Timing and dynamics of the last deglaciation from European and North African δ¹³C stalagmite profiles—comparison with Chinese and South Hemisphere stalagmites

Abstract

The last deglaciation and its climatic events, such as the Bølling–Allerød (BA) and the Younger–Dryas (YD), have been clearly recorded in the δ¹³C profiles of three stalagmites from caves from Southern France to Northern Tunisia. The three δ¹³C records, dated by thermal ionization mass spectrometric uranium–thorium method (TIMS), show great synchroneity and similarity in shape with the Chinese cave δ¹⁸O records and with the marine tropical records, leading to the hypothesis of an in-phase (between 15.5 and 16 ka ∼±0.5 ka) postglacial warming in the Northern Hemisphere, up to at least 45°N. The BA transition appears more gradual in the speleothem records than in the Greenland records and the Allerød seems warmer than the Bølling, showing here close similarities with other marine and continental archives. A North–South gradient is observed in the BA trend: it cools in Greenland and warms in our speleothem records. Several climatic events are clearly recognizable: a cooler period at about 14 ka (Older Dryas (OD)); the Intra-Allerød Cold Period at about ∼13.3 ka; the YD cooling onset between 12.7 and 12.9±0.3 ka. Similar to the BA, the YD displays a gradual climate amelioration just after its onset at 12.75±0.25 ka, up to the Preboreal, and is punctuated by a short climatic event at 12.15 ka. Even though the Southern Hemisphere stalagmite records seem to indicate that the postglacial warming started about ∼3 ka±1.8 ka earlier in New Zealand (∼41 °S), and about ∼1 to ∼2 ka earlier in South Africa (24.1 °S), large age uncertainties, essentially due to slow growth rates, make the comparison still perilous. The overall δ¹³C speleothem record seems to follow a baseline temperature increase controlled by the increase in insolation and punctuated by cold events possibly due to the N-America freshwater lake discharges.

Introduction

The last deglaciation is punctuated by several climate events (i.e. transitions toward warm periods like the Bølling–Allerød or abrupt cooling such as the OD and YD), whose timing, amplitude, and distribution on the Earth are of primary importance if we want to understand their causes. It is certainly the best documented period of major climatic changes due to high resolution records from the ice core records (Alley et al., 1993; Dansgaard et al., 1993; Thompson et al., 1995; NorthGRIPmembers, 2004; EPICA, 2004), the ocean core records (Flower et al., 2004; Hendy et al., 2002; Hughen et al., 1995; Waelbroeck et al., 2001), the continental marine records (Turon et al., 2003) and the lake records (Allen et al., 1999; Grafenstein Von et al., 1999; Magri and Sadori, 1999; Stager et al., 2002; Zolitschka, 1992). Despite all these, there are still some uncertainties about the exact origin of the climate events of this period (i.e. the abrupt YD onset, the Older Dryas, the Inter Allerød Cold Period), the difference in the onset of deglaciation between the Northern and Southern Hemispheres, the occurrence and amplitude of the YD in the Southern Hemisphere, the climate variability during events such as the BA or the YD, etc. Difficulties in obtaining accurate absolute ages is certainly the main obstacle to resolving these questions : ¹⁴C ages are complicated by the occurences of ¹⁴C plateaus during the deglaciation and by dead carbon fraction uncertainties ; ice core layer counting appears difficult during cold and low accumulation periods like the YD; and finally, climatic signals are extremely various, from water stable isotopes to pollen in marine cores, that any comparison has to seriously consider the complexity of each proxy. It is likely that, depending on geographical location, latitude, and distance from the ocean, the deglaciation generated different imprints in palaeoclimatic records. Consequently, in order to get a better pattern of this period, the study of many geographically widespread records is necessary. Indeed, the goal behind an improved understanding of the BA/YD events is also to improve our understanding of the Dansgaard–Oeschger events that occurred during the last glacial period and whose origin might be similar (Blunier et al., 1998; Bond et al., 2001; Rahmstorf, 2003). In both cases, there are still questions about the origin of such sudden climatic shifts. Although the idea that large inputs of fresh water due to the melting of ice caps and discharge of water stored in huge North American lakes strongly reduced the North Atlantic ocean circulation is still the consensual hypothesis for events such as the YD, other hypotheses, involving a bipolar seesaw of a complex ocean-ice-atmosphere system (Blunier and Brook, 2001), stochastic resonance in an ocean-atmosphere climatic model (Ganopolski and Rahmstorf, 2002) or external forcing like solar activity variation are also evoked (Bond et al., 2001; Goslar et al., 1999; Hughen et al., 2000; Rahmstorf, 2002; Renssen et al., 2000; van Geel et al., 2003). The uncertainty in the causes is essentially due to the lack of accurate and absolute dated records, and to the uneven distribution of these records, especially on the continent (besides polar ice core archives), in the Southern Hemisphere and in low latitudes.

Well dated records of the last deglaciation on the European continent are rare, especially those whose dating is absolute thanks to annual layers counting or uranium–thorium dating and which possess a continuous record. Among them, there is the Lake Monticchio record where varve counting coupled with ¹⁴C measurements on organic matter allowed pollen–climate reconstructions of the last 102 ka (Allen et al., 1999). The last deglaciation and the YD are clearly recorded in this record, although with a low resolution due to pollen sampling constraints. Some stalagmites from Germany and Austria have partly recorded the last deglaciation period, but there was either a hiatus during the glacial-interglacial transition (Sauerland caves, western Germany) (Niggemann et al., 2003), or low resolution which prevented a good time constraint through this period (Hölloch Cave stalagmite where the Younger-Dryas and the Bølling Allerød occupy a few millimeters and where only two U/Th dated points could be made; (Wurth et al., 2004). The continuous Soreq Cave record (Israel) has brought precious information about the millennial scale climatic variation of the last 250 ka, as well as the Sapropel events linked to changes in the humidity and temperature of the Mediterranean Sea; the last deglaciation period was also recorded here in a composite record but the resolution for this particular period needs to be improved (Bar-Matthews et al., 2000; Bar-Matthews et al., 1999, Bar-Matthews and Ayalon, 2003b). A very recent study shows that the last deglaciation was recorded in a 16.5 ka continuous stable isotopic record of a stalagmite from the Savi Cave (South-Eastern Alps of Italy) (Frisia et al., 2005). But here too, the transition from glacial to the beginning of the warming is not recorded and the transitions to the BA and to the YD are not clear because of a very slow growth rate during this period and very few U–Th ages (no age between 13.5 and 10.7 ka).

Outside of Europe there are also few speleothem records that cover the last deglaciation or a part of it : a stalagmite from the Onondaga Cave (Missouri, US) recorded a part of the Allerod-YD period in its stable isotopic profiles. Its very short growing period, about 0.8 ka long, and the uncertainties in the chronology do not allow any solid comparison with other records (Denniston et al., 2001). Chinese caves have recently brought well U/Th dated speleothem records: Hulu Cave (or Tangshan Cave) in East China and Dongge Cave in South China which both show striking similarities with the Greenland ice core records (Dykoski et al., 2005; Wang et al., 2001; Zhao et al., 2003). In the Southern Hemisphere, the New Zealand speleothems (∼41 °S) have recorded the last deglaciation linked with changes in the convergence of subtropical and sub-Antarctic waters (Hellstrom and McCulloch, 2000; Hellstrom et al., 1998; Williams et al., 2004). South Africa speleothem records also display a clear pattern of the last deglaciation in an area where other palaeoclimate proxies are very rare; they have shown that the temperature changed by about 6 °C during this period (Holmgren et al., 2003; Talma and Vogel, 1992). From the above examples, it appears that speleothems can unravel the timing and the climatic structure of the last deglaciation in specific areas and thus help to better understand them: the chronology that relies on the U–Th analyses avoids the problems inherent to the ¹⁴C dating methods and the climatic signal extracted from the calcite stable isotopes is relatively well understood. But, up to now, there have been no such speleothem records in Europe that allow to decipher climatic events in the BA or YD periods, the timing of the beginning of the warming following the pleniglacial. The results we present in this study show a continuous speleothem record of the last deglaciation, and of the associated events such as the YD and the BA) on a NW-SE transect, in three different sites (Fig. 1): Villars Cave (SW-France, 45.30 °N, 0.50 °E, 175 m asl; Vil-stm11 stalagmite), Chauvet Cave (S-France, 44.23 °N; 4.26 °E; 240 m asl; Chau-stm6 stalagmite) and La Mine Cave (Central Tunisia , 35 °N, 9.5 °E, 1000 m asl; Min-stm1 stalagmite) (Fig. 1). The chronology was constructed through 29 thermal ionisation mass spectrometric uranium-thorium datings and climatic variations were characterized by 418 δ¹³C and δ¹⁸O measurements made along the growth axis of the stalagmites. All these sites are situated relatively close to the North Atlantic bassin and thus have likely been influenced by any changes in this key area (i.e. ocean circulation, ice-sheet flooding dynamics, ocean–atmosphere interactions). The novelty of this study is also the use of the calcite δ¹³C instead of the calcite δ¹⁸O as a palaeoclimatic signal, which here appears much less variable from one site to another and also much more coherent when compared with the other records.