Introduction
The efficacy of erosion by High Arctic glaciers remains poorly understood. Little erosion is predicted by a theoretical, maximum ice cover over the Queen Elizabeth Islands, Arctic Canada (Fig. 1), because much of the ice was cold-based (Reference SugdenSugden, 1978). Nevertheless, the channelling of ice streams down pre-glacial fluvial valleys presumably caused “selective linear erosion” (Reference SugdenSugden, 1978) that overdeepened these valleys, producing fiords (cf. Reference MüllerPeltier, 1966; Reference BlakeBlake, 1970). At present, neither the occurrence nor the effects of such a maximum ice cover are documented for the Queen Elizabeth Islands, although erratics and melt-water channels at high elevations have been reported for many areas (Reference ShawSim, 1961; Reference Hattersley-SmithHattersley-Smith, 1969; Reference EnglandEngland, 1976, Reference England1978; Reference BlakeBlake, 1977; Reference England, England, Bradley and StuckenrathEngland and others, 1981; Reference HodgsonHodgson, 1985). On the other hand, during the last glaciation of northern Ellesmere Island, most glaciers advanced only 20–40 km beyond their present margins due to the constraints imposed by extreme aridity and the calving of glaciers within fiords (Reference EnglandEngland, 1983, in press; Reference BednarskiBednarski, in press). Therefore, the extent to which glaciers are responsible for the major valleys and fiords of the Queen Elizabeth Islands is unknown (Reference EnglandEngland, 1985).
Fig. 1. Location map of the Queen Elizabeth Islands. The field area shown in Figure 2 is outlined by the rectangle.
Field Work
During the summer of 1984, I investigated the Quaternary deposits of a glaciated valley that extends from the ice-covered Kreiger Mountains to the mouth of Tanquary Fiord, north–central Ellesmere Island (Fig. 2). The valley is 18 km long, 2–3 km wide, and up to 1100 m in relief. Today, the head of the valley is occupied by a large outlet glacier that advanced to Tanquary Fiord during the last glaciation. At that time, lateral moraines were deposited in the lower part of the valley (8 km from sea-level) at 300 m a.s.l. marking a local ice thickness of at least 250 m. In this vicinity, up to 6 m of till are exposed along the present river, blanketing the bedrock. The till, in turn, is overlain by marine silts which contain in-situ shells at 76 m a.s.l. dated 7805 ± 125. (S-2649).
Fig. 2. Field area of the glaciated valley extending from the Krieger Mountains to Tanquary Fiord. Glacially abraded bedrock in the lower valley shows former north–south flow (arrow) immediaely down-valley from the glaciated alluvial fan. The fan was cut by melt-water channels (dashed lines) when ice retreated from site A to site B.
Localized glacial erosion is manifested, c. 3 km from the coastline where a diabase dike has been plucked, grooved, and striated by a southerly ice flow toward Flora Island, Tanquary Fiord (Figs 2 and 3). This flow diverges somewhat from the axis of the valley (north-west to south-east), becausethe ice which deposited the moraines was being deflected by a large gypsum hill along the lower east side of the valley while spreading out into Tanquary Fiord. Immediately up-valley from the abraded diabase dike are abundant crag-and-tail features comprised of low ridges of limestone extending from the lee sides of more resistant chert nodules (Fig. 4). The crag-and-tail is aligned with the sculptured diabase dike and therefore they were eroded along a similar flow line. The limestone ridges are commonly <5 cm high and from 10 to >100 cm long.
Fig. 3. Plucked, grooved and striated diabase dike in the lower valley. Former direction of ice flow was north–south. Bedrock outcrop covers an area of c. 400m2. The view is taken in down-ice direction.
Fig. 4. Small crag-and-tail composed of limestone ridges on the lee sides of more resistant chert nodules. The ridges are commonly <5 cm high and often >25 cm long.
During deglaciation of the lower valley (7800 B.P.), ice-contact deltas were deposited into the sea, marking the local marine limit at 116 m a.s.l. (site A, Fig. 2). By the time the ice front reached site В (Fig. 2), 2 km farther up-valley, relative sea-level had fallen from 116 to 108 m a.s.l., indicating that 8 m of restrained rebound (Reference AndrewsAndrews, 1970) occurred beneath the retreating ice across this distance. This amount of emergence (8 m) could have been accomplished in 200–300 years of ice retreat given a rate of initial emergence of ~3 m/100 year (Reference EnglandEngland, 1983). Therefore, the average rate of retreat between sites A and В (Fig. 2) was c. 10 m year–1.
Between sites A and B, two former melt-water streams have incised a large alluvial fan, producing two sloping escarpments c. 75 cm high and >100 m long. These former channels are nested in the lower valley at the base of the fan and their gradients represent successive positions of the retreating ice margin (Fig. 5a). Because these escarpments were cut during deglaciation, the alluvial fan must pre-date the last glaciation. The degree of alteration of the fan caused by the over-riding ice is minimal and the final glaciofluvial incision along the ice margin is estimated to be similar to the present depth of the active layer in this area (~75 cm). Moreover, tundra polygons on the original surface of the fan cross the escarpments indicating that the melt water incised, but did not remove, their underlying ice wedges (Fig. 5b). Conversely, since deglaciation, these escarpments have not been muted or buried by subsequent mass wastage (solifluction) or alluviation.
Fig. 5. 5. (a) View from the ice-contact, marine limit (site A) westward to the glaciated alluvial fan (centre background). The two linear incisions from the melt-water channels occur on the lower right side of the fan. (b) Telephoto view of the same two escarpments cut by lateral melt-water streams across the pre-existing alluvial fan. The escarpments are estimated to represent incision similar to the depth of the modern active layer (~75 cm). The upper (oldest) escarpment is c. 100 m long and it has not been altered by subsequent mass movement or alluviation. Note that the original ice-wedge polygons extend across the upper escarpment.
Discussion
Abraded bedrock and the subsequent deposition of up to 6 m of till indicate that some erosion of the valley occurred during the last glaciation. Although such abraded bedrock cannot be dated directly, it is assumed to be the product of the last glaciation because: (1) it occurs inside the last ice limit; (2) the lateral moraines marking this limit are bordered down-valley by a gypsum hill that would align the flow of ice with the observed striations and crag-and-tail features; and (3) the abrasion is freshly inscribed and laterally capped by till, in turn, overlain by deglacial marine silts dated c. 7800 B.P. The extent of the abraded bedrock indicates that during the last glaciation the ice remained grounded at least to within 3 km of the present coastline (Fig. 2). Although the areas of abraded bedrock within the valley are generally small (<2500 m2), they nonetheless represent a minimum estimate for such abrasion because much of the bedrock is buried beneath till, colluvium, or marine sediments. However, compared to the abraded bedrock, a larger area of the valley (1 km2) is occupied by an alluvial fan that pre-dates the last glaciation. Although the fan occurs immediately up-stream from the abraded bedrock, it has been little altered by the over-riding ice. It is apparent, therefore, that in this valley, different erosional regimes have characterized the glacier bed and, as is the case today, such sub-polar glaciers probably varied from cold-based to warm-based conditions on a local scale (Reference KlassenMüller, 1976; Reference Hambrey and MüllerHambrey and Müller, 1978).
Today, in the Canadian High Arctic, the most effective erosional processes associated with sub-polar glaciers include: (1) pro-glacial thrusting of permafrost blocks commonly composed of glaciofluvial or glaciomarine sediments that are subsequently over-ridden and entrained by the advancing ice (cf. Reference KälinKälin, 1971; Klassen, 1982; Reference Stewart and EnglandStewart and England, 1983); and (2) pro-glacial melt water that incises both unconsolidated sediment and bedrock, producing channels nested in the direction of ice retreat (cf. Reference BlakeBlake, 1981; Reference DykeDyke, 1983). Little is known about subglacial abrasion, although it is recognized that cold-based ice can erode bedrock providing it has sufficient basal debris (cf. Reference SugdenSugden, 1978). Reference BoultonBoulton (1979 has also observed frozen eolian sands that were deformed and plucked beneath over-riding cold-based ice in Wright Valley, Antarctica. Nevertheless, moraines and till blankets (>1 m thick) are rare over large areas of the Canadian High Arctic, suggesting that erosion and re-deposition by these glaciers was slight.
Modern sub-polar glaciers that are advancing often do so by over-riding ice-rich debris aprons that accumulate by dry calving along oversteepened ice fronts (Reference ShawShaw, 1977). Consequently, new basal ice is passively incorporated such that little active erosion can occur at the glacier bed. This is illustrated eloquently on east-central Ellesmere Island, where a retreating glacier is exposing dead, intact plant communities previously over-run by ice at least 400 years ago (Reference Bergsma, Bergsma, Svoboda and FreedmanBergsma and others, 1984). In the same area, the retreat of a glacier, following the most recent Neoglacial advance, has exposed an alluvial fan covered by “coarse bouldery ablation till” (Reference BlakeBlake, 1981, Fig. 10, p. 211). Finally, on north-central Baffin Island, Reference FalconerFalconer (1966) also reported preserved vegetation dated c. 330 B.P. that was exposed by the retreat of thin ice caps during the first half of the present century.
The observations presented here indicate that, although a glacier at least 250 m thick advanced down a valley ideally located for selective erosion, this did not remove or significantly alter a large alluvial fan that lay in its path. The fan occurs adjacent to abraded bedrock that covers less area of the valley. It is apparent that the glacier advanced passively over the fan, presumably by over-riding a frontal ice apron that covered the undeformed (non-thrusted) fan. Subsequently, the strength provided by permafrost within the over-ridden fan exceeded the forces of entrainment or deformation exerted by the ice. If the fan had not been incised by lateral melt-water streams during deglaciation, one would normally assume that the fan was of post-glacial age. However, because the fan pre-dates the last glaciation, it is apparent that there has been a net increment of alluvium versus glacial erosion or deposition at this site. Similar observations have been made on Somerset Island, central Arctic Canada, where colluvial slopes with solifluction features and patterned ground cover thousands of square kilometers. Although these landforms have been reactivated during the post-glacial, they are developed nonetheless on old colluvium and residual terrain that pre-date at least the late Wisconsinan glaciation (personal communication from Reference DykeA.S. Dyke, 1985). Therefore, in permafrost terrain, one must reject the assumption that all sediments mantling deglaciated valleys are of post-glacial age.
The question that follows is whether the erosive capacity of High Arctic glaciers is responsible for the profiles of such valleys, let alone the much larger fiords commonly occupied by water depths of 300–900 m (Reference Hattersley-SmithHattersley-Smith, 1969). The concept of selective linear erosion has been widely used to explain similar valleys. Most notably, zones of selective linear erosion by warm-based ice have been predicted for the Laurentide ice sheet at its maximum (including Arctic Canada), based on a pervasive ice cover with a theoretical profile and basal thermal regime (Reference SugdenSugden, 1977, Reference Sugden1978). On a more local scale, Reference BlakeBlake (1978, p. 176) invoked selective linear erosion to explain the topography of the Makinson Inlet area, south-east Ellesmere Island, stating that … “This type of landscape is characterized by valleys in which tremendous amounts of scouring take place along pre-existing drainage lines of structural lineaments …”. In the case of the valley discussed in this paper, the proponents of selective linear erosion could argue that its profile was cut during an earlier glaciation when pervasive ice presumably over-rode both the valley and its adjacent interfluves. Subsequently, during less extensive glaciations, it also could be argued that selective linear erosion did not occur beneath isolated outlet glaciers because they would have been thinner, and thereby mainly cold-based and protective. If so, it could be concluded that the evidence presented in this paper does not conflict with the concept that selective linear erosion produced the valley’s profile.
So far, there is no such evidence for an earlier, pervasive ice cover causing the scale of erosion presently attributed to selective linear erosion in the Canadian High Arctic. To the contrary, above and beyond the limit of the last glaciation in this valley, there is widespread weathered bedrock without till sheets and certainly with few erratics. Farther out in Greely Fiord (into which the valley in question leads) there is also little evidence of glaciation other than the expansion of local ice caps. Therefore, evidence for massive glacial erosion and re-deposition, as would be required if the size of these valleys and fiords was caused by such a pervasive glaciation, is lacking. It is unacceptable to assume that all the moraines and till sheets re-deposited from such erosion are below present sea-level, and that this pervasive ice also crossed interfluves without depositing similar material! Indeed, in the central part of the Canadian Arctic Archipelago, well-preserved river valleys, likely of Tertiary age, are preserved on the floor of Barrow Strait, arguing against significant glacial erosion of that channel (Reference Bornhold, Bornhold, Finlayson and MonahanBornhold and others, 1976; Reference DykeDyke, 1983).
Finally, although the valley glacier discussed in this paper was much smaller than the regional ice sheet invoked for selective linear erosion, this does not mean that the erosional effects of the smaller glacier are not diagnostic of the effects of earlier glaciations. For example, other Arctic valley glaciers of similar size and thickness (e.g. White Glacier, Axel Heiberg Island) may locally reach pressure melting-point and therefore contribute to basal erosion (Reference MüllerMüller, 1976; Reference Hambrey and MüllerHambrey and Müller, 1978). Furthermore, it is inconsistent to dismiss the erosive power of this valley glacier due to its smaller size, because it has been shown that effective abrasion does not require excessively thick ice (Reference Boulton and CoatesBoulton, 1974). Therefore, because this glacier (250 m thick) did little to alter this valley’s profile during the last glaciation, it is considered just as valid to assume that earlier glaciations did no more!
In conclusion, selective linear erosion persists in the literature as a fundamental concept used to explain the development of glacial landscapes. Despite the widespread acceptance of selective linear erosion, one must not lose sight of the fact that in Arctic Canada it is based solely on a theoretical model which assumes a pervasive ice cover whose very presence and dynamics remain to be demonstrated. Moreover, it has not been demonstrated that this concept takes precedence over alternative causes for such valleys; for example, faulting (Reference BirdBird, 1967; Reference EnglandEngland, 1985). Therefore, further consideration of the erosive capacity of glaciers is warranted, particularly as it pertains to the evolution of the fiords and inter-island channels of Arctic Canada which have been explained routinely by glacial over deepening. This, in turn, has created a circular view of high-latitude ice sheets that may not be valid.
Acknowledgements
The field work was logistically supported by the Polar Continental Shelf Project, Energy, Mines and Resources, Ottawa. Additional financial support for the research was provided by the Natural Science and Engineering Research Council of Canada (NSERC grant A 6680). Thanks are expressed to K. Lindstrom, University of Alberta, who assisted me in the field. G. Lester, Head of the Cartographic Division, Department of Geography, University of Alberta, and R. Pakan, of the same Division, provided the maps and photographs, respectively. S. Fowler and F. Metcalfe typed the manuscript. Finally, A.S. Dyke, Terrain Sciences, Geological Survey of Canada, brought to my attention similar observations he had made on the Somerset Islands. Two anonymous reviewers also provided helpful comments that led the author to refine his objections to selective linear erosion.
