive ion mode at EDSRC, NCKU. Several reference materials, including synthetic aragonites (AS10, AS30, and AS50), KTP-1 (in-house coral standard), JA-A, and JA-B, were prepared for cross-calibration of δ¹¹B values measured by MC-ICP-MS and P-TIMS.

δ¹⁸O and δ¹³C of corals are common tools in environmental proxy studies. The δ¹⁸O of corals is mainly controlled by seawater temperature and salinity, whereas δ¹³C is related to skeletal extension rate, dissolved inorganic carbon, and light availability (Shimamura et al., 2008). In this study, δ¹³C ranged between −3.9‰ and −1.7‰ and δ¹⁸O varied from −5.7‰ to −4.3‰. A clear annual cycle can be seen in both δ¹⁸O and δ¹³C; however, the correlation is out-of-phase (Figure 3).

The Lanyu corals grew at a rate of 12 mm year⁻¹, implying that kinetic isotope effects would be relatively constant due to the high skeletal extension rate, exceeding 5 mm year⁻¹ (Suzuki et al., 2005). The negative relationships (r = −0.73) between δ¹⁸O and δ¹³C are consistent with the previous results by Omata et al. (2005), suggesting that high solar radiation corresponds to higher seawater temperature, resulting in the out-of-phase pattern for δ¹⁸O and δ¹³C in rapidly growing corals. This indicates that the coral skeleton was strongly affected by metabolic effects (photosynthesis) over kinetic effects at Lanyu. Despite the potential metabolic isotope effect playing a main factor to affect δ¹³C, the possible influence by seawater δ¹³C_DIC was needed for further evaluation. Lin et al. (1999) pointed out that the δ¹³C_DIC in surface seawater was slightly lower during the cold season than the warm season around southern Taiwan. It is possible that the δ¹³C was affected by seawater δ¹³C due to vital effect (Pelejero et al., 2005). However, more detailed studies are required to further evaluate the relationship between coralline δ¹³C and seawater δ¹³C_DIC at decadal timescales.

Based on the close relationships between marine carbonate chemistry and pCO₂ concentrations, boron isotopic composition of marine biogenic carbonates is the most important geochemical proxy for reconstructing paleo-pH in the ocean (Hönisch et al., 2007; De'ath et al., 2009). Here, we discuss δ¹¹B values of different marine biogenic carbonates and interpret the boron isotope excursion in the Lanyu corals.

The δ¹¹B in marine carbonates fall within the theoretical curve for the borate species in seawater ( Figure 6 ). This observation has been supported by a number of empirical calibration works on planktonic foraminifers (Sanyal et al., 1996; Sanyal et al., 2001), scleractinian corals (Hemming et al., 1998; Hönisch et al., 2004; Pelejero et al., 2005), and inorganic precipitations (Sanyal et al., 2000) over a pH range of 7.7 to 9.0. Previous research has shown that variation of boron isotopic composition in marine carbonates was controlled by ambient seawater. Nevertheless, the systematic offset from the theoretical curve suggests that potential effects on the boron isotopic composition may have occurred, possibly due to variation in pH in calcification fluids (Hemming et al., 1998; Wei et al., 2009). This indicates that coralline δ¹¹B may not record ambient seawater pH, but these processes only resulted in slight constant offset between corals and seawaters due to direct seawater supply to the calcification site in corals.

Although the boron isotope in biogenic marine carbonates has been used for constructing seawater pH in the past, detailed studies about coralline δ¹¹B are scarce, especially for high-resolution seasonal variation in δ¹¹B records (Pelejero et al., 2005). It is necessary to understand the controlling factors on boron uptake and the isotopic fractionation before applying the coralline δ¹¹B to reconstruct the seawater pH in the past.

Hemming et al. (1998) investigated the seasonal variation in the boron isotopic composition in coral skeletons. Paired analyses of carbon and boron isotopes revealed a clear seasonal variation, displaying the high-density bands having heavier δ¹¹B and δ¹³C in the coral P. lobata from Fanning Island, probably reflecting high productivity during the period of high solar insolation. However, the boron and carbon isotope ratios in our corals did not show a similar correlation. Heavier δ¹¹B together with lighter δ¹³C is exhibited in the high-density bands which grew in the cold season (Figure 3). Based on the different patterns of the seasonal fluctuation between two study areas, this effect cannot be used to explain the variation in coralline δ¹¹B in Lanyu. Culture experiments with two surface corals, Porites and Acropora, have confirmed that coral skeletons can record the variation of seawater pH (Hönisch et al., 2004; Reynaud et al., 2004). The experimental studies showed that the δ¹³C in cultured corals can be significantly affected by symbiont photosynthetic activity, which is related to light intensity, feeding, and habitat depth. This result is inconsistent with the δ¹¹B values in the same corals, suggesting that pH variations at the calcification site are rather small relative to the pH-dependent isotopic fractionation. As mentioned above, our coral samples do faithfully record the seawater signal on the seasonal timescale and support the potential use of coral skeleton boron isotopes for reconstructing seawater pH in the past.

Boron isotopic compositions of the Lanyu corals, ranging from 23.6‰ to 25.1‰, correspond to seawater pH of 8.0–8.2 (Figure 3). These δ¹¹B-derived pH values from the Lanyu corals are identical to the modern surface seawater pH in the western Pacific. There is a clear decreasing δ¹¹B trend of −0.16‰ if the El Niño 1991 data are excluded, and it implies that ocean acidification has occurred in the study area (pH change of −0.02 unit). The δ¹¹B records obtained from the Lanyu corals provide the first detailed and high-resolution record of seawater pH in the region, which display low and high values during warm and cold seasons, respectively. This can be explained by seasonal variation in water masses with different pH values, caused by the monsoon climate (Liu et al., 2009; Wei et al., 2009) or pCO₂ variations (Takahashi et al., 2002).

The Kuroshio Current is the most important current around Lanyu Islet. Major water masses at the west side and the east side of 121°E are the South China Sea surface water and Kuroshio Current (KC), respectively. The typical water properties of the two water masses are significantly different. The main water mass close to Lanyu Islet is Kuroshio waters, based on the θ–S plot (Chen et al., 2007). Chou et al. (2007) studied the carbon chemistry of Kuroshio waters off SE Taiwan. Seawater from stations off southeast Taiwan has intermediate values between stations at the West Philippine Sea and the SCS. The pH value of the surface seawater around Lanyu Islet is approximately 8.15, significantly higher than the pH at the SCS (pH = 7.96 to 8.10) (Chen et al., 2006).

The seasonal variability in KC can be linked to the East Asia Monsoon (EAM) climate in the study area. The seawater chemistry is affected by changes in the EAM, which causes SCS seawater entering the KC from the loop current around southern Taiwan. Nevertheless, most KC transport is associated with the path contributed by the east flow of the Lanyu (Hsin et al., 2008). The loop current probably carries the low pH seawater from the SCS (pH~8.0) to the main KC (West Philippine Sea water, pH ~ 8.2); thus, the pH value of surface seawater will decrease in the winter due to the greater contribution from the SCS. However, this is inconsistent with the seasonal variation in δ¹¹B-based seawater pH, which is characterized by high pH (high δ¹¹B) during the cold season. A more detailed assessment on the influence of the SCS waters is necessary for understanding the seawater pH variation around Lanyu.

Another possible reason for the change of seawater pH is seasonal pCO₂ variation. Culture experiments on the boron isotopic composition of coral skeletons were carried out at two pCO₂ conditions (~440 vs. ~725 ppm) (Reynaud et al., 2004). Their results indicated that seawater pH changes due to pCO₂ variations can be recorded in coral skeletons. In comparison with the global sea–air CO₂ flux study by Takahashi et al. (2002), low and high pCO₂ were found during the cold and warm seasons, respectively, and pCO₂ is mainly controlled by the solubility of CO₂ and can be related to seawater temperature. For further evidence of the relationship between pCO₂ and seawater temperature in tropical western Pacific, the variation of pCO₂ caused by temperature effect was estimated using equation I (Takahashi et al., 2002).

The changes of the estimated pCO₂ from the Lanyu coral record are plotted in Figure 7, showing the correlation of reconstructed pCO₂ record in the study area on the basis of the two estimates derived from SST and TA. Seasonal variability in seawater pCO₂, with higher pCO₂ at warm seasons (higher SST), can also be found. This observation agrees well with the estimated seawater pH in the Lanyu corals, showing low pH value (or higher pCO₂) in the warmer SST. Furthermore, the variation of seawater pH is approximately 0.2 units between the cold and warm seasons, which is similar to the previous studies in the western Pacific (Pelejero et al., 2005). This study suggests that the boron isotopic composition of coral skeletons can be used for studying the relationship between changes in surface seawater chemistry and climate change and further provide important information on ocean acidification and abrupt climate change.