Licht et al. Eolian cannibalism: Reworked loess and fluvial sediment as the main sources of the Chinese Loess Plateau A. Licht1,2,3,†, A. Pullen1,4, P. Kapp1, J. Abell1, and N. Giesler1 1Department of Geosciences, University of Arizona, Tucson, Arizona 85721, USA 2Biodiversity Institute, University of Kansas, Lawrence, Kansas 66045, USA 3Institut für Erd- und Umweltwissenschaften, Potsdam Universität, 14476 Potsdam, Germany 4Department of Earth and Environmental Sciences, University of Rochester, Rochester, New York 14627, USA ABSTRACT propose that sediment reworking of Yellow 2009; Vandenberghe, 2013). In the modern inter- River sediment and older loess deposits by glacial climate, dust storms originate north and The loess and paleosol sequences of the wind on the Chinese Loess Plateau homoge- northwest of the Chinese Loess Plateau, in the Chinese Loess Plateau are composed of Qua- nized the eolian zircon populations toward Gobi Desert, commonly during the breakdown ternary dust, the origin of which has been a glacial provenance due to higher (2–20 of the Siberian High in Spring (Roe, 2009), and the subject of considerable debate. Some times) dust accumulation rates during gla- they track across the Badan Jaran, Tengger, and recent U-Pb geochronological studies of eo- cials. These findings suggest that the Chinese Mu Us Deserts (Wang et al., 2004). These des- lian zircons have proposed the existence of Loess Plateau has evolved as a more dynamic ert areas are thought to be the main source areas two major wind pathways: from the north landform than previous thought, where wind for the coarse fraction of the modern dust to the and northwest, through the Badan Jaran, deflation, fluvial input, lateral transport, and plateau, though the exact locus of dust genera- Tengger, and Mu Us Deserts during inter- accumulation of sediment are equally impor- tion in these regions—piedmont alluvial fans, glacials, and from the west, through the tant. These internal reworking effects would dried lake basins, or sand dune fields—is widely Qaidam Basin during glacials. Others have then significantly bias the paleoclimatic inter- debated (Pye, 1995; Sun, 2002; Amit et al., emphasized the importance of Yellow River pretations based on eolian dust properties of 2014). Additionally, recent provenance studies sediment supply in the Chinese Loess Plateau the Chinese Loess Plateau. of loess have also highlighted the importance sediment budget. However, tracking dust of reworking of local substratum rocks (mostly source regions through U-Pb dating in a sta- INTRODUCTION pre-Quaternary terrestrial sediment) and of flu- tistically robust manner is particularly com- vial detritus brought by the Yellow River, with plex given the similar age peaks in the age The Chinese Loess Plateau (Fig. 1) is mainly headwaters in northeastern Tibet (Fig. 1), in the probability distributions of potential source composed of eolian dust deposits that are pre- loess sedimentary budget (Stevens et al., 2013; regions in Central Asia. This paper pre sents dominantly brought by spring storms sweeping Che and Li, 2013; Bird et al., 2015; Nie et al., 2410 new U-Pb ages of detrital zircons from clastic material from the deserts of the Asian 2015; Kapp et al., 2015). wind-eroded strata, Quaternary eolian de- interior (Roe, 2009). Though the oldest loess The geographic origin of pre-Holocene gla- posits, and modern river sands in central deposits within the southernmost and western- cial loess is more controversial. Wind-eroded China in order to increase the robustness and most Chinese Loess Plateau have been dated to landforms in the Qaidam Basin (Kapp et al., the spatial resolution of zircon age distribu- 8 Ma (An et al., 2001), 22–25 Ma (Guo et al., 2011; Rohrmann et al., 2013), U-Pb geochro- tions in dust source regions. We then propose 2002; Qiang et al., 2011), and 41 Ma (Licht nology of eolian zircons (Pullen et al., 2011), a new mixture modeling technique to statisti- et al., 2014), most of the Chinese Loess Pla- and grain-size data and climate simulations cally address the contribution of these differ- teau is composed of Quaternary loess-paleosol (Vandenberghe et al., 2006) have suggested the ent sources to the Chinese Loess Plateau sedi- sequences. Loess layers were deposited during existence of a different dust-storm track during mentary budget. Our contribution estimates glacial periods under a colder and drier Asian glacial periods. These observations point to the indicate that eolian supply to the Chinese climate, whereas paleosol layers developed dur- loci of source areas located south of the Qilian Loess Plateau is dominated (60%–70%) by ing warmer and moister interglacial periods with Shan mountain ranges, in the Qaidam Basin reworking of Yellow River sediment. More- enhanced summer monsoons (An et al., 1990; and northern Tibetan Plateau (Fig. 1). This arid over, evidence of Qaidam Basin–sourced zir- Porter and An, 1995; Kang et al., 2011, 2013). region would have been deflated by surface cons (15%–20%) in both loess (glacial) and Loess deposits consist of a coarse (12–70 mm) westerly winds, in response to either enhanced paleosol (interglacial) layers corroborates the primary fraction, transported via saltation by penetration of Atlantic westerlies during glacial existence of an erosive wind pathway through dust storms in near-surface suspension clouds, periods (Vandenberghe et al., 2006), and/or to the Qaidam Basin during glacials and implies and a fine (1–3 mm) secondary fraction, the ori- the shift of the whole midlatitude atmospheric that a substantial portion of the interglacial gin of which is still controversial, either pedo- system by ~10° of latitude equatorward (Togg- dust is recycled from older glacial loess. We genic, supplied by upper-level airflow, and/or weiler and Russell, 2008; Kapp et al., 2011). In by adherence to the coarse fraction (Derbyshire deeper time, this westerly wind pathway would †licht@ uw .edu et al., 1998; Sun et al., 2006; Stevens and Lu, have been favored by a lower-elevation north- GSA Bulletin; May/June 2016; v. 128; no. 5/6; p. 944–956; doi: 10.1130/B31375.1; 9 figures; 2 tables; Data Repository item 2016026; published online 8 January 2016. 944 GeologicalFor permission Society to of copy, America contact [email protected] Bulletin, v. 128, no. 5/6 © 2016 Geological Society of America Reworked loess and fluvial sediment as the main sources of the Chinese Loess Plateau Alta i Figure 1. Schematic map of Central Asia, showing the modern, simplified dust-storm T Gobi 40° iansha 1 n Desert 30°E tracks and near-surface wind patterns (red arrows), through- 120° Taklimakan Desert C out the (A) Mu Us, (B) Tengger, Pamir gh and (C) Badan Jaran Deserts, n Ta Qilia B A Altu n S and the hypothetical “glacial” han iver Qaidam dust-storm track (blue arrows), Kunlun w R llo Basin Shan e Y through the Qaidam Basin and QinlingShan across northern and central 0°N 3 Tibet (Pullen et al., 2011). Chinese LoessPlateau Tibet Sand desert Modern dust-storm track Glacial dust-storm track 500 km ern Tibetan Plateau until the late Neogene–early et al., 2011; Lease et al., 2007, 2012; Weislogel Plateau, we sampled modern eolian sand dunes Quaternary (Sun et al., 2008; Nie et al., 2014). et al., 2010). However, the number (n) of U-Pb (one sample in the Mu Us Desert, four in the However, the relative contributions from the ages per published loess sample (commonly n < Tengger Desert, and two in the Badan Jaran Qaidam Basin compared with that of the more 150) is so far too low to accurately constrain Desert). We also sampled the tops of yardangs proximal desert regions and the Yellow River the relative contribution of each age group (see (three samples in the Mu Us Desert and four in to the north in the eolian sedimentary budget next section; e.g., Andersen, 2005; Pullen et al., the Badan Jaran Desert), streamlined landforms are poorly constrained, given the complexity of 2014). Some workers have been using the multi- sculpted by windblown sand and composed of the provenance problem with multiple potential dimensional scaling (MDS) statistical technique Pliocene–Quaternary to Cretaceous sedimen- source regions. of Vermeesch (2013), based on the less n-depen- tary bedrock. In the western, deflationary, and Single-grain U-Pb dating of detrital zircons dent Kolmogorov-Smirnov (KS) statistic, as a bedrock-floored Qaidam Basin, we sampled is an efficient and effective technique to dif- visualization tool to qualitatively estimate the Miocene–Pliocene strata that have been sculpted ferentiate the contribution of discrete sources similarity (or dissimilarity) between loess sam- into yardangs (three samples, which were previ- in sediment samples (Gehrels, 2000). U-Pb ples and potential sources (Stevens et al., 2013; ously assessed using cosmogenic nuclides by analy ses of detrital zircon grains produce age Che and Li, 2013; Bird et al., 2015). However, Rohrmann et al., 2013). To represent the aver- distributions that reflect the ages of zircons in this approach does not provide quantitative esti- age influx of sediment from northeast Tibet, the source rocks for the sediment. The pres- mates of the relative contribution of the different we sampled modern fluvial deposits from the ence or absence of a particular age group in a dust source regions. Yellow River at the western edge of the Mu Us sample is used as a test to determine if a par- This paper presents new U-Pb ages of detrital Desert (one sample). We also determined new ticular source region contributes to the sample zircons from eolian and fluvial deposits as well U-Pb ages on zircons from paleosol layers in the (Gehrels et al., 2011). Recent developments in as from wind-eroded strata in central China, in Heimugou section (layers S0, S1, S9, S15 and laser-ablation techniques allow for U-Pb analy- order to better illuminate the zircon age dis- S22; after Porter, 2001) of the central Chinese ses of small (<40 µm) zircons in eolian dust tributions in potential dust source regions.