Available online at www.sciencedirect.com

2008) 237–248www.elsevier.com/locate/geomorph

Geomorphology 97 (

Ice-walled-lake plains: Implications for the origin of hummockyglacial topography in middle North America

Lee Clayton a,⁎, John W. Attig a, Nelson R. Ham b, Mark D. Johnson c,Carrie E. Jennings d, Kent M. Syverson e

a Wisconsin Geological and Natural History Survey, 3817 Mineral Point Road, Madison, Wisconsin 53705, USAb Geology Department, St. Norbert College, 100 Grant Street, DePere, Wisconsin 54115, USA

c Department of Geology, Earth Science Centre, Box 460, SE-405 30 Göteborg, Swedend Minnesota Geological Survey, 2642 University Avenue, St. Paul, Minnesota 55114, USA

e Geology Department, University of Wisconsin, Eau Claire, Wisconsin, 54702, USA

Received 12 September 2006; received in revised form 4 December 2006; accepted 27 February 2007Available online 23 June 2007

Abstract

Ice-walled-lake plains are prominent in many areas of hummocky-till topography left behind as the Laurentide Ice Sheet meltedfrom middle North America. The formation of the hummocky-till topography has been explained by: (1) erosion by subglacialfloods; (2) squeezing of subglacial till up into holes in stagnant glacial ice; or (3) slumping of supraglacial till. The geomorphologyand stratigraphy of ice-walled-lake plains provide evidence that neither the lake plains nor the adjacent hummocks are of subglacialorigin. These flat lake plains, up to a few kilometers in diameter, are perched as much as a few tens of meters above surroundingdepressions. They typically are underlain by laminated, fine-grained suspended-load lake sediment. Many ice-walled-lake plainsare surrounded by a low rim ridge of coarser-grained shore sediment or by a steeper rim ridge of debris that slumped off thesurrounding ice slopes. The ice-walled lakes persisted for hundreds to thousands of years following glacial stagnation. Shells ofaquatic molluscs from several deposits of ice-walled-lake sediment in south-central North Dakota have been dated from about13500 to 10500 B.P. (calibrated radiocarbon ages), indicating a climate only slightly cooler than present. This is confirmed byrecent palaeoecological studies in nearby non-glacial sites. To survive so long, the stagnant glacial ice had to be well-insulated by athick cover of supraglacial sediment, and the associated till hummocks must be composed primarily of collapsed supraglacial till.© 2007 Elsevier B.V. All rights reserved.

Keywords: Ice-walled-lake plain; Glacial hummocks; Supraglacial debris; Pleistocene–Holocene transition

⁎ Corresponding author. Tel.: +1 608 263 6839; fax: +1 608 2628086.

E-mail addresses: lclayton@wisc.edu (L. Clayton),jwattig@wisc.edu (J.W. Attig), nelson.ham@snc.edu (N.R. Ham),markj@gvc.gu.se (M.D. Johnson), carrie@umn.edu (C.E. Jennings),syverskm@uwec.edu (K.M. Syverson).

0169-555X/$ - see front matter © 2007 Elsevier B.V. All rights reserved.doi:10.1016/j.geomorph.2007.02.045

1. Introduction

In many places in the mid-continent area of NorthAmerica, ice-walled-lake plains are a conspicuouscomponent of tracts of hummocky-till topography. Thepurpose of this paper is to detail evidence that the ice-walled-lake plains are of supraglacial origin, includingpalaeontological evidence that they could not haveformed subglacially. Because of the close association of

238 L. Clayton et al. / Geomorphology 97 (2008) 237–248

ice-walled-lake plains and the hummocky-till topogra-phy, we argue that the hummocky till was also primarilyof supraglacial origin.

Till hummocks have been interpreted to have formedin several different ways, including by erosion insubglacial floods (Munro and Shaw, 1997), by squeezingof subglacial material up into holes through stagnantglacial ice (Stalker, 1960; Eyles et al., 1999; Boone andEyles, 2001), and by slumping of supraglacial materialdown into holes through stagnant glacial ice (Gravenorand Kupsch, 1959; Johnson et al., 1995; Johnson andClayton, 2003).

The ice-walled-lake plains in these tracts of hum-mocky topography have typically been overlooked orunrecognized, and they are rarely mentioned in regionalreports or textbooks on glacial geology. Yet they arewidespread in areas of thick glacial sediment along thesouthern and southwestern marginal zone of theLaurentide Ice Sheet (Fig. 1). Here we review the mor-phology and sedimentology of ice-walled-lake plains inmid-continental North America and evaluate what theytell us about the origin of hummocky till.

Many ice-walled-lake plains can be identified withconfidence because they have distinctive characteristics.They typically consist of offshore sediment surrounded bya ring of shore-zone sediment (Fig. 2a,b). Commonly, the

Fig. 1. Locations of the known ice-walled-lake plains in North Dakota, MinEach is represented by a spot that has been exaggerated by adding 0.5 km to iline, and the solid line on the inset map is the approximate farthest extent of th

outward-facing ice-contact face at the edge of the plainslumped when the support was removed as the surround-ing icemelted (Fig. 2b). Inmany places, the lake sedimentis tens ofmeters thick. In contrast to typical lake sediment,the ice-walled-lake sediment today is not in a basin, but iselevated above the surrounding topography.

Avariety of names has been applied to these features,typically using various combinations of words such asdead-ice or stagnant-ice, ice-walled or perched, lakeor lacustrine, and plains or plateaus. Parizek (1969)called them moraine lake plateaus.

2. Sources of information

Much of the information about ice-walled-lake plainsis scattered in local reports on the glacial geology ofspecific areas, many not widely available. Our intenthere is to summarize observations from several decadespast and bring them up to date with more recentfindings. Our emphasis is on North Dakota, Minnesota,and Wisconsin, because this area has abundant ice-walled-lake plains, most of which have been inventoriedin reports of the three state geological surveys (Fig. 1),although we know of numerous examples in Iowa,South Dakota, Illinois, Manitoba, Saskatchewan, andAlberta as well as in Denmark, Sweden, Norway,

nesota, and Wisconsin, compiled largely from references cited in text.ts periphery. Major scarps are indicated on the inset map by a hachurede Laurentide Ice Sheet during the last part of the Wisconsin Glaciation.

Fig. 2. Schematic cross sections of ice-walled-lake plain while still confined by ice (a) and after ice melted. (b). Vertical ruling in sloughs represents faintlybedded Holocene sediment; horizontal ruling in non-glacial lake represents latest Pleistocene and earliest laminated fossiliferous Holocene sediment.

239L. Clayton et al. / Geomorphology 97 (2008) 237–248

Poland, Lithuania, Germany, and Russia (Johnson andClayton, 2003; Johnson et al., 2004).

The purpose of those reports was to map and describethe geology of an area, generally a county or regionalstudy. Some of these contain detailed discussions of ice-walled-lake plains and were based on interpretations ofaerial photographs and large numbers of observations ofshallow exposures, generally roadcuts, with some drillholes and excavations.

3. Description

3.1. Abundance

In North Dakota (Fig. 1), 170 ice-walled-lake plainslarger than about 1 km have been shown on a 1:500000state-wide compilation (Clayton, 1980; many smallerones exist) derived from 1:125000 maps in bulletinspublished by the North Dakota Geological Survey in the1960s and 1970s. All occur on either the TurtleMountains, a 60-km-wide upland in northern NorthDakota and southern Manitoba, or on the Coteau duMissouri (Missouri Coteau), an upland 20 to 50 kmwide,extending 1000 km southeast across Saskatchewan, NorthDakota, and SouthDakota (Fig. 1). Numerous ice-walled-lake plains are shown on 1:125000 maps in bulletins ofthe South Dakota Geological Survey (Koch, 1975;Christensen, 1977; Leap, 1988; Hammond 1991). Allare on either the Coteau du Missouri or the Coteau desPrairies (Prairie Coteau), a triangular upland that extendsinto Minnesota (Fig. 1).

In Minnesota, hundreds of ice-walled-lake plainsformed in broad zones along the west flank of the Des

Moines Lobe on the Coteau des Prairies in southwesternMinnesota (Patterson, 1992, 1995, 1997; Patterson et al.,1999) and along its east flank in the southeast and south-central part of the state (Patterson and Hobbs, 1995;Lusardi, 1997; Lusardi et al., 2002). They also occur, butless abundantly, in other parts of the state, where they tendto be more subtle and of lower relief than those formed onthe flanks of the Des Moines Lobe (Fig. 1; Meyer, 1985,1993a, 1993b; Hobbs et al., 1990; Meyer et al., 1990,2001; Patterson, 1992; Meyer and Hobbs, 1993; Harrisand Knaeble, 1999; Meyer and Lusardi, 2000; Pattersonand Knaeble, 2001; Lusardi et al., 2002; Syverson et al.,2005). About 330 ice-walled-lake plains have beenidentified and shown on 1:100000maps of theWisconsinGeological and Natural History Survey (Clayton, 1986,1987, 2001; Johnson, 1986, 2000;Mickelson, 1986; Attigand Muldoon, 1989; Attig, 1993; Ham and Attig, 1997;Mickelson and Syverson, 1997).

3.2. Morphology

Like present-day lakes, ice-walled lakes tend to haverounded outlines, especially the smaller ones. Few ice-walled-lake plains are more than 5 km in diameter. Manyless than 0.5 km have been recognized but have generallynot been mapped. Many ice-walled-lake plains are ashigh as, or even twice as high as, the surrounding tillhummocks (in the concluding remarks, we argue that thecorrelation between the height of hummocks and theheight of ice-walled-lake plains is evidence that thehummocks are of supraglacial origin).

Ice-walled-lake plains form a morphologic continu-um with low to high relief (Clayton and Cherry, 1967;

240 L. Clayton et al. / Geomorphology 97 (2008) 237–248

Ham and Attig, 1996, 1997; Syverson et al., 2005). Thelower ones, less than about 20 m high, tend to beslightly dished, gently sloping downward from an outerrim into a central level area. Many low ones have adistinct ridge around the edge of the plain, some ofwhich are several meters above the middle of the plain(Figs. 3 and 4). The higher ice-walled-lake plains (20 to50 m high) tend to be flat to slightly convex, gently

Fig. 3. Two low-level ice-walled-lake plains with rim ridge beaches, surroucontact faces; 8 km north of Medford, Taylor County, Wisconsin; sec. 36, T(b) Part of the U.S. Geological Survey Medford 7.5-minute topographic quadthe view of the photograph. Star symbols show location drill holes mentioned

sloping from a central flat area down past a rim thatlacks a ridge (Fig. 5a,b). Rim ridges might have origi-nally existed but slumped away from the higher ice-contact faces.

Some ice-walled-lake plains appear to have had morethan one phase of development (Johnson, 1986; Attig andMuldoon, 1989; Attig, 1993). For example, in Fig. 4a, icewalls apparently at first surrounded the lake plain above

nded by hummocky collapsed till. Dotted lines mark the crest of ice-. 32 N., R. 1 E.; from Attig, 1993, Figs. 24 and 25. (a) View to west.rangle; contour interval 3 m (10 ft). The large dashed V symbol marksin text. North is to right in the photograph. Photograph by John Attig.

Fig. 4. (a) Two-phase terraced ice-walled-lake plain. Part of the U.S. Geological Survey Rosholt NW 7.5-minute quadrangle. Contour interval is 10 ft(∼3 m). Located in sec. 34, T. 26 N., R. 9 E. in Marathon County, Wisconsin (Attig and Muldoon, 1989). (b) Ice-walled-lake plain, cultivated,surrounded by hummocky till. Dotted line marks the crest of the ice-contact face around the lake plain. Located 19 km southwest of Ross, NorthDakota, in sec. 6 and 7, T. 154 N., R. 94 W. (Clayton, 1972, Plates 1 and 2). U.S. Department of Agriculture aerial photograph BAL-4 V45.(c) Oblique aerial photograph of ice-walled-lake plain surrounded by till hummocks with central depressions, located 11 km south-southwest ofStanley, North Dakota, in sec. 23 and 26, T. 155 N., R. 92W. The plain is 0.5 km wide. Dotted line marks the crest of the ice-contact face. Photographby Lee Clayton.

241L. Clayton et al. / Geomorphology 97 (2008) 237–248

Fig. 5. Diagram (a) and photograph (b) of a section through a small delta in the rim of an ice-walled-lake plain. The topset and foreset beds are slightlygravelly sand, and the bottomset beds are silty. This section was exposed in a sand and gravel pit 2 km south of the town of Turtle Lake, BarronCounty, Wisconsin, in the SE1/4 NW1/4 SE1/4 sec. 6, T. 33 N., R. 14 W. From Johnson (1986, Fig. 31). (c) Small ice-walled-lake plain, 18 m high,11 km northeast of Napoleon, Logan County, North Dakota, in sec. 24, T. 136 N., R. 72 W. (Clayton, 1961, Plate 1). Photograph by Lee Clayton.

242 L. Clayton et al. / Geomorphology 97 (2008) 237–248

the 1280 ft (390 m) contour. The ice walls receded and alower lake plain formed at the 1230 ft contour.

3.3. Sedimentology

The central part of most low-lying lake plains isunderlain by offshore clay, silt, and fine sand, grading tocoarser near-shore sand, with beach sand and gravel atthe rim (Figs. 5a,b and 6a,b; Clayton and Cherry, 1967;Johnson, 1986; Attig, 1993; Ham and Attig, 1997;Johnson, 2000; Syverson et al., 2005). The offshoresediment is horizontally laminated, and near the marginthe bedding in the coarser sediment dips at a few degreestoward the centre. Faults are commonly observed nearthe ice-contact faces of the lake plains.

Some of the lower-level ice-walled-lake plains haverim ridges with steep outer slopes (ice-contact faces) butwith gentle inner ones. These rim ridges are typicallycomposed of beach sediment, but some have poorlysorted debris flows interbedded with the shore or near-

shore sediment (Fig. 6a,b; Attig, 1993; Attig et al.,1998). The gently sloping rims of some ice-walled-lakeplains have bodies of sand and gravel interpreted to bedeltas with high-angle foreset beds dipping toward thecentre of the lake plain (Fig. 5a,b). A delta extendinginto a lake plain is visible in the southwest part of thelake plain shown in Fig. 4a.

Other low-lying ice-walled-lake plains have rimridges that are steep on both sides. They are composedof material that is similar to the till surrounding the ice-walled-lake plain and is thought to be material slumpedoff the surrounding ice.

3.4. Thickness of lake sediment

Few ice-walled-lake plains have been systematicallyinvestigated, but the thickness of lake sediment inoutcrops or drill holes has been noted in several regionalstudies. In North Dakota and Wisconsin, the lakesediment is often reported to be as thick as or thicker

Fig. 6. (a) Cross section of the rim ridge of an ice-walled-lake plain. Slumped ice-contact face is on the right (southeast) side of ridge. Shore and near-shoregravelly sand, till, and offshore silt and fine sand interfinger on the left (northwest) side of ridge. (b) Close-up view of till on top of offshore sediment, locatedat the rectangle in themiddle of the outcrop. (c) Part of U.S. Geological Survey Perkinstown 7.5-minute topographic quadrangle. This exposure is in a smallgravel pit indicated by a star. The crest of the ice-contact face around the lake plain is marked with a dotted line. Contour interval is 3 m (10 ft). Located1.5 km south of Perkinstown, Taylor County, Wisconsin. From Attig (1993, Fig. 19) and Attig et al. (1998, Fig. 1.10). Photographs by Nelson Ham.

243L. Clayton et al. / Geomorphology 97 (2008) 237–248

than the height of the plain above adjacent inter-hummockdepressions. The following paragraphs detail the observa-tions of thickness known to us in these two states. Lakesediment is 15 to 30 m thick in drill holes in ice-walled-lake plains of North Dakota where typical till hummocksare 5 to 15 m high, with some 30 m high (Clayton, 1962;Clayton and Cherry, 1967; Clayton and Freers, 1967).Clayton (1962) mapped an ice-walled-lake plain in thesame area with more than 23 m of horizontally laminatedclay and silt exposed in a roadcut on one side and a similarthickness of stratified sand, gravel, and lacustrine clay andsilt with numerous gravity faults in gravel pits on its otherside. Royse and Callender (1967) found about 5 m of lake

sediment in the concave middle of three large but low-relief ice-walled-lake plains in northwest North Dakota.They found more than 17 m in the middle of a large high-level ice-walled-lake plain in the same area, where thesurrounding till hummocks are up to 20 m high.

In northwest Wisconsin, Johnson (1986) reporteddrill holes with 7 to 10 m of lacustrine silt with inter-bedded debris-flow sediment in low-level ice-walled-lake plains, where the surrounding till surface has about10 to 15 m of local relief. Cahow (1976) reported 23 mof lake silt in a drill hole in an ice-walled-lake plain innorthwestern Wisconsin, where the local till relief is‘tens of feet’. Attig (1993; unpublished data) drilled

Fig. 7. Chronology of the Pleistocene transition in south-central NorthDakota based on calibrated radiocarbon dates (refer to text).

244 L. Clayton et al. / Geomorphology 97 (2008) 237–248

through 3 to 4 m of laminated clayey silt and fine sandoverlying gravelly sand and till at three sites in themiddle part of the lake plain shown in Fig. 3, and 4 m ofgravelly sand in the rim ridge. Twelve additional siteswith 3 to more than 20 m of laminated lake sedimentobserved in drill holes in ice-walled-lake plains areknown to us (Ham, 1994; Attig et al., 1998).

The presence of thick offshore sediment in high-relief ice-walled-lake plains indicates an abundantsource of silt and clay up the slope above lake level.These observations of lake-sediment thickness confirmthat lake processes shaped these features, which are notflat-topped till hills with only a thin layer of postglacialsediment on top.

4. Identification

Ice-walled-lake plains generally have a distinctivemorphology (Fig. 2). They are commonly first recog-nized because they are much flatter than surroundingareas of hummocky till (Figs. 2, 3, 4, 5c and 6c), andbecause they are perched well above the present-daywetlands in the surrounding depressions, and becausefarm fields are free of pebbles, cobbles, and boulders(Figs. 3a,b, 4a–c, 5c and 6c). Many are conspicuousbecause they are cultivated land surrounded by grazingland (Fig. 4b) or by woodland (Fig. 3a).

Hummocky collapsed supraglacial outwash or supra-glacial lake deposits are common in some hummocky-tillregions (Attig, 1993). Their morphologies are similarand may be hard to distinguish from each other and fromhummocky till without subsurface information. In oneroadcut through till hummocks in north-central Wiscon-sin, Attig (1993) noted that crude bedding plains couldbe seen to dip downward toward the high points in thelandscape. He interpreted this to indicate that at least theupper parts of the hummocks were of supraglacial originand that all the hummocks are gradational with ice-walled-lake plains.

5. Fossils, climate, and chronology

The period of treeless tundra vegetation and perma-frost in the proglacial environment during active-icedeglaciation at the end of the Wisconsin Glaciation ispoorly known in the mid-continent area, because of thescarcity of wood for radiocarbon dating (Clayton et al.,2001). However, this generally frigid scene ended duringthe better-dated Pleistocene–Holocene transition, per-haps several hundred years after active-ice deglaciation,with a marked warming of the climate (Fig. 7; Yansa,2006). This is significant in understanding ice-walled-

lake plains becausemany persisted through the transitionand into earliest Holocene time.

In hummocky-till areas of the northeast Great Plainsthis striking change has been documented in two sets ofpalaeoecologic studies. One is based on mollusc shellspreserved in the youngest ice-walled-lake deposits in thehummocky-till area of south-central North Dakota. Theother is based on organic remains in non-glacial lakedeposits of the same age and in the same region as theice-walled-lake plains.

Together these two suites of fossils show that some ofthe ice-walled lakes persisted for hundreds or thousandsof years in a temperate climate. This indicates aninsulating blanket of thick debris on top of thesurrounding stagnant glacial ice.

5.1. Ice-walled lake suite

At the time of the Pleistocene–Holocene transition, theclimate warmed, and the ice-walled lakes that survivedwere colonized by a variety of organisms. The shells of 28species of mollusc were recognized in ice-walled-lakesediment, as well as in hummocky collapsed supraglaciallake sediment and in hummocky collapsed supraglacialoutwash at 40 sites in south-central North Dakota(Clayton, 1961, 1962; Tuthill, 1967). Most of thesefossils are unleached and retain well-preserved morpho-logic details, permitting identification to species level.Also present are the shells of at least ten species ofostracods and the oogonia (reproductive bodies) ofcharophytes (a bushy green alga). Generally, onlycalcareous fossils have been preserved; lime has beenleached from only the upper several centimeters of soil inthis area. Most non-calcareous fossils have beendestroyed because all the fossil sites are above the watertable and are well oxidized, but the presence of fish isindicated by the presence of the glochidia (larvae) ofmussels, which are obligate parasites of fish. Recently,Brandon Curry (Illinois State Geological Survey) has

245L. Clayton et al. / Geomorphology 97 (2008) 237–248

found shells of ostracods that lived in ice-walled lakes inIllinois.

Most of these fossil sites were roadcuts, and all werewell above any nearby bodies of water. All the fossilswere found in sediment that could only have been de-posited in depressions that were walled or floored bystagnant glacial ice. Few of the shells show any evidenceof transport; many of the mussel shells were still articu-lated, indicating they remain where they lived, died, andwere subsequently buried by lake or stream sediment.

According to Tuthill et al. (1964) and Tuthill (1967),the climate at the time molluscs lived in the ice-walledand supraglacial environments in south-central NorthDakota was moist enough to support a mixed coniferand deciduous woodland and was only slightly coolerthan at present. It may have been like present-day north-central Minnesota, 400 km to the northeast, where fresh-water lakes have outlets and stable water levels. Incontrast, the environment of the Coteau du Missouritoday and through most of Holocene time had dry hotsummers with prairie vegetation. Modern sloughs andmost lakes are ephemeral, lack outlets, and are alkaline,and harbour a different association of molluscs thanthose that lived in the ice-walled lakes.

Mussel shells in seven of the ice-walled deposits havebeen radiocarbon dated: 10123±399; 10966±365;11298±195; 11531±494; 11711±505; 12944±279;13497±302 B.P. (Moran et al., 1973). In addition,radiocarbon dates of 12088±482 and 12119±477 B.P.from white-spruce wood in hummocky till on the Coteaudu Missouri in northwestern North Dakota probablymark the time of tree burial during the collapse ofsupraglacial till (Pettyjohn, 1967). All dates presentedhere have been calibrated using the method described byFairbanks et al. (2005).

5.2. Non-glacial-lake suite

The fossils just described from the ice-walled andsupraglacial stream and lake deposits on the Coteau duMissouri are now in high positions in the landscape,well above, and unrelated to, present-day sloughs, lakes,and streams. However, complementary climatic evi-dence comes from studies of pollen and macrofossilscollected from non-glacial lake sediment underlying themodern slough deposits, in the same general area as theice-walled-lake plains.

Most sloughs in south-central North Dakota containseveral meters of Holocene sediment, which consists offaintly bedded or non-bedded silt and clay with few non-calcareous fossils as a result of bioturbation andoxidation during widely fluctuating Holocene water

levels. Below the faintly bedded sediment in manysloughs is finely laminated organic lake clay, containingabundant well-preserved plants and animals that lived atthe same time as the molluscs in the ice-walled lakes.

This lake sediment occurs under the slough sedimentin less-hummocky places where the stagnant glacial icemelted earlier than the ice around nearby better-insulated ice-walled lakes. This lake sequence wasdeposited in deeper water with more-stable water levelsand less-oxidizing bottom conditions, resulting in betterfossil preservation than in the later Holocene sloughs. Atone site in south-central North Dakota (Cvancara et al.,1971), more than 160 species have been identified,including beaver, muskrats, frogs, molluscs, numerousyellow perch and other fish, 81 species of beetles, aspenleaves, and white-spruce cones and needles, with beaver-gnawed wood radiocarbon dated at 11155±182B.P.(calibrated date). Yansa (2002, 2006) studied severalsites like this and concluded that south-central NorthDakota may have had tundra vegetation for a short timefollowing deglaciation, which was replaced by a white-spruce parkland (savanna) from about 14500 to about13500 B.P. (calibrated radiocarbon dates), followed bya deciduous parkland until about 11500 B.P., which inturnwas replaced by grassland through theHolocene. Theparkland vegetation resembled the aspen parkland exist-ing today 400 km to the north and northeast in northwestMinnesota and southern Manitoba.

This corroborates Tuthill's (1967) interpretation thatthe climate that exists today just northeast of the prairie–woodland boundary is similar to the climate that existed insouth-central North Dakota during at least the final stagesofmany of the ice-walled lakes. This is also the conclusionreached by Florin andWright (1969) andWright (1980) innorthern Minnesota, where forest-floor litter occurs at thebottom of Holocene pond and lake sequences, indicatingthat buried ice lasted into a time of moderate climate andthat trees growing on the debris-covered ice were let downinto the final collapse depressions as the ice melted.

Although the fossil preservation in these non-glaciallakes was very different from the preservation in the ice-walled lakes, the two suites of organisms were con-temporaneous and lived in the same general area, andfor that reason their ecological conditions were verysimilar. Together they indicate that the ice-walled lakespersisted into the early Holocene warm period and werenot subglacial.

6. Discussion

Reasoning by analogy with permafrost thaw lakesgives some insight into the formation of supraglacial

246 L. Clayton et al. / Geomorphology 97 (2008) 237–248

and ice-walled lakes (Attig, 1993; Ham, 1994; Ham andAttig, 1998). Both ice-walled lakes and permafrost thawlakes (thermokarst lakes) occur where ice is presentbelow the surrounding land surface. The basin of a thawlake deepens because the ground ice continually thawsbeneath the lake if it is too deep to freeze solid duringthe winter. Where unfrozen, the average annual temper-ature at the bottom must stay above the freezing point.The flow of heat downward from the lake water andupward from within the earth will cause the permafrostor stagnant glacial ice to gradually melt away beneaththe lake. However, if everything is frozen solid up to theland surface or lake surface during the winter, conduc-tion of heat to the atmosphere will prevent the perma-frost or glacier ice from melting. Therefore, once asupraglacial lake is deep enough to remain unfrozenduring the winter, the lake will eventually melt its waythrough underlying stagnant glacial ice, initially restingon a small area of solid ground, eventually becoming anice-walled lake, and then perhaps merging with adjacentice-walled lakes before final drainage.

An explanation is needed for the way the lake isinitiated and becomes deep enough for the bottom waterto remain unfrozen in the winter. A starter lake mightappear where the thickness of supraglacial debris isirregular, causing differential insulation and melting,which could result in depressions with lakes too deep tofreeze solid. However, the distribution of individual ice-walled-lake plains within a group of ice-walled-lakeplains generally has no obvious relationship to anyknown pattern of debris irregularity. Some ice-walled-lake plains are arranged in rows parallel to rows of lakesin ice-block depressions. However, their distributiongenerally seems to be haphazard, depending on theirregular arrangement of glacial debris on stagnantglacial ice. If the initial surface of the supraglacial debrisis too flat for supraglacial starter lakes, such as on asupraglacial outwash plain, no ice-walled lakes canform. The presence of collapse hummocks composed ofsupraglacial lake sediment between many ice-walled-lake plains indicates that these supraglacial lakes weretoo shallow or short lived to melt their way to the bottomof the ice (Attig, 1993).

Permafrost may have played an important role in thedevelopment of ice-walled-lake plains by inhibiting themelting of the debris-covered ice (Ham and Attig,1996). For example, Attig (1993), Clayton et al. (2001)and Attig et al. (2003) argued that some of the highestice-walled-lake plains, 4 km from the outermost limit ofthe late Wisconsin glacial advance, at least 60 m abovethe limit, could not have existed unless the drainage-ways through the glacier and through the coarse-grained

till under the ice were frozen shut. Ham et al. (1993) andHam and Attig (1996) also proposed a landscape modelfor the deglaciation of hummocky areas in northernWisconsin, suggesting that permafrost delayed themelting of stagnant buried glacial ice for a few thousandyears after active glaciation and stabilized the ice-coredlandscape, allowing the formation of large, well-developed ice-walled lakes.

7. Conclusions

In the previous sections, we have described themorphology, sedimentology, palaeontology, and chro-nology of these flat-topped hills in enough detail todemonstrate that they are ice-walled-lake plains. Theirelevated position above the surrounding till hummocksand the distribution of lake-sediment facies within themare possible only if the sediment was deposited withinice walls that have subsequently melted. The presence ofmolluscs in the lakes and the radiocarbon dates indicatethat the glacial ice was well-insulated from thesurrounding ice. The insulating material could onlyhave been the same supraglacial till that makes up thehummocks around the lake plains.

Any explanation of hummocky-till topography mustaccommodate associated ice-walled-lake plains. Mol-luscs could not live under a glacier, and ice walls couldnot have survived during hundreds or thousands of yearsof temperate climate unless they were well-insulated.Although it is certainly possible that there is somesubglacial component of hummocky-till topography inthe mid-continent area, the geomorphology and stratig-raphy of ice-walled-lake plains argue for a primarilysupraglacial origin for the lake plains and much of thesurrounding hummocky topography.

Acknowledgments

We thank Armand LaRoque and John Hiemstra fortheir many suggestions for improving the manuscript.Lee Clayton thanks Jon Rau for introducing him to ice-walled-lake plains.

References

Attig, J.W., 1993. Pleistocene geology of Taylor County, Wisconsin.WisconsinGeological andNatural History SurveyBulletin 90, 25 pp.

Attig, J.W., Muldoon, M.A., 1989. Pleistocene geology of MarathonCounty, Wisconsin. Wisconsin Geological and Natural HistorySurvey Information Circular 65, 27 pp.

Attig, J.W., Ham, N.R., Mickelson, D.M., 1998. Environments andprocesses along the margin of the Laurentide Ice Sheet in north-central Wisconsin (Guidebook for the 44th Friends of the

247L. Clayton et al. / Geomorphology 97 (2008) 237–248

Pleistocene Field Conference). Wisconsin Geological and NaturalHistory Survey Open-File Report 1998–01. 62 pp.

Attig, J.W., Clayton, Lee, Johnson, M.D., Patterson, C.J., Ham, N.R.,Syverson, K.M., 2003. Ice-walled-lake plains in themid-continent—what they tell us about late glacial ice marginal processes and envi-ronments. Geological Society of America Program and Abstracts 35,61.

Boone, C.H., Eyles, N., 2001. Geotechnical model for Great Plainshummocky moraine formed by till deformation below stagnant ice.Geomorphology 38, 109–124.

Cahow, A.C., 1976. Glacial Geology of the Southwest Segment of theChippewa Lobe Moraine Complex, Wisconsin. Ph.D. dissertation,Michigan State University, East Lansing, 227 pp.

Christensen, C.M., 1977. Geology and water resources of McPherson,Edmunds, and Faulk Counties, South Dakota. South DakotaGeological Survey Bulletin 26, Part 1, 58 pp.

Clayton, L., 1961. Late Wisconsin Molluska from ice-contact depositsin Logan County, North Dakota. Proceedings of the North DakotaAcademy of Science 15, 11–18.

Clayton, L., 1962. Glacial geology of Logan and McIntosh Counties,North Dakota. North Dakota Geological Survey Bulletin 37, 84 pp.

Clayton, L., 1972. Geology of Mountrail County, North Dakota. NorthDakota Geological Survey Bulletin 55-IV, 70 pp.

Clayton, L., 1980. Geologic Map of North Dakota. U.S. GeologicalSurvey.

Clayton, L., 1986. Pleistocene geology of Portage County, Wisconsin.Wisconsin Geological and Natural History Survey InformationCircular 56, 19 pp.

Clayton, L., 1987. Pleistocene geology of Adams County, Wisconsin.Wisconsin Geological and Natural History Survey InformationCircular 59, 14 pp.

Clayton, L., 2001. Pleistocene geology of Waukesha County,Wisconsin. Wisconsin Geological and Natural History SurveyBulletin 99, 33 pp.

Clayton, L., Freers, T.F., 1967. Roadlog. North Dakota GeologicalSurvey Miscellaneous Series 30, 1–24.

Clayton, L., Cherry, J.A., 1967. Pleistocene superglacial and ice-walled lakes of west-central North America. North DakotaGeological Survey Miscellaneous Series 30, 47–52.

Clayton, L.,Attig, J.W.,Mickelson,D.M., 2001. Effects of late Pleistocenepermafrost in the landscape ofWisconsin, USA. Boreas 30, 173–188.

Cvancara, A.M., Clayton, L., Bickley Jr., W.B., Jacob, A.F., Ashworth,A.C., Brophy, J.A., Shay, C.T., Delorme, L.D., Lammers, G.E.,1971. Paleolimnology of late Quaternary deposits. Seibold Site,North Dakota. Science 171, 172–174.

Eyles, N., Boyce, J.I., Barendregt, R.W., 1999. Hummocky moraine:sedimentary record of stagnant Laramide Ice Sheet lobes resting onsoft beds. Sedimentary Geology 123, 163–174.

Fairbanks, R.G., Mortlock, R.A., Chiu, T.-C., Cao, L., Kaplan, A.,Guilderson, T.P., Fairbanks, T.W., Bloom, A.L., Grootes, P.M.,Nadeau, M.-J., 2005. Radiocarbon calibration curve spanning 0 to50,000 years BP based on paired 230Th/234U/238U and 14C dates onpristine corals. Quaternary Science Reviews 24, 1781–1796.

Florin, M.B., Wright Jr., H.E., 1969. Diatom evidence for thepersistence of stagnant glacial ice in Minnesota. Bulletin of theGeological Society of America 80, 695–704.

Gravenor, C.P., Kupsch, W.O., 1959. Ice-disintegration features inwestern Canada. Journal of Geology 67, 48–64.

Ham, N.R., 1994. Glacial Geomorphology andDynamics ofWisconsinValley Lobe of the Laurentide Ice Sheet, Lincoln County,Wisconsin. Ph.D. dissertation. University of Wisconsin, Madison,235 pp.

Ham, N.R., Attig, J.W., 1996. Ice wastage and landscape evolutionalong the southern margin of the Laurentide Ice Sheet, north-central Wisconsin. Boreas 25, 171–186.

Ham, N.R., Attig, J.W., 1997. Pleistocene geology of Lincoln County,Wisconsin. Wisconsin Geological and Natural History SurveyBulletin 93, 31 pp.

Ham, N.R., Attig, J.W., 1998. Conceptual model for the developmentof ice-walled lakes along the southern margin of the Laurentide IceSheet: implications for deglacial dynamics. Geological Society ofAmerica Program and Abstracts 30, 113.

Ham, N.R., Attig, J.W., Clayton, L., 1993. The role of permafrost inthe development of hummocky disintegration topography with ice-walled-lake plains in north-central Wisconsin, U.S.A. INQUACommission on the Formation and Properties of Glacial Deposits,Field Conference and Workshop Program and Abstracts, p. 6.

Hammond, R.H., 1991. Geology of Lake and Moody Counties, SouthDakota. South Dakota Geological Survey Bulletin 35, 49 pp.

Harris, K.L., Knaeble, A.R., 1999. Surficial geology of OttertailCounty, Minnesota. Minnesota Geological Survey RegionalHydrologic Assessment 5 part A, plate 1.

Hobbs, H.C., Aronow, S., Patterson, C.J., 1990. Surficial geology.Minnesota Geological Survey County Atlas Series, p. C6. plate 3.

Johnson, M.D., 1986. Pleistocene geology of Barron County,Wisconsin. Wisconsin Geological and Natural History SurveyInformation Circular 55, 42.

Johnson, M.D., 2000. Pleistocene geology of Polk County, Wisconsin.WisconsinGeological andNatural History SurveyBulletin 92, 70 pp.

Johnson, M.D., Clayton, L., 2003. Supraglacial landsystems in lowlandterrain. In: Evans,D.J.A. (Ed.),Glacial Landsystems.Arnold, London,pp. 228–258.

Johnson, M.D., Mickelson, D.M., Clayton, L., Attig, J.W., 1995.Composition and genesis of glacial hummocks, western Wiscon-sin. Boreas 24, 97–116.

Johnson, M.D., Attig, J.W., Clayton, L., Patterson, C.J., Ham, N.R.,Syverson, K.M., 2004. Ice-walled-lake plains in North America andEurope—description, genesis, and paleoglaciological implications.26th Nordic Winter Meeting, GFF, Programme with Abstracts,vol. 126, p. 120.

Koch, N.C., 1975. Geology and water resources of Marshall County,South Dakota. South Dakota Geological Survey Bulletin 23, 76 pp.

Leap, D.I., 1988. Geology and hydrology of Day County, SouthDakota. South Dakota Geological Survey Bulletin 24, 117 pp.

Lusardi, B.A., 1997. Surficial geology of Waseca County, Minnesota.Minnesota Geological Survey Miscellaneous Map Series M-84.

Lusardi, B.A., Hobbs, H.C., Patterson, C.J., 2002. Surficial geology of theFaribault 30×60 minute quadrangle, south-central Minnesota.Minnesota Geological Survey Miscellaneous Map Series M-130.

Meyer, G.N., 1985. Quaternary geologic map of the Minneapolis–St.Paul urban area, Minnesota. Minnesota Geological SurveyMiscellaneous Map Series M-54.

Meyer, G.N., 1993a. Surficial geology of parts of Koochiching, Itascaand Beltrami Counties, Minnesota. Minnesota Geological SurveyMiscellaneous Map Series M-76.

Meyer, G.N., 1993b. Quaternary geologic map of Chisago County,Minnesota. Minnesota Geological Survey Miscellaneous MapSeries M-78.

Meyer, G.N., Hobbs, H.C., 1993. Quaternary geology map ofSherburn County, Minnesota. Minnesota Geological Survey MapSeries M-77.

Meyer, G.N., Lusardi, B.A., 2000. Surficial geologic map ChicagoCounty, Minnesota. Minnesota. Geological Survey Map Miscel-laneous Map Series M-106.

248 L. Clayton et al. / Geomorphology 97 (2008) 237–248

Meyer, G.N., Baker, B.R., Patterson, C.J., 1990. Surficial geology.Minnesota Geological Survey County Atlas Series C-5.

Meyer, G.N., Knaeble, A.R., Ellingson, J.B., 2001l. Surficial geologyof the St. Cloud 30×60 minute quadrangle. Minnesota GeologicalSurvey Miscellaneous Map Series M-115.

Mickelson, D.M., 1986. Glacial and related deposits of LangladeCounty, Wisconsin. Wisconsin Geological and Natural HistorySurvey Information Circular 52, 30 pp.

Mickelson, D.M., Syverson, K.M., 1997. Quaternary geology ofOzaukee and Washington Counties, Wisconsin. Wisconsin Geo-logical and Natural History Survey Bulletin 91, 56 pp.

Moran, S.R., Clayton, L., Scott, M.W., Brophy, J.A., 1973. Catalog ofNorth Dakota radiocarbon dates. North Dakota Geological SurveyMiscellaneous Series 53, 51 pp.

Munro, M., Shaw, J., 1997. Erosional origin of hummocky terrain insouth-central Alberta, Canada. Geology 25, 1027–1030.

Parizek, R.R., 1969. Glacial ice-contact ridges and rings. GeologicalSociety of America Special Paper, vol. 123, pp. 49–102.

Patterson, C.J., 1992. Surface geology. In: Meyer, G.N., Swanson, L.M.(Eds.), Geologic Atlas of Ramsey County, Minnesota. MinnesotaGeological Survey County Atlas C-7.

Patterson, C.J., 1995. Surface geologic map. In: Setterholm, D.R.(Ed.), Regional Hydrologic Assessment: Quaternary Geology—Southwestern Minnesota. Minnesota Geological Survey RHA-2.

Patterson, C.J., 1997. Surface geology of southwestern Minnesota.In: Patterson, C.J. (Ed.), Contributions to the Quaternary Geology ofSouthwestern Minnesota. Minnesota Geological Survey Report ofInvestigations 47, 1–45.

Patterson, C.J., Hobbs, H.D., 1995. Surface geology. In: Hobbs, H.C.(Ed.), Geologic Atlas of Rice County, Minnesota. MinnesotaGeological Survey County Atlas C-9.

Patterson, C.J., Knaeble, A.R., 2001. Surficial geology. Geologic atlasof Pine County, Minnesota. Minnesota Geological Survey CountyAtlas C-13.

Patterson, C.J., Knaeble, A.R., Gran, S.E., Phippen, S.J., 1999.Surficial geology. In: Patterson, C.J. (Ed.), Regional Hydrogeolo-

gic Assessment: Quaternary Geology—Upper Minnesota RiverBasin. Minnesota Geological Survey Regional HydrogeologicAssessment Series RHA-4.

Pettyjohn, W.A., 1967. Multiple drift sheets in southwestern WardCounty, North Dakota. North Dakota Geological Survey Miscel-laneous Series 30.

Royse, C.F., Callender, E., 1967. A preliminary report on some ice-walled-lake deposits (Pleistocene), Mountrail County, NorthDakota. North Dakota Geological Survey Miscellaneous Report30, 53–61.

Stalker, A.M., 1960. Ice-pressed drift forms and associated deposits inAlberta. Geological Survey of Canada Bulletin 57, 38 pp.

Syverson, K.M., Baker, R.W., Kostka, S., Johnson, M.D., 2005. Pre-Wisconsinan and Wisconsinan glacial stratigraphy, history, andlandscape evolution, western Wisconsin. In: Robinson, J.E. (Ed.),Field Trip Guidebook for Selected Geology in Minnesota andWisconsin.Minnesota Geological SurveyGuidebook 21, 238–278.

Tuthill, S.J., 1967. Late Pleistocene Mollusca of the Missouri CoteauDistrict, North Dakota. North Dakota Geological Survey Miscel-laneous Series, vol. 30, pp. 73–82.

Tuthill, S.J., Clayton, L., Laird, W.M., 1964. A comparison of a fossilPleistocene molluscan fauna from North Dakota with a recentmolluscan fauna from Minnesota. American Midland Naturalist71, 344–362.

Wright, H.E., 1980. Surge moraines of the Klutlan Glacier, YukonTerritory, Canada: origin, wastage, vegetation, succession, lakedevelopment, and application to the Late Glacial of Minnesota.Quaternary Research 14, 2–17.

Yansa, C.H., 2002. Late-glacial and Early Postglacial Vegetation andClimate Change in the Northeastern Great Plains: Evidence fromPollen and Plant Macrofossil Studies. PhD Thesis, University ofWisconsin, Madison, 337 pp.

Yansa, C.H., 2006. The timing and nature of Late Quaternaryvegetation changes in the northern Great Plains, USA and Canada:a re-assessment of the spruce phase. Quaternary Science Reviews25, 263–281.