Australian Summer Monsoon variability in the past 14,000 years revealed by IODP Expedition 356 sediments

The International Ocean Discovery Program (IODP) Expedition 356 Site U1461 cored a Miocene to Holocene sedimentary sequence in the upper bathyal carbonate offshore northwestern Australia (NWA). The siliciclastic component of these strata is primarily derived from the Australian continent. Radiocarbon dating on macrofossils and planktonic foraminifera shows that the upper 14 m section at Site U1461 preserves Holocene sediments, recording regional climate variability. K/Ca ratios determined by X-ray fluorescence elemental analyses and %K determined by shipboard natural gamma ray analysis are interpreted as indicators of riverine run-off from the Australian continent. We document the consequences of the variability of the Australian Summer Monsoon (ASM) on the continental shelf of NWA. We report an increase in terrigenous input due to a riverine run-off after 11.5 ka, which reaches a maximum at ~ 8.5 ka. This maximum is the result of the enhanced ASM-derived precipitation in response to the southern migration of the Intertropical Convergence Zone (ITCZ). A decrease in riverine run-off due to a weakening of precipitation in the NWA region after 8.5 ka was caused by the northern migration of the ITCZ. We conclude that the ITCZ reached its southernmost position at 8.5 ka and enhanced precipitation in the NWA region. This Holocene record shows that even during interglacial periods, monsoonal variability was primarily controlled by the position of the ITCZ.


Introduction
Northwestern Australia (NWA) is located at the edge of the Indo-Pacific Warm Pool (IPWP), which is the largest source of moisture and heat on Earth (Gagan et al. 2004;De Deckker et al. 2014;De Deckker 2016). NWA shelf underlies the path of the Leeuwin Current, a regional south-flowing current and branch of the Indonesian Throughflow, which brings warm and low salinity water from the IPWP to the Indian Ocean (Gallagher et al. 2009(Gallagher et al. , 2014. A strong rainfall seasonality characterizes NWA climate variability during the Australian Summer Monsoon (ASM), which is one of the largest monsoons similar to the Indian Monsoon (Betzler et al. 2017) and the Asian Monsoon (Wang et al. 2001;Tada et al. 2016). ASM rainfall plays an important role for the hydrology of the equatorial Pacific and Indian Ocean through the Indonesian archipelago deriving freshwater discharge from rivers and changing water salinity in this region (De Deckker 2016). During glacial-interglacial cycles, the intensity of the ASM changed on~100-1000-year time scales (Kuhnt et al. 2015;Eroglu et al. 2016). The ASM is strongly influenced by the position of the Intertropical Convergence Zone (ITCZ), which migrated with warming and cooling conditions during the deglaciation period (Kuhnt et al. 2015;Mohtadi et al. 2011). Indonesian stalagmite δ 18 O data records the precipitation pattern of the ASM during the last deglaciation, suggesting that the ASM was synchronized with northern high-latitude climate events (Ayliffe et al. 2013;Denniston et al. 2013).
Modelling studies show that the exposure of the Sunda Shelf during the Last Glacial Maximum (LGM) and its flooding during the Holocene global sea level rise have influenced the hydrological cycle in the tropical Pacific and the eastern Indian Ocean and have caused rainfall variation in NWA (Di Nezio and Tierney 2013).
Terrigenous input from riverine run-off is a key proxy for monsoonal rainfall variation in the NWA region Gallagher et al. 2014;Kuhnt et al. 2015;Stuut et al. 2014). Previously, analyses of cores in the Timor Sea used riverine run-off as a key proxy for monsoonal rainfall to determine the relative ASM activity (Kuhnt et al. 2015). However, the source of sediment in the Timor Sea is both from distal locations of the Australian continent and the Indonesian archipelago. The International Ocean Discovery Program (IODP) Expedition 356 Indonesian Throughflow recovered a proximal ASM terrigenous proxy record by coring a series of sites in the shallow continental shelf to the upper bathyal region part of offshore NWA (Fig. 1), recovering a sequence of Miocene to Holocene sediments at Site U1461 over 1 km thick (Gallagher et al. 2015(Gallagher et al. , 2017a. Due to their proximity to the Australian coast, the siliciclastic component of the sediments is derived from the NWA continent via the Fortescue and De Grey rivers ( Fig. 1; James et al. 2004). We date the upper 14 m of Site U1461 using radiocarbon dating and describe the past 14,000 years of climate variability in NWA. The strata reveal evidence of the linkage between the ASM and global climate changes, enhancing our understanding of the Holocene history of the ITCZ migration across the continental shelf of NWA. Fig. 1 The location map of site U1461 with 200 m water depth contour from ETOPO dataset (Amante and Eakins 2009) and a monthly rate of precipitation in the tropical Indo-Pacific region. Rivers are shown with thin black lines (data obtained from a Global Self-consistent, Hierarchical, High-resolution Geography Database after Wessel and Smith 1996). Austral summer daily precipitation data were obtained from the Global Precipitation Climatology Project (Huffman et al. 1997), indicating that Site U1461 locates the southern limit of rain belt. Locations of the discussed records from Ball Gown Cave (Denniston et al. 2013), MD00-2361, andFR10/19-GC17 (De Deckker et al. 2014;Stuut et al. 2014 gher et al. 2015). This study focuses on the top 14 m of the core, which yielded well-preserved macrofossils (bivalves) and planktonic foraminifera for radiocarbon dating. The upper part of Site U1461 is divided into two units: unit I and subunit IIa (Gallagher et al. 2017a). Unit I extends from the core top to 12.4 m CCSF-A and unit IIa from 12.4 to 42.09 m CCSF-A, and subunit IIa was investigated to a depth of 14.0 m CCSF-A for this study. Unit I is a homogeneous olive-grey to greenish grey packstone ( Fig. 2) with planktonic and benthic foraminifera, sponge spicules, and aragonitic pteropod shells (Gallagher et al. 2017a). The main carbonate mineral is aragonite (average~35% of the sediment obtained from five samples in unit I) and the main non-carbonate mineral is quartz (average~27% of the sediment) based on the shipboard semi-quantitative XRD analysis (Gallagher et al. 2017a). Subunit IIa is a creamy-grey mudstone to wackestone ( Fig. 2) with glauconite and abundant peloids (Gallagher et al. 2017a). Macrofossils such as bivalves, gastropods, scaphopods, pteropods, echinoderms, bryozoans, small benthic foraminifers, and solitary corals are common (Gallagher et al. 2017a). The average content of aragonite of this interval is 79%. Quartz is rare, and clay minerals are absent in subunit IIa (Gallagher et al. 2017a).

Physical properties and major elemental analysis
Natural gamma radiation (NGR) is emitted when 40 K, 232 Th-series, and 238 U-series nuclides decay to their daughter isotopes. Shipboard measurements of NGR detect these emissions and provide the means to quantify the K, U, and Th contents of the recovered sediments (Gallagher et al. 2017b;). In the upper part of the cored sequence considered in this study, shipboard NGR measurements were carried out at a 10 cm resolution in Holes U1461B and D and at a 20 cm resolution in Holes U1461A and C. We use the freely available MATLAB algorithms of  to decompose the NGR energy spectra and quantify the K (wt.%), U (mg/kg), and Th (mg/kg) contents (https://doi.org/10.1594/IEDA/100668).
The top 14 m CSF-A (equivalent to 15 m CCSF-A) of Hole U1461C was also subject to an Avaatech X-ray fluorescence (XRF) scanning at 5 cm intervals, which was conducted at the International Ocean Discovery Program's Gulf Coast Repository. The core sections were covered with 4 μm-thick ultralene, and the samples were analysed twice at 9 kV and 30 kV. Here, we interpret only calcium (Ca), potassium (K), and titanium (Ti), which are more robust compared to lighter elements, as the latter is affected by the water content and the split-core surface (Kido et al. 2006;Ishiwa et al. 2016;Nakamura et al. 2016).

Age-depth model based on radiocarbon dating
Radiocarbon dating of macrofossils and planktonic foraminifera (Table 1) was carried out using single-stage accelerator mass spectrometry at Atmosphere and Ocean Research Institute, The University of Tokyo ). Macrofossils were etched by 10 M HCl to remove contamination and secondary carbonate (Ishiwa et al. 2016). Foraminifera were washed in an ultrasonic bath prior to graphitization. The method of graphitization follows Yokoyama et al. (2007).
Calendar ages were obtained using a Bayesian age modelling approach (Bchron: Haslett and Parnell 2008;Parnell et al. 2008) with Marine 13 (Reimer et al. 2013) as a calibration curve. We set the local reservoir correction to 0 at Site U1461, since a local reservoir correction in the NWA continental shelf was previously reported to be minor (O'Connor et al. 2010;Bowman 1985). The Bchron algorithm uses Monte Carlo methods to produce a continuous age-depth model and interpolates ages of outlier intervals using obtained ages.

%K variation by natural gamma radiation analysis
The %K is~0.2% from 14.0 to 12.4 m CCSF-A and increases to 9.6 m CCSF-A. The maximum %K is~0.6% at~9.6 m CCSF-A. Above 9.6 m CCSF-A, the %K decreases to 0.2% to the core top (Fig. 2). The %K value at 8.31 m CCSF-A in Hole U1461C is interpreted to be an outlier because there was an insufficient volume of sediments in the core liner at this level.

Major elemental analysis by XRF scanning
The Ca counts are~800,000 from 14.0 to 11.5 m CSSF-A and gradually decrease from 11.5 to 9.4 m CCSF-A. There is a marked reduction at 1.6 m CCSF-A (Fig. 2). The K counts are~2000 from 14.0 to 11.5 m CCSF-A, increasing to 10,600 counts at 10.2 m CCSF-A. The K counts gradually decrease from 8.2 m CCSF-A to 4000 counts in the core top (Fig. 2). The Ti counts arẽ 500 from 14.0 to 12.4 m CCSF-A, with a marked increase to 6000 counts that occurs from 12.4 to 10.2 m CSF-A. The Ti counts gradually decrease to 1000 counts from 8.2 m CCSF-A to the core top. K and Ti counts show a similar pattern of variability through the studied interval.

Age-depth model
Radiocarbon dating results are shown in Table 1. The age-depth model established by Bchron suggests that the interval from 15.0 m CCSF-A to the core top corresponds to the last 25,000 years (Fig. 3). An extremely low sedimentation rate (~0.07 m/kyr) occurred before 13.5 ka, suggesting that the past climate variability may not be recording at the accurate timing. Subsequently, relatively high sedimentation (~1.0 m/kyr) is observed to the core top. Ages obtained from foraminifera at 7.81 m CCSF-A and macrofossil ages at 7.62 m CCSF-A show an inverse age-depth relationship. We interpret radiocarbon date at 0.8 m CCSF-A as an outlier due to the secondary deposition or coring disturbance, since the date at this interval shows an extremely older age than proximal ages.

Discussion
Riverine run-off records at Site U1461 The %K contribution to the total NGR on the shelf of NWA is interpreted to indicate the presence of fluvial-derived clays and feldspar, reflecting the relative variability of continental moisture Groeneveld et al. 2017). The %K from NGR data shows a similar pattern to K counts derived from XRF scanning (Figs. 2 and 3). %K variation from NGR complements the gap of K counts from 8.17 to 9.45 m CCSF-A (Fig. 2). The K and Ti counts also exhibit a similar pattern of variability (Fig. 2). Biogenic or precipitated carbonate dominates at Site U1461, with high Ca intensity and minor variation. We focus on the pattern of K normalized against Ca. It has been previously reported that K/Ca ratios are interpreted to reflect the variation in riverine run-off from the Australian continent in NWA (Kuhnt et al. 2015;Stuut et al. 2014). Fig. 3 Site U1461 age-depth model by radiocarbon dating. Bchron Parnell et al. 2008) age-depth models with calibrated ages using Marine 13 (Reimer et al. 2013

Variability in the Australian Summer Monsoon from the Pleistocene and the Holocene
It has been suggested that the ASM was weak or inactive during glacial periods (Wyrwoll and Miller 2001;Gallagher et al. 2014;De Deckker et al. 2014). This relative inactivity explains the extremely low sedimentation rate prior to 13.5 cal kyr BP at Site U1461 (Figs. 3 and 4), suggesting a weak or absence of terrigenous input from the continent during this period. The rapid increase in the sedimentation rate is at1 3.5 ka (Figs. 3 and 4), when the %K and K/Ca ratios at Site U1461 remain relatively low from~13.5 to 11.5 ka (Fig. 4), which reveal that the ASM intensification occurred at 11.5 ka during the transition from the Late Pleistocene to the Holocene. This intensification of the ASM corresponds to a change in lithological units from subunit IIa to unit I, marked by an increased quartz content. These increases in quartz content and %K also suggest that the terrigenous input from the continent increased due to the enhanced ASM-driven precipitation. A marine sediment core off the Cape Range region (core Fr10/95-GC17, Fig. 1) records evidence of the ASM intensification/onset at~13 ka, which was interpreted to be due to a southward shift in the ITCZ ). In addition, an increase in the chlorite content in core Fr10/95-GC17 indicates that Leeuwin Current strengthened at this time since chlorite originated from the Fortescue and De Grey rivers (Gingele et al. 2001a(Gingele et al. , 2001b, and Leeuwin Current transported these clays westward to the Cape Range region ). Wyrwoll and Miller (2001) analysed the lacustrine and fluvial sediments from the Kimberley region and suggested that ASM was active after 14 ka. Similarly, peat sediments in north Kimberley record the onset of monsoonal activity after~14 ka, as evidenced by the increasing organic content in a spring (Field et al. 2017). The Kimberley region is the northern region compared with Site U1461, suggesting that the timing of onset/intensification of ASM due to the southern shift of ITCZ was earlier than Site U1461.
The transgression associated with the deglacial sea level rise shifted Site U1461 from an inner-(0-50 m) to a mid-ramp (50-120 m) position at around 14 ka (Fig. 4). On the modern shelf, the siliciclastic content decreases from the inner-to the mid-ramp due to the impact of storm waves and tidal currents (James et al. 2004). The observed increase in K/Ca ratios therefore cannot be explained by a change in the depositional environment due to sea level variations.

Climate variability in the NWA during the Early Holocene
Increasing K/Ca ratios and %K from 11.5 to 8.5 ka reveal the terrigenous input from the Australian continent (Fig. 4). Intensified precipitation during the ASM would have enhanced erosion, caused greater drainage by rivers, and increased the amount of riverine run-off bringing terrigenous components to the NWA continental shelf. Site U1461 directly receives sediments from the NWA continent through the Fortescue and De Grey rivers (James et al. 2004). The enhanced precipitation in the drainage catchment of the Fortescue and De Grey rivers is interpreted to have been caused by the southward migration of the ITCZ that began during the Early Holocene.
Stalagmite records from Ball Gown Cave (Denniston et al. 2013) (Fig. 1) reveal the trend of enhanced ASM from 11 to 8.4 ka, whereas the variation in the terrigenous input from marine sediment off the Cape Range region ( Fig. 1, MD00 -2361- , Stuut et al. 2014 suggests an increase to 8 ka (Fig. 4). The ITCZ southern migration shifted the rain belts to the south, explaining the increased pattern of precipitation in the NWA region. This ITCZ migration is consistent with the records from the latitudinal cores across the Timor Sea (Kuhnt et al. 2015).

Climate variability in the NWA after the Early Holocene
The decrease in the K/Ca ratios and %K content revealed a reduction in terrigenous input after 8.5 ka at Site U1461 (Fig. 4), indicating a weakening of erosion and riverine run-off in the drainage area due to reduced precipitation. The northern migration of the ITCZ moved the rain belt away from NWA after 8.5 ka, while the northern cores across the Timor Sea show the shift to dry condition between 8.1 and 7.3 ka (Kuhnt et al. 2015). However, this apparent dry shift is not strongly observed in the record from the Cape Range region (Fig. 4, Stuut et al. 2014). The stronger covariation between the record at Site 1461 with continental archives may be related to its proximal position at the continental shelf, which is directly influenced by the continental climate variability in NWA.

Conclusions
The IODP Expedition 356 cored the upper slope to the outer shelf carbonates at Site U1461 at a present-day latitude of 20°S. The site is close to the Australian coast, which makes Site U1461 ideally located to record the ASM intensity. Indeed, this area has received fluvially sourced siliciclastic sediment over the last five million years. The focus of this work is the upper 14 m, where a near-complete 14,000-years record of the variability of the ASM is preserved. Shipboard NGR (%K) and XRF-derived K/Ca ratios chart a detailed LGM to Holocene record of terrigenous input through riverine run-off.
Increased %K and K/Ca ratios from 11.5 to 8.5 ka reveal the enhanced ASM precipitation due to the ITCZ Fig. 4 Comparison of paleoclimatic records related to ASM variability in the past 15,000 years. a Ball Gown Cave stalagmite records (Denniston et al. 2013). b Fe/Ca records of core MD00-2361 recovered from the continental slope of Western Australia (Stuut et al. 2014 southward shift. This increased precipitation trend is also observed in a speleothem record from Ball Gown Cave and sediment records from the Cape Range region. This trend is caused by the southern shift of the ITCZ, which brings the seasonal rain belt to the NWA. A reduction of %K and K/Ca ratios after 8.5 ka is interpreted as a shift to more arid conditions in the NWA, caused by the northward shift of the ITCZ.
Abbreviations ASM: Australian Summer Monsoon; IODP: International Ocean Discovery Program; ITCZ: Intertropical Convergence Zone; LGM: Last Glacial Maximum; NGR: Natural gamma radiation; NWA: Northwestern Australia thank the handling editor Ken Ikehara and the two anonymous reviewers whose comments greatly improved the manuscript.

Funding
Funding was provided by the Australian IODP office and the ARC Basins Genesis Hub (IH130200012) to S.J.G., the Japan Society for the Promotion of Science (JSPS) KAKENHI (16J04542) to T.I. and KAKENHI (15KK0151, 17H01168) to Y.Y., the DFG (German Science Foundation, Project 320220579) to L.R., NSF -United States Science Support Program (USSSP), OCE-1450528 for C.M., and the Deutsche Forschungsgemeinschaft (VL96/1-1) to D.D.V.. This work is part of a DAAD project DANA between S.J.G. and D.D.V.. The production of this paper was supported by a National Institute of Polar Research publication subsidy.
Availability of data and materials Radiocarbon date are available in Table 1. Please contact author for other data requests such as NGR and XRF.
Authors' contributions TI carried out this research with input from the co-authors. CMM and LR conducted the XRF core scanning, and YY proceeded with the radiocarbon measurements. YY, CMM, LR, DDV, and SJG contributed to the discussions and improved the manuscripts. TI, CMM, LR, DDV, and SJG were part of the IODP Expedition 356 Science Party. All authors approved the final manuscript.