Article body

Introduction

The mapping and altimetry of raised shorelines, both marine and lacustrine, have been standard components of Quaternary studies in glaciated North America since their beginning. The highest shorelines have received particular attention. Several regional to continental summaries of data have appeared as these data became increasingly available and better understood in terms of ice-sheet geometry. However, thirty-five years have now lapsed since the last continental synthesis (Andrews, 1972) and hundreds of additional measurements, about two-thirds of the data, are available. Assembling of these data is not a simple task, because authors rarely report precise locations of measured features and they commonly plot data on maps at scales that make it difficult or impossible to derive accurate co-ordinates. Nevertheless, this paper presents a reasonably complete inventory of available measurements (Appendix), a more detailed and better constrained contour map of marine- and lake-limit elevations, and an interpretation of the map as a reflection of the history of deglaciation. The map, and particularly the primary data that are contoured, should be useful replication targets for glacioisostatic rebound modelling. Some basic concepts are reviewed before presenting and interpreting the marine-limit and lake-limit data.

Background

Marine Limits

A marine limit, as used here, is the highest position (latitude, longitude and elevation) reached by the postglacial sea at a site, the elevation being measured with respect to present sea level (Andrews, 1970a). The Earth’s crust in heavily glaciated regions was strongly depressed by its load of thick ice, by approximately one-third the ice thickness. Depression extended beyond the ice margin due to crustal stiffness. Although global sea level was lowered during the last glaciation by about 120 m (Fairbanks, 1989; Clark and Mix, 2000), the beds of large ice sheets were depressed by much more than that amount a short distance inside of the ice-sheet limits. Consequently, the sea flooded the depressed areas upon deglaciation. However, crustal uplift due to ice-load removal exceeded global (eustatic) sea-level rise in most areas thereafter, thus causing relative sea level to fall. Therefore, with few exceptions in North America and Greenland, the marine limit is the local position of the sea surface at the time of deglaciation of a site. Hence, the marine limit does not form a synchronous shoreline at a continental or even at a regional scale. Rather, it varies in age almost as widely as does the timing of deglaciation. Known marine-limit shoreline features in North America thus range in age from about 14 500 14C BP to about 4000 14C BP (Dyke, 2004a).

Marine-limit ages do not span the full range of postglacial time. Where deglaciation occurred shortly after the last glacial maximum, crustal uplift initially caused emergence (Fig. 1, curves C, H, J). However, the net postglacial (eustatic) sea-level rise (ca. 120 m; Fig. 1, curve B) exceeded that initial emergence, leaving these earliest deglacial shorelines below present sea level just inboard of the glacial limit. There are thus no postglacial shorelines higher than present sea level in the area from New Jersey to the Atlantic coast of Nova Scotia, where deglaciation occurred between 18 000 and 14 500 14C BP. Similarly, all postglacial shorelines apparently are below present sea level in the area between the Yukon coast and western Amundsen Gulf, where deglaciation occurred between 18 000 and 12 000 14C BP. Areas deglaciated after these times have raised shorelines (Fig. 1, curves M and higher, except for H), because crustal depression, hence rebound, was larger toward the ice-sheet centre and postglacial eustatic sea-level rise (Fig. 1, curve B) was less the later the date of deglaciation.

Figure 1

The relationship between the present elevation of a shoreline formed at the time of deglaciation and the relative sea-level history of a site. Where the deglacial shoreline is above present sea level, it is referred to as the marine limit. Shown are approximate (Dyke, unpublished relative sea-level database) and hypothetical (dashed lines) relative sea-level curves for (P) Poste-de-la-Baleine, southeast Hudson Bay, at the postglacial uplift centre, and a sequence of curves for sites progressively closer to the limit of glaciation (less isostatically depressed): (L) Lac Saint-Jean, Québec, (N) St.Anthony, Newfoundland, (PM) Portland, Maine, (S) St. John’s, New Brunswick, (PE) northwestern Prince Edward Island, (H) Halifax, Nova Scotia, (M) Boston, Massachusetts, (C) southern Connecticut, (J) New Jersey, and (B) Barbados (Fairbanks, 1989). Relative sea-level rise at Halifax during the last 7000 years has exceeded the eustatic sea-level rise at Barbados because of crustal subsidence due to collapse of the glacioisostatic forebulge, which migrates inward as the radius of the area of postglacial uplift shrinks.

Relation entre l’altitude actuelle de la ligne de rivage formée au cours de la déglaciation et l’historique du niveau marin relatif d’un site. Dans les cas où le trait de côte durant la déglaciation est au-dessus du niveau marin actuel, on y réfère comme étant la limite marine. Les courbes du niveau marin relatif illustrées sont soit approximatives (Dyke, base de données non publiée du niveau marin relatif), soit hypothétiques (lignes pointillées) pour (P) Poste-de-la-Baleine, le sud-est de la Baie d’Hudson, au centre du soulèvement post-glaciaire, et pour une séquence de courbes pour des sites progressivement plus proches de la limite glaciaire (moins déprimées isostatiquement) : (L) Lac Saint-Jean, Québec, (N) St. Anthony, Terre-Neuve, (PM) Portland, Maine, (S) St. John’s, Nouveau-Brunswick, (PE) nord-ouest de l’Île-du-Prince-Édouard, (H) Halifax, Nouvelle-Écosse, (M) Boston, Massachussetts, (C) sud du Connecticut, (J) New Jersey et (B) Barbade (Fairbanks, 1989).

-> See the list of figures

A marine-limit elevation can be measured only where it is recorded by a shoreline feature, such as a delta, a beach, or an upper limit of wave erosion. Thus, in many places the position of the marine-limit shoreline is undefined, as is typical of steep rocky slopes. Probably the chief source of error in our current compilation of marine-limit elevations is the misinterpretation of a high shoreline feature in an area as a marine limit, where the true limit lies higher but may be weak or not detectible. It is also possible to overestimate marine-limit elevations, as where marine sediments or fossils are glacially thrust or redeposited above the contemporaneous sea level. Note that our definition of marine limit differs from some earlier ones, which, for example, define it as the “highest evidence in the form of a marine/littoral deposit or erosional form, such as a wave-cut terrace, reached by the sea in areas affected by glacioisostatic loading and unloading...” (Andrews, 1973), because we wish to emphasize that the highest “evidence” does not necessarily represent the true marine limit. Fortunately, many marine-limit shoreline features are unambiguous, such as ice-contact deltas (built at the receding glacier margin and graded to sea level of the time) and outwash sediment or meltwater channels that terminate at a raised beach. Andrews (1970a) provides a discussion of criteria for recognizing the marine limit in the field as do many regional geological reports. Because the marine limit is a fundamental and powerfully informative feature of regional Quaternary geology, numerous investigators have focussed on properly identifying and surveying it and we consequently now have a large body of information pertaining to it.

The known exceptions, where marine inundation occurred after, rather than precisely at the time of, deglaciation in Canada (see above), are as follows: (1) In James Bay and along the south shore of Hudson Bay, the ice front retreated northward in glacial Lake Agassiz-Ojibway to a position north of the subsequent marine-limit shoreline. When the sea later flooded in from the north, it thus formed the marine-limit shoreline synchronously across the entire southern part of Hudson basin (Dredge and Cowan, 1989; Veillette, 1994; Clarke et al., 2004; Dyke, 2004a). (2) Similarly, in the St. Lawrence and Ottawa valleys the ice front retreated northward in a proglacial lake. When the sea flooded in from the Gulf of St. Lawrence, it formed a synchronous shoreline south of the contemporaneous ice front (Occhietti et al., 2001). (3) On the Queen Charlotte Islands of British Columbia, deglacial shorelines are far below present sea level. The sea then rose from ‑150 m to a maximum of about 16 m elevation well after deglaciation (Clague et al., 1982; Clague, 1989; Josenhans et al., 1997). Relative sea level has subsequently fallen due to tectonic, as opposed to glacioisostatic, uplift (Clague et al., 1982). The Queen Charlotte Islands marine-limit data are not considered here, because they are not of deglacial age. The Hudson Bay and St. Lawrence-Ottawa data are shown, because marine-limit shorelines formed there within a century or two after deglaciation. Nevertheless, in these two regions where the marine limit forms a synchronous shoreline, the marine-limit surface is steeper than it would have been had it formed diachronously.

Isolines or profiles drawn through marine-limit elevations define a virtual topography, which we here call the marine-limit surface (Fig. 2). That surface is everywhere a direct measure of the net change of elevation since deglaciation, regardless of whether a site is at the coast, inland, or offshore. The surface is the net result of two opposing variables that determine relative sea-level history: (1) the amount of depression of Earth’s crust by the ice sheets during the last glaciation, and (2) the amount of uplift (emergence plus eustatic sea-level rise) that occurred prior to deglaciation. The depression varied spatially from nil to an order of 10m, being about one-third of ice thickness. Postglacial emergence was initially rapid, with a half-time of 1000-2000 years at most sites (Dyke and Peltier, 2000). Therefore, the marine-limit surface reflects both gross ice-sheet geometry in its overall form, and the rates of ice recession in its details (Fig. 2). Note on this figure that the marine-limit gradient is always less than the gradient of any contemporaneous shoreline in its vicinity although it approaches the shoreline gradient as the rate of ice recession increases, as shown for the interval 9000 to 8000 BP. When ice recession is relatively slow, as shown for the interval 11 000 to 10 000 BP, the marine-limit surface dips in the direction opposite to that of the shorelines. This is referred to as a negative marine-limit gradient (dipping toward the ice-load centre), all others being positive or zero. Had the ice margin readvanced between 11 000 and 10 000 BP to the position shown by the open vertical bar on the 10 000 BP shoreline, a cliff with a negative gradient of 90° would have been formed on the marine-limit surface, as indicated by the vertical dashed line.

Figure 2

The schematic relationship between raised shoreline gradient, marine limit gradient, and rate of ice marginal recession. Numbered solid lines are the profiles of hypothetical raised shorelines with the numbers indicating age in thousands of years. Thick vertical bars represent ice margins. The open bar represents a stillstand or readvance of the ice margin. Dashed extensions of the shoreline profiles represent the increase in elevation of the glacier bed since the times indicated. The heavy line connecting ice marginal positions is the marine-limit surface.

Relation schématique entre le gradient des lignes de rivages, le gradient des limites marines et le taux de récession des marges glaciaires. Les lignes numérotées représentent les profils hypothétiques des lignes de rivage soulevées, les chiffres indiquant l’âge en milliers d’années. Les barres verticales représentent les marges glaciaires et la barre vide représente un arrêt ou une avancée de la marge glaciaire. Les pointillés représentent l’augmentation de l’altitude du lit glaciaire depuis le temps indiqué. La ligne épaisse reliant la position des marges glaciaires délimite la surface marine.

-> See the list of figures

Lake Limits

The maximum water levels that occurred in large glacial lakes as ice fronts retreated across their basins define lake-limit surfaces. These are similar to the marine-limit surface as long as the lake used the same outlet, the outlet was not substantially lowered by erosion, and the outlet was located at the least uplifted end of the lake. All other outlet positions would cause shoreline transgressions ice-distal from the outlet. In other words, had the sea invaded the glacial lake basin, the marine-limit isolines would have trended parallel to the lake-limit isolines, although the marine-limit isolines would have been lower (Fig. 3). According to published interpretations, these conditions pertain to glacial Lake McConnell and Lake Mackenzie in the Mackenzie River drainage basin (Smith, 1992, 1994), to the Ojibway phase of glacial Lake Agassiz-Ojibway (Veillette, 1994), and to glacial Lake Hitchcock in the Connecticut River valley in New England (Koteff et al., 1993).

There is, however, a difference between a lake-limit surface and a marine-limit surface. Although both were affected equally in neighbouring localities along the same isobase by the isostatic uplift that occurred as the ice front retreated, the marine-limit surface was additionally affected by eustatic sea-level rise. Because emergence is isostatic uplift minus eustatic sea-level rise, marine-limit elevations and gradients were reduced by eustatic rise. Although the elevation of a lake basin was also reduced by eustatic sea-level rise, the tilting of the basin with respect to its outlet was not. The consequence is that for adjacent lake and marine basins, the lake-limit surface should rise more steeply than the marine-limit surface.

Figure 3 compares shoreline and water-limit profiles in adjacent marine and lake basins with identical deglaciation and uplift histories. Part A shows hypothetical emergence curves for site A (curve Ae) and site B (curve Be) and uplift curves for the same sites, adding the eustatic corrections from curve B in Figure 1. Site A, at marine limit, and the outlet sill (A’) of the glacial lake have identical uplift histories (part B of Fig. 3). Site B is at marine limit 100 km closer to the uplift centre and it shares an identical uplift history with a site (B’) in the lake basin 100 km in the same direction from the outlet. The lower part of part B of Figure 3 shows profiles of the 9, 10, and 11 ka BP shorelines (solid lines labelled 9 and 10) or their projections behind the contemporaneous ice margins (dashed line labelled 11) resulting from curves Ae and Be. It also shows the amount of uplift since 9, 10, and 11 ka BP (lines labelled 9u, 10u, 11u) resulting from curves Au and Bu. The ice front receded across point A at 11 ka BP and across point B at 9 ka BP, with a constant rate of recession during the interim. Thus at the 50-km mark, the 10-ka shoreline is at 25 m. The resulting marine-limit surface (bold line), which starts at 20 m at point A, rises to 25 m at the 50-km mark and intersects the 9-ka shoreline at point B, returning to an elevation of 20 m (positive and negative gradients of 1:10 000). The upper part of part B of Figure 3 (note change of vertical scale) shows lake shoreline profiles for the same times or their projections behind the contemporaneous ice margins (dashed lines). Point B’ has been uplifted 40 m more than point A’ (lake sill) since 11 ka BP, 28 m more since 10 ka BP, and 20 m more since 9 ka BP, according to curves Au and Bu in part A of Figure 3. The lake-limit surface rises from the sill to intersect the 10-ka shoreline midway between points A and B at 13 m above the sill. It intersects the 9-ka shoreline 20 m above the sill. Therefore, the lake-limit surface is steeper (net rise of 20 m) than the marine-limit surface (net rise 0 m), because only the latter is dampened by eustatic sea-level rise during the interval over which it formed.

Figure 3

A

B

Relationship between a lake-limit surface and an adjacent marine-limit surface formed over the same time interval. (A) Emergence (Ae, Be) and uplift (Au, Bu) curves for sites A and B. (B) Site B is 100 km closer to the uplift centre than site A. Sites A and B are at marine limit. Site A’ is at a glacial lake outlet sill on same isobase as site A. Site B’ is on same isobase as site B. Numbered lines (9, 10, 11) are shoreline profiles; lines labelled 9u etc. show uplift profiles. Heavy lines represent marine-limit (lower) and lake-limit profiles (upper).

Relation entre la surface d’une limite lacustre et celle d’une limite marine adjacente qui a été formée en même temps. (A) Courbes d’émergence et de soulèvement (Au, Bu) pour les sites A et B. (B) Le site B est 100 km plus près du centre de soulèvement que le site A. Les sites A et B occupent la limite marine. Le site A’, à l’exutoire d’un lac glaciaire, est sur la même isobase que le site A. Le site B’ est sur la même isobase que le site B. Les lignes numérotées (9, 10, 11) sont les profils des lignes de rivage; les lignes identifiées par 9u, etc., représentent les profils de soulèvement. Les lignes épaisses représentent les profils des limites marines (bas) et des limites lacustres (haut).

-> See the list of figures

It is evident, therefore, that marine-limit and lake-limit topographies are features that need to be reconciled with each other and with the pattern and chronology of deglaciation, on the one hand, and with geophysical models of glacioisostatic adjustment on the other.

Previous Marine-Limit Maps

The most recent marine-limit map of Canada was prepared by John Andrews for the National Atlas of Canada using data available as of 1969 (Andrews, 1972; scale 1:30 000 000). The map was also included in a journal article that reviewed previous renderings of marine-limit isolines (Andrews, 1973). The earliest maps recognized a prominent region of maximum emergence over Québec-Labrador or southeastern Hudson Bay and placed the zero isoline over Atlantic Canada and New England with reasonable accuracy (De Geer, 1892; Fairchild, 1918; Daly, 1934). The first continental-scale map, which however excluded the Pacific coast, was by Farrand and Gajda (1962). It was made possible by the great expansion of field observations throughout the Arctic after the second world war, largely by the officers of the Geological Survey of Canada and the former Canadian Geographical Branch (Prest et al., 1968). The map was reasonably accurate in placing centres of maximum emergence in southeastern Hudson Bay and a second centre in the High Arctic, as well as in the values shown there and elsewhere. Also, it illustrated the tilted shorelines of glacial Lake Algonquin in the Great Lakes region and of Lake Agassiz west of there, so as to indicate the limit of the uplifted region and the trend of isobases. The map was unfortunately marred by misleading data (terraces and other features of non-marine origin [Ives, 1963]) from eastern Baffin Island, showing emergence of 150-180 m in regions now known to have emerged by less than 50 m. Andrews (1972) constructed his map for the National Atlas by selecting the maximum recorded marine-limit elevation within each of 125 grid cells, placing these values at the cell centres, and contouring them, thus filtering out lower values. No lake-limit data were used, nor were data from the Pacific coast included.

Method

Isolines on the current map are simple objective contours of all available measurements of marine-limit elevations (n = 929) for which reasonably accurate co-ordinates are available (Fig. 4; Appendix). Minimum and maximum elevations of marine limits are not knowingly included, because these are not amenable to straightforward contouring. Inland trends of marine-limit isolines, where not guided by lake-limit isolines, are based on the general form of the continental ice sheet. Published statements that geological features represent local marine limits are accepted at face value, as they were in previous syntheses. Primary references are given wherever possible. For several sites, we have retained information from the Glacial Map of Canada (Prest et al., 1968) for which we were unable to identify a primary source. These were probably supplied directly to the authors (Prest et al., 1968) from unpublished field notes, primarily by officers of the Geological Survey of Canada. Data points are not plotted for the lake limits, but these may be found in Koteff et al. (1993) for Lake Hitchcock, in Veillette (1994) for Lake Ojibway, in Dredge (1983a) and Klassen (1983) for Lake Agassiz, and in Smith (1992, 1994) for lakes Mackenzie and McConnell. The Hitchcock and Ojibway limits are very well constrained by measurements.

Pattern

Marine Limits

The zero isoline (Fig. 5) represents the geographical limit of net postglacial emergence. Its position resembles that shown by Andrews (1972) in roughly mimicking the limit of glaciation. It is fixed just south of Boston in the northeastern U.S.A. (Stone and Peper, 1982) and trends across central Nova Scotia (Prest et al., 1968, 1972; Stea et al., 2001) and Prince Edward Island (Prest et al., 1968). The location of the zero isoline across the Gulf of St. Lawrence is problematic, depending on acceptance or rejection of a 37-m feature shown on the Magdalen Islands. This feature was first proposed as the postglacial marine limit by Robert Chalmers (see Goldthwait, 1915) and figured as such by Prest et al. (1968). However, that interpretation was questioned by Grant (1989), who placed the zero isoline north of these islands and the deglacial shoreline on the Magdalen Shelf at ‑75 m. Shaw et al. (2002) followed Grant’s interpretation but did not explicitly consider the problem. The 37-m feature is, however, accordant with marine-limit elevations in southwestern Newfoundland and in northwestern Prince Edward Island, on either side of the Gulf directly to the east and west. It is considered to be a postglacial marine feature by M. Parent (Geological Survey of Canada, personal communication, 2004) based on elevations of probable raised beach sands. It is, therefore, tentatively accepted here. The zero isoline is placed offshore from southern and southeastern Newfoundland, in contrast to earlier renditions, because postglacial raised shorelines have been reported throughout that region (Brookes, 1977: 9 m at Wreckhouse; Tucker et al., 1982: 6 m at Fortune; Rogerson and Tucker, 1972: 11 m at Spear Island; Catto et al., 2000: 6 m at Horse Cove). It is far offshore from northeastern Newfoundland to Hudson Strait, as required by the high marine limits along the Labrador coast, which however decline to 15 m in the north (Løken, 1962; Ives, 1963). The zero isoline intersects Resolution Island at the mouth of Hudson Strait, but marine limit rises to 21 m on adjacent Edgell Island (Kaplan and Miller, 2003). It is then located offshore along most or all of eastern Baffin Island but onshore in southeastern Devon Island, where deglacial meltwater channels can be traced down to the present shoreline (Dyke, 1998). It is apparently located offshore around the entire length of Greenland (Funder, 1989) and off the northwestern Canadian polar margin as far as its landfall at Cape Bathurst, east of the Mackenzie Delta (Prest et al., 1968). The position of the zero isoline near Banks Island remains somewhat problematic, because no radiocarbon dates are available to document postglacial (as opposed to earlier) marine overlap of the western part of that island. Nevertheless, because the last glacial limit has been placed at the moraines flanking the north and south coast of that island (Dyke and Prest, 1987; Dyke, 2004a), the associated raised shorelines at 20 m (e.g. the ice-contact delta at Sachs Harbour; Vincent, 1983) are necessarily of the same age as the moraines and are so plotted here. In the Pacific, the deglacial marine-limit zero isoline passes approximately through Seattle, Washington (Thorson, 1980), is offshore of Vancouver Island, but evidently lies landward of the Queen Charlotte Islands (Josenhans et al., 1997). It is undefined along the southeast coast of Alaska.

Figure 4

Sites of recorded marine-limit elevations listed in the Appendix.

Altitude des limites marines présentées en annexe.

-> See the list of figures

The most salient features of the marine-limit topography (Fig. 5) are similar to those on the map of Andrews (1972). These features are (1) the cell of highest values in southeastern Hudson Bay, here reaching 270 m (Hillaire-Marcel, 1980; Vincent, 1989), (2) the cell of high values near Bathurst Inlet in the western arctic mainland, reaching 228 m (Kerr, 1996), and (3) the cell of high values over the Eureka Sound region between Ellesmere and Axel Heiberg islands in the Canadian High Arctic, reaching 156 m (Bell, 1996; Bednarski, 1998). The shapes of these cells differ from the earlier rendition, particularly that of the Bathurst Inlet cell, which is shown as more elongate southward, the trend of the isolines being guided by the lake-limit isolines of glacial Lake McConnell. Two additional high cells on the marine-limit surface are located over eastern Boothia Peninsula to northeastern Keewatin, reaching 255 m (Dyke, 1984; Giangioppi et al., 2003), and in southern Foxe Basin, reaching 195 m (Bird, 1970; Laymon, 1988). These cells are less clear on the earlier map of Andrews (1972) because of the fewer data then available. Also similar to the map of Andrews (1972), is the region of high marine limits along the Ottawa-St. Lawrence valley, reaching 250 m (Fulton, 1987; Occhietti, 1989). Although Vincent (1989) showed the highest marine limit in the Ottawa region as 275 m in the Gatineau valley, Occhietti (1989) considered that area to have been inundated by fresh water, and the 275-m site of Vincent is, therefore, not included here. Andrews (1972) showed the high values of the Ottawa region as an extension of the southeast Hudson Bay cell. However, on Figure 5 these are shown as a closed cell of high values with a negative (downward) northwestward slope, because the trend of lake-limit isolines across Lake Ojibway (Veillette, 1994) does not allow connection of high marine-limit contours between the two regions.

Figure 5

Deglacial marine- and lake-limit surfaces, North America and Greenland. Dashed isolines are more tentatively placed than others. The framed insets show lake-limit isolines for glacial Lake Mackenzie (smaller) and Lake McConnell (larger) in the Northwest Territories, a portion of the Ojibway phase of Lake Agassiz in northern Manitoba, a portion of Lake Ojibway in Québec and Ontario, and Lake Hitchcock in New England. The marine-limit isolines are at 25-m intervals starting at 0 m. Lake Mackenzie isolines are at 20-m intervals starting at 80 m elevation, Lake McConnell at 20-m intervals starting at 180 m, Lake Agassiz at 20-m intervals starting at 200 m, Lake Ojibway at 20-m intervals starting at 245 m, and Lake Hitchcock at 25-m intervals starting at 45 m.

Surfaces marines et lacustres de déglaciation, pour l’Amérique du Nord et le Groenland. Les isolignes pointillées sont plus incertaines que les autres. Les encadrés montrent les isolignes des limites lacustres du lac glaciaire Mackenzie (petit) et du lac McConnell (grand) dans les Territoires du Nord-Ouest, une partie de la phase Ojibway du lac Agassiz au nord du Manitoba, une partie du lac Ojibway au Québec et en Ontario, et le lac Hitchcock en Nouvelle-Angleterre. Les isolignes des limites marines sont espacées de 25 m et débutent à 0 m. Les isolignes du lac Mackenzie sont espacées de 20 m et débutent à 80 m, celles du lac McConnell sont espacées de 20 m et débutent à 180 m, celles du lac Agassiz sont espacées de 20 m et débutent à 200 m, celles du lac Ojibway sont espacées de 20 m et débutent à 245 m et celles du lac Hitchcock sont espacées de 25 m et débutent à 45 m.

-> See the list of figures

Prominent regions of low marine-limit elevations located well inboard of the limit of glaciation occur in north-central Keewatin, where elevations fall below 100 m (Lee, 1968; Prest et al., 1968; Dredge and McMartin, 2005), on the Ungava Peninsula of Québec, where elevations fall below 125 m (Lauriol, 1982) and over the spine of Baffin Island, where elevations fall below 50 m. The last are the lowest values to be found inboard of higher values in the glaciated region (Blake, 1966; Andrews, 1970b; Dyke, 1979b). Also of note are the negative slope of 100 m northward across Foxe Basin and the 100 m positive (upward) slope northeastward across northern Ontario.

The edge of a final cell of high values in Canada, the centre of which must well exceed 230 m (Miller, 1973; Clague, 1989), is located along the Pacific coast. These high values might reasonably be extrapolated across the central Cordilleran Ice Sheet region. Although we currently lack lake-limit data to guide the trend of the isolines there, lines of 100 m and greater value must close to make a separate cell over the Cordilleran region if the 100-m line over the Laurentide Ice Sheet region is correctly tied into the lake-limit isolines of Lake McConnell from the north.

Three cells with marine-limit elevations exceeding 100 m are shown in Greenland (Funder, 1989). One of these, reaching 140 m in the northwest (England, 1985), is an extension of the cell over the Canadian High Arctic. The other two, reaching maximum values of 140 m (Kelly, 1973) and 115 m (Funder, 1990), coincide with the widest strips of ice-free land in southwestern and central-eastern Greenland, respectively. Lower ridges of marine-limit values, exceeding 50 m, extend along the coast from these cells.

Lake Limits

Lake-limit surfaces generally rise, as expected, toward the centre of the glaciated region. The Lake McConnell and Lake Ojibway (in Québec and easternmost Ontario) surfaces are particularly useful in guiding the trends of marine-limit isolines through interior regions. Unfortunately, similar data are not available for the vertical limits of Lake Ojibway in most of its extent in Ontario.

In the case of Lake Hitchcock and Lake Ojibway, the lake-limit surfaces have gradients that are clearly greater than those of the adjacent marine-limit surfaces, which is to be expected as discussed above. Both these lakes show only positive northward gradients, with that of Lake Hitchcock being essentially constant and that of Ojibway convex upward. Lakes Mackenzie and McConnell have entirely positive lake-limit gradients. However, the McConnell gradient decreases toward the centre of the ice sheet. The McConnell gradient exceeds the gradient of the adjacent marine-limit surface to the north, as expected. The lake-limit surface of the Ojibway phase of Lake Agassiz is evidently more complex with a positive northward gradient followed by an irregular negative gradient at its north end.

Discussion

Marine Limits

In Canada, most of the cells of high marine limits correspond approximately to areas of inferred maximum former ice thickness. The cells over southeastern Hudson Bay and Québec, over Bathurst Inlet and areas to the south, and over southern Foxe Basin roughly reflect the areas of maximum ice thickness of the Labrador Sector, the Keewatin Sector, and the Baffin Sector of the Laurentide Ice Sheet, respectively (Dyke and Prest, 1987; Dyke et al., 2002; Dyke, 2004a). The cell over the High Arctic corresponds to the area of maximum thickness of the Innuitian Ice Sheet (Blake, 1970; Dyke, 1998; Atkinson and England, 2004). Finally the cell over the Cordillera is related to uplift engendered by removal of the Cordilleran Ice Sheet, which had its thickest ice in the region between the coastal mountain ranges and the Rocky Mountains (Clague, 1989). If the marine-limit surface along the Pacific coast were extrapolated a short distance inland, the indicated postglacial increase in elevation over central British Columbia would exceed that of southeast Hudson Bay, which is generally taken to be the highest in the world.

In contrast to the interpretation above, the high marine-limit elevations over the St. Lawrence-Ottawa region and over the Gulf of Boothia region are interpreted as resulting from incursion of the sea into areas that were still greatly depressed at time of deglaciation rather than former ice-load centres (Occhietti, 1989; Dyke and Dredge, 1989). This interpretation is compatible with the most recent interpretation of pattern of deglaciation (Dyke, 2004a).

A trough of low marine limits (<150 m) over western Hudson Bay may indicate a separation of Keewatin and Labrador sector uplifts that was not previously recognized. This feature is defined by a series of marine-limit determinations in northern Manitoba that decline to a low of 125 m (Klassen, 1986; Dredge and Nixon, 1992) and by a single determination of 124 m on Coats Island (Aylsworth and Shilts, 1991). In the former area, the apparent marine limit is just above a prominent marine feature known as the Great Beach. All water-laid sediment above it apparently was deposited in glacial Lake Agassiz. The Great Beach is traceable for over 180 km and rises from 125 m at its south end to about 145 m at its north end. In addition to its large form, it is conspicuous because the Lake Agassiz basin inland of it lacks beaches (Klassen, 1986; Dredge and Nixon, 1992) due to the sudden final drainage of the lake (Clarke et al., 2004). The apparent marine limit on Coats Island is a clear upper limit of raised beaches, above which is evidently unmodified till. Thus, the trough of low marine-limit values appears to be a valid feature. It is not readily interpretable as a product of late deglaciation.

In Greenland, the highest marine limits correspond to areas of greatest ice-load reduction, rather than to areas of greatest former ice thickness, because most of that ice load is still present. Thus, the highest raised shorelines are in areas where the coastal ice-free zone is widest and values decline from the outer or central coasts toward the present ice margin.

Other elements of the marine-limit topography can best be interpreted in terms of the pattern of deglaciation. Among these are the following clear examples: (1) The low values over northern Keewatin and the spine of Baffin Island, discussed above, correspond to areas of late deglaciation, as does the long negative northward slope of values across Foxe Basin. (2) The steep negative eastward slope on the marine-limit surface on western Melville Peninsula, one of the most prominent features of the map and where values decrease eastward from 225 m to 125 m, corresponds with a large regional end moraine, the Melville Moraine, from which the ice retreated slowly enough for 100 m of emergence to occur (Dredge, 1990). (3) A westward drop in marine-limit elevation of similar magnitude at the head of Frobisher Bay on southeastern Baffin Island corresponds to a major belt of end moraines there (Blake, 1966; Miller, 1980), and the southwestward decline of values from a ridge over the northeast Baffin Island fiords similarly corresponds with a dense belt of end moraines (Andrews et al., 1970; Hodgson and Haselton, 1974). (4) The abrupt drop of marine limit on southern Melville and northwestern Victoria islands from 90-120 m on the distal side of Winter Harbour Till to 30-55 m on the proximal side records emergence that occurred between the onset of ice recession prior to the Winter Harbour Advance of the Laurentide Ice Sheet and the onset of recession after the advance (Hodgson and Vincent, 1984). (5) The decrease of values from the outer coasts to the spine of Devon Island by 25-50 m reflects inland recession of the ice cap on that island (Dyke, 1998). (6) The negative marine-limit slope of about 100 m northwest of the St. Lawrence lowlands indicates slow ice recession across that region. (7) The decline of marine-limit elevations by as much as 50-60 m inland from the southeast coast of Hudson Bay reflects emergence that occurred while ice stood at and retreated behind the Sakami Moraine (Allard and Seguin, 1985). (8) Smaller negative gradients related to slow ice recession occur along central, northern, and east-central Ellesmere Island fiords (Hodgson, 1985; Smith, 1999; England et al., 2000, 2004) and along fiords in western Newfoundland (Grant, 1987) and British Columbia (Friele and Clague, 2002). More detailed mapping of marine limits will probably bring to light other areas where strong negative marine-limit gradients can be related to the history of ice recession.

Lake Limits

A problematic feature of the lake-limit data is the difference between the lake-limit elevation of northernmost Lake Hitchcock in the Connecticut River valley and the marine limit in the adjacent Lake Champlain valley. The northernmost Lake Hitchcock limit, at 245 m, is uplifted by 200 m more than its southern outlet, at 45 m (Koteff et al., 1993). The 0-m marine-limit isoline near Boston (which is also the 0-m isobase on the 14 500 14C BP marine shoreline) is here extrapolated westward to correspond to the 70-m level of Lake Hitchcock, leaving northernmost Lake Hitchcock 175 m higher than that. The marine limit in that part of the Lake Champlain valley that lies along the same isobase as northernmost Lake Hitchcock and 100 km to the west-southwest is at 67 m (West Bridport site of Stewart and MacClintock, 1969; correlated using the isobase trend of Parent and Occhietti, 1988). If both elevations and the inferred history of Lake Hitchcock are correct, the difference of about 108 m can only be accommodated by the rise of eustatic sea level between the time that the Lake Hitchcock outlet in Connecticut was deglaciated (15 400 14C BP based on the age of the basal varve in southernmost Lake Hitchcock [Ridge, 2004]; note that the age of this varve has been revised from 14 820 14C BP of Ridge et al. [1999]) and the time of formation of the marine limit in the Lake Champlain valley. Dyke (2004a) placed the latter at about 11 500 14C BP, although it may date as late as 11 100 14C BP (Richard and Occhietti, 2005; recall that marine limit is synchronous in this region). Relative sea level at Barbados, the most commonly cited modern eustatic sea-level reference, was at about ‑68 m and ‑72 m at 11 100 and 11 500 14C BP, respectively (Fairbanks, 1989). Because eustatic sea level at 15 400 14C BP was at ‑113 m, only 45 m of eustatic sea-level rise occurred between that time and 11 100 14C BP. The problem cannot be resolved even by allowing deglaciation of the lake outlet to be much earlier than thought because eustatic sea-level rise between 18 000 and 11 100 14C BP was only 60 m. Thus, an error in either the interpretation of the lake history or marine limits is indicated. An attempt to reconcile these data might start with evaluating the assumption that Lake Hitchcock was still using its southernmost outlet when its northernmost lake-limit shorelines were forming.

A further issue involves the negative slope on the lake limit of northernmost Lake Agassiz, here shown as less steep than would be indicated by several of the lower sets of shorelines assigned to Lake Agassiz by Dredge (1983a) and Dredge et al. (1986). The pattern of deglaciation in this region, a simple northward recession of the ice margin toward the 60th parallel, is not controversial, because it seems clear from the landform record (Dredge et al., 1986). The fragment of the Lake Agassiz record that is preserved in northernmost Manitoba is probably related to the Ojibway phase of Lake Agassiz when outflow was via the Ottawa River headwater in Québec and when the southern shore of Lake Agassiz in Manitoba either barely enclosed Lake Winnipeg or later ended north of Lake Winnipeg (Klassen, 1983, 1986; Thorleifson, 1996; Dyke, 2004a). The Ottawa River was the last available outlet of Lake Agassiz prior to catastrophic northward drainage to Hudson Bay (Clarke et al., 2004). In that reconstruction, northward declining lake limits and shorelines below the lake limit within northern Lake Agassiz can only relate to differential uplift unless numerous subglacial drainage events occurred. Ignoring lake stages lower than 340 m and assuming an Ottawa River outlet throughout, data indicate a negative northward lake-limit slope of about 100 m across about 1.5° of latitude at the north end of the basin. Thus, during the time the ice margin retreated across that part of the basin, 100 m more uplift occurred there than occurred at the Ottawa River outlet. Lake Agassiz was emptied by northward drainage at about 7700 14C BP, since which time 180 m of emergence (about 200 m of uplift) has occurred on the adjacent coast of Hudson Bay. Current reconstruction of ice retreat (Dyke, 2004a) places the beginning of ice recession from the highest Agassiz shorelines north of Reindeer Lake (442 m) at 8000-8200 14C BP. Thus in a span of 300-500 years, that region was uplifted 100 m more than was the Ottawa River outlet. Because the implied uplift rate is much larger than others currently known for areas of Laurentide glaciation, the pattern of lake limits shown may be a misleading target for uplift modelling. Alternative interpretations of the lake shoreline data might involve (a) considering whether multiple subglacial drainages or partial drainages of Lake Agassiz occurred (as implicit in Dredge, 1983a), or (b) considering whether these shorelines were formed in lakes independent of Lake Agassiz, such as a lake in the Reindeer Lake basin.

Conclusion

Marine-limit and certain lake-limit surfaces can now be reconstructed reasonably well for much of Canada and Greenland and in greater detail for the northern, central and eastern Arctic, as well as for southeastern Canada and New England. In these details can be recognized the influences of maximum ice-load distributions and the pattern and relative rates of deglaciation as currently understood. These virtual topographies form an independent set of data that already have been largely reconciled with the pattern of deglaciation, because marine-limit ages known from radiocarbon dating are a large part of the age control used for deglaciation maps (Dyke, 2004a). Hence, they are valuable targets for replication by geophysical models of glacioisostatic adjustment, which traditionally have chosen other and perhaps more straightforward targets, such as relative sea-level curves. Nevertheless, certain aspects of the primary data, lake-limit data in particular, are problematic and alternative interpretations might be considered. Modelling may bring to light other currently accepted interpretations that are problematic.

Future syntheses of marine- and lake-limit surfaces should aim at capturing full local details of variability. Researchers would greatly contribute to these syntheses by reporting geographic co-ordinates of their measured sites.