Introduction
In order to obtain accurate radiocarbon (14C) ages of marine samples, one needs to consider the marine reservoir effect that causes a difference between the 14C content of the surface ocean and contemporary atmosphere (Craig, Reference Craig1954). This difference is expressed as the 14C age offset between samples formed in seawater and the contemporary atmosphere (Stuiver and Polach Reference Stuiver and Polach1977). This offset varies because the surface ocean 14C budget varies through time and location (Bard Reference Bard1988). We can calibrate 14C ages precisely for samples that derived their carbon directly from the atmosphere by means of tree-ring records, however we must correct for the 14C offset in marine samples before we can perform a calibration to determine a calendar age. The modeled global ocean marine reservoir age (MRA) value in the marine calibration curve serves as a baseline to understand the location-dependent nature of surface seawater 14C ages. The first marine calibration curve was published in 1986 from 0 to 9000 cal BP (Stuiver et al. Reference Stuiver, Pearson and Braziunas1986), in which a constant 14C offset, or MRA between the atmosphere and the global surface ocean was assumed; and the 14C age of the surface layer was determined using an ocean-atmosphere box-diffusion model. The same model was still in use in the Marine13 calibration curve, which extended to 50,000 cal BP (Reimer et al. Reference Reimer, Bard, Bayliss, Beck, Blackwell, Ramsey, Buck, Cheng, Edwards, Friedrich, Grootes, Guilderson, Haflidason, Hajdas, Hatté, Heaton, Hoffmann, Hogg, Hughen, Kaiser, Kromer, Manning, Niu, Reimer, Richards, Scott, Southon, Staff, Turney and van der Plicht2013a) but has been superseded by the modified Box-Isotopic Carbon cYCLE (BICYCLE) model, which is adopted to account for temporal MRA variability in the latest marine calibration curve, Marine20 (Heaton et al. Reference Heaton, Köhler, Butzin, Bard, Reimer, Austin, Ramsey, Grootes, Hughen, Kromer, Reimer, Adkins, Burke, Cook, Olsen and Skinner2020).
An additional regional reservoir correction (ΔR) accounts for the difference between the global MRA and regional reservoir age. This value needs to be known in order to correct marine 14C ages since the reservoir age at different locations may deviate from the modeled global MRA (Stuiver and Braziunas Reference Stuiver and Braziunas1993). Marine20 provides more reasonable time-variable MRA, but the values are generally higher than the Marine13 constant global MRA about 150 14C yr for the past 10,500 cal BP (Heaton et al. Reference Heaton, Bard, Ramsey, Butzin, Hatté, Hughen, Köhler and Reimer2023). This difference of the global reservoir age derived from Marine20 means that new ΔR must be recalculated even though regional values remain the same. For the above reason, most of the previously reported ΔR (pre-bomb values) have been updated and are maintained on the website (https://calib.org/marine/; Reimer and Reimer Reference Reimer and Reimer2001). Due to the limited availability of data, often confined to specific regions, ΔR has been calculated as constant, both for 1950 and throughout the Holocene, in practical applications. However, temporal ΔR variations have been observed in many locations (Burr et al. Reference Burr, Haynes, Shen, Taylor, Chang, Beck, Nguyen and Zhou2015; Hirabayashi et al. Reference Hirabayashi, Yokoyama, Suzuki, Esat, Miyairi, Aze, Siringan and Maeda2019; Hua et al. Reference Hua, Webb, Zhao, Nothdurft, Lybolt, Price and Opdyke2015; Hua et al. Reference Hua, Ulm, Yu, Clark, Nothdurft, Leonard, Pandolfi, Jacobsen and Zhao2020; McGregor et al. Reference McGregor and Abram2008; Wan et al. Reference Wan, Meltzner, Switzer, Lin, Wang, Bradley, Natawidjaja, Suwargadi and Horton2020; Weisler et al. Reference Weisler, Hua, Zhao, Nguyen, Nothdurft, Yamano and Mihaljević2018; Yang et al. Reference Yang, Wang, Burr, Liu and Fan2019; Yu et al. Reference Yu, Hua, Zhao, Hodge, Fink and Barbetti2010), and the magnitude of the ΔR shift based on Marine 20 in the Holocene is not known. Thus, before applying updated ΔR values acquired from modern-day marine samples, we need to re-evaluate whether this site-specific constant ΔR is consistent through the Holocene. This is the main focus of this study.
Regional MRA can be determined from local shells with known calendar ages and paired terrestrial-marine samples by 14C measurements, and also from scleractinian corals dated with accurate and precise U-Th and 14C measurements (Alves et al. Reference Alves, Macario, Ascough and Ramsey2018; Ascough et al. Reference Ascough, Cook and Dugmore2005; Durand et al. Reference Durand, Deschamps, Bard, Hamelin, Camoin, Thomas, Henderson, Yokoyama and Matsuzaki2013; Petchey et al. Reference Petchey, Dabell, Clark and Parton2023). Paired 230Th and 14C ages of corals have long been applied to construct calibration curves (Bard et al. Reference Bard, Arnold, Hamelin, Tisnerat-Laborde and Cabioch1998; Burr et al. 1998; Reference Burr, Galang, Taylor, Gallup, Edwards, Cutler and Quirk2004; Cutler et al. Reference Cutler, Gray, Burr, Edwards, Taylor, Gabioch, Beck, Cheng and Moore2004; Edward et al. Reference Edward, Beck, Burr, Donahue, Chappell and Bloom1993; Durand et al. Reference Durand, Deschamps, Bard, Hamelin, Camoin, Thomas, Henderson, Yokoyama and Matsuzaki2013; Fairbanks et al. Reference Fairbanks, Mortlock, Chiu, Cao, Kaplan, Guilderson, Fairbanks, Bloom, Grootes and Nadeau2005), and coral diagenesis that might alter both ages can be identified by various techniques. 230Th ages can be screened by δ234U measurements to evaluate whether the coral skeleton was preserved in a closed system (Edwards et al. Reference Edward, Beck, Burr, Donahue, Chappell and Bloom1993, Reference Edwards, Gallup and Cheng2003). Scanning electron microscopy (SEM) and X-ray diffraction (XRD) can help to detect secondary aragonite and calcite (Grothe et al. Reference Grothe, Cobb, Bush, Cheng, Santos, Southon, Edwards, Deocampo and Sayani2016 and references therein). Secondary calcite has a potentially large impact on 14C ages since it may reflect exchange with atmospheric carbon during times of former subaerial exposure. This can often be removed by stepped-dissolution of the coral skeleton under vacuum (Bard et al. Reference Bard, Arnold, Fairbanks and Hamelin1993; Burr et al. Reference Burr, Edwards, Donahue, Druffel and Taylor1992; Edwards et al. Reference Edward, Beck, Burr, Donahue, Chappell and Bloom1993), but the outcome may be different among different samples (Burr et al. Reference Burr, Haynes, Shen, Taylor, Chang, Beck, Nguyen and Zhou2015; Durand et al. Reference Durand, Deschamps, Bard, Hamelin, Camoin, Thomas, Henderson, Yokoyama and Matsuzaki2013; Reimer et al. Reference Reimer, Bard, Bayliss, Beck, Blackwell, Ramsey, Brown, Buck, Edwards, Friedrich, Grootes, Guilderson, Haflidason, Hajdas, Hatté, Heaton, Hogg, Hughen, Kaiser, Kromer, Manning, Reimer, Richards, Scott, Southon, Turney and van der Plicht2013b; Yokoyama et al. Reference Yokoyama, Esat, Lambeck and Fifield2000). XRD analysis is thus necessary to identify secondary calcite in corals before accelerator mass spectrometry (AMS) measurement, and a target criterion of less than 1% calcite in coral samples has been suggested (Reimer et al. Reference Reimer, Hughen, Guilderson, McCormac, Baillie, Bard, Barratt, Beck, Buck, Damon, Friedrich, Kromer, Ramsey, Reimer, Remmele, Southon, Stuiver and van der Plicht2002, Reference Reimer, Bard, Bayliss, Beck, Blackwell, Ramsey, Brown, Buck, Edwards, Friedrich, Grootes, Guilderson, Haflidason, Hajdas, Hatté, Heaton, Hogg, Hughen, Kaiser, Kromer, Manning, Reimer, Richards, Scott, Southon, Turney and van der Plicht2013b).
The ΔR values around Taiwan need to be updated, in part due to the changes introduced with the Marine20 calibration, but also because emerging studies about extreme climate change and natural hazards require more accurate age control during the late Holocene (Dezileau et al. Reference Dezileau, Lehu, Lallemand, Hsu, Babonneau, Ratzov, Lin and Dominguez2016; Ota et al. Reference Ota, Shyu, Wang, Lee, Chung and Shen2015; Yu et al. Reference Yu, Yen, Chen, Yen and Liu2016). Current MRA information around Taiwan is very limited, based on 6 shells collected at eastern Taiwan around 1900 AD from a museum collection (Yoneda et al. Reference Yoneda, Uno, Shibata, Suzuki, Kumamoto, Yoshida, Sasaki, Suzuki and Kawahata2007), and 11 pairs of charcoal and shell samples collected from sediment cores at a river mouth in western Taiwan (Yang et al. Reference Yang, Wang, Burr, Liu and Fan2019). Although the latter study reported an average MRA during the mid-late Holocene, the MRA variation of 82–1035 14C year was large. Factors that may affect MRA values in coastal areas, such as freshwater input and redeposition of charcoals, cannot be ignored (Ascough et al. Reference Ascough, Cook and Dugmore2005). Hence, corals in the Penghu Islands were chosen to obtain MRA information for western Taiwan for the following reasons: 1) the Penghu Islands are in the Taiwan Strait and there are no rivers developed on the islands, and 2) the Taiwan Strait is an important conduit of seawater exchange between the South China Sea and East China Sea.
In this study, we provide U-Th and radiocarbon ages of scleractinian corals collected from the Penghu Islands (Figure 1). We adopted a thorough physical cleaning process before any analyses and established an XRD measurement procedure that can detect calcite in our samples to less than 1%, with a relatively constant 0.2% uncertainty. After XRD screening, the coral samples with less than 1% calcite were sent for U-Th and 14C measurements. The marine reservoir age and the regional reservoir correction are calculated based on IntCal20 and Marine20 via the ResAge and the delta programs respectively (Reimer and Reimer Reference Reimer and Reimer2017; Soulet Reference Soulet2015). Those previously reported ΔR values inside the South China Sea and several available data from offshore eastern Taiwan were also recalculated in the same manner for comparison. The purpose of this study is to determine the best ΔR value for age calibration of marine samples from the late mid-Holocene around Taiwan and the northeastern South China Sea. Observed ΔR variations, which may be climate-driven, and comparisons with ΔR values from other Pacific Ocean locations will be discussed in a separate study.
Figure 1. Sampling sites of coral samples from the Penghu Islands (upper insert) and ΔR values from the Penghu Islands and other locations shown in three different time periods: 0-50, 50-5500, 5500-9000 cal BP. “x” indicates no data available in the period of time. References for dataset: 1. Wan et al. (Reference Wan, Meltzner, Switzer, Lin, Wang, Bradley, Natawidjaja, Suwargadi and Horton2020); 2. Bolton et al. (Reference Bolton, Goodkin, Druffel, Griffin and Murty2016); 3. Yu et al. (Reference Yu, Hua, Zhao, Hodge, Fink and Barbetti2010); 4. Hua et al. (Reference Hua, Ulm, Yu, Clark, Nothdurft, Leonard, Pandolfi, Jacobsen and Zhao2020); 5. Southon et al. (Reference Southon, Kashgarian, Fontugne, Metivier and Yim2002); 6. Hirabayashi et al. (Reference Hirabayashi, Yokoyama, Suzuki, Esat, Miyairi, Aze, Siringan and Maeda2019); 7. Hirabayashi et al. (Reference Hirabayashi, Yokoyama, Suzuki, Miyairi and Aze2017); 8. Hideshima et al. (Reference Hideshima and Matsumoto2001); 9. Zeng et al. (Reference Zeng, Yokoyama, Hirabayashi, Miyairi, Suzuki, Aze and Kawakubo2024); 10. Glynn et al. (Reference Glynn, Druffel, Griffin, Dunbar, Osborne and Sanchez-Cabeza2013); 11. Andrews et al. (Reference Andrews, Asami, Iryu, Kobayashi and Camacho2016). The currents, North Equatorial Current (NEC), North Equatorial Counter Current (NECC), Mindanao Current (MC) and Kuroshio Current (KC) follow Hu et al. (Reference Hu, Wu, Cai, Gupta, Ganachaud, Qiu, Gordon, Lin, Chen, Hu, Wang, Wang, Sprintall, Qu, Kashino, Wang and Kessler2015), and the currents through the Luzon Strait (LS) and the Penghu Channel (PC) were modified from Liang et al. (Reference Liang, Tang, Yang, Ko and Chuang2003) and Jan et al. (Reference Jan, Wang, Chern and Chao2002). Changyun Rise is a seafloor ridge with a depth of 30–40 mbsl perpendicular to western Taiwan coastline, denoted as CR (Jan et al. Reference Jan, Wang, Chern and Chao2002).
Methods
Notation
All ages are reported relative to 1950 CE (or 0 year before present, BP). Conventional 14C ages are henceforth referred to as “BP” and calibrated/calendar ages as “cal BP”. The marine reservoir age (R (t, site)) is defined as the difference in 14C yr between the radiocarbon age of dissolved inorganic carbon (DIC) of seawater and the radiocarbon age of CO2 in the atmosphere in Heaton et al. (Reference Heaton, Köhler, Butzin, Bard, Reimer, Austin, Ramsey, Grootes, Hughen, Kromer, Reimer, Adkins, Burke, Cook, Olsen and Skinner2020):
R (t, site) = 14C age of coral (from DIC) – 14C age of the atmosphere (from CO2)
Thus, in this study, we use “MRA” from Heaton et al. (Reference Heaton, Köhler, Butzin, Bard, Reimer, Austin, Ramsey, Grootes, Hughen, Kromer, Reimer, Adkins, Burke, Cook, Olsen and Skinner2020) to represent the marine reservoir “age” and discuss the 14C difference between the Northern Hemispheric atmosphere and surface seawater (surface to 100 m depth), as modeled in Marine20 (Heaton et al. Reference Heaton, Köhler, Butzin, Bard, Reimer, Austin, Ramsey, Grootes, Hughen, Kromer, Reimer, Adkins, Burke, Cook, Olsen and Skinner2020) and for the most commonly identified coral taxa (Hibbert et al. Reference Hibbert, Rohling, Dutton, Williams, Chutcharavan, Zhao and Tamisiea2016). The regional adjustment to the MRA uses ΔR values, defined as the difference between the regional and global MRA (Stuiver and Braziunas Reference Stuiver and Braziunas1993):
ΔR (t, site) = regional 14C age of coral – modeled global 14C age of surface seawater
Physical cleaning of coral samples
The Penghu Islands are about 50 km away from the western coast of Taiwan (Figure 1). The highest peak on the islands is 56 m above sea level, located in the center of the island, and no rivers drain the Penghu islands. Scleractinian coral samples were collected from 11 locations on the Penghu Islands (shown in the upper insert of Figure 1). Except Xixi (XX), all coral blocks were collected near the coast. After sampling, the weathered parts of the corals, yellow, black and visually recrystallized portions, were removed, and only the inner portions were used for the next step. The inner materials were crushed into pieces less than 3 mm and scrutinized under an optical microscope. Those pieces that exhibited signs of recrystallization under the microscope, e.g. secondary deposits on coral surfaces and within pores (McGregor et al. Reference McGregor and Abram2008; Nothdurft and Webb Reference Nothdurft and Webb2009), were discarded. A very small amount of secondary deposit was still observed in thin section images of coral samples with ≤1% calcite (McGregor et al. Reference McGregor and Abram2008), so we proceeded with an additional step. For the coral genera Cyphastrea, Dipsastraea, Platygyra, and Favites, septa were removed, and only the corallite walls were kept, and a scalpel was used to scrape out the surface of the corallite walls for possible secondary deposits that were not identified under an optical microscope. For the coral genera Porites and Acropora, it is difficult to clean the surface of the small pores, so we crushed the sample fragments into an even smaller size (∼1–3 mm) and analyzed only those fragments with clean pores. Photos of sample fragments before and after physical cleaning are shown in the Supplementary Material S1 (SM1). Finally, samples were ultrasonically cleaned and rinsed with milli-Q until no suspended particles were observed, and then the samples were dried in an oven at 50°C.
Powder X-ray diffraction (XRD) analysis
Powder XRD analysis was performed to scrutinize the level of secondary calcite in our coral pieces after physical cleaning as described above. A portion of the cleaned fragments were ground to a powder by agate pestle and mortar and sieved to less than 63 µm to homogenize the samples. The powdered samples were kept dry in the oven until pressed into the Bruker Si holder for XRD analysis. This special holder uses silicon wafers cut on the (911) plane to minimize diffraction signals and provide a “zero background” effect, and the quality of X-ray diffraction data can be enhanced accordingly. Next, the samples were scanned by an XRD (Bruker D2 Phaser with a LYNXEYE XE-T detector) in the Department of Geosciences, National Taiwan University (NTU). The amount of calcite in samples was measured, and the percentage was determined by TOPAS software (TOPAS V.6 Bruker AXS Inc. Karlsruhe, Germany), based on the method of Rietveld refinement.
The intensity of the largest calcite peak (104) was often used to detect the calcite content in carbonate samples (Chiu et al. Reference Chiu, Fairbanks, Mortlock and Bloom2005; Douka et al. Reference Douka, Hedges and Higham2010). However, inconsistent XRD results from preferred orientation (caused by pressing powder on slides with inconsistent force) was observed in our earlier experiments of standard powders and also in previous study (Sepulcre et al. Reference Sepulcre, Durand and Bard2009). Thus, the Bruker D2 Phaser was chosen for its horizonal sample stage with a theta/theta design, which allowed for a consistent pressing. Next, we adopted the Rietveld refinement, which estimates and matches all peaks of calcite and aragonite diffraction pattern to reduce the uncertainties from preferred orientation and other effects for precise quantification (Bish and Post Reference Bish and Post1993). This method has been used to evaluate secondary calcite in coral samples (Grothe et al. Reference Grothe, Cobb, Bush, Cheng, Santos, Southon, Edwards, Deocampo and Sayani2016). Moreover, we found that it is not proper to report percent calcite with the uncertainty only from the fitting result of the Rietveld refinement via single measurement. Hence, we utilized the sample holder that allowed us to do repeated measurements to evaluate the uncertainties from other factors such as the size of sample powders and sample preparation by different persons.
A calibration line of calcite weight percent vs. percent calcite fitting by the Rietveld refinement was constructed via standard powders. The standard powders mixing modern coral (100% aragonite) with calcite crystals (100% calcite) were made in the proportions: 0.2, 0.5, 0.75 and 1 weight % calcite. The importance of screening coral samples containing less than 1% calcite is explained below. The lowest detection limit (LDL), reproducibility and overall uncertainty of the repeated measurements were determined from our calibration line (in Result, Figure 2a). The same instrument settings and the same scan range parameters, 20°–80° 2θ, 0.01° step size, and 0.3 s per step, from the measurements of standards, were carried out for samples.
Figure 2. (a) The calibration line of calcite weight% vs. calcite% fitting by the Rietveld refinement method. (b) Marine reservoir age (MRA) and (c) marine reservoir age correction (ΔR) derived from 11 of Penghu coral samples that were processed with physical cleaning and XRD screening (After) and without (Before).
U-Th and 14 C measurements
Approximately 50 mg of coral fragments were collected for U-Th and 14C measurements. The chemical procedure of the U-Th dating method in the clean room and the analytical protocol using the Thermo Electron NEPTUNE multi-collector-inductively coupled plasma-mass spectrometer (MC-ICP-MS) were developed by the High-precision mass Spectrometry and Environment Change Laboratory (HISPEC) in the Department of Geosciences, NTU. The details of experiments are described in Shen et al. (Reference Shen, Li, Sieh, Natawidjaja, Cheng, Wang, Edwards, Lam, Hsieh, Fan, Meltzner, Taylor, Quinn, Chiang and Kilbourne2008, Reference Shen, Wu, Cheng, Edwards, Hsieh, Gallet, Chang, Li, Lam, Kano, Hori and Spötl2012). The measured seawater δ234U with seawater values of 147.4±2.0‰ (Shen et al. Reference Shen, Li, Sieh, Natawidjaja, Cheng, Wang, Edwards, Lam, Hsieh, Fan, Meltzner, Taylor, Quinn, Chiang and Kilbourne2008) from southern Taiwan is used for the post-analysis screening criteria, and the weighted mean δ234U value of 146.6±1.5‰ in Penghu corals matches this local value. The initial 230Th/232Th atomic ratio of 4 (±2) × 10–6 (Shen et al. Reference Shen, Li, Sieh, Natawidjaja, Cheng, Wang, Edwards, Lam, Hsieh, Fan, Meltzner, Taylor, Quinn, Chiang and Kilbourne2008) was used to calculate corrected U-Th ages (cal BP, relative to 1950 CE). The 50 mg coral pieces for the 14C measurements were performed by the BETA Analytic. The procedure can be found in the company’s website and is briefly described here. About 10–30% of the surface material is further removed by 1N HCl solution at room temperature and then rinsed and dried in 90°C oven for 12–14 hr. Next, the sample powder is hydrolyzed by 85 wt.% H3PO4 (99.9% purity) and heated at 90°C for 4 hr or more until complete dissolution. CO2 is collected through cryogenic purification and then transferred to a cell with cobalt powder (Co:C∼3:1 in weight). Hydrogen is added subsequently (H2:CO2∼3:1 in volume) for the process of graphitization. The reaction cell is heated at 550–660°C until the reduction reaches an 80–100% completion, and the sample graphites are measured with standards in BETA’s AMS facility.
R and ΔR calculation
Measured 14C and U-Th ages were combined to compute MRA values with 1σ uncertainties using ResAge software (Soulet Reference Soulet2015) and IntCal20 data (Reimer et al. Reference Reimer, Austin, Bard, Bayliss, Blackwell, Ramsey, Butzin, Cheng, Edwards, Friedrich, Grootes, Guilderson, Hajdas, Heaton, Hogg, Hughen, Kromer, Manning, Muscheler, Palmer, Pearson, van der Plicht, Reimer, Richards, Scott, Southon, Turney, Wacker, Adolphi, Büntgen, Capano, Fogtmann-Schulz, Friedrich, Köhler, Kudsk, Miyake, Olsen, Reinig, Sakamoto, Sookdeo and Talamo2020). The method of data input and adopted R programs were described in Soulet (Reference Soulet2015). The probability density function of the 14C reservoir age offset derived from R program in Soulet (Reference Soulet2015) is likely not symmetric, so a mid-range value and the range at 68% (1σ) is chosen for the purpose of calculating the following weighted mean and uncertainties. We also compare the global MRA derived from IntCal13 and Marine13 data (Reimer et al. Reference Reimer, Bard, Bayliss, Beck, Blackwell, Ramsey, Buck, Cheng, Edwards, Friedrich, Grootes, Guilderson, Haflidason, Hajdas, Hatté, Heaton, Hoffmann, Hogg, Hughen, Kaiser, Kromer, Manning, Niu, Reimer, Richards, Scott, Southon, Staff, Turney and van der Plicht2013a) to the MRA from IntCal20 and Marine20 for the past 9000 cal BP, designated as R13 and R20 and illustrated in Figure 3a. ΔR20 was calculated via the online program deltar (Reimer and Reimer Reference Reimer and Reimer2017) with Marine20 (Heaton et al. Reference Heaton, Köhler, Butzin, Bard, Reimer, Austin, Ramsey, Grootes, Hughen, Kromer, Reimer, Adkins, Burke, Cook, Olsen and Skinner2020), indicating the difference between the site-specific MRA and R20. ΔR13 are values from previous studies that are derived from Marine13. Those ΔR without a subscript indicate ΔR20 values.
Figure 3. (a) Marine reservoir ages (MRA) from Penghu and the other five locations in the SCS: Natuna, Nansha, Xisha, Leizhou-Hainan and Luzon Islands. Marine reservoir age curves derived from Marine 20 (R20) and Marine13 (R13) are plotted for comparison. (b) Regional reservoir corrections (ΔR) of Penghu and the same five SCS locations based on Marine20.
The equations used for calculating the weighted mean and the estimated uncertainty for the weighted mean of a group of MRA or ΔR are from Bevington and Robinson (Reference Bevington and Robinson2003). The reported uncertainty for the weighted mean is the larger one of Equation (2) and Equation (3).
$$Weighted\;mean\;of\;a\;group\;\Delta R\;or\;MRA = \mu = {{\mathop \sum \nolimits_i {{{{X_i}}}\over{{\sigma _i^2}}}}\over{{\mathop \sum \nolimits_i {{1}\over{{\sigma _i^2}}}}}}\;$$
$$Error\;in\;the\;weighted\;mean\;from\;unequal\;uncertainties\;of\;a\;group\;\Delta R\;or\;MRA\!:$$
$$\;{\sigma _\mu } = \sqrt {{1}\over{{\mathop \sum \nolimits_i {{1}\over{{\sigma _i^2}}}}}} \;$$
$$Standard\;deviation\;of\;a\;group\;\Delta R\;or\;MRA\!:$$
$$\;{\sigma _{}} = \sqrt {{{{n\;\mathop \sum \nolimits_i {{{{{\left( {{{\rm{{\rm X}}}_i} - \mu } \right)}^2}}}\over{{\sigma _i^2}}}}}\over{{\left( {n - 1} \right)\mathop \sum \nolimits_i {{1}\over{{\sigma _i^2}}}}}}}$$
Where Xi indicates the value of data point X, σi is the uncertainty of data point Xi, and n is the sample numbers of a group of MRA or ΔR.
Results
The necessary of 1% calcite criterion
The effect of secondary calcite on the 14C age of coral samples depends on both the age of calcite and coral aragonite. The age of secondary calcite is difficult to know, so we could assume an extreme case of modern calcite. The apparent age of coral samples, a mixture of calcite and aragonite, can then be estimated via a simple mixing calculation (Supplementary Material 2, SM2). Considering the oldest age of coral in our study, 7000 14C years, 1% modern carbon incorporation from the secondary calcite would reduce 110 14C years from the original 14C age of aragonite (Figure S2). From 2% to 5%, the 14C ages are reduced from 220 to 540 years, which is much larger than the total range of R13 and R20 variation, 206 and 282 14C years respectively in 0-10,500 cal BP (Heaton et al. Reference Heaton, Bard, Ramsey, Butzin, Hatté, Hughen, Köhler and Reimer2023). It appears that the natural local MRA variation would be difficult to observe if the 14C age of coral is affected by secondary calcite of modern carbon larger than 2%. Although keeping calcite content as low as not detected (i.e. <0.2±0.2%) would be ideal for 14C measurements on coral samples, this strict requirement would eliminate 3/4 of all samples in this study. Hence in our study, 1% calcite criterion is adopted following the suggestion of Reimer (Reference Reimer, Hughen, Guilderson, McCormac, Baillie, Bard, Barratt, Beck, Buck, Damon, Friedrich, Kromer, Ramsey, Reimer, Remmele, Southon, Stuiver and van der Plicht2002, Reference Reimer, Bard, Bayliss, Beck, Blackwell, Ramsey, Brown, Buck, Edwards, Friedrich, Grootes, Guilderson, Haflidason, Hajdas, Hatté, Heaton, Hogg, Hughen, Kaiser, Kromer, Manning, Reimer, Richards, Scott, Southon, Turney and van der Plicht2013b).
XRD and sampling process
The purpose of the calibration line developed from carbonate standards is to optimize the screening process for coral samples that meet the requirement of <1% calcite. The result shows that the linear fit between calcite weight% and fitting% from the Rietveld refinement is significant (R2=0.84, Figure 2a). The LDL is about 0.2%, as in Chiu et al. (Reference Chiu, Fairbanks, Mortlock and Bloom2005) and Sepulcre et al. (Reference Sepulcre, Durand and Bard2009). The overall uncertainty for all standard powders with different proportions is approximately ±0.2% based on repeated measurements on different days and by different operators. Hence, the 0.2% error is adopted with percent calcite obtained from the Rietveld refinement fitting for all coral samples. Only the powder samples that contained less than 1% calcite, i.e. <0.8±0.2%, passed the criteria (Reimer et al. Reference Reimer, Hughen, Guilderson, McCormac, Baillie, Bard, Barratt, Beck, Buck, Damon, Friedrich, Kromer, Ramsey, Reimer, Remmele, Southon, Stuiver and van der Plicht2002, Reference Reimer, Bard, Bayliss, Beck, Blackwell, Ramsey, Brown, Buck, Edwards, Friedrich, Grootes, Guilderson, Haflidason, Hajdas, Hatté, Heaton, Hogg, Hughen, Kaiser, Kromer, Manning, Reimer, Richards, Scott, Southon, Turney and van der Plicht2013b) and then the coral pieces with the same physical conditions proceeded with U-Th and 14C measurements.
All coral samples in this study followed the same protocol: physical cleaning and then XRD screening. Only one sample that didn’t meet the 1% calcite requirement was discarded and the rest passed. The XRD results show all passed samples contain less than 0.5% calcite (Table S2), and the values from repeated measurements are all within 0.2% uncertainty. 53 MRA and ΔR, including duplicates for both U-Th and 14C measurements, were obtained subsequently. The results of XRD analysis, 230Th and 14C ages of cleaned coral pieces are presented in Supplementary Tables S1 and S2, respectively.
Among all, 11 samples were inspected under the optical microscope and then analyzed with and without physical cleaning and XRD examination, and the influence of two different processes on MRA and ΔR is illustrated in Figures 2b and 2c. Comparing “after” to “before” physical cleaning, the calcite% was reduced from various values (0.5–2%) to all less than 0.5% consistently. The range of MRA and ΔR are decreased from 506 to 298 14C yr and from 366 to 227 14C yr, respectively. These large ranges in MRA and ΔR before physical cleaning were likely caused by diagenetic calcite and aragonite, which affected both the radiocarbon and U-Th ages. Our method successfully removes those secondary deposits that are not easily detected under the optical microscope and provides more reliable and consistent data that may reflect the true variation of MRA and ΔR in this area.
Marine reservoir age (MRA)
MRAs derived from measured 14C ages and 230Th ages as calendar ages of coral samples from the Penghu Islands range from 202 to 620 14C yr for all samples studied (total range from 50 and 6700 cal BP) (Table S2, Figure 3a). MRA values range from 202 to 468 14C yr, with an average of 335±72 14C yr from 50 to 5500 cal BP; and from 5500 to 6700 cal BP, MRA varies from 291 to 620 14C yr, with an average of 456±108 14C yr. The lowest MRAs were observed around 1300–1400 and 3500–3800 cal BP, and the highest MRA was observed near 5800 cal BP.
Regional reservoir correction (ΔR)
ΔR reflects the trends observed for local MRA, with a change before and after 5500 cal BP at Penghu Islands. ΔR varies from –254 to 49 14C yr before 5500 cal BP, and from –271 to –47 14C yr after (Table S2, Figure 3b). The ΔR variation is much smaller in 50–5500 cal BP than that prior to 5500 cal BP. The two low MRAs correspond to low (more negative) ΔR (–271 and –242 14C yr), and another low ΔR value (–233 14C yr) was observed near 340 cal BP. The weighted means of ΔR from 50–5500 and 5500–6700 cal BP are –155±59 14C yr and –107±96 14C yr, respectively.
Discussion
Holocene Marine reservoir age (MRA) for the Taiwan Strait and the SCS (4°N–24°N) (Figure 3a)
The range of MRA derived from the Penghu Islands corals were compared with coral results from other SCS sites. Their locations are shown in Figure 1, and recalculated MRA and ΔR values, using IntCal20 and Marine20 respectively, are presented in Table S3. The southernmost site is Natuna Island at 4°N (N = 12, Wan et al. Reference Wan, Meltzner, Switzer, Lin, Wang, Bradley, Natawidjaja, Suwargadi and Horton2020). Moving northward, the next site is at 10.7°N Nansha Island (N = 4, Yu et al. Reference Yu, Hua, Zhao, Hodge, Fink and Barbetti2010; N = 9, Hua et al. Reference Hua, Ulm, Yu, Clark, Nothdurft, Leonard, Pandolfi, Jacobsen and Zhao2020) and at 16.2°N, Xisha Island (N = 8, Hua et al. Reference Hua, Ulm, Yu, Clark, Nothdurft, Leonard, Pandolfi, Jacobsen and Zhao2020). At 20°N, results are given for the Leizhou Peninsula and Hainan Island in the northern SCS (N = 23, Yu et al. Reference Yu, Hua, Zhao, Hodge, Fink and Barbetti2010). Four coral samples from the eastern portion of the SCS come from Luzon Island at about 16.8–18°N (Hirabayashi et al. Reference Hirabayashi, Yokoyama, Suzuki, Esat, Miyairi, Aze, Siringan and Maeda2019).
The weighted means of MRA in the period 0-5500 cal BP from the above locations are: 341±37 (Natuna), 357±64 (Nansha), 425±48 (Xisha), 417±77 (Leizhou-Hainan), 318±32 (Luzon), and 335±72 14C yr (Penghu). The Xisha and Leizhou-Hainan islands have slightly higher average MRA, drawn by the high MRA values around 2000 and 2500 cal BP. The MRA at the other four sites are consistent within uncertainties. Compared to the more stable MRA during 0-5500 cal BP, larger MRA variation was observed at all locations prior to 5500 cal BP; from about 290 (Penghu) to 800 (Nansha) 14C yr. The weighted MRA means are also calculated for all locations in the same order: 401±127 14C yr (Natuna), 737±68 14C yr (Nansha), 518±31 14C yr (Xisha), 530±86 14C yr (Leizhou-Hainan), 640±134 (Luzon), and 456±108 14C yr (Penghu).
Holocene regional reservoir correction (ΔR) for the Taiwan Strait and the SCS (4°N–24°N) (Figure 3b)
Calculated ΔR for the Penghu Islands (Table S2) and recalculated ΔR for the SCS locations based on Marine20 (Table S3) are compared next. Most of the ΔR derived from the Penghu Islands are synchronous with published SCS ΔR values, and minor differences reflect site-specific ocean circulation (Figure 3b). The weighted mean ΔR from the same locations in the period 0–5500 cal BP are –139±22 14C yr (Natuna), –132±76 14C yr (Nansha), –119±119 14C yr (Xisha), –46±82 14C yr (Leizhou-Hainan), –154±39 (Luzon), and –155±59 14C yr (Penghu). The surface seawater affected by upwelling near Xisha and Leizhou-Hainan evidently influences the average ΔR values at those sites (Hua et al. Reference Hua, Ulm, Yu, Clark, Nothdurft, Leonard, Pandolfi, Jacobsen and Zhao2020; Yu et al. Reference Yu, Hua, Zhao, Hodge, Fink and Barbetti2010), but the high ΔR values around 2000 and 2500 cal BP are not observed at the other locations. In the southern SCS, ΔR values from Natuna and Nansha are slightly higher but statistically consistent with those from the Penghu and Luzon Islands at the northern SCS.
Prior to 5500 cal BP, data are limited for most locations, except Leizhou-Hainan and Penghu Islands. The ΔR values at both locations have greater variability before 5500 cal BP than after. The range of ΔR values at Leizhou-Hainan is about 380 and 194 14C yr before and after 5500 cal BP respectively, and those corresponding to the same period for Penghu Islands are about 300 and 215 14C yr. The larger ΔR variation during the period 5.5–8 cal BP in the SCS likely resulted from combined local upwelling with more frequent or strong upwelling of cold and 14C-depleted water from the tropical east Pacific and then transported to the west Pacific Ocean (Hua et al. Reference Hua, Ulm, Yu, Clark, Nothdurft, Leonard, Pandolfi, Jacobsen and Zhao2020). A similar observation at Penghu Islands reflects that the seawater in the Taiwan Strait is also likely influenced by the Pacific seawater through the Kuroshio Current (KC) intrusion over the Luzon Strait (discussed below), in addition to the local factor, unstable seawater stratigraphy caused by sea level rise.
The ΔR variation in the Taiwan Strait before 5500 cal BP
An upward trend, increasing ΔR values from 6700 to 5800 cal BP, is observed for the Penghu Islands in the Taiwan Strait. The Taiwan Strait was dry during the LGM and then gradually filled up during deglaciation. Hence, this increasing trend might indicate gradual seawater intrusion before reaching the regional sea level high stand between 5000 and 7000 cal BP in the Taiwan Strait (Yang et al. Reference Yang, Liu, Fan, Burr, Lin and Chen2017; Zong et al. Reference Zong2004). More detailed studies are needed to evaluate site-specific ΔR variation in this period of variable MRA, and it would be interesting to see the difference in ΔR variation between global ocean circulation and climate change as driving forces offshore eastern Taiwan and the additional influence of rising sea level offshore western Taiwan during this period of time. Until then, an average ΔR of –107±96 14C yr in the period 5500–6700 cal BP may be applied around Taiwan, but it should be used with caution.
Factors that affect the MRA and ΔR in the southern Taiwan Strait and the northeastern SCS between 0–5500 cal BP
The surface currents surrounding the Penghu Islands are mainly from the southeast, through the Penghu Channel (PC in Figure 1; Hu et al. Reference Hu, Kawamura, Li, Hong and Jiang2010; Jan et al. Reference Jan, Wang, Chern and Chao2002; Jan and Chao Reference Jan and Chao2003; Liang et al. Reference Liang, Tang, Yang, Ko and Chuang2003; Wang et al. 2004). The Taiwan Strait is affected by the East Asian monsoon from the northeast during September–May, and from the southwest during June–August (Jan et al. Reference Jan, Wang, Chern and Chao2002, Reference Jan, Sheu and Kuo2006). While the China coastal current along the west margin of the Taiwan Strait flows southward in winter and northward in summer, consistent with direction of winds, the current along the eastern side of the Taiwan Strait flows northward all year, bringing water from the northeastern SCS through the Penghu Channel (Jan and Chao; Liang et al. Reference Liang, Tang, Yang, Ko and Chuang2003; Wang et al. Reference Wang, Chiao, Lwiza and Wang2004). The northward current velocity through the Penghu Channel varies seasonally, lower in winter (<10 cm/s) and higher in summer (up to 100 cm/s) (Jan et al. Reference Jan and Chao2003, Reference Jan, Sheu and Kuo2006). Even when impeded by the strong northeastern monsoon, the northward current can still reach the Changyuen Ridge (CR in Figure 1), the seafloor ridge north of the Penghu Islands; and turn to the northwest around the islands. The Changhuen Ridge, perpendicular to Taiwan’s west coast, also blocks most of the China coastal water from the north during the regular winter monsoon (Chang et al. Reference Jan and Chen2009; Jan et al. 2002; Reference Jan, Sheu and Kuo2006; Lin et al. Reference Lin, Tang, Jan and Chen2005). Despite higher current speeds and larger flowthrough observed in the Penghu Channel during summer, the southwestern monsoon is still much weaker (<5 m/s) than the northeastern monsoon (∼10 m/s) (Hu et al. Reference Hu, Kawamura, Li, Hong and Jiang2010; Jan and Chao Reference Jan and Chao2003). Hence, another forcings, such as pressure gradient and the influence of KC, other than local winds must be involved for the constant norward flow (Jan et al. Reference Jan, Wang, Chern and Chao2002, Reference Jan, Sheu and Kuo2006; Jan and Chao Reference Jan and Chao2003; Liang et al. Reference Liang, Tang, Yang, Ko and Chuang2003; Lin et al. Reference Lin, Tang, Jan and Chen2005).
In addition to the pressure gradient, with sea surface height generally lower at the northern entrance of the Taiwan Strait, the northeastern SCS waters that feed the Penghu Channel feature some of the largest nonlinear internal waves observed in the global ocean (Alford et al. Reference Alford, Peacock, MacKinnon, Nash, Buijsman, Centurioni, Chao, Chang, Farmer, Fringer, Fu, Gallacher, Graber, Helfrich, Jachec, Jackson, Klymak, Ko, Jan, Johnston, Legg, Lee, Lien, Mercier, Moum, Musgrave, Park, Pickering, Pinkel, Rainville, Ramp, Rudnick, Sarkar, Scotti, Simmons, St Laurent, Venayagamoorthy, Wang, Wang, Yang, Paluszkiewicz and Tang2015). These internal waves are generated from the Luzon Strait, caused by the interaction between internal tides, seafloor topography and water column stratification, associated with the KC waters. The strong internal wave energy can propagate westward to the continental shelf of the northern SCS (117°E) from the Luzon Strait (121°E) in days (LZ in Figure 1; Lien et al. Reference Lien, Tang, Chang and D’Asaro2005; Ramp et al. Reference Ramp, Yang and Bahr2010; St. Laurent et al. Reference St. Laurent, Simmons, Tang and Wang2011), as well as to the 200-meter isobath at the southeastern Taiwan Strait, where the Penghu Channel is (Bai et al. Reference Bai, Liu, Li and Hu2014; Beardsley et al. Reference Beardsley, Duda, Lynch, Senior Member, Irish, Ramp, Chiu, Tang, Yang and Fang2004). The propagation speed of the waves is higher in spring and summer (∼100cm/s) than in winter (∼30cm/s), and the frequency of wave generation is also higher in summer than winter (Ramp et al. Reference Ramp, Yang and Bahr2010; Shaw et al. Reference Shaw, Ko and Chao2009). Thus, the constant northward flow through the Penghu Channel is enhanced by the radiation of internal wave energy (Beardsley et al. Reference Beardsley, Duda, Lynch, Senior Member, Irish, Ramp, Chiu, Tang, Yang and Fang2004; Lien et al. Reference Lien, Tang, Chang and D’Asaro2005). The vertical displacement of seawater caused by internal waves can be larger than 150 m (Alford et al. Reference Alford, Lien, Simmons, Klymak, Ramp, Yang, Tang and Chang2010; Ramp et al. 2010), and the subsurface cold and nutrient-rich water brought up to the surface by internal wave movement were also observed (Jan and Chen Reference Jan and Chen2009; Reid et al. 2019; Wang et al. Reference Wang, Dai and Chen2007). Moreover, surface water circulation in the northeastern SCS due to the KC intrusion in different seasons also produces vertical motions induced by eddies (Hu et al. Reference Hu, Kawamura, Hong and Qi2000; Nan et al. Reference Nan, Xue and Yu2015). Another turbulent mixing of water masses occurs when current flows through the Penghu Channel due to its funnel-shaped topography (Shao et al. Reference Shao, Tseng, Lien, Chang and Chen2018). Given the previous findings, the surface 14C value is mixed with the subsurface one at several locations, including the Luzon Strait, northeastern SCS and the Penghu Channel. Hence, the temporal ΔR variation recorded in corals at the Penghu Islands may imply the regional ocean circulation, even though the water depth around the Penghu Island is shallow (∼60 m).
All in all, several driving forces are combined for transporting seawater from the Luzon Strait to the southern Taiwan Strait. The vertical mixing of seawater enhanced by the internal wave propagation likely provides similar water source for the northeastern SCS and the southern Taiwan Strait, supported by the comparable weighted mean ΔR values between the Penghu Islands and the Luzon Island. Other locations in the SCS may involve local upwelling and other local surface flows, resulting in different weighted mean ΔR (Yang et al. Reference Yang, Zhou, Cheng, Zhang, Wei, Zhou, Yan, Chen and Hou2025). Moreover, the similar ΔR temporal variation between the Penghu Islands and the other sides in the SCS also indicates parallel forcings in the region, such as East Asian monsoon and El Niño/Southern Oscillation (ENSO). More thorough discussion including a comparison with the ΔR fluctuation in the south Pacific Ocean will be provided elsewhere.
The Holocene average ΔR for the marine reservoir age correction around Taiwan during the period 0–200 cal BP
Age control is important for natural hazard studies around Taiwan, such as paleotsunamis. Big tsunami events can be found in historical documents (Cheng et al. Reference Cheng, Shaw and Yeh2016; Lau et al. Reference Lau, Switzer, Dominey-Howes, Aitchison and Zong2010), and recorded in the coastal deposits, such as the events at 1782 CE in southwestern Taiwan, and 1867 CE in northeastern Taiwan (Li et al. Reference Li, Switzer, Wang, Weiss, Qiu, Chan and Tapponnier2015; Liu et al. Reference Liu, Wu and Hsu2022; Yu et al. Reference Yu, Yen, Chen, Yen and Liu2016). However, the recurrence interval of tsunamis is not easy to estimate since other possible tsunami records have been observed in a number of locations (Yu et al. Reference Yu, Yen, Chyi, Lu, Matsuta and Yen2025). Radiocarbon ages of shells and corals have often been used to constrain the ages of tsunami events (Ota et al. Reference Ota, Shyu, Wang, Lee, Chung and Shen2015; Yu et al. Reference Yu, Yen, Chen, Yen and Liu2016). Although the old ΔR13 of 73±17 14C yr (Yoneda et al. Reference Yoneda, Uno, Shibata, Suzuki, Kumamoto, Yoshida, Sasaki, Suzuki and Kawahata2007) was replaced by the recalculated ΔR20 of –73±39 14C yr from the shell record, more accurate age control is necessary to refine the tsunami studies in this short period of time. In this study, there are only seven ΔR data points available between 0 and 200 cal BP (Figure 4b). Hence, we propose to use the weighted mean ΔR of –155±59 14C yr from 50–5500 cal BP for age correction of marine samples in this short period until more data become available.
Figure 4. A comparison of regional reservoir correction (ΔR) obtained from coral samples between Penghu and (a) Guam and Palau, (b) Hon Tre and Xisha, (c) Ryukyu Islands including Kikai, Okinawa and Ishigaki. The shaded area indicates the weighted mean ΔR of –155±59 (1σ) 14C yr, derived from the Penghu Islands in 50–5500 cal BP.
In order to assess the suitability of this value, we compare available modern-day regional MRA and ΔR, also recalculated using IntCal20 and Marine20 (Table S4). Sites include: (A) Guam and Palau in the path of the North Equatorial Current (NEC) and North Equatorial Countercurrent (NECC), respectively (Andrews et al. Reference Andrews, Asami, Iryu, Kobayashi and Camacho2016; Glynn et al. Reference Glynn, Druffel, Griffin, Dunbar, Osborne and Sanchez-Cabeza2013; Southon et al. Reference Southon, Kashgarian, Fontugne, Metivier and Yim2002; Yoneda et al. Reference Yoneda, Uno, Shibata, Suzuki, Kumamoto, Yoshida, Sasaki, Suzuki and Kawahata2007), (B) The Philippines, Xisha and Hon Tre islands, in the SCS (Bolton et al. Reference Bolton, Goodkin, Druffel, Griffin and Murty2016; Southon et al. Reference Southon, Kashgarian, Fontugne, Metivier and Yim2002), and (C) eastern Taiwan and the Ryukyu Islands, including Ishigaki, Okinawa, Kikai and Amami-Oshima (Hideshima et al. Reference Hideshima and Matsumoto2001; Hirabayashi et al. Reference Hirabayashi, Yokoyama, Suzuki, Miyairi and Aze2017; Yoneda et al. Reference Yoneda, Uno, Shibata, Suzuki, Kumamoto, Yoshida, Sasaki, Suzuki and Kawahata2007; Zeng et al. Reference Zeng, Yokoyama, Hirabayashi, Miyairi, Suzuki, Aze and Kawakubo2024). Although species-specific feeding habits might affect the 14C ages of shells as compared to corals (Ulm et al. Reference Ulm, O’Grady, Petchey, Hua, Jacobsen, Linnenlucke, David, Rosendahl, Bunbury, Bird and Reimer2023), it remains useful to consider the full natural range of age values. Indeed, data from some of the shells deviate from coral ages in the figures, so our subsequent discussion focuses on coral ages.
In spite of the temporal variability, our weighted mean ΔR of –155±59 14C yr overlaps with the coral data from Palau and Guam, an average ΔR value of –149±17 and –154±38 14C yr, respectively, between 1949 CE and 1939 CE (Figure 4a). The average ΔR of –135±14 14C yr at Hon Tre Island and –136±47 at Xisha Island in the southern SCS from 1949 CE to 1900 CE is slightly higher than the ΔR value for Palau, Guam and Penghu (Figure 4b). In addition, this average ΔR of modern-day values at Hon Tre Island is consistent with the average ΔR of mid-late Holocene values at Natuna (–139±22 14C yr) and Nansha (–132±76 14C yr). The Ryukyu Islands are in the path of the KC further downstream, but their recent ΔR records are not sufficient to display KC variation in this short period of time (Figure 4c). Only the ΔR value of –149±74 14C yr at Kikai between 1947 CE and 1950 CE is close to our weighted mean ΔR value (Zeng et al. 2024). Although the ΔR variation offshore eastern Taiwan is yet to be determined, the weighted mean ΔR value around the Penghu Islands is comparable to the ΔR value at Palau and Guam, indicating a strong contribution of KC waters upstream from the Taiwan Strait. Therefore, a ΔR of –155±59 14C yr could be used for the age correction in the period of 0–200 cal BP, not only in the Taiwan Strait and offshore western Taiwan, but also offshore eastern Taiwan.
Summary
-
1. We determined updated ΔR values based on Marine20 for the Penghu Islands in the Taiwan Strait during the past 6700 cal BP. Meticulous physical cleaning and XRD analyses were employed, and secondary calcite in coral samples was effectively removed. MRA and ΔR values were then obtained from paired 14C and U-Th analyses on cleaned samples. The temporal variation of ΔR in the Penghu Islands was compared with published data from the SCS and surrounding region. Slight differences among sites reflect the variability of ocean circulation at each site.
-
2. The construction of time variable ΔR can improve the accuracy of radiocarbon ages of marine samples at specific sites. An assumption of a time-invariant ΔR is still necessary where continuous radiocarbon reservoir age data is not yet available. A single average ΔR value is suitable during the period 0-5500 cal BP with relatively stable MRA variation observed at several locations in this study. The weighted mean ΔR for the Penghu Islands between 0 and 5500 cal BP is –155±59 14C yr, and this value can be applied around Taiwan, including the Taiwan Strait and offshore eastern Taiwan.
-
3. The average ΔR value of –155±59 14C yr in the Penghu Island is close to the value at Luzon, but slightly lower than the values at other sites in the SCS and fits better with modern values from Palau and Guam in the path of the NEC and NECC. This implies that despite local upwelling and seawater mixing with other currents in the SCS, temporal ΔR variations in the Taiwan Strait are more closely related to temporal changes in upstream surface waters of the southwest North Pacific Ocean.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/RDC.2025.10146
Acknowledgments
We are grateful to the National Science and Technology Council in Taiwan (NSTC 110-2116-M-002-004, NSTC 111-2116-M-002-033 and NSTC 112-2116-M-002-028 to S.L. Wang and G.S. Burr) for funding. The U-Th dating was supported by grants from the NTU Core Consortiums Project (113L891902), Higher Education Sprout Project of the Ministry of Education (112L894202), and the projects under the National Science and Technology Council (111-2116-M-002-022-MY3, 113-2926-I-002-510-G) and Academia Sinica (AS-TP-113-L04). We also thank Yu-Hsiu Chiu for logistic support, Sze-Chieh Liu and Aileen Hoyle for their hard work in the field, Sze-Chieh Liu and Chun-Yuan Huang for their assistance of the laboratory work, and Mr. John Chang from Bruker Taiwan Co., Ltd. for his assistance in the operation and maintenance of the XRD instrument. The authors have no competing interests to declare.
Open access
