Abstract
An archaeological site at Anshan in the coastal area of Fujian province, southern China, was excavated in 2007, 2009 and 2015. Abundant artefacts including adzes, cores, bronze fishhook, pottery and bone arrowheads are found in the aeolian sediments. The aim of this article is to understand the geomorphological backdrop and process of Anshan site, and the coupling relationship between human activity and environmental evolution. In this study, optically stimulated luminescence (OSL) technique was employed to establish the chronological framework of the site. Samples from the top and bottom of cultural layer yield OSL ages ranging from 1.4 to 6.1 ka, providing a systematic geochronological evidence for the development of ‘Anshan culture’ in coastal area of southern Fujian province and eastern Guangdong province. In the meantime, there is a clear link between the varying regional sea levels, the chronology of regional wind-sand deposition and the period of Anshan culture since the mid-Holocene.
Keywords
Introduction
The Anshan site (N24°39′22.5724″, E118°37″21.1704″) is located in the southern coastal area of Fujian province in southeastern China about 1.0 km far away from Taiwan Strait rim (Figure 1). The Anshan site was discovered in 2007 and excavated in 2007, 2009 and 2015, respectively. It is a new and formal excavation of Neolithic and Bronze Age site in recent years in Fujian province, which filled the gaps of Neolithic and Bronze Age archaeology in coastal area of Fujian province. We named it as ‘Anshan Cultural Type’ because it is one of the aboriginal cultural representatives of Neolithic and Bronze Age in coastal area of southern Fujian province and eastern Guangdong province. The archaeological site flanking the coastline of East China Sea and deposit of sea shells are the evidence for the long-term use of marine resource and coastal environments in civilization development and subsequent evolution, which is important to understanding patterns of human subsistence.

The geographical location of Anshan site.
Settlements of Fujian’s ancient humans usually located in the top of hilly or bench terrace, so did Anshan site (Figure 2). The section of stratum in archaeological excavation at Anshan site is aeolian sand deposit with a total thickness of ca. 7 m, the bottom of which is red sand terrace. The site contains rich sandy pottery, hard pottery, proto porcelain and bronze ware, and the remains of building material are ‘old red sand’ from the second terrace. The cultural strata is covered by a sediment sequence up to 1.2 m thick, extending from 22 to 23.2 m a.s.l., and a large number of shells are continuously distributed over the cultural layer vertically which are used for 14C dating. The previous results show that the cultural strata sustained over a long period from 3.4 to 2.7 ka (Fan et al., 2014), but the geomorphological backdrop and process of Anshan site are not clear, which is epochmaking to understanding the coupling relationship between human activity and environmental evolution.

Geomorphological sketch map of Anshan Neolithic and Bronze Age site: (a) front view, (b) side view, (c–f) excavations are bronze fishhook, pottery Fu, pottery jar and bone arrowheads, respectively, and (g) a photography of research site in front view.
The aeolian sand deposit lacked suitable material for 14C dating, but optically stimulated luminescence (OSL) technique overcomes the shortcoming of radiocarbon method and has already become a widely used method for estimulating the ages of aeolian sediment (Athanassas and Wagner, 2016; Cunningham and Wallinga, 2009; Jacobs, 2008; Li et al., 2015; Madsen and Murray, 2009; Murray and Funder, 2003; Murray and Olley, 2002; Peng et al., 2016; Reimann et al., 2012). In this study, OSL signals from quartz by the single-aliquot regenerative-dose (SAR) protocol were applied on the samples to acquire the age of the Anshan aeolian dune site (Table 1).
SAR protocol used for age determination.
SAR: single-aliquot regenerative-dose.
Sample collection and measurements
Luminescence dating offers one benefit that allows the direct determination of buried ages for sediments from a wide variety of depositional environments (Murray and Olley, 2002). SAR dating protocol for quartz is used to estimate the equivalent dose (De) values (Wintle and Murry, 2006). In this study, OSL dating has been employed to date the aeolian sand collected from the Anshan site. This method has been widely applied to date coastal sand and aeolian deposits throughout the world (Hilgers et al., 2001; Tamura et al., 2011; Zhang et al., 2008). The aim of this study is to examine the periods of sand accumulation based on quartz OSL dating. In addition, we also propose to discuss the relationship between human activities and their associations with sand activities and climate change.
Nine OSL samples were collected to setup a chronological framework for the sediment of the Anshan profile in 2015, and the approximate positions of each sample are shown as white filled circles in Figure 2. Three samples (ASS1-05, ASS2-01 and ASS2-02) covering the time interval of cultural layer and other samples from overlying and underlying sediment were all taken with stainless steel tube and black plastic tape to provide protection against light and breakage during collection procedure and transportation.
All preparations were carried out under the subdued red-light conditions using standard methods in the laboratory (Aitken, 1985, 1998). Medium-grained quartz was extracted and used for equivalent dose test. The chemical treatment with hydrochloric acid and hydrogen peroxide was employed to dissolve carbonate and organic matter, respectively. The separated 58- to 63-µm fraction was then corroded using hydrofluoric acid for 30 min to leach quartz fractions, and then dissolved with 10% HCl to remove any fluorides and rinsed with distilled water several times. The purity of quartz for each aliquot was checked using the infrared (IR) depletion ratio (Duller, 2003), and no significant IR signal was observed, indicating that the isolation of quartz was successful.
An automated Risø TL/OSL DA-20 reader installed with a 90Y/90Sr beta source (Bøtter-Jensen et al., 2003) was used for quartz equivalent dose test. The OSL signals were stimulated with blue LEDs (470 ± 30 nm) and the detection of the corresponding quartz OSL signals was in the ultraviolet (UV) through a 7.5-mm Hoya U-340 filter in front of an EMI 9235QA photomultiplier tube (Duller, 2003). The SAR protocol (Murray and Wintle, 2000) was applied to the quartz OSL measurement. In order to obtain a suitable thermal treatment for the SAR protocol, a series of conditional tests, such as preheat plateau, dose recovery and thermal transfer tests, were applied to three representative samples (L201572, L201573 and L201576) previous to dating (Roberts, 2006). Based on the results of these tests (Figure 3), a preheat temperature of 260°C and a cutheat at 220°C were used for three samples (L201571–L201573), and a preheat temperature of 200°C and a cutheat at 160°C were selected for other samples (L201574–L201579), and then quartz OSL samples were stimulated by blue light at 125°C for 40 s. In this study, 16–30 aliquots for each sample were employed for De measurement. The first 0.8-s integral of the initial OSL signal minus a background estimated form of the last eight integrals was used for De estimation (Jin et al., 2016).

Solid circles display equivalent dose as a function of preheat temperature (cutheat temperature equal to preheat temperature minus 40°C, and except for the 180°C preheat where the cutheat temperature tracks the preheat temperature by −20°C). Open squares and circles refer to the recycling ratios and recuperation values.
Dose rates of the samples were calculated by measuring the activities of 238U, 232Th and 40K in the surrounding sediment by neutron activation analysis (NAA) method. The conversion factors (Guérin et al., 2011) and the beta-attenuation factors (Mejdahl, 1979) were used to calculate the beta and gamma dose rates of the sediment surrounding. An α-value of 0.04 ± 0.02 adopted for quartz was assumed to estimate the alpha contribution to the dose rate (Rees-Jones and Tite, 1997). A water content of 8 ± 5% was assumed for the total dose rate calculation (Aitken, 1985). The contribution of the cosmic dose rate to the total dose rate was calculated according to Prescott and Stephan (1982) and Prescott and Hutton (1994).
Besides, nine samples were used for grain size measuring. The grain size of samples was measured using a Malvern laser grain size analyser with a range of measurement of 0.02–2000 µm (Jin et al., 2016). The final results of grain size distribution are shown in Figure 4.

Grain size distribution frequency and triangular plot of nine OSL samples.
Results
All results of the preheat plateau tests, dose recovery tests and thermal transfer tests for sample L201572, L201573 and L201576 are presented in Figure 3. For sample L201572 (Figure 3a), the Des are very sensitive to the temperature variation and no marked De plateau was observed. The Des increased with increasing preheat temperature until 200°C and thereafter decreased again up to 260°C. The dose recovery ratios show a similar tendency as those of the preheat plateau test. The recuperation was consistently lower than 3% for the entire temperature range. The recycling ratios are usually near by one within 5% of unity. For sample L201573 (Figure 3b), the mean Des showed a wide plateau under the condition of different preheat temperatures. The recycling ratios rather focus on scope between 0.99 and 1.07. And the recuperation was less than 2% for the whole temperature range. For sample L201576 (Figure 3c), the mean Des showed a wide plateau across the whole temperature range of 180–280°C. The recycling ratios rather focus on scope between 0.92 and 1.06. It is noted that the OSL intensity was not remarkably changed when the preheat temperature is above 220°C. The high preheat temperature caused no conspicuous sensitivity increase in the quartz OSL signal for sample L201576. The recuperation was less than 4% for the whole temperature range.
Table 2 shows a summary of dose rate, De values and ages of the nine quartz OSL samples with 16–30 aliquots measured of each sample. As shown in Table 1, the dose rates do not change significantly with sampling depth except samples (L201578 and L201579) giving a total dose rate of 0.98 ± 0.04 and 0.93 ± 0.05 Gy/ka, respectively, which are lower than that of the other samples ranging from 1.34 ± 0.05 to 1.80 ± 0.07 Gy/ka.
U, Th and K concentration and depth of the samples collected from the Anshan site.
The Des for all samples range between 0.32 ± 0.01 Gy and 196.73 ± 6.65 Gy. The OSL ages are consistent with the stratigraphic order for the section and vary between 0.15 ± 0.01 and 127.32 ± 6.62 ka. A total of seven OSL ages from the Anshan site correlate to the Holocene. Two samples from the top of the first terrace and the bottom of the second terrace yielded ages of 127.32 ± 6.62 ka (L201571) and 49.32 ± 2.05 ka (L201572). The modern aeolian sand samples L201574–L201577 and L2.1579 were dated to be 0.44 ± 0.03, 0.31 ± 0.03, 0.35 ± 0.02, 0.15 ± 0.01 and 0.34 ± 0.02 ka, respectively, which are correlated to the ‘Little Ice Age’ (LIA).
At the same time, a sample of shell from the top of the cultural layer was dated by AMS 14C method at the Department of Archaeology, Peking University. Table 3 provides the details of the data. The sample fielded age of 1340 ± 35 ka, and the ages after tree-ring calibration are correlated to AD 630 (95.4%) AD 780, also calling for warm period of Sui-Tang Dynasty in China (Ge, 2011).
The AMS 14C age of shell on the top of culture layer.
Discussion
OSL characteristics
Figure 5 shows the OSL decay, dose response curves and frequency distribution histograms for three representative samples (L201572, L201573 and L201576). Most of quartz OSL dose response curves can be usually well fitted either with a linear or a single saturation exponential function (Wintle and Murray, 2006). OSL decay curves and dose response curves for these three representative samples show the luminescence characteristics. Figure 5a–c show the natural OSL decay curves for three representative samples. The OSL signals decrease very quickly during the first second of stimulation and seemingly dominated by the fast component (Jain et al., 2003). The representative growth curves are well represented by exponential function (Figure 5d–e) or linear function (Figure 5f) with five regeneration dose points, including a zero dose for the measurement of recuperation. Figure 5g–i presents the De distributions for the three samples, which suggest that the central age model (CAM) can be used for age calculation (Galbraith et al., 1999). In addition, sample L201576 has low intensity with a natural signal less than 1000 counts/0.16 s. Previous studies show that the lower OSL sensitivity of unheated quartz grains is probably related to a short sedimentation history of the particles from the source region to depositional site (Pietsch et al., 2008; Fitzsimmons et al., 2010). Nevertheless, there are no spatial and temporal patterns on the variations of OSL sensitivity (Zhang et al., 2015).

The OSL characteristics of samples (L201572, L201573 and L201576) from Anshan site: (a)–(c) component separation of CW-OSL signals from natural quartz grains. F, M and C are the fast, medium and constant components, respectively; (d)–(f) dose response curve obtained with quartz SAR protocol. Lx/Tx represents the sensitivity-corrected luminescence intensity to the test dose. The open circle and square symbols represent regeneration doses and the equivalent dose, respectively. The curves fitted with a single saturating exponential function (I = I0(1 − exp[−(D)/D0])) or a linear function (I = I0 + kD); (g)–(i) frequency distribution histogram of Des with 16–30 aliquots measured for each sample.
In our study, sample L2015071 yielded De above 120 Gy, reaching the critical dose range which is strongly inclined to underestimate the true ages according to the previous studies (Zhang et al., 2008). The dose response curve can be fitted better with a single saturating exponential plus linear function (I = I0(1 − exp[−(D)/D0]) + kD) better with a single saturating exponential function. The additional component in the high-dose region has been considered not reliable in producing exact equivalent dose estimates and resulting in age underestimation (Zhang et al., 2015). So far, the mechanisms of the additional linear growth related to age underestimation are still in controversy. De < 2D0 can be used as a boundary line for reliable age estimates, while the OSL signal should be about 15% below the saturation value obtained in the laboratory dose response curve (Wintle and Murray, 2006). The 2D0 from the exponential fitting part is 36 Gy for sample L201571, which is much lower than the corresponding natural De. The nature doses derived from an additional linear component in the high-dose region, which might cause an age underestimation according to the previous studies (Zhang et al., 2015). So, we should employ this age cautiously for interpreting the background of wind-sand activity.
Chronology of deposition
OSL dating at Anshan site allows the chronologic reconstruction of the widespread accumulation processes during the late Pleistocene at coastal area of Fujian province, China. The morphological units formed between 0.15 ± 0.01 and 127.32 ± 6.62 ka.
The outcrop located at the southwest of Shenhu Bay contains three major units of different sediments (Figure 2). The whole outcrop formed over a time horizon of ~130 ka, with the youngest age of 0.15 ± 0.01 ka derived from the top sample of the modern aeolian sand unit. Ages increase with the depth up to 127.32 ± 6.62 ka for the lowermost sample of the first terrace.
Based on sedimentological and grain size analyses (Figure 4), it is likely that the first terrace was not transported and deposited by wind, but instead within a coastal environment. And this unit consists of 70–90% fine sand with greyish yellow colour. The clay particle of this unit is the product of feldspar weathering (Li et al., 1988). The OSL sample L201571 (~19.5 m a.s.l.) taken from this unit was dated to 127.32 ± 6.62 ka. At that time, the eustatic sea level was above the present level. There is a large-scale marine transgression about −125 ka BP in many parts of the world (Church et al., 2008; Lambeck et al., 2002), but not any sedimentological evidence were recorded in China’s Fujian province so far. Maybe the ‘old red sand’ from the first terrace is the record of high sea levels at conversion period of marine isotope stage (MIS) 5e/6. It is likely that the ‘old red sand’ from the first terrace was accumulated under an estuarine environment, by a combination of marine and fluvial processes.
The sediment of the second terrace mainly composed of medium sand and fine sand with brownish red colour. The samples (L201572, L201573 and L201578) were taken from a homogenous sandy layer in heights of 20.5, 21.5 and 22.5 m a.s.l. and yielded depositional ages of 49.32 ± 2.05, 6.12 ± 0.52 and 1.68 ± 0.12 ka, respectively. The well-sorted, sandy material shows accumulation by aeolian processes. The provenance of the second terrace could be the former beach sand (Li et al., 1988). The depositional process indicates that the second terrace was probably deposited in a beach environment, while the lowest limit of ‘old red sand’ accumulation can indicate the former sea level approximately. In this sense, the bottom margin of ‘old red sand’ regional distribution can provide a gist to determine the former sea levels. In addition, it is the first ever Holocene ‘old red sand’ to be detected from South China. The minimum OSL data from published reports are 12.03 ± 1.60 ka (Hu et al., 2013). The sediment formed in period of Holocene is usually eroded away by wind and rain fall in coastal area of South China. But to a certain degree, human activities (including shell pits, foundations of the buildings, old red sand mounds and activity area) can protect the Holocene ‘old red sand’ from erosion.
The top of the site was deposited by aeolian processes between 0.44 ± 0.03 and 0.15 ± 0.01 ka, and the ages are concentrated in the period of the 16th century and the 17th century, which corresponds to the maximum stage of LIA, that is, the aeolian sand was very likely formed in association with the sediment supply and climatic changes of LIA. Grain size analysis exhibits absolute advantage of medium grained sand of at least 70%; finer material has most presumably been picked out.
It can be seen from the above analysis that nine ages correspond to the high sea levels of MIS5e, MIS3 and MIS1, respectively, which can verify the results on the deposition of old red sand (Zhang et al., 2008). And yet, more details need to do more OSL experiments in the future.
Relationship between human coastal migrations and wind-sand activities
Until recently, archaeologists and historians have been short of information about short-term climate change and wind-sand activities during the period of human societal evolution. High-resolution paleoclimate records from ice cores (Bond et al., 1993; Dansgaard et al., 1993; Heinrich 1988), tree rings (Briffa et al., 1990; Fang et al., 2015), stalagmites (Jiang et al., 2012; Wang et al., 2008) and some deep-sea sediments (Tamburini et al., 2003) make it clear that climatic shifts did occur within the Holocene and that often coincided with twists and turns in human history (Haug et al., 2003). Figures 6 and 7 and Table 4 show a clear link between the varying regional sea levels (Zeng, 1991), the chronology of regional wind-sand deposition and the period of Anshan culture since the Holocene.

The relationship of OSL ages and the horizons of cultural layer at Anshan site. Open circles represent the ages obtained by the SAR protocol.

The relationship of the duration of Anshan culture, regional sea levels and East Asia Winter Monsoon. Open circles and squares represent the ages from Anshan site; solid circles and squares represent the ages from Changle site (Jin et al., 2015); (a, b) sea levels of south and north of Fujian province, China (Zeng, 1991); (c, d) susceptibility of the sediments of Lake Huguang Maar in coastal southeast China (Yancheva et al., 2007).
Chronological sequence of historic cultures in coastal area of Fujian (the symbol ‘?’ represents controversial cultural period).
The duration of Anshan culture sustained a long time. The Anshan site is close to the coastal area of Taiwan Strait and is the well-dated shell mound/aeolian sand dune site in coastal area of China at present, which had powerful effects on human migrations, transmission routes of southland phylum and cultural exchange among southeast China, southeast Asia islands area and Pacific region, providing a suitable living environment for living beings during the era of five emperors (6–4 ka) and Xia-Shang-Zhou period (4–2 ka). So, based on the existing research results (Fan et al., 2014; Fujian Museum, 2008), the Anshan Culture has very likely endured from ~4000 BC to AD ~600 with a gap period of about 1.4 ka from 800 BC to AD 600. There is a period of human activities after AD 600, and the duration is not clear and should not exceed AD 1600. Late period of five emperors (~2000 BC) is referred to as Chalcolithic. Amount of cultural relics (including polished stone implements, chipped stone implements and bronzes) excavated from Anshan site are in accordance with the time feature of Chalcolithic. The period (1400–800 BC) is good time for living beings and contains rich shells, artefacts and bronzes.
The rapid falling of regional sea level is very likely the key factor of cultural discontinuity in research area after 3 ka. The crux of survival strategy was acclimatization when the human’s influence on natural environment was not so strong. Viewed from the excavations (stones with artificial pit, stone plummets of fishing net with trace by the cord, implements and adornments made of sea shells and high amount of shells), the food-getting behaviours of ancient humans are fishing and shell dredging (Fan et al., 2014). The sea level was higher than now, and the settled terrace was bordering the Shenhu bay, which was the best place for fishing and shell dredging. Subsequent rapid falling of sea level made the coastline away from the site about 5 km. It created inconvenience to ancient human in getting food. We hypothesize that it is this change of living conditions that the ancient inhabitants of Anshan site were forced to migrate the settlements. The next continuous period of low sea level is likely the major reason of the cultural gap about 1.4 ka from 800 BC to AD 600. The gap does not necessarily means that there is no civilization in this period. It is also possible that these people migrated from the research site to the zones near the coastline of the time. It is likely that as a result of transition of coastline, that is, to say, dwellings and other relics were levelled or severely damaged by the next immediate transgression.
The site was utilized again under the influences of emigration and the change of farming methods. This cultural deposition started the Wei, Jin, South and North Dynasties (AD 220–589) and ended the middle period of Tang Dynasty (AD 630–780). The Wei, Jin, and South and North Dynasties is a rare era of turbulence and is one of the three immigrant peaks in Chinese history (the other two are the later period of Tang Dynasty and the switching period of North and South Song Dynasties). The Han ancestors were the main force of southward migration. In the meantime, the agricultural productivity was improved. ‘Qi-people Essential’ precisely was a remarkable work of this period, which systematically summarized the producing skills of dealing with climate changes after later period of North Dynasty. The dependency of ocean was reduced to a certain extent based on the development of agricultural technology. And ancient human have the habit of living on the top of hills or terraces (Wu, 2004); therefore, the old red sand terrace was utilized for living again despite away from coastline.
In the mid-late period of Tang Dynasty (AD 741–907), a cold climate accompanied social instability, and vegetation fraction was decreasing rapidly (Ge et al., 2003), which is the main abandonment cause of ancient site. After Song Dynasty, this site gradually turned into an area with wind erosion and reactivation of sand dunes, combined with the rapid growth of population density. Then, the site was buried by sand and discarded.
Conclusion
The Anshan site built upon the ‘old red sand’ terrace was formed since 127 ka. The aeolian deposition of this site can be clearly divided into three sedimentary units corresponding to three high sea-level periods of MIS5e, MIS3 and MIS1, respectively. The human activities among the site’s duration indirectly protect the Holocene ‘old red sand’ from erosion, which is the most intact and systematic chronosequence of ‘old red sand’ in coastal area of southern China. Based on both the data, this research and previous research, the duration of Anshan archaeological culture continued at least in Wei, Jin, South and North Dynasties and can be divided into two stages: one is the high sea-level stage about 3 ka and the other is the lower sea-level stage between 1.8 and 1.4 ka. The agricultural production level had a visible difference in these two stages, that is, the human gradually reduced the dependence on ocean in the last stage compared with the former one. The site was gradually becoming extinct after Wei, Jin, South and North Dynasties, and the reasons were not clear. Next, bedded aeolian sand piled up on the abandoned site due to the enhanced sand-drift activity of LIA. More age data of this area should be added in the future if we want to see more details.
Footnotes
Funding
This work was financially supported by the National Natural Science Foundation of China (41301012, 41271031 and U1405231).
