High-resolution Late-holocene Biomarkers Records in Southern Continental Shelf of the Korean Peninsula: Implications for Climate Variations Over the Last 4 Kyr

High-resolution records for carbon isotopes of organic matter and n-alkane compounds were investigated in two gravity cores (SJP15-2 and SJP15-4) taken from the southern continental shelf of the Korean peninsula to evaluate the inuxes of terrestrial biomarkers and their linkage to paleoclimate and marine environmental changes since the last 4 kyr. The total organic carbon contents were < 1%, and the carbon isotope ratio of organic matters (d 13 C org ) ranged from approximately −21 ‰ to -22 ‰ and, they did not highly uctuate throughout the two cores. However, the vertical distributions of total terrestrial biomarkers, long-chain n-alkanes (nC 25-35 ), and individual n-alkane compounds exhibited distinctive uctuations. There are two switching points that discriminate patterns of excursion and distribution at ca. 2.5 ka, and 0.5 ka. Several n-alkane combined indices such as average chain length (ACL), carbon preference index (CPI), and paleovegetation index (Paq) were coincident with these switching points, implying that the supply of terrestrial biomarkers was strongly associated with environmental changes at the source area. In particular, variations of compound-specic n-alkanes isotope and the ratios of nC 31 /nC 27 and nC 31 /nC 29 follow those of n-alkanes indices, implying that this millennium records were associated with wetter climate conditions, and thus paleovegetation and paleoclimate variation. Comparison with previous data of the detrital quartz from the East China Sea and aeolian dust in the Cheju (Jeju) Island, South Korea, and Dongge cave oxygen isotope records indicates strong synchronicity with gradual paleoclimate degradation between 2.5 ka and 0.5 ka. Therefore, our high-resolution n-alkane data are very useful for reconstructing past climatic variability, suggesting that paleoclimate system of the East Asian region may have inuenced the sediment records of study area since the last 4 kyr.


Introduction
Continental shelf and coastal areas are important locations where accumulated terrestrial materials and in situ marine materials can be easily preserved. This means that terrestrial and oceanic records are mixed and re ect past climatological and oceanographic variations. Identi cation of the sources of terrestrial biomarkers in marine environments and the factors controlling their distributions is important for understanding the global biogeochemical cycle of biomarkers as well as local oceanographic and paleoclimatic environmental changes (Hu et al. 2006;Hyun et al. 2006;Jeng 2006). Such sediments are likely to contain high concentrations of oceanic and terrestrial materials (Stein 1990). However, it is di cult to determine the contributions of certain types of biomarkers to bulk organic matter because most biomarkers are mixed with terrestrial and marine materials. Owing to the complex nature of organic compounds in shallow continental marine environments, various parameters (stable isotope ratios of organic carbon [δ 13 C org ] and total nitrogen [δ 15 N TN ], carbon/nitrogen ratio [C/N ratio], and n-alkanes have been widely used to determine the sources of organic matter and related environmental interpretation (Minoura et al. 1997;Naidu et al. 2000;Cloern et al. 2002;Usui et al. 2006;McKirdy et al. 2010).
As one of the most signi cant terrestrial organic materials, n-alkanes have been widely used to determine organic matter sources and climatic variation in various locations (Jeng 2006; Eglinton and Eglinton 2008;Yamamoto et al. 2010;Chu et al. 2014). In particular, because high-molecular-weight long-chain nalkanes such as nC 27 , nC 29 , and nC 31 are major constituents of vascular land plants (Sikes et al. 2009;Yamamoto et al. 2010;Bush and McInerney 2013), sedimentary n-alkanes are potentially important for determining the sources of organic matter and related environmental changes, including paleoclimate variation. The abundance of these waxy hydrocarbons re ects the amount of organic matter transported to lake beds, alluvial deposits, marine sediment, and rivers from the surrounding land, making it possible to determine their contribution to the terrestrial organic matter ux. Therefore, these proxies provide insight into organic matter sources, past environmental changes, including vegetation changes Yamamoto et al. 2010), and local and global paleoclimate environmental variations (Horikawa et al. 2010;Seki et al. 2006;Street et al. 2013;Li et al. 2013;.
As a component of global climate changes, the East Asian monsoon system and its variation have been investigated through the various proxy data (Nagashima et al. 2013;Wang et al. 2014;Tada et al. 2016;Wang et al. 2020). In East Asia, variation of the East Asian Summer Monsoon (EASM) and East Asian Winter Monsoon (EAWM) has been a focused because many studies have shown that the monsoon signal and its intensity differs from among scienti c datasets (Li et al. 2013;Wang et al. 2014;. Some signals show coincident variations, while others are inconsistent in terms of EASM and EAWM intensity (An 2000;Wang et al. 2020). Regarding the Holocene climatic variation, it is strongly coupled with enhanced and/or reduced precipitation due to the latitudinal shift of EASM intensity. This earlier study indicates that the oscillation of EASM front results in deepest northwestward penetration make enhanced EASM during the period 3.5-2.0 kyr, and southeastward retreat (weak EASM) occurred during the period 2.0-0 kyr (Wang et al. 2020). Also this precipitation changes supported by the other proxies of eolian dust that transported in accordance to the westerly jet (WJ) path over the East Asia region (Nagashima et al. 2013).
Our study area is within the East Asian monsoon belt; therefore, the sediments of this study may re ect past climatic variations in the scope of the East Asian region. As long-chain n-alkanes from this studied core are likely to reveal changes in terrestrial in ux by river and atmosphere under the in uence of East Asian monsoon climate regime, we investigate n-alkanes variability recovered from southern inner continental shelf. The purpose of the present study is the reconstruction of past n-alkane in uxes and their relationships with in situ paleoceanographic and paleoclimate variabilities since the mid-Holocene.

Geological and oceanographic setting
The southern coastal area of the Korean peninsula is characterized by a complex geomorphology and shallow water depth. It is expected that large in uxes of terrestrial material have been supplied from large and small river systems such as the Seomjin River in study area (Fig. 1). The Seomjin River, located on the extreme southern coast of Korea, is characterized by a large outward ow of fresh water throughout the year, with an annual discharge of approximately 7.2 × 10 8 m 3 of fresh water to Gwangyang Bay (Korea Ministry of Construction, KMC 1980). The average suspended sediment accumulation rate in the river mouth is ~ 17.8 mg/L in the bottom layer and 4.8 mg/L in the surface layer, which results in an estimated deposited mass of 0.8 × 10 6 tons per year (Kim and Kang 1991). Considering the sea level changes of southern coastal area of Korean peninsula, it has been considered that the sea level was reached at present level around the 6 ~ 7 ka before (Park et al. 2000;Yoo et al. 2014), therefore Yeosu Bay, a major catchment area of the Seomjin River, has been subject to sediment accumulation since the early Holocene.
Besides direct supply from the neighboring inland area, there are other sediment transportations resulting from the regime of the East Asian monsoon belt. Materials could be transported through air, and suspended materials could be transported by the Korean Coastal Current (Fig. 1). Therefore, the high sedimentation rate of the studied cores makes it possible to reconstruct high-resolution records re ecting past monsoon signals as well as in situ local paleoceanographic and paleoclimate variations.
As the two cores used in this study were taken from nearby locations, the lithological characters were very similar and reserve paleoclimatic and paleoceanographic information. To acquire information and improve our understanding of the Late-Holocene paleoclimate variation, we gathered Holocene paleoclimatic data from cores acquired from the Cheju Island, South Korea (Lim and Matsumoto 2008), and a coastal area of the Yangtze River, China (Wang et al. 2020).

Sampling
A total two gravity piston core samples were collected from the outer part of Seomjin River, inner shelf of the Southern Sea of Korean Peninsula using a gravity piston core (Fig. 1). After sedimentological study of the two piston core, total 57 samples were collected for the analysis of total organic carbon (TOC), and carbon and nitrogen isotopes, n-alkanes. TOC, carbon (δ 13 C org ) isotopes in organic matter, and n-alkanes were analyzed to determine the time-dependent variation of the source of the organic matter and its distribution. Also sampling for carbon 14 C ages sediment samples were collected. Based on the sedimentary feature, a total 17 radio carbon 14 C ages samples were determined using sedimentary organic matters and carbonate shell fragments at Beta-isotope (Table 1).

Carbon isotope of organic matter and n-alkane
Isotopic composition of sedimentary organic matter was determined after treatment with 1M HCl overnight to remove carbonate materials and then subsequently rinsed with deionized water three times before freeze drying. The percentage of total carbon and nitrogen as well as their isotopic composition (δ 13 C org ) from the carbonate free samples were determined using EA-CF-IRMS. The isotope ratios of standards in the laboratory was measured three times on three separate occasions and quoted an instrumental measurement precision of ± 0.01‰ for carbon and ± 0.03‰ for nitrogen. To ensure quality control, each analysis run included international standards. All analytical process was conducted at the Iso-analytic in UK.
Gas chromatography (GC) was used for the analysis of n-alkanes in the F1 fraction using a Hewlett Packard 6890 series GC unit, with a split/splitless injector operated in a pulsed splitless mode, and electronic pressure control systems and a ame ionization detector (FID). A CP Sil5CB column (60 m × 0.32 mm i.d.) was used with the oven temperature programmed at 50-120ºC at 30ºC/min, 120-320ºC at 7ºC/min, and then held at 320ºC for 50 min. Prior to the quanti cation of lipids, a known aliquot of an internal standard, 5α-Cholestane, was added to the hydrocarbon fraction. Individual aliphatic (n-alkane) compounds were identi ed and quanti ed using the retention times of n-alkanes and the mass spectrum of selected ions from calibration standards at concentrations ranging from 10 to 200 µg/ml. Analytical quality control was carried out to make sure that the results are consistent, comparable, within speci ed limits of precision and accurate. n-alkane standard mixtures were injected to the GC/MS three times to have an absolute retention time of each individual compounds. Individual compounds of nalkanes in samples were identi ed and quanti ed based on the retention time and mass spectrum of selected ions with the calibration standards. Samples were spiked with a surrogate standard (n-Triacontane d62). After nal concentration, the extracts were spiked with internal standard (5α-Cholestane). A 70-90% surrogate recovery was seen from all samples.
After n-alkane analysis, the fractions were used for the individual n-alkane carbon isotope (δ 13 C ALK ) analysis using the GC/IRMS at Woods Hole Oceanographic Institute (WHOI), USA. A total of 1 µl of the nalkane concentration was injected into a fused silica capillary column (30 m * 0.32 mm i.d.), with helium used as carrier gas. The GC oven temperature was programmed with a stepwise increase from 50 to 310ºC at rates of 30ºC, 6ºC, and 9ºC/min. All δ 13 C ALK values were checked for precision and accuracy to interpret small differences in δ 13 C ALK values. The standard material (Methyl heneicosanoate, C 22 H 44 O 2 ) was used and the precisions of its δ 13 C ALK values obtained by GC/IRMS analyses were reported as the standard deviations (σ) of populations of at least three analyses. The standard deviation of GC/IRMS data for n-alkanes was usually less than 0.3‰.

Sediment characteristics and age model
In order to track any past records from the core sediment, it is fundamental to understand the sediment characteristics. This means sedimentological examination is the rst step to identifying past environmental changes. Based on the detailed examination of the sedimentary structure of the two core, the core sediments reveal typical hemipelagic sediment characteristics, having massive structures and being composed of homogenous clay materials. The examination begins with age determination on the sediment column, which may clarify high-resolution time-dependent environmental records in terms of terrestrial in ux from nearby land areas as well as in situ oceanographic variations. The sedimentary features and sedimentation rates preserved in these two cores suggesting that the sedimentary environment did not change signi cantly throughout deposition since the last 4 ka. This may imply that the sediments preserve continuous in situ oceanographic and climatic records. In particular, coring site belongs to East Asian monsoon belt, it can provide monsoon climatic signal as well as local coastal environmental changes.
According to the carbon-14 ( 14 C) age results, which were obtained by radiocarbon age dating at Beta Analytic Inc. (Miami, FL, USA), the age model for the SJP15-2 core was determined using 14 C bulk organic matter (BOM) and carbonate shell fragment (CSF) as reference data produce two regression lines. The slopes of the BOM-and CSF-based regression lines were nearly parallel, showing larger age differences toward surface sediments ( Fig. 2A). To avoid the contamination of BOM-based 14 C ages by the 14 C reservoir effect caused by older terrestrial carbon (Ikehara, 2000;Zhou et al. 2018), these age differences were transformed into real ages for detailed comparisons. Because CFS-based 14 C ages are also susceptible to contamination by the reservoir effect, we assumed a global average reservoir effect of 2-300 years, and a surface sediment age of 0 years (i.e., the present year), despite an intercept of 0.11 ka for the SJP15-2 core ( Fig. 2A).
Similarly, regression lines for the SJP15-4 core age model were constructed using BOM data, followed by CSF data. The SJP15-4 core was estimated to be older according to the BOM-based regression line than by the CSF-based line, with an age difference of about 1863 years at a depth of 136 cm. Therefore, we drew hypothetical lines within the age gap between the BOM-and CSF-based regression lines at a depth of 136 cm, using the SJP15-2 regression lines as a reference. The slopes of the new regression lines for the SJP15-4 core were similar to those of SJP15-2, with surface sediment age differences between the BOM-and CSF-based lines of about 1.69 ka and 2.04 ka for the SJP15-2 and SJP15-4 cores, respectively. This result suggests higher contamination by older carbon in the SJP15-4 core (Fig. 2B). However, the intercept ages of surface sediments in both cores (0.11 and 0.35 ka for the SJP15-2 and SJP15-4 cores, respectively) were offset to 0 years for convenient age model establishment. Thus, the age data have an uncertainty of 100-300 years, which is similar to the global average reservoir effect of 2-300 years (Alves et al. 2018).
Based on this 14 C age determination, the average sedimentation rates of the two cores were quite high, it was 120 cm/kyr for core SJP15-2 and 159 cm/kyr for core SJP15-4, respectively, exhibiting typical features of coastal and hemipelagic sediment. This high sedimentation rate makes it possible to track the in ux of terrestrial biomarker since the mid-Holocene. The results of 14 C ages for the two cores are summarized in Table 1.
3.2 Variations of total organic carbon (TOC, %), δ 13 C org , and δ 15 N TN The sediment of SJP15-2 (inner shelf) and SJP15-4 (outer shelf) showed a narrow range of TOC variation, remaining almost constant throughout except for several upper samples, which showed a slight decrease in SJP15-2, and increase in SJP15-4. Overall, TOC contents in the sediments of the two cores show narrow range of uctuation, ranging from 0.6-0.9% (average = 0.71%, n = 53) (Fig. 3).
A C/N ratio has been used to determine the source of organic matter even though it could not provide critical criteria where they are come from. For example, when the C/N ratio of organic matter > 12 indicates a terrigenous source, whereas a ratio < 12 indicates a marine source (Stein 1990). Given that the C/N ratios of organic matter in both cores were < 10, larger portion of organic matters were derived from in situ bio-productivity. Meanwhile, the surface distribution of TOC from the nearby areas re ects a mixture of marine and terrestrial sources (Badejo et al. 2017). Also large portion of terrestrial organic matter were contained in the sediment of the southern part of the coastal area (Hyun et al. 2011). Therefore, the TOC contents of the two cores sediment may come from a mixture of terrestrial and marine sources.
Like TOC contents and C/N ratios, δ 13 C org and δ 15 N TN values have also been used to determine the sources of organic carbon. In present study, δ 13 C org values ranged from − 22‰ to − 21‰ (average = − 21.4‰, n = 41) in both core sediments, showing typical marine productivity-related organic carbon sources when considering its endmember in neighboring sites (Hyun et al. 2011). The δ 13 C org value of marine particulate organic matter ranges from − 22‰ to − 18‰ (Middelburge and Nieuwenhuize, 1998), whereas lower values between − 33‰ and − 22‰ are considered to re ect terrestrial origins (Pancost and Boot 2004). Because lighter ratios indicate marine sources (the marine endmember value is approximately − 20‰ and the terrestrial endmember value is − 27‰), the values in the two cores sediment indicating that most of the organic carbon was produced by in situ marine bioproductivity.
Surface sediments from the catchment areas showed that relatively high amounts of terrestrial organic carbons were supplied (Hyun et al. 2011;Badejo et al. 2017). The heavier values of δ 13 C org upward in SJP15-4 may be attributable to a higher portion of marine sources or to diminishing terrestrial organic carbon. However, in the inner shelf, core SJP15-2 showed a relatively narrow range of δ 13 C org , implying a higher portion of terrestrial organic carbon compared to SJP15-4. Similar δ 13 C org values were recognized in nearby surface sediments, indicating mixtures of terrestrial and marine organic matter, while the δ 13 C org values of the Seomjin River predominantly indicate a terrestrial endmember (Badejo et al. 2017).
The results indicate that the organic matter deposited in the study area was delivered not directly from the Seomjin River but was mixed with biogenic sources and other origins. A notable depletion of δ 13 C org in surface sediments was observed around the catchment areas of Gwangyang Bay and Yeosu Bay, which had relatively low C/N ratios of < 10 (Hyun et al. 2011); data from the neighboring areas indicates that organic carbon has been deposited from various sources, and the mixture of terrestrial and oceanic sources differed from their site.
The uppermost lighter ratios of δ 13 C org and δ 15 N TN in SJP15-2 are associated with lower organic carbon contents. One simple explanation for this is the presence of inorganic nitrogen, presumably as ammonium absorbed onto clay, in the surface sediments, which can interfere with the identi cation of organic matter sources when applying the C/N ratio (Kang et al. 2007;Hu et al. 2009). This assumption is supported by the δ 15 N TN values (Fig. 3), which had a similar distribution to that of the C/N ratio. The observed variation in δ 15 N TN could be the result of a change in the nitrate supply, which could be in the form of nitrate fertilizer around the catchment area and a marine nitrate source in the inner shelf study area. Because the onset of δ 13 C org increase has occurred at 2.5 ka and 3.5 ka in SJP15-4 and SJP15-2, respectively, it may re ect a gradational increase of bioproductivity. The TOC and δ 13 C org excursions coincide, the lower C/N ratio of ~ 8 supports in situ bioproductivity. Even though gradational increases of δ 13 C org in both core were observed and interpreted as bioproductivity variation, terrestrial biomarker such as n-alkanes also could be supplied. Because δ 13 C org values are not an end-member of biogenic source, therefore, it was mixed with mainly marine and minor terrestrial organic matter.

Concentration and distribution of terrestrial n-alkanes
In addition to δ 13 C org and δ 15 N TN , terrestrial n-alkanes provide detailed information on source-related organic compound characteristics because each n-alkane compound can re ect a speci c source. For example, among the long-chain n-alkanes, nC 29 and nC 31 are predominately derived from terrestrial plant waxes, whereas the mid-chain n-alkanes nC 20 and nC 23 are indicative of different oating macrophyte groups (Visco et al. 1993;Fricken et al. 1998). This means that both middle (nC 21 , nC 23 ) and longer (nC 29 , nC 31 ) chain lengths can be traced to aquatic and terrestrial sources with excellent con dence. In a broad sense, the overall distribution of n-alkane concentrations is consequence of vegetation changes, thus climatic changes in source area.
An overall distribution of n-alkanes is characterized by predominant odd-number long-chain alkanes concentration (nC 27 − 33 ). The total n-alkane concentration (nC 25 -nC 35 ) showed large uctuations, ranging from 1272 to 274 µg/g (average = 728 µg/g, n = 27) in core SJP15-2, and from 3056 to 339 µg/g (average = 773 µg/g, n = 30) in core SJP15-4, respectively (Fig. 4). The vertical distribution of n-alkanes between the two cores was harmonious, but it was slightly different in terms of the concentration of midchain n-alkanes (nC 22 , nC 24 , nC 25 and nC 26 ). However, in detail, the vertical distribution was characterized by a distinctive increase in the total concentration from 2.8 ka to 1.3 ka in core SJP15-2, and from 3.0 ka to 0.5 ka with abrupt increase at the layers of 0.2 ka in core SJP15-4.
The timing of onset of increase in several individual n-alkanes are different from each other. Individual nalkane such as nC 29 , nC 31 , nC 33 , nC 35 in two core start to increase at 2.5 ka, but these four n-alkanes group differ from other group (nC 28 , nC 30 , nC 32 ) that shows distinctive increase from 2.0 ka. Given that long-chain n-alkanes such as nC 29 and nC 29 + nC 31 are considered proxies for forest and/or grassland vegetation, the excursions of these n-alkanes can provide information on paleoclimate variation because they are derived from terrestrial higher plant waxes (Eglinton and Hamliton 1967). Therefore, the major changes in above mentioned long-chain odd-carbon-numbered n-alkanes re ect shifts in terrestrial vegetation at the source area, as well as shifts in the photosynthetic pathway. This variation directly can be interpreted as vegetation community changes, thus climate changes.
The abrupt increasing of even numbered n-alkanes, nC 22 and nC 24 occurred around 2.5 ka in the cores SJP15-4, is very meaningful (Fig. 4). This may imply some submerged vegetation has been ourished in this interval. Also, the strong increase of nC 24 and nC 26 appeared at 1.2 ka in core SJP15-2 can be explained as ourish of oating macrophyte. Similarly, the increase of mid-chain n-alkanes (nC 21 , nC 23 ) has been observed during mid-late Holocene at the Swamp Lake of USA, and this mid-chain increase was explained by an increase in the abundance of aquatic plants and availability of shallow water habitat in rising lake level (Street et al. 2013). Therefore, increase of mid-chain n-alkanes occurred around 2.5 ka in core SJP15-4 might be associated with temporal environmental changes due to the ourishing of aquatic plants under the shallow water environment.
The comparison of each numbered n-alkane concentration of the two core can provide more detailed time-dependent variations (Fig. 5). Long-chain n-alkane such as nC 29 , nC 31 are predominantly present throughout the two core. However, mid-chain n-alkane such as nC 24 ~ 26 are differ from each other. The concentration of mid-chain n-alkane in SJP15-4 is relatively higher than that of SJP15-2 at around of 0.5 ka, whereas it is less at 1.2 ka than that of SJP15-2 at 1.12 ka (Fig. 5). Therefore, as shown in Fig. 4 and Fig. 5, each n-alkane concentration has been differentiated from time to time in each core site, implying time-dependent mid-chain n-alkane uctuation in accordance to climatic and vegetation changes. Badejo et al. (2017) examined n-alkane concentrations from the surface sediment in nearby area, and suggested large inputs of materials from the mouth of the Seomjin River to the inner shelf area, with a slight variation that could be due to an indirect supply. According to that results, the in ux of terrestrial material was higher near the Seomjin River mouth, decreasing further offshore. Therefore, various nalkane compounds have been expected to being supplied from a nearby source by Seomjin River and possible faraway mainland sources.
As combined proxies, spatiotemporal distribution of average chain length (ACL) and carbon preference index(CPI) of the n-alkanes also have been used to determine the characteristics of organic carbon since they can provide valuable information on n-alkane sources (Sikes et al. 2009;Jia et al. 2015;Li et al. 2015). In particular, the distribution of ACL has been linked to the geographical distribution of uvial, aeolian inputs and source regions (Poynter and Eglinton 1990). Therefore, it is possible to delineate nalkane sources by identifying environmental differences, even though the usefulness of this approach is limited to a source-speci c n-alkane distribution (Jeng 2006). In our study, there was a narrow and distinctive variation in the ACL with a similar trend between 3.0 ka and 1.0 ka in two core sediments. A clear difference between the two cores seem to appear around ca. 0.8 ka (Fig. 6). Overall, values of ACL in our core start to increase at 3.5 ka, and it describe a parabola until 1.0 ka with a peak around at 2.0 ka. Given that ACL refer to average number of carbon atoms per molecule based on the abundance of the odd-carbon-numbered higher plant n-alkane (Poynter and Eglinton 1990), this gradual increase and decrease interval implies the variabilities of alkane distribution, i.e., it might be related to gradational drier and wetter climatic condition. One possible reason for the variation in the distribution is the ACL variation of n-alkanes in different plant species. Based on this decreased ACL interval from 2.0 ka to 1.0 ka, we infer that our study area also experienced a susceptible East Asian monsoon climatic changes within this millennium scale.
CPI values of the waxes also provide additional information on n-alkane source because it originates from terrestrial plant community. Usually it shows wide range of variation; terrestrial higher plants usually have values of 5 ~ 10 (Hedges and Prahl 1993), lower values < 1 is thought to oil contamination in modern sediments (Jeng 2006;Seki et al. 2006). In the present study, the average CPI of 3.06 (range = 1.74-4.93) in core SJP15-2, and 3.2 (range = 1.77-4.38) in core SJP15-4, respectively. The highest value (4.93) occurred at 0.5 ka at SJP15-2 close to Seomjin River/inner shelf area, possibly associated with an in ux of n-alkanes from a terrestrial leaf wax source. However, CPI values in both core are not show any particular trends, so it is di cult to evaluate whether core sediments were contaminated or not, even though CPI values < 3 indicate oiled sediments (Farrington and Tripp, 1977).
Like ACL and CPI, the P aq is indicative of paleovegetation changes (Yamamoto et al., 2010). Sike et al.
(2009) used this paleo n-alkane index to detect the pattern of paleovegetation as follows: P aq = (C 23 + C 25 )/(C 23 + C 25 + C 29 + C 31 ). This formula indicates the relative abundance of C 3 plants with respect to C 4 plants. Increasing P aq values indicate an increasing ratio of C 23 and C 25 . A P aq value < 0.1 corresponds to terrestrial plants; 0.1-0.4 corresponds to emergent macrophytes; and 0.4-1 corresponds to submerged macrophytes (Ficken et al. 2000). In our study, the values of P aq does not exceed 0.4, but it includes increased interval from 2.0 ka to 0.8 ka in two cores with a peak at 0.8 ka (Fig. 6). Thus, the increasing tendency from 2.0 ka to 0.8 ka indicates an expansion of C 3 plants. This means that the P aq is strongly associated with the ACL and CPI, both of which indicate climatic and paleo-vegetation changes.
Similarly, nC 31 /nC 23 , nC 31 /nC 25 , nC 31 /nC 27 and nC 31 /nC 29 ratios also widely used as an indicator of past paleovegetation and paleoclimate variation (Hanisch et al. 2003;Xie et al. 2003;Zech 2006;Horikawa et al. 2010). As a whole, these indices show very similar excursion with ACL and opposite trend with respect to P aq between 3.5 ka and 1.0 ka (Fig. 6). This interval from 3.5 ka to 1.0 ka corresponds well with the increase of the concentration of long-chain n-alkanes, heavier trending δ 13 C (‰). All of these proxies show a clear change between 3.5 ka and 1.0 ka, supporting vegetation and climate changes. This climate condition supported by expansion of C 4 plant between 3.5 ka and 2.5 ka, and again relative expansion of C3 from 2.5 ka to 1.0 ka, as increase of Paq as shown in Fig. 6. All of multiple proxies of these ACL, CPI and Paq are very useful to understand paleovegetation and paleoclimate variations in study area.
3.4 Individual carbon isotopes of n-alkanes (δ 13 C ALK ) and sources The compound-speci c isotopic ratio in n-alkane (δ 13 C ALK ) has higher potential for complementing nalkane distributions and providing more reliable n-alkane provenances because n-alkane isotopes show no fractionation during photo-oxidation and microbial degradation (Collister et al. 1994;Ahad et al. 2011). This means that different sources result in different characteristics for δ 13 C ALK values. Previous research showed that δ 13 C ALK values for C 4 -derived n-alkanes were all within the range of − 18.5‰ to − 24.5‰, and those for C 3 -derived n-alkanes were more depleted, ranging from approximately − 31.4‰ to − 38.6‰ (Collister et al. 1994). For this reason, δ 13 C ALK has been widely used for compound source discrimination in various environments (e.g., Jeng 2006;Ahad et al. 2011).
In this study, the entire range of odd-numbered n-alkane carbon isotopes (δ 13 C ALK ) showed overall comparable excursions (Fig. 7). The distribution and the entire range of odd-numbered δ 13 C ALK does not show homogenous excursion, but it has a switching points at around 1.0 ka. Judging from the criteria reported previous result, our δ 13 C ALK ratio ranged from about − 21‰ to -31‰ is regarded as intermediate ratio of C 3 and C 4 plants, thus as mixture of C 3 and C 4 -drived consequence. In particular, the δ 13 C ALK excursions are very complicated and characterized by an abrupt negative shift of δ 13 nC 21 , δ 13 nC 23 , δ 13 nC 25, δ 13 nC 33 in SJP15-4 at 1.0 ka. Whereas, a positive shift of δ 13 nC 23 , δ 13 nC 25, δ 13 nC 33, δ 13 nC 35 in SJP15-2 at 1.0 ka can be recognized. The excursions of these individual δ 13 C ALK values were decoupled in several compounds such as nC 23 and nC 33 . Among oating macrophytes groups, such as nC 21 , nC 23 , and nC 25 , seems to differ from each other, and also the excursion of this group can be distinguished from emerged land macrophytes, such as nC 29 , nC 31 , and nC 33 (Fig. 7). As mentioned previously, because of the different sources of n-alkane isotopes have different values for δ 13 C ALK , the wide variation of δ 13 C ALK values in our study suggests major paleo-vegetation community change. As far as we focused on compound-speci c δ 13 C ALK between 3.0 ka and 1.0 ka, it is not complicated to compare each n-alkane compound provenance. However, δ 13 nC 29 shows lighting trend from 3.0 ka to 1.0 ka, indicating decrease of C4 plant. As indicated in Fig. 7, compound-speci c δ 13 C ALK may show some switching points at about 1.0 ka and 0.2 ka. This means that n-alkane compound changes would be matched well with its corresponding δ 13 C ALK .
The excursions of δ 13 nC 29 and δ 13 nC 31 are quite different from those of nC 29 and nC 31 even though both concentrations of nC 29 and nC 31 show rather homogenous excursion between 3.0 ka and 0.5 ka in two core site (Fig. 8). The difference of compound-speci c n-alkane source can be expressed as δ 13 C ALK .
Relatively large difference in δ 13 C ALK (around − 1.5) appeared between 3.0 ka and 2.0 ka may imply differences of vegetation community at this interval, and correspond to paleovegetation change from C 4 plants to C 3 plants. As mentioned, this interval is regarded as wetter climatic interval, so depleted δ 13 nC 31 in SJP15-4 may support paleo-vegetation community changes. As mentioned paleovegetation index, Paq, also indicate C 3 plant expansion is observed between 2.5 ka and 1.0 ka. Therefore, compound-speci c δ 13 nC ALK strongly indicative of variations of compound source and supports individual n-alkane excursion roughly.

Implication for paleoclimate variations in East Asia
In East Asian region, the most well-known studies on paleoclimate variation during the past glacialinterglacial periods were carried out using stalagmites, which showed evidence of strong coincident variation with climatic variation in high-latitude ice core studies and East Asian monsoon variation (Xiao et al. 1999;Wang et al. 2005;Li et al. 2015). In these studies, n-alkanes proxies were also widely used to track past paleoclimate variation during the Quaternary period (Li et al. 2013;Street et al. 2013;Hyun et al. 2017). Several previous records have focused on Holocene climate changes by using indirect proxies such as detrital quartz in Yangtze River off inner shelf area (Wang et al. 2020), and in ux of quartz as an indicator of aeolian dust in Cheju Island, Korea (Lim et al. 2005;Lim and Matsumoto 2008). These studies used proxies associated with the East Asian monsoon system. Therefore, detecting any of these climatic events at a high-resolution scale is important for detailed assessments of paleoclimate variation in the East Asian monsoon area.
The concentration of n-alkanes, the nC 31 /nC 27 proxy, and the δ 13 C org values in this study show a distinguished variation since the mid-Holocene. Because these kinds of combined proxies imply humid conditions in East Asia (e.g., Li et al. 2013), our data showing a decreasing interval between 2.5 ka and 0.5 ka may correspond to a wetter climatic condition as indicated by Li et al. (2013) in East Asian region. In this study, a climatic event-like 2.0 ka duration wetter condition during the late Holocene (2.5-0.5 ka) is strongly coupled with decreasing ne quartz ux in Cheju Island (Lim et al. 2005). One of similar result can be found from the detrital quartz from the Yangtze River, off East China Sea (Wang et al. 2020) ( Fig. 9B). This earlier study indicated that enhanced and/or reduced EASM intensity strongly associated with southward shift of EASM precipitation during this interval. General heavier trend of δ 13 C org between 2.5 and 0.5 ka in our core records (Fig. 9E) may support this EASM intensity.
As a paleoclimate monsoon variation, our n-alkane data clearly show gradational variation during 2.5 ka to 0.5 ka. The total n-alkane concentration shows upward increasing from 2.5 ka, then peaked at 1.2 ka in SJP15-2 and at 0.2 ka in SJP15-4, and then start to decrease at 0.2 ka. Individual nC 22 and nC 24 records showed an abrupt increase between 2.5 ka and 0.2 ka in SJP15-4. These mid-chain n-alkanes that originate from oating plants indicate that oating macrophytes ourished at these periods. Lim et al. (2008; demonstrated that paleoclimatic changes in Korea may be associated with the East Asian monsoon and the D-O cycle based on the input of aeolian quartz records from the Cheju Island, South Korea. In these result, the paleovegetation of Cheju Island was strongly linked with East Asian monsoon climatic variation, and the quartz in ux has been in uenced by the pathway change of the westerlies. In our study, gradational changes between 2.5 ka and 0.5 ka may correspond to long range of global climatic variation, and might be associated with East Asian monsoon environmental changes since the mid-Holocene (Fig. 9C). The clear decrease of in ux of aeolian quartz begin at about 2.3 ka, and it show gradational decrease until 0.5 ka. The timing of this ne quartz input is rather later than our nalkane record. Nevertheless, assuming that decrease of ne quartz in ux as diminished westerlies, and thus more moisture, in turn, this decrease in ux of ne quartz may bring vegetation changes; the changes of total concentration of n-alkanes.
Large amounts of terrestrial leaf-wax n-alkanes would have been transported, particularly during glacial periods and the Holocene. Due to the dry conditions during glacial periods in East Asia, terrestrial nalkanes were transported far across the Paci c Ocean (Ohkouch et al. 1997) and the East Sea (Yokoyama et al. 2006). Also, aeolian dust from mainland China could be transported via the westerlies to Cheju Island (Lim et al. 2005;Lee et al. 2014) and even in present study area. A large amount of terrestrial leaf wax could have been derived from nearby land areas. Therefore, our record of n-alkane concentrations and their carbon isotopes re ects information about not only the in situ paleo plant community but also allochthonous terrestrial n-alkanes which were associated with east Asian monsoon climatic variation. Allochthonous leaf wax could be transported from nearby and faraway places, and these delivery variations could have been associated with local climatic changes, as well as East Asian monsoon variation.

Conclusions
The time-dependent distributions of sedimentary δ 13 C org (‰) and n-alkanes of organic matter were investigated to link paleoclimate and oceanographic changes since the mid-Holocene. The proxies (TOC, δ 13 C org , n-alkanes, δ 13 C ALK ), representative terrestrial biomarkers, and long-chain n-alkanes exhibited gradational variations. In particular, 2.5 ka, 0.5 ka are switching points, indicating that the input of terrestrial biomarkers has been changed in accordance with paleoclimate variation in East Asia.
The proxies, including ACL, CPI, and P aq , also strongly coincide with the excursions of long-chain nalkanes. In particular, nC 31 /nC 27 describe a parabola during 3.5 ka to 1.0 ka, implying gradual and wetter climatic conditions. Compound-speci c δ 13 C ALK indicates more clearly distinguished switching boundaries, and show clear difference of compound source between 3.0 ka and 1.0 ka. A synthetic comparison with previous works indicated that most paleoclimate proxies of stalagmites, ne quartz, and detrital quartz show coincident variation, and are well matched with our n-alkanes proxy. Therefore, this study provides information on paleoclimate variation that might have been regulated by centennial-scale dynamics such as the East Asian monsoon as well as local paleoclimate effects since the mid-Holocene in the southern inner continental shelf of the Korean peninsula. In this study, specially, distinctive wetter climate condition during Late-Holocene (2.0 ka ~ 0.5 ka) has been prevailed in southern inner shelf of Korean peninsula.   (Wang et al. 2020), and (C) is Cheju Island, Korea (Lim et al. 2005). KCC is Korean Coastal Current, TWC is Tsushima Warm Current, respectively.

Figure 2
Sedimentation rate of the two cores. Both core show homogenous mud sediment and have a similar sedimentological characteristics and high sedimentation rate. A hypothetical line (broken line) in (B) was dawn by the age gab between bulk organic carbon and shell fragments (see text).

Figure 3
Vertical distributions of TOC (%), C/N ratio, δ13Corg (‰) and δ15NTN (‰) in two cores. Data were drawn by three-point moving average. Distinctive two switching points (2.5, and 0.5 ka) in their excursion are shown by thin lines.

Figure 4
Vertical n-alkane distribution in two cores. The total concentration of n-alkanes (nC25-35) is calculated by sum from each nC25-nC35. Due to rather speci c excursion in even-numbered n-alkanes distribution, odd and even number n-alkanes pro les are illustrated together. Representative individual n-alkane concentrations in studied two core. These time-dependent differences in concentration were attributed to changes in paleo-plant population and to supply changes of n-

alkanes.
Page 24/27 Figure 6 Average chain length (ACL), carbon preference index (CPI) and paleo-vegetation index (Paq). Several representative combined n-alkanes indices were also illustrated. These distinctive changes in proxies were very clear between 2.5 ka and 0.5 ka, suggesting wetter climate condition and increase of C3 plant during this time.

Figure 7
Vertical pro les of odd-numbered compound-speci c carbon isotope. The excursions are quite different in δ13nC21, δ13nC23, δ13nC27, δ13nC33, δ13nC35 suggesting provenance changes due to the paleovegetation change or/and supply change due to climatic changes.

Figure 8
Comparison of single n-alkane compounds (nC29, nC31) and their compound-speci c carbon isotope.