I thank Magali Delmas and one anonymous reviewer for their constructive comments on this review article, and the editorial team for their efficient work.
1Quaternary geochronology, fluvial (palaeo-)environments/archives, relative and numerical dating methods, fluvial chronologies, cosmogenic nuclides, surface exposure dating, weathering indicators, multi-methods approaches
2Quaternary geochronology is the research field committed to generating chronologies of Earth’s history over the last 2.6 Ma. This definition is slightly modified from that of Noller et al., (2000), who, twenty years ago, considered the last 1.8 Ma accordingly. The early stages of this discipline occurred at the onset of the xxth century: varve- and dendrochronology saw the day of light more than one century ago thanks to the pioneering works of De Geer (1912) and Douglass (1919), respectively. As stated by Walker (2005, p. 4): “Varve chronology was the first dating technique to provide a realistic estimate of Quaternary time”. From that moment on, this discipline has experienced spectacular progress. This is obviously attested by the ever-growing number of dating methods that have been developed over one century, including of course the development of radiocarbon (14C) dating a little more than 70 years ago (Libby et al., 1949). This evolution was well expressed by Noller et al. (2000, p. 2) for whom Quaternary geochronology “(…) can be seen as a collection of strategies or approaches that apply one or more methods in the quest to estimate the age of a material and/or landform. (…) No one approach or method can provide reliable age estimates in all contexts; thus the researcher must be familiar with as many methods as possible”.
3Importantly, these strategies and approaches are underlain by very diverse founding principles (fig. 1). They include, among others, the annual increment of clastic material (e.g., varve chronology) or organic tissue (e.g., dendrochronology), the operation of bio-chemical processes (e.g., amino-acid racemization or diagenesis) or the radioactive decay of unstable isotopes (e.g., 14C). They accordingly require very different analytical efforts (Noller et al., 2000), ranging from straightforward field measurements (e.g., lichenometry) to highly sophisticated laboratory devices such as accelerator mass spectrometry (e.g., in situ cosmogenic nuclides). Moreover, their applicability in the past covers quite different periods of time (fig. 1), ranging from a few decades (e.g., Caesium-137) to the whole Quaternary and way beyond it (e.g., palaeomagnetism). Interestingly, although authors tend to agree upon to what extent a specific method can be used, some divergences subsist in terms of chronological applicability (fig. 1). Time ranges of well-established Quaternary dating methods, such as U-series dating (Duval et al., 2020), are not set in stone among the different authors (Walker, 2005; Jull, 2007).
4A key subject in geomorphology, and more generally in all disciplines dealing with the shaping of the Earth surface (“surface geosciences”), addresses the study of processes and rates of landscape evolution (e.g., Anderson & Anderson, 2010). Noller et al.’s (2000) formulation above fits in particularly well with geomorphology’s aim of understanding Earth surface processes and reconstructing rates of landscape evolution. Owing to its ability of providing the element of time, that is assessing the age of geomorphological markers (see e.g., chapters 2 and 3 of Burbank & Anderson, 2012), Quaternary geochronology is no less than crucial and unavoidable. This was already acknowledged more than 120 years ago: Davis’s ground-breaking concept of cycle of erosion set the notion of time at the very heart of the discipline (Davis, 1899). It has remained there since then, as illustrated by Schumm & Lichty’s (1965) seminal paper conceptually discussing the respective role of time, space and causality in geomorphology about 55 years ago. The same holds for subsequent well-known concepts and theories that have been developed in fluvial geomorphology, including, for instance, the fluvial system of Schumm (1977) or sediment budgeting (Dietrich and Dunne, 1978). More recently, in the wake of physically-based, numerical landscape evolution models developed in the late nineties such as CASCADE (Braun & Sambridge, 1997), the number of newly developed landscape evolution models spectacularly grows each year. The two last decades have thus witnessed the establishment of this prominent research field in geomorphology and, for almost each of them, time plays a central role.
Fig. 1: Selection of relative, correlation and numerical dating methods spanning the Quaternary.
Although their time range of application according to two different authors is globally concordant, some significant discrepancies are reported.
Fig. 1 : Sélection de méthodes de datation du Quaternaire (relatives, corrélatives, numériques). Bien que les périodes de temps pour l’applicabilité de ces méthodes convergent globalement entre les deux auteurs, certaines divergences importantes sont observées.
5Quantitative assessment of processes has in fact become a central task in fluvial geomorphology, either on (very) short (e.g., modern bedload transport rates: 101 yrs), intermediate (e.g., rates of overbank sedimentation/floodplain formation: 102-3 yrs) or (very) long (e.g., fluvial incision rates along valleys: 104-6 yrs; fig. 2) timescales. Dating landforms and associated fluvial material forming floodplains and/or terraces sequences has represented a persistent scientific challenge (fig. 2; Rixhon et al., 2017). Some of the main advances in fluvial geomorphology were therefore closely intertwined with methodological developments in Quaternary geochronology. First applications to date organic material embedded in fluvial sediments by radiocarbon (e.g., Fairbridge, 1962) or to date alluvium itself by optically-stimulated luminescence (e.g., Perkins & Rhodes, 1994) epitomise this well. Importantly, these developments allowed gaining valuable new insights into the drivers of landscape evolution: “Continuing advances in Quaternary geochronology have enhanced our ability to determine when events occurred, and thus to understand interactions between landscapes and potential controlling actors such as climate [, tectonics] and land use” (Wohl, 2013, p. xxix; foreword).
Fig. 2: Terrace staircase of the Lower Meuse Valley (Belgium-Netherlands border) with available chronological information (correlation and numerical age estimates). Modified from Bridgland & Westaway, 2014.
Fig. 2 : Séquence de terrasses alluviales de la basse vallée mosane (à la frontière belgo-néerlandaise) avec le cadre chronologique existant (datations par corrélation et numériques). Modifié de Bridgland & Westaway, 2014.
6This contribution starts by briefly overviewing the most usual Quaternary dating methods, without being exhaustive. Comprehensive works detailing these methods can be found in the volume of papers edited by the AGU dedicated to Quaternary Geochronology (Noller et al., 2000) or the book of Walker (2005). The first goal of this contribution is to propose an historical contextualisation of how Quaternary geochronology allowed the production of fluvial chronologies over the last century. Whereas age controls on fluvial landforms and deposits were mostly achieved via relative and correlation dating until the late xxth century, methods providing numerical age estimates have progressively taken over the prevailing role over the last two or three decades (Rixhon et al., 2017). Two main explanatory causes are then tackled to tentatively analyse this gradual shift in the use of dating methods. The bipartite evolution of relative dating methods over the last decades is explored: whilst some methods were further developed in the fluvial context (e.g., amino-acid racemization), some others were vehemently questioned (e.g., lichenometry). The late flourishing of radiometric dating methods during the second half of the xxth century is also considered. The second goal is to propose a contemporary reflection on Quaternary dating methods by asking two provocative questions:
7(i) Is the die cast for “disused” relative dating methods?
8(ii) Can radiometric dating methods solve all chronological issues related to fluvial environments?
9To answer the second question, surface exposure dating based on cosmogenic nuclides measured in alluvial surfaces is then thoroughly discussed, considering both its main strengths and weaknesses. It is suggested here to rehabilitate some methods fallen into disuse, such as weathering indicators, to support numerical dating methods, such as surface exposure dating. The final part of this contribution aims at promoting the use of geochronological approaches based on multiple dating tools, either by crosschecking the outputs of different methods or coupling numerical dating methods.
10Two main types of classification were proposed to inventory the wide array of Quaternary dating methods that have been developed over the last century (Noller et al., 2000). They are either based on the type of method itself or on the type of results that these methods yield (tab. 1). Whereas the former contains six types of methods, which are sidereal, isotopic, radiogenic, biologic/chemical, geomorphologic and correlative, the latter recognizes four “classes” of age production: relative, correlated, calibrated or numerical (tab. 1). Here, the second classification is followed as it opportunely describes the level of information as well as the degree of confidence that these methods yield (Noller et al., 2000). The methods hereinafter will be presented accordingly. Whilst relative and correlation methods are presented alongside, numerical methods are considered separately.
Tab. 1: Two classifications of Quaternary dating methods according to the kind of the method itself or the kind of results that these methods generate. Modified from Noller et al., 2000.
Tab. 1 : Deux classifications des méthodes de datation du Quaternaire, selon le type de méthodes ou le type de résultats générés. Modifié de Noller et al., 2000.
11The founding principles of relative and correlation methods are not quite identical and are thus reminded here. Relative dating methods can be considered as a derivative of stratigraphy and the law of superposition in geology dating back to Steno and Lyell: they determine whether an object (or feature) is younger or older than a similar object (Walker, 2005). Those objects or features, which can be fossils, artefacts or sedimentary layers/units/horizons, are thus ranked in relative order of age. In other terms, an ordinal measurement or age sequence is provided by relative dating methods and some of them can yield an assessment of the magnitude of age differences between those objects/features in the sequence (Noller et al., 2000). Interestingly, although stratigraphy remains the very first principle to produce relative chronologies in the Quaternary, other processes or phenomena can be relatively classified when one purposely studies Quaternary environments. “Many of these involve the operation of physical or chemical processes that are wholly or partially time-dependent (…): sub-aerial weathering of rock surfaces and pedogenesis will bring about gradual changes on rock and ground surfaces (…) (Walker, 2005, p. 165-166). Relative dating methods used in the Quaternary are based on biological, chemical and geomorphological processes and comprise, among others, amino-acid racemization, lichenometry, mineral and/or rock weathering (including obsidian hydration), Schmidt hammer and soil development/soil chronosequence (tab. 1). Importantly, the calibration, or “tuning” some might say, of relative dating methods by independent age information enables the production of calibrated ages using a ratio scale (Noller et al., 2000). When calibrated, relative dating methods thus can yield approximate numerical ages (tab. 1). This facet of these methods will be tackled in section 4.1.2.
12By contrast, correlation methods mostly are nominal-scale methods. Instead of directly measuring ages, they produce ages by evidencing an equivalence using time-independent properties (Noller et al., 2000), hence their designation under “techniques for establishing age equivalence” by Walker (2005). In other terms, “to obtain an age estimate, a geologic unit is correlated, using a variety of properties, to another independently dated geologic unit (Noller et al., 2000, p. 8)”. Many of these methods are regularly used for “long” geological timescales, well beyond the Quaternary timespan such as palaeomagnetism or tephrochronology (fig. 1). However, some of them have a peculiar relevance for Quaternary studies, such as archaeology or climatic correlations/oxygen isotope stratigraphy (tab. 1).
13Before going any further, an important semantic remark must be done. Numerical dating methods, in the broadest sense, are those methods enabling numerical age estimates to be produced (Walker, 2005). In other terms, they yield a quantitative assessment not only of “the age” but also of associated uncertainties, thereby allowing a degree of confidence to be produced (Noller et al., 2000). An alternative designation in the scientific literature under “absolute dating methods and absolute ages or chronologies” is regularly encountered (e.g., Broecker, 1963; Johnson & Knoll, 1975; Partridge et al., 1984; Zolitschka, 1991; Andriessen et al., 1993; Watchman & Twidale, 2002; Haeberli et al., 2003; Cossart et al., 2010; Tassy et al., 2013; Ruiz and Rowe, 2014; Rosenwinkel et al., 2015). However, as claimed by Walker (2005; p. 15), “there is (…) nothing ‘absolute’ about a date, and it should not be referred to as such ” because “[this term] implies a level of accuracy and precision that can seldom, if ever, be achieved in reality”. Examining this same question, Noller et al. (2000, p. 5) provided more details: “Absolute commonly has been used to describe the results of isotopic dating methods, but variation in estimates of analytical precision, decay constants (…), or half-lives (…) invalidates the “absoluteness” of the age estimates derived from these methods. In addition, undetected contamination and geologic uncertainties, such as the delay between a geologic event and the “time zero” used by a dating method, commonly render isotopic ages (indeed, all age estimates) less than absolute”. This designation is thus purposely avoided here; instead “numerical dating methods” and “numerical age estimates” will be used hereinafter.
14Among the wide breadth of numerical dating methods, a primary subdivision into two main categories owing to the core principle enabling the estimation of numerical ages can be achieved. This principle is either an annual increment or a radiogenic/isotopic process (Walker, 2005). The first group of methods takes advantage of the yearly addition of material to:
15(i) sedimentary sequences, either in lakes/sea: varve chronology, or karstic environments: annual banding in speleothems;
16(ii) ice accumulation: annual layering in glacier ice;
17(iii) organic tissue, either the annual tree-ring growth: dendrochronology, or annual growth bands in molluscs and corals in marine environments: sclerochronology.
18These methods are also referred to as sidereal methods (tab. 1).
19The second group of methods can be referred to as radiometric dating sensu lato (Walker, 2005) and they will be named “radiometric dating methods” hereinafter. However, different designations can be found in the literature: Noller et al. (2000) distinguish the isotopic methods from the radiogenic methods (tab. 1). As for the first ones, age production relies on changes in isotopic composition as a result of radioactive decay of unstable isotopes and/or growth. In Quaternary studies, the most “emblematic” of these methods is radiocarbon (14C) dating. Further methods are either based on long-lived isotopes such as Argon-isotope dating (K-Ar or 39Ar-40Ar), Uranium-series methods (e.g., 230Th/U) and Uranium-Lead (U-Pb), and cosmogenic nuclide dating (e.g., 10Be), or on short-lived isotopes such as Lead-201 and Caesium-137. Radiogenic methods, also referred to as radiation exposure dating (Walker, 2005), assess the cumulative effects of radioactive decay on the crystal structure of certain minerals and electron energy traps (Noller et al., 2000). They include thermo-, optically- and infra-red stimulated luminescence (TL, OSL and IRSL respectively), electron spin resonance (ESR) and fission-track dating (tab. 1). At this stage, narrowing down the scope of this contribution appears necessary. Despite their undoubted usefulness, numerical dating methods based on annual increments are not considered hereinafter, mostly because they are slightly off topic, except perhaps dendrochronology, when it comes to date fluvial archives or landforms.
20Establishing chronologies for Quaternary fluvial (palaeo-)environments has long been achieved via diverse relative and correlation dating methods for most of the xxth century (see the thorough review of Bridgland & Maddy, 2002). They allowed the production of chronological frameworks for a wide array of fluvial records or landforms. This diversity is reflected in three main aspects:
21(i) the size of the investigated river, valley or catchment, ranging from small systems such as mountainous torrential streams (e.g., Crook, 1986) to large rivers such as the Rhine (e.g., Van den Boogard et al., 1989);
22(ii) the time period of interest, ranging from short-lasting climatic events such as the Little Ice Age (e.g., Jacob et al., 2002) to the whole Quaternary timespan and even beyond (e.g., Van den Berg and Van Hoof, 2001);
23(iii) the nature of the investigated fluvial record and/or landforms.
24Note that, regarding the second point, application time spans of relative and correlation dating methods largely differ too, ranging from modern times (i.e., historical data) to geological timescales (i.e., palaeomagnetism). As for the third point, whereas well-preserved terrace sequences were by far the most commonly studied geomorphological feature/sedimentological record (fig. 2; e.g., Caillier et al., 1986; Macaire, 1986; Knuepfer, 1988; Adams et al., 1992; Bates, 1993), outwash or fluvio-glacial deposits (e.g., Nelson & Shroba, 1998) as well as alluvial fan surfaces were investigated as well (e.g., McFadden et al., 1989).
25Although they are not comprehensively listed given the manifoldness of applications and case studies, relative and correlation dating methods applied to Quaternary fluvial archives include (fig. 3):
26- aminostratigraphy in terrestrian molluscs (e.g., Seine, Somme and tributaries; Bates, 1993);
27- archaeology, through the finding of diverse objects such as artifacts, ceramics, potteries (e.g., Toth, 2008);
28- clast seismic velocity (e.g., San Gabriel canyon, California; Crook, 1986);
29- climatic correlation to glacial chronologies, for instance that of the Alps in northern Europe (e.g., Rhine, Germany; Brunnacker et al., 1982);
30- correlation to marine isotopic stages via mammal (e.g., British rivers; Schreve, 2001) or mollusc (e.g., Thames; Preece, 1999) biostratigraphy;
31- historical data and artifacts (e.g., see review of Trimble, 1998);
32- lichenometry (e.g., Corsican river; Gob et al., 2003);
33- obsidian hydration (e.g., Madison River, Yellowstone-USA; Adams et al., 1992);
34- palaeomagnetism (e.g., Meuse, The Netherlands; Van den Berg & Van Hoof, 2001);
35- Schmidt hammer (e.g., alluvial fan system in Tunisia ; White et al., 1998) ;
36- soil profile development and soil chronosequence (e.g. Moselle, France ; Caillier et al., 1986) ;
37- tephrochronology (e.g., Rhine, Germany ; Van den Boogard et al., 1989) ;
38- weathering rind analysis of fluvial clasts (e.g., Loire and tributaries; Macaire, 1986, fig. 2.5c).
Fig. 3: Array of relative dating methods based on surficial/shallow weathering processes.
Among the variety of weathering parameters that can be assessed at the surface of a clast in fluvial deposits, rind thickness and fracture width (bold) will be discussed in detail in this contribution. Modified from Birkeland & Noller, 2000.
Fig. 3 : Méthodes de datation relative basées sur des processus/paramètres d’altération en (sub-)surface. Parmi la variété de paramètres d’altération qui peuvent être estimés à la surface d’un galet ou bloc dans des dépôts fluviaux, l’épaisseur du cortex d’altération et la largeur des fractures seront discutées en détail dans cet article. Modifié de Birkeland & Noller, 2000.
39Importantly, one must bear in mind that these methods are not incompatible and that their advantageous combination allowed producing useful relative chronologies. This was achieved by Knuepfer (1988) who reliably assessed the age of late Quaternary terrace sequences in New-Zealand through a mix of weathering rind analysis and the study of soil chronosequence. In addition, studying a sequence of alluvial fans in an arid environment (California), McFadden et al. (1989) purposely designed an interesting multi-parameter approach for age estimation. It involves a set of easily measurable rock-weathering ratios (i.e., fracture, grain relief, pit, roundness, rind, ring, rubification; fig. 3) and varnish cover/ratio. Note finally that this kind of geochronological approaches based on a combination of relative and/or correlative dating methods was not restricted to fluvial settings. They were also applied to coastal (e.g., marine terraces: Muhs, 2000), glacial (e.g., moraine complexes; Shiraiwa & Watanabe, 1991) and periglacial environments (e.g., rock glaciers; Haeberli et al., 2003) as well as slope deposits (e.g., debris flow; Boelhouwers et al., 1999).
40Over the last two decades, radiometric dating methods sensu lato have spectacularly progressed owing to their very attractive potential of delivering numerical age estimates and have progressively supplemented the use of relative and correlation dating methods (see section 4.2). Dating methods yielding numerical age estimates applied to fluvial archives or landforms include (fig. 4):
41- dendrochronology (e.g., Siret and Moldova Rivers, Romania; Rădoane et al., 2015);
42- electron spin resonance (ESR) dating (e.g., Moselle River; Harmand et al., 2015);
43- fission-track dating (e.g., Rhône and Rhine; Bernet et al., 2004);
44- luminescence, either via thermoluminescence dating (e.g., Alligator River, Australia; Murray et al., 1992) or optically stimulated luminescence (OSL) dating (e.g., Rhine-Meuse, the Netherlands; Wallinga et al., 2001);
45- Potassium/Argon (K/Ar) dating (e.g., Gediz River, Turkey; Westaway et al., 2006) and Argon-Argon (Ar/Ar) dating (e.g., Tigris River, Turkey; Bridgland et al., 2007);
46- radiocarbon (14C) dating (e.g., Nile; Fairbridge, 1962);
47- short-lived isotopes (137Cs and 210Pb) dating (e.g., Geul catchment, Belgium/ Netherlands; Stam, 1999);
48- terrestrial or in situ-produced cosmogenic nuclides (CN), either via surface exposure dating or burial dating (e.g., Meuse catchment, Belgium; Rixhon et al., 2011, 2020 for exposure and burial dating, respectively);
49- Uranium-series, either via 230Th/U dating (Wind River, Wyoming; Sharp et al., 2003) or U-Pb dating (e.g., Qaidam basin, Tibet; Song et al., 2019);
Fig. 4: Sketch highlighting the wide-ranging dating applications of six first-order radiometric methods in fluvial settings, i.e. 14C, OSL-IRSL, ESR, 230Th/U, CN and K/Ar or Ar/Ar.
It depicts (i) the dateable landforms and correlated deposits, (ii) the pathways of dateable material for these six methods and (iii) the transport pathways and provisional storage of both organic material (upper right corner) and coarse- and fine-grained clastic materials on hillslopes and the fluvial system itself. Modified from Rixhon et al., 2017.
Fig. 4 : Croquis synthétisant la grande diversité de formes et archives alluviales ciblées pour une datation numérique via six méthodes radiométriques couramment utilisées (radiocarbone, luminescence, résonance de spin électronique, série de l’Uranium, nucléides cosmogéniques et Potassium ou Argon/Argon). Il montre (i) les morphologies et dépôts associés datables, (ii) les trajectoires du matériel datable pour ces six méthodes et (iii) les transferts ainsi que les stockage temporaire de matière organique (coin supérieur droit) et des matériaux clastiques fins et grossiers sur les versants et dans le système fluvial. Modifié de Rixhon et al., 2017.
50Importantly, among the diverse geomorphological settings, fluvial (palaeo-)environments represented a primary repository for new dating applications via these methods (Rixhon et al., 2017). This is true regardless the size of the investigated system, ranging from small headwaters (e.g., Bartz et al., 2015, 2017) to the largest rivers (e.g, Fairbridge, 1962). Numerical dating methods were successfully applied to a very wide array of fluvial landforms and correlated deposits (fig. 4), thereby providing age controls from which various rates of landscape evolution could be inferred. In floodplain environments for instance, the most common use consists in dating overbank fine-grained deposition via radiocarbon and/or luminescence (e.g., Preusser, 1999; Hormes et al., 2003). This allows reconstructing mean accretion or aggradation rates and gaining insights into the drivers of sedimentation (e.g., Zielhofer et al., 2008). Sedimentary bodies forming specific floodplain landforms such as natural levees (e.g., Pierik et al., 2017), point bars/scroll bars (e.g., Quik & Wallinga, 2018) and palaeochannels/oxbow lakes (e.g., Rowland et al., 2005) were successfully targeted as well (fig. 4). Radiogenic or palaeodosimetric methods (OSL/IRSL, ESR) along with isotopic methods using cosmogenic nuclides enabled the numerical dating of more ancient fluvial landforms or sequence of landforms, i.e., alluvium-mantled terraces and terrace staircases in particular (fig. 4; Rixhon et al., 2017).
51A remark on the list above should be made at this stage. Fission-track dating in Quaternary studies was most commonly applied to volcanic materials, in particular obsidian and tephras (Walker, 2005). Moreover, when applied to detrital apatite or zircon minerals collected from river sediments, fission-track dating is rather used to determine changes in sediment provenance (e.g., Bernet et al., 2004) or to reconstruct sediment budget (e.g., Resentini & Malusà, 2012). In spite of their usefulness in many fields of the geosciences, these applications are beyond the scope of this contribution: we refer instead to the authoritative book of Braun et al. (2006) for a comprehensive review on that topic.
52Broadly speaking, a bipartite evolution regarding the aforementioned relative and correlation dating methods has been witnessed over the last three decades.
53On the one hand, some of the relative/correlation dating methods listed above (see section 3.1) have continued to evolve owing to significant methodological improvements that have been accomplished since then. Aminostratigraphy of fluvial deposits, mostly based on terrestrial molluscs, epitomises this well. This is demonstrated by a number of studies carried out in terrace sequences, such as the ostracod-based aminostratigraphy developed for central Spain (Torres et al., 2005). The most significant results, however, were obtained on terrace sequences in Great Britain (fig. 5a). Firstly, beyond the use of a single amino acid in freshwater shells (i.e., isoleucine and its inter-conversion to alloisoleucine), Penkman et al. (2007) analysed the intra-crystalline fraction behaving as a closed system, thereby providing a greater potential for enhanced temporal resolution. Secondly, the aminostratigraphy proposed by Penkman et al. (2011), based on the analysis of multiple amino acids contained in the opercula of a specific gastropod (Bithynia), provided a remarkable chronological framework for Quaternary continental deposits of Britain, including several emblematic terrace sequences in England (e.g., Avon/Severn, Thames) reaching back to MIS 11 (fig. 5a).
Fig. 5: Relative dating of terrace sequences via amino-acid racemization (top) and weathering rind analysis (bottom).
a - Cross section through the terrace sequence of the Lower Thames (left) and plot of THAA vs FAA D/L Ala (right); Ala, FAA and THAA refer to alanine, the free amino acid and the total hydrolysable amino acid fraction, respectively. Older, higher terraces show more degraded proteins within the shell’s opercula; this allows allocating the successive terrace deposits to specific MIS. Modified from Penkman et al., 2007; b - Sequence of Quaternary fill terraces (Qt1-3) preserved along the Pacific coast (Costa Rica): the higher the degree of alteration, assessed here via the mean rind thickness, the older the terrace. Modified from Sak et al., 2004.
Fig. 5 : Datation relative de séquences de terrasses via la racémisation d’acides aminés (dessus) et l’épaisseur du cortex d’altération (dessous). a - Coupe transversale dans la séquence de terrasses de la basse vallée de la Tamise (gauche) et graphe THAA vs FAA D/L Ala (droite) où Ala, FAA and THAA désignent respectivement l’alanine, les acides aminés libres et la fraction totale hydrolysable d’acides aminés. Les niveaux de terrasses plus élevés -donc plus ancients - présentent des protéines plus dégradées au sein des opercules de mollusques, ceci permettant d’attribuer les niveaux successifs à des stades isotopiques marins spécifiques. Modifié de Penkman et al., 2007 ; b - Séquence de terrasses quaternaires (Qt1-3) préservées le long de la côte pacifique du Costa Rica : plus le degré d’altération est élevé - estimé ici d’après l’épaisseur moyenne du cortex d’altération - plus la terrasse est ancienne. Modifié de Sak et al., 2004.
54On the other hand, the reliability of the key assumptions lying at the heart of other relative and/or correlation dating methods was either critically assessed (e.g., soil chronosequences: Huggett, 1998; Birkeland, 1999) or vehemently questioned. This is especially true for dating methods based on surficial processes from which calibrated ages were obtained (Ridings, 1996; Anovitz et al., 1999; Osborn et al., 2015). According to Wayne (1984), note that these methods not only include morphological changes and variations caused by weathering (e.g., clast weathering rind analysis, obsidian hydration, soil profile development) but also the growth of plants (e.g., lichenometry). The first key assumption for using surficial processes as a geochronologic tool is that the object that has to be dated starts to alter as soon as it becomes exposed (Wayne, 1984), i.e., “zeroing the clock”. This implies that, in a chronostratigraphic sequence of e.g., fluvial deposits (typically a terrace flight), the oldest deposit will be affected by the highest degree of alteration or change, and gradually younger deposits will exhibit lesser degrees of change (fig. 5b). Time therefore is the prevailing factor from the onset of exposure to the moment of observation or sampling. The second key assumption is that the physical or (bio-)chemical process must continuously operate over time. This also implies that if any change in the rate at which it operates should occur, e.g., the growth rate of lichens over time, this variation must be properly considered.
55Representative issues inherent to relative dating methods based on surficial processes, which were regularly used in fluvial settings, are tackled here. A significant body of scientific literature, either reporting unsuccessful dating applications (e.g., Ridings, 1996) or critical methodological reviews (e.g., Anovitz et al., 1999; Osborn et al., 2015), showed that one of these assumptions, or both, can be regularly violated, thereby casting doubt on the usefulness of these methods in Quaternary studies. This is highlighted here by two debated dating methods: WR analysis (e.g., Veldkamp et al., 1990) and lichenometry (e.g., Osborn et al., 2015).
56Weathering rind (WR) analysis based on rind thickness was most commonly used to infer glacial chronologies (e.g., Porter, 1975) but was also employed to investigate alluvial bodies (e.g., Macaire, 1986) or, interestingly, to correlate moraines with fluvial terraces (e.g., Pinter et al., 1994). This is for instance well exemplified by WR analysis of basaltic clasts included in three consecutive Quaternary fill terraces preserved along the Pacific coast (Costa Rica; Sak et al., 2004). The increasing values of mean rind thickness observed between the youngest (~0.9 cm) and the oldest (~6.9 cm) terraces allowed inferring a chronostratigraphic framework for this sequence (fig. 5b). These observations are well in line with the first assumption for this kind of geochronometers: the degree of alteration steadily increases from the youngest to the oldest fluvial deposit in the terrace flight.
57However, beyond the fact that rock weathering obviously is dependent on lithology, serious complications can affect WR analysis when it is used as a geochronological tool. Three of them are briefly tackled here. Firstly, as for the spatial variability in surficial weathering processes at the local scale, Thorn (1975) identified the important role played by local factors such as late-lying snow. This results in substantial variations of rind thickness, i.e., the key parameter usually measured for this method (fig. 3), in the same glacial landform which, in turn, lead to potential significant bias in relative age dating. In his case study, Thorn (1975, p. 373) stated that, “Differences of the magnitude [occurring in rock-weathering rind thickness linked to the presence of late-lying snow] are comparable to those used to distinguish between glaciations. It is recommended that rock-weathering rind thickness be identified as a very sensitive indicator of environment, or paleoenvironment”. Secondly, according to Gordon & Dorn’s review (2005) dedicated to in situ erosion of weathering rind, this process was only marginally considered in previous studies using rind thickness as a geochronological tool. Considering the usual field technique to measure the rind thickness, which originally consists in hitting and breaking the clast with a rock hammer (Colman & Pierce, 1981), the loss of matter, at least in the weakened, outer part of the rind, is highly plausible. This can obviously result in a significant bias for chronological assessment if this process is not properly taken into account. Thirdly, an issue related to lithology was reported by a case study using WR analysis of alkali basalt gravels included in three Quaternary terrace deposits of various ages along the Allier Valley in Limagne (France; Veldkamp et al., 1990). Unlike the previous study on the terrace sequence in Costa Rica (Sak et al., 2005), neither the terrace age (at least for the two oldest terraces) nor the sampling depth could explain the variation in weathering intensity. Instead, it likely results from the variable chemical composition of the alkali basalt pebbles (Veldkamp et al., 1990). The key factor controlling basalt weathering in this case is not time but mineralogy.
58The discussion engendered by lichenometric dating among Quaternary scientists, especially geochronologists, perhaps best synthesises most of the fundamental critics addressed against these methods. Both aforementioned key assumptions primarily underlie the use of the largest lichens settled on a deposit as a geochronological tool (fig. 6a,b). The method doubly assumes that the lichen’s colonisation begins soon after deposition and that the lichen growth has continuously occurred in the time interval comprised between settlement and sampling (McCarthy, 2007; Osborn et al., 2015; fig. 6a). Since its first dating application to moraine deposits about 70 years ago (Beschel, 1950), lichenometric dating has become a much used and widely accepted method in a broad range of lichen-rich, polar and alpine regions (e.g., Matthews & McCarroll, 1994; fig. 6a). Along with its applications in palaeosismology (e.g., Bull, 1996), it also has spread to other geomorphological environments to assess the timing of coastal (e.g., raised beach), hillslope (e.g., debris flows, rockfall deposits) or fluvial features (e.g., bedrock channels; Jacob et al., 2002).
Fig. 6: Illustration of two main issues related to lichenometry (Rhizocarpon geographicum agg).
a - Thallus monitored on a moraine ridge in British Columbia (Canada): after a healthy and growing phase (1996-2002), a quick dieback (2007) followed by a complete disappearance (2012) subsequently took place. At the same site, a recovery and a regrowth of other thalli occurred after dieback, showing a poor understanding of lichen’s mortality rates. Modified from Osborn et al., 2015; b - Growth curves reconstructed for two different settings: rock glacier deposits (Tian Shan Mountains) and high-energy river (Corsica), modified from Gob et al., 2003 (Corsica) and Rosenwinkel et al., 2015 (Tian Shan). The vast majority of measured lichens are smaller than 4 cm in diameter and are younger than 200 years; the few older individuals result in different curve’s shapes for the same lichen. This highlights variable growth rates as a function of the geographic location but also questions the robustness of lichenometric dating at the same time.
Fig. 6 : Illustration de deux difficultés majeures en lien avec la lichenométrie (Rhizocarpon geographicum agg). a - Suivi temporel d’un thalle sur une crête morainique en Colombie-Britannique (Canada) : après une phase de croissance (1996-2002), un dépérissement rapide (2007) suivi d’une disparition totale (2012) se produisirent. D’autres thalles sur le même site, après avoir été affectés par un dépérissement similaire, ont pourtant connu une nouvelle phase de croissance, démontrant une connaissance insuffisante des taux de mortalité de lichens. Modifié de Osborn et al., 2015 ; b - Courbes de croissance reconstituées pour deux environnements différents : dépôts de glaciers rocheux (Massif du Tian Shan) et rivières de haute-énergie (Corse). Dans la mesure où la grande majorité des lichens mesurés ont un diamètre inférieur à 4 cm et sont donc jeunes (<200 ans), les rares individus plus âgés influent largement sur les formes nettement différentes des courbes de croissance du même lichen. Outre le fait que le taux de croissance soit dépendant du contexte géographique, ceci remet en cause (dans une certaine mesure) la fiabilité de la lichénométrie. Modifié de Gob et al., 2003 (Corse) et Rosenwinkel et al., 2015 (Tian Shan).
59The lack of substantial methodological progress, however, was emphasised by McCarthy (2007). He argued that the way practitioners employ lichenometric dating has very little evolved over the past few decades. McCarthy (2007, p. 1399) critically added that: “During this time, few papers have offered constructive criticism of lichenometry and its questionable biological foundation”. Osborn et al.’s (2015) very provocative review goes one step further by clearly casting doubt on the method’s soundness by factually discussing a series of methodological pitfalls. Regularly invalidating one or both key assumption(s), these pitfalls can be classified into two main categories. The first one deals with fundamental ecological considerations, including lichen’s mortality rates and the usually problematic but essential reconstruction of growth curves (fig. 6a,b). As for the latter, no less than six specific issues are listed, such as the poor understanding of their shape which entails an initial “great exponential growth” followed by a “slow linear growth” (fig. 6b). The second concerns the problematic lack of agreement on practice, including a missing consensus on the object that should be actually measured in the field (e.g., exclusion of coalescing thalli?) as well as disputed sampling strategies. As for the latter, dissimilar and disputed approaches were proposed to infer the substrate age: sampling of the single largest lichen or of five to ten lichens versus a size-frequency approach entailing the measurements of hundreds to thousand lichens. This lack of agreement also includes divergences on how the data and errors should be handled. Following on from this, Osborn et al. (2015, p. 10) state that “Lichenometry rests on several assumptions, rarely articulated, that are unverified or patently incorrect” (p. 4), and harshly conclude that “it only appears to be a science”.
60A brief historical perspective helps contextualising the slow-starting development of radiometric dating methods before they eventually took over the main role in Quaternary geochronology in the late xxth century and the beginning of the xxist century. Firstly, key findings in physics and chemistry from the end of the xixth to the first half of the xxth century laid the theoretical foundations for many of these radiometric dating methods (fig. 7a-c). Some of the most significant ones are chronologically presented below:
61- the physical phenomenon known as luminescence was firstly referred to as such by Q.C. Lum in 1888;
62- radioactive decay was discovered by H. Becquerel in 1896, closely followed by the discovery of radium by M. Curie two years later;
63- cosmic rays were discovered by V. Hess in 1912 through in situ measurements of atmospheric radiation during his balloon ascents (fig. 7a);
64- W. Pauli with others physicists in the 1920’s demonstrated the existence of electron spin angular momentum;
65- the isotope Carbon-14 was discovered in 1940 by M. Kamen and S. Ruben.
Fig. 7: Three “big names” of Quaternary geochronology:
Victor Franz Hess (left), Willard Frank Libby (middle) and Martin Jim Aitken (right) due to their essential (direct or indirect) contributions in cosmogenic nuclide, radiocarbon and luminescence dating, respectively.
Fig. 7 : Trois “grands noms” de la géochronologie du Quaternaire : Victor Franz Hess (gauche), Willard Frank Libby (centre) and Martin Jim Aitken (droite) grâce à leurs contributions fondamentales (directes ou indirectes) pour, respectivement, la datation par nucléides cosmogéniques, radiocarbone et luminescence. Sources : http://physik.uibk.ac.at/hephy/Hess/fig3.gif (left/gauche) ; https://www.nationalgeographic.fr/ sciences201907la-datation-au-carbone-est-un-outil-utile-mais-imparfait (centre) ; https://www.sciencephoto.com/media/75357/view/martin-aitken-british-physicist (right/droite)
66Secondly, “the most significant advance in Quaternary chronology [i.e., the development of radiometric dating methods] came during and immediately after the Second World War, with the discovery that the decay of certain radioactive elements could form a basis for dating (Walker, 2005, p. 5).” Originally outlined in 1946, Willard Libby’s ground-breaking idea to employ 14C as a geochronometer materialises this (fig. 7b): radiocarbon dating was made a reality three years later (Libby et al., 1949). Since that time, and even more since the foundation of the eponymous scientific journal Radiocarbon in 1959, it is not exaggeration to say that this dating method has revolutionised several scientific disciplines, including archaeology, geomorphology and Quaternary science (e.g., Taylor, 1995; Wood, 2015). Despite its undisputed usefulness, the applicability time span of 14C dating, however, covers the last ~50 ka at the most, owing to the nuclide’s half-life (5730 years) and inherent measurement limitations, even using the later accelerator mass spectrometry (AMS). Further efforts were thus needed to develop other radiometric dating methods which allow going further back in time.
67In contrast with radiocarbon dating, which somehow represents an exception in the rapid use of theoretical physical knowledge to set up a geochronological tool, this knowledge transfer for other radiometric dating methods took place with significant delays. From the previous list in section 3.2, the two next methods to have been developed were U-series and Potassium (K)-Argon (Ar). They both largely benefited from a growing knowledge on nuclear decay processes along with improving analytical instrumentation during the 1960s and 1970s. This resulted, among others, in the development of one of the most widely used dating methods of the Quaternary: the 230Th/U-dating method (Broecker, 1963). As for electron spin resonance (ESR), luminescence and cosmogenic nuclides (CN), all routinely used nowadays, the delays were even longer:
68(i) The first successful application of ESR as a dating method on speleothem (Ikeya, 1975) largely postdates ESR spectroscopy which was already developed about forty years earlier (Gorter, 1936).
69(ii) While thermoluminescence was originally turned into a successful geochronological tool in the 1960’s (fig. 7c; Aitken et al., 1964), optically stimulated luminescence dating of sediments “saw the light of day” about 20 years later with the seminal paper of Huntley et al. (1985). It has become, however, a routinely applied method only since the development of the single aliquot regenerative dose (SAR) protocol developed by Murray & Wintle (2000).
70(iii) Despite the discovery of cosmic rays in 1912 and the pioneering work of Davis & Schaeffer (1955) outlining the basic principles of surface exposure dating via in situ-produced cosmogenic nuclides, it is only “in 1986/87 [that] terrestrial in situ cosmogenic nuclides finally came of age (Dunai, 2010, p. 25)”. This was directly related to a major analytical progress in applied physical sciences: the development of accelerator mass spectrometry (AMS, e.g., Elmore & Phillips, 1987).
71To sum up, radiometric dating methods yielding numerical age controls, when they are considered in an historical perspective, experienced a late breakthrough that took place at the end of the xxth and the beginning of the xxist century. Importantly, the latter goes hand in hand with a declining use of relative and correlative dating methods in fluvial (palaeo-)environments. It is no exaggeration to say that the vast majority of fluvial archives and/or landforms are primarily dated by a handful of radiometric methods s.l. nowadays. Let’s conclude this section with a provocative question asked by Gordon & Dorn (2005, p. 109): “Why fuss with guesses, when radiometric control can be had?” In other terms, is it necessary to continue using relative dating methods when a chronological framework based on radiometric age estimates can be achieved instead?
72At a time where the use of relative/correlative and numerical dating was at a crossroads, Gordon & Dorn (2005) delivered a thoughtful message on how Quaternary geochronology was about to evolve by questioning the future of WR analysis. “The use of cosmogenic nuclide dating methods place in doubt the long-term future of weathering rinds (WRs) as a chronometric tool” (Gordon & Dorn, 2005, p. 97). Interestingly, they added: “Future refinement of WR dating methods have been largely halted by the advent of cosmogenic nuclide measurements. In just a few years, precise radiometric ages have become a far more appealing metric than convoluted calibrated ages providing only rough age assignments.” (Gordon & Dorn, 2005, p. 97).
73Three compelling elements from this message should be briefly discussed. Firstly, these authors perceived with very good timing that radiometric dating methods s.l. were about to take over the prevailing role, by far, in Quaternary geochronology. When we look back on what has happened over the last two decades, Quaternary scientists and researchers in geomorphology have manifestly opted en masse for these methods delivering numerical age estimates (section 4.3). Secondly, they address the fact that the use of dating methods initially conceived to infer relative age controls, such as lichenometry (Beschel, 1950) or WR analysis (Porter, 1975), has gradually shifted towards the intricate production of “numerical” chronologies based on non-straightforward calibration processes. By stretching the applicability of these methods to their very limits, many justified critics were raised in return (see section 4.1.2.). Thirdly, as largely pointed out in the same previous section, they emphasize the fact that these same dating methods definitely require significant methodological adjustments to say the least. The implementation of these necessary improvements was considerably impeded by the huge amount of work that has instead been achieved for the rapid development of numerical dating methods over the last three decades. Against this background, let’s ask two provocative questions:
74Is it definitely over and done with those disused relative dating methods?
75Are the prevailing and highly efficient radiometric/numerical dating methods self-sufficient?
76The abundant contributions delivered by numerical dating methods over the last decades have indisputably revolutionised the field of Earth and Quaternary sciences. However, Rixhon et al. (2017, p. 108) recently stated that: “[…], there is no ideal numerical dating method that can provide accurate age results on any kind of sample and in any context. The use of a dating method, even the most established one such as 14C, is limited by a range of intrinsic constraints and based on some implicit assumptions. Because the latter are rarely openly stated, expectations regarding numerical dating methods from non-geochronologists are sometimes unreasonable.”
77On the one hand, geomorphologists, field geologists or archaeologists, and more generally any researcher aiming at obtaining age control, should be thus aware of the main potential pitfalls of the most commonly used numerical dating methods. Despite the overwhelming number of case studies based on radiometric dating of alluvial archives or landforms in fluvial geomorphology and Quaternary studies, one must bear in mind that these methods can be affected by multiple problems in this kind of settings (fig. 8). Inherent methodological issues can bias the “real” age, either towards age over- or underestimation, of the fluvial event that has to be dated. Let’s consider 14C, OSL/IRSL, ESR and cosmogenic nuclide dating of a depositional event for instance. All methods are prone to age overestimation (Rixhon et al., 2017) due to possible:
78reworking of organic material, e.g., 14C-dated charcoal included in overbank fines;
79partial/incomplete bleaching of the quartz (OSL and ESR) or feldspar (IRSL) dosimeter;
80inherited concentrations of in situ-produced cosmogenic 10Be and 26Al in quartz or 36Cl in calcite (see section 5.2.3.);
81By contrast, the same methods are prone to age underestimation due to possible:
82contamination of more ancient organic material by modern carbon (even at very low levels; fig. 8);
83thermally unstable OSL signal components and/or weak fast component of the quartz dosimeter (e.g., Steffen et al., 2009; fig. 8);
84poor or absence of correction for anomalous fading occurring in feldspar (IRSL) or for thermal signal loss from the Ti centre (ESR; Richter et al., 2020);
85undetected post-depositional reworking/weathering and/or gradual exhumation of the sampled material for cosmogenic nuclide dating (see section 5.2.3).
Fig. 8: Two methodological difficulties (among others) inherently related to 14C and OSL/IRSL dating.
In black (left) - Effect of contamination by modern carbon on measured 14C ages; thin curves range from 0.25 to 2% of contamination by weight vs uncontaminated bold line (1:1). Note that, the older the sample, the more significant the bias. Modified from Pigati et al., 2007; In red (right) - Systematic discrepancy between OSL and IRSL ages; all samples collected from sand lenses plot below the 1:1 line (bold) without overlapping. This discrepancy is attributed to poor characteristics of the quartz dosimeter, i.e. weak fast component along with thermally unstable signal components. Modified from Steffen et al., 2009
Fig. 8 : Deux problèmes méthodologiques inhérents à la datation par radiocarbone et par luminescence optiquement stimulée (OSL et IRSL). En noir (gauche) - impact de la contamination par du carbone moderne sur des âges 14C mesurés. Les lignes fines représentent 0,25 à 2% de contamination par poids de l’échantillon par rapport à la ligne 1:1 (en gras). Remarquez que, plus l’échantillon est âgé, plus le biais est significatif. Modifié de Pigati et al., 2007 ; En rouge (droite) - divergence systématique entre les âges OSL et IRSL. Tous les échantillons collectés dans des lentilles sableuses se trouvent sous la ligne 1:1 (gras), sans chevauchement. Ce décalage est attribué aux mauvaises caractéristiques du quartz en tant que dosimètre (faible composante rapide et composantes du signal thermiquement instables). Modifié de Steffen et al., 2009.
86On the other hand, geochronologists must always be carefully and methodically informed on the context in which the samples were collected: “Absolute age determinations must be consistent with the stratigraphic and geomorphologic settings. The best results are obtained when physicists and earth scientists pool their knowledge and experience. A background in local and regional geology [and geomorphology] is especially important” (Watchman & Twidale, 2002, p. 1). In fact, field scientists and geochronologists working hand-in-hand during sampling is the best way to avoid some basic pitfalls mentioned above. We now thoroughly consider the case of surface exposure dating via concentration measurements of cosmogenic nuclides (classically 10Be) of alluvial archives to highlight this statement.
87In situ-produced cosmogenic-nuclides (CN) exposure dating has allowed huge progress on providing numerical age controls for a wide array of landforms since the first applications more than thirty years ago (e.g., Phillips et al., 1990). Since then, it has surely represented a breakthrough in the glacial realm by dating both moraine emplacement along with ice retreat worldwide (Balco, 2011). Two key assumptions lie at the heart of surface CN exposure dating involving the sampling of material from the surficial landform for which numerical chronological control is sought. Explicitly formulated for glacial landforms by Balco (2011, p. 12), these are: “(…) (i) that the rock surfaces to be dated lacked any inherited nuclide inventory when first exposed at the ice margin, and (ii) that they have remained uncovered, uneroded, and in their original configuration until the time of sampling”. Beyond its predominant use on glacial landforms, typically on moraines, both assumptions can be in fact extended to any depositional surfaces, making this geochronological tool so versatile. It was successfully employed to date, among others, coastal (e.g., tsunamigenic boulders; Rixhon et al., 2018) or lacustrine landforms (e.g., palaeo-shorelines; Rades et al., 2013) as well as earthquakes along fault scarps (e.g., Palumbo et al., 2004), mass-wasting processes (e.g., landslides, Ivy-Ochs et al., 2009) and volcanic events (e.g., basalt flows and cinder cones; Heineke et al., 2016). Importantly, fluvial deposits were extensively dated by this method as well. The dating approach here consists either in sampling material lying at or protruding the surface (alluvial fans in many instances; e.g., Siame et al., 1997) or performing depth profiles along terrace sediments (e.g., Rixhon et al., 2014).
88However, it rapidly turned out, from manifold studies carried out in a wide array of settings worldwide, that these assumptions were regularly transgressed. A scatter in apparent exposure ages resulting from different nuclide concentrations measured in individual clasts collected from the same landform were frequently reported (Balco, 2011). Moreover, the older the landform in the stratigraphic sequence, the larger the scatter in apparent exposure ages (Balco, 2011). These issues, characterised as the geological/natural source of uncertainty by Dunai (2010) and referred to as geologic (or geological) scatter by Balco (2011), are well-known processes and were abundantly discussed within the cosmogenic community. “Geological uncertainties arise mostly from our intrinsically limited a priori knowledge of the exposure history of individual samples (pre-exposure, burial) and the uncertainty in the shielding history of a sampling site (snow, soil, ash, vegetation cover)” (Dunai, 2010, p. 239).
89Following on from this, main uncertainties inherent to surface cosmogenic-nuclide (CN) exposure affect all depositional surfaces, including fluvial terraces (Anderson et al., 1996) or alluvial fan surfaces (fig. 9a-c). As for the latter, two representative case studies are briefly outlined here: they both show intra-surface scatter along their respective studied fan sequence. The first one focuses on fan surfaces at the active mountain front of the eastern Andes (Schmidt et al., 2011). The authors revealed that the clasts transported in the fluvial system, whose grain size range from pebbles to boulders, had accumulated a substantial pre-exposure nuclide inventory (equivalent to an apparent age of ~90 ka) on their way to the Andean footslopes. The most plausible reason was intermediate storage in intramontane basins. This process, referred to as inheritance, undoubtedly invalidates the first key assumption above. In the second case study, samples collected from the oldest and youngest surface of a large fan structure located in the Death Valley showed a variability in 10Be concentrations exceeding a factor four and two, respectively (Dühnforth et al., 2017). This accordingly results in large scatters in apparent 10Be exposure ages: ~270-990 and ~125- 275 ka for the oldest and youngest surface, respectively (fig. 9a-c). Interestingly, the much larger scatter for the oldest surface is well in line with the observation made above for glacial settings (fig. 9b-c). This highlights the fact that pristine landforms, such as mid-Pleistocene fan surfaces, should be seen as rare rather than common and makes the second assumption above invalid. Therefore, understanding the post-depositional evolution (i) of the landform, e.g., surface lowering over time, and (ii) of the dated material, e.g., boulder weathering or gradual exhumation, is a key element in determining the “true” exposure age of fan surfaces and, more generally, any exposure-dated landform (Balco et al., 2011).
Fig. 9: Geological scatter in 10Be surface-exposure ages from an alluvial fan system with different generations of surfaces in an arid setting.
a - Geomorphological map of the Warm Springs fan in Death Valley (California) with relative chronological classification (i.e. coloured surfaces) based on geomorphological markers. Full red dots and red/white dots refer to sampling location on the fan surfaces and the active channel, respectively. Numerical age estimates with 1σ uncertainty; those framed in magenta on each surface refer to the minimal and maximal exposure age; b - View of the oldest surface (Qg2) at the 10Be sampling spot yielding an “old” age estimate (~925 ka); c - Box-and-whisker plots of 10Be age estimates: note (i) the significant age scatter for each fan surface and (ii) the much larger scatter for the oldest surface. Modified from Dühnforth et al., 2017
Fig. 9 : Dispersion des estimations d’âge numérique basées sur des mesures de concentration en 10Be réalisées sur trois générations de surfaces d’un cône de déjection en milieu aride. a - Carte du cône Warm Springs dans la Vallée de la Mort (Californie) avec classification chronologique relative basée sur des marqueurs géomorphologiques (surfaces colorées). Les cercles rouges et rouges/blancs localisent les échantillons prélevés respectivement sur les différentes surfaces et dans le chenal actif. Les estimations d’âge sont données avec une incertitude de 1σ, celles encadrées en magenta se référant aux valeurs minimale et maximale pour chaque surface. b - Vue de la surface supérieure (Qg2) à l’emplacement de l’échantillon produisant un âge “ancien” (~925 ka); c - Diagramme “en boîte à moustaches” des âges 10Be d’exposition: remarquez que la dispersion des âges est (i) significative pour chaque surface et (ii) d’autant plus importante que la surface est ancienne. Modifié de Dühnforth et al., 2017.
90A key question arising from the remarks above is how to deal with this scatter in nuclide concentrations to produce a reliable chronological control based on exposure ages (fig. 9c). On the one hand, post-depositional surficial erosion, clast weathering or gradual boulder exhumation over time result in younger apparent ages and, if not properly considered in “true age” calculation, in underestimating the age of the landform. On the other hand, an inherited nuclide inventory result in older apparent ages and, if not properly considered, leads to age overestimation. Balco (2011) lists and discusses a set of approaches to overcome geological scatter. Only some of them are briefly presented here and we refer to this review paper for further details.
91A first, commonly used strategy consists in subsampling the dataset based on geomorphological arguments with two end-members (Heyman et al., 2011). It argues either that the true age of the landform is best approximated by the oldest sample provided that post-depositional boulder/clast weathering and/or landform erosion is considered as predominant (e.g., Schaeffer et al., 2006). On the contrary, if inheritance counterbalances erosion or boulder/clast exhuming and largely prevails, the youngest sample yields the best estimation (e.g., Benson et al., 2004). However, this subsampling uses the extreme values of the distribution, i.e., the minimal or maximal apparent exposure ages. It is thus highly sensitive and entails that all samples with no exception obey the geomorphological assumptions. A second strategy consists in applying statistical criteria (fig. 10), such as Rinterknecht et al. (2006) who used the Chauvenet’s criterion. It allows producing a subset which can be appropriately averaged by detecting and removing outliers whose probability of belonging to the same population as the remaining data is low. The effectiveness of this approach however requires that only a few values out of the complete dataset are outliers reflecting geological scatter (fig.10). A third strategy consists in matching observed and predicted distributions of exposure ages, such as the approach developed by Applegate et al. (2010). “The various geomorphic-argument-based estimators (…) are all special cases of the idea that a physical model of the processes that deliver boulders [or clasts] to a moraine [or an alluvial fan], and subsequently act to move, bury, exhume or weather them, can be used to quantitively predict the distribution of exposure ages on the moraine [or fan]” (Balco, 2011, p. 17).
Fig. 10: Statistically dealing with geological scatter in datasets of CN exposure ages.
White circles are (almost) normally distributed samples and are used to determine the exposure time (black line and 1σ uncertainty) of the landform (i.e. the Pommerian moraine in N. Germany), whereas black circles are outliers (~15% of the dataset) that can be reliably excluded (Chauvenet’s criterion) because of post-depositional processes (erosion and/or shielding) or inheritance. Data from Rinterknecht et al., 2006 ; diagram modified from Dunai, 2010.
Fig. 10 : Approche statistique pour déterminer un âge d’exposition à partir d’un large jeu de données de concentration en 10Be. Les cercles blancs représentent les échantillons avec une distribution (presque) normale sur base desquels le temps d’exposition (ligne noire avec incertitude de 1σ) de la moraine poméranienne (Allemagne) est déterminé. Les cercles noirs sont des outliers (~15% du set de données) qui peuvent être exclus de manière fiable (critère de Chauvenet) pour cause de processus postérieurs au dépôt (par ex. érosion) ou d’héritage. Données de Rinterknecht et al., 2006; diagramme modifié de Dunai, 2010.
92Importantly, whatever the strategy employed to cope with geological scatter, an in-depth knowledge/understanding of the sampling context at various spatial scales is key (Watchman & Twidale, 2002), from surface processes acting throughout the whole catchment to local-scale weathering processes. This is obviously relevant for the first and second strategies but also for the approach matching observed and predicted distributions as geological processes involved in the model can be defined and calibrated thanks to field observations (Balco, 2011). In this respect, a first unavoidable step is to routinely document as many field information as possible in a sampling checklist and field-note template. Such a thorough template was developed by Dunai (2010) for CN dating. Duval et al. (2017) explicitly reported the necessary amount of information required for ESR dating of optically bleached quartz, including fieldwork and sample collection. Here, the main message is to go beyond these highly necessary but mere field observations: in addition to numerical dating tools, we strongly suggest applying some of the relative dating methods discussed above, including some of the disused ones. Indeed, these methods deliver chronological information that can greatly help gaining highly valuable insights into geological scatter related to a dataset of surface-exposure ages. More generally, they can be useful for all radiometric dating methods and this aspect is developed in the next section.
93There are opposite answers to the provocative question at the very end of section 4 (i.e., “Why should we continue using relative dating methods when a chronological framework based on radiometric age estimates can be achieved instead?). It is a fact that (i) some of the debated relative methods, such as WR analysis, have progressively fallen into disuse, if not completely passed into oblivion, (ii) there is an undisputed prevailing use of radiometric dating methods in a wide array of fluvial contexts nowadays. The option retained here suggests a positive outcome for some “ancient”, somehow disused methods whose forgotten or underrated benefits have been neglected over the last decades. Rather than stretching the applicability of these methods to their limits to obtain convoluted calibrated ages, thereby provoking justified critics, it is argued here to let these methods do what they were basically made for: the production of relative age controls. Even if new methodological developments are still needed (Gordon & Dorn, 2005), their most up-to-date usage warrants a consistent relative chronological information. For instance, the still-in-use lichenometric dating continues delivering useful relative outputs, as recently showed by Rosenwinkel et al. (2015, p. 237): “Given these constraints, we favour the use of lichenometry for establishing a relative - rather than an absolute - chronology of rock-glacier lobes in the northern Tien Shan”. A relative chronology via WR analysis of successive generations of moraines and rock glaciers in the French Alps was recently achieved as well (Cossart et al., 2010) potential of this method to support CN exposure ages is discussed in the next section.
94Measuring rind thickness along a sequence of landforms can advantageously help radiometric dating methods. Minor lithological variations between the different generations of the investigated landform would consequently be a prerequisite. This assumption is likely in the case of successive fan surfaces/terraces located at the outlet of small mountainous catchments as the source area of eroded material should remain quasi-identical over time (fig. 9a,b). In this respect, WR analysis conveniently matches the CN sampling procedures of these surfaces as they both target similar coarse grain sizes. Since the first applications of CN exposure dating to these landforms about 25 years ago (e.g., Siame et al., 1997), sampling strategies have varied, mostly depending on the grain size distribution at the fan surface. They most commonly consist in (i) collecting single pebbles or cobbles lying at the surface (e.g., Le Dortz et al., 2009); (ii) amalgamating several dozens of pebbles or cobbles (e.g., Dühnforth et al., 2017) or (iii) chiseling off material from boulders protruding the surface, in particular from flat-top areas (e.g., Siame et al., 1997). Whilst sub-surface sampling (e.g., Mériaux et al., 2004) or depth profiles were sometimes carried out (Le Dortz et al., 2012), Schmidt et al. (2011) compared these sampling approaches. Importantly, many of the case studies performed in arid environments worldwide reported geological scatters in the exposure-ages datasets (sometimes very significant, see section 5.2.3), whatever the spatial context of sampling.
Fig. 11: Rejuvenating WR analysis to support CN exposure dating.
a – Road cut in a marine terrace (N Chile) embedding weathered boulders. Note the different degrees of weathering (partially vs complete-weathered) and the preservation of the in situ WR of a granitic boulder at depth. These observations can be helpful for depth profile dating. Photos from G. Rixhon; b - Four-step evolution of WR integrating erosion processes over time. Based on observations made on moraine boulders in Iceland, for which flaking of the rind was already observed within 150-200 years after weathering had begun under periglacial conditions, this evolutionary trend was also acknowledged in arid settings over much longer timescales (Ajo Mountains, Arizona). This notion of time can be much helpful for surface exposure dating. Modified from Etienne, 2002 and Gordon & Dorn, 2005.
Fig. 11 : Utiliser l’analyse des cortex d’altération pour consolider des datations par nucléides cosmogéniques. a - Coupe dans une terrasse marine (Chili du Nord) incluant des blocs altérés. Remarquez les différents degrés d’altération (partiel et complet) et la préservation in situ d’un cortex d’altération d’un bloc granitique en profondeur. Ces observations peuvent être utiles pour un profil vertical de concentrations. Photos de G. Rixhon ; b - Evolution en quatre étapes du cortex d’altération intégrant son erosion au cours du temps. Basé sur des observations réalisées sur des blocs morainiques en Islande, où l’écaillage du cortex se produisait endéans 150-200 ans après le début de l’altération en contexte périglaciaire, ce schéma évolutif fut aussi reconnu dans un contexte aride sur des périodes de temps beaucoup plus longues (Monts Ajo, Arizona). Cette notion de temps apparaît centrale pour des datations par exposition. Modifié de Etienne, 2002 et de Gordon & Dorn, 2005.
95It is argued here that systematic WR analysis could have several benefits for CN exposure dating (fig. 11a,b). Firstly, beyond those clasts exhibiting severe weathering making them unsuitable for sampling, WR analysis would detect potential weathering variability on a single landform, or within its deposits (fig. 11a). In particular, individual clasts that are likely to have undergone sustained alteration, materialized by a significantly thicker WR than that from the rest of the population, could be isolated. It would be thus highly interesting to test whether these clasts contain higher nuclide concentrations as a result of longer pre-exposure. If yes, higher WR thickness could be used as a proxy to detect higher inherited CN inventories. Note that this approach could be likewise useful to depth profiles based on cosmogenic nuclides. Secondly, one should consider in situ erosion of WR (fig. 11a). Increasing thickness of WR over time most probably is the outcome of a dynamic equilibrium between rind growth towards the clast’s core and surficial erosion processes (Gordon & Dorn, 2005; fig. 11b). Interestingly, a case study on basaltic clasts in Iceland revealed that, subsequently to weathering initiation (“zero clock”, stage 1), three consecutive stages were identified (Etienne, 2002): growth only (stage 2), growth and onset of (micro-)erosion (stage 3) and growth and flaking (stage 4). Resulting from combined physical and biochemical processes, sustained exfoliation and flaking of the rind had already occurred within 200 years after weathering initiation (fig. 11b; Etienne, 2002). A similar evolutionary trend was also acknowledged in arid settings but over much longer timescales, i.e., 103-104 years (Ajo Mountains in Arizona; Gordon & Dorn, 2005). This could be very useful for further constraining geological scatter in datasets by exploring whether the younger apparent exposure ages in the distribution match those clasts having attained a “maturity” degree in erosion of the WR (typically stages 3-4 described above). If yes, such chronological information would greatly help assessing how much the CN exposure age is biased towards age underestimation and possibly corrected for. Thirdly, Ma et al. (2012) successfully quantified the rate of rind formation under tropical conditions using U-series analysis on basaltic and andesitic clasts. This can be usefully employed as a potential supplementary chronological constrain. Finally, late Quaternary fluvial terrace flights from the Southern Island of New-Zealand were extensively dated by a combination of relative age information (i.e., WR analysis and Schmidt hammer rebound values) with numerical age control (i.e., radiocarbon, luminescence and cosmogenic nuclides; Stahl et al. 2013). In that respect, multi-method geochronological approaches are thoroughly discussed in the section 7.
96As highlighted in figure 3, several weathering processes surficially affect boulders and can be turned into relative dating methods. One of them is the development of vertical/meridional weathering fractures. Their growth or expansion results from various erosional processes, such as salt shattering and freeze-thaw action (Goudie, 2013). Their potential as a dating tool (fig. 12a-c), though long disregarded, was successfully demonstrated in an enlightening case study performed on alluvial fan surfaces in an arid environment of California (Owens Valley, D’Arcy et al., 2015). Similar to WR analysis along a sequence of landforms, increasing fracture widths reflecting higher degree of alteration are indicative of increasing ages of fan surfaces (fig. 12a,b). The main novelty here is that this study took advantage of several existing CN exposure-age datasets for these fan surfaces (e.g., Zehfuss et al., 2001; Dühnforth et al., 2007). Owing to a remarkable consistency along the fan sequence between mean 10Be exposure age and mean fracture width highlighted by a textbook linear regression (R²=0.99; fig. 12c), D’Arcy et al. (2015) consider this weathering parameter as an ideal tool for surface dating and achieved to:
97(i) accurately calibrate the widening rate (~1 mm/ka) of fractures in granitic boulders over a long timespan (~ 140 ka);
98(ii) improve the uncertainties related to both 10Be age estimates and surface erosion rate for the sequence of studied fans;
99(iii) develop a robust age calibration for fracture widths in clasts or boulders of non-dated fan surfaces in Owens Valley, and to thereby spread the chronostratigraphy to further fan systems in the Sierra Nevada.
100“This novel approach to dating sedimentary surfaces is inexpensive and easily applied in the field, and has the potential to significantly increase the temporal and spatial density of age constraints available for a particular study area” (D’Arcy et al., 2015, p. 487). It perfectly epitomises what could be achieved by WR analysis as a complementary method to CN surface-exposure dating.
Fig. 12: Weathering fracture width in boulders as an efficient dating tool for alluvial surfaces.
a - Field photos exhibiting fractures throughout granitic boulders embedded in a sequence of fan terraces in Owens Valley (Sierra Nevada, California). Note that, along the sequence of fan surfaces of increasing ages (Q3= youngest; Q1 =oldest), the fracture width consistently increases; b - Variations of fracture width (mean value=dashed black line) on the same fan surfaces as above, verifying the aforementioned assumption. n refers to the number of observed fractures; c - Positive linear correlation between mean surface age of alluvial fan (inferred from 10Be dating) and mean fracture width. Note the remarkable fit of linear regression (R2 = 0.99!) intersecting the origin and having a gradient equal to the time-averaged rate of fracture widening. Modified from D’Arcy et al., 2015.
Fig. 12 : La largeur des fractures d’altération dans des blocs reposant sur des surfaces alluviales comme méthode de datation efficace. a - Photos montrant des fractures d’altération dans des blocs granitiques reposant sur des “terrasses” de cônes de déjection dans la vallée d’Owen (Sierra Nevada, Californie). Remarquez que la largeur de la fracture augmente en relation avec l’âge croissant des surfaces (Q3= plus jeune; Q1 =plus âgé) ; b - Variation de la largeur de la fracture (valeur moyenne=ligne hachurée) sur les mêmes surfaces, vérifiant la relation “fracture/âge”. n correspond au nombre de fractures observées ; c - Corrélation linéaire positive entre l’âge moyen de la surface du cône de déjection (basé sur des âges 10Be d’exposition) et largeur moyenne des fractures. Remarquez le R² de 0,99 pour la régression linéaire qui intersecte l’origine et dont la pente est égale au taux d’élargissement de la fracture moyenné dans le temps. Modifié de D’Arcy et al., 2015.
101Already tackled in the previous section, the final part of this contribution aims at promoting the use of geochronological approaches based on multiple dating tools. These approaches can be roughly subdivided into two groups:
102(i) the comparison of different dating methods implemented in the same setting;
103(ii) the combined use of dating methods on the same sampling material, including those approaches which merge the core principles of different dating methods.
104The comparison of different numerical dating methods, already outlined in the 1960s (e.g., Kaufman & Broecker, 1965), primarily consists in producing independent chronologies and checking whether these methods yield consistent age results. Note that, in many cases, the chronological information varies by nature because (i) the methods frequently require different dateable material and (ii) their core principle basically differs (fig. 1 & tab. 1). Let’s consider first one the most common combination of numerical dating methods, especially in fluvial settings: 14C and OSL/IRSL (fig. 13). The first method fundamentally constrains the moment during which the carbon-rich organism ceased exchanging carbon with its environment. The subsequent radioactive decay of 14C allows then inferring the timespan elapsed since the death of plant or animal. By contrast, the second method records the burial time of the quartz- and/or feldspar-bearing material in a sediment sequence since the last time it was exposed to sunlight. In certain instances, both methods support each other by showing a good agreement between their respective dating results. This was the case for the dating of a palaeosol bearing archaeological remains embedded in a fluvial/aeolian sequence along the coast of Central Chile (Longotoma): the three consistent 14C age estimates are particularly well matched by the three IRSL-dated sand layers sandwiching the palaeosol (May et al., 2015). In other instances, chronological outcomes of both methods show (more or less) large discrepancies. Interestingly, these inconsistencies can then provide valuable insights into methodological biases encountered by dating methods (Rixhon et al., 2017). This was the case for coupled 14C and OSL age estimates of Late Pleistocene deposits of fill terraces from the Nene and Welland Valleys (eastern England; Briant & Bateman, 2009). 14C dating, beyond 29-35 ka, produces systematic and significant age underestimations, which were attributed to secondary contamination of older organic material by low levels of modern carbon (fig. 13).
Fig. 13: On the usefulness of cross-checking numerical age estimates from independent methods.
Comparison of OSL and 14C ages (uncertainties=1σ) from identical fluvial deposits. Note that, beyond 30-35 ka, 14C dating yields systematic and significant age underestimations, attributed to contamination by modern carbon (see fig. 8). Modified from Briant & Bateman, 2009.
Fig. 13 : Utilité de croiser les estimations d’âges numériques obtenues par des méthodes indépendantes. Comparaison d’âges OSL et 14C (incertitude de 1σ) produits pour les mêmes dépôts fluviaux ; remarquez la sous-estimation systématique et significative des âges 14C à partir de 30-35 ka, probablement due à une contamination par du carbone moderne (voir fig. 8). Modifié de Briant & Bateman, 2009.
105Further combinations are of course conceivable. To remain in line with the case studies involving the dating of alluvial fans via 10Be surface-exposure ages in California (see sections 5.2.3 and 6.3; D’Arcy et al., 2015; Dühnforth et al., 2017), we briefly present the outcomes of Blisniuk et al.’s (2012) study. They applied a combination of cosmogenic nuclide (10Be) and U-series dating to constrain the deposition of Holocene and Late Pleistocene fans in the Anza Borrego Desert. The first method, via the threefold sampling of individual boulders, amalgamated clasts and depth profile (see section 6.2), is supposed to constrain the timing of surface abandonment. The second method provide minimum age estimates for surface abandonment since post-depositional pedogenic carbonate from subsurface clast-coating was sampled for 230Th/U dating. Age results regularly met this chronological assumption: “concordant 10Be and U-series dates were measured for most fans, thereby providing robust age estimates [for fan deposition]” (Blisniuk et al., 2012, p. 26).
106Finally -and importantly-, provided that the nature, composition and grain size distribution of the fluvial sediments are favourable, these multi-method approaches can -and should- compare numerical age estimates inferred from different radiometric dating methods with age information delivered by relative and correlation dating methods. Such a successful approach was carried out by Bartz et al. (2018, 2019) to date fill terrace deposits along the lower Moulouya River in north-eastern Morocco (fig. 14). They employed a combination of ESR, luminescence and palaeomagnetic dating in the same fined-grained deposits. They convincingly assigned terrace formation to the Early Pleistocene (fig. 14). The external crosschecking between ESR and thermally transferred-OSL dating is not only consistent, yielding finite age estimates older than 1 Ma, but is also in good agreement with the reverse polarities (older than 0.78 Ma) along the sequence (fig. 14).
Fig. 14: Example of a successful multi-method geochronological approach (numerical and relative dating) to date Quaternary terrace deposits.
Left - Terrace deposits of the Lower Moulouya (NE Morocco) made of massive and coarse gravel layers embedding fine-grained lenses; the latter were specifically targeted for the geochronological approach. Photos from G. Rixhon; Right - simplified log of the terrace with the chronological outputs. Note the twofold consistency: concordant numerical dating of two independent methods (TT-OSL and ESR) points to Early Pleistocene deposition, corroborated by the reverse normal polarity (palaeomagnetism). Modified from Bartz et al., 2019.
Fig. 14 : Exemple d’une approche géochronologique multi-méthodes (numériques et relatives) fiable pour dater des dépôts de terrasse quaternaires. Gauche - Dépôts de terrasse de la Basse Moulouya (NE Maroc), les unités principales et massives de galets incluant des lentilles sableuses qui furent ciblées pour l’échantillonnage multi-méthodes. Photos de G. Rixhon ; Droite : log simplifié avec résultats chronologiques. Remarquez la double cohérence : la concordance des âges numériques indépendants (TT-OSL et ESR) indique un dépôt lors du Pléistocène ancien, corroboré par la polarité inverse des mesures paléomagnétiques. Modifié de Bartz et al., 2019
107Beyond the simultaneous use of two or more dating methods in the same setting or sedimentary sequence, geochronological approaches based on coupled methods which investigate the same sampling material also exist. Originally developed by Grün et al. (1988), the combined U-series/electron spin resonance (US-ESR) dating approach is one of them. It has successfully yielded numerous age estimates in archaeological sites containing fossil teeth since then, including those sites linked to fluvial archives (e.g., Barranco León at Orce in Andalusia: Duval et al., 2012; fig. 4). The main advantage of this method, which relies on ESR measurements and modern U-concentrations (i.e., 230Th/234U and 234U/238U) measured in each dental tissue, does not a priori assume the U-uptake history but assesses it mathematically instead. Furthermore, a very few pioneering studies further explored the potential of merging the dating principles of independent numerical dating methods. One of them is the innovative approach developed by Guralnik et al. (2011). It uses a mathematical framework for consistently incorporating 10Be concentration data along a depth profile with OSL ages from a single alluvial section (fig. 15). This model is based on three parameters and solves an integrated, co-dependent and self-consistent set of equations and assumes fluvial aggradation at a constant rate, with uniform cosmogenic inheritance, followed by terrace abandonment and subsequent preservation and exposure of its surface. This scenario of terrace evolution may be validated or rejected by comparing model depth concentration data.
Fig. 15: Example of a geochronological approach merging the dating principles of two independent numerical methods.
Integrated 10Be depth profile and OSL model results in the model parameter space of deposition time (t1), exposure time (t2), and 10Be inheritance for a single alluvial sequence. The cosmogenic nuclide and OSL model best fits and their respective 68% confidence level envelopes are shown (CN: green dot and surface ; OSL: blue surface), along with the intersection of the two confidence surfaces (deep blue). Modified from Guralnik et al., 2011.
Fig. 15 : Exemple d’une approche géochronologique fusionnant les principes de base de deux méthodes de datation numériques indépendantes. L’intégration du profil de concentration en 10Be et du modèle OSL permet de modéliser simultanément le temps d’enfouissement (dépôt, t1), le temps d’exposition (t2) et l’héritage en 10Be pour une même séquence alluviale. Les meilleurs fits du modèle ainsi que les enveloppes de confiance à 68% sont montrées pour les nucléides cosmogéniques (point et surface en vert) et la luminescence (surface en bleu), ainsi que l’intersection des deux surfaces (bleu foncé). Modifié de Guralnik et al., 2011
108Instead of a lengthy text (the manuscript being already long enough), let’s exceptionally conclude this contribution by a portfolio of photos showing the diversity of fluvial settings that can be dated by one or multiple - even better as suggested above - method(s) (fig. 16). Here, the “dateable” material is subdivided into three categories
109(i) bedrock landforms:
110- surface of bedrock strath terraces (CN and/or OSL surface exposure dating);
111- lava flow (e.g., basaltic; K/Ar or Ar/Ar dating) sealing the fluvial deposits (CN isochron burial dating)
112(ii) clastic fluvial sediments:
113- either coarse- (e.g., pebbles; CN depth profile dating) or fine-grained alluvium (OSL/IRSL, ESR, palaeomagnetic dating);
114- either from subaerial deposits (e.g., alluvium-mantled terraces) or endokarstic deposits (cave-deposited alluvium; CN burial dating)
115(iii) material included in the sedimentary sequence:
116- inorganic material such as secondary carbonates (U-series dating) or stone/lithic artefacts (CN burial dating);
117- organic material such as embedded trunk (14C dating, dendrochronology) or fossil teeth (U-series coupled to ESR dating).
118As a final sentence, to take full advantage of dating these fluvial landforms and/or deposits: be curious, observe, and document all information in the field…and keep being innovative and well-informed.
Fig. 16: Portfolio of dateable material in fluvial settings for a combined geochronological approach (the symbology refers to fig. 4).
a - Bedrock strath terrace along the Lower Moulouya Valley (Morocco; photo: G. Rixhon); b - Basaltic lava flow sealing terrace deposits of the Tigris River (Turkey; modified from Bridgland et al., 2007); c & d - Alluvium-mantled terraces of the Lower Moulouya and Middle Rhine (Germany), respectively (photos: G. Rixhon); e - Cave-deposited alluvium in the Chawresse multi-level cave system in the lower Ourthe (Belgium, photo: A. Peeters); f - Embedded trunk in floodplain deposits of the Siret River (Romania; modified from Rădoane et al., 2015); g - Pedogenic clast-coating made of secondary carbonates occurring in terrace deposits of the Wind River (USA, modified from Sharp et al., 2003); h - Acheulean handaxe buried in the fluvial sequence at the La Noira site along the Cher River (France, modified from Despriée et al., 2017); i. Macrommal molar buried in the infill of the La Belle-Roche cave along the Amblève River (Belgium, photo: G. Rixhon).
Fig. 16 : Portfolio des matériaux datables en contexte fluvial en utilisant des approches géochronologiques couplées (symbologie : voir fig. 4). a - Terrace d’érosion sur bedrock dans la basse vallée de la Moulouya (Maroc; photo: G. Rixhon) ; b - Lave basaltique scellant des dépôts de terrasse du Tigre (Turquie; modifié de Bridgland et al., 2007) ; c & d - Terrasses d’accumulation de la Moulouya et du Rhin Moyen (Allemagne), respectivement (photos: G. Rixhon) ; e - Dépôts alluviaux piégés en grotte au sein du système multi-étages de la Chawresse dans la vallée de l’Ourthe Inférieure (Belgique, photo: A. Peeters) ; f - Tronc d’arbre inclus dans des dépôts fins de plaine alluviale de la Siret (Roumanie; modifié de Rădoane et al., 2015) ; g - Précipitation pédogénique de carbonates secondaires enrobant des galets de terrasse de la Wind River (USA, modifié de Sharp et al., 2003) ; h - Biface acheuléen enfoui dans la séquence fluviale de la Noira, vallée du Cher (modifié de Despriée et al., 2017) ; i. Molaire de macro-mammifère enfouie dans le remplissage karstique de la Belle-Roche, vallée de l’Amblève (Belgique, photo: G. Rixhon).