In-Situ Silt Generation in the Taklimakan Desert Evidenced by Uranium Isotopes

mechanism of is critical for evaluating the global cycles of nutrient elements and for interpreting paleorecords derived from aeolian deposits. Here, we use a novel geochemical tracer, the 234 U/ 238 U activity ratio that reveals particle comminution age, and Sr-Nd isotopes to investigate how silt-sized particles are produced in the Taklimakan Desert, a major dust source region having effects on global ecosystems. Based on the results from 20- to 25-μm size fractions, we find that approximately 40% of the Taklimakan Desert silts and approximately 30% of the silts in the aeolian dust flux from the area are produced by in-situ desert processes (e.g., abrasion). The silt-sized materials in the Taklimakan Desert are mainly sourced from the eastern Kunlun timespan the the in the the the is a the Loess and a significant amount of the in the and the dust from the which gets transported by the and

produce silts (Nie et al., 2018;Smalley, 1995;Smalley and Vita-Finzi, 1968;Smalley, 1966). In turn, mountainous regions have been proposed as the locus of silt production (L. Li et al., 2018;Lu et al., 2019;J. Sun, 2002) and we refer to the associated silt-producing processes as "mountainous processes." Other studies propose that silts are also produced in arid, lowland settings by nonglacial/nonfluvial processes, especially in desert regions (Lancaster, 2020, and references therein). Specifically, the proposed silt-producing mechanisms in lowland environments include salt weathering, insolation weathering, frost weathering, combined frost, and salt weathering, aeolian abrasion and attrition, abrasion and fracture during fluviatile transport, hydration and crack exploitation by expanding lattice clays, and direct release from siltstones (Muhs, 2013;Pye, 1995;Smalley & Krinsley, 1978;Smith et al., 2002). Those processes can reduce coarse particles to silt sizes, which is important for supplying coarse-silt-sized particles (20-63 μm) to loess deposits (Amit et al., 2014;Crouvi et al., 2010;Lancaster, 2020). Hereafter, we refer to these lowland silt-producing processes as "in-situ desert processes." Separating the locus of silt production is important for evaluating how silts impact the cycling of nutrient elements. Mountain-produced silts experience fluvial transit, during which nutrient elements are often weathered and liberated in dissolved loads. In contrast, desert-produced silts are readily transported through atmospheric circulation, acting to fertilize ecosystems at global scales with minimal loss of nutrient during transport Mahowald et al., 2005). However, the relative contributions of mountains and lowlands to silt production and the aeolian dust flux have remained poorly constrained, especially in mountain-bounded deserts where silts can be produced in both mountains and deserts (Smalley, 1995).

Silt Production in Northwest China and Transport Histories
The geomorphic systems in Northwest China, including the Taklimakan Desert (TD), the Chinese Loess Plateau (CLP), and the adjacent central Asian mountain, provide a unique setting to investigate silt production and transport in arid environments ( Figure 1). Previous studies have found that both in-situ and mountainous processes can produce silts and supply aeolian dust (Amit et al., 2014;X. Chen et al., 2020;Jiang & Yang, 2019;Rao et al., 2015;Rittner et al., 2016;Yang et al., 2007;Zhu et al., 1981), but the relative proportions of silts produced from the two processes remains unclear. The TD silts are mainly sourced from the surrounding mountains, meaning the silts produced in mountains and deserts have the same petrological origin and geochemical-mineralogical compositions. This makes it challenging to distinguish silts produced from different mechanisms using conventional geochemical tracers (e.g., major and trace elements, Sr-Nd isotopes, mineralogy, and detrital-zircon U-Pb ages) (Jiang & Yang, 2019;Rao et al., 2015;Rittner et al., 2016). However, the silts produced by in-situ and mountainous processes have different life histories, providing a new clue for tracing silt production. For example, the silts produced by in-situ desert processes are readily transported by near-surface winds to the Kunlun Shan mountains and by the westerly jet stream to further areas (J. Sun, 2002). Modern meteorological observations show that dust storms in the TD can easily entrain and lift particles to elevations >5,000 m and transport dust particles to the north-northwest, ultimately reaching 50°N, where the dust particles are transported by the westerly jet stream to the remote Pacific Ocean (Gao & Washington, 2010;Iwasaka et al., 2003;J. Sun, 2002). Considering the high frequency of dust storms in the TD and their high efficiency in mobilizing dust particles (Qian et al., 2002;J. Sun et al., 2001;Uno et al., 2009), the silts produced by in-situ desert processes are expected to have short residence times in the TD since production (e.g., days to years) (Belmaker et al., 2011). In contrast, the silts produced in the mountains experience temporary storage in weathering profiles, down-hillslope and fluvial transport, and temporary storage in alluvial plains, and thus likely have longer residence times (e.g., thousands to tens of thousands of years), especially in the flat landscape with lower erosion rate (Dosseto & Schaller, 2016;L. Li et al., 2018;Repasch et al., 2020). Therefore, a tracer of particle ages is promising for distinguishing silts produced by in-situ versus mountainous processes.  Sun and Liu (2006). The white arrow in the inset shows the upper westerly jet (Fang et al., 2020

of 11
Recent studies show that the activity (product of nuclide concentration and decay constant) ratio between 234 U and 238 U, or ( 234 U/ 238 U), provides a promising tool to distinguish particles with similar petrological origins but different production-transport histories (Aciego et al., 2015;L. Li et al., 2018L. Li et al., , 2019. The ( 234 U/ 238 U) can be converted to a particle comminution age, or the length of time since the particles were separated from the bedrock, based on a theoretical decay model (DePaolo et al., 2006). The ( 234 U/ 238 U) of fine particles (≤50 μm) decreases progressively after separating from the parent material as a result of the accumulative recoil ejection of the 234 U precursor out of the particle surface and the associated α decay of 238 U (details in DePaolo et al., 2006). The U-isotope chronometer starts once the fine particles are generated and is not significantly affected by the mineral compositions (DePaolo et al., 2006) due to the similar initial ( 234 U/ 238 U) values among different minerals (Bosia et al., 2018). In the TD setting (Figure 1), the desert silts produced by the in-situ desert processes likely have a younger comminution age and a higher ( 234 U/ 238 U) value than those produced by the mountainous processes (L. , considering their distinct production-transport histories. In this study, we combined ( 234 U/ 238 U) with Sr-Nd isotopes to investigate silt production in the TD, the CLP, and the central Asian mountain ranges, to better understand the aeolian dust cycle in the arid region of Northwest China. We measured ( 234 U/ 238 U) and Sr-Nd isotopes of 20-25-μm fractions on a comprehensive set of samples collected from the TD and the surrounding mountainous catchments. The 20-to 25-μm fraction was chosen for four reasons. First, converting ( 234 U/ 238 U) to a comminution age requires knowing the recoil factor f, which is a function of particle size (DePaolo et al., 2006). Thus, targeting a fixed size fraction (e.g., 20-25 μm) allows to better constrain f. Second, the 20-to 25-μm fraction is relatively fine and can be readily transported by wind over long distances (D. Sun et al., 2003), integrating provenance information over broad spatial scales. Third, this size fraction allows both precise measurement and practical separation (through sieving), although smaller grains would yield a more significant ( 234 U/ 238 U) depletion signal but are more difficult to isolate (L. Li, Liu, et al., 2017). Fourth, the 20-to 25-μm fraction represents a typical size class for the Tarim Basin sediment and Kunlun Shan loess (Honda & Shimizu, 1998;Zan et al., 2015). We used the results from this 20-to 25-μm fraction to infer the behavior of silt-sized particles in the region. We compared the measured U-Sr-Nd isotopes among the samples to identify the source of the TD silts and conducted end-member mixing calculations to quantify the contributions of the in-situ desert and mountainous processes to produce silts. Lastly, we compared the geochemical data set of the TD samples to those from the CLP to evaluate whether the TD could be a potential material source of the CLP.

Study Area
The deserts in Northwest China and the adjacent Tibetan Plateau are the main source region of Asian dust, a major component in the global dust system (Dong et al., 2016;Shao et al., 2011;Uno et al., 2009;W. Zhang et al., 2016;Zheng et al., 2015). The TD is located in the Tarim Basin of Northwest China, and is bounded by a series of mountain belts: the Tian Shan to the north, the Pamir mountains to the west, and the Kunlun Shan and Altun Shan to the south ( Figure 1). The sediments from the Tarim Basin have been well studied for mineral composition, which is dominated by quartz and feldspar minerals and shows limited spatial variations (Honda & Shimizu, 1998;Rittner et al., 2016;J. Sun, 2002). The TD-regional aeolian transport system is characterized by the prevailing near-surface winds that blow from the north in the northern part and from the east in the central and southern parts of the TD (Figure 1) (J. Sun & Liu, 2006). The TD and the CLP are both located in the path of a high-level westerly jet with the CLP downwind of the TD (Figure 1). Such geographic configurations make the TD a possible material source for the CLP (Liu et al., 1994;Meng et al., 2019), which remains to be better understood.

Samples and Methods
A total of 40 surface dune and fluvial sediment samples were collected within the Tarim Basin, including 13 TD desert sand samples and 27 fluvial sediment samples from active channels draining the adjacent mountain ranges (9, 4, and 14 from fluvial channels originated from the Tian Shan, the Pamir mountains, and the Kunlun Shan, respectively) ( Figure 1 and Table 1). Desert samples were taken from the dune crests and fluvial sediment samples were collected from recently deposited riverbed materials in active channels on the piedmont alluvial fans and in mountainous river catchments. We found aeolian deposits (loess) on the Kunlun Shan (downwind from the TD) and collected six loess samples at depths of 5-10 cm from the surface at an elevation of >2,300 m, with one surface sample from a 671-m drill core (Fang et al., 2020). Studies of the wind current circulation, dust storm tracks and deposition, and the geochemical/geological characteristics of dust and loess show that the prevailing near-surface winds can carry silts from the TD to the northern slope of the Kunlun Shan (Bai & Xu, 1988), forming a continuous loess deposit (Fang et al., 2020;Zan et al., 2010). Based on these findings, we considered the Kunlun Shan loess as a first-order representation of the silt-sized materials in the Tarim Basin. All sampling sites where we collected the dune, fluvial sediment, and loess samples were far away from anthropogenic features such as settlements and infrastructures.
To quantify the contribution of the in-situ desert processes to silt production, the ( 234 U/ 238 U) ratios of freshly produced silts need to be constrained. Theory predicts this ratio to be one, but other studies have observed non-one values (DePaolo et al., 2012;Martin et al., 2019). To better constrain the ( 234 U/ 238 U) of fresh silts, we sieved out particles coarser than 100 μm from four sand samples located in the central TD and ground the >100-μm particles to silt sizes using a clean tungsten carbide mortar and pestle. We treated the ( 234 U/ 238 U) of those >100 μm particles as that of the parent rock for the desert silts because (a) the depletion in the ( 234 U/ 238 U) of >100 μm particles caused by recoil can be neglected (DePaolo et al., 2012), and (b) directly sampling the parent rock of the TD silts was difficult in the field. Grinding the samples helped to constrain whether abrasion fractionated ( 234 U/ 238 U).
Next, we used electroformed sieves to accurately separate the 20-to 25-μm size fraction from all samples, including those collected from the TD, the adjacent mountains, and those ground in the laboratory. The sieved 20-to 25-μm fractions were further cleaned, following the protocol by L. . The ( 234 U/ 238 U) ratio and the Sr-Nd isotopic compositions ( 87 Sr/ 86 Sr and 143 Nd/ 144 Nd) of the 20-25-μm fractions were measured using an MC-ICP-MS (Neptune Plus, Thermo-Fisher Scientific) at Nanjing University following established protocols (L. Li, Liu, et al., 2017;W. Zhang et al., 2016). For convenience, the 143 Nd/ 144 Nd ratios are reported in the epsilon notation (ε Nd ), where ε Nd = ( 143 Nd/ 144 Nd sample / 143 Nd/ 144 Nd sample − 1) × 10,000, CHUR stands for Chondritic Uniform Reservoir with 143 Nd/ 144 Nd ratio of 0.512638 (Jacobsen & Wasserburg, 1980). We also measured the U concentrations of the 20-to 25-μm fractions of the TD and mountainous fluvial sediment samples using a quadrupole ICP-MS (Agilent 7900) at Nanjing University.

The Provenance of the Taklimakan Desert Silts
The ( 234 U/ 238 U) and Sr-Nd isotopic compositions between the 20-and 25-μm fractions of the samples from different locations show clear differences (Figures 2 and 3). Based on the Sr-Nd isotopic compositions of the 20and 25-μm fractions, we separate the Kunlun Shan data into the eastern Kunlun Shan group (low 87 Sr/ 86 Sr and high ε Nd ) and the western group (high 87 Sr/ 86 Sr and low ε Nd ) (Figure 2a). The Sr-Nd isotopic compositions of the 20-to 25-μm fractions from the TD samples are similar to those of the eastern Kunlun Shan, suggesting a major    Although the eastern Kunlun Shan is located downwind of the TD (Figure 1), two lines of evidence suggest that the TD contributes a minimal amount of sediment to the Kunlun Shan fluvial sediments through aeolian transport. First, the carbonate grains are highly angular in the Kunlun Shan fluvial sediments, but are well-rounded in the TD sediments (Rittner et al., 2016), where aeolian-transported carbonate grains are often rounded quickly (Garzanti et al., 2015). Second, the eastern Kunlun Shan fluvial sediments have a broader ( 234 U/ 238 U) distribution than the TD sediments (Figure 3b).

Reservoirs of Silt in the Taklimakan Desert and Adjacent Mountains
Here, we use the measured ( 234 U/ 238 U) data set to unravel the production mechanism of the TD silts. As explained in Section 1.2, we expect (a) the freshly produced silts by the in-situ processes to have ( 234 U/ 238 U) values close to freshly broken particles and young comminution ages, considering their short residence time in the TD, and (b) the mountainous-processes-produced silts to have low ( 234 U/ 238 U) and old comminution ages due to the transient storage in sediment routing systems.
To delineate silt cycling in our study area, we consider four material reservoirs (Figure 4): (a) fresh silts produced from the in-situ desert processes; (b) aged silts produced in the mountains; (c) silts deposited in the TD; and (d) silts entrained in aeolian dust. In theory, the silts deposited in the TD (c) and entrained in the aeolian dust (d) are both the mixtures of the TD desert-produced (a) and mountain-derived silts (b). Next, we explain how we approximate these reservoir using our collected samples.
Reservoir (1) is approximated by the ground particles obtained from our laboratory experiment, since it difficult to collect fresh silt particles in the field. Reservoirs (2) and (3) are represented by the fluvial bed sediment samples from the eastern Kunlun Shan and the TD-desert silt samples, respectively. Reservoir (4) is approximated by the Kunlun Shan loess samples, since it is difficult to collect dust particles under active aeolian transport. Note that the silt-sized materials in the Kunlun Shan loess represent the average composition of the same-sized materials in the Tarim Basin (including the TD and the piedmonts of the surrounding mountains) as evidenced by Sr-Nd isotopes (Figure 2a). These approximations provide the basis for quantifying the contributions of silts produced by in-situ versus mountainous processes to the silts deposited in the TD and entrained in aeolian dust.

U Isotopic Mass Balance and Silt Production Mechanisms
In this section, we first quantify the mass balance of the U element using U-isotope-based end-member mixing calculations and then resolve the contributions to the mass of silt-sized materials using the measured U concentrations. Note that our mass balance calculations are based on the results measured for the 20-to 25-μm fractions of the collected samples, and we use these results to infer the behavior of silt-sized particles in the study region.
Considering the mass balance of U in the four reservoirs, we establish end-member mixing equations using ( 234 U/ 238 U): where the subscriptions DP, MP, TD, and KSL represent silts produced by in-situ desert processes (reservoir 1), silts produced by mountainous processes (reservoir 2), the TD silts (reservoirs 3), and the Kunlun Shan silt-sized loess (reservoirs 4), respectively. k DP-TD and k DP-KSL denote the fractions of U from the silts produced by the in-situ desert processes (reservoir 1) in the TD silts and Kunlun Shan silt-sized loess, respectively.
To resolve Equations 1 and 2, we use the measured ( 234 U/ 238 U) of the 20to 25-μm fractions of the collected TD silt (0.991 ± 0.001, mean ± 1 SE hereafter), the Kunlun Shan loess (0.988 ± 0.001), and fluvial sediment samples from eastern Kunlun Shan (0.978 ± 0.004) for ( 234 U/ 238 U) TD , ( 234 U/ 238 U) KSL , and ( 234 U/ 238 U) MP , respectively ( Figure 3b). The 20-to 25-μm fraction of the ground particles and the >100 μm coarse particles have almost identical ( 234 U/ 238 U) values (1.012 ± 0.001, n = 4, and 1.010 ± 0.002, n = 4, for the 20-to 25-μm and >100-μm samples, respectively, Table 2), suggesting that abrasion would not fractionate ( 234 U/ 238 U) and the ( 234 U/ 238 U) of fresh silt particles produced by in-situ desert processes in the filed should be consistent with that of parent rock. The ( 234 U/ 238 U) DP is chosen as 1.012 ± 0.001 from the measurements of the 20-to 25-μm fractions of the laboratory-ground TD grains, instead of the theoretical value of one. The greater-than-one ( 234 U/ 238 U) values have been reported in the ( 234 U/ 238 U) data sets from unweathered rock, glacial outwash, marine sediments and fluvial sediments, with plausible explanations of influences from authigenic phases, incongruent weathering, and recoil from the U-rich phase (DePaolo et al., 2012;Handley et al., 2013;C. Li et al., 2016;Martin et al., 2019;Thollon et al., 2020). Further work is required to unravel those unusually high ratios, but this does not affect our interpretation that high ( 234 U/ 238 U) values correspond to young comminution ages. Notably, according to the U-isotope comminution age model, the greater-than-one ( 234 U/ 238 U) of freshly broken particles will lead to older comminution ages, but will not change the relative differences in comminution ages between samples (see Text A1 in Supporting Information S1). With these considerations, we resolve Equations 1 and 2 and obtain k DP-TD and k DP-KSL as approximately 37% and 27%, respectively.
We then quantified the mass contributions of the silt-sized materials for the in-situ and mountainous processes from k DP-TD and k DP-KSL . Considering the TD silts (reservoir 3), approximately 37% (k DP-TD ) of U in the reservoir comes from the silts produced by the in-situ desert processes, and the remaining approximately 63% (1 − k DP-TD ) U comes from the silts produced by mountainous processes. To obtain the relative contributions of the silt mass from the in-situ and mountainous processes, we need to normalize k DP-TD and 1 − k DP-TD by the U concentrations of reservoirs (1) and (2), respectively (Text A2). We have measured the U concentrations of the 20-to 25-μm fractions of the mountainous fluvial (reservoir 2) and TD (reservoir 3) sediment samples (3.20 ± 0.37 ppm, n = 7 and 3.34 ± 0.20 ppm, n = 5, respectively, Table 3). The U concentrations of the two groups are statistically indistinguishable as evidenced by the Mann-Whitney U test (p = 0.3939). Whereas reservoir (3) represents a mixture of reservoirs (1) and (2) and the U concentrations of reservoirs (2) and (3) are the same, mass conservation requires that the U concentration of reservoir (1) must be the same as reservoirs (2) (Text A2 in Supporting Information S1). Note the uniform U concentrations among reservoirs (1)-(3) is expected due to the same petrological origin (the eastern Kunlun Shan) (Jiang & Yang, 2019;Rittner et al., 2016). The equivalent U concentrations of reservoirs (1) and (2) mean that k DP-TD and 1 − k DP-TD equate to the mass contributions of silts, and the same   Table 3 The U Concentration of the 20-25 μm of the Mountainous Fluvial Sediment Samples and the TD Silt Samples for k DP-KSL and 1 − k DP-KSL . Thus, the in-situ desert processes contribute approximately 37% (k DP-TD ) and 27% (k DP-KSL ) to the TD silts and the Kunlun Shan silt-sized loess, respectively.
An earlier study found that the ( 234 U/ 238 U) of the Alxa arid lands (including Badain Jaran Desert and Tengger Desert), which is the main material source region of the loess deposits on CLP, is close to that of the piedmont fluvial sediment derived from the surrounding mountains, suggesting the limited contribution of in-situ silt production to the silt fraction in those deserts (L. Li et al., 2018). So, why can the TD produce a significant amount of silts through the in-situ desert processes? We speculate that climatic factors may play a primary role here. The TD is characterized by an extremely arid climate (mean annual precipitation <100 mm), large diurnal temperature variations (over 40°C), and frequent, high-magnitude dust storms (Qian et al., 2002). These factors can promote the exfoliation and dismantling of bedrock and coarser particles, thus facilitating the in-situ desert processes (Glennie, 1987). Future studies are needed to better understand the influences of those factors and investigate the contribution of different silt production mechanisms in other deserts all over the world, such as Sahara and Negev desert.

Implication for the Provenance of the Chinese Loess Plateau
Several studies have argued whether the TD provides a major material source to the CLP (J. Chen et al., 2007;Liu et al., 1994;Meng et al., 2019;Rittner et al., 2016). Comparing our data set from this work to a comprehensive U-Nd isotopic data set of the CLP and adjacent arid lands (L. Li et al., 2018), we find significant differences in the U-Nd isotopic compositions between the CLP and TD sediments, which does not support the direct supply of the TD silts to CLP by the high-level westerly jet (Figure 3a). The 20-to 25-μm fractions of the samples from the Alxa arid lands along the direct route from the TD to the CLP also show significantly different isotopic compositions from the same size fractions of the TD sediment samples (Figure 3a). These differences weaken the possibility of transporting the TD silts to the CLP through the Hexi Corridor. Thus, we postulate the high-level westerly jet that carries the TD silts has a minor influence in delivering dust to the CLP, which is consistent with the provenance studies that are based on other integrated multiproxy approaches (e.g., detrital-zircon U-Pb ages) (Bird et al., 2020;Rittner et al., 2016;H. Zhang et al., 2021).

Conclusions
In this work, we measured the uranium comminution ages and Sr-Nd isotopes of the 20-25-μm fractions of the sediment and loess samples from the Taklimakan Desert and adjacent mountains in Northwest China to investigate the sources and production mechanisms of the silt-sized particles in the region. Based on our results, we conclude that: 1. the silts in the TD are mainly sourced from the eastern Kunlun Shan and partly from the Pamir mountains; 2. both in-situ and mountainous processes contribute to silt production in the TD. The efficacy of silt production by in-situ desert processes, however, is significantly higher in the TD than other deserts in Northwest China; and 3. the silts in the TD are not a major material source of the CLP.
Overall, our study provides new insights into silt production in arid environments in Northwest China by demonstrating the importance of in-situ desert processes in breaking down coarse particles. Our findings hold far-reaching implications for the global dust cycle, the material sources of the CLP, and the reconstruction of past environmental changes from the loess archive.