Development of a protocol to obtain the composition of 1 terrigenous detritus in marine sediments -a pilot study from 2 International Ocean Discovery Program Expedition 361 3

Abstract The geochemical and isotopic composition of terrigenous clays from marine sediments can provide important information on the sources and pathways of sediments. International Ocean Discovery Program Expedition 361 drilled sites along the eastern margin of southern Africa that potentially provide archives of rainfall on the continent as well as dispersal in the Agulhas Current. We used standard methods to remove carbonate and ferromanganese oxides and Stokes settling to isolate the clay fractions. In comparison to most previous studies that aimed to extract the detrital signal from marine sediments, we additionally applied a cation exchange wash using CsCl as a final step in the sample preparation. The motivation behind the extra step, not frequently applied, is to remove ions that are gained on the clay surface due to adsorption of authigenic trace metals in the ocean or during the leaching procedure. Either would alter the composition of the detrital fraction if no cation exchange was applied. Moreover, using CsCl will provide an additional measure of the cation exchange capacity (CEC) of the samples. However, no study so far has evaluated the potential and the limitations of such a targeted protocol for marine sediments. Here, we explore the effects of removing and replacing adsorbed cations on the clay surfaces with Cs+, conducting measurements of the chemical compositions, and radiogenic isotopes on a set of eight clay sample pairs. Both sets of samples underwent the same full leaching procedure except that one batch was treated with a final CsCl wash step. In this study, organic matter was not leached because sediments at IODP Site U1478 have relatively low organic content. However, in general, we recommend including that step in the leaching procedure. As expected, significant portions of elements with high concentrations in seawater were replaced by Cs+ (2SD 2.8%.) from the wash, including 75% of the sodium and approximately 25% of the calcium, 10% of the magnesium, and 8% of the potassium. Trace metals such as Sr and Nd, whose isotopes are used for provenance studies, are also found to be in lower concentrations in the samples after the exchange wash. The exchange wash affected the radiogenic isotope compositions of the samples. Neodymium isotope ratios are slightly less radiogenic in all the washed samples. Strontium and Pb isotopes showed significant deviations to either more or less radiogenic values in different samples. The radiogenic isotopes from the CsCl-treated fractions gave more consistent correlations with each other, and we suggest this treatment offers a superior measure of provenance. Although we observed changes in the isotope ratios, the general trend in the data and hence the overall provenance interpretations remained the same. However, the chemical compositions are significantly different. We conclude that a leaching protocol including a cation exchange wash (e.g. CsCl) is useful for revealing the terrestrial fingerprint. CEC could, with further calibration efforts, be useful as a terrestrial chemical weathering proxy.


Abstract
The geochemical and isotopic composition of terrigenous clays from marine sediments can provide important information on the sources and pathways of sediments.International Ocean Discovery Program Expedition 361 drilled sites along the eastern margin of southern Africa that potentially provide archives of rainfall on the continent as well as dispersal in the Agulhas Current.We used standard methods to remove carbonate and ferromanganese oxides and Stokes settling to isolate the clay fractions.In comparison to most previous studies that aimed to extract the detrital signal from marine sediments, we additionally applied a cation exchange wash using CsCl as a final step in the sample preparation.The motivation behind the extra step, not frequently applied, is to remove ions that are gained on the clay surface due to adsorption of authigenic trace metals in the ocean or during the leaching procedure.Either would alter the composition of the detrital fraction if no cation exchange was applied.Moreover, using CsCl will provide an additional measure of the cation exchange capacity (CEC) of the samples.
However, no study so far has evaluated the potential and the limitations of such a targeted protocol for marine sediments.Here, we explore the effects of removing and replacing adsorbed cations on the clay surfaces with Cs + , conducting measurements of the chemical compositions, and radiogenic isotopes on a set of eight clay sample pairs.Both sets of samples underwent the same full leaching procedure except that one batch was treated with a final CsCl wash step.In this study, organic matter was not leached because sediments at IODP Site U1478 have relatively low organic content.However, in general, we recommend including that step in the leaching procedure.
As expected, significant portions of elements with high concentrations in seawater were replaced by Cs + (2SD 2.8%.) from the wash, including 75% of the sodium and approximately 25% of the calcium, 10% of the magnesium, and 8% of the potassium.
Trace metals such as Sr and Nd, whose isotopes are used for provenance studies, are also found to be in lower concentrations in the samples after the exchange wash.
The exchange wash affected the radiogenic isotope compositions of the samples.
Neodymium isotope ratios are slightly less radiogenic in all the washed samples.
Strontium and Pb isotopes showed significant deviations to either more or less radiogenic values in different samples.The radiogenic isotopes from the CsCltreated fractions gave more consistent correlations with each other, and we suggest this treatment offers a superior measure of provenance.Although we observed changes in the isotope ratios, the general trend in the data and hence the overall provenance interpretations remained the same.However, the chemical compositions are significantly different.We conclude that a leaching protocol including a cation exchange wash (e.g.CsCl) is useful for revealing the terrestrial fingerprint.CEC could, with further calibration efforts, be useful as a terrestrial chemical weathering proxy.

Introduction
IODP Expedition 361, with sites located along Africa's southeastern and southern margin, sought to assess secular variability in Agulhas Current properties and southern African hydroclimates (Fig. 1).To approach documenting changes in either the spatial distribution of rainfall variability or to assess the relationship between sediment sources around southern Africa and Agulhas Current flow, measurement of radiogenic isotope ratios on the clay size fraction of marine sediment was part of the research plan.
Strontium (Sr) and neodymium (Nd) isotopic variations produced by long-lived radioactive decay are well known as powerful tools for the study of provenance of silicate detritus in marine sediment cores.Rb-Sr and Sm-Nd isotopic systems, reflecting the weighted average of detrital components generated in different catchments, provide information on past events of crustal generation and thus represent a powerful means to investigate the provenance of river sediments (Allègre et al., 1996;Goldstein et al., 1984).Sediments derived from older crust have less radiogenic Nd and often more radiogenic Sr isotopes than those derived from younger crust, and a broad inverse relationship between 87 Sr/ 86 Sr and Nd is thus observed.Prior studies have successfully used the 87 Sr/ 86 Sr and εNd signatures of lithogenic fractions of terrigenous detritus to infer provenance, transport patterns and continental weathering states on the African continent during the Pleistocene and Holocene (Franzese et al., 2009;Franzese et al., 2006;Hahn et al., 2016;Jung et al., 2004;Rutberg et al., 2005;van der Lubbe et al., 2016).However, none of these studies aiming to target the composition of the detrital fraction have applied a cation exchange wash, which highlights the need for this methodological study.Several processes may make the relationship of the geology in the source area and the ultimate end-members observed at the sedimentary sink difficult to deconvolve, including sediment recycling (e.g.Haughton et al., 1991) differential weathering (Jiménez-Espinosa et al., 2007) sediment sorting by the energy of the transport medium, and diagenetic alteration in the deposited sediments (Morad et al., 2000;Zabel et al., 1999).
Cation exchange washes (e.g.MgCl2) are employed in the preparation of marine sediment to desorb elements from clay surfaces (Gutjahr et al., 2007;Tessier et al., 1979).Gutjahr et al. (2007) employed this method to remove clay-adsorbed ions before leaching the authigenic ferromanganese fraction which was then used for investigating the composition of bottom seawater in the North Atlantic, thus allowing the reconstruction of past water mass signatures (e.g.Gutjahr et al., 2007).
For the provenance work that is the goal of this project, both the adsorbed and authigenic fractions need to be removed, and we do not plan to use the dispersed ferromanganese fraction in the sediment fines for water mass proxy measurement.
Thus, we decided to perform all of the leach steps as well as the clay separation before the ion exchange washing of the sample.We chose to use CsCl for the cation exchange wash because our goal is to obtain major and trace element chemistry as well as radiogenic isotopes from the clay fraction, and Mg is a major element that we would like to quantify.In contrast, Cs is a very minor element in clays and thus using it as an exchange reagent allows a robust measurement of the Mg concentration of the clays.CsCl is an adequate reagent for the extraction of exchangeable ions as caesium has a large ionic radius but a low charge.This gives it a small hydration sphere, allowing it to form strong bonds with clay surfaces better than any other monovalent or divalent cations.Furthermore, because the Cs exchanged onto the surface will be at levels equivalent to a major element, its abundance in the exchanged clay samples will provide a measure of the concentration of ions displaced from the clay surface, as well as the cation exchange capacity (CEC) of the sample.We reasoned that the CEC of the clays may be related to their weathering history although this is not necessarily a simple relationship.
Although this is a minor procedural change, for completeness we decided to determine how the application of a cation exchange wash may impact radiogenic isotopes and other geochemical measurements, as well as clay mineralogy on the same sediment fraction during sample preparation.To our knowledge, this is the first such study to be reported.We selected samples from one of the Expedition 361 marine sediment cores in the vicinity of the Limpopo River mouth as this location receives significant detrital input from land.(Hall et al., 2017).

Oceanographic
The Delagoa Bight is a distinct indentation of the continental margin (Lamont et al., 2010) into which the Limpopo River, the second-largest eastward-draining river in Africa, deposits sediment.However, terrigenous material in sediments in that regime can have several potential other source areas including the Incomati, Matola, and Lusutfu river catchments.Surface sediment analysis at a core site very close to Site U1478 demonstrates that the heavy mineral assemblages can be attributed to source rocks in the Limpopo catchment.However, a contribution of Matola and Lusutfu end-members were also detectable (Schüürman et al., 2019).
The Delagoa Bight´s hydrography is influenced by the southerly flowing waters from the Mozambique Channel and the East Madagascar Current (Lutjeharms, 2006) the confluence of which forms the Agulhas Current.The Limpopo catchment is a little over 410,000 km 2 and its watershed includes parts of Botswana, Zimbabwe, southern Mozambique and northern South Africa (shaded yellow; Fig. 1).The Limpopo River has a mean annual discharge of ~170 m 3 /s and delivers annual sediment loads of 33 Mt/y (Milliman and Meade, 1983).Its system comprises several tributaries with the most significant rivers being the Pienaars; Olifants, Serobaneng, Diep, Upper Mzingwane and Ngez.
Under modern climate, the Delagoa Bight´s catchment lies in the transition between tropical and subtropical climate zones, just south of the subtropical ridge between the southern Hadley and the Ferrel cell (Tyson and Preston-Whyte, 2000).Most of the year, surface airflow is from east to west and is stronger during the summer.
During winter, the South Indian Ocean anticyclone moves west and the region is under strong subsidence and anticyclonic circulation with very little rainfall.The topographic configuration of the high interior, the escarpment and the coastal lowland and the eastward propagating migratory anticyclones during the synoptic cycle create coastal shallow low-pressure cells associated with Berg winds blowing down the mountains in an offshore direction (Reason and Jury, 1990).
The Limpopo River drains the oldest terrane on the African continent.The sediment load (~35 x~ 10 6 ton/yr) is supplied by southern tributaries draining the Kaapvaal Craton (Eglington and Armstrong, 2004) and northern tributaries draining the Zimbabwe Craton (Jelsma and Dirks, 2002) as well as Karoo siliciclastic and basaltic cover.The geology of the highlands is Archean, composed of the Zimbabwe and Kaapvaal Cratons joined by the Limpopo Belt.Bedrock in the cratons, emplaced as early as 3.8 Ga, is characteristically tonalite-trondhjemite-granodiorite, an intermediate, phaneritic, intrusive igneous rock rich in plagioclase and mafic minerals (Kreissig et al., 2000).The Limpopo Belt is a greenstone belt, composed of amphibolite-granulite grade metasediments formed between 2.9 and 2.7 by the collision of the Zimbabwe and Kaapvaal Cratons, the root of a neo-Archean orogenic complex (Kreissig et al., 2000).(Garzanti et al., 2014).We note that Sr isotope ratios are affected by grain size and that the study conducted by Garzanti et al. (2014) used a higher size fraction (<32 μm fraction) than targeted here.

Sediment leach
The detailed laboratory protocol can be found in the research method journal MethodsX.Eight samples were selected from sedimentary depths between 198 and 214 m core depth, which approximately, spans the time interval 2.4-2.8Ma (Hall et al., 2017).In general, the effects of sediment diagenesis under methanogenic conditions in the anoxic zone of the sedimentary column should be considered when sampling to analyse the authigenic phases of marine sediments.This is crucial to consider as, within the methanogenic zone of the sedimentary column, which is anoxic, the originally present authigenic Fe-Mn oxyhydroxides are expected to be at least partially and likely completely removed or altered into other diagenetic mineral phases such as pyrite.Hence the authigenic phases still present in these sediments will have quite different chemical properties and reactivities than sediments in, for example, the sulphate-bearing zone of the sedimentary column.
The samples were sieved to separate the >63μm fraction, to save foraminifera for further analyses.The bulk fine sediment (<63μm), (1.7-2.1gmaterial per sample) was decarbonated using 1 M buffered acetic acid solution following the procedure outlined in Gutjahr et al. (2007).This step was repeated 1-3 times until once beyond the application where the sediment no longer appeared to react with the acid.
Following carbonate removal the Fe-Mn oxyhydroxide coatings were dissolved by leaching the samples for 24 h, with agitation on a rocker table, at room temperature in a 0.05 M hydroxylamine hydrochloride (HH)-15% distilled acetic acid solution, buffered to pH 4 with analytical grade NaOH.The HH leaching was followed by centrifugation and triple rinses in deionized water.This sequence of procedures deviates from that of Gutjahr et al. (2007) who followed the initial carbonate leaching with a cation exchange step with MgCl2 to desorb cations before leaching the dispersed Fe-Mn oxy-hydroxide for authigenic Nd isotope measurements.
Furthermore, we did not use the complexing reagent EDTA as originally suggested in the protocol of Gutjahr et al. (2007) during the leaching of the authigenic Fe-Mn oxyhydroxide fraction.
The separation of particles smaller than 2-micron diameter was carried out by settling in a column of water and adding 2mL of 0.5% sodium metaphosphate (calgon) to each tube for the first 2 settling rounds.Particles are assumed to settle in the water according to Stokes Law.For the clay fraction (<2 m) samples were settled for at least 3 hours and 52 minutes.Settling rounds were performed until the water was clear, usually 15 to 20 iterations.Clods (that is small chunks of clay that are stuck together so the sample can be weighed as a single grain) for the analysis of 40 Ar* ages (Gombiner et al., 2016;Hemming et al., 2002) were created after the first settling round by pipetting approximately 300 g of suspended sediment solution into a micro-centrifuge tube, centrifuging it at high speeds for 30 minutes, and then drying it in the oven at 60°C.
After the clay separation, a Cs + cation exchange wash was performed on aliquots of the 8 samples using a 0.1 M CsCl solution.From here forward, this group of samples is referred to as the washed group.After centrifugation and removal of water, 10 mL of ethyl alcohol was added to the sample and vortexed to rinse away excess salt, then the samples were centrifuged, and decanted and the rinsing was repeated a second time.Following the rinsing, samples were dried in an oven at 60°C.Subsamples for measuring 40 Ar* were also taken from the aliquot containing CsCl after drying.
In this study, organic matter (OM) was not leached because sediments at IODP Site U1478 have relatively low organic content, varying from 0 to 1.25 wt% with an average of 0.45 wt% for the shipboard samples (Hall et al., 2017).However, we acknowledge that even in trace amounts in the clay fraction, OM could significantly impact CEC (pH-dependent).Thus we include a step to leach organic matter in the suggested laboratory protocol.However, it should be considered that in some cases, depending on the sediment properties, an organic oxidative leaching step with hydrogen peroxide (3%) would still not be sufficient to remove various organic phases.Stronger treatments for OM removal can, however, attack the clays and leach away soluble elements including the REE that would affect the trace element measurements.Hence this step might have to be adapted individually according to the setting and sediment composition.

Sediment dissolution
About 20 mg samples of homogenized and powdered material from the clay fraction were weighed and completely dissolved using a mixture of concentrated HF-HNO3-HClO4.
The dissolution of the sediment was conducted by pipetting 500µl concentrated nitric acid (double-distilled 16N HNO3) into the sample beaker and sonicating for ~20 minutes to 1 hour to disaggregate the material.Thereafter 200µl of perchloric acid (optima grade) was added to the sample to remove organic matter and renewed sonication of the solution in the beakers was performed.Afterward, 600 µl of concentrated HF (optima grade) was pipetted into the sample and the beakers were heated on a hot plate at 150°C overnight (>12 hours).After 12 hours, the solution should be completely clear and have no suspended solids.In some cases, the sample had a black ring around the edge of the beaker likely resulting from undigested organic matter.These samples required the addition of further perchloric acid and heating.We performed a first evaporation step by heating the beakers uncapped on the hot plate at 200°C.Thereafter 1 mL of 8N HNO3 was added to the beaker to perform the second dissolution step.Samples were heated in capped beakers at 150°C on the hot plate for 2 or more hours to ensure full dissolution.Thereafter beakers were uncapped to perform the second evaporation and heated at 175-200°C.The final dissolution of the samples was performed by adding 1mL 4N HNO3 and heating the solution at 150°C and occasional sonification until samples have gone into full solution.
Sediment standard from the Japanese Geologic Survey, JLk-1 (Japanese lake sediment) was dissolved alongside the 16 unknown aliquots used in this study (Imai et al. (1996).

Major/trace element analysis
Major and trace element analyses were carried out using an Agilent Technologies  (1996).Precision is estimated by comparing sample replicates.For precision, we report the sample averaged percent standard deviation for each element (2SD).
Precision for the major elements except Ca is better than 2%.The precision for trace elements reported here is better than 4%.Due to high Ca background and low sample Ca abundance, precision for that element is 12%.For one run, Ca had to be discarded.To achieve an accurate measurement of Cs, we applied the standard addition technique.We added 40 l of a 100 ppm Cs solution (4 g of Cs) to an aliquot of each Cs washed sample.To determine sensitivity to Cs, this was measured alongside an unspiked aliquot of that sample.
Concentrations of Cs determined using standard addition were converted to cation exchange capacities using the equation: Where [Cs] is the concentration of Cs in a sample, in ppm.MCs/QCs represents the equivalent weight of a Cs atom, where MCs is the molar mass of Cs, and QCs is the valence of Cs (+1).The 10 is a conversion factor to convert equivalent weight, usually reported in grams/equivalent, to 100g/milliequivalent (meq).Using this equation will result in units of meq/100g, a common unit for reporting cation exchange capacity.

Radiogenic isotope analysis
The Sr, Nd, and Pb isotopic measurements were performed at the Lamont-Doherty Earth Observatory (New York) using a ThermoScientific Neptune Plus multicollector ICP-MS in static mode.Pb was extracted using Bio-rad AG®1-X8 resin and HBr and HCl column chromatography.Nd was extracted using Tru-Spec® resin followed by Eichrom Ln-Spec® resin and HNO3.Sr was extracted using Eichrom Sr-Spec® resin.(Jacobsen and Wasserburg, 1980).

K/Ar age analysis
Aliquots of the clay fraction were removed from the <2 m fraction of both washed and unwashed samples and dried in 1 mL micro-centrifuge tubes after the settling process.Samples were removed from micro-centrifuge tubes, weighed, and loaded into a 32-spot copper disk -typically in pairs of duplicates for 14 samples -plus 4 GLO-1 glauconite standard (Odin, 1976).To obtain a reproducible age for a sample, we found it important to weigh and load the entire clod.Sample weights in this study ranged between 0.1 and 0.5 mg.
For each measurement session, the pattern of runs includes blanks at the beginning and end of the session bracketing the entire sequence of runs.Each air standard was bracketed by a blank measurement, and each set of 3 unknowns or GLO-1 measurements is bracketed by a blank/air/blank sequence.Signal intensities of the airs, unknowns, and GLO-1 are well above the background levels, and blank corrections are estimated from the most recently measured preceding blank.Air standards are used for quality control, to correct for any instrument drift, and to estimate the correction for the atmospheric component in the samples to yield the 40 Ar* (that is the radiogenic argon from the decay of K in the sample).GLO-1 standards are used to quantify the sensitivity of the instrument during the session.
GLO-1 has 1.109e -12 moles of 40 Ar* per mg of sample (Odin, 1976).The sensitivity of the mass spectrometer/extraction line system for argon is ~5.13e 13 nannoamps/mole.Signals are measured on an analogue multiplier with high voltage set such that 1 nannoamp is approximately 10 mv on a 10 11  resistor.The reproducibility of the age of GLO-1 is between 1 and 2 % for most sessions (2 SD).
Reproducibility of the sediment clods is not generally as good as for the GLO-1, and when the difference between aliquots is >10%, additional aliquots of the sample are repeated in another session.

Clay mineral analyses
Clay mineralogy determinations were performed under three conditions, i.e., airdrying, ethylene-glycol solvation for 24 h, and heating at 490 °C for 2.5 hours, by standard X-ray diffraction (XRD) on a PANalytical X'Pert Pro diffractometer in Key Laboratory of Submarine Geosciences, China, SOA with CuKα radiation and Nifilter, under a voltage of 45 kV and an intensity of 40 mA.Identification of clay minerals and calculation of their relative proportions were made according to the position and area of the (001) series of basal reflections on the three XRD diagrams (not shown), i.e., smectite (001) including illite/smectite mixed-layers at 17Å, illite (001) at 10 Å, and kaolinite (001) and chlorite (002) at 7 Å (Biscaye, 1965).Relative proportions of kaolinite and chlorite were determined based on the ratio from the 3.57/3.54Å peak areas.Additionally, illite crystallinity was obtained from the full width at half maximum (FWHM) of the 10 Å peak (Esquevin, 1969).Semiquantitative calculations of all the included peak parameters were performed on the glycolated curves using Jade6.Clay mineral analyses were performed on oriented mounts.Results on the washed fraction can be found in the supplementary information.

Major and Trace Elements
Here we assess which elements were exchanged during the procedure as well as the effectiveness of the exchange wash.Figure 2A depicts how the wash changes the composition of the samples via cation exchange.The differences in concentrations, including uncertainties, between washed and unwashed aliquots are shown.Note the log scale, used to accentuate changes in concentration.All of the data reported here are from samples dissolved with HF, so Si was lost.Most notable here is that by far and away, Cs is replacing Na on the clay surfaces.Sodium concentration is reduced by 75% on average.Less outstanding is the loss of other soluble major elements: Ca (22%), Mn (14%), Mg, (14 %), and K (8%).Ca concentrations in the clay are extremely low (~0.12 g/100g), and tuning the quadrupole mass spectrometer to achieve precise measurements on this low abundance major element, the other major elements, and trace elements simultaneously was a challenge.Concentrations of elements predominantly integrated into the structure of clay minerals, such as Al, Fe, and Ti, are not changed by the wash outside of the reported analytical uncertainty.Concentrations of trace metals Sr and Nd, used by geochemists as isotopic tracers, appear to decrease as a result of the exchange wash by 25%.
We test the effectiveness and quality of the exchange using a charge balance.
However, this balance is not trivial because mass is not conserved in the exchange reaction.Replacing light elements like sodium with the heavier element caesium creates an appreciable weight difference between the washed and unwashed samples.So first we simulate a "mass balance" by estimating what percent of each major element is lost, on average.We determine what fraction that element contributed to the total losses by summing those percents and normalizing them to 100.For example, about 75% of Na is lost in each sample during the washing, but this represents only ~60% of total ions exchanged.To simulate this "mass balance," we model the weight change of replacing each element (e.g.Na, 23g/mol) with Cs (133 g/mol) at the percent it contributed to the cations exchanged.The above calculations reveal each sample is ~5g/100g heavier after the exchange wash.The calculation process artificially removed this extra weight.
To complete the charge balance, we take exchangeable major element oxides (Na, Mg, K, Ca, and Mn) and convert them first to moles, and then to equivalents using their atomic charge.Ideally, the charge of the exchangeable cations lost and the charge of Cs + ions gained should sum to zero.Therefore, we visualize the charge balance as a cumulative sum (Fig. 2B).The bulk of the charge displacement comes from Na and Mg, as noticed previously.Of the 8 samples, 4 return to zero within error.3 samples gained more Cs than they lost exchangeable cations.One sample does not appear to have gained enough Cs.See the discussion section for potential explanations for this apparent charge imbalance.We admit that these results are troubling, as they would cause us to question the accuracy of our Cation Exchange Capacity measurements.We believe this is a function of accumulated uncertainties, but more tests should be performed to increase the confidence in the CEC.
40 K/ 40 Ar ages from the washed samples are systematically younger than from the unwashed aliquot (Fig. 3C) and they have lower %radiogenic 40 Ar.Of 8 sets of samples, 7 unwashed samples are 4-10% older than their washed counterparts.
Two unwashed sample, MS22 and MS 34 yielded ages which are 15 -22 % older, respectively.The unwashed samples, in general, are less reproducible (Fig. 3C), and subsequently we found that because of this small-scale heterogeneity it is necessary to weigh the entire clod to get consistent results.
There are many reasons why different isotope systems used as provenance indicators may yield different results from each other, such as the varying temperatures at which different systems close and how they are affected by diagenesis, weathering, and clay mineral formation.However, in principle, we would expect that they show some similarities, reflecting the chemistry of their source regions.Therefore, we compare trends in isotope systems and quantify these using linear regressions.We argue that if the correlation changes, it is most likely the result of authigenic phases biasing the radiogenic isotope signature.
We present a linear regression of each isotope ratio used for provenance to all the others for both the unwashed (Fig. 4A-F, left panel) and the washed group (Fig. 4A-F, right panel).The application of our procedure resulted in trends in isotopic compositions within the washed group that more resemble each other: the regressions from the washed group have higher correlation coefficients.These data demonstrate how the wash affected the composition of radiogenic Sr in our samples.
Before the wash, εNd does not have a strong correlation with 87 Sr/ 86 Sr (R 2 = 0.25) (Fig. 4A).There is weak correlation between the 40 Ar model age and 40 K/ 40 Ar with 87 Sr/ 86 Sr (R 2 = 0.2, 0.04)(Fig.4B-C).When radiogenic isotope measurements are compared amongst the washed group, the trends between each isotope system are more obviously systematic.The relationship of 87 Sr/ 86 Sr to εNd and 40 K / 40 Ar is most improved with an R 2 improving from 0.25 to 0.6 and from 0.05 to 0.49 respectively (Fig. 4A; C).
As expected, the 40 Ar model age, calculated assuming a uniform 2% K, correlates strongly with the 40 K/ 40 Ar age, which is calculated with the measured K (R 2 = 0.87).
The 40 K/ 40 Ar age correlates more strongly with εNd than the 40 Ar model age (Fig. 4D, E), (R 2 = 0.62 vs R 2 = 0.56).The relationships of all of these isotopic tracer proxies are improved by the wash, but less dramatically than for Sr isotope composition.
Matrices of correlation coefficients were generated to evaluate the effects of the procedure.Fig. 4G shows the R 2 coefficients before the procedure, amongst the unwashed group, and Fig. 4H shows the R 2 coefficients after.The effects of the procedure are best visualized by subtracting the before (Fig. 4G) and after (Fig. 4H) matrices to generate a matrix of anomalies (Fig. 4I).Correlations of 87 Sr/ 86 Sr to

Clay mineralogy
The major clay minerals in the samples were identified by X-ray diffraction techniques (Fig. 5).In general we find that illite is the most abundant, smectite is second, and kaolinite is third.Chlorite could not be detected in the 8 samples.Illite values range from 29 to 50 % in the unwashed fraction and tend to increase downcore.Smectite has a range of values from 24 to 42% and showing a decreasing trend over time.Kaolinite seems to vary around a mean between 21-30% (Fig. 5).Further details on the clay minerals in the washed aliquot can be found in the SI.

Chemical Index of Alteration (CIA) and Cation exchange capacity (CEC) of the clay fractions
The Chemical Index of Alteration (CIA), (Fig. 3D) estimates the extent of weathering of continental rocks and sediments, where higher values indicate greater alteration, and is very useful in the interpretation of marine sediments (Nesbitt and Young, 1982b).
CIA = 100 *  2  3 /( 2  3 +  +  2  +  2 ) Cation exchange capacity (CEC) is used in agriculture as an assessment of soil fertility, indicating the capacity of the soil to retain several nutrients (e.g.K + , NH4 + , Ca 2+ ) in plant-available form.The extent of weathering of soil can be estimated from its CEC, as the availability of exchangeable cations depends on clay mineralogy, which is a function of the substrate composition and degree of chemical weathering.
We calculated the index for both sets of samples and find that the CIA based on the CsCl treated samples has higher values than the non-treated aliquots (Fig. 3D).CEC of samples has a range from 42 to 49 over the core depth covered in this study (Fig. 3E).Measurements of CEC and CIA have an inverse relationship (R 2 =0.47)(Fig.6).
High measurements of CIA indicate high chemical weathering in rocks, given that aluminum remains in the weathering residue.Low CEC indicates a breakdown of primary minerals to secondary clay minerals as a consequence of weathering.Here, a lower CIA, suggesting lower chemical weathering extent, correlates negatively with higher measurements of CEC, suggesting lower degrees of secondary mineral formation.

Major and trace elements abundant in seawater removed by exchange wash
Our geochemical results demonstrate that Na concentrations are most affected by the CsCl rinse, reduced by 75% (Supplementary Table, Fig. 2A).This suggests that much of the material bound to the clay surface is Na + ions, far and away the most abundant ion in seawater.Cs + replaces Na + , bound to the surface of the clay, during the washing procedure due to the powerful hydration sphere of the Cs + and the high concentration of ions available for the exchange in the 0.1 molar solution.Also affected are Ca, Mg, and K concentrations, the 3 rd , 5 th , and 6 th most abundant elements dissolved in seawater.
This change in our perception of major element chemistry has consequences.The CIA of marine sediment, dependent on concentrations of major elements, is commonly used as both a tracer and to appraise changes in weathering on the landscape (Garzanti et al., 2014)].Here, our CIA measurements are higher using the Cs + treated value compared to the non-treated aliquots (Fig. 3D).This pattern results from the 75% reduction in the Na values in the Cs + washed sediment.We argue that this CIA measurement after the exchange wash provides a more realistic terrestrial signal, appraising the concentrations of only the structural components of the clay.The composition and rate of authigenic cation binding to clay minerals could depend on salinity and pH of seawater, transport pathways, and depositional environment, factors that could change on varying timescales.
We find reduced concentrations of strontium and neodymium in all samples that have undergone the Cs + cation exchange wash.Sr concentration in seawater is relatively high for a trace element, at ~10 ppm (Angino et al., 1966).The source of strontium is rocks of the continental crust delivered to the ocean in sediments, but it is highly soluble like its alkaline earth relative calcium and has a long residence time in seawater of ~5 Myr [Palmer and Elderfield, 1985].Nd is also sourced from the continental crust and sediments but has a much lower concentration (~0.1-0.3 ppb) and much shorter residence time (200-1000 years) (Goldstein and O'Nions, 1981;Tachikawa et al., 2003;van de Flierdt et al., 2016).We attribute the loss of these elements to the fact that seawater derived Sr and Nd that has been adsorbed on the clay's surface in the marine environment are exchanged away with the CsCl solution.The isotopic composition of this material displaced by Cs + ions may have the potential to bias our measurements if not accounted for or removed.
How and when are cations usually adsorbed onto the surface of the clay?Studies have shown that diagenesis of clay minerals may occur when they are transferred from freshwater to seawater via the river system.The mechanism involves reactions of the cations and anions in seawater with the minerals through exchange, whereby a rearrangement of the cations in the exchange positions takes place (Carroll and Starkey, 2013).
Accounting for the conservation of charge in our data was a challenge: 4 of the 8 samples failed to achieve balance from the exchange.3 samples appeared to collect more Cs than equivalents of soluble major elements were displaced, a positive balance.We offer a likely explanation for this set: organic matter in the finest fraction may have absorbed additional Cs in a way that we did not account for.This yields an important takeaway: given the very high cation exchange capacity of organic matter, it could contribute to CEC significantly even with relatively low concentrations of organic matter in bulk sediment.Therefore, we advise a step to oxidize organic matter to be introduced before the cation exchange wash.
Other problems contributing to the imbalance may be rooted in 1) analytical uncertainty in major elements and/or Cs leading to an accumulated error larger than we thought.Ca concentrations were difficult to measure because of its low concentration but Ca did not contribute to the charge balance enough to explain the observed imbalance.2) the partial dissolution of clay minerals during the cation wash, limiting the ability of the clay to hold charge, contributing to a negative balance.This is also highly unlikely given the reagent was a low concentration (0.1 N) salt solution, with a pH of 4.5.More work would be needed to fully evaluate the charge balance.

Effects on radiogenic isotopes
The cation exchange wash may improve the fidelity of the radiogenic isotope proxies.Our logic is that if the different provenance proxies correlate more strongly with one another after the wash, that regression line must be pointing more accurately towards the isotopic composition of the source, or mixing sources.
The present-day seawater 87 Sr/ 86 Sr is 0.70918 (Henderson et al., 1994).In this study, we find 87 Sr/ 86 Sr between 0.74 and 0.77 (Supplementary Table ), which points to a continental detritus origin of the signal.Two out of six samples treated with the Cs + wash yield more radiogenic ratios than before the treatment (beyond the measurement error of 0.00001), which could be explained by the removal of seawater derived Sr from the clay surface, thus altering the ratio towards more radiogenic values (Fig. 3A).Conversely, the remaining samples show the opposite 87 Sr/ 86 Sr trend, which cannot be explained by the removal of seawater derived Sr.
However, in all cases, the removal of adsorbed Sr, whatever its isotopic composition,  Thompson, 2015).Exchangeable cations thus form part of the diffuse layer above the charged surface.These cations can easily be displaced from the surface by other cations from a surrounding solution, as the binding is relatively weak.Cationexchange capacity is measured by displacing all the bound cations with a concentrated solution of another cation, and then measuring either the displaced cations in solution or the amount of added cation that is retained on the solid (Bache, 1976).In this study, we used a CsCl solution because Cs + is a trace element in the natural solids and thus its concentration in the washed sample is a potential measure of the CEC.
CIA and CEC have an inverse relationship and are moderately correlated (Fig. 6).
However, a poor correlation might be expected because the proxies are not sensitive to exactly the same processes.CIA was designed to quantify degree of bedrock weathering, and thus would be sensitive to sediment source sorting processes.A study from Changjiang River basin, China has questioned whether the CIA reflects silicate weathering, and concluded that it probably indicates the integrated weathering history in the river basin, and thus, cannot be used as a reliable proxy of instantaneous chemical weathering (Shao and Yang, 2012).Furthermore, the hydrodynamic sorting also influences the CIA values.On the other hand, the CEC may be sensitive to the climate of the clay minerals' formation, and the residence time of the sediment on the landscape.In one endmember scenario, extremely high rates of physical weathering and erosion could result in the deposition of primary minerals with low CEC, such as the glacial flour in Owens Lake (Bischoff et al., 1997).
In another endmember scenario, low rates of erosion coupled with high soil flushing could produce laterite soils, rich in Fe and Al oxides, which would also have extremely low CEC.
This study shows that a CsCl exchange wash can be applied to marine sediments (sensitive to terrestrial input from land) to derive the cation exchange capacity in clay minerals.The CEC measurement gave a qualitative prediction of the clay mineralogy of the sample, as the proportion of kaolinite to smectite appeared to control CEC.Because clay mineralogy relates to terrestrial weathering processes, there is a potential for CEC to indicate climate conditions and chemical weathering extent.In our data, the CEC is inversely proportional to the weathering index CIA, with more aluminous, high CIA samples correlating with low CEC samples that have more simple clay minerals.

Conclusions
We present a recommended sediment leach protocol for an improved terrigenous signature extraction from marine sediments that includes a cation exchange wash.
We find that ions having higher concentrations in seawater are predominantly displaced by the cation exchange wash, and conclude that the derived geochemical data represents a more accurate terrestrial signal.Therefore, this procedure is favourable for major and trace elements and leads to Nd and Sr isotope ratios that are more truly representative of the terrestrial detritus.The observed magnitude of change in the Sr and Nd isotopes between washed and non-washed aliquots is large.
Nevertheless, whether the Cs + -treatment was used or not, the overall downcore trends and correlations between isotopes are similar, and hence whether or not it is applied, general climate reconstruction conclusions are likely to be similar.
Due to flocculation and rapid settling of clay minerals in Cs-washed samples, it is evident that a dried Cs-washed aliquot is more homogenous, and that 40 K/ 40 Ar age measured on it will be more reproducible and reflect the bulk age of the sediment.
For the accurate determination for the CEC of the clay fraction of marine sediment, we recommend adding to the procedure a step that oxidizes organic matter.This may be important, even if the concentration of organic matter in the sediment is low.
Because of the relationship between cation exchange capacity, clay mineralogy, soil flushing, and ultimately annual precipitation and climate conditions, CEC could, with further calibration efforts, be useful as a terrestrial chemical weathering proxy.
However, CEC's potential in this capacity is difficult to tell due to the limited range of CEC (42-50 meq/100g) in this study.A detailed survey of CEC of clay minerals in different depositional and climate settings would be necessary to relate the CEC of the clay mineral fraction to chemical weathering regimes.Figure 2: A) The magnitude of the reduction in the concentration of an element as a result of the cation exchange wash, expressed as a percent of the concentration of the unwashed group.Sample IDs are listed in table 2-4 for reference.The elements bound to the clay surface in the highest relative concentrations are Na, Ca, Mn, Ba, and Sr. B) The equivalents of cation displaced must equal the equivalents of Cs measured to conserve charge.The residual between the concentrations of the major elements in the test group (washed) and the control group (unwashed) was calculated, then converted to equivalents.Here, we visualize the summation of these charges, element-wise, for each sample.The average of all samples is plotted behind as a thick black line, with a ±1σ standard deviation envelope surrounding it (light grey).Ideally, the cumulative sum of these losses should equal the equivalents of Cs.The more equivalents of Na, Mg, K, and Ca a sample lost, the higher its Cs concentration and cation exchange capacity should be.

Figure captions
P u blis h e r s p a g e : h t t p:// dx.doi.o r g/ 1 0. 1 0 1 6/j.ch e m g e o. 2 0 1 9. 1 1 9 4 4 9 Pl e a s e n o t e: C h a n g e s m a d e a s a r e s ul t of p u blis hi n g p r o c e s s e s s u c h a s c o py-e di ti n g, fo r m a t ti n g a n d p a g e n u m b e r s m a y n o t b e r efl e c t e d in t hi s v e r sio n.Fo r t h e d efi nitiv e v e r sio n of t hi s p u blic a tio n, pl e a s e r ef e r t o t h e p u blis h e d s o u r c e .You a r e a d vis e d t o c o n s ul t t h e p u blis h e r's v e r sio n if yo u wis h t o ci t e t hi s p a p er.This ve r sio n is b ei n g m a d e a v ail a bl e in a c c o r d a n c e wi t h p u blis h e r p olici e s. S e e h t t p://o r c a .cf. a c. u k/ p olici e s. h t ml fo r u s a g e p olici e s.Co py ri g h t a n d m o r al ri g h t s fo r p u blic a tio n s m a d e a v ail a bl e in ORCA a r e r e t ai n e d by t h e c o py ri g h t h ol d e r s .radiogenic isotopes, Highlights • application of CsCl-wash leads to significant change in geochemical signature of the sediment • elements in high concentrations in seawater are replaced by Cs + from the wash • cation exchange wash reveals a truer terrigenous signal • the cation exchange capacity of the sediment may provide additional environmental information quadrupole ICP-MS at the Facility for Isotope Research and Student Training at Stony Brook University.The 16 samples were measured alongside sediment standard JLk.Major and trace elements were measured simultaneously from solutions with 20,000 dilution.A mixture of all of the diluted samples was concocted as a drift solution to monitor and correct for instrumental drift.The pattern of measurements was blank, drift solution, unknown, unknown.Standards were measured with the pattern blank, drift solution, standard, twice each at the beginning, middle, and end of the analytical sessions.Samples were replicated over two sessions that took place within 24 hours.The drift solution was also used to correct the offset between runs.Signal intensities were converted using the signal intensity from 12 measurements on JLk and the concentrations from Imai et al.
40 K/ 40 Ar and Nd are notably improved by the CsCl wash, with the highest anomaly and warmest colors (Fig 4I).
reveals87 Sr/ 86 Sr values that correlate better with other isotopic measurements on the sample, as the CsCl wash improved correlations of Sr isotopes with other sourcebased proxies (εNd and 40 K/ 40 Ar age)(Fig.4).Because of these improvements, the application of the cation wash is recommended for provenance studies employing Sr isotopes on clay minerals.While the observed magnitude of change on 87 Sr/ 86 Sr ratios from the CsCl washed aliquots is large (0.0025-0.08), and the correlations with the other isotopes are better, the overall trends in the correlations are the same, indicating that the general provenance interpretations of studies lacking a cation exchange wash are nevertheless valid.All Cs + -treated samples yield less radiogenic εNd values compared to their counterparts of 0.2-0.7 εNd units.The offsets are beyond the analytical uncertainty endemic to the clays of the sample.Variations in local dissolved seawater cations, the aqueous geochemistry of the river catchment, changes in salinity or pH of seawater, transport processes and duration, and landscape storage are among the list of many potential factors that could affect the chemical composition of cations bound to clay minerals.These might have the effect of biasing that terrestrial signal.ions are adsorbed to soil surfaces by the electrostatic interaction between their positive charge and the negative charge of the surface (Schaetzl and

Figure 1 :
Figure1: Location map of IODP Site U1478 with main surface currents (arrows) in the southwest Indian Ocean and atmospheric circulation over southern Africa during austral summer (December, January, February) with approximate position of the Intertropical Convergence Zone (ITCZ) and Congo Air Boundary (CAB) (dashed lines; adapted fromReason et al. (2006).AC = Agulhas Current, SEC = South Equatorial Current, SEMC = South East Madagascar Current, NEMC = North East Madagascar Current, EACC = East Africa

Figure 4 :
Figure 4: (Left set of panels) Left panels show correlations between the control group (non treated aliquots) and right panels display correlations between test groups (CsClwashed).R 2 correlation coefficients displayed at the bottom of each plot A) Correlations of Nd and 87 Sr/ 86 Sr and (Ma) B) Correlations of 40 Ar model age (Ma) and 87 Sr/ 86 Sr C) Correlations 40 K/ 40 Ar (Ma) and 87 Sr/ 86 Sr. D) Correlations of 40 K/ 40 Ar (Ma) and 40 Ar model age (Ma).E) Correlations of Nd and 40 Ar model age (Ma).F) Correlations of Nd and 40 K/ 40 Ar (Ma).(Right) Correlation matrices were generated to evaluate the effect of CsCl wash holistically.Warmer colors represent higher degrees of correlation.G) R 2 values without the CsCl wash.H) R 2 values of clays washed with CsCl.I) The effects of the procedure are better visualized by subtracting the washed (H) and unwashed (G) matrices, visualizing the difference before and after the procedure.Correlations of 87 Sr/ 86 Sr to 40 K/ 40 Ar and Nd are notably improved by the CsCl wash, depicted with warmer colors.

Figure 5 :
Figure 5: Clay mineral content down core of the 8 test samples from IODP Site U1478 without CsCl treatment.

, climatic, and geologic setting
(Abouchami and Galer, 1998) for Pb in hotplate digestions range from 21-30 pg and the blanks for Nd and Sr are negligible.The MC-ICP-MS instrumental drift is monitored by standard-sample bracketing using JNdi-1 for Nd, NBS 987 for Sr, and NBS 981 for Pb.Typically 30-70 standards and 20-30 samples are analysed during a session and reported 2SD external errors are estimated from the reproducibility of the standards.These are 0.2 for Nd and 0.000019 for 87 Sr/ 86 Sr.For 206 Pb/ 204 Pb we report a standard reproducibility error of 0.001795; 0.0024297 for 207 Pb/ 204 Pb and 0.007317 for 208 Pb/ 204 Pb.We collected 60 ratios for each run using 200 ppb solutions for Sr and Pb and 100 ppb solutions for Nd.Reported values are corrected to those of the international standards.Samples were corrected to the accepted value of the JNdi-1 standard of 0.512115(Tanaka et al., 2000)for143Nd/ 144 Nd, NBS 987 standard of 0.710248 for 87 Sr/ 86 Sr(Weis et al., 2006), and NBS 981 for 16.9405 for 206 Pb/ 204 Pb; 15.4963 for 207 Pb/ 204 Pb and 36.7219 for 208 Pb/ 204 Pb(Abouchami and Galer, 1998).In-run mass fractionations was corrected using 146 Nd/ 144 Nd = 0.7219 and 86 Sr/ 88 Sr = 0.1194 Pb isotopes were doped with Tl in a proportion of Pb/Tl = 5.In-run mass fractionations were corrected using205Tl/ 203 Tl value of 2.388.Quality control was provided by measurements of rock standards BCR-2 and JLK for Pb, BCR-2, JLK and BHVO-2 for Sr, BCR-2, JLK and BHVO-2 for Nd, all of which were repeatedly run through the entire dissolution and chemical separation procedures (Supplementary Table4).The 143 Nd/ 144 Nd ratios of the samples are reported as Nd, which is the parts per 10,000 deviation of the Comparison for 87 Sr/ 86 Sr, average (n=16) 87 Sr/ 86 Sr 2 SE measurement error: 0.000012; B: εNd; average (n=16) 2 SE measurement error: 0.07;C: 40 Ar model age in Ma calculated using the 2% K, error is the ± 1s (Ma); D: Chemical Index of Alteration (CIA, (Nesbitt and Young, 1982a)); E: Cation exchange capacity (CEC,); F: 206 Pb/ 204 Pb, average (n=16) 2 SE measurement error 0.0004 G: 207 Pb/ 204 Pb, average (n=16) 2 SE measurement error: 0.00036 H: 208 Pb/ 204 Pb, average (n=16) 2 SE measurement error: 0.001044; I: 208 Pb/ 206 Pb, average (n=16) 2 SE measurement error: 0.00023; J: 207 Pb/ 206 Pb, average (n=16) 2 SE measurement error: 0.00010.