1. Introduction
The process of chemical weathering of silicate minerals forms a critical part of the Earth’s carbon cycle because the atmospheric greenhouse gas CO2 is consumed during their breakdown and the subsequent fixture of carbon into carbonate sediment in the ocean. Once deposited in this way, carbon can be stored over long periods of geologic time (Berner & Berner Reference Berner, Berner and Ruddiman1997; Kump & Arthur Reference Kump, Arthur and Ruddiman1997) and even returned to the mantle (Kelemen & Manning Reference Kelemen and Manning2015; Clift Reference Clift2017). In order to estimate how much CO2 is being consumed, it is important to know not only the volumes of material involved but also the change in the major element chemistry of the degraded sediment compared to its original fresh bedrock composition (France-Lanord & Derry Reference France-Lanord and Derry1997; Clift & Jonell Reference Clift and Jonell2021). However, an accurate estimate of both sediment and source rock chemistry can be difficult to ascertain. In this study, we examine potential uncertainties in the composition of the weathering products (i.e. the sediment) and in particular look at the Indus submarine fan of the western Indian Ocean (Fig. 1A), which is the primary repository for sediments eroded from the Western Himalaya and Karakoram (Naini & Kolla Reference Naini and Kolla1982; Droz & Bellaiche Reference Droz, Bellaiche, Weimer and Link1991; Clift et al. Reference Clift, Shimizu, Layne, Gaedicke, Schlüter, Clark and Amjad2001).
The Indus Fan is the second largest sediment body in the world, falling only slightly behind its Eastern neighbour, the Bengal Fan, located in the Bay of Bengal. The Indus Fan’s impressive size (4.5 × 106 km3) is owed to the uplift and erosion of the Himalayas, which has for many years been invoked as one of the primary controls over global weathering fluxes and in turn on atmosphere CO2 concentrations (Raymo & Ruddiman Reference Raymo and Ruddiman1992). However, a recent compilation by Clift et al. (Reference Clift, Jonell, Du and Bornholdt2024b) argues differently. Using published geochemical data (Zhou et al. Reference Zhou, Ireland, Murray and Clift2021) in the aftermath of a campaign of scientific drilling conducted by the International Ocean Discovery Program (IODP) in the Asian marginal seas, the team was able to estimate how much CO2 was consumed during the Neogene by the major drainage systems of south and southeast Asia. That work argued that the Himalayas have not dominated the global carbon budget, except potentially since 2−3 Ma, when an increasing weathering flux contributed to the intensification of cooling linked to northern hemisphere glaciation (NHG)(Clift et al. Reference Clift, Jonell, Du and Bornholdt2024b). If this assessment is correct, then other parts of the world may be contributing to an enhanced weathering flux (Bayon et al. Reference Bayon, Patriat, Godderis, Trinquier, De Deckker, Kulhanek, Holbourn and Rosenthal2023; Martin et al. Reference Martin, Macdonald, McQuarrie, Flowers and Maffre2023; Clift et al. Reference Clift, Du, Mohtadi, Pahnke, Sutorius and Böning2024a), but in all cases, any quantitative estimate critically depends both on accurate determination of the sediment volumes and also their chemical composition.
Several important uncertainties still exist that inhibit the construction of accurate regional weathering and carbon budgets. A recent study showed that the estimated efficiency of CO2 consumption in sediments from the Bengal Fan was critically dependent on them being rigorously decarbonated (Tachambalath et al. Reference Tachambalath, France-Lanord, Galy, Rigaudier and Charreau2023). Because calcium (together with Na, K and Mg) is an important element used in calculating the amount of CO2 consumed per unit volume of sediment (France-Lanord & Derry Reference France-Lanord and Derry1997), it is essential that sediment analysed for this purpose has all of its biogenic carbonate removed prior to analysis. More recent studies often omit the decarbonization process or fail to fully decarbonate their samples, making it difficult to integrate older data into regional syntheses.
Chemical analyses of Indus Fan sediments published by Zhou et al. (Reference Zhou, Ireland, Murray and Clift2021) demonstrate some of the potential complexities. Figure 2 shows the temporal trend in the Chemical Index of Alteration (CIA) (Nesbitt et al. Reference Nesbitt, Markovics and Price1980) at IODP Sites U1456 and U1457 reconstructed by Zhou et al. (Reference Zhou, Ireland, Murray and Clift2021). CIA is a common proxy for the degree of alteration in sediments (Nesbitt & Young Reference Nesbitt and Young1982). These data were used to argue for a decrease in the intensity of chemical weathering starting ∼8 Ma and that continued to the present day, correlating with decreasing global temperatures at the same time (Westerhold et al. Reference Westerhold, Marwan, Drury, Liebrand, Agnini, Anagnostou, Barnet, Bohaty, De Vleeschouwer, Florindo, Frederichs, Hodell, Holbourn, Kroon, Lauretano, Littler, Lourens, Lyle, Pälike, Röhl, Tian, Wilkens, Wilson and Zachos2020). However, the reconstructed trend is noisy, and even the five-point running average employed by the study shows a rather erratic decrease in the value of CIA through time (Fig. 2). This trend contrasts with the regular and smooth decrease in CIA seen in the hemipelagic sediment at Ocean Drilling Program (ODP) Site 1146 in the South China Sea (Clift & Jonell Reference Clift and Jonell2021). Is there something intrinsically noisy about environmental records stored in submarine fan complexes, potentially driven by recycling and the wide range of grain sizes, or could this noisy record be a reflection of sampling and sample processing?
Two potential complexities may be causing the poor quality of this submarine fan record. One is whether the decarbonation process was done sufficiently well, while the other questions the potential role of grain size. In general, fine-grained sediments are more altered than coarser material because small grains have a higher surface area to volume ratio, which allows the silicate material to be broken down more rapidly (He et al. Reference He, Zheng, Clift, Tada, Wu and Luo2015). A lack of sand in the dominantly silt-clay sediments from ODP Site 1146 in the South China Sea may be one of the reasons that this reconstruction is much less noisy than that found in the Arabian Sea. In contrast, the Zhou et al. (Reference Zhou, Ireland, Murray and Clift2021) study was done with samples taken at regular depth intervals within the Indus Fan, but without any systematic selection for certain size fractions. This is even though the vast majority of the volume of large submarine fans comprises silt-sized material (Stow et al. Reference Stow, Howell, Nelson, Bouma, Normark and Barnes1985; Bouma Reference Bouma2001). Sand is also present in the fan but is more common in channel and proximal levee complexes. These are volumetrically small compared to the overall stratigraphic architecture, despite being more common in proximal settings (McHargue & Webb Reference McHargue and Webb1986; Kolla & Coumes Reference Kolla, Coumes, Weimer and Link1991).
Consequently, in this study, we preferentially resampled muddy sediments from IODP Sites U1456 and U1457 (Pandey et al. Reference Pandey, Clift, Kulhanek, Andò, Bendle, Bratenkov, Griffith, Gurumurthy, Hahn, Iwai, Khim, Kumar, Kumar, Liddy, Lu, Lyle, Mishra, Radhakrishna, Routledge, Saraswat, Saxena, Scardia, Sharma, Singh, Steinke, Suzuki, Tauxe, Tiwari, Xu and Yu2016a; Pandey et al. Reference Pandey, Clift, Kulhanek, Andò, Bendle, Bratenkov, Griffith, Gurumurthy, Hahn, Iwai, Khim, Kumar, Kumar, Liddy, Lu, Lyle, Mishra, Radhakrishna, Routledge, Saraswat, Saxena, Scardia, Sharma, Singh, Steinke, Suzuki, Tauxe, Tiwari, Xu and Yu2016b) in order to test whether grain size or carbonate removal may have affected the reconstruction of chemical weathering and in turn influenced estimates of the amount of atmosphere CO2 removed during the erosion and weathering of the Western Himalaya and Karakoram.
2. Geologic setting
IODP Sites U1456 and U1457 recovered sediments from the Laxmi Basin in the eastern part of the Arabian Sea (Pandey et al. Reference Pandey, Clift, Kulhanek, Andò, Bendle, Bratenkov, Griffith, Gurumurthy, Hahn, Iwai, Khim, Kumar, Kumar, Liddy, Lu, Lyle, Mishra, Radhakrishna, Routledge, Saraswat, Saxena, Scardia, Sharma, Singh, Steinke, Suzuki, Tauxe, Tiwari, Xu and Yu2016a; Pandey et al. Reference Pandey, Clift, Kulhanek, Andò, Bendle, Bratenkov, Griffith, Gurumurthy, Hahn, Iwai, Khim, Kumar, Kumar, Liddy, Lu, Lyle, Mishra, Radhakrishna, Routledge, Saraswat, Saxena, Scardia, Sharma, Singh, Steinke, Suzuki, Tauxe, Tiwari, Xu and Yu2016b). This basin is located immediately adjacent to the passive margin of Western India, offshore the Western Ghats (Fig. 1). The Laxmi Basin is separated from the main Arabian Basin by the Laxmi Ridge (Fig. 1), interpreted as a fragment of continental crust that rifted away from the western Indian margin prior to a jump in the locus of extension and the onset of sustained seafloor spreading to the west of Laxmi Ridge (Talwani & Reif Reference Talwani and Reif1998; Pandey et al. Reference Pandey, Pandey, Clift, Nair, Ramesh, Kulhanek and Yadav2020). In general, the sediment in the Laxmi Basin has been interpreted to be mostly derived by supply from the Indus River, located to the north (Clift et al. Reference Clift, Zhou, Stockli and Blusztajn2019; Andò et al. Reference Andò, Aharonovich, Hahn, George, Clift and Garzanti2020; Zhou et al. Reference Zhou, Stockli, Ireland, Murray and Clift2022), although there is evidence for some sediment supply from the east, from the passive margin of the Indian peninsula (Yu et al. Reference Yu, Colin, Wan, Saraswat, Song, Xu, Clift, Lu, Lyle, Kulhanek, Hahn, Tiwari, Mishra, Miska and Kumar2019; Cai et al. Reference Cai, Xu, Clift, Khim, Lim, Yu, Kulhanek and Li2020; Carter et al. Reference Carter, Griffith, Clift, Scher and Dellapenna2020; Garzanti et al. Reference Garzanti, Andò and Vezzoli2020). The drilled fan sequence is a continuous succession of hemipelagic muds and graded silty-sand turbidites and was truncated by a major mass wasting complex (MWC) emplaced just prior to 10.8 Ma (Dailey et al. Reference Dailey, Clift, Kulhanek, Blusztajn, Routledge, Calvès, O’Sullivan, Jonell, Pandey, Andò, Coletti, Zhou, Li, Neubeck, Bendle, Bratenkov, Griffith, Gurumurthy, Hahn, Iwai, Khim, Kumar, Kumar, Liddy, Lu, Lyle, Mishra, Radhakrishna, Saraswat, Saxena, Scardia, Sharma, Singh, Steinke, Suzuki, Tauxe, Tiwari, Xu and Yu2019). The fan section pre-dating the MWC has been removed by erosion at Site U1457, while drilling recovered only a short sequence under the MWC at Site U1456 dated at ∼15 Ma (Routledge et al. Reference Routledge, Kulhanek, Tauxe, Scardia, Singh, Steinke, Griffith and Saraswat2020).
The sediments analysed here were taken from sequences dominated by mudstones, but with interbedded silt and sand layers that were preferentially avoided (Fig. 3). Most of the sediments are hemipelagic, but the coarser-grained material is often graded and is interpreted to have been deposited from turbidity currents either on lobes or within channels of the submarine fan (Andò et al. Reference Andò, Aharonovich, Hahn, George, Clift and Garzanti2020). Sediment from Site U1456 is sandier than that at Site U1457, especially in Upper Pliocene and Lower Pleistocene sediments (Fig. 3). There are significant sections where the sediment is uniformly muddy in character, but there are also carbonate or carbonate-rich mudstone layers, testifying to episodes of pelagic sedimentation and abandonment of the fan, at least locally. The biostratigraphic-magnetostratigraphic framework indicates that there are significant time gaps (disconformities) within the sequence above that are associated with the MWC (Routledge et al. Reference Routledge, Kulhanek, Tauxe, Scardia, Singh, Steinke, Griffith and Saraswat2020). These disconformities are interpreted to reflect the avulsion of the primary depositional lobe both within Laxmi Basin and away from Laxmi Basin and into the main part of the Arabian Basin, west of Laxmi Ridge.
Zircon U-Pb and apatite fission track dating of sandy materials also indicate that most of the coarse-grained sediment has been supplied from the Indus River (Zhou et al. Reference Zhou, Carter, Li and Clift2020; Zhou et al. Reference Zhou, Stockli, Ireland, Murray and Clift2022), but the origin of the fine-grained material has been disputed. Upper Miocene sediments from Site U1457 yielded pollen and biomarkers that were indicative of some of the sediment coming directly from peninsular India (Feakins et al. Reference Feakins, Liddy, Tauxe, Galy, Feng, Tierney, Miao and Warny2020). Likewise, in the younger part of the drilled section, studies have argued for alternating supplies from the peninsula and from the delta (Cai et al. Reference Cai, Xu, Clift, Lim, Khim, Yu, Kulhanek, Li, Chen and Sun2019; Clift et al. Reference Clift, Zhou, Stockli and Blusztajn2019; Carter et al. Reference Carter, Griffith, Clift, Scher and Dellapenna2020; Garzanti et al. Reference Garzanti, Andò and Vezzoli2020). Resolving these contrasting supplies is not entirely straightforward because many of the proxies used offer non-unique solutions.
During the 10.8 Ma of sedimentation recorded at the drill sites since the MWC, the source of the sediments has seen a progressive evolution towards more Himalayan supply as the focus of erosion within the Indus Basin and shifted away from sediment derived from the Karakoram (Clift et al. Reference Clift, Zhou, Stockli and Blusztajn2019; Zhou et al. Reference Zhou, Stockli, Ireland, Murray and Clift2022). In particular, the uplift of the Lesser Himalaya impacted the supply of sediment to the Arabian Sea by increasing the amount coming from the radiogenic ancient Indian continental crust (Najman et al. Reference Najman, Bickle, Garzanti, Pringle, Barfod, Brozovic, Burbank and Ando2009; Exnicios et al. Reference Exnicios, Carter, Najman and Clift2022) versus the more primitive rocks of the Karakoram. The latter were more dominant before ∼8 Ma. Zhou et al. (Reference Zhou, Stockli, Ireland, Murray and Clift2022) used detrital zircon U-Pb age data combined with numerical unmixing methods (Sundell & Saylor Reference Sundell and Saylor2017) to estimate that around 70% of the sand reaching the drill sites before 8 Ma was from the Karakoram but that this had fallen to ∼35% by 5.7 Ma. Associated bulk sediment Nd isotope analysis was consistent with this estimate and supported a progressive increase in flux from the Lesser Himalaya from 5.7 Ma until the present day (Clift et al. Reference Clift, Zhou, Stockli and Blusztajn2019).
The change in the provenance of the Indus River is important for chemical weathering records because the sediments eroded from the Himalayas have bedrock compositions that are more consistent with average upper continental crust (UCC) (Hodges Reference Hodges2000), while those derived from the Karakoram or Kohistan have more primitive, mafic compositions (Searle et al. Reference Searle, Rex, Tirrul, Rex, Barnicoat and Windley1989; Khan et al. Reference Khan, Stern, Gribble and Windley1997; Jagoutz et al. Reference Jagoutz, Müntener, Ulmer, Pettke, Burg, Dawood and Hussain2007). Because the amount of CO2 consumed during chemical weathering is controlled by the difference between the sediment composition and the original bedrock, alteration of reactive Ca–Mg-bearing mafic rocks, such as those exposed in the Karakoram-Kohistan, typically results in higher degrees of atmospheric CO2 consumption (Caves Rugenstein et al. Reference Caves Rugenstein, Ibarra and von Blanckenburg2019). It is this compositional difference between the source rocks of the sediments of the Arabian Sea that makes them more effective in sequestering CO2 from the atmosphere than the sediments of the Bengal Fan (Clift & Jonell Reference Clift and Jonell2021; Clift et al. Reference Clift, Jonell, Du and Bornholdt2024b).
3. Sampling
Fifty-two samples were selected from both Sites U1456 and U1457 to assess whether earlier work was biased by grain-size issues and/or incomplete decarbonation. Thirty-one samples were taken from Site U1456 and 21 from Site U1457. Site U1456 was favoured because of its location in the central part of the Laxmi Basin where it is more likely to be the recipient of sediment transported from the Indus Delta to the north, whereas Site U1457 is located on the flanks of the Laxmi Ridge and could be less strongly Indus River influenced. In this work, we specifically targeted fine-grained lithologies, typically described as various types of mudstones (silt-clay mixtures) by the shipboard core descriptions (Pandey et al. Reference Pandey, Clift, Kulhanek, Andò, Bendle, Bratenkov, Griffith, Gurumurthy, Hahn, Iwai, Khim, Kumar, Kumar, Liddy, Lu, Lyle, Mishra, Radhakrishna, Routledge, Saraswat, Saxena, Scardia, Sharma, Singh, Steinke, Suzuki, Tauxe, Tiwari, Xu and Yu2016a; Pandey et al. Reference Pandey, Clift, Kulhanek, Andò, Bendle, Bratenkov, Griffith, Gurumurthy, Hahn, Iwai, Khim, Kumar, Kumar, Liddy, Lu, Lyle, Mishra, Radhakrishna, Routledge, Saraswat, Saxena, Scardia, Sharma, Singh, Steinke, Suzuki, Tauxe, Tiwari, Xu and Yu2016b). We attempted to achieve the best temporal control as possible by employing the revised age models from Routledge et al. (Reference Routledge, Kulhanek, Tauxe, Scardia, Singh, Steinke, Griffith and Saraswat2020). All samples were analysed for grain size and decarbonated sediment chemistry.
4. Methodology
4.a. Grain-size analysis
A total of 52 samples from Sites U1456 and U1457 had their grain size determined following the procedure of Hülse & Bentley (Reference Hülse and Bentley2012). About 2 g of sample were placed in a 50 ml plastic centrifuge tube with 5.75 ml of NaPO3 solution. Each tube was capped and vortexed to deflocculate clay-sized sediment and separate organic particles. The sample was poured through an 850 µm sieve to help remove large particles of organic matter. The polypropylene tube was placed into a centrifuge to settle the sediment at the bottom of the vial. The vials were then run at ∼25 rotations per minute for 60 minutes, and the supernatant was then removed. About 2 ml of NaPO3 was then added to the centrifuged sample, as well as 5 ml of H2O2. Small amounts of acetone were sprayed in the vials to stabilize the reaction, following which the samples were placed in a hot bath set at 70°C overnight. Once removed from the hot bath, the supernatant was removed from the vials, and the samples were transferred into plastic tubes with caps to be vortexed for analysis. The samples were then analysed using a Beckman-Coulter (LS 13-320SW Laser Diffraction Particle Size Analyzer Single Wavelength) at Louisiana State University (LSU). Results, including full grain-size spectra up to 282 µm, mean, median, kurtosis and skewness, are shown in Table S1.
4.b. Major and trace element analysis
Sediments were decarbonated with 20% acetic acid for at least two days until no further bubbles were noted, then washed with distilled and deionized water with a purity of 9–12 megaohms and hand powdered. Decarbonated powder samples for major element geochemistry were processed and analysed in the Chevron Geomaterials Characterization Laboratory (CGCL) in the Department of Geology and Geophysics at LSU. The samples were ground in a mortar and pestle and then transferred into a hardened steel vial to be milled to a grain size of <30 μm using a SPEX® SamplePrep 8000M mixer/mill. Approximately 4.00 ± 0.04g of powder from each sample and 1.00 ± 0.01g of SPEX® SamplePrep UltraBind® powder was weighed and mixed using the 8000M mixer/mill. The mixture was then pressed into a pellet using a bench top 20 tonne hydraulic pellet press and a 31 mm pellet die.
Analysis was carried out using the Bruker S2-PUMA energy-dispersive X-ray fluorescence (XRF) instrument in the CGCL. All analyses were performed under a vacuum using a Pd X-ray source. An accelerating voltage and current of 20 kV and 0.25 μA with no primary filter were used for Na2O, MgO, Al2O3, SiO2 and P2O5, while 40 kV and 0.5 μA and a 0.5 mm Al filter was used for K2O, CaO, TiO2, MnO and Fe2O3, as well as the remaining trace elements. The analyses were calibrated using 19 international standard reference materials, including AGV-2, BIR-1a, G-2, GSP-2, RGM-1, SDC-1 and W-2 from the United States Geological Survey (USGS); JA-3, JB-2, JB-3, JG-2, JG-3, JR-1 and JP-1 from the Geophysical Survey of Japan (GSJ); and BCS-309, BCS-313-2, BCS-375, BCS-376/1 and BCS-388 from British Chemical Standard. Replicate analysis (N = 10) of the USGS SGR-1b and GSJ JB-3, JA-3, JSd-1 and JLk-1 standards was performed to check for drift and assess the quality of the calibration prior to analysis. Secondary standards (e.g. JA-3, SCo-1) were analysed during every analytical run to monitor for instrumental drift and assess the continued quality of the calibration. Concentrations of Na2O, MgO, Al2O3, SiO2, P2O5, K2O, CaO, TiO2, MnO and Fe2O3, as well as the loss on ignition, are shown in Table S2. CIA (Nesbitt et al. Reference Nesbitt, Markovics and Price1980) is calculated using the following molar equation.
5. Results
Grain-size analysis reveals that the sediments selected were largely successful in sampling only fine-grained material. Figure 4 shows the triangular diagram from Folk (Reference Folk1974). Most of the sediments considered here are defined as mud or silt under this scheme, although two samples from Site U1456 were defined as sandy silt. This represents true siliciclastic sandy silt and is not the product of carbonate shells being present. It is noteworthy that the sediments from Site U1456 are mostly finer grained than those from Site U1457, which tend to plot more in the silt field. When compared with the sediments analysed in the earlier Zhou et al. (Reference Zhou, Ireland, Murray and Clift2021) study, we note that many of their analyses were made on sandy-silts or silty-sand material, consistently coarser grained than the material examined here.
Decarbonated sediment major element chemistry reflects not only the source bedrock composition but also the influence of sediment transport on the material being analysed. In the discrimination diagram of Singh et al. (Reference Singh, Sharma and Tobschall2005), our samples plot towards the lower left of the diagram (Fig. 5A), away from the UCC average (Rudnick & Gao Reference Rudnick, Gao and Rudnick2003) and implying preferential enrichment in quartz, with associated relative loss of material such as biotite, muscovite and clay. The fact that many of these sediments are quite fine grained makes this a little surprising, although it emphasizes the importance of quartz-dominated silt in the make-up of these materials. Enrichment in quartz reflects the breakdown of chemically unstable phases, such as the micas and the concentration of quartz by hydrodynamics sorting. We note that the sediments in this study show this effect even more than the sediments in the Zhou et al. (Reference Zhou, Ireland, Murray and Clift2021) work, although there is significant overlap.
The concentrations of some water-immobile elements are closely tied to certain minerals and thus to provenance. These potentially act as tracers that might quantify the competing influence of the Indus River and peninsular India. The latter is dominated by the Deccan Traps in the immediate onshore region, as drained by the Narmada, Tapti and Mahi rivers (Fig. 1). Zr concentration is largely a reflection of zircon abundance. This mineral is most common in felsic igneous rocks and in siliciclastic sedimentary rocks but is less common in mafic rocks. Zr concentrations might be expected to rise with greater erosion from older continental (i.e. Himalayan or Indian cratonic) rocks. There is a weakly defined trend to higher Zr concentrations through time at both sites, rising from ∼130 ppm at 10 Ma to ∼160 ppm in recent times (Fig. 6B). This trend is not linked to grain size as mean grain size and Zr concentrations show an R2 = 0.16 correlation. Ti is concentrated in rutile and ilmenite grains, which are found in high-grade metamorphic rocks and intrusive igneous rocks (Zack & Kooijman Reference Zack and Kooijman2017). Ti abundances in fine-grained sediments are also strongly controlled by the presence of clays, but there is no correlation between TiO2 and mean grain size in this dataset (R2 = 0.001). Ilmenite is especially associated with cumulate igneous intrusions (Anthony et al. Reference Anthony, Bideaux, Bladh and Nichols2022), although it is not exclusive to felsic or mafic rocks. The concentrations of Ti do not change much over time, although sediments from Site U1456 have lower average contents (0.97%) compared to those at Site U1457 (1.24%) (Fig. 6A).
Cr is primarily found in chromite, which in turn is associated with mafic and ultramafic intrusions (Dawson & Smith Reference Dawson, Smith, Ahrens, Dawson, Duncan and Erlank1975; Lorand & Ceuleneer Reference Lorand and Ceuleneer1989), so its abundance might be associated with erosion of ophiolite and/or arc lower crustal rocks, such as found in the Karakoram, Kohistan or in the ophiolites in the Indus Suture Zone. However, Cr contents in the Laxmi Basin sediments show no significant temporal evolution (Fig. 6C). None of these stable element provenance proxies correlate very much with the Nd isotope record from the sites. Nd is generally considered to be a good provenance proxy linked to the age of the source rocks and is not affected by chemical weathering (Goldstein et al. Reference Goldstein, O’Nions and Hamilton1984). Earlier Nd isotope work featured a cluster of high εNd value samples, which were associated with run-off from the Indian peninsula (Fig. 6D), as well as a trend to decreasing values linked sediment supply from the Indus River and reflecting greater degrees of Himalayan erosion through time (Clift et al. Reference Clift, Zhou, Stockli and Blusztajn2019).
6. Discussion
The state of chemical weathering of the sediments considered here can be assessed in the (CaO* + Na2O), K2O and Al2O3 (CN-K-A) triangular diagram of Fedo et al. (Reference Fedo, Nesbitt and Young1995), where CaO* represents the CaO associated with silicate (Fig. 5B). In this plot, freshly eroded continental material typically lies at the bottom of the triangle, while progressive chemical weathering displaces the composition towards the top as water-mobile components are removed and Al2O3 is correspondingly relatively enriched. The sediments considered in this work form part of an array stretching from the bottom left-hand corner towards the mid-right-hand side of the triangle, towards an illite and mica end member. This is caused by the breakdown of the original rock fragments into clay minerals.
The new data are largely consistent with those in the earlier study (Zhou et al. Reference Zhou, Ireland, Murray and Clift2021), although they tend to plot a little below the previously analysed material, implying less chemical weathering than previously measured. This is despite the material being mostly finer grained than that considered before. The reduced range of compositions on this plot reflects the more limited grain size considered in this study. The same plot allows the degree of weathering to be assessed according to the CIA proxy. Samples from Site U1456 have slightly lower CIA values, a minimum of ∼59, while Site U1457 samples have higher values, up to a maximum of 73.
6.a. Grain size and alteration
We investigate how the grain size of the sediment may be linked to its degree of chemical alteration. Considering the full range of grain sizes analysed by Zhou et al. (Reference Zhou, Ireland, Murray and Clift2021), it is possible to see that there is a rough correlation whereby sediments with the largest grain size (lowest φ values) generally have the lowest CIA values (Fig. 7A), whereas the finest grained sediments (highest φ values) have the highest CIA values. This is consistent with the idea of the fine-grained material being more altered than the coarse grained. However, if we only consider the dataset generated in this study, then we see that no clear relationship is apparent (Fig. 7B). The highest CIA values are not present in our dataset. Overall, the material from Site U1457 is coarser grained than that found at Site U1456, but the Site U1457 sediment has higher average CIA values (70.6 vs 68.0), the opposite of what was observed in the previous study. Although this may seem counter-intuitive, this may point to the material at Site U1457 being preferentially derived from a more weathered area than that sourcing Site U1456. Furthermore, within each site, it is possible that there is insufficient grain size variability to make much difference to the alteration state.
6.b. Evolution in weathering
We now investigate whether the new dataset provides a more coherent long-term evolution in chemical weathering trends. The long-term trend in CIA (Fig. 8A) is much better defined than observed in the previous study (Fig. 2) (Zhou et al. Reference Zhou, Ireland, Murray and Clift2021). During the Pleistocene (<2.58 Ma), our new data indicate CIA averaging 61.8 with a standard deviation (SD) of 2.5; in contrast, the Zhou et al. (Reference Zhou, Ireland, Murray and Clift2021) study estimated CIA averaging 67.6 and with an SD of 5.4. During the Pliocene (2.58–5.3 Ma), CIA values are 69.0 with an SD of 2.1, while Zhou et al. (Reference Zhou, Ireland, Murray and Clift2021) estimate the average CIA at 70.9 with SD of 6.9. Late Miocene (5.3–11.6 Ma) CIA values are 71.5 with an SD of 3.4, while Zhou et al. (Reference Zhou, Ireland, Murray and Clift2021) estimate CIA averaging 72.8 with an SD of 5.3.
At 10 Ma, CIA was ∼75, but this falls to values ∼65 with a lot of variability by the present day.
It is possible that the strong recent variability may be linked to high amplitude climatic cycles following the onset of NHG. Rates of weathering are dependent on temperature and humidity (West et al. Reference West, Galy and Bickle2005). We also note that the highest CIA values in the last 4 Ma are almost exclusively found at Site U1457.
If we consider the alternative proxy CIX (Garzanti et al. Reference Garzanti, Vermeesch, Padoan, Resentini, Vezzoli and Andò2014), which is calculated in the same way as CIA except excluding Ca, then there is a more clear-cut difference between the two sites. Material from Site U1456 shows a trend from values ∼77 at 10 Ma, falling to ∼68 in recent times. In contrast, the material from Site U1457 maintains a relatively constant CIX value after 10 Ma (Fig. 8B). In both cases, it is possible to compare these values with recent measurements from the Holocene Indus River and its continental shelf (Limmer et al. Reference Limmer, Boening, Giosan, Ponton, Köhler, Cooper, Tabrez and Clift2012), as well as rivers draining the Indian peninsula (Kurian et al. Reference Kurian, Nath, Kumar and Nair2013). As might be expected, alteration of sediment in the Indus River drainage is weaker than that seen in the Indian peninsula, reflecting faster erosion driven by active rock uplift and thus little time for weathering in the Himalayan source region. In general, there is a tendency of the Site U1457 material to trend towards peninsular values, while the data from Site U1456 mostly coincide with recent values from the Indus River.
We also examine the degree of chemical alteration using two widely applied proxies involving the ratio of water-mobile alkali elements versus immobile Al. Figure 8C shows that values in K/Al at Site U1456 mostly increase from 10 Ma, when values were ∼0.27, and reached ∼0.3 in recent geological time. In contrast, a second trend can be seen, dominated by material from Site U1457, which started at the same value at 10 Ma but instead slightly decreased to ∼0.2 at the present. A very similar arrangement can be seen with Na/Al (Fig. 8D), which on average increases to the present day at Site U1456, indicative of less chemical weathering through time, as predicted by Zhou et al. (Reference Zhou, Ireland, Murray and Clift2021), but again with the most weathered material being preferentially found at Site U1457.
In order to assess whether there might be a relationship between the extent of chemical weathering and the source of sediments, we plot the CIA against the Nd isotope character (εNd). Figure 9A shows that there is a positive correlation between Nd isotopic composition, which is entirely determined by sediment provenance (Goldstein et al. Reference Goldstein, O’Nions and Hamilton1984), and the extent of chemical weathering at both drilling sites, although the correlation coefficient is low (R2 = 0.15 at Site U1456 and 0.11 at Site U1457). This implies that more weathered sediment is preferentially derived from younger more primitive bedrock terrains, although these could be represented by either the Deccan Plateau flood basalts or the magmatic arc units in the Indus Suture Zone, including Kohistan. Because erosion is rapid and the sediment in the Indus River is relatively chemically fresh, as demonstrated by the overall low CIA, it seems more likely that strongly weathered, primitive material would be coming from the Deccan Plateau where chemical weathering has been impacting the landscape since the end of the eruption in the Paleocene (Widdowson & Cox Reference Widdowson and Cox1996; Dessert et al. Reference Dessert, Dupré, Francois, Schott, Gaillardet, Chakrapani and Bajpai2001).
Considering only the new data from this study (Fig. 9B), the same positive correlation is noted, but the correlation coefficient is substantially better, especially at Site U1456 (R2 = 0.56 compared to 0.27). This result provides further support for the importance of using fine-grained samples for CIA-based weathering reconstructions. Nonetheless, this plot has to be interpreted with some caution because the Nd isotope data is from an earlier work (Clift et al. Reference Clift, Zhou, Stockli and Blusztajn2019) and the two datasets are not from the same samples. To be able to make any comparison, we must assume that analyses made on samples taken from close to the earlier Nd isotope measurements (<10 m) would have similar isotope character. This is reasonable because provenance changes slowly through time as the source ranges uplift. The improved correlation mirrors the temporal trend to lower CIA, which is particularly pronounced at Site U1456 compared to Site U1457. As well as reflecting varying input from the Deccan Plateau, some of the trend may be related to progressively faster erosion of old continental rocks uplifted in the Himalayas, which would reduce εNd values while also decreasing the CIA.
7. Different sediment sources to different drilling sites
The new data presented here, together with a reassessment of the previously published data, now allows us to better explain why the chemical trend highlighted by Zhou et al. (Reference Zhou, Ireland, Murray and Clift2021) was so noisy. Part of this scatter is related to grain size. A larger range of grain sizes was used in the earlier record but results in the greater spread in values of various chemical weathering proxies such as CIA (Fig. 2). Because the sediment is not consistently fining or coarsening up-section, this means that there is variability introduced into the earlier record because of randomly analysing coarse or fine-grained material. This reinforces the need to use a uniform grain-size range when attempting to reconstruct palaeo-weathering conditions.
In the reconstruction presented above, it is also clear that sediment is coming into the Laxmi Basin from more than one source. One source is the Indus River, and the other is Peninsular India, largely dominated by the Deccan Plateau. Mixing calculations based on an average εNd value of −2.4 for the Deccan Plateau and the evolving average for the Indus River (Clift et al. Reference Clift, Zhou, Stockli and Blusztajn2019) indicate that ∼20–30% of the Upper Miocene at Site U1457 is from the Deccan Plateau, rising to 24–37% in the Pliocene and 31–35% in the Pleistocene. This is the other important factor causing a noisy record in the earlier study. Because the source compositions and weathering conditions are quite different in these two areas, mixing sediments derived from different regions results in a more random pattern than might be otherwise expected. In this work where we focus on muddy sediment, there is a relatively gradual and more orderly evolution in the intensity of chemical alteration of the sediment especially at Site U1456 and after 8 Ma (Fig. 8). As a result, two types of sedimentary systems can be recognized. One is finer grained and is derived from the Indus River. It shows decreasing degrees of chemical alteration through time and is preferentially deposited in the centre of the basin at Site U1456. The second system is slightly coarser grained and is derived from peninsular India. It is generally more strongly weathered and does not evolve much through time. Sediment from this source is deposited at both sites but is preferentially preserved at Site U1457. This is slightly surprising because Site U1457 is on the western side of Laxmi Basin, further from the Indian coast. Although aeolian dust is blown east from Arabia into the basin, the difference in transport distance between Sites U1456 and U1457 is not great, and earlier analysis of the isotopes has argued that influence from Arabian is insignificant (Clift et al. Reference Clift, Zhou, Stockli and Blusztajn2019). We infer that sediment coming from the north, from the Indus River mouth, is funnelled into the deepest axial part of the basin. Sediment coming from peninsular India must cross the basin in order to be deposited at Site U1457. We note that material from the peninsula is not exclusively found at Site U1457. Our reassessment of the provenance is consistent with the palynology data from Feakins et al. (Reference Feakins, Liddy, Tauxe, Galy, Feng, Tierney, Miao and Warny2020), which focused on the older part of the section but also recognized the alternation of sediment supply from the north and from the east at Site U1457, albeit only in the Miocene.
8. Calculating CO2 consumption
We use the new geochemical data to reassess the impact of chemical weathering in the Indus Basin on the consumption of carbon dioxide by the breakdown of silicate minerals. The Indus Basin represents a significant component of the Himalayan weathering budget that was proposed by Raymo and Ruddiman (Reference Raymo and Ruddiman1992) to have driven Cenozoic global cooling. In a recent synthesis, Clift et al. (Reference Clift, Jonell, Du and Bornholdt2024b) suggested that the Indus Basin dominated the Asian carbon budget prior to the Late Miocene. The new dataset is focused on a limited grain-size variation more typical of the bulk volume of the submarine fan and has been more carefully decarbonated before analysis. By estimating the change in composition of sediment relative to fresh source bedrock, the amount of CO2 consumed per unit weight can be estimated (France-Lanord & Derry Reference France-Lanord and Derry1997).
Changes in the molar ratios of Mg/Al, Ca/Al, Na/Al and K/Al are used to derive the ΔCO2 values for each sample, that is, how much CO2 is consumed per unit weight of sediment. We use the Nd isotopic data as interpreted by Clift et al. (Reference Clift, Zhou, Stockli and Blusztajn2019) to determine if sediments were Indus River or peninsula-derived; however, because the isotope data come from samples analysed in a previous study rather than this one, we had to assume that sources did not change rapidly with time in order to assign our new data to an appropriate source, as explained above. After eliminating sediment eroded from the peninsula, we were able to estimate the composition of the evolving Indus River’s bedrock sources following the existing model. This involves using the εNd value to estimate the contributions from primitive sources within the Indus Suture Zone including the Karakoram (average εNd value of -5 (Clift & Jonell Reference Clift and Jonell2021)), compared to older continental rocks in the Himalayas (average εNd value of −15.7), and then using average crustal compositions from these areas to calculate an average source major element composition at any one particular time. The Himalayas are assumed to have an average UCC (Rudnick & Gao Reference Rudnick, Gao and Rudnick2003), while Kohistan-Karakoram values are derived from the Georock database. The differences between the bedrock and the sediment Mg/Al, Ca/Al, Na/Al and K/Al values are then used to calculate the ΔCO2 values using the method of France-Lanord and Derry (Reference France-Lanord and Derry1997).
Some samples show a negative difference in K/Al values, which should be impossible because this implies that the sediment is less depleted than the bedrock source. This discrepancy is caused by reverse weathering, which is a process in which K is fixed into the deposited marine sediment in the form of diagenetic clay minerals, for example, smectite (Mackenzie & Kump Reference Mackenzie and Kump1995; Dunlea et al. Reference Dunlea, Murray, Santiago Ramos and Higgins2017). This process is hard to correct accurately, although this is not a serious source of error because the amount of this negative difference is quite small and because K contributes a relatively moderate amount to the total CO2 consumption. Its contribution is weighted as just 10% compared to the CO2 consumption by Mg and Ca (France-Lanord & Derry Reference France-Lanord and Derry1997).
Figure 10A shows the results of the recalculation of the ΔCO2 based on the new geochemical analyses. Our analysis shows that the ΔCO2 values started at around 0.8–1.0 mol/y at 10 Ma and subsequently decreased to the present day, reaching ΔCO2 values ∼0.6–0.8 mol/y. This means that the amount of CO2 consumed per unit weight of rock weathered decreased through time to the present. There is short-term variability related to changes in the sediment geochemistry, so a more representative long-term estimate may be derived using a five-point running average. The long-term changing chemical reactivity of rocks in the Indus Basin is linked to lower degrees of alteration and the reducing input from the more mafic rocks of the Karakoram, Kohistan and suture zone since ∼8 Ma (Clift et al. Reference Clift, Zhou, Stockli and Blusztajn2019; Zhou et al. Reference Zhou, Stockli, Ireland, Murray and Clift2022). These rocks break down more readily and have a bigger impact of CO2 consumption because of their high Mg contents.
We compare our estimates with the long-term estimates of ΔCO2 values based on the Zhou et al. (Reference Zhou, Ireland, Murray and Clift2021) data and recently published by Clift et al. (Reference Clift, Jonell, Du and Bornholdt2024b). Although the ΔCO2 values were slightly higher according to Zhou et al. (Reference Zhou, Ireland, Murray and Clift2021) at 10 Ma, 1.0–1.2 mol/y (Fig. 10A), our new values were lower and showed a decrease from that time towards the present. In general, ΔCO2 values were 1.0–0.8 mol/y and then dipped below 0.7 mol/y in the latter part of the Pleistocene. This means that the amount of CO2 consumed per unit weight of sediment and thus the total CO2 budget based on the earlier work yielded estimates of CO2 consumption that are ∼30% higher during the Miocene and ∼60–70% in the Pleistocene than those based on data from this work. Given the large size of the Indus submarine fan and the prolific erosional flux from the Himalayas, this would result in a significant overestimate of the carbon sequestering power of chemical weathering in this region. In this case, it is less likely that chemical weathering in the Western Himalaya-Karakoram would be a critical factor in driving global cooling in the Plio-Pleistocene. This result emphasizes the importance of choosing appropriate material and undertaking rigorous carbonate removal prior to major element analysis if representative estimates of CO2 sequestration are to be made over significant periods of geological time.
We infer that grain size is the more important of these factors distinguishing this and earlier work. Although finer grained, the sediments considered here were less altered than the sandier sediments analysed previously (Fig. 5). Sandy sediments are richer in quartz but depleted in minerals rich in Ca, Mg, K and Na that determine the amount of CO2 consumed during weathering. In contrast, incomplete decarbonation would tend to result in elevated Ca contents and thus less CO2 reduction, which is the opposite of what we see in these data. Total CO2 consumption is more strongly controlled by the rates of sediment supply to the ocean rather than the degree of alteration (Clift et al. Reference Clift, Jonell, Du and Bornholdt2024b), and the temporal trends reconstructed for Asia are not greatly altered by this work, but the total amount of CO2 consumed is reduced.
9. Conclusions
Calculating chemical weathering fluxes is critically dependent on accurate estimates of the average composition of sediments deposited in the depocentres of major drainage systems. While scientific drilling of the Indian Ocean submarine fans has provided samples of the largest depositional bodies on Earth, the resultant temporal histories of weathering history have yielded variable records often with poorly defined trends, especially when compared to hemipelagic sites. In this work, we assessed whether grain size or incomplete decarbonation prior to major element analysis might be affecting the results from the Indus submarine fan. We show that when fine-grained muddy sediments are preferentially selected and carefully decarbonated, the trends in chemical weathering intensity are more clearly defined. The data also highlight that sediments from the Laxmi Basin are derived from both the Indus River and Peninsular India. The former are in general less weathered than those eroded from the Craton or Deccan Plateau. Sediments deposited at Site U1456 are finer grained and preferentially more Indus River-derived than those at Site U1457. If these groups are treated separately, then a clear trend to less chemical alteration over time since ∼8 Ma is visible in the Indus River.
The major element composition of the Indus River-derived sediments can be used to recalculate the amount of CO2 consumed per unit weight during chemical degradation. The value was moderately lower than earlier estimates at ∼10 Ma but decreased over time, especially after ∼8 Ma, so that in the recent geologic past, the estimates were ∼60–70% lower than values based on earlier geochemical work. This result does not change the overall trend of CO2 consumption in the Western Himalaya since the Middle Miocene, but it does reduce the total amount of CO2 removed and further weakens arguments that Himalaya erosion and weathering drove Neogene global cooling.
Supplementary material
The supplementary material for this article can be found https://doi.org/10.1017/S0016756824000499.
Acknowledgements
We thank the Charles T. McCord Jr Chair in Petroleum Geology at Louisiana State University (LSU) for the financial support needed to undertake this study. We further thank Dr Matthew Loocke (LSU) for his help in undertaking the major element analysis and Ms. Wanda LeBlanc for her help and advice in sample preparation. The manuscript was improved by comments from two anonymous reviewers.