ALTERNATIVE RADIOCARBON AGE-DEPTH MODEL FROM LAKE BAIKAL SEDIMENT: IMPLICATION FOR PAST HYDROLOGICAL CHANGES FOR LAST GLACIAL TO THE HOLOCENE

Abstract

We present an alternative radiocarbon (¹⁴C) age-depth model using IntCal20 to calibrate new accelerator mass spectrometry (AMS) data applied to a Lake Baikal sediment core (VER99G12) in southern Siberia. ¹⁴C dating showed that the core extends to 31 ka. To take into account uncertainties in ¹⁴C age and sedimentation depth in the core, a new age-depth modeling routine, undatable, was used in this study. Undatable revealed that significant changes in sedimentation rate correspond to global climate events, either warm or cold, which periods are likely close to the timing of the occurrence of the Meltwater pulses (MWP) at 19 and 14 ka, and the Last glacial Maximum (LGM) at 21–20 ka. Since the Selenga River accounts for 50% of the total river inflow to Lake Baikal, we interpret that these changes in sedimentation rate could be signals of significant changes in Selenga River discharge to the lake, which is expected to be affected by global climate change. Based on pollen analysis, it is highly probable that the sudden influx of the Selenga River to Lake Baikal, particularly at 19 ka, was due to the thawing of permafrost water through the Selenga River, which had developed in the region. Total organic carbon content and mean grain size increases concurrent with sedimentation rate, suggesting river inflow increased available nutrients for biological activity. Our results indicate that hydrological changes corresponding to MWP events can be observed in continental areas of the Northern Hemisphere.

INTRODUCTION

Lake Baikal, which is located in the south Siberian region (Figure 1), is the world’s oldest (at least 30Ma) and deepest (1648 m) lake with the largest water volume (23,000 km³), which represents ∼20% of the total unfrozen freshwater on the earth. Because long, continuous past environmental records (Kashiwaya et al. 2001) in the south Siberian region are preserved at its basin, numerous studies using lake sediment cores from Lake Baikal have been carried out to understand the past climate and environmental histories in the south Siberian region (Colman et al. 1995; Horiuchi et al. 2000; Kashiwaya et al. 2001; Karabanov et al. 2004; Prokopenko et al. 2006; Shichi et al. 2007, 2013; Tani et al. 2009). The inseparable linkage to the global climate changes and orbital climate forcing, such as glacial and inter glacial climate cycles and the Milankovitch cycles, respectively, have been revealed (Colman et al. 1995; Kashiwaya et al. 2001; Ochiai and Kashiwaya 2003, 2005; Prokopenko et al. 2006). Therefore, Lake Baikal has been regarded as an iconic site in the Siberian region for scientific study (Arzhannikov et al. 2018).

Figure 1 Map showing (A) the Eurasian continent, (B) Lake Baikal and its watershed and (C) Lake Baikal and the sampling site of the core VER99G12. The maps of Lake Baikal and its watershed were created with reference to Kuzumin et al. (2000). Lake Baikal is located at middle latitudes of the Eurasian continent (51.5–55.8°N, 103.7–109.0°E) and stores the largest volume of freshwater on Earth.

To reconstruct the paleoenvironmental changes using a lacustrine sediment core, the establishments of a precise age model is essential. Approaches for Lake Baikal age-depth models, especially for the late Quaternary period, are based on radioactive nuclides, such as ¹⁰Be, ¹⁴C, ¹³⁷Cs, ²¹⁰Pb, ²³⁷Am, and U/Th (Horiuchi et al. 2003; Chebykin et al. 2007; Watanabe et al. 2009a; Nara et al. 2010; Swann et al. 2018), paleomagnetic records (Antipin et al. 2001; Demory et al. 2005), and orbital tuning (Ochiai and Kashiwaya 2005). Among of them, radiocarbon (¹⁴C) age models have been widely applied to determine the deposition age of the Lake Baikal sediment during the last glacial to the Holocene (Colman et al. 1996; Horiuchi et al. 2000; Prokopenko et al. 2001; Soma et al. 2006; Nara et al. 2014). Lake Baikal sediments lack biogenic and authigenic carbonate for ¹⁴C measurement, as well as the plant macrofossils, leading to ¹⁴C dating approaches based on using total organic carbon (TOC).

The first reported ¹⁴C age-depth model of the Lake Baikal sediment cores (Colman et al. 1996) showed the ¹⁴C age-depth profiles measured by accelerator mass spectrometer (AMS) at two iconic sampling sites from Academician Ridge and Buguldeika Saddle (Figure 1C). These sites are topographically high and divided the basins into the Northern and the Central Basins for Academician Ridge, and into the Central and Southern Basin for Buguldeika Saddle (Figure 1C). Colman et al. (1996) concluded that the reservoir effect is limited to about 1000 ± 500 ¹⁴C yr for these sites, because of the most of organic carbon in Lake Baikal is autochthonous. The ¹⁴C age-depth profiles at these sites described the constant linear sedimentation rates of about 4 cm/kyr and 14 cm/kyr from Academician Ridge and Buguldeika Saddle, respectively (Colman et al. 1996). Further intensive study with high temporal resolution ¹⁴C analysis from three Lake Baikal sediment cores at Academician Ridge was conducted by Watanabe et al. (2009a) to develop a precise ¹⁴C age model of TOC. Their work revealed the known ¹⁴C activity plateau at the Younger Dryas (YD)/Preboreal (PB) boundary, which resulted from the changes in the atmospheric radiocarbon concentration during the cooling period. This ¹⁴C plateau allowed for the application of radiocarbon wiggle-match dating, providing an estimation of the reservoir effect of ∼2100 ¹⁴C yr for the site at Academician Ridge. Since ¹⁴C ages of TOC in the lacustrine sediments can be influenced by the lake reservoir effect, the hard-water effect, and the effect of the terrestrial organic matter (Watanabe et al. 2009a), it is very important to consider their effects on the ¹⁴C ages of TOC in order to obtain and accurate ¹⁴C age-depth model from the Academician Ridge sediment core.

Nara et al. (2010) previously showed a calibrated ¹⁴C age-depth profile measured from TOC and pollen grains from core VER99G12 which was retrieved from Buguldeika saddle and spanned for the past 33 cal kyr BP (Figure 1C). Although changes in the organic carbon source at the corresponding periods would also alter the ¹⁴C age, the ¹⁴C ages of pollen and total lipids were not significantly different from those of TOC in the core VER99G12 (Table 1; Watanabe et al. 2009a; Nara et al. 2010), suggesting that TOC is a suitable material for ¹⁴C dating in this case. Furthermore, it is reported that TOC and total nitrogen were significantly positively correlated with zero intercept throughout the core, meaning that TOC was negligibly contributed by allochthonous sources such as lignin and cellulose (Nara et al. 2014). These results indicate the significant reservoir effect on TOC in the core VER99G12 would be negligible.

Table 1 Conventional ¹⁴C ages and calibrated ages of TOC, total lipids, and pollen fractions from the core VER99G12 from Lake Baikal.

In this study, we present additional ¹⁴C of TOC results (20 samples) from the VER99G12 to establish the alternative ¹⁴C age model for the core VER99G12 based on IntCal20 (Reimer et al. 2020), especially for the climate transition period from the Last Glacial Maximum to the onset of the Holocene (20.8–11.7 ka; Ishiwa et al. 2016), using Undatable age model approach. The routine Undatable uses a deterministic approach to fold the sampling distance into total uncertainty and is designed to efficiently model age-depth relationships using an iterative procedure to explore multiple model settings. Undatable has previously been successfully to establishing accurate age-depth models for marine and lacustrine sedimentary achieves (Obrochta et al. 2018; Lougheed and Obrochta 2019; Waelbroeck et al. 2019). Recent refinements in the calibrations curve (Reimer et al. 2020) that have been made available since the last publication of the age model of the VER99G12 (Nara et al. 2010) reveal that the sedimentation processes on the Buguldeika saddle in Lake Baikal have been strongly influenced by the global climate events, resulting in the rapid changes in the sedimentation rate of the core.

STUDY AREA AND SEDIMENT SAMPLES

The location of Lake Baikal in Eurasian continent, the Lake Baikal watershed and the sampling site of the core VER99G12 are shown in a map in Figure 1. Owing to this large watershed, more than 80% of the water entering the lake comes from rivers (Osipov and Khlystov 2010); and less than 2% is contributed by precipitation on the lake surface. The sediment core sample in this study (VER99G12; 52º31'36"N, 106º09'08"E) was extracted from the Buguldeika saddle in Lake Baikal, which is geomorphologically separated from the central and southern basins. The sediment core sample was mainly formed by deposition materials from Selenga River (Kuzumin et al. 2000), which provides the largest inflow to Lake Baikal (ca. 50% of the total river input) (Osipov and Khlystov 2010). Therefore, the VER99G12 core records the climate and environmental changes not only in the Lake Baikal water column but also in its watershed, including the semi-arid region in Mongolia (Figure 1B). A number of paleoenvironmental studies using the core VER99G12 have been carried out (Soma et al. 2006; Nara et al. 2010; Shichi et al. 2013; Nara et al. 2014; Katsuta et al. 2018). These studies confirmed that the core VER99G12 is an excellent archive for understanding biological and hydrological changes in the south Siberian region caused by global climate changes, such as the last glacial to the post glacial change.

MATERIAL AND METHODS

Radiocarbon Measurements

The bulk sediment samples, retrieved from the core VER99G12 (Table 1) for the ¹⁴C measurements, were aliquoted into individual glass vials and stored in a freezer at −20ºC until the time of analysis. Samples were treated with 1.2 N-HCl to remove any carbonates. All the treated ¹⁴C samples were combusted at 850ºC for 6 hr in evacuated tubes with CuO and Ag wire. The resulting CO₂ was collected and purified in a vacuum line and reduced to graphite using an iron catalyst and hydrogen at 650ºC for 6 hr. The ¹⁴C measurements were performed by an accelerator mass spectrometer (JAEA-AMS-TONO-5MV; 15SDH-2, National Electrostatics Corporation) in the TONO Geoscience Center, Japan Atomic Energy Agency. Calibration to calendar ages for the ¹⁴C ages were performed using MatCal (Lougheed and Obrochta 2016) and the IntCal20 calibration curve (Reimer et al. 2020). Although the ¹⁴C age-depth model of the VER99G12 core has been previously inferred from the previous IntCal datasets (Nara et al. 2010; Katsuta et al. 2018), we recalibrated the ¹⁴C age of the core based on the updated IntCal20 dataset (Reimer et al. 2020; see Table 1).

Age Model Construction Using Undatable

Age modeling using Undatable was established following the previously reported studies in Lougheed and Obrochta (2019) and Obrochta et al. (2018). Undatable was developed out of a need to handle datasets with both high age-depth scatter and a disturbed depth scale by being less optimistic regarding uncertainty than the typical Bayesian models. The initial version was developed to model age in highly expanded cores from the Baltic Sea (Obrochta et al. 2017), was further modified to model coral growth rates (Webster et al. 2018), and further refined by Obrochta et al. (2018). Lougheed and Obrochta (2019) then substantially optimized the routine to be currently the most efficient model (to our knowledge) available for determining age in geological archives. Model parameters were 10⁵ simulations, 0.1 sedimentation rate uncertainty factor, and 30% bootstrapping, excluding intervals with dates in stratigraphic order in the upper ∼1 m, at ∼170 cm and below ∼300 cm. Although the reservoir effect by older carbon on total organic carbon date for the Buguldeika saddle in Lake Baikal could be significantly small (Nara et al. 2014), we used 380 ¹⁴C yr for the reservoir effect in Undatable to establish the age model because of its water residence time (Shimaraev et al. 1993).

RESULTS

All ¹⁴C data of the core VER99G12 are summarized in Table 1. The depth profiles of combined all calibrated age data of the core VER99G12 alongside Undatable were shown in Figure 2. The ¹⁴C age of the bottom layer (461–460 cm) was determined as ∼31,000 cal years BP (Table 1), which spans the marine isotope stage 3. The calibrated age depth profiles show the linear sedimentation rates (ca. 13.3 cm/kyr) with the high correlation coefficient through the core (R² = 0.974) through the core (Figure 2B). Nevertheless, the high-time-resolution ¹⁴C dating from 300 to 100 cm depth in the core VER99G12 (Figure 2C) studied here showed notable fluctuations at the layer from 260 to 120 cm depth.

Figure 2 Calibrated age model for the core VER99G12. (A) Undatable age model of core VER99G12, (B)Vertical distributions of calibrated ages for core VER99G12, and (C) limited to a sediment depth of 100–300 m of calibrated ages for core VER99G12. The dashed line in Figure 2B is a regression line for the calibrated age through the core. Correlation coefficient between the calibrated age and the depth through the core was calculated in 0.974. Dark and light shading in calendar age probability density functions (PDFs), indicates the calibrated 68.2% and 95.4% age ranges, respectively. The modeled median age and 95.4% range are indicated by the red solid and black dashed lines, receptively. The shaded density cloud reflects the 1 to 99th percentile range.

The set of assumptions built in to an age-depth model affects the resulting estimate of uncertainty. While Bacon (Blaauw and Christen 2011) treats age-depth determinations that are not in stratigraphic order as outliers that should not contribute to uncertainty, Undatable includes information from such determinations by increasing age-depth model uncertainty, particularly if bootstrapping (i.e., the number of dates to randomly exclude from a single Monte Carlo iteration), is relatively high. Reworking and bioturbation in marine and lacustrine settings mixes older and younger sediment, which, combined with sampling resolution, can lead to apparent age-depth reversals and increasing uncertainty. Undatable also increases uncertainty at sediment rate inflection points because the depth of the change in rate is dependent on the depth and number of age determinations.

Thus, we find the uncertainty produced by Undatable to be more appropriate to our setting, because it incorporates uncertainty due to both age-depth reversals as well as inflections in sedimentation rate. Figure 3 shows Bacon and Undatable age models run with parameters producing very similar median modeled ages. The primary difference is the increased uncertainty in the Undatable model near the large changes in sedimentation rate, which we believe is most appropriate given the nature of the data.

Figure 3 VER99G12 age models produced with Undatable (left panel; using parameters described in the Methods) and Bacon (center panel; accumulation mean shape: 50 and 1.5, respectively; memory mean and strength: 0.5 and 10, respectively). Both produce similar median modeled ages, but Undatable produces wider uncertainty (right panel) near sedimentation rate inflections that are characterized by scatter in age-depth determinations (shaded region).

DISCUSSION

Rapid Changes of Sedimentation Rates Corresponding the MWPs

The variation of the sedimentation rate of core VER99G12, is characterized by striking increases at 19.5, 18.3, 13.5, 10.2, and 9.5 ka (Figure 4), which are corresponding to the age reversal layers (Table 1). Apparent age reversals were observed in these layers, corresponding to ca. 19, 14, and 11.5 ka. Such age reversals at the same periods (ca. 19 and 14 ka) have been reported in the sediment core from Qinghai Lake (Zhou et al. 2016). Zhou et al. (2016) pointed out that old organic radiocarbon inputs to the Qinghai Lake sediment core at ca. 19 and 14 cal kyr BP could be caused by significant inputs of the meltwater from the glaciers around Qinghai Lake.

Figure 4 The profiles of sedimentation rate with calibrated age in the core VER99G12. The periods of global climate events for the Last Glacial Maximum, the MWP events and Younger Dryas event were highlighted in blue and yellow bars.

These significant increases in the sedimentation rates at 18.3 and 13.5 ka in the core VER99G12 happen just after the remarkable climatic warming events associated with the millennia of deglacial global sea level rise of 120 m in maximum (Yokoyama and Esat 2011). These sea level rises were caused by the large volume meltwater input to the oceans at 19 ka (meltwater pulse; MWP-1A0; Yokoyama et al. 2000; Clark et al. 2004; Yokoyama and Esat 2011) and 14.2–13.7 ka (MWP-1A; Deschamps et al. 2012). On the other hand, the rapid change in the sedimentation rate at 9.5 ka in the core VER99G12 was ca. 2000 years younger than the MWP-1B at 11.5 ka (Bard et al. 2010; Figure 4). Since the sampling site of the core VER99G12 was faced to the Selenga River inflow (Figure 1B and 1C), the deposition materials in the core VER99G12 could be strongly influenced from the Selenga River input (Kuzumin et al. 2000). Since Lake Baikal has large lake watershed (Figure 1B) and the Selenga River inflow has ca. 50% of the total river input, the lake level has been mainly controlled by the Selenga River inflow and the evaporation at the warm and cold climate stage, respectively (Urabe et al. 2004; Osipov and Khlystov 2010). Seismic and geophysical analysis have revealed drastic lake-level changes repeatedly at the climate transition between the last glacial and the post-glacial period (Urabe et al. 2004; Nara et al. 2014). The lake level of Lake Baikal at MIS2 was estimated at 11–15 lower than present level (Urabe et al. 2004). However, the detection of pollen in core VER99G12 during MIS2 means that the Selenga River inflow did not cease (Shichi et al. 2013). Based on the pollen analysis (Shichi et al. 2013), tundra steppe vegetation was established in the Lake Baikal watershed around 18.5 ka during the first MWP period. This means that the vegetation in the Lake Baikal watershed at 18.5 ka was characterized by the expansion of permafrost and dominance of grassland. In contrast, during the second phase of rapid sedimentation at 13.5 ka witnessed the growth of aquatic vegetation such as spruce, alder and willow along streams in the Baikal Lake lowlands (Shichi et al. 2013) indicating a wet climate condition attributed to increased precipitation. Therefore, the rapid sedimentation rate at 13.5 ka was influenced not only by inflow water resulting from the permafrost thawing but also from precipitation in the Lake Baikal watershed.

During MIS2, the major inflow through the Selenga River to Lake Baikal was the melt water from the local glacier around the lake, although the precipitation was 25–50% lower than modern-day precipitation (Osipov and Khlystov 2010). Today, permafrost expands over the whole Lake Baikal watershed and thermokarst lakes have developed in the modern Lake Baikal watershed, indicating repeated permafrost thawing during the glacial period (Törnqvist et al. 2014). A recent study showed the dominance of icesheets in Scandinavia and North America as a source of meltwater during MWP-1A, rather than from those of the Antarctic (Lin et al. 2021). Also, significant increase in river water inflow to Qinghai Lake in China at the MWP events have been reported (Zhou et al. 2016). Our result suggested the observation for the change in the hydrological systems corresponding with the MWP events should be applicable to the lake with a significant melting glacier around the lake.

Re-Evaluating the Biological Activity in Lake Baikal for 31 ka with the Alternative Age-Depth Model

Based on the alternative age-depth model of core VER99G12 using Undatable, we reformed the age profiles of total organic carbon (TOC) and the mean grain size (MGS) from the core VER99G12, which represent for the biological activity and the hydrological change, respectively (Figure 5; Nara et al. 2014). Biological activity in Lake Baikal for the last 32 ka inferred from TOC variations showed the synchronous variation with the summer isolation at 51°N on the whole (Figure 5). But also, the TOC profile of VER99G12 has small fluctuations as superimposed variations. Corresponding with the periods of increase in the sedimentation rates at 18.0 and 13.5 ka, the positive peaks of TOC concentration are also observed (Figure 5D). As well as TOC variation, despite of the sparse profile, the MGS variation from the Lake Baikal sediment core shows temporal increases at the above periods, which result from the high flux of the Selenga River input into the Lake Baikal (Figure 5C). The significant inflow of the Selenga River into the lake brings a significant nutrient load into the lake, resulting in the large biological activity in Lake Baikal (Nara et al. 2014). As well as warm periods, the temporal decreases in TOC of the core were observed at the LGM and YD (Figure 5). Therefore, the variations of biological activity and the grain size agree with the significant input of the Selenga River water into the lake at after the MWPs. These results mean that the alternative ¹⁴C age calculated by IntCal20 with the Undatable age model for the core VER99G12 can be used to exploit the new findings of the environmental and biological changes in the Eurasian continental area inferred from the lake sediment core corresponding with the global climate changes.

Figure 5 The comparison of the core VER99G12 records with other climate records. (A) δ¹⁸O record from NGRIP ice core (blue; Rasmussen et al. 2014)); (B) δ¹⁸O records from Sanbao and Hulu cave stalagmite (orange and red, respectively; Wang et al. 2005, 2001); (C) mean grain size and (D) TOC from the core VER99G12 in Lake Baikal (black and green, respectively, data from Nara et al. 2014); (E) July 51°N integrated irradiation for summer (JJA; day of year 152–243) (red; calculated following Lougheed (2022) using the orbital parameters of Laskar et al. (2004) and solar constant of 1361 Wm^–2).

SUMMARY

The rapid changes in the sedimentation rate corresponding with the global climate events, such as the MWPs, are observed from the Lake Baikal sediment core. The striking increase in the sedimentation rate were recorded at 18.3 and 13.5 ka, which are just after the periods of the MWP events. The signal of the melting glacier around the lake at 18.3 and 13.5 ka manifested as these rapid increases in the sedimentation rates. These sedimentation rate changes at 18.3 and 13.5 ka from the core VER99G12 were distinct variations comparing with other sediment cores from the Academician Ridge in Lake Baikal. The topographical feature of the Buguldeika Saddle, which is the sampling site of this study, manifested the change of inflow volume of Selenga River into the lake. The concurrent increases in TOC and the MGS at the 18.3 and 13.5 ka also support our interpretation that the high load of the nutrient for the biological activity to the lake caused by the high input of the river water at these periods.