phase of geomorphic quiescence (51–39ka)asinferredfromacalcretisedpalaeosol horizon (Oldknow and Hooke, 2017;Oldknow et al., 2020).Well-preserved exposures of this sequence provide an opportunity toinvestigate the allogenic and autogenic controls that may have facilitatedlong-term landscape stability during this interstadial period, and theirimplications for subsequent landscape development.The specific aims of this paper are to: 1) investigate thepedosedimentary architecture of the MIS 3 calcretized palaeosol;2) infer the primary controls on pedogenesis, embedding results ina discussion of regional geomorphic processes and their potentialrelationship to climate; and 3) assess the impact of pedosedimentaryprocesses during MIS 3 on subsequent landscape development.2. Site contextThe Wilgerbosch catchment includes several bedrock-controlledpartly confined (Fryirs and Brierley, 2013) low-order tributaries(Fig. 1B) which comprise the headwaters of the Sundays River(Fig. 1A). The Sundays River drains the Sneeuberg,flowing southeastfor ~650 km to the Indian Ocean. The bedrock lithology of theSneeuberg is dominated by sedimentary rocks belonging to thePermian/Triassic Karoo Supergroup (Turner, 1978), which areintrudedbyDrakensbergGroupdoleritesillsanddykes(Neumannet al., 2011). The dolerites have been classified as subalkalinetholeiitic basalts (Cox et al., 1967;Neumann et al., 2011). Duringthe Quaternary, South Africa has been largely tectonically stable,except for localised uplift (Andreoli et al., 1996).2.1. Africanders Kloof study siteThefindings presented here are drawn principally from detailedpedosedimentary investigations conducted on a calcretized abandonedfloodplain preserved within the widest reaches of Africanders Kloof, theGanora gorge (Figs. 1, 2D–F,3) and Wilgerbosch Kloof in the north(Oldknow and Hooke, 2017).Situated at or just above bedrock, the modern channels are‘perennial’withflow sustained by aquifer-fed groundwater dis-charge. The dolerite imposes strong lithological controls on channelbehaviour resulting in structurally controlled straight reaches,abrupt meander bends (up to 90o), knickpoints at various degreesof incision and riffle-pool sequences. Present-day average annualrainfall is ~423 mm, concentrated in the late summer/early autumn.Seasonal temperaturesfluctuate from 30 °C in summer to below−10 °Cin winter (Schultz, 1980). The vegetation of the study area is charac-terized by‘Eastern Upper Nama-Karoo (NKu2)’on gently slopinghills, and‘Upper Karoo Hardeveld (NKu4)’on steeper sandstoneslopes and dolerite ridges (Mucina et al., 2006). Modernfloodplainsare colonized by shrubs (e.g.Acacia karoo) and grasses, butPhragmitesaustralisdominates channel bars andfloodplains in zones offlowconvergence.Deep-channel incision and donga formation following the 19thcentury European occupation (Boardman, 2014) has resulted inflood-plain abandonment, the desiccation of former late Holocene valley-floor wetlands (Fig. 2A, B), and exposure of extensive alluvial cut andfill sequences which contain palaeosols and calcretes.Oldknow andHooke (2017)proposed that the patterns of cut andfill in the higherorder channels were controlled by basin-scale changes to structuralconnectivity, and that gaps within the alluvial record were the resultof varied preservation, rather than autogenic processes which producepocket aggradation (Fryirs and Brierley, 2013). Notable exceptionsinclude the catchment headwaters, where a geological barrier acted asa local discontinuity on sedimentflux, resulting in terminal channelandfloodout deposition, the deposits of which are diachronous withthose preserved in the rest of the system (Oldknow et al., 2020).Recent luminescence dating has shown that the cut andfill succes-sion dates back to MIS 3 (Oldknow et al., 2020). Of key interest fromthis new chronology is an apparently prolonged sedimentary hiatusfrom 51 to 39 ka, which was associated with palaeosol formation andcalcium carbonate (CaCO3,calcrete) accumulation within thefloodplainsediments. Calcretes formed infloodplain settings have been reportedand analysed elsewhere in southern Africa, with formation typicallyattributed to climatically-driven changes to groundwater (Netterberg,1969, 1978;Gobagoba et al., 2005). Prior to the present study, thecontrols on the pedosedimentary architecture of this MIS 3 calcretizedalluvial sequence at Wilgerbosch, its palaeoenvironmental significanceand implications for long-term landscape development were notestablished.3. Methods3.1. Pedosedimentary analysesThe sedimentology and stratigraphy of the alluvial succession atWilgerbosch was reported in detail byOldknow and Hooke (2017).The alluvial succession and chronology are portrayed at transect 1(Figs. 1–3). We sampled profile AK12 at 10 cm intervals (n = 31samples), and collected a block of the calcrete and overlying red palaeosol.Thin sections of the calcrete and overlying red palaeosol hori-zon at profile AK12 were made at the University of Stirling (fulldetails provided athttp://www.thin.stir.ac.uk/2008/06/03/methods-impregnation/). Micromorphological analyses were car-ried out using optical microscopy, cathodoluminescence micros-copy and scanning-electron microscopy (Hitachi TM4000benchtop SEM) at Royal Holloway, University of London, and theUniversity of Liverpool (Fig. 4).The 31 samples were subjected to grain size (Section 3.2),geochemical (Sections 3.3–3.4) and mineral magnetic analyses(Section 3.5). Dry sieving isolated the 0–63μm fractions for mineralmagnetic and geochemical analyses. 11 previously collected samplesfrom the wider catchment (including AK12) were recommissioned forAtterberg grain size separation to assess whether particle-size distinc-tions in the geochemical and magnetic properties are masked by thebulk (0–63μm) measurements (Hao et al., 2008).3.2. Laser diffraction particle size analysisOrganic residue was removed using 10% v/v H2O2and sodiumhexmetaphosphate (NaPO3)6applied to disperse silts and clays. Grainsize analysis was conducted using a Beckman Coulter LS 13320 laserdiffraction particle size analyser, equipped with an aqueous liquidmodule (ALM) and a Polarization Intensity Differential Scatter (PIDS)unit capable of detecting sub-micron clay.3.3. Mineralogical analysis by XRDXRD analyses were conducted on two samples XRD-1 and XRD-2(Fig. 5), to assess the clay mineralogy of the soils below and above thecalcrete and constrain interpretation of the magnetic properties(Section 3.5).3.4. Elemental analysis by XRFXRF analyses were undertaken to assess the dependence of thesediment magnetic properties on mafic element concentrations (e.g. Fe, Mg, Mn and Ni). Measurements were carried out using a BrukerS2 Ranger energy-dispersive X-rayfluorescence analyser, using a PdX-ray tube and Peltier-cooled silicon drift detector. The instrumentwas run under three different measurement conditions (20, 40 and50 keV tube excitement) on loose powder (1–3 g) under helium.Powder cups were prepared with spectroscopic grade 6μm polypro-pylenefilm (Chemplex Cat. No. 425). Calibration used a set of up to18 certified reference materials.2C.J. Oldknow et al. / Sedimentary Geology 405 (2020) 105698