Introduction
Ice shelves in the Canadian and Greenland High Arctic are presently formed by either land-fast sea ice or floating glacier margins. Ice shelves created by the accretion of land-fast sea ice are well documented from northernmost Ellesmere Island (cf. the Ward Hunt Ice Shelf; Reference Koenig, Koenig, Greenaway, Dunbar and Hattersley-SmithKoenig and others, 1952; Reference Hattersley-Smith and CraryHattersley-Smith and others, 1955; Reference CraryCrary, 1960; Reference Lyons and MiclkcLyons and Mielke, 1973). Radiometric dates on the youngest driftwood trapped behind the Ward Hunt Ice Shelf and on organic debris incorporated within it suggest initial formation$$ c. 3 000-4 000 B.P., hence, following the Holocene Climatic Optimum (Reference CraryCrary, 1960; Reference Hattersley-SmithHattersley-Smith, 1969; Reference Lyons and MiclkcLyons and Mielke, 1973). The build-up of land-fast sea ice is due to the freezing of low-salinity sea-water at its base and, to a lesser extent, to the periodic surface accumulation of iced firn {Reference MarshallMarshall, 1955). The growth of such ice shelves is favoured by severe winter cold, low precipitation, and limited summer melting (Reference Hattersley-SmithHattersley-Smith, 1960). Depositional features along the landward margins of these ice shelves have been briefly described and are restricted to shattered boulders transported down ice ramps from the adjacent land (Reference Hattersley-Smith and CraryHattersley-Smith and others, 1955) or to "debris-laden ice ridges" where the ice shelf is grounded (Reference Lyons and MiclkcLyons and Mielke, 1973, p. 315). Morphological evidence for more extensive, pre-Holocene ice shelves has not been cited either due to its absence or a lack of systematic investigation. It seems likely that such land-fast sea ice was more extensive in the past, perhaps during the cold phase between 65 000 and 10 000 $$B.P. as recorded in the "Camp Century" and Devon Island ice cores (Reference Dansgaard, Dansgaard, Johnsen, Clausen and GundestrupDansgaard and others, 1973; Reference Paterson, Paterson, Koerner, Fisher, Johnsen, Clausen, Dansgaard, Bûcher and OeschgerPaterson and others, 1977)·
Ice shelves, formed by floating glacier ice discharging from land-based ice sheets, have been extensively reported from Antarctica (Reference SwithinbankSwithinbank, 1957; Reference Crary, Crary, Robinson, Bennett and BoydCrary and others, 1962; Reference Swithinbank, Zumberge and HathertonSwithinbank and Zumberge, 1965; Reference BuddBudd, 1966; Reference HoldsworthHoldsworth, 1969; Reference ThomasThomas, 1973, 1976). In addition, the occurrence of present-day, floating glacier termini in arctic Canada and Greenland is widely recognized (Reference KochKoch, 1928; Reference SorgeSorge, 1933; Reference Krinsley and RaaschKrinsley, 1961; Reference Carbonnell and BauerCarbonnel and Bauer, 1968; Reference Feazel, Kollnieycr and KarlssonFeazel and Kollmeyer, 1972; Reference Leken, Loken, Ornrnanney, Holdsworth and KarlssonLoken and others, 1972). The dimensions of such features are generally limited by the rate of ice flow from the adjacent land, rates of melting and calving, submarine topography, and additional forces such as the bending stresses at the hinge line (Reference HoldsworthHoldsworth, 1969). Their principal morphological distinction is a comparatively flat upper surface with a limited amount of freeboard (ratio of total ice thickness to front height is$$$ c. <8.5 (Reference BuddBudd, 1966; Reference ReehReeh, 1968, Reference Reeh1969; Reference ThomasThomas, 1973)).
Several authors have speculated on the presence of extensive ice shelves during former glaciations of the High Arctic (Reference Hattersley-SmithHattersley-Smith, 1960; Reference MercerMercer, 1969, 1970; Reference Grosval'dGrosval'd, 1972; Reference BroeckerBroecker, 1975; Reference Hughes, Hughes, Denton and GrosswaldHughes and others, 1977). Stratigraphie evidence has also been cited which suggests the presence of shelf ice bordering the Magdelen Islands, Gulf of St. Lawrence, during the Wisconsin glaciation (Reference Prest, Prest, Terasmae and MathewsPrest and others, 1976). Although submarine and near sea-level moraines have been described from arctic Canada and Greenland (Reference CraryCrary, 1956; Loken, 1973; Reference Ten Brink and A.Ten Brink and Weidick, 1974; Reference BlakeBlake, 1977), no documentation of former ice-shelf moraines has been made. In addition, Smith (unpublished) showed that the moraines in Sam Ford Fiord, eastern Baffin Island, descended to sea-level with a gradient of 1 : 100.
Evidence for Former Ice Shelves
During 1975-76, studies were conducted on the surficial geology of Judge Daly Promontory, north-eastern Ellesmere Island (Fig. 1). In this area, moraines deposited by the outermost Ellesmere Island ice advance cross-cut an older and more extensive zone of Greenland till (Reference England and BradleyEngland and Bradley, 1976). The distribution and gradients on the Ellesmere Island moraines indicate the presence of thin, topographically controlled ice lobes draining southeastward across the promontory to sea-level along western Kennedy Channel. The termini of two ice lobes were investigated, one originated from the interior of Judge Daly Promontory and extended to Cape Defosse, whereas the second, 20 km to the north-east, represented tributary ice flowing south-eastward out of Lady Franklin Bay into lower "Beethoven Valley" Footnote * (Fig. 1). Both ice lobes crossed an interior lowland $$$(c. 200-300 m a.s.l.) $$ before descending into narrow valleys, 2-3 km in width, which lead to Kennedy Channel. Relief in the valleys is >500 m and bathymétrie soundings 5 km offshore show water depths of >36o m (Canada. Hydrographie Service, 1973). Due to glacio-isostatic depression of these valleys, both outlet glaciers were forced to float in the resulting embayments along western Kennedy Channel. Under these conditions, it is suggested that small ice shelves developed but it is not known to what extent these would have been sustained by bottom accretion of ice or by surface accumulation. Evidence in support of such ice shelves is based on morphology, stratigraphy and the relative sea-level at the time of their formation. The chronology of these ice shelves is also discussed.
Morphology
The uppermost Ellesmere Island moraines in the interior of Judge Daly Promontory occur at c, 500 m a.s.l. and descend to c. 260 m a.s.l. at the entrance to lower "Beethoven Valley" (Fig. 1). Within 0.5 km of this latter point, a well-developed system of conical kames and lateral moraines descends steeply down-valley to an elevation of 200 m a.s.l., where they become abruptly horizontal for a distance of c. 2 km. Immediately up-slope from these horizontal lateral moraines one encounters a sharp break in weathering characterized by deeply oxidized bedrock, tors, and a sparse distribution of crystalline erratics previously deposited by the Greenland ice sheet. The horizontal moraines are considered to represent the lateral margin of a floating outlet glacier whose grounding line was located at the lowermost sector of steeply sloping lateral moraines up-valley. Vertical and ground-level view of these moraines on the south-west side of "Beethoven Valley" are shown in Figures 2 and 3, respectively. The slope of the moraines shown in Figure 3 compares with observations made on the Ross Ice Shelf, Antarctica, whose profile is characterized by "(i) the abrupt increase in elevation as one goes from the ice shelf on to the continental ice sheet and (ii) the depression associated with the juncture of the ice shelf and ice sheet" (Reference Theil and OstensoTheil and Ostenso, 1961, p. 825). Several altimeter transects, corrected for both temperature and pressure, were run along the entire horizontal sector of the "Beethoven Valley" moraines and no elevation differences > 1 m were detected. Similar, but less extensive, lateral moraines occur on the opposite side of the valley and suggest an ice-shelf width of $$$c. 2 km.
20 km to the south-west of "Beethoven Valley" an outlet glacier from the interior of Judge Daly Promontory formerly reached Kennedy Channel via the lower Daly River valley (Fig. 1). Along the western slope of this valley another prominent horizontal moraine system occurs at $$$c. 195 m a.s.l. and is bordered up-slope by the same deeply weathered bedrock and sparse Greenland erratics. This moraine system and the morphology of its upper surface are shown in Figures 4 and 5, respectively. The surfaces of the horizontal moraines in both valleys show little relief (<o.5 m), are 5 m in width and covered by a highly frost-shattered lag gravel lying on a stony/sand weathering profile. On the basis of the valley size and the distribution of depositional features, the ice shelf in lower Daly River was not more than 2 km in length and width.
Stratigraphy
In "Beethoven Valley", the ice-shelf moraines are occasionally fossiliferous, containing fragments of marine shells. That these shells have not been ice-transported across the interior of Judge Daly Promontory is clearly evidenced by their termination up-valley at a point coinciding with the apparent grounding line. In addition, no shells were found in the interior of Judge Daly Promontory despite extensive traverses of this area. The manner by which these shells have been incorporated in the moraines is unclear; they may have been scoured up by the advancing ice at its grounding line or, alternatively, they may have been gradually transferred to the surface by the accretion of freezing sea-water at the base of the floating glacier. This latter process has been well documented from the 40 m thick Ward Hunt Ice Shelf where organic material (siliceous sponges, sea worms, pelecypods, and remains of Arctic cod) has been transported to the surface (Reference Lyons and MiclkcLyons and Mielke, 1973). Lyons and Mielke (1973, p, 315) also reported a rich biota in both the ice shelf's debris ridge and in the ice moat between this ridge and the shore. As regards glacier flow as a mechanism for transporting these shells into the moraines, it is of interest that particle trajectories through the Brunt Ice Shelf, Antarctica, do not reveal basal flow lines returning to the surface (Thomas,1973}.
Adjacent to the ice-shelf moraines in lower "Beethoven Valley" are massive pro-glacial terraces at c. 175 m a.s.l. (Fig. 3). These terraces are capped with coarse till and/or ice-rafted debris overlying thickly bedded and poorly sorted outwash sands. The terraces occur at similar elevations on both sides of the valley and they are considered to represent rapid sedimentation along the retreating Ellesmere Island ice margin following the removal of the ice shelf. These terraces are fossiliferous throughout; although most shells are fragmented, complete valves also occur. Terraces up-valley from the former grounding line, however, are not fossiliferous.
Along the west side of lower Daly River, a massive section of unconsolidated material occurs down-slope from the ice-shelf moraines (Fig. 4). The truncated face of this 50 m thick section reveals till overlying bedded sands containing preserved plant debris. The till is fossiliferous and its deposition is considered to have taken place during the formation of the ice shelf. The underlying fluvial (?) sands stratigraphically pre-date the till and their preservation puts maximum limits on the depth to which the floating ice shelf extended. Adjacent to this section are fossiliferous pro-glacial terraces which occur up to c. 105 m a.s.l. (Fig. 4). Additional fossiliferous terraces and raised beaches occur at the same elevation farther down-valley (Reference EnglandEngland, 1974). The uppermost post-glacial marine deposits, on the other hand, occur below these features at 90 m a.s.l. These fossiliferous terraces from "Beethoven" and Daly River valleys have been dated both by$$$ l4C and the amino-acid method, and are discussed under the section on chronology.
Former Relative Sea-Level
Estimates can be made on the thickness of these former ice shelves and, hence, the water depths required to float them. In "Beethoven Valley", a maximum estimate of ice thickness is based on the difference in elevation between the bedrock floor $$(c. go m a.s.l.) and the horizontal moraines (200 m a.s.l.). This suggests a maximum thickness of < 110 m for the ice shelf since it is assumed to be floating. As a result, the associated water depth must also be somewhat less than 0.88 X 110 m (Reference ReehReeh, 1969) or <97 m above the bedrock. This results in a maximum relative sea-level of < 187 m a.s.l. As discussed under the section on stratigraphy, pro-glacial terraces c. 175 m a.s.l. occur adjacent to these ice-shelf moraines in "Beethoven Valley”. The elevation of these graded terraces suggests water depths of c. 85 m above bedrock which would be capable of floating an ice thickness of e. 100 m. Such an ice thickness would extend from the horizontal moraines to within 10 m of the bedrock floor. It is suggested that these 175 m terraces, containing fragments of marine shells, represent the approximate relative sea-level that existed during the formation and break-up of the associated ice shelf. The deposition of these terraces clearly necessitated the removal of this ice shelf from the valley.
Along lower Daly River a maximum estimate on the ice thickness is c. 150 ni based on the difference in elevation between the preserved bedded sands (containing organic debris at c. 45 m a.s.l.) and the horizontal moraines (195 m a.s.l.). This would suggest a required water depth off. 130 m above the bedded sands (c. 45 m), hence a similar relative sea-level of c. 175 m at the time of ice-shelf formation. The absence of terraces at this elevation in the valley may be due to the fact that (1) they never formed, (2) they have been removed by subsequent erosion, or (3) the ice tongue stagnated in the lower valley, i.e. if the ice shelf remained in place during c. 15-20 m of emergence, it would have become grounded. Moraines on the valley sides at 150 m a.s.l. (below the horizontal moraines) may reflect such stagnation and grounding as the sea-level dropped from the ice-shelf stage. Following the déglaciation of" lower Daly River valley, there is evidence of a pre-Holocene shoreline at 105 m a.s.l. (Fig. 4) which occurs above the upper post-glacial marine deposits at c. 90 m a.s.l.
Chronology
In lower Daly River valley, fragmented shells collected from a pro-glacial terrace at 105 m a.s.l. dated 27 950+5 400 B.P. (St. 4325; Reference EnglandEngland, 1974). Amino-acid age estimates on the same shell sample, and from a subsequent collection from the same site, yielded ages >35 000 B.P. (Table I). 3 km up-valley there is an exposed section of fossiliferous till overlying bedded sand along the west side of the Daly River. The fossiliferous till is considered to represent deposition in a marine environment as it occurs below both the local ice-shelf moraine (Fig. 4) and the former relative sea-level at 175 m. In addition, no shells were observed up-valley from the ice-shelf moraines which precludes transport and re-deposition of these shells by ice from the interior of the promontory. Shell fragments from this marine till dated 28 610 + 1 170/-2 180 B.P. (DIC-550). The bedded sands underlying this till contain remnants of the locally extinct plant species Dryas octopetaia (persona] communication from J. Packer, 1976), a sample of which dated >25 000 B.P. (DIC-584). Adjacent to this section is a second fossiliferous terrace at c. 105 m a.s.l. (Fig. 4) containing shells dated at >35 000 B.P. by the amino-acid method.
Along the south-western slope of "Beethoven Valley" two samples of fragmented shells were collected in 1975. The first sample, incorporated in the horizontal ice-shelf moraines, dated $$$23 no__ B.P. (DIC-544), whereas the second, collected down-slope and washing out of the 175 m terrace, dated $$$22 780^ B.P. (DIC-546). X-ray analysis of both shell samples, however, revealed that they were encrusted with 50% calcite and 50% silica. These contaminants could not be completely removed and may date from the recrystallization of the shells following their initial deposition. Hence, it was concluded that both dates were minimum estimates. Amino-acid analyses of the same samples indicated ages >35 000B.P. (Table I).
During 1976 the 175 m terrace was re-visited and a second sample was collected. These shells occurred in the same location as sample DIC-546; however, they were not encrusted by contaminants. This sample dated $$$29 670 B.P. (DIG-738) and it takes precedence over the first date of $$$22 780 B.P.Footnote * This most recent date closely coincides with the other finite $$$l4C dates along the ice margin at Cape Defosse (27950-5400 and 28 610_ « B.P.; St. 4325 and DIC-550, respectively). However, as there are problems in obtaining reliable dates on marine shells >2θ 000 B.P. (Reference Olsson and BlakeOlsson and Blake, 1961 ; personal communication from M. Stuiver, 1977), it is most realistic to treat these $$$14C dates as minimum estimates. Amino-acid age estimates on the same samples are all >35 000 B.P, (Table I) and their true age may be as old as 70 000 B.P. Consequently, these c. 28 000-30 OOO B.P. dates on pro-glacial terraces should be considered as minimum estimates on initial recession from the ice margin which previously formed the ice shelves. On the other hand, the 175 m sea-level appears to be consistent with the water depths required to float the calculated ice thicknesses in both valleys and, hence, the ice shelves may not be substantially older than the pro-glacial terraces. Finally, it is apparent that the 175 m sea-level associated with these former ice shelves is substantially older than the highest post-glacial marine deposits (110 m.a.s.l.) observed on northern Judge Daly Promontory $$$(8 380-105 B.P. ; DIC-737).
Discussion
During the maximum ice advance on north-eastern Ellesmere Island, outlet glaciers formed ice shelves along western Kennedy Channel when the relative sea-level was c. 175 m above present. Evidence in support of these ice shelves is principally morphological and is based on the horizontality of two separate moraine systems which extend for 2 km beyond steeply descending moraines up-valley. These ice shelves are useful stratigraphically in that they delimit the extent of former ice margins and, chronologically, because they formed in a marine environment which favours the deposition of dateable fossiliferous units. In addition, these ice shelves provide estimates on the former relative sea-level at the time of their formation, since this can either be observed (pro-glacial terraces) or calculated, given the local ice thickness. On the basis of both the amino-acid and $$$I4C datings, it is presently concluded that the maximum north-eastern Ellesmere Island ice advance occurred within the period of the Wisconsin/Wurm glaciation. Because of the known problems in dating inorganic shell carbonate >20 000 B.P., we favour the amino-acid age estimates which place this ice advance at >35 000 B.p.
Throughout the Canadian and Greenland Arctic, numerous outlet glaciers descend from upland icefields into prominent fiords and embayments. In addition, glacio-isostatic depression during former glaciations resulted in the inundation of many low coastal valleys that are presently above sea-level. Therefore, during the advance of Arctic glaciers in the past, many ice shelves similar to those described from eastern Judge Daly Promontory must have formed. Where such ice shelves were constrained between steep valley sides, remnant depositional features, such as horizontal moraines, should be preserved. Air-photograph analysis of the coastline south of Cape Defosse, Judge Daly Promontory, indicates the presence of additional ice-shelf moraines. At present the extent of former ice shelves in the North American Arctic is open to considerable speculation (Reference MercerMercer, 1970; Reference BroeckerBroecker, 1975; Reference Hughes, Hughes, Denton and GrosswaldHughes and others, 1977) and hence the mapping, dating and analysis of their related deposits is pertinent to the understanding of high-latitude Quaternary environments.
Acknowledgements
This research was supported by the Climate Dynamics Program, Climate Dynamics Research Section, Division of Atmospheric Sciences, National Science Foundation (Grant OCD-00975). Additional logistic support was provided by the Polar Continental Shelf Project, Department of Energy, Mines and Resources, Ottawa. Radiocarbon dates were provided by the National Science Foundation Grant OCD-00975 and by a National Science Foundation Grant for Improving Doctoral Dissertations, NSFGA-35562. Mrs Irene Stehli, DICAR Corporation, Cleveland, Ohio, provided X-ray analyses of several shell samples. Dr J. Packer, Department of Botany, University of Alberta, identified the sample of Dryas octopetala. Dr V. K. Prest, Geological Survey of Canada, and Dr J. T. Andrews, INSTAAR, University of Colorado, provided helpful comments on the manuscript. Amino-acid dates were obtained by support of the National Science Foundation (Grant DES-74-01857).