Contents lists available at ScienceDirect

Palaeogeography, Palaeoclimatology, Palaeoecology

journal homepage: www.elsevier.com/locate/palaeo

3500-year western Pacific storm record warns of additional storm activity ina warming warm pool

Yuanfu Yuea, Kefu Yua,⁎, Shichen Taob, Huiling Zhangc, Guohui Liub, Ning Wanga, Wei Jianga,Tianla Fana, Wuhui Lina, Yinghui Wanga

aGuangxi Laboratory on the Study of Coral Reef in the South China Sea, Coral Reef Research Centre of China, School of Marine Sciences, Guangxi University, Nanning530004, PR Chinab Key Laboratory of Marginal Sea Geology, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou 510301, PR Chinac Department of Ocean Engineering, Faculty of Ocean Engineering, Guangdong Ocean University, Zhanjiang 524088, PR China

A R T I C L E I N F O

Keywords:Coral reef lagoonStorm reconstructionGrain sizeSouth China SeaWestern PacificLate Holocene

A B S T R A C T

Frequent storm surges often cause catastrophic impacts on human lives and the global economy; however, thesephenomena are not well understood. In this study, a regional storm reconstruction is performed based on a grain-size analysis and stratigraphic modelling of the accelerator mass spectrometry radiocarbon dates of benthicforaminifera from two neighbouring lagoon cores from Lingyang Reef in the Xisha Islands located in thenorthern South China Sea of the western Pacific. The dating results from the lagoon cores reveal a ~3500-yeardepositional history. Three different depositional units are recognized based on a time series of distinct grain-sizevariations that correspond to the following three stages of storm activity: intense and frequent storms from~3500 to 3100 cal yr BP and ~1800 cal yr BP to present and weak and infrequent storms from ~3100 to1800 cal yr BP. A high sedimentation rate remarked by reverse and chaos age is observed from ~2800 to2600 cal yr BP in both cores, and it was likely associated with a tsunami event. In addition, grain-size variabilitymay be associated with changes over time caused by the synchronous Asian monsoon and may also be correlatedwith climate records retrieved from the ice cores from Greenland; thus, this variability could indicate pervasiveglobal climatic teleconnections. The overall temporal patterns of the isolated coral branches and shells from thesediment sequences are well correlated with the high sea surface temperatures in the western tropical Pacific.We suggest that increasing sea surface temperatures in the future may lead to more intense storm activity in thewestern Pacific warm pool as the planet warms.

1. Introduction

The frequent occurrence of strong storm surges in coastal regionsrepresents a marine disaster commonly associated with low-pressureweather systems, such as tropical cyclones (e.g., typhoons and hurri-canes) and strong extra-tropical cyclones (cold currents). These surgesfrequently have a catastrophic impact on human lives, the globaleconomy (Blake et al., 2011; Needham et al., 2015), and increasinglyvulnerable ecosystems (Yu et al., 2004, 2009, 2012; Scheffers et al.,2009; Zhao et al., 2009; Sun et al., 2013). The observational data de-rived from the western Pacific over the past seven decades (CE1945–2018) demonstrate that 2161 storms originated from this regionat an average of ~29 per year (Table 1), which suggests a high fre-quency and intensity of storm occurrence in this region. Moreover, 672major storms (Joint Typhoon Warning Centre (JTWC) Category 3–5)

were observed, with an average number of ~nine per year. Such highfrequent storm activity attracts public attention and has led to intensedebates. Can such storm activity be predicted and does the storm ac-tivity present any trends? Furthermore, what are the forcing factors andwill global warming cause an increase in storms? Therefore, a scientificinvestigation of historical strong storms (including hurricanes, ty-phoons, and tropical cyclones) and their impact on humans must beperformed (Williams, 2013).

Reconstructing the long-term variability in the frequency andmagnitude of strong storms will help us understand regional cata-strophic events induced by internal and external mechanisms and cli-mate change and assess the contribution of anthropogenic forcing toglobal warming since the Industrial Revolution (Yu et al., 2009).However, few continuous and high-resolution records are available forthe western Pacific region, which limits a better understanding of the

https://doi.org/10.1016/j.palaeo.2019.02.009Received 10 November 2018; Received in revised form 30 January 2019; Accepted 14 February 2019

⁎ Corresponding author.E-mail address: kefuyu@scsio.ac.cn (K. Yu).

Palaeogeography, Palaeoclimatology, Palaeoecology 521 (2019) 57–71

Available online 18 February 20190031-0182/ © 2019 Elsevier B.V. All rights reserved.

T

long-term trends in the frequency and intensity of strong storm activity.The observational records (~166 years, as far back as CE 1851) are toofew and insufficiently reliable to reveal the trends in the frequency oftropical cyclone activity (Landsea et al., 2006), although these data areof great value for simulating the frequency and intensity of storm ac-tivity (Donnelly and Woodruff, 2007; Liu, 2007). Moreover, long his-torical documentary records of such events are too short (usually<1000 years) and sparse, and they are limited to only a few countrieswith older civilizations, such as China (Liu et al., 2001; Louie and Liu,2003). In addition, geological evidence derived from coastal lakes, la-goons, swamps or marshes may provide new potential insights into thelink between climatic conditions and storm activity over long time-scales as well as the progress and mechanisms of storms (Liu and Fearn,1993, 2000; Nott and Hayne, 2001, Noren et al., 2002; Nott, 2004; Yuet al., 2004, 2006, 2009, 2012; Lu and Liu, 2005; Donnelly andWoodruff, 2007; Woodruff et al., 2009, 2015; Lane et al., 2011; Chenet al., 2012; Williams, 2013; May et al., 2017). Nevertheless, this typeof research has primarily been performed in the Gulf of Mexico, Car-ibbean Sea, North Atlantic Ocean, western Northwest Pacific, e.g.China, and Japan, and areas surrounding Australia. In comparison tothese storm reconstructions from the landfall region, fewer numbers ofstudies have been performed in the western tropical Pacific, e.g. thearea of origin of storm.

Therefore, systematic high-resolution reconstructions of such stormevents and their impact on humans (Williams, 2013) must be performedusing natural long timescale archives with precise dating (Yu et al.,2006, 2009). Such investigations are vital for understanding the long-term variability of tropical cyclone activity and its possible global tel-econnections (Liu et al., 2001), and they are imperative for improvedpreparedness for future events (Leroy and Niemi, 2009). Such effortswill allow more accurate risk assessments of the future impacts of theseevents on coastal communities (Nott, 2004), and they are timely be-cause of the controversy on the nature global warming's impacts on thefrequency and magnitude of hurricanes worldwide (Leroy and Niemi,2009).

Tropical coral reef lagoons in the northern South China Sea (SCS) ofthe western Pacific are predominated by unconsolidated deposits ofbroken coral skeletal materials, such as storm-induced fragile coralbranches around the reef lagoon (Yu et al., 2004). Reports have in-dicated that high sedimentation rates (Yu et al., 2006, 2009) mayprovide a potential long-term, high-resolution, and continuous recordof environmental change (Yu et al., 2006, 2009; Donnelly andWoodruff, 2007). Thus, in this study, we present two neighbouringcores from a coral reef lagoon that span the past ~3500 years and offeran extraordinary opportunity to provide a new record for high-resolu-tion tropical cyclone-generated storm reconstructions of the Xisha Is-lands in the northern SCS. Moreover, such reconstructions could pro-vide insights into storm trends and the underlying mechanisms.

Yu et al. (2006, 2009) performed storm reconstructions based ongrain size in the Nansha Islands in the southern SCS over the last4000 years in the last decades. Although Yu et al.'s work and the current

study both focus on lagoon sediment profiles derived from the SCS, thestudy sites were in different locations with distinct regional differences,with Yu et al. (2006) collecting data from a site in the southern tropicsin the SCS, which is located approximately 800 km away from our site.Furthermore, we discuss the significance of the time series for stormreconstruction by comparing other climate proxy records from theNorthern Hemisphere. Consequently, our findings and conclusions arealso somewhat different compared with the previous study.

In addition, different storm-related terminology (e.g., storm, cy-clone, typhoon and hurricane) is used in different parts of the world; forexample, hurricane is used in North America, cyclone is used in SouthAsia and Australia, and typhoon is often used in the SCS. In this study,“storm” is used to represent all of these terms.

2. Regional setting

The SCS is the largest Chinese semi-enclosed marginal sea in thewestern Pacific (Fig. 1A). The Xisha Islands (15°40′-17°10′N, 111°-113°E) are located in the northern SCS in the western Pacific (Fig. 1A)within the northern tropical zone, and they are adjacent to the edge ofthe most northerly latitude of the Intertropical Convergence Zone(ITCZ) migration in the Northern Hemisphere (Fig. 2). This area isconnected to the western Pacific warm pool, which is a key pathway ofthe Asian monsoon in summer and represents the major moisturesource for Asian monsoon (AM) precipitation and typhoons frequentlyoriginate from this area. Instrumental data from the Xisha Observatoryindicates that Xisha Islands is a weak tide sea area, e.g., the mean tidalrange is approximately 90 cm in this region, while the average annualsea wave height is approximately 140 cm. According to the observa-tional data derived from the western Pacific for the past seven decades(CE 1945–2018), > 1/5 of the storms recorded in the West Pacificstrike originate in the vicinity of 10–20°N, 110–115°E (based on thedata from http://weather.unisys.com/hurricane/index.php). In addi-tion, detailed instrumental data from the Xisha Observatory for theperiod of CE 1958–2001 show that the mean annual temperature (MAT)is 26.6 °C (Fig. 2) and that the sea surface temperature (SST) is rela-tively high and presents a monthly average value above 29 °C andfluctuates within the range of 29.8–30.2 °C. The mean annual humidityis ~81.6%, the mean annual precipitation (MAP) is approximately1500mm, and most of the precipitation occurs from June to November(Fig. 2). The mean surface salinity in this area is ~33.5‰.

3. Materials and methods

3.1. Materials

The Lingyang Reef (16°28′N, 111°35′E) is a small coral atoll that islocated to the southwest of the Yongle Atoll (Fig. 1B & C) off the XishaIslands, and it is ~300 km southeast of Hainan Island and ~100 kmsouthwest of Yongxing Island in the SCS. This reef is an open spindle-shaped atoll that is approximately 6 km long in the NW-SE direction

Table 1Storm summary in the western Pacific over the past 74 years (CE 1945–2018) (calculated based the data listed at http://weather.unisys.com/hurricane/index.php).

Period Storm sum Unknown storm Tropical depression Subtropical storm Tropical storm Typhoon Super typhoon JTWC category 1–5 JTWC category 3–5

1945–1954 221 0 0 0 67 134 20 153 851955–1964 292 35 0 0 59 134 64 198 1191965–1974 329 32 0 9 108 137 43 176 971975–1984 270 0 0 27 97 121 25 146 711985–1994 311 0 0 24 96 151 40 191 941995–2004 350 0 78 8 96 110 58 168 932005–2014 270 0 32 5 99 83 51 134 812015–2018 118 0 24 0 45 32 17 49 32Sum 2161 67 134 73 667 902 318 1215 672Average 29 1 2 1 9 12 4 16 9

Y. Yue, et al. Palaeogeography, Palaeoclimatology, Palaeoecology 521 (2019) 57–71

58

and 3.5 km wide in the W-E direction, and it covers an area of ap-proximately 21 km2 (Fig. 1C). A closed lagoon (~5 km in length and~3 km in width) is situated in the centre of Lingyang Reef (Fig. 1C). Theshallow area on the reef flat and even the upper region of the slope isthe principal zone of the storm effect and represents the area where thereef will be exposed under low tidal levels and suffer severe damagefrom wave and current energy.

In 2013, two neighbouring cores (LYJ2 and LYJ3) at a distance of150m apart (Fig. 1C) were collected completely without any loss fromthe closed lagoon of Lingyang Reef using the approach described by Yuet al. (2006). The LYJ2 (16°27′59.32″N, 111°36′09.62″E, 150 cm belowrelative sea level) and LYJ3 (16°27′07.44″N, 111°36′11.11″E, 450 cmbelow relative sea level) sediment cores were 287 cm and 438 cm inlength, and they were retrieved from the inner reef flat and lagoonslope, respectively. Core LYJ2 is 150 cm below relative sea level, andthis area is vulnerable and likely exposed under storm-induced seawave even low tide. Considering the influence of the sea wave or tideand the potential exposure, an overview of the stratigraphy andchronology from LYJ2 is thus presented for comparison with core LYJ3.

3.2. Grain size

Combined lagoon sediment and transported coral block recordshave been used for storm reconstruction (Yu et al., 2009). In this paper,samples for the grain-size analysis were collected at 0.5 cm intervalsfrom the two LYJ (LYJ2 and LYJ3) profiles, and a total of 1450 sedi-ment samples were analysed. All samples were first weighed and thenoven dried for 48 h at 45 °C to achieve a constant weight and dry se-diment weight record. The dried samples were manually sieved through2mm and 1mm meshes, and then three fractions (> 2mm, 1–2mm,and<1mm) were reweighed. The>1mm grain-size fraction herebywas referred to as “coarse-grained fraction” in this paper. In addition,the isolated or mixed and disorderly coral branch or large shell in the

samples was separated, counted and weighed. Consequently, the drymass content of the different grain sizes can be calculated. According tothe lithology variations and grain-size distribution, different deposi-tional units for the LYJ cores can be recognized.

3.3. Sediment chronologies

To establish the chronology of the sedimentary sequences and ob-tain a better understanding of the coral reef lagoon sequences, 19 in-situ foraminifera samples from the two parallel cores. 8 samples fromLYJ2 and 11 samples from LYJ3, were selected for AMS 14C analyses(Table 2). The sediment samples were selected from the profile at10–60 cm intervals, and> 120 foraminifera (Calcarina sp.) with well-preserved fine thorns were handpicked under the stereomicroscope.The selected foraminifera samples were washed using an ultrasonicvibrator, oven dried and sent to the Beta Laboratory (USA) for AMS 14Cdating in October 2015 and December 2016. Radiocarbon ages of thetwo cores were calculated with a half-life of 5568 years. All AMS 14Cdates were reported with the 2-sigma (95.4%) probability range andconverted to calendar years before present (cal yr BP relative to CE1950) after calibration using the programme CALIB REV7.0.1 (Stuiverand Reimer, 1993) with the calibration dataset MARINE13 (Reimeret al., 2013), and also adjusted for local reservoir correction (Delta±R). A Delta±R correction is applied to the sample that has alreadybeen corrected with the global marine reservoir correction. Here, theDelta±R value is 89 ± 59 that was calculated by Yu et al. (2010).

Corals were often chosen for thermal ionization mass spectrometric(TIMS) U-series dating which indicated the absolute age of coral sam-ples, e.g., the ages of these storm-relocated coral blocks (or branches),should approximate the times when the storms occurred (Yu et al.,2004). In this paper, measurement of TIMS U-series dating was used asa supplementary method for examining the reliability of AMS 14C. Thus,a reloaded fresh coral branch (16.6 g) with sharp corners was collected

Fig. 1. Location of the LYJ lagoon cores (LYJ2 and LYJ3) from Lingyang Reef off the Xisha Islands in the South China Sea.

Y. Yue, et al. Palaeogeography, Palaeoclimatology, Palaeoecology 521 (2019) 57–71

59

at 53 cm from LYJ3, and sent to Advanced Centre for QueenslandUniversity Isotope Research Excellence (ACQUIRE), the University ofQueensland for TIMS U-series dating (Table 3).

Furthermore, eight samples collected from the upper 35 cm of LYJ3were conducted for 210Pb and 137Cs dating in Guangxi University(China) (as seen in Table 4) in order to establish the modern sedimentchronologies. For radionuclides analysis, approximate 8.0 g of pow-dered sediment samples were placed in 7ml plastic tube and counted bya well-type High Purity Gamma Spectrometry (Ortec) for 72–120 h. Theactivities of 210Pb, 226Ra, and 137Cs were computed spectroscopicallyfrom the photopeaks of 46.5 keV, 609.3 keV, and 661.7 keV. The IrishSea marine sediment provided by International Atomic Energy Agency(IAEA-385) was used as the standard material to calculated the 210Pb,226Ra, and 137Cs activities.

In this paper, we developed an age model by assuming that thesediment-water interface is CE 2013 when the cores were collected. Thechronological sequences for the two LYJ cores were established basedon the mean sedimentation rate between two adjacent calibrated datesvia linear interpolation, and the age-depth curve was based on the meanage values [(agemax+agemin)/2].

4. Results

4.1. Lithology and grain-size distribution

The two neighbouring lagoon sediments were primarily grey-whiteunconsolidated skeletal materials composed predominantly of coralsand, coral branches, and shells (Figs. 3 and 4). A total number of 380isolated or mixed and disorderly coral branches or shells (164 in LYJ2and 216 in LYJ3, respectively) were counted in the LYJ cores (DatasetGrainsize), which distributed throughout the whole sediment sequenceand led to sharp increases in the coarse-grained fraction (Figs. 3 and 4).The sediment grain-size variations in both cores generally followchanges in the lithology. According to the variations in sediment grainsize, two distinct depositional units for LYJ2 and three distinct de-positional units for LYJ3 were recognized (Figs. 3 and 4). Here, wesummarized the lithology and grain-size distribution for the two coresin order of depth from bottom to top.

4.1.1. LYJ2The whole deposition sequence in the LYJ2 core was primarily

dominated by the fine-grained fraction (< 1mm), which accounted fora high mean value of 67.56%, although it displayed a progressive de-clining trend as shown in Fig. 3. However, the coarse-grained fraction

Fig. 2. Mean monthly temperature, precipitation and humidity of the Xisha Islands (data from Xisha Observatory during CE 1958–2001) and the mean latitude of theIntertropical Convergence Zone (ITCZ) migration in the South China Sea from CE 1991–2001.

Y. Yue, et al. Palaeogeography, Palaeoclimatology, Palaeoecology 521 (2019) 57–71

60

Table2

Rad

iocarbon

dating

resultsof

benthicforaminiferaCalcarina

sp.from

theco

ralreeflago

onprofi

lesof

LYJ2

andLY

J3retrieve

dfrom

Ling

yang

Reefin

theSC

S.Allda

teswerecalib

ratedto

calend

arye

arsbe

fore

presen

t(cal

yrBP

relative

to19

50CE)

usingCALIB

REV

7.0(Stuiver

andReimer,19

93)withthecalib

ration

datasetMARIN

E13(R

eimer

etal.,20

13),

andallag

eswereco

rrectedforalocalmarinereservoireff

ectwith

Delta

±R=

89±

59(Y

uet

al.,20

10).

Coresite

Samplenu

mbe

rDep

th(cm)

Labnu

mbe

rDated

material

δ13C(‰

)Con

ventiona

lag

e(yr

BP)

Adjustedforlocalreservoirco

rrection

(Delta

±R=

89±

59)(yrBP

)Calibratedag

erang

e(2σ,

95.4%,

calyr

BP)

Med

ianprob

ability

(cal

yrBP

)

LYJ2

LYJ2

–50

2544

7,16

1Calcarina

sp.

-189

0±

3087

0±

5055

1–45

450

3LY

J2–1

1055

447,16

2Calcarina

sp.

−0.6

1330

±30

1310

±50

941–

785

863

LYJ2

–180

9044

7,16

3Calcarina

sp.

−0.4

2340

±30

2320

±50

2058

–186

619

62LY

J2–3

0015

044

7,16

4Calcarina

sp.

0.5

2770

±30

2750

±50

2652

–236

625

09LY

J2–4

0020

044

7,16

5Calcarina

sp.

0.1

2900

±30

2880

±50

2750

–256

926

60LY

J2–4

5022

545

0,79

5Calcarina

sp.

0.3

2920

±30

2900

±50

2766

–264

027

03LY

J2–4

8024

044

7,16

6Calcarina

sp.

−0.3

2810

±30

2790

±50

2690

–244

025

65LY

J2–5

5027

545

0,79

6Calcarina

sp.

−0.8

2890

±30

2870

±50

2743

–255

926

51LY

J3LY

J3–1

0050

418,15

9Calcarina

sp.

−0.1

1610

±30

1590

±50

1256

–108

411

70LY

J3–2

0010

041

8,16

0Calcarina

sp.

0.2

2440

±30

2420

±50

2183

–197

120

77LY

J3–3

0015

041

8,16

1Calcarina

sp.

−0.3

2770

±30

2750

±50

2652

–236

625

09LY

J3–4

0020

041

8,16

2Calcarina

sp.

0.2

3020

±30

3000

±50

2862

– 272

527

94LY

J3–5

0025

041

8,16

3Calcarina

sp.

−0.3

2940

±30

2920

±50

2783

–266

527

24LY

J3–6

0030

041

8,16

4Calcarina

sp.

−0.1

3020

±30

3000

±50

2862

–272

527

94LY

J3–6

4132

0.5

429,32

3Calcarina

sp.

0.3

3020

±30

3010

±50

2862

–272

527

94LY

J3–6

7733

8.5

418,16

6Calcarina

sp.

030

80±

3030

60±

5029

37–2

758

2848

LYJ3

–697

348.5

429,32

2Calcarina

sp.

0.3

3200

±30

3180

±50

3121

–289

030

06LY

J3–7

7738

8.5

418,16

5Calcarina

sp.

0.1

3600

±30

3580

±50

3581

–340

034

91LY

J3–8

5042

545

0,79

7Calcarina

sp.

0.5

3560

±30

3540

±50

3544

–336

634

55

Table3

TIMSisotop

icda

taan

dU-seriesag

esforaco

ralbran

chwithsharpco

rnersfrom

LYJ3

,Lingy

angReef,Xisha

Island

s,in

theSC

S.Th

eTIMSU-seriesag

es(yrBP

)arerelative

to20

16CE.

Samplena

me

Dep

th(cm)

Sampleweigh

t(g)

U(ppm

)±

2s

232T

h(ppb

)±

2s

(230

Th/2

32Th

)±

2s

(230Th

/238U)

±2s

(234U/2

38U)

±2s

Unc

orrected

age(yrBP

)Corrected

age(yrBP

)Yearin

AD

LYJ3

–53

530.14

948

3.06

210.00

120.21

40.00

154

5.8

30.01

259

0.00

006

1.14

580.00

112

04±

612

02±

681

4±

6

Y. Yue, et al. Palaeogeography, Palaeoclimatology, Palaeoecology 521 (2019) 57–71

61

(> 2mm) that marked by 164 isolated or mixed and disorderly coralbranches or shells, showed a gradual increasing trend and a mean valueof 19.31%, and the 1–2mm fraction was relatively low with a meanvalue of 13.13% (Fig. 3). Overall, the 287 cm sediment profile of LYJ2was roughly divided into two depositional units according to the li-thology and grain size as shown in Fig. 3 and Dataset Grainsize.

Unit 2 (287–93 cm) is the basal deposit of the LYJ2 core, and it wasmainly composed of 194 cm white fine-grained coral sand (< 1mm,74.96% in average) and included coral branches and shells. The per-centage of the 1–2mm fraction varied between 7.07% and 20.4% andpresented a low value of 12.57%. Although the content of the> 2mmfraction had a low mean value of 12.47%, it presented frequent fluc-tuations in the range of 0.99% to 72.75% due to the occurrences of 71coral branches or shells.

Compared with Unit 2, Unit 1 (93–0 cm) was primarily pre-dominated by a long layer of white coarse-grained coral sand withabundant coral branches and shells for the topmost 93 cm. 93 isolatedor mixed and disorderly coral branches or shells were observed in thisunit. The>2mm fraction (33.69%) presented a distinct increase thatfluctuated within the range of 12.08–75.64% and reached a peak(75.64%) within the core at 51.5 cm because of a 42.69 g coral branch(with a size of 2 cm*6.5 cm) (Fig. 3). Compared with the other fractions,the 1–2mm fraction (14.31%) presented a stable level with less change.

4.1.2. LYJ3LYJ3 was primarily composed of rapidly accumulated coral sand

intercalated with coarse-grained deposits, such as coral branch andisolated shell, as shown in Fig. 4 and Dataset Grainsize. The basal de-posit (438–358 cm) of core LYJ3 was mainly composed of 180 cm ofwhite fine-grained coral sand with frequent occurrences of coral bran-ches. From 358 to 86 cm, the core was also dominated by white fine-grained coral sand with sporadic coral branch and shell. The followingsection from 86 to 0 cm presented a distinct lithological change markedby white coarse-grained coral sand with abundant coral branches andshells compared with the sections below.

The whole deposition sequence of the LYJ3 core was predominantlycomposed of the fine-grained fraction (< 1mm), which had a highmean of 74.08% and presented frequent fluctuations in the range of14.29% to 98.61%, although it displayed a progressive declining trendas shown in Fig. 4. The coarse-grained fraction (> 1mm), which in-cluded the sum of the fractions> 2mm (15.56%) and 1–2mm(10.36%), averaged 25.92% and showed a gradual increasing tendency(Fig. 4). In addition, almost half of the samples (402 samples) had ahigh content of coarse-grained sand at more than the average of 25.92%for the whole sediment sequence. Dramatic fluctuations were observedbecause of the frequent occurrence of isolated coral branches and shells(216 samples), which were mainly distributed in the lower(438–358 cm) and upper (86–10 cm) sections (Fig. 4). The percentageof the>2mm fraction had a mean value of 15.56% and ranged from0.09% to 85.45%. The percentage of the 1–2mm fraction varied be-tween 0.26% and 25.60%, with a mean value of 10.36%. Overall, ac-cording to the lithology, grain size and the abundance of coral branches

Table 4210Pb, 226Ra and 137Cs activities in the core of LYJ3 (Unit: Bq/kg).

Samplenumber

Depth (cm) 210Pbactivity

Uncertainty 226Ra Uncertainty 137Csactivity

LYJ3–6 3 6.94 1.37 3.41 0.24 <MDALYJ3–14 7 10.11 1.40 4.25 0.26 <MDALYJ3–18 9 6.60 1.46 4.99 0.33 <MDALYJ3–22 11 10.31 2.28 4.25 0.41 <MDALYJ3–26 13 6.78 1.72 6.96 0.40 <MDALYJ3–30 15 3.00 2.09 2.81 0.26 <MDALYJ3–40 20 7.93 1.41 3.20 0.21 <MDALYJ3–70 35 4.11 1.38 3.26 0.21 <MDA

Fig.

3.Litholog

ical

descriptionan

dsedimen

tgrain-size

variationof

core

LYJ2

from

theLing

yang

Reefin

theno

rthe

rnSo

uthChina

Sea.

Y. Yue, et al. Palaeogeography, Palaeoclimatology, Palaeoecology 521 (2019) 57–71

62

Fig.

4.Litholog

ical

descriptionan

dsedimen

tgrain-size

variationof

core

LYJ3

from

theLing

yang

Reefin

theno

rthe

rnSo

uthChina

Sea.

Y. Yue, et al. Palaeogeography, Palaeoclimatology, Palaeoecology 521 (2019) 57–71

63

and sporadic shells, three depositional units can be recognized (Fig. 4).The sediment grain size variations are described as follows.

Unit 3 (438–358 cm) had high percentages of the< 1mm fraction(85.99% average value), which included fine-grained coral sand withfrequent isolated broken coral branches. The 1–2mm fraction had arelatively lower percentage (3.02%) and did not present distinct var-iations. The content of the coarse-grained fraction (> 2mm) accountedfor 11% and exhibited a strong fluctuation within the range of0.09–85.45%, which produced many sharp peaks that were mainlyattributed to the frequent presence of isolated coral branches or shellsin certain samples. 48 isolated coral branches or shells with size of>0.5 cm*0.5 cm were observed in this interval. For example, two iso-lated 408 and 405 cm coral branches weighed 77.5 g (size of5 cm*6 cm) and 48.2 g (size of 2.5 cm*5 cm), respectively, and a 380 cmshell was 26.2 g (size of 2.5 cm*10 cm) (Fig. 4).

Unit 2 (358–86 cm) mainly consisted of fine-grained coral sand ofthe<1mm fraction (77.02%), and this sand presented a rather highsedimentation rate. In contrast, the> 2mm fraction reached its lowestvalue (10.36%) in the profile, especial for the interval from 358 to223 cm (3.05%), and it presented with less obvious fluctuations. Thecontent of the 1–2mm fraction was characterized by an increasingtendency up to 12.62% on average and reached its profile peak (25.6%)at 209.5 cm.

Unit 1 (86–0 cm) was represented by a different lithology from theprevious units, and it was marked by a distinct increase of coarse-grained coral sand, which presented the highest content (38.8%) in thewhole sequence except for the upper 11 cm which was disturbed duringthe drilling and it was therefore not used in this paper. All the samples(except for the upper 11 cm) in this section presented coarse-grainedfraction (> 1mm) contents that were more than the average value of25.92%. The> 2mm fraction abruptly increased to its profile peak andpresented abundant coral branches and shells (108 in total). However,the<1mm (50.86%) and 1–2mm (10.06%) fraction showed a de-creasing trend towards the top (Fig. 4).

4.2. Chronology

Modern sediment chronologies were obtained for surface cores bygamma spectrometry. However, the results showed that the activities of210Pb (~6.97 Bq/kg) and 137Cs (< 0.1 Bq/kg, Minimum DetectionActivity, MDA) from the calcium carbonate-dominated sediment in LYJcores (as seen in Table 4). It was extremely low in comparison with thatin other sediments (~100 Bq/kg and 1–10 Bq/kg for 210Pb and 137Cs)from northern South China Sea, or East China Sea and continental shelf(Qiao et al., 2017; Yang et al., 2018; Lin et al., 2018 (in Press)). The low137Cs activity was attributed to low global fallout of artificial radio-nuclides from nuclear weapon testing in the low latitude regions re-lative to the middle latitude regions and no record of local input ofartificial radionuclides from nuclear weapon testing in the SCS, whilethe low 210Pb activity was determined by mineral component (carbo-nate). The high specific area of small particle and strong surface ad-sorption of clay could result in high 210Pb in the mud regions of con-tinental shelf (Qiao et al., 2017). By contrast, large particle andcarbonate component of coral sand sediments would significantly de-press the 210Pb activity in our samples (< 10% of 210Pb activity in otherstudy). Therefore, it suggests that the coral sand sediments collectedfrom the enclosed LYJ coral reef lagoon in this region are not suitablefor 210Pb and 137Cs dating.

The AMS 14C results for LYJ2 and LYJ3 are listed in Table 2.Eight samples from LYJ2 were selected for the AMS 14C analyses

(Table 2, Fig. 5). Four samples from 275 cm, 240 cm, 225 cm, and200 cm, indicating the base and middle of the LYJ2 sediment sequencewere selected for dating. The age ranges for these analyses were closeand overlapped at a median of ~2650 cal yr BP suggesting that thesediment appeared to accumulate rapidly within this period. The resultsfor the sample collected from 150 cm indicate an age of approximately

2652–2366 cal yr BP. The age range for the sample collected at 90 cmfrom the middle of the deposit was 2058–1866 cal yr BP. The upperdeposits were accurately dated using two samples. The age range for thesample collected at 55 cm was 941–785 cal yr BP, and the sample col-lected at 25 cm from the uppermost deposits was deposited almost fourhundred years later at 551–454 cal yr BP.

A total of 11 samples were selected for AMS 14C dating from theLYJ3 sediment sequence (Table 2, Fig. 5). Two samples dating the lowerpart of the core were collected from 425 cm and 388.5 cm, and theywere close in age at ~3500 cal yr BP. The date for the sample from348.5 cm was almost 500 years later (3121–2890 cal yr BP). Five sam-ples from the middle part of the profile collected from 338.5 cm,320.5 cm, 300 cm, 250 cm and 200 cm were dated and produced agenerally close age that overlapped at a median value of ~2860 cal yrBP. The combined age range for the 138.5 cm sequence presented clo-sely overlapping ages, which suggests a rather high sedimentation rate,and this was also observed in the neighbouring LYJ2. Three sampleswere selected for the upper dating. The sample collected at 150 cm,yielded a date of 2652–2366 cal yr BP. The results for the sample from100 cm were within the range of 2183–1971. The uppermost samplecollected from 50 cm indicated an age of ~1256–1084 cal yr BP (~CE694–866, with a mean age of CE 780).

The sample of coral branch collected from 53 cm of LYJ3 core forTIMS U-series dating, yielded a date of CE 814 ± 6 (relative to CE2016, uncalibrated age is 1204 ± 6 yr BP, as seen in Table 3 andFig. 5). Results for the two adjacent samples from the upper part ofLYJ3 core (53 cm and 50 cm, respectively) are identical and theyoverlap at around 1200 yr BP (~CE 800) within analytical un-certainties, indicating our AMS 14C data are credible.

The uncalibrated AMS 14C ages for the foraminifera (Calcarina sp.)retrieved from the two sediment cores varied from 890 ± 30 to3600 ± 30 yr BP. All samples were interpolated based on the estab-lished AMS 14C chronology. Based on the relationship between age anddepth, the basal ages of the LYJ2 and LYJ3 cores were determined toextend to ~2700 and ~3500 cal yr BP, respectively. The age-depth plotshowed that the estimated average sedimentation rates of the LYJ2 andLYJ3 profiles were approximately 1.35mm a−1 and 1.27mm a−1 re-lative to the drilling depth, respectively, whereas the AMS 14C resultsrevealed an uneven sedimentation rate with differences between thelower and upper sections of the profile. Moreover, the foraminiferaAMS 14C dates from the two cores were almost in the correct strati-graphic order, and the sedimentation rates for both LYJ cores wererelatively lower in the upper part (~0.3mm a−1) and increased in thelower portions of the record (as high as ~8.8 mm a−1) as shown inFig. 5. Each 0.5 cm grain-size sample therefore integrates ~1–16 yearsof lagoon accumulation.

5. Discussion

5.1. Indicator of storm events

Interpreting proxy records is always a major issue for paleoclimateand paleo-environment studies (Yao et al., 1997; Chen et al., 2004).Sediment grain-size analyses often provide direct information onchanges in the transporting mechanism and sedimentary environment(Chen et al., 2004; Yu et al., 2009; Yan et al., 2011; Yue et al., 2015; Liuet al., 2016). Coral reef lagoons represent the main accumulation areafor bioclastics and protozoan sediments, which are often deposited withthe rubble of coral skeletons, shells, coral sand, and other lithic clasts.In general, lagoon hydrodynamics gradually weaken from the edgetowards the centre, and the sediment changes from coarse to fine be-cause of the coral reef geomorphology and depth (Yu et al., 2004).Because storms are brief, during and immediately after storm events,strong storm-induced wave and current energy quickly spreads alongthe coral reef. Consequently, a large number of coral branches as well assand and gravel primarily from the reef flat and the reef front slope

Y. Yue, et al. Palaeogeography, Palaeoclimatology, Palaeoecology 521 (2019) 57–71

64

living coral zones are directly transported into the lagoon and chaoti-cally deposited. Of particular interest are isolated or mixed and dis-orderly coral branches. Based on our field survey in 2013, there was fewlive coral in the enclosed Lingyang Reef lagoon, and instrumental dataindicated that this area was a weak tide sea area. From an interpretiveperspective, we infer that these broken coral branches or shell frag-ments derived from the enclosed coral lagoon should correspond tostrong storm-induced wave and current energy. In general, coralbranches with rounded corners suggest repeated transport, whereaslarge fresh branches with sharp corners indicate rapid transport anddeposition without obvious abrasion by intense storm-generated wavesand tides. Therefore, in this context, the occurrence of isolated coralbranch (fresh or not) from the enclosed lagoon cores must have beentransported by storms from the living sites, can serve as indicators ofstrong storm events (even tsunami event), and the content of coarse-grained fraction suggest a strong hydrodynamic environment. For thisreason, the ages of these storm-relocated coral branches, e.g., the freshcoral branch with sharp corners, should approximate the times whenthe storms occurred, and the size/weight of a single coral branch isassociated with the intensity of the storm. By this assumption, thefrequent occurrences of coral branches have been successfully used forthe reconstruction of long-term storm history (Yu et al., 2004, 2006,2009, 2012; Zhao et al., 2009).

The Lingyang Reef is far from the mainland; therefore, it is rela-tively less artificially disturbed and likely provides potential evidencefor use in storm reconstructions for this region. Broken coral branchesfrom the LYJ cores were transported from the vicinity (e.g., the reef-front living coral zone, where it is optimum habitat for coral growth) tothe lagoon by storm-induced water currents and thus led to increases ofthe coarse-grained sediments. However, we noticed that simultaneousdeposits from the two LYJ cores showed somewhat different char-acteristics and presented visible variations marked by the abundance ofisolated coral branches or shells in the Unit 2 from both profiles (Fig. 6).As shown in Fig. 6, LYJ2 was collected from the inner reef flat at adepth of 1.5 m below sea level, and coral branches were frequentlyobserved throughout the entire profile, whereas LYJ3 was collectedfrom the lagoon slope at a depth of 4.5 m below sea level, coral bran-ches frequently occurred in the upper and lower parts. As previouslymentioned, changes in the spatial distribution of the cores could

indicate the distribution of grain sizes across the lagoon. Therefore,compared with LYJ3, LYJ2 was found to be much more easily impactedby the dynamic effects of waves and tides. Therefore, the coarse-grainedsediments from core LYJ3 should inspire more confidence for recordingstorm events and the natural deposition of this region.

5.2. Evolution of the Lingyang Reef lagoon depositional environment andstorm activity since the late Holocene

Based on the sedimentary chronological sequence, the grain-sizevariations of the two LYJ profiles were established as shown in Fig. 7. Itshows that the basal deposits of the LYJ3 core roughly spanned from~3500 to 3100 cal yr BP, and they were marked by dramatic fluctua-tions of the coarse fraction (> 1mm) because of the frequent occur-rence of storm-induced deposition, e.g., isolated coral branch or shell(Fig. 7). In fact, a total of 49 isolated coral branches or shells can beidentified within this period at an average recurrence rate of ~8 years.Such heavy coral branches, such as the isolated coral branches at 77.5 gand 48.2 g from 408 cm and 406 cm respectively dated to ~3350 cal yrBP (~1400 BCE, Figs. 6 and 7), indicate a strong hydrodynamic en-vironment associated with the frequent occurrence of strong stormsduring this period. Historically, earthquake-generated tsunamis havegenerated higher-magnitude coastal flooding events than storm surgesin Japan (Needham et al., 2015). For example, Woodruff et al. (2009)provided a 6400-year record of episodic coastal flooding using sedimentdeposits from two coastal lakes located on the remote island of Kami-koshiki in southwestern Japan for exploring typhoon variability, whichoffered a good opportunity for comparison between our storm re-constructions. This frequent storm scenery observed in our LYJ corescan be also recorded in south-western Japan that associated with morefrequent marine-sourced deposition (Woodruff et al., 2009), suggestinga period of more storm activity in both western Northwest Pacific andwestern Pacific. However, this frequent storm scenario was not ob-served in the NY4 lagoon core from the Yongshu Reef from the lowerlatitude region in the Nansha Islands in the southern SCS, except for theextremely stormy periods centering around ~1200 BCE (~3200 cal yrBP) (Yu et al., 2009). This phenomenon was likely because of thenorthward shift of the zonal storm track, the regional difference be-tween the south and north SCS and the location of NY4, which was too

Fig. 5. Age-depth model for LYJ profiles. The foraminifera AMS 14C dates from the two cores were almost in the correct stratigraphic order. Results for the twosamples from the upper part of LYJ3 core (U-series age from 53 cm and AMS 14C dating from 50 cm, respectively) are almost similar, indicating our AMS 14C data arecredible.

Y. Yue, et al. Palaeogeography, Palaeoclimatology, Palaeoecology 521 (2019) 57–71

65

far away at ~800 km southward. Therefore, the frequency and intensityof storm activity during this period in the north SCS was likely muchhigher than those in the south SCS.

During the period from ~3100 to 1800 cal yr BP, the high content ofthe fine fraction (< 1mm) from the LYJ3 profile appeared to suggest astable and quiet depositional environment with an ~1300-year gap instrong tropical storm activity. However, the AMS 14C data indicated arather high sedimentation rate throughout the entire profile, especiallyduring the period from ~2800 to 2600 cal yr BP which correspondingto Unit 2 of LYJ3 (Figs. 5 & 6). This phenomenon was also observed inthe basal part of the neighbouring LYJ2 core sediment profile (Unit 2),but it was marked by the frequent occurrence of mixed and disorderedcoral branches. An interpretation for the different sediment componentbetween LYJ2 and LYJ3 during this period could be as result of thechange in geomorphology of the reef, even changes in the spatial dis-tribution of the cores could indicate the distribution of grain sizes

across the lagoon as above-mentioned. For instance, the reef ridge beingraised or the expansion of shallow reef-flat areas by its natural growththat greatly slowed down the storm-induced seawater velocity towardsthe centre of lagoon and therefore tend to reduce the size of the re-sultant coarse-fraction content peak. There is evidence that the ex-pansion of shallow reef flat may reduce the capacity for giant waves topenetrate into the lagoon sediment region and therefore limit the fre-quency and magnitude of coarse-fraction content peaks (Yu et al.,2009). Furthermore, 71 isolated coral branches or shells were foundwithin this period from the lower part (Unit 2) of LYJ2 core. Previousstudy confirmed that the frequent occurrence of coarse-grained fractionmay be partially related to a high deposition rate and undoubtedlytransported by extreme wave events such as strong storms or tsunamis(Yu et al., 2009). Such episodes of accelerated deposition were wellcorrelated with the strong storm events that were identified by U-seriesdates of storm-transported coral blocks in the Nansha Islands (Yu et al.,

Fig. 6. Schematic cross section of the Lingyang Reef (A) and lithological comparison of the LYJ3 and LYJ2 cores (B). Based on our systematic field investigations in2013, the following six biogeological-biogeomorphological and sedimentary zones were identified around the Lingyang Reef lagoon: reef front zone, outer reef flatzone, reef ridge zone, inner reef flat zone, lagoon slope, and lagoon basin from the outer to inner zones. The occurrence of isolated or mixed and disordered isolatedcoral branches from the enclosed lagoon cores must have been transported by storms from the living sites, can serve as indicators of strong storm events (eventsunami event).

Y. Yue, et al. Palaeogeography, Palaeoclimatology, Palaeoecology 521 (2019) 57–71

66

2009). Therefore, an alternative interpretation for the results may bethat storm were not absent from 3100 to 1800 cal yr BP. A comparisonbetween the grain-size distributions of the both LYJ cores (Fig. 6) in-dicated that such synchronous episodes of accelerated sedimentationrates and the frequent occurrence of mixed and disordered coralbranches with chaotic age were likely indicative of a storm-induceddeposit even redeposit associated with a tsunami during this period. Forexample, a big and heavy isolated coral branch (14.7 g, with a size of3.5 cm*4 cm in size) from LYJ3 core was observed in approximately2500 cal yr PB (inferred from the closed AMS 14C data) (Figs. 6 and 7),temporally consistent with a record characterized by a 80.1 g coral

branch (3.5 cm*6.5 cm) from LYJ2 core (Fig. 6). It probably indicated astrong hydrodynamic environment which likely associated with astrong storm (even tsunami event). However, due to incomplete ob-servation data, e.g., tsunami observations (National Geophysical DataCentre/World Data Service, 2015), this hypothesis requires further re-search.

Intense and frequent storms can be identified according to the highpercentage and dramatic fluctuations of the coarse fraction over the last1800 years, which were inferred from distinct lithology changes withfrequent occurrences of isolated or mixed and disorderly coral branchesobserved in the upper part of both LYJ cores (Fig. 6). Here, we observed

Fig. 7. Grain-size variation records from the LYJ3 lagoon profile over time and the tropical sea surface temperatures (SSTs) from Core MD81 in the western tropicalPacific (Stott et al., 2004) and δ18O records (‰, VSMOW) from the stalagmite of Dongge Cave (Wang et al., 2005) in the Northern Hemisphere. The isolated or mixedand disorderly coral branches or shells provide information on the spatial and temporal variability of storm (even tsunami) events, and the content of coarse-grainedfraction suggest a strong hydrodynamic environment.

Y. Yue, et al. Palaeogeography, Palaeoclimatology, Palaeoecology 521 (2019) 57–71

67

the transition between depositional units 1 and 2, with AMS datesshowing the change occurred closed to around 1800 cal yr BP (CE 150),was characterized by the frequency of isolated coral branches or shellsthat was exceptionally low between 3000 and 1800 years ago, and itsuddenly jumped to very high at 85 cm and upwards of LYJ3 as seen inFigs. 6 and 7. We interpret this as the reappearance of storm deposits. Incomparison with its lower depositional units of LYJ cores, a plausibleinterpretation for the re-appearance of storm deposits in the upper90 cm of LYJ2 and upper 85 cm of LYJ3 was a result of the reef ridgebeing eroded by a tsunami, which opened a path for the broken coralbranches and allowed them be washed to the two core sites. If this is thecase, the overall sediment grain size (e.g., > 1mm %) should be abruptchanged and the age will be chaotic in the transitional between Unit 1and 2. However, the average value of the coarse-grained content(> 1mm) from LYJ3 showed a progressively upwards increase, in-dicating an enhancing storm-induced hydrodynamic environment overthe last 1800. In addition, the foraminifera AMS 14C dates from the twocores were almost in the correct stratigraphic order. Furthermore, basedon our field surveys in 2013, we did not found any potential trenchessurrounding the core location.

On the other hand, the peak coarse-fraction (> 1mm) contentsgreater than the average value can be used to provide a reliable as-sessment of strong storms and suggest good correlations between thecoarse fraction and storms (Yu et al., 2006, 2009). For example, thecriteria for mean coarse-fraction content peaks> 9.2% were identifiedand considered to be related to strong storms (Yu et al., 2009). Based onthese criteria, the average value of the coarse-grained content (25.92%)from LYJ3 suggest a strong hydrodynamic environment marked by thefrequent occurrence of broken coral branches as well as a progressiveincrease in storm frequency. Such episodes likely indicate a period ofintense storms over the last two millennia.

In fact, some strong storm events were observed from LYJ3 duringthis period can be also identified in other regions in the SCS.Radiochronological dating (AMS 14C and U-series) together with his-torical record allows dating the record of past storm event. For ex-ample, our LYJ data suggested the storm periods (within the interval of54–60 cm from LYJ2 and 40–44 cm from LYJ3, respectively) centeringaround CE 1024 was temporally consistent with the disastrous eventhappened at ~CE 1024 that was reported in Dongdao Island in thisregion (Sun et al., 2013). It was also correlated with in timing withstrong storm/tsunami event dated at CE 1064 ± 30 in YN4 from thesouthern SCS (Yu et al., 2009). Interestingly, the age of the storm re-corded in the three sediment cores roughly coincided with the docu-mented record of disastrous event on the coast of Chaoang in the pro-vince of Guangdong (South China Sea) that characterized by a highwave in CE 1076 (National Geophysical Data Centre/World DataService, 2015). It is therefore reasonable to link the coarse-grainedfraction content to strong storm activity, and suggests that our hy-pothesis is credible: the occurrence of isolated coral branches from theenclosed lagoon cores can serve as indicators of strong storm events(even tsunami event), its size/weight closely ties to the intensity ofstorm, and the content of coarse-grained fraction indicates a stronghydrodynamic environment at this site.

What is more, it was also worthwhile to note the regional differentiaconcerning the historical storm reconstructions. A comparison betweenour findings and the storm records NY4 from south SCS (Yu et al., 2009)showed different findings with regard to the intensity and frequency ofstrong storms. For instance, the 107 isolated storm-transported coralbranches and shells observed in 151 samples from our LYJ3 sedimentcore over the last 1800 suggested a mean storm recurrence rate of~17–20 years, which was much higher than the ~77 year recurrencerate in NY4 (Yu et al., 2009). Even the average value of the coarse-grained content (25.92%) from LYJ3 that showed a progressively up-wards increase, was higher than that of NY4 (9.2%) (Yu et al., 2009). Incontrast, the latter appeared to be a weakly increasing trend in coarse-fraction contents (over the last 4000 years) towards present time, but

this trends not obvious within the last millennium (Yu et al., 2009). Inaddition, comparisons between our Linyang Reef storm proxy recordsand the Kamikoshiki (Japan) storm reconstruction from western NorthPacific (Woodruff et al., 2009) showed an inverse correlation over thepast millennia. However, documented typhoon landfalls to theGuangdong Providence in the northern coast of SCS over the pastmillennia exhibited an increase in typhoon occurrences (Elsner and Liu,2003), which were concurrent with periods of more frequent and strongstorm activity in the Xisha Islands during this period. It may indicate anoscillating pattern in storm activity between the SCS and western NorthPacific or regional shift/differentia in the preferred paths for storms.

In addition, Xisha Islands where storms frequently originate fromthis area, is frequently influenced by tropical storms that will increasethe coarse-grained fraction contents (especially for the isolated ormixed and disorderly coral branches) in the lagoon sediments. Therewas evidence that coral blocks were undoubtedly transported by ex-treme wave events, and the content peaks of coarse-grained fraction inthe lagoon core should reliably record extreme wave events such asstrong storms or tsunamis (Yu et al., 2009). As described above, eachidentified content peak (marked by the isolated or mixed and disorderlycoral branches) probably corresponded to an episode of relatively highstorminess rather than an individual strong storm (or tsunami) event bycomparison with modern observations from the western Pacific over thepast several decades (CE 1945–2018). If this is the case, it may easilyoverwrite the sign of individual storm and lead to the failure for thecomparison with modern storm records due to the AMS 14C error andresolution, and provide a reasonable explanation for the low sedimen-tary rate (even hiatus) in the upper part of LYJ cores. However, it doesnot rule out the possibility that a storm or tsunami may have depositeda thick layer of sediment. For example, a possible tsunami-inducedsediment, which inferred by reverse or chaotic age and acceleratedsedimentation rates within the period of ~2800–2600 cal yr BP wasfound in both LYJ cores.

5.3. Probable impact of tropical storms

Tropical storm activity often develops over oceans in regions whereSSTs exceed 26 °C in the current climate (Lighthill et al., 1994) becausea high-temperature ocean represents a huge energy base for the gen-eration and maintenance of storms. Several recent publications havesparked debate over whether a causal relationship occurs between in-creasing tropical storm frequency and intensity and increasingly warmclimates, such as increasing SSTs (Trenberth, 2005; Emanuel, 2005;Webster et al., 2005; Christopher and Landsea, 2005; Klotzbach, 2006;Knutson et al., 2010). Evidence from an analysis of global hurricanedata indicates a 30-year trend against a background of increasing SSTs(Webster et al., 2005). The trends in global tropical cyclone activityover the past twenty years (1986–2005) indicate that global net tropicalcyclone activity has not significantly changed (Klotzbach, 2006).Nevertheless, evidence from the North Atlantic suggests that greaterstorm activity is caused by simultaneous increases in SST and decreasesin vertical wind shear (Goldenberg et al., 2001; Mann and Emanuel,2006; Arpe and Leroy, 2009). However, the cited studies are based oninstrumental records of tropical cyclone activity that only span the pastfew decades. Therefore, this hypothesis presents considerable un-certainty and would require a longer global natural record for furtherconfirmation.

Our grain-size results identified three distinct units in the transfor-mation of the coral reef lagoon of the Yongshu Atoll of the XishaIslands. We interpreted these findings based on a comparison with theresults of previous studies from Yongshu Reef (Yu et al., 2006, 2009)and Dongdao Island (Yan et al., 2011; Sun et al., 2013) in the SCS, δ18Orecords (‰, VSMOW) from the stalagmite of Dongge Cave (Wang et al.,2005) in the Northern Hemisphere.

The coarse-grained fraction (> 1mm) content peaks (more than theaverage value) from the lagoon core NY4 were successfully used as

Y. Yue, et al. Palaeogeography, Palaeoclimatology, Palaeoecology 521 (2019) 57–71

68

storm indicators and provided a reliable assessment of strong storms(Yu et al., 2006, 2009). Accordingly, the long-term reconstruction ofstorm variability based on our coral lagoon core LYJ3 obtained from thetropical SCS may be used in evaluations of the dominant processes thatcontrol storm activity in the western Pacific over longer timescales. Incoral lagoon core LYJ3, 404 samples with coarse-fraction contents>25.92% were identified from the last ~3500 years, which indicatedthat intense storms occurred at a frequency of approximately a decade.At least 216 major peaks marked by isolated coral branches or shellscan be counted in the LYJ3 profile. As discussed above, these isolatedcoral branches or shells were likely transported by strong hydro-dynamic energy, and they serve as indicators of a series of strongstorms. The age distribution and relative probability frequency weremainly centered on the two broad periods from ~3500 to 3100 cal yrBP and ~1800 cal yr BP to present, respectively. As shown in Fig. 7, thetime series for major peaks of the coarse-grained fraction content abovethe average value over the last ~3500 years were consistent with thetime series of SST variability (Stott et al., 2004). Both curves presenteda roughly similar trend that likely indicated a correlation between in-creases in tropical storms and the relatively warming climate. Our re-sults are likely relevant to the question of whether global warming leadsto an increase in cyclone frequency or intensity. One plausible ex-planation for this pattern is that increased tropical SSTs promotevariability in storm activity. This hypothesis was supported by our dataand correlated with observations over the past several decades(Emanuel, 2005; Webster et al., 2005; Arpe and Leroy, 2009). Therewas also evidence that the warming of the tropical ocean actuallypromoted vertical mixing as may form more storms (Fedorov et al.,2009). Romps and Kuang (2009) found that storms brought moistureinto the stratosphere, and the increase in stratospheric water vapourmay cause the atmosphere to warm up (Shindell, 2001), thereby in-creasing the global warming trend (Romps and Kuang, 2009). Globalwarming, in turn, will further affect the frequency and intensity of thestorm (Emanuel, 2005), thus forming a possible positive feedback me-chanism between storm activity and temperature. Therefore, the long-term frequency and intensity of tropical storms may be attributed to therelative high SSTs in western Pacific over the last ~3500 years. In ad-dition, an upward trend in the coarse-grained fraction and SSTs hasoccurred in recent decades. These findings indicate that increasing SSTsin the future may lead to more storm activity in a warming warm poolcaused by global warming and the rapid increase in atmosphericgreenhouse gas concentrations attributed to intense human impacts(IPCC, 2007) and that such changes might also lead to increases instorm activity in this region. What is more, recent numerical models offuture typhoon and storm surges under different global warming sce-narios also indicates that climate change, e.g., is considered to onlyincrease SST, its intensity and storm surge will be larger than under thepresent climate (Nakamura et al., 2016). Therefore, greater attentionshould be focused on potential upward trends of storm-related losses incoastal regions of western Pacific.

The potentially high sedimentation rate and decade-long timescalein the LYJ3 coral reef lagoon profile can be used to study naturalchanges under different boundary conditions, such as the relativewarming of the Medieval Warm Period (MWP, CE 800–1300) and thecooling period of the Little Ice Age (LIA, CE 1400–1850) (Yan et al.,2011). As shown in Fig. 7, our data suggest less variation in the contentof the coarse-grained fraction associated with storm frequency duringthe two anti-phases. One plausible explanation for this reduced varia-tion is the wetter conditions in the SCS during the LIA which wasmarked by a strengthened and perhaps westward-shifted Pacific WalkerCirculation (PWC) (Yan et al., 2011) as well as the southward shift ofthe ITCZ (Sachs et al., 2009; Yan et al., 2011, 2015). In this case, thesouthward migration of the ITCZ likely resulted in drier conditions inEast Asia compared with the conditions in the SCS during the LIA. Inaddition, high SSTs and storms usually favour more intense rainfall. Thedecline in the fine-grained fraction and relative increase in the coarse-

grained fraction suggest the strengthening of hydrodynamic forcingassociated with intense storm activity and considerable precipitation.Therefore, an increasing trend of storm frequency likely occurredduring the LIA because of the influence of a strengthened PWC, whichcaused considerable precipitation associated with storms and promotedincreases in the coarse-grained fraction. Such conclusion also was ob-tained from lacustrine sediments that suggested larger sediment grainsize reflected more rainfall and higher lake level for short time-scale(Chen et al., 2004). Similar findings in which precipitation was in-herently associated with tropical storms were observed in the neigh-bouring Dongdao Island (Yan et al., 2011).

The lagoon core LYJ3 was extracted from a location in the tropicalSCS, where is strongly influenced by AM. As shown in Fig. 7, we notedthat the long-term trend of the coarse-fraction content in the LYJ lagooncore displayed a progressive increasing trend over the last ~3500 yearsthrough the late Holocene. Interestingly, we found that the variationtrend of the coarse-grained fraction was generally synchronous withweakening AM intensity (Wang et al., 2005). It was also significantlycorrelated with the North Atlantic climate recorded in the Greenlandice core (North Greenland Ice Core Project Members, 2004). Thesefindings likely suggest climatic similarities and correlations betweenthese areas, thus indicating pervasive global climatic teleconnections inthis time interval. Similarly, Donnelly and Woodruff (2007) also sug-gested that intense hurricane activity in the Caribbean area over thepast 5000 years was closely related to the West African monsoon.Nevertheless, changing temporal resolution, due to varying sedi-mentation rate, is always an unavoidable problem for time-series studybased on archive such as sediment core. Our coral lagoon cores (LYJ2 &LYJ3) are not an exception. Due to a great uncertainty exists in as-suming linear compaction rates in the LYJ cores and linear sedi-mentation rates between dated intervals, as well as the analytical un-certainty in the AMS 14C and TIMS U-series dates, our data do notprovide a sufficiently detailed picture to extract the definitive relationsbetween them but more research is needed. Although the broad declinein AM intensity during the late Holocene has been reported to be as-sociated with the southward shift of the Northern Hemisphere summerposition of the ITCZ, which occurs in response to the orbitally inducedlowering of summer insolation (Wang et al., 2005; Fleitmann et al.,2007; Yancheva et al., 2007; Wanner et al., 2008). Therefore, in ourfuture research, we will investigate the extent to which changes in themonsoon cycle are reflected in our lagoon cores via additional radio-carbon and TIMS U-series dating and further grain-size analyses.

The last but not the least is the effect of Holocene sea-level changeon storm. The sea-level change during the late Holocene, for instance,Lam and Boyd (2001) and Tanabe et al. (2006) suggested that it was asmuch as ~1.5–3m higher between 2000 and 3500 yr BP. By contrast,clear evidences from Zong (2004) and Xiong et al. (2018) indicated thatthe last 3000 years saw little movement of the relative sea level (RSL)for the low latitude part of the China coast or the northern coast ofSouth China Sea. e.g., in geologically stable coastal sites, the highstandis recorded at the same altitude as the present-day sea level (Zong,2004). On the other hand, the recent research on the sea level varia-bility over the last 2000 years suggested that global sea level varied by∼ ± 8 cm over the pre-Industrial Common Era (Kopp et al., 2016).There is no evidence that the gradually increasing coarse-grained se-diments from LYJ cores were consistent with sea level changes.Therefore, we infer that sea level changes during the past 2000 yearsmay have little effect on LYJ lagoon sediments. Similarly, a previousstudy also suggested that cyclone frequency might not have been af-fected by sea level change (Hayne and Chappell, 2001). In addition, wedid not discuss the influence of ENSO or other not mentioned aspectsdue to space constraints.

In sum, the variability in the grain sizes may be linked to changesover time in the synchronous AM intensity, and the results are similar toclimate records obtained from ice cores from Greenland because of theorbitally induced lowering of summer insolation associated with the

Y. Yue, et al. Palaeogeography, Palaeoclimatology, Palaeoecology 521 (2019) 57–71

69

southern shift in the ITCZ. On the other hand, the inverse correlationbetween the coral storm record and the AM strength, together with theinverse correlation between the coral storm record and the storm recordof Japan, probably revealed a very important mechanism operating inthis region. In fact, the coral storm record indicated no fewer typhoonshitting the northern SCS during the LIA, during which a lower numberof typhoons was expected to be generated. Thus, the following sce-narios can be observed. During a period of warmer climate, more ty-phoons were generated, but more of them travelled north hitting Japanrather than SCS. During a period of cooler climate, fewer typhoons weregenerated, but most of them travelled into SCS.

6. Conclusions

Well-dated grain-size records were generated from the neighbouringLYJ2 and LYJ3 cores which collected from the same coral reef lagoon atthe Yongle Atoll off the Xisha Islands in the northern SCS, and providenew insights into the tropical storm history in this area over the last~3500 years. The results suggest that coarse-grained deposits, whichinclude isolated or mixed and disorderly coral branches, provide in-formation on the spatial and temporal variability of extreme events, andthis information is essential for understanding past storm activity andrelated dynamics as well as predicting possible future impacts. Themain conclusions are as follows.

The sediment grain-size analyses show that strong and frequenttropical storm activity occurred over the period from ~3500 to3100 cal yr BP and the last 1800 years, and weak storm stage occurredfrom ~3100 to 1800 cal yr BP. However, a tsunami event likely oc-curred within the period from ~2800 to 2600 cal yr BP.

It was evidenced that the warming of the tropical ocean actuallypromoted vertical mixing as may form more storms, while stormsbrought moisture into the stratosphere, and the increase in strato-spheric water vapour may cause the atmosphere to warm up, therebyincreasing the global warming trend. Global warming, in turn, willfurther affect the frequency and intensity of the storm, thus forming apossible positive feedback mechanism between storm activity andtemperature. The overall temporal pattern obtained from coral bran-ches, which were used as an indicator of strong storms, resembled therelatively high SST records of the western Pacific. Our results furtherpredict that increasing SSTs in the future might lead to an increase instorm activity in the western Pacific warm pool as the planet warms.

Acknowledgments

Drs. Hong Yan, Rui Wang, Tao Liu, Juan Li, Yu Zhang and BrianChase provided helpful comments. This work was supported jointly byNational Key R&D Program of China (Grant No. 2017YFA0603300),National Natural Science Foundation of China (Grant Nos. 91428203,41702182, and 41302281), the Guangxi scientific projects (Nos.AD17129063, AA17204074 and 2018GXNSFAA281293), and the BaguiFellowship of Guangxi Province (Grant No.2014BGXZGX03).

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.palaeo.2019.02.009.

References

Arpe, K., Leroy, S.A.G., 2009. Atlantic hurricanes—testing impacts of local SST, ENSO,stratospheric QBO—implications for global warming. Quat. Int. 195 (1–2), 4–14.

Blake, E.S., Landsea, C.W., Miami, N., Gibney, E.J., Group, I.M.S., Asheville, N, 2011. TheDeadliest, Costliest, and Most Intense United States Tropical Cyclones From 1851 to2010 and Other Frequently Requested Hurricane Facts. NOAA TechnicalMemorandum NWS NHC-6.

Chen, J., Wan, G., Zhang, D.D., Zhang, F., Huang, R., 2004. Environmental records oflacustrine sediments in different time scales: sediment grain size as an example. Sci.

China Ser. D 47 (10), 954.Chen, H.-F., Wen, S.-Y., Song, S.-R., Yang, T.-N., Lee, T.-Q., Lin, S.-F., Hsu, S.-C., Wei, K.-

Y., Chang, P.-Y., Yu, P.-S., 2012. Strengthening of paleo-typhoon and autumn rainfallin Taiwan corresponding to the Southern Oscillation at late Holocene. J. Quat. Sci. 27(9), 964–972.

Christopher, W.L., Landsea, C.W., 2005. Meteorology: hurricanes and global warming.Nature 438 (7071), E11–E12.

Donnelly, J.P., Woodruff, J.D., 2007. Intense hurricane activity over the past 5,000 yearscontrolled by El Niño and the West African monsoon. Nature 447 (7143), 465–468.

Elsner, J.B., Liu, K., 2003. Examining the ENSO–typhoon hypothesis. Climate Res. 25,43–54.

Emanuel, K., 2005. Increasing destructiveness of tropical cyclones over the past 30 years.Nature 436 (7051), 686–688.

Fedorov, A.V., Brierley, C.M., Emanuel, K., 2009. Tropical cyclones and permanent ElNiño in the early Pliocene epoch. Nature 463, 1066–1070.

Fleitmann, D., Burns, S.J., Mangini, A., Mudelsee, M., Kramers, J., Villa, I., Neff, U., Al-Subbary, A.A., Buettner, A., Hippler, D., Matter, A., 2007. Holocene ITCZ and Indianmonsoon dynamics recorded in stalagmites from Oman and Yemen (Socotra). Quat.Sci. Rev. 26 (1–2), 170–188.

Goldenberg, S.B., Landsea, C.W., Mestas-Nuñez, A.M., Gray, W.M., 2001. The recent in-crease in Atlantic hurricane activity: causes and implications. Science 293 (5529),474–479.

Hayne, M., Chappell, J., 2001. Cyclone frequency during the last 5000 years at CuracoaIsland, North Queensland, Australia. Palaeogeogr. Palaeoclimatol. Palaeoecol. 168(3–4), 207–219.

IPCC, 2007. Climate Change 2007: The Physical Science Basis: Contribution of WorkingGroup I to the Fourth Assessment Report of the Intergovernmental Panel on ClimateChange. Cambridge University Press, Cambridge and New York, pp. 1–996.

Klotzbach, P.J., 2006. Trends in global tropical cyclone activity over the past twenty years(1986-2005). Geophys. Res. Lett. 33 (10), L10805.

Knutson, T.R., McBride, J.L., Chan, J., Emanuel, K., Holland, G., Landsea, C., Held, I.,Kossin, J.P., Srivastava, A.K., Sugi, M., 2010. Tropical cyclones and climate change.Nat. Geosci. 3 (3), 157–163.

Kopp, R.E., Kemp, A.C., Bittermann, K., Horton, B.P., Donnelly, J.P., Gehrels, W.R., Hay,C.C., Mitrovica, J.X., Morrow, E.D., Rahmstorf, S., 2016. Temperature-driven globalsea-level variability in the Common Era. Proc. Natl. Acad. Sci. U. S. A. 113 (38),E1434–E1441.

Lam, D.D., Boyd, W.E., 2001. Some facts of sea-level fluctuation during the latePleistocene–Holocene in Ha Long Bay and Ninh Binh area. J. Sci. Earth 23, 86–91 (inVietnamese with English abstract).

Landsea, C.W., Harper, B.A., Hoarau, K., Knaff, J.A., 2006. Can we detect trends in ex-treme tropical cyclones? Science 313 (5786), 452–454.

Lane, P., Donnelly, J.P., Woodruff, J.D., Hawkes, A.D., 2011. A decadally-resolved pa-leohurricane record archived in the late Holocene sediments of a Florida sinkhole.Mar. Geol. 287 (1–4), 14–30.

Leroy, S.A.G., Niemi, T.M., 2009. Hurricanes and typhoons: from the field records to theforecast. Quat. Int. 195 (1–2), 1–3.

Lighthill, J., Holland, G., Gray, W., Landsea, C., Graig, G., Ecans, J., Kurihara, Y., Guard,C., 1994. Global climate change and tropical cyclones. Bull. Am. Meteorol. Soc. 75(11), 2147–2157.

Lin, W., Yu, K., Wang, Y., Liu, X., Wang, J., Ning, Q., Li, Y., 2018. Extremely lowradioactivity in marine sediment of coral reefs and its mechanism. Chin. Sci. Bull. 63(21), 2173–2183.

Liu, K., 2007. Uncovering prehistoric hurricane activity. Am. Sci. 95 (2), 126–133.Liu, K., Fearn, M.L., 1993. Lake-sediment record of late Holocene hurricane activities

from coastal Alabama. Geology 21 (9), 793–796.Liu, K., Fearn, M.L., 2000. Reconstruction of prehistoric landfall frequencies of cata-

strophic hurricanes in northwestern Florida from lake sediment records. Quat. Res.54 (2), 238–245.

Liu, K., Shen, C., Louie, K.-S., 2001. A 1,000-year history of typhoon landfalls inGuangdong, southern China, reconstructed from Chinese historical documentary re-cords. Ann. Assoc. Am. Geogr. 91 (3), 453–454.

Liu, X., Vandenberghe, J., An, Z., Li, Y., Jin, Z., Dong, J., Sun, Y., 2016. Grain size of LakeQinghai sediments: implications for riverine input and Holocene monsoon variability.Palaeogeogr. Palaeoclimatol. Palaeoecol. 449 (1), 41–51.

Louie, K., Liu, K., 2003. Earliest historical records of typhoons in China. J. Hist. Geogr. 29(3), 299–316.

Lu, H., Liu, K., 2005. Phytolith assemblages as indicators of coastal environmentalchanges and hurricane overwash deposition. Holocene 15 (7), 965–972.

Mann, M.E., Emanuel, K.A., 2006. Atlantic hurricane trends linked to climate change. EOSTrans. Am. Geophys. Union 87, 233–244.

May, S.M., Brill, D., Leopold, M., Callow, J.N., Engel, M., Scheffers, A., Opitz, S., Norpoth,M., Brückner, H., 2017. Chronostratigraphy and geomorphology of washover fans inthe Exmouth Gulf (NW Australia) – a record of tropical cyclone activity during thelate Holocene. Quat. Sci. Rev. 169, 65–84.

Nakamura, R., Shibayama, T., Esteban, M., Iwamoto, T., 2016. Future typhoon and stormsurges under different global warming scenarios: case study of typhoon Haiyan(2013). Nat. Hazards 82 (3), 1645–1681.

National Geophysical Data Center/World Data Service, 2015. Global Historical TsunamiDatabase. National Geophysical Data Center, NOAAhttps://doi.org/10.7289/V5PN93H7. Available at. http://ngdc.noaa.gov/hazard/tsu_db.shtml.

Needham, H.F., Keim, B.D., Sathiaraj, D., 2015. A review of tropical cyclone-generatedstorm surges: global data sources, observations, and impacts. Rev. Geophys. 53,545–591.

Noren, A.J., Bierman, P.R., Steig, E.J., Lini, A., Southon, J., 2002. Millennial-scale stor-miness variability in the northeastern United States during the Holocene epoch.

Y. Yue, et al. Palaeogeography, Palaeoclimatology, Palaeoecology 521 (2019) 57–71

70

Nature 419 (6909), 821–824.North Greenland Ice Core Project members, 2004. High-resolution record of Northern

Hemisphere climate extending into the last interglacial period. Nature 431 (7004),147–151.

Nott, J., 2004. Palaeotempestology: the study of prehistoric tropical cyclones–a reviewand implications for hazard assessment. Environ. Int. 30 (3), 433–447.

Nott, J., Hayne, M., 2001. High frequency of 'super-cyclones' along the Great Barrier Reefover the past 5,000 years. Nature 413 (6855), 508–512.

Qiao, S., Shi, X., Wang, G., Zhou, L., Hu, B., Hu, L., Yang, G., Liu, Y., Yao, Z., Liu, S., 2017.Sediment accumulation and budget in the Bohai Sea, Yellow Sea and East China Sea.Mar. Geol. 390, 270–281.

Reimer, P.J., Bard, E., Bayliss, A., Beck, J.W., Blackwell, P.G., Ramsey, C.B., Buck, C.E.,Cheng, H., Edwards, R.L., Friedrich, M., Grootes, P.M., Guilderson, T.P., Haflidason,H., Hajdas, I., Hatté, C., Heaton, T.J., Hoffmann, D.L., Hogg, A.G., Hughen, K.A.,Kaiser, K.F., Kromer, B., Manning, S.M., Niu, M., Reimer, R.W., Richards, D.A., Scott,E.M., Southon, J.R., Staff, R.A., Turney, C.S.M., van der Plicht, J., 2013. IntCal13 andmarine13 radiocarbon age calibration curves 0–50,000 years cal BP. Radiocarbon 55(4), 1869–1887.

Romps, D.M., Kuang, Z., 2009. Overshooting convection in tropical cyclones. Geophys.Res. Lett. 36, 269–277.

Sachs, J.P., Sachse, D., Smittenberg, R.H., Zhang, Z., Battisti, D.S., Golubic, S., 2009.Southward movement of the Pacific intertropical convergence zone AD1400–1850.Nat. Geosci. 2, 519–525.

Scheffers, S.R., Haviser, J., Browne, T., Scheffers, A., 2009. Tsunamis, hurricanes, thedemise of coral reefs and shifts in prehistoric human populations in the Caribbean.Quatern. Int. 195 (1–2), 69–87.

Shindell, D.T., 2001. Climate and ozone response to increased stratospheric water vapor.Geophys. Res. Lett. 28 (8), 1551–1554.

Stott, L., Cannariato, K., Thunell, R., Haug, G.H., Koutavas, A., Lund, S., 2004. Decline ofsurface temperature and salinity in the western tropical Pacific Ocean in theHolocene epoch. Nature 431 (7004), 56–59.

Stuiver, M., Reimer, P.J., 1993. Extended 14C database and revised CALIB radiocarboncalibration program. Radiocarbon 35 (1), 215–230.

Sun, L., Zhou, X., Huang, W., Liu, X., Yan, H., Xie, Z., Wu, Z., Zhao, S., Da, S., Yang, W.,2013. Preliminary evidence for a 1000-year-old tsunami in the South China Sea. Sci.Rep. 3, 1655.

Tanabe, S., Saito, Y., Lan Vu, Q., Hanebuth, T.J.J., Lan Ngo, Q., Kitamura, A., 2006.Holocene evolution of the Song Hong (Red River) delta system, northern Vietnam.Sediment. Geol. 187 (1–2), 29–61.

Trenberth, K., 2005. Uncertainty in hurricanes and global warming. Science 308 (5723),1753–1754.

Wang, Y., Cheng, H., Edwards, R.L., He, Y., Kong, X., An, Z., Wu, J., Kelly, M.J., Dykoski,C.A., Li, X., 2005. The Holocene Asian monsoon: links to solar changes and NorthAtlantic climate. Science 308 (5723), 854–857.

Wanner, H., Beer, J., Bütikofer, J., Crowley, T.J., Cubasch, U., Flückiger, J., Goosse, H.,Grosjean, M., Joos, F., Kaplan, J.O., Küttel, M., Müller, S.A., Prentice, I.C., Solomina,O., Stocker, T.F., Tarasov, P., Wagner, M., Widmann, M., 2008. Mid- to Late Holoceneclimate change: an overview. Quat. Sci. Rev. 27 (19–20), 1791–1828.

Webster, P.J., Holland, G.J., Curry, J.A., Chang, H.-R., 2005. Changes in tropical cyclone

number and intensity in a warming environment. Science 309 (5742), 1844–1846.Williams, H.F.L., 2013. 600-year sedimentary archive of hurricane strikes in a prograding

beach ridge plain, southwestern Louisiana. Mar. Geol. 336 (1), 170–183.Woodruff, J.D., Donnelly, J.P., Okusu, A., 2009. Exploring typhoon variability over the

mid-to-late Holocene: evidence of extreme coastal flooding from Kamikoshiki, Japan.Quat. Sci. Rev. 28, 1774–1785.

Woodruff, J.D., Kanamaru, K., Kundu, S., Cook, T.L., 2015. Depositional evidence for theKamikaze typhoons and links to changes in typhoon climatology. Geology 43, 91–94.

Xiong, H., Zong, Y., Qian, P., Huang, G., Fu, S., 2018. Holocene sea-level history of thenorthern coast of South China Sea. Quat. Sci. Rev. 194, 12–26.

Yan, H., Sun, L., Oppo, D.W., Wang, Y., Liu, Z., Xie, Z., Liu, X., Cheng, W., 2011. SouthChina Sea hydrological changes and Pacific Walker Circulation variations over thelast millennium. Nat. Commum. 2, 293.

Yan, H., Wei, W., Soon, W., An, Z., Zhou, W., Liu, Z., Wang, Y., Carter, R.M., 2015.Dynamics of the intertropical convergence zone over the western Pacific during theLittle Ice Age. Nat. Geosci. 8 (4), 315–320.

Yancheva, G., Nowaczyk, N.R., Mingram, J., Dulski, P., Schettler, G., Negendank, J.F.,Liu, J., Sigman, D.M., Peterson, L.C., Haug, G.H., 2007. Influence of the intertropicalconvergence zone on the East Asian monsoon. Nature 445 (7123), 74–77.

Yang, W., Zhao, X., Zhang, F., Fang, Z., Ma, H., Chen, M., Qiu, Y., Zheng, M., 2018.Identification of the earlier human-induced sedimentation change in Daya Bay,northern South China Sea using 210Pb and 137Cs. Mar. Pollut. Bull. 126, 334–337.

Yao, T., Shi, Y., Thompson, L.G., 1997. High resolution record of paleoclimate since theLittle Ice Age from the Tibetan ice cores. Quat. Int. 37, 19–23.

Yu, K., Zhao, J., Collerson, K.D., Shi, Q., Chen, T., Wang, P., Liu, T., 2004. Storm cycles inthe last millennium recorded in Yongshu Reef, southern South China Sea.Palaeogeogr. Palaeoclimatol. Palaeoecol. 210 (1), 89–100.

Yu, K., Zhao, J., Wang, P., Shi, Q., Meng, Q., Collerson, K.D., Liu, T., 2006. High-precisionTIMS U-series and AMS 14C dating of a coral reef lagoon sediment core fromsouthern South China Sea. Quat. Sci. Rev. 25 (17–18), 2420–2430.

Yu, K., Zhao, J., Shi, Q., Meng, Q., 2009. Reconstruction of storm/tsunami records overthe last 4000 years using transported coral blocks and lagoon sediments in thesouthern South China Sea. Quat. Int. 195 (1–2), 128–137.

Yu, K., Hua, Q., Zhao, J., Hodge, E., Fink, D., Barbetti, M., 2010. Holocene marine14Creservoir age variability: evidence from230Th-dated corals in the South China Sea.Paleoceanography 25, PA3205.

Yu, K., Zhao, J., Roff, G., Lybolt, M., Feng, Y., Clark, T., Li, S., 2012. High-precision U-series ages of transported coral blocks on Heron Reef (southern Great Barrier Reef)and storm activity during the past century. Palaeogeogr. Palaeoclimatol. Palaeoecol.337-338, 23–36.

Yue, Y., Zheng, Z., Rolett, B.V., Ma, T., Chen, C., Huang, K., Lin, G., Zhu, G., Cheddadi, R.,2015. Holocene vegetation, environment and anthropogenic influence in the FuzhouBasin, southeast China. J. Asian Earth Sci. 99 (1), 85–94.

Zhao, J.X., Neil, D.T., Feng, Y.X., Yu, K.F., Pandolfi, J.M., 2009. High-precision U-seriesdating of very young cyclone-transported coral reef blocks from Heron and Wistarireefs, southern Great Barrier Reef, Australia. Quat. Int. 195 (1–2), 122–127.

Zong, Y., 2004. Mid-Holocene sea-level highstand along the Southeast Coast of China.Quat. Int. 117 (1), 55–67.

Y. Yue, et al. Palaeogeography, Palaeoclimatology, Palaeoecology 521 (2019) 57–71

71