Reorganization of ice sheet flow patterns in Arctic Canada and the mid-Pleistocene transition



[1] Evidence for the evolution of Laurentide Ice Sheet (LIS) basal thermal regime patterns during successive glaciations is poorly preserved in the geologic record. Here we explore a new approach to constrain the distribution of cold-based ice across central Baffin Island in the eastern Canadian Arctic over many glacial-interglacial cycles by combining till geochemistry and cosmogenic radionuclide (CRN) data. Parts of the landscaped with geomorphic evidence for limited glacial erosion are covered by till characterized by high chemical index of alteration (CIA) values and CRN concentrations requiring complicated burial-exposure histories. Till from regions scoured by glacial erosion have CIA values indistinguishable from local bedrock and CRN concentrations that can be explained by simple exposure following deglaciation. CRN modeling results based on these constraints suggest that the weathered tills were deposited by 1.9 to 1.2 Ma, and by that time the fiorded Baffin Island coastline must have developed close to its modern configuration as piracy of ice flow by the most efficient fiord systems resulted in a major shift in the basal thermal regime across the northeastern LIS. The resultant concentration of ice flow in fewer outlet systems may help explain the cause of the mid-Pleistocene transition from 41- to 100-kyr glacial cycles.

1. Introduction

[2] Since the late Pliocene, global climate variability has been dominated by the cyclical waxing and waning of continental ice sheets, initially following the obliquity orbital frequency of 41 kyr. Between ca. 1.25 and 0.7 Ma [Clark et al., 2006], the cyclicity transitioned to a higher-amplitude 100-kyr periodicity. This enigmatic mid-Pleistocene transition (MPT) is of particular interest because it reflects the development of a non-linear climate system response independent of orbital forcing. Possible explanations include mechanisms related to gradual atmospheric and/or deep ocean cooling [Raymo, 1997; Tziperman and Gildor, 2003], the skipping of obliquity-driven terminations [Bintanja and van de Wal, 2008; Huybers, 2007], and changes in ice sheet stability due to regolith removal [Clark and Pollard, 1998; Clark et al., 2006; Roy et al., 2004]. However, evidence suggests that a change in ice sheet dynamics is required to fully explain the MPT [Bintanja and van de Wal, 2008; Clark et al., 2006; Sosdian and Rosenthal, 2009].

[3] Spatial and temporal variations in the basal thermal regime of an ice sheet exert a fundamental control on basal sliding, basal sediment deformation, ice sheet geometry, the response of ice sheets to climate forcing, and the extent to which landscapes are modified by glacial erosion [Clark, 1994; Marshall and Clark, 2002; MacAyeal, 1993]. The Foxe Sector of the Laurentide Ice Sheet (LIS; Figure 1a) experienced a complex and dynamic interplay between non-erosive, cold-based ice, erosive, fast-moving outlet glaciers that carved deep fiords through Canada's eastern Arctic rim, and even more erosive ice streams that occupied large straits and sounds. These glaciers and ice streams transported ice from the Foxe Dome to calving margins in Baffin Bay and the Labrador Sea [Andrews et al., 1985; De Angelis and Kleman, 2007].

Figure 1.

Study area and till geochemistry results. (a) Satellite image of central Baffin Island with generalized bedrock geology. Black boxes show NTS grids. Inset shows study area relative to the LGM LIS extent. (b) Till CIA values; due to limitations in the available data, CIA values represent the silt fraction for areas 37E-H, and 47E and the clay fraction for areas 27B, 27C, 37A, and 37D. (c) Till Ca/Na ratios. Values for clay fraction data must be divided by 100. (d) Till CIA values excluding samples with excessive CaO. Open circles denote samples with inherited weathering characteristics from metasedimentary bedrock. Black squares show sample locations from Miller et al. [2006]. (e) Inset shows sample locations from Staiger et al. [2006]. Colors correspond to groups 1–4 in Figure 2b. (f) Inset shows till CIA values for the area northwest of the Barnes Ice Cap.

[4] Megascale geomorphic features across Baffin Island illustrate variable LIS erosion, with evidence for the preservation of till and bedrock over multiple glacial cycles and possible relict pre-glacial surfaces [De Angelis and Kleman, 2007; Staiger et al., 2006]. In contrast, areas of the landscape covered by warm-based, fast-moving ice were extensively eroded during the last glacial maximum (LGM) [Briner et al., 2006]. The intermittent till mantle on the interior plateaux of Baffin Island may hold the key to understanding the evolving relationship between ice-sheet erosion and the dynamics of the Foxe Dome over Quaternary time scales. We use previously published till geochemistry and cosmogenic radionuclide (CRN) data to explore this evolution.

2. Chemical Index of Alteration

[5] To evaluate the extent to which surface tills have weathered, we use the chemical index of alteration (CIA) [Nesbitt and Young, 1982]. This index reflects the proportion of primary minerals to chemically-weathered products in a sample according to

equation image

where oxides are expressed as molar proportions and CaO* is CaO from non-carbonate phases. We utilize geochemistry data from till samples collected by Dredge [2004] and Utting et al. [2008] from 20 to 50 cm depth. We first calculate CIA values for the silt- and/or clay-sized fractions of till samples using CaO rather than CaO* (Figure 1b) because of limited carbonate content data. Many samples contain CaO from carbonate bedrock sources (Figure 1a), and Ca/Na ratios (Figure 1c) help identify samples without excessive CaO (Figure 1d). Samples with excessive CaO are defined as those with >1% evolved CO2 based on till carbonate content data from National Topographic Series (NTS; Figure 1a) area 37G in the silt-sized fraction, which corresponds to Ca/Na > 40 and 0.2 for the silt and clay fractions, respectively.

2.1. Till Weathering

[6] The CIA approach is particularly advantageous in this region because the majority of the study area is underlain by gneissic Rae Craton bedrock (Figure 1a; CIA values 40–55). Carbonate bedrock crops out to the west and north (CIA values <30), and metasedimentary bedrock is exposed to the south (CIA values 50–90). Till CIA values on the Rae Craton range from 20–90 (Figure 1b) and generally decrease inland from Foxe Basin, but several plumes of higher Ca/Na ratios extend from Foxe Basin to Cambridge and Clyde Fiords and Home Bay (Figure 1c). The latter is consistent with a train of carbonate boulders [Tippett, 1985], and all higher Ca/Na plumes correspond to a high concentration of lakes in ice-scoured bedrock [Andrews et al., 1985].

[7] After removing the CIA values affected by excessive CaO, more subtle patterns become apparent (Figure 1d). In the northern end of the study area, high CIA values (>70) occur on most of the interior highlands, whereas lower values (<60) are predominantly found at lower elevations in ice-scoured terrain. Another concentration of similarly high CIA values occurs across a surface covered by a thick till blanket in terrain devoid of lakes and scoured bedrock joints near the Barnes Ice Cap (Figure 1f).

[8] The distribution of till CIA values is consistent with a variety of other lines of evidence previously used to identify LIS basal thermal regimes. Notable differences in the surficial weathering characteristics of bedrock have been identified along fiord walls across the region [Ives, 1975], as warm-based ice was restricted to fiords and the broad valleys in fiord onset zones [Briner et al., 2006]. CIA values are consistently lower in fiord and fiord-onset-zone tills and higher on interfluves that were likely covered by slow-moving or cold-based ice between adjacent onset zones (Figure 1d). Low CIA values correspond with high scour-lake densities [Andrews et al., 1985]. Similarly, mega-scale geomorphic features have been used to identify the locations of erosive, fast-flowing ice [De Angelis and Kleman, 2007] and indicate erosive ice crossing Baffin Island toward Home Bay and Clyde and Inugsuin Fiords, corresponding to low till CIA values.

[9] We interpret the CIA values to reflect the residence time of till on the landscape. Low CIA values in the range of local bedrock are due to the erosion fresh bedrock and deposition of relatively unweathered material upon deglaciation. High CIA values reflect prolonged subaerial weathering that must have occurred over multiple interglaciations, and overriding ice has subsequently been cold-based and only minimal till transport and deposition has taken place.

3. Detrital CRN Data

[10] Detrital CRN data from Staiger et al. [2006] provide a unique opportunity to compare CIA values to CRN inventories and infer exposure-burial histories in the same samples (Figure 1e). CRN concentrations were measured in quartz from the 250–355 μm size fraction and are adjusted for a 10Be half-life of 1.36 Myr and rescaled according to Desilets et al. [2006] using a 10Be production rate of 4.41 atoms g SiO2−1 yr−1, a 26Al/10Be production ratio of 6.75 following the approach of Balco et al. [2008] assuming a sediment density of 1.6 g/cm3, an attenuation length of 150 g/cm2 for nucleon spallation, and a 26Al half-life of 0.71 Myr.

3.1. Till Exposure History

[11] Regression slopes for 10Be and 26Al concentrations against CIA values for silt and clay fractions are significant at the 0.05 critical α level (Figure 2a), confirming our hypothesis that weathered till has been at the landscape surface longer than unweathered till. More telling relationships emerge when clay fraction CIA values are plotted on a two-isotope diagram (Figure 2b), in which samples cluster into four groups.

Figure 2.

Relationships between till CIA and CRN data. (a) Till clay data with error bars showing 1σ uncertainties. (b) Two-isotope diagram showing till CRN data from Staiger et al. [2006] (groups 1–4; excludes 5 samples without Al measurements) and bedrock CRN data from Miller et al. [2006] (group 5). Circle diameters correspond to till clay fraction CIA values from the same samples. The solid black curve and associated ages track the CRN inventory derived from steady surface exposure in the absence of erosion. Upon burial, samples will track downward along the dotted lines, and the dashed lines indicate the 26Al/10Be ratio after a period of continuous burial. The three jagged black lines depict the CRN inventory evolution for a sample subjected to 1.0, 1.5, and 2.0 Myr of glacial-interglacial cycles.

[12] Group 1 includes samples from low-elevation, glacially sculpted terrain with low CIA values similar to local bedrock CIA values and CRN inventories consistent with minimal burial and ≤10 kyr of surface exposure, representing warm-based LGM tills [Staiger et al., 2006]. Group 3 samples are from upland surfaces (low-relief terrain 600–1000 m asl) and fiord rims, have moderately high CIA values, and have been buried for at least 1 Myr (longer if burial was episodic). Two of these three samples were potentially covered for a considerable part of past interglaciations by the largest ice cap in the study area. Group 4 samples were collected from upland surfaces adjacent to fiords and have CRN inventories consistent with burial in excess of 2 Myr and likely contain inherited pre-glacial CRNs [Staiger et al., 2006]. CRN data from bedrock knolls between fiord onset zones imply cover by only cold-based ice during glaciations in the past 200 to 500 kyr (group 5 in Figure 2b) [Briner et al., 2008].

[13] Of particular importance are group 2 samples, which have the highest CIA values and come from upland surfaces. It is likely that these uplands were deglaciated completely during most interglaciations, and CRN inventories are consistent with exposure only during interglaciations for the past 1.9–1.2 Myr (Figure 2b). This exposure history is based on a model with 5/95 kyr of exposure/burial during 800 kyr of 100-kyr glacial cycles and 3/38 kyr of exposure/burial for 41-kyr cycles. Production rates are scaled for sampling depths (ca. 20 cm), but CRN concentrations alone will provide limited information about the exposure-burial history of till on this landscape because of mixing by cryoturbation. However, the 26Al/10Be ratio is constant for near-surface production, so we use this value to evaluate modeled exposure-burial histories, and 1.2–1.9 Myr provides the best agreement with the 26Al/10Be ratios of group 2 samples. While it is not possible to rigorously test our assumption of negligible CRN inheritance, an inherited CRN inventory developed over only a few interglacial cycles, equivalent to the most inheritance commonly seen in recent LIS deposits on Baffin Island [Briner et al., 2006; Davis et al., 2006; Staiger et al., 2006], would result in our modeled till age being ca. 100 to 300 kyr too old.

4. Discussion

[14] The samples in group 2 come from topographic settings and erosional regimes that are similar to samples with equally-high CIA values adjacent to the Barnes Ice Cap (Figures 1d and 1f). We postulate that each of these areas can be characterized by similar exposure-burial histories, implying that (1) these areas have experienced minimal glacial erosion since 1.9 to 1.2 Ma, and (2) flow paths along which ice from Foxe Basin was channeled to outlet glaciers on the Baffin Island coast have been largely invariant since 1.9–1.2 Ma, but that prior to that time outlet glacier flow through the eastern Baffin Island rim was more uniform. Consequently, valley and fiord erosion in some areas was greatly reduced after 1.9–1.2 Ma as ice flow was captured by a smaller number of more efficient conduits such as Cambridge, Clyde, Inugsuin, and McBeth Fiords, and Home Bay, as well as the larger sounds north and south of Baffin Island. This mechanism of preferentially concentrating flow and enhancing erosion through fiord systems has been illustrated by numerical ice-sheet model simulations [Kessler et al., 2008].

[15] Additionally, fiords with high till CIA values near their heads (e.g., Clark, Dexterity, and Tromso Fiords, Ayr Lake Trough) have abrupt decreases in elevation from the onset zone into the trough itself, and these troughs extend only to the inland edge of the mountains (Figure 3). Fiords acting as more favorable conduits are fed by scoured, lake-dotted, broad valleys extending through the entire breadth of the mountains and into the interior of the island (e.g., Oliver Sound, Cambridge and Inugsuin Fiords). The onset zones of these fiords are characterized by low CIA values, low CRN concentrations [Briner et al., 2008], and elevations decrease gradually (Figure 3).

Figure 3.

Long profiles of troughs from fiord head at sea level to trough head inland. Fiords with high CIA values in and adjacent to their onset zones have characteristically different profiles than fiords with low CIA values in and adjacent to their onset zones.

[16] The combination of till CIA data and modeled CRN burial-exposure histories provides strong evidence for a shift in basal thermal regimes across the interior plateaux of Baffin Island by 1.9–1.2 Ma. While it may be coincidence that this time interval abuts the onset of the MPT (∼1.2–0.7 Ma), it has been hypothesized that changes in subglacial conditions were potentially an important mechanism in altering LIS dynamics across the MPT [Clark and Pollard, 1998; Clark et al., 2006]. We suggest that prior to this time, ice was warm-based and erosive across the majority of the Baffin Island interior, and widespread glacial erosion left behind ice-sculpted bedrock and a nearly continuous blanket of till. By 1.9–1.2 Ma, some parts of the landscape became perpetually covered by cold-based ice during glaciations, a pattern that persisted through the LGM.

[17] Preferential channeling of ice flow into major fiords may have been sufficient to effectively shut off ice flow across the landscape between outlet glaciers. Fiord development likely increased both the efficiency of ice delivery to the oceans by calving outlet glaciers and the climatic sensitivity of the ice sheet [Kessler et al., 2008]. However, it is possible that the evacuation of ice to Baffin Bay became less efficient as the aerial extent of cold-based ice across Baffin Island increased between 1.9 and 1.2 Ma. Additional factors may have contributed to this expansion of cold-based ice, including a gradual decline in global temperature [Sosdian and Rosenthal, 2009], changes in bed composition, or the formation of large ice streams in sounds to the north and south. The associated decrease in the extent of basal sliding would require a greater ice thickness for the basal ice to be at the pressure melting point. Such a feedback could potentially result in the Foxe Dome becoming less sensitive to climate forcing, contributing to the internal changes within the climate/ice sheet system required to explain the MPT [Clark et al., 2006; Huybers, 2007; Sosdian and Rosenthal, 2009]. Although the Foxe Dome and Baffin Island coastline comprise a relatively small percentage of the total LIS volume and perimeter, ice-sheet modeling may clarify whether the reconstructed changes had a disproportionate impact on the behavior of the LIS as a whole.

5. Conclusions

[18] Till deposits on central Baffin Island contain a geochemical record of landscape and LIS evolution. Using constraints from major oxide and CRN data, we model the burial and exposure history of the oldest of these tills and suggest that they were deposited by 1.9 to 1.2 Ma. Cover by cold-based ice during subsequent glaciations has preserved these tills. We interpret this shift in basal thermal regime to be the result of the preferential funneling of ice into the most efficient fiord systems, which were apparently developed by 1.9 to 1.2 Ma. If changes in the dynamics of the Foxe Dome slowed its response to climate forcing, the maturation of the fiorded coastline may help explain the MPT.


[19] This researched was supported by NSF award ARC-0903024 and benefitted from discussions with Robert Anderson, Suzanne Anderson, and John Andrews and helpful comments from two anonymous reviewers.