Onshore–offshore correlation of central Lake Erie glacial deposits Open Access

The Great Lakes of North America occupy bedrock basins that have been excavated by glacial plucking and abrasion during millions of years and tens of ice ages during the Quaternary Period beginning about 2.6 million years ago (Hough 1958; Prest 1970; Flint 1971; Eschman and Karrow 1985; Muller and Prest 1985; Barnett 1992, and references within). Although their glacial origin is widely understood, and they were ice-covered during the last glaciation by the Laurentide Ice Sheet (LIS) about 25 000–15 000 years ago, few studies, apart from those in Lakes Superior and Michigan (Lineback et al. 1979), for example, have identified and named offshore glacial units in their basins. Direct offshore extrapolation of well-known onshore units exposed in shore bluffs may be possible where erosion-resistant bedrock occurs below lake level and where thick sequences of glacial deposits have been eroded and exposed by waves of the Great Lakes. However, such expansion of areal knowledge has not been demonstrated to date, and is rarely accomplished in nearshore marine investigations.

In this paper, we show that a well-known sequence of glacial deposits north of central Lake Erie can be extrapolated directly offshore to identify glacial units exposed in an extensive unconformity beneath north-central Lake Erie. It is widely understood that these units were deposited by readvances from the east of the Ontario–Erie lobe of the LIS through the basins of Lakes Ontario and Erie.

Regional characteristics of Lake Erie and its 400 km long basin (Fig. 1) are described by Sly (1976), Thomas et al. (1976), Herdendorf (1989), Bolsenga and Herdendorf (1993), and Holcombe et al. (1997, 2003, 2005). The three main basins of Lake Erie with their maximum water depths below a mean low water datum of 173.3 m above sea level (asl) (International Great Lakes Datum 1985), are the eastern Erie basin (63–64 m), the central Erie basin (24–25 m), and the western Erie basin (10–11 m). Hydrographic survey data have been compiled into a detailed map of bathymetry by the United States National Oceanic and Atmospheric Administration (NOAA) (NOAA National Geophysical Data Center 1999) (Fig. 1). From the Lake Erie bathymetry map, and from initial sediment sampling reconnaissance (Lewis 1967; Lewis et al. 2012), ridge-shaped recessional glacial moraines, signifying ice sheet marginal positions, have been identified beneath the lake and its Holocene sediments from west to east as the Pelee–Lorain (P), and Erieau–Cleveland (E) moraines of the Port Bruce phase, the Norfolk (N) morainic ridges of the following early Mackinaw phase, and the Maitland (M) moraine of the subsequent Port Huron phase (Fig. 1).

A recent account of the glacial history of the central Erie basin is provided in Barnett and Karrow (2018). The Ontario–Erie lobe of the LIS readvanced westward and began an oscillatory retreat through the Lake Erie basin about 17.71 cal ka (mean of 2σ range) before present (BP) (18 039–17 372 2σ, ¹⁴C 14 500 ± 100 years) (Calkin and Feenstra 1985) or possibly earlier at 18.06 cal ka BP (19 090–17 022, ¹⁴C 14 900 ± 450 years) (Schooler 1974) in the Port Bruce phase. Here, all radiocarbon dates both original, and those previously published were calibrated or recalibrated using the IntCal20 data set and the Calib 8.1 program (Stuiver and Reimer 1993; Reimer et al. 2020). The Port Bruce phase of glaciation followed the first major retreat of the LIS (Erie phase) after the Late Wisconsinan glacial maximum in east-central North America. Dalton et al. (2020) questioned the Port Bruce readvance, but Sodeman et al. (2021) provided support for it. The readvance is also interpreted as ice stream 49 (the eastern branch of the Huron Erie ice stream) in the LIS by Margold et al.(2015a,b, 2018), directed by the large elongate basins of Lakes Ontario and Erie; ice stream existence and flow direction is inferred from glacial lineations and drumlin fields, yet such flow events have also been related to meltwater enhanced surging and flooding (Kamb 1985, 1987), specifically in this region (Sharpe and Russell 2023). Port Bruce phase ice advanced to a position in the Ohio–Indiana border region before beginning an oscillatory recession eastward that lasted up to almost 2000 years until about 16.45 cal ka BP (17 912–14 978, 13 600 ± 100 ¹⁴C years) (Calkin and Feenstra 1985, fig. 2) (Fig. 2). The Late Wisconsinan epoch was termed the Michigan subepisode by Karrow et al. (2000). The sequence and age data of the ancestral Erie basin lakes, particularly in the southwestern (Ohio) sector of the Erie basin, have been reviewed and augmented by Fisher et al. (2015) using geophysical survey methods and optically stimulated luminescence (OSL) dating. Retreat of the ice sheet margin after each readvance left recessional moraines in the basin which were commonly buried or succeeded by lacustrine sediments deposited in glacial lakes Maumee at levels of 244–232 m asl, and Arkona at a level of 216 m asl (nonuplifted elevations), and successor lakes impounded in the basin (Calkin and Feenstra 1985; Barnett 1979, 1992). Lake Maumee may have first overflowed into the Wabash River in Indiana and later via the Imlay channel in southeastern Michigan and along the ice margin to Grand Valley to the Lake Michigan basin as ice receded. Final retreat of the Port Bruce ice was relatively rapid, as shown by the dates of proglacial lake ages. In Ohio the approximate OSL age of Lake Maumee is 16.7 ± 0.8 ka (Fisher et al. 2015), and the mean radiocarbon age of Lake Maumee commencement is given as 17.75 cal ka BP (18 174–17 327, ¹⁴C 14 500 ± 150) (Calkin and Feenstra 1985, p. 150). The Imlay channel outlet in southeastern Michigan was abandoned by 16.70 cal ka BP (17 301–16 091, ¹⁴C 13 770 ± 210 years) (Eschman and Karrow 1985), a date later confirmed by Luczak et al. (2022). Investigations by Luczak et al. (2022) suggested that all phases of Lake Maumee could have used the Imlay Channel, but Barnett (1992, p. 1045) found that an ice readvance during Maumee III to the St. Thomas moraine in Erie basin and possibly to the Mitchell moraine in Huron basin raised Lake Maumee to the Wabash outlet again. Imlay channel final abandonment was followed by Lake Arkona and successor lakes as the ice-free area expanded northward in the Lake Huron basin. A lagoon at Cleveland, Ohio (210 m asl) has been attributed to the lowest Arkona level and dated about 16.45 cal ka BP (17 912–14 978, ¹⁴C 13 600 ± 500 years) by Hough (1958) (Calkin and Feenstra 1985, fig.2, p. 156).

The Port Bruce glacial readvances and retreats deposited the Port Stanley Drift north of the central basin of Lake Erie, comprising ice-deposited diamictons termed Port Stanley Till, and proglacial lake sediments (Barnett 1978, 1982, 1987, 1992; Barnett et al. 1991). Lake Ypsilanti low phase, identified on the basis of a pre-Lake Whittlesey infilled river channel at Ypsilanti, Michigan, was graded to a sub-Lake Erie level (Kunkle 1963). The low phase is commonly linked with the Pennsylvania channel in Lake Erie bathymetry (Fig. 1) (Calkin and Feenstra 1985; Lewis et al. 2012). However, Lake Ypsilanti’s existence in Ohio has been challenged by Fisher et al. (2015). Port Bruce ice deposition and ice retreat ended with ice readvances of the LIS Mackinaw Phase into eastern Lake Erie from the northeast that deposited Wentworth Till in the Paris, Galt, and Moffat moraines in Ontario (Table 1) (Taylor 1913; Karrow 1974; Barnett 1992). In assessing depositional elements of the long north-to-south Paris and Galt moraines, Russell et al. (2009, 2013, 2015) noted, in addition to relief differences and their interbedded Wentworth Till and gravel composition, a distinct southward decrease in diamicton and an increase in sand content at elevations below about 300 m asl, suggesting possible deposition in a glacial lake (Lake Arkona?) in the Erie basin. Ice readvances of the Mackinaw phase also formed the Norfolk morainic ridges across Lake Erie (Lewis 1967; Holcombe et al. 2003, 2005), probably the Girard member of the Ashtabula morainic system, and possibly other members of the Lake Escarpment moraine on the south side of Lake Erie (Leverett 1902; Muller 1963, 1977; Cadwell 1988). These moraines probably formed in the early part of the subsequent Mackinaw phase before 15.96 cal ka BP (17 372–14 547, ¹⁴C 13 360 ± 440 years) when runoff was flowing in front of the Galt moraine (a Mackinaw ice front) into the Jacksonburg delta in the Erie basin (Barnett 1985) (Figs. 1 and 2).

The younger Mackinaw Wentworth Till (Karrow 1974) overlies the Port Bruce units of Port Stanley Drift in the eastern sector of the central Lake Erie basin (Barnett 1978, 1987, 1992; Barnett et al. 1991). Facies descriptions and models for deposition of these units, based on exposures in the Lake Erie shore bluffs, are presented by Barnett and Karrow (2018).

The terrace on which the offshore glacial units crop out is a part of the “b” seismic reflection horizon mapped in the central Erie basin by Wall (1968) using a horizontal-plate sonar thumper, type St-8 (also called a “boomer” source, made by Edgerton, Germerhausen, and Grier, Inc. 1960) in a sub-bottom seismic reflection system. The strong reflections from the “b” horizon were interpreted to represent an “erosional unconformity cutting across glaciolacustrine clays and tills” (Wall 1968). However, boundaries of individual glacial units, as determined here by higher frequency echo sounders, were not resolved by Wall (1968).

The bedrock surface is deeply buried (34–89 m) below Lake Erie along the north shore of the central basin (Caley and Sanford 1952; Sanford 1953, 1954; Gao 2011; Carter et al. 2019; Morgan et al. 2020). As a result, wave action on the present lake rapidly eroded the unlithified shore exposures, leaving beds of glacial sediment exposed in bluffs up to several tens of metres high.

Study of the sediment stratigraphy and sedimentology in bluff exposures and mapping of the surficial sediments in the onshore region north of Lake Erie has established the framework of glacial lakes and glacial deposition along the north shore of central Lake Erie (Dreimanis and Reavely 1953; Dreimanis,1964, 1970, 1992; Barnett 1978, 1982, 1985, 1987, 1992, 1993, 1998; Dreimanis and Barnett 1985; Dreimanis et al. 1987; Barnett and Karrow 2018). The general distribution of these sediments, deposited approximately 18.20–15.96 cal ka BP (18 830–17 572 to 17 372–14 547, ¹⁴C 15 000 ± 250 to 13 360 ± 440 years), was studied in detail along the bluffs by one of the authors (P.J. Barnett), and is summarized in Fig. 3 (Barnett 1985, 1987, 1993, 1998). Offshore glacial units were identified in this study by direct correlation to these exposed shore bluff deposits as they dipped below Lake Erie level. Older glacial sediments, such as the Catfish Creek Till deposited during the last glacial maximum (Nissouri phase, Fig. 2), that outcrop farther west in the Lake Erie coastal zone near and west of Port Talbot (Fig. 1) (Dreimanis and Goldthwait 1973; Dreimanis et al. 1987) were not encountered, but are probably present west of the sites studied.

The Port Stanley Drift exposed along the north shore of Lake Erie consists of three massive diamicton layers each of which have been interpreted as subglacial till (Fig. 3), and generally consist of 6% sand and 39% clay with minor amounts of coarser particles to boulder-sized sediments (Table 1) (Barnett 1993; Barnett and Karrow 2018). These massive layers interfinger with glaciolacustrine sediments and are arranged in an off-lapping sequence. Each massive layer is commonly associated with units of thin-bedded diamictons that have been interpreted as glacially derived debris flows or flow tills. The massive diamicton layers extend northwestward in the onshore landscape to end moraines where they terminate, and are thought to merge into one layer towards the east (below lake level). The glaciolacustrine sediments between diamicton layers consist of thick sand and silt rhythmites and thin silt and clay rhythmites (Fig. 3) (Barnett 1985, 1987, 1993, 1998). Offshore, these stratified sediments can be recognized in acoustic and seismic records by their internal reflections. South of Lake Erie, the Hiram Till in northwestern Pennsylvania and northeastern Ohio (White et al. 1969; Dreimanis and Goldthwait 1973; White and Totten 1979) is considered equivalent to the Port Stanley Till.

The Wentworth Till (Karrow 1974) is exposed along the lake bluffs east of Jacksonburg, Ontario (Figs. 1 and 3) where the Paris and Galt moraines are intersected by the Lake Erie shoreline (Barnett and Karrow 2018). Diamicton facies associations and texture of the Wentworth Till are similar to those of the Port Stanley Till onshore (Barnett 1993, 1998). South of Lake Erie, sediment comprising the Lake Escarpment moraines (Muller 1977; Cadwell 1988), and the Girard Moraine, youngest member of the Ashtabula Moraine in northeastern Ohio and northwestern Pennsylvania and closest to Lake Erie (Leverett 1902; Mickelson et al. 1983) may correlate with the Wentworth Till in moraines in Ontario (Barnett 1993, 1998).

Stratified Mackinaw sediments associated with the Wentworth Till interfinger with organic-bearing upper sediments of a delta near Jacksonburg, Ontario (Figs. 1 and 3). The Jacksonburg delta is marked DDD in Fig. 3. These sediments include delicate leaves of Dryas integrifolia and Salix herbacea dated at 15.96 cal ka BP (17 372–14 547, ¹⁴C 13 360 ± 440 years) (Barnett 1985; Warner and Barnett 2008). The relatively low elevation of the delta (lower than Lake Arkona level) indicates that its formation followed lakes Maumee and Arkona during the Port Bruce phase and its upper beds were deposited during the Mackinaw phase. Rivers draining the onshore area, possibly also carrying overflow waters from lakes in the developing Oak Ridges moraine north of western Lake Ontario (Fig. 1 index map) (Barnett et al. 1998), and flowing in front of the Galt moraine transported sediments and plant fragments to the Jacksonburg delta (Barnett 1985). Later, after ice had receded, rivers broke through the Galt moraine, and formed another delta in the Erie basin due south of Simcoe, Ontario. This delta was later incised 8 m and graded to a low lake in the Erie basin (Barnett 1985). This lake may have declined further to the level of Lake Ypsilanti and lower (Kunkle 1963) before its shores and inflowing river channels were submerged by the subsequent high-level Lake Whittlesey.

The reprofiled Ontario–Erie lobe of the LIS next was active (surged) in the western end of the Lake Ontario basin effecting deposition of Halton Till (Sharpe and Russell 2016), as far west as eastern Lake Erie (Barnett 1979; Feenstra 1981). This Port Huron readvance stopped eastward overflow of the previous Lake Ypsilanti and lower lakes. The readvance dammed this overflow and raised Lake Whittlesey (to ∼226 m asl) to fill the Erie basin about 15.55 cal ka BP (15 846–15 253, ¹⁴C 13 000 ± 100, estimated from four dates) (Barnett 1979). Lake Whittlesey overflowed westward at Ubly, Michigan, across the “thumb” of Michigan, into Lake Saginaw (Huron basin) and the Grand Valley into the Lake Michigan basin (Barnett 1979; Calkin and Feenstra 1985; Eschman and Karrow, 1985).

Offshore sediments in Lake Erie (Table S1, Thomas et al. 1976, and Lewis and Todd 2021) were sampled, cored, and profiled extensively in the 1960s using the research vessel CNAV Porte Dauphine programmed by the University of Toronto (Fig. 4) (Lewis 1967). A clam-shell type grab sampler of 2.5 L capacity modelled after one designed by Franklin and Anderson (1961) sampled surface sediments whose distribution is shown in Thomas et al. (1976, fig. 2) and Lewis and Todd (2021, fig. 6). A weighted 3 m tube of AX casing (internal diameter of 4.8 cm) with a check valve on top, sharpened cutter at bottom, and steel-plate fins affixed to its upper part, served as a gravity corer, and recovered sediment sequences up to 1–2 m long depending on sediment resistance to penetration (Table S1). The recovered cores were extruded and commonly described immediately. A Kelvin Hughes MS 26B sounder (14.25 kHz) operated continuously between sampling stations to record the acoustic character and depth of surface and subsurface reflections (Fig. 4, Table 2) for correlation to sediment units and their boundaries observed in the cores (Table S1 and Fig. 5).

A standard piston corer as used for oceanographic sedimentary studies (Lewis 1967; Rechnitzer and Baker 2022) recovered cores up to 6–7 m long in areas covered by soft lacustrine silt and clay (Table S1). The core lengths were 35%–40% shorter than the length of sediment adhering to the exterior of the coring barrels, thought to be due to sediment compression and/or sediment bypassing (Lewis and Todd 2021). The core lengths were reported as measured. The lake bottom was examined visually using self-contained underwater breathing apparatus at some corer sites in the northern sector of the central Erie basin and farther west (Lewis 1967). Long sediment sequences in boreholes, to bedrock in most places, were recovered in the western part of the central Erie basin through collaborative studies with The Consumers Gas Company of Toronto (Lewis et al. 1973; Lewis and Todd 2021).

A section between central Lake Erie and Long Point was interpreted from a 78 km long 1967 Geological Survey of Canada (GSC) seismic reflection line (Todd et al. 2023) to show the depth distribution of many of the offshore sediment units (Fig. 6). A multifrequency boomer sound source (∼2–20 kHz) and a receiving eel of hydrophones were employed. Almost 11 m of seismic record was compressed horizontally to page-width to enhance the recognition and correlation of sediment reflections using seismic processing software JP2Viewer written by R.C. Courtney (GSC Atlantic) (Courtney 2012).

An acoustic transect was obtained in 1987 through most of the central basin (Figs. 5 and 7) from CCGS Limnos using its Kelvin Hughes 26B sounder and to a limited extent a portable Raytheon sounder (7 kHz) (Figs. 5 and 8) (Cameron 1991). Profiles from these instruments confirmed the distribution of subsurface reflections and their relationship to mapped glacial sediment boundaries.

Selected samples were analysed for grain size using sieves for the sand portion and hydrometer with Stokes Law of Settling Velocities for the finer fraction (Lewis 1967; Hossain et al. 2021; ASTM, 2021). Selected cores were analysed for pollen content using standard methods for correlation to onshore sites and their chronology (Fig. 9). The pollen was extracted by the standard acetolysis/hydrofluoric acid procedure (Faegri and Iversen 1964) and mounted in silicone oil, or in Hoyer’s solution (Anderson 1954). Counts of at least 200 arboreal grains were made for each sample, and were used as the pollen sum; all other counts of pollen and spores were calculated as percentages of this sum. The pollen diagrams were divided into recognizable pollen zones characterized by diagnostic pollen assemblages. Summary pollen diagrams were prepared using key taxa for correlation to onshore ¹⁴C-dated profiles. All radiocarbon dates, including original dates, those transferred from onshore sediment sequences to pollen profiles, and those from published sources were calibrated or recalibrated using the calibration software Calib 8.1 with the IntCal20 data set (Stuiver and Reimer 1993; Reimer et al. 2020).

Diving observations revealed that parts of the lake bed were covered by isolated sand-rich symmetrical (wave-formed) ripples separated by stiff clay swept nearly clean of sand (Lewis 1967). As a consequence, much of the northern central Lake Erie region, covered intermittently with a rippled gravelly sand lag, is interpreted as a wave-cut terrace eroded by storm waves and currents. Repeated lake bed observations in the northwestern part of the central basin confirmed that the terrace of stiff glacial sediments was undergoing erosion today by actively moving ripples of sand and gravel, driven by waves and currents, probably forced by storm winds. Core samples showed that the terrace and lag extended to >30 km offshore to depths of 28–36 m below lake level beneath soft Holocene mud between 80°30′W and 82°W longitude. The offshore limit of the wave-cut terrace extends approximately to the 30–35 m contours below lake level on the surface of glacial deposits (Fig. 5). At core C-2 (Figs. 5, 7, and 9), pollen analysis shows that the sediments were disturbed (i.e., some sediments were removed or displaced) at water depths of ∼40 m, possibly in a low lake-level wave base or by wind-driven pressure keels of lake ice (e.g., Grass 1984). The presence of the unconformity beneath offshore Holocene mud indicates that the terrace was being eroded during earlier and lower stages of Lake Erie.

Offshore glacial sedimentary units crop out on the eroded surface of the wave-cut terrace situated below the rippled lag concentrate and below Holocene mud farther offshore (Table 3). The relatively high-frequency and short wavelength of the echo sounder’s acoustic pressure waves enabled resolution of the contacts between eroded glacial sedimentary units near the outcrop surface (Figs. 4, 5, 7, and 8; Table 2). Lower frequency and more powerful seismic reflection systems such as that used to produce the section from central Lake Erie to Long Point, mentioned above (Fig. 6), tended to achieve deeper penetration with less vertical resolution. Short reflections of the glacial sedimentary bed boundaries are evident just below the terrace surface in the 14.25 kHz Kelvin Hughes 26B echo sounder profiles (Figs. 4, 5, and 7; Table 2), that are interpreted to define the boundaries between stratigraphic layers. The relative sequences of sediment unit deposition are also evident in the Raytheon sounder records (Fig. 8) and seismic profile (Fig. 6). Glacial sedimentary unit boundaries on the terrace are subtly marked also, in places, by small steps or relief changes where sediments of different erosion resistance meet (Figs. 4 and 5). The principal acoustic facies are

• 1) structureless layers lacking internal reflections and

• 2) structured layers with parallel internal reflections, discontinuous or weak in places.

The location of acoustic profiles and selected echograms for illustration is shown by the inset map in Fig. 4, and by the grey track in Fig. 5 labelled C to H. The offshore units are identified by large red letters M to S (Fig. 5). The boundaries of the offshore units were traced from profile to profile on the terrace surface (Fig. 5). Layers of unstructured and structured acoustic facies were confirmed by core sampling of the terrace surface, and the acoustic facies were found to correlate with massive and laminated sedimentary facies, respectively. Matrix grain sizes of the offshore till units are listed in Table 1. Although the grain size compositions of the two tills are similar onshore, the offshore samples of Wentworth Till are finer in grain size (Table 1), and their colour in some samples is slightly redder in tone than the offshore Port Stanley Till samples (Table S1). In this region, the red colour of a glacial deposit may indicate a possible contribution of particles eroded from the red Ordovician Queenston shale exposed to the north along the southern edge of the Lake Ontario basin (A. Dreimanis (personal communication, 1984)).

Sediment units M to R comprise an alternating sequence of acoustically structureless and structured layers of decreasing age. Core samples from the structureless layers (units M, O, and Q) were commonly massive with a sparse coarse fraction dispersed in a clay matrix (Tables 1, 3, and S1) and are interpreted to be diamicton (in places interbedded with debris flows and scattered zones of ice-rafted debris, and glaciolacustrine sediment). Samples from the structured layers (units N, P, and R) were commonly laminated clay and are interpreted to be glaciolacustrine sediment (Tables 3 and S1). Unit S, in the western part of the central Erie basin, is a well-structured acoustic unit with many parallel reflections which core and borehole samples show to be laminated reddish grey clay. Unit T, recovered in cores and shown on acoustic profiles (Figs. 4, 7, and 8) consists of Holocene surficial soft lacustrine mud (Tables 3 and S1).

A transect through the central Erie basin was conducted by CCGS Limnos in 1987 using the onboard Kelvin Hughes 26B echo sounder (14 kHz) mostly at full speed (∼10 knots, 18.5 km/h), and a side-mounted Raytheon sounder (7 kHz) at a reduced speed. The results confirmed the distribution of glacial units as profiled and sampled in the basin during the 1960s and mapped during the 1980s. A grey line in Fig. 5, labelled C to H, shows the track of the confirming acoustic transect, marked and labelled every 10 km. The echo sounding transect extends through the central Erie basin from C south of Erieau, Ontario, through some borehole locations, and across the sediment unit boundaries to the edge of the eastern Erie basin south of the base of Long Point at H. The acoustic transect is described in the following paragraphs.

From 0 km (Fig. 7a), unit M (lower and middle layers of Port Stanley Till as in Fig. 3) rises eastward through a thin cover of unit T (Holocene mud) to an exposed elevated position at about 3 km on the Erieau–Cleveland moraine (and lake bed) before dipping into the central Erie basin beneath unit T. Unit M continues descending beneath unit T, then beneath unit N at about 20 km. The contact between M and N is at the base of unit N. Unit N has numerous chaotic internal reflections that darken the echogram record. Unit S appears between units T and N from 20 to 25 km. This sequence of deposits passes through borehole 13194 and beyond. The surfaces of units T and N occur at 24.7 and 36.9 m depth, respectively, below mean lake level (∼174 m asl) at borehole 13194. Unit N rises slightly at the C-2 site shown by its probable reflection at the base of the record (Fig. 7). Only the surficial unit T is visible at 44 km and beyond Turning Point 1. Near borehole 13193 at 66 km where the lake bed occurs 23.8 m below mean lake level, the upper surface of unit N reappears about 42 m below mean lake level in borehole 13193.

As the ship travelled northeastward from borehole 13193 to TP2, the sequence of units T over S over N was recorded from 66 km to E (not shown). As the ship travelled eastward after TP2 (E), the lower part of the sediment sequence was largely nonreflective, and only appears in the Raytheon record (Fig. 8). At about 99 km, unit O rises to appear below units S and T (Fig. 7a). Unit S appears beneath unit T and above unit O at TP3 (F) (Fig. 5) where the ship turned to travel north-northwestward. At about 105 km, after TP3, unit S pinches out as unit O rises under a northward-thinning unit T, to appear at the lake bed at about 112 km (Figs. 7a and 8). Unit O pinches out as Unit N rises to the lake bed at about 116 km and continues to 125 km where older unit M rises to the lake bed (Fig. 7b). Relief on unit M (probably the middle layer of Port Stanley Till at this location) between 127 and 130 km and again near 135 km is possibly marking a morainic ridge on the trend of the onshore Lakeview moraine (Fig. 5).

Unit M continues through TP4 (G) (ship turned to travel eastward) to 140 km where it is overlain for a short distance by unit N again (Fig. 7b). Upper Port Stanley Till in unit O with some stratified debris flows appears on the lake bed from 150 to 160 km (Fig. 7b). The boundary between unit M and overlying unit O is not visible after 140 km, and may indicate that the Port Stanley Till onshore layers have merged offshore. Then the glaciolacustrine unit P appears. Unit P fades (interfingers?) into unit Q at about 168 km. East of unit P, the extensive Wentworth Till and its debris flows in unit Q, underlie the Norfolk morainic ridges. The several ridges of the Norfolk moraine probably represent several ice readvances or stillstands. Unit S seems to be absent near the northern sector of Lake Erie, possibly because it was eroded in the shallower water. The reflective ridge surfaces are probably draped with beach sand as interpreted by Holcombe et al. (2003, 2005). Glaciolacustrine sediments with internal parallel reflections lap onto the eastern margin of unit Q and the Norfolk morainic ridges starting at 191 km as unit R (Fig. 7b).

A part of the confirming transect from about TP2 eastward to TP3 and north-northeastward to TP4 was also recorded with a portable Raytheon echo sounder operating at 7 kHz (Figs. 5 and 8). From about 89 to 93 km, irregular reflections possibly from unit O apparently rise to the unconformity beneath Holocene unit T. The Raytheon sounder recorded similar sediment sequences as the Kelvin Hughes instrument but with one-half the frequency and double the wavelength of the acoustic pressure pulse of the latter sounder. As expected, the Raytheon signal penetrated farther into the lake bed. This effect is most evident in the Raytheon record from 104 to 114 km (Fig. 8) where reflections from unit N beneath unit O are more clearly expressed when compared with the Kelvin Hughes record (Fig. 7a).

Two boreholes and one piston core from central Lake Erie (boreholes 13193, and 13194, and piston core C-2) were analysed for pollen content (Fig. 9). The pollen diagrams show only those taxa which best document the overall pollen zonation at each site. The pollen diagrams were correlated with ¹⁴C-dated pollen profiles at nearby terrestrial sites (Yu 2000) to establish the time horizons shown alongside the sediment columns on each diagram.

The lowest pollen zone, the sagebrush (Artemisia) zone in borehole 13193, is characterized by high percentages of herb pollen, mainly Artemisia, grass, and sedge. Herb pollen declines and is replaced by an increase in spruce at about 13.84–14.69 cal ka BP (14079-13607–15128-14243 ¹⁴C 12 000 ± 100–12 500 ± 100). The high percentages of spruce pollen in unit S (up to 80% in borehole 13194) represent the spruce (Picea) pollen zone at about 11.63 cal ka BP (¹⁴C 10 100 ± 100) (Yu 2000). The spruce–pine transition corresponds with the sediment change from laminated reddish grey silty clay (unit S) to overlying massive grey silty clay (unit T) in boreholes 13193 and 13194. Spruce, in turn, gives way to a pine (Pinus) zone at the base of unit T marked by maximum values in pine pollen (boreholes 13193 and 13194), or to the oak–maple (Quercus–Acer) pollen zone in unit T of core C-2, but here the top of the spruce zone and the entire pine zone are missing, probably indicating erosion and/or removal or displacement of sediments containing these pollen zones during early formation of the wave-cut terrace in lower stages of Lake Erie.

The seismic section between Long Point and the central Erie basin reveals a ∼40 m rise in bedrock elevation from the eastern to central Erie basins from ∼120 to ∼80 m below the Lake Erie level (Fig. 6). Thick accumulations of Port Bruce glacial units O, and lower P, Mackinaw phase units, upper P, Q, and R, and unidentified early glacial deposits were deposited as the glacial advances from the east encountered the bedrock rise. Unit O pinched out in a till tongue (King et al. 1991) at ∼40 m depth on the western side of the rise in the central basin shown at about 10 km in the seismic record (Fig. 6). Two earlier till tongues below the unit O till tongue, delineated by the red and blue horizons in Fig. 6, may have originated as part of the initial Port Bruce and earlier readvances. The stratified unit P thickens eastward under Q and may be a deposit of a glacial lake that formed in the eastern Erie basin during the recession between the Port Bruce and Mackinaw glacial readvances. The lacustrine units S and T overlie unit O in the central basin as shown in Fig. 6. Laminated unit S was recognized in the western section of the seismic section (0–15 km, Fig. 6), in the central Erie basin.

The seismic profile (Fig. 6) shows a massive deposit of Wentworth Till beneath the Norfolk morainic ridges (small peaks on the lake bed reflection at about 55–60 km in Fig. 6). The ridges are covered with sand which Holcombe et al. (2003, 2005) attributed to beach development during the mid-Holocene period of Nipissing Great Lake inflow into the Erie basin (Thompson et al. 2011). Some of the sand could have also originated with an outburst of stored subglacial sediment-laden meltwater from the eastern basin during retreat of the Mackinaw ice, but a thick reflective sand apron is not evident. The seismic profile in Fig. 6 shows seismic wave energy penetrating and reflecting from horizons (unit P) in the subsurface. This is only consistent with a thin cover of sand, as sand is a strong reflector. Some reduction in seismic wave energy is indicated by the gap in the bedrock surface reflection below the Wentworth Till deposit. Also, the multiple reflection of the lake bed is strong in the area. These observations suggest there is some sand over the till but not enough to completely blank out the subsurface.

Correlation of the offshore glacial units was made by projecting their boundary trends into known lithostratigraphic unit boundaries exposed at lake level in the northern shore bluffs of the central Erie basin (Figs. 3 and 5). The summary onshore stratigraphy shown in Fig. 3 is based on numerous detailed logs of the sediment facies exposed in the shore bluffs between the base of Long Point and Port Bruce, Ontario (Barnett 1985, 1987, 1993, 1998) (Figs. 1 and 3). In short, the offshore units could be related to known time-stratigraphic units, ice margins, and lake phases of the Erie basin by interpreting their connections with the better-known onshore stratigraphy as mapped from the shore bluff exposures (Barnett 1985, 1993, 1998; Barnett and Karrow 2018). By this process, the offshore unit M is equivalent to the lower and middle layers of Port Stanley Till with their associated interbedded glaciolacustrine sediment (Fig. 3 and Table 3).

Glaciolacustrine clay and silt of the next younger unit N were deposited before deposition of the upper layer of Port Stanley Till, unit O, which is confined to the eastern half of the mapped area (Fig. 5 and Table 3). The subsequent glaciolacustrine clay, unit P, is transitional into the overlying Wentworth Till, unit Q, and its subaquatic flow tills (Table 3) (Barnett 1985). This till, unit Q, forms the body of Wentworth Till that underlies ridges of the Norfolk moraine and is succeeded by glaciolacustrine clay, unit R, near the base of Long Point (Figs. 5 and 6; Table 3). The laminated clay of unit S in the western sector of the central basin overlies unit N conformably, and could represent offshore lacustrine deposition during overflow from the Lake Huron basin (from proglacial Lake Algonquin) to just prior to deposition of unit T, the offshore postglacial Lake Erie lacustrine mud (Table 3). In addition, erosion of Lake Erie shore bluffs by storm waves during the ice-free season may have introduced coarse sediment to effect lamination in unit S sediments.

The mapped portion of unit M (lower layer of Port Stanley Till) broadens to the south of Erieau into the offshore Erieau–Cleveland moraine (Figs. 1 and 5). Relief on this moraine appears not to have been developed in the deeper parts of the central Lake Erie basin, suggesting formation by a relatively thin glacier that was partially afloat in Lake Maumee and was grounded only on the margins of the basin (Fig. 1). Calculation of the probable thickness of the submerged portion of the floating ice shelf requires comparison of the pre-uplift elevation of the grounding line with the original elevation of the lake surface. An estimate of the total ice shelf thickness may be obtained by dividing the submerged ice thickness by the ice density, 0.9 g/cm³.

The apparent grounding line south of Erieau, Ontario, is 30 m below Lake Erie (174 m asl) (Fig. 5) at a present-day elevation of 174 − 30 m = 144 m asl. Maumee beaches rise eastward about 9 m to the Ohio–Pennsylvania border near Ashtabula County, Ohio, on the south side of Lake Erie (Totten 1985, p. 179). At Ashtabula, Ohio, approximately on the Maumee uplift isobase through the Erieau region (assumed to be similar in orientation to the Whittlesey isobase through Erieau (Barnett 1979)), the upper Maumee beach rises ∼8.5 m due to glacial differential uplift (Totten 1985, fig. 6). Assuming uplift of the Erieau area to have been similar, the initial elevation of the grounding line (before uplift) at the time of Lake Maumee would have been 8.5 m less, or at an elevation of 144 − 8.5 m = 135.5 m asl. With the glacier front near the Erieau–Cleveland position (Barnett 1992, p. 1046), Lake Maumee (Maumee III) was overflowing to the Wabash River near Fort Wayne, Indiana, at its highest elevation of 244 m asl (Eschman and Karrow 1985, p. 80). These inferences imply that the thickness of the submerged portion of the ice shelf was 244 − 135.5 m = 108.5 m, and that the overall thickness of the ice shelf was 108.5/0.9 = 120.6 m.

Later, the upper layer of Port Stanley Till, unit O in Fig. 6, terminated in a till tongue at about 40 m depth in Lake Erie, near the ice margin in Lake Maumee IV (Barnett 1992, fig. 21.39, p. 1047). The offshore position of the ice front in Lake Maumee IV correlates northwestward to a Maumee IV beach bar at an elevation of 262 m asl about 1.5 km northeast of Delmer, Ontario, on the Norwich moraine (Fig. 5) (Barnett 1985, p. 189) using the trend of the Whittlesey isobases (Barnett 1979, fig. 4). This beach elevation suggests the offshore ice-front area of Maumee IV was uplifted about 262 − 232 m = 30 m. This calculation assumes Lake Maumee IV overflowed near Imlay, Michigan, with a lake level of 232 m asl (Eschman and Karrow 1985; Barnett 1992, p. 1047). Thus, the base of the glacier near the till tongue was lower by 30 m and about 40 + 30 m = 70 m lower than Lake Erie level (174 m asl) at the time of Maumee IV. The basal elevation of the ice shelf would have been about 174 − 70 m = 104 m asl elevation at the time of Maumee IV. The thickness of the submerged portion of the ice below lake level was ∼232 − 104 m = 128 m, and the total ice thickness would have been 128/0.9 m = 142.2 m.

That the LIS Ontario–Erie lobe was grounded onshore in the central Erie basin is well understood. However, the above inferences and computations suggest that the ice sheet retreat in the deeper parts of the central Erie basin during the Port Bruce phase was by a floating ice shelf with thicknesses in the order of 120–150 m.

Chapman and Putnam (1984, p. 94) observed that steep-sided gullies of many of the creeks draining into the north shore of central Lake Erie are constantly growing headwards. They inferred that the creeks formerly extended downstream beyond the present shoreline. In particular, former tributary branches of Kettle Creek now enter the lake separately near Port Stanley, and suggest that the lake shoreline had receded at least 10 miles (16 km) from its original position (probably since lake level rose during mid-Holocene Nipissing Great Lake inflow from the Huron basin). This amount of recession, and more, is abundantly confirmed by the evidence in this paper of a wave-cut terrace extending >30 km offshore.

An abandoned channel beneath Clear Creek between Jacksonburg and the base of Long Point was cut into the erosional platform at least 5 m below the level of Lake Erie and was subsequently infilled (Barnett et al. 1985). A study of the palynology of the sediment fill suggests that the infilling began approximately 10.80–10.08 cal ka BP (11167–10515 to 10404–9761, ¹⁴C 9500–9000 years B.P.) with the onset of low-level Early Lake Erie (Barnett et al. 1985). Infilling was complete sometime after the rise of Lake Erie level with Nipissing Great Lake overflow from the Lake Huron basin 4.39 cal ka BP (4781–3988, ¹⁴C 3900 ± 100 years B.P.) based on a radiocarbon date on wood near the top of the channel fill (Barnett et al. 1985).

The lacustrine mud in unit T that overlies the deepest and farthest offshore parts of the terrace (e.g., at core site C-2, Figs. 5 and 7) indicates that terrace formation began when lake level was several metres lower than at present. A consideration of lake level history suggests that lake bed erosion and sediment removal or displacement occurred by wave base, or by grounded lake ice pressure keel movement (e.g., Grass 1984), possibly during the closed basin lowstand when water levels were up to 16 m below the present Lake Erie level. At that time, the potential overflow control of the lake waters was the Lyell–Johnson bedrock ridge across the Niagara River about 11.93–3.78 cal ka BP (12459–11401 to 4077–3490, ¹⁴C 10200 ± 100 to 3500 ± 100) (Pengelly et al. 1996; Lewis et al. 2012, fig. 5). Lake bed erosion and formation of the outer terrace would certainly have been underway during rising Lake Erie as overflow from the Nipissing transgression and Nipissing Great Lake in Huron basin was being increasingly diverted into Lake Erie after ca. 8.04 cal ka BP (8285–7792, ca. ¹⁴C 7200 ± 100), and continuing to the present (Lewis and Anderson 2012, fig. 5; 2017, fig. 4).

A groundwater study commissioned by the Ontario Geological Survey drilled 21 boreholes to investigate the subsurface beneath the Norfolk Sand Plain in the area of the Paris and Galt moraines in the Delhi–Simcoe region, Ontario (Fig. 5) (Marich 2014). Five samples containing leaf fragments of Dryas integrifolia were obtained and dated (Table 4). One older sample (18.03 cal ka BP) that was originally attributed to younger lakes Whittlesey or Warren was deemed to have been recycled on the landscape and not considered further (Table 4). Four other samples, dating 17.25–17.16 cal ka BP by the accelerated mass spectrometry (AMS) method (Table 4), were enclosed in deposits attributed to Lake Ypsilanti. Although the sediments were originally interpreted to be deposits in channels leading to low-level Lake Ypsilanti, co-deposited sediments of the Dryas leaves exhibited ripples and/or rhythmites, and deposition in a lake environment was preferred (Table 4). Also, their elevations exceeded the expected elevation limit (165.5 m asl) of the nonuplifted Lake Ypsilanti surface (Kunkle 1963), even when the sample elevations in the Delhi–Simcoe area were adjusted (−26 m) for post-Ypsilanti (post-Arkona) uplift. The uplift was calculated as Arkona shore elevation near Delhi minus the elevation of the shore farther west where it becomes horizontal, i.e., 235 − 209 m = 26 m (Barnett 1982, 1985; Eschman and Karrow 1985; Lewis et al. 2022). The four dated samples (17.25–17.16 cal ka BP, Table 4) are younger than Lake Maumee (Calkin and Feenstra 1985 p. 150, and younger than the maximum of the uncertainty range for Lake Maumee found by Fisher et al. 2015, table 1). The four samples may reflect the age of the younger Lake Arkona, and could be from Arkona deposits that were overridden by Mackinaw ice that readvanced to the Paris moraine (Table 4). The youngest date, 15.96 cal ka, obtained from Dryas integrifolia and Salix leaves in the Jacksonburg delta (Table 4) (Barnett 1985), may have been deposited by rivers flowing in front of the Galt moraine fed by overflow water from lakes during formation of the Oak Ridges moraine near western Lake Ontario (Fig. 1 index map) (Barnett et al. 1998).

In addition to direct correlation of offshore units to adjacent shore bluff sediment units, parts of the offshore sequences may also indicate other lacustrine events. For example, the missing pollen zones (upper spruce and entire pine zones) in core C-2 (Fig. 9) may be related to wave action or to scour by lake ice pressure ridges (similar to modern ice scour described by Grass (1984)) in the low, closed basin phase of Early Lake Erie after Lake Algonquin overflow to the Erie basin was diverted northward, after 12.36 cal ka BP (12712–12007, ¹⁴C 10500 ± 100) (Lewis et al. 2012, fig. 5). The spruce–pine pollen transition dated about 12.36 cal ka BP (12712–12007, ¹⁴C 10500 ± 100) in Karrow et al. (1975), or later about 11.63 cal ka BP (11993–11267, ¹⁴C 10100 ± 100) in Yu (2000) corresponds with the sediment change from laminated reddish-grey silty clay, unit S, to overlying massive grey silty clay, unit T, in boreholes 13193 and 13194. The upper boundary of the spruce zone therefore dates the end of postglacial reddish grey clay (Early Lake Erie) deposition in central Lake Erie, possibly owing to the diversion of Huron-basin Lake Algonquin overflow from Lake Erie to lower outlets southeast of North Bay, ON, draining to Ottawa River (Eschman and Karrow 1985; Lewis et al. 2012).

The Norfolk moraine comprises the Long Point–Erie and Clear Creek ridges which are draped with covers of sand and gravel (Holcombe et al. 2003, 2005). The material underlying the ridges constitutes the bulk of the Wentworth Till, unit Q, as illustrated in the seismic profile of Fig. 6. The glacial origin of the southern end of the Norfolk morainic ridges was indicated during inventory studies of the sand resource on their surfaces by seismic reflection surveys, and by vibracoring that recovered both unstratified and stratified sediment beneath the surficial sand (Williams and Meisburger 1982). The underlying material is interpreted here as glacial diamicton (till), and glaciolacustrine sediment and/or debris flows (flow till), respectively. Wood fragments in firm clay were dated 12.63 cal ka BP (13156–12104, ¹⁴C 10800 ± 190 years) (Williams and Meisburger 1982). This wood may be from tree growth on the moraine during the lowstands of the lake prior to the overflow of Lake Algonquin through the Erie basin (Tinkler et al. 1992; Lewis et al. 2012). Three other younger wood pieces in sand (Williams and Meisburger 1982) may represent tree growth during the Erie lowstand between the Algonquin and Nipissing periods of overflow into the Lake Erie basin (Lewis et al. 2012). All of this tree growth was several thousand years younger than the Wentworth Till.

The Port Bruce sediments are expected to be fully present in the eastern basin as their depositing ice lobe advanced from the up-ice direction (i.e., from the east–northeast). There is a possibility that forcing of the Mackinaw stade was different than for the glacial deposition of the Port Bruce phase. The Mackinaw phase may have included processes that eroded former glacial deposits in the eastern Erie basin. The Mackinaw phase is poorly understood and is regarded as a period of ice sheet reorganization (Barnett 1992 p. 1047). The Mackinaw phase is consistent with the notion that the eastern Lake Erie glacial deposits may contain erosion surfaces that can be considered a southern extension of the regional LIS unconformity proposed by Sharpe et al. (2004). In that event high-velocity subglacial meltwater flows eroded tunnel channels and showed evidence of hydraulic jumps and related high-velocity sediment deposition (by jökulhlaups) north of the western area of Lake Ontario before construction of the Oak Ridges moraine (ORM) (Fig. 1 inset) north (up-ice) of eastern Lake Erie (Russell and Arnott 2003; Russell et al. 2003, 2004). These events occurred while the LIS was intact across the western Lake Ontario basin, the Niagara Peninsula, and the eastern Lake Erie basin before deposition of the Oak Ridges moraine in the area north of western Lake Ontario. Clearly, further investigation of the eastern basin environment during the Mackinaw phase is needed to understand the processes operating there.

The northern portion of the central Lake Erie basin is a wave-cut terrace, presumably eroded by storm wave and current action. The terrace is an erosional unconformity, evidenced by wave-formed ripples of sand and gravel where it is exposed at the lake bed. The terrace is covered intermittently with a rippled lag of sand and gravel down to lake depths of 20–22 m south of the base of Long Point in the east, decreasing to depths of 15–16 m in the west near Pelee Point. Beyond these depths, the lag-covered erosion surface extends >30 km off the northern shore of the central basin beneath a cover of Holocene mud to depths of 28–36 m and possibly deeper. The deeper occurrence suggests that erosion of the terrace was underway at least during the early to mid-Holocene lowstand.

Relatively high frequency Kelvin Hughes echograms (14.25 kHz), supplemented by Raytheon echograms (7 kHz), and sampling of the terrace by gravity and piston corers, revealed alternating truncated beds of glacial diamicton (till) and glaciolacustrine sediments. The former, with massive acoustic facies, and the latter with parallel reflection acoustic facies, crop out on the erosion surface (wave-cut terrace). Diamictons of the Port Bruce westward glacial readvances, and glaciolacustrine sediments during glacial retreats, some deposited as rhythmites in proglacial lakes, were mapped in the northern part of the central Erie basin. Unit M, a Port Bruce diamicton, parallels the north shore, constitutes the Erieau–Cleveland Moraine, and correlates with the lower and middle layers of Port Stanley Till exposed in the adjacent shore bluffs. Glaciolacustrine unit N overlies unit M which is succeeded, in turn, by diamicton unit O that correlates with the upper layer of Port Stanley Till.

Unit O and two other earlier readvance deposits ended in till tongues after grounding on a bedrock rise at the eastern margin of the central Erie basin. These features were revealed in a seismic profile between the central basin and the base of Long Point.

Lack of ridging in the deep-water part of the Erieau–Cleveland moraine indicates a possible grounding line at about 30 m depth in Lake Erie, and suggests that the Port Bruce deposits in the central basin were laid down by a westward advancing ice shelf. The 120–150 m thick shelf grounded near shore, and its ice front terminated at onshore moraines. The advancing Port Bruce glacier became a floating ice shelf after grounding on a bedrock rise between the eastern and central Erie basins.

Laminated sediments of lower unit P overlie the preceding Port Bruce unit O, and thicken at depth in the offshore. Sediments of the later Mackinaw glacial phase were identified in the eastern part of the central basin, and cast new information on the chronology and nature of Mackinaw sedimentation. Debris flows from Mackinaw unit Q (Wentworth Till) interfinger with the upper strata of unit P, and onshore, Wentworth debris flows interfinger with the upper layers of a delta in the central Erie basin near Jacksonburg, Ontario. The upper strata of the delta, correlative with the latter part of the Mackinaw glacial phase, contain delicate plant fragments well-dated at 15.96 cal ka BP. Four other samples of delicate leaf fragments dating 17.25–17.16 cal ka BP in the Delhi–Simcoe, Ontario, area, probably date Lake Arkona sediments and suggest the Mackinaw ice readvance was in progress during or slightly after that time (Table 4).

Unit P may represent a proglacial lake following Lake Arkona that existed between the Port Bruce and Mackinaw ice readvances. The offshore extensions of the Galt and Moffat onshore Mackinaw moraines project into ridges on Unit Q which constitute the offshore Norfolk morainic ridges between the base of Long Point, Ontario, and Erie, Pennsylvania. The offshore extension of the onshore Paris moraine may only appear on the south shore of Lake Erie about 26 km west of Erie, Pennsylvania. The northern part of the moraine may have been eroded during formation of the wave-cut terrace in northern central Lake Erie. A later glaciolacustrine unit R overlies unit Q on its eastern side, and may represent the deposits of a proglacial lake as the Mackinaw ice melted away.

A younger unit S occurs in the western part of the central Erie basin beneath the soft lacustrine Holocene mud unit T, and was encountered in boreholes and a piston core. Dated pollen stratigraphy showed that deposition of this unit in central Lake Erie ended when Lake Algonquin overflow from the Huron basin was diverted northward to lower outlets that drained the Lake Huron basin to the Ottawa River.

This contribution is dedicated to the memory of Dr. Thane W. Anderson (Geological Survey of Canada (GSC), retired), who passed away during preparation of this paper. Dr. Anderson, an expert palynologist, undertook the analysis of pollen in the Lake Erie core and boreholes, illustrated in Fig. 9. We are indebted to the officers and crew of the research vessel CNAV Porte Dauphine for their assistance in obtaining offshore samples, acoustic profiles, and lake bed samples from Lake Erie during the 1960s. We thank George D. Hobson and crew (GSC) who acquired the seismic record across the eastern margin of the central Erie basin in 1967. Similarly, we thank the officers and crew of CCGS Limnos for their assistance in 1987. We are grateful to Dr. R.L. Thomas (Canada Centre for Inland Waters, CCIW) for his assistance in obtaining the Limnos ship time. We thank the Technical Operations group (CCIW) for their expert assistance with sediment sampling and acoustic profiling, and The Consumers Gas Company and its geotechnical contractor for expert assistance with the western central basin boreholes. Staff of the former Illustrations Unit at Bedford Institute of Oceanography are thanked for preparing some of the figures in this paper. We appreciate reviews by S.M. Blasco (GSC Atlantic) and D.R. Sharpe (GSC) of this manuscript, and Walta-Anne Rainey (GSCA) for improvement of Figs. 1, 3, and 5. We appreciate the thorough reviews and comments by the Associate Editor (O. Lian) and reviewers (G. Brooks and T. Fisher) that helped us improve the paper. Onshore data are used with the permission of the Director, Ontario Geological Survey. This is Natural Resources Canada contribution number 20210264.

Offshore sediment data generated or analyzed during this study are provided in full within the published article and its supplementary materials (Table S1, file: cjes-2023-0017suppla). Offshore sediment sample data for the whole lake and some interpreted echogram data are available in Geological Survey of Canada Open File 8818 via GEOSCAN on the internet. Echograms are archived at the Geological Survey of Canada Atlantic, Curator K. Jarrett (email: [email protected]). Lake Erie seismic reflection data Line LE 19 are available from GSC Open File 8947 (Todd et al. 2023) via GEOSCAN on the internet.

Conceptualization: CFML

Data curation: CFML

Formal analysis: CFML, GDMC, PJB, BJT

Investigation: CFML, GDMC

Methodology: BJT

Validation: PJB

Writing – original draft: CFML

Writing – review & editing: CFML, GDMC, PJB, BJT

This research was supported by the University of Toronto and the Geological Survey of Canada.

Supplementary data are available with the article at https://doi.org/10.1139/cjes-2023-0017.