Search

Quaternary interglacials provide key observations of the Earth system's responses to orbital and greenhouse gas forcing. They also inform on the capabilities of Earth system models, used for projecting the polar ice-sheet and sea-level responses to a regional warmth comparable to that expected by 2100 C.E. However, a number of uncertainties remain regarding the processes and feedbacks linking climate, ice-sheet and sea-level changes during past warm intervals. Here, we delineate the major research questions that need to be resolved and future research directions that should be taken by the paleoclimate, sea-level and ice-sheet research communities in order to increase confidence in the use of past interglacial climate, ice-sheet and sea-level reconstructions to constrain future predictions. These questions were formulated during a joint workshop held by the PAGES-INQUA PALSEA (PALeo constraints on SEA level rise) and the PAGES-PMIP QUIGS (QUaternary InterGlacialS) Working Groups in September 2018.

The Arctic Ocean is a siliciclastic depositional environment which lacks any rock-forming biogenic calcareous and siliceous components during large parts of its Quaternary history. These hemipelagic sediments are nevertheless suitable for the study of organic-walled microfossils of which the fossil remains of dinoflagellates - dinoflagellate cysts - are the most important group. Dinoflagellate cysts have become an important tool in paleoceanography of the high northern latitudes, but their potential for Quaternary biostratigraphy has remained largely unexplored. Dinoflagellate cysts are the dominant marine palynomorph group which is more continuously present in the marginal seas (e.g. Barents Sea, Bering Sea) than in the Arctic Ocean itself throughout the Quaternary. Most species have long stratigraphic ranges, are temporary absent and show abundance variations on glacial-interglacial timescales. Of the more than 30 taxa recorded, only Habibacysta tectata and Filisphaera filifera became extinct in the Pleistocene. The highest persistent occurrence of H. tectata at ca. 2.0 Ma and the top of F. filifera acme at ca. 1.8 Ma can be used for supra-regional stratigraphic correlation between the Arctic Ocean and adjacent basins. These events corroborate a slow sedimentation rate model for the Quaternary section on the central Lomonosov Ridge, but a combination of different methods will have to be applied to provide a detailed chronostratigraphy. The occurrence of cysts of phototrophic dinoflagellates in certain stratigraphic intervals on Lomonosov Ridge supports published evidence of episodic opening of the multiyear Arctic sea ice cover during the Quaternary probably related to a stronger inflow of Atlantic water. This contradicts the hypothesis of a permanently ice covered central Arctic Ocean in the Quaternary.

•A synthesis of existing literature on Holocene climate change in Arctic Canada and Greenland.•Warmest-to-coldest millennium temperature change in the Holocene in Arctic Canada and Greenland is 3.0 ± 1.0 °C.•Thermal maximum conditions occurred earlier in high latitudes and at sites less influenced by North Atlantic circulation.

To investigate the mechanisms behind Holocene regional climate trends from north of 58°N in the North Atlantic-Fennoscandian region Principal Component Analysis (PCA) was performed and a temperature anomaly stack produced from 81 proxy derived summer temperature time series from 74 sites. The PC results show distinctly different trends for near-surface versus surface temperatures, demonstrating the importance of handling these separately. The first PC of weighted sea surface summer temperature time series and continental time series explains 45 ± 8% of the variance, where the uncertainty is the standard deviation of the distribution of variance explained across the 1000 age-uncertain ensemble members. PC1 has a relatively uniform expression over the whole region, closely following the summer insolation at 65°N. The second PC explains 22 ± 4% of the variance and shows a non-uniform expression, with loadings in opposite directions in the northern and southeastern parts of the region. Comparing the PC time series with model runs and with the timing of the demise of the Laurentide Ice Sheet (LIS), suggest that this pattern reflects both topographic and albedo effects of the LIS as well as release of meltwater into the North Atlantic and Arctic Oceans. Comparing the stack of gridded records with published global stacks reveals an unusual Holocene temperature development in the North Atlantic-Fennoscandian region most likely resulting from the location relative to the decaying LIS.

Sea surface temperature (SST), salinity and density gradients in the upper water column of the northwest Labrador Sea have been reconstructed based on high resolution analysis of a core (HU2008-029-004PC) spanning the last ∼36 ka, raised off Hudson Strait. The modern analogue technique was applied to dinocyst assemblages and combined with stable isotope data from Neogloboquadrina pachyderma left-coiled (Npl) for this purpose. Three oceanographic regimes were identified, broadly corresponding to the “glacial”, “deglacial” and “postglacial” intervals. The site remained under the direct influence of the Laurentide Ice Sheet (LIS) margin until the postglacial and did not record the Bolling-Allerod warming and weakly recorded the Younger Dryas event. The “glacial” regime lasted until ∼12.2 cal ka BP. It was characterized by generally low concentrations of dinocysts within an assemblage indicative of quasi-perennial sea ice. The “deglacial” regime (ca 12.2–8.3 cal ka BP) was marked by increased biogenic fluxes and more diversified dinocyst assemblages and possibly an enhanced subsurface inflow of North East Atlantic Deep Water. Warm summer (∼11 °C) but low winter (∼0 °C) sea surface temperatures, sea ice cover during about 3 months per year, and low summer salinity (∼28) suggest strong stratification in the upper water layer in relation to meltwater supply from the LIS. Following the final drainage of glacial Lake Agassiz through Hudson Strait, which is dated here at ∼8.3 cal ka BP, and the subsequent LIS collapse, increased summer salinity (up to ∼35) was accompanied by a reduced seasonal gradient of sea surface temperature from winter (∼3.8 °C) to summer (∼8.6 °C) suggesting enhanced penetration of North Atlantic Water. Weakened stratification of the surface water layer then allowed for winter convection and Labrador Sea Water formation, which is consistent with increased Npl-δ13C values in response to higher ventilation of the subsurface water layer.

Sea ice cover extent expressed in terms of mean annual concentration was reconstructed from the application of the modern analogue technique to dinocyst assemblages. The use of an updated database, which includes 1492 sites and 66 taxa, yields sea ice concentration estimates with an accuracy of ±1.1/10. Holocene reconstructions of sea ice cover were made from dinocyst counts in 35 cores of the northern North Atlantic and Arctic seas. In the Canadian Arctic, the results show high sea ice concentration (>7/10) with little variations throughout the interval. In contrast, in Arctic areas such as the Chukchi Sea and the Barents Sea, the reconstructions show large amplitude variations of sea ice cover suggesting millennial type oscillations with a pacing almost opposite in western vs. eastern Arctic. Other records show tenuous changes with some regionalism either in trends or sea ice cover variability. During the mid-Holocene, and notably at 6 ± 0.5 ka, minimum sea ice concentration is recorded in the eastern Fram Strait, northern Baffin Bay and Labrador Sea. However, this minimum cannot be extrapolated at the scale of the Arctic and circum-Arctic. The comparison of recent observations and reconstructions suggests larger variations in the Arctic sea ice cover during the last decades than throughout the Holocene.

Dinocysts occur in a wide range of environmental conditions, including polar areas. We review here their use for the reconstruction of paleo sea ice cover in such environments. In the Arctic Ocean and subarctic seas characterized by dense sea ice cover, Islandinium minutum , Islandinium ? cezare , Echinidinium karaense , Polykrikos sp. var. Arctic, Spiniferites elongatus–frigidus and Impagidinium pallidum are common and often occur with more cosmopolitan taxa such as Operculodinium centrocarpum sensu Wall & Dale, cyst of Pentapharsodinium dalei and Brigantedinium spp. Canonical correspondence analyses conducted on dinocyst assemblages illustrate relationships with sea surface parameters such as salinity, temperature, and sea ice cover. The application of the modern analogue technique permits quantitative reconstruction of past sea ice cover, which is expressed in terms of seasonal extent of sea ice cover (months per year with more than 50% of sea ice concentration) or mean annual sea ice concentration (in tenths). The accuracy of reconstructions or root mean square error of prediction (RMSEP) is ±1.1 over 10, which corresponds to perennial sea ice. Such an error is close to the interannual variability (standard deviation) of observed sea ice cover. Mismatch between the time interval of instrumental data used as reference (1953–2000) and the time interval represented by dinocyst populations in surface sediment samples, which may cover decades if not centuries, is another source of error. Despite uncertainties, dinocyst assemblages are useful for making quantitative reconstruction of seasonal sea ice cover.

Sea ice is an important component of the Earth system with complex dynamics imperfectly documented from direct observations, which are primarily limited to the last 40 years. Whereas large amplitude variations of sea ice have been recorded, especially in the Arctic, with a strikingly fast decrease in recent years partly attributed to the impact of anthropogenic climate changes, little is known about the natural variability of the sea ice cover at multi-decadal to multi-millennial time scales. Hence, there is a need to establish longer sea ice time series to document the full range of sea ice variations under natural forcings. To do this, several approaches based on biogenic or geochemical proxies have been developed from marine, ice core and coastal records. The status of the sea ice proxies has been discussed by the Sea Ice Proxy (SIP) working group endorsed by PAGES during a first workshop held at GEOTOP in Montréal. The present volume contains a set of papers addressing various sea ice proxies and their application to large scale sea ice reconstruction. Here we summarize the contents of the volume, including a table of various proxies available in marine sediments and ice cores, with their possibilities and limitations.

Geological records of past sea ice, such as those contained in Arctic marine sediments, offer an opportunity to strengthen our understanding of long-term sea ice variability, provided unambiguous paleo-sea ice proxies can be developed. One such recently proposed proxy is IP25, a highly branched isoprenoid alkene biosynthesized exclusively by sea-ice dwelling diatoms (Haslea spp.), which is well preserved in marine sediments and could be used to reconstruct past changes in spring sea-ice extent. However, little is known about regional-scale controls on IP25 production in sea ice, limiting its wider applicability as a paleo-sea-ice proxy. To address this issue we examined the distributions of IP25 and the marine productivity biomarkers dinosterol and brassicasterol in a suite of surface sediment samples distributed across the Arctic. We find a statistically significant, logarithmic relationship between IP25 and spring sea ice cover in sediment samples from arctic and subarctic sites in the Pacific (n = 96, R2 = 0.67, P

Deglaciation of the James Bay region was highly dynamic, with the occurrence of ice (Cochrane) readvances into glacial Lake Ojibway around final deglaciation time, which culminated with the drainage of Ojibway waters into Hudson Bay and subsequent incursion of the Tyrrell Sea at ∼8 ka. Renewed interest on these events comes from the possible link between the drainage of the ice-dammed Lake Agassiz-Ojibway and a major climate deterioration known as the 8.2-ka cooling event. Recent glaciological modeling suggests that this drainage may have occurred subglacially, a mechanism that can accommodate more than one lake discharge, as suggested by marine records. The exact number and timing of drainage events, as well as location of the lake discharge pathway(s) remain, however, largely unconstrained. Here we focus on the events that led to the drainage of Lake Ojibway by documenting late-glacial sedimentary sequences located east of James Bay. Our investigations indicate that the deglacial sequence consists of a readvance till, extensive Ojibway rhythmites, and thick marine sediments. The glaciolacustrine and marine units are separated by a 60 cm-thick horizon composed of laminated silt beds containing rounded clay balls and disseminated clasts resulting from the abrupt drainage of the lake. Radiocarbon dating of marine fossils lying above the drainage horizon indicates that the glaciolacustrine episode ended around 8128–8282 cal yr BP. Micropaleontological analyses reveal that freshwater ostracods ( Candona sp.) and marine microfossils (foraminifers, dinocysts) occur together in the upper part of the Ojibway sediments. Analysis of oxygen isotopes (δ 18 O) of ostracods and foraminifers originating from the same stratigraphic position show highly contrasting values that suggest possible subglacial exchanges between Lake Ojibway and Tyrrell Sea waters prior to the final drainage event. The complexity of the deglacial events is further indicated by radiocarbon dating of marine shells retrieved from a Cochrane till that suggests that the last ice readvance occurred almost simultaneously with the final lake discharge. These results bring additional constraints on the drainage mechanism of the coalesced Lake Agassiz-Ojibway and indicate that the James Bay region formed an important drainage pathway for meltwaters at the end of the last deglaciation.

A series of cores (box and piston) were collected at 2 key locations in Lancaster Sound (cores 2004-804-009 BC and PC) and Barrow Strait (cores 2005-804-004 BC and PC) in the Canadian Arctic Archipelago to document the evolution of sea-surface conditions in the main axis of the Northwest Passage during the Holocene time period. Reconstruction of sea-surface parameters (summer temperature and salinity, duration of sea-ice cover) were estimated based on transfer functions using dinoflagellate cyst assemblages as proxy indicators. The chronology of these cores is based on calibrated AMS-14C dates, 210Pb analyses, and correlations between paleomagnetic secular variations of the geomagnetic field and a predicted spherical harmonic model of the geomagnetic field (CALS7 K.2). Our age models for both cores indicate that 009 PC spans the last 11,100 cal BP, while 004 PC encompasses the last 10,800 cal BP. Calculated sedimentation rates are in the range of 43–140 cm/kyr for core 009 PC and 15–118 cm/kyr for core 004 PC, allowing for a millennial time scale resolution in each core. Our results indicate relatively harsh conditions in Lancaster Sound between 10,800 and 9000 cal BP (summer temperatures 2 °C cooler than at present), which we associate with the presence of active ice-streams in northernmost Baffin Bay and a weak West Greenland Current. This is followed by a warming trend (up to 3 °C higher than present) that took place between ∼8500 and 5500 cal BP, which we associate with the Holocene thermal maximum and to a large scale atmospheric pattern such as the Arctic Oscillation operating at the millennial time scale. This is concomitant with an increase in the relative abundance of phototrophic dinoflagellate cyst taxa. A gradual cooling of sea-surface temperature and increased sea-ice follow this from 5500 cal BP until the modern period. In Barrow Strait, harsh sea-surface conditions prevailed from 11,100 to 5500 cal BP, with summer temperatures as low as 4 °C cooler than at present. The warming trend occurred later in this region (between 5500 and ∼1000 cal BP), after which a gradual cooling is observed until the modern period. The apparent shift, or opposite warming trends, between Lancaster Sound and Barrow Strait after 5500 cal BP could be the result of a change in atmospheric circulation patterns related to a possible shift in the AO mode (from AO+ to AO−). Comparison of ice core δ18O record from Devon Island ice cap and the reconstructed sea-surface temperature from core 009 PC suggests a strong atmospheric-oceanic coupling throughout the Holocene in this area.

Proxy evidence suggests that a mean atmospheric state, reminiscent of the positive phase of the North Atlantic/Arctic Oscillation (NAO), persisted throughout the early Holocene and resulted in a dipole pattern in sea-ice concentration between the north-eastern and north-western North Atlantic. A dynamic thermodynamic coupled sea-ice–ocean model is used to simulate the sea-ice concentration and thickness in the Arctic during the early Holocene. It is forced with winds, ocean currents and surface air temperatures (SAT) from recent years with a positive phase of the NAO in conjunction with altered long and shortwave radiation and surface air/ocean temperatures. The simulation reproduces an east/west dipole in sea-ice cover of the Arctic and is compatible with reconstructed sea-ice conditions in the Chukchi Sea (inferred from dinoflagellate cyst analysis of ocean sediment cores). Sensitivity studies were performed to investigate the individual effects of radiation, ocean forcing, SAT and winds on sea-ice cover. Results show that in the East Siberian Sea, SAT is the dominant forcing for changes in sea-ice thickness, whereas winds and SAT are the dominant factors in sea-ice concentration anomalies. In the Barents Sea, sea-ice anomalies are influenced by air and sea surface temperatures, and ocean currents.

A new calibration database of census counts of organic-walled dinoflagellate cyst (dinocyst) assemblages has been developed from the analyses of surface sediment samples collected at middle to high latitudes of the Northern Hemisphere after standardisation of taxonomy and laboratory procedures. The database comprises 940 reference data points from the North Atlantic, Arctic and North Pacific oceans and their adjacent seas, including the Mediterranean Sea, as well as epicontinental environments such as the Estuary and Gulf of St. Lawrence, the Bering Sea and the Hudson Bay. The relative abundance of taxa was analysed to describe the distribution of assemblages. The best analogue technique was used for the reconstruction of Last Glacial Maximum (LGM) sea-surface temperature and salinity during summer and winter, in addition to sea-ice cover extent, at sites from the North Atlantic ( n =63), Mediterranean Sea ( n =1) and eastern North Pacific ( n =1). Three of the North Atlantic cores, from the continental margin of eastern Canada, revealed a barren LGM interval, probably because of quasi-permanent sea ice. Six other cores from the Greenland and Norwegian seas were excluded from the compilation because of too sparse assemblages and poor analogue situation. At the remaining sites ( n = 54), relatively close modern analogues were found for most LGM samples, which allowed reconstructions. The new LGM results are consistent with previous reconstructions based on dinocyst data, which show much cooler conditions than at present along the continental margins of Canada and Europe, but sharp gradients of increasing temperature offshore. The results also suggest low salinity and larger than present contrasts in seasonal temperatures with colder winters and more extensive sea-ice cover, whereas relatively warm conditions may have prevailed offshore in summer. From these data, we hypothesise low thermal inertia in a shallow and low-density surface water layer.

In Arctic and sub Arctic seas, shell growth and/or secondary calcite overgrowth of Neogloboquadrina pachyderma (left coiled)—Npl—occur along the pycnocline, and their δ13C and δ18O-values are size and weight dependent. However, whereas the Npl 18O data from the NW Atlantic indicate near-equilibrium conditions with ambient waters and a positive relationship between shell weight and 18O-content, assemblages from box-cored sediments of the Chukchi Sea (western Arctic) are depleted by ∼2‰ with respect to equilibrium values with modern conditions, and depict a negative relationship between shell weight and its δ18O-value (−0.15±0.03‰/μg on VPDB scale). A similar feature is also depicted by the dextral form of N. pachyderma (Npd). We associate the reverse shell-size or weight vs. δ18O relationship to the reverse temperature gradient observed along the thermocline between the surface cold and dilute water layer, and the underlying near 3°C-warmer saline North Atlantic water mass. The analysis of two late to post-glacial sedimentary sequences from the Chukchi Sea indicates that such a water mass stratification with a reverse thermocline persisted throughout the Holocene, thus reflecting an early onset of the modern-like linkage between the Arctic Ocean and the North Atlantic. Moreover, lower δ18O-values in both Npl and Npd together with larger δ18O-gradients between the different shell sizes at ca 9–7 ka BP suggest ∼3°C higher temperatures in the upper North Atlantic water mass, in comparison with the present (approximately +1°C, at the study site), thus likely a higher inflow rate of this water mass during the early Holocene.

The variation of North Atlantic Deep Water (NADW) formation over the Last Glacial cycle, from Oxygen Isotopic Substage 5e (OIS-5e; the Eemian) to future global warming projections, is investigated using the UVic Earth System Climate Model. The results are compared with available micropaleontological and stable isotope proxy paleo-reconstructions. Equilibrium simulations for the Eemian ( 125 kyr BP) and the Last Glacial Maximum (LGM— 21 kyr BP) both reveal the absence of Labrador Sea Water (LSW) formation although NADW formation still occurs, albeit at a reduced rate relative to the modern times. For the Eemian, the location of convection in the eastern North Atlantic is similar to the present, although it is generally shallower and less extensive. In the case of the LGM, deep convection has moved southward to the western coast of Europe and is much more localised. The inferred inception of a modern-like circulation slightly before 7 kyr BP revealed by proxy reconstructions is not captured by the model unless the meltwater forcing from the Laurentide ice sheet is applied in a long transient simulation. This raises questions concerning the applicability of equilibrium simulations in capturing the early Holocene climate. In all global warming projections, the LSW formation initially ceases as atmospheric CO 2 rises, but recovers once the level is held fixed in the atmosphere. Convection in the north extends further into the Nordic Seas as the sea ice edge retreats. In all simulations convection remains active in the eastern North Atlantic, with its latitude depending on the position of the sea ice edge, suggesting that the formation of lower NADW is a robust feature of Late Quaternary climate. As the Labrador Sea is found to be very sensitive to the freshwater forcing, it suggests that this region represents an ideal location for the concentration of observational studies to monitor a possible oceanic response to anthropogenic climate change.

Two sediment cores from the Labrador Sea and one from the Iceland Basin were analysed in order to compare Holocene sea-surface conditions across the northern North Atlantic. δ18O measurements on meso- and epi-planktonic foraminifera (Neogloboquadrina pachyderma left-coiling and Globigerina bulloides), along with sea-surface condition reconstructions using transfer functions based on dinoflagellate cyst assemblages, were used to document thermohaline properties of the upper water column. At the Labrador Sea sites, a decreasing trend in sea-surface temperature following an early Holocene maximum, is observed. At the Iceland Basin site, a hiatus between 13 and 8 ka BP prevents us from documenting the entire Holocene, but a stable to slightly decreasing trend is observed from 8 ka BP onwards. Sea-surface salinity and potential density (σθ) show little variation at the Orphan Knoll site in the Labrador Sea since 7 ka BP, whereas they depict a decreasing trend at the two other sites since ca 8 ka BP. This shift in σθ values suggests a progressive enhancement of Labrador Sea Water formation and a relative decrease in the Nordic seas components of the North Atlantic Deep Water (i.e., Denmark Strait Overflow Water and Norwegian Sea Overflow Water). Discrepancies are also observed with regard to millennial-frequency oscillations in sea-surface conditions. At the easternmost site, they likely match previously documented cold water pulses in the north-eastern Atlantic, whereas they do not show any clear pattern at the westernmost sites. Hence, the western and eastern areas show different behaviours with respect to both long-term and high-frequency fluctuations.

International audience; no abstract

Seasonal sea-ice develops along the eastern continental margins in the northern North Atlantic, where freshwater and/or meltwater outflow are responsible for relatively low salinity in surface waters and very pronounced water mass stratification. Sea-ice constitutes a major parameter in the marine ecosystem since the duration and extent of its seasonal spreading constrain the plankton distribution and the related microfossil assemblages on the sea floor. Organic-walled dinoflagellate cysts that are highly resistant to dissolution were recovered from surface sediments of the northern North Atlantic, and used to develop transfer functions (best analogue method) for the reconstruction of the seasonal spreading and duration of sea-ice cover, in addition to salinity and temperature of the warmest month of the year. Application of the best analogue approach to cores from the Labrador Sea reveals large variations in sea-ice cover and sea-surface conditions throughout the last glacial stage and during the early Holocene. Isotopic analyses in epipelagic and mesopelagic planktonic foraminifers also suggest important changes in salinity and temperature gradients between the surface and sub-surface water masses. Specific study of the last glacial maximum LGM time slice (16–20 ka on a 14 C time scale) in the northwestern North Atlantic shows much more extensive sea-ice than at present, with perennial sea-ice lying along the continental margins of eastern Canada. Seasonal spreading of the cover of sea-ice offshore was accompanied by large seasonal contrasts in temperature, with very cold winters but relatively warm summers, a pattern linked to strong stratification between a buoyant low saline surface layer having a low thermal inertia, and the underlying intermediate oceanic waters.

The application of quantitative and semiquantitative methods to assemblage data from dinoflagellate cysts shows potential for interpreting past environments, both in terms of paleotemperature estimates and in recognizing water masses and circulation patterns. Estimates of winter sea-surface temperature (WSST) were produced by using the Impagidinium Index (II) method, and by applying a winter-temperature transfer function (TFw). Estimates of summer sea-surface temperature (SSST) were produced by using a summer-temperature transfer function (TFs), two methods based on a temperature-distribution chart (ACT and ACTpo), and a method based on the ratio of gonyaulacoid:protoperidinioid specimens (G:P). WSST estimates from the II and TFw methods are in close agreement except where Impagidinium species are sparse. SSST estimates from TFs are more variable. The value of the G:P ratio for the Pliocene data in this paper is limited by the apparent sparsity of protoperidinioids, which results in monotonous SSST estimates of 14–26°C. The ACT methods show two biases for the Pliocene data set: taxonomic substitution may force ‘matches’ yielding incorrect temperature estimates, and the method is highly sensitive to the end-points of species distributions. Dinocyst assemblage data were applied to reconstruct Pliocene sea-surface temperatures between 3.5−2.5 Ma from DSDP Hole 552A, and ODP Holes 646B and 642B, which are presently located beneath cold and cool-temperate waters north of 56°N. Our initial results suggest that at 3.0 Ma, WSSTs were a few degrees C warmer than the present and that there was a somewhat reduced north-south temperature gradient. For all three sites, it is likely that SSSTs were also warmer, but by an unknown, perhaps large, amount. Past oceanic circulation in the North Atlantic was probably different from the present.