Assessing the late Holocene14C reservoir age of theChukchi Sea with the AniakchakCFE II tephra 3.6 kyr BP

Master’s thesis

Physical Geography and Quaternary Geology, 30 Credits

Department of Physical Geography

Assessing the late Holocene 14C reservoir age of the

Chukchi Sea with the Aniakchak CFE II tephra 3.6 kyr BP

Alexis Geels

NKA 239

2019

Preface

This Master’s thesis is Alexis Geels’s degree project in Physical Geography and Quaternary Geology at the Department of Physical Geography, Stockholm University. The Master’s thesis comprises 30 credits (one term of full-time studies).

Cooperation with Århus University.

Supervisors have been Stefan Wastegård at the Department of Physical Geography, Stockholm University and Christof Pearce at the Department of Geoscience, Århus

University. Examiner has been Benedict Reinardy at the Department of Physical Geography, Stockholm University.

The author is responsible for the contents of this thesis.

Stockholm, 1 July 2019

Björn Gunnarson Vice Director of studies

Abstract

Tephrochronology is a powerful tool to correlate and improve the chronology of sedimentary archives in the Arctic Ocean. The Aniakchak Caldera Forming Eruption (CFE) in Alaska at 3.6 cal kyr BP ejected ash that were found in a widespread layer in Alaska, and as cryptotephra in the Chukchi Sea, Newfoundland, and Greenland. This study presents data from the core SWERUS-L2-4-PC1 (4PC) taken at a water depth of 120 m in the Chukchi Sea. The sharp peak in tephra shards concentration permitted to clearly place the isochron. Unfortunately, the mi- croprobe analyses were unsuccessful, however measurements of trace elements were performed with Laser Ablated-Inductively Coupled Plasma-Mass Spectrometry (LA-ICP-MS). The geo- chemical signature of the Aniakchak 3.6 eruption was ensured with significant trace element ratios. The isochron of the eruption combined with the radiocarbon dates from 4PC permitted to calculate the local marine radiocarbon reservoir age offset ∆R=364±46 years. This value is relatively low compared to recent estimates in the Chukchi Sea, especially to the neighbouring core SWERUS-L2-2-PC1 were ∆R=477±60. The ∆R value of this study is explained by the influence of the "young" Atlantic water mixing with the "old" Pacific water at the depth where the core was taken.

Keywords

Cryptotephra, Aniakchak CFE II, Arctic Ocean, Chukchi Sea, Marine Reservoir Effect

Contents

1 Introduction 4

1.1 Aim of the study . . . 4

1.2 The use of tephrochronology . . . 4

1.3 Marine reservoir effect . . . 7

2 Materials and methods 10 2.1 Study site and coring . . . 10

2.2 Sample processing and tephra quantification . . . 10

2.3 Microprobe analysis . . . 11

2.4 Laser Ablation-Inductively Coupled Plasma-Mass Spectrometry (LA-ICP-MS) 11 2.5 Radiocarbon chronology and reservoir age determination . . . 12

3 Results 15 3.1 Tephra counting . . . 15

3.2 Geochemistry . . . 16

3.3 Radiocarbon chronology . . . 18

4 Discussion 20 4.1 Defining the isochron . . . 20

4.2 Geochemical context . . . 22

4.3 The radiocarbon reservoir age of the Chukchi sea . . . 26

4.4 Comments on tephra sample preparation . . . 28

5 Conclusion 30

Acknowledgements 30

References 31

Appendix 39

A Supplementary tables 39

B Original OxCal code by C. Pearce 42

C Lab manuals 43

1 Introduction

1.1 Aim of the study

The primary aim of this study is to process samples of the core SWERUS-L2-4-PC1 from the Chukchi sea, mainly sections that represent the Mid- to Late Holocene, to find the Aniakchak 3.6 kyr BP tephra. When found, this layer will be put in a radiocarbon timescale to deduct the local marine¹⁴C reservoir age. This reservoir age will be compared to other reserevoir age estimates available for the Chukchi Sea.

The secondary aim of this study is to bring back the method of tephrochronology to the Department of Geoscience of Aarhus University, which has not been practised there for about ten years. Under the supervision of Christof Pearce, the laboratory manuals will be updated to facilitate forthcoming studies.

1.2 The use of tephrochronology

Volcanic eruptions are an almost instantaneous processes on a geological scale. They will pro- duce a certain amount of ashes, depending on the intensity of the eruption. The ashes will be deposited downwind (Walker, 2005). Thorarinsson (1944) was one of the first to make a detailed work about tephrochronology and he introduced the use of the word Tephra (τεφρα meaning ash in Greek). Tephrochonology developed further when Turney et al. (1997) presented a good technique to find rhyolitic cryptotephra (microscopic tephra that are not visible to the naked eye).

This meant that cryptotephra could potentially be found in many stratigraphic horizons where they were invisible (Walker, 2005; Lowe, 2011). In a later article, Turney (1998) detailed the methods for density separation and studies about cryptotephra started (Wastegård et al., 2000;

Davies et al., 2003; Bergman et al., 2004; Pearce et al., 2004).

Eruption of Volcanic Explosivity Index (VEI) 3 or more are likely to produce enough ma- terial to have an extensive proximal tephra layer. Eruptions of VEI 5 or more have a sufficient explosivity to eject ash in the stratosphere that will travel much further, forming distal tephra layers (Newhall and Self, 1982; Abbott and Davies, 2012). The airborne ashes would fallout within hours or days, unlike the aerosols that can stay several years in the atmosphere (Lowe, 2011). The ashes will settle everywhere and can be subsequently found in peat, ice and sedimen- tary environments. The greater the distance between the volcano and the deposit, the lower the tephra concentration (Griggs et al., 2014). The rapid deposition will create a widespread layer of tephra with the same origin and age that is called an isochron (Lowe, 2011). If the isochron

CHAPTER 1. INTRODUCTION

is from a short-lived eruption, the tephra layers are isochronous and can be used to correlate different stratigraphic archives.

However, in the marine environment, tephra are subject to secondary transport and deposi- tion. In high latitudes, tephra might fall on sea-ice or ice sheets and will have a delayed release onto the ocean floor (Griggs et al., 2014). Brendryen et al. (2010) demonstrated that the main mean of transportation of tephra in the Norwegian Sea was by iceberg rafting. The tephra were found in Ice Rafted Debris (IRD) rich layers that were lagging several centuries after the erup- tion. Yet, those layers contained ashes from geochemicaly distinct sources and poorly-sorted deposits, they were thus identifiable. Sea-ice rafting of tephra presents similar characteristics (Austin et al., 2004). But the lag would be much smaller, from a season to a couple of decades depending on the location and climate. Therefore, it would still make appropriate isochrons (Griggs et al., 2014). The ocean currents will also have an effect during the deposition of tephra.

Nevertheless, Wiesner et al. (1995) showed that the sinking speed of the ash from the 1991 Pinatubo eruption in the South China Sea was about 2 cm/s, thus minimising the influence of ocean currents in the deposition. After deposition the tephra layer can also be reworked by bottom currents and bioturbation (Austin et al., 2004; Lowe, 2011; Griggs et al., 2014). The concentration of tephra in a core should come out as a peak associated with the primary airfall and the isochron, followed by a tail associated with the secondary transport and post-deposition processes. Bioturbation will also lead to a small concentration of tephra below the peak (Griggs et al., 2014, 2015).

Geochemistry is used to distinguish between the different tephra layers. Indeed, volcanoes have their own geochemical signature of major elements (Cole, 1979; Jacobsson, 1979; Walker, 2005). This signature is linked to their geographical settings and how the magma is formed (Lacasse et al., 2006; Sigmarsson and Steinthórsson, 2007). However, to distinguish different eruptive phases of one volcano, the major elements are not sufficient (Reiners, 2002). The analy- sis of trace elements (Wood, 1976, 1981; Frey, 1979; Lacasse et al., 2006) as well as the analysis of isotope composition (Wood, 1981; Hanan and Schilling, 1997) would permit to differentiate different eruptive phases.

Tephra layers can be dated via indirect method such as radiocarbon age of trapped organic material, varve chronology or ice-core age (Walker, 2005; Abbott and Davies, 2012). Tephra shards can also be directly dated with radiometric method such as fission track or ⁴⁰Ar/³⁹Ar (Walker, 2005).

The Aniakchak

The Aniakchak volcano is located in the Aleutian range in Alaska (Figure 1.1). The stratovol- cano has had two Caldera Forming Eruptions (CFE) during the Holocene and numerous smaller eruption including a Plinian eruption at 400 yr BP and a subplinian eruption in 1931 as the most recent one (Miller, 1998; Bacon, 2014). The Aniakachak CFE I is dated at 9500-7000 yr BP, this wide age range make this eruption less useful as a potential isochron (Miller and Smith, 1987). CFE II was a large Plinian eruption at around 3600 yr BP with a VEI of 6 and extruded

>50 km³of andesitic and rhyodacitic material with ash flows (ignimbrite) reaching as far as 80 km from the caldera (Miller and Smith, 1987; Miller, 1998; Dreher et al., 2005; Bacon, 2014).

CHAPTER 1. INTRODUCTION

This eruption left a 10 km wide caldera with rim up to 1.3 km high, later eruptions left several smaller cones, lava flows and ash deposits in the caldera (Bacon, 2014).

Ash deposits have been widely found as visible layers in Western Alaska (Riehle et al., 1987;

Begét et al., 1992; Kaufman et al., 2012; Blackford et al., 2014) and as cryptotephra in New- foundland (Pyne-O’Donnell et al., 2012), Greenland (Pearce et al., 2004; Abbott and Davies, 2012) and in the Chukchi Sea (Pearce et al., 2017; Ponomareva et al., 2018).

CFE II has been dated with direct and indirect methods on many occasions (Pearce et al., 2004; Blackford et al., 2014; Kaufman et al., 2012; Abbott and Davies, 2012; Davies et al., 2016; McAneney and Baillie, 2019). Nevertheless, there is no agreement for a unique date range.

Pearce et al. (2017) reported that the incompatible ages from the radiocarbon dates (Davies et al., 2016) and the Greenland ice-core timescale (Abbott and Davies, 2012) could be unified. Indeed, the offset between the IntCal 13 radiocarbon timescale and the GICC05 ice-core timescale is - 19±3 yr at 3600 cal yr BP (Adolphi and Muscheler, 2016). This offset is applied to the CFE II age of 3591±3 cal yr BP in the GRIP ice-core (Abbott and Davies, 2012). The combination of these two dates gives the age 3572±4 cal yr BP for the Aniakchak CFE II. This age can be used in a timescale with radiocarbon ages (Pearce et al., 2017).

Figure 1.1. Map of the study area and site mentioned. Red: subject of this study, green: completing tephra infor- mation, turquoise: Marine ∆R sites. Star: volcano, triangle: ash site, pentagon: sediment core. Cartographic data from http://www.naturalearthdata.com

CHAPTER 1. INTRODUCTION

1.3 Marine reservoir effect

14C is a naturally occurring isotopes of carbon that is assimilated along with the other two iso- topes of carbon,¹²C and¹³C (Anderson et al., 1947). ¹⁴C is mainly formed in the upper atmo- sphere by the interaction of cosmic ray neutrons (n) and ¹⁴N to form the radiogenic¹⁴C and a proton (p) (Eq. 1.1). This reaction will generate further cascading reactions among atmospheric nuclei (Gosse and Phillips, 2001; Alves et al., 2018). The ¹⁴C is radioactive thus will decay over time and produce¹⁴₇ N, an electron (e⁻) and an electron antineutrino (ν_e) (Eq. 1.2) (Walker, 2005). The half life of radiocarbon is 5730 years but before the refining of the methodology 5568 yr was used. To avoid confusion with many old studies, the convention is to use the old half life (Mook, 1986; Stuiver and Polach, 1977).

147 N + n →¹⁴6 C + p (1.1)

146 C →¹⁴7 N + e⁻+ν_e (1.2)

The¹⁴C will then bond with oxygen to form carbon monoxide (¹⁴CO) and then carbon diox- ide (¹⁴CO₂) (Pandow et al., 1960). The¹⁴CO₂will mix with normal CO₂and will be absorbed by plants via photosynthesis. It will reach animals via the ingestion of plants (Aitken, 1990;

Walker, 2005). The atmospheric ¹⁴C will dissolve into the oceans and will be absorbed by the aquatic life. As the ¹⁴C is constantly replenished to thwart the decay, the terrestrial organisms are in an isotopic equilibrium with the atmosphere. At the death of the organism the¹⁴C contin- ues decaying but there are no more input, thus with the appropriate half-life and measurement of the remaining¹⁴C, the death age of the organism can be inferred.

However, to get accurate results, some more parameters need to be taken in account. The

14C rate of production in the upper atmosphere is not constant. It will depend on the cosmic ray intensity, the altitude and the intensity of the Earth and Sun magnetic fields (Gosse and Phillips, 2001; Muzikar et al., 2003). Furthermore, since humans started burning fossils fuels (coal, oil, gas) exempt of ¹⁴C, it disrupted the atmospheric isotopic ratio (Alves et al., 2018).

This is called the Suess effect after the first report of the imbalance (Suess, 1955). Finally, the nuclear tests from the 1950s and 1960s produced considerable amounts of ¹⁴C generating a disequilibrium in the global radiocarbon budget (Levin and Hesshaimer, 2000; Alves et al., 2018). Hence, the radiocarbon ages are calculated before the year 1950 as a reference, Before Present (BP) means before 1950. Precise calibration curves must be used to accurately handle radiocarbon dating. The IntCal13 is an example of curve for the atmosphere and Marine13 for the marine environments (Reimer et al., 2013). Dendrochronology is the main tool to calibrate the atmospheric curve, it is available up to 12.4 kyr (Blockley and Housley, 2009). To calibrate older part, other methods are needed such as U-Th dating of speleothems and corals or foraminifera from varved sediments. However, using organisms from the marine environment bring an other issue such as bioturbation or the marine reservoir effects.

The mixing of CO₂from the atmosphere to the ocean is not an instantaneous process (Mangerud, 1972). The CO₂will dissolve in the water and will sink slowly from the top of the water column

CHAPTER 1. INTRODUCTION

to the bottom. As it is away from a constant replenishment of ¹⁴C, it will be decaying along the migration in the water bodies. The organisms in the ocean will feed on carbon that has al- ready been decaying, hence that would have an older apparent age (Mangerud, 1972). Stuiver and Polach (1977) defined this difference between terrestrial¹⁴C ages and age from the marine reservoir the reservoir age (R(t)). This means that, due to the R(t) effects, a marine organism that would die today would have an apparent radiocarbon age different than zero. On average on the oceans of the whole planet, the marine reservoir age is 400 radiocarbon years (Ascough et al., 2005) .

Additionally R(t) varies with space and time, this offset is called the Delta R (∆ R) (Stuiver et al., 1986; Austin et al., 1995; Ascough et al., 2005). For example, in parts of the oceans where water has had a long residence time, it can be 1000 yr old, and thus have a ∆R of +600 yr (offset of 600 yr to the average 400 yr). Furthermore, due to the nuclear tests in the 50s and 60s, the amount of¹⁴C in the atmosphere considerably increased and additional dating methods or samples retrieved before 1950 are needed to calibrate the ∆R in the different basins and seas.

The difference in water residence time in the deep oceans is due mainly to the thermohaline circulation (Figure 1.2), also called the Great Conveyor Belt (Marshak, 2015). The motor of these currents is the change in density around the poles where the water gets colder and saltier, and sinks. The cold and dense water will travel around the globe along deep currents and resur- face in upwelling zones. When the water is deep in the oceans there are no more mixing with the ¹⁴C from the atmosphere, the water will get older. In contrary with surface waters where the atmospheric ¹⁴C has just been mixed, the water is younger. For example, the water in the North Atlantic is considered young as it comes directly from surface waters (Fig. 1.2) where

∆R is about 0 (Ascough et al., 2005). On the opposite, in the Northern Pacific Ocean, where water arise from the depth after a long residence time, the surface water can be up to thousands of years old (Southon and Fedje, 2003; Ohkushi et al., 2007; Kuzmin et al., 2007) as it has been away from any source of¹⁴C for a long time.

Figure 1.2. Thermohaline circulation, modified from Marshak (2015)

CHAPTER 1. INTRODUCTION

The Arctic ocean and the Chukchi sea

The Fram strait, between Greenland and Svalbard, is the main passage for the inflow and outflow of the Arctic Ocean (Rudels, 2015; Rudels et al., 2015). The inflow of warm Atlantic water is balanced by the outflow of cool Arctic water. There is also an input of Pacific Water via the Bering Strait, and freshwater coming from large rivers in Siberia, Alaska and North-Western Canada.

These waters slowly mix with the cold less salty surface water of the Arctic (Stranne et al., 2017). The fresh water coming from the Arctic rivers such as the Ob or the Lena helps main- taining the upper halocline (Janout et al., 2017). The influence of Pacific water is not yet fully understood but it would be constrained to the Chukchi sea and would vary seasonally (Linders et al., 2017). Figure 1.3 summaries the current agreement concerning the Pacific water entering the Chukchi Sea. It goes mostly into Barrow canyon along the Alaskan coast with the Alaskan Coastal Current (Corlett and Pickart, 2017; Spall et al., 2018). But also through Herald Canyon and the Central Channel East of Herald Shoal (Rudels, 2015; Linders et al., 2017).

Figure 1.3. The main oceanic currents in the Chukchi Sea, from Corlett and Pickart (2017)

Incidentally, all those different water bodies have different ∆R. For example the At- lantic water ∆R is low (Austin et al., 2011;

Eiríksson et al., 2011) compared to the ∆R of the Pacific water coming through the Bering strait (Southon and Fedje, 2003; Ohkushi et al., 2007). The water from the Siberian rivers might actually have a high ∆R com- pared to normal freshwater. The watershed of these rivers contains substantial amount of permafrost that would contain thousands years old ¹⁴C (Gustafsson et al., 2011; Mc- Clelland et al., 2016). Furthermore, the Chukchi sea and most of the arctic shelf were emerged during the Last Glacial Maximum (LGM), complicating the local history of¹⁴C (Martens et al., 2019).

2 Materials and methods

2.1 Study site and coring

The examined core was taken out by the ice breaker Oden during the SWERUS-C3 (Swedish- Russian-US Arctic Ocean Investigation of Climate-Cryosphere-Carbon interaction) expedition in 2014 (Jakobsson et al., 2017). Two cores were taken from the Eastern side of the Herald Canyon to investigate the Pacific water influence and the temporality of the Bering Strait flooding after the LGM. The first core, SWERUS-L2-2-PC1 (2PC) is a piston core 8.2 m long taken at a depth of 57 m at72^◦52.O’N,175^◦19.2’W (Pearce et al., 2017). The second core, SWERUS-L2- 4-PC1 (4PC) is a piston core 6.2 m long taken at a depth of 120 m at72^◦50.3’N,175^◦43.6’W.

4PC is the core analysed in this work.

2.2 Sample processing and tephra quantification

The core was divided in 5 sections, sections 1 to 4 of 4PC (0-471 cm deep), corresponding roughly to the Late Holocene, were sampled. The sampling was done continuously on the four sections, extracting 50 mm x 5 mm x 5 mm samples one after the other. They were frozen and then freeze dried.

The method used for processing the samples is an adaptation of the method described by Turney (1998). This method aims on isolating the rhyolitic ash from the rest of the sediments.

It is calibrated to extract particles with a size from 25 to 80 µm and a density between 2.3 and 2.5 g/cm³, the characteristics of the rhyolitic cryptotephra fall into these ranges.

When dry, the samples were weighted. A granulometry separation was done using two sieves of 80 µm and 25 µm mesh aperture. The fraction bigger than 80 µm was kept in tubes (>80 µm) and were not analysed further. The fraction smaller than 25 µm went down the drain.

The fraction between 25 and 80 µm was then separated using Sodium Polytungstate (SPT, Na₆[H₂W₁₂O₄₀]) which was calibrated to have a density of 2.3 g/cm³ first and then 2.5 g/cm³. The part of the sample of a density over 2.5 g/cm³, which contains the potential basaltic tephra was labelled B tephra. The part of the sample of a density between 2.3 g/cm³ and 2.5 g/cm³, which contains the potential rhyolitic tephra was labelled R tephra.

Lycopodiumspore tablets (batch no. 938934) were added to the R tephra tubes and dissolved using HCl 10% solution. A drop of the sample was then mounted on a glass slide using Canada Balsam resin and covered with a cover slip. The method is described in details with two manuals in Appendix C.

CHAPTER 2. MATERIALS AND METHODS

2.3 Microprobe analysis

The Electron Microprobe analysis determines the composition of a material using an electron beam. The results give the major elements oxides composition in weight %. It is extremely useful in tephrochronology to deduct the provenance of tephra by linking their geochemistry with the geochemistry of a volcano.

These analyses were planned on a JEOL JXA 8600 Superprobe located at the Department of Geosciences of Aarhus University. Unfortunately, the electron beam was slightly drifting thus making the analysis of the tiny shards impossible. Lack of time prevented to send samples for analyses in an other university.

2.4 Laser Ablation-Inductively Coupled Plasma-Mass Spectrometry (LA-ICP-MS)

The LA-ICP-MS measurements were executed on the rhyolitic fraction (between 2.3 and 2.5 g/cm³) of sample 373.5 (cm) from 4PC. The sample was dried then cased in epoxy. The surface with the sample was polished using 15, 6, 3 and 1µm diamond spray then coated with carbon.

The sample was analysed for trace elements and additionally for major elements. The analysis were carried on in the AGIR facility at Aarhus University. A S155 sample cell (Laurin Technic) was used for laser ablation and a 7900-ICP-MS (Agilent Technologies) for mass spectrometry analysis. The laser ablation was carried on with 5 hz, 80.0 Mj laser beam on a 18 µm wide ablation zone. The particles were carried with He (375 ml/min) mixed with N₂(3 ml/min). For the mass spectrometry they were mixed with Ar (1 l/min). For each sample, the MS analysis took 25 seconds of background analysis then 30 seconds of the ablated sample analysis. Out of the 9 analyses, 8 were successful and 1 was discarded because there were not enough counts. Before and after the 9 samples, two runs were done for calibration on NIST-610 and NIST-612 synthetic glass of known trace elements concentration. For each sample, 37 elements were measured with each one measured for 10 ms before measuring the next element.

The raw data was then transformed with the Iolite software. The background noise was removed and the integration time was defined following the method described by Longerich et al. (1996). As there were no specific value of major elements to calibrate the analysis, the percentage of Si was arbitrarily defined as 72 %. The data was then normalised with²⁸Si.

CHAPTER 2. MATERIALS AND METHODS

2.5 Radiocarbon chronology and reservoir age de- termination

The chronology of the core has been constructed with the program OxCal 4.3.

OxCal is a calibration program created in 1994 by Christopher Bronk Ramsey to facilitate the conception of age model without the mathematical troubles. It is used to calibrate the chronol- ogy of stratigraphic sequences with primarily radiocarbon dates (Ramsey, 1995). Furthermore, OxCal use Bayesian statistics to tests the accuracy of the model and how trustworthy it is con- sidering all the dates. The model will give a probability density for each given radiocarbon age and will allow to probe the likely stratigraphy of the sequence (Ramsey, 1995).The prior con- cept is important to understand the outcome of the model. It follows the observation of physical characteristics, which sample is after which in the sequence thus wich sample should be older than which. The other conept is the likelihood of a sample, the age measured with its prob- abilities (Ramsey, 2001). With the help of the Markov Chain Monte Carlo (MCMC) method (Gilks et al., 1995), those two parameters are put together to obtain the posterior distribution probability, this is the Bayes theorem. The boundaries are another important concept to take in account. They will define how the probabilities will behave at the beginning and the end of the sequence. Boundaries can also be put in the sequence if there are different group of sample ages (Ramsey, 2001). Ramsey (2001) assumes that the boundaries are uniformly distributed in the geological sequence defined with a Poisson interval distribution and that sample are also uniformly distributed between the boundaries. There are several possible boundaries options such as a Boundary, that has a uniform probability until the boundary, or a Zero_Boundary, that has a linearly decreasing probability with the boundary at zero (Buck et al., 1992) (Figure 2.1).

Figure 2.1. An example of two boundaries and their probabilities visually represented. From http://c14.arch.ox.ac.uk/oxcalhelp/

In 2007, Ramsey implemented a wider range of deposition Bayesian models to improve the interpolation between different sequences. There are different function in OxCal to represent the depositional rate. For example, if the deposition rate is theoretically known, for example with ice-core, the function will be a defined deposition rate (D_Sequence) (Ramsey, 2008). On the contrary, if the deposition rate is unknown, the function P_Sequence will be used where the deposition rate is considered random and defined by a Poisson distribution (Ramsey, 2008).

The latest calibration curve (0-50 cal kyr BP) are IntCal13 for the atmospheric ¹⁴C and Marine13 for the marine ¹⁴C (Reimer et al., 2013). The IntCal13 depict the reservoir of the mid-latitudes from the Northern an Southern hemisphere while Marine13 depict a global marine reservoir. The IntCal13 is an improvement of the IntCal04 and IntCal09 curves, new treerings data sets have been added to reach 13.9 cal kyr BP with dendrochronology (Reimer et al., 2013).

Plant macrofossils data from the varved sediments of the Lake Suigetsu bring an additional chronology from 13.9 to 50 cal kyr BP (Ramsey et al., 2012; Reimer et al., 2013). Measurements

CHAPTER 2. MATERIALS AND METHODS

of¹⁴C in U-dated series of speleothems in the Bahamas have also been added starting 14 cal kyr BP (Reimer et al., 2013). For the period of 0-10.5 cal kyr BP, the Marine13 curve is based on the box diffusion model from Oeschger et al. (1975). Beyond the limit, the curve is calibrated via datasets of U-Th dated corals and non-varved marine sediments from the Cariaco Basin, the Iberian and Pakistani margins (Reimer et al., 2013).

In this study, five radiocarbon dates of sections three and four of core 4PC have been used (Table 2.1). Additionally, the layer comprising the Aniakchak CFE II isochron has been set as a numerical age.

Different parameters were used in OxCal to explore the potential changes in ∆R. The isochron has been established at 373.5 cm (Section 4.1). As mentioned in the Introduction, the Aniakchak CFE II eruption date was set at 3572±4 cal yr BP (Pearce et al., 2017). Another date, 1627±1 BC (McAneney and Baillie, 2019), has also been used to explore the sensitivity of the model.

Table 2.1. Radiocarbon dates of section 3 and 4 of 4PC with their respective position down the core in cm below sea floor (cmbsf). Beta dates are from molluscs and NOSAM date is from benthic foraminifera

Core depth (cmbsf) ¹⁴C Lab ID ¹⁴C age (yrs BP) ¹⁴C st. dev. Ref.

333 Beta-512060 3510 30 Unpublished

334 Beta-490777 3350 30 Unpublished

337 NOSAMS133772 3490 25 Jakobsson et al. (2017)

345 Beta-512061 3530 30 Unpublished

370 Beta-512062 3980 30 Unpublished

Furthermore, to contribute to the discussion about the reservoir age, a set of calibrated ∆R from the Marine Database is used (see Table 2.2), along with a new set of ∆R (Pearce et al., unpublished).

CHAPTER 2. MATERIALS AND METHODS

Table 2.2. Different ∆R and their location in the Chukchi Sea, the Laptev Sea and around the Bring Strait. Data from calib.org/marine.

Longitude Latitude Delta R DeltaR Err. Reference

-166.67 65.25 350 50

McNeely (2006)

-166.67 65.25 520 20

-166.37 65.27 580 40

-166.37 65.27 450 20

-161.42 70.40 610 40

-161.42 70.40 310 50

-161.42 70.40 320 25

-156.48 71.40 420 40

-156.48 71.40 560 30

-156.48 71.40 470 40

-156.48 71.40 560 60

118.33 76.63 37 40

Bauch et al. (2001)

118.33 76.63 -8 45

124.68 75.70 162 45

130.18 75.60 27 50

134.57 74.70 42 60

-177.08 62.65 558 34

Pearce, unpublished

-173.38 67.07 488 26

-172.05 64.87 532 28

115.5 76.67 6 30

144.33 73.83 642 26

164.17 70.47 422 26

174.45 69.93 400 30

3 Results

3.1 Tephra counting

Figure 3.1. View of the smearslide under the 100x mag- nification of a polarising microscope. The tephra are the clear glass-like particle and the Lycopodium spores are the roundish orange particle.

After the processing of the samples, the fin- ished product for analyses was a smearslide with Lycopodium spores and tephras. The Figure 3.1 is the typical view on the mi- croscope of a slide with many tephras glass shards, some Lycopodium spores and a few other minerals. The weight of the sample (W ) and the number of Lycopodium spores (L) per tablet, and thus per sample, is known. The tephra (T_s) and Lycopodium spores (L_s) are counted at the same time under the micro- scope. The concentration of tephra per gram of dry sediment (C) is found when these pa- rameters are inserted in the following equa- tion:

C = T_s L_s ∗ L

W (3.1)

Rhyolitic tephra are transparent glass-like shards (Figure 3.2a). They usually have sharp edges that can be straight or roundish. Tephra can also have a tubular shape (Figure 3.2b). In general, the colour ranges from transparent for rhyolitic tephra to slightly maroon for basaltic tephra. In this study, all the tephra found were transparent or with a slightly green tint. Theo- retically, tephra is isotrope thus completely opaque under cross-polarised light. However, they can be slightly birefringent and present a greyish colour under cross-polarised light, or some sediment can be under the tephra giving the impression of a birefingence.

40 samples of the rhyolitic fraction (density between 2.3 and 2.5 g/cm³) were counted after being processed. In addition, 8 samples of the fraction >2.5 g/cm³were taken around the peak and counted. The 40 rhyolitic samples represent the lower part of section 3 and section 4 of 4PC, spanning from 271 to 471 cmbsf, the 8 >2.5 g/cm³samples span from 361 to 401 cmbsf. The name of the samples represent the mid core depth between the two extremes of the sample.

Figure 3.3 summarises the concentration of tephra along the core depth. In the bottom part of section 4, from sample 468.5 to sample 383.5, the concentration of tephra is low, this is

CHAPTER 3. RESULTS

Figure 3.2. Microscopic pictures of the tephra glass shards from sample 378.5

the background level of tephra that is constantly present in the system. A sharp increase in concentration of tephra starts at sample 378.5 to reach maximums with sample 373.5 and 368.5.

After the peak, the concentration varies much, but is always above background level. A second peak might be present in sample 318.5.

Figure 3.3. Tephra concentration from 271 to 471 cmbsf of 4PC

3.2 Geochemistry

The trace elements analysis was conducted on sample 373.5 from which 8 shards were analysed and one shard analysed twice. Major elements analysis were also conducted on these shards, however the results are not reliable (Thomas Ülrich, personnal communication). From the 9

CHAPTER 3. RESULTS

measurements only 8 were successful. The results in ppm for each element and shard are shown in Table 3.1 and with a spider graph in Figure 3.4.

The numbering of the shards refers to the order they were measured in. It is a coincidence that the ppm of most of the elements increases after each elements. In comparison, shard-8 have some of the concentration up to ten times the concentration of shard-1. The spider graph shows that the curves have mostly the same shape, there are some discrepancies with a few elements, mostly Er, Lu and Ta.

Table 3.1. Results of 8 out of 9 analyses for the major and trace elements with the LA-ICP-MS, all concentrations are in ppm. The elements in the upper part are considered as major elements, trace elements are in the lower part of the table.

shard-1 shard-1b shard-3 shard-4 shard-5 shard-6 shard-7 shard-8

Na 27900 32900 44400 37600 32780 48300 34860 43200

Mg 926 1393 735 922 986 1603 1620 3040

Al 55700 56000 81300 67700 57100 80300 65500 74500

P 81 20 -8.2 10.5 55 -9.3 48 na

K 2016 3185 4870 5440 5770 9040 8670 20600

43Ca 22400 22100 16200 17900 10600 13700 14700 na

44Ca 18700 18620 13300 14000 11470 10600 12700 12500

Ti 395 576 401 487 483 718 746 1216

Mn 127.2 208.9 158 217.4 247.6 337 380 636

Fe 5540 12870 5890 11790 19590 111000 na na

Li 1.45 2.75 4.01 5.43 5.52 8.07 8.807 20.2

Rb 2.97 5.5 7.7 10.2 12.06 20.2 21.6 48.4

Sr 31.6 47.8 42.3 52.1 52.7 78.2 85.4 127.8

Y 3.09 4.66 5.98 7.4 7.63 11.8 14.18 25.2

Zr 16.4 26.6 39.5 50.3 51.6 77.9 98.7 189

Nb 0.55 0.92 2.01 2.68 2.89 4.98 5.51 11.1

Cs 0.33 0.43 0.64 0.81 0.84 1.23 1.02 2.42

Ba 67.7 111.4 136 195 206.3 334 386 762

La 1.44 2.24 3.57 4.88 4.95 7.98 9.83 18.8

Ce 3.45 5.3 7.99 10.32 10.61 16.2 19.2 36.4

Pr 0.53 0.62 0.94 1.12 1.27 2.07 2.53 5.27

Nd 2.08 3 3.7 4.8 5.17 7.8 11 18.9

Sm 0.86 1.12 1.22 1.15 1.25 1.88 4.5 4.8

Eu 0.118 0.183 0.23 0.25 0.26 0.56 0.67 0.81

Gd 0.75 0.81 0.76 1.5 1.58 2.2 2.9 4.9

Tb 0.062 0.147 0.16 0.23 0.225 0.23 0.22 0.78

Dy 0.86 0.8 0.74 1.05 1.16 2.03 2.72 5

Ho 0.082 0.195 0.286 0.26 0.3 0.550 0.47 1.27

Er 0.29 0.48 0.35 1.25 0.87 0.91 2.04 2.5

Tm 0.062 0.053 0.113 0.2 0.163 0.214 0.15 0.4

Yb 0.3 0.5 0.4 0.8 0.72 1.24 1.69 2.3

Lu 0.031 0.08 0.103 0.109 0.3 0.227 0.17 0.4

Ta 0.026 0.055 0.13 0.23 0.145 0.22 0.32 na

Pb 1.13 1.94 1.49 1.89 2.37 3.71 4.98 6.7

Th 0.33 0.53 0.8 1.31 1.04 1.99 2.4 3.66

U 0.125 0.233 0.294 0.51 0.51 0.87 0.73 1.46

CHAPTER 3. RESULTS

Figure 3.4. Results of 8 out of 9 analyses for the trace elements with the LA-ICP-MS. Note the logarithmic scale.

3.3 Radiocarbon chronology

Different parameters have been used in OxCal to explore the different ∆R outcomes (Ramsey, 2008, 2009; Ramsey and Lee, 2013; Reimer et al., 2013). Two ages of the eruption are used, the first is the age mentioned earlier (3572±4 cal yr BP) which is mainly used for comparison between the core 2PC and 4PC. The second age (1627±1 BC) is from McAneney and Baillie (2019), it was determined indirectly with dendrochronology and used here to investigate the sensitivity of the model to slight changes in the Aniakchak CFE II age. The preset of ∆R are imposed by the user of the program. A preset with U(.,.) indicates a uniform distribution of

∆R possibilities. For example, the preset U(0,800) indicates that there is an equal probability of

∆R being anything between 0 and 800. The other preset x±x is written in a slightly different way than in the program for an easier understanding. This preset means that there is a higher probability of having the first number plus or minus a certain age with probabilities decreasing to the extremities. For example, the preset 400±400 means that the probability of having a ∆R equal to 400 is the highest, but the possible ∆R span form 0 to 800, while the extremes have the lowest probability of being true. The last parameter that has been tested is if whether all the radiocarbon dates (Table 2.1) should be taken in account. This will be discussed why in Section 4.3.

CHAPTER 3. RESULTS

Table 3.2. This table is a summary of each run performed, where the ∆R and itsσ (68.2%) in the last two columns

Aniakchak Age DeltaR preset Dates DeltaR obtained Sigma 68,2%

3572±4 cal yr BP U(0,800) All 346 46

3572±4 cal yr BP U(200,600) All 347 41

3572±4 cal yr BP U(300,500) All 347 39

3572±4 cal yr BP 400±400 All 349 91

3572±4 cal yr BP 400±200 All 349 50

3572±4 cal yr BP U(0,800) No 333cm 361 47

3572±4 cal yr BP 400±200 No 333cm 363 56

3572±4 cal yr BP U(200,600) No 333cm 364 46

3572±4 cal yr BP U(200,600) No 334cm 343 43

3572±4 cal yr BP U(200,600) No 333/334 cm 356 45

3572±4 cal yr BP 400±200 No 333/334 cm 361 58

1627±1 BC 400±200 All 347 50

1627±1 BC U(200,600) All 344 43

1627±1 BC U(200,600) No 333cm 361 47

4 Discussion

4.1 Defining the isochron

Defining the isochron can be difficult. Nevertheless, the results of this study are clear ( Sect.

3.1). The sharp increase in tephra per gram of dry sediment at 373.5 cm (Fig. 3.3) should clearly indicate the isochron.

In the bottom of the oceans there are many benthic animals that dwell on and in the sea floor.

They move sediment up and down and will cause bioturbation (Austin et al., 1995; Clough et al., 1997). Griggs et al. (2015) have demonstrated that tephra can be moved up and down the sea floor by the benthic dweller, thus bioturbation reflects the benthic life activity (Austin et al., 1995).

Figure 4.1 shows a CT scan of a sediment core where tubular structures are clearly visible below the blocky structure, there bioturbation reaches about 3 cm down-core. Austin et al. (1995) found that the bioturbation would be of no more than 10 cm on the shelf off the West coast of Scotland.

Clough et al. (1997) found similar characteristics for the Arctic ocean and insist on the fact that water depth is a primordial parameter as this will greatly influence the density of the infaunal biomass. With the depth increasing the infaunal density decreases. Furthermore, there is about 5 to 10 cm of bioturbation, deducted from²¹⁰Pb, in the Trigger Core right next to 4PC (Pearce, personal communication).

In addition with bioturbation, there are also secondary transports and reworking as men- tionned in the Introduction (Lowe, 2011; Griggs et al., 2014). It will give the pyramidal structure (Figure 4.1) to the tephra. This is reflected with the concentration of tephra slowly decreasing after the peak.

The data from this study (Figure 4.2) fit well with the assumptions of bioturbation and re- working. The primary airfall would be the sharp peak in concentration of samples 373.5 and 368.5. The medium concentration of sample 378.5 would be the effect of bioturbation and the decreasing concentration of the samples above the peak would be a tail induced by reworking and secondary transport. The isochron is then put at the start of the peak corresponding to the primary airfall on sample 373.5, spanning 371 to 376 cm. There is a temptation to put it on a more precise centimetre for example at 376 cm. However, it is not possible to do that without more precise analyses, the isochron will have to be on the middle of one of the 5 cm sample, here at 373.5 cm.

1 cm thick samples should be taken around the peak to precise the onset of the primary airfall.

There would be no point in taking subcentimetre samples as it would be below the resolution of the current methodology.

Kaufman et al. (2012) speculated on the presence of a second tephra signal coming from an

CHAPTER 4. DISCUSSION

Figure 4.1. CT scan of a core with bioturbation within affecting the tephra location, from Griggs et al. (2015)

eruption of the Aniakchak volcano at 3.1 kyr BP, and that eruption might be visible in the results of this study. Indeed, it seems that a second peak is perceptible with sample 318.5. Nevertheless, it might be an artifact from the noise of the methodology (Section 4.4). It could also be the effect of increased reworking from the bottom currents. The eruptions of Aniakchak seem to be slightly different in their trace elements (Section 4.2). It would be interesting to analyse tephra from the second peak to determine if they are geochemically different.

The analysis of the >2.5 g/cm³samples around the peak (orange in Figure 3.3) did not give additional information concerning the onset of the eruption. Furthermore, these samples were supposed to contain basaltic tephra but were not different from the rhyolitic fraction. Indeed, they presented all the same characteristics and colour as the rhyolitic tephras. It is likely that the tephra found in the >2.5 g/cm³ fraction are actually rhyolitic tephra that did not separate well during the processing of the samples. Thus the peak in "basaltic" tephra is only a mirror of the peak in rhyolitic tephra.

CHAPTER 4. DISCUSSION

Figure 4.2. Tephra concentration from 271 to 471 cmbsf of 4PC. The green line rpresent the isochron, the red star represent the sample taken for the LA-ICP-MS analysis.

4.2 Geochemical context

Since the electron microprobe analysis was unsuccessful, the LA-ICP-MS analysis was not ac- curately calibrated. Indeed, to give reliable results, a concentration of a reference major element, such as Si, needs to be introduced in the analysis software. Ponomareva et al. (2018) found that the rhyodacitic tephra SiO₂wt.% is between 71 and 73%, we applied 72% for every shard anal- ysed. Furthermore, the shards encased in the epoxy were quite small. The laser ablated most of the material, usually with a bit of the encasing epoxy (Ülrich, personal communication). This would give some noise to the obtained data. Pearce (2014) gives extensive information about the effect of the laser beam width, energy and time of ablation and their effect on measurement.

Comparison with other Aniakchak CFE II studies

A spider graph with the trace elements concentration for this study and other studies concerning the Aniakchak CFE II tephras is shown in Figure 4.3. While the three previous studies (Pearce et al., 2004; Kaufman et al., 2012; Ponomareva et al., 2018) have extremely similar concentration of trace elements, the present study has lower concentrations for every elements. Even one standard deviation does not reach similar concentrations, bearing in mind the logarithmic scale.

However, this gap is most likely due to the imprecise calibration mentioned above. Indeed, the shapes of the curves are remarkably similar.

The ratio between different trace elements are used to investigate further the similarity of the results. These ratios have been chosen based on the potential discrepancy with the involved elements (Pearce et al., 2004; Pearce, 2014). When looking at the ratio, the imprecise calibration should disappear as it is the same imprecision on both measurements. To facilitate comparison this study will be first confronted with Ponomareva et al. (2018) as it is the most recent study.

CHAPTER 4. DISCUSSION

Figure 4.3. Comparison of trace elements of the Aniakchak CFE II from different studies. This study, on single shard analysis with LAC-ICP-MS, is represented as an average of the 8 shards analysed. The error bar shows the spread of the data with one standard deviation, for a graph with every analysed shard, see Figure 3.4. Kaufman et al. (2012) is the mean of single shard analyses (n=15) with LA-ICP-MS (UA 1963 High SiO₂). Ponomareva et al. (2018) is the mean of rhyodacitic single shard (n=12) analyses with LA-ICP-MS. Pearce et al. (2004) is the mean of bulk glass (n=4) analyses with Solution ICP-MS.

Tephra-1 and Tephra-1b correspond to two analyses made on a single shard, they will be referred as T1.

In table 4.1, T1 appears to be of a different composition than the rest of the tephra (T_r). Most of T1 ratios are slightly different, there is only the ratio La/Lu that is similar to the other analyses.

T1 seems most closely related to the Dacitic group from Ponomareva et al. (2018). However, T1 might be another mineral.

The mean of T_ris very similar to the rhyodacitic group from Ponomareva et al. (2018). There are two ratios that are slightly off, La/Lu again and Ba/Rb. Nevertheless, the ratio similarity between T_r and the rhyodacitic group should be sufficient to say that they are actually of the same origin. To ensure the tephra provenance of this study, they will be compared with studies of other eruption and volcanoes.

Comparison with other major Mid- to Late Holocene eruptions

During the Mid to Late-Holocene there were many eruption in the Northern Hemisphere big enough (VEI> 5) to spew tephra on extremely long distance (Abbott and Davies, 2012; Kaufman et al., 2012; Mackay et al., 2016; Ponomareva et al., 2017; Abbott et al., 2018). For example, the Thera eruption, 3577–3559 cal yr BP (Sulpizio et al., 2013), on the island of Santorini happened almost at the same time as the Aniakchak CFE II, 3572±4 cal yr BP. It was previously tought that tephra from the Thera eruption was found in the GRIP ice core (Hammer et al., 2003). But Pearce et al. (2004) showed, using trace elements, that the tephra layer was actually from the Aniakchak CFE II. In the Greenland ice cores most of the ash layers found come from Iceland and Jan Mayen, some come from North America (Oregon, Alaska), Kamchatka and the provenance of a few layers has not been found yet (Abbott and Davies, 2012). In the Arctic, Pearce et al. (2017)

CHAPTER 4. DISCUSSION

Table 4.1. Different significant ratios to establish the likeness of the datasets. The first part of the Table shows the direct ratio for each tephra analysis of this study. The second part shows different means and their standard deviations. The third part is the mean and standard deviation of the ratios for the different groups of tephra from Ponomareva et al. (2018) where n equals the number of shards analysed.

Zr/Nb Ce/Nb Nb/Y La/Lu Yb/U Ba/Rb Rb/Sr La/Nb La/Sm Shard-1 29.82 6.27 0.178 46.45 2.40 22.79 0.094 2.62 1.67

Shard-1b 28.91 5.76 0.197 28 2.15 20.25 0.115 2.43 2

Shard-3 19.65 3.98 0.336 34.66 1.36 17.66 0.182 1.78 2.93 Shard-4 18.77 3.85 0.362 44.77 1.57 19.12 0.196 1.82 4.24 Shard-5 17.85 3.67 0.379 24.87 1.41 17.11 0.229 1.71 3.96 Shard-6 15.64 3.25 0.422 35.15 1.43 16.53 0.258 1.60 4.24 Shard-7 17.91 3.48 0.389 57.82 2.32 17.87 0.253 1.78 2.18

Shard-8 17.03 3.28 0.440 47 1.58 15.74 0.379 1.69 3.92

Mean 20.70 4.19 0.338 39.84 1.78 18.39 0.213 1.93 3.14

St. Dev. 5.48 1.16 0.098 11.06 0.44 2.28 0.090 0.38 1.08

Mean T_r 17.81 3.59 0.388 40.71 1.61 17.34 0.249 1.73 3.58

St. Dev. 1.39 0.30 0.038 11.55 0.36 1.17 0.070 0.08 0.84

Mean T1 29.37 6.02 0.188 37.23 2.27 21.52 0.105 2.53 1.84

St. Dev. 0.64 0.36 0.014 13.05 0.18 1.80 0.015 0.13 0.23

Andesitic (n=3) 15.91 4.27 0.252 32.10 2.50 14.32 0.074 2.04 3.08

St Dev 0.98 0.11 0.015 5.42 0.49 1.52 0.007 0.06 0.15

Dacitic (n=3) 23.97 4.81 0.242 32.51 2.35 14.30 0.174 2.24 3.21

St Dev 7.15 1.07 0.085 5.96 0.84 3.14 0.039 0.47 0.15

Rhyodac. (n=12) 17.81 3.44 0.337 36.34 1.75 12.79 0.354 1.65 3.60

St Dev 1.53 0.18 0.033 5.07 0.18 1.15 0.024 0.10 0.35

have found a layer from the Aniakchak CFE II in a core (2PC) nearby the core of this study.

In another core (HLY0501-01) from the Chukchi sea, Ponomareva et al. (2018) found a tephra layer with the Aniakchak CFE II, and additional tephras from other Alaskan and Kamtchakan volcanoes.

Figure 4.4 compiles in a spider graph the trace elements concentrations from different vol- canoes. Kaufman et al. (2012) found the Aniakchak 5.8 kyr BP tephra in only one site but that layer might potentially be correlated with tephra layers of several other sites. Its geochemical signature in major elements is similar to the Aniakchak CFE II and is probably from a smaller eruption. The Thera eruption in the island of Santorini in 3577–3559 cal yr BP sent ash north in Europe (Sulpizio et al., 2013), and it might have gone further north. This eruption is almost synchronous with the Aniakachak CFE II. The Khangar is a volcano in Kamchatka, its eruption in 7620-7920 cal yr BP, referred here as KHG, spewed ash to the North and has been identi- fied in the NGRIP ice-core (Cook et al., 2018). The Ksudach is another volcano in Kamchatka.

The eruption in 1800¹⁴C yr BP, referred as KS₁(Kyle et al., 2011), is the second largest from Kamchatka during the Holocene. Cryptotephra from KS₁where found as far as Eastern Canada (Mackay et al., 2016). The White River Ash (WRA), is a tephra layer that has been found in Alaska and North-Western Canada but nowhere as cryptotephra (Preece et al., 2014). The WRA North (WRAn) is most likely coming from Mt Churchill, Alaska and its age is 1605-1805 cal yr BP (Davies et al., 2016).

The shape of the different curves of Figure 4.4 are similar to an extent, to this study and

CHAPTER 4. DISCUSSION

Ponomareva et al. (2018). For example, the Thera and KS₁ have some clear discrepancy with the shape around some elements. The shape for Aniakchak 5.8 and KHG are much more similar.

Figure 4.4. Mean of single shard trace elements analyses of different volcano studies. Aniakchak CFE II: this study, n=8 and Ponomareva et al. (2018) rhyodacitic, n=12; Aniak 5.8, n=6 (Kaufman et al., 2012); Thera, n=10 (Sulpizio et al., 2013); KHG, n=7 (Cook et al., 2018); KS1, n=7 (Kyle et al., 2011); WRA, n=7 (Preece et al., 2014).

Table 4.2 summarises the ratio from the Aniakchak CFE II studies and ratios of the volcanoes and eruptions mentioned in Figure 4.4. Some studies do not have all the same trace elements analysed, thus some ratios are not available (na).

Table 4.2. Means of trace element ratios of different volcano studies.

n Zr/Nb Ce/Nb Nb/Y La/Lu Yb/U Ba/Rb Rb/Sr La/Sm

Aniak.

CFE II

All¹ 8 20.70 4.19 0.338 39.84 1.78 18.39 0.213 3.14

T_r¹ 6 17.81 3.59 0.388 40.71 1.61 17.34 0.249 3.58

T1¹ 2 29.37 6.02 0.188 37.23 2.27 21.52 0.105 2.53

Andesitic² 3 15.91 4.27 0.252 32.10 2.50 14.32 0.074 3.08

Dacitic² 3 23.97 4.81 0.242 32.51 2.35 14.30 0.174 3.21

Rhyodac.² 12 17.81 3.44 0.337 36.34 1.75 12.79 0.354 3.60

Bi-modal³ 15 19.67 3.41 0.290 33.16 2.39 17.63 0.252 4.01

Rhyodac.⁴ 4 17.23 3.70 0.335 35.68 1.68 12.95 0.334 3.45

Aniak. 5,8 Bi-modal³ 6 21.78 3.19 0.421 30.22 2.67 25.67 0.275 3.91

Thera Rhyolitic⁵ 10 25.95 na 0.302 na na 4.38 1.926 na

KHG Rhyolitic⁶ 7 13.40 3.44 0.742 62.98 0.95 12.54 0.321 6.57

KS1 Rhyolitic⁷ 7 na na na 11.38 10.72 17.88 na 1.56

WRA Adakitic⁸ 7 31.83 5.44 0.686 116.82 0.61 19.05 0.076 8.68

1This study; ²Ponomareva et al. (2018); ³Kaufman et al. (2012); ⁴Pearce et al. (2004);

5Sulpizio et al. (2013); ⁶Cook et al. (2018); ⁷Kyle et al. (2011); ⁸Preece et al. (2014)

The Thera, KS₁ and WRA have most of their ratios largely different from all the different studies of the Aniakchak CFE II. It is clear that the analysed shards of the Aniakchak CFE II eruption from this study have a different composition and do not come from any of those two volcanoes. Whether the KHG curve looks similar to the Aniakchak CFE II curves, most of the ratios are sufficiently distinct to assume that the shards from this study are different and thus

CHAPTER 4. DISCUSSION

do not come from the KHG. Even if the Aniakchak 5.8 was found in one place and thus was probably not a large eruption, it is likely that it was from the Aniakchak itself (Kaufman et al., 2012). Most of the ratios are slightly different from the Aniakchak CFE II studies, but statistical analysis would be needed to confirm it. Nevertheless, the ratio Ba/Rb is definitely different and that alone would lead to believe that the analysed shards in this study are different from the Aniakchak 5.8.

4.3 The radiocarbon reservoir age of the Chukchi sea

The∆R deducted from core 4PC

Figure 4.5. Chronology of the core 4PC using OxCal. a) Age-depth model for sections 3 and 4.

b) Detailed probabilities for the ∆R

Looking back at all the runs summarised in Table 3.2 it appears that changing the ∆R preset does not make any significant differences, therefore the preset U(200,600) will be kept. This preset would also fit with the Late Holocene ∆R result of 477±60 yr from the neighbouring core 2PC (Pearce et al., 2017). It seems that changing the eruption age does not give a significant difference. The age of 1627±1 BC does not give a more precise ∆R and its standard deviation is similar to the ∆R of the age 3572±4 BP. It is likely due to the fact that the high resolution of the two ages is absorbed by the lower resolution of the radiocarbon ages (Table 2.1). Therefore, the age 3572±4 BP will be kept.

Table 2.1 shows the dates that were used in OxCal. There is one sequence of dates that seems slightly offset. Indeed the date Beta-512060 (3510±30, 333 cmbsf) is significantly older than Beta-490777 (3350±30, 334 cmbsf) which was taken only 1 cm below and thus should, theoretically, be older. The two dates have a difference of almost 200 yr ¹⁴C age. Maybe something went wrong in the sampling or the lab, resulting in these seemingly odd ages. It might also be the results of reworking on the sea-floor (Heier-Nielsen et al., 1995). The model in OxCal was run with all dates, without Beta- 512060, without Beta-490777 and without both of them. There are significant differences between these runs. Deleting both of them does not make much

sense. The date that would most likely be wrong is the older one (Pearce, personal commu-

CHAPTER 4. DISCUSSION

nication). Then, in this case, Beta-512060 (333 cmbsf) is put aside.

The depth curve with these parameters is showed in Figure 4.5a and the calculated ∆R of 364±46 yr is represented in Figure 4.5b. The overall agreement is very good (>100%) for all the dates and for the ∆R (Appendix A).

Comparison with other∆R of the Chukchi sea

The Figure 4.6 is a collection of the ∆R available around the Chukchi sea from the Marine Reservoir Database (Table 2.2), a new set of ∆R (Pearce, unpublished) and the ∆R from the cores of this study and Pearce et al. (2017). It is essential to keep in mind that ∆R vary spatially but also temporally. The ∆R from the two cores (2PC and 4PC) reflect the conditions at the time of their deposition (∼ 3.6 kyr BP). The other ∆R (from the database and Pearce, unpublished) reflect the condition when the samples were taken, around the 1800’ and 1900’. Nevertheless, the assumption is made that the conditions in the Chukchi sea and the Arctic ocean did not drastically change in the last couple thousands years, thus they are comparable.

Figure 4.6. Collection of the ∆R available around the Chukchi sea from the Marine Reservoir Database (yellow dots), a new set of ∆R (red dots, Pearce, unpublished) and the ∆R from the cores of this study (red star) and Pearce et al. (2017) (yellow star). Bathymetry and land topography data from are the International Bathymetric Chart of the Arctic Ocean (IBCAO) (Weatherall et al., 2015) and General Bathymetric Chart of the Ocean (GEBCO) (Jakobsson et al., 2012).

The ∆R from Figure 4.6 reflect the¹⁴C age of the water body the original samples were found in. For example, the ∆R around the Bering Strait are the oldest because the North Pacific water has an "old"¹⁴C age due to the thermohaline circulation and the long residence time and thus have high ∆R (Southon and Fedje, 2003; Ohkushi et al., 2007).

The Pacific water goes mostly along the Alaskan coast (Corlett and Pickart, 2017; Spall et al., 2018) with the Alaskan Coastal Current (Figure 1.3), it is reflected with the ∆R from Point Barrow (PB). The Pacific water also goes through Herald Canyon (HC) (Linders et al., 2017).

The dates on the shelf along the coast of the East Siberian Sea are a bit more complicate to

CHAPTER 4. DISCUSSION

interpret, but they probably are a product of the mixing between the Arctic Surface water, the Atlantic water and some runoff freshwater. In the Laptev Sea, most of the water comes from the Lena, a large river draining a large part of Siberia. Theoretically, the carbon from the rivers should be very young, it would have just mixed with the water and thus have a significantly lower

∆R of approximately -400 yr as there would be close to no reservoir age (R(t)=0) (Alves et al., 2018). However, the carbon contained in the Siberian permafrost can be thousands of years old, and is now more actively released due to global warming. The Lena, as well as the other Arctic rivers, would contain a fair amount of very old carbon (Gustafsson et al., 2011; McClelland et al., 2016). Thus, the ∆R from the Laptev Sea, instead of being extremely negative, would reflect the mixing between carbon recently incorporated in the rivers, and carbon eroded from permafrost in the catchment. In Figure 4.6 the ∆R of the Laptev Sea are comprised between -8 and 162 years.

All those ∆R are located on the Arctic shelf and reflect the¹⁴C age of the water body present there. The younger Atlantic water is kept under the Arctic halocline, at depths lower than the shelf (Spall, 2013). The core 2PC is located at a depth of 57 m, just on the fringe of Herald Canyon. Its ∆R (477±60 yr) reflects the influence of Pacific water and is similar to the ∆R of the shelf. The core from this study (4PC), was retrieved at a depth of 120 m. Linders et al. (2017) reported Atlantic water in Herald Canyon up to a depth of 120 to 100 m and in a modelling study Corlett and Pickart (2017) report that the Atlantic water would reach depth of 150 to 120 m. Furthermore, Woodgate et al. (2005) and Pickart et al. (2009) observed upwelling of the Atlantic water along the shelf and more specifically in Herald Canyon. Therefore, the ∆R obtained in this study would be the result of the mixing between "old" Pacific water and "young"

upwelling Atlantic water. It is important to still keep in mind that the ∆R from this study reflects former conditions.

4.4 Comments on tephra sample preparation

In the first part of the process, there are many potential sources of contamination and/or loss of a part of the samples. First, for the sampling, the contamination will be avoided by cleaning the trowel between every sample and by sampling away from the edge of the core. Second, the manipulation of dry sample for weighing until sieving is where a loss might happen. A small air flow or an unintended blow might send some sediment flying, it is then lost.

Concerning the sieving, some tephra might get stuck in the mesh which would be a loss.

Then the mesh should be cleaned between each sample so the previously stuck tephra might not contaminate the next sample.

The density separation could be a step where a lot of error is accumulated especially if the homogenisation is not done properly. However, with repeating three times each separation and carefully shaking the sample before centrifuging there shouldn’t be any problems. The only issue would come from a big sample that cannot be homogenised by shaking. It would have to be done with a stick to unclog it and allow every part of the sample to be in contact with the heavy liquid. Making the slide itself shouldn’t bring any more errors.

Counting is definitely one of the troublesome part. For example, it is very unlikely that two