1. Introduction
Landscape evolution is linked to the interplay of influences over physical erosion and chemical weathering that are in turn governed by tectonics and climate change (Burbank et al. Reference Burbank, Blythe, Putkonen, Pratt-Sitaula, Gabet, Oskins, Barros and Ojha2003; Gabet & Mudd, Reference Gabet and Mudd2009; Reiners et al. Reference Reiners, Ehlers, Mitchell and Montgomery2003; Riebe et al. Reference Riebe, Kirchner and Finkel2004). Changes in these processes result in alteration of sediment compositions that may be preserved in the final depocenter, albeit modulated during transportation along the pathway between source and sink (Allen, Reference Allen2008; Clift & Jonell, Reference Clift and Jonell2021; Kuehl et al. Reference Kuehl, Alexander, Blair, Harris, Marsaglia, Ogston, Orpin, Roering, Bever, Bilderback, Carter, Cerovski-Darriau, Childress, Reide Corbett, Hale, Leithold, Litchfield, Moriarty, Page, Pierce, Upton and Walsh2016). Sediments may be stored and later released from floodplains and dryland regions, including deserts, which may affect the rate of supply and composition of sediment supply to the final depocenter. Depending on the duration of sediment transport, significant chemical weathering may occur in floodplains along the course between mountainous sources and the lower reaches and river mouth (Lupker et al. Reference Lupker, France-Lanord, Galy, Lave, Gaillardet, Gajured, Guilmette, Rahman, Singh and Sinha2012). Reworking of floodplain sediments can play a vital role in controlling river sediment composition, as can recycling from desert regions that are proximal to the river (Alizai et al. Reference Alizai, Carter, Clift, VanLaningham, Williams and Kumar2011a; Clift & Jonell, Reference Clift and Jonell2021). Erosion patterns and rates are also controlled by tectonic processes that uplift source terrains over millions of years, but here we consider specifically the influence of climate-modulated surface processes changing over orbital and shorter timescales (<105 years) (Clift et al. Reference Clift, Giosan, Carter, Garzanti, Galy, Tabrez, Pringle, Campbell, France-Lanord, Blusztajn, Allen, Alizai, Lückge, Danish, Rabbani, Clift, Tada and Zheng2010a; Colin et al. Reference Colin, Siani, Sicre and Liu2010; Fildani et al. Reference Fildani, McKay, Stockli, Clark, Dykstra, Stockli and Hessler2016; Mason et al. Reference Mason, Romans, Stockli, Mapes and Fildani2019).
In Earth’s history, changing climate and sediment supply have been factors behind the formation of deserts associated with drainage basins (Blum & Törnqvist, Reference Blum and Törnqvist2000). Deserts are found preferentially at the mid-latitudes, where descending dry air masses lead to increased aridity (Bostock et al. Reference Bostock, Opdyke, Gagan, Kiss and Fifield2006; Cronin, Reference Cronin1999). Global climate change can increase the extent and location of arid regions (Zeng & Yoon, Reference Zeng and Yoon2009). Deserts also form in the rain shadow of tectonically generated mountain ranges (Galewsky, Reference Galewsky2009; Garzanti et al. Reference Garzanti, Pastore, Stone, Vainer, Vermeesch and Resentini2022) and can potentially accumulate large volumes of detritus that buffer the sediment flux between mountain sources and their final depocenter.
Here, we investigate the associations between the Thar Desert of South Asia, the Indus River and climate change since the Last Glacial Maximum (LGM, ∼20 ka) (Fig. 1). The semi-arid Thar Desert of the northwestern Indian subcontinent is affected by the southwesterly summer monsoon, so named because this is the direction of the dominant winds. Thar Desert sediments are predominantly derived from the western Tibetan-Himalaya orogen, with supply from the mainstream Indus River, as well as from its Himalayan tributaries from the east (i.e., Punjab) (Clift et al. Reference Clift, Campbell, Pringle, Carter, Zhang, Hodges, Khan and Allen2004; Garzanti et al. Reference Garzanti, Vezzoli, Ando, Paparella and Clift2005, Usman et al. Reference Usman, Clift, Pastore, Vezzoli, Andò, Barbarano, Vermeesch and Garzanti2024). The main Indus River carries detritus eroded from the Karakorum, Kohistan and Nanga Parbat ranges to the lower reaches (Fig. 1).
Digital Elevation Model (DEM) of Pakistan and adjacent regions (from https://download.gebco.net/). The map shows sampling locations in the Cholistan (blue triangles) and Sindh (red squares) deserts. In the map, blue curves show major rivers, and dotted curves show palaeorivers in the desert sides, and the map is adapted from Usman et al. (Reference Usman, Clift, Pastore, Vezzoli, Andò, Barbarano, Vermeesch and Garzanti2024).
The Thar Desert contains aeolian dunes built by monsoon and, to a lesser extent, westerly winds over a variety of timescales (Glennie et al. Reference Glennie, Singhvi, Lancaster, Teller, Clift, Kroon, Gaedicke and Craig2002; Kar et al. Reference Kar, Felix, Rajaguru and Singhvi1998; Singhvi & Kar, Reference Singhvi and Kar1992). These winds bring rain that is largely focused on the southern flank of the western Himalaya, resulting in high erosion rates (Bookhagen & Burbank, Reference Bookhagen and Burbank2006). In this study, we present multi-proxy datasets that test this monsoon-dominated sediment supply model. We employ a series of geochemical and isotopic methods that have been proven to be effective provenance proxies within the Indus River system to constrain the source of sediment in the modern Thar Desert. In particular, we investigate if systematic differences occur across the northern and southern parts of the western Thar Desert, respectively found in the Pakistani provinces of Cholistan and Sindh, as well as in Indian Rajasthan’s Eastern Thar Desert (Bhattacharyya et al. Reference Bhattacharyya, Singh, Qasim and Chandrashekhar2024) (Fig. 1). Besides utilizing the unique chemical and isotopic signatures of the primary sediment sources as reference, we also consider the role played by other parameters that may influence the composition of dune sand, including grain size and weathering. Contrasting degrees of weathering can be used to constrain the sediment sources, although additional weathering may also occur in floodplains, and it should be remembered that this may change as the monsoon strengthened and declined through time (Clift et al. Reference Clift, Giosan, Blusztajn, Campbell, Allen, Pringle, Tabrez, Danish, Rabbani, Carter and Lückge2008; Clift et al. Reference Clift, Giosan, Carter, Garzanti, Galy, Tabrez, Pringle, Campbell, France-Lanord, Blusztajn, Allen, Alizai, Lückge, Danish, Rabbani, Clift, Tada and Zheng2010a). More broadly, warmer and more humid conditions enhance chemical alteration and clay mineral formation as a result of the close relationship between climate and weathering rates (Kump et al. Reference Kump, Brantley and Arthur2000; West et al. Reference West, Galy and Bickle2005).
Chemical alteration results in a loss of water-mobile elements (e.g., Na, K, Ca, Mg, Sr) over immobile elements (e.g., Si, Al, Ti) compared to the original composition and thus allows the intensity of alteration to be quantified (Nesbitt et al. Reference Nesbitt, Markovics and Price1980). Changing climate also influences the speed and duration of sediment transport between the source and the final depocenter (Herman & Champagnac, Reference Herman and Champagnac2016; Huntington et al. Reference Huntington, Blythe and Hodges2006; Neubeck et al. Reference Neubeck, Carter, Rittenour and Clift2023; West et al. Reference West, Galy and Bickle2005), which in turn influences the degree of alteration that detrital grains encounter. We use a combination of grain-size analysis, bulk-sediment major and trace element geochemistry with Sr and Nd isotopes to investigate the desert sand origin and the processes that have allowed the desert to form.
2. Geological evolution and Thar Sand dune accretion
Quaternary aeolian sediment deposits in the Thar Desert are interspersed between low hills of Cenozoic rocks and rest upon a substratum of Archean gneiss covered by Proterozoic sedimentary rocks and recent alluvium. The wind remobilized fluvial sediments and then created sand ridges that characterize the present desert landscape, occasionally separated by interdune clay deposits. Ahmad (Reference Ahmad2008) had hypothesized that the sand in the northern part of the desert (i.e., in Cholistan) was predominantly derived from the Sutlej River. However, detrital U-Pb zircon dating and Nd isotope data, presented by East et al. (Reference East, Clift, Carter, Alizai and VanLaningham2015), refuted this model by indicating that sediment in the region is sourced from the modern Indus delta. Sediment in the southern desert has been supplied by recycling from the mid-Holocene delta (East et al. Reference East, Clift, Carter, Alizai and VanLaningham2015). In contrast, studies by Clift et al. (Reference Clift, Lee, Hildebrand, Shimizu, Layne, Blusztajn, Blum, Garzanti and Khan2002), Garzanti et al. (Reference Garzanti, Padoan, Setti, Najman, Peruta and Villa2013b) and Usman et al. (Reference Usman, Clift, Pastore, Vezzoli, Andò, Barbarano, Vermeesch and Garzanti2024) suggested that the southern desert region (i.e., Sindh) sand is primarily sourced from the recent Indus River. Recent analysis of the Eastern Thar Desert in India employed Nd and Sr isotopes to argue that this region is also primarily Indus-supplied (Bhattacharyya et al. Reference Bhattacharyya, Singh, Qasim and Chandrashekhar2024), suggesting sediment supply from the exposed Indus Shelf during and immediately after the LGM. The pre-industrial sediment load of the Indus River was large (250–450 Mt/y) (Milliman & Farnsworth, Reference Milliman and Farnsworth2011) and could have served as the primary source of sand for the southern reaches of the Thar Desert.
The dynamics of the Indus River and its potential influence on the Thar Desert have been debated over the years. The significant role that river systems play in shaping desert landscapes, particularly in regions where fluvial processes interact with arid environments, has been recognized (Bookhagen & Burbank, Reference Bookhagen and Burbank2010; Clift et al. Reference Clift, Carter, Giosan, Durcan, Tabrez, Alizai, Van Laningham, Duller, Macklin, Fuller and Danish2012; Clift et al. Reference Clift, Lee, Hildebrand, Shimizu, Layne, Blusztajn, Blum, Garzanti and Khan2002; Garzanti et al. Reference Garzanti, Liang, Andò, Clift, Resentini, Vermeesch and Vezzoli2020; Usman et al. Reference Usman, Clift, Pastore, Vezzoli, Andò, Barbarano, Vermeesch and Garzanti2024). These studies underscored the importance of understanding sedimentary processes within the Indus River basin if their implications for the evolution of adjacent arid regions are to be appreciated. Both climatic variability and tectonic activity modulate sediment fluxes from the Himalayas to the Indus River and ultimately influence depositional patterns in the Thar region (Clift et al. Reference Clift, Giosan, Blusztajn, Campbell, Allen, Pringle, Tabrez, Danish, Rabbani, Carter and Lückge2008; Tandon & Sinha, Reference Tandon and Sinha2022). A comprehensive understanding of transport and depositional processes in the Indus River is essential for elucidating its relationship with the Thar Desert.
The arid and hot climate of the Thar Desert (annual summer temperature: 50℃, winter: 10℃ and annual rainfall of 100–200 mm) reflects its location between the Himalaya and the adjacent floodplains (Fig. 1). Furthermore, the desert is at the edge of the influence of monsoonal rains (Bookhagen & Burbank, Reference Bookhagen and Burbank2006). Previous studies of the Eastern Thar Desert showed that sand dunes have accreted in multiple phases over the last 200,000 years (Singhvi et al. Reference Singhvi, Williams, Rajaguru, Misra, Chawla, Stokes, Chauhan, Francis, Ganjoo and Humphreys2010). Desert expansion is a function of both the amount of sediment supplied and wind intensity carrying material across the region. It might be expected that periods of weak monsoonal rain, such as during the LGM (Clift & Plumb, Reference Clift and Plumb2008; Zhisheng et al. Reference Zhisheng, Clemens, Shen, Qiang, Jin, Sun, Prell, Luo, Wang, Xu, Cai, Zhou, Liu, Liu, Shi, Yan, Xiao, Chang, Wu, Ai and Lu2011), would be periods of desert expansion. However, existing provenance analysis suggests that much of the sediment supply occurred when the summer monsoon was strong, during the end of the LGM and when the climate shifted to become wetter, which highlights the strong monsoonal influence over Thar Desert sedimentation (Clift & Giosan, Reference Clift and Giosan2014; East et al. Reference East, Clift, Carter, Alizai and VanLaningham2015; Garzanti et al. Reference Garzanti, Liang, Andò, Clift, Resentini, Vermeesch and Vezzoli2020; Glennie & Singhvi, Reference Glennie and Singhvi2002; Singh et al. Reference Singh, Wasson and Agarwal1990). A recent re-examination of detrital zircon single-grain U-Pb dates suggests that much of the sediment was supplied from the lower reaches to the desert after the LGM, during the latest Pleistocene, and in the early to mid-Holocene (Usman et al. Reference Usman, Clift, Pastore, Vezzoli, Andò, Barbarano, Vermeesch and Garzanti2024).
Thermoluminescence dating of aeolian sediments in the Thar Desert has yielded evidence of multiple phases of dune accretion during the past 200 k.y., punctuated by interludes of low or weak sediment supply related to orbital precessional forcing (Nitundil et al. Reference Nitundil, Stone and Srivastava2023; Singhvi et al. Reference Singhvi, Williams, Rajaguru, Misra, Chawla, Stokes, Chauhan, Francis, Ganjoo and Humphreys2010). The last major phase of dune growth in the Eastern Thar Desert took place during a transitional climate, when the southwesterly monsoon winds were strengthening following an aridity peak during the LGM (Gebregiorgis et al. Reference Gebregiorgis, Hathorne, Sijinkumar, Nath, Nürnberg and Frank2016; Srivastava, Reference Srivastava, Lu, Gaur and Squires2023). Sand aggradation in the Eastern Thar Desert started between 17 ka and 14 ka and lasted until 9 ka, at the onset of the early Holocene wet phase (Dhir et al. Reference Dhir, Singhvi, Andrews, Kar, Sareen, Tandon, Kailath and Thomas2010; Singhvi et al. Reference Singhvi, Williams, Rajaguru, Misra, Chawla, Stokes, Chauhan, Francis, Ganjoo and Humphreys2010). In contrast, the western Thar Desert is argued to have been supplied by Indus Delta sediment since the onset of wetter, windier conditions during the Holocene (Usman et al. Reference Usman, Clift, Pastore, Vezzoli, Andò, Barbarano, Vermeesch and Garzanti2024). The desert has expanded further towards the west as the climate dried following the mid-Holocene, potentially bringing it into proximity with the Indus River (Alizai et al. Reference Alizai, Carter, Clift, VanLaningham, Williams and Kumar2011a; East et al. Reference East, Clift, Carter, Alizai and VanLaningham2015).
Although the subsidence of the Himalayan foreland requires the Indus River to flow in the deepest part of this flexural basin, there has been some evolution of the river courses through time. The delta itself has migrated to the west since the early Holocene and appears to have reached the sea near the Rann of Kutch (Fig. 1) early in the Holocene, before moving to its modern location (Inam et al. Reference Inam, Clift, Giosan, Tabrez, Tahir, Rabbani, Danish and Gupta2007). It is possible that the lower Indus used to flow through the Nara Valley, west of its present course, but provenance work was unable to resolve between this possible earlier course and infilling of a separate channel by Thar Desert sands (Alizai et al. Reference Alizai, Clift, Giosan, VanLaningham, Hinton, Tabrez and Danish2011b). Further north, the location where the Indus and its Himalayan tributaries reached the flood plains has been stable since they are fixed in deep rocky gorges, although there is evidence for the migration of the Sutlej to the NW to its modern location during the Holocene (Mehdi et al. Reference Mehdi, Pant, Saini, Mujtaba and Pande2016; Saini et al. Reference Saini, Tandon, Mujtaba, Pant and Khorana2009) and the cessation of flow in an ephemeral Ghaggar-Hakra stream by ∼4–8 ka (Clift et al. Reference Clift, Carter, Giosan, Durcan, Tabrez, Alizai, Van Laningham, Duller, Macklin, Fuller and Danish2012; Khan et al. Reference Khan, Sinha, Murray and Jain2024). Although the Yamuna used to flow to the west into the Indus in the past, this connection was likely lost before ∼20 ka (Clift et al. Reference Clift, Carter, Giosan, Durcan, Tabrez, Alizai, Van Laningham, Duller, Macklin, Fuller and Danish2012).
3. Methods
3.a. Grain size and geochemical analyses
The grain-size distribution of 27 sand samples (collected from the surface of sand dunes in the Western Thar Desert (Sindh and Cholistan, Pakistan)) was quantified at the University of Milano-Bicocca by employing standard wet sieving techniques; textural parameters were recalculated using the Folk & Ward (Reference Folk and Ward1957) classification (Table 1). For geochemical analysis, samples were ground in a mortar and then loaded into a hardened steel vial and milled to a grain size of < 30 μm using a SPEX Sample Prep 8000M Mixer/Mill. Approximately 2.00 ± 0.02 g of powder from each sample was weighed and loaded into a furnace at 900°C for two hours. Following extraction from the furnace, the sample powders were re-weighed and their loss on ignition (LOI) calculated. Geochemical analysis was conducted with a Bruker S2-PUMA energy-dispersive X-ray fluorescence (XRF) instrument at the Chevron Geomaterials Characterization Laboratory at Louisiana State University (LSU), after calibration against 19 international standards. Analytical uncertainties calculated as a percentage of the content were ∼15% for Na2O and <2% for the other elements, reflecting the higher volatility and mobility of Na during analysis. The results are presented in Tables 1 and 2.
Grain-size analyses of studied samples and major-element distribution in aeolian sand of the Sindh and Cholistan deserts are determined by X-ray fluorescence, with different weathering proxies
Trace-element distribution with alpha values normalized to non-mobile Al calculated in aeolian sand of the Sindh and Cholistan deserts determined by X-Ray Fluorescence
3.b. Sr and Nd Isotopes
Sr and Nd isotopes have been widely used as provenance proxies (Goldstein et al. Reference Goldstein, O’Nions and Hamilton1984) and were also found to be successfully effective in the Indus River system (Clift et al. Reference Clift, Lee, Hildebrand, Shimizu, Layne, Blusztajn, Blum, Garzanti and Khan2002). Their contents were measured in 11 representative powdered bulk-sediment samples based on geochemical content. After decarbonation with 10% acetic acid and dissolution, Sr and Nd were concentrated by standard column extraction techniques, and isotopic compositions were measured by the Thermo ‘Neptune’ multi-collector inductively coupled plasma mass spectrometer (MC-ICP-MS) at the Woods Hole Oceanographic Institution. Analytical methods are described in the Supplementary Information. However, analytical uncertainties are extremely low, and Jonell et al. (Reference Jonell, Li, Blusztajn, Giosan and Clift2018) noted that within the Indus drainage system, bulk isotopic compositions may vary up to ±1.04 units for ϵNd and ±0.0099 for 87Sr/86Sr values in any sediment because of mineralogy and grain size distribution. This real-world uncertainty is much greater than the analytical error. The results are presented in Table 3.
87Sr/86Sr and 143Nd/144Nd isotopic ratios determined by Thermo the ‘Neptune’ multi-collector inductively coupled plasma mass spectrometer (MC-ICP-MS) at Woods Hole Oceanographic Institution
4. Geochemical variability of the Thar Desert
4.a. Major and trace elements
The variability of major-element concentrations in Thar dune sand is moderate. SiO2 (Sindh 62–72%; Cholistan 64–71%), Al2O3 (Sindh 5–10%, Cholistan 9–10%), Fe2O3 (Sindh 5–10%, Cholistan 9–10%), MgO (Sindh 5–10%, Cholistan 9–10%), CaO (Sindh 5–10%, Cholistan 9–10%), Na2O (Sindh 1–2%, Cholistan 2–3%), K2O (Sindh 1.5–1.8%, Cholistan 1.6–2.0%), TiO2 (Sindh 0.2–0.7%, Cholistan 0.3–0.6%), P2O5 (Sindh 0.1–0.3%, Cholistan 0–0.2%) and MnO (Sindh and Cholistan <0.1%) (Fig. 2). Such chemical variability is partly grain-size dependent. Sindh dune sand is significantly coarser than Cholistan sand and has mainly lower SiO2 content (Fig. 3a) and higher CaO (Fig. 3b). Differences and significant variabilities in the distribution of major oxides between Cholistan and Sindh desert sands, together with their correlation trends, are illustrated in Figure S1.
Diagram showing major-element variability and grain-size characteristics of studied aeolian-dune sediments from the Sindh and Cholistan deserts. The range from the Eastern Thar Desert is from Bhattacharyya et al. (Reference Bhattacharyya, Singh, Qasim and Chandrashekhar2024).
Cross-plots showing that Sindh Desert sand is coarser but has lower SiO2 (A) and higher CaO (B) than Cholistan Desert sand.
In general, Sindh Desert sand is richer in Sr, S, V, Cr, Nb and Sb than that from Cholistan. Trace element contents of Thar Desert sand do not show major systematic differences with both the Indus and Himalayan tributary sands (Fig. 4b). Normalizing trace-element concentration against the Upper Continental Crust (UCC) (Taylor & McLennan, Reference Taylor and McLennan1995) allows us to compare the Thar Desert sands with the signatures of Upper and Lower Indus sediments together with Himalayan tributary sands determined by earlier studies (Garzanti et al. Reference Garzanti, Liang, Andò, Clift, Resentini, Vermeesch and Vezzoli2020; Liang et al. Reference Liang, Garzanti, Andò, Gentile and Resentini2019) (Figs. 4a, 4b). Analyses from the Eastern Thar Desert (Bhattacharyya et al., Reference Bhattacharyya, Singh, Qasim and Chandrashekhar2024) do not include as many elements as provided by this work, but when comparison is possible, the different sands are seen to be similar, except that Zr contents are much lower in the east (Fig. 4a).
Trace element compositions normalized to the Upper Continental Crust (UCC) standard for (A) Sindh and Cholistan dune sand, compared with sand of the Upper and Lower Indus River and Eastern Thar Desert (Bhattacharyya et al. Reference Bhattacharyya, Singh, Qasim and Chandrashekhar2024); (B) the Thal Desert and major Punjabi tributaries; and (C) river sands derived from end-member sources (data from Garzanti et al. (Reference Garzanti, Liang, Andò, Clift, Resentini, Vermeesch and Vezzoli2020)).
4.b. Sr and Nd Isotopes
Sindh Desert sand yields generally less negative ϵNd (−9.0 to -12.0) values compared to the Cholistan sand (−11.8 to −14.0) (Fig. 5), which also has slightly higher 87Sr/86Sr values (0.71769 to 0.72528 in Sindh versus 0.72235 to 0.73087 in Cholistan). Sindh Desert sand also has less negative ϵΝd values compared to the Holocene Indus post-LGM delta (ϵNd values of -11.8 to -10.8 before 12 ka) (Clift et al. Reference Clift, Tada and Zheng2010b). In comparison with potential source regions, Thar Desert sands are similar to Holocene Indus River and delta sands (ϵNd values of -12.9 to -15.4 since 9 ka) (Clift et al. Reference Clift, Giosan, Blusztajn, Campbell, Allen, Pringle, Tabrez, Danish, Rabbani, Carter and Lückge2008), with intermediate values between Karakorum and Himalayan end-member sources (Fig. 5).
Cross-plot of Sr and Nd isotope values for Sindh and Cholistan dune sands compared to end-member sources and post-15 ka Indus Delta sediments (Garzanti et al. Reference Garzanti, Liang, Andò, Clift, Resentini, Vermeesch and Vezzoli2020). Data sources: Transhimalayan: Rolland et al. (Reference Rolland, Picard, Pecher, Lapierre, Bosch and Keller2002), Singh et al. (2002) and Khan et al. (Reference Khan, Stern, Gribble and Windley1997); Greater Himalaya: Ahmad et al. (Reference Ahmad, Harris, Bickle, Chapman, Bunbury and Prince2000), Deniel et al. (Reference Deniel, Vidal, Fernandez, Lefort and Peucat1987), Inger et al. (1993) and Parrish & Hodges (Reference Parrish and Hodges1996); Karakorum: Crawford & Searle (Reference Crawford and Searle1992) and Schärer et al. (Reference Schärer, Copeland, Harrison and Searle1990); Eastern Thar Desert: Bhattacharyya et al. (Reference Bhattacharyya, Singh, Qasim and Chandrashekhar2024).
5. Compositional signatures of the Thar Desert
The geochemical signatures of the Thar Desert sands (Fig. S2a) are here compared with those of mainstream Indus River sediments and its Himalaya-draining Punjabi tributaries (Fig. S2b) and details in Table 4. Sindh Desert sands are similar to the lower Indus River and the Indus in Ladakh, India. Himalayan-derived river sediment contains significantly less CaO and more Na2O and TiO2 than the Sindh average. Cholistan sands are similar to the Himalayan bedrock and the Sutlej River, especially with regard to Ni, Cu, Zn, Al2O3 and Ga contents. As far as grain size is concerned, Sindh dunes tend to be coarser-grained than Cholistan dunes (Fig. 3), which is ascribed to deflation by strong southwesterly summer monsoon winds that blow sand from the Indus delta inland (East et al. Reference East, Clift, Carter, Alizai and VanLaningham2015; Usman, Reference Usman2024; Usman et al. Reference Usman, Clift, Pastore, Vezzoli, Andò, Barbarano, Vermeesch and Garzanti2024). On the contrary, the prevalence of finer-grained sediments in Cholistan dunes is attributed to a significant amount of detritus supplied by Himalayan rivers.
5.a. Geochemical proxies and grain size in the Thar Desert
Geochemical indices are widely used to estimate weathering intensity (Price, Reference Price1995; Minyuk et al. Reference Minyuk, Borkhodoev and Wennrich2014; Duzgoren-Aydin & Aydin, Reference Duzgoren-Aydin and Aydin2003; Bloemsma et al. Reference Bloemsma, Zabel, Stuut, Tjallingii, Collins and Weltje2012; Guo et al. Reference Guo, Yang and Deng2021: Maslov & Podkovyrov, Reference Maslov and Podkovyrov2023; Price & Velbel, Reference Price and Velbel2003), even though these proxies are also strongly controlled by grain size, source-rock lithology and hydraulic sorting (Dinis et al. Reference Dinis, Garzanti, Vermeesch and Huvi2017; Garzanti & Resentini, Reference Garzanti and Resentini2016). The Chemical Index of Alteration (CIA) (Nesbitt & Young, Reference Nesbitt and Young1982) is widely used and is calculated from molar proportions with the following equation:
$$CIA = (Al_2O _3 / (Al_2O_3 + CaO + NaO + K_2O))*100$$
CIA only considers CaO hosted in silicate minerals (if the number of CaO moles after correcting for CaO in phosphate is greater than that of Na2O, CaO in silicates can be assumed as = Na2O; CIA*, (McLennan, Reference McLennan1993)). CIA values for unweathered source rocks range from 30−40 for basalt to 45−55 for granite and granodiorite, whereas they are 75−85 for illite, ∼75 for muscovite, and ∼100 for kaolinite and chlorite.
An alternative proxy is the weathering index of Parker (Reference Parker1970) (WIP). WIP is calculated with the following molar equation:
$$WIP = \left({{2Na_2O\over 0.35} + {MgO\over 0.9} + {2K_2O\over 0.25} + {CaO\over 0.7}}\right) *100$$
Other geochemical indices have also been used as proxies to track weathering intensity, e.g., K/Al and K/Rb (Nesbitt & Young, Reference Nesbitt and Young1982; Price & Velbel, Reference Price and Velbel2003). However, all such proxies are also partially controlled by grain size (von Eynatten et al. Reference von Eynatten, Tolosana-Delgado and Karius2012; von Eynatten et al. Reference von Eynatten, Tolosana-Delgado, Karius, Bachmann and Caracciolo2016), hydraulic sorting or quartz addition by recycling of older sediments (Fig. 6). Rb is not as water-immobile as Al, but it is less water-mobile than K in micas and K-feldspar grains (Nesbitt et al. Reference Nesbitt, Fedo and Young1997). K/Rb has been used to effect in several studies in Asia and was tested against an array of environmental proxies by Hu et al. (Reference Hu, Clift, Wan, Böning, Hannigan, Hillier, Blusztajn, Clift, Harff, Wu and Qiu2016) in a synthesis that showed K/Rb to be more sensitive to chemical weathering than K/Al, at least in the Pearl River. There is a common relationship between different weathering proxies (e.g., CIA*, K/Al, K/Rb) and mean grain size, because finer sediments are generally expected to show a higher degree of alteration. Coarser sediments tend to have higher CIA* values (Fig. 6a), implying a stronger control by source-rock lithology and sediment recycling rather than weathering over this proxy. Eastern Thar sediments plot closest to Sindh sediments (Fig. 7) in being coarser and having higher CIA* values and even exceed these, trending towards Himalayan tributary or Upper Indus sediment compositions. Finer-grained Thar sediments have a lower K/Rb ratio (Fig. 6b), indicating that this proxy may be more sensitive to grain size. Finally, no relationship is observed between LOI, which is expected to be higher when alteration is greater, and grain size (Fig. 6c).
Cross plots, showing the relationship between mean grain size and a variety of chemical weathering indices for sediment from both the Sindh and Cholistan deserts, as well as from the Upper and Lower Indus (Garzanti et al. Reference Garzanti, Liang, Andò, Clift, Resentini, Vermeesch and Vezzoli2020) and major Punjabi tributaries (Clift et al. Reference Clift, Tada and Zheng2010b). Mean grain size versus (A) CIA*, (B) K/Rb, (C) LOI and (D) Mg/Al.
Geochemical signatures. A) CN-A-K ternary diagram (Fedo et al. Reference Fedo, Nesbitt and Young1995) comparing studied samples with the Eastern Thar Desert (Bhattacharyya et al. Reference Bhattacharyya, Singh, Qasim and Chandrashekhar2024), Holocene sediments from the Indus Canyon (Li et al. Reference Li, Clift, Böning, Blusztajn, Murray, Ireland, Pahnke, Helm and Giosan2018) and onshore delta (Clift et al. Reference Clift, Tada and Zheng2010b). CN, A and K are the mole weights of Na2O and CaO* (CaO associated with silicates only), Al2O3 and K2O, respectively. CIA values are shown on the left side: sm, smectite; pl, plagioclase; ksp, K-feldspar; il, illite; m, muscovite. B) Cross plot of Fe2O3/SiO2 vs. Al2O3/SiO2 used as a proxy of grain size (Singh et al. Reference Singh, Sharma and Tobschall2005). Data sources: Indus Canyon from Li et al. (Reference Li, Clift, Böning, Blusztajn, Murray, Ireland, Pahnke, Helm and Giosan2018), Indus Delta from Clift et al. (Reference Clift, Tada and Zheng2010b), Siwalik Group from Vögeli et al. (Reference Vögeli, van der Beek, Huyghe and Najman2017) and Exnicios et al. (Reference Exnicios, Carter, Najman and Clift2022), and Himalaya from Galy & France-Lanord (Reference Galy and France-Lanord2001). C) CIA* vs. WIP plot was plotted for the Sindh and Cholistan dune sands, which are indicating slight quartz addition and less weathering intensity for the studied aeolian sands.
Chemical weathering proxies (e.g., CIA*, K/Al, Mg/Al) have different sensitivities to weathering intensity, which reflects the mineralogy of the sediment sources and the relative mobility of the elements within them (Hu et al. Reference Hu, Clift, Wan, Böning, Hannigan, Hillier, Blusztajn, Clift, Harff, Wu and Qiu2016). We compare the different weathering proxies by cross-plotting them against one another. In both Sindh and Cholistan, high Mg/Al is correlated with low CIA* (Fig. 6d). This implies that Mg/Al is not primarily a weathering proxy but is linked to the influence of Mg-rich mafic rocks widely exposed in Kohistan and the Karakorum.
5.b. Provenance versus weathering effects
The Thar Desert sands are mostly enriched in the Ca-Na endmember in the (Ca+Na)-Al-K diagram (Nesbitt & Young, Reference Nesbitt and Young1989) (Fig. 7a), reflecting the prevalence of plagioclase over K-feldspar in the sources. Average CIA values (51 for Cholistan sand, 54 for Sindh sand) indicate low weathering intensity. Sediments from the Holocene Indus Canyon (Li et al. Reference Li, Clift, Böning, Blusztajn, Murray, Ireland, Pahnke, Helm and Giosan2018) and post-glacial Indus Delta (Clift et al. Reference Clift, Giosan, Carter, Garzanti, Galy, Tabrez, Pringle, Campbell, France-Lanord, Blusztajn, Allen, Alizai, Lückge, Danish, Rabbani, Clift, Tada and Zheng2010a) are finer grained than Thar Desert sand and show higher CIA indices. The Al2O3/SiO2 ratio, which is often used as a grain-size proxy, and the Fe2O3/SiO2 exhibits a wide variation in Indus River sediments (Fig. 7b). Holocene and modern river sediments are invariably enriched in quartz relative to their sources and this enrichment is more pronounced in sediments of the Indus delta and the Thar Desert, especially the Sindh Desert. Weathering indices such as the CIA or WIP indicate a slightly higher of weathering intensity for sand carried by the Himalayan Punjabi tributaries and the upper Indus than eolian sediments in the Sindh, Cholistan and eastern deserts (Fig. 7c).
The weathering effect is best resolved from other controls over geochemical composition if mobile elements (e.g., Na, Ca, Sr, Mg, K and Ba) are considered individually (Table 2). This can be done by using αAlE values, defined as (Al/E)sample/(Al/E)UCC (Garzanti et al. Reference Garzanti, Padoan, Setti, López-Galindo and Villa2014a; Garzanti et al. Reference Garzanti, Vermeesch, Padoan, Resentini, Vezzoli and Andò2014b), a parameter that compares the concentration of any mobile element E with reference to non-mobile Al in our samples compared to the UCC. Aluminium, which is hosted in a wide range of rock-forming minerals, including phyllosilicates (concentrated in mud) and feldspars (concentrated in sand), is used as a reference for all elements rather than Ti, Nd, Sm or Th (Gaillardet et al. Reference Gaillardet, Dupré and Allègre1999). Those immobile elements are preferentially hosted in ultra-dense minerals and may thus reach anomalous concentrations as a result of hydrodynamic processes (Garzanti et al. Reference Garzanti, Padoan, Andò, Resentini, Vezzoli and Lustrino2013a; Garzanti et al. Reference Garzanti, Padoan, Setti, Najman, Peruta and Villa2013b).
Sediments from the Thar Desert, Indus River and its Punjabi tributaries show a weak depletion in highly mobile Na (αAlNa 0.9 to 1.6 in Sindh and 0.8 to 1.3 in Cholistan) compared to the Indus River (Fig. 8a, b). Cholistan and Eastern Thar Desert sands are enriched in Ca, Sr and Mg compared to the Indus, while Sindh sands are generally more depleted than the other desert areas and the Indus, except in regard to Mg. αAlCa = 0.2−0.6 and 0.4−0.9, αAlK = 0.6−1.2 and 1.1−1.3, αAlSr = 0.2−1.4 and 0.5−3.2 and αAlMg = 0.6−3.8 and 1.1−4.0 in Sindh and Cholistan sands, respectively. Instead, Ba is more strongly depleted in both western and eastern deserts (αAlBa 0.6−1.2). The Thar Desert has comparable αAlE values to many of the Himalayan tributaries, except for the Ravi, which is enriched in Na, Ca and Sr. The variability between the Sindh and Cholistan deserts in concentration factors indicates differences in the weathering influence (Viers et al. Reference Viers, Dupré and Gaillardet2009); however, the lower αAlE values in Sindh imply stronger weathering.
Weathering indices of AlphaAlE of sand fractions in the Thar (Sindh and Cholistan) Desert. Elemental data in previous studies were plotted for comparison, including bulk sediment (Garzanti et al. Reference Garzanti, Liang, Andò, Clift, Resentini, Vermeesch and Vezzoli2020). αAl E values (Garzanti et al. Reference Garzanti, Padoan, Setti, López-Galindo and Villa2014a; Garzanti et al. Reference Garzanti, Vermeesch, Padoan, Resentini, Vezzoli and Andò2014b) indicate negligible weathering intensity, especially for Sindh Desert sand displaying the same fingerprint as Upper Indus, Thal Desert and Lower Indus sands (A, data from Garzanti et al. (Reference Garzanti, Liang, Andò, Clift, Resentini, Vermeesch and Vezzoli2020)). B) Cholistan sand is slightly more depleted in Sr and Mg, which is an inherited effect consequence of greater supply from Himalayan Punjabi tributaries (data from Garzanti et al. (Reference Garzanti, Liang, Andò, Clift, Resentini, Vermeesch and Vezzoli2020)).
To explore these conclusions further, we have used the cross plot of K/Si versus Al/Si (Lupker et al. Reference Lupker, France-Lanord, Galy, Lave, Gaillardet, Gajured, Guilmette, Rahman, Singh and Sinha2012) (Fig. 9). We compare the Indus and Thar Desert sediments with material from the Indus Canyon (Li et al. Reference Li, Clift, Böning, Blusztajn, Murray, Ireland, Pahnke, Helm and Giosan2018) and Indus Delta (Clift et al. Reference Clift, Giosan, Carter, Garzanti, Galy, Tabrez, Pringle, Campbell, France-Lanord, Blusztajn, Allen, Alizai, Lückge, Danish, Rabbani, Clift, Tada and Zheng2010a) to understand the weathering intensity while accounting for grain size effects. Weathering intensity affects the gradient of the array, with finer-grained sediment typically showing higher Al/Si values. The sediments show an overall coherent array between the offshore fine-grain sediments, the modern river and the Sindh and Cholistan desert sediments. This indicates that they are part of a coherent sediment grouping, but that the offshore sediments are generally finer-grained and further transported than those seen onshore in rivers and deserts. The Thal Desert and some of the mainstream Indus River depart most from this general trend. In contrast, this figure implies that marine sediment may be being reworked from the Thar Desert prior to transport offshore.
Cross plot of K/Si versus Al/Si for samples from the offshore submarine canyon and the Holocene Indus delta compared to the modern desert sands. This plot reveals differences in overall weathering intensity based on the gradient of the array (Lupker et al., Reference Lupker, France-Lanord, Galy, Lave, Gaillardet, Gajured, Guilmette, Rahman, Singh and Sinha2012). The gradient defined by the offshore fine-grained sediments is consistent with the desert sediments as well as the Upper and Lower Indus and the major Punjabi tributaries, indicating that they are part of a coherent sediment grouping. Canyon data are from Li et al. (Reference Li, Clift, Böning, Blusztajn, Murray, Ireland, Pahnke, Helm and Giosan2018). Delta data are from Clift et al. (2010).
5.c. Mineralogical fingerprints and provenance studies
In the Indus River catchment, mafic rocks are widely exposed in the Kohistan Range and to a lesser extent in the Karakorum, both drained by the Upper Indus, whereas they are scarce in the Himalayan thrust belt drained by the Punjabi tributaries. Cholistan dune sand is feldspatho-litho-quartzose with dominant monocrystalline quartz and subequal amounts of plagioclase and K-feldspar (Usman et al. Reference Usman, Clift, Pastore, Vezzoli, Andò, Barbarano, Vermeesch and Garzanti2024). Sindh dune sand is litho-felspatho-quartzose, with sedimentary rock fragments prevailing over metapelite, metapsammite and metavolcanic grains. Heavy mineral assemblages in all Thar Desert sands consist of hornblende, subordinate epidote and garnet, and minor clinopyroxene, hypersthene, staurolite, titanite, kyanite and fibrolitic sillimanite, a typical association of erosion from an orogen (Garzanti & Andò, Reference Garzanti, Andò, Mange and Wright2007). Sindh sands have less quartz than those from Cholistan (53 ± 2% vs. 60 ± 2%), are more sedimentary than metamorphic lithics (Lm 39 ± 6 Ls 58 ± 5 vs. Lm 62 ± 2 Ls 37 ± 2) and have higher heavy-mineral concentrations than Cholistan sand (Usman et al. Reference Usman, Clift, Pastore, Vezzoli, Andò, Barbarano, Vermeesch and Garzanti2024). Detrital zircon U-Pb ages are dominantly Palaeozoic and Neoproterozoic, and this resembles the ages of basement rocks of the Karakorum and Himalayan sources (details in Usman et al. (Reference Usman, Clift, Pastore, Vezzoli, Andò, Barbarano, Vermeesch and Garzanti2024), Section 6.2 and Figs. 6 and 7). Mesoproterozoic to Paleoproterozoic detrital zircons are primarily derived from the Greater Himalaya, Lesser Himalaya and the Nanga Parbat Massif. Younger Cretaceous to Oligocene grains are predominantly derived from the Karakorum (130–99 Ma and 43–24 Ma) and Transhimalayan arcs (96–43 Ma), whereas the Baltoro granite within the Karakorum is likely the source of Miocene grains (21–17 Ma) (Mahar et al. Reference Mahar, Mahéo, Goodell and Pavlis2014). The combination of petrographic, heavy-mineral and detrital-geochronology methods indicates that aeolian dunes in the northern Cholistan area contain greater proportions of sediment delivered by Punjabi tributaries sourced in the Himalayan belt compared to aeolian dunes in the southern Sindh area. Southern dune sediments have a greater compositional affinity to Indus Delta sediments dated between ∼7 and 14 ka, which are relatively enriched in Karakorum-Transhimalayan sediment. This could imply a higher sediment supply to the Sindh Desert in the early Holocene and more supply to the Cholistan Desert after the early Holocene when the Indus River became more Himalayan in character (Clift et al. Reference Clift, Giosan, Blusztajn, Campbell, Allen, Pringle, Tabrez, Danish, Rabbani, Carter and Lückge2008).
5.d. Sr and Nd isotope fingerprints
Sindh sand has less negative ϵNd values not only compared to Cholistan and Eastern Thar sand but also to any sediment within the Indus River basin, apart from Thal Desert dunes (where ϵNd values are as high as -8.7 and even -3.5) (Fig. 10a). Sindh sands are closest to the modern Upper Indus River at Tarbela Dam (Clift et al. Reference Clift, Lee, Hildebrand, Shimizu, Layne, Blusztajn, Blum, Garzanti and Khan2002). The ϵNd values of Cholistan sand, instead, largely overlap with ϵNd values of Holocene Indus Delta sediments and with Eastern Thar sands (Bhattacharyya et al. Reference Bhattacharyya, Singh, Qasim and Chandrashekhar2024). Cholistan sands are slightly more ϵNd negative than LGM sediments at the Indus River mouth (Fig. 10a). These data indicate a more primitive source for the Sindh Desert compared to Cholistan and the Eastern Thar.
A) KDE plot of ϵNd values of aeolian sand from Sindh and Cholistan deserts compared with sand carried by Sutlej and Jhelum rivers (Clift et al. Reference Clift, Lee, Hildebrand, Shimizu, Layne, Blusztajn, Blum, Garzanti and Khan2002), Eastern Thar Desert (Bhattacharyya et al. Reference Bhattacharyya, Singh, Qasim and Chandrashekhar2024), Holocene sediments of Punjabi floodplain (Alizai et al. Reference Alizai, Carter, Clift, VanLaningham, Williams and Kumar2011a; East et al. Reference East, Clift, Carter, Alizai and VanLaningham2015), post-LGM Indus delta (Clift et al. Reference Clift, Giosan, Blusztajn, Campbell, Allen, Pringle, Tabrez, Danish, Rabbani, Carter and Lückge2008), Upper Indus River upstream of Tarbela Dam (Garzanti et al. Reference Garzanti, Liang, Andò, Clift, Resentini, Vermeesch and Vezzoli2020) and river mouth/delta sediments from LGM to present (Clift et al. Reference Clift, Giosan, Blusztajn, Campbell, Allen, Pringle, Tabrez, Danish, Rabbani, Carter and Lückge2008; Clift et al. Reference Clift, Lee, Hildebrand, Shimizu, Layne, Blusztajn, Blum, Garzanti and Khan2002). B) Range of ϵNd values characterizing bedrock in main geological units drained by the Indus River. Data sources: Kohistan from Petterson et al. (Reference Petterson, Crawford and Windley1993), Khan et al. (Reference Khan, Stern, Gribble and Windley1997) & Jagoutz et al. (Reference Jagoutz, Bouilhol, Schaltegger and Müntener2019)); Ladakh batholith from Rolland et al. (Reference Rolland, Picard, Pecher, Lapierre, Bosch and Keller2002); Karakorum from Schärer et al. (Reference Schärer, Copeland, Harrison and Searle1990), Crawford & Searle (Reference Crawford and Searle1992), Mahéo et al. (Reference Mahéo, Blichert-Toft, Pin, Guillot and Pêcher2009) and Jagoutz et al. (Reference Jagoutz, Bouilhol, Schaltegger and Müntener2019); Nanga Parbat from George et al. (Reference George, Harris and Butler1993), Gazis et al. (Reference Gazis, Blum, Chamberlain and Poage1998), Whittington et al. (Reference Whittington, Foster, Harris, Vance and Ayres1999), Foster (Reference Foster2000) and Argles et al. (Reference Argles, Foster, Whittington, Harris and George2003); Tethys Himalaya from Whittington et al. (Reference Whittington, Foster, Harris, Vance and Ayres1999), Ahmad et al. (Reference Ahmad, Harris, Bickle, Chapman, Bunbury and Prince2000) and Robinson et al. (Reference Robinson, DeCelles, Patchett and Garzione2001); Greater Himalaya from Deniel et al. (Reference Deniel, Vidal, Fernandez, Lefort and Peucat1987), Stern et al. (Reference Stern, Kligfield, Schelling, Virdi, Futa, Peterman and Malinconico1989), Bouquillon et al. (Reference Bouquillon, France-Lanord, Michard, Tiercelin, Cochran, Stow and Auroux1990), France-Lanord et al. (Reference France-Lanord, Derry, Michard, Treloar and Searle1993), Parrish & Hodges (Reference Parrish and Hodges1996), Ahmad et al. (Reference Ahmad, Harris, Bickle, Chapman, Bunbury and Prince2000), Miller et al. (Reference Miller, Thöni, Frank, Grasemann, Klotzli, Guntli and Draganits2001), Robinson et al. (Reference Robinson, DeCelles, Patchett and Garzione2001) and Martin et al. (Reference Martin, DeCelles, Gehrels, Patchett and Isachsen2005); Lesser Himalaya from Bouquillon et al. (Reference Bouquillon, France-Lanord, Michard, Tiercelin, Cochran, Stow and Auroux1990), Parrish & Hodges (Reference Parrish and Hodges1996), Ahmad et al. (Reference Ahmad, Harris, Bickle, Chapman, Bunbury and Prince2000) and Robinson et al. (Reference Robinson, DeCelles, Patchett and Garzione2001); Siwaliks from Huyghe et al. (Reference Huyghe, Galy, Mugnier and France-Lanord2001) and Chirouze et al. (Reference Chirouze, Huyghe, Chauvel, van der Beek, Bernet and Mugnier2015).
The Sr and Nd isotope composition of the Indus has changed since the LGM caused by variations in erosion patterns driven by monsoon intensification (Clift et al. Reference Clift, Giosan, Blusztajn, Campbell, Allen, Pringle, Tabrez, Danish, Rabbani, Carter and Lückge2008). Sindh, Cholistan and Eastern Thar desert sands display ϵNd values less negative than Himalayan-derived sediments transported by Himalayan tributaries and deposited in the Punjab plains (Tripathi et al. Reference Tripathi, Bock, Rajamani and Eisenhauer2004). This implies that the Thar Desert sands are mainly derived from the Indus River, in which Himalayan sources are subordinate, but these are more significant for the Cholistan and Eastern Thar Desert than in Sindh (Figs. 10a, b).
Because Sindh sand is not directly delivered from the upstream Indus but is first transported into the lower reaches, it seems most likely that Sindh received more material eroded from the exposed continental shelf and delta during the glacial era when the Indus was less ϵNd negative. In contrast, Cholistan and the Eastern Thar are similar to the post-glacial isotopic fingerprint. This would seem to indicate that sediments in Cholistan and the Eastern Thar Desert were transported to these locations in more recent times than those found in Sindh. However, our results imply a complete bypass of Sindh in the middle to late Holocene, although this process should not be interpreted by the source rock perspective only.
Grain size holds the key to this mismatch. Jonell et al. (Reference Jonell, Li, Blusztajn, Giosan and Clift2018) showed that during the Holocene, sediment >125 µm tends to have less negative ϵNd values by 1–2 ϵNd points than bulk sediment at the delta. The higher ϵNd values in Sindh are consistent with the generally coarser grain size of those samples, while the finer sediment in Cholistan has more negative ϵNd values. Lack of grain size information from the Eastern Thar Desert prevents testing whether the same process may be affecting the more negative ϵNd values there, although the longer transport distance from the delta source would favour a preponderance of finer-grained sediment in that region too. We suggest that the finer-grain sediment being supplied from the delta is transported further and preferentially deposited in the north and the eastern parts of the desert, while coarser material remains closer to the source. A simple grain size sorting by deflation could account for the observed isotopic difference. It is, however, noteworthy that even Cholistan is less ϵNd negative than the post-LGM sediments and much more positive than the modern river mouth. This implies that the sediments reaching the Thar Desert were mobilized in the Early Holocene when the river had ϵNd values around -12, which was also a time of strong summer monsoon winds.
5.e. Recycling of sand from the Indus River to the Thar Desert
The degree of recycling of sediment can be partly constrained through consideration of the ‘transparent Heavy Mineral Concentration index’ (tHMC) (Garzanti & Andò, Reference Garzanti, Andò, Mange and Wright2007). This is calculated with the following equation:
$$\rm tHMC = HMC(1-opaque-turbid)$$
where % opaque and % turbid are the percentages of opaque and turbid heavy minerals over total heavy mineral concentration (HMC). The ZTR index is also a useful proxy and was defined by Hubert (Reference Hubert1962) as the percentage of chemically ultra-stable species (zircon, tourmaline and rutile) among transparent detrital heavy minerals.
The low proportion of durable zircon, tourmaline and rutile (ZTR<4) argues in favour of limited recycling of older sedimentary rock detritus, which should be richer in ZTR because of the removal of less robust phases during transport or decay during diagenesis. In contrast, the high concentration of transparent heavy minerals in both regions (tHMC = 6–15%) and the common presence of pyroxene (7 ± 2% of the total heavy minerals (tHM)), including both clinopyroxene and orthopyroxene), implies significant erosion from mafic rocks such as those exposed in the Kohistan and Karakorum (Liang et al. Reference Liang, Garzanti, Andò, Gentile and Resentini2019). Himalayan tributaries carry sand with notably lower tHMC concentrations (Sutlej = 6%, Ravi, Chenab and Beas = 1%) and with much less pyroxene (Beas = 5%, Sutlej, Ravi and Chenab = 1%) (Garzanti et al. Reference Garzanti, Liang, Andò, Clift, Resentini, Vermeesch and Vezzoli2020; Usman, Reference Usman2024; Usman et al. Reference Usman, Clift, Pastore, Vezzoli, Andò, Barbarano, Vermeesch and Garzanti2024). Using heavy-mineral suites alone could result in underestimation of the contribution from the Himalayan tributaries, although heavy minerals may be concentrated by hydrodynamic sorting.
Thermoluminescence dating of aeolian sediments from the Thar Desert has revealed multiple phases of dune accretion, sand recycling and accumulation that may have taken place in the last 200 k.y. (Nitundil et al. Reference Nitundil, Stone and Srivastava2023; Singhvi et al. Reference Singhvi, Williams, Rajaguru, Misra, Chawla, Stokes, Chauhan, Francis, Ganjoo and Humphreys2010). Zircon grains with U-Pb diagnostic signatures of Upper Indus provenance (i.e., 43–96 Ma for Transhimalayan arcs and 130–99 Ma, 43–24 Ma and 21–17 Ma for Karakorum and Baltoro Granite) occur both in Sindh and Cholistan sands but not in Himalayan-derived sand carried by the eastern tributaries draining the edges of the Cholistan Desert. This suggests that Indus sand has been extensively blown northward, especially in the Sindh Desert, by wind transportation during the latest glacial-interglacial cycles (Usman et al. Reference Usman, Clift, Pastore, Vezzoli, Andò, Barbarano, Vermeesch and Garzanti2024). Sand transported by the eastern Himalayan tributaries tends to exhibit depletion in most elements compared to the UCC (this study and Garzanti et al. (Reference Garzanti, Liang, Andò, Clift, Resentini, Vermeesch and Vezzoli2020)), which indicates much more extensive recycling from sedimentary and metasedimentary rocks (Figs. 4b, 8 and S2).
Garzanti et al. (Reference Garzanti, Liang, Andò, Clift, Resentini, Vermeesch and Vezzoli2020) further noted that Lower Indus sand is depleted in most elements except Si, Ca and P when contrasted with Upper Indus sand. This points to a significant additional contribution from erosion of quartz-rich metasedimentary and siliciclastic rocks. Enhanced reworking by river incision of older floodplain sediments occurred during the Holocene (Giosan et al. Reference Giosan, Clift, Macklin, Fuller, Constantinescu, Durcan, Stevens, Duller, Tabrez, Adhikari, Gangal, Alizai, Filip, VanLaningham and Syvitski2012), as well as after the onset of extensive agricultural activities (Li et al. Reference Li, Clift, Murray, Exnicios, Ireland and Böning2019). Furthermore, relative to the Lower Indus River sand, deltaic sediments from the LGM to the Holocene period and Thar Desert sediments display enrichment in most elements (Fig. S2).
5.f. Comparative provenance signatures of the Thar Desert and sources
The two distinct provenance patterns of the Thar Desert sands are relative to potential sources in the Upper and Lower Indus, in the Punjab tributaries and the Indus Delta, as shown in the conceptual Figure 11 and Table 4. The Sindh Desert sands are coarser, have lower SiO2, higher CaO, Sr-Cr-Nb enrichment and less negative ϵNd values (−9 to −12), which closely resemble Indus Delta and Upper Indus sediments derived from Karakorum and Transhimalayan arcs. Sands of the Cholistan Desert are finer and more quartz- and Al2O3-enriched, have lower ϵNd values (−12 to −14) and greater proportions of metamorphic lithics and Proterozoic zircon populations, suggesting more Greater and Lesser Himalayan sources via Punjab tributaries. Upper Indus sands are more mafic-enriched, while Punjab tributaries deliver more Al2O3, transition metals and Proterozoic zircons, with Indus Delta deposits having a composite signal as might be expected. In general, the Thar Desert records a bimodal provenance pattern: southern Sindh is linked to Indus/Transhimalayan sources, and northern Cholistan is linked to Himalayan Punjabi tributary sources.
This conceptual diagram visually explains the provenance and transport history of sand in the Thar Desert, demonstrating why the southern Sindh Desert and the northern Cholistan Desert sand have different compositions.
Comparison of geochemical, isotopic, mineralogical and provenance features of sediments from the Sindh Desert, Cholistan Desert and potential sediment sources (Upper Indus, Punjab tributaries, Indus Delta). The dataset integrates major and trace element geochemistry, Sr–Nd isotopic signatures, mineralogy, detrital zircon U–Pb age spectra and weathering proxies, highlighting compositional overlaps and contrasts that help discriminate source contributions and post-depositional processes
6. Conclusions
This study compares elemental geochemistry, isotope geochemistry and mineralogical and geochronological signatures of aeolian sand in the Sindh and Cholistan regions of the western Thar Desert. Closer affinity to sediment carried during the Early Holocene by the Indus River compared to by its major Himalayan tributaries helped us to determine the provenance of the dune sand and to investigate the processes leading to the growth of the Thar Desert through geological time. Elemental and isotope geochemistry are consistent with petrographic and mineralogical evidence in indicating that sand in the southern Sindh region is largely derived from deflation of the Indus River floodplain and delta, especially during the post-LGM period. Sand in the northern Cholistan region shows greater compositional similarity with Himalayan tributaries. Although this would imply delivery of sediment during the mid-late Holocene as the river evolved to more Himalayan compositions, this is inconsistent with the longer distance between the delta and the desert in Cholistan. The less negative ϵNd values in Sindh compared to Cholistan or the modern river mouth require preferential supply from primitive sources.
The early post-LGM delta has the closest match in ϵNd values to sands in Cholistan, but this may simply reflect the fact that the sediment is coarser, indicating a grain-size control biased in favour of Karakorum sources (Jonell et al. Reference Jonell, Li, Blusztajn, Giosan and Clift2018). The compositional differences between the south and the north can therefore be understood as reflecting deflation of coarser sediment in the south and the concentration of finer Himalaya sediments in the north. This assessment is confirmed by the higher concentration of elements such as Mg and Cr preferentially hosted in mafic rocks, implying that a larger proportion of Sindh sand is supplied from erosion of mafic rocks in the Upper Indus River catchment. Mafic source rocks are widely exposed in these regions, especially in the Kohistan and Karakorum, but are scarce in the Himalayan Belt drained by Punjabi tributaries. The dominant supply from the post-LGM lower Indus River to the Thar Desert suggests that most aeolian sediment transport is linked to southwesterly summer monsoon winds blowing from the delta region inland in recent times.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0016756825100320.
Acknowledgements
MU would like to thank the PhD programme of the Department of Earth and Environmental Sciences, University of Milano-Bicocca, Italy, for supporting his research and the International Union for Quaternary Research (INQUA) for receiving a research grant as an Early Career Scientist. PC thanks the Charles T. McCord Jr chair in petroleum geology at LSU for support during this work. We are also thankful to Dr Saif Ur Rehman for help during fieldwork. This article is an outcome of project MIUR – Dipartimenti di Eccellenza 2023–2027, Department of Earth and Environmental Sciences, University of Milano-Bicocca, Italy.
Open access
