En hommage à Jean-Pierre Lautridou.
In tribute to Jean-Pierre Lautridou.
The authors are deeply grateful to Paul F. Karrow of the University of Waterloo for his help during field investigation and his meaningful comments on several drafts that greatly improved the manuscript. Sincere thanks to Alayn Larouche who helped with the Tilia and TGView programs. The authors thank very much Maria Fernanda Sánchez Goñi who provided new references and contributed to revise the timing of the Last Interglacial. Many thanks also to the anonymous reader for his comments which lead to a deeper analysis of the global stratigraphy of the Upper Pleistocene.
1The time scale of the worldwide climatic fluctuations during the last interglacial Marine Isotope Stage 5e (MIS 5e) and following episodes of MIS 5d to 5a, is now precisely delimited by the marine and glacial isotopic records. In Europe, this time scale is applied to several terrestrial sequences, being aware that the duration of continental climatic episodes is variable and depends on the geographic settings of latitude, elevation and continentality.
2In North America, MIS 5 continuous terrestrial sequences are rare and with variable climatic accuracy (see section 5.10). However, tens of units are related to the climatic optimum of MIS 5e, the Sangamonian Interglacial sensu stricto, equivalent to the Eemian in continental Europe. The lower beds of the Don Formation, in the Toronto area (fig. 1 and 2), described as early as 1894 by Coleman and in 1895 by Chamberlin, are the most studied continental sediments of this interglacial in North America (review by Kelly et al., 1987). The entire Don Formation rests on the York Till, related to the Illinoian (Saalian) Glacial Stage and is overlain by the Scarborough Formation deposited in a glacial lake. Some valleys are carved into these two formations, partly filled in by the Pottery Road Formation (fig. 3 and 4). This succession is then overlain by several stadial and interstadial units of Wisconsin age (Karrow & Occhietti, 1989; Karrow, 2004).
3The Don Formation was deposited as lower shoreface beds (Eyles & Clark, 1988), near the mouth of a paleoriver named the Laurentian River (White & Karrow, 1971), in the non-glacial Lake Coleman (Terasmae, 1960; Karrow, 1990). Most of the studies conducted before 1950 on the exposures of the Don Valley Brickyard in Metropolitan Toronto (fig. 5) focused on the lower fossiliferous beds, with a fauna including Bison, Ursus, Castoroides and two types of Cervidae (Coleman, 1933, 1941). These lower beds correspond to the interglacial climatic optimum. Gray (1950) was the first to describe all the lithologic parts of the Don Formation. His unpublished thesis, difficult to access and with obsolete correlations, contains original lithologic and lithostratigraphic data that have not been subsequently taken up. Later, the climatic sequence constructed by Terasmae (1960) from the pollen content and other fossils of the formation became the reference frame for numerous further studies and remained so. However, the pollen content of the lowermost and uppermost beds was not identified, the Non Arboreal Pollen (NAP) and the spores were not counted, and the number of samples was limited. Eyles & Clark (1988, 1989) made a detailed facies analysis before the closure of the Don Valley Brickyard in 1984, without stratigraphy. Some of their conclusions were debated by Karrow (1989). Thus, up to now, the studies after 1950 did not use nor define a lithostratigraphic framework for the Don Formation.
4Based mostly on the sequence of Terasmae (1960), climatostratigraphic charts have been used for several decades. A long hiatus between the Don Formation and the overlying glaciolacustrine Scarborough Formation was inferred, which corresponds to an interval between respectively the Sangamonian Interglacial and related transitional episode equivalent to MIS 5e, and an early Wisconsinan cold stage attributed either to MIS 5b or to MIS 4 (Prest, 1970; Terasmae et al., 1972; Karrow & Occhietti, 1989). Karrow et al. (2000) propose an event chart with the same long hiatus for the Eastern and Northern Great Lakes area (see chapter 5.8). Conversely, based on facies analysis, Eyles & Clark (1988) conclude to a continuous sedimentation in a deepening Coleman Lake. Without reference to other regions, the Don and Scarborough Formations are then related to the entire MIS 5 (Eyles & Williams, 1992). Later, based on the same facies analysis, a contradictory chart is proposed by Berger & Eyles (1994) from thermoluminescence (TL) ages: the Don Formation, with calculated ages circa 80 ka is related to substage MIS 5a, and the Scarborough Formation with calculated ages circa 68 ± 9 ka corresponds to MIS 4. All those divergent interpretations concern the climatic variations and the inception and fluctuations of the Laurentide Ice Sheet during MIS 5. Regarding the major influence of the continental glacier of North America on global change, the significance of the Don Formation is of international importance, and clear litho- and biostratigraphy of this formation are prerequisites for any additional dating and finer correlations.
5The objective of this paper is to provide a renewed look at the published data and present new field observations and biological data on the Don Formation. The first part applies an allostratigraphic approach to the formation, followed by a detailed pollen analysis including NAP countings. After the reassessment of the biotic content established by previous studies and of the sedimentation processes involved, the different allozones of the formation are correlated to dated units of the St. Lawrence River Valley and Estuary, and alternative chronologies are proposed. Variations of the level of lakes Coleman and Scarborough through time are then tentatively reconstructed. Finally, the interglacial and transitional sequence of units of the Toronto area and St. Lawrence Valley is compared to other sequences in North America.
Fig. 1: Location of Toronto and USA sites with Sangamonian soil, beds or speleothems referred to in Section 5.10. The limits of interglacial Lake Coleman changed during the interglacial and transitional time; the limit on the figure is tentative.
Fig. 2: Simplified path of the ancestral Laurentian River and location of Toronto and other sites with Interglacial Sangamonian beds in eastern Canada (from Richard et al., 1999) including the Moose River site (Mott & Dilabio, 1990), and in New York State (Karrow et al., 2009).
Fig. 3: Upper part of the Don Valley Brickyard working face, Toronto (enlargement of a color photo of Paul Karrow, 1957): planar geometry of the Don Formation with continuous allozones and a gentle slope from the East (right) to the West (left).
Pale bands (even numbers) indicate more sandy and drier groups of beds. Darker bands correspond to moister groups of beds. The continuous and thin pale band IV is related to an erosional discontinuity and indicates the upper limit of the climatic optimum of the Sangamonian Interglacial.
Fig. 4: Lower units of the Toronto Quaternary sequence and Allozones D1 to D5 of the Don Formation, at the Don Valley Brickyard (enlargement of a color photo of Paul Karrow, 1984).
Fig. 5: Location of the local sites in the Toronto area, referred to in the text.
6This study takes into account the various approaches developed in previous work (Occhietti, 1990; Richard, 1994; Clet-Pellerin & Occhietti, 2000) and in the studies of other authors (Gray, 1950; Karrow, 1967, 1990; Eyles & Clark, 1988; Kelly & Martini, 1986; Lamothe, 1989). Given that the Don Formation accumulated at the margin of a fluvial system, we propose a stratigraphy of the formation based on allostratigraphic and sequence analyses (Bhattacharya, 2001) coupled with biostratigraphy. From the sequence stratigraphy principles (see Catuneanu et al., 2011) and their application to sequences of the Toronto area by Martini & Brookfield (1995), attention is given to discontinuities (fig. 3, 4, 6 and 7) and sedimentary system tracks. The paleoenvironmental interpretations and paleoclimatic reconstructions are based on converging facts at local, regional and global scales.
Fig. 6 : Detailed view of the Don Formation at the Don Valley Brickyard (sample log by S. Occhietti, 1998). The lower Allozones D1 and D2a were covered by debris.
7Former pollen analyses by members of our team, mainly on the Scarborough Formation, revealed the necessity to tally the NAP grains and to analyse more samples within a given unit (Richard et al., 1999). With sandy and gravelly beds of the Don Formation usually sterile in pollen and spore grains, a close and selective sampling had to be applied to beds with silt and clay when possible. Therefore, the sampling intervals in the open part of the Don Formation are not regular, with most of them close to 10 cm. Gravel beds and several sandy beds were not sampled (fig. 8). The sandy upper part, poorly studied so far, was closely sampled, every 5 cm or less. The lower part of the overlying Scarborough Formation was sampled every 10 or 20 cm, for a total of 18 samples.
8The samples (20 g) were sieved between 100 µm and 10 µm after removal of carbonates with hydrochloric acid and removal of silica with hydrofluoric acid. They were prepared without acetolysis. The concentrations (number of pollen grains per gram of sediment) were calculated using the weighted aliquot method of Jørgensen (1967). The pollen percentages were calculated using a standard sum of all the pollen of terrestrial vascular plants counted (300-400 grains per sample). The representation of spores (Pteridophyta, Sphagnum) was expressed over this Pollen Sum. Identification was made according to the criteria of Richard (1970); however, because of morphological uncertainties, Pinus banksiana/divaricata and Pinus resinosa pollen grains were grouped within the Diploxylon pines and Pinus strobus was specified as an Haploxylon pine. Besides, identification of spruce pollen to the species level is tentative: Picea glauca type is nevertheless distinguished from Picea mariana/rubens type. Regarding alder identification, Alnus incana ssp. rugosa type is clearly distinguished from Alnus viridis ssp. crispa type. The pollen taxa are grouped into the following categories: thermophilous trees, other trees, shrubs, herbs, spores (fig. 8). The diagram was prepared with Tilia, version 1.7.16 (2011), and plotted with TGView, version 2.0.2 (2004), developed by Dr Eric C. Grimm, from the Illinois State Research and Collections Center at Springfield, Illinois.
9The climatic significance of the Late Pleistocene units is assessed according to the pollen and spore content, the inferred past vegetation and its evolving trend. The pollen and spore assemblages are compared to the Holocene assemblages and their associated climatic significance, as expressed in modern vegetation (Ritchie, 1987; Anderson et al., 1991; Richard, 1994).
10From our previous work mostly in the St. Lawrence River Valley and Basin (Clet & Occhietti, 1994; 1995; Occhietti & Clet, 1989; Richard et al., 1999), distinct pollen and spore assemblages occurred during Late Pleistocene and characterized the regional paleoenvironmental dynamics (Clet-Pellerin & Occhietti, 2000). These are as follows, from warm temperate to subarctic assemblages (tab. 1): 1) the Mixed Hardwoods (MW) assemblage (MWa, MWb, MWc) which contains trees and shrubs found in temperate deciduous forests; 2) the Southern Boreal Forest (SBF) which is subdivided into SBFa: a balsam fir (Abies balsamea) dominated assemblage, and SBFb: a pine (Pinus diploxylon e.g. P. banksiana and/or P. resinosa) dominated assemblage; 3) the Northern Boreal Forest (NBF): a spruce (Picea cf. mariana) dominated assemblage; 4) the Forest Tundra (FT) with evidences of an open tree canopy with a subarctic character: Alnus cf. crispa, Ericaceae, Sphagnum. These pollen and spore assemblages are related to vegetation types which correspond in most cases to the canadian biomes (Strong et al., 1989; Dyke, 2005).
11Pollen or spores from species living along the banks of lakes and rivers were grouped in an azonal group labelled RB (River Banks). It comprises mainly Alnus rugosa, and part of Cyperaceae, Poaceae, other herbs and Pteridophyta. This group is present throughout the sedimentary sequence but Sphagnum is however best represented in the overlying Scarborough Formation.
Fig. 7: Log of the section of the Don Formation sampled in 1998: allozones, pollen zones and fossil assemblages from other studies referred to in tab. 4.
The covered basal part of the section (D1 and D2a) is simplified from Karrow (1969). HK: Hann & Karrow, 1993; R: Richard et al., 1999; T: Terasmae,1960.
12Using a system track approach, the correlations between the Don Formation and units of the St. Lawrence River Valley are based on the stratigraphic relative position, the nature of the units and their elevation, the climatic conditions deduced from the biotic content, and the inferred sea level at the time of deposition established from the global climatic conditions.
Fig. 8: Pollen and spore diagram of the sampled parts of the Don and Scarborough Formations at the Don Valley Brickyard, Toronto.
Tab. 1: Upper Pleistocene pollen assemblages identified in Eastern Canada, from Clet-Pellerin & Occhietti (2000).
13Changes in orbital forcing is the primary mechanism of climate changes, at the scale of thousands of years. Due to changes of the insolation through time and latitudes, to complex retroactions of the global climatic system and to variable inertia, the climatic responses in oceans, continents and glaciers are not perfectly synchronous (see the report of Wohlfarth, 2013). Nevertheless, significant climatic changes are recorded within a defined time span. For example, continental beds with markers of the climatic optimum of the last interglacial can be correlated with marine beds with benthic foraminifers and pollen which indicate a marine high stand and elevated surface temperature. Despite some differences in ages given to the limits of MIS 5e and clear differences in the duration and representation of the Eemian - Sangamonian Interglacial according to the regional setting, a careful correlation between the continental and oceanic stratigraphic systems gives a global significance to local data. The Clear Lake pollen record in California (Adam et al., 1981) and the reassessed chronology of the Devils Hole samples of calcite (Moseley et al., 2016) are examples of the global response within North America. Shorter climatic events (in broad sense), identified on continents and in Greenland ice cores may be related, or at least compared as similar types of events. This is usually applied for example to the late-glacial Younger Dryas “event” on both sides of the Atlantic Ocean, even if this cold episode is not directly linked to orbital forcing. The reassessed continental record of the Don Formation can therefore be compared to the European, Greenland and marine series, using an unambiguous terminology.
14Marine Isotope Stage 5 (MIS 5) occurred between 135-130 and 72-70 ka (Shackleton et al., 2003; Sánchez Goñi et al., 2012; Capron et al., 2014; Simon et al., 2016). From 18O and other markers, in concordance with the orbital forcing, this stage comprises five main climatic substages: the interglacial climatic optimum (MIS 5e) and subsequent colder or mild substages (MIS 5d to 5a). In Europe, the pollen content of several continuous continental series, mainly from peat bogs (La Grande Pile, Woillard, 1978; Les Echets, de Beaulieu & Reille, 1984) and lake fillings (Reille & de Beaulieu, 1990), and from discontinuous fluvio-deltaic accumulations (Rhine delta, Zagwijn, 1996), indicates the same climatic pattern. The climatic optimum is related to the Eemian, considered as the interglacial sensu stricto, and the subsequent substages are referred to the Early Weichselian. Nevertheless, as mentioned, the boundaries of the continental and oceanic records are not exactly synchronous. From Shackleton et al. (2003), the base of the Eemian is younger than the base of MIS 5e and falls within the isotopic “plateau” of the marine isotope stage (tab. 2).
15In continental North America, the direct equivalent to MIS 5, both in time span and climate, was the Sangamonian Stage in its former definition. This period was defined, in Illinois, from a paleosol developed in Illinoian drifts and related colluvial deposits, and overlain by Wisconsinan till or loess (Willman & Frye, 1970). According to this definition, which was in use for a long time (Fulton, 1989; Zhu & Baker, 1995), the Sangamonian Stage (sensu lato) was equivalent to the European Eemian and Early Weichselian altogether. The usage of Sangamonian Stage in its broad sense came also from a tacit assessment that the development of glaciers in North America was limited during substages MIS 5d and 5b, and very limited or even close to interglacial conditions during respectively substages MIS 5c and 5a. Nevertheless, this model with a short glacial time (MIS 4 to 2) was debated mostly in Canada (Dreimanis, 1977; Karrow & Occhietti, 1989; Grant, 1989; Occhietti et al., 1995), and the extent of the glaciers in North America during substages 5d and 5b is still to be assessed from direct continental evidence. Sangamonian Stage sensu stricto was also used by some authors in order to distinguish the warmer interglacial phase equivalent to the Eemian. Richmond & Fullerton (1984) proposed to use Sangamonian Stage as the equivalent to the Eemian, and Eowisconsinan Stage as the equivalent to MIS 5d to 5a substages. The Sangamonian sensu stricto is now used as the interglacial stage (see 6. Delineation of the Last Interglacial - Last Glacial stage boundary, in Otvos, 2015). Using a diachronic nomenclature, Karrow et al. (2000) distinguishes the Sangamon Episode related to MIS substage 5e, and the Ontario Subepisode of the Wisconsin Episode as the equivalent of MIS substages 5d to 5a and MIS 4, in the eastern and northern Great Lakes area.
16In this paper on continental deposits, the climatic optimum of the last interglacial is related to the beds with indicators of climatic conditions warmer than today and to the beds which directly precede and follow those beds with indicators of climatic conditions as warm as today. Some beds or palynozones which record very short cooler events can be included in the series of beds or palynozones of the climatic optimum. The climatic optimum beds are attributed to the main part of the Sangamonian Interglacial and to the “plateau” of MIS 5e (Shackleton et al., 2003). Without a complete series such as the one at La Grande Pile (Woillard, 1978) and because of the homotaxy of beds with interstadial or stadial markers (Karrow, 1989; Occhietti, 1990), the overlying units or beds with transitional and colder indicators of climate have an equivocal stratigraphic position, as observed by Mott & Dilabio (1990) in the James Bay Lowlands. In this study, all the transitional beds of the Don Formation are revealed to have an equivocal position. For this reason, they are temporarily not related to the Sangamonian (sensu stricto), until univocal evidence can be obtained.
17In many papers, Last Interglacial (LIG), Eemian, Sangamonian and MIS 5e refer to the same event. Nevertheless, the given limits of the last interglacial and transitional phases which precede the major glaciation related to MIS 4 are somehow changing from one reference to the other. For example, the age given to the lower limit of Stage 5e varies from 135 ka (Sánchez Goñi et al., 2012) to 132 ka (Shackleton et al., 2003), and to the widely used value of 130 ka (Simon et al., 2016). Usually, MIS 5d is related to the Melisey I continental cold phase, from 115-112 ka to 105 ka, but covers also the end of the Eemian in Sánchez Goñi et al. (2012), from 120 ka to 105 ka. From a wide compilation including Antarctic ice cores, Capron et al. (2014) define the LIG period between 129 and 116 ka, and Veres et al. (2013) develop a time scale for the last 120 ka. Establishing formal limits to MIS 5 and to the ice core and continental phases/episodes is out of the scope of this study. In tab. 2, the data from the Toronto lower sequence of units are compared to a perfectible climatostratigraphic and chronostratigraphic framework based on several sources, keeping in mind that the limits given to the different phases/events can vary by some thousands of years (see Tables 6-2 and 8-1 in Wohlfarth, 2013). Toronto is located at the same latitude than Nice, France; therefore, despite some continentality buffered by the proximity of the Great Lakes, the Toronto units are compared to the climatostratigraphic chart of central and southern Europe (tab. 2).
Tab. 2: MIS 5 chronology in Europe and Greenland, and age hypotheses of the allozones of the Don Formation.
* 135 ka: MIS 5 lower limit by Sanchez Goni et al. (2012). ** Lower limit of MIS 4 above Ognon proposed by Sanchez Goni et al. (2013).***Limits of cold events Melisey I and Montaigu from Drysdale et al. (2007).
18In the Don Valley Brickyard, Gray (1950) recognized a decimetric Basal Clay and three other units in the Don Formation, a lower deltaic sand (Unit A) up to 5.4 m thick, an intermediate lacustrine unit up to 1.4 m thick, with some stones, deposited in deeper waters (Unit B), and an upper weathered sand (Unit C) up to 1.8 m thick. The total thickness reaches 7.6 m, but the thickness of each unit is variable. Gray (1950) stated (p. 12) that “sections fail to correlate in detail; no on (one?) stratum can be said to extend completely around the quarry in a consistent fashion, except for the basal clay”. According to Coleman (1941), the “basal blue clay with logs of wood and unios (clams)... are of warm climate”. From Karrow (1969), the top of the Don Formation slopes southeasterly from 102 m a.s.l. in central Toronto to 93 m at the Don Bri-ckyard and 70 m a.s.l. below the Scarborough Bluffs along the present Lake Ontario shore. The sedimentary structures and changes of facies were related to a river entering the shallow interglacial Lake Coleman and the position of the river channel (Terasmae, 1960). Eyles & Clark (1988) described the “Don Beds” as sands interbedded with bioturbated peaty muds. They stated that “Most sand beds are composite in origin,… each in erosional contact with the underlying unit.” These beds were deposited in a lacustrine lower shoreface environment subject to episodic storms and to increasing depth, from 2 to about 18 m, that is 12 to 28 m above present level of Lake Ontario. They did not recognize any major erosional discontinuity within the Don Formation or between the top of it and the overlying Scarborough Formation.
19Numerous studies of the biotic content were done (reviewed in Kerr-Lawson et al., 1992). Terasmae (1960) established the pollen sequence of the Don Formation from 33 pollen assemblages. According to his sequence and reassessing previous works (Coleman, 1933; Gray, 1950; Watt, 1953), the lower part of the Don Formation was deposited during the climatic optimum of the Sangamonian Interglacial, and the upper part of the formation is related to a change towards a boreal climate. This is confirmed by other biota (see chapter 5.3), by a stratigraphic diagram of prominent tree pollen by McAndrews (in Westgate et al., 1999), and by a preliminary study of the pollen content of a part of the Don Formation by Richard et al. (1999).
20The stratified sand beds with some pebbles and cobbles, located at the top of the unit (Gray, 1950; Terasmae, 1960) were interpreted as evidence for a further shallowing of the lake accompanied by a cooling of the climate (Terasmae, 1960). Their significance has been controversial. Gray (1950) concluded to a low lake level and even to the exposure of the formation, while Terasmae (1960) proposed a considerable hiatus between the Don and the Scarborough Formations. Karrow (1969) concluded there were deposition in shallow waters and a disconformity. During the study of fossil caddishflies, Williams & Morgan (1977) identified the following sequence of sediments: lowermost beds deposited on the substrate, lower quiet-water deposits with current-bedding structures and some large logs, a buff-grey coarse cross-bedded sand unit which suggests a more littoral facies, clay laminae and ripple-marked structures with detrital organics related to deeper water conditions, and at the top of the sequence, limonite stained sands. One of two groups of caddishflies is similar to assemblages found on exposed shores of large lakes, the second group is related to large rivers. Morgan (1979) evoked a significant interval of weathering. On the west side of the Don Valley Bri-ckyard, Poplawski & Karrow (1981) observed coarse, rust-coloured sand and gravel below 1 m of iron-stained deposits. According to Hann & Karrow (1984), the organic matter content of about 4 % in the lower fossiliferous part decreases abruptly to 1 and 2 % upwards. The percentages of carbonate content decrease slowly, from 12 to 8 %, in the lower beds, and rapidly to 2 % in the overlying part. Conversely, for Eyles & Clark (1988) the rusty upper sand bed is related to post-sedimentation underground waters and the sedimentation is continuous between the Don and Scarborough units.
21Scattered findings indicate that a large sedimentary body of the order of one hundred to several hundred km2 was deposited in the Toronto area, in the lower reaches of the Laurentian River valley at the edge of Lake Coleman. The Don Formation was exposed naturally in the banks of the Don River and widely accessible in the former Sun Brickyard (Coleman, 1933) and, since 1899, in the now abandoned Don Valley Brickyard (fig. 5). From the pre-sence of abundant mollusc shells, and sometimes large pieces of wood, the part of the formation deposited during the interglacial climatic optimum was recognized in many temporary excavations downtown Toronto (Karrow, 1990). The largest excavations were done during the construction of subways in the 1950 and 1960’s in Metropolitan Toronto. These data and subsequent excavations were compiled by Sharpe (1980) and Eyles (1987) as a series of cross sections with a total length of nearly 105 km. In the Scarborough area, downstream the Metropolitan Toronto on the North shore of Lake Ontario, the Don Formation was reached by coring (Terasmae, 1960) and by drilling (Hann & Karrow, 1993) under the Scarborough Formation.
22A major interglacial and early stadial river, with no modern equivalent, is inferred from this large sedimentary body (White & Karrow, 1971; Eyles, 1987; Martini & Brookfield, 1995) and from the biotic studies. The river flowed along a topographic low in the bedrock identified as the Laurentian River channel by Spencer (1890). The bedrock contours of the buried valley are established from drilling and water wells (White & Karrow, 1971; Westgate et al., 1999). From quartzite cobbles in the Don Formation, the interglacial drainage is thought to have connected Lake Huron to the Lake Ontario basin (Gray, 1950), over an approximate distance of 110 km (fig. 9; see Fig. 1 in Eyles, 1987).
23Most of the knowledge of the Don Formation is based on the exposures followed through time at the Don Valley Brickyard. The distal part of the sedimentary body was deposited in deeper waters, as suggested by cladocerans in logs of the Brimley Road site (Hann & Karrow, 1993) (fig. 5). This part, equivalent to prodelta and bottom deposits, was eroded during subsequent glacial phases (Martini & Brookfield, 1995) and the post-glacial Lake Iroquois low phase, as indicated by a prominent escarpment on offshore seismic profiles (Eyles & Clark, 1989; Anderson & Lewis, 2012). Upstream from the Don Valley Brickyard, shallower shoreface deposits are observed on the Leaside site (Hann & Karrow, 1993) (fig. 5), and further northward in the Toronto vicinity, sandy facies correspond to fluvial deposits. If not erased by the Wisconsinan glaciers, concealed fluvial sediments of the interglacial Laurentian River are potentially present in the middle and upper reaches toward the Georgian Bay, as suggested by the interglacial beds of the Woodbridge Cut, 21 km northwest of the Don Brickyard (Sharpe, 1987; Karrow et al., 2001) (fig. 5).
24The status of the Don beds has changed since the early works of Coleman (1894, 1933) and Chamberlin (1895), regarding the interpretation given to the Don-Scarborough succession of formations (Eyles & Clark, 1989; Karrow, 1990). Either, the two units are in continuity (Coleman, 1933, who maintained the name of Toronto Formation; as reused by Eyles & Clark, 1988), or they represent distinct units (Don and Scarborough Formations, Karrow, 1967, 1990).
Fig. 9: Inferred drainage pattern between the Great Lakes during the Last Interglacial and transitional phase, from Spencer (1890), White & Karrow 1971) and Westgate et al. (1999).
25In the Toronto area, a set of thermoluminescence dates by Berger & Eyles (1994) gives an age of 80 ± 19 ka at the base of the Don Formation, 78 ± 17 ka in the middle part, and 67.6 ± 9 ka near the top of the formation. The same set provides also TL ages of 59.8 ± 8.5 and 54.1 ± 8.2 ka at the base of the Scarborough Formation.
26Several early correlation charts between the units of the Lake Ontario Basin and the St. Lawrence River Valley, referred to in the introduction, were based on the 14C age (see discussion in Dreimanis, 1977) of the St. Pierre Sediments found in the central St. Lawrence River Valley (Gadd, 1971; Ferland & Occhietti, 1990a). From the age of 74.7 ka + 2.7/-2.1 (Stuiver et al., 1978), these sediments are related to MIS 5a. The contradictory charts proposed by Eyles & Clark (1988); Eyles & Williams (1992), and Berger & Eyles (1994) do not take notice of stratigraphic successions of other areas. Meanwhile, the stratigraphy of the central St. Lawrence River Valley was revised, with the identification of an early Wisconsinan glacial and non glacial succession and the reassessment of several units (Lamothe, 1985, 1989; Occhietti & Clet, 1989; Occhietti, 1990; Besré & Occhietti, 1990; Ferland & Occhietti, 1990b; Bernier & Occhietti, 1991). Preliminary correlations between the lower units of the two areas were proposed by Clet & Occhietti (1994) and by Richard et al. (1999).
27On the working face of the former Don Valley Brickyard, the formation presented eight distinct metric to submetric continuous bands, either pale or dark grey (fig. 3, 4 and 6). Light bands are composed of drier sandy beds (bands with even numbers II to VIII ; fig. 3 and 4). The darker bands (bands with odd numbers) are more clayey and moister, the basal water-saturated sandy band I being the exception. In the Don Formation, the moisture varies within a range of 22 to 6 % of water content (Hann & Karrow, 1984). The bands show a progressive thickness variation at the scale of a single working face (fig. 3) and stronger variations at the scale of the past working faces and between working faces with different orientations (Gray, 1950). The visual bands are approximate and informal units, as the moisture limits are only indicative of the dominant grain size and may vary with the seasons. Thus, this visual approach provides only a schematic view of the tabular geometry and gentle slope of the Don Formation. From a detailed analysis, each continuous band corresponds to a sequence of discontinuous beds with roughly the same dominant grain size or same dominant lithofacies.
28Based on this general geometry, together with sampling logs and published detailed studies, the Don Formation can be subdivided into five allozones (D1 to D5), separated mostly by erosional limits (fig. 3,4 and 6).
29The lowermost Allozone D1 corresponds to the discontinuous “Basal Clay” described by Coleman (1933) and Gray (1950), and mentioned by Terasmae (1960). The unit is composed of layered clay, interbedded with coarse dark sand. The thickness is decreasing (40 to 15 cm) from the early working faces of the brickyard to those observed by Gray (1950). On the late exposures, the clay beds were apparently discontinuous. This zone belongs to the lower grey band I. Some cobbles and blocks lay on the York Till at the bottom of the lowermost beds D1 or D2a (fig. 7).
30The lower Allozone D2a is mainly sandy and corresponds approximately to the lower grey band I (fig. 4), with an irregular erosional upper limit with ripples and lenses of gravel (Allozone D2b; the unconsolidated “pebble conglomerate”of Gray, 1950; see also the lower Gr beds in Fig. 5 of Eyles & Clark, 1988). Just above, Allozone D2c (fig. 4 and 6) comprises the sequence of sand beds related to the pale band II. Allozone D2d (fig. 6) corresponds to the conformable grey band III (fig. 3 and 4) composed of planar silt and fine sand beds, and is overlain by the pale grey sand beds of Allozone D2e (fig. 6). The erosional upper limit is irregular.
31Allozone D3 is limited to coarser beds with gravel related to the end of a major erosional hiatus in the pale band IV. This allozone represents a continuous marker unit on the faces of the brickyard (fig. 3 and 4). On the sample log (fig. 6), the thickness of these beds varies between 20 and 40 cm. This unit corresponds to the horizon with unconsolidated “pebble conglomerate” and gravel described on several sections by Gray (1950) at the top of his Unit A. Allozone D3 is overlain by distinct finer deposits.
32Allozone D4 comprises the sequence of silt and fine sand beds of the grey bands V and VII (fig. 4), and the intermediate sandy beds of pale band VI which is thinning from the West to the East on the brickyard working face (fig. 3). This unit corresponds to the ‘’lacustrine’’ Unit B of Gray (1950).
33At the top of the formation, Allozone D5 (fig. 4 and 6) is over 1.2 m thick, heterogeneous and corresponds roughly to the upper visual band VIII (fig. 4 and 6). During sampling, we observed a basal bed (10 cm) with green shale fragments (D5a) (uppermost part of grey band VII, fig. 4 and 6), covered by a thin bed of green sand (D5b). Above it (fig. 7), a pavement of green shale flat cobbles and wood fragments (D5c) underlies beds of greenish sand (D5d) and white sand and fine gravel beds (D5e). The uppermost beds (D5f and D5g) consist respectively of grey sand with frequent flat cobbles of green shale and of strongly weathered coarse brown sand.
34The seasonal, annual or multi-year beds of the formation are discontinuous, as stated by Gray (1950) and Eyles & Clark (1988). These inner discontinuities are the rule in every log of the formation and every continuous allozone. For this type of sediments, the thickness of the five allozones and their subdivisions varies within the entire sedimentary body. Nevertheless, despite these lateral thickness variations, the relative position of each allozone is constrained within the formation, as it can be pointed out in all the photos of the former Don Valley Brickyard (see photos by Coleman, 1941, p. 74; Coleman in Eyles & Clark, 1988; Eyles, 1987; Karrow, 1990; this study).
35The exposed middle and upper parts of the Don Formation (Allozones D2b to D5) were resampled (1994 and 1998), resulting in 16 sterile samples and 42 pollen assemblages. From this analysis, five palynozones can be distinguished in the formation, with a close but not perfect relationship between allostatigraphy and biostratigraphy (tab. 3; fig. 7 and 8).
36The pollen assemblages are essentially interpreted as indicating climatic changes. Given the fact that the sedimentological sequence represents shoreface deposits and is not deposited in an enclosed lake basin, the pollen and spore content of the formation should be affected by changes of catchment size in the river system feeding the shoreface, or by changes in discharge resulting in sorting. In such a situation, it may appear unwise to attribute change in palynological assemblages directly to climatic change when variations arising from changing taphonomy could potentially explain much of the identified variability. The interpretations of climate instability from our record, presented below, are nevertheless valid because there is no evidence of a reaction to obvious changing taphonomy throughout the record. An exception may be found in the behavior of the Monolete spores (most probably derived from erosion of river banks occupied by ferns) that parallels the total pollen concentration. However this does not affect the sequence of changes in the mainland vegetation, as translated by the successive pollen assemblages.
37Terasmae (1960) did not sample the basal clay, and these beds were buried in 1994 and 1998. Therefore, the pollen content of Allozone D1 is unknown, but Hann & Karrow (1984) identified in the lowermost beds of the Don Formation a cladoceran fauna that can be found today in glacial lakes and boreal to subarctic habitats. This cladoceran biozone and the equivalent virtual pollen zone Don 1 would correspond to a transition after the Illinoian glaciation. Nevertheless, trunks and branches of red cedar and Unio shells with both valves united observed in the lowermost blue clay by Coleman (1941) correspond to warm interglacial conditions. This contradiction is not resolved.
38Allozone D2a was buried in the 1990's and only one sample could be obtained from the top of the unit. The analyses of Terasmae (1960: Fig. 19) are used to define the type of assemblage related to these beds. The pollen content of 10 samples from Terasmae and one from this study (fig. 8; Palynozone Don 2.1) can be related to a zonal Mixed Hardwood Forest (MWb) assemblage with Quercus, Carya, Tilia, Acer and Ulmus, and without Liquidambar.
Tab. 3: Processes and conditions related to the deposition of the Don and Scarborough Formations, from the Illinoian/Sangamonian (MIS 6/5) transition to an early Wisconsinan stadial (MIS 5d or b).
39The new pollen spectrum contains Southern Boreal Forest (SBF) element with diploxylon pine pollen (probably Pinus cf. resinosa) and Picea cf. mariana. This is related to a short cooler episode.
40Palynozone Don 2.3 is characterized by the occurrence of Liquidambar and Nyssa along with most of the other Mixed Hardwood Forest (MWa) species. The basal part comprises dominant Carya (37 %) and Quercus, then dominant Quercus, Pinus diploxylon (cf. P. resinosa) and Carya. In the upper part, the forest contains dominant Quercus, Ulmus, Pinus diploxylon (cf. P. resinosa), Tilia and Tsuga. The River Bank (RB) species with Larix and mostly fern monolete spores and herbaceous plants (Poaceae, Cyperaceae) are present (between 7 and 10 %). Pollen and spore concentration is low (2000 grains/g.), with increasing values in the upper part. The diversity is high (fig. 8).
41Increased representation of small size Betula pollen grains (cf. Betula glandulosa), Alnus cf. crispa and Ericaceae pollen characterize palynozone Don 2.4, indicating an important cooling. The palynozone is related to the upper half of sandy Allozone D2c (fig. 8). In the lower samples, elevated pollen representation of Quercus is maintained along with that of Juglans, but increasing of Picea cf. glauca, albeit low amounts, is in harmony with the presence of the mentioned shrubs. The corresponding vegetation may still be a deciduous forest, but experiencing a harsher climate. Some thermophilous tree pollen may be reworked from underlying beds of the shoreface. The upper sample is characterized by a Southern Boreal Forest assemblage with a low pollen representation of thermophilous trees and increasing amounts of Abies, Pinus diploxylon (cf. Pinus resinosa/banksiana) and Picea cf. mariana.
42Diploxylon pine (cf. Pinus resinosa) and Quercus are dominant with Carya and Ulmus of the Mixed Hardwood Forest. Then the vegetation evolves to Quercus, Pinus cf. resinosa, Carya, Ulmus, Fagus and Acer. Palynozone Don 2.5 corresponds to a zonal Hardwood Forest (MWb). Picea cf. mariana is present (7 to 10 %) and the Picea rubens type (Richard, 1970) is identified almost exclusively in this zone. The total concentration lowers progressively upwards the section (13,000 to 7000 grains/g.). The uppermost assemblage with low concentration (3000 grains/g.) is composed of Quercus, Carya, Pinus diploxylon (cf. Pinus resinosa) and Picea cf. mariana; the diversity of trees and herbaceous plants is significantly poorer than in the rest of palynozone Don 2 with the exception of the colder episodes Don 2.2 and 2.4.
43The gravelly Allozone D 3 is sterile. The assemblage immediately above the gravel corresponds to a harsher Southern Boreal Forest (SBFb) with Pinus diploxylon (P. cf. resinosa/banksiana), Picea cf. mariana and Quercus. The low concentration could indicate leaching of the sediment, but the abundance of Picea cf. mariana (34 %) would confirm real harsher conditions for the vegetation than the overlying zonal Southern Boreal Forest of Palynozone Don 4. Palynozone Don 3 follows a gap and represents a drastic change in comparison to the preceding zonal Mixed Hardwood Forest; the pollen grains were deposited at the end of an unknown vegetation episode, but the harsher assemblage is representative of an intermediate colder episode. The observed discontinuity in the studied log may be different laterally: some beds, truncated at the sampled section, may be present and contain pollen assemblages from a part of the intermediate vegetation between the two zonal forests.
44Palynozone Don 4 is characterized by Southern Boreal Forest assemblages with Abies balsamea, Pinus diploxylon (cf. Pinus resinosa/banksiana), Picea cf. mariana, and a sharp decrease of deciduous trees (Quercus, Carya). The palynozone is subdivided in three parts.
45Palynozone D 4.1 is characterized by a rapid increase of Abies (from 8 to 14 %), with Pinus diploxylon and Picea cf. mariana, and represents the zonal Southern Boreal Forest (SBFa) with some Tsuga. This vegetation developed under wetter conditions than during palynozones Don 2 and Don 3. The highest concentrations in the pollen diagram of the formation are observed in this palynozone (up to 70,000 grains/g.).
46A peak of Picea cf. mariana pollen, at the base of this palynozone, corresponds to a more sandy band (D4c, fig. 6 & 8), translating a vegetational response to a climate change promoting erosion. Opening of the forest cover in a colder environment is inferred.
47Abies balsamea decreases slowly (7 to 2 %), indicating the extent of a harsher Southern Boreal Forest (SBFb) with dominant Picea cf. mariana and Pinus diploxylon. The pollen concentration lowers to 2000 grains/g.
48Palynozone Don 5 is characterized mostly by Northern Boreal Forest (NBF) assemblages (Picea cf. mariana, Pinus cf. banksiana), with intermediate to very low pollen concentrations. The River Banks species are very rare, mainly in palynozone Don 5.2, even though tundra species are present. In the two upper palynozones Don 5.2 and Don 5.3, Betula papyrifera (25 micrometer class) indicates a boreal forest. In those palynozones, the occurrence of the small pollen size-class of Betula (Betula cf. glandulosa) and of Salix is related to a more open Northern Boreal Forest. This promotes an increased representation of the long-distance transported pollen of trees.
49Pinus cf. banksiana and Picea cf. mariana pollen are dominant but Abies is still present, being notably underrepresented by its pollen. The concentration is intermediate (20,000 grains/g.) and associated to a low diversity of trees and herbaceous species. These pollen assemblages are likely representative of the vegetation of a fairly closed Northern Boreal Forest.
50Despite a lower concentration (1100 to 3000 grains/g.), the diversity of tree pollen is higher than in palynozone D 5.1. The relative percentage of Picea cf. mariana increases in relation to the decrease of Pinus cf. banksiana and Abies.
51The pollen assemblages have the same diversity and concentration than those of Don 5.2. The percentages of diploxylon Pinus (Pinus cf. banksiana) increase with related Picea cf. mariana decrease. This is attributed to a stronger eolian transport of pollen grains in an open Forest Tundra.
52The pollen assemblages of the lower part of the Scarborough Formation are characterized by Betula cf. glandulosa (20 microns size-class), Alnus crispa, shrubs of the Ericacea family, Artemisia, Ambrosia and Sphagnum spores, all indicative of a very open and harsh Forest Tundra. Picea cf. mariana is present but most pollen of diploxylon pine is likely long-distance transported. Pollen zone Scar 1a simply represents a transition to Scar 1b.
53This new pollen analysis extends the preliminary results of Richard et al. (1999) for the Don Formation and confirms their harsh Forest Tundra vegetational interpretation for the Scarborough Formation exposed at the Scarborough Bluffs. It refines the reconstruction of past regional vegetation proposed by Terasmae (1960) and Westgate et al. (1999).
54In this paper, we maintain the use of two distinct units. The Don Formation consists fondamentally of discontinuous seasonal or event beds with irregular thickness and no lateral continuity. The continuous allozones are distinct and separated in most cases by an erosional discontinuity. Conversely, the Scarborough Formation has all the lithological characters of lake bottom, prodelta and delta sediments deposited in a glacial lake, with continuous lower rhythmites, the classical gradient of grain size within the formation and channeling structures in the upper beds (Kelly & Martini, 1986). These lithic differences are sufficient to consider two distinct units, according to the INQUA stratigraphic guide (INQUA 2015).
55The first sand beds of the Scarborough Formation are conformable on top of the Don Formation (fig. 3, 4 & 6). The main erosional discontinuities in the upper part of the Don Formation are observed at the base of the coarse beds D5a and D5c, over Allozone D4 (fig. 7). The origin of the green shale fragments of zones D5a and D5c is attributed to erosional processes at the bottom of streams or on exposed slopes of the local Ordovician shale (Georgian Bay Formation), and to seasonal fluvial transportation and deposition in shallow waters (tab. 3). Seasonal ice on the shore could explain the D5a and D5c pavements. A low level of Lake Coleman is unavoidable during the deposition of these beds, with probably some channels on the shoreface. In the watershed, frost action on slopes with a Northern Boreal Forest plant cover and spring thaw floods are inferred.
56The rusty facies of the uppermost sand bed (Allozone D5g) is likely related to underground waters. With no distinctive soil horizons at the top or within the upper Allozone D5, the hypothesis of exposures of the Don Formation is not yet demonstrated, and the duration of the intraformational erosional episodes is not known. The beds of Allozone D5.2 & 3 are decarbonated and altered with a very low concentration of pollen. Tentatively, two types of processes may have altered these beds. In the case of transitional exposures of the formation, weathering and early stages of leaching can occur rapidly, in several centuries on sands. Without exposure, underground water flows between the silty beds of Allozone D4 and the overlying clayey rhythmites of the Scarborough Formation, during the Wisconsinan and Holocene time, could explain the altered appearance of the upper beds of the Don Formation and the decarbonated Allozones D4 and D5. In both cases, long hiatus in the order of one to several MIS 5 substages during the deposition of Allozone D5 are not necessarily involved but could have occurred (tab. 1).
57The insertion of the published biotic data into the proposed allostratigraphic subdivisions (tab. 4) needed a review of the sampling logs co-authored by Karrow or a careful reading of the other studies, since the position of the samples was indicated only by the depth on the sampling log or core.
58The lowermost Cladoceran boreal and subarctic assemblages correspond to the glacial-interglacial transition. The bio-member related to Allozone D2 contains irrefutable evidence of interglacial climatic conditions as warm or warmer than today. Within the complete sequence of the Don Formation, the discontinuity D2-D3 and Allozone D3 can be stratigraphically positioned from the major change of biota between Allozones D2 and D4. In Allozone D4, the distribution of the biotic remains, pollen grains excluded, is poor, not homogenous, and most of the fossils indicate a climatic cooling or at least an ecology compatible with a fir-dominated forest. The heterogeneous upper Allozone D5 contains rare plant macrofossils (uppermost samples of Kerr-Lawson, 1985) which correspond to a taiga and subarctic conditions (tab. 3 & 4).
59The published biotic analyses lead also to a fair reconstruction of the past environmental conditions. The fluvial system of the Laurentian River Valley was maintained during the interglacial and the following early cold episode. The biotic changes sustain the model of sedimentary processes with a climatic forcing.
60From the present study, the sedimentary processes invoked by Eyles & Clark (1988) are to be extended. Instead of a simple continuous shoreface deposition depending on water depth, distality from sand source, storms and clearweather, the deposits show sedimentary discontinuities and are the product of three main factors: fluvial load or debris input, fluvial and shoreface dynamics, and lake level variations. The fluvial load is linked to the vegetation cover and weathering processes in the watershed, in covariance with the climate. The fluvial dynamics depend on the hydric regime in covariance with the climate, at the origin of flows, floods and low waters, and on moving river channels of the distal ancestral river. The lake level controls the potential depth of deposition and the accommodation space (see Brookfield & Martini, 1999), the distance from the mouth of the river, and the shoreface processes.
61The lateral continuity of the allozones is in favour of distinct phases of shoreface sedimentation near the mouth of the ancestral river (tab. 3). A finer load occurs during phases of stable vegetation and climate on the watershed, and/or corresponds to a distal source of debris and/or to deeper waters. These conditions can be applied to the sequence of beds of Allozone D2d, and to most Allozone D4 with the highest lake level. A coarser load can be related either to shallow waters and coarse supply, as lithozones D2b, D2c, or to a coarser load, as in the Allozone D4c with boreal forest and deeper waters.
62In most of the allozones, the inner discontinuities are related to autocyclic variations (Beerbower, 1964; Blaine, 2003) due to the processes of a river-lake sedimentary system. Conversely, the main discontinuities related to the gravel lenses D2b, to the D2-D3 discontinuity, and to the coarse beds of Allozone D5 correspond to allocyclic variations with external forcing (climate change and continental consequences) on the Lake Coleman-Laurentian River sedimentary system. The gravel lenses D2b seem to correspond to a single short erosional episode or a series of flood and/or storm waves episodes within shallow waters. The cross-bedded coarse sands of the upper part of D2c (fig. 6) are related to a cool event. The phase related to D3 is a major episode, confirmed by a significant gap in the pollen diagram. The visible coarse beds D3 are laid down at the end of the erosional phase. They record ultimate major storm waves or floods of a long erosional phase. Most of the D5 phase corresponds to a drastic drop of the level of Lake Coleman as well as new continental processes on the watershed and seasonal ice processes on the shores of Lake Coleman (tab. 3).
63In summary, the debris supplied by the ancestral river are spread at the bottom of the shoreface and partly reworked within the accommodation space at the margin of Lake Coleman. As a result, the bottom tends to be roughly levelled at the scale of the shoreface, with bottom microforms inferred from the sedimentary structures described by Eyles & Clark (1988), some moving shore channels, variable bevelled discontinuities and a gentle irregular slope to deep waters. The relative weight of the variables is changing through time (tab. 3), at the origin of distinct phases of deposition and changes of the accommodation space. These processes are compatible with the depositional environments reconstructed by Willams & Morgan (1977) and biologists to whom they refer to: "studies reveal an unstable depositional environment with fluctuating water levels such as those seen in certain littoral areas in the lower Great Lakes today."
Tab. 4: Correlation of the published biotic assemblages with the allozones of the Don Formation.
64The stratigraphic reassessment of the Don Formation now allows closer correlations (tab. 5). The main reconstructed system-tracks and sedimentary settings are the following.
65The Illinoian Ice sheet had deeply eroded the previous Quaternary units and deposited a till cover (York Till vs Rigaud, Becancour, Odanak, Portneuf and Baie-Saint-Paul tills, referred to respectively by Karrow, 1967; Anderson et al., 1990; Gadd, 1971; Lamothe, 1989; Bernier & Occhietti, 1991; Occhietti et al., 1995). The St. Lawrence Lowlands, depressed by glacio-isostatic subsidence, were rapidly filled in by unnamed post-Illinoian rhythmites (Lamothe, 1989; Occhietti & Clet, 1989; Besré & Occhietti, 1990) and by the lower clayey beds of the Ile aux Coudres Formation (Occhietti et al., 1995). With the Late Wisconsinan Champlain Sea analog (Richard & Occhietti, 2005), this transitional sedimentary phase would have lasted about 2000 years.
66During the climatic optimum, the Laurentian River built a large sedimentary body at the margin of Lake Coleman. At the same time, a major fluvial plain (Clet-Pellerin & Occhietti, 2000) is deposited in the St. Lawrence River Valley (deducted from Anderson et al., 1990) and middle Estuary (Clet & Occhietti, 1995), up 4 to 6 m metres above the present sea level. By analogy with the Holocene fluvial system, the stepped plain is likely incised in post-Illinoian clays and sandy marine terraces. The delta of the interglacial ancestor of the St. Lawrence River reached at least the Ile aux Coudres area (Ile aux Coudres Formation, Occhietti et al., 1995)(fig. 2). From the elevation of the overlying deposits, most of the channels of the middle estuary, between the Ile d’Orléans and the Saguenay River mouth (fig. 2), were filled in by prodelta and delta sediments.
67Cooling conditions in the Toronto area means a severe cooling over the Québec-Labrador northern and central area. For the St. Lawrence sedimentary system, it means a lowering sea level and strong regressive erosion. The erosion phase is stratigraphically recorded (tab. 5) at the Pointe-Fortune section (Anderson et al., 1990), in the central St. Lawrence Valley (Lamothe, 1989; Ferland & Occhietti, 1990) and at the base of the Ile aux Coudres sections (Clet & Occhietti, 1995).
68Allozone D4 can be correlated with the Lotbinière Sand characterized by a pollen content of boreal forest (Clet & Occhietti, 1994). The higher level of the Lake Coleman basin related to D4 is probably linked to isostatic readjustments at a sub-continental scale to early ice accumulation over northeastern Canadian Arctic and the Québec-Labrador area. This episode could have lasted 5 ka in the hypothesis of a long substage MIS 5e transition and much longer if it corresponds to one of the mild substages MIS 5c or 5a (tab. 1 & 5).
69The stony beds of Allozone D5 record cold but non-glacial conditions. The drop of lake level represents either drier conditions and/or isostatic sub-continental movements. This episode of heterogeneous sedimentation is related to a phase of glacierisation over Québec-Labrador, preceding the first glacial invasion of the central St. Lawrence River Valley. At least 3800 varves were deposited in glacial Lake Deschaillons (Besré & Occhietti, 1990; Hillaire-Marcel & Pagé, 1981) before this first glacial invasion. Allozone D5 deposited at the margin of Lake Coleman is likely contemporaneous to these glaciolacustrine deposits. The uppermost sandy bed D5g represents the transition to the following inundation by Lake Scarborough.
Tab. 5: Correlation of the interglacial and early glacial units of the Toronto and St. Lawrence River Valley areas
70The Levrard Till records the first post-Sangamonian glacial event in the central St. Lawrence River Valley (Lamothe, 1989). The ice dammed the main outlet of the Lake Ontario basin, at the origin of the Lake Scarborough episode (Karrow, 1967). The probable position of the ice front in the Lake Ontario basin area is drawn by Kelly and Martini (1986).
71The discontinuity and lower beds of the Pottery Road Formation are tentatively correlated to the non-glacial subsequence of the central St. Lawrence River Valley (Clet & Occhietti, 1996) including the La Pérade marine Clay, an erosional discontinuity, the St. Pierre Sediments, the Saint-Maurice Rhythmites and the lower part of the Vieilles-Forges Sands. The coarse upper beds of the Pottery Road Formation (fig. 4), probably deposited by distal melt-waters, correspond to the proglacial upper beds of the Vieilles-Forges Sands.
72From the biotic content, Palynozone Don 2.3 of the Don Formation is firmly related to the world wide climatic optimum of MIS 5e substage, dated circa 125 ka. The interglacial units of two other groups of sites in Eastern Canada (fig. 2) gave absolute Th/U ages compatible with MIS 5e, in the James Bay Lowlands (Allard et al., 2012: River Nottaway) in agreement with amino acids ages (Andrews et al., 1983) and in Cape Breton (de Vernal et al., 1986: East Bay). The luminescence dates of Berger & Eyles (1994) from the beds of the Don Formation related to the interglacial climatic optimum are therefore undoubtedly too young by about 40 ka, because of the lack of fading correction (see Lamothe & Auclair, 1997; Huntley & Lamothe, 2001). This is confirmed by preliminary TL results (Lamothe et al., 1998), referred to in Karrow et al. (2000).
73Minimum ages of the uppermost Allozone D5 and the related stadial episode are given by the 79.8 ± 1.4 Th/U age of the Deschaillons Varves and the 98 ± 8 ka luminescence age of the marine La Pérade Clay deposited after the Levrard Till glacial event (Occhietti et al., 1996). The Th/U age of concretions of the Deschaillons Varves (Hillaire-Marcel & Causse, 1989) gives the age of the post-deposition process of CaCO3 precipitation, and a minimum absolute age to the last varves which were deposited before the glacial event in the central St. Lawrence River Valley and the correlative glacial Lake Scarborough. Therefore, the early glacial dam downstream the central St. Lawrence River Valley (Lake Deschaillons) and the correlated late phase of Lake Coleman (Allozone D5) are older than 80 ka and occurred during stage MIS 5, either MIS 5d or MIS 5b, as well as the first Upper Pleistocene glacial invasion of the central St. Lawrence River Valley and the Lake Scarborough flooding (tab. 1).
74The accuracy of the pollen content of the shoreface beds of the Don Formation is poorer than that of lacustrine or peat deposits. The deposition of the formation lasted at least 20,000 years ("minimum" hypothesis D, tab. 1) for less than 8 meters of sediments, with irregular rates of sedimentation and several discontinuities. Moreover, the sampled beds contain an average proportion of pollen grains and spores mostly transported by the Laurentian River, deposited and reworked for one season to several years at the bottom of the shoreface of Lake Coleman. For all these reasons, the pollen content of the formation indicates the climatic trends that prevailed on the regional watershed, but a complex transfer function would be necessary to infer precisely the temperature changes.
75Being aware of those limitations, the new pollen diagram of the Don Formation, together with the allostratigraphy, is compared to the detailed charts of central and southern Europe (Guyot et al., 1993), of the Iberian margin (Sánchez Goñi et al., 1999) and Greenland (NGRIP, 2004) (tab. 1), and to some events in Northern Europe. With the large uncertainties related to both the available ages and the duration of the erosional discontinuities and cooler phases revealed by this study, several hypotheses on the ages of the pollen zones can presently be retained (tab. 1 & 5; fig. 10).
76From Palynozone Don 2.2, a cool and short event is clearly recorded within the interglacial warmth, shortly after an apparent drop in the level of Lake Coleman (Allozone D2b). The maximum warmth of the Sangamonian Interglacial (Palynozone Don 2.3) follows this event. The same pattern is observed in northern Finland about the Tunturi event (Helmens et al., 2015). Without numerical age, the correlation of Palynozone Don 2.2 with the Tunturi event and the inferred event in Norwegian Sea circa 120 ka (see Fig. 3 in Helmens et al., 2015) remains tentative, as well as with some higher δ18O values observed in the MIS 5 beds of marine cores referred to in Wohlfarth (2013). The same uncertainties can be applied to Palynozone Don 2.4.
77The significance of Palynozone Don 2.4, after the warm maximum of the interglacial, is puzzling. The pollen content indicates clearly a cooling event, from the conditions of a harsher Mixed Hardwood Forest (MWc) to those of a Southern Boreal Forest (SBF). Presently, the fir forest biome, similar to the SBF, extends 500 km North of Toronto, according to the ecoclimatic map of Canada (Strong et al., 1989). The cross-bedded sands of the related lithologic unit (upper beds of D2c) are devoid of silty bed for 35 cm above the sample with the fir forest assemblage. What was the type of vegetation during the deposition of these sands? What is the duration of the full event? The event corresponds either to the cold stadial MIS 5d-Melisey I (hypothese C on tab. 1) or to the group Melisey I, St. Germain 1a and Montaigu (hypothese A), or to a shorter cool event within the interglacial, possibly related to one of the Greenland Stadials (hypotheses B, D & E). The short interglacial event hypothesis is supported by the previous biotic studies (tab. 4).
78The D2-D3 erosional discontinuity and Palynozone Don 3 could represent another short cool event within the climatic optimum, followed by the progressive transition of the end of MIS 5e until circa 115-112 ka (hypotheses D & E), or conversely be the equivalent of the first cold stadial related to the group Melisey I, St. Germain 1a and Montaigu (hypothese B) or to the Montaigu event (hypothese C).
79Palynozone Don 4 can be related either to the transitional end of the interglacial (hypotheses D & E) or to one of the two mild interstadials Saint-Germain Ic (hypotheses B & C) or II (hypothese A).
80With the coarse deposits at the base of Allozone D5 and a Northern Boreal Forest pollen content, the Palynozone Don 5 could be equivalent to one of the Stadials Melisey I (hypothese D of an almost continuous unit) or Melisey II (hypotheses B & C), or could represent a poorly recorded cooler long episode (hypothese E). The lower fluvial beds of the Pottery Road Formation, inset into the Don and Scarborough Formations, are apparently the only indicators of another mild episode prior to the Holocene. From the available ages referred to in chapter 5.6, they would correspond to MIS 5a (hypotheses B, C & E), but they could represent a long erosional episode (MIS 5c to MIS 5a).
81In summary, direct correlation of the Don Formation allozones and palynozones to units of Europe is equivocal. The hypothesis A (tab. 1) of a “long” Don Formation (70,000 years) seems to be the less valuable, because of the lack of fossil evidence of cold stadial conditions in Allozone D2 (tab. 4). Given the ages presently available, of the order of 80 ka, the hypotheses B, C and E of an “intermediate” Don Formation (40,000 years) would be the most relevant, but these hypotheses are not based on direct ages of the upper part of the Don Formation. The hypothesis D of a "short" Don Formation (20,000 years) implies an almost continuous sedimentation on the northeastern shore of Lake Coleman and an extensive early glaciation in Northeastern Canada. Extra-regional indicators support this hypothesis: for example, the drop of sea level during MIS 5d and "the most positive benthic 18O values" implies "that the Laurentide ice sheet had grown considerably." (Shackleton et al., 2003), or the ice sheet limits in Scandinavia (Mangerud, 2004). Again, direct ages on the Don Formation are necessary to retain this hypothesis.
Fig. 10: Contribution to the time event regional stratigraphy of the northern and eastern Great Lakes area: four hypotheses of time-distance diagram of the diachronic events of the Last Interglacial and early Wisconsinan.
a and b: Hypotheses adapted from Karrow et al. (2000), with a long (hypothesis a) or short (hypothesis b) hiatus inferred between the Don and Scarborough Formations. The chronological frame is not revised. c and d: The chronology of the MIS 4 is updated from Sanchez et al. (2005). Hypothesis c: Long discontinuities within Allozone D5 and unstable interglacial Sangamon Episode. Hypothesis d: Palynozones Don 3 and 4 correspond to unnamed phases of the Wisconsin Episode (MIS 5d and 5c).
82The time event classification in the Southern and Eastern Great Lakes area proposed by Karrow et al. (2000) (fig. 10a & b) is based on the units of the Toronto area for the Sangamonian interglacial and the early Wisconsinan episodes. The Don Formation is related to the Sangamon Event (MIS 5e), the Scarborough Formation to the Greenwood Phase (MIS 5b or d), and the fluvial beds of the Pottery Road Formation to the Willowvale Phase (MIS 5a or 5abc). The overlying Sunnybrook Drift is the reference unit for the Guildwood Phase and "probably corresponds to OIS 4" (Karrow et al., 2000). These authors mentioned that new findings could foster a revision of their time event classification.
83From the reassessed bio- and allostratigraphy of the Don Formation, two changes are proposed regarding the two hypotheses of Karrow et al. (2000), the Allozone D5/Palynozone Don 5 represents the onset of the Greenwood cold Phase (fig. 10 a & b), and the limits of the Guildwood Phase should be updated to the MIS 4 chronology (Sánchez Goñi et al., 2005). In the absence of finite ages, the Allozones D2 to D4 of the Don Formation could also correspond either to the interglacial stage ("plateau" of MIS 5e) with three climatic coolings (Don 2.2, Don 2.4 and Don 3) before the transitional end (Don 4) (fig. 10c), or to the interglacial stage (MIS 5e - D2/Don 2) and two additional unnamed phases (MIS 5d - D3/Don 3 and MIS 5c - Don 4) that would represent the earliest phases of the Wisconsin Episode (fig. 10d), prior to the Greenwood Phase materialized by Allozone D5/Palynozone Don 5 and the Scarborough Formation.
84Thus, the study of the Don Formation reveals one to several local minor fluctuations during the interglacial. This relative instability is recorded in ice cores of Greenland (Dansgaard et al., 1993; Johnsen et al., 2001; NorthGRIP Members, 2004; Barker et al., 2011; NEEM, 2013), in continental sequences (Clear Lake, Adam et al.,1981); and in the Iberian margin (Shackleton et al., 2003).
85Due to contradictory elements about the lowermost clayey Allozone D1, the hypothesis of a transitional Illinois/Sangamon lacustrine phase based on the cold cladoceran fauna (Don 1) is only tentative (tab. 3 to 5).
86With the approximate time scales of hypotheses B to E (tab. 1), a tentative reconstruction of the variations of the relative level of Lakes Coleman and Scarborough, in the Don Valley Brickyard and Scarborough Bluffs areas respectively, is summarized in figure 11. The trends are established from different authors (Terasmae, 1960; Karrow, 1969; Kelly & Martini, 1986; Eyles & Clark, 1988) and the present study. Direct evidence of secondary fluctuations of the level of Lake Scarborough seems to be difficult to infer from the data (see Brookfield & Martini, 1999).
87At least two episodes of low level, during the D2-D3 discontinuity and the deposition of D5a and D5c, are indicated by coarser deposits, erosional features and, for the D2-D3 discontinuity, by a lag in the pollen record. Both are correlated to climatic change. Their amplitude is unknown. A minor drop is inferred from the coarser beds D2b. According to Terasmae (1960) and Eyles & Clark (1988), the maximum depth registered at the Don Brickyard during the D4 deposition reached 18 m above the bottom of the shore. With a thickness of sediments of 3 m, this is the equivalent to a high level of 37 m above the current level of Lake Ontario. At least one minor variation or incursion of fluvial waters (tab. 4) is indicated by the biotic content (Poplawski & Karrow, 1981) during the D2 episode, which would correspond to the Don 2.4 episode. From a peak of Picea cf. mariana pollen, the fine-sand phase recorded during the D4 sedimentary episode is probably linked to an opening of the Southern Boreal Forest, but could correspond to drier conditions on the watershed and/or to a change of the bathymetry.
88The Lake Scarborough invasion is rapid. A minimal relative level is inferred from the top of the delta deposits. Post-Scarborough erosional valleys, filled in by the Pottery Formation, indicate a drastic drop in the level of the former Lake Ontario.
89From these reconstructions, the presupposition of a continuous deepening of the bathymetry on the shore of Lake Coleman taken by Eyles & Clark (1988) and Eyles & Williams (1992) during the Don Formation episode, as well as the apparent long discontinuity between the Don and the Scarborough Formations inferred by Terasmae (1960) and used later by most of the authors are to be abandoned.
90The lake levels in the Great Lakes area depend on several factors, as developed on the post-glacial lakes evolution by Sly and Prior (1984), Kelly & Martini (1986), Pair et al. (1988), Clark et al. (1994), Coakley & Karrow (1994), and Anderson & Lewis (2012). These factors include: - regional isostatic equilibration and tilting, - elevation of the outlets with or without ice damming, -glacio-isostatic response to remote continental ice over Québec-Labrador, the St. Lawrence Valley and Hudson Bay, - local isostatic response to the accumulation of major sedimentary bodies, - and hydrological regime. At this stage of the study, the level of Lake Coleman depends mainly on global isostatic movements. The depth of the sill at the outlet of Lake Coleman in the upper St. Lawrence Valley may have been higher than today (see the study during the Champlain Sea episode by Pair et al., 1988), but this explains neither the slow rising trend of the level, nor the different level drops during the Lake Coleman episode.
91In comparison with the present small Humber and Don Rivers which flow through the Toronto vicinity (fig. 5), the interglacial and transitional Laurentian River drained a larger watershed (White and Karrow, 1971; Eyles, 1987). The hypothesis of being a second outlet of Lakes Huron-Michigan is therefore logical. Presently, the buried sill between the Georgian Bay of Lake Huron and Lake Ontario is comprised between 120 and 150 m a.s.l. (Fig. 4 in White & Karrow, 1971) or close to 120 m a.s.l. (Fig. 1.6 in Westgate et al., 1999). During the Sangamonian and early transitional phase, the gradient between the interglacial-transitional Lake Huron and Lake Coleman was therefore probably less than what it would be today (close to 102 m) for two reasons: the sill was probably less buried than today after the very erosive Illinoian glaciation, with the consequence of a lower water level of the interglacial Lakes Huron-Michigan (between 176 and 120 m a.s.l.), and the water level of Lake Coleman was higher (93 m + 2 to 18 m a.s.l.) than the water level of Lake Ontario today (74,2 m a.s.l.). The Laurentian River was a shortcut for the drainage system of the Great Lakes (fig. 9).
92When the St. Lawrence outlet of Lake Scarborough was dammed by the expansion of the Québec-Labrador ice dome in the central St. Lawrence River Valley, the lacustrine waters overflowed along the Mohawk River Valley toward the Hudson River (Karrow, 1969), through the natural outlet located at Rome (New York State). A differential trend of the global glacio-isostatic response in the Lake Scarborough basin to the advancing and retreating Québec-Labrador glacier can be inferred but is hard to quantify (see Brookfield & Martini, 1999).
93The exposures of the Don Valley Brickyard and Scarborough Bluffs are located back from the ultimate delta front in glacial Lake Scarborough. The observed gullies and valleys, related to the Pottery Road Formation, incised into the delta and underlying Don Formation deposits are similar to a regressive fluvial system, fed either by melt waters from a remote retreating ice margin (Sharpe & Barnett, 1985) or by a non-glacial drainage system (Karrow, 1990), or by the succession of the two. In both cases, a drastic drop of the level of Lake Scarborough is inferred, related to the ice retreat and reopening of the St. Lawrence River outlet.
Fig. 11: Inferred changes in bathymetry on the north shore of the Lake Ontario basin, in the Toronto area, from the Illinoian/Sangamonian transitional episode to the interstadial erosional episode (MIS 5 cba or a?) which follows the early glacial event (MIS 5d or b) in the central St. Lawrence River Valley and the Lake Scarborough episode. Lake Scarborough lasted some thousand years, with the hypothesis of a continuous event. The duration of the pre-Pottery Road erosional episode is unknown
94A comparison with other transitional sequences of non-tropical North America is tentatively done in order to date the different allozones of the Don Formation and the overlying units. Within the domain covered by the Wisconsinan Laurentide Ice Sheet in Northeastern North America, most of the units and beds with biotic markers indicating climatic conditions as warm or warmer than today are located on figure 2. These units were deposited or trapped in various settings: peat bogs in situ or collapsed in gypsum karstic hole, fluvial deposits with reworked plant markers, and seashore deposits (Skinner, 1973; Brookes et al., 1982; Mott, 1990; Mott & Dilabio, 1990; see Richard et al., 1999). The associated sequences have in common an erosional disconformity over the interglacial warm and sometimes early transitional units. The overlying units correspond either to cool interstadial conditions, with a return to a boreal forest biome or a milder fir forest (Cape Breton, Fréchette & de Vernal, 2013), or to a glacial unit. Up to now, only one interstadial event with mild or cool conditions is recorded between the interglacial climatic optimum ("plateau" of the MIS 5e) and the onset of the main glacial stage considered commonly as the equivalent of MIS 4.
95Outside the domain of the last Laurentide Ice sheet, continuous sequences have been recognized (fig. 1). The pollen sequence from Clear Lake, California, is the only continuous pollen record obtained in North America which shows a response to climatic changes as detailed as the records in Europe for the last 130,000 years (Adam et al. 1981). The pollen preserved in the lake sediment comes from a large surrounding area devoid of glaciers. The pollen diagram, based mainly on the representation of Quercus, Pinus and a TCT group including Taxodiaceae, Cupressaceae and Taxaceae, is subdivided in pollen zones very similar to the Grande Pile pollen zones in France (Woillard, 1978). The discontinuities observed within the pollen diagram of the Don Formation preclude, for now, a valid correlation with the Clear Lake sequence.
96In south-central Illinois, several authors described an interglacial vegetation which changed from a mesic deciduous forest to a prairie and back to a deciduous forest. Teed (2000) compared her results in Pittsburg Basin to previous works and datings on several local basins (Grüger,1972; King & Saunders, 1986; Curry & Baker, 2000; Zhu & Baker, 1995) and ascribed the "Sangamonian" pollen zones PB-2, PB-3 and PB-4 to MIS 5e. The subsequent pollen zone PB-5 corresponds to a prairie and indicates drier conditions correlated to substages MIS 5d to 5a. Compared to the Don Formation sequence, the pollen record of Illinois is related to variations of dryness and humidity with a poor climatic accuracy. The correlation between the units of the two areas is equivocal.
97From isotopic and other proxy data, a core of the Owens Basin, California, shows that the lake overflowed during wet/cold conditions of MIS 6, 5b and 4, partly overflowed during MIS 5d and was closed during dry conditions during MIS 5e, c and a (Li et al., 2004). The dry conditions corresponds to warm periods dated circa 126, 102 and 80 ka on the speleothems of Devils Hole, Nevada (Winograd et al., 1992). The chronological anomaly on the timing of the last interglacial measured at this site is now resolved, with a fair time coincidence of the ages from local calcite with the global boreal summer insolation and marine record (Moseley et al., 2016).
98The pollen content (Anderson et al., 2014) and multi-proxy data from the high-elevation (2705 m a.s.l.) Ziegler Reservoir site (Snowmastodon site), Colorado, suggest both good correlations with MIS 6 and 5e, and unexpected local climatic trends during MIS 5d and b. Other papers emphasize the synchroneity of climatic changes in continental United States with the oceanic isotope phases, either from speleothems in South Dakota (Reed's Cave; Serefiddin et al., 2004) and Nevada (Goshute Cave; Denniston et al., 2007), or from cores in western Washington State (Hamptulipe mire; Heusser & Heusser, 1990), in Oregon (Carp Lake; Whitlock & Bartlein, 1997), and in Utah-Idaho (Bear Lake; Jiménez-Moreno et al., 2007). Compared to these proxy data, the continental record of the MIS 5d-a period in the Lake Ontario-St. Lawrence River Valley area is discontinuous and does not allow a fair correlation with the episodes recorded in Europe and Greenland. This confirms the statements by Otvos (2015) that ''Delineation of the boundary between the Last Interglacial (LIG) and the last (Wisconsinan) Glacial Stage in North America represents a critical, yet unresolved issue.'' and ''In the absence of adequate pollen-stratigraphic documentation, Pleistocene subdivision boundaries were harder to establish in North America than in Europe.''
99From this study, the Don Formation is heterogeneous, comprises inner major erosional discontinuities and five distinct planar allozones, and is overlain conformably by the Scarborough Formation. The continuous allozones correspond to stacked discontinuous beds, with dominant sedimentological characters and with distinct pollen assemblages. The thicknesses of the allozones and subdivisions are basically variable at the scale of the Don Formation. The debris supplied by the ancestral Laurentian River were spread and partly reworked at the bottom of the shoreface and littoral which tended to be continuously levelled with a gentle slope.
100The new pollen and spore analysis, coupled to the allostratigraphy and published data, extends the climatic signal of the Don Formation from the interglacial optimum to one of the two early cold Wisconsinan stadials. It reveals the occurrence of cool events during the interglacial episode and a major discontinuity. Several hypotheses can be considered about the age and duration of these events. From the available data, the tentative climatic sequence is as follows:
101- The climatic optimum of the Sangamonian Interglacial ("plateau" of MIS 5e) is represented by a rich biota identified since the 19th Century, by Allozone D2 and a pollen and spore content of Mixed Hardwood and Deciduous Forests. The interglacial event can be subdivided into: 1) an early phase as warm as today with a zonal Mixed Hardwood Forest with Quercus, Caria, Tilia, 2) a cool event characterized by the pollen content of a Southern Boreal Forest with Pinus cf. banksiana and Picea cf. mariana; this event could correspond to the Tunturi event recorded in Finland, 3) the climatic maximum warmth with a Deciduous Forest including Liquidambar and Nyssa, 4) a cool event with a Southern Boreal Forest assemblage that could be related to a Greenland Stadial, and 5) a return to zonal Mixed Hardwood Forest conditions.
102- A transitional part of the end of the interglacial is missing, as indicated by an erosional episode (the D2-D3 discontinuity), the coarse beds of Allozone D3, and a gap in the vegetation record. This interval is equivocal, as either a relatively short cold event within the Sangamonian Interglacial (end of MIS 5e) or more likely the early stadial related to MIS 5d, which would correspond to the Melisey I episode (or both Melisey I and Montaigu episodes).
103- A younger mild climatic episode is inferred from the pollen content of Allozone D4 deposited in deeper waters. Palynozone Don 4 is characterized by a rapid improvement to zonal Southern Boreal Forest assemblages and then, by a decrease of deciduous trees and Abies balsamea. A peak of Picea cf. mariana indicates the extent of a harsher Southern Boreal Forest. The fir forest mild episode was interpreted as the transitional end of the last interglacial; the alternative hypothesis of an interstadial phase, equivalent of MIS 5c (the St. Germain I phase) can also be inferred.
104- One or two erosional phases are related to a drop in the lake level and the display of coarse clasts at the beginning of the last climatic episode recorded by the Don Formation (Allozone D5 and Palynozone Don 5). Coeval boreal to sub-arctic conditions, with Northern Boreal Forest assemblages which evolve to Forest Tundra assemblages, prevailed on the Laurentian River watershed. This episode is attributed to the first part of a stadial, either the early part of MIS 5d (Melisey I-Montaigu) or of MIS 5b (Melisey II). The second part of this stadial corresponds to the closure of the outlet of the Lake Ontario basin by an extent of the continental ice in the St. Lawrence River Valley, at the origin of glacial Lake Scarborough and its related formation.
105Based on the reassessed litho- and biostratigraphy and system tracks, a finer correlation can be established with the units of the St. Lawrence River Valley. The uppermost Allozone D5 of the Don Formation is correlated to the older than 80 U/Th ka Deschaillons Varves which were deposited in the St. Lawrence River central Valley prior to the first regional Late Pleistocene glacial invasion and Lake Scarborough episode. The Allozone D5 would be also older than the 98 ka ± 8 TL ka isostatic Cartier Sea invasion which followed the deglaciation of the St. Lawrence River Valley. From these two ages, the cold event recorded by the uppermost beds of the Don Formation occurred during one of the two cold substages of MIS 5, as well as the subsequent glacial Lake Scarborough event.
106The subdivisions of the Don Formation represent a solid reference frame for further dating programs which would allow stronger correlations to other interglacial, transitional and early stadial sequences of North America, Greenland and Europe, resolving the question of the age of the Late Pleistocene glacial inception in Northeastern North America.