- © 2013 by the Mineralogical Society of America
The Quaternary Period, by virtue of the near-surface preservation and widespread accessibility of its environmental archives, provides fundamental data to test models of climate change, sea level variation, geomagnetic field variation, human and faunal migration, cultural evolution and more. Spatially disparate records of past environmental change with subannual to multimillennial temporal resolution are compared to examine the relative timing of events and consider causal mechanisms, and this analysis puts great demands on the chronological tools available. Highly precise and accurate age estimates are required, in concert with correlative tools or chronostratigraphic markers. We focus on radioisotope chronometers (e.g. U-series, 40Ar/39Ar and 14C) and illustrate their application in three vignettes for which different strategies are required: (1) the dramatic decades of the last deglaciation (~14.7 ka), (2) before and after one of the last geomagnetic excursions (~41 ka) and (3) the glacial–interglacial cycles of the Middle Pleistocene (125–780 ka).
- Quaternary geochronology
- uranium-series dating
- radiocarbon dating
- chronostratigraphic markers
- early modern humans
Environmental archives deposited during the Quaternary Period, that is, the last 2.6 million years, provide an extremely important test bed for theories of Earth system change and human evolution because they are the most accessible, spatially comprehensive and highest-temporal-resolution records of Earth's geo logical history. The Quaternary Period encompasses both the Pleistocene and the Holocene epochs and is characterised by large global and regional climatic swings, most notably the waxing and waning of the Northern Hemisphere ice sheets. Accurate time constraints are crucial for understanding connections between different processes and quantifying rates of change across the Earth system during this time period. The complex pattern of environmental change provides important boundary conditions and direct comparative evidence for modelling efforts that assess the sensitivity of climate to both natural and anthropogenic forcing. Time slices and temporal records of Quaternary ice sheet dimensions, sea surface temperatures, vegetation patterns, greenhouse gas concentrations, atmospheric dust concentrations, and much more, provide the fundamental physical evidence interpreted in successive Intergovernmental Panel on Climate Change reports. Importantly, the Quaternary Period provides all the available evidence for hominin evolution from the earliest species of the Homo genus (Homo habilis at 2.3 Ma) to anatomically modern humans. The range of materials and settings studied in Quaternary science is vast, but chronological control underpins nearly all efforts. There are two major challenges: effectively synthesising widely spaced data, and matching age-estimate uncertainty with ever-improving sampling resolution.
DEVELOPING QUATERNARY CHRONOLOGIES
The past century saw a tremendous effort by scientists working at local or regional scales with a wide range of sedimentary archives, including ocean and lake sediments, ice cores, glacial moraines, and aeolian and cave deposits. They have amassed a vast array of information using a combination of numerical, correlative and stratigraphical tools to provide a complex global chronostratigraphical framework for the Quaternary Period. A century ago, Quaternary geochronology in Europe and beyond was reliant on correlation with the fourfold Alpine glacial scheme of Penck and Brückner (1909) – from oldest to youngest, the Günz, Mindel, Riss and Würm – which was based on mapping of glacial moraines, deltaic sequences and river terraces (Fig. 1). For many years, age constraints for Quaternary material from many different locations and settings across the globe were achieved by recognising depositional or erosional units in the stratigraphical record, then “counting from the top” before making poorly constrained correlations with the Alpine scheme. Increasingly complex records of environmental change relied on this simple and persistent model, the temporal axis of which was based on a crude estimate of sedimentation rates in Alpine glacial lake deltas. The duration of interglacials was based on relative ages, calibrated against process rates. While the Alpine glacial scheme took hold, astronomical theories of climate change that were developed by Köppen and Milankovitch, building on the earlier work by Croll, gained strength, and it was recognised that ages based on orbital parameters could be applied to the ice ages. By the mid 1940s, the groundwork had been laid for a move to radioisotopic methods that could provide robust age estimates, especially radiocarbon (14C) and potassium–argon techniques via decay counting or mass spectrometry. A major proponent of the combined use of radioisotope and astronomical methods at this time, F. Zeuner of the Department of Geochronology, University of London, was sufficiently confident to state in 1946 that “though many adjustments will be made necessary by future research, the story revealed by the ‘calendar’ of the Pleistocene is extremely consistent” (Zeuner 1946). The underlying tenets of his argument remain, but there has indeed been much refinement, and today's picture is significantly more complex (Fig. 1).
Dating methods based on radioactive decay are part of what has been termed the numerical, absolute or chronometric branch of geochronology. There are also many important relative dating techniques that rely on calibration of process rates to assign temporal constraints. We do not review these, but acknowledge the continued advances in, for instance, amino acid racemisation and biomolecular clocks. Also not represented here are the cosmogenic nuclide burial and exposure dating techniques that have revolutionised studies of landscape evolution and glacial history, and optically stimulated luminescence dating, which is widely applied to aeolian and fluvial sediments.
Critical for all geochronological methods is the combination of precision and accuracy (Schoene et al. 2013 this issue) and, in ideal circumstances, these measures should match achievable sampling resolution. Since the advent of radioisotopic dating techniques for the Quaternary Period in the 1940s, instrumental precision has improved sizeably, and the extent to which this is matched by improvements in accuracy is dependent upon the techniques and materials under investigation. By way of example, Arnold and Libby (1949) used a technique that consisted of “combustion of about 1 ounce of wood” (i.e. 28.3 g) to achieve specific 14C activities with uncertainties of 2–10%, whereas routine accelerator mass spectrometry methods during the past two decades have precisions less than 5‰ (parts per thousand) for samples less than 50 mg – an improvement of four orders of magnitude.
We cannot fully illustrate the vast array of methods used to “date the past” in this short review, nor all of the geological deposits that record temporal patterns of change or events. Rather, we focus on three time windows for which high-resolution geochronology at different scales is fundamental to current debate. Within these time windows, we illustrate how state-of-the-art technology has improved our understanding of the dynamics of the Earth system, highlight the accuracy of respective methods, and stress the importance of multiproxy evidence, time-stratigraphic markers and a combination of synchronous geochronometers.
Time Window 1: Towards Decadal Resolution during the Last Deglaciation?
Unravelling the driving mechanisms for the repeated pattern of transitions, or deglaciations, from cold glacial periods with large continental ice sheets across much of North America and Eurasia to a warmer state with ice sheets restricted to the high Arctic and Antarctica, is one of the main challenges to our understanding of past and future climate. The last deglaciation, between 20 and 10 thousand years ago (ka), has been intensely studied (see Clark et al. 2012 for a recent synthesis), and a large number of high-resolution climate records have provided crucial information about changes in atmospheric CO2, CH4, sea level, routing of glacial meltwater into the ocean, temperature and many other parameters. The initiation of this deglaciation was triggered by an increase in Northern Hemisphere solar radiation (insolation), causing the extensive ice sheets to reduce in area and volume, passing a threshold before major collapse. Although Northern Hemisphere insolation variation was gradual and monotonic during the transition to the warm interglacial climate of the Holocene, there were abrupt changes in climate states in both directions at the centennial to subdecadal scale.
Greenland ice cores provide the highest-resolution, continuous records of climate change during the past 100 thousand years, and we know, on the basis of hydrogen and oxygen isotope evidence from annually banded layers, that one of the highest-amplitude warmings during the last deglaciation occurred in this region over a period as short as ~3 years at ~14.7 ka (Steffensen et al. 2008). However, a precise and accurate absolute age determination for this sharp transition to the warm period known as the Greenland Interstadial 1 (GI-1) (see Fig. 2) is hampered by uncertainties associated with counting these layers downwards from the top, which can incrementally accumulate uncertainties of as much as ~200 years at such depths. In Antarctica, where snow accumulation rates are much lower, uncertainties are greater still. Therefore, matching age uncertainty with sampling resolution in Greenland ice cores is not yet achievable. It is possible, however, to synchronise polar records from both hemispheres at higher resolution by correlating the measured signals of globally well-mixed gases, such as CH4, in trapped bubbles in the ice, and this has yielded very useful information about the relative timing of temperature shifts and greenhouse gases (e.g. Stenni et al. 2011). But to extend such temporal comparison globally, it is crucial that we fold in independent evidence from alternative climate-sensitive sediments, such as lake, ocean and cave deposits from lower latitudes. Currently, this temporal comparison is restricted to the centennial scale during the last deglaciation, due to the combination of uncertainties in radioisotope dating methods and annual-layer counting in respective deposits, in spite of much higher subsampling resolutions achievable for proxy climate evidence.
At lower latitudes, U–Th-dated calcium carbonate deposits offer some of the best evidence for climate change because centennial or better age uncertainties can be achieved. Data obtained from these deposits can be compared with polar ice cores to detect synchroneity, causal links and feedback in the Earth system. Figure 2 illustrates a range of such evidence from the marine and terrestrial realm of the last deglaciation, including deep-sea corals from the Southern Ocean, which record the extent of carbon exchange between the atmosphere and ocean water masses (Burke and Robinson 2012); oxygen isotope composition (δ18O) in speleotherm calcite from the Hulu cave, China, which reflects the intensity of the Asian monsoon (Wang et al. 2001); and fossil surface-ocean corals, which function as direct sea level markers (Deschamps et al. 2012).
Of particular importance to the last deglaciation are north–south changes in ocean and atmospheric circulation and feedbacks associated with greenhouse gases. It is suggested that the transition to the GI-1 warm period was initiated by a stronger meridional overturning circulation that provided more effective heat transport to the Northern Hemisphere high latitudes. This could have triggered a major Northern Hemisphere ice sheet melting event with an initially rapid sea level rise. This change in sea level led to the collapse of an unstable portion of the West Antarctic Ice Sheet, resulting in further large-scale meltwater release. Increases in atmospheric CO2 at this time, associated with reorganisation of water masses in the Southern Ocean, would have accelerated the melting. Concurrent with these high-latitude shifts, synchronous behaviour is observed at lower latitudes – for example the Asian monsoon – with evidence of global reorganisation of atmospheric circulation. The extent to which changes in East Asia are driven by sea surface temperature, wind regimes and changing coastal morphology during sea level change remains debated, but the key advantage here is that the U–Th chronology of this region is amongst the most intensively studied for this period, with uncertainties of only decades.
Time Window 2: Chronology Across the Earth System at 40 ka
Quaternary geochronology cuts across the disciplines of geophysics, geochemistry, archaeology and palaeoclimatology to provide critical information about the past dynamics of the Earth system and human evolution. Here, we focus on a combination of numerical age estimates and correlative tie-points associated with events at ~41 ka: the demise of the Neanderthals, a global geomagnetic excursion, the Campanian Ignimbrite caldera-forming volcanic supereruption in the Mediterranean and rapid, large-amplitude swings in climate.
One of the most important questions about the evolution of our own species is the relationship between early modern humans, after their exodus from Africa at ~60 ka (based on the mutation rate of mitochondrial DNA), and more archaic humans. To investigate patterns and timing of migration, cultural or genetic exchange, and species contraction and disappearance, we need tight chronological constraints on the death and burial of associated organic material (e.g. bones) and cultural artefacts (e.g. cave art). Unfortunately, the period of 60 to 35 ka is near the limit of 14C dating techniques, and we are hindered by a combination of theoretical and practical barriers, including calibration and closed-system behaviour. Recent developments in 14C calibration and methodology and complementary evidence from chronostratigraphic markers and U–Th methods have provided improved constraints.
Radiocarbon dating is by far the most widely used radioisotopic chronometer for the Holocene and Late Pleistocene epochs and is based on the extent of radioactive decay from an assumed initial 14C concentration. However, the 14C timescale must be calibrated using independent chronometers because of past variations in 14C production in the atmosphere by as much as a factor of two during the last glacial period. This variation is a function of secular changes in helio- and geomagnetic intensity and the partitioning of 14C in the various reservoirs of the ocean–atmosphere–terrestrial system. Overlapping annually banded tree ring records of 14C provide a detailed curve to 12.55 ka. However, extension to the practical 14C dating limit, which is 55–60 ka after consideration of detection limits, backgrounds and reproducibility, relies on (1) organic and inorganic forms of calcium carbonate, including corals, marine foraminifera and cave calcites, each corrected for reservoir age (or initial conditions), and (2) organic carbon, such as leaves and twigs in sedimentary archives.. Over the course of the last decade, broad consensus has emerged regarding past 14C variation, and the international working group INTCAL periodically reports updated calibration curves for widespread use after careful scrutiny. Crucially, calibration efforts rely on a combination of independent numerical ages (U–Th, varve counts) and correlative ages (based on comparison with alternative dated climate records). This is not an optimal solution, and further refinements will emerge as new data with tighter constraints on the atmospheric signal come to the fore.
While secular variation in atmospheric 14C might be considered an unavoidable obstacle to chronological efforts, it provides valuable information about the state of the Earth's geomagnetic field strength, which is a useful global chronostratigraphic tool itself. The most outstanding feature in the atmospheric 14C concentration curve is a large peak representing a >50% increase between 43 and 40 ka, coincident with large-amplitude peaks in the flux of the cosmogenic isotopes 10Be and 36Cl in polar ice cores (Fig. 3). The cause of these peaks is the reduced shielding of the Earth by its magnetic field during the Laschamp geomagnetic excursion, a field-intensity low originally observed in a lava flow near a village of this name in the Massif Central, France. The precise timing, magnitude and duration of this event are still subject to investigation from an empirical and theoretical stance. For example, the duration is important for the possible explanations for the differences between magnetic excursions and reversals, where the former might be related to reversals in the fluid outer core alone. Data constraints are provided by sparsely distributed volcanic fields that trap the magnetic signal at high resolution, combined with lower-resolution sedimentary archives. The best current estimate for the Laschamp excursion is 40.7 ± 0.9 ka, based on a combination of 40Ar/39Ar, unspiked K–Ar ages and 230Th/238U data from three lava flows in France and one in Auckland, New Zealand (Singer et al. 2009). This age also agrees with the peak of the atmospheric 14C increase observed in organic material in the Lake Suigetsu sediment core, Japan (Bronk Ramsey et al. 2012). Ultimately, it is envisaged that the Laschamp excursion will become sufficiently well constrained in terms of both age and duration that it will become a robust and precise tie-point with accuracy and precision at the subcentennial scale. This is particularly important in sediment records that are lacking material suitable for radioisotopic age determination.
Attempts to constrain ages at the limit of 14C dating have been improved greatly by the calibration efforts discussed above, but also important is the isolation of the original geochemical signal in organic remains, assuming closed-system behaviour since the time of death. In archaeological contexts, the dating of bone is extremely important, be it human, animal with cut marks, or cultural artefacts, but providing a reliable 14C age is challenging. Successive methodological advances in separating collagen and associated biomarkers from bone are consistently yielding older ages, suggesting earlier determinations were subject to the effects of contamination. This has resulted in a reassessment of numerous sites. For example, a maxilla (jaw bone) from what is currently thought to have been an anatomically modern human, found in Kents Cavern, UK, was originally estimated to be ~36 ka old, but its age is now considered to be >41 ka, based on measurements after application of ultrafiltration methods (Higham et al. 2011). This pushes back the dates for the dispersal of modern humans into the new world of Europe and demands much faster dispersal rates across Europe after arrival from Africa at ~60 ka.
An alternative approach to dating human presence and artistic culture is determining the age of cave art. Again, this is fraught with difficulties because of the potential for contamination and post-depositional alteration, but recent attempts using U–Th methods are extremely encouraging. Thin layers of calcite covering various stages of artistic development in caves of the Iberian Peninsula (Fig. 4) have yielded U–Th ages of >41 ka, pushing back the age of the oldest art in Europe and leaving open the possibility that symbolic expression was not restricted to modern humans but may have been adopted earlier by Neanderthals (Pike et al. 2012).
Additional chronological constraints used for archaeological and climate records during the late Pleistocene are based on volcanic tephra and their eruption ages. These may have only a regional fingerprint in sedimentary archives but are proving to be critical in establishing links between terrestrial, ocean and ice core records (e.g. the INTIMATE initiative; Blockley et al. 2012). Iceland-sourced tephras are particularly useful in the North Atlantic, and material from central and southern Italy, the Hellenic Arc and Anatolia have proved crucial in Mediterranean settings. Particularly important for the time window of interest here is the Campanian Ignimbrite tephra layer, with an 40Ar/39Ar age estimate of 41.1 ± 2.1 ka (2σ) (Ton-That et al. 2001). Crypto-tephra (volcanic ash layers that are not visible to the naked eye) associated with this eruption are emerging in disparate locations across the Mediterranean (Lowe et al. 2012) (Fig. 4c). Researchers use these layers to test hypotheses such as whether the combined influence of widespread ash and a synchronous Northern Hemisphere cold climate caused the demise of the Neanderthals. These chronostratigraphic markers – globally well-mixed gases, geomagnetic intensity variations and volcanic tephra – are fundamental to understand phasing relationships across the Earth system. The key for the future is to provide increasingly precise and robust constraints on their timing using radioisotope ages.
Time Window 3: The Combination of Isotope, Orbital and Magnetic Chronometers in the Middle Pleistocene
Establishing precise and accurate timescales of environmental change during the Early to Middle Pleistocene (2.6 to 0.126 Ma) is challenging and has a long history of revision based on a combination of astronomical, geomagnetic and isotope chronometers. When a signal response can be unequivocally ascribed to changes in insolation forcing, records can be aligned and tuned to provide orbital chronologies based on a priori assumptions about phasing relationships. However, important caveats are associated with this method because of the potential for diachronous response, problems with interpolation between tie-points, and tuning to parameters using phasing relationships that are themselves under scrutiny by the palaeoclimate research community.
A host of strategies are available for confirming the accuracy of timescales, and critical among them is independent corroboration between isotope chronometers (see also Schmitz and Kuiper et al. 2013 this issue, for older examples). As illustrated in the preceding vignette (~40 ka), the role of independently dated tie-points (or “golden spikes”) and correlation via volcanic tephra and geomagnetic reversals or excursions is extremely important, and a key “golden spike” for the Quaternary is the age of the Matuyama-Brunhes (M-B) geomagnetic reversal. Channell et al. (2010) provide the most precise and arguably the most accurate age for the M-B boundary, that is, 773.1 ± 0.8 ka (2σ), determined on the basis of astrochronology and magnetic records from five widely dispersed North Atlantic Ocean sediment cores (Fig. 1). This astronomically derived age compares well with an 40Ar/39Ar age for the M-B boundary recorded in lava flows from Maui (Coe et al. 2004); the 10Be spike evidenced in the EPICA Dome C ice core, East Antarctica (Raisbeck et al. 2006); and associated prominent lows in geomagnetic intensity from a stacked ocean sediment record (PISO-1500; Channell et al. 2009) (Fig. 1).
The M-B reversal provides a useful tie-point for ocean and terrestrial records of the Early and Middle Pleistocene, but we require continuous records with a high density of radioisotopic ages to test astronomically tuned chronologies for this time period. Much of the Middle Pleistocene period is within the range of U–Th dating methods (where the oldest age estimates possible depend on the U concentration and instrumental procedure, but can be as high as ~750 ka); thus corals and speleothems from this time period may offer absolutely dated climate markers. Corals formed during high sea stands offer excellent opportunities to constrain ice-volume change, but they are limited to short periods of the Middle Pleistocene and often experience disturbance of the U–Th isotope system, compromising age determinations in all but a few cases (Stirling and Andersen 2009). Speleothem calcite, however, holds excellent promise because it is often well preserved and provides unambiguous stratigraphy (Cheng et al. 2009) (Fig. 1). In the future, we also expect to see the presentation of similar records based on U–Pb techniques, which can improve upon U–Th disequilibrium methods where applied to material older than 500 ka. U–Pb-dated continental records of glacial–interglacial change based on δ18O in speleothem calcite offer much promise, and Bajo et al. (2012) illustrate a chronology for a sample from Corchia Cave, Italy, with uncertainties less than ±5 ka (2σ). This sample preserves a record of glacial terminations to beyond 750 ka and offers an excellent opportunity for comparison with ice and ocean-sediment cores. We expect speleothem records to become extremely important to “plug the gap” in numerical age constraints for sedimentary climate archives between the M-B reversal and the U–Th-dated records of the Late Pleistocene (Fig. 1).
We have focussed here on state-of-the-art provision of numerical ages for records of Quaternary climate and human evolution, some one hundred years after the pioneering work by Penck and Brückner (1909) on the chronology of European glaciations. We have provided examples of chronological constraint with local, regional and global importance. Where global chronostratigraphic markers are apparent, they are extremely useful, but they are few and far between and are subject to intense scrutiny and inevitable controversy. Often, one has to rely on site-specific or, at best, regional tie-points, and derive ages for chronologies based on multiple chronometers, some with lower precisions and based on relative methods, to gain confidence in the ages ascribed. Also, where continuous records are involved, one must apply age–depth models to infer ages of material deposited between tie-points. The tools of event stratigraphy, age–depth modelling and correlation are becoming ever more prevalent as we continue to increase the resolution of past environmental records. Age estimates of any materials have their own ”half-life,” or measure of longevity for providing salient information, which depends on the quality of the material under investigation, its context and preservation, understanding of the geochemical history, an accurate reflection of the uncertainties involved, and full documentation of all information that has been used in its derivation. We put emphasis on the last of these because the adjustment of fundamental parameters or algorithms used to calculate ages is inevitable and continual update of data is required. By focussing on the legacy of chronological data, all syntheses of past Earth system states will be much better served.
Postscript: What has become of Penck and Brückner's (1909) chronology of glacial advance in the Alps? Global correlation of such regional climate events is no longer attempted, but their pattern, timescales and extent remain a major focus. Dating glacial landforms, however, is challenging and even recent results offer only tentative conclusions. In a recent “multidating” study of the Most Extensive Glaciation in Switzerland by Dehnert et al. (2010), ages from 170 to 140 ka, including luminescence ages on a sequence of deltaic foreset beds, cosmogenic 26Al/10Be burial ages on quartz, and U–Th ages of a carbonate crust, are correlated with the Riss glaciation of Penck and Brückner.
The authors have benefitted from wider discussions at meetings associated with INTCAL, INTIMATE, ESF-Earthtime and PALSEA, a PAGES/WUN working group. We are grateful for the suggested improvements to the manuscript by Brad Singer and an anonymous reviewer and the thorough editorial guidance of Dan Condon, Mark Schmitz and John Valley.