Laser diffraction size analysis of loess-paleosol sequences– pretreatment, calculation, interpretation

Von der Fakultät für Georessourcen und Materialtechnik der Rheinisch -Westfälischen Technischen Hochschule Aachen

zur Erlangung des akademischen Grades eines

Doktors der Naturwissenschaften

genehmigte Dissertation

vorgelegt von

M.Sc. Philipp Schulte

aus Lennestadt

Berichter: Univ.-Prof. Dr. rer. nat. Frank Lehmkuhl Prof. Dr. rer. nat. Bernhard Diekmann

Tag der mündlichen Prüfung: 18.04.2017

Diese Dissertation ist auf den Internetseiten der Hochschulbibliothek online verfügbar

II

Zusammenfassung

Die Korngrößenverteilung ist eine der wesentlichsten Eigenschaften von Korngemischen und wird daher seit Jahrhunderten in vielen technischen und wissenschaftlichen Bereichen mit unterschiedlichen Methoden analysiert. In der Industrie werden Korngrößenverteilungen von zum Beispiel Baustoffen, Farbpigmenten, Lebensmitteln oder Medikamenten bestimmt, um die Bestandteile und Eigenschaften der Produkte besser zu verstehen und die Produktionsprozesse zu optimieren (Produkt und Prozessoptimierung) sowie eine gleichbleibende Qualität zu sichern (Compliance).

In den Geowissenschaften werden Korngrößenanalysen angewendet um Sedimente und Böden zu klassifizieren (Granulometrie). Die granulometrischen Eigenschaften (Form und Größe) erlauben Rückschlüsse auf die Entstehungsgeschichte der Boden- und Sedimentpakete. Die gewonnenen Informationen können als Proxy (indirekter Indikator) zur Rekonstruktion der Klima und Umweltbedingungen genutzt werden, welche zur Zeit der Entstehung geherrscht haben. Das Wissen über die känozoische und vor allem über die quartäre Umweltentwicklung hilft uns, aktuelle Prozesse und zukünftige Veränderungen besser einschätzen zu können.

Korngrößenanalysen können mit diversen Methoden (Mikroskopie, Siebanalyse, Sedimentationsverfahren, Laser-Streulicht-Analyse) an unterschiedlichen Archiven (lakustrine, marine, (peri-)glaziale, äolische und kolluviale Sedimente) durchgeführt werden. Aktuell werden häufig Löss-Paläoboden-Sequenzen untersucht, da diese große Bereiche der Erdoberfläche bedecken und teilweise weit in der Zeit zurückreichen. Eine hochaufgelöste Bestimmung der Korngrößenverteilung solcher Sequenzen wurde erst durch die Entwicklung der Laser-Streulicht-Analyse ermöglicht. Obwohl dieses Verfahren deutlich zeit- und kosteneffektiver ist als alle verfügbaren Alternativen, ist es insbesondere in der Bodenkunde bis heute umstritten und wird daher nicht standartmäßig angewendet.

Die vorliegende Dissertation liefert einen Beitrag zum Verständnis der Einschränkungen der Laser- Streulicht-Analyse bezüglich der Erfassung und Interpretation von Korngrößendaten und liefert Ansätze, diesen zu begegnen. Dabei liegt ein besonderer Fokus auf der Bestimmung der vertikalen Korngrößenvariabilität in Löss-Paläoboden-Sequenzen sowie die Etablierung derselben als Proxy zur Rekonstruktion vergangener Klima und Umweltbedingungen.

Im Rahmen einer Literaturübersicht wurden die wichtigsten etablierten Methoden, welche zur Korngrößenbestimmung von Lösssedimenten eingesetzt werden können, zusammengefasst um im Hinblick auf ihre Eignung bewertet. Eine detaillierte Validierung der Laser-Streulicht-Analyse wurde im Rahmen der einzelnen bereits publizierten wissenschaftlichen Artikel durchgeführt.

Bei der Untersuchung eines Schlüsselprofils der Gebirgsregion auf dem nordöstlichen Tibet Plateau wurde eine klassische granulometrische Analyse durchgeführt, um die Korngrößenvariabilität innerhalb dieser Löss-Paläoboden-Sequenz zu bestimmen. Basierend auf einer kombinierten Betrachtung der horizontalen (innerhalb der gemessenen Klassen einer Probe) und vertikalen (über alle gemessenen Proben) Variabilität

I

mittels verschiedener etablierter Parameter, konnten Aussagen über holozäne Klima und Umweltveränderungen getroffen werden. Durch die Generalisierung der horizontalen Variabilität in Form eines einzelnen Wertes (z.B. eines Korngrößenverhältnisses) werden geringe aktive Korngrößenänderungen verschleiert. Solche können als Folge leichter klimatischer Veränderungen in schwach frequentierten Korngrößenklassen auftreten und werden durch die relative Veränderungen der hoch frequentierten Klassen maskiert. Dies wurde unter anderem anhand des Vergleichs zweier Fallstudien aus Rumänien und der Ukraine nachgewiesen. Im Rahmen dieser Arbeit wurde die Heatmap als Darstellungsform für hochaufgelöste (große Anzahl an Größenklassen) Korngrößendatensätze von langen und dicht beprobten Sediment Sequenzen vorgeschlagen. Dies ist die einzige Möglichkeit um die Variabilität innerhalb von Löss- Paläoboden-Sequenzen ohne horizontalen oder vertikalen Datenverlust darzustellen.

Mittels einer Versuchsreihe wurde der Einfluss der Probenvorbehandlung, insbesondere der Zugabe von Salzsäure, auf das Ergebnis der Laser-Streulicht-Analyse evaluiert. Es ist festzuhalten, dass mit zunehmender post-depositioneller Veränderung der analysierten Proben die Wirkung der einzelnen Vorbehandlungsschritte undurchsichtiger wird. Besonders die Vorbehandlung mit Salzsäure kann zu missverständlichen Interpretationen führen. Generell ist jede Vorbehandlung eine zum Teil selektive und unergründliche Modifikation der natürlichen Sedimentprobe. Daher sollte die Probenvorbehandlung nur die essentiellsten Schritte umfassen (Oxidation der Organik und Dispergierung).

Nach der eigentlichen Messung wird das detektierte Laser-Streulicht-Muster mithilfe verschiedener optischer Modelle in eine Korngrößenverteilung umgerechnet. Die gängigsten Modelle sind die Fraunhofer Annäherung und das Lorenz-Mie Modell. Obwohl mit letzterem, vor allem für sehr feine Partikel (< 7,8 µm), deutlich validere Ergebnisse berechnet werden können, wird es bis heute nicht standardmäßig angewandt. Der Nachteil dieses Modells besteht darin, dass zur Berechnung der komplexe Brechungsindex, welcher die optischen Eigenschaften der in der Probe enthaltenden Minerale beschreibt, benötigt wird. Da dieser bei polymineralischen Gemischen nicht eindeutig zu bestimmen ist und eine Bestimmung extrem zeit- und kostenintensiv ist, wird in der Regel ein Näherungswert angenommen. Letztendlich liefert keines der beiden Modelle die tatsächliche Größe sehr feiner Partikel. Allerdings konnte nachgewiesen werden, dass die Differenz zweier Korngrößenverteilungen jeweils berechnet mit einem der beiden Modelle von der mineralischen Zusammensetzung der Feinkomponente Probe abhängig ist. Hiervon ausgehend, wurde im Rahmen dieser Arbeit ein neuer Korngrößenparamter, welcher vom Autor als delta grain size distribution (ΔGSD) bezeichnet wird, entwickelt, evaluiert und bereits vielfach zur Charakterisierung von Löss- Paläoboden-Sequenzen angewandt. Der Parameter reagiert sensitiv auf die post-depositionelle Bildung von Tonmineralen als Folge chemischer Verwitterung und kann als Proxy für fossile Bodenbildung und die damit assoziierten Paläoklima- und Paläoumweltbedingungen genutzt werden.

II

Summary

Grain size distribution is one of the most important properties of particle mixtures, and, for centuries, it has been analyzed in various technical and scientific fields using different techniques. In industry, grain size distributions of, for example, building materials, color pigments, foodstuffs, or medicines are identified in order to better understand the ingredients and the properties of products, to optimize the production processes (product and process optimization), and to ensure a consistent quality (compliance). In the geosciences, grain size analysis are used to classify sediments and soils (granulometry). The granulometric properties (degree of roundness, shape, and size) allow conclusions about the genesis of the sediment and soil packages. The information obtained can be used as a proxy (indirect indicator) for the reconstruction of climate and environmental conditions which prevailed at the time of its origin. Knowledge about the Cenozoic Era, especially Quaternary period environmental development, helps us to better assess current processes and possible future changes.

Grain size analysis can be conducted using various methods (microscopy, sieve analysis, sedimentation methods, and laser diffraction size analysis) at different archives (lacustrine, marine, (peri-glacial, and aeolian and colluvial sediments). Currently, loess-paleosol sequences are regularly investigated because they cover large areas of the earth’s surface and some date back far in time. Highly resolved determinations of the grain size distribution of such sequences was enabled by the development of laser diffraction size analysis. Although this technique is significantly more time- and cost-effective than all available alternatives, especially in soil sciences, it remains controversial, and was, therefore, not applied by default until recently.

This study intends to disclose and defuse the constraints of the laser diffraction technique concerning grain size date acquisition, subsequent analysis, and interpretation. Especial focus is on the methodological progress of the investigation of vertical grain size variability within loess- paleosol sequences and its validation as a proxy to reconstruct past climate and environmental conditions.

In the literature review, the most important and well-established methods used for grain size analysis of loess sediments were summarized and assessed with regard to their susceptibility. A detailed validation of laser diffraction size analysis was carried out in the framework of the individual previously published chapters.

For the investigation of a key section of the mountain region on the northeastern Tibet plateau, a classical granulometric analysis was conducted to determine grain size variability within this loess- paleosol-sequence. Based on a combined analysis of the horizontal (within the measured classes of

III

a sample) and vertical (across all measured samples) variability, using different established parameters, conclusions about the Holocene climate and environmental changes were possible. Due to the generalization of horizontal variability in the form of a single value (e.g., a grain size ratio), small active grain size changes are obscured. Such variations can occur as a result of slight climatic changes in low frequented grain size classes and are masked by the relative changes of the highly frequented classes. This was demonstrated, among other things, by the comparison of two case studies, one from Romania and one from Ukraine. In the framework of this study, the visualization of highly resolved (large number of size classes) grain size data, obtained for long and densely sampled sediment sequences in the form of a heatmap signature, has been established. This is the only way to present the variability within loess-paleosol sequences without horizontal or vertical data loss.

In an analytical test series, the influence of pretreatment agents, particularly hydrochloric acid (HCl), on the result of laser diffraction size analysis was evaluated. It should be noted that with increasing post-depositional alteration of the analyzed sample, the effect of the individual steps of pretreatment becomes more inscrutable. Especially, pretreatment with HCl can result in misleading interpretations. In general, each pretreatment is a partially selective and inscrutable modification of the natural sediment sample. Therefore, sample pretreatment should include only the essential steps (oxidation of the organic matter and dispersion).

After the actual measurement, the detected laser scattering pattern is converted into a grain size distribution by means of various optical models. The most common are the Fraunhofer approximation and the Lorenz-Mie models. Although, by means of the latter, it is possible to calculate more valid results, especially for very fine particles (< 7.8 μm), it has not been used by default until recently. The disadvantage of this model is that the complex refractive index, which describes the optical properties of the minerals contained in the sample, is needed for calculation. Since this cannot be determined clearly in polymineral mixtures and a determination is extremely time-consuming and cost-intensive, an estimated value is generally assumed. Finally, neither of the two models delivers the actual size of very fine particles. However, it was demonstrated that the difference between two grain size distributions, each calculated with one of the two models, is dependent on the mineral composition of the fine component of the sample. Based on this, a new grain size parameter, which was designated by the author as the delta grain size distribution (ΔGSD) signature, was developed, evaluated, and used for the characterization of loess-paleosol sequences. The parameter reacts sensitively to the post-depositional formation of clay minerals as a result of chemical weathering and can be used as a proxy for fossil soil formation and the associated paleoclimatic and paleoenvironmental conditions.

IV

Table of contents

ZUSAMMENFASSUNG ...... I

SUMMARY ...... III

TABLE OF CONTENTS ...... V

LIST OF FIGURES ...... VIII

LIST OF TABLES ...... XI

INTRODUCTION ...... 1

1.1. FRAMEWORK AND OBJECTIVES ...... 1

1.2. OUTLINE ...... 2

1.3. ASSOCIATED RESEARCH PROJECTS ...... 4

FUNDAMENTALS OF LOESS‐PALEOSOL‐SEQUENCE INVESTIGATION ...... 6

2.1. LOESS AS AN AEOLIAN SEDIMENT ...... 6

2.2. POST‐DEPOSITIONAL ALTERATION OF LOESS SEDIMENTS ...... 7 2.2.1. Grain size reduction ...... 7 2.2.2. Aggregation and agglomeration of loess sediments ...... 8

2.3. GRAIN SIZE PARAMETERS AND RATIOS FOR LOESS‐PALEOSOL SEQUENCES ...... 9

ESTABLISHED METHODS FOR GRAIN SIZE ANALYSIS OF LOESS ...... 12

3.1. MICROSCOPIC ANALYSIS ...... 13

3.2. SIEVE ANALYSIS ...... 15

3.3. SEDIMENTATION / GRAVITATION ANALYSIS ...... 16 3.3.1. Areometer Method ...... 18 3.3.2. Photometer Method ...... 18 3.3.3. Sedimentometer Method (SediGraph) ...... 19 3.3.4. Pipette analysis ...... 20 3.3.5. Atterberg sedimentation ...... 20 3.3.6. Limitations of gravitational techniques ...... 21

LOESS‐PALEOSOL‐SEQUENCES DEVELOPED UNDER DIFFERENT PALEOENVIRONMENTAL CONDITIONS . 22

MANUSCRIPTS ...... 24

5.1. TIMING AND SPATIAL DISTRIBUTION OF LOESS AND LOESS‐LIKE SEDIMENTS IN THE MOUNTAIN AREAS OF THE

NORTHEASTERN TIBETAN PLATEAU ...... 24 Abstract ...... 24 5.1.1. Introduction ...... 25 5.1.2. Regional setting and distribution of loess and loess‐like sediments ...... 27 V

5.1.3. Material and methods ...... 30 5.1.3.1. Geomorphological mapping, site selection and sampling strategy ...... 30 5.1.3.2. Sedimentological and geochemical methods ...... 30 5.1.3.3. Dating methods ...... 31 5.1.4. Results ...... 33 5.1.4.1. Sediments and grain size distribution ...... 34 5.1.4.2. Geochemical results of SHD section ...... 37 5.1.4.3. Geochronological results ...... 39 5.1.5. Discussion ...... 41 5.1.5.1. Late Glacial to early Holocene (15 ka to 8.5 ka) ...... 41 5.1.5.2. Mid‐Holocene (8.5 to 4.0 ka) ...... 42 5.1.5.3. Late Holocene (4.0 – 0 ka) ...... 43 5.1.6. Conclusion ...... 43

5.2. ENVIRONMENTAL CHANGE INDICATED BY GRAIN SIZE VARIATIONS AND TRACE ELEMENTS: EXAMPLES FROM TWO

DIFFERENT SECTIONS ‐ THE SANDY‐LOESS SEDIMENTS FROM THE DOROSHIVTSY SITE (UKRAINE) AND THE LOESS SECTION SEMLAC

(ROMANIA) ...... 45 Abstract ...... 45 5.2.1. Regional Setting ...... 46 5.2.1.1. Western Romania: the section Semlac ...... 46 5.2.1.2. South‐western Ukraine: the section Doroshivtsy ...... 46 5.2.2. Methods ...... 47 5.2.3. Results ...... 47 5.2.3.1. Semlac ...... 47 5.2.3.2. Doroshivtsy ...... 49 5.2.4. Conclusion ...... 52

5.3. INFLUENCE OF HCL PRETREATMENT AND ORGANO‐MINERAL COMPLEXES ON LASER DIFFRACTION MEASUREMENT OF

LOESS‐PALEOSOL SEQUENCES ...... 53 Abstract ...... 53 5.3.1. Introduction ...... 54 5.3.2. Material and methods ...... 58 5.3.2.1. Sample Preparation ...... 59 5.3.2.2. Laser diffraction measurement ...... 60 5.3.2.3. Optical model ...... 60

5.3.2.4. Content of CaCO3 and Corg ...... 61 5.3.2.5. Inductively coupled plasma‐optical emission spectrometry ...... 61 5.3.2.6. Scanning electron microscopy ...... 62 5.3.3. Results ...... 62 5.3.3.1. Effect of acetic acid treatment ...... 62 5.3.3.2. Comparison of HCl treated and untreated samples ...... 62

5.3.3.3. MODGSD dependence on CaCO3 and Corg content ...... 66 5.3.3.4. Element composition of the HCL leachable fraction ...... 67

VI

5.3.3.5. Scanning electronmicroscopy (SEM) ...... 69

5.3.3.6. MODGSD, CaCO3 and Corg at section DDG ...... 71 5.3.4. Discussion ...... 73 5.3.4.1. MODGSD of the samples from the SHD section ...... 74 5.3.4.2. (a) Disintegration of aggregates and organo‐mineral complexes ...... 75 5.3.4.3. (b) Formation of cSi‐ to fS‐sized aggregates ...... 76 5.3.4.4. Comparison of SHD and DDG results ...... 76 5.3.5. Conclusion ...... 78

5.4. THE DIFFERENCE OF TWO LASER DIFFRACTION PATTERNS AS AN INDICATOR FOR POST‐DEPOSITIONAL GRAIN SIZE

REDUCTION IN LOESS‐PALEOSOL SEQUENCES ...... 80 Abstract ...... 80 5.4.1. Introduction ...... 81 5.4.2. Materials and Methods ...... 82 5.4.2.1. Sample selection ...... 83 5.4.2.2. Laser diffraction measurement ...... 84 5.4.2.3. Optical model ...... 85 5.4.2.4. Data preparation and illustration ...... 87 5.4.3. Results ...... 88 5.4.3.1. Submicron fraction of single samples ...... 88 5.4.3.2. Loess sections ...... 90 5.4.4. Discussion ...... 93 5.4.4.1. Quality check of the clay measurement ...... 93

5.4.4.2. Dependency of the ΔGSDraw on the mineral composition ...... 94

5.4.4.3. ΔGSDclr as proxy in loess research ...... 96

5.4.4.4. ΔGSDclr variability within the loess‐paleosol sequences ...... 97 5.4.4.5. Potential sources of error ...... 99 5.4.5. Conclusion ...... 99

SYNTHESIS ...... 101

6.1. SAMPLE PRETREATMENT PRIOR TO LASER DIFFRACTION SIZE ANALYSIS ...... 101

6.2. GRAIN SIZE MEASUREMENT AND CALCULATION ...... 104

6.3. THE ANALYSIS OF GRAIN SIZE DATA ...... 107

DANKSAGUNG / ACKNOWLEDGMENTS ...... 112

REFERENCES ...... 116

APPENDIX A ...... 140

APPENDIX B ...... 148

VII

List of Figures

Fig. 3.1: Two exemplary samples illustrated as cumulative percentage frequency curves (A and B), as percentage frequency histogram (A), as distribution density curve (B), as colorline signature (C) and as a heatmap signature (D)...... 12

Fig. 5.1.1: Map of the northeastern Tibetan Plateau. The box shows the location of the main section Suohuduo at the southern margin of the Anyemaqen Shan. The numbers indicate further sections mentioned in the text...... 26

Fig. 5.1.2: NW–SE-cross section indicating the distribution of loess and loess-like sediments in eastern Tibet...... 27

Fig. 5.1.3: (A) Situation of the valley close to the settlement Suohuduo. (B) Suohuduo section indicating the two paleosols. Photos: F. Lehmkuhl, August 2006...... 29

Fig. 5.1.4: (A) Mean grain-size distributions of selected units from the Suohuduo on log-normal scale, (B) stratigraphy, (C) depth profiles of selected grain-size fractions (C: 0.04–2 μm; fSi: 2–6.3 μm; mSi: 6.3– 20 μm; cSi b 36: 20–36 μm; cSi > 36: 36–63 μm; fS: 63–200 μm; mS: 200–630 μm; cS: 630–2000 μm) and (D) depth profiles of grain-size mean and mode...... 34

Fig. 5.1.5: Triangle diagram of the textural classes (classification according FAO, 2006) of all Suohuduo samples and of the three comparative sections (Aba, Jiukehe and Dari) as mean values. In addition, the classification of loess, sandy loess and sand from the Donggi Cona catchment are shown...... 36

Fig. 5.1.7: Comparison of paleoenvironmental reconstructions by this study with other results from the Donggi Cona, the Chinese Loess Plateau and the Lake...... 41

Fig. 5.2.1: Loess distribution in Europe (modified according to Haase et al., 2007), maximum extent of the Weichselian ice sheet, coastline during LGM and location of the sections Doroshivtsy in the western Ukraine and Semlac in south-western Romania...... 46

Fig. 5.2.2: Lithostratigraphy, sensitive grain size ranges and selected element concentrations of the Semlac section. OSL dating was carried out on silty quartz samples (grain size 40-63 µm) and a standard SAR protocol (Murray and Wintle, 2003). The central age model was used for De calculation...... 48

Fig. 5.2.3: Lithostratigraphy, specific GS-ratio, and U-ratio of the Semlac section...... 49

Fig. 5.2.4: Lithostratigraphy, sensitive grain size ranges, and Al/Zr ratio of the Doroshivtsy section...... 50

VIII

Fig. 5.2.5: Lithostratigraphy, specific GS-ratio, U-ratio, Sedimentology, OSL ages, and Radiocarbon ages of the Doroshivtsy section. OSL dating was carried out using OSL and pIRIR stimulation (Klasen et al., 2015)...... 51

Fig. 5.3.1: Stratigraphy and location of the Suohuduo section at the northeastern margin of the Tibetan Plateau (Lehmkuhl et al., 2014)...... 56

Fig. 5.3.2: Stratigraphy and location of the Düsseldorf-Grafenberg coring situated at on the right side of the Rhine River...... 57

Fig. 5.3.3: Heatmap showing HCl-induced MODGSD (frequency of HCl treated subsamples – frequency of untreated subsamples) of all samples from the Suohuduo section. This difference is shown for each of the 116 grain size classes resulting from the laser diffraction analysis...... 63

Fig. 5.3.4: Depth profiles of the parameters mean, mode and median for the MODGSD of the Suohuduo section...... 64

Fig. 5.3.5: GS distribution curves of untreated samples (top), the HCl-treated samples (upper middle), calculated differences (lower middle), and mean SD of all aliquot measurements (bottom) for three pedogenic units of the Suohuduo section: (A) primary loess, (B) modern soil, (C) paleosols...... 65

Fig. 5.3.6: Dependence of the HCl-induced MODGSD on (A) the content of CaCO3 and on (B) the content

of Corg for the pedogenic units primary loess, modern soil, and paleosol...... 67

Fig. 5.3.7: Two aliquots of the same samples after the first 5 h run of the H2O2 pretreatment cycle. The right glass tube was previously treated with HCl, the attempt in the left glass tube only pre-moistened with aqua dest...... 69

Fig. 5.3.8: SEM images of a sample from paleosol S01 at different magnifications: (A) and (B) show the

untreated sample, (C) and (D) the sample pretreated with H2O2, and (E) and (F) the sample pretreated

with H2O2 und HCl...... 70

Fig. 5.3.9: Heatmap showing HCl-induced MODGSD (frequency of HCl treated subsamples – frequency of untreated subsamples) of all samples from the Düsseldorf-Grafenberg section. This difference is shown for each of the 116 grain size classes resulting from the laser diffraction analysis...... 71

Fig. 5.3.10: Depth profiles of the statistical parameters mean, mode and median of the MODGSD for the Düsseldorf-Grafenberg coring...... 72

Fig. 5.3.11: Comparison of mean MODGSD and standard deviation of both Suohuduo (SHD) and Düsseldorf-Grafenberg (DDG) sequences, showing heterogeneous effects of HCl treatment on different GS spectra...... 77

IX

Fig. 5.4.1: Grain size distribution curves of four standard mineral samples and two Russian standard soils reduced to the submicron fraction (left column). The specifications of these samples are given in Tab. 5.4.1. The right column shows the results of five natural clay mineral samples...... 89

Fig. 5.4.2: Optical microscope photograph showing the submicron fraction after separation by Atterberg sedimentation of (A) a sample from the S1 palaosol of the Semlac sequence and (B) a pure quartz sample ...... 90

Fig. 5.4.3: Heatmap showing the ΔGSDraw (A) and the ΔGSDclr (B) of all samples from the loess section Semlac (Romania). The vertical curves represent the contents the fraction 15-30 µm and <1µm, the CaCO3 content and the magnetic susceptibility (MS) ...... 91

Fig. 5.4.4: Heatmap showing the ΔGSDclr of all samples from the Düsseldorf-Grafenberg sequence. The vibracore consist of calcareous loess and intercalated interglacial, interstadial and periglacial soils and soil sediments. The vertical curves represent the content of submicron particles...... 92

Fig. 5.4.5: Heatmap showing the ΔGSDclr of all samples from the Suohuoduo section, containing pure loess (white), slightly weathered loess (light grey) and chernozem-like paleosols (S01; S02: dark grey). The

vertical curves represent content of submicron particles and Corg...... 93

Fig. 6.1: GS distribution curves of three loess samples on a different state of pedogenesis. Prior to measuring each sample was dispersed in tetrasodium pyrophosphate solutions of different concentrations (given in g/L)...... 102

Fig. 6.2: GS distribution curves of two silt dominated sediment samples. The samples were measured several times. Prior to each repetition ultrasonic treatment was applied for 120 seconds respectively...... 103

Fig. 6.3: Illustration of different equivalent spheres of the same “submicron” irregularly shaped particle (in accordance to Malvern, 2012)...... 105

Fig. 6.4: Heatmap showing the grain size distribution of all samples from the Suohuoduo section, containing pure loess (white), slightly weathered loess (light grey) and chernozem-like paleosols (dark grey). The vertical curves represent classical grain size parameters and ratios ...... 108

X

List of Tables

Tab. 5.1.1: Radionuclide concentrations derived from laboratory high resolution γ-spectrometry, measured water contents, calculated dose rates, equivalent doses and the resulting ages of the 125–180 μm quartz fraction. Ages in italics are calculated with the finite mixture model after Galbraith and Green (1990), all others are with the weighted mean age model including 5% instrumental error and given with 1σ age deviation...... 40

Tab. 5.1.2: Results of the radiocarbon dating of section 2 (2.11), section 3 (3.3) and section 7 (7.11, 7.19)...... 40

Tab. 5.3.1: Precision of particle size measurements of a modern top soil, a paleosol and a loess sample of the SHD section. Precision was assessed by four aliquots of the same bulk sample and is gives as % CV...... 59

Tab 5.3.2: CaCO3 contents [%] and concentrations of HCl-leachable elements [mg/l] within the supernatant after centrifugation ...... 68

Tab 5.3.3: Elemental concentrations [mg/kg] of the stratigraphic units of the SHD section; aqua regia extract

(3:1 concentrated HCl and concentrated HNO3) ...... 68

Tab. 5.4.1: Properties and references of the selected standard material samples...... 83

XI

XII

Introduction

1.1. Framework and objectives

The application of laser diffraction particle size analysis is widespread in earth sciences (Konert and Vandenberghe, 1997; Pye and Blott, 2004; Blott and Pye, 2006; Machalett et al., 2008; Roberson and Weltje, 2014; Újvári et al., 2016). Large amounts of grain size data can be generated in a short time with satisfying precision and accuracy (Blott and Pye, 2006; McCave et al., 2006; Miller and Schaetzl, 2012). Due to technical advancements since the 1980s, granulometric analysis of sediments from Pleistocene and Holocene loess-paleosol sequences are increasingly used in paleoclimate research (Vandenberghe, 1997; Antoine et al., 2009a; Lehmkuhl et al., 2014). The reconstruction of past environmental and climate conditions is important for the general understanding of climate processes. Loess-paleosol sequences are often investigated since they are one of the most widespread terrestrial archives that record environmental developments (Sun et al., 2006; Antoine et al., 2009a,b; Bokhorst and Vandenberghe, 2009; Antoine et al., 2013; Gocke et al., 2014; Meszner et al., 2014; Vandenberghe et al., 2014; Obreht et al., 2016). Through the investigation of information about the Cenozoic and especially the Quaternary, climate and environmental evolution can be obtained.

Various methods exist to analyze the particle size distribution of natural sediments, including dry or wet sieving, several gravitational techniques, or direct microscopic measurement by microscopy. Each technique is appropriate for a specific grain size range. Sieving is most useful for sand or gravel and gravitational techniques for coarse clay and silt, while a combination of several microscopic techniques allows the entire range from clay to sand to be analyzed. However, the latter technique is very time- and cost-intensive.

The advantage of the laser diffraction technique is that it is suitable for a wide range of grain sizes (fine clay to coarse sand) and is extremely time- and cost-effective. For the investigation of temporal trends within huge sample sets, as is common in loess-paleosol-sequence studies, this method provides grain size results in a high resolution (116 classes in case of the LS 13320) with sufficient precision (Chapter 5.3.) (Blott and Pye, 2006). Recently, laser diffraction has been widely used to characterize fine sediments (< 2 mm). However, especially in soil science, this technique remains controversial since the assumption of a significant underestimation of the clay fraction (in comparison to gravitational techniques) is frequently discussed (Loizeau et al., 1994; Konert and Vandenberghe, 1997; Beuselinck et al., 1998; Roberson and Weltje, 2014; Újvári et al., 2016).

1

This study intends to disclose and defuse the constraints of the laser diffraction technique concerning grain size date acquisition and subsequent analysis and interpretation. For this purpose, the main objectives of this dissertation were the following:

 The evaluation of different grain size analysis techniques to characterize loess-

paleosol sequences;

 The evaluation of the effect of pretreatment agents on grain size data processed for

samples from loess-paleosol sequences developed under different weathering

conditions. In particular, the influence of HCl treatment should be tested;

 The identification of the advantages of grain size calculation by the application of

the Lorenz-Mie theory;

 The improvement of the visual presentation of grain size data processed for loess-

paleosol sequences;

 The evaluation of commonly used grain size parameters and ratios as proxy data

for paleoenvironmental variations; and,

 The introduction and interpretation of a novel reliable grain size parameter.

1.2. Outline

Initially, Chapter 2 presents a brief overview of the state of the research concerning loess-paleosol- sequence investigation. A special focus is on the processes which influence the grain size distribution subsequent to the accumulation of the aeolian dust.

Chapter 3 reviews the most frequently used methods for grain size analysis which have been used through the present time. The respective methods are assessed according to their suitability for the characterization of loess sediments.

Chapter 4 offers an overview of the locations of the loess-paleosol sequences investigated in this dissertation. The respective sequences have been selected in such a way that they cover a wide spectrum of varying degrees of weathering and pedogenesis. Each sequence consists of loess sediments which experienced different post-depositional alterations as a function of different paleoenvironmental conditions. The Suohuduo sequence (Tibetan Plateau, ) developed 2

during the Holocene. Under predominantly dry and cold environmental conditions, dark and slightly weathered Eurasian steppe soils developed. Under distinctly warmer and moister conditions near Semlac (Arad Plain, Romania), the local loess plateau covers sediments accumulated relatively slowly over the past 350,000 years; therefore, intensively weathered soils were formed and preserved. The Grafenberg sequence (Germany) is located in the oceanic influenced lower Rhine embayment. Due to high humidity and periglacial processes, erosional processes are comparatively enhanced. The Doroshivtsy sequence represents a special case in this selection as it consists of former aeolian sediment accumulated by fluvial and periglacial processes in a foot slope position.

Chapter 5 comprises the manuscripts of three original research articles and one extended abstract showing two case studies. The content of the respective subchapters follows the form as it was published in the corresponding scientific journals. Only editorial modifications have been made to integrate the individual articles into the dissertation.

Chapter 5.1. shows a classic granulometric-based case study focused on the reconstruction of grain size variability within a loess-paleosol-sequence. The paleoclimatic interpretation is based on a combination of different established parameters to visualize the variability within a primary grain size data set. Both the horizontal (over all calculated size classes) variation of grain size distribution curves averaged for the respective sediment units and the vertical variability (over all individual samples of the sediment sequence) of common statistical parameters result in a generalization, and, hence, in a loss of information.

Chapter 5.2. shows the problems occurring when common grain size parameters and ratios of fixed grain size ranges are applied as proxy data. The advantages of individualized grain size ratios, which are independent of firm boundaries and highly sensitive to individual sedimentary or post- sedimentary processes, are discussed. Such ratios react sensitively in individual grain size influencing processes but remain a generalization of the horizontal variability.

Predominantly based on the results of Chapter 5.1., in an analytical test series, the influence of hydrochloric acid treatment prior to laser diffraction size analysis was evaluated. The expertise and interpretations gained from these experiments are described in Chapter 5.3. Additionally, in this chapter, the visualization of grain size data in the form of a heatmap signature has been established. In this way, both primary and differential (e.g., GSD without HCl treatment minus GSD after HCl treatment) grain size data sets of sediment sequences can be visualized without any loss of horizontal or vertical variability.

Chapter 5.4. focusses on the calculation of the primary laser diffraction scattering pattern by the means of two different optical models (Fraunhofer approximation and Lorenz-Mie theory). In this 3

study, a novel grain size parameter, the ΔGSD, has been developed, evaluated, and interpreted. This parameter can be used as a suitable indicator of the degree of post-depositional chemical weathering in loess-paleosol sequences, and, consequently, as a proxy for past climatic and environmental evolution. The ΔGSD signature is unaffected by the constraints outlined in the previous chapters and by other grain size modifying processes, such as cryogenesis, or the input of further synsedimentary deposits like coarse material transported by saltation, reptation, creeping, or sheet wash.

Chapter 6, the final chapter of this dissertation, represents an overall synthesis based on the major findings of the individual chapters.

1.3. Associated research projects

The presented research was conducted as part of two research projects. The first is called Landscape and Lake-System Response to Late Quaternary Monsoon Dynamics on the Tibetan Plateau - Northern Transect and was funded by the Deutsche Forschungsgemeinschaft (DFG) between 2008 and 2014. The research area of the project comprises the two lake catchments on the north-eastern margin of the Tibetan Plateau: These catchments are the Donggi Cona (4100m asl.) in the north-east and the Lake Heihai in the in the Mountain range at the northern margin (4420m asl.) of the Plateau. The project aims to reconstruct complete sediment cascades from the high mountain area to the lakes and their relation to the late Quaternary climate evolution (monsoon versus Westerlies). The Sino‐German research group consisted of scientists from the RWTH Aachen University, the Alfred‐Wegener Institute in Potsdam, the Free University of Berlin who collaborated with Nanjing University. The research strategy of the RWTH Aachen team reflects the identification of climato-sensitive geomorphic process regions and the timing of sediment-mobilization and deposition periods of glacial, aeolian and fluvial sediments in the context of the late Quaternary climate and landscape evolution. For the reconstruction detailed sedimentological and geochemical analysis were used. The project results demonstrate that the two study areas have different sediment sources and varying sediment cascades. While the Donggi Cona system is dominated by a large alluvial fan with a strong recycling of older sedimentary deposits (IJmker et al., 2012a; IJmker et al., 2012b; Stauch et al., 2014), in the Heihai catchment the local lithology is the main influencing factor for the sedimentary system. The analysis of some samples which was carried during the first funding period of this project was finished at the beginning of the second phase. The results of the sample set from the Suohuoduo section, mainly processed by the author and his supervisor Frank Lehmkuhl, the full professor for Physical Geography and Geoecology at RWTH Aachen University, are part of this dissertation thesis. 4

The second research project in which framework this dissertation was conducted was the Collaborative Research Center (CRC): OUR WAY TO EUROPE - Culture-Environment Interaction and Human Mobility in the Late Quaternary funded by the Deutsche Forschungsgemeinschaft (DFG) between 2009 and 2017.

The research project deals with the history of mankind and specifically with the understanding of human-environment interaction at the time of Europe’s colonization by first anatomically modern humans and their population mobility. This objective was pursued with a combination of geoscientific and archaeological methods. The chair of Physical Geography and Geoecology at RWTH Aachen University was involved in the subproject B1: The "Eastern Trajectory": Last Glacial Paleogeography and Archaeology of the Eastern Mediterranean and of the Balkan Peninsula. Although a lot of natural factors like the access to raw material, water and food and booty are of importance for a corridor which was preferred during the spread of early modern humans on their way to Western Europe, we believe that local climate acts as one of the main steering factors for human dispersal, presence and absence. A climatically more favorable region with more moisture resulting in a different vegetation pattern close to the border of the Carpathian Arch could be the reason why especially here higher densities of remains from early anatomically modern humans have been found. Further, the chair of Physical Geography and Geoecology investigated the subproject D1: Analysis of Migration Processes due to Environmental Conditions between 40,000 and 14,000 a BP in the Rhine Catchment and Adjacent Areas. Around 40,000 B.P the early modern humans arrives the western central European habitat. After their initial dispersal into Europe they emigrate around 30,000 a BP. Subsequent to the last glacial maximum (LGM), a resettlement by the Homo sapiens started. The combined geoscientific and archaeological research concentrated on the two time slices: MIS 3 and LGM/post-LGM.

One main task of these two subproject was to combine granulometric and geochemical methods for selected key sections to get proxy data to correlate loess-paleosol sequences in different geomorphological positions. In the framework of this dissertation the results of the loess-paleosol sequence Semlac (Arad Plain, Romania) and Düsseldorf Grafenberg (Lower Rhine Embayment, Germany) was used to improve and precise the laser diffraction size analysis.

5

Fundamentals of loess-paleosol-sequence investigation

2.1. Loess as an aeolian sediment

Loess is defined as an aeolian sediment predominantly consisting of silt size particles (Pye, 1995; Stauch et al., 2012; Muhs et al., 2014). Aeolian sediments cover approximately 30% of the earth’s surface (Thomas and Wiggs, 2008).

The formation of loess is controlled by various factors. The most important are sediment source, chain of transport processes, wind as the final transport agent (Wright, 2001; Smalley et al., 2009), an area favorable for accumulation due to a reduction of wind speed or sediment trapping by vegetation or gravel layer (Bagnold, 1941; Lancaster and Baas, 1998), and an adequate timespan, which is necessary for the initial preservation of the deposited dust particles (Pécsi, 1990; Smalley and Marković, 2014). Generally, aeolian sediment transport occurs during dry and/or glacial periods (Tsoar and Pye, 1987). However, for example, even an increase in moisture availability can result in the accumulation of loess due to the fluvial transport of silty and sandy sediments to storage sites where it is easily available for aeolian deflation (Bullard and McTainsh, 2003; Stevens et al., 2013; Nottebaum et al., 2014; Nie et al., 2015). Further problems in the interpretation of aeolian sediments as paleoproxy are the frequent reworking of primary aeolian sediments and inherited weathering products due to (glacio-) fluvial processes, solifluction, or cryogenic processes (Stevens et al., 2006; Schirmer, 2016; Lehmkuhl et al., 2016) and gaps or discordances in loess sections due to erosional events (Thomas, 2013; Schirmer, 2016; Lehmkuhl et al., 2016).

During the erosion, transport, and accumulation processes, sediment sorting occurs. Variations of the grain size distribution of loess-paleosol sequences are frequently used for the reconstruction of the climatic and environmental conditions during aeolian sedimentation (Ding et al., 1995; Muhs and Bettis, 2003; Nugteren and Vandenberghe, 2004). If the accumulation rate is greatly reduced while moisture availability is increased, soil formation is enhanced (e.g., Stevens et al., 2011; Sprafke and Obreht, 2016). Primary loess consists mainly of coarse silt (Pécsi, 1990; Pye, 1995; Muhs and Bettis, 2003), whereas submicron particles (< 1 µm) are greatly underrepresented during accumulation (Qiang et al., 2010; Újvári et al., 2016). Soil formation results in post-depositional grain size fractionation (e.g., Újvári et al., 2016). As a function of physical and chemical weathering processes, the dominating medium and coarse silt fractions are reduced in favor of the clay and fine silt fractions.

6

2.2. Post-depositional alteration of loess sediments

2.2.1. Grain size reduction After the sedimentation and initial fixation of atmospheric mineral dust particles, a first weak sediment alteration process, often called loessification, begins (Berg, 1916; Pécsi, 1990; Smalley et al., 2011; Svirčev et al., 2013; Smalley and Marković, 2014; Sprafke and Obreht, 2016). Such processes are not sufficiently understood, and it is a matter of debate whether they should be assigned to pedogenic or diagenetic process spheres or to a kind of transition zone (Sprafke and Obreht, 2016). However, there is a consensus that the typical structure of a loess sediment is generated by these loessification processes, whereby the loess is differentiated from primary airborne dust (Pécsi, 1990; Sprafke and Obreht, 2016).

If the sediment accumulation rate is strongly reduced while moisture availability is increased, these initial and weak alteration processes are enhanced and result in actual pedogenesis (e.g., Stevens et al., 2011).

Due to soil formation, the grain size of the non-carbonate part of the loess sediment is reduced (e.g., Újvári et al., 2016). In the near surface range, these post-depositional fractionation processes occur both in carbonate containing and in carbonate-free milieus. If decalcification takes place, grain size reduction follows downwards as much as several meters (Rohdenburg and Meyer, 1966) and results in a reduction of the dominating coarse silt fractions in favor of an increase in the clay fractions (fine and medium silt fractions also increase in most cases). In fact, the post-depositional variation of the sand fraction is negligible.

Generally, there are two possible causes for fractionation: the chemical weathering of silt-sized minerals, such as mica and feldspar, as a result of hydration and hydrolysis (Schaetzl and Thompson, 2015) and the physical breakdown of particles due to cryogenic processes. In contrast to chemical weathering, in physical weathering, all contained mineral types are affected. During the Pleistocene, within the near surface part of loess sequences, both processes emerge in parallel, whereas the physical breakdown decreases with the increasing depth of the weathering front (Rohdenburg and Meyer, 1966).

Due to the relative absence of fine and medium clay during the deposition of loess, the enrichment of these fractions is often used as a proxy for pedogenesis and climate variations (e.g., Nugteren and Vandenberghe, 2004; Antoine et al., 2009b; Terhorst et al., 2012). For example, if physical processes prevail, cryosols or chernozems can be formed within loess sediments. Under dominant chemical weathering conditions, the formation of cambisols, luvisols, and planosols is more likely (Schaetzl and Thompson, 2015).

7

Former soils buried and preserved by overlying sediments are called as paleosols (Sheldon and Tabor, 2009). During their formation, the soils are exposed at the earth’s surface and interact with the contemporary prevailing climatic and environmental conditions. Consequently, they are frequently used as a powerful archive to reconstruct past environmental conditions (Sheldon and Tabor, 2009). To this end, loess-paleosol sequences are increasingly investigated in all regions were loess and loess derivatives are distributed and preserved (Kukla and An, 1989; Muhs and Bettis, 2003; Sun et al., 2006; Smalley et al., 2011; Gocke et al., 2014; Marković et al., 2015; Lehmkuhl et al., 2016).

2.2.2. Aggregation and agglomeration of loess sediments Post-depositional processes not only result in a general size reduction of the sediment components, but, during pedogenesis particularly, the fine materials of loess sediments tend to aggregate into larger flocks with varying grain sizes and resistances (Schulten and Leinweber, 2000; Balabane and Plante, 2004; Six et al., 2004; Lützow et al., 2008; Stamati et al., 2013). These soil aggregates are cemented by agents such as calcium carbonate, clay minerals, organic matter, or metallic oxides and hydroxides (Oades, 1988; Wiseman and Püttmann, 2006; Kögel-Knabner et al., 2008). Especially within temperate soils on loess sediments, various stabilizing mechanisms are evident, reducing or preventing the degradation of the organic matter (Boudot et al., 1989; Lützow et al., 2006; Kögel-Knabner et al., 2008; Lützow et al., 2008).

These mechanisms include: (i) occlusion in aggregates, (ii) aromatic compounds/black carbon, (iii) cation sorption, and (iv) anion sorption.

(i) Within soil aggregates, organic matter (OM) is protected against decomposition by physical separation of the organic material and the decomposer (Oades, 1995; Lützow et al., 2006). The majority of the soil organic carbon (Corg) in silty soils is aggregated with clay minerals and organo- mineral complexes of the 2–20 µm size fraction (Van Gestel et al., 1996; Schulten and Leinweber,

2000; Lützow et al., 2006). The lowest content of Corg is associated with the 20–50 µm fraction which is typically composed predominantly of non-aggregated mineral particles. The largest grain size fraction, 50–200 µm, falls between these extremes regarding its Corg content (Van Gestel et al., 1996). Large fragments of plant residue microaggregates, organo-mineral complexes, and mineral particles are bound together into macroaggregates by transient interaction with fungal hyphae and roots or by polysaccharides (Six et al., 2000, 2004; Stamati et al., 2013). In addition, inorganic binding agents such as iron oxides or carbonates play an important role in the formation of micro- and macroaggregates. During sample pretreatment, the OM in such stable complexes will only be decomposed if the surrounding clay minerals or iron oxides have been destroyed previously. At 8

lower pH values (< 4.8), some clay minerals and iron oxides can be destroyed, and weak clay-humic complexes and microaggregates can be split up (Kögel-Knabner et al., 2008; Vaasma, 2008).

(ii) Particularly in steppe soils, a high proportion of Corg is present as black carbon (highly-dispersed charcoal) (Rodionov, 2010). This is caused by natural vegetation fires and has often been observed in chernozems and similar soils (e.g., Haumaier and Zech, 1995; Glaser et al., 2001). Such black carbon is assumed to form aromatic compounds (Lützow et al., 2006). If present, these occur in different grain size ranges, their detection is complicated (Schmidt et al., 2001), and the compounds are very resistant to degradation (Schmidt and Noack, 2000; Lützow et al., 2008).

(iii) Cations are selectively bound to negatively charged humic substances and clay minerals (Edwards and Bremner, 1967; Wiseman and Püttmann, 2006). The tendency to sorb cations increases with higher pH values and cation charge (Na+/K+ < Ca2+/Mg2+ < Al3+/Fe3+). In particular, cations with a higher charge form covalent bonds with an OH- group of adjacent chemical compounds (Lützow et al., 2006; Nieder and Benbi, 2008). Organic compounds form clay-humic complexes (Baldock and Skjemstad, 2000; Laird et al., 2001) with clay minerals and organo-metallic complexes through interaction with iron oxides (Lützow et al., 2006; Lützow et al., 2008).

2- 3- (iv) Humic substances and clay minerals adsorb anions (SO4 , Cl-, NO ) if Fe or Al ions are bound on their surface. The adsorbed amount increases with decreasing pH value (Hingston et al., 1974; Nieder and Benbi, 2008). The sorption increases with the specific surface area of the sorbents (Baldock and Skjemstad, 2000; Kaiser and Guggenberger, 2003). Furthermore, pedogenic chlorites and poorly crystalline pedogenic oxides can be formed owing to pH reduction when sufficient H+ and Cl- ions are supplied. The chlorites and pedogenic (hydro-) oxides tend to coagulate to stable aggregates because of their large surface area and high electrostatic charge (Schulten and Leinweber, 2000).

During sample preparation, aggregates bound by calcium carbonate or clay minerals can be easily decomposed, whereas organic matter can be embedded within stable aggregates, in which it is preserved and protected against chemical decomposition by natural leaching processes or by classic sample preparation techniques (Chapter 5.3.).

2.3. Grain size parameters and ratios for loess-paleosol sequences

9

Over the past decades, several parameters and ratios were introduced and suggested as proxy data for Quaternary paleoclimate and paleoenvironmental evolution. The vertical grain size variability within loess-paleosol sequences serves as a proxy for reconstruction of, e.g., the energy of the wind, moisture availability, distance and direction from the dust source, vegetation cover, and the type of transport process (saltation or suspension) (e.g., Prins et al. 2007, Újvári et al. 2016).

Classical statistic parameters, such as the mean, median, mode, or selected grain size fractions, are calculated from the distribution density curves to function as proxy data (Újvári et al., 2016). Liu et al. (1985) suggested the mean and median as indicators for paleowind strength, while An et al. (1991) concluded that these parameters reflect the aridity of the source region and the frequency of dust storm events. Prins et al. (2007) suggest the median value as a proxy for the dynamic of the East Asian winter monsoon on the Chinese Loess Plateau. In more detail, the variability of the median is interpreted as a function of wind intensity during the winter monsoon, secondary to the distance to the source (Sun et al., 2012). In other studies, specific grain size classes are interpreted as proxy data. For example, the fine sand fraction (63–200 µm) is used as an indicator of the frequency of stronger winds and dust storms (Antoine et al., 2013; Zeeden et al., 2016). Furthermore, the fraction < 20 µm is often suggested as a proportion which was transported as background sedimentation over long distances and was deposited far from the source area (Pye, 1995; Sun et al., 2002). The disadvantage of such single value parameters is their susceptibility to post-depositional or synsedimentary processes (e.g., pedogenesis or aeolian short distance transport) which indirectly affect the primary dust signal.

To avoid this problem, Vandenberghe et al. (1985) introduced the first widely used grain size ratio, the U-Ratio (16–44 µm/5.5–16 µm), and evaluated it for a section of the Chinese Loess Plateau (Luochuan et al., 1997; Nugteren and Vandenberghe, 2004). Machalett et al. (2008) developed a similar ratio using a loess-paleosol-sequence from Kazakhstan (Remisowka) called the Twin-Peak ratio (30.1–63.4/11.8–27.4 µm). Both ratios exclude the fine range where pedogenically formed minerals occur and the sand fraction which is affected by short distance transport (saltation, reptation, creeping, or sheet wash). High values of these ratios indicate relatively coarser, silty material and are interpreted as the influence of higher wind speeds under cold climatic conditions. Conversely, low values are connected to less intense winds and warmer climatic conditions. Today, the most frequently used grain size ratio is the grain size index (GSI), first suggested by Rousseau et al. (2002). The coarse silt fraction (20–50 µm) is divided by the entire finer range, including the pedogenically affected clay. This proxy was modified to adjust to data sets from other measuring techniques (26–52.6 µm/< 26 µm, for the Nussloch section, Germany) (Antoine et al., 2009a) or to other loess regions (20.7-63.4 µm/< 20.7 µm, for the Eustis section, Nebraska, USA) (Rousseau et al., 2007) and (22.73-63 µm/ < 22.73 µm, for the Surduk section, Serbia) (Antoine et al., 2009b). 10

High values are interpreted as a result of a period with a high frequency of intense dust storm events. However, Rousseau et al. (2011) stated that the grain size distribution of loess is the result of a combination of changing wind and moisture conditions, both in the source and the deposition area of the aeolian dust. As the GSI includes the pedogenically affected clay fraction, this proxy is suggested to reflect all these environmental parameters and not just the wind speed (Rousseau et al., 2011; Újvári et al., 2016).

11

Established methods for grain size analysis of loess

The grain size analysis comprises the destruction and dispersion of granular soil sample packages into individual grains and their separation into distinct classes. The percentage of each individual grain size class of the total aliquot volume or weight is determined to present the measurement results as cumulative percentage frequency curves or as particle size distribution density curves (Fig. 3.1).

Fig. 3.1: Two exemplary samples illustrated as cumulative percentage frequency curves (A and B), as percentage frequency histogram (A), as distribution density curve (B), as colorline signature (C) and as a heatmap signature (D).

12

There are various grain sizing techniques; the most important can be separated into two main categories (Gee and Orr, 2002; Roberson and Weltje, 2014): (i) Direct measuring of individual grains a. Microscopic analysis b. Sieve analysis (ii) Indirect measuring of collectives or suspensions a. Sedimentation/gravitation analysis b. Laser diffraction size analysis.

The importance of knowledge about the size of individual grains to characterize the properties of a composite particle has been appreciated for a long time. It is said that the Egyptians used sieves to sort metal ores over 4,000 years ago (Miller and Lines, 1988). The technical development of sizing methods continues to the present day. The first technical devices were sieves made of textile fabric (Kinsman, 1978). Since the early 20th century, a combination of sieving and the pipette method is the certified standard for industrial applications (ISO 11277, 2002), and most methods are established in academics (Konert and Vandenberghe, 1997; Buurman, 2001; Müller et al., 2009). Furthermore, sophisticated optical and electron microscopy techniques are used (Allen, 1990). As this procedure is conducted manually and is very time-consuming, cost-intensive, and prone to operator errors (Syvitski et al., 1991), several automated electronic grain sizing techniques which are more rapid, accurate, and require significantly less sample material were developed and established in the previous decades (e.g., Roberson and Weltje, 2014). The first major innovation was a modification of the existing gravitational sedimentation methods in the early 1970s: X-ray granulometry with sedimentometers (Hendrix and Orr, 1972). In the late 1970s, the first authors described the angular dependency of forward scattered coherent light on the size of suspended particles (Cornillault, 1972; Weiss and Frock, 1976; McCave and Syvitski, 1991). The first commercial laser diffraction particle size analyzer was offered by the French company CILAS in 1968 and was followed by instruments developed by Leeds and Northrup, Malvern Instruments, Horiba, Fritsch, and Beckman Coulter (Agrawal et al., 1991).

3.1. Microscopic analysis

Particle sizing by microscopy is often used as an absolute detection of grain size distributions since it is the only method in which the individual particles can be observed directly. Furthermore, microscopic investigation permits observation of the shape and composition of particles and aggregates in a better resolution than any other technique. There are different commonly used microscopic devices: optical microscope, transmission electron microscope (TEM) and scanning 13

electron microscope (SEM). The selection of the most suitable depends on the considered grain size range and the availability of time and money (Allen, 1990).

Optical microscopes are suitable for a grain size range from 0.8 to 150 µm (Allen, 1990). Possible measuring parameters are Feret’s diameter (distance between two parallel tangents at opposite particle edges), Martin’s diameter (length of a line dividing the particle in a fixed direction into two segments with same area), equivalent projected area (and the related diameter) and longest dimension (Kinsman, 1978; Allen, 1990). Generally, the projected area diameter is the most suitable estimator of the true cross-sectional area of the particle (Allen, 1990). It is the most suitable for comparison with sieving results, whereas for comparison with laser diffraction size analysis an average of all possible Feret’s diameters is best suited.

Submircon particles (colloides) cannot be detected directly using optical microscope. Siedentopf and Zsigmondy (1903) demonstrated that such small particles can be made visible by lateral lighting, as a light scattering cone spreads out from them. In contrast to laser diffraction techniques, the size and angle of theses diffraction cone is not indicative for the size of the particles (Lagaly et al., 1997).

Scanning electron microscopy (SEM) is suitable for particle characterization in a size range of 0.1 to 15 µm. The sample is scanned line by line by an electron beam. The secondary electrons emitted by atoms excited by the electron beam are collected by a detector. Due to the low emitted energy, only atoms located near the sample surface are exited and thus provide an image of the sample topography (Goldstein et al., 2007). The advantage of SEM is the high optical depth, so that the images of particles can be sharp focused from the center to the edges. The disadvantage for particle sizing of loess is the charging effect due to the low electric conductivity of dry terrestrial sediments. To avoid electric charge, the top of the sample surface has to be connected to sample holder. To achieve this, the electrically non-conducting material such as carbon, Au-Pb or Pt-Pd mixtures is dusted with a thin layer (Goldstein et al., 2007, Halfpenny, 2010). However, for layer silicates with a porous structure and a large inside surface the coating can be is partially incomplete (see Chapter 5.3.). Modern techniques as low vacuum method (Halfpenny, 2010), applying of biasing voltage to the sample surface or surface charge neutralization by an ion gun (Titze, 2013) allow the microscopy of non-conducting materials without coatings.

Using Transmission electron microscopes (TEM) a resolution from 0.001 to 5 µm can be achieved (Allen, 1990; Reimer, 2013). The sample is set within a thin membrane on a grid made usually by copper. Electrons are transmitted throw the sample and provide an image on a photographic plate or on a fluorescent screen. This technique is expensive, time consuming and rarely statistically significant (Kinsman, 1978; Allen, 1990; Reimer, 2013). 14

Grain size analysis by any of these microscopic techniques require a careful slide preparation, which include the selection of a representative aliquot, a proper dispersion and a minimum of particle stacks and overlaps. Generally, the higher the resolution of the microscopic technique, the lower is the representativeness of the investigated aliquot, since sample quantities are extremely small compared to other techniques. For adequate particle separation a wet dispersed suspension can be transferred onto a microscopic slide. After evaporation of the fluid the dry dispersed particles can be investigated. Alternatively, the dispersion can be fixed by very fast freezing (Lagaly et al., 1997).

Beside the slide preparation, the statistical significance of the counted particle amount is most important to obtain an accurate grain size distribution representing the investigated sediment sample. Kinsman (1978) stated that a minimum of 2000 counts are required for narrow size distributions or previously fractionated subsamples and accordingly more for broader distributions as loess sediments. Manually microscopic investigations are time consuming and prone to deviations due to different operating persons. Modern automated image scanning techniques have been developed to increase the speed and accuracy of microscopic grain size characterization (Kinsman, 1978; Allen, 1990; Holdich, 2002; Francus, 2005; Jouve et al., 2013).

3.2. Sieve analysis

Particles that pass through a standardized sieve are defined as particles having a sieve diameter just below the length of a square hole (Konert and Vandenberghe, 1997). The sieve diameter is defined as the width of the minimum square aperture through which the particle will pass (Allen, 1990). Since Egyptian times different types of sieves have been used. The first known sieves were made of woven material, but the ancient Egyptians already used perforated plate sieves and woven wire sieves, as they still in use nowadays in a similar form (Allen, 1990). The first sieves with standardized meshes were proposed by P. Ritter von Rittinger (1967). From this point sieves were used for particle classification and test sieving. Following, an increasing number of different classification systems were developed (Blott and Pye, 2012). The technical committee ISO/TC 24 “Particle characterization including sieving” of the international standard organization (ISO) had the nominal sizes of sieve apertures committed to 125 to 0.02 mm in the ISO 565 (1983). In this standard the gradation of the sieve meshes is derived from the preferred numbers R20/3 (Renard series) which is based on a sieve opening of 1 mm in a twenties root of ten progression. The 3 after the slash means a reduction of the entire series to every third value. For more narrow size distributions there is an intermediate series based on R20 numbers in the same standard. In Germany the corresponding standard is the DIN 4188-1 which defines the grading of analytic sieves openings since 1977. Here the main scale is defined as the R10 number series which is closer 15

to the Atterberg size scale (Atterberg, 1905). In particle size measurements by sieve analysis the produced grain size classes are constrained by sieves with opening sizes which are produced by the manufactures. Today, the most widely used sieves refer to the standard DIN ISO 3310-1 (2001) which has evolved from the DIN 4188-1 or the Atterberg scale which is incorporate in the international standard DIN EN ISO 14688-1 (2016). The sieves are stacked and sorted with opening size ascending from bottom to top. After shaking over a defined intensity and time each sieve with the oversize grain is weighed. The mass of each grain size fraction is calculated by subtracting the weight of the empty sieve respectively. The result of an individual fraction is expressed as the percentage of each fraction of the total sample mass. The percentage mass fractions of the sediment sample can be visualized as a histogram. By interpolation between the measurement points a cumulative distribution curve can be calculated (see. Fig. 3.1) (Allen, 1990; Retsch 2015). Depending on the general fineness of the sediment sample, different sieving techniques should be preferred: dry sieving is used for particles between 40 µm and 125 mm in size, wet sieving is appropriate in a size range between 20 µm and 20 mm and air jet sieving should be used in a range between 10 µm and 200 µm if dry measuring conditions are essential (Retsch, 2015). The resultant size values are to be considered as the minimum square opening through which the particle can pass, named sieve diameter. This value is only a function of the two shortest dimensions, whereas the longest dimension is not crucial for the passage of the particle (Allen, 1990).

Several limitations in sieving methods have been known for many years. One problem is that for an individual particle the probability of passing through a sieve opening depends on the size distribution of the loaded sample (especially the content of very fine particles), the number of particles which are close to the opening size, the physical properties of the particles (especially the shape) and the geometry of the sieve (relation of openings to woven metal mesh) (Gee and Or, 2002). Since not all openings have the defined size, but are slightly smaller, and irregular shaped particles can pass only in one orientation through the meshes, particularly the particles with sizes near the opening size have a limited chance of passage (Gee and Or, 2002).

Nevertheless, sieve analysis is one of the most widely used methods of particle size analysis since particles are fractionated just on the basis of size, independently of their density and surface properties and the results are quite reproducible even when different sets of sieves are used (Allen, 1990).

3.3. Sedimentation / Gravitation analysis

There are several sedimentation techniques which are all based on Stokes’ Law (Kinsman, 1978; Allen, 1990; Újvári et al., 2016), which is used to calculate the sedimentation velocity of spherical 16

bodies in a fluid. If gravitational forces prevail, the operating temperature remain constant and vibrations are avoided, it is assumed that the terminal velocity of a particle depends on the mass of the particle and the viscosity of the fluid (Stokes 1851). The distance (or height h) which a particle settles over time t is used to determine the terminal velocity V:

(1)

Furthermore the terminal velocity of a particle is a function of gravitation g, the fluid density ρfluid, the fluid viscosity η, the particle density ρparticle and the particle diameter d and can be calculated by:

² (2)

After equating the equation can be solved for d:

∗ (3)

For a single particle or a monodisperse suspension, d corresponds to the Stokes diameter (diameter of a spherical particle and settling velocity in the laminar flow region (Allen, 1990)) and can be estimated if settling height and time are measured experimentally (Kinsman, 1978).

Oden (1915) stated that a settled material of a suspension consisting of a homogeneous mixture of different sizes is divided into two fractions after time t in a tube of height h (McCave and Syvitski, 1991):

1. The fraction of the bulk material with V(t) < h/t  the partially sedimented fraction 2. All fractions with V ≥ h/t  the fully sedimented fraction

By considering the mass of the particles within the suspension exhibiting settling velocities < V(t) a cumulative settling velocity distribution can be obtained. This requires experimental analysis as measuring given fractions, which are separated by wet sieving, in settling tubes over defined time 17

intervals (Brozek and Surowiak 2005). For a known density of all measured particles by means of Stokes’ Law the particle size frequency curve can be calculated from the settling velocity weight frequency curve (a further derivation from the cumulative settling velocity distribution) (McCave and Syvitski, 1991).

Since the early 20th century several sedimentation techniques were established. The methods can be divided into two main categories: The incremental methods, which determine the change of concentration or density with time in a constant depth (Areometer/Hydrometer, Photometer, Sedimentometer) and the cumulative methods, which determine the mass of particles settled after a defined time and distance (Atterberg sedimentation, Pipette analysis). Different grain size ranges are suggested in the literature for the sedimentation methods. Kinsman (1978) suggests a particle size range of 2 to 40 µm, Mingard (2009) suggests 2 to 50 µm and Fan and Zhu (2005) set the lower limit to 5 µm. In the submicron fraction the settling velocity is affected by Brownian movement. Therefore 1 µm is the generally accepted lower limit. The international standard organization defined the grain size range for sedimentation technique from 2 to 63 µm, which is equivalent to the silt fraction (ISO 11277, 2002).

3.3.1. Areometer Method Bouyoucos (1927) and Casagrande and Loos (1934) developed a method to determine the grain size of dispersed fine-particles by measuring density changes of the suspension as a function of time using an aerometer (also referred to as hydrometer). The procedure is described in the German standard DIN 18123 (2011). The international replacement document DIN EN ISO 17892-4 (2014) is in preparation. At the beginning of the method all particles have to be homogenously suspended within a fluid before leaving them to the gravitational forces. Depending on the grain size varies the settling velocity according to Stokes’ Law (equation 2). Since the suspension density depends on the particle concentration, it also changes with time and can be measured using an aerometer (Casagrande and Loos, 1934). For determination of the respective grain size frequencies there are different methods. The most common method is the deduction of the results using a nomogram (DIN 18123, 2011). Due to the simplicity and the time- and cost- efficiency this method was widely used since the 1980’s years. But in recent times, it is rarely used, because it is not very accurate compared to modern standards (McCave and Syvitski, 1991)

3.3.2. Photometer Method

18

The transmissivity of a homogenous particle suspension changes over time, as the particle settling progresses. By measuring the changes of the amount of light transmitted through the suspension using a photometer or hydrophotometer, the grain size distribution of silt can be calculated. The transmitted light energy is received on a photocell. If time and particle density are given, the maximum grain size at the point of detection can be calculated by Stokes’ Law (equation 2). In the metal and cement industries this method have been commonly used until the 1970’s (Kinsman, 1978). Since its accuracy is poor for polymodal samples, in the years thereafter it was decreasingly used (Singer et al., 1988).

3.3.3. Sedimentometer Method (SediGraph) As in any aforementioned classical method based on gravitational settling in accordance to Stokes’ Law for the sedimentometer technique the dispersed clay-silt sample is homogenously suspended within a sedimentation tube. The method is defined for opaque and spherical particles with the same chemical composition (ISO 13317-3, 2001). The particle concentration in a specific depth is determined by detecting the increasing signal absorption from a beam of X-rays. This absorption is proportional to the concentration (C [g/ml]) of sample suspended within the tube area penetrated by the X-ray beam. Following Stokes’ Law (equation 2) after time ti all particles coarser than size xi are settled from the top of the suspension tube below the height h where the concentration Ci is measured by X-ray absorption (Coakley and Syvitski, 1991). The weight percentage pi of the fraction finer than xi can be calculated as follows:

100 ∗ (4)

The concentration is measured continuously and after certain time intervals values are obtained.

By using the calculated Pi values, which are decreasing with time, and the corresponding sizes xi (Stokes’ or Free-falling diameter) a cumulative grain size distribution curve can be plotted (Olivier et al., 1970; Micromeritics, 1982). By subtraction the weight percentages of two adjacent grain size classes (P1 – P2) the content of a specific grain size class (x1 to x2) can be calculated. The values of the specific classes can be used to plot a distribution density curve.

To obtain a more rapid measurement, the sedimentation tube is moved downwards successively with decreasing grain size classes to reduce the height h, and thereby, the time t (Kinsman 1978, Micromeritics, 1982). In comparison with other Sedimentation analysis the advantages of the

19

sedimentometer method are: the automated procedure, the digital output of the results, the reduction of analysis time and the small aliquot size required.

3.3.4. Pipette analysis The combined sieve-pipet analysis is the traditional, widely accepted and certified technique for grain size analysis (McCave and Syvitski, 1991; Beuselinck, 1998; ISO 11277, 2002; Kowalenko and Babuin, 2013). The grain size fractions coarser than 63 µm are determined by sieving as described above. The clay and silt fractions are measured by the pipette analysis according to Köhn (1928).

Using Stokes’ Law following equation (2), for time t at depth h the maximum grain size of the particles suspended in a fluid with given density can be calculated (McCave and Syvitski, 1991; Gee and Or, 2002).

Aliquots are taken by extraction of definite volumes by means of a pipette from a thin layer of a settling suspension in definite time intervals (from few minutes to several hours or even days) (ISO 11277, 2002). After the evaporation of the fluid the solid material with known maximum grain size

(xn) remains in the sample pan and can be weighed. A cumulative distribution is generated by the weights of the successively later extractions. The content of a specific grain size class (x1 to x2) can be calculated by subtraction (x2 – x1), while the last fraction (usually <2µm) is the weight of the remnant minus the proportionate weight of the crystallized dispersant (Köhn, 1928; ISO 11277, 2002). This principles only work, on the condition that the sample is perfectly dispersed, no particle interactions occur beside the free Stokesian settling and above all that the starting suspension is perfectly homogeneous (McCave and Syvitski, 1991).

3.3.5. Atterberg sedimentation The Atterberg sedimentation method allows the determination of absolute contents of specific grain size classes and the total grain size separation of a bulk sample by following Stokes’ Law (Equation 2) (Atterberg, 1914; Müller and Engelhardt, 1967). After separation the subsample can be further characterized by other analysis as X-ray diffraction (XRD) (Lehmann, 2004; Blume et al., 2011). Similar to the pipette method, the sample has to be homogeneously suspended within a so called Atterberg cylinder. When the settling time for the finest desired grain size (usually 1 or 2 µm) is reached at the point of the cylinder where the release opening is placed, the supernatant suspension containing only finer particles is discharged into a sample pan or vessel. Following, the remaining suspension is filled up to the starting level and is strongly shaken until homogeneity of the suspension is reached and the settling process begins again. To gain the entire fraction, the

20

procedure has to be repeated until the supernatant is perfectly clear (usually 10 to 15 times) (Stein, 1985). Subsequently, the further fractions are separated in the same way, whereby the required time decreases with increasing grain size classes.

3.3.6. Limitations of gravitational techniques Generally, for all sedimentation techniques one restriction is that the terminal velocity of a particle through the suspending fluid is not only subjected to gravitational force, but other factors as well: - The autumn leaf effect: The principle of all gravitational techniques requires that the particle density is known and equal for all containing particles (usually the density of quartz is assumed), and that all particles are spherically shaped. While the density of the major component of sedimentary particles is a narrow range (2.5–2.8 g/cm3 for quartz, feldspars, and common clay minerals), the shape of natural particles is usually asymmetrical and heterogeneous (Gee and Or, 2002). The nonsphericity increases with increasing post- depositional alteration and with decreasing grain size. Hence, such irregular shaped particles have a lower settling velocity than a volume- or area-equivalent sphere (Matthews, 1991; Allen, 2003). - The pirate effect: Due to electrostatic or van der Waals forces, small particles can be attached to significantly greater particles (Qiang et al., 2010; Újvári et al., 2016). - Particle interaction: Between single particles random movements (Brownian motion or upward current) result in a random settling velocity independent of the physical grain size (Kinsman, 1978; Allen, 1990; Allen, 2003). Especially in the submicron range, this effect results in increasing unreliability with decreasing grain size (Allen, 1990; Eshel et al., 2004; ISO 13317-1, 2001).

21

Loess-paleosol-sequences developed under different paleoenvironmental conditions

This dissertation focuses on the methodological progress of the investigation of grain size variability within loess-paleosol sequences. To cover a wide range of influencing factors, test sites with different environmental conditions and at varying geomorphological positions were chosen:

The Suohuoduo section (34°28’ N, 99°50’ E; 4016 m a.s.l.; China) located on the northeastern margin of the Tibetan Plateau. It developed during the late Pleistocene and Holocene and consists of loess, Chernozem-like paleosols and a slightly weathered modern soil (Lehmkuhl et al., 2014). The grain size distribution is characterized by a gradual fining upwards tendency since the late glacial due to increasing temperature and moisture availability. During periods which were favorable for soil formation the post-depositional grain size modification is affected by the formation and preservation of stable organo-mineral complexes.

The Semlac section is a natural outcrop exposed at a loess plateau (46° 7'12.97"N / 20°56'54.70"E; ~100 m a.s.l.; Romania) consisting of very homogenous and relatively fine loess sediments likely accumulated in the past 350,000 years. During interglacial and interstadial periods well developed paleosol complexes were formed. The intense soil formation in the surrounding mountains of the Carpathian Basin, in comparison to sections in the central part, is due to the higher moisture availability and the relatively low accumulation rate during the Pleistocene (Zeeden et al., 2016). The GS-variation is mainly influenced by post-depositional processes such as clay mineral translocation and –formation (Schulte et al., 2014; Zeeden et al., 2016).

The Düsseldorf-Grafenberg sequence (51°14´49´´N, 6°50´58´´E; 83 m a.s.l.; Germany) is located on the eastern side of the Rhine River and represents a special geomorphological position (interfluve (recently)) with high accumulation rates and favorable sediment preservation characteristics (Lehmkuhl et al., 2016). It consists of calcareous loess accumulated during cold and semiarid conditions of the last glacial cycle and intercalated soils and soil sediments which have developed under interglacial, interstadial and periglacial conditions (Schulte et al., 2016). Large parts of the loess sequence are affected by secondary sedimentation processes (e.g. fluvial reworking), which had considerable impact on the final grain size distribution.

Finally, the section Doroshivtsy (48°35’37.6’’N, 025°52’10.7’’E; 150 m a.s.l.; Ukraine) is presented in the framework of Chapter 5.2. as a further case study to demonstrate the problems occurring, when common grain size parameters are applied as proxy data. The section is situated in the right bank of the middle course of the Dniester River (Kulakovska et al., 2014). More specifically, it is exposed within a slope sediment which covers the lower terrace. It represents a more than 9 m 22

sequence of predominantly fluvial relocated sandy loess with intercalated weak humic horizons. The sediment consist of formerly aeolian transported loess and coarser slope-wash material. In comparison to the Semlac sequence (located 470 km southwest) the Doroshivtsy sequence covers only the time span since the last glacial maximum, but due to the high accumulation rate and the resulting good preservation conditions, these period is recorded by the sediment in an extremely high resolution.

23

Manuscripts

5.1. Timing and spatial distribution of loess and loess-like sediments in the mountain areas of the northeastern Tibetan Plateau

Frank Lehmkuhla, Philipp Schultea, Hui Zhaob, Daniela Hüllec, Jens Protzea, Georg Staucha

a Department of Geography, RWTH Aachen University, Templergraben 55, D-52056 Aachen, Germany b Key Laboratory of Desert and Desertification, Cold and Arid Regions Environment and Engineering Research Institute, Chinese Academy of Science, Lanzhou 730000, China c Institute for Geography, University of Cologne, Albertus-Magnus-Platz, D-50923 Cologne, Germany

Published 2014 in Catena, 117, 23–33

Abstract Most studies on landscape evolution on the Tibetan Plateau during the late Quaternary have mainly focussed on using lacustrine records. However, mantles of sandy silt and paleosols also provide valuable archives for reconstructing Holocene paleoenvironmental change. Yet little is known about the distribution and timing of these late Quaternary aeolian sediments. To enhance understanding and knowledge of aeolian sediments in Tibet and to help reconstruct the nature of Late Glacial through mid-Holocene landscape development, a loess-paleosol sequence, the Suohuduo section, located at the eastern margin of the Tibetan Plateau at about 4,000 m above sea level is examined using sedimentological, geochemical and geochronological methods. A chronostratigraphy is established using nine optically stimulated luminescence and one radiocarbon ages. Sedimentation increased during the Late Glacial and the early Holocene with an upwards fining of sediments in the lower part of the section. Two mid Holocene paleosols that date to about 8.5 ka to 7 ka and ~5.5 ka to 4 ka reflect more humid climate conditions during the mid-Holocene than earlier times. The upper part of the section is mainly silt, which reflects a more open landscape with higher aridity since 4 to 3 ka this time. These data support evidence for similar climate/paleoenvironmental change in adjacent regions on the Tibetan Plateau.

24

5.1.1. Introduction Loess sequences provide important and in some cases an almost continuous record of Quaternary paleoenvironmental change; some of sequences even provide a record extending into the Pliocene (e.g. Ding et al., 1992). These loess sequences are particularly impressive on the Chinese Loess Plateau (CLP) and have been studied in great detail (e.g. An et al., 1991; Pye, 1995 and references therein; Buylaert et al., 2008, 1989; Liu et al., 1985; Vriend et al., 2011). Loess and associated aeolian and slope deposits are also present in the high mountains at the northeastern margin of the Tibetan Plateau (Fig. 5.1.1). These deposits have the potential to provide valuable archives for reconstructing Quaternary landscape development in Tibet and they also provide important comparisons for the lacustrine archives, which have been regularly studied on the Tibetan Plateau. At the eastern margin of the Tibetan Plateau in basins at elevations ranging from 2,000 m to 3,500 m above sea level (asl), the slopes are covered with silt deposits with thicknesses up to several meters; at an elevation above 3,500 m asl, thinner deposits of silty sand (< 1 m thick) exist (Fig. 5.1.2; Lehmkuhl, 1997). These aeolian sediments, together with slope wash deposits, contain paleosol horizons and ash layers from burning that provide evidence for alternating times of landscape stability and geomorphic activity throughout the late Quaternary. To illustrate the potential and importance of these aeolian archives in Tibet, we examine a section loess-paleosol sequence, the Suohuduo section, located at the eastern margin of the Tibetan Plateau at about 4,000 m asl. We compare this section with other aeolian mantles along a transect from the Sichuan Basin towards the upper reaches of the Huang He (Fig. 5.1.1 and 5.1.2).

Accumulations of loess in the valleys in eastern Tibet around 3,000 m asl were first described by Tafel (1914), an early German explorer. He observed loess on the mountain slopes close to Songpan (see Fig 5.1.1) and suggested that the predominate winds from NW to SE, especially winter storms, caused accumulation of aeolian dust (loess) on the northwestern slopes. In addition, Hövermann (1987) reported that loess is the dominant surface cover on the eastern slopes of the Anyemaqen Shan at elevations of 3,500 - 3,900 m asl where alpine meadows exist. Lehmkuhl (1995, 1997; Lehmkuhl et al., 2000) supported the view that these sands and silts cover in several mountain areas of the Tibetan Plateau are aeolian in origin. The aeolian cover is predominantly sandy-loess in the areas above about 3,600 m to 4,300 m asl in eastern Tibet and up to more than 5,000 m asl in western Tibet as well as in Mongolia (Lehmkuhl, 1997). Kaiser et al. (2007, 2009) and IJmker et al. (2012) published studies on mountain silts on the Tibetan Plateau. Yet most recently, Stauch et al. (2012) presented 51 new OSL ages from aeolian sediments from the Donggi Cona area.

25

Fig. 5.1.1: Map of the northeastern Tibetan Plateau. The box shows the location of the main section Suohuduo at the southern margin of the Anyemaqen Shan. The numbers indicate further sections mentioned in the text.

26

Fig. 5.1.2: NW–SE-cross section indicating the distribution of loess and loess-like sediments in eastern Tibet.

Most papers dealing with late Quaternary landscape evolution on the Tibetan Plateau are based mainly on lacustrine records with studies focussing on lake level changes (e.g. Herzschuh, 2006; Lehmkuhl and Haselein, 2000) and pollen data (e.g. Herzschuh, 2006; Schlütz and Lehmkuhl, 2009). With respect to recently published paleoclimatic records, the lake nearest to our research area south of the Anyemachin Shan is the Donggi Cona, about 150 km northwest of the study area (Opitz et al., 2012, see Fig. 5.1.1). is about 250 km further north (An et al., 2006; Fig. 5.1.1). In a summary based on several selected lacustrine records from the mid-latitude arid Asian region, Chen et al. (2008) postulate that the paleoclimate records in arid central Asia are predominately influenced by today’s mid-latitude westerlies. Further results concerning the late Quaternary evolution are derived from cosmogenic surface exposure dating (Owen et al., 2003; Lehmkuhl and Owen, 2005).

5.1.2. Regional setting and distribution of loess and loess-like sediments Our key sections are located in the vicinity of the Anyemaqen Shan (or Amne Machin, Anye Machin) in the upper reaches of the Huang He (Yellow River, Fig. 5.1.1). The Anyemaqen Shan range trends northwest stretching from 34° to 35°N and 99° to 100°30’E and covers an area of roughly 190 km by 80 km. With the highest peak, Anye Machin, is 6,282 m asl. the Anyemaqen Shan rises above the present day snowline and boasts modern valley glaciers on the different slopes. Pleistocene glacial landforms such as moraines, erratics, cirques, and trough valleys are well developed and allow reconstructions former glacier extents throughout the late Quaternary (Lehmkuhl 1995). Owen et al. (2003) dated three terminal moraines in the eastern Anyemaqen

27

Shan at 45.5 ka, 16 ka and 9 ka and suggested that glaciers in the more monsoon-affected regions of Tibet (such as the Nianbaoyeze and the Anyemaqen Shan; see Fig. 5.1.1 and 5.1.2) advanced during times of increased insolation, such as MIS-3 and the early Holocene. Huge active gravel accumulations and inactive older gravel terrace sequences formed from glacial-fluvial streams and glaciers at an elevation ranging from 4,000 m to 5,000 m asl provide a source for dust production.

The climate of the study area is monsoonal to continental and is characterized by a wide annual range of temperature and summer precipitation. In the higher parts of the Tibetan Plateau near its eastern margin, the mean annual temperature (MAAT) is rather low, for example, it is 0.1°C in Jiuzhi (3,629 m asl), -1.3 in Dari (3,968 m asl), and -4.1 in Madoi (4,272 m asl) (Fig. 5.1.1; Domrös and Peng, 1988). The average environmental dry lapse rate is ~0.55°C/100 m. In addition, more than 80% of the total precipitation occurs from May to October. Monthly precipitation is < 10 mm in the winter season, which is very dry due to winter winds (winter monsoon) controlled by the Siberian and Tibetan high-pressure system. In general, there is a decrease in (summer) precipitation from >1,000 mm at the southeastern margin of the Tibetan Plateau towards the interior in the west to values of less than 300 mm/a. The mean annual rainfall in valleys and basins ranges from 300 mm up to more than 1,000 mm (estimated) in the summit areas. Even though the Jiuzhi, Dari and Madoi climate stations are situated in the valleys bottoms they show a decrease in precipitation towards the west: Jiuzhi 765 mm, Dari 537 mm, and Madoi 306 mm/yr. However, the precipitation in the mountains should be much higher.

The vegetation in the study area, from the Sichuan Basin towards the Tibetan Plateau, is among the most diverse in the Holarctic region (Mutke and Barthlott, 2005). During the past several centuries, the natural vegetation at elevations below about 3,300 m asl has been converted from forests to agricultural land and, in the upper parts, to pastures. More descriptions about the vegetation belts are summarized in Schlütz and Lehmkuhl (2009). Soil types differ due to slope, aspects and elevation. However, the prevalent soil types are chestnut soils (kastonozems) followed by phaeozems or mountain tschernozems occurring in the higher catchment areas. Furthermore, shallow luvisols, gleysols, regosols and local peat bogs are present at even higher elevations.

Loess deposits, several decimeters thick, have accumulated in basins below 3,400 m asl, e.g. the Basin of Aba (Fig. 5.1.1, No. 2). Holocene luvisols are present on top of these loess deposits. At higher elevation, up to the boundary of alpine meadows, the surfaces are covered by sandy silt up 50 to 60 cm thick (Lehmkuhl and Lui; 1994, Lehmkuhl, 1995). An example is the Jiukehe section on the northern slope of the Nianbaoyeze Shan (Fig. 5.1.1, No. 3). Sand and dune fields close to Madoi, Dari (Fig. 5.1.1, No. 4) and in the Zoige Basin (Fig. 5.1.2) are composed dominantly of fine and medium sand indicating higher wind speeds. The lowermost parts in the Sichuan Basin, below

28

1,000 m asl, are covered by the locally called Chengdu clay. Yang et al. (2010) show that these sediments in the Huagai section (Fig. 5.1.1, No.1) are aeolian in origin. In addition, Yang et al. (2010) demonstrate that there is a eastward change in particle size, from ~ 4Φ (64 µm) to below 6Φ (16 µm), in the loess deposits from the Tibetan Plateau to the Sichuan Basin, which suggests that the Tibetan Plateau could be a potential dust source for the CLP and even for the Chengdu clay. The different grain-size pattern of the aeolian mantles has already been described in Lehmkuhl (1995).

Fig. 5.1.3: (A) Situation of the valley close to the settlement Suohuduo. (B) Suohuduo section indicating the two paleosols. Photos: F. Lehmkuhl, August 2006.

The key section examined in this current study is located at 4,016 m asl close to a little village called Suohuduo (SHD, 34°28’ N and 99°50’ E) in the Anyemaqen Shan. The studied section is an exposure along a county road (Fig. 5.1.3). The aeolian sediments were deposited on a terminal moraine several tens of meters above the valley floor. Further terminal moraines are present downstream of the section. The source area of the glaciers is ~40 km upstream. The flanks of the 1 km-wide valley are covered by several sets of lateral moraines. The valley floor is completely covered by fluvial sediments that make up the main inactive river terrace. The present river bed is deeply incised in these sediments.

The Suohuduo section is compared with three other aeolian sections at Ake River, Jiukehe and Dari (Fig. 5.1.1 and 5.1.2). The section at Aba (32°57´N/ 101°35´E; 3,330 m asl) is situated on top of the first river terrace of the Ake River. The surrounding areas above the modern floodplain are cultivated with barley. The section at Jiukehe (33°32´N/ 101°05´E; 3,860 m asl) is situated on top of the first river terrace of the Jiukehe River in an environment with alpine meadows. Further descriptions are provided in Lehmkuhl (1995). The section at Dari (34°09´N / 99°17´E; 4,320 m asl) is along a road cut in a small natural depression of a fossil dunefield buried under the modern 29

alpine meadow environment close to the settlement of Dari. In all probability, the origin of the fine sand of this dunefield is the nearby floodplain of the Huanghe River.

These aeolian sediments are also compared with those in the catchment of the Donggi Cona (Fig. 5.1.1, No. 5) described by IJmker et al. (2012) and Stauch et al. (2012). Loess and loess-like sediments as well as sand dunes can be distinguished in this area, at elevations ranging from 4,050 m to 4,300 m asl. Sand dunes and sand sheets are mostly present in the basins. When a lake is present, sand dunes and sand fields are predominately found in the eastern parts of basins as they originated from paleoshores and were accumulated by westerly wind systems.

5.1.3. Material and methods 5.1.3.1. Geomorphological mapping, site selection and sampling strategy Geomorphological mapping and the analysis of landforms and sediments, aided by the use of global positioning systems, and satellite images, were used to provide an overview of the general geomorphology and to identify the areas for detailed study. Designations of soil horizons and classification of textural classes are given using FAO (2006) and AG Boden (2005). Samples from the SHD-section were taken at 10 cm intervals in the upper 35 cm and at 5 cm intervals from 40 to 340 cm for sedimentological analysis. OSL samples were confined in selected horizons.

5.1.3.2. Sedimentological and geochemical methods

Particle size, CaCO3–content and elemental composition were determined for 64 samples. For particle-size analysis, the samples were air-dried. To remove the organic matter, the samples were treated with 0.70 ml 20 % H2O2 at 70 °C for several hours. This process was repeated four times over a period of two days. To keep particles dispersed, the samples were treated with 1.25 ml

Na4P207 for 12 hours (Pye and Blott, 2004; DIN ISO 11277, 2002). Particle size was measured with a Laser Diffraction Particle Size Analyzer (Beckman Coulter LS 13 320) by calculating the mean diameters of the particles within a size range of 0.04 - 2000 μm with an error of 2 % (see Appendix Tab. A1). Each sample was measured four times in two different concentrations to increase accuracy. To determine the grain-size distribution the Fraunhofer theory was used for particles with a diameter greater than 36 µm, while the Mie theory was used for particles with a diameter of less than 36 µm (Fluid RI: 1.33; Sample RI: 1.55; Imaginary RI: 0,1) (Özer et al. 2010; ISO 13320- 1, 1999). For the section at Dari, samples were sieved with nets of 630 µm, 200 µm and 63 µm mesh width, respectively, to separate the sandy fraction from the smaller grains of the sample. The remaining fraction of the sample was measured with a Sedigraph micromeritics 5100 (Coakley and Syvitski, 2007). 30

Three sediment types were differentiated: 1) Samples with a silt and clay content of 80% or more and a homogenous sediment structure are denoted as ‘loess; 2) Samples with a silt and clay content of 50 - 80% are denoted as ‘sandy loess’ (Stauch et al., 2012); and 3) Samples with a silt and clay content of less than 50 % are denoted as ‘silty sand’ (Fig. 5.1.4 and 5.1.5).

To determine the element concentrations of the fine-grained fractions, the <63- µm fraction was sieved and dried at 105°C for 12 hours. An 8g-quantity of the sieved material was mixed with 2g Fluxana Cereox, homogenized and pressed to a pellet with a pressure of 20 tons for 120 s. All samples were measured twice with the XRF device. Mean values were calculated from the two measurements (according to SPECTRO, 2007).

Element concentrations of magnesium (Mg), aluminium (Al) calcium (Ca), titanium (Ti), zirconium (Zr), phosphorous (P) and manganese (Mn) were determined using the XRF device (see Appendix Tab. A2). To differentiate the paleosols and the surrounding loess-like sediments, two element ratios were calculated: i.e., the Ti/Zr ratio indicating changes in provenience and weathering (e.g.

Zech et al., 2008); and the modified MgO/(CaO/Al2O3) ratio based on Sverdrup et al.(1992), which is referred as ‘WP’, a proxy for weathering. The CaCO3 content (see Appendix Tab. A1) was determined volumetrically with the SCHEIBLER-method (Schaller, 2000; ISO 10693, 1995).

5.1.3.3. Dating methods 5.1.3.3.1. Luminescence dating To provide information on the geochronological context, optically stimulated luminescence (OSL) dating was applied to nine samples from the SHD section (Tab. 5.1.1). With OSL dating, the time elapsed since the last exposure of quartz or feldspar grains to sunlight during transport processes can be determined. For a summary of the technical basics and a review of application studies see Preusser et al. (2008), Aitken (1998) or Bøtter-Jensen et al. (2003).

Nine OSL samples were collected from section SHD. The sampling points were selected on the basis of subdivision boundaries observed in the field (Fig. 5.1.4). All samples were obtained by hammering 5 cm-diameter iron tubes into cleaned vertical sections. The tubes were covered with a lid immediately after they had been taken from the section, and then sealed inside black plastic bags with tape to ensure the samples retained their natural water content.

OSL dating procedure for three samples (T06-1 to T06-3) was undertaken at the Luminescence Dating Laboratory at the University of Cologne. The OSL measurements on the other 6 samples SHD 19 – SHD 24) were carried out in the luminescence laboratory of the Cold and Arid Environment and Engineering Research Institute, CAS, China.

31

In the laboratory, the material at each end of the tube that might have been exposed to light was scraped away and used for water content and dose rate measurements. All laboratory processes of sample preparation and luminescence measurement were carried out in subdued red light. Raw samples were treated with 10% HCl and 20% H2O2 to remove carbonate and organic matter. The samples were then sieved in water to select the grain size interval of 125-180 m. To obtain quartz, this grain fraction was separated by solutions of sodium polytungstate of densities 2.62 g/cm3 and 2.75 g/cm3. After drying, the quartz grains were treated with 40% HF for 40 minutes to remove the outer layer irradiated by alpha particles and any remaining feldspars. The grains were then treated with 1 M HCl for 10 minutes to remove fluorides created during the HF etching.

These quartz grains were mounted with silicone oil on aluminum discs as aliquots for measurement. In Cologne (T06-1 to T06-3), 25–50 aliquots with 2 mm diameter quartz grains (~100–500 grains, Duller 2008) were measured for each sample. At CAS, 24 aliquots with 4 mm diameter quartz grains were measured for each sample. Before measurement of the OSL signals, the room temperature IRSL signal and the 110 °C TL peak were monitored for each aliquot to check for feldspar contamination (Li et al., 2002). Only the aliquots with IRSL signal intensities of less than 3 % of the natural OSL signals, with recuperation (the corrected OSL signal of the regeneration dose at 0 Gy) of less than 3% of the natural signal and with a recycling ratio of between 0.90-1.10 were used for the De determination (Wintle and Murray, 2006).

OSL measurements of all samples were carried out by using the automated Risø TL/OSL-DA-15 reader (Markey et al., 1997). The OSL signal was detected through two 3 mm thick Hoya U-340 filters. Laboratory irradiation was performed using 90Sr/90Y sources mounted within each reader, with dose rates of 0.105 Gy/s and 0.104 Gy/s in the Cologne and CAS laboratory, respectively.

The De value for each aliquot was determined using a modified single-aliquot regenerative-dose procedure (SAR) (Wintle and Murray, 2006). All aliquots were preheated at 260 °C for 10 s according to results of preheat-tests, and the OSL signals (Ln the natural OSL signal and Lx created by laboratory regenerative doses) were read at 125 °C. Sensitivity monitoring was achieved by measuring the OSL signal, Tn and Tx, obtained after a cut heat to 220 °C; these signals were created by a fixed test dose after every Ln and Lx measurement, respectively. The De values were calculated by comparing the sensitivity-corrected OSL intensities Ln/Tn with the values of Lx/Tx measured for each dose point on each aliquot.

The environmental dose-rate originates from the radioactive elements existing in grains of the sample and surrounding sediment, with a contribution from cosmic rays. For the samples measured in the University of Cologne, the HP-Ge-Gamma-Spectrometry method was used to obtain the contribution from the uranium (U), thorium (Th) decay chains and from potassium (K). Indications

32

of radioactive disequilibria were not observed. For the samples measured in the Cold and Arid Regions Environment and Engineering Research Institute, U, Th concentrations and K contents were determined by means of Neutron Activation Analysis (NAA). All measurements were converted to alpha, beta and gamma dose rates according to the conversion factors of Guérin et al. (2011). The dose rate from cosmic rays was calculated on the basis of sample burial depth and the altitude of the section (Prescott and Hutton, 1994). The water content was calculated as the ratio of moist weight over dried sample weight, obtained from sample weights before and after drying in an oven. Due to the uncertainty in the mean water content during the sediment burial period, an error of 5 percent was added to water contents when the ages were calculated.

5.1.3.3.2. Radiocarbon dating Five samples were dated using radiocarbon methods, all obtained from bulk material of buried paleosols and dated by AMS in the Erlangen laboratory (Tab. 5.1.2). All data were calibrated with CalPal-online and the CalPal2007-HULU curve (Weninger and Jöris, 2008). In the following, radiocarbon ages are presented as calibrated years (cal years BP) unless otherwise noted.

5.1.4. Results The key section, SHD, is composed of two distinctive paleosol units separated by two layers of sandy loess (Fig. 5.1.4), whereby the late Pleistocene loess layer (L1) is located at the bottom and the modern soil covers the section on the top. The stratigraphic subdivision was conducted by examining of the color, texture and structure of the sediment in the field. The stratigraphic subdivisions of section SHD are described as follows from the top to bottom:

(1) TS, 0-20 cm: modern top soil (phaeozem or mountain tschernozem), with multiple plant roots

(2) L01, 20-90 cm: yellow sandy-silt, loose crumbly structure

(3) S01, 90-120 cm: paleosol, dark grey silt

(4) L02, 120-150 cm: pale yellow sandy-silt, loose crumbly structure.

(5) S02, 150-200 cm: paleosol, dark grey, nearly black silt, more weathered than S01

(6) L03, 200-265 cm: pale yellow sandy-silt, interbedded strata of several thin dark laminations indicating initial soil formation

(7) L1, 265-340 cm: yellowish silty sand at the base which fines upwards to sandy silt.

33

The sections at Aba and Jiukehe did not contain any paleosols, and they were composed of mostly silty sediments (silty loam, Fig. 5.1.5). The section at Dari was composed of coarse silty to clay loam (Fig. 5.1.5) and two small humic horizons are present at a depth of 85 to 145 cm.

5.1.4.1. Sediments and grain size distribution The grain-size distribution curves of the stratigraphic subdivisions for the SHD section (Fig. 5.1.4) are unimodal and have peaks either in the coarse silt or in the lower fine sand fraction. The stratigraphic subdivisions of section SHD are described as follows from the base to the top.

Fig. 5.1.4: (A) Mean grain-size distributions of selected units from the Suohuduo on log-normal scale, (B) stratigraphy, (C) depth profiles of selected grain-size fractions (C: 0.04–2 μm; fSi: 2–6.3 μm; mSi: 6.3–20 μm; cSi b 36: 20–36 μm; cSi > 36: 36–63 μm; fS: 63–200 μm; mS: 200–630 μm; cS: 630–2000 μm) and (D) depth profiles of grain-size mean and mode.

The base of the key section SHD (Fig. 5.1.4, L1, 265 – 340 cm) is composed of yellowish silty sand and fines upwards to sandy loess at 265 cm. The mean values of 18 well-sorted samples decreases from 78 µm to 56 µm and the mode values from 80 µm to 66 µm. The overlying pale yellow unit (L03, 200 – 265 cm) consists of sandy loess and interbedded strata of several thin dark laminations between a depth of 215 cm and 235 cm.

34

Twenty-one sediment samples taken from the two paleosols S01 (90 – 120 cm) and S02 (150 - 200 cm) are poorly sorted. The mean grain-size distributions have a mode of 50 µm each and a mean of 34 µm. In addition, they show a flattening between 6 µm and 22 µm. Paleosol S02 can be divided into two subunits on the basis of the grain-size variations. Furthermore, mean and mode grain-size of unit S01 increases from the top towards the base of this paleosol (mean 27.6 µm to 39.6 µm and mode 38 µm to 60.5 µm).

The two paleosol units are separated by a pale yellow transition unit L02 (120 – 150 cm). The twelve sediment samples of this unit are better sorted and coarser than the surrounding paleosols. The mode of the mean grain-size distribution curves is 66.4 µm, and the mean is 54.8 µm.

The stratigraphic unit L01 (20 – 90 cm) above the buried soil complexes consists of yellowish sandy loess and fines upwards to the top of the section. The grain-size distribution is similar to other unweathered units below (mean: 48.8 and mode: 66.4 µm). Within this unit the mean increases between a depth of 15 cm to 75 cm (41.6 µm to 50.1 µm), while the mode is relatively constant.

The described stratigraphic units are covered by the modern topsoil (TS, phaeozem or mountain tschernozem). This unit consists of finer sandy loess (mean: 42.39 µm and mode: 60.53 µm) than that of the underlying unit L01 and is crossed by multiple plant roots.

In addition Figure 5.1.4 shows the general fining upwards of the sediments within the grain-size fractions. The clay fraction (< 2 µm) reaches values of 14 % within the two dark paleosols S01 and S02, nevertheless, there are minima of clay-content below and above the paleosols within the brighter stratigraphic subdivisions L03, L02 and L01. The fractions fine silt (2 – 6.3 µm), middle silt (6.3 – 20 µm) and lower coarse silt (20 – 36 µm) are nearly concurrent to the clay fraction (correlation coefficient R = 0.94, R = 0.85 and R = 0.65). By contrast, the correlation of the upper coarse silt fraction (36 – 63 µm) and the fine sand fraction (63 – 200 µm) with the clay fraction is negative (R = -0.66 and R = -0.85). The coarser sand fractions are underrepresented. The middle sand fraction (200 – 630 µm) is in all samples below 2 % and can be neglected. Moreover, the coarse sand fraction (630 – 2000 µm) does not exist in any sample and the predominant grain-size fraction is fine sand. In almost all samples, fine sand predominates (mean: 32 %) and has the largest variation (standard deviation: 11 %). The fine sand fraction may be considered as a proxy for moderate paleowind velocity.

Five samples were collected from the Aba section (Fig. 5.1.2, No. 2). The form and statistical parameters (mode: 42 µm; mean 30 µm) of the sediment are comparable to those of the buried sol- complexes of the SHD section, whereas the sediments of the Jiukehe section are better sorted and coarser (mode: 55 µm; mean: 46 µm). This section can be compared with the unweathered units of

35

the SHD key section. The Dari section (Fig. 5.1.2, No. 4), a sand pit close to the road, is composed of fine sand. Buried fossil soils are present at a depth of 85 cm to 145 cm.

Fig. 5.1.5: Triangle diagram of the textural classes (classification according FAO, 2006) of all Suohuduo samples and of the three comparative sections (Aba, Jiukehe and Dari) as mean values. In addition, the classification of loess, sandy loess and sand from the Donggi Cona catchment are shown.

Typical loess is distributed at elevations below 3,600 m to 3,400 m asl (sample from Aba Basin, see Fig. 5.1.2, No. 2 and Fig. 5.1.5). In alpine meadow environments at elevations > 3,600 m asl, the aeolian mantles are composed of coarser and sandy loess dominates (section Jiukehe = sandy loess from the foothills of Nianbaoyeze Shan, Fig. 5.1.1, No. 3 and Fig. 5.1.5). Dune sands are mainly present close to the Huang He, the major stream of the region (dune sand from Zoige Basin, dune sand close to Madoi, Dari (Fig. 5.1.1 No. 4) and within the Donggi Cona catchment (Fig. 5.1.1, No. 5 and Fig. 5.1.5).

36

5.1.4.2. Geochemical results of SHD section In Unit L1 (265 – 340 cm) and L03 (200 – 265 cm) the WP index and the P- and Mn-values only vary slightly (Fig. 5.1.6). Carbonate eluviations and secondary enrichment are visible in Figure 5.1.6 via the CaCO3 – content and the WP index. The Ti/Zr ratio increases at 220 cm and decreases again in the lower part of this initial soil complex.

Unit S02 (150 - 200 cm) is a well-developed paleosol and was divided into two subunits. With respect to a maximum peak of the Ti/Zr ratio and P-value in the lower subunit S02, a stronger weathering degree can be verified here. In addition the P- and Mn-values run similarly but are slightly offset. The variation in the ratios is accompanied by decreasing grain-size mean and mode in the upper part and increasing ones in the lower part of the unit.

Within unit L02 (120 – 150 cm), the Ti/Zr ratio and the WP index increase, while the grain-size contribution decreases. Furthermore, the CaCO3 – content increases from 1 % to 10 % in the upper part in about 130 cm and drops again by 5% in the lower part of the unit. This implies carbonate precipitation in the upper part and dislocation towards the lower parts. In addition, there are minimum peaks of the P- and Mn-values.

37

Fig. 5.1.6: (A) Suohuduo stratigraphy and (B) depth profiles of selected element ratios and CaCO3- content.

Unit S01 (90 – 120 cm) is another dark silty paleosol. Within this unit, the Ti/Zr ratio and the P- and Mn-values increase strongly and decrease again in the lower part of the soil complex. The WP- index increases significantly within this unit, while the CaCO3 – content decreases. There are two parallel peaks of the P- and Mn-values.

The upper part of unit L01 (20 – 90 cm) is characterized by high values of the WP-index, whereby the index strongly varies in the lower part of this unit (below 60 cm). The P- and Mn-values run predominately contrary to one another. Only when the WP-index reaches a minimum do the two elements show a parallel trend. Seen in the field, secondary carbonate precipitation at the base of this unit is also shown by the decreasing WP-index and the increasing CaCO3 content. It can be assumed that the underlying paleosol acts like a kind of barrier for leachate.

Unit TS (0 – 20 cm) is the modern topsoil. Here, the P- and Mn-values show a reverse trend. The Ti/Zr ratio decreases while the WP index increases from the top down. The WP-index shifts abruptly at the unit boundary TS/L01, and there is a minimum of CaCO3 content within and below

38

this unit. Furthermore, the topsoil shows trends of soil degradation. Nowadays the land surface is stable.

In addition to the grain-size distribution the buried soil complexes is identified by the relative trend of all selected ratios, compared to the horizons below and above. The increased Ti/Zr ratio and a low CaCO3 content indicate a phase of landscape stability with initial soil formation (Fig. 5.1.6).

5.1.4.3. Geochronological results The OSL data provides a depositional age, while the radiocarbon data for the fossil soils provides an age close to soil formation process. All the dating results show consistent ages for all stratigraphic layers except sample SHD 22. The three basal OSL ages for the SHD section (SHD 24, SHD 23, T06-3; Fig. 5.1.4 and Tab. 5.1.1) show that the sedimentation began in the early Holocene and, concerning the sedimentological and geochemical data, this is the transition from the Late Glacial L1 to the Holocene L03.

The sample SHD 22 was collected within the paleosol S02, and yields an age of 5.8 ± 0.5 ka. In comparison with the ages in the L02 (T06-2 of 6.8 ± 0.8 ka, SHD 21 of 5.8 ± 0.5 ka) above this age is reversed and too young. This age inversion may have been caused by infiltration of younger sediments by animals or along roots. T06-2 was sampled at the base of the L02 indicating an onset of fresh aeolian material covering the paleosol S02 at around 6.8 ka. This is in accordance with the sample SHD 21 at the top of the L02 with an age of 5.8 ± 0.5 ka. The radiocarbon age for the humic material in S01 is 5.56 ± 0.05 cal ka BP (Erl-10668, Tab. 5.1.2) and is consistent with the luminescence data bracketing the time of the soil S01 formation. The three topmost luminescence samples (SHD20, SHD19, T06-1) in the uppermost L01 provide ages of between 3.7 ± 0.2 ka and 3.8 ± 0.4 ka indicating minimum termination of the paleosol S01. Based on these results, it can be summarized that the main paleosol formations date from ca. 7 ka to ca. 8.5 ka (S02) and from ca. 4 ka to ca. 5.5 ka (S01). In addition, initial soil formation processes occurred in the L03 between 8.8 ± 0.8 ka and 10.7 ± 1.0 ka.

The buried fossil soils of the Dari section in 85 cm to 145 cm generate late Holocene radiocarbon ages ranging from 897 ± 47 to 3,363 ± 17 cal years BP (Tab. 5.1.2).

39

Tab. 5.1.1: Radionuclide concentrations derived from laboratory high resolution γ-spectrometry, measured water contents, calculated dose rates, equivalent doses and the resulting ages of the 125– 180 μm quartz fraction. Ages in italics are calculated with the finite mixture model after Galbraith and Green (1990), all others are with the weighted mean age model including 5% instrumental error and given with 1σ age deviation.

Depth Water Dose rate Sample De [Gy] U [ppm] Th [ppm] K [%] Age [ka] [cm] content [%] [Gy/ka] SHD 19 55 11.55 ± 2.33 ± 10.6 ± 1.97 ± 10.0 3.02 ± 3.82 ± 0.43 1.20 0.10 0.25 0.11 0.21 T06-1 86 11.37 ± 2.24 ± 10.30 ± 1.65 ± 10.0 3.00 ± 3.79 ± 0.36 0.60 0.08 0.49 0.04 0.21 SHD 20 92 10.50 ± 2.02 ± 10.2 ± 1.84 ± 10.0 2.83 ± 3.71 ± 0.26 0.39 0.10 0.25 0.12 0.22 SHD 21 120 16.95 ± 2.29 ± 11.5 ± 1.76 ± 8.8 2.94 ± 5.78 ± 0.45 0.46 0.10 0.26 0.12 0.21 T06-2 150 19.41 ± 2.20 ± 9.83 ± 1.70 ± 13.8 2.85 ± 6.81 ± 0.75 1.36 0.08 0.47 0.04 0.20 SHD 22 165 17.88 ± 2.74 ± 12.7 ± 2.00 ± 13.8 3.10 ± 5.76 ± 0.47 0.31 0.07 0.25 0.07 0.24 SHD 23 230 27.11 ± 2.68 ± 11.2 ± 1.95 ± 11.1 3.08 ± 8.80 ± 0.77 1.26 0.08 0.28 0.10 0.24 SHD 24 255 30.88 ± 2.50 ± 9.57 ± 1.97 ± 13.3 2.88 ± 10.72 ± 1.63 0.11 0.25 0.11 0.22 0.99 T06-3 260 29.42 ± 2.40 ± 10.87 ± 1.78 ± 13.3 3.00 ± 9.81 ± 0.96 1.58 0.09 0.52 0.04 0.21

Tab. 5.1.2: Results of the radiocarbon dating of section 2 (2.11), section 3 (3.3) and section 7 (7.11, 7.19).

Age BP Age BP Depth Section Sampe Lab.-No. Material (uncalibrated (calibrated Calender years [cm] years BP) years BP) Suduhuo P 5-12 Erl-10668 Humic acid 110 4814 ± 25 5545 ± 47 cal BC 3595 ± 47 Dari P 11-11 Erl-10669 Humic acid 160 3127 ± 24 3363 ± 17 cal BC 1413 ± 17 Dari P 11-17 Erl-10670 Humic acid 140 1153 ± 28 1069 ± 56 cal AD 881 ± 56 Dari P 12-12 Erl-10671 Humic acid 85 996 ± 30 897 ± 47 cal AD 1053 ± 47 Dari P 12-17 Erl-10672 Humic acid 110 1130 ± 29 1030 ± 35 cal AD 920 ± 35

40

5.1.5. Discussion 5.1.5.1. Late Glacial to early Holocene (15 ka to 8.5 ka) The SHD sections show at the basis a sedimentation of coarser fine sand to coarse silt-sized particles (L1; > 10 ka, Tab. 5.1.1, Fig. 5.1.4, 5.1.7). This is an indication for higher wind speed in the Late Glacial. There is a fining upwards towards the L03 in the Holocene. The timing of this sediment sequence is similar to those recorded by other studies for example from the CLP (e.g. Jia et al., 2011; see Fig. 5.1.7) and the beginning of aeolian sedimentation in the steppe regions of northern Mongolia (Lehmkuhl et al., 2011, 2012). Using pollen data from the nearby Nianbaoyeze Shan, Schlütz and Lehmkuhl (2009) showed that there was a generally more open landscape during this time. This pioneering phase of the vegetation combined with simultaneous local loess sedimentation (fed by dust from riverbeds) seems to be characteristic of the early Holocene in eastern Tibet (Owen et al., 2003, 2006; Küster et al., 2006; Lu et al., 2011; Stauch et al., 2012).

Fig. 5.1.7: Comparison of paleoenvironmental reconstructions by this study with other results from the Donggi Cona, the Chinese Loess Plateau and the Qinghai Lake.

41

The aeolian mantle in L03 shows an initial soil formation in this period, thus indicating slightly wetter climate conditions than in L1 but more arid conditions than those of the mid-Holocene, which was predominantly humid. This is not in accordance with the highest lake level of Qinghai Lake (Fig. 5.1.7). However, this high lake level might be attributed to the melting of glaciers and permafrost (An et al., 2006) as well as an overall reduced evaporation during that relatively cold period.

5.1.5.2. Mid-Holocene (8.5 to 4.0 ka) The mid-Holocene is characterized by the first paleosol S02 dated to ca. 8.5 ka to 7 ka and a second paleosol S01 dated to 5.5 ka to 4 ka and interrupted by a period of sandy loess accumulation to form L02 at around 6 ka. Several other buried A-horizons from other regions, such as the CLP could be dated to the mid-Holocene (Huang et al., 2002; Jia et al., 2011; Li et al., 2012), even though the time windows for soil formation as set by the over- and underlying luminescence and radiocarbon ages, are quite large. As Figure 5.1.6 illustrates, the soil complex S02 is characterized by two significant peaks of the WP-index. The phase of steppe vegetation and sedimentary stagnation is interrupted by a phase of clastic sedimentation, suggested by a minimum peak of the WP-index. The soil complex is an almost climax-like soil formation of the Eurasian steppe. According to the flattening of the grain-size distribution between 6 µm and 22 µm (see Fig. 5.1.4, S01 and S02), fixation and immobilization of the fine sediments caused by warmer and moister environmental conditions and vegetation covering is assumed. This soil formation period can be associated with higher humidity resulting from an intensifying monsoonal system also indicated by higher lake levels of the Qinghai Lake (Liu et al., 2012; An et al., 2006 – see Fig. 5.1.7). An et al. (2006) reported humidity changes in the arid and semi-arid regions in China and pointed out that this mid-Holocene moisture period ended at about 4 ka when the climate became drier. The parallel peak of P-and Mn-values in the S01 can be attributed to forest or steppe fires in the environment triggered by either natural (wildfires) or human (slash and burn) effects (Eckmeier et al., 2007; Lehmkuhl et al., 2011; Misra et al., 1993). Thus, it is assumed that this had been a typically more arid steppe soil (dark kastanozem, phaeozem or even tschernozem type).

Summarizing, we suggest that these phases of soil development during the mid-Holocene represented periods of stable land surfaces due to dense vegetation cover indicating relatively moist climate conditions. According to the parallel P- and Mn-value trends within the paleosol S01, we assume that steppe fires occurred repeatedly (P- and Mn-values) due to fluctuations in weather conditions. Some elements of the burned plants were leached out and shifted vertically. The CaCO3 enrichment within the L02 acted like a chemical element barrier. Moreover, the WP-values indicate

42

a vertical transport through the S01 and an enrichment of the eluted elements above the L02 unit. The mid-Holocene period of soil formation is in accordance with humid periods as shown by high lake level stands in the Donggi Cona (Opitz et al., 2012) or the Qinghai Lake (An et al., 2006, see Fig. 5.1.7), thus indicating wetter climate conditions at around 8 ka to 7 ka and 5 ka to 4 ka. OSL ages or reworked sediments from the Donggi Cona area (Fig. 5.1.7) indicate a phase of enhanced fluvial erosion from 9 ka to at least 5.7 ka that is also attributed to more humid conditions (Stauch et al., 2012). Similar ages of reworked sediments are also documented for other areas on the NE Tibetan Plateau (e.g. Owen et al. 2003; Kaiser et al. 2007, 2009).

Most researchers (e.g. Chen et al., 2008 and references therein), conclude that especially mid- Holocene vegetation changes stem from climatic instead of human impact. However, Miehe et al. (2007) and Schlütz and Lehmkuhl (2009) argue that humans have greatly influenced forest decline and the spreading of sagebrush and pastures supposedly since 6,000 years but at least since 4,500 years.

5.1.5.3. Late Holocene (4.0 – 0 ka) The particle-size distribution (Fig. 5.1.4) and several elements (Fig. 5.1.6) show the transition from the paleosol S01 to the sedimentation of sandy silt L01 in the upper part of SHD section, thereby indicating stronger aeolian activity. A continuous decrease in vegetation or change by fire and/or grazing may have affected the mobilization of coarser material. There are similar results from the Donggi Cona, where the sand dunes were remobilized from 3 ka until present (Stauch et al., 2012). This is in accordance with pollen and other records from the area indicating less humidity (e.g. Herzschuh, 2006; Colman et al., 2007). The decrease in lake levels in China (An et al., 2006; Chen et al., 2008) clearly verifies also an increased aridity in the environment. In addition, Kaiser et al. (2009), Huang et al. (2009), Jia et al. (2011) and Schlütz and Lehmkuhl (2009) contend that there was a stronger human impact since 3 ka. The buried paleosols in the Dari section at the transition from the mid-Holocene to the late Holocene (~ 3.3 ka) and around 1 ka indicate higher wind speed and accumulation of fine sand in a more open landscape interrupted by moderate soil formation processes with higher vegetation cover. A similar soil might be fragmentarily preserved in the L01 unit below 60 cm according to the WP-index. For eastern Tibet, a cooling period starting at ~2 ka and caused by periglacial mass movements demonstrates increased surface activity especially during the Little Ice Age (Schlütz and Lehmkuhl, 2009). In addition, this proves a stronger fluvial sedimentation and an increase in erosion and slope wash starting already at ~4 ka with the onset of the Neoglacial period as well as stronger grazing influence.

5.1.6. Conclusion 43

Aeolian sedimentation in the study area began during the Late Glacial with the deposition of sandy loess. This concurs with the beginning of aeolian sedimentation on the eastern margin of the Tibetan Plateau and also in the steppe environments in Mongolia. This Late Glacial coarser material (L1/ L03) of sandy silt shows a fining upwards that started during the early part of the Holocene. The aeolian mantle in L03 shows an initial soil formation indicating slightly wetter climatic conditions. The mid-Holocene paleosols S02 and S01 suggest a more dense vegetation cover during this time, likely caused by higher humidity and being most distinct and widespread beginning at around 8 ka to 7 ka. The second Holocene paleosol S01 from ~5.5 ka to 4 ka provides evidence for (steppe) fires. The sedimentological, geochemical and geochronological results suggest several phases of landscape stability as well as active phases of aeolian accumulation. An et al. (2006) and Chen et al. (2008) also showed moisture climatic condition during these time periods using lacustrine records. These periods of soil formation were then followed by accumulation of sandy silt in the late Holocene, indicating an increase in aeolian accumulation and slope wash in a more open landscape caused by higher aridity and more grazing activities. The local dust source for this section are the braided river systems and the periglacial environment during drier and cooler climate periods with higher aridity and sparse vegetation cover. Furthermore, especially during the late Holocene, dust emission might have been enhanced by overgrazing and localized soil erosion. This study shows that even such relatively short loess sections provide valuable information concerning the paleoclimatic and landscape evolution. Coarser sediments and silt size particles indicate higher wind speeds and steppe environments with the accumulation of dust interrupted by more stable and moisture conditions with different soil formation processes during the Holocene.

44

5.2. Environmental change indicated by grain size variations and trace elements: examples from two different sections - the sandy-loess sediments from the Doroshivtsy site (Ukraine) and the loess section Semlac (Romania)

Philipp Schultea, Frank Lehmkuhla, Holger Kelsa, Christa Loibla, Nicole Klasenb, Thomas Hauckc

aDepartment of Geography, RWTH Aachen University,52056 Aachen, Germany, bInstitute for Geography, University of Cologne, 50923 Cologne, Germany, cInstitute for Prehistoric Archaeology, University of Cologne, 50923 Cologne, Germany

Published 2014 in DUST2014 - 1st International Conference on Atmospheric Dust, 106–112

Abstract Loess sequences provide important and at least a partial continuous record of Quaternary paleoenvironmental change; some of the sequences even bury archaeological remains. In addition, loess-paleosol sequences provide valuable information concerning environmental change and climate evolution. In this study, we compare two sections: (1) The Middle to Late Pleistocene loess- paleosol section of Semlac in western Romania (MIS 10 – 1), and (2) the sandy-loess section of Doroshivtsy in western Ukraine (MIS 2) (Fig. 5.2.1). To characterize these quite different sections we calculated individualized grain size (GS) ratios and compared them to the common U-ratio (Vandenberghe et al.,1985) and selected geochemical parameters.

45

5.2.1. Regional Setting

Fig. 5.2.1: Loess distribution in Europe (modified according to Haase et al., 2007), maximum extent of the Weichselian ice sheet, coastline during LGM and location of the sections Doroshivtsy in the western Ukraine and Semlac in south-western Romania.

5.2.1.1. Western Romania: the section Semlac The loess section Semlac (46°7'12.97"N / 20°56'54.70"E / ~100 m a.s.l.) is situated in the Arad Plain (Câmpia Aradului) in western Romania at an undercut slope position on the right bank of the Mureş River (Fig. 5.2.1). The more than 10 m thick section contains four main paleosol complexes developed in very homogenous loess with high silt content without any major discordance. A first chronological model which is based on luminescence dating and rock magnetics suggests an age from the MIS 10 to MIS 1 (Kels, 2012; Kels and Hambach, pers. comm.).

5.2.1.2. South-western Ukraine: the section Doroshivtsy The section Doroshivtsy (025°52’10.7’’E, 48°35’37.6’’N, Fig. 5.2.1) is situated in the south-western part of the Ukraine at the right bank of the middle course of the Dniester River. Bearing seven Upper Paleolithic levels in its lower part, the Doroshivtsy sequence is archaeologically important

46

as it documents repeated use of the site by Gravettian hunter-gatherers during the Last Glacial Maximum (LGM) (Kulakovska et al., 2012). The section represents a more than 9 m sequence of sandy loess with intercalated weak humic horizons. It is situated in a flat gully at an undercut slope of the Dniester River close to the village Doroshivtsy. The less steep slope of the terrace-covering sandy loess indicates sediment accumulation of slope material in this position; therefore the section most probably represents a combination of formerly aeolian transported loess and of slope-wash material.

5.2.2. Methods The particle size was measured with a Laser Diffraction Particle Size Analyzer (Beckman Coulter LS 13 320 PIDS) by calculating the mean diameters of the particles within a size range of 0.04 - 2000 μm with an error of 2 %. To remove the organic matter, the samples were treated with 0.70 ml 30 % H2O2 at 70 °C for several hours. To keep particles dispersed, the samples were treated with 1.25 ml Na4P2O7 x 10 H2O for 12 hours. The Mie theory was applied to determine the grain- size distribution (Fluid RI: 1.33; Sample RI: 1.55; Imaginary RI: 0.1). To determine the element concentrations of the fine-grained fractions, the <63-μm fraction was sieved and dried at 105 °C for 12 hours. An 8 g-quantity of the sieved material was mixed with 2 g Fluxana Cereox, homogenized and pressed to single pellets with a pressure of 20 t for 120 s. To determine the element concentrations by x-ray fluorescence, a Spectro Xepos was used. Every sample was measured twice. Mean values were calculated from the two measurements. To characterize the sediment layers of the two sections we calculated a section specific GS-ratio respectively (0.04 to 5.88 µm / 11.29 to 26.15 µm for Semlac and 3.5 to 8.1 µm / 69.6 to 161.1 µm for Doroshivtsy). In order to ensure that each parameter of the ratio represents just one sedimentary or post sedimentary process, we narrow down the respective GS-range to the GS-classes which show constant variations with depth. For reasons of comparability we calculated the U-ratio of 44 to 16 µm versus 5.5 to 16 µm (Vandenberghe et al., 1985).

5.2.3. Results 5.2.3.1. Semlac

For the 10.70 m thick sequence, 10 different main units can be distinguished (Fig. 5.2.2). The documented loess body contains four loess-paleosol-complexes with individual features. Generally, the loess is comparably homogenous with a distinctive dominance of silt for the whole section, higher amounts of clay in-between the paleosols and two major events with an increase of sand.

47

The lower part of the sequence from 7.10 m on was decalcified. The section is labelled in the loess units from L1 to L4 and in the fossil soil complexes from S0 to S3 following the Serbian loess classification (Marković et al., 2008, 2009). The individual GS-ratio shows clear variations between the weakly weathered loess units and the paleosol complexes (Fig. 5.2.3). The numerator of the ratio appears to be related to pedogenic processes and particularly to the relocation and new formation of clay minerals. The denominator is obviously related to the strength of the loess accumulation (Fig. 5.2.2). In general, the Semlac-specific GS-ratio is lower in the loess sequences (especially in L1L1, L2, L3 and L4) and higher in the paleosols (S0, S1, S2, S3). The L1S1 paleosol, which was formed during MIS 3, is an exception; there is no distinct difference to the surrounding loess units. The results of XRF analysis showed comparable results to the GS-ratio. In particular, the Al/Na ratio which is shown in Fig. 5.2.2 showed a similar curve progression as a function of the loess units and the paleosol complexes.

Fig. 5.2.2: Lithostratigraphy, sensitive grain size ranges and selected element concentrations of the Semlac section. OSL dating was carried out on silty quartz samples (grain size 40-63 µm) and a standard SAR protocol (Murray and Wintle, 2003). The central age model was used for De calculation.

48

Fig. 5.2.3: Lithostratigraphy, specific GS-ratio, and U-ratio of the Semlac section.

5.2.3.2. Doroshivtsy

Based on the field description and the grain size distribution the profile can be divided into four main genetic units (I-IV) representing changes during the deposition (Fig. 5.2.4). Part I (1-3.2 m) of the profile is characterized by aeolian loess and sand deposition intercalated with a few gravel lines. Part II (3.2-6.1 m) of the profile is influenced by aeolian, denudative and weak soil forming processes. Part III (6.1-7.8 m) of the section represents a combination of aeolian loess and re-

49

deposited slope material. Part IV (7.8-9.1 m) is mainly composed of clayey and silty layers of a tundra gley. The whole parts III and IV are affected by hydromorphic influences. As most of the sediment is rather homogeneous sandy silt the U-ratio did not show any distinct variations. However, the Doroshivtsy-specific GS-ratio shows clear peaks which represent environmental changes (Fig. 5.2.5). These variations of the Doroshivtsy-specific GS-ratio represent environmental changes which were also observed by structures and weak soil formations in the section during field work.

In addition, geochemical analysis show comparable results to the GS-ratio and provide further evidence for the differentiation of the stratigraphic units (Fig. 5.2.4). Summarizing we can detect 4 different main units and 11 sub-units which are related to paleoclimatic and environmental conditions.

Fig. 5.2.4: Lithostratigraphy, sensitive grain size ranges, and Al/Zr ratio of the Doroshivtsy section.

50

Fig. 5.2.5: Lithostratigraphy, specific GS-ratio, U-ratio, Sedimentology, OSL ages, and Radiocarbon ages of the Doroshivtsy section. OSL dating was carried out using OSL and pIRIR stimulation (Klasen et al., 2015).

Radiocarbon and luminescence ages from the lower part of the section were conclusive within the expected age range of the Gravettian period (22-28 ka). However, optically stimulated luminescence (OSL) ages of quartz minerals were younger than radiocarbon and post infrared stimulated luminescence ages (pIRIR, measured at 290 °C). This indicated sediment relocation for this part of the profile (Klasen et al., 2015). The loess section is composed mainly of sandy silt and covers the time span from about 26 to 16 ka. This is one of the very few sections in Europe which provides a high resoluted sedimentary record including traces of human occupation during the cooling maximum of the last glacial cycle.

51

5.2.4. Conclusion The two loess-paleosol sequences Doroshivtsy and Semlac differ especially in the geomorphological situation, the grain size composition and the temporal resolution. Therefore, both the GS composition of the individual units as well as the degree of weathering of the preserved buried sol-complexes is different.

In Semlac the natural outcrop is exposed at a plateau (Arad Plain) with likewise horizontal layers. The results from grain size measurements show that there was a very continuous accumulation of dust that took place under highly consistent circumstances of accumulation (source area and wind direction) since the MIS 10. During interglacial and interstadial periods well developed paleosol complexes were formed. The GS-variation is mainly influenced by post-depositional processes (clay mineral relocation and -formation).

The Doroshivtsy section was developed with a high sedimentation rate in a relatively short period of time (MIS 2) on a terrace step in beneath the Canyon of the Dniester. The sedimentation was affected by both, aeolian and slope wash processes. The section was partly influenced and modified by hydromorphic and cryogenic processes. During short phases of warming some initial paleosols were formed. In addition there are some layers of reworked soil material, which were eroded probably at the transition to dryer periods from the slopes. Essentially the Doroshivtsy section is dominated by sedimentary processes. Hence, GS and geochemical data show a high correlation.

For the calculation of a specific GS-ratio the ranges of the entire GS spectrum should be selected which react as sensitively as possible on individual processes. The advantage of the individualized GS-ratios compared to the U-ratio is their independency of firm boundaries and their higher sensitivity to individual sedimentary or post-sedimentary processes.

52

5.3. Influence of HCl pretreatment and organo-mineral complexes on laser diffraction measurement of loess-paleosol sequences

Philipp Schultea, Frank Lehmkuhla, Florian Steiningerb, David Loibla, Gregori Lockotc, Jens Protzea,d, Peter Fischere, Georg Staucha a Department of Geography, RWTH Aachen University, Templergraben 55, D-52056 Aachen, Germany b Institute for Geography, University of Cologne, Albertus-Magnus-Platz, D-50923 Cologne, Germany c Institute of Geographical Sciences, Free University of Berlin, Maltesertraße 74-100, D-12249 Berlin, Germany d Division of Academic and International Affairs, University of Konstanz, Universitätsstraße 10, D-78457 Konstanz, Germany

Published 2016 in Catena, 137, 392–405

Abstract The influence of different sample pretreatment methods on grain size distributions in particle size analysis has been subject to controversial discussions. Standard sample preparation typically comprises the disaggregation of aggregated and agglomerated particles into single primary particles, i.e., the organic binding material is oxidized by hydrogen peroxide (H2O2) and the contained carbonates are dissolved by hydrochloric acid (HCl). The aim of this study is to evaluate the effects of HCl treatment on grain size analysis of Late Pleistocene and Holocene loess-paleosol sequences investigated by a Beckman Coulter LS 13320 laser particle analyzer. For this purpose, samples from two different sections with different weathering degrees and sedimentary genesis were measured: (1) the Suohuoduo section on the northeastern fringe of the Tibetan Plateau (China) containing loess and paleosols, and (2) a vibracore from Düsseldorf-Grafenberg (Germany) containing calcareous loess and intercalated interglacial, interstadial and periglacial soils and soil sediments. All samples were pretreated with hydrogen peroxide and sodium pyrophosphate. Subsequently, the samples were prepared with and without the addition of HCl. There is no significant association of the HCl-induced grain size modifications after the HCl treatment with the calcium carbonate content. Conversely, a distinct dependence of the modification of grain size distributions on the content of organic matter, the weathering degree of the sediment, and the presence of stable

53

aggregates as well as organo-mineral complexes was observed. Consequently, pretreating post- depositionally modified aeolian sediments with HCl may result in misleading grain size distributions and should be avoided in standard analysis of loess-paleosol sequences. However, the HCl-induced modification of grain size distributions provides an indication of the existence of stable aggregates or organo-mineral complexes.

5.3.1. Introduction The application of laser-diffraction-based particle size analysis is widespread in earth sciences. Large amounts of grain size data can be generated in a short time and with satisfying precision and accuracy, provided that the method is applied properly (Blott and Pye, 2006). In this context, sample pretreatment is often applied to reconstruct the characteristics of the particles at the time of accumulation, typically by dissolving aggregates and removing binding agents (Murray, 2002). Common guidelines for the application of sample pretreatment are, however, lacking. This is particularly problematic since knowledge regarding the influence of these procedures on the measurement results is still patchy (Lu and An, 1997; Vaasma, 2008).

Previous technical studies associated with laser granulometric data are commonly investigating medium- to coarse-grained sediments, e.g., sand or gravel, which are mostly poor in carbonates and organic matter (OM) and hardly tend to aggregation (Chappell 1998; Blott and Pye, 2006; Storti and Balsamo, 2010; Treadwell-Steitz and McFadden, 2000; Vandenberghe, 2013). Additionally, several studies on very fine grained sediments in the clay and fine silt fractions have been conducted (Konert and Vandenberghe, 1997; McCave et al., 1986; Murray, 2002).

High resolution grain size data obtained from Pleistocene and Holocene loess-paleosol sequences (LPS) are of increasing significance in paleoclimate research (e.g.,Vandenberghe, 1997; Antoine et al., 2009a; b; Kaiser et al., 2009; Machalett et al., 2008; Sun et al., 2006; Prins et al., 2007; Nugteren et al., 2004). These sediments typically consist of a mixture of clay-, silt- and sand-sized particles. Particularly the fine materials tend to aggregate, resulting in larger flocks with varying grain size and resistance (Schulten and Leinweber, 2000; Balabane and Plante, 2004; von Lützow et al., 2008; Six et al., 2000, 2004; Stamati et al., 2013). In addition, close interactions between mineral surfaces and organic matter (OM) within soils are evident (Tiessen et al., 1984; Oades 1988; Wiseman and Püttmann, 2006; Kögel-Knabner et al., 2008). The sorption of OM predominantly occurs on minerals with large specific surfaces, i.e., minerals from the clay and fine silt fractions. Clay minerals and other phyllosilicates are associated with OM as clay-humic complexes, as iron (hydr-)oxides, or as organo-metallic complexes (hereafter collectively referred to as organo-mineral complexes). As a consequence of the occlusion in microaggregates and organo-mineral complexes, the OM is 54

protected from biological and chemical degradation or oxidation (von Lützow et al., 2006; Wiseman and Püttmann, 2006; Bachmann et al., 2008; Kögel-Knabner et al., 2008; Eckmeier et al., 2010). The amount of stabilized oxidation-resistant OM increases with increasing age and depth within the sediment section (von Lützow et al., 2006; Kögel-Knabner et al., 2008).

In order to reconstruct the characteristics of the particles at the time of accumulation and to dissolve post-depositionally formed aggregates (mainly due to soil formation processes and weathering), the binding agents have to be removed (Murray, 2002). Typically, OM is destroyed by hydrogen peroxide (H2O2; e.g., Rowell, 1994) and calcium carbonate (CaCO3) is dissolved by hydrochloric acid (HCl; e.g., Konert and Vandenberghe, 1997). The timing and relevant sedimentological processes of aggregate formation are, however, still under debate. Menendez (2014) concluded that carbonates and clay minerals are partly transported and deposited as aggregates. Pesci (1990) suggested that the aeolian accumulation of siliceous deposits and the deposition of detrital carbonate in form of particulate crystals and amorphous powder are occurring simultaneously. Smalley (1971, 1978, 2011) postulated that the precipitation of carbonates occurs subsequent to the accumulation of siliceous particles, resulting in incrustations or coatings on quartz grains and pore fillings. Nevertheless, the HCl treatment removes all carbonates, including the primary allochthonous material that was part of the sediment during the accumulation process (Machalett et al., 2008; Antoine et al., 2009a). Furthermore, HCl treatment not only dissolves carbonates (CaCO3 and MgCO3) but also small amounts of Corg (Midwood and Boutton 1998; Harris et al., 2001; Schumacher, 2002; von Lützow et al., 2007) as well as pyroxenes, amphiboles, clay minerals and metal (hydr-) oxides (Petersen et al., 1966; Schwertmann et al., 1987; Scheffer et al., 2002; Vaasma, 2008; Li et al. 2008). However, by treatment with H2O2, most of the carbonate- bound aggregates are already dissolved (Pingitore et al., 1993; Allen and Thornley, 2004; Mikutta et al., 2005). This dissolution is increased further by adding Na4P2O7 for dispersion of the particles during the measurement process (Mikutta et al., 2005). In contrast to removing OM, which is conducted in almost every study, decalcification is not a standardized procedure and often omitted in the context of grain size analysis (Buurman et al., 1997a; Beuselinck et al., 1998; Chappell, 1998; Allen and Thornley, 2004; Buurman et al., 2004; Eshel et al., 2004; Pye and Blott, 2004; Zobeck, 2004; Özer et al., 2010). Conversely, some studies apply HCl-treatment to the entire sample set, regardless of the origin and the type of the carbonates therein (Konert and Vandenberghe, 1997; Buurman, 2001; Sun et al., 2006; Van der Veer, 2006; Kaiser et al., 2009; Daut et al., 2010; Dietze et al., 2012; IJmker et al., 2012; Qiang et al., 2013). The effect of different pretreatment methods including the HCl treatment on grain size (GS) measurement has been analyzed by Lu and An (1997) on loess paleosol samples and by Vaasma (2008) on lacustrine sediments. According to Lu and An (1997) different pretreating methods can provide great differences in the GS results of 55

loess paleosol samples. Vaasma (2008) concluded that pretreatment with HCl and H2O2 is the fastest method for the oxidation of OM.

A quantitative assessment of the effects of HCl pretreatment on GS distributions is, however, still lacking. Due to the specific GS distribution of primary loess and the aggregations during pedogenesis, the analysis of LPS allows the investigation of the effect of the acid pretreatment on both GS results and aggregate stability under comprehensible conditions.

Fig. 5.3.1: Stratigraphy and location of the Suohuduo section at the northeastern margin of the Tibetan Plateau (Lehmkuhl et al., 2014).

The aim of this study is to evaluate the effects of HCl treatment on grain size analysis of Late Pleistocene and Holocene LPS investigated by a Beckman Coulter LS 13320 laser particle analyzer. For this purpose, two sections from different environments, representing widely different sedimentation regimes and stages of weathering and soil development, are evaluated to disentangle

56

the effect of sample preparation with HCl. The Suohuoduo section (SHD; 34°28’ N, 99°50’ E, Fig. 5.3.1) is located on the northeastern margin of the Tibetan Plateau (China). It consists of loess and paleosols which have accumulated and developed during the late Pleistocene and Holocene (Lehmkuhl et al., 2014).

Fig. 5.3.2: Stratigraphy and location of the Düsseldorf-Grafenberg coring situated at on the right side of the Rhine River.

The second archive is a vibracore from Düsseldorf-Grafenberg (DDG; 51°14´49´´N, 6°50´58´´E; 83 m asl; Fig. 5.3.2) on the right side of the Rhine River (Germany). It contains both unweathered and weakly weathered loess deposits which have accumulated during cold and semiarid conditions of the last glacial cycle. The grain size distribution of each sample was measured twice; once on a 57

subsample pretreated with HCl, once on a subsample were this process was omitted. The differences between both measurements were then analyzed to quantify the effect and disentangle the relevant processes.

5.3.2. Material and methods Grain size analysis of all investigated sediment and soil samples was conducted using a Beckman Coulter LS 13320 (Beckman Coulter GmbH, Krefeld, Germany) laser diffraction particle size analyzer, equipped with an aqueous liquid module (ALM) and a Polarisation Intensity Differential Scatter (PIDS unit). This combination of techniques allowed establishing grain size distribution curves composed of 116 logarithmic grain size classes, covering a range of 0.04–2000 µm. The main diffraction system uses a 5 mW monochromatic laser diode with a wavelength of 780 nm. The light source of the PIDS unit is a tungsten-halogen lamp which is transmitted in three wavelengths (450 nm, 600 nm and 900 nm) through a horizontal and vertical polarizer (Beckman Coulter, 2011).

In accordance with the standard ISO 13320 (2009), the absolute measurement accuracy of the LS 13320 is within acceptable limits if the deviations between measured and certified values of reference material do not exceed 3 % in the range between the 10th and the 30th percentile of the entire GS distribution (D10 to D30), 2.5 % in the range of D30 to D70 and 4 % in the range of D70 to

D90. The accuracy of the LS 13320 used in this study is checked by periodic test measurements using 15 µm Garnet beads (Coulter LS CONTROL G15) and 0.3 µm latex beads (Coulter LATRON 300LS). In addition, the accuracy of the device was certified by an international collaborative study of the BAM Federal Institute for Materials Research and Testing (RV BAM- 5.5-2014). The measurement accuracy varies between different instruments (Singer et al. 1988; Syvitski et al. 1991; Etzler and Sanderson 1995, Pye and Blott, 2004, Roberson and Weltje, 2014). However, it is impossible to evaluate the accuracy in case of natural sediments with a large portion of fine grains (Roberson and Weltje, 2014). Therefore, measurement precision is more important than the absolute accuracy in the context of particle size analysis. Concerning the International Standard (ISO 13320, 2009), the precision for reference materials with a D90/D10 ratio of 1.5 to 10 is below 3 % for the median value and below 5 % for D10 and D90 values.

Additionally, measurement precision increases with better sorting of the samples, mostly resulting from inevitable deviations from the original sample’s composition when taking multiple aliquots (Pye and Blott, 2004). Blott and Pye (2006) found that the precision is higher than 1.5 % for poorly sorted sediment samples and much higher than 1 % for standard material and well sorted sediment samples. 58

To evaluate material dependence regarding the repeatability of single aliquot measurements, the precision of one representative sample from each of the pedogenic units (loess, modern top soil and paleosol) was calculated in accordance to ISO 13320 (2009; Tab. 5.3.1). The precision is expressed by the coefficient of variation (CV; i.e., the standard deviation divided by the mean, reported as a percentage). To distinguish between effects originating from measurement uncertainties and modifications induced by the HCl treatment, the mean SD of the aliquot measurements was calculated for each of the 116 GS classes for all the samples (Fig. 5.3.5).

Tab. 5.3.1: Precision of particle size measurements of a modern top soil, a paleosol and a loess sample of the SHD section. Precision was assessed by four aliquots of the same bulk sample and is gives as % CV.

mean mode median D10 D90

modern top soil 0.51 0.00 0.98 1.23 0.27

paleosol 3.61 4.89 6.28 10.62 2.98

loess 0.15 0.00 0.07 2.66 0.27

5.3.2.1. Sample Preparation Prior to the analysis all samples were dried at 35 °C, homogenized and passed through a 2 mm sieve. To evaluate the effects of HCl, each sample was split into two subsamples (0.3 g each) out of which one was treated for 12 hours with 10% HCl and the other was not. Prior to the addition of HCl, the subsample was suspended with deionized water (0.3 g sediment:1.5 ml aqua dest.). All samples were treated with 0.15 ml of HCl. In cases where the CaCO3 content of the sample was higher than 2%, further 0.015 ml HCl were added for each percent CaCO3. The GS distributions of all selected subsamples were measured and compared. In addition, the weaker acetic acid

(CH3COOH, 10%) was applied to some subsamples to evaluate the effect of this alternative solving agent on GS distribution in comparison to the treatment with HCl. Subsequent to the acid treatment, the soluble compounds were washed out by repeated centrifugation.

All samples were treated with 0.70 ml 30 % hydrogen peroxide (H2O2) at 70 °C for several hours to remove the OM (ISO 11277, 2002). Dry subsamples were suspended with 0.70 ml aqua dest. prior to the addition of H2O2. The treatment was repeated until a bleaching of the sediment occurred (Allen and Thornley, 2004). Treatment was stopped after the sixth run at latest. 59

The presence of aggregates and the formation of flocculation are major obstacles for robust measurements of the suspension. Flocculation can be avoided by various methods. The admixture of calgon, sodium hexametaphospate, tetra-sodium pyrophosphate or similar substances disperses the suspension (Chappell, 1998; Kaiser and Guggenberger, 2003). Alternatively or additionally, an ultrasonic treatment may be performed immediately before the measurement (Konert and Vandenberghe, 1997; Lu and An, 1997; McCave et al., 2006). However, the ultrasonic treatment may also yield undesired effects, including formation of air bubbles, breakup of quartz grains, or immediate reaggregation (Chappell, 1998; Machalett et al., 2008; Asano and Wagai, 2013).

In this study, the samples were solely dispersed with distilled water and dissolved tetra-sodium pyrophosphate (Na4P2O7 * 10 H2O), applied for at least 12 hours in an overhead shaker (Pye and Blott, 2004; DIN ISO 11277, 2002).

5.3.2.2. Laser diffraction measurement Two to four aliquots per sample were weighted in plastic test tubes and filled into the ALM container by an auto-prep station, facilitating equal measuring conditions. The amount of subsample aliquots depended on the material properties and how the material was affected by light. For dark/fine and bright/coarse material subsamples, 0.1–0.5 g and 0.5–2 g were used, respectively (Zobeck, 2004). While the suspension circulates through the sample cell, every second a measure value is generated to ensure that all particles are detected. After 90 s an average is being calculated from the individual grain size distributions. Each aliquot was measured twice to improve accuracy.

The light intensity adsorbed by the suspension is measured as obscuration and indicates the ratio of subsample and fluid. The LS 13320 software provides two values determining the level of obscuration, one for the diffraction module (total obscuration) and one for the PIDS module. Values of 8–12 % are satisfactory for the total obscuration. Even though PIDS obscuration values of 40–60 % are advised, higher values are acceptable in favor of the more important total obscuration in cases of wide grain size distributions (Beckman Coulter, 2011). In the next step, runs which met the obscuration criteria were averaged.

Finally, the results are provided as a distribution density curve. The lower limits of the calculated 116 GS classes are plotted on the x-axis given in µm. On the y-axis, the grain size distribution is plotted as volumetric distribution density curve. The curve is based on percentage frequencies of the respective GS-classes referring to the volume of the aliquot (Roberson and Weltje, 2014).

5.3.2.3. Optical model

60

There are two different common optical models to calculate the grain size distribution from the diffraction pattern: the Fraunhofer diffraction theory and the Lorenz-Mie theory (hereafter “Mie theory”; de Boer et al., 1987). The underlying light scattering theory assumes that the angle dependent scattering from natural sediment is similar to the scattering pattern of homogenous spheres (ISO 13320, 2009; Lagasse and Richards, 2003). According to the Fraunhofer theory, some of the light hitting a particle is diffracted in a grain size dependent angel. The angle of diffraction increases with decreasing grain size of the particles (Blott and Pye, 2006). Contrary to the Fraunhofer theory, which solely bases on diffraction at the particle surface, Mie theory also considers complex refractive indices of the particles and the suspension fluid (ISO 13320, 2009).

A user specific Mie model can be implemented in the device by defining three parameters: the refractive index of the fluid (here fluid RI: 1.33), the “real” component of the sample refractive index (here sample RI: 1.55) and the “imaginary” component of the sample refractive index (here sample AI: 0.01) (Buurman, 1997b; Murray, 2002; Özer et al., 2010).

Provided that perfect spheres are measured, the angle-dependent scattering pattern calculated by the complex Mie theory approaches the pattern calculated by the Fraunhofer theory with increasing grain size. Specifically, the angule-dependent scattering at particles larger than ten times the lightsource wavelength (in the case of the Beckman Coulter LS 13320: 10*780 nm = 7.8 µm) is virtually independent of the optical properties of the sediment particles (McCave and Syvitski, 1991; Bayvel and Jones, 1981; Lagasse and Richards, 2003; Loizeau et al., 1994). In this study, Mie theory was used exclusively owing to the large portion of particles < 8 µm.

5.3.2.4. Content of CaCO3 and Corg Total carbon content was quantified by thermal conductivity measurements using an EuroEA3000

CHNS analyzer (HEKAtech GmbH, Wegberg, Germany). The content of CaCO3 was determined by the volumetric Scheibler approach (Leser, 1977; Schaller, 2000; ISO 10693, 1995). Total Corg contents are commonly determined by loss on ignition (Dean, 1974). However, this method is problematic in calcareous environments, possibly yielding overestimations (Froelich, 1980; Nieuwenhuize et al., 1994). Consequently, the CHNS results for total carbon and Scheibler results for inorganic carbon were used to determine the contents of Corg by difference. Nieuwenhuize et al. (1994) showed that this approach yields reliable and reproducible results, particularly for carbonate-rich sample material.

5.3.2.5. Inductively coupled plasma-optical emission spectrometry

61

After HCl dissolution and centrifugation, the supernatant was decanted to analyze the element concentration within the HCl-leachable fraction of six selected samples. The supernatant of the soil-solution suspension (1g soil:5.5 ml solution) was measured by an Optima 2000 (PerkinElmer LAS GmbH, Rodgau, Germany) inductively coupled plasma-optical emission spectrometry (ICP- OES) (c.f. Cazes, 2004). For a seimiquantitative comparison the total elemental contents of the same samples were measured after the leaching 3 g bulk sediment in aqua regia (21 ml HCl + 7 ml

HNO3) at 120 °C for 120 minutes.

5.3.2.6. Scanning electron microscopy Scanning electron microscopy (SEM) observation was performed using a ZEISS DSM 962 device (Carl Zeiss Microscopy GmbH, Oberkochen, Germany) with a Tracor Northern energy dispersive X-ray spectroscopy (EDX). To avoid electric charge, electrically non-conducting material is dusted with a thin Au-Pb layer (Goldstein et al., 2007). Layer silicates, particularly clay minerals, have a porous structure and a large inside surface. Therefore, the Au-Pb layer on these minerals is partially incomplete. Consequently, clay minerals tend to become electrically charged during the SEM procedure. This is reflected in artifacts on the SEM-picture, such as blooming and edge effects, whereby the contrast is greatly exaggerated (c.f. Suito, 1990).

5.3.3. Results In the following, the results from the SHD section from the northeastern fringe of the Tibetan Plateau are presented in detail. For comparison, the key results from the DDG section in western Germany are provided in Chapter 3.6.

5.3.3.1. Effect of acetic acid treatment

To evaluate the effect of acetic acid (CH3COOH) in comparison to HCl, subsamples of six samples

2 were treated with both acids. The reaction with CH3COOH led to the same GS result as HCl (r =

0.994). However, the reaction was slower and the double amount of CH3COOH was required owing to its lower concentration of H+ ions.

5.3.3.2. Comparison of HCl treated and untreated samples Prior to the GS measurement, subsamples from all 57 samples of the SHD LPS were pretreated both with and without the addition of HCl. The HCl-induced modifications of GS distributions (MODGSD) are shown as differences between the GS frequency of HCl-treated and untreated 62

subsamples in a heatmap (Fig. 5.3.3). Each cell of the heatmap represents the MODGSD given by the colorbar (values between -0.8 and 1.2 vol. %) of one sample within one of the 116 GS classes. In Fig. 5.3.4 the central statistical parameters mean, mode and median of the GS distributions are plotted against the depth.

Fig. 5.3.3: Heatmap showing HCl-induced MODGSD (frequency of HCl treated subsamples – frequency of untreated subsamples) of all samples from the Suohuduo section. This difference is shown for each of the 116 grain size classes resulting from the laser diffraction analysis. Red colors indicate a decrease and blue colors an increase of the frequency within the respective GS class after HCl treatment. The dotted lines separate the classic GS fractions C: 0.04 – 2 µm, fSi: 2 – 6.3 µm, mSi: 6.3 – 20 µm, cSi: 20 – 63 µm, fS: 63 – 200 µm, mS: 200 – 630 µm, cS: 630 – 2000 µm. The vertical curves represent the content of CaCO3 and Corg.

The base of the SHD section (L1, 312.5–265 cm; Fig. 5.3.1, Fig. 5.3.5 A, Fig. 5.3.3) is composed of yellowish silty sand and shows continuous fining upward towards sandy loess at 265 cm (c.f. Lehmkuhl et al., 2014). Treatment with HCl generally led to an increasing proportion of cSi, widely on the expense of a decreasing proportion of fS (Fig. 5.3.3). The variation amounted to a shift of about 2.76 % on average between these two fractions (n = 10; SD = 1.05 vol. %; mean aliquot SD: 1.35 vol. %). The central statistical parameters mean, mode and median were not significantly affected by the HCl treatment. 63

Fig. 5.3.4: Depth profiles of the parameters mean, mode and median for the MODGSD of the Suohuduo section.

The overlying pale yellow unit (L03, 265–200 cm) consists of sandy loess and interbedded strata of several thin dark laminations between 225 and 240 cm depth (Fig. 5.3.1, Fig. 5.3.5 A, Fig. 5.3.3). The MODGSD of this unit was slightly stronger than in the underlying unit, even though trends were similar (n = 13; mean = 4.67 vol. %; SD = 1.58 vol. %; mean aliquot SD: 1.4 vol. %). A slight decrease of the fSi and mSi fractions in favor of the cSi-fraction was evident in several samples between 225 and 240 cm depth (humic zones in Fig. 5.3.3). The mean, mode and median values were slightly reduced, with the exception of the humic zones in which the parameters remained the same or increased slightly (Fig. 5.3.4).

64

Fig. 5.3.5: GS distribution curves of untreated samples (top), the HCl-treated samples (upper middle), calculated differences (lower middle), and mean SD of all aliquot measurements (bottom) for three pedogenic units of the Suohuduo section: (A) primary loess, (B) modern soil, (C) paleosols.

Prior to the HCl treatment, the 13 sediment samples taken from the two paleosols S01 (120–90 cm) and S02 (200–150 cm) were poorly sorted (Fig. 5.3.1, Fig. 5.3.3, Fig. 5.3.5 C). The respective mean GS distributions had a mode of 50.2 µm each and mean values of 33.7 and 33.8 µm. Furthermore, they showed fluctuations between 2 and 22 µm (Fig. 5.3.3). As a consequence of HCl treatment, these broad shoulders disappeared and the cSi mode peak became better defined (Fig. 5.3.5 C). A considerable shift of the GS distribution from the fSi, mSi and lower cSi fractions to the upper cSi and fS fractions (6.73 vol. % in average) was observed for the sediments of S02 (n = 65

8; SD = 2.39 vol. %; mean aliquot SD: 5.59 vol. %). The sediments of this paleosol can be divided into two subunits: The upper four and the lower four samples were separated by two samples which show only a weak reaction to the HCl treatment. They showed similar characteristics as the sample of the units above and below (mean deviation: 2.3 vol. %; SD = 0.36 vol. %; Fig. 5.3.3). The changes within the upper paleosol (S01) showed a similar but even more pronounced trend. The upper cSi and the fS fractions decreased in favor of the fSi and mSi fractions, on average by 11.59 vol. % (n = 4; S.D = 3.56 vol. %; mean aliquot SD: 5.89 vol. %). The lower cSi fraction showed no significant variation (Fig. 5.3.5 C, Fig. 5.3.3). Within this intensively weathered Holocene paleosol, the greatest MODGSDs of the entire sample set occured. Specifically, the mean and median values of all paleosol samples are increased after HCl treatment (Fig. 5.3.4). At two samples from S01 even the mode values increased significantly.

The two paleosols are separated by a pale yellow transition unit L02 (150–120 cm; Fig. 5.3.1). The sediment samples from this unit were better sorted and slightly coarser than the surrounding paleosols. The MODGSD was similar to those of L1 and L03 though the effect was less pronounced, showing an average shift of 2.4 vol. % (n = 7; SD = 1.06 vol. %; mean aliquot SD: 1.76 vol. %) from upper fS to lower fS after HCl treatment (Fig. 5.3.5 A, Fig. 5.3.3).

The uppermost stratigraphic unit (L01, 90–5 cm) above the buried soil complexes is a weak cambic horizon (modern top soil; Fig. 5.3.1). The weak soil formation of the parent sandy loess material is evident from brownish discoloration and carbonate removal. Within this unit, hardly any MODGSD was evident (Fig. 5.3.5 B, Fig. 5.3.3). With exception of the two calcareous samples from its base (Fig. 5.3.3), the MODGSDs showed a shift from cC and fSi to cSi and fS by 2.08 vol. % on average (n = 11; SD = 0.88 vol. %; mean aliquot SD: 1.74 vol. %).

The mean, mode and median diameters of the samples from the units L02 and L01 are not significantly affected by the HCl treatment (Fig. 5.3.4).

Figure 5.3.5 compares the GS distributions of all samples from SHD section, grouped by the units ‘primary loess’, ‘modern top soil’, and ‘paleosol’. Mean aliquot SD of the respective pedogenic units are plotted to illustrate their heterogeneity.

5.3.3.3. MODGSD dependence on CaCO3 and Corg content

The CaCO3 content varied between 1.0 and 4.6 % in the paleosol samples and between 5.4 and 14.3 % in almost all loess samples (Fig. 5.3.3). Two samples classified as loess –originating from layers situated directly subjacent to paleosol S01 in the section– and most of the modern soil samples were free of carbonates. No linear relationship between CaCO3 content and MODGSD

66

for the samples from modern soil and paleosol (modern soil: r = -0.236; paleosol: r = 0.096) and only a weak linear relationship for the loess samples (r = 0.344) was observed (Fig. 5.3.6 A).

Fig. 5.3.6: Dependence of the HCl-induced MODGSD on (A) the content of CaCO3 and on (B) the content of Corg for the pedogenic units primary loess, modern soil, and paleosol.

Conversely, a distinct dependence of the MODGSD on the Corg content and the weathering degree of the sediment was evident (Fig. 5.3.6 B). The Corg content varied between 1.6 and 4.1 % in the paleosol samples and between 1.2 and 4.8 % in the modern soil samples. Most of the loess samples contain less than 1% Corg (Fig. 5.3.3). A strong linear relationship between the MODGSD of intensively weathered samples from paleosols S01 and S02 and their content of Corg was found (r = 0.924). This relationship was weaker for the slightly weathered modern soil (r = 0.653) and absent with the loess samples (r = -0.209).

5.3.3.4. Element composition of the HCL leachable fraction To examine the effect of HCl treatment on the elemental composition of the sediment samples, the eluate of 6 selected samples was decanted after leaching with HCl and measured with an ICP- OES. The resulting concentrations of Ca, Mg, Al, Fe, Mn, K, Na, Ba, Cu, Ni, and Zn of the HCl leachable fraction are shown in Table 5.3.2, element concentrations of the aqua regia extract in Table 5.3.3. Total concentrations (aqua regia extract) of Ca, Mg and Na in the loess samples L02, L03 and L1 were high in comparison to the other samples (mean values: Ca = 44783, Mg = 7860, Na = 343.7 mg/kg). Accordingly, these elements were highly concentrated in the HCl leachable fraction (mean values: Ca = 6357, Mg = 120.33, Na = 7.6 mg/l). In addition, the concentrations of K, Fe and Mn were comparatively high in the HCl eluate (mean values: K = 8.55, Fe = 16.24, 67

Mn = 11.58 mg/l), even though the total concentrations of these elements were lower than in the other samples analyzed (mean values: K = 2387, Fe = 22370, Mn = 423.53 mg/kg).

Tab 5.3.2: CaCO3 contents [%] and concentrations of HCl-leachable elements [mg/l] within the supernatant after centrifugation

Sample CaCO3 Ca Mg Al Fe Mn K Na Ba Cu Ni Zn

L01 0 227.9 16.64 50.67 4.02 4.19 5.45 4.91 3.56 0.34 0.36 0.12

S01 1.02 2933 20.58 56.59 5.76 4.14 6.05 5.31 5.57 0.31 0.38 0.24

L02 8.18 4808 74.28 44.7 14.3 8.37 7.68 6.49 3.57 0.52 0.18 0.17

S02 3.66 4028 26.6 80.18 4.86 3.18 7.32 5.58 6.12 0.21 0.4 0.17

L03 11.11 6891 116.6 31.35 10.6 7.49 8 7.47 4.3 0.45 0.13 0.13

L1 12.44 7371 170.1 26.69 23.81 18.89 9.97 8.86 3.22 0.18 0.13 0.17

Tab 5.3.3: Elemental concentrations [mg/kg] of the stratigraphic units of the SHD section; aqua

regia extract (3:1 concentrated HCl and concentrated HNO3)

Sample Ca Mg Al Fe Mn K Na Ba Cu Ni Zn

L01 4470 5950 15560 25320 603.4 2349 269.3 106.7 19.8 24.1 54.1

S01 19150 6160 16540 25820 641.8 2862 283.8 149.2 26.1 21.9 57.8

L02 35440 6900 13170 22280 463.7 2150 315.7 86.5 18.6 18.6 39.5

S02 23620 6660 17500 26650 509.3 3205 278 130.5 31 22.4 53.9

L03 46040 8380 14240 23950 448.8 2490 398.2 94.7 19.4 20 44.2

L1 52870 8300 11500 20880 358.1 2521 317.2 59.3 26.4 13.7 36.8

The exchangeable bases Ca, Mg and Na were leached by HCl from the paleosols S01 and S02 to a lesser degree (mean values: Ca = 3480, Mg = 23.59, Na = 5.44 mg/l) but analogous to the lower levels in the aqua regia extract (mean values: Ca = 21385, Mg = 6410, Na = 280.9 mg/kg). In addition, a comparatively little amount of K, Mn and Fe ions was removed from the paleosols (mean values: K = 6.68, Fe = 5.31, Mn = 3.33 mg/l). However, the aqua regia-extractable content of these elements was higher than in the other samples (mean values: K = 3033, Fe = 26235, Mn

68

= 575.55 mg/kg). The concentrations of Al and Ba in the paleosol samples were comparatively high in both the HCl eluate and in the aqua regia extract. The sample from the modern soil (L01) showed the smallest affect by the HCl treatment with the exception of Al, Cu and Ni which showed slightly higher values than the calcareous loess samples. The H2O2 treatment was much more effective when applied subsequent to HCl treatment as indicated by faster and stronger bleaching of the sample (Fig. 5.3.7).

Fig. 5.3.7: Two aliquots of the same samples after the first 5 h run of the H2O2 pretreatment cycle. The right glass tube was previously treated with HCl, the attempt in the left glass tube only pre- moistened with aqua dest.

5.3.3.5. Scanning electronmicroscopy (SEM) The untreated subsample from paleosol S01 was characterized by large particles (300–900 µm), exhibiting an aggregated form (Fig. 5.3.8 A). This demonstrated that a considerable portion of macroaggregates prevailed within this Holocene paleosol, even after mechanical pestling. After treatment with H2O2 (Fig. 5.3.8 C) and H2O2 + HCl (Fig. 5.3.8 E), single grain minerals without the adhesion of finer grains and aggregates up to a size of 250 µm were observed at this magnification.

69

Fig. 5.3.8: SEM images of a sample from paleosol S01 at different magnifications: (A) and (B) show the untreated sample, (C) and (D) the sample pretreated with H2O2, and (E) and (F) the sample pretreated with H2O2 und HCl.

Further assessment of the same sample at higher magnification revealed the presence of silt-sized aggregates and layer silicates (Fig. 5.3.8 B, D, F). Within the untreated sample (Fig. 5.3.8 B), the clay minerals were predominantly agglomerated to aggregates of 10–50 µm. In contrast, particles were mostly separated into single grains after H2O2 treatment (Fig. 5.3.8 D). After H2O2 + HCl treatment, however, most of the small particles were coagulated and showed aggregates up to a size of 150 µm (Fig. 5.3.8 F).

The most obvious difference between the raw sample, the H2O2 treated sample, and the H2O2 + HCl treated samples was the electric charge of the small particles at high magnification induced by the primary electron beam. In case of the untreated sample, the small particles and the silt sized 70

aggregates showed a slight electric charge (Fig. 5.3.8 B). As a result of the H2O2 treatment, corresponding particles were disaggregated to single grains which exhibited a strong electric charge

(Fig. 5.3.8 D). In contrast, virtually no electric charge was detectable after H2O2 + HCl treatment (Fig. 5.3.8 F).

5.3.3.6. MODGSD, CaCO3 and Corg at section DDG

Fig. 5.3.9: Heatmap showing HCl-induced MODGSD (frequency of HCl treated subsamples – frequency of untreated subsamples) of all samples from the Düsseldorf-Grafenberg section. This difference is shown for each of the 116 grain size classes resulting from the laser diffraction analysis. Red colors indicate a decrease and blue colors an increase of the frequency within respective GS class after HCl treatment. The dotted lines separate the classic GS fractions C: 0.04 – 2 µm, fSi: 2 – 6.3 µm, mSi: 6.3 – 20 µm, cSi: 20 – 63 µm, fS: 63 – 200 µm, mS: 200 – 630 µm, cS: 630 – 2000

µm. The vertical curves represent the contents of CaCO3 and Corg.

The HCl-induced MODGSDs of all 79 samples taken from the DDG coring are shown as differences between the HCl-treated and the untreated subsamples in a heatmap (Fig. 5.3.9). The statistical parameters mean, mode and median of the MODGSD are plotted versus depth in Fig. 5.3.10. Generally, the variations were significantly lower in comparison to the SHD section. The 71

mean values of several samples between 3.5 and 6.5 m depth were distinctly reduced after treatment with HCl. Conversely, the treatment had only little effect on the mean values of the remaining samples. Modal and median values were not significantly affected by the acid treatment. The contents of CaCO3 and Corg are shown in Fig. 5.3.9. With the exception of the topmost modern soil sample, the entire section was free or contained very little remnants of Corg (<0.4 %).

Fig. 5.3.10: Depth profiles of the statistical parameters mean, mode and median of the MODGSD for the Düsseldorf-Grafenberg coring.

The base of the analytical sequence of vibracore DDG (18.2–17.4 m) is a slightly oxidized loess layer composed of reworked sediment (Fig. 5.3.2, Fig. 5.3.9). The MODGSD of the remaining samples (n = 6; SD = 0.84 vol. %) only slightly exceeded the corresponding mean aliquot SD (1.52 vol. %). From 17.4 m to 13.6 m, reddish, decalcified, clayey soil sediments occur (Fig. 5.3.2, Fig. 5.3.9). In contrast to the lower part of this unit which showed hardly any GS variation, the MODGSD of the upper part is stronger, exhibiting a distinct shift of 3.55 vol. % on average from the fSi and mSi to the cSi and fS fractions (n = 5; SD = 1.64 vol. %; mean aliquot SD: 2.26 vol. 72

%). The overlying unit (13.6–12.9 m) consists of weakly calcareous, yellowish loess which is slightly reworked (Fig. 5.3.2, Fig. 5.3.9). Here, MODGSD showed a shift of 2.72 % on average from the upper mSi and cSi to all other fractions (n = 4; SD = 0.78 vol %; mean aliquot SD: 1.26 vol. %). This sediment is followed by a reworked calcaric cambisol (12.9–11.2 m) which is composed of brownish, partially laminated silt and numerous mollusks and mollusk fragments (Fig. 5.3.2, Fig. 5.3.9). The mean MODGSD (2.42 vol. %) of most of the samples exhibited a slight shift from the upper fS and mS to the mSi, cSi and lower fS fractions (n = 7; SD = 0.43 vol. %; mean aliquot SD: 2.04 vol. %). The MODGSD was stronger in the top unit than in the underlying sediment, yielding a distinct shift of 4 vol. % on average from the fSi fractions to the coarser fractions (n = 2; SD = 0.9 vol. %; mean aliquot SD: 1.72 vol. %).

The unit from 11.2–2.6 m consists of loess and intercalated, weakly developed cryosols which are entirely calcareous. Finely laminated loess and soil sediments indicate reworking processes. Especially within some of the cryosols, features of initial oxidation and secondary carbonate precipitation were detected. Within most of the samples, a MODGSD decrease of 3.16 vol. % on average of the fine fractions (C, fSi and cSi) and the upper fS and mS fractions in favor of the cSi and lower fS fractions was evident (n = 34; SD = 0.68 vol. %; mean aliquot SD: 1.99 vol. %). The MODGSD of the samples between 7–8.3 m differed from the rest of the unit and indicated a shift of 2.70 vol. % on average from the fine fraction (until lower fSi) to the coarser fractions (n = 8; SD = 0.53 vol. %; mean aliquot SD: 1.18 vol. %). The uppermost unit (0–2.6 m) is the modern top soil which is completely decalcified and influenced by illuviation processes (Fig. 5.3.2). Within this unit, hardly any variation of the HCl-induced MODGSD was observed (Fig. 5.3.9).

5.3.4. Discussion The samples comprise weathered and unweathered loess and loess derivates, exhibiting a composition which is representative for many other Late Pleistocene and Holocene loess-paleosol sequences. The SHD section is composed of pure aeolian silty sediments in which two distinct, climax-like paleosols have developed. These paleosols are rich in Corg, whereas the primary loess units are rich in CaCO3. In contrast, the DDG section is mainly composed of reworked aeolian loess and soil sediments. Former distinct soil complexes are severely degraded (<1% Corg) or not preserved due to erosion (Fischer et al., 2012). The following discussion focuses on the GS variations within the SHD section. Subsequently, the results from DDG and SHD sections are compared to evaluate (i) whether there are any residues of OM preserved by stabilization mechanisms and (ii) whether there are indications of GS variations caused by HCl treatment.

73

5.3.4.1. MODGSD of the samples from the SHD section The distribution density curve is based on percentage frequencies of the respective GS classes. Therefore, variations in a certain GS range affect all other GS classes. Hence, it is difficult to differentiate between absolute GS changes and relative increase or decrease of individual classes. The mean aliquot SD is a convenient parameter to differentiate the variations related to HCl pretreatment from variations caused by the heterogeneity of bulk samples and instrument-related uncertainties. Hereafter, the results of the SHD section are interpreted separately by pedogenic units (i) primary loess, (ii) modern top soil, and (iii) paleosols.

(i) In the primary loess units, average MODGSD of the unweathered loess samples (3.5 vol. %, n=30) exceeded the mean SD of the aliquot measurements (1.5 vol. %) in all GS classes. However, the deviations appear to be GS-dependent. The GS decrease in the range of 0.04 to ~10 µm is a consequence of the solution of finely divided CaCO3, primary or secondary (Nettleton, 1991). The decrease in the range of ~83 to ~256 µm could either be attributed to a size reduction of single grains owing to removal of carbonate or iron oxide incrustations at larger quartz particles (Smalley, 1971) or to a solution of aggregations with carbonate binding agent (Cantoni et al., 2012; Six et al., 2004). The reductions at 0.04 to ~10 µm and ~83 to ~256 µm are both in favor of a relative increase of the mode fraction, i.e., ~10 to ~83 µm. No significant relationship between the HCl- induced MODGSD and the CaCO3 content was observed (Fig. 5.3.6). This indicates that carbonates are predominantly present as primary particles in the clay and fine silt fraction or as thin incrustations on larger particles. Only in a few samples carbonates seem to act as a binding agent within aggregates which have a significant influence on the measured particle size range. The humic residues between 225 and 240 cm depth were macroscopically detected as thin bands during the field observation but could not be detected by the conventional laboratory analysis (e.g., Corg). Nevertheless, the MODGSDs of these samples are comparable to the results from the paleosols.

There is no relationship between the GS decrease of the fine fractions and the content of CaCO3 within the corresponding samples. This could be an indication of the presence of OM embedded in stable compounds. The Corg content in the total bulk samples was, however, small; accordingly, no significant difference to the surrounding samples is evident.

(ii) The average MODGSD of the recent soil material (2.08 vol. %, n=11) only slightly exceeded the mean SD calculated for all aliquot measurements of the modern top soil samples (1.74 vol. %; Fig. 5.3.5). The HCl treatment has weak systematic effect on the results of the GS analysis as a function of the Corg content (Fig. 5.3.6). The appearance of MODGSD is generally similar to the results from the paleosol samples. Nevertheless, modifications are significantly less pronounced, owing to the lower degree of weathering as well as the lower age of the stable microaggregates and organo-mineral complexes (cf. Kögel-Knabner et al., 2008). 74

(iii) In average, the strongest HCl-induced MODGSDs occured in the paleosols. The fSi and mSi fractions, i.e., ~1.8 to ~30 µm, strongly decreased while the upper cSi and fS fractions, i.e., ~30 to ~213 µm, increased (Fig. 5.3.5 C and Fig. 5.3.3). Additionally, a slight increase of the C fraction, i.e., 0.04 to ~1.8 µm, was evident. A distinct dependence of the MODGSD on the content of Corg was observed (Fig. 5.3.6). We assume two processes as reasons for the GS variations within the paleosols: (a) The disintegration of micro- and marcoaggregates or organo-mineral complexes accompanied by the loss of several HCl-leachable elements due to acid treatment, and (b) the formation of upper coarse silt- and fine sand-sized aggregates composed of micro-aggregates, quartz and clay minerals. A relative increase of the upper cSi and fS fractions, i.e., ~30 to ~213 µm, as a reaction to the decrease of the fSi and mSi fractions, i.e., ~1.8 to ~30 µm, can be neglected in this case, since the major component of the untreated samples (38–55 µm) hardly changes after the HCl treatment (Fig. 5.3.5 C).

5.3.4.2. (a) Disintegration of aggregates and organo-mineral complexes

By pretreatment with H2O2, some weaker bound aggregates are split up and the embedded OM is decomposed (cf. von Lützow et al., 2007, 2008). The separation of the macroaggregates into microaggregates and individual particles was revealed by the SEM images (Fig. 5.3.8D). Most of these fine particles were clay minerals or organo-mineral complexes as indicated by the strong electrical charging of the SEM preparation. However, stabilizing mechanisms reducing or preventing the destruction of the OM by interactions with the mineral surfaces were evident within paleosol complexes, such as S01 and S02 of the SHD section (cf. Kögel-Knabner et al., 2008).

For the following reasons, it is concluded that within certain microaggregates and particularly organo-mineral complexes the embedded OM is preserved and protected against decomposition by H2O2 (cf. Boudot et al., 1989; Kleber et al., 2005):

 There is a strong correlation (r² = 0.924) between HCl-induced MODGSD in addition to

the H2O2 treatment and the content of Corg (Fig. 5.3.6). This indicates that the H2O2 treatment does not result in a complete oxidation of the OM. An incomplete oxidation of OM was also detected at lacustrine sediments from two Estonian lakes (Vaasma, 2008) and at estuarine silts from southwest Britain (Allen and Thornley, 2004).

 The most significant GS decrease after HCl treatment occurs in the GS range 1.8–30 µm.

The majority of the soil organic carbon (Corg) in silty soils is aggregated with clay minerals and organo-mineral complexes in the size 2–20 µm fraction (Van Gestel et al., 1995; Schulten and Leinweber, 2000; von Lützow et al., 2006).

75

 The highest total concentrations (aqua regia extract; Tab 5.3.3) of the elements which are associated with organo-metallic complexes (Al, Fe, Mn) are evident within the paleosols. Nevertheless, Fe and Mn are dissolved only to a small extent by HCl, suggesting that these metal ions are present in stable complexes (cf. von Lützow et al., 2008).

 H2O2 treatment is much more effective if applied subsequent to HCl treatment, as indicated by faster and more intense bleaching of the samples (Fig. 5.3.7).

In summary, it can be assumed that HCl treatment reduces or even prevents aggregation, destabilizes complexes, and destroys or diminishes susceptible clay minerals.Therefore, the HCl- induced MODGSD must be considered selective and inscrutable. The comparison of the GS distributions with and without HCl pretreatment provides an indication of the existence of stable microaggregates and organo-mineral complexes. It also indicates GS ranges in which these structures are present and the relative age of the H2O2 resistant features (c.f. Kögel-Knabner et al., 2008; von Lützow, 2008).

5.3.4.3. (b) Formation of cSi- to fS-sized aggregates As shown in the SEM images, not all particles were present as single grains; instead, many particles are bound in aggregates to a size of 200 µm (Fig. 5.3.8E, F) (c.f. Fernandez-Ugalde et al., 2011). After the dissolution of most silt-sized aggregates and organo-mineral complexes, some of the separated particles probably polymerize directly to larger stable aggregates. Stable intra-spherical complexes may occur due the interaction of electrostatic attraction and covalent bonds (Stumm, 1992) and due to anion sorption which increases with decreasing pH value (Hingston et al., 1972; Nieder and Benbi, 2008) as well as with increasing specific surface area of the sorbents (Baldock and Skjemstad, 2000; Kaiser and Guggenberger, 2003).

5.3.4.4. Comparison of SHD and DDG results The MODGSD of the DDG samples shows similar dependencies on sediment types as the samples from SHD section (Fig. 5.3.9). Again, no significant correlation between HCl-induced MODGSD and CaCO3 content was observed. The entire coring contained only very low contents of Corg which had no detectable influence on the processes within the sample. The HCl-induced MODGSD only slightly exceeded the GS variation of single aliquot measurements of the same attempt (mean aliquot SD). Nevertheless, the variations showed a dependence on the sediment type.

Most of the samples taken from the calcareous loess sediment and intercalated cryosols between 11.2 and 2.6 m show a similar MODGSD as the primary loess samples from the SHD section. The 76

decrease in the fine fraction within some subunits (e.g., between 7 and 7.8 m) provides an indication of disintegration of silt-sized aggregates and organo-mineral complexes occurring in cryosols.

Knowledge on the role of clay humic complexes and their contribution to Corg stabilization in cryosols is, however, still sparse (Zech et al., 2014). Similar to the SHD section, no significant HCl- induced MODGSD is evident in the modern soil.

Most of the soil sediments from 17.4–13.6 m and from 12.9–11.2 m are degraded (OM is decomposed by bacterial activities) so far that no pedogenic features reacting notably to the HCl treatment are preserved. However, within the upper parts of both soil units distinct HCl- induced MODGSDs occur whose patterns are very similar to those detected in the paleosols of the SHD section.

Fig. 5.3.11: Comparison of mean MODGSD and standard deviation of both Suohuduo (SHD) and Düsseldorf-Grafenberg (DDG) sequences, showing heterogeneous effects of HCl treatment on different GS spectra. (A) Disintegration of silt-sized microaggregates and organo-mineral complexes owing to destabilization of the pedo-features and the loss of several HCl leachable elements, especially in the paleosol samples. (B) Relative increase of the mode fraction in the entire sample set. (C) GS decrease due to the dissolution of weak aggregates and GS reduction caused by the dissolution of carbonate and iron oxide coatings (mainly in the primary loess samples). The GS increase due to the formation of fS-sized aggregates composed of micro-aggregates, quartz and clay minerals. (D) Disintegration of weakly bound S-sized aggregates which are largely dissolved just by the pretreatment with H2O2 and Na4P2O7 accompanied by a gentle measurement. (E)

Disintegration of infrequent macroaggregates which are not previously separated by the H2O2 treatment.

77

The mean values are sensitive to the edge regions of the distribution density curves. Grain size reductions of coarser particles are particularly reflected by the mean values. Median and mode diameters are suitable for the interpretation of GS data since they are robust against outliers. By comparison to mean values, median and mode values of several samples between 3.5 and 6.5 m depth (Fig. 5.3.10) indicate that small amounts of coarse aggregates are split up after HCl treatment. However, measurement precision and accuracy are lowest at the coarse edge of the GS distribution curve. In general, it is evident that median and mode diameters are relatively insensitive to MODGSD variations. Consequently, the effects of HCl treatment on the validity of mean and mode values of GS distributions seem to be negligible in many cases.

5.3.5. Conclusion

 There is no significant association of HCl-induced MODGSD and CaCO3 content.

Conversely, a distinct dependence of MODGSD on the content of Corg and the weathering degree of the sediment is evident.

 Omitting HCl treatment does not result in misleading GS distributions. Although the loess samples contain 5.42–14.31 % detrital and authigenic carbonate particles, the dispersive

effect of Na4P2O7 and H2O2 was sufficient to achieve full dispersion of the particle suspension.

 The comparison of GS distributions with and without HCl pretreatment provides an indication of the existence of stable microaggregates and organo-mineral complexes as well as an indication of the GS ranges in which these structures are present.

 The HCl treatment of primary loess samples predominantly results in dissolution of

contained CaCO3. However, the nonlinear relationship between MODGSD and CaCO3 content provides an indication of the modifications being not exclusively a function of carbonate content. The HCl treatment has no significant impact on the validity of GS measurement as indicated by lacking changes of the central statistical parameters of the distribution density curve.

 The pretreatment with HCl at post-depositional modified aeolian sediments, particularly at intensively weathered paleosols, is reflected in substantially differing GS distributions measurements. In such settings HCl treatment may lead to misleading results.

 For the GS analysis of a sediment sequence, all samples should be pretreated uniformly in order to ensure comparability. As each pretreatment is a modification of the original sample

78

and our results show that HCl pretreatment is particularly selective and inscrutable, we recommend omitting decalcification in the context of common particle size analysis of a LPS. In general, the sample pretreatment should be matched to the research question of the study and should be as gentle as possible to minimize GS modification and hence possibly misleading effects on the measurement results.

79

5.4. The difference of two laser diffraction patterns as an indicator for post- depositional grain size reduction in loess-paleosol sequences

Philipp Schultea and Frank Lehmkuhla a Department of Geography, RWTH Aachen University, Templergraben 55, D-52056 Aachen,

Germany

Under Review in Paleogeography, Paleoclimatology, Paleoecology

Abstract Chemical and physical processes within aeolian sediments result in a reduction of the particle grain size. In loess sediments the post-depositional grain size variation is due to a reduction of the dominating coarse silt fraction in favor of the clay and fine silt fraction. Generally, there are two post-depositional fractionation processes: (1) the chemical weathering of silt sized minerals like mica and feldspar as a result of hydration and hydrolysis, (2) the physical weathering of all minerals contained in the primary loess sample by cryogenic and fluvial relocation processes. There are many widely used proxies to estimate their vertical variation in thick sediment columns. However, there are complicating factors related to aeolian sorting effects, the distance to the source regions and carbonate dynamics, which reduces the sensitivity of common proxies to the chemical weathering. In this study, we present a simple and quick method using laser diffraction calculations of grain size distribution obtained by two optical models (the Fraunhofer approximation and the Lorenz- Mie theory) to highlight the enrichment of fine grained material by post-depositional chemical weathering processes. In contrast to the Fraunhofer approximation, the Lorenz-Mie theory considers the complex refractive index which depends on the mineral properties. Thus, the difference of the grain size distributions (ΔGSD) calculated with both models is sensitive to the mineral composition of the sample. In separated submicron mineral suspensions we found that different crystalline properties are reflected by repeatable signatures of the ΔGSD. This specific ΔGSD-signature also occurs in the submicron grain size range of bulk measurements of weathered loess samples. In contrast to previous studies, we obtain reliable laser diffraction results in the submicron range. Summarizing, we present the ΔGSD within the submicron range of bulk measurements as a suitable indicator for the chemical weathering degree of loess-paleosol sequences, which is virtually unaffected by cryogenic processes, weak relocation of inherited 80

weathering products and synsedimentary processes (sheet wash, saltation or enhanced background sedimentation).

5.4.1. Introduction The formation of secondary clay minerals in loess deposits is a common indicator for environmental conditions which are favorable for soil development. Loess is defined as an aeolian deposit predominantly consisting of silt size particles (Pye, 1995; Stauch et al., 2012; Muhs et al., 2014; Sprafke and Obreht, 2016) and is consolidated due to loessification (Pecsi, 1990; Smalley et al., 2011; Svircev et al., 2013; Smalley and Markovic, 2014; Sprafke and Obreht, 2016). Loess consists mainly of silt sized particles (Pecsi, 1990; Pye, 1995; Muhs and Bettis, 2003), whereas submicron clay sized particles (< 1 µm) transported as aggregates or quartz grain coats (Mason et al. 1999, 2003) are extremely underrepresented during accumulation (Qiang et al., 2010; Ujvari et al., 2016). Therefore, we suggest the occurrence of submicron secondary clay minerals as a strong estimator for the degree of post-depositional grain size (GS) fractionation due to chemical weathering processes (silicate weathering and soil formation).

The allogenic mineral fraction of loess is composed of quartz, feldspar (K-feldspar, plagioclase), phyllosilicates (biotite, muscovite, clay minerals) carbonates (calcite, dolomite) and sometimes small amounts of iron-oxides and –hydroxides, amphibole, heavy minerals and volcanic ashes (Pecsi, 1990; Pye, 1995; Jeong et al., 2008, 2011; Smalley et al., 2011; Ujvari et al., 2012). Loess is predominantly accumulated during glacial periods. If the accumulation rate is strongly reduced while moisture availability is increased, soil formation is enhanced (e.g. Stevens et al., 2011; Sprafke and Obreht, 2016). Weathering and pedogenesis result in a finer grain size (e.g. Ujvari et al., 2016). In loess sediments the post-depositional grain size variation is due to a reduction of the dominating coarse silt fraction in favor of an increase in the clay fraction. The fine and medium silt fractions also increase in most cases. The dominant post-depositional fractionation process is the chemical weathering of silt sized minerals like mica and feldspar as a result of hydration and hydrolysis (Schaetzl and Thompson, 2015). Loess-paleosol sequences can be investigated to get information about the Cenozoic and especially Quaternary climate evolution. The amount of clay sized particles could be considered as a meaningful proxy for pedogenesis. However, it is widely accepted that this fraction is underestimated by standard laser diffraction measurements (e.g. Konert and Vandenberghe, 1997; Beuselinck et al., 1998; Mason et al., 2011). Hence, there are many other widely used proxies, such as geochemical weathering indices (Buggle et al., 2011; Ujvari et al., 2014), grain size indices (Vandenberghe et al., 1997; Antoine et al., 2009a,b) or rock magnetism (Buggle et al., 2014; Hosek et al., 2015), to estimate the vertical variation and the different intensities of the

81

post-depositional weathering. However, there are complicating factors related to these proxies including aeolian sorting, the distance to the source regions, cryogenic processes, reworking and carbonate dynamics (Zhao et al., 2005; Buggle et al., 2011; Peng et al., 2014).

Due to the relative absence of fine and medium clay in many deposits of loess, the enrichment of these fractions is often used as a proxy for pedogenesis and climate variations (e.g. Nugteren and Vandenberghe, 2004; Antoine et al., 2009a; Terhorst et al., 2012). In this study, we present a simple and quick method for quantifying post-depositional enrichment of pedogenic clay minerals using the difference of two grain size distributions (GSD) obtained by two optical models, the Fraunhofer approximation (FH) and the Lorenz Mie Theory (Mie) from the same laser diffraction raw data (ΔGSD = GSDFH – GSDMie).

The commonly used optical model to estimate a GSD on the basis of the detected scattering pattern is the Fraunhofer approximation (Konert and Vandenberghe, 1997; Beuselinck et al., 1998; Jones, 2003, Pye and Blott, 2004). Currently, the Lorenz Mie Theory including the complex index of refraction is increasingly applied (de Boer et al., 1987; Jones, 2003; Eshel et al., 2004; Schulte et al., 2016). The Mie Theory is derived from atmospheric physics where it is used to calculate light scattering by mineral aerosols in the Earth’s atmosphere (Volten et al., 2001; Klüser et al., 2016). Özer et al. (2010) measured different soil samples with a Malvern Mastersizer X and demonstrated that there are only very minor differences between GSDs calculated with the different optical models. In contrast, if the Beckman Coulter LS 13320 is used with exactly the same optical models, considerable differences in the clay size range may occur (Schulte et al., 2016). The detector arrangement of the LS 13320 with additional PIDS technology (polarization intensity differential scattering system) is particularly sensitive to changes within the submicron grain size range.

In this study, the submicron difference between the two optical models (ΔGSD, GS frequency calculated with FH – GS frequency calculated with Mie) is evaluated as a proxy for post- depositional enrichment of pedogenic clay minerals. For general evaluation, several single samples with known mineral composition were analyzed. In order to test the applicability of the ΔGSD proxy, three loess-paleosol sequences were investigated. The variability of low concentrated grain size ranges (submicron) within percentage-frequency distributions can be misleading due to the compositional data effect (Aitchison 1986; Bloemsma et al. 2012; Roberson and Weltje 2014). Therefore, the vertical variability of bulk sample ΔGSDs is analyzed as centered log-ratio transformed grain size differences (ΔGSDclr), as these are more robust against other grain size influencing processes especially during sediment accumulation (sheet wash, saltation or enhanced background sedimentation).

5.4.2. Materials and Methods 82

5.4.2.1. Sample selection As illite is one of the predominant clay minerals pedogenetically formed in loess sediments (Bronger and Heinkele, 1990), the ΔGSD was calculated for several illite containing mineral samples and soil standards (Tab. 5.4.1), as well as for five natural kaolinite dominated clay mineral samples with known mineralogical composition (K1 and K4 from Czech Republic and K2 and K3 from Germany (Weber et al. 2014)). For comparison the ΔGSD was calculated for quartz and hematite as common minerals in loess sediments. To concentrate on the grain size range which is in loess sediments naturally dominated by post-depositionally formed particles (< 1µm (Schaetzl and Thompson, 2015), the standard samples were milled and reduced to the submicron fraction.

The investigated loess-paleosol sequences represent widely different sedimentation regimes and stages of weathering and soil development: The Semlac section is a natural outcrop exposed at a loess plateau (46° 7'12.97"N / 20°56'54.70"E / ~100 m a.s.l.; Arad Plain, Romania) consisting of very homogenous and relatively fine loess sediments accumulated under rather constant sedimentary conditions since the MIS 10 (Middle to Late Pleistocene). During interglacial and interstadial periods (MIS 3, 5, 7 and 9) well developed paleosol complexes were formed. The intense soil formation in the surrounding mountains of the Carpathian Basin, in comparison to sections in the central part, is due to the higher moisture availability during the Pleistocene (Zeeden et al., 2016). The GS-variation is mainly influenced by post-depositional processes such as clay mineral relocation and –formation (Schulte et al., 2014; Zeeden et al., 2016). The second example is a vibracore from Düsseldorf-Grafenberg (51°14´49´´N, 6°50´58´´E; 83 m a.s.l.) on the eastern side of the Rhine River (Germany). It consists of calcareous loess accumulated during cold and semiarid conditions of the last glacial cycle and intercalated soils and soil sediments which have developed under interglacial, interstadial and periglacial conditions (Schulte et al., 2016). The third archive is the Suohuoduo section (34°28’ N, 99°50’ E; 4016 m a.s.l.) located on the northeastern margin of the Tibetan Plateau (China). It developed during the late Pleistocene and Holocene and consists of loess, Chernozem-like paleosols and a slightly weathered modern soil (Lehmkuhl et al., 2014).

For grain size analysis each of these loess-paleosol sequences was sampled in the same resolution (1 sample every 5 cm). In rare cases individual samples are missing.

Tab. 5.4.1: Properties and references of the selected standard material samples.

83

Sample Specification Pretreatment Reference washed with >3 M Illite Fithian (89 % Illite, 9 % KCl (see Weber et al., Illite Weber et al.2014 Quartz) 2014 for further details) milled for 16 min by a planetary mill; the SiO2, washed and calcined for fraction < 1 µm was Quartz Merck KGaA analysis, CAS-No: 14808-60-7 separated by Atterberg sedimentation the fraction < 1 µm NCS DC 14033, Certified China National was separated by Hematite Reference Material (61.68 % total Analysis Center for Atterberg Fe) Iron and Steel, 2005 sedimentation Dorfner Industrial sample (81 % washed with >3 M Kaolinite, 0.08 % Quartz, 0.06 % KCl (see Weber et al., H1, Kaolinite Weber et al.2014 Illite, 0.05 % Microcline, 18.81 % 2014 for further unknown rest) details) Certified Reference Materials, organic matter was CRM No 5358-90, (Bulk clay oxidized by H2O2; the Caspian Light- mineralogy: 8-12 % Illite, 5-10 % fraction < 1 µm was State University Chestnut Chlorite, 10-15 % Kaolinite, 3-5 % separated by Irkutsk, 1990 Colored Soil Monmorillonite, 30-35 % Quartz, Atterberg other minerals: feldspar, sedimentation carbonates, oxides, heavy minerals)

Certified Reference Materials, organic matter was CRM No 5359-90, (Bulk clay oxidized by H2O2; the Kursk mineralogy: 12-15 % Illite, 4-6 % fraction < 1 µm was State University Chernozemic Chlorite, 4-8 % Kaolinite, 12-15 % separated by Irkutsk, 1990 Soil Monmorillonite, 35-40 % Quartz, Atterberg other minerals: feldspar, mica, sedimentation oxides, heavy minerals)

5.4.2.2. Laser diffraction measurement For particle size analysis, all samples were dried at 35 °C, homogenized and passed through a 2 mm sieve. Organic matter was removed by treating the samples with 0.70 ml 30% H2O2 at 70 °C for several hours. This process is repeated for a maximum of three days until a bleaching of the sediment occurs (Allen and Thornley, 2004). To keep particles dispersed, the samples were treated with 1.25 ml Na4P207 for 12 hours in an overhead shaker (Pye and Blott, 2004; DIN ISO 11277, 2002). Particle size was measured with a Laser Diffraction Particle Size Analyzer (Beckman Coulter

84

LS 13 320 with additional PIDS technology) calculating the percentaged size frequency within a size range of 0.04–2000 μm with an error of 2%. The light scattering is detected at 134 detectors which are arranged in angles between 0° and 144°. By matrix transformation of the scattering pattern a particle size distribution with 116 data points is calculated.

To increase accuracy two different subsamples were analyzed for each sample. Each subsample was measured twice. The results are commonly shown as frequency distribution density curves, based on the proportion of particles in each size class. Further details concerning the measurement and the sample preparation are given by Schulte et al. (2016). For several samples the submicron (<1µm) part was separated by sedimentation in Atterberg cylinders (Atterberg, 1911). Subsequently, this submicron suspension was viewed under an optical microscope with a 100x objective (Carl Zeiss Microscopy GmbH, Axiostar Plus).

5.4.2.3. Optical model In order to compute the grain size distribution (GSD), the detected light scattering matrix is transformed into a set of individual GS classes, and the relative frequency of volume of equivalent spherical particles of the respective GS range is determined. This transformation includes the detection of the additional polarization intensity differential scattering system (PIDS) which measures the scattering of polarized (horizontal and vertical) light with additional wavelengths (450nm, 600nm and 900nm). Since the analysis is completely integrated, a single solution is obtained although two methods are used (ISO 13320, 2009).

There are two different recently common optical models to calculate the GSD from the light scattering pattern: the Fraunhofer approximation (FH) and the Lorenz-Mie Theory (hereafter “Mie Theory”). According to the Fraunhofer approximation, some of the light hitting a particle is diffracted in a grain size dependent angle. The angle of diffraction increases with decreasing grain size (Blott and Pye, 2006). Contrary to the Fraunhofer approximation, which takes only the diffraction at the particle surface into account, the Mie Theory also considers the complex refractive indices of the particles and the suspension fluid (ISO 13320, 2009). The Mie Theory was deduced from the theory of movement of light inside optically heterogeneous and translucent objects devised by Lorenz (1883). Maxwell Garnet (1906) adopted this theory to colloidal metal suspensions considering the wavelength of light and the distance between the particles. Using the Maxwell equations (Maxwell 1898) Gustav Mie (1908) demonstrated the dependency of light scattering in gold suspensions on the wavelength of light and the particle diameter. Debye (1909) described the interaction of incident light and reflection, scattering and absorbing at spherical 85

particles. These authors showed that the light scattering within a particle suspension depends on the wavelength of the light, the complex refraction index of the suspended minerals and the particle size. If the first two arguments are given, the scattering pattern can be used to calculate the grain size of the suspended particles by mathematical transformation of the diffraction pattern (Kerker, 1969; de Boer, 1987).

For the application of the Mie Theory, two indices have to be defined: the refractive index of the fluid (here 1.33 for water) and the complex refraction index (m = n – ki) of the sample composed of the real component (n) reflecting the refraction within the particle and the imaginary component (ki) which considers the light absorption (Buurman, 1997b; Murray, 2002; Özer et al., 2010). Since the complex refraction index varies for different minerals (Haynes, 2015) and natural samples represent a polydisperse mineral mixture, a suitable medium value has to be implemented. For most of the common soil forming minerals the real component varies between 1.4 and 3.2 and the imaginary component between 0.01 and 0.2 (Özer et al., 2010). In this study, the complex refraction index is fixed to m = 1.55 – i0.1 (Mishchenko, 1993; Özer, 2010).

Nevertheless for several natural sediment samples, in particular in the case of post-depositional altered minerals, the optical properties may differ considerably. Except when very specific analysis of the required parameters are conducted a priori, which is almost unfeasible for most users, the indices are set as generalized unique values. Introduction of inaccurate parameters may lead to undesired artefacts in the obtained GSD (Konert and Vandenberghe, 1997; Keck, 2006). Hence, in many studies the Fraunhofer approximation is used (Loizeau et al., 1994; Konert and Vandenberghe 1997; Xu and Di Guida, 2003; Blott and Pye, 2006), especially if the precision of the measurement is more important than the absolute accuracy (Pye and Blott, 2004).

The angle-dependent scattering pattern calculated by the Mie Theory approaches the scattering pattern calculated by the Fraunhofer approximation with increasing grain size. Specifically, the scattering at particles larger than ten times the light source wavelength (in the case of the Beckman Coulter LS 13320: 10*780 nm = 7.8 µm) is virtually independent of the optical properties (described by the complex refraction index) of the sediment particles (McCave and Syvitski, 1991; Bayvel and Jones, 1981; Lagasse and Richards, 2003; Loizeau et al., 1994). In contrast, in the submicron range substantial differences occur between the calculated GSDs of both models.

In fact, neither of the common optical models provides the true size of the submicron particles. However, since the complex refractive index depends on the mineral properties, the difference of the GSDs calculated with both models is sensitive to the mineral composition of the sample:

86

ΔGSD GSD – GSD (1)

5.4.2.4. Data preparation and illustration Grain-size distributions are compositional in nature, which means that a GSD of one sample is mathematically defined as a positive vector with a constant sum (e.g. 1 for proportions or 100 for percentages) over all grain size classes (Aitchison 1986; Roberson and Weltje 2014). Following, the dataset is in a closed number space (positive and ≤ 100). Multivariate and most bivariate statistics, such as the Pearson's correlation coefficient (Pearson 1896), require data sets in a real number spaces. To solve these problems, a data transformation is applied. Following Aitchison (1986) each single grain size value (xi) is divided by the geometric mean of the respective GSD (g(GSD)), to minimize the relative dependency of each grain size class value on all other classes (the constant sum problem). Due to their asymmetry, if xi < g(GSD), the ratio is in the range of 0 to 1 and if xi

> g(GSD), the ratio is in the range of 1 to xmax/gGSD), such ratios are mathematically doubtful (c.f. Weltje and Tjallingii 2008; Weltje et al. 2015). Aitchison (1986) solved this problem by calculating the natural logarithms (log) of the ratios, so that the calculated values are symmetric:

(2) and spread around 0, which corresponds to the g(GSD):

log log1 0 (3)

This centered-log-ratio transformation (clr) was established by Aitchison (1982) and is suitable to transform grain size distributions from closed to real number space (Weltje and Tjallingii, 2008; Roberson and Weltje, 2014). The clr-transformation is applied to both GSDs prior to the calculation of the differential:

ΔGSD clrGSD – clrGSD (4) for i = grain size classes 1 to 116:

ΔGSDclr log log ,…,log log (5)

For the illustrations of the ΔGSDs of the standard and natural clay mineral samples (Fig. 5.4.1C/D) the differences are calculated from the raw data following equation (1). The advantage is that the absolute differences between the respective results obtained by the two optical models are

87

emphasized. Here, this approach is suitable because large proportions of the entire distributions are in the submicron range, where we focus on in this study. In Figure 5.4.3 the ΔGSD for the

Semlac section is shown both as clr-transformed (ΔGSDclr) and as raw data differences (ΔGSDraw).

Figure 5.4.3A illustrates the relative dependency of variations within a single grain size class on all other classes. As the submicron grain size range constitute only a small proportion of the bulk GSD, slight submicron variations can be masked by relative high absolute grain size frequencies within coarser classes (15-30µm fraction in Fig. 5.4.3A). Therefore, the results of the loess-paleosol sequences (Fig. 5.4.3B, 5.4.4 and 5.4.5) are depicted as clr-transformed grain size differentials according to equation (5). For comparison, the heatmaps depicting the ΔGSDraw for the sequences Düsseldorf-Grafenberg and Suohuoduo are presented in the supplementary material (Appendix Fig. B1 and B2).

5.4.3. Results 5.4.3.1. Submicron fraction of single samples The grain size distributions of submicron suspensions calculated with the two different optical models show considerable differences. The deviations of both distributions can be emphasized by the ΔGSDraw. The results of the laser diffraction analysis of several samples of known mineral composition are shown in Fig. 5.4.1 as frequency distribution density curves and heatmap signatures.

The ΔGSDraw of submicron hematite reflects more fine particles for the GSDs obtained by the Fraunhofer approximation than by the GSD based on the Mie Theory. Furthermore, the mode of the GSD is lower if the Fraunhofer approximation is applied (FH_mode: 0.29 µm; Mie_mode: 0.52 µm). In the case of the milled and separated submicron quartz samples the GSD obtained by the Mie Theory is wider, whereas the mode of the distribution remains unchanged at 0.62 µm. The GSDs of industrial kaolinite H1 (submicron) and Illite (bulk) show distinct differences in the entire distributions of these two samples calculated with the Fraunhofer approximation (Fig. 5.4.1 A). In contrast, the GSDs calculated with the Mie Theory (Fig. 5.4.1 B) reflect only slight differences in the cC and fSi fractions. For the submicron fractions of the two Russian soil standards (illite contents of the bulk samples: 8-12% and 12-15 %), the ΔGSDraw s show a distinct peak within the mC fraction. Remarkably, the systematic signature in the medium clay fraction is very similar to the

ΔGSDraw of the illite bulk sample (89 % illite) and the submicron part of the industrial kaolinite (H1, bulk mineralogy: 0.06 % illite), as illustrated by the heatmap pattern (Fig. 5.4.1, left column). This systematic peak is visible for all measured illite containing clay mineral samples (Fig. 5.4.1, right column). Additionally, the lows of the mC peaks show a distinct relationship with the 88

respective illite contents of these kaolinite dominated clay mineral samples. Further the GSDs are dominated by the submicron GS range (in case of the Mie calculations: between 90.76 and 93.01 vol. %). An exception is the industrial kaolinite H1 where the proportion < 1µm contains only 72.89 vol. %. The subsequent microscopic examination of the separated submicron suspension of a clay rich paleosol sample revealed the presence of several rod and plate shaped particles with sizes up to 25 µm in one dimension (Fig. 5.4.2A). In contrast, within the submicron fraction of a pure quartz sample only particles around 1 µm and smaller are detectable (Fig. 5.4.2B).

Fig. 5.4.1: Grain size distribution curves of four standard mineral samples and two Russian standard soils reduced to the submicron fraction (left column). The specifications of these samples are given in Tab. 5.4.1. The right column shows the results of five natural clay mineral samples. The mineral composition of the clay mineral samples are published in Weber et al. (2014). The GSDs are presented as: (A) calculation with Fraunhofer approximation, (B) calculation with Mie

Theory, (C) ΔGSDraw as curves and (D) ΔGSDraw as a heatmap. The dotted lines separate the classic

89

GS fractions fC: 0.04 – 2 µm, fSi: 2 – 6.3 µm, mSi: 6.3 – 20 µm, cSi: 20 – 63 µm, fS: 63 – 200 µm, mS: 200 – 630 µm, cS: 630 – 2000 µm.

It has been shown that for some specific mineral groups the scattering behavior has an impact on the calculated GSD which can be emphasized by the differential consideration of the results obtained by the two optical models (the ΔGSDraw signature).

Fig. 5.4.2: Optical microscope photograph showing the submicron fraction after separation by Atterberg sedimentation of (A) a sample from the S1 palaosol of the Semlac sequence and (B) a pure quartz sample (s.f. Tab. 5.4.1).

5.4.3.2. Loess sections

In the Semlac section, the ΔGSDraw and ΔGSDclr signals for the submicron range are considerably prominent (negative values) within the interglacial and interstadial paleosols, and weak (values around zero) in the carbonate containing samples (Fig. 5.4.3). The carbonate free loess units L3 and L4 show weak signatures within both of the presented ΔGSD heatmaps, whereas unit L1L2 shows a similar signature as the paleosols. 90

Fig. 5.4.3: Heatmap showing the ΔGSDraw (A) and the ΔGSDclr (B) of all samples from the loess section Semlac (Romania). The vertical curves represent the contents the fraction 15-30 µm and <1µm, the CaCO3 content and the magnetic susceptibility (MS) (Zeeden et al., 2016).

A comparison of the ΔGSD signatures for the clay fraction to the classic pedogenic proxies, such as clay content (<2µm), lightness L* and magnetic susceptibility (Zeeden et al., 2016) indicates a general interrelation between parts of strong GSD differences, and classic proxies of post-

91

depositional chemical weathering and pedogenesis. However, whereas, especially the ΔGSDclr shows sharp transitions between the loess and the soil units, the variations of the other proxies are not abrupt but gradual.

Generally, the ΔGSDclr values of the Düsseldorf-Grafenberg sequence are significantly weaker in comparison to the Semlac section. The Holocene soil and the possibly interstadial cambisol show the highest differences in the clay fractions, whereas for the clayey soil sediment from 13.6 m to

17.4 m the ΔGSDclr is weaker (Fig. 5.4.4). The samples between 2.6 m and 12.2 m which are correlated with the MIS2 show consistently the slightest ΔGSDclr values. For comparison the classic clay values are given in Fig. 5.4.4 as depth functions. The increased values of the submicron fraction between 4.8 m and 6 m and between 9.8 m and 10.7 m are not reflected by the ΔGSDclr in the clay fractions. Excluding the corresponding samples the submicron fraction and the ΔGSDclr exhibit good correlation (R² = 0.79; n = 284).

Fig. 5.4.4: Heatmap showing the ΔGSDclr of all samples from the Düsseldorf-Grafenberg sequence. The vibracore consist of calcareous loess and intercalated interglacial, interstadial and periglacial soils and soil sediments. The vertical curves represent the content of submicron particles.

92

In the Suohuoduo section the ΔGSDclr values (Fig. 5.4.5) for the submicron GS range are elevated both in the weak cambic horizon (L01, 90-5 cm) and in the two chernozem paleosols S01 (120–90 cm) and S02 (200–150 cm). The extent of the deviation is comparable to the weathered sediments of the Grafenberg core. The unweathered loess sediments (L1, 312.5–265 cm and L03, 265– 200 cm) and the transition unit L02 (150–120 cm) in between the two paleosols show comparably weak ΔGSDclr values. Furthermore, the behavior of the Corg contents and the frequencies of the submicron particles show a substantial correlation with the ΔGSDclr (R² = 0.75 and R² = 0.8; n = 56).

Fig. 5.4.5: Heatmap showing the ΔGSDclr of all samples from the Suohuoduo section, containing pure loess (white), slightly weathered loess (light grey) and chernozem-like paleosols (S01; S02: dark grey). The vertical curves represent content of submicron particles and Corg.

5.4.4. Discussion 5.4.4.1. Quality check of the clay measurement In order to evaluate the extent of the generally accepted underestimation of the submicron percentage in comparison with settling techniques (e.g. Loizeau et al., 1994; Konert and Vandenberghe, 1997; Roberson and Weltje, 2014), we measured the submicron fraction exclusively

93

after prior separation by Atterberg sedimentation. The expectation, that there is a distinct overestimation of the submicron particle sizes in particular for the samples from clay rich soils was not confirmed. If the Mie Theory is applied for calculation, the GSDs of the separated particle suspensions are predominant in the submicron fraction (< 1µm, Fig. 5.4.1 B). The subsequent microscopic examination of a submicron suspension extracted by Atterberg sedimentation from a paleosol sample revealed the presence of several rod and plate shaped particles with sizes up to 25 µm in one dimension (Fig. 5.4.2A). As shown by several authors (e.g. Konert and Vandenberghe, 1997; Buurman et al., 2001; Eshel et al., 2004) neither the settling technics nor the laser diffraction detects the correct volume based equivalent spherical diameter of such irregularly shaped particles. With respect to our observations, we assume that the proportion > 1 µm, detected by the laser diffraction analyzer, is the result of higher content of rod/plate shaped particles. Furthermore, there is an overestimation of the clay fraction in sedimentation based techniques (cf. Eshel et al., 2004). We can conclude that the phenomenon of systematic underestimation of the submicron percentage in comparison with settling techniques is negligible, if the LS 13320 with additional PIDS technology is used for laser diffraction measuring and the Mie Theory is applied for GS calculation.

5.4.4.2. Dependency of the ΔGSDraw on the mineral composition Light scattering is a function of particle size, particle shape and optical properties described by the complex index of refraction (m = n + ik) (e.g. Müller et al., 2009). In this section, we focus on the dependence of the ΔGSDraw from (1) the particle shape and (2) the optical parameters.

The parameters of the Mie Theory (real and imaginary part of the refractive index) were fixed to (1.55 + i0.1) and the particle suspensions of the standard samples were reduced to the submicron fraction. Consequently, the deviations of individual modes of the GSD calculated with the Mie

Theory and the Fraunhofer approximation (ΔGSDraw) can be discussed as a signature for different scattering properties of the mineral suspensions.

(1) The shape of the mineral particles depends on their mineral composition (e.g. Veghte and Freedman, 2014), on the process regime during transport, sedimentation and post-depositional alteration (Volten, 2001). The particle shape has an effect on the optical properties and consequently on the scattering pattern of the particle suspension (Mishchenko, 1993; Veghte and Freedman, 2014; Munoz, 2006; Lindqvist, 2014; Alexander, 2015).

Veghte and Freedman (2014) measured the aspect ratio of submicron dust minerals using a scanning electron microscope in top-down and side-on orientation. They found aspect ratios of 1.3 to 1.64 for calcite and quartz particles and considerably larger ratios for aluminosilicate clay 94

minerals (1.35 to 1.44 in top orientation and 4.8 to 9.14 in side orientation). Non-spherical shaped clay minerals with such high aspect ratios have different optical properties and result consequently in different scattering patterns than particles which are more sphere-like shaped (Volten et al., 2001; Veghte and Freedman, 2014; Klüser et al., 2016).

The shape effect particularly affects the scattering properties of dust particles coarser than 1 µm (Lindqvist et al., 2014). In contrast, within the submicron range especially if the particles are particularly smaller than the wavelength of the incident light, the scattering behavior is much less dependent on the true particle shape (Vengte et al., 2015). This is because the scattering in this GS range depends more on particle volume than on surface area (Mishchenko, 1993; Klüser et al., 2016). Despite potentially high aspect ratios of the clay minerals in comparison to the more spherical shaped quartz particles (Fig. 5.4.2A/B; cf. Veghte and Freedman, 2014), the grain size calculation using the Mie Theory provides appropriate volume equivalent sphere diameters for the submicron particle suspensions.

Furthermore, for most of the submicron particles which are in the scope of this work, the different shapes play a subordinate role. To this end, in spite of the likelihood of shape differences, the extracted submicron fraction from the Russian standard soil samples, the industrial kaolinite and the Illite Fithian show strikingly similar curve progressions of the ΔGSDraw (Fig. 5.4.1 C).

(2) The second factor influencing the scattering behavior is the complex refractive index. Particularly, the imaginary part describing the light absorption has a distinct influence on the scattering behavior (Munoz et al., 2006; Müller et al., 2009). Müller (2009) found that the scattering at dust particles is influenced by light absorption in the GS range of 0.5 to 10 µm. There is no interrelation detectable below 0.5 µm. Furthermore, according to Veghte et al. (2015) at particle sizes smaller 0.4 µm, the result of the Mie Theory is more scattering dominated, while at sizes larger than 0.4 µm the results are dominated by absorption. Consequently, we suggest that the significantly different ΔGSDraw of the hematite samples (Fig. 5.4.1) is a function of the irregularly shaped submicron particles rather than a direct reaction to the higher absorption capability of the hematite minerals. Accordingly, in contrast to non-absorbing submicron particles where the light scattering is not predominantly shape dependent (see above), for absorbing particles (e.g. the analyzed hematite) the shape has a substantial impact on the scattering pattern of submicron particles. Similarly, Veghte et al. (2015) showed that the extinction efficiency, which influences the scattering behavior, is more shape dependent for absorbing particles. Additionally, metallic effect and birefringence could enhance the complexity of the light scattering pattern of iron minerals (cf. Volten et al. 1999). For the GS calculation using the Fraunhofer approximation it is assumed that the particles absorb the light totally and diffraction is the only interaction leading to the scattering

95

pattern (ISO 13320, 2009). The complex scattering pattern may cause the submicron peak position of the hematite measurement at smaller GS classes when applying the Fraunhofer approximation (Fig. 5.4.1 A).

The comparison of the calculations for the quartz measurements exhibited variation of steepness and narrowness of the peaks, but the peak position remained constant (Fig. 5.4.1). Presumably, the higher similarity of the two calculations is a function of the regular and sphere-like shape of the particles (Fig. 5.4.2B). The light scattering pattern is not as complex and can therefore be appropriately estimated with the Fraunhofer approximation.

The absorption und scattering behavior of clay minerals are still insufficiently studied (Linke et al.,

2006; Veghte and Freedman, 2014; Veghte et al., 2015). However, the similarity of the ΔGSDraw signatures of all clay mineral containing samples indicates that the effect of the respective optical properties of different clay mineral species is negligible.

The smaller the particles the lesser is the influence of the shape and absorption on the scattering behavior (Mishenko, 1993; Linke et al., 2006; Veghte et al., 2015). In the submicron range the light scattering seems to be most influenced by the mineral specific refraction index. However, under certain conditions there is an additional influence of the shape or absorption capacity of the particle suspension on the scattering pattern. For some mineral groups, such as clay minerals, these dependencies are reflected by the GSD and can be emphasized by the calculation of the ΔGSDraw.

Summarizing the results of the clay-size mineral analysis suggests that especially the crystalline properties of the polydisperse mineral suspension are reflected by the curve progression of the

ΔGSDraw. This hypothesis is supported by the relationship of the intensity of the mC peak, whose position is comparable to the illite sample (Fig. 5.4.1, left column), and the illite content of the kaolinite-dominated clay mineral samples (Fig. 5.4.1, right column).

5.4.4.3. ΔGSDclr as proxy in loess research Loess consists mainly of coarse silt (Pecsi, 1990; Pye, 1995; Muhs and Bettis, 2003), whereas clay sized particles < 1 µm are extremely underrepresented during accumulation (Qiang et al., 2010;

Ujvari et al. 2016). Therefore, we suggest the variation of the ΔGSDclr as a strong estimator for the degree of post-depositional GS fractionation due to chemical weathering processes (silicate weathering).

The aeolian sediments presented in this study have experienced varying degrees of post- depositional alterations under fluctuating climatic conditions during up to the last 400 ka years 96

(Lehmkuhl et al., 2014; Schulte et al., 2016; Zeeden et al., 2016). During periods in which water is present in the aeolian sediment body, the mineral compositions, or, at least, the properties of single minerals, are altered by chemical weathering processes – especially by silicate weathering and the formation of new clay minerals (Sun et al. 2006; Schaetzl and Thompson, 2015).

There are many widely used proxies to estimate the vertical variation and the different intensities of the post-depositional weathering: geochemical weathering indices (Buggle et al., 2011; Ujvari et al., 2014), grain size indices (Vandenberghe et al., 1997; Antoine et al., 2009a,b) or rock magnetism (Buggle et al., 2014; Hosek et al., 2015). Sun et al. (2006) isolated the quartz minerals from the bulk by means of chemical pretreatment and discussed it as an indication of pure windblown dust deposit. Furthermore, they suggested the correlation coefficient of the bulk GSD and the isolated quartz GSD as an indicator for the relative intensity of pedogenesis (Sun et al. 2006).

As discussed above the ΔGSDraw appears to be highly depended on the mineral composition of the analyzed particle suspension. The same signature as interpreted above as a signal for the illite concentration is visible in the ΔGSDraw heatmaps of the three loess-paleosol sequences in different intensity (Fig. 5.4.3A and Appendix Fig. B1 and B2). Chemical weathering comprises the decomposition of primary silicate minerals (in particular mica and feldspar), the release of weathering products and the formation of new minerals such as secondary clay minerals (Schaetzl and Thompson, 2015), which results in a post-depositional modified sediment with deviating refraction properties within the submicron fraction in comparison to the parent loess. When analyzing polyprocessual bulk samples, there are considerable constraints due to the compositional data effect (see Chapter 2.4). Therefore, for discussing the variability within the loess-paleosol sequences we use the clr-transformed ΔGSDs, as these are more robust against other grain size influencing processes as an enhanced input of fine grained background sediment (2 – 30 µm) or local inputs of sand due to saltation or sheet wash (c.f. Zeeden et al., 2016). The ΔGSDclr variability with depth is supposed to be a function of the degree of post-depositional silicate weathering of loess sediments (Fig. 5.4.3 – 5.4.5).

5.4.4.4. ΔGSDclr variability within the loess-paleosol sequences In comparison to almost unweathered parts of the Semlac section (e.g. from 410 cm to 480 cm depth (Zeeden et al., 2016)) within the soil complexes the mSi fraction which contains mica and feldspar (Nemecz et al., 2000; McCave et al., 2006) is reduced whereas the clay fraction is enhanced.

When comparing the ΔGSDraw signature of the clay mineral analysis above (Fig. 5.4.1) and the

ΔGSDraw heatmap of the Semlac section (Fig. 5.4.3A), the clay enhancement seems to be due to 97

the existence of illite or similar secondary clay minerals with similar mineral properties. The interrelation of the ΔGSDraw and ΔGSDclr signals and the other weathering proxies (magnetic susceptibility and soil color, (Zeeden et al., 2016)) supports the pedogenic origin of the fine clay minerals highlighted by the ΔGSDs. The deviating course of the submicron GS fraction could be a function of the interaction of sedimentation and subsequent pedogenesis (Zeeden et al., 2016). Furthermore Zeeden et al. (2016) stated that the soil complexes developed during the last interstadials and interglacials are comparatively intensely weathered which is due to the high moisture availability at this position. This causes the considerably distinct values of the ΔGSDs for the Semlac section in comparison to the other section presented in this study.

The ΔGSDclr of the Grafenberg sequence indicates a similar interrelation with the degree of in-situ soil formation. However, the generally less intense ΔGSDclr values of the Grafenberg sequence reflect the absence of interglacial in-situ soil complexes. Only the reworked soil sediment in the lower part of the section (17.4 m to 13.6 m) might be altered by pedogenesis during the last interglacial (P. Fischer, personal communication). However, there are clear indications for two in- situ soil formations: the modern soil (plough horizon) on top of the section and a calcaric cambisol (12.9 m to 11.2 m) which developed in reworked material during the last interstadial (Schulte et al.,

2016). Both soil formations are clearly displayed by the ΔGSDclr signal (Fig. 5.4.4). In contrast, in the entire range between 11.2 m and 2.6 m depth the ΔGSDclr signature is consistently around zero. For the reworked soil sediment, indicated by the sediment color and elevated submicron particle frequencies (Fig. 5.4.4), no in-situ soil formation is supposed (P. Fischer, personal communication). The units interpreted as loess and gleysols should also not be altered by intense silicate weathering (Buggle et al., 2014; Gocke et al., 2014; Schaetzl and Thompson, 2015). This indicates that the

ΔGSDclr is independent of initial alteration processes like weak sediment relocation, decarbonisation and gleyzation and the mere presence of submicron particles. In fact, it corroborates the hypothesis that the ΔGSDclr is a function of in-situ silicate weathering.

In accordance with the previously presented sections, the ΔGSDclr signal of the Suohuoduo section emphasizes the parts of the section that experienced in-situ soil formation and remains around zero in the unaltered loess parts (Fig. 5.4.5). Despite the indications for a stronger weathering in the paleosols S01 and S02 (Lehmkuhl et al., 2014) reflected by dark soil color and maxima of the

Corg contents, the ΔGSDclr signal is not prominent in comparison to the recent soil formation. Under semiarid climatic conditions, when these steppe soils were formed, no intense silicate weathering occur (Antoine et al, 1999; Schaetzl and Thompson, 2015).

Summarizing, the ΔGSDclr signal within the submicron range of bulk measurements can be established to function as a proxy to characterize the intensity of post-depositional GS fractionation

98

due to chemical weathering processes (silicate weathering). The ΔGSDclr signal is virtually unaffected by increasing of submicron particles due to cryogenic processes, secondary carbonate dynamics, weak relocation of inherited weathering products and synsedimentary processes (sheet wash, saltation or enhanced background sedimentation).

5.4.4.5. Potential sources of error The sources of error involve a possible deposition of secondary clay minerals which are formed in the source area of the aeolian deposit. Such allochthone clay minerals could be deposited as (1) single particle or (2) aggregate which endured the aeolian erosion and transport.

1. Zhang and Shao (2014) developed a new model for dry particle deposition and concluded that in the range of 0.1 to 1 µm hardly any particles are deposited. In the case of wet deposition the “Greenfield Gap” describes the range between 0.1 and 1 µm as a range where the coagulation of raindrops with particles distributed in the atmosphere is minimal (Greenfield, 1957; Ardon-Dryer et al., 2015; Xu et al., 2016). Particles in this GS range are less affected by washout processes, thus, they exhibit long atmospheric residence times (Pranesha and Kamra, 1997). Ujvari et al. (2016) confirms this concept, experimentally determined, by changes of the GSD in the atmosphere after several rainfall events of different durations. Thus, clay sized particles < 1 µm are extremely underrepresented in loess sediments during deposition (Qiang et al., 2010; Ujvari et al., 2016).

2. Previously weathered clay sized particles could also be deposited in the form of aggregates or by adhering to larger mineral grains during erosion, transport and accumulation (Pye, 1995; Volten et al., 1999; Qiang et al., 2010). According to Ujvari et al. (2016) most of the aggregates break up into single grains during erosion or transportation by collisions. Fine grained pedogenic clay minerals adhering to larger particles do not survive the aeolian transport in a substantial quantity. However, a preservation of stable microaggregates containing inherited pedogenic clay minerals cannot be ruled out (Schulte et al., 2016) and may have an effect on the GSD of analyzed loess samples.

5.4.5. Conclusion It has been shown that in the submicron range of bulk sample measurements the deviations of the

GSD calculated with the Mie Theory and the Fraunhofer approximation (the ΔGSDraw) can be used as a signature for the scattering properties of polydisperse mineral suspensions. For some specific mineral groups the scattering behavior has an impact on the calculated GSD which can be emphasized by the calculation of the ΔGSDraw. The submicron mineral analysis (Fig. 5.4.1, left column) show that different crystalline properties are reflected by different signatures of the 99

ΔGSDraw. The intensity of the mC peak, whose position is comparable to those of the pure illite sample, is dependent on the illite content of the natural clay mineral samples (x-ray diffraction in

Fig. 5.4.1 right column). The prominent values of the ΔGSDraw signature of the mC fraction within the paleosols of the loess sequences reflect the existence of illite or similar secondary clay minerals with similar mineral properties.

During accumulation loess consists mainly of silt sized particles, whereas clay sized particles < 1 µm are extremely underrepresented. The variability of low concentrated grain size ranges within percentage-frequency distributions can be misleading due to the compositional data effect (constant sum of 100%). Therefore, the vertical variability of bulk sample distributions within loess- paleosol sequences are analyzed as clr-transformed grain size differences (ΔGSDclr), as these are more robust against other processes influencing the entire grain size distribution especially during sediment accumulation (sheet wash, saltation or enhanced background sedimentation).

Summarizing, we present the ΔGSDclr as a comparatively time- and cost-effective estimator for secondary clay mineral formation as a function of post-depositional chemical weathering (silicate weathering).

This proxy can be used as a benchmark for paleopedogenesis within loess-paleosol sequences and for inter-sequence comparison if the same laser diffraction device is used for GS measuring. In comparison to classic grain size parameters, it is virtually unaffected by aeolian sorting effects, the distance to the source regions, cryogenic processes, weak reworking of inherited weathering products and carbonate dynamics. Furthermore, this study shows that there is a significant post- depositional grain size shift due to pedogenesis. Consequently, grain size based time series analysis to reconstruct quaternary sedimentary environments must be considered in a critical way, especially if the clay content is involved as a grain-size parameter.

In addition, the initial tests of this study show that the phenomenon of systematic underestimation of the submicron percentage in comparison with settling techniques is negligible, if the LS 13320 with additional PIDS technology is used for laser diffraction measuring and the Mie Theory is applied for GS calculation.

100

Synthesis

The synthesis section summarizes the results that were obtained in the manuscripts (Chapter 5) and gives general implications for the individual steps of grain size data collection for loess and loess-like sediments.

6.1. Sample pretreatment prior to laser diffraction size analysis

Sample pretreatment is often applied to reconstruct the characteristics of the particles at the time of accumulation, typically by dissolving aggregates and removing binding agents (Murray, 2002). With increasing post-depositional alteration (loessification, chemical and physical weathering, aggregate formation, and pedogenesis), this intension becomes progressively difficult or even impossible. Thus, for post-depositional altered samples, only the absolutely essential pretreatment steps (wet dispersion and oxidation of organic matter) should be applied (Chapter 5.3.). For grain size analysis of a loess-paleosol-sequence, all samples should be pretreated uniformly to ensure comparability. In general, as each pretreatment is a modification of the original sample and as our results show that pretreatment is particularly selective and inscrutable, the sample pretreatment should be as gentle as possible to minimize grain size modification and, hence, any misleading effects on the measurement results (Chapter 5.3.).

During the first step of laboratory sample preparation, the air-dried macroaggregates are disaggregated using a pestle and mortar. This must be done carefully so that the abrasion of contained minerals is as little as possible, but enough so that the material passes through a 2-mm sieve. Subsequent analysis were undertaken on this < 2 mm fine fraction. Problems can occur if the sample contains fragile components such as volcanic glass particles.

The oxidation of organic components and binding argents is mandatory prior to grain size analysis.

Typically, organic matter is destroyed by hydrogen peroxide (H2O2) (e.g., Rowell, 1994; DIN ISO 11277, 2002). For intensively weathered soils containing stable post-depositional formed aggregates, H2O2 treatment does not result in a complete oxidation of the organic matter. However, oxidation is much more effective if applied subsequent to HCl treatment as indicated by faster and more intense bleaching of the samples (Chapter 5.3.).

Furthermore, the formation of flocculation is a major obstacle for robust measurements of the prepared sediment suspensions. This depends on the forces between the suspended particles. Due to Brownian motion, they approach close together. Whether two particles agglomerate depends on the potential barrier between them. The potential energy consists of two forces: the van der Waals

101

forces effect attraction, while the electrical double layers around the particles cause repulsion. When two adjacent particles approach close enough to cross the potential barrier (attractive forces exceed repulsive forces), they collide and probably remain attached due to the strong van der Waals forces (Allen, 1990; Rowell, 1994). Generally, flocculation can be avoided or reduced by chemical or physical means. If the sample contains fine silt and clay particles (high surface area), the admixture of a dispersing agent, such as calgon, sodium hexametaphosphate, tetrasodium pyrophosphate, or similar substances, is mandatory to facilitate proper dispersion by surfactant coating on particle surfaces (Chappell, 1998; Kaiser and Guggenberger, 2003). By such chemical dispersion the surface energy and following the attraction between the touching particles is reduced. Figure 6.1 shows the results of a test in which the concentration of tetrasodium pyrophosphate as a dispersing agent was evaluated. The concentration should be at least 2 g/L to achieve a sufficient dispersion.

Fig. 6.1: GS distribution curves of three loess samples on a different state of pedogenesis. Prior to measuring each sample was dispersed in tetrasodium pyrophosphate solutions of different concentrations (given in g/L).

102

Additionally, an ultrasonic treatment may be performed to effect a short-time separation of the flocculations and facilitate chemical wetting (Allen, 1990; McCave et al., 2006). However, the ultrasonic treatment may also yield undesirable effects, including the formation of air bubbles, breakup of quartz grains, or immediate reaggregation (Chappell, 1998; Machalett et al., 2008; Asano and Wagai, 2014). This constraint is demonstrated in Figure 6.2 for two exemplary samples.

Fig. 6.2: GS distribution curves of two silt dominated sediment samples. The samples were measured several times. Prior to each repetition ultrasonic treatment was applied for 120 seconds respectively.

In contrast to the oxidation of organic matter, which is conducted in almost every study including grain size analysis, dissolution of carbonatic binding agents is not a standardized procedure and is often omitted (Buurman et al., 1997a; Beuselinck et al., 1998; Chappell, 1998; Allen and Thornley, 2004; Buurman et al., 2004; Eshel et al., 2004; Pye and Blott, 2004; Zobeck, 2004; Özer et al., 2010). Conversely, some studies effect decalcification by HCl treatment to the entire sample set, regardless of the origin and the type of the carbonates therein (Konert and Vandenberghe, 1997; Buurman, 103

2001; Sun et al., 2006; Van der Veer, 2006; Kaiser et al., 2009; Daut et al., 2010; Dietze et al., 2012; IJmker et al., 2012a; Qiang et al., 2013). To evaluate the effects of HCl treatment on the grain size analysis of different loess archives containing samples with different weathering degrees and sedimentary genesis, the samples were measured in two different ways: with and without the addition of HCl (Chapter 5.3.). No significant association of the HCl-induced grain size modifications (MODGSD) after HCl treatment with the calcium carbonate content could be detected. However, a distinct dependence of MODGSD on the content of Corg and the weathering degree of the sediment is evident. Consequently, pretreating post-depositional altered loess sediments with HCl may result in misleading grain size distributions and should be avoided in standard analysis of loess-paleosol sequences.

6.2. Grain size measurement and calculation

Grain size analysis is frequently applied in a wide range of industries and research institutions. However, the results obtained from different grain sizing techniques and laboratories are rarely comparable. The most challenging factors limiting the comparability are: (i) the expression of a 3- dimensional body as a single number or ratio and (ii) the different principles of the applied measuring techniques (see Chapter 3). Particles described by the same diameter can occur in many different shapes. Furthermore, an irregularly shaped particle can have an infinite number of single diameters (Fig. 6.3); hence, equivalent diameters are only meaningful when a significant number of particles have been measured to provide average statistical diameters for the respective grain size ranges.

By means of direct measuring, techniques such as microscopy or sieve analysis offer the best approximations of the actual grain size. Following such analysis, techniques can be used to calibrate the indirect techniques. Nevertheless, for fine silt and clay sized particles, the direct techniques are technically complex, time- and cost-intensive, or, in the case of submicron particles, even impossible. Consequently, the main difficulties and differences occur between the indirect measuring techniques (e.g., gravitational techniques and laser diffraction size analysis) in the clay and fine silt grain size ranges (see Chapter 3 and 5.4.).

104

Fig. 6.3: Illustration of different equivalent spheres of the same “submicron” irregularly shaped particle (in accordance to Malvern, 2012).

The most often discussed disadvantage of the laser diffraction technique is the purported overestimation of the size of clay and fine silt particles in comparison with gravitational techniques (e.g., Loizeau et al., 1994; Konert and Vandenberghe, 1997; Roberson and Weltje, 2014). Since by laser diffraction size analysis a particle is theoretically detected in all possible orientations (Fig. 6.3), the distribution of a sample containing several irregularly shaped particles having the same equivalent diameter is broadened. Due to the logarithmic scale, the averaged results tend higher of all possible equivalent diameters. Particularly, due to the autumn leave effect (see Chapter 3.3.6. and 5.4.4.) in all gravitational techniques, the size of the irregularly shaped fine particles is underestimated (Fig. 6.3). If a sample contains numerous pedogenically formed clay minerals, this effect may result in the massive overestimation of the clay fraction. Hence, compensations by shifting the upper edge of the clay fraction, obtained by laser diffraction size analysis, to coarser grain sizes to match the pipette clay, as suggested by Konert and Vandenberghe et al. (1997), to 8

105

µm, or to 4.6 µm, as suggested by Antoine (2009a,b), are not appropriate. Employing separation by Atterberg sedimentation prior to measurement with the laser diffraction size analyzer, the overestimation of the < 2 µm fraction can be directly evaluated. We have shown that the phenomenon of systematic overestimation of the submicron grain sizes in comparison with settling techniques is negligible if the LS 13320 with additional polarization intensity differential scattering (PIDS) technology is used for laser diffraction size analysis, and the Mie theory is applied for grain size calculation (Chapter 5.4.4.). During Atterberg or pipette analysis, the < 2 µm fractions are weighted, whereas during laser diffraction the scattered light is detected. The probability that single, distinctly coarse particles cross the measuring range of the laser diffraction size analysis is relatively low. By contrast, in the weight aliquot of the respective Atterberg or pipette fraction, they are definitely recorded.

In summary, it can be stated that neither of the recently available methods for grain size analysis result in the perfect “real life” size of the particles as the methods measure different properties of the same sediment sample (Fig. 6.3).

The entire sample set presented in this study was measured with a laser diffraction particle size analyzer (Beckman Coulter LS 13 320; Beckman Coulter Life Sciences, Indianapolis, IN, USA). The application includes the PIDS system which measures the scattering of polarized (horizontal and vertical) light with additional wavelengths (450 nm, 600 nm, and 900 nm). Since the analysis is completely integrated, a single solution is obtained although two methods are used (ISO 13320, 2009). Without such extension of the main scattering pattern (780 nm), the information of the submicron range is potentially unreliable.

In order to perform a mathematical transformation of the diffraction pattern, there are two different common optical models: the Fraunhofer approximation and the Lorenz-Mie theory (ISO 13320, 2009). Unlike the Fraunhofer approximation, which considers only the diffraction at the particle surface, the Lorenz-Mie theory takes mineral specific refraction into account (cf. Chapter 5.4.). The angle-dependent scattering pattern calculated by the Lorenz-Mie theory approaches the scattering pattern calculated by the Fraunhofer approximation with increasing grain size. Specifically, in the submicron range, substantial differences occur between the calculated grain size distributions of these two models.

In fact, neither of the models provides the true size of the submicron particles. However, since the complex refractive index depends on the mineral properties, the difference of the grain size distributions calculated with both models is sensitive to the mineral composition of the sample (Chapter 5.4.4.4.).

106

For standard grain size analysis of loess-paleosol sequences, the Lorenz-Mie theory should be used exclusively owing to the large portion of fine particles (< 8 µm). However, when analyzing sample sets consisting solely of coarse silt and sand, both models provide the similar results.

6.3. The analysis of grain size data

In studies on loess-paleosol sequences, different grain size proxies are commonly used to reconstruct the environmental and climatic conditions during the aeolian accumulation of dust particles suspended in the atmosphere. These variations are analyzed on glacial/interglacial to millennial timescales (Újvári et al., 2016). Moreover, these proxies are a function of the complex interaction of all factors influencing the final shape/composition of the grain size distribution (cf. Chapter 2.3.): wind speed, frequency of storm events, moisture availability in the source and sink areas, source and content of different proportions (long distance/short distance), the transport process (suspension, saltation, reptation), sediment availability (can be affected by the activity of proximal river beds), vegetation cover in the source and sink areas, mineral composition, preferred occurrence of certain minerals in certain size classes, and, especially, the post-depositional processes (loessification, pedogenesis, and cryogenic processes). Each grain size proxy describing the distribution as a single value depends on a combination of some of these influencing processes and parameters. Therefore, it is virtually impossible to disentangle how a single factor influences the respective proxy. The analysis obtained in this study show that although the respective proxies (U-ratio, GSI, GS-mean, clay content, etc.) are associated with different single parameters of the above-mentioned, their depth variation within a loess-paleosol-sequence is very similar (see Fig. 6.4).

Generally, three constraints limit the use of such proxies to reconstruct past environmental and climatic conditions:

First, grain size distributions are compositional in nature, which means that a grain size distribution of one sample is mathematically defined as a positive vector with a fixed sum (e.g., 1 for proportions or 100 for percentages) over all grain size classes (see Chapter 5.4.2.4.). Consequently, absolute variations within a grain size range with a low occurrence rate (e.g., < 5.5 µm, which is excluded in cases of the U-ratio) are commonly masked by relative high frequencies within the major component of the distribution (Fig. 6.4). Therefore, the variation of the common grain size proxies does not generally indicate a specific traceable parameter of the past environmental conditions.

107

Fig. 6.4: Heatmap showing the grain size distribution of all samples from the Suohuoduo section, containing pure loess (white), slightly weathered loess (light grey) and chernozem-like paleosols (dark grey). The vertical curves represent classical grain size parameters and ratios (for their explanation see Chapter 2.3.).

The second constraint is the simplicity of the common models describing transport and sorting mechanisms suggested by Pye (1987), Pye and Tsoar (1987), and Pye (1995). For coarse grains, short distance transport is generally assumed as a result of their mass and rapid gravitational deposition (Crouvi et al., 2010; Vandenberghe, 2013). However, during extreme storm events, particles of up to 250 µm can be transported over comparatively long distances (Betzer et al., 1988). Menendez et al. (2014) detected quartz particles with sizes coarser than 150 µm in recent Saharan dust collected near the Canary Islands. Such so-called “giant particles” are measured in several studies of long-range transported aeolian dust (Betzer et al., 1988; Middleton and Goudie, 2001; Fiol et al., 2005; von Suchodoletz et al., 2009; Alcántara-Carrió et al., 2010). It is widely accepted that particles < 20µm are transported as background dust over long distances and are deposited far from their sources (Pye, 1995; Sun et al., 2002). Rosenberg et al. (2014) measured size-resolved modern atmospheric dust and found no correlation between grain size fraction up to 40 µm, wind speed, or the distance to the assumed source region. Due to turbulences, vertical mixing, and the generally decreasing deposition velocity (cf. Chapter 5.4.) with decreasing grain sizes, transport over 108

several hundred kilometers is obviously possible (Goossens, 2008; Újvári et al., 2016). In summary, it can be stated that the source of the aeolian dust, distance from source to sink, and, to a certain extent, wind strength cannot be reconstructed by a single value describing the general fineness (mean, median, fraction > 63 µm or < 20 µm, etc.) or the relation of fine to coarse fractions (U- ratio, Twin-Peak ratio, and twin).

The third constraint is the post-depositional grain size fractionation due to physical and chemical weathering (see Chapter 2.2.1. and Chapter 5.4.) or due to the formation of stable aggregates and organo-mineral complexes (see Chapter 2.2.2. and Chapter 5.3.). As these processes have the final influence on the composition of the recent sediment sample, their impact has to be identified and quantified before interpretations concerning previous sedimentation, or even transport processes, are possible.

Consequently, there is a need for appropriate visualization techniques and powerful proxies of post-depositional alteration processes, such as pedogenesis and the formation of stable soil aggregates, for the interpretation of loess-paleosol sequences. In this study, we suggest the visualization of high resolved grain size data obtained for a loess-paleosol-sequence as a heatmap (e.g. Fig. 6.4). In this way, the entire data set can be visualized without the loss of vertical (samples with depth) or horizontal (the number of detected grain size classes) information. Whereas, by the calculation of single value parameters “true” variations within sensitive grain size classes are masked by relative changes of the more frequent classes. If single values are required, e.g., for correlation analysis, a section-specific grain size ratio can be calculated. To ensure that each parameter of the ratio represents just one fractionation process, the two grain size ranges for ratio calculation were selected so that respective grain size classes would show similar depth variations (Chapter 5.2.).

For classical studies, a multi parameter approach comprising mean, median, and mode values, common classified values (clay, silt, sand, and respective subclasses), and the shape of the distribution density curves of selected representative samples is appropriate. That was demonstrated, for instance, for the Suohuduo section (Chapter 5.1.).

To solve the general problem of disentangling post-depositional processes from other grain size influencing processes, especially during sediment accumulation, a new proxy using laser diffraction calculations of grain size distribution obtained by two optical models, called the ΔGSD (Chapter 5.4.), was presented. As during accumulation, loess consists mainly of coarse silt, whereas particles < 1 µm are extremely underrepresented. This proxy is focused on the submicron grain size range. To deal with the compositional data effect, the distributions are analyzed as centered log ratio (clr)- transformed grain size differences (ΔGSDclr). Summarizing, the ΔGSDclr is a comparatively time- and cost-effective estimator for secondary clay mineral formation as a function of post-depositional 109

chemical weathering (silicate weathering). This proxy can be used as a benchmark for the paleopedogenesis of loess-paleosol sequences and for inter-sequence comparison, if the same laser diffraction device is used for grain size measuring. In comparison to the above-mentioned grain size parameters, it is virtually unaffected by aeolian sorting effects, distance to the source regions, cryogenic processes, gentle reworking of inherited weathering products, and carbonate dynamics. This new proxy was applied and evaluated for several loess-paleosol sequences originating from different loess regions (Lower Rhine Embayment: Zens et al., in review, Chapter 5.4.; Northern foreland of Harz Mountains: Krauß et al., 2016; Neckar Basin: Krauß et al., in review; northeastern Tibetan Plateau: Chapter 5.4. and Carpathian Basin: Chapter 5.4.).

Pretreatment with HCl is commonly used to separate aggregates bound by calcium carbonates. In Chapter 5.3., we evaluate the effects of HCl treatment on grain size analysis of Late Pleistocene and Holocene loess-paleosol sequences. Remarkably, a distinct dependence of HCl-induced modifications of grain size distributions on the content of organic matter, weathering degree of the sediment, and presence of stable aggregates was observed. This knowledge can be used to identify stable microaggregates and organo-mineral complexes as well as the grain size ranges in which these structures are present.

Summarizing, in all presented case studies (Chapters 5.1. to 5.4..; Zens et al., in review; Krauß et al. 2016; Krauß et al., in review) a significant post-depositional grain size shift due to pedogenesis is observed. Consequently, grain size-based time series analysis to reconstruct Quaternary sedimentary environments must be considered critically, especially if clay content is involved as a grain size parameter. The common grain size ratios can only be used sensibly to reconstruct paleosedimentation conditions if the sediment has not markedly changed subsequent to initial loessification (see Chapters 5.2. and 5.4.).

Taking into account the constraints outlined in this dissertation, by using laser diffraction size analysis, large data sets can be obtained for loess-paleosol sequences with greater reliability and time- and cost-efficiency than by any other available means.

110

111

Danksagung / Acknowledgments

Im Verlaufe der vergangenen Jahre wurde ich auf verschiedene Art und Weise von zahlreichen Personen unterstützt, wodurch die Erstellung dieser Dissertation erst möglich wurde.

Ohne die vorbehaltlose Unterstützung, das entgegengebrachte Vertrauen, die zahlreichen Motivationsgespräche und die fortwährend eingebrachten frischen Denkanstöße meines Doktorvaters Prof. Dr. Frank Lehmkuhl wäre die Vorlage der Dissertation nicht möglich gewesen.

Ein ganz besonderer Dank gilt Dr. Georg Stauch; nicht nur für intensive und fruchtbare wissenschaftliche, gesellschaftliche und nicht zuletzt moralische Diskussionen, sondern auch für zahlreiche kritische Anmerkungen und Revisionen. Außerdem danke ich für diverse gemeinsame Geländeaufenthalte, egal ob in den Weiten des Tibetplateaus oder im südlichen Umland von Aachen.

Marianne Dohms danke ich dafür, dass sie sich nicht nur um die primäre Bearbeitung zahlreicher Sedimentproben, sondern vor allem um gebetsmühlenartig wiederkehrende Sonderwünsche meinerseits gekümmert hat und auch die 56. Nachmessung einer einzelnen Probe mit beständiger Besonnenheit durchgeführt hat.

Den Freunden und Kollegen Dr. Veit Nottebaum, Dr. David Loibl und Jörg Zens möchte ich herzlich danken. Sie haben mir nicht nur als Sparringspartner beim Diskutieren gedient sondern haben maßgeblich dazu beigetragen, dass ich das Geographische Institut in den letzten 5 Jahren nahezu jeden Morgen mit einem Lächeln betreten habe.

Auch meinen ehemaligen Kollegen und Freunden Dr. Jens Protze, Dr. Holger Kels und Prof. Dr. Eileen Eckmeier möchte ich besonders dafür danken, dass sie mir immer ein Vorbild waren und mich dank ihrer enormen Erfahrung in zahlreichen Situationen geerdet und auf dem richtigen Weg gehalten haben.

Dank auch an alle weiteren studentischen, wissenschaftlichen und nicht wissenschaftlichen Mitarbeiterinnen und Mitarbeitern des Geographischen Instituts. Ihr habt durch euer offenes Gemüt und euer unkonventionelles Betragen eine außergewöhnliche Arbeitsatmosphäre geschaffen.

Apl. Prof. Dr. Wolfgang Römer danke ich dafür, dass er sich in den vergangenen zehn Jahren immer wieder Zeit für Fragen der besonderen Art genommen hat und vor allem dafür, dass er trotz allen Dingen, die er in der Vergangenheit erlebt hat, einfach Wolfgang geblieben ist.

112

Mein aufrichtiger Dank gilt auch apl. Prof. Dr. Bernhard Diekmann; nicht nur dafür dass er sich bereits frühzeitig zur Co-Betreuung dieser Arbeit bereit erklärt hat, sondern vor allem, für wissenschaftliche Denkanstöße und wertvolle Erinnerungen an unsere gemeinsame Geländekampagne.

Ich danke auch einigen externen Kollegen und Freunden, darunter Dr. Gregori Lockot (Berlin), Dr. Arne Ramisch (Potsdam) und vor allem Simon Mayer Heintze (Würzburg), welche auf die ein oder andere Weise einen Beitrag zu den hier dargelegten Arbeiten geleistet haben.

Ich danke meinen Freunden und Wahlbrüdern aus der Barage, und meiner Familie für ihre Geduld, vereinzelte Ermunterungen und vor allem eine ganze Menge Spaß und Geborgenheit.

Ganz herzlich möchte ich meiner damaligen Freundin Rachel Hennecken für ihre Geduld und ihr Verständnis und insbesondere dafür danken, dass sie mir immer wieder das Gefühl gegeben hat nach Hause zu kommen.

Nicht zuletzt möchte ich der Deutschen Forschungsgemeinschaft danken. Durch die Förderung der Projekte SFB 806 „Unser Weg nach Europa“ und SPP 1372 „Tibet Plateau – Entstehung, Klima, Ökosysteme” wurden die Grundvoraussetzungen zur Erstellung dieser Dissertation geschaffen.

Acknowledgments for “Timing and spatial distribution of loess and loess-like sediments in the mountain areas of the northeastern Tibetan Plateau” (Chapter 5.1.)

This paper is a contribution to the occasion of the 80th birthday celebration of the accomplishments of E. Derbyshire who accompanied the first discussions with the first author concerning the aeolian mantles in the high mountain areas of Tibet and Mongolia in 1996. We wish to thank the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG) for funding of the project as well as Fahu Chen, who had performed fieldwork also supported by the Lanzhou University. This research has been conducted within the scope of the project ‘Landscape and Lake- System Response to Late Quaternary Monsoon Dynamics on the Tibetan Plateau - Northern Transect’ which was established by the German Science Foundation (DFG) as part of the SPP 1372: Tibetan Plateau – Formation, Climate, Ecosystems. We also are most grateful to two anonymous reviewers, the third reviewer, J. Grunert, and the editor L.A. Owen for detailed comments that led us to improve the manuscript.

113

Acknowledgments for “Environmental change indicated by grain size variations and trace elements: examples from two different sections - the sandy-loess sediments from the Doroshivtsy site (Ukraine) and the loess section Semlac (Romania)” (Chapter 5.2.)

The authors would like to thank the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG) for funding of the project in the context of the CRC 806 "Our way to Europe", subproject B1 "The „Eastern Trajectory“: Last Glacial Palaeogeography and Archaeology of the Eastern Mediterranean and of the Balkan Peninsula". We would like to thank several colleagues for support during the field work, especially Larissa Kulakovska (leader of the team) and Vitaly Usik (Kiew University, Ukraine), Paul Haesaerts (Bruxelles, Belgium), Valéry Sitlivy (former Institute of Prehistoric Archaeology, University of Cologne, Germany), Cristian Ţuţu (Târgovişte, Romania); and Thomas Felauer (former PhD at the Department of Geography, RWTH Aachen University, Germany).

Acknowledgments for “Influence of HCl pretreatment and organo-mineral complexes on laser diffraction measurement of loess-paleosol sequences” (Chapter 5.3.)

We wish to thank the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG) for funding. Part of this research has been conducted within the scope of the project ‘Landscape and Lake-System Response to Late Quaternary Monsoon Dynamics on the Tibetan Plateau - Northern Transect’ (Le 730/22-1,2) which was established by the DFG as part of SPP 1372 “Tibetan Plateau – Formation, Climate, Ecosystems”. The sampling at site Grafenberg was carried out in the frame of DFG-funded CRC 806 “Our way to Europe”, subproject D1. We are grateful to Janneke IJmker, Marianne Dohms, Irene Knisch and Uwe Wollenberg for the help with field work, laboratory and SEM analysis and to Sören Lehmkuhl for language corrections. Jörg Zens and Holger Kels are thanked for the ideas and cartography on Fig. 2.

Acknowledgments for “The difference of two laser diffraction patterns as an indicator for post-depositional grain size reduction in loess-paleosol sequences” (Chapter 5.4.)

We wish to thank the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG) for funding. The Grafenberg and Semlac samples were carried out in the frame of the DFG funded CRC 806 “Our way to Europe”, subprojects D1 and B1). The Suohuoduo section was investigated within the scope of the project ‘Landscape and Lake-System Response to Late Quaternary Monsoon Dynamics on the Tibetan Plateau - Northern Transect’ which was established by the DFG as part of the SPP 1372: Tibetan Plateau – Formation, Climate, Ecosystems. We are grateful

114

to Holger Kels, Peter Fischer, Florian Steininger, Georg Stauch, Janneke IJmker and Marianne Dohms, for the help with field work and laboratory analysis. Data presented in this work are available upon request from the corresponding author.

115

References

AG Boden (Ed.), 2005. Bodenkundliche Kartieranleitung. KA5. Schweizerbart Science Publishers, Stuttgart, Germany. Agrawal, Y.C., 1991. Laser diffraction size analysis, in: Syvitski, J.P.M., McCave, I.N., Riley, J.B. (Eds.), Principles, Methods, and Application of Particle Size Analysis. Cambridge University Press, pp. 119–128. Aitchison, J., 1986. The Statistical Analysis of Compositional Data. Chapman & Hall, Ltd., London, UK, UK. Aitchison, J., 1982. The statistical analysis of compositional data. Journal of the Royal Statistical Society. Series B (Methodological) 139–177. Aitken, M.J., 1998. Introduction to Optical Dating: The Dating of Quaternary Sediments by the Use of Photon-stimulated Luminescence. Oxford University Press, Oxford, New York. Alcántara-Carrió, J., Fernández-Bastero, S., Alonso, I., 2010. Source area determination of aeolian sediments at Jandia Isthmus (Fuerteventura, Canary Islands). Journal of Marine Systems, Models and observations of Marine Systems 80, 219–234. doi:10.1016/j.jmarsys.2009.10.011 Alexander, J., 2015. Optical properties of mineral dust aerosol including analysis of particle size, composition, and shape effects, and the impact of physical and chemical processing. Theses and Dissertations. Allen, J.R.L., Thornley, D.M., 2004. Laser granulometry of Holocene estuarine silts: effects of hydrogen peroxide treatment. The Holocene 14, 290–295. doi:10.1191/0959683604hl681rr Allen, T., 2003. Powder Sampling and Particle Size Determination. Elsevier New York, Amsterdam ; Boston. Allen, T., 1990. Particle Size Measurement. Springer Netherlands, Dordrecht. An, C.-B., Feng, Z.-D., Barton, L., 2006. Dry or humid? Mid-Holocene humidity changes in arid and semi-arid China. Quaternary Science Reviews 25, 351–361. doi:10.1016/j.quascirev.2005.03.013 An, Z., Kukla, G., Porter, S.C., Xiao, J., 1991. Late quaternary dust flow on the chinese Loess Plateau. CATENA 18, 125–132. doi:10.1016/0341-8162(91)90012-M An, Z., Wu, X., Lu, Y., Zhang, D., Sun, X., Dong, G., 1991. Paleoenvironmental changes of China during the last 18,000 years, in: Liu, T. (Ed.), Quaternary Geology and Environment in China. Science Press, Beijing, pp. 228–236. Antoine, P., Rousseau, D.-D., Degeai, J.-P., Moine, O., Lagroix, F., kreutzer, S., Fuchs, M., Hatté, C., Gauthier, C., Svoboda, J., Lisá, L., 2013. High-resolution record of the environmental response to climatic variations during the Last Interglacial–Glacial cycle in Central Europe: the loess-palaeosol sequence of Dolní Věstonice (Czech Republic). Quaternary Science Reviews 67, 17–38. doi:10.1016/j.quascirev.2013.01.014 Antoine, P., Rousseau, D.-D., Fuchs, M., Hatté, C., Gauthier, C., Marković, S.B., Jovanović, M., Gaudenyi, T., Moine, O., Rossignol, J., April 1, 2009b. High-resolution record of the last climatic cycle in the southern Carpathian Basin (Surduk, Vojvodina, Serbia). Quaternary International, 116

Loess in the Danube Region and Surrounding Loess Provinces: The Marsigli Memorial Volume 198, 19–36. doi:10.1016/j.quaint.2008.12.008 Antoine, P., Rousseau, D.-D., Lautridou, J.-P., HATTé, C., 1999. Last interglacial-glacial climatic cycle in loess-palaeosol successions of north-western France. Boreas 28, 551–563. doi:10.1111/j.1502-3885.1999.tb00241.x Antoine, P., Rousseau, D.-D., Moine, O., Kunesch, S., Hatté, C., Lang, A., Tissoux, H., Zöller, L., Dezember 2009a. Rapid and cyclic aeolian deposition during the Last Glacial in European loess: a high-resolution record from Nussloch, Germany. Quaternary Science Reviews 28, 2955– 2973. doi:10.1016/j.quascirev.2009.08.001 Ardon-Dryer, K., Huang, Y.-W., Cziczo, D.J., 2015. Laboratory studies of collection efficiency of sub-micrometer aerosol particles by cloud droplets on a single-droplet basis. Atmospheric Chemistry and Physics 15, 9159–9171. doi:10.5194/acp-15-9159-2015 Asano, M., Wagai, R., 2014. Evidence of aggregate hierarchy at micro- to submicron scales in an allophanic Andisol. Geoderma 216, 62–74. doi:10.1016/j.geoderma.2013.10.005 Atterberg, A.M., 1914. Die Konsistenzkurven der Mineralböden. Internationale Mitteilungen für Bodenkunde 4, 418–432. Atterberg, A.M., 1911. Über die physikalische Bodenuntersuchung und über die Plastizität der Tone. Internationale Mitteilungen für Bodenkunde 1, 10–43. Atterberg, A.M., 1905. Die rationelle Klassifikation der Sande und Kiese. Chemiker-Zeitung 29, 195–198. Bachmann, J., Guggenberger, G., Baumgartl, T., Ellerbrock, R.H., Urbanek, E., Goebel, M.-O., Kaiser, K., Horn, R., Fischer, W.R., 2008. Physical carbon-sequestration mechanisms under special consideration of soil wettability. Z. Pflanzenernähr. Bodenk. 171, 14–26. doi:10.1002/jpln.200700054 Bagnold, R., 1941. The Physics of Blown Sand and Desert Dunes. Chapman & Hall, London. Balabane, M., Plante, A.F., 2004. Aggregation and carbon storage in silty soil using physical fractionation techniques. European journal of soil science 55, 415–427. Baldock, J.A., Skjemstad, J.O., 2000. Role of the soil matrix and minerals in protecting natural organic materials against biological attack. Organic Geochemistry 31, 697–710. Bayvel, L.P., Jones, A.R., 1981. Electromagnetic Scattering and its Applications. Springer, London ; Englewood, N.J. Beckman Coulter, 2011. LS 13 320 Laser Diffraction Particle Size Analyzer - Instructions For Use. Berg, L.S., 1916. The origin of loess. Communications Russian Geographical Foundation 52, 579–646. Betzer, P.R., Carder, K.L., Duce, R.A., Merrill, J.T., Tindale, N.W., Uematsu, M., Costello, D.K., Young, R.W., Feely, R.A., Breland, J.A., Bernstein, R.E., Greco, A.M., 1988. Long–range transport of giant mineral aerosol particles. Nature 336, 568–571. doi:10.1038/336568a0

117

Beuselinck, L., Govers, G., Poesen, J., Degraer, G., Froyen, L., 1998. Grain-size analysis by laser diffractometry: comparison with the sieve-pipette method. CATENA 32, 193–208. doi:10.1016/S0341-8162(98)00051-4 Bloemsma, M.R., Zabel, M., Stuut, J.B.W., Tjallingii, R., Collins, J.A., Weltje, G.J., 2012. Modelling the joint variability of grain size and chemical composition in sediments. Sedimentary Geology 280, 135–148. doi:10.1016/j.sedgeo.2012.04.009 Blott, S.J., Pye, K., 2012. Particle size scales and classification of sediment types based on particle size distributions: Review and recommended procedures: Classification of particle size and sediment type. Sedimentology 59, 2071–2096. doi:10.1111/j.1365-3091.2012.01335.x Blott, S.J., Pye, K., 2006. Particle size distribution analysis of sand-sized particles by laser diffraction: an experimental investigation of instrument sensitivity and the effects of particle shape: Particle size distribution analysis of sands by laser diffraction. Sedimentology 53, 671–685. doi:10.1111/j.1365-3091.2006.00786.x Blume, H.-P., Stahr, K., Leinweber, P., 2011. Bodenkundliches Praktikum: eine Einführung in pedologisches Arbeiten für Ökologen, insbesondere Land- und Forstwirte, und für Geowissenschaftler, 3., neubearb. Aufl. ed. Spektrum, Akad. Verl, Heidelberg. Bokhorst, M.P., Beets, C.J., Marković, S.B., Gerasimenko, N.P., Matviishina, Z.N., Frechen, M., 2009. Pedo-chemical climate proxies in Late Pleistocene Serbian–Ukranian loess sequences. Quaternary International, Loess in the Danube Region and Surrounding Loess Provinces: The Marsigli Memorial Volume 198, 113–123. doi:10.1016/j.quaint.2008.09.003 Bokhorst, M.P., Vandenberghe, J., 2009. Validation of wiggle matching using a multi-proxy approach and its palaeoclimatic significance. Journal of Quaternary Science 24, 937–947. doi:10.1002/jqs.1271 Bøtter-Jensen, L., McKeever, S.W.S., Wintle, A.G. (Eds.), 2003. Front Matter A2 - Bøtter- Jensen, L., in: Optically Stimulated Luminescence Dosimetry. Elsevier, Amsterdam, p. 355. Boudot, J.P., Bel Hadj^Brahim, A., Steiman, R., Seigle-Murandi, F., 1989. Biodegradation of synthetic organo-metallic complexes of iron and aluminium with selected metal to carbon ratios. Soil Biology and Biochemistry 21, 961–966. doi:10.1016/0038-0717(89)90088-6 Bouyoucos, G.J., 1927. The Hydrometer as a new and rapid method for determining the colloidal content of soils. Soil Science 23, 319–331. Bronger, A., Heinkele, T., 1990. Mineralogical and clay mineralogical aspects of loess research. Quaternary International 7, 37–51. doi:10.1016/1040-6182(90)90037-5 Brożek, M., Surowiak, A., 2005. The dependence of distribution of settling velocity of spherical particles on the distribution of particle sizes and densities. Physicochemical Problems of Mineral Processing 39, 199–210. Buggle, B., Glaser, B., Hambach, U., Gerasimenko, N., Marković, S., 2011. An evaluation of geochemical weathering indices in loess–paleosol studies. Quaternary International, The Second Loessfest (2009) 240, 12–21. doi:10.1016/j.quaint.2010.07.019 Buggle, B., Hambach, U., Müller, K., Zöller, L., Marković, S.B., Glaser, B., 2014. Iron mineralogical proxies and Quaternary climate change in SE-European loess–paleosol sequences.

118

CATENA, Loess and dust dynamics, environments, landforms, and pedogenesis: a tribute to Edward Derbyshire 117, 4–22. doi:10.1016/j.catena.2013.06.012 Bullard, J.E., McTainsh, G.H., 2003. Aeolian-fluvial interactions in dryland environments: examples, concepts and Australia case study. Progress in Physical Geography 27, 471–501. doi:10.1191/0309133303pp386ra Buurman, P., de Boer, K., Pape, T., Juli 1997a. Laser diffraction grain-size characteristics of Andisols in perhumid Costa Rica: the aggregate size of allophane. Geoderma 78, 71–91. doi:10.1016/S0016-7061(97)00012-8 Buurman, P., Garcıá Rodeja, E., Martıneź Cortizas, A., van Doesburg, J.D.., 2004. Stratification of parent material in European volcanic and related soils studied by laser-diffraction grain-sizing and chemical analysis. CATENA 56, 127–144. doi:10.1016/j.catena.2003.10.007 Buurman, P., Pape, T., Muggler, C.C., 1997b. Laser grain-size determination in soil genetic studies. 1. practical problems. Soil Science 162 (3), 211–218. Buurman, P., Pape, T., Reijneveld, J.A., Jong, F. de, Gelder, E. van, 2001. Laser-diffraction and pipette-method grain sizing of Dutch sediments: correlations for fine fractions of marine, fluvial, and loess samples. Netherlands Journal of Geosciences 80, 49–57. doi:10.1017/S0016774600022319 Buylaert, J.P., Murray, A.S., Vandenberghe, D., Vriend, M., De Corte, F., Van den haute, P., 2008. Optical dating of Chinese loess using sand-sized quartz: Establishing a time frame for Late Pleistocene climate changes in the western part of the Chinese Loess Plateau. Quaternary Geochronology 3, 99–113. doi:10.1016/j.quageo.2007.05.003 Cantoni, C., Luchetta, A., Celio, M., Cozzi, S., Raicich, F., Catalano, G., 2012. Carbonate system variability in the Gulf of Trieste (North Adriatic Sea). Estuarine, Coastal and Shelf Science 115, 51–62. doi:10.1016/j.ecss.2012.07.006 Casagrande, A., Loos, N.A., 1934. Die Aräometer-Methode zur Bestimmung der Kornverteilung von Böden und anderen Materialien. Springer-Verlag. Cazes, J., 2004. Analytical Instrumentation Handbook. CRC Press. Chappell, A., 1998. Dispersing sandy soil for the measurement of particle size distributions using optical laser diffraction. CATENA 31, 271–281. doi:10.1016/S0341-8162(97)00049-0 Chen, F., Yu, Z., Yang, M., Ito, E., Wang, S., Madsen, D.B., Huang, X., Zhao, Y., Sato, T., John B. Birks, H., Boomer, I., Chen, J., An, C., Wünnemann, B., 2008. Holocene moisture evolution in arid central Asia and its out-of-phase relationship with Asian monsoon history. Quaternary Science Reviews 27, 351–364. doi:10.1016/j.quascirev.2007.10.017 Coakley, J.P., Syvitski, J.P.M., 2007. SediGraph Technique, in: Syvitski, J.P.M. (Ed.), Principles, Methods and Application of Particle Size Analysis. Cambridge University Press. Coakley, J.P., Syvitski, J.P.M., 1991. SediGraph Technique, in: Syvitski, J.P.M. (Ed.), Principles, Methods, and Application of Particle Size Analysis. Cambridge University Press, pp. 129–142. Colman, S.M., Yu, S.-Y., An, Z., Shen, J., Henderson, A.C.G., 2007. Late Cenozoic climate changes in China’s western interior: a review of research on Lake Qinghai and comparison with other records. Quaternary Science Reviews 26, 2281–2300. doi:10.1016/j.quascirev.2007.05.002

119

Cornillault, J., 1972. Partile size analyzer. Applied Optics 11, 265–268. Crouvi, O., Amit, R., Enzel, Y., Gillespie, A.R., 2010. Active sand seas and the formation of desert loess. Quaternary Science Reviews 29, 2087–2098. doi:10.1016/j.quascirev.2010.04.026 Daut, G., Mäusbacher, R., Baade, J., Gleixner, G., Kroemer, E., Mügler, I., Wallner, J., Wang, J., Zhu, L., 2010. Late Quaternary hydrological changes inferred from lake level fluctuations of Nam Co (Tibetan Plateau, China). Quaternary International 218, 86–93. doi:10.1016/j.quaint.2010.01.001 de Boer, G.B., de Weerd, C., Thoenes, D., Goossens, H.W., 1987. Laser diffraction spectrometry: Fraunhofer diffraction versus Mie scattering. Particle & Particle Systems Characterization 4, 14– 19. Dean, W.E., 1974. Determination of carbonate and organic matter in calcareous sediments and sedimentary rocks by loss on ignition; comparison with other methods. Journal of Sedimentary Research 44, 242–248. doi:10.1306/74D729D2-2B21-11D7-8648000102C1865D Debye, P., 1909. Der Lichtdruck auf Kugeln von beliebigem Material. Ann. Phys. 335, 57–136. doi:10.1002/andp.19093351103 Dietze, E., Hartmann, K., Diekmann, B., IJmker, J., Lehmkuhl, F., Opitz, S., Stauch, G., Wünnemann, B., Borchers, A., 2012. An end-member algorithm for deciphering modern detrital processes from lake sediments of Lake Donggi Cona, NE Tibetan Plateau, China. Sedimentary Geology 243–244, 169–180. doi:10.1016/j.sedgeo.2011.09.014 Dietze, E., Wünnemann, B., Diekmann, B., Aichner, B., Hartmann, K., Herzschuh, U., IJmker, J., Jin, H., Kopsch, C., Lehmkuhl, F., Li, S., Mischke, S., Niessen, F., Opitz, S., Stauch, G., Yang, S., 2010. Basin morphology and seismic stratigraphy of Lake Donggi Cona, north-eastern Tibetan Plateau, China. Quaternary International 218, 131–142. doi:10.1016/j.quaint.2009.11.035 DIN 4188-1, 1977. Siebböden; Drahtsiebböden für Analysensiebe, Maße. DIN 18123, 2011. Soil, investigation and testing – Determination of grain-size distribution. DIN EN ISO 14688-1, 2016. Geotechnical investigation and testing - Identification and classification of soil - Part 1: Identification and description. DIN EN ISO 17892-4, 2014. Geotechnical investigation and testing – Laboratory testing of soil – Part 4: Determination of particle size distribution. DIN ISO 3310-1, 2001. Test sieves - Technical requirements and testing - Part 1: Test sieves of metal wire cloth. International Organization for Standardization, Geneva. Ding, Z., Liu, T., Rutter, N.W., Yu, Z., Guo, Z., Zhu, R., 1995. Ice-Volume Forcing of East Asian Winter Monsoon Variations in the Past 800,000 Years. Quaternary Research 44, 149–159. doi:10.1006/qres.1995.1059 Ding, Z., Rutter, N., Jingtai, H., Tungsheng, L., 1992. A coupled environmental system formed at about 2.5 Ma in East Asia. Palaeogeography, Palaeoclimatology, Palaeoecology 94, 223–242. Domrös, M., Peng, G., 1988. The Climate of China. Springer. Duller, G. a. T., 2008. Single-grain optical dating of Quaternary sediments: why aliquot size matters in luminescence dating. Boreas 37, 589–612. doi:10.1111/j.1502-3885.2008.00051.x

120

Eckmeier, E., Egli, M., Schmidt, M.W.I., Schlumpf, N., Nötzli, M., Minikus-Stary, N., Hagedorn, F., 2010. Preservation of fire-derived carbon compounds and sorptive stabilisation promote the accumulation of organic matter in black soils of the Southern Alps. Geoderma 159, 147–155. doi:10.1016/j.geoderma.2010.07.006 Eckmeier, E., Rösch, M., Ehrmann, O., Schmidt, M.W.I., Schier, W., Gerlach, R., 2007. Conversion of biomass to charcoal and the carbon mass balance from a slash-and-burn experiment in a temperate deciduous forest. The Holocene 17, 539–542. doi:10.1177/0959683607077041 Edwards, A.P., Bremner, J.M., 1967. Microaggregates in Soils1. Journal of Soil Science 18, 64–73. doi:10.1111/j.1365-2389.1967.tb01488.x Eshel, G., Levy, G.J., Mingelgrin, U., Singer, M.J., 2004. Critical evaluation of the use of laser diffraction for particle-size distribution analysis. Soil Science Society of America Journal 68, 736– 743. Etzler, F.M., Sanderson, M.S., 1995. Particle size analysis: a comparative study of various methods. Particle & particle systems characterization 12, 217–224. Fan, L.-S., Zhu, C., 2005. Principles of gas-solid flows. Cambridge University Press. FAO, 2006. Guidelines for soil description, 4th ed. ed. Food and Agriculture Organization of the United Nations, Rome. Fernández-Ugalde, O., Virto, I., Barré, P., Gartzia-Bengoetxea, N., Enrique, A., Imaz, M.J., Bescansa, P., 2011. Effect of carbonates on the hierarchical model of aggregation in calcareous semi-arid Mediterranean soils. Geoderma 164, 203–214. doi:10.1016/j.geoderma.2011.06.008 Fiol, L.A., Fornós, J.J., Gelabert, B., Guijarro, J.A., 2005. Dust rains in Mallorca (Western Mediterranean): Their occurrence and role in some recent geological processes. CATENA 63, 64–84. doi:10.1016/j.catena.2005.06.012 Fischer, P., Hilgers, A., Protze, J., Kels, H., Lehmkuhl, F., Gerlach, R., 2012. Formation and geochronology of Last Interglacial to Lower Weichselian loess/palaeosol sequences–case studies from the Lower Rhine Embayment, Germany. E&G Quaternary Science Journal 61, 48–63. Francus, P. (Ed.), 2005. Image Analysis, Sediments and Paleoenvironments, 1st ed. 2004. 2nd printing 2007 edition. ed. Springer, Dordrecht ; Norwell, MA. Froelich, P.N., 1980. Analysis Of Organic-Carbon In Marine-Sediments. Limnology And Oceanography 25, 564–572. doi:10.4319/lo.1980.25.3.0564 Galbraith, R.F., Green, P.F., 1990. Estimating the component ages in a finite mixture. International Journal of Radiation Applications and Instrumentation. Part D. Nuclear Tracks and Radiation Measurements 17, 197–206. Gee, G.W., Or, D., 2002. Particle-Size Analysis, in: Dane, J.H., Topp, C. (Eds.), Methods of Soil Analysis: Part 4. Physical Methods, Soil Science Society of America Book Series. Soil Science Society of America, Madison, Wis., USA. Glaser, B., Haumaier, L., Guggenberger, G., Zech, W., 2001. The “Terra Preta” phenomenon: a model for sustainable agriculture in the humid tropics. Naturwissenschaften 88, 37–41. doi:10.1007/s001140000193

121

Gocke, M., Hambach, U., Eckmeier, E., Schwark, L., Zöller, L., Fuchs, M., Löscher, M., Wiesenberg, G.L.B., 2014a. Introducing an improved multi-proxy approach for paleoenvironmental reconstruction of loess–paleosol archives applied on the Late Pleistocene Nussloch sequence (SW Germany). Palaeogeography, Palaeoclimatology, Palaeoecology 410, 300–315. doi:10.1016/j.palaeo.2014.06.006 Gocke, M., Hambach, U., Eckmeier, E., Schwark, L., Zöller, L., Fuchs, M., Löscher, M., Wiesenberg, G.L.B., 2014b. Introducing an improved multi-proxy approach for paleoenvironmental reconstruction of loess–paleosol archives applied on the Late Pleistocene Nussloch sequence (SW Germany). Palaeogeography, Palaeoclimatology, Palaeoecology 410, 300–315. doi:10.1016/j.palaeo.2014.06.006 Goldstein, J., Newbury, D.E., Joy, D.C., Lyman, C.E., Echlin, P., Lifshin, E., Sawyer, L., Michael, J.R., 2007. Scanning Electron Microscopy and X-ray Microanalysis, 3rd ed. Springer, New York. Goossens, D., 2008. Relationships between horizontal transport flux and vertical deposition flux during dry deposition of atmospheric dust particles. Journal of Geophysical Research 113. doi:10.1029/2007JF000775 Greenfield, S.M., 1957. Rain scavenging of radioactive particulate matter from the atmosphere. J. Meteor. 14, 115–125. doi:10.1175/1520-0469(1957)014<0115:RSORPM>2.0.CO;2 Guérin, G., Mercier, N., Adamiec, G., 2011. Dose-rate conversion factors: update. Ancient TL 29 (1), 5–8. Halfpenny, A., 2010. Some important practical issues for the collection and manipulation of electron backscatter diffraction (EBSD) data from geological samples. Journal of the Virtual Explorer 35. Harris, D., Horwáth, W.R., van Kessel, C., 2001. Acid fumigation of soils to remove carbonates prior to total organic carbon or CARBON-13 isotopic analysis. Soil Science Society of America Journal 65, 1853. doi:10.2136/sssaj2001.1853 Haumaier, L., Zech, W., 1995. Black carbon—possible source of highly aromatic components of soil humic acids. Organic Geochemistry 23, 191–196. doi:10.1016/0146-6380(95)00003-W Haynes, W.M. (Ed.), 2015. CRC Handbook of Chemistry and Physics, 96th Edition, 96 edition. ed. CRC Press, Boca Raton, Florida. Hendrix, W.P., Orr, C., 1972. Automatic sedimentation size analysis instrument, in: Groves, M.J., Wyatt-Sargent, J.L. (Eds.), Particle Size Analysis. Society of Anatytical Chemistry, London, pp. 133–146. Herzschuh, U., 2006. Palaeo-moisture evolution in monsoonal Central Asia during the last 50,000 years. Quaternary Science Reviews 25, 163–178. doi:10.1016/j.quascirev.2005.02.006 Hingston, F.J., Posner, A.M., Quirk, J.P., 1974. Anion Adsorption by Goethite and Gibbsite. Journal of Soil Science 25, 16–26. doi:10.1111/j.1365-2389.1974.tb01098.x Holdich, R.G., 2002. Fundamentals of Particle Technology. Midland Information Technology & Publishing. Hošek, J., Hambach, U., Lisá, L., Grygar, T.M., Horáček, I., Meszner, S., Knésl, I., 2015. An integrated rock-magnetic and geochemical approach to loess/paleosol sequences from Bohemia

122

and Moravia (Czech Republic): Implications for the Upper Pleistocene paleoenvironment in central Europe. Palaeogeography, Palaeoclimatology, Palaeoecology 418, 344–358. doi:10.1016/j.palaeo.2014.11.024 Hövermann, J., 1987. Morphogenetic regions in northeast Xizang (Tibet), in: Hövermann, J., Wang, W. (Eds.), Reports on the Northeastern Part of the Qinghai-Xizang (Tibet) Plateau. Science Press, Beijing, pp. 112–139. Huang, C.C., Pang, J., Huang, P., Hou, C., Han, Y., 2002. High-resolution studies of the oldest cultivated soils in the southern Loess Plateau of China. CATENA 47, 29–42. doi:10.1016/S0341- 8162(01)00169-2 Huang, C.C., Pang, J., Su, H., Li, S., Ge, B., 2009. Holocene environmental change inferred from the loess–palaeosol sequences adjacent to the floodplain of the Yellow River, China. Quaternary Science Reviews 28, 2633–2646. doi:10.1016/j.quascirev.2009.05.024 IJmker, J., Stauch, G., Dietze, E., Hartmann, K., Diekmann, B., Lockot, G., Opitz, S., Wünnemann, B., Lehmkuhl, F., 2012a. Characterisation of transport processes and sedimentary deposits by statistical end-member mixing analysis of terrestrial sediments in the Donggi Cona lake catchment, NE Tibetan Plateau. Sedimentary Geology 281, 166–179. doi:10.1016/j.sedgeo.2012.09.006 IJmker, J., Stauch, G., Hartmann, K., Diekmann, B., Dietze, E., Opitz, S., Wünnemann, B., Lehmkuhl, F., 2012b. Environmental conditions in the Donggi Cona lake catchment, NE Tibetan Plateau, based on factor analysis of geochemical data. Journal of Asian Earth Sciences, Asian Climate & TectonicsAsian Climate & Tectonics 44, 176–188. doi:10.1016/j.jseaes.2011.04.021 ISO 565, 1983. Test sieves; Woven metal wire cloth, perforated plate and electroformed sheet; Nominal sizes of openings. International Organization for Standardization, Geneva. ISO 10693, 2014. Soil quality - Determination of carbonate content - Volumetric method. ISO 10693, 1995. Soil quality - Determination of carbonate content - Volumetric method. International Organization for Standardization, Geneva. ISO 11277, 2002. Soil quality - Determination of particle size distribution in mineral soil material - Method by sieving and sedimentation. International Organization for Standardization, Geneva. ISO 13317-1, 2001. Determination of particle size distribution by gravitational liquid sedimentation methods - Part 1: General principles and guidelines. ISO 13317-3, 2001. Determination of particle size distribution by gravitational liquid sedimentation methods — Part 3: X-ray gravitational technique. ISO 13320, 2009. Particle size analysis — Laser diffraction methods. International Organization for Standardization, Geneva. Jeong, G.Y., Hillier, S., Kemp, R.A., 2011. Changes in mineralogy of loess–paleosol sections across the Chinese Loess Plateau. Quaternary Research 75, 245–255. doi:10.1016/j.yqres.2010.09.001

123

Jeong, G.Y., Hillier, S., Kemp, R.A., 2008. Quantitative bulk and single-particle mineralogy of a thick Chinese loess–paleosol section: implications for loess provenance and weathering. Quaternary Science Reviews 27, 1271–1287. doi:10.1016/j.quascirev.2008.02.006 Jia, Y., Huang, C., Mao, L., 2011. OSL dating of a Holocene loess-paleosol sequence in the southern Loess Plateau, China. Environ Earth Sci 64, 1071–1079. doi:10.1007/s12665-011-0925- z Jones, R.M., 2003. Particle size analysis by laser diffraction: ISO 13320, standard operating procedures, and Mie theory. American Laboratory 35, 44–47. Jouve, G., Francus, P., Lamoureux, S., Provencher-Nolet, L., Hahn, A., Haberzettl, T., Fortin, D., Nuttin, L., 2013. Microsedimentological characterization using image analysis and μ-XRF as indicators of sedimentary processes and climate changes during Lateglacial at Laguna Potrok Aike, Santa Cruz, Argentina. Quaternary Science Reviews, Potrok Aike Maar Lake Sediment Archive Drilling Project (PASADO) 71, 191–204. doi:10.1016/j.quascirev.2012.06.003 Kaiser, K., Guggenberger, G., 2003. Mineral surfaces and soil organic matter. European Journal of Soil Science 54, 219–236. Kaiser, K., Lai, Z., Schneider, B., Reudenbach, C., Miehe, G., Brückner, H., 2009. Stratigraphy and palaeoenvironmental implications of Pleistocene and Holocene aeolian sediments in the Lhasa area, southern Tibet (China). Palaeogeography, Palaeoclimatology, Palaeoecology 271, 329–342. doi:10.1016/j.palaeo.2008.11.004 Kaiser, K., Schoch, W.H., Miehe, G., 2007. Holocene paleosols and colluvial sediments in Northeast Tibet (Qinghai Province, China): Properties, dating and paleoenvironmental implications. CATENA 69, 91–102. doi:10.1016/j.catena.2006.04.028 Keck, C.M., 2006. Cyclosporine nanosuspensions: optimised size characterisation & oral formulations: = Ciclosporin Nanosuspensionen: optimierte Partikelgrößencharakterisierung & orale Formulierungen. Berlin. Kels, H., 2012. Stepping through the palaeolandscape: A transect from the Carpathian foothills to the lowland of the Romanian Banat. Quaternary International 279–280, 240–241. Kerker, M., 1969. The Scattering of Light and Other Electromagnetic Radiation: Physical Chemistry: A Series of Monographs. Academic Press. Kinsman, S., 1978. Particulate Characterization, in: Peterson, R.V., Richardson, J.H. (Eds.), Systematic Materials Analysis. Academic Press, London, pp. 183–203. Klasen, N., Loibl, C., Rethemeyer, J., Lehmkuhl, F., 2015. Testing feldspar and quartz luminescence dating of sandy loess sediments from the Doroshivtsy site (Ukraine) against radiocarbon dating. Quaternary International. doi:10.1016/j.quaint.2015.05.036 Kleber, M., Mikutta, R., Torn, M.S., Jahn, R., 2005. Poorly crystalline mineral phases protect organic matter in acid subsoil horizons. European Journal of Soil Science 56, 717–725. doi:10.1111/j.1365-2389.2005.00706.x Klüser, L., Di Biagio, C., Kleiber, P.D., Formenti, P., Grassian, V.H., 2016. Optical properties of non-spherical desert dust particles in the terrestrial infrared – An asymptotic approximation approach. Journal of Quantitative Spectroscopy and Radiative Transfer, Electromagnetic and

124

light scattering by nonspherical particles XV: Celebrating 150 years of Maxwell’s electromagnetics 178, 209–223. doi:10.1016/j.jqsrt.2015.11.020 Kögel-Knabner, I., Guggenberger, G., Kleber, M., Kandeler, E., Kalbitz, K., Scheu, S., Eusterhues, K., Leinweber, P., 2008. Organo-mineral associations in temperate soils: Integrating biology, mineralogy, and organic matter chemistry. Journal of Plant Nutrition and Soil Science 171, 61–82. doi:10.1002/jpln.200700048 Köhn, M., 1928. Bemerkungen zur mechanischen Bodenanalyse. III. Ein neuer Pipettapparat. Z. Pflanzenernaehr. Dueng. Bodenk. 11, 50–54. doi:10.1002/jpln.19280110104 Konert, M., Vandenberghe, J.E.F., 1997. Comparison of laser grain size analysis with pipette and sieve analysis: a solution for the underestimation of the clay fraction. Sedimentology 44, 523–535. Kowalenko, C.G., Babuin, D., 2013. Inherent factors limiting the use of laser diffraction for determining particle size distributions of soil and related samples. Geoderma 193–194, 22–28. doi:10.1016/j.geoderma.2012.09.006 Krauß, L., Kappenberg, A., Zens, J., Kehl, M., Schulte, P., Zeeden, C., Eckmeier, E., Lehmkuhl, F., submitted. Reconstruction of Late Pleistocene Paleoenvironments in Southern Germany using two high-resolution loess-paleosol records. Palaeogeography, Palaeoclimatology, Palaeoecology,. Krauß, L., Zens, J., Zeeden, C., Schulte, P., Eckmeier, E., Lehmkuhl, F., 2016. A Multi-Proxy Analysis of two Loess-Paleosol Sequences in the Northern Harz Foreland, Germany. Palaeogeography, Palaeoclimatology, Palaeoecology 461, 401–417. doi:10.1016/j.palaeo.2016.09.001 Kukla, G., An, Z., 1989. Loess stratigraphy in Central China. Palaeogeography, Palaeoclimatology, Palaeoecology 72, 203–225. Kulakovska, L., Usik, V., Haesaerts, P., Ridush, B., Uthmeier, T., Hauck, T.C., 2015. Upper Paleolithic of Middle Dniester: Doroshivtsi III site. Quaternary International 359–360, 347–361. doi:10.1016/j.quaint.2014.10.034 Kulakovska, L.V., Usik, V.I., Haesaerts, P., 2012. Doroshivtsy III - Gravettian Site in the Dniester Valley (Ukraine). Stratum plus 1, 131–150. Küster, Y., Hetzel, R., Krbetschek, M., Tao, M., 2006. Holocene loess sedimentation along the Qilian Shan (China): significance for understanding the processes and timing of loess deposition. Quaternary Science Reviews 25, 114–125. doi:10.1016/j.quascirev.2005.03.003 Lagaly, G., Schulz, O., Zimehl, R., 1997. Dispersionen und Emulsionen: eine Einführung in die Kolloidik feinverteilter Stoffe einschließlich der Tonminerale. Steinkopff, Darmstadt. Lagasse, R.R., Richards, D.W., 2003. Determining the size distribution of core–shell spheres and other complex particles by laser diffraction. Journal of Colloid and Interface Science 267, 65–73. doi:10.1016/S0021-9797(03)00669-6 Lancaster, N., Baas, A., 1998. Influence of vegetation cover on sand transport by wind: field studies at Owens Lake, California. Earth Surf. Process. Landforms 23, 69–82. doi:10.1002/(SICI)1096-9837(199801)23:1<69::AID-ESP823>3.0.CO;2-G

125

Lehmann, M., Berthold, C., Pabst, W., Gregorová, E., Nickel, K.G., 2004. Particle Size and Shape Characterization of Kaolins-Comparison of Settling Methods and Laser Diffraction. Key Engineering Materials 264–268, 1387–1390. doi:10.4028/www.scientific.net/KEM.264-268.1387 Lehmkuhl, F., 1997. The spatial distribution of loess and loess-like sediments in the mountain areas of Central and High Asia. Zeitschrift fur Geomorphologie, Supplementband 111, 97–116. Lehmkuhl, F., 1995. Geomorphologische Untersuchungen zum Klima des Holozäns und Jungpleistozäns Osttibets. Goltze, Göttingen. Lehmkuhl, F., Haselein, F., 2000. Quaternary paleoenvironmental change on the Tibetan Plateau and adjacent areas (Western China and Western Mongolia). Quaternary International 65–66, 121– 145. doi:10.1016/S1040-6182(99)00040-3 Lehmkuhl, F., Hilgers, A., Fries, S., Hülle, D., Schlütz, F., Shumilovskikh, L., Felauer, T., Protze, J., 2011. Holocene geomorphological processes and soil development as indicator for environmental change around Karakorum, Upper Orkhon Valley (Central Mongolia). CATENA 87, 31–44. doi:10.1016/j.catena.2011.05.005 Lehmkuhl, F., Hülle, D., Knippertz, M., 2012. Holocene geomorphic processes and landscape evolution in the lower reaches of the Orkhon River (northern Mongolia). CATENA 98, 17–28. doi:10.1016/j.catena.2012.06.003 Lehmkuhl, F., Klinge, M., Rees-Jones, J., Rhodes, E.J., 2000. Late Quaternary aeolian sedimentation in central and south-eastern Tibet. Quaternary International, Nat Rutter Honorarium 68–71, 117–132. doi:10.1016/S1040-6182(00)00038-0 Lehmkuhl, F., Liu, S., 1994. An outline of physical geography including Pleistocene glacial landforms of Eastern Tibet (provinces Sichuan and Qinghai). GeoJournal 34, 7–30. doi:10.1007/BF00813966 Lehmkuhl, F., Owen, L.A., 2005. Late Quaternary glaciation of Tibet and the bordering mountains: a review. Boreas 34, 87–100. doi:10.1111/j.1502-3885.2005.tb01008.x Lehmkuhl, F., Schulte, P., Zhao, H., Hülle, D., Protze, J., Stauch, G., 2014. Timing and spatial distribution of loess and loess-like sediments in the mountain areas of the northeastern Tibetan Plateau. CATENA, Loess and dust dynamics, environments, landforms, and pedogenesis: a tribute to Edward Derbyshire 117, 23–33. doi:10.1016/j.catena.2013.06.008 Lehmkuhl, F., Zens, J., Krauß, L., Schulte, P., Kels, H., 2016. Loess-paleosol sequences at the northern European loess belt in Germany: Distribution, geomorphology and stratigraphy. Quaternary Science Reviews 153, 11–30. doi:10.1016/j.quascirev.2016.10.008 Leser, H., Feld, K., 1977. Feld- und Labormethoden der Geomorphologie. De Gruyter, Berlin ; New York. Li, G., Ji, J., Zhao, L., Mao, C., Chen, J., 2008. Response of silicate weathering to monsoon changes on the Chinese Loess Plateau. CATENA 72, 405–412. doi:10.1016/j.catena.2007.07.006 Li, S.-H., Sun, J.-M., Zhao, H., 2002. Optical dating of dune sands in the northeastern deserts of China. Palaeogeography, Palaeoclimatology, Palaeoecology 181, 419–429. doi:10.1016/S0031- 0182(01)00443-6

126

Li, Y., Wang, N., Li, Z., Zhang, H., 2012. Basin-wide Holocene environmental changes in the marginal area of the Asian monsoon, northwest China. Environ Earth Sci 65, 203–212. doi:10.1007/s12665-011-1083-z Lindqvist, H., Jokinen, O., Kandler, K., Scheuvens, D., Nousiainen, T., 2014. Single scattering by realistic, inhomogeneous mineral dust particles with stereogrammetric shapes. Atmospheric Chemistry and Physics 14, 143–157. doi:10.5194/acp-14-143-2014 Linke, C., Möhler, O., Veres, A., Mohácsi, Á., Bozóki, Z., Szabó, G., Schnaiter, M., 2006. Optical properties and mineralogical composition of different Saharan mineral dust samples: a laboratory study. Atmos. Chem. Phys. 6, 3315–3323. doi:10.5194/acp-6-3315-2006 Liu, T., Xitao, Z., Jiamao, H., Honghan, Z. (Eds.), 1985. Loess and the environment, 1st ed. ed. China Ocean Press, Beijing : Liu, X., Lai, Z., Yu, L., Sun, Y., Madsen, D., 2012. Luminescence chronology of aeolian deposits from the Qinghai Lake area in the Northeastern Qinghai-Tibetan Plateau and its palaeoenvironmental implications. Quaternary Geochronology, 13th International Conference on Luminescence and Electron Spin Resonance Dating - LED 2011 Dedicated to J. Prescott and G. Berger 10, 37–43. doi:10.1016/j.quageo.2012.01.016 Loizeau, J.-L., Arbouille, D., Santiago, S., Vernet, J.-P., 1994. Evaluation of a wide range laser diffraction grain size analyser for use with sediments. Sedimentology 41, 353–361. Lorenz, L., 1883. Theorie der Dispersion. Ann. Phys. 256, 1–21. doi:10.1002/andp.18832560902 Lu, H., An, Z., 1998. Pretreated methods on loess-palaeosol samples granulometry. Chinese Science Bulletin 43, 237–240. doi:10.1007/BF02898920 Lu, H., Zhao, C., Mason, J., Yi, S., Zhao, H., Zhou, Y., Ji, J., Swinehart, J., Wang, C., 2011. Holocene climatic changes revealed by aeolian deposits from the Qinghai Lake area (northeastern Qinghai-Tibetan Plateau) and possible forcing mechanisms. The Holocene 21, 297–304. doi:10.1177/0959683610378884 Machalett, B., Oches, E.A., Frechen, M., Zöller, L., Hambach, U., Mavlyanova, N.G., Marković, S.B., Endlicher, W., 2008. Aeolian dust dynamics in central Asia during the Pleistocene: Driven by the long-term migration, seasonality, and permanency of the Asiatic polar front. Geochemistry, Geophysics, Geosystems 9, 1–22. doi:10.1029/2007GC001938 Malvern Instruments Limited, 2012. A basic guide to particle characterization. Markey, B.G., Bøtter-Jensen, L., Duller, G.A.T., 1997. A new flexible system for measuring thermally and optically stimulated luminescence. Radiation Measurements, Proceedings of the 8th International Conference on Luminescence and Electron Spin Resonance Dating 27, 83–89. doi:10.1016/S1350-4487(96)00126-6 Marković, S.B., Bokhorst, M.P., Vandenberghe, J., McCoy, W.D., Oches, E.A., Hambach, U., Gaudenyi, T., Jovanović, M., Zöller, L., Stevens, T., Machalett, B., 2008. Late Pleistocene loess- palaeosol sequences in the Vojvodina region, north Serbia. Journal of Quaternary Science 23, 73– 84. doi:10.1002/jqs.1124 Marković, S.B., Stevens, T., Kukla, G.J., Hambach, U., Fitzsimmons, K.E., Gibbard, P., Buggle, B., Zech, M., Guo, Z., Hao, Q., Wu, H., O’Hara Dhand, K., Smalley, I.J., Újvári, G., Sümegi, P., Timar-Gabor, A., Veres, D., Sirocko, F., Vasiljević, D.A., Jary, Z., Svensson, A., Jović, V., 127

Lehmkuhl, F., Kovács, J., Svirčev, Z., 2015. Danube loess stratigraphy — Towards a pan- European loess stratigraphic model. Earth-Science Reviews 148, 228–258. doi:10.1016/j.earscirev.2015.06.005 Mason, J.A., Greene, R.S.B., Joeckel, R.M., 2011. Laser diffraction analysis of the disintegration of aeolian sedimentary aggregates in water. CATENA 87, 107–118. doi:10.1016/j.catena.2011.05.015 Mason, J.A., Jacobs, P., 1999. High resolution particle siz e analysis as a tool for interpreting incipient soils in loess. Chinese Science Bulletin 44(S1), 70–74. Mason, J.A., Jacobs, P.M., Greene, R.S.B., Nettleton, W.D., 2003. Sedimentary aggregates in the Peoria Loess of Nebraska, USA. CATENA 53, 377–397. doi:10.1016/S0341-8162(03)00073-0 Matthews, I.N., 1991. The effect of grain shape and density on size measurenment, in: Syvitski, J.P.M. (Ed.), Principles, Methods, and Application of Particle Size Analysis. Cambridge University Press, pp. 22–33. Maxwell Garnett, J.C., 1906. Colours in Metal Glasses, in Metallic Films, and in Metallic Solutions. II. Philosophical Transactions of the Royal Society of London A: Mathematical, Physical and Engineering Sciences 205, 237–288. doi:10.1098/rsta.1906.0007 Maxwell, J.C., 1898. Über physikalische Kraftlinien. Leipzig, W. Engelmann, Leipzig. McCave, I.N., Bryant, R.J., Cook, H.F., Coughanowr, C.A., 1986. Evaluation of a Laser- Diffraction-Size Analyzer for Use with Natural Sediments. Journal of Sedimentary Research 56, 561–564. McCave, I.N., Hall, I.R., Bianchi, G.G., 2006. Laser vs. settling velocity differences in silt grainsize measurements: estimation of palaeocurrent vigour. Sedimentology 53, 919–928. doi:10.1111/j.1365-3091.2006.00783.x McCave, I.N., Syvitski, J.P.M., 1991. Principles and methods of geological particle size analysis, in: Syvitski, J.P.M. (Ed.), Principles, Methods, and Application of Particle Size Analysis. Cambridge University Press, pp. 129–142. Menéndez, I., Pérez-Chacón, E., Mangas, J., Tauler, E., Engelbrecht, J.P., Derbyshire, E., Cana, L., Alonso, I., 2014. Dust deposits on La Graciosa Island (Canary Islands, Spain): Texture, mineralogy and a case study of recent dust plume transport. CATENA, Loess and dust dynamics, environments, landforms, and pedogenesis: a tribute to Edward Derbyshire 117, 133–144. doi:10.1016/j.catena.2013.05.007 Meszner, S., Kreutzer, S., Fuchs, M., Faust, D., 2014. Identifying depositional and pedogenetic controls of Late Pleistocene loess-paleosol sequences (Saxony, Germany) by combined grain size and microscopic analyses. Zeitschrift für Geomorphologie, Supplementbände 63–90. doi:10.1127/0372-8854/2014/S-00169 Micromeritics (Ed.), 1982. Instruction manual for Sedigraph 5000D particle-size analyzer with X-ray shutte. Norcross. Middleton, N.J., Goudie, A.S., 2001. Saharan dust: sources and trajectories. Transactions of the Institute of British Geographers 26, 165–181. doi:10.1111/1475-5661.00013

128

Midwood, A.J., Boutton, T.W., 1998. Soil carbonate decomposition by acid has little effect on δ13C of organic matter. Soil Biology and Biochemistry 30, 1301–1307. doi:10.1016/S0038- 0717(98)00030-3 Mie, G., 1908. Beiträge zur Optik trüber Medien, speziell kolloidaler Metallösungen. Ann. Phys. 330, 377–445. doi:10.1002/andp.19083300302 Miehe, G., Miehe, S., Kaiser, K., Jianquan, L., Zhao, X., 2008. Status and Dynamics of the Kobresia pygmaea Ecosystem on the Tibetan Plateau. AMBIO: A Journal of the Human Environment 37, 272–279. doi:10.1579/0044-7447(2008)37[272:SADOTK]2.0.CO;2 Mikutta, R., Kleber, M., Kaiser, K., Jahn, R., 2005. Review: Organic Matter Removal from Soils using Hydrogen Peroxide, Sodium Hypochlorite, and Disodium Peroxodisulfate. Soil science society of America journal 69, 120–135. Miller, B.A., Schaetzl, R.J., 2012. Precision of Soil Particle Size Analysis using Laser Diffractometry. Soil Science Society of America Journal 76, 1719. doi:10.2136/sssaj2011.0303 Miller, B.V., Lines, R.W., 1988. Recent Advances in Particle Size Measurements: A Critical Review. CRC Critical Reviews in Analytical Chemistry 20, 75–116. doi:10.1080/00078988808048808 Mishchenko, M.I., 1993. Light scattering by size–shape distributions of randomly oriented axially symmetric particles of a size comparable to a wavelength. Applied optics 32, 4652–4666. Misra, M.K., Ragland, K.W., Baker, A.J., 1993. Wood ash composition as a function of furnace temperature. Biomass and Bioenergy 4 (2), 103–116. Muhs, D.R., Bettis, A.E., 2003a. Quaternary loess-Paleosol sequences as examples of climate- driven sedimentary extremes. Geological Society of America Special Papers 370, 53–74. doi:10.1130/0-8137-2370-1.53 Muhs, D.R., Bettis, E.A., 2003b. Quaternary loess-paleosol sequences as examples of climate- driven sedimentary extremes. Special Papers-Geological Society of America 53–74. Muhs, D.R., Cattle, S.R., Crouvi, O., Rousseau, D.-D., Sun, J., Zárate, M.A., 2014. Loess Records, in: Knippertz, P., Stuut, J.-B.W. (Eds.), Mineral Dust. Springer Netherlands, pp. 411–441. doi:10.1007/978-94-017-8978-3_16 Müller, G., Engelhardt, W. von, 1967. Methods in sedimentary petrology. E. Schweizerbart Hafner, Stuttgart; New York. Müller, T., Schladitz, A., Massling, A., Kaaden, N., Kandler, K., Wiedensohler, A., 2009. Spectral absorption coefficients and imaginary parts of refractive indices of Saharan dust during SAMUM- 1. Tellus B 61, 79–95. doi:10.1111/j.1600-0889.2008.00399.x Muñoz, O., Volten, H., Hovenier, J.W., Min, M., Shkuratov, Y.G., Jalava, J.P., van der Zande, W.J., Waters, L.B.F.M., 2006. Experimental and computational study of light scattering by irregular particles with extreme refractive indices: hematite and rutile. Astronomy and Astrophysics 446, 525–535. doi:10.1051/0004-6361:20053727 Murray, M.R., 2002. Is laser particle size determination possible for carbonate-rich lake sediments? Journal of Paleolimnology 27, 173–183.

129

Mutke, J., Barthlott, W., 2005. Patterns of vascular plant diversity at continental to global scales. Biologiske Skrifter 55, 521–531. Nemecz, E., Pécsi, M., Hartyáni, Z., Horváth, T., 2000. The origin of the silt size quartz grains and minerals in loess. Quaternary International 68, 199–208. Nettleton, W.D. (Ed.), 1991. Occurrence, Characteristics, and Genesis of Carbonate, Gypsum, and Silica Accumulations in Soils: Proceedings of a Symposium Sponsored by Division S. Soil Science Society of Amer, Madison, Wis., USA. Nie, J., Stevens, T., Rittner, M., Stockli, D., Garzanti, E., Limonta, M., Bird, A., Andò, S., Vermeesch, P., Saylor, J., Lu, H., Breecker, D., Hu, X., Liu, S., Resentini, A., Vezzoli, G., Peng, W., Carter, A., Ji, S., Pan, B., 2015. Loess Plateau storage of Northeastern Tibetan Plateau- derived Yellow River sediment. Nature Communications 6, 8511. doi:10.1038/ncomms9511 Nieder, R., Benbi, D.K., 2008. Carbon and Nitrogen in the Terrestrial Environment. Springer Science. Nieuwenhuize, J., Maas, Y.E.M., Middelburg, J.J., 1994. Rapid analysis of organic carbon and nitrogen in particulate materials. Marine Chemistry 45, 217–224. doi:10.1016/0304- 4203(94)90005-1 Nottebaum, V., Lehmkuhl, F., Stauch, G., Hartmann, K., Wünnemann, B., Schimpf, S., Lu, H., 2014. Regional grain size variations in aeolian sediments along the transition between Tibetan highlands and north-western Chinese deserts – the influence of geomorphological settings on aeolian transport pathways. Earth Surf. Process. Landforms 39, 1960–1978. doi:10.1002/esp.3590 Nugteren, G., Vandenberghe, J., 2004. Spatial climatic variability on the Central Loess Plateau (China) as recorded by grain size for the last 250 kyr. Global and Planetary Change, Human Dimensions and Natural Processes in Environmental Change 41, 185–206. doi:10.1016/j.gloplacha.2004.01.005 Oades, J., 1995. An overview of processes affecting the cycling of organic carbon in soils. John Wiley. Oades, J.M., 1988. The retention of organic matter in soils. Biogeochemistry 5, 35–70. doi:10.1007/BF02180317 Obreht, I., Zeeden, C., Hambach, U., Veres, D., Marković, S.B., Bösken, J., Svirčev, Z., Bačević, N., Gavrilov, M.B., Lehmkuhl, F., 2016. Tracing the influence of Mediterranean climate on Southeastern Europe during the past 350,000 years. Scientific Reports 6, 36334. doi:10.1038/srep36334 Olivier, J.P., Hickin, G.K., Orr, C.J., 1970. Rapid, automatic particle-size analysis in the subsieve range. Powder Technology 257–263. Opitz, S., Wünnemann, B., Aichner, B., Dietze, E., Hartmann, K., Herzschuh, U., IJmker, J., Lehmkuhl, F., Li, S., Mischke, S., Plotzki, A., Stauch, G., Diekmann, B., 2012. Late Glacial and Holocene development of Lake Donggi Cona, north-eastern Tibetan Plateau, inferred from sedimentological analysis. Palaeogeography, Palaeoclimatology, Palaeoecology 337–338, 159–176. doi:10.1016/j.palaeo.2012.04.013

130

Owen, L.A., Finkel, R.C., Haizhou, M., Barnard, P.L., 2006. Late Quaternary landscape evolution in the and Qaidam Basin, Northern Tibet: A framework for examining the links between glaciation, lake level changes and alluvial fan formation. Quaternary International, Quaternary palaeoenvironmental change in Tibet and the bordering mountainsQuaternary palaeoenvironmental change in Tibet and the bordering mountains 154–155, 73–86. doi:10.1016/j.quaint.2006.02.008 Owen, L.A., Finkel, R.C., Haizhou, M., Spencer, J.Q., Derbyshire, E., Barnard, P.L., Caffee, M.W., 2003. Timing and style of Late Quaternary glaciation in northeastern Tibet. Geological Society of America Bulletin 115,11, 1356–1364. Özer, M., Orhan, M., Işik, N.S., 2010. Effect of Particle Optical Properties on Size Distribution of Soils Obtained by Laser Diffraction. Environmental & Engineering Geoscience 16, 163–173. doi:10.2113/gseegeosci.16.2.163 Pearson, K., 1896. Mathematical contributions to the theory of evolution. On a form of spurious correlation which may arise when indices are used in the measurement of organs. Harrison and Sons, London. Pécsi, M., 1990. Loess is not just the accumulation of dust. Quaternary International 7–8, 1–21. doi:10.1016/1040-6182(90)90034-2 Peng, S., Hao, Q., Oldfield, F., Guo, Z., 2014. Release of iron from chlorite weathering and links to magnetic enhancement in Chinese loess deposits. CATENA 117, 43–49. doi:10.1016/j.catena.2013.07.005 Petersen, G.W., Chesters, G., Lee, G.B., 1966. Quantitative Determination of Calcite and Dolomite in Soils. Journal of Soil Science 17, 328–338. doi:10.1111/j.1365-2389.1966.tb01477.x Pingitore, N.E., Fretzdorff, S.B., Seitz, B.P., Estrada, L.Y., Borrego, P.M., Crawford, G.M., Love, K.M., 1993. Dissolution kinetics of CaCO3 in common laboratory solvents. Journal of sedimentary petrology 63, 641–645. Pranesha, T.S., Kamra, A.K., 1997. Scavenging of aerosol particles by large water drops: 3. Washout coefficients, half-lives, and rainfall depths. J. Geophys. Res. 102, 23947–23953. doi:10.1029/97JD01835 Prescott, J.R., Hutton, J.T., 1994. Cosmic ray contributions to dose rates for luminescence and ESR dating: Large depths and long-term time variations. Radiation Measurements 23, 497–500. doi:10.1016/1350-4487(94)90086-8 Preusser, F., Degering, D., Fuchs, M., Hilgers, A., Kadereit, A., Klasen, N., Krbetschek, M., Richter, D., Spencer, J.Q., 2008. Luminescence dating: basics, methods and applications. Eiszeitalter und Gegenwart / Quaternary Science Journal 57, 95–149. Prins, M.A., Vriend, M., Nugteren, G., Vandenberghe, J., Lu, H., Zheng, H., Jan Weltje, G., 2007. Late Quaternary aeolian dust input variability on the Chinese Loess Plateau: inferences from unmixing of loess grain-size records. Quaternary Science Reviews 26, 230–242. doi:10.1016/j.quascirev.2006.07.002 Pye, K., 2015. Aeolian Dust and Dust Deposits. Elsevier. Pye, K., 1995. The nature, origin and accumulation of loess. Quaternary Science Reviews, Aeolian Sediments in the Quaternary Record 14, 653–667. doi:10.1016/0277-3791(95)00047-X 131

Pye, K., 1987. Aeolian Dust and Dust Deposits. Academic Press, London. Pye, K., Blott, S.J., 2004. Particle size analysis of sediments, soils and related particulate materials for forensic purposes using laser granulometry. Forensic Science International 144, 19–27. doi:10.1016/j.forsciint.2004.02.028 Pye, K., Tsoar, H., 1987. The mechanics and geological implications of dust transport and deposition in deserts with particular reference to loess formation and dune sand diagenesis in the northern Negev, Israel. Geological Society, London, Special Publications 35, 139–156. doi:10.1144/GSL.SP.1987.035.01.10 Qiang, M., Chen, F., Song, L., Liu, X., Li, M., Wang, Q., 2013. Late Quaternary aeolian activity in Gonghe Basin, northeastern Qinghai-Tibetan Plateau, China. Quaternary Research 79, 403–412. doi:10.1016/j.yqres.2013.03.003 Qiang, M., Lang, L., Wang, Z., 2010. Do fine-grained components of loess indicate westerlies: Insights from observations of dust storm deposits at Lenghu (Qaidam Basin, China). Journal of Arid Environments 74, 1232–1239. doi:10.1016/j.jaridenv.2010.06.002 Reimer, L., 2013. Transmission Electron Microscopy: Physics of Image Formation and Microanalysis. Springer. Rittinger, P. ritter von, 1867. Lehrbuch der aufbereitungskunde in ihrer neuesten entwicklung und ausbildung systematisch dargestellt: Atlas. Ernst & Korn (Gropius’sche buch- und kunsthandlung). Roberson, S., Weltje, G.J., 2014. Inter-instrument comparison of particle-size analysers. Sedimentology 61, 1157–1174. doi:10.1111/sed.12093 Rohdenburg, H., Meyer, B., 1966. Zur Feinstratigraphie und Paläopedologie des Jungpleistozäns nach Untersuchungen an südniedersächsichen und nordhessischen Lössprofilen, Mitteilungen der Deutschen Bodenkundlichen Gesellschaft. Eigenverlag Deutsche Bodenkund Ii ehe Gesellschaft, Göttingen. Rosenberg, P.D., Parker, D.J., Ryder, C.L., Marsham, J.H., Garcia-Carreras, L., Dorsey, J.R., Brooks, I.M., Dean, A.R., Crosier, J., McQuaid, J.B., Washington, R., 2014. Quantifying particle size and turbulent scale dependence of dust flux in the Sahara using aircraft measurements. J. Geophys. Res. Atmos. 119, 2013JD021255. doi:10.1002/2013JD021255 Rousseau, D.-D., Antoine, P., Gerasimenko, N., Sima, A., Fuchs, M., Hatté, C., Moine, O., Zoeller, L., 2011. North Atlantic abrupt climatic events of the last glacial period recorded in Ukrainian loess deposits. Clim. Past 7, 221–234. doi:10.5194/cp-7-221-2011 Rousseau, D.D., Antoine, P., Hatté, C., Lang, A., Zöller, L., Fontugne, M., Othman, D.B., Luck, J.M., Moine, O., Labonne, M., Bentaleb, I., Jolly, D., 2002. Abrupt millennial climatic changes from Nussloch (Germany) Upper Weichselian eolian records during the Last Glaciation. Quaternary Science Reviews 21, 1577–1582. doi:10.1016/S0277-3791(02)00034-3 Rousseau, D.-D., Antoine, P., Kunesch, S., Hatté, C., Rossignol, J., Packman, S., Lang, A., Gauthier, C., 2007. Evidence of cyclic dust deposition in the US Great plains during the last deglaciation from the high-resolution analysis of the Peoria Loess in the Eustis sequence (Nebraska, USA). Earth and Planetary Science Letters 262, 159–174. doi:10.1016/j.epsl.2007.07.016 132

Rowell, D.L., 1994. Soil science: methods and applications. Longman Scientific & Technical. Schaetzl, R.J., Thompson, M.L., 2015. Soils: Genesis and Geomorphology, 2nd ed. Cambridge University Press, New York, NY. Schaller, K., 2000. Praktikum zur Bodenkunde und Pflanzenernährung, Auflage: 8., Aufl. ed. Forschungsanstalt Geisenheim, Geisenheim. Scheffer, F., Schachtschabel, Blume, H.-P., Brümmer, G., Schwertmann, U., Horn, R., Kögel- Knabner, I., Stahr, K., Wilke, Beyer, L., Auerswald, K., Stanjek, H., Fischer, W.R., Renger, M., Strebel, O., 2002. Lehrbuch der Bodenkunde. Spektrum Akademischer Verlag, Heidelberg. Schirmer, W., 2016. Late Pleistocene loess of the Lower Rhine. Quaternary International, Middle Palaeolithic in North-West Europe: Multidisciplinary Approaches 411, Part A, 44–61. doi:10.1016/j.quaint.2016.01.034 Schlütz, F., Lehmkuhl, F., 2009. Holocene climatic change and the nomadic Anthropocene in Eastern Tibet: palynological and geomorphological results from the Nianbaoyeze Mountains. Quaternary Science Reviews 28, 1449–1471. doi:10.1016/j.quascirev.2009.01.009 Schmidt, M.W.I., Noack, A.G., 2000. Black carbon in soils and sediments: Analysis, distribution, implications, and current challenges. Global Biogeochem. Cycles 14, 777–793. doi:10.1029/1999GB001208 Schmidt, M.W.I., Skjemstad, J.O., Czimczik, C.I., Glaser, B., Prentice, K.M., Gelinas, Y., Kuhlbusch, T.A.J., 2001. Comparative analysis of black carbon in soils. Global Biogeochem. Cycles 15, 163–167. doi:10.1029/2000GB001284 Schulte, P., Lehmkuhl, F., submitted. The difference of two laser diffraction patterns as an indicator for post-depositional grain size reduction in loess-paleosol sequences. Paleogeography, Paleoclimatology, Paleoecology. Schulte, P., Lehmkuhl, F., Kels, H., Loibl, C., Klasen, N., Hauck, T., 2014. Environmental change indicated by grain-size variations and trace elements: examples from two different sections - the sandy-loess sediments from the Doroshivtsy site (Ukraine) and the loess section Semlac (Romania), in: DUST2014 - 1st International Conference on Atmospheric Dust. pp. 106–112. doi:10.14644/dust.2014.017 Schulte, P., Lehmkuhl, F., Steininger, F., Loibl, D., Lockot, G., Protze, J., Fischer, P., Stauch, G., 2016. Influence of HCl pretreatment and organo-mineral complexes on laser diffraction measurement of loess–paleosol-sequences. CATENA 137, 392–405. doi:10.1016/j.catena.2015.10.015 Schulten, H.-R., Leinweber, P., 2000. New insights into organic-mineral particles: composition, properties and models of molecular structure. Biology and Fertility of Soils 30, 399–432. Schumacher, B.A., 2002. Methods for the determination of total organic carbon (TOC) in soils and sediments. Ecological Risk Assessment Support Center 1–23. Schwertmann, U., Süsser, P., Nätscher, L., 1987. Protonenpuffersubstanzen in Böden. Zeitschrift für Pflanzenernährung und Bodenkunde 150, 174–178. doi:10.1002/jpln.19871500310

133

Sheldon, N.D., Tabor, N.J., 2009. Quantitative paleoenvironmental and paleoclimatic reconstruction using paleosols. Earth-Science Reviews 95, 1–52. doi:10.1016/j.earscirev.2009.03.004 Siedentopf, H., Zsigmondy, R., 1903. Über Sichtbarmachung und Größenbestimmung ultramikoskopischer Teilchen, mit besonderer Anwendung auf Goldrubingläser. Ann. Phys. 315, 1–39. doi:10.1002/andp.19023150102 Singer, J.K., Anderson, J.B., Ledbetter, M.T., McCave, I.N., Jones, K.P.N., Wright, R., 1988. An assessment of analytical techniques for the size analysis of fine-grained sediments. Journal of Sedimentary Research 58, 534–543. doi:10.1306/212F8DE6-2B24-11D7-8648000102C1865D Six, J., Bossuyt, H., Degryze, S., Denef, K., 2004. A history of research on the link between (micro)aggregates, soil biota, and soil organic matter dynamics. Soil and Tillage Research 79, 7– 31. doi:10.1016/j.still.2004.03.008 Six, J., Elliott, E.T., Paustian, K., 2000. Soil macroaggregate turnover and microaggregate formation: a mechanism for C sequestration under no-tillage agriculture. Soil Biology and Biochemistry 32, 2099–2103. doi:10.1016/S0038-0717(00)00179-6 Smalley, I., Marković, S.B., Svirčev, Z., 2011. Loess is [almost totally formed by] the accumulation of dust. Quaternary International 240, 4–11. doi:10.1016/j.quaint.2010.07.011 Smalley, I., O’Hara-Dhand, K., Wint, J., Machalett, B., Jary, Z., Jefferson, I., 2009. Rivers and loess: The significance of long river transportation in the complex event-sequence approach to loess deposit formation. Quaternary International, Loess in the Danube Region and Surrounding Loess Provinces: The Marsigli Memorial Volume 198, 7–18. doi:10.1016/j.quaint.2008.06.009 Smalley, I.J., 1971. “In-situ” theories of loess formation and the significance of the calcium- carbonate content of loess. Earth-Science Reviews 7, 67–85. doi:10.1016/0012-8252(71)90082-1 Smalley, I.J., Leach, J.A., 1978. The origin and distribution of the loess in the Danube basin and associated regions of East-Central Europe — A review. Sedimentary Geology 21, 1–26. doi:10.1016/0037-0738(78)90031-3 Smalley, I.J., Marković, S.B., 2014. Loessification and hydroconsolidation: There is a connection. CATENA, Loess and dust dynamics, environments, landforms, and pedogenesis: a tribute to Edward Derbyshire 117, 94–99. doi:10.1016/j.catena.2013.07.006 SPECTRO, 2007. Analysis of trace elements in geological materials, soils and sludges prepared as pressed pellets, ReportXRF. SPECTRO. Sprafke, T., Obreht, I., 2016. Loess: Rock, sediment or soil – What is missing for its definition? Quaternary International 399, 198–207. doi:10.1016/j.quaint.2015.03.033 Stamati, F.E., Nikolaidis, Ν.P., Banwart, S., Blum, W.E.H., 2013. A coupled carbon, aggregation, and structure turnover (CAST) model for topsoils. Geoderma 211–212, 51–64. doi:10.1016/j.geoderma.2013.06.014 Stauch, G., IJmker, J., Pötsch, S., Zhao, H., Hilgers, A., Diekmann, B., Dietze, E., Hartmann, K., Opitz, S., Wünnemann, B., Lehmkuhl, F., 2012. Aeolian sediments on the north-eastern Tibetan Plateau. Quaternary Science Reviews 57, 71–84. doi:10.1016/j.quascirev.2012.10.001

134

Stauch, G., Pötsch, S., Zhao, H., Lehmkuhl, F., 2014. Interaction of geomorphological processes on the north-eastern Tibetan Plateau during the Holocene, an example from a sub-catchment of Lake Donggi Cona. Geomorphology 210, 23–35. doi:10.1016/j.geomorph.2013.12.014 Stein, R., 1985. Rapid Grain-size Analyses of Clay and Silt Fraction by Sedigraph 5000d: Comparison with Coulter Counter and Atterberg Methods: RESEARCH METHOD PAPER. Journal of Sedimentary Research 55. Stevens, T., Armitage, S.J., Lu, H., Thomas, D.S.G., 2006. Sedimentation and diagenesis of Chinese loess: Implications for the preservation of continuous, high-resolution climate records. Geology 34, 849–852. doi:10.1130/G22472.1 Stevens, T., Carter, A., Watson, T.P., Vermeesch, P., Andò, S., Bird, A.F., Lu, H., Garzanti, E., Cottam, M.A., Sevastjanova, I., 2013. Genetic linkage between the Yellow River, the Mu Us desert and the Chinese Loess Plateau. Quaternary Science Reviews 78, 355–368. doi:10.1016/j.quascirev.2012.11.032 Stevens, T., Marković, S.B., Zech, M., Hambach, U., Sümegi, P., 2011. Dust deposition and climate in the Carpathian Basin over an independently dated last glacial–interglacial cycle. Quaternary Science Reviews 30, 662–681. doi:10.1016/j.quascirev.2010.12.011 Stokes, G.G., 1851. On the effect of the internal friction of fluids on the motion of pendulums. Pitt Press. Storti, F., Balsamo, F., 2010. Particle size distributions by laser diffraction: Sensitivity of granular matter strength to analytical operating procedures. Solid Earth 1, 25–48. Stumm, W., 1992. Chemistry of the Solid-Water Interface: Processes at the Mineral-Water and Particle-Water Interface in Natural Systems. Wiley-Interscience, New York. Suito, E., 1990. Electron Microscopic Investigations of Powders: The Shape, Crystal Habit and Microstructure of Fine Particles, in: Sōmiya, S., Moriyoshi, Y. (Eds.), Sintering Key Papers. Springer Netherlands, pp. 689–706. Sun, D., Bloemendal, J., Rea, D.K., Vandenberghe, J., Jiang, F., An, Z., Su, R., 2002. Grain-size distribution function of polymodal sediments in hydraulic and aeolian environments, and numerical partitioning of the sedimentary components. Sedimentary Geology 152, 263–277. doi:10.1016/S0037-0738(02)00082-9 Sun, Y., Clemens, S.C., Morrill, C., Lin, X., Wang, X., An, Z., 2012. Influence of Atlantic meridional overturning circulation on the East Asian winter monsoon. Nature Geosci 5, 46–49. doi:10.1038/ngeo1326 Sun, Y., Lu, H., An, Z., 2006. Grain size of loess, palaeosol and Red Clay deposits on the Chinese Loess Plateau: Significance for understanding pedogenic alteration and palaeomonsoon evolution. Palaeogeography, Palaeoclimatology, Palaeoecology, Monsoon and Tectonics of AsiaIGCP 476, Monsoons and Tectonics 241, 129–138. doi:10.1016/j.palaeo.2006.06.018 Sverdrup, H., Warfvinge, P., Rosén, K., 1992. A model for the impact of soil solution Ca: Al ratio, soil moisture and temperature on tree base cation uptake. Water Air Soil Pollut 61, 365– 383. doi:10.1007/BF00482616 Svirčev, Z., Marković, S.B., Stevens, T., Codd, G.A., Smalley, I., Simeunović, J., Obreht, I., Dulić, T., Pantelić, D., Hambach, U., 2013. Importance of biological loess crusts for loess formation in 135

semi-arid environments. Quaternary International, Closing the gap - North Carpathian loess traverse in the Eurasian loess belt 6th Loess Seminar, Wrocław, Poland Dedicated to Prof. Henryk Maruszczak 296, 206–215. doi:10.1016/j.quaint.2012.10.048 Syvitski, J.P.M., LeBlanc, K.W.G., Asprey, K.W., 1991. Interlaboratory, interinstrument calibration experiment, in: Syvitski, J.P.M. (Ed.), Principles, Methods, and Application of Particle Size Analysis. Cambridge University Press. Tafel, A., 1914. Meine Tibetreise: eine Studienfahrt durch das Nordwestliche China und durch die innere Mongolei in das östliche Tibet. Union deutsche Verlagsgesellschaft, Stuttgart, Berlin, Leipzig. Terhorst, B., Ottner, F., Wriessnig, K., 2012. Weathering intensity and pedostratigraphy of the Middle to Upper Pleistocene loess/palaeosol sequence of Wels-Aschet (Upper Austria). Quaternary International, Paleosols in soilscapes of the past and present 265, 142–154. doi:10.1016/j.quaint.2011.08.042 Thomas, D.S.G., 2013. Reconstructing paleoenvironments and palaeoclimates in drylands: what can landform analysis contribute? Earth Surf. Process. Landforms 38, 3–16. doi:10.1002/esp.3190 Thomas, D.S.G., Wiggs, G.F.S., 2008. Aeolian system responses to global change: challenges of scale, process and temporal integration. Earth Surf. Process. Landforms 33, 1396–1418. doi:10.1002/esp.1719 Tiessen, H., Stewart, J.W.B., Hunt, H.W., 1984. Concepts of soil organic matter transformations in relation to organo-mineral particle size fractions. Plant Soil 76, 287–295. doi:10.1007/BF02205587 Titze, B., 2013. Techniques to prevent sample surface charging and reduce beam damage effects for SBEM imaging. Ruprecht-Karls-Universität Heidelberg, Heidelberg. Treadwell-Steitz, C., McFadden, L.D., 2000. Influence of parent material and grain size on carbonate coatings in gravelly soils, Palo Duro Wash, New Mexico. Geoderma 94, 1–22. Tsoar, H., Pye, K., 1987. Dust transport and the question of desert loess formation. Sedimentology 34, 139–153. doi:10.1111/j.1365-3091.1987.tb00566.x Újvári, G., Kok, J.F., Varga, G., Kovács, J., 2016. The physics of wind-blown loess: Implications for grain size proxy interpretations in Quaternary paleoclimate studies. Earth-Science Reviews 154, 247–278. doi:10.1016/j.earscirev.2016.01.006 Újvári, G., Varga, A., Ramos, F.C., Kovács, J., Németh, T., Stevens, T., 2012. Evaluating the use of clay mineralogy, Sr–Nd isotopes and zircon U–Pb ages in tracking dust provenance: An example from loess of the Carpathian Basin. Chemical Geology 304–305, 83–96. doi:10.1016/j.chemgeo.2012.02.007 Újvári, G., Varga, A., Raucsik, B., Kovács, J., 2014. The Paks loess-paleosol sequence: A record of chemical weathering and provenance for the last 800 ka in the mid-Carpathian Basin. Quaternary International, Loess Research and Lithostratigraphy in Hungary 319, 22–37. doi:10.1016/j.quaint.2012.04.004 Vaasma, T., 2008. Grain-size analysis of lacustrine sediments: a comparison of pre-treatment methods. Estonian Journal of Ecology 57, 231. doi:10.3176/eco.2008.4.01 136

van der Veer, G., 2006. Geochemical soil survey of the Netherlands: atlas of major and trace elements in topsoil and parent material; assessment of natural and anthropogenic enrichment factors. Koninklijk Nederlands Aardrijkskundig Genootschap, Utrecht. Van Gestel, M., Merckx, R., Vlassak, K., 1996. Spatial distribution of microbial biomass in microaggregates of a silty-loam soil and the relation with the resistance of microorganisms to soil drying. Soil Biology and Biochemistry 28, 503–510. Vandenberghe, J., 2013. Grain size of fine-grained windblown sediment: A powerful proxy for process identification. Earth-Science Reviews 121, 18–30. doi:10.1016/j.earscirev.2013.03.001 Vandenberghe, J., Markovič, S.B., Jovanovič, M., Hambach, U., 2014. Site-specific variability of loess and palaeosols (Ruma, Vojvodina, northern Serbia). Quaternary International 334–335, 86– 93. doi:10.1016/j.quaint.2013.10.036 Vandenberghe, J., Meijer, T., Kolfschoten, M. van, 1985. Sedimentary processes, periglacial activity and stratigraphy of the loess and fluvial deposits at Maastricht‑Belvédère (The Netherlands). Vandenberghe, J., Zhisheng, A., Nugteren, G., Huayu, L., Van Huissteden, K., 1997. New absolute time scale for the Quaternary climate in the Chinese loess region by grain-size analysis. Geology 25, 35–38. Veghte, D.P., Freedman, M.A., 2014. Facile Method for Determining the Aspect Ratios of Mineral Dust Aerosol by Electron Microscopy. Aerosol Science and Technology 48, 715–724. doi:10.1080/02786826.2014.920484 Veghte, D.P., Moore, J.E., Jensen, L., Freedman, M.A., 2015. Influence of shape on the optical properties of hematite aerosol: optical properties of heamaite aerosol. Journal of Geophysical Research: Atmospheres 120, 7025–7039. doi:10.1002/2015JD023160 Volten, H., Jalava, J.-P., Lumme, K., De Haan, J.F., Vassen, W., Hovenier, J.W., 1999. Laboratory measurements and T-matrix calculations of the scattering matrix of rutile particles in water. Applied optics 38, 5232–5240. Volten, H., Muñoz, O., Rol, E., de Haan, J.F., Vassen, W., Hovenier, J.W., Muinonen, K., Nousiainen, T., 2001. Scattering matrices of mineral aerosol particles at 441.6 nm and 632.8 nm. J. Geophys. Res. 106, 17375–17401. doi:10.1029/2001JD900068 von Lützow, M., Kögel-Knabner, I., Ekschmitt, K., Flessa, H., Guggenberger, G., Matzner, E., Marschner, B., 2007. SOM fractionation methods: Relevance to functional pools and to stabilization mechanisms. Soil Biology and Biochemistry 39, 2183–2207. doi:10.1016/j.soilbio.2007.03.007 von Lützow, M., Kogel-Knabner, I., Ekschmitt, K., Matzner, E., Guggenberger, G., Marschner, B., Flessa, H., 2006. Stabilization of organic matter in temperate soils: mechanisms and their relevance under different soil conditions - a review. European Journal of Soil Science 57, 426– 445. doi:10.1111/j.1365-2389.2006.00809.x von Lützow, M., Kögel-Knabner, I., Ludwig, B., Matzner, E., Flessa, H., Ekschmitt, K., Guggenberger, G., Marschner, B., Kalbitz, K., 2008. Stabilization mechanisms of organic matter in four temperate soils: Development and application of a conceptual model. Journal of Plant Nutrition and Soil Science 171, 111–124. doi:10.1002/jpln.200700047 137

von Suchodoletz, H., Faust, D., Zöller, L., 2009. Geomorphological investigations of sediment traps on Lanzarote (Canary Islands) as a key for the interpretation of a palaeoclimate archive off NW Africa. Quaternary International, Lower Latitudes Loess-Dust Transport Past and Present 196, 44–56. doi:10.1016/j.quaint.2008.03.014 Vriend, M., Prins, M.A., Buylaert, J.-P., Vandenberghe, J., Lu, H., 2011. Contrasting dust supply patterns across the north-western Chinese Loess Plateau during the last glacial-interglacial cycle. Quaternary International, The Second Loessfest (2009) 240, 167–180. doi:10.1016/j.quaint.2010.11.009 Weber, C., Heuser, M., Mertens, G., Stanjek, H., 2014. Determination of clay mineral aspect ratios from conductometric titrations. Clay Minerals 49, 17–26. doi:10.1180/claymin.2014.049.1.02 Weiss, E.L., Frock, H.N., 1976. Rapid analysis of particle-size distributions by laser light scattering. Powder Technology 14, 287–293. Weltje, G.J., Bloemsma, M.R., Tjallingii, R., Heslop, D., Röhl, U., Croudace, I.W., 2015. Prediction of Geochemical Composition from XRF Core Scanner Data: A New Multivariate Approach Including Automatic Selection of Calibration Samples and Quantification of Uncertainties, in: Croudace, I.W., Rothwell, R.G. (Eds.), Micro-XRF Studies of Sediment Cores, Developments in Paleoenvironmental Research. Springer Netherlands, pp. 507–534. doi:10.1007/978-94-017-9849-5_21 Weltje, G.J., Tjallingii, R., 2008. Calibration of XRF core scanners for quantitative geochemical logging of sediment cores: Theory and application. Earth and Planetary Science Letters 274, 423– 438. doi:10.1016/j.epsl.2008.07.054 Weninger, B., Jöris, O., 2008. A 14C age calibration curve for the last 60 ka: the Greenland-Hulu U/Th timescale and its impact on understanding the Middle to Upper Paleolithic transition in Western Eurasia. Journal of Human Evolution, Chronology of the Middle-Upper Paleolithic Transition in Eurasia 55, 772–781. doi:10.1016/j.jhevol.2008.08.017 Wintle, A.G., Murray, A.S., 2006. A review of quartz optically stimulated luminescence characteristics and their relevance in single-aliquot regeneration dating protocols. Radiation Measurements 41, 369–391. doi:10.1016/j.radmeas.2005.11.001 Wiseman, C.L.S., Püttmann, W., 2006. Interactions between mineral phases in the preservation of soil organic matter. Geoderma 134, 109–118. doi:10.1016/j.geoderma.2005.09.001 Wright, J.S., 2001. “Desert” loess versus “glacial” loess: quartz silt formation, source areas and sediment pathways in the formation of loess deposits. Geomorphology 36, 231–256. doi:10.1016/S0169-555X(00)00060-X Xu, J., Zhang, J., Yu, Y., Meng, Q., Zhong, H., 2016. Characteristics of Vapor Condensation on Coal-Fired Fine Particles. Energy & Fuels 30, 1822–1828. doi:10.1021/acs.energyfuels.5b02200 Xu, R., Di Guida, O.A., 2003. Comparison of sizing small particles using different technologies. Powder Technology 132, 145–153. doi:10.1016/S0032-5910(03)00048-2 Yang, S., Fang, X., Shi, Z., Lehmkuhl, F., Song, C., Han, Y., Han, W., 2010. Timing and provenance of loess in the Sichuan Basin, southwestern China. Palaeogeography, Palaeoclimatology, Palaeoecology 292, 144–154. doi:10.1016/j.palaeo.2010.03.039 138

Yu, L., Lai, Z., 2014. Holocene climate change inferred from stratigraphy and OSL chronology of aeolian sediments in the Qaidam Basin, northeastern Qinghai–Tibetan Plateau. Quaternary Research 81, 488–499. doi:10.1016/j.yqres.2013.09.006 Yu, L., Lai, Z., 2012. OSL chronology and palaeoclimatic implications of aeolian sediments in the eastern Qaidam Basin of the northeastern Qinghai-Tibetan Plateau. Palaeogeography, Palaeoclimatology, Palaeoecology 337–338, 120–129. doi:10.1016/j.palaeo.2012.04.004 Zech, M., Zech, R., Zech, W., Glaser, B., Brodowski, S., Amelung, W., 2008. Characterisation and palaeoclimate of a loess-like permafrost palaeosol sequence in NE Siberia. Geoderma 143, 281–295. doi:10.1016/j.geoderma.2007.11.012 Zech, W., Schad, P., Hintermaier-Erhard, G., 2014. Böden der Welt: Ein Bildatlas. Springer Spektrum, Berlin. Zeeden, C., Kels, H., Hambach, U., Schulte, P., Protze, J., Eckmeier, E., Marković, S.B., Klasen, N., Lehmkuhl, F., 2016. Three climatic cycles recorded in a loess-paleosol sequence at Semlac (Romania) – implications for dust accumulation in south-eastern Europe. Quaternary Science Reviews. Zens, J., Schulte, P., Klasen, N., Krauß, L., Pirson, S., Burow, C., Brill, D., Eckmeier, E., Kels, H., Zeeden, C., Spagna, P., Lehmkuhl, F., submitted. OSL chronologies of paleoenvironmental dynamics recorded by loess-paleosol sequences from Europe: Case studies from the Rhine- Meuse area and the Neckar Basin. Palaeogeography, Palaeoclimatology, Palaeoecology,. Zhang, J., Shao, Y., 2014. A new parameterization of particle dry deposition over rough surfaces. Atmospheric Chemistry and Physics 14, 12429–12440. doi:10.5194/acp-14-12429-2014 Zhao, L., Ji, J., Chen, J., Liu, L., Chen, Y., Balsam, W., 2005. Variations of illite/chlorite ratio in Chinese loess sections during the last glacial and interglacial cycle: Implications for monsoon reconstruction. Geophys. Res. Lett. 32, L20718. doi:10.1029/2005GL024145 Zobeck, T.M., 2004. Rapid soil particle size analyses using laser diffraction. Applied engineering in agriculture 20, 633–640.

139

Appendix A

Tab. A1: Grain size details and CaCO3 content of the loess-paleosol sequence Suohuduo

Grain-size fractions [%] Grain-size (clay, silt, sand) [%] sample depth [cm] CaCO3 [%] C fSi mSi cSi<36 cSi>36 fS mS gS ∑C ∑Si ∑S 4173 5 14.90 9.59 13.90 12.50 23.00 25.70 0.23 0 14.90 58.99 25.93 0.00 4174 15 14.60 9.43 12.90 11.90 23.10 27.70 0.15 0 14.60 57.33 27.85 4175 25 13.70 8.84 12.40 11.40 22.80 30.60 0.26 0 13.70 55.44 30.86 0.17 4176 35 13.40 9.20 12.70 11.30 22.30 30.40 0.21 0 13.40 55.50 30.61 4177 42.5 11.80 7.88 11.00 10.40 23.50 35.00 0.25 0 11.80 52.78 35.25 0.00 4178 47.5 11.90 7.98 11.30 10.70 23.60 34.10 0.21 0 11.90 53.58 34.31 4179 52.5 13.00 8.82 12.70 11.60 23.00 30.40 0.16 0 13.00 56.12 30.56 0.00 4180 57.5 11.90 7.91 12.00 11.50 23.70 32.60 0.20 0 11.90 55.11 32.80 4181 62.5 1.19 4182 67.5 10.60 7.78 11.60 10.70 22.90 35.00 0.79 0 10.60 52.98 35.79 4183 72.5 11.30 7.91 11.70 11.10 23.50 33.80 0.35 0 11.30 54.21 34.15 0.68 4184 77.5 10.70 7.50 11.10 10.80 23.80 35.30 0.50 0 10.70 53.20 35.80 4185 82.5 8.96 6.27 9.43 9.51 23.40 40.40 1.33 0 8.96 48.61 41.73 8.46 4186 87.5 10.80 7.83 11.20 10.50 22.20 35.70 0.98 0 10.80 51.73 36.68 4187 92.5 13.40 11.20 17.30 14.20 21.60 21.60 0.19 0 13.40 64.30 21.79 4.58 4188 97.5 13.90 12.80 20.40 16.10 19.80 16.50 0.07 0 13.90 69.10 16.57 4189 102.5 14.30 12.10 20.10 16.40 20.20 16.60 0.04 0 14.30 68.80 16.64 1.02 4190 107.5 13.60 11.40 17.10 13.70 21.50 22.10 0.06 0 13.60 63.70 22.16 4191 112.5 12.90 10.70 16.50 13.40 21.70 24.10 0.08 0 12.90 62.30 24.18 0.68 4192 117.5 10.20 6.84 10.80 10.70 25.30 35.90 0.09 0 10.20 53.64 35.99 4193 122.5 7.99 5.20 8.12 8.88 25.10 44.10 0.40 0 7.99 47.30 44.50 4194 127.5 6.72 4.46 6.85 7.85 26.60 46.50 0.47 0 6.72 45.76 46.97 9.82 4195 132.5 6.80 4.66 7.50 8.47 26.90 44.40 0.51 0 6.80 47.53 44.91 10.14 140

4196 137.5 8.36 5.98 9.67 11.00 28.00 36.00 0.25 0 8.36 54.65 36.25 4197 142.5 11.00 8.14 12.70 12.30 24.50 30.30 0.38 0 11.00 57.64 30.68 6.27 4198 147.5 12.10 9.39 14.00 11.80 21.90 29.40 0.79 0 12.10 57.09 30.19 4199 152.5 11.70 10.50 16.50 12.70 21.40 26.10 0.44 0 11.70 61.10 26.54 5.59 4200 157.5 13.70 12.40 20.60 15.10 20.00 17.60 0.10 0 13.70 68.10 17.70 4201 162.5 13.60 12.30 20.00 14.30 19.60 19.60 0.12 0 13.60 66.20 19.72 4.06 4202 167.5 4203 172.5 12.90 10.10 15.00 13.20 24.30 23.90 0.06 0 12.90 62.60 23.96 1.36 4204 177.5 14.10 10.90 17.10 15.00 23.20 19.30 0.13 0 14.10 66.20 19.43 4205 182.5 14.40 12.60 20.20 16.10 20.40 15.80 0.04 0 14.40 69.30 15.84 1.70 4206 187.5 14.00 15.20 25.90 17.60 16.30 10.60 0.05 0 14.00 75.00 10.65 4207 192.5 13.50 13.40 22.70 18.20 19.60 12.00 0.02 0 13.50 73.90 12.02 4.23 4208 197.5 11.20 10.20 17.90 18.30 24.90 17.00 0.02 0 11.20 71.30 17.02 4209 202.5 9.28 7.56 13.20 15.00 27.60 26.00 0.44 0 9.28 63.36 26.44 5.42 4210 207.5 8.78 6.30 10.90 13.60 28.20 30.70 0.75 0 8.78 59.00 31.45 4211 212.5 9.08 6.54 11.30 13.70 28.10 29.50 0.97 0 9.08 59.64 30.47 11.18 4212 217.5 9.92 7.00 11.50 14.10 28.10 27.80 0.78 0 9.92 60.70 28.58 4213 222.5 9.69 7.06 11.40 12.70 27.10 30.30 0.84 0 9.69 58.26 31.14 12.70 4214 227.5 11.70 8.94 14.50 15.20 25.00 22.80 0.82 0 11.70 63.64 23.62 4215 232.5 12.10 9.55 15.40 15.80 24.40 21.10 0.62 0 12.10 65.15 21.72 13.36 4216 237.5 10.20 8.39 12.60 12.60 23.90 30.50 0.63 0 10.20 57.49 31.13 4217 242.5 11.10 9.08 14.60 15.00 23.70 25.20 0.50 0 11.10 62.38 25.70 14.31 4218 247.5 8.84 7.00 12.10 12.40 24.00 33.70 0.76 0 8.84 55.50 34.46 4219 252.5 8.06 5.92 10.60 12.20 25.80 35.90 0.61 0 8.06 54.52 36.51 13.55 4220 257.5 8.67 6.77 14.00 15.80 26.00 27.20 0.66 0 8.67 62.57 27.86 4221 262.5 7.05 4.87 9.28 11.90 27.80 37.80 0.58 0 7.05 53.85 38.38 12.76 4222 267.5 7.12 5.10 10.20 13.20 28.10 35.10 0.41 0 7.12 56.60 35.51 4223 272.5 6.86 4.90 9.90 13.00 27.60 36.80 0.46 0 6.86 55.40 37.26 12.35 4224 277.5 7.43 5.36 10.30 12.60 28.00 35.10 0.45 0 7.43 56.26 35.55 4225 282.5 7.05 5.07 10.20 13.10 28.50 35.00 0.40 0 7.05 56.87 35.40 12.06

141

4226 287.5 6.99 5.00 9.69 12.20 28.30 36.70 0.43 0 6.99 55.19 37.13 4227 292.5 7.21 5.08 9.19 11.70 28.70 37.00 0.39 0 7.21 54.67 37.39 12.81 4228 297.5 7.49 5.06 9.04 11.80 29.00 36.60 0.27 0 7.49 54.90 36.87 4229 302.5 6.43 4.27 7.64 10.20 27.70 42.60 0.60 0 6.43 49.81 43.20 13.04 4230 307.5 6.16 4.25 7.75 10.30 27.80 42.60 0.55 0 6.16 50.10 43.15 4231 312.5 5.75 3.94 6.93 8.99 26.70 46.30 0.77 0 5.75 46.56 47.07 12.44 4232 317.5 5.09 3.20 5.28 7.19 25.50 51.80 1.40 0 5.09 41.17 53.20 4233 322.5 5.06 3.16 5.32 7.56 26.50 51.10 0.78 0 5.06 42.54 51.88 12.10 4234 327.5 4.55 2.74 4.21 5.79 22.70 58.60 1.13 0 4.55 35.44 59.73 4235 332.5 4.28 2.47 3.65 5.01 22.30 60.30 1.74 0 4.28 33.43 62.04 11.55 4236 337.5 4.79 2.64 3.79 4.91 21.50 60.70 1.59 0 4.79 32.84 62.29 Standard deviation 3.04 2.98 4.66 2.93 2.89 11.07 0.39 0 3.48 13.19 12.56 5.30 Mean 10.08 7.66 12.34 12.21 24.39 32.27 0.49 0 9.76 54.83 31.74 7.17

142

Tab. A2: XRF details of the loess-paleosol sequence Suohuduo

Depth Element concentration [ppm] sample [cm] Mg Mg mean Al Al mean P P mean Ca Ca mean Ti Ti mean Mn Mn mean Zr Zr mean 4173 5 11210 11225 64040 64275 1136 1141.5 13110 13135 4085 4083.5 740.1 744.85 345.9 346.3 4173 11240 64510 1147 13160 4082 749.6 346.7 4174 15 11720 11805 67610 67540 1111 1109.5 11740 11735 4131 4137 763.2 768.6 369.2 369.4 4174 11890 67470 1108 11730 4143 774 369.6 4175 25 13090 13065 73580 73265 1059 1060 11500 11465 4549 4579 818 816.75 387.3 387.3 4175 13040 72950 1061 11430 4609 815.5 387.3 4176 35 11860 12010 67690 67820 1118 1124 12860 12935 4085 4089 778.1 781.2 343 340.8 4176 12160 67950 1130 13010 4093 784.3 338.6 4177 42.5 12760 12725 70970 71040 1099 1111 12330 12315 4369 4347.5 765.9 770.95 406.5 407.6 4177 12690 71110 1123 12300 4326 776 408.7 4178 47.5 12440 12430 70420 70450 1031 1028.5 12180 12265 4406 4384.5 792.1 794.2 394.7 395.4 4178 12420 70480 1026 12350 4363 796.3 396.1 4179 52.5 12810 12795 70820 70785 1032 1041 12350 12420 4275 4292 841.3 836 366.8 368.45 4179 12780 70750 1050 12490 4309 830.7 370.1 4180 57.5 12730 12675 70760 70755 919.4 908.1 11220 11210 4415 4394 781.3 782 375.7 372.25 4180 12620 70750 896.8 11200 4373 782.7 368.8 4181 62.5 13020 13095 71650 71605 805.9 810.45 16290 16180 4330 4355.5 765.1 766.25 393.2 391.5 4181 13170 71560 815 16070 4381 767.4 389.8 4182 67.5 12350 12440 63890 63755 902.7 923.95 38460 38315 4011 4019 693.1 693.15 353.5 353.55 4182 12530 63620 945.2 38170 4027 693.2 353.6 4183 72.5 12640 12515 69700 69715 881.3 868.05 13020 13005 4407 4420.5 777.9 774.8 405.5 406.6 4183 12390 69730 854.8 12990 4434 771.7 407.7 4184 77.5 11740 11810 64410 64280 927.1 934.45 26400 26660 4195 4204 751.1 752.1 394.6 396.85 4184 11880 64150 941.8 26920 4213 753.1 399.1 4185 82.5 11570 11575 58320 58155 866.9 847.6 48850 49060 3937 3938.5 606.1 607.65 442.3 441.9 4185 11580 57990 828.3 49270 3940 609.2 441.5 143

4186 87.5 11800 11850 61470 61475 858 850.55 41660 41790 4153 4198.5 742.6 747.05 429.3 429.9 4186 11900 61480 843.1 41920 4244 751.5 430.5 4187 92.5 11640 11440 62960 63080 1110 1112.5 37850 37850 4053 4057 833.5 827.8 321.4 326.7 4187 11240 63200 1115 37850 4061 822.1 332 4188 97.5 11970 12090 65840 66180 1333 1341 29930 29835 4094 4102 886.6 884.25 295.2 296.75 4188 12210 66520 1349 29740 4110 881.9 298.3 4189 102.5 12460 12405 68120 68135 1468 1459.5 21650 21700 4217 4240.5 926.1 934.6 290.3 292.35 4189 12350 68150 1451 21750 4264 943.1 294.4 4190 107.5 12890 12810 69800 69270 1448 1446.5 20130 20145 4251 4250.5 910.4 911.55 310.3 310.45 4190 12730 68740 1445 20160 4250 912.7 310.6 4191 112.5 12710 12815 70180 69990 1393 1388 18370 18390 4239 4248.5 897.8 903.7 331.4 332.8 4191 12920 69800 1383 18410 4258 909.6 334.2 4192 117.5 12740 12680 70290 70265 1022 1026 12290 12320 4354 4384 769.8 767.9 405.5 406.6 4192 12620 70240 1030 12350 4414 766 407.7 4193 122.5 11830 11920 63640 63715 777.7 785.7 21600 21525 4024 4050 659.2 657.3 420.3 420.6 4193 12010 63790 793.7 21450 4076 655.4 420.9 4194 127.5 12380 12390 55520 55310 854.6 847.05 60700 61035 3479 3490.5 534 537.6 349.8 344.2 4194 12400 55100 839.5 61370 3502 541.2 338.6 4195 132.5 13230 13360 59090 59320 829.4 817.05 64670 64525 3829 3871.5 602 601.4 380.3 378.95 4195 13490 59550 804.7 64380 3914 600.8 377.6 4196 137.5 12850 12735 60570 60570 769.7 794 47870 47815 3898 3892 557.5 557.95 340 343.4 4196 12620 60570 818.3 47760 3886 558.4 346.8 4197 142.5 12640 12745 65140 65540 735.7 737.75 33130 33085 4026 4053 733.5 728.1 354.2 354.05 4197 12850 65940 739.8 33040 4080 722.7 353.9 4198 147.5 13650 13635 66540 66460 876.8 875.2 43840 43830 4226 4223 977.3 975.8 328.3 324.25 4198 13620 66380 873.6 43820 4220 974.3 320.2 4199 152.5 12340 12145 63900 63865 956.8 945.7 37570 37535 4092 4115 804.5 803.35 336.3 340.35 4199 11950 63830 934.6 37500 4138 802.2 344.4 4200 157.5 12310 12425 66210 66265 1149 1143.5 34130 34155 4139 4139.5 887.6 895.25 324.6 321.7 4200 12540 66320 1138 34180 4140 902.9 318.8

144

4201 162.5 12370 12310 65690 65460 1141 1138.5 34360 34445 4190 4190 890 889.2 327.4 325.9 4201 12250 65230 1136 34530 4190 888.4 324.4 4202 167.5 12120 12155 66930 66975 999.4 999.7 26620 26645 4208 4236.5 869.8 867.9 335.4 337.3 4202 12190 67020 1000 26670 4265 866 339.2 4203 172.5 12350 12230 68370 68285 835.7 840.85 19900 19940 4393 4388 794.8 793.7 377.5 380.2 4203 12110 68200 846 19980 4383 792.6 382.9 4204 177.5 12590 12530 69980 69840 741.8 737.35 17640 17620 4424 4463.5 814.8 818.4 349 349.15 4204 12470 69700 732.9 17600 4503 822 349.3 4205 182.5 12690 12805 69130 69215 1156 1139 21880 21820 4254 4240.5 839.8 839.3 296.5 302.35 4205 12920 69300 1122 21760 4227 838.8 308.2 4206 187.5 12320 12315 65670 65665 1377 1380 34050 34075 4054 4063.5 726.4 726.95 263.4 264.25 4206 12310 65660 1383 34100 4073 727.5 265.1 4207 192.5 12450 12425 66660 66670 1129 1127.5 32350 32315 4173 4150.5 627.2 627.85 265.6 266.6 4207 12400 66680 1126 32280 4128 628.5 267.6 4208 197.5 13040 13085 69690 69800 1074 1075 20510 20470 4245 4244 743.9 743.85 299.6 304.1 4208 13130 69910 1076 20430 4243 743.8 308.6 4209 202.5 13160 13175 65960 65730 830.5 825.65 33510 33545 4083 4106 793 793.3 309.4 310.35 4209 13190 65500 820.8 33580 4129 793.6 311.3 4210 207.5 13830 13880 62370 62215 808.2 808.7 56240 56180 4031 4025 646.2 644.45 327.4 327.65 4210 13930 62060 809.2 56120 4019 642.7 327.9 4211 212.5 13970 13885 60530 60400 830.4 818.7 57350 57405 3820 3821 595.7 602 307.9 306.75 4211 13800 60270 807 57460 3822 608.3 305.6 4212 217.5 13720 13700 58330 58345 812.6 810.2 68850 68660 3822 3820.5 585.9 588.85 296.3 297 4212 13680 58360 807.8 68470 3819 591.8 297.7 4213 222.5 14230 14195 59380 59210 859 865.65 70550 70555 3890 3865.5 565.3 566.05 308.5 310.2 4213 14160 59040 872.3 70560 3841 566.8 311.9 4214 227.5 13690 13700 58260 58315 770.2 768.7 76870 76855 3753 3753.5 595.6 597.2 269 267.8 4214 13710 58370 767.2 76840 3754 598.8 266.6 4215 232.5 13190 13145 57890 57860 814.2 799.9 75840 76030 3827 3794 682.8 685.45 272.5 273 4215 13100 57830 785.6 76220 3761 688.1 273.5

145

4216 237.5 12910 12885 60200 60345 791.9 799.6 61700 61470 3898 3895 603.9 604.1 302.5 300.8 4216 12860 60490 807.3 61240 3892 604.3 299.1 4217 242.5 12730 12715 57640 57620 755.8 770.2 67920 67990 3780 3778.5 669.4 672.75 270.8 274.65 4217 12700 57600 784.6 68060 3777 676.1 278.5 4218 247.5 14420 14445 62820 62810 854 858.2 54320 54220 4130 4145 643.2 642.45 316.6 312 4218 14470 62800 862.4 54120 4160 641.7 307.4 4219 252.5 12390 12445 53360 53125 705.4 735.2 70720 71210 3734 3738 580.5 582.35 342.2 348.75 4219 12500 52890 765 71700 3742 584.2 355.3 4220 257.5 13260 13460 57580 57590 745.2 736.45 62840 62390 3701 3679.5 617.1 613.6 313.1 314.2 4220 13660 57600 727.7 61940 3658 610.1 315.3 4221 262.5 12930 12845 55580 55580 759.7 755.3 62180 62460 3807 3786 583.3 588.15 369.6 370.1 4221 12760 55580 750.9 62740 3765 593 370.6 4222 267.5 13790 13790 58750 58745 802.8 821.15 59990 59970 3790 3804.5 606.5 608.4 328.6 329.45 4222 13790 58740 839.5 59950 3819 610.3 330.3 4223 272.5 13590 13625 58100 58440 720.3 732.1 56070 56305 4006 3996 629.2 627 336 338.6 4223 13660 58780 743.9 56540 3986 624.8 341.2 4224 277.5 13630 13550 55330 55690 709.1 734.2 66820 67430 3707 3728.5 578 577.75 340.4 340.45 4224 13470 56050 759.3 68040 3750 577.5 340.5 4225 282.5 13400 13515 57270 57605 740.9 730.4 63070 63145 3916 3913.5 629.8 633.65 347.8 348.3 4225 13630 57940 719.9 63220 3911 637.5 348.8 4226 287.5 13550 13500 56490 56400 771.3 772.1 66600 66745 3755 3747.5 645.4 645.5 332.7 332.25 4226 13450 56310 772.9 66890 3740 645.6 331.8 4227 292.5 13670 13740 56660 56850 742.9 744.55 67110 67035 3741 3763.5 630.9 627.05 339.4 337.9 4227 13810 57040 746.2 66960 3786 623.2 336.4 4228 297.5 14650 14445 56890 56605 783 781.25 73540 73750 3628 3654 569.2 568 313.2 315 4228 14240 56320 779.5 73960 3680 566.8 316.8 4229 302.5 14330 14170 55450 55240 729 746.1 74820 74925 3641 3647 580.1 581.4 295.2 297.2 4229 14010 55030 763.2 75030 3653 582.7 299.2 4230 307.5 13910 13950 54780 54685 801.5 800.15 70240 70420 3535 3522.5 575.3 575.7 308.8 308.45 4230 13990 54590 798.8 70600 3510 576.1 308.1

146

4231 312.5 13800 13650 54580 54085 754 759 74520 74530 3745 3728 635 637.05 354.1 358.7 4231 13500 53590 764 74540 3711 639.1 363.3 4232 317.5 12780 12815 51930 51910 725 719.5 73360 73610 3598 3622.5 624.4 623.6 323.6 328.4 4232 12850 51890 714 73860 3647 622.8 333.2 4233 322.5 13070 12860 52210 51785 787.6 764.55 73990 74520 3887 3823.5 610.3 612.45 394.3 394.55 4233 12650 51360 741.5 75050 3760 614.6 394.8 4234 327.5 12920 12955 50670 51005 825.1 803.85 74180 74190 3548 3607.5 593.6 589.5 368.2 369.15 4234 12990 51340 782.6 74200 3667 585.4 370.1 4235 332.5 13530 13280 52740 52380 747.2 736.2 74980 74440 3579 3580.5 585.8 592.8 325.3 330.65 4235 13030 52020 725.2 73900 3582 599.8 336 4236 337.5 13380 13425 54200 54115 735.4 691.3 75470 75535 3505 3476.5 574.8 579.55 314.6 313.75 4236 13470 54030 647.2 75600 3448 584.3 312.9 Standard deviation 746.96 743.69 6090.53 6111.42 196.54 196.94 23253.7 23345.07 271.27 271.49 114.70 115.10 41.21 41.29 Mean 12893.91 12893.91 62335.55 62335.55 915.70 915.70 44360.7 44360.70 4005.61 4005.61 706.08 706.08 340.42 340.42

147

Appendix B

Fig. B1: Heatmap showing the ΔGSDraw of all samples from the Düsseldorf-Grafenberg sequence.

148

Fig. B2: Heatmap showing the ΔGSDraw of all samples from the Suohuoduo section.

149