Trabaque tufa record
Trabaque Canyon (40.36° N; 2.26° W; 840 m above sea level) is located in the central Iberian Peninsula (Fig. 1). At this site, tufa deposits precipitate as freshwater carbonates downstream of overflow karst springs. During the last interglacial period, tufa precipitated continuously at the studied site while water level of the aquifer was high enough for upstream springs to discharge13. Outcrops of the studied tufa deposit are preserved in the margins of Trabaque River over a distance of 500 m downstream of overflow karstic springs. The studied tufa deposit is 12 m thick, with a gentle ramp morphology, and a simple stratigraphy of sub-horizontal tufa beds that covered the full section of the narrow canyon. The accumulation of tufa created a small lake upstream the ramp, which prevented erosive events while the deposit was active, because most of the river bedload was accumulated in the basin of the lake. This configuration favoured the lack of erosive episodes in the tufa and the deposition of a continuous record. The tufa deposit was partially eroded by subsequent fluvial incision once the tufa accretion ceased and detrital sediments filled the lake basin and started to flow over the ramp during floods. The tufa deposit is mostly composed of well-cemented intra-clastic and peloidal carbonate particles13. The deposit comprises tufa beds 0.02–1 m thick that typically extend tens of metres downstream. At the base of the section, the tufa lies over loose fluvial sediments of sandy silt, whereas at the top of the section there is an erosive scar, and recent gravitational deposits overlay the tufa preserved in the slopes of the canyon.
The base of the deposit section is characterized by nearly 4 m of tufa sediments in the centre of the valley, laterally interdigitating with gravitational deposits towards the slopes (Fig. 1). These gravitational deposits partially invaded the bottom of the valley during three distinct pulses. These gravitational deposits occurred during periods of enhanced slope processes due to the decrease in vegetation cover on the canyon slopes during prolonged dry periods. The evidence of local erosion recorded by the gravitational deposits is consistent with other proxies that record local and regional erosion and that are displayed in Fig. 2. Thus, independent evidence of erosion is also recognized from the increase of insoluble residue (IR) particles in the tufa, recorded by the percentage of silt IR. IR particles were transported to the tufa by the river or by the action of wind. The increase of these particles in the tufa is interpreted as enhanced erosion, not only from the catchment but also from outside the basin. Higher concentrations of Si and Al are also interpreted as proxies of soil erosion from areas with silicate substrates inside or outside the catchment. The increase of micro-charcoal particles in the tufa is also interpreted as a sign of enhanced soil erosion. Charcoals were incorporated to the tufa during floods or transported by the wind after the occurrence of fires, as well as from the erosion of soils that accumulated charcoals from previous fire events. In any case, the increase of micro-charcoals in the tufa record suggests soil erosion due to the lost of vegetation cover. Major events of local and regional erosion occurred synchronously (Fig. 2), supporting that the common decreases in vegetation cover that resulted in erosion events were related to periods of reduced precipitation.
The middle section of the tufa deposit (i.e., 4–11 m from the base), does not record gravitational deposits in the centre of the valley. In agreement with this retreat of the gravitational deposits to the slopes of the canyon, other proxies of erosion record minimum values along most of the middle section of the tufa. This supports the expansion of forest and/or vegetation cover in the region, which protected soils from erosion. The top section of the tufa (i.e., 11–12 m from the base) records an alternation of well-cemented and loose tufa beds (Fig. 1). The loose beds were formed when karst groundwater discharge upstream of the studied site occurred as overflow instead of permanent springs, as a response to a drop in the level of the aquifer14, which limited the formation of cements. The decrease in precipitation that caused this drop of the groundwater level was also responsible for the progressive increase in percentage of Pinus sylvestris/nigra, a tree with high tolerance to cold and dry conditions (Fig. 2). Therefore, the tufa deposit records periods of significant hydrological deficit in the system at the base and the top of the section.
The tufa δ18O and δ13C values have a clear response to climate during the early phase of deglaciation, while forest cover in the Iberian Peninsula was variable. However, once the forest cover expanded in the Iberian Peninsula15, additional factors in the water and carbon cycles complicated the link between climate and these proxies. The δ18O oscillations recorded at the base of the section, are interpreted as changes in the amount of precipitation (i.e., more negative δ18O values were recorded with increased precipitation). During this period, amount of precipitation and the ratio of recycled precipitation, two main controls on the δ18O values of precipitation in the Iberian Peninsula16, co-varied due to the positive feedback between precipitation and forest coverage, which determined the ratio of recycled precipitation. Once forest expanded in most of the Iberian Peninsula, limited changes in forest cover caused the ratio of recycled precipitation to be independent of the amount of precipitation, and recycled precipitation instead of amount of precipitation dominated the δ18O signal17. Temperature is not a significant control on Trabaque tufa δ18O values at inter-annual timescales, because of isotope fractionation at the time of raindrops formation and atmosphere equilibration is counteracted by a similar isotope fractionation of opposite sign when calcite precipitates18,19. On the other hand, the large δ13C oscillations recorded at the base of the record were controlled by the variability of CO2 degassing (i.e., more negative δ13C values occurred during periods of increased precipitation that raised the level of the aquifer and limited the air space for degassing). Once the forest cover stabilized in the catchment area, enhanced soil activity increased the importance of biological controls on the carbon cycle and limited an unequivocal interpretation of the δ13C signal. The base of the tufa deposit records three δ18O and δ13C oscillations that represents periods of drier and wetter conditions. Drier conditions according to the isotope records occurred at the time of enhanced local and regional erosion recorded in other Trabaque proxies. The replication of climate and environmental signals from multiple proxies supports the robust interpretation of the record.
A chronological study based on radiometric dates from this deposit confirmed that the tufa was mostly formed during the last interglacial period13, although its uncertainty prevents discernment of the age of events at a millennial timescale. Since the morpho-stratigraphical evidence supports that Trabaque sequence records a continuous time series, we use the age model of Corchia speleothems, located in central Italy8 to improve Trabaque chronology. This synchronization was conducted using δ18O anomalies that are clearly identifiable in both records (Supplementary Figs. S1–Fig. S4; supplementary Table S1). We compare the Trabaque record to a key ocean sediment core (ODP 984) located south of Iceland (61.25° N; 24.02° W) since this location is particularly sensitive to record ice rafted debris (IRD) events during the deglaciation11. We also synchronize the age model of ODP984 record to the Corchia chronology (Supplementary Figs. S5–S6; supplementary Table S1), based on the strong connection between the Mediterranean climate and the North Atlantic during periods of ice-sheet instability20.
First stage of T-II
During a first stage of T-II, three millennial oscillations are recorded in the North Atlantic and the Mediterranean (Fig. 3). In Corchia Cave, less negative speleothem δ18O values during these oscillations represent drier conditions8, that correspond with dry and erosive conditions in Trabaque record, and the deposition of IRD events at ODP 984 site. According to an independent synchronized chronology6, the first meltwater pulse of the T-II (MWP-2A), occurred during the cold/arid period of the first millennial oscillation, at the time when the integrated solar insolation in the Northern Hemisphere started to rise7. The initial ablation of Northern Hemisphere continental glaciers provided freshwater to the ocean surface and enhanced the halocline in the high latitudes of the North Atlantic, while the North Atlantic Current still flowed into the Nordic Seas10. The sharp salinity gradient in surface waters of the Nordic Seas caused salty waters from the North Atlantic Current to flow under the halocline, preventing the release of heat from those relatively warm waters to the atmosphere. This decoupling of the North Atlantic Current from the atmosphere enhanced the cold conditions in the high latitudes of the Northern Hemisphere, triggered the formation of ice rafted debris (IRD) events in the North Atlantic21, and the related decrease of precipitation in the Mediterranean region. Two periods of enhanced glacier ablation recorded south of Greenland22, are likely related to episodes of enhanced IRD deposition at ODP 984 site subsequent to MWP-2A and arid/erosive conditions recorded in the Mediterranean region. The cold climate in the Nordic Seas, caused by the decoupling of the North Atlantic Current from the atmosphere, led to a negative feedback that progressively reduced the rate of glacier ablation and eventually disrupted the stratification of surface ocean waters. Without a significant stratification in the Nordic Seas, heat was released from the North Atlantic Current to the atmosphere in the region, which enhanced the ablation of continental glaciers and increased the input of freshwater to the ocean. These interactions resulted in climate oscillations at millennial scale. During this first stage of the deglaciation, no shutdown of the Thermohaline Circulation was recorded23. The mechanism causing these millennial climate oscillations clearly originated from modifications of the ocean–atmosphere interactions at high latitude of the North Atlantic. The impact of these anomalies in the ocean circulation seems to be limited to the uppermost section of the water column as recorded by planktonic assemblages11, while deep or intermediate waters kept their flow10. Therefore, these anomalies were not propagated by the ocean circulation to the Southern Hemisphere23 and consequently they differ from other millennial scale climate interactions that involve the BSM such as Heinrich or Dansgaard-Oeschger events.
Second stage of T-II
The second stage of T-II started with the onset of a fourth IRD event recorded at ODP 984 site at 134.5 ± 0.9 ka BP. This IRD event, known as Heinrich event 11 (H11)24, was much larger than previous IRD events of T-II, although its onset was still in phase with the pace of the millennial oscillations. H11 significantly contributed to the larger meltwater pulse within T-II (MWP-2B)6. However, the glacier decay that resulted from the progressive increase of integrated summer insolation in the Northern Hemisphere7 was disproportionate to the magnitude and duration of H11 and MWP-2A, even accounting for the enhanced freshwater supplied to the North Atlantic by ice-sheet decay that resulted from the millennial oscillations. An event or a surpassed threshold in deglaciation occurred that triggered the observed non-linear response. Ocean sediments from Bermuda Rise recorded an increase in the flux of clay from Canada just before the shutdown of the Thermohaline Circulation25. This suggests a significant outburst from North American proglacial lakes, a scenario supported by climate models26. The outburst occurred in phase with the enhanced Northern Hemisphere ice-sheet decay related to the millennial oscillations, which controlled its timing. The large amount of freshwater released to the North Atlantic in relation to the H11 caused the collapse of Thermohaline Circulation23 and triggered the BSM that forced the Southern Hemisphere to react for the first time since the onset of T-II.
The activation of the BSM initiated the second stage of T-II. The erosion in the Trabaque catchment during H11 was limited, especially during the first half of this cold period (Fig. 2; Supplementary Fig. S7). So, unlike previous IRD events, H11 was not a particularly dry event in the Iberian Peninsula. During H11, Trabaque deposit recorded its lowest growth rate (Supplementary Fig. S4). At this time, the North Atlantic off the Portuguese margin had a winter temperature drop of 6 °C27. The continental and mountainous location of Trabaque Canyon results in nowadays mean temperature of the coldest month (December) < 3 °C. Therefore, the drop of temperature during H11 likely resulted in seasonal freezing of the lake surface upstream the studied site, and limited the growth rate of tufa during the H11.
The suppression of the Thermohaline Circulation caused the expansion of the Antarctic Bottom Water through the North Atlantic basin23. Temperature in Antarctica28 started to increase once the BSM was activated, and ablation of glaciers from the Southern Hemisphere contributed in part to the MWP-2B (Fig. 4). The freshwater released around Antarctica enhanced wind activity and caused a northward shift of the Antarctic Polar Front that favoured the upwelling of deep water from the Southern Ocean and the release of CO2 to the atmosphere12,29. When H11 finished, Northern Hemisphere continental glaciers outside Greenland were mostly ablated, MWP-2B ended, the Thermohaline Circulation was re-established and most of the Northern Hemisphere had a climate typical of an interglacial period2,23,30. At this time, Trabaque tufa recorded limited regional erosion because forest dominated the landscape (Supplementary Fig. S8). As a result of the BSM, the deglaciation in the Southern Hemisphere continued during two more thousand years before T-II was completed.
The role of millennial climate oscillations on terminations
Previous studies have pointed out that millennial oscillations probably played an important role in terminations5,31. The occurrence of H11 was in-phase with the millennial oscillations that dominated the climate of the North Atlantic and the Mediterranean, which constrained the time frame for the trigger of the BSM. The climate interactions that operated during the first stage of T-II, (i.e., millennial oscillations of heat release to the atmosphere at high latitudes of the Northern Hemisphere from the North Atlantic Current), were not observed during Termination I (T-I)30, because of North Atlantic Current did not flowed into the Nordic Seas during T-I10 and the rates of insolation change were very different7. The heat transported by the ocean into high latitudes triggered the millennial climate oscillations recorded during the first stage of T-II as a result of the coupling/decoupling of ocean–atmosphere heat fluxes in the Nordic Seas. This particular climate mechanism was absent during T-I, even if the structure of T-I was also dominated by millennial climate oscillations (e.i., Mistery Interval-Bølling/Allerød-Younguer Dryas). The millennial oscillations of T-I were controlled by different climate mechanisms and resulted in an early BSM that caused both hemispheres to respond synchronously at a millennial timescale32,33. The sequence of events and the evolution of the sea level rise clearly differed in the last two terminations, which is reflected in their structure, duration and timing23,30,34,35. Thus, together with the variable rates of insolation change, the climate mechanisms that control the millennial oscillations could account for the different structure observed between T-I and T-II2. The structure of T-II is very similar to other terminations such as T-IV, T-V or T-VI9,36, suggesting that the operation of millennial oscillations controlled by similar climate mechanisms as during the first stage of T-II could have occurred during those deglaciations. Therefore, we suggest that the difference in structure between terminations and the precise timing of the major events in the deglaciations could be controlled not only by the rates of insolation change, but also by the climate mechanisms behind the millennial oscillations within the terminations, which characterizes different modes of deglaciation.
Inter-hemispheric asynchrony of stages during T-II
Description of two stages or pulses were previously reported during T-II6,10,34,35,37, although these stages are defined by different events and consequently they are often not equivalent stages. We define two stages during T-II easily identified in marine and continental records of the North Hemisphere. The duration as well as the onset and demise of T-II were not synchronous in both hemispheres. The Southern Hemisphere only recorded the second stage of T-II, and the clear evidence of regional deglaciation started 7 ka after the onset of T-II. However, the end of deglaciation during T-II lasted 2 ka more in the Southern Hemisphere, while in the Northern Hemisphere full interglacial conditions were already established. The sequence of events here reported shows that the first stage of T-II was initiated with the progressive decay of the large ice-sheets in the Northern Hemisphere responding to the increase in summer insolation. However, the deglaciation did not occur at a steady rate, and complex ocean–atmosphere interactions in the high latitudes of the Northern Hemisphere caused millennial climate oscillations that paced the decay of ice-sheets. These millennial oscillations controlled the timing of a large glacier outburst that triggered the H11, and the collapse of the Thermohaline Circulation that initiated the BSM at the onset of the second stage of T-II. Although most drastic deglaciation events occurred with the onset of the second stage of T-II, the termination already started 7 ka earlier. To understand the complete sequence of events that enabled the completion of T-II, it is essential to cover not only the most significant changes of the deglaciation, but also those smaller oscillations that eventually were responsible of triggering the larger climate changes.