Hydroclimatic variations in southeastern China during the 4.2 ka event reflected by stalagmite records

Although the collapses of several Neolithic cultures in China are considered to have been associated with abrupt climate change during the 4.2 ka BP event (4.2-3.9 ka BP), the timing and 20 nature of this event and the spatial distribution of precipitation between northern and southern China are still controversial. Until now, the hydroclimate of this event in southeastern China is still poorly known, except for a few published records from the lower reaches of Yangtze River. In this study, a high-resolution record of monsoon precipitation between 5.3 and 3.57 ka BP based on a stalagmite from Shennong Cave, Jiangxi Province, southeast China, is presented. Coherent variations in δO 25 and δC reveal that the climate in this part of China was dominantly wet between 5.3 and 4.5 ka BP and mostly dry between 4.5 and 3.57 ka BP, interrupted by a wet interval (4.2-3.9 ka BP). A comparison with other records from monsoonal China suggests that summer monsoon precipitation decreased in northern China but increased in southern China during the 4.2 ka BP event. We propose that the weakened East Asian summer monsoon controlled by the reduced Atlantic Meridional 30 Overturning Circulation resulted in this contrasting distribution of monsoon precipitation between northern and southern China. During the 4.2 ka BP event the rain belt remained longer at its southern position, giving rise to a pronounced humidity gradient between northern and southern China.

studied in the past 20 years. It was identified as an abrupt (mega)drought and/or cooling event in a variety of natural archives including ice cores, speleothems, lake sediments, marine sediments, and loess. This climate episode was associated with the collapse of several ancient civilizations and human migrations in many sites worldwide (e.g., Egypt, Greece, the Indus Valley and the Yangtze Valley) (Weiss et al., 1993;Cullen et al., 2000;Gasse, 2000;DeMenocal, 2001;Weiss and 40 Bradley, 2001;Thompson et al., 2002;Booth et al., 2005;Bar-Matthews and Ayalon, 2011;Berkelhammer et al., 2012;Ruan et al., 2016). Recently, the 4.2 ka event was defined as the lower boundary of the Meghalayan Stage by the International Commission on Quaternary Stratigraphy.
The chronology of this geological boundary is defined at a specific level in a stalagmite from northeast India (http://www.stratigraphy.org/). 45 The climate change associated with the 4.2 ka event also significantly affected ancient civilizations in China, resulting in the collapses of Neolithic cultures (Jin and Liu, 2002;Huang et al., 2010Huang et al., , 2011Wu et al., 2017). Most of these studies imply a temperature drop in continental China at about 4.2 ka BP (Yao and Thompson, 1992;Jin and Liu, 2002;Zhou et al., 2002;Xu et al., 2006;Yao et al., 2017;Zhao et al., 2017), but changes in the spatial distribution of 50 precipitation are also discussed (Tan et al., 2008;Huang et al., 2010Huang et al., , 2011Tan et al., 2018aTan et al., , 2018bWu et al., 2017). For example, a grain-size record from Daihai Lake, north China, suggests a decrease in monsoon precipitation between 4.4 and 3.1 ka BP with a very dry interval between 4.4 and 4.2 ka BP (Peng et al., 2005). Extreme flooding during the 4.2 ka event was identified by paleoflood deposits in the middle reaches of the Yellow River (Huang et al., 2010(Huang et al., , 2011. Wu et al. 55 (2017) reported evidence of two extraordinary paleoflood events in the middle reaches of the Yangtze River at 4.9-4.6 ka BP and 4.1-3.8 ka BP, closely related to the expansion of the Jianghan lakes. These extreme hydroclimate events may have accelerated the collapse of the Shijiahe Culture in the middle reaches of the Yangtze River (Wu et al., 2017). Multiple proxies in four stalagmites from Xianglong cave, south of the Qinling Mountains, indicate that the upper 60 Hanjiang River region experienced a wet climate during the 4.2 ka event (Tan et al., 2018a). Peat records provide a broad picture of climate variations in southeast China (SEC) during the Holocene (Zhou et al., 2004;Zhong et al., 2010aZhong et al., , 2010bZhong et al., , 2010cZhong et al., , 2015Zhong et al., , 2017. The resolution of these records, however, is not high enough to study the detailed structure of the 4.2 ka event. Until now, there is only one published stalagmite record from SEC (Xiangshui cave; Fig. 1), indicating a 65 wet interval during the 4.2 ka event (Zhang et al., 2004).
The aim of this study was to obtain a high-resolution stalagmite-based record from SEC to explore the hydroclimatic variations during the 4.2 ka event and to compare them to records in northern China in order to explore the possible north-south precipitation gradient during this event. 70 Clim. Past Discuss., https://doi.org/10.5194/cp-2018-116 Manuscript under review for journal Clim. Past Discussion started: 30 August 2018 c Author(s) 2018. CC BY 4.0 License.

Study area and sample
Shennong Cave (117°15' N, 28°42' E, 383 m a.s.l.) is located in the northeast of Jiangxi Province, SEC (Fig. 1), a region in the mid-subtropical zone that is strongly influenced by the East Asian summer monsoon (EASM). Mean annual precipitation and temperature at the nearest 75 meteorological station (Guixi;1951 are 1857 mm and 18.5 °C , respectively (Fig. 2a).
The rainy season includes both summertime monsoon rainfall and spring persistent rain (Tian and Yasunari, 1998;Wan et al., 2008;Zhang et al., 2018). The latter is a unique synoptic and climatic phenomenon that occurs from March until mid-May, mostly south of the Yangtze River (about 24°N to 30°N, 110°E to 120°E -Tian and Yasunari, 1998;Wu, 2007, 2009). EASM 80 precipitation lasts from mid-May to September (Wang and Lin, 2002). Shennong cave is located in the region of the spring persistent rain. Data from the nearest GNIP station in Changsha, also located in the region of the spring persistent rain, indicate that the EASM (May to September) precipitation with lower δ 18 O values (Figs. 2a,b) accounts for 54% of the annual precipitation and the non-summer monsoon (NSM) (October to next April) precipitation with higher δ 18 O values 85 (Figs. 2a, b) accounts for 46% (Zhang et al., 2018).
The cave was discovered by local people in 1998 after several days of continuous heavy rain.
The cave developed in Carboniferous limestone of the Chuan-shan and Huang-long Groups, which are mainly composed of limestone and interbedded dolostone. The thickness of the cave roof ranges from about 20 to about 80 m, with an average of ~50 m. The overlying vegetation consists 90 mainly of secondary forest tree species such as Pinus, Taxodiaceae and bamboo and shrub-like Camelliaoleifera and Ilex bioritsensis which are C 3 plants . According to two years ' monitoring (2011-2013), the mean temperature in the cave is 19.1 °C with a standard deviation of 2.5 °C (Fig. 2c), consistent with mean annual air temperature outside the cave (Fig.   2a). The relative humidity in the interior of the cave approaches 100% during the most of the year 95 ( Fig. 2c). Abundant aragonite and calcite speleothems are present in the cave. Their mineralogy is likely controlled by the Mg/Ca ratio of the drip water reflecting the variable dolomite content of the limestone (De Choudens-Sanchez and Gonzalez, 2009;Zhang et al., 2014Zhang et al., , 2015. All aragonite stalagmites were deposited within ~1.5 km of the cave entrance where the bedrock is dolomite and all calcite stalagmites were deposited in more distal parts of the cave where limestone constitutes 100 the bedrock. In November 2009 stalagmite SN17 (Fig. 3), 320 mm in length, was collected 200 m behind the cave entrance where the bedrock is dolomite. X-ray diffraction (XRD) analyses suggest that the stalagmite is composed of aragonite, except for the bottom section below 318 mm which is composed of calcite (Fig. 3). The calcite section was not included in the present study. 105 Clim. Past Discuss., https://doi.org/10.5194/cp-2018-116 Manuscript under review for journal Clim. Past Discussion started: 30 August 2018 c Author(s) 2018. CC BY 4.0 License.

230 Th dating
Eleven subsamples for 230 Th dating were drilled along the growth axis of SN17 (Fig. 3) with a hand-held carbide dental drill, and were dated on a multi-collector inductively coupled plasma 110 mass spectrometer (MC-ICP-MS, Neptune-Plus) at the Department of Earth Sciences, University of Minnesota and the Institute of Global Environmental Change, Xi'an Jiaotong University (Cheng et al., 2000(Cheng et al., , 2013. The chemical procedure used to separate uranium and thorium followed those described by Edwards et al. (1987).

Stable isotope analyses
120 sub-samples for stable isotope analyses were drilled along the central axis of SN17 at intervals of 1 mm between 0 and 70 mm and 5 mm between 70 and 320 mm distance from top, respectively. The samples were analysed using a gas-source stable isotope ratio mass spectrometer (Isoprime100), equipped with a MultiPrep system at the Institute of Earth Environment, Chinese 120 Academy of Sciences (IEECAS). The international standard NBS19 and the laboratory standard HN were analysed after every 10 sub-samples to monitor data reproducibility. All oxygen and carbon isotope compositions are reported in per mil relative to the Vienna Pee Dee Belemnite (VPDB). Reproducibility of δ 18 O and δ 13 C values was better than 0.1‰ and 0.08‰ (2), respectively. 125

Chronology
The 230 Th dates are all in stratigraphic order and no hiatus was observed ( Table 1). Because of the high uranium concentrations (1120-6380 ppb) and relative low thorium concentrations 130 (56-195 ppt), most of the dating errors are less than 6‰. We used a linear interpolation method to establish a depth-age model of stalagmite SN17 which grew from 3643 to 5299 a BP (Fig. 3).

δ 13 C and δ 18 O records
The temporal resolution of the δ 13 C and δ 18 O records ranges from 6 to 21 years. The δ 13 C 135 values fluctuate around -9.18‰ (mean value) during the period 5.3 to 4.5 ka BP, and increase to -8.69‰ between 4.5 and 3.6 ka BP (Fig. 4a). δ 18 O fluctuates around -6.75‰ (mean value) during the period 5.3 to 3.6 ka BP, without a significant long-term trend (Fig. 4c). The δ 18 O record is broadly similar to the δ 13 C record between 5.3 and 3.6 ka BP, with a significantly positive correlation (Fig. 5a covariance along the growth axis suggesting that the speleothem was not deposited under isotopic equilibrium (Hendy, 1971;Dorale and Liu, 2009). Several studies, however, demonstrated that stalagmites showing a significant correlation between δ 18 O and δ 13 C can also have formed under isotopic equilibrium, indicating they are controlled by the common influencing factors (Dorale et al., 1998;Dorale and Liu, 2009;Tan et al., 2018a). Another, more robust test is the replication of 145 δ 18 O records from different caves (Dorale et al., 1998;Wang et al., 2001;Dorale and Liu, 2009;Cai et al., 2010). The δ 18 O records of SN17 and of stalagmites from Dongge (Wang et al., 2005) and Xiangshui (Zhang et al., 2004) caves, located southwest of Shennong cave ( Fig. 1), show remarkable similarities during the overlapping interval (Fig. 6). The replication of these records further confirms that aragonite of stalagmite SN17 was most likely deposited close to isotopic 150 equilibrium, i.e., the δ 18 O variation primarily reflects climatic changes. The growth rate of SN17 shows a persistently decreasing trend from 0.62 to 0.034 mm/yr between 5.2 and 4.5 ka BP, followed by low values during the period 4.5 to 3.6 ka BP with relatively higher values between 4.25 and 4.0 ka BP (Fig. 4e).

Interpretation of δ 18 O and δ 13 C
The climatic significance of the δ 18 O parameter in speleothems from the monsoon region of China has been intensively debated in recent years, because it is influenced by several, partly competing factors including precipitation amount, moisture source and transport pathway, 160 upstream depletion and changes in the ratio of the amount of summer to winter precipitation, as well as the cave temperature (Maher, 2008;Clemens et al., 2010;Dayem et al., 2010;Pausata et al., 2011;Caley et al., 2014;Tan, 2014). Most scientists agree that speleothem δ 18 O values in monsoonal China represent variations in the EASM intensity and/or changes in spatially-integrated precipitation between different moisture sources and the cave site on orbital to 165 millennial timescales rather than local precipitation amount (Cheng et al., 2016). Some researchers suggest that the speleothem δ 18 O value in south China is not influenced by precipitation amount or EASM intensity but by moisture circulation on interannual to interdecadal (He et al., 2009;Tan, 2009Tan, , 2014 or even on centennial to millennial timescales (Tan, 2016). According to the specific seasonal distribution of precipitation amount and δ 18 O values in the region of spring persistent rain, 170 we suggest that the speleothem δ 18 O from E'mei cave, located at 160 km northwest of Shennong cave, is primarily controlled by the ratio of EASM precipitation amount to NSM precipitation amount (EASM/NSM) on interannual timescales (Zhang et al., 2018). In addition, the speleothem interannual to centennial timescales for the interval 1810-2000 AD (Zhang et al., 2018). This indicates that the speleothem δ 18 O values in Shennong cave, similar to that in E'mei cave, can be also influenced by the EASM precipitation amount on interannual to centennial timescales.
Speleothem δ 13 C is influenced by many processes from the soil via the vadose zone to cave environment (Genty et al., 2003;McDermott, 2004;Fairchild et al., 2006 and references therein). 180 Many studies suggest that changes in speleothem δ 13 C are generally controlled by vegetation density and composition in the catchment which vary according to the hydroclimate (Genty et al., 2001(Genty et al., , 2006Baldini et al., 2005;Cruz Jr et al., 2006;Fleitmann et al., 2009). Soil CO 2 , which is produced by root respiration and organic matter decomposition, dissolved in the drip water drives speleothem δ 13 C. In regions where the vegetation type is predominantly C 3 or C 4 plants, a dry 185 climate will lead to a reduction of the vegetation cover, density and soil microbial activity as well as an increase in the groundwater residence time allowing more δ 13 C-enriched bedrock to be dissolved. In addition, Prior calcite precipitation (PCP) in the vadose zone will result in higher δ 13 C values accompanied by increased Mg/Ca ratios in speleothems (Baker et al., 1997). Slow drip rates and increased evaporation and/or ventilation inside the cave will lead to higher δ 13 C values, 190 usually accompanied by kinetic isotopic fractionation (Fairchild et al., 2000;Frisia et al., 2011;Li et al., 2011;Tremaine et al., 2011).
In Jiangxi Province, a region presently occupied by mostly C 3 plants, no evidence has been found for a general pattern of replacement of C 3 plants by C 4 plants during the late Holocene (Zhou et al., 2004;Zhong et al., 2010b). There is no significant positive correlation between δ 13 C 195 values and Mg/Ca ratios between 5.3 and 3.6 ka BP (Fig. 5b). In Shennong cave ventilation is weak and relative humidity remains close to 100% throughout the year. Rapid CO 2 degassing is inhibited under these conditions. Stalagmite SN17 was likely deposited close to isotopic equilibrium, as confirmed by the replication test (section 4.2). The δ 13 C variations in this stalagmite were therefore not controlled by PCP or rapid CO 2 degassing. Rather, changes in δ 13 C 200 were driven by vegetation density and soil bioproductivity associated with hydroclimatic variations, with lower δ 13 C values corresponding to a dense vegetation cover associated with a wet climate and vice versa .

Hydroclimate between 5.3 and 3.6 ka BP 205
Previous speleothem studies from monsoonal China suggested a coherent trend of decreasing precipitation from the early to the late Holocene Wang et al., 2005;Hu et al., 2008;Cai et al., 2010Cai et al., , 2012Dong et al., 2010Dong et al., , 2015Jiang et al., 2013;Bai et al., 2017;Tan et al., 2018a), which follows the gradually decreasing Northern Hemisphere summer insolation. Our δ 13 C record exhibits a similarly increasing trend (Figs. 4a, b), indicating that the climate in our  (Figs. 2a, b) accounts for 46% (Zhang et al., 2018). Therefore, we found that speleothem δ 18 O in our study area are significantly influenced by the changes in the EASM/NSM ratio on interannual timescales (Zhang et al., 2018). A decreasing intensity of EASM might lead to changes in the EASM/NSM ratio and in turn result in variations in speleothem δ 18 O, however, it remains unclear how the EASM and NSM precipitation amount varied on centennial to millennial 220 timescale, respectively. Therefore, the long-term trend in δ 18 O is less significant than that in δ 13 C.
But we still find that there were more wet intervals (z-scored δ 18 O values > 0) between 5.3 and 4.5 ka BP than between 4.5 and 3.6 ka BP on centennial timescales (Fig. 4d). The δ 18 O record shows similar variations as the δ 13 C record (Figs. 4b, d), which are not related to seasonal variations of δ 18 O values in precipitation on centennial to millennial timescales. In addition, the SN17 δ 18 O 225 record is remarkably similar to the δ 18 O record from Dongge cave (Fig. 6), which is primarily controlled by summer monsoon precipitation Wang et al., 2005). These observations indicate that variations in SN17 δ 18 O might reflect changes in summer monsoon precipitation on decadal to millennial timescales, with higher (lower) δ 18 O values representing decreased (increased) summer monsoon precipitation. The conspicuously decreasing trend in 230 growth rate further confirms that the monsoon precipitation gradually decreased from 5.3 to 3.6 ka BP (Fig. 4e). During 4.5-3.6 ka BP, a wet interval between 4.3 and 4.05 ka BP can be identified in our δ 18 O record (Fig. 4d) on centennial timescales, consistent with high values of growth rate between 4.25 and 4.0 ka BP. The low values of δ 13 C record between 4.15 and 3.95 ka BP (Fig. 4b), however, lags the decreased δ 18 O record by 150 years. The asynchronously decreased variations 235 between the δ 13 C and δ 18 O may be ascribed to the delayed response of enhanced vegetation density to increased monsoon precipitation. Two decadal intervals with decreased δ 18 O values during the period 3.95 to 3.8 ka BP cannot be found in the δ 13 C record and the growth rate of SN17 (Fig. 4), indicating they are two insignificantly wet events. Therefore, we suggest that the climate in our study area between 5.3 and 4.5 ka BP was dominantly wet, and changed to a rather 240 dry climate between 4.5 and 3.6 ka BP with one wet interval between 4.3 and 4.05 ka BP (Fig. 4).
Our δ 18 O and δ 13 C records exhibit a two-stage structure during the 4.2 ka event, characterized by a wet phase with high variability between 4.3 and 4.05 ka BP and a distinctly dry phase from 4.05 to 3.6 ka BP. south of Shennong cave, indicates a dry climate between 6.0 and 4.0 ka BP, with an short-lived wet event at 4.1 ka BP (Zhou et al., 2004). Subsequent multi-proxy records of several new cores from this site also revealed a prevailingly dry climate between 6.0 and 3.0 ka BP with a wet interval at 4.2-3.9 ka BP (Zhong et al., 2010a(Zhong et al., , 2010b(Zhong et al., , 2010c) Speleothem δ 18 O record from Xiangshui cave also exhibits a remarkable similarity to the SN17 δ 18 O record, showing a wet 250 interval between 4.2 and 4.0 ka BP (Fig. 6). These published records from SEC are consistent with our isotope records within error, indicating a wet climate in SEC during 4.2 ka event.

Comparison with other records in monsoonal China covering the 4.2 ka event
The remarkable drought during 4.2 ka event was recorded by various archives from different 255 sites in northern and southwestern China (Tan et al., 2008;2018;Zhao et al., 2010;Chen et al., 2015;Goldsmith et al., 2017 and references therein). In contrast, the extraordinary flood during the 4.2 ka event was identified in the middle reaches of the Yellow River in north-central China (Huang et al., 2010(Huang et al., , 2011 and the middle reaches of the Yangtze River in south-central China (Wu et al., 2017). For SEC, δ 15 N and δ 13 C records from the Daping swamp (Fig. 1)

reveal a wet 260
interval of 4.5-4.0 ka BP (Zhong et al., 2017), however, other proxies from the same site indicate cool and dry climate between 4.5 and 3.5 ka BP (Zhong et al., 2015). Pollen data of two sediment profiles from the Daiyunshan Mountain (Fig. 1), SEC, indicate a centennial-scale wet event at 4.4 ka BP (Zhao et al., 2017). A dry and cold period during 4.2 ka event was identified in a sediment profile near Taihu Lake (Fig. 1), East China (Yao et al., 2017). To sum up, records from lacustrine, 265 peat and paleoflood sediments suggest that the climate during the 4.2 ka event was dry in northern and southwestern China and wet in central and southern China (Fig. 8), although there are discrepancies in some records from southern China, which might be caused by the large dating uncertainties and the low resolution. Tan et al. (2018) has already proposed the same -north-dry and south-wet‖ pattern by reviewing records from the monsoon region of China; however, more 270 speleothem records from SEC are needed to further confirm this pattern.
In the following discussion only high-precision and high-resolution speleothem records are used. Two stalagmite δ 18 O records from Nuanhe and Lianhuadong caves in northern China indicate that the climate between 4.0 and 3.5 ka BP was very dry (Tan and Cai, 2005;Dong et al., 2015). Another two stalagmite δ 18 O records from Jiuxian and Xianglong caves south of the 275 Qinling Mountains revealed increased monsoon precipitation in central China during the 4.2 ka event (Fig. 7a, b; Cai et al., 2010;Tan et al., 2018a). Stalagmite δ 18 O records from Sanbao ( Fig. 7c; Dong et al., 2010), Heshang ( Fig. 7d; Hu et al., 2008) and Lianhua (Zhang et al., 2013) caves in the middle reaches of Yangtze River, south-central China, also indicate a wet interval between 4.15 and 3.9 ka BP, which is consistent with a δ 13 C record of peat from the Dajiuhu basin (Figs. 7e;280 Clim. Past Discuss., https://doi.org/10.5194/cp-2018-116 Manuscript under review for journal Clim. Past Discussion started: 30 August 2018 c Author(s) 2018. CC BY 4.0 License. Ma et al., 2008) in the same region. The SN17 δ 18 O record also reveals a wet interval between 4.3 and 4.05 ka BP on centennial timescales (Fig. 7g). These records reveal a coherently dry period in northern China but a wet climate in south-central and southeastern China during 4.2 ka event (Fig.   7).  (Wang et al., 2001). During these 290 intervals, the rain belt migrated southward and remained longer at this position, which reduced rainfall in northern China but enhanced rainfall in central and southern China (Tan et al., 2011(Tan et al., , 2018aZhang et al., 2014;Chiang et al., 2015).
The SN17 δ 18 O record exhibits a coherent variation with the δ 18 O record from Dongge cave within error, showing four wet intervals between 4.6 and 3.8 ka BP on centennial timescales (Figs. 295 7f, g). These four wet intervals were also identified in other stalagmite δ 18 O records, although there are some differences in the amplitude (Fig. 7). For example, the first wet interval in the SN17 δ 18 O record corresponds to the dry interval in the δ 18 O records from Jiuxian and Xianglong caves. A reason for this discrepancy might be chronology offsets, because these records are constrained by two dates only. Alternatively, monsoon precipitation in our study area during this 300 period was out of phase with those in the region around Jiuxian and Xianglong caves. Because at 4.55 ka BP the AMOC was in a very weak stage, this could have resulted in a weakened EASM and a further southward shift of the monsoonal rain belt, possibly causing a dry climate in south-central China and a wet climate in SEC. The different amplitudes of the other three wet intervals in these speleothem records from central to southern China might have been caused by 305 the spatial distribution of monsoon precipitation, reflecting the position and residence time of the rain belt associated with variations in EASM intensity.

Conclusions
We reconstructed monsoon precipitation variations in the lower Yangtze River region, SEC, 310 conducted the 230 Th dating. All authors discussed the results and provided input on the manuscript.