Article body

Introduction

The Cypress Hills of southeastern Alberta and southwestern Saskatchewan are an area of unique physiographic and ecological diversity within the Canadian prairie ecozone. The resulting abundance and diversity of resources has attracted numerous aboriginal groups to this region throughout much of the Holocene. The location of archeological sites within the Cypress Hills is ultimately dependent on factors affecting site selection and, subsequently, to processes of site formation and preservation, each of which are inherently linked to geomorphic processes and climatic conditions.

Current paleoenvironmental research in the Cypress Hills is focused on the area around Elkwater Lake (Fig. 1). The Stampede archeological site (DjOn-26) is located approximately 750 m from the eastern shore of Elkwater Lake on the proximal end of an alluvial fan and adjacent to a deeply incised channel descending from the northern flank of the Cypress Hills (Fig. 2). The distal end of the fan extends northwest across the Elkwater Channel.

It is hypothesized that initial and subsequent occupations of the Stampede site were a function of its proximity to Elkwater Lake at different periods in the Holocene and that other sites may exist in similar nearshore environments elsewhere in the study area. Furthermore, variations in the rate of deposition and textural composition of sediments deposited at the site are hypothesized to be a function of one or more of the following factors: 1) adjustment of the fluvial geomorphic system to changing climatic conditions; 2) change in the position of the site on the fan as it formed; and 3) changing loci of deposition on the fan controlled by variations in the distance to the shoreline of Elkwater Lake. These hypotheses can only be tested by establishing when the lake formed and how the level and extent of the lake has varied throughout the Holocene.

Objectives

The objectives of this research are to: 1) evaluate the viability of the slump-dammed lake hypothesis as a possible explanation for the formation of Elkwater Lake and establish the age of lake formation; 2) evaluate the viability of alternative outlets from Elkwater Lake northward through the Green Lake moraine; 3) determine the probable elevation and extent of the lake at different periods throughout the Holocene; and 4) reconstruct paleoenvironmental conditions at Elkwater Lake based on these results.

Figure 1

Cypress Hills region and Elkwater Lake study area.

Localisation de la région des Cypress Hills et de l’Ellkwater Lake.

-> See the list of figures

Figure 2

Elkwater Lake study area map showing location of the East Slump Block (ESB), West Slump Block (WSB), Stampede site, core extracted by Vreeken (1986), alluvial fan, and Ross Creek outlet. Orthorectified photo-mosaic created from conventional aerial photography courtesy Alberta Environment, Airphoto Services.

Région de l’Elkwater Lake montrant la localisation du East Slump Block (ESB), du West Slump Block (WSB), du site Stampede, de la carotte extraite par Vreeken (1986), du cône de déjection et de l’exutoire de Ross Creek. Photo-mosaïque créée à partir de photographies aériennes conventionnelles (Alberta Environment, Airphoto Services).

-> See the list of figures

The Cypress Hills

Geologic Description

The Cypress Hills are located in southeastern Alberta and southwestern Saskatchewan (Fig. 1), extending approximately 130 km east-west and 25 to 40 km north-south (Jungerius, 1966). Two north-south trending valleys divide the Cypress Hills into three units referred to as the East, Centre, and West Blocks.

The Cypress Hills are a bedrock-supported plateau capped by the Upper Eocene to Miocene age Cypress Hills formation, a resistant semi-arid braidplain deposit (Leckie and Cheel, 1989). The plateau rises to the west, attaining a maximum elevation of 1 465 m asl just southwest of Elkwater Lake, where it rises more than 500 m above the surrounding plains. Approximately 300 km2 of the West Block (above an elevation of 1 250 m asl) remained unglaciated throughout the Pleistocene, forming a nunatak rising 90 m above the surrounding Laurentide Ice Sheet (Stalker, 1965).

Approximately 20 000 BP, during the height of Late Wisconsinan glaciation in this area (Kulig, 1996), the Laurentide Ice Sheet abutted the northern flanks of the West Block, impounding several small proglacial lakes. These lakes drained to the west and east along the ice margin, cutting more than 60 m into the preglacial topography and forming the Elkwater and Battle Creek channels (Vreeken, 1990; Kulig, 1996). The Elkwater Channel (Fig. 2) was established during the Middle Creek Advance ca. 15 000 BP (Kulig, 1996) and carried meltwater westward to the Medicine Lodge Channel, which skirts the western perimeter of the Cypress Hills plateau, then turns to the southeast. Deglaciation of the area north of the West Block at the conclusion of the Middle Creek Advance (ca. 13 500 BP) resulted in the deposition of hummocky disintegration moraine and a series of recessional moraines including the Green Lake Moraine (Westgate, 1968; Vreeken, 1986; Kulig, 1996).

Ecological Diversity

The comparatively high ecological diversity of the area is the result of a combination of unique physiography and postglacial climate. The resultant variety and abundance of resources has attracted numerous aboriginal groups to the area throughout much of the Holocene (Bonnichsen and Baldwin, 1978).

Historically, mean annual temperatures are 2.8 °C lower, and mean annual rainfall 109 mm higher than in Medicine Hat located 75 km to the northwest, resulting in a cooler and wetter environment than the surrounding plains (Cowell, 1982).

Microclimatic effects as a result of variations in slope and aspect are also considerable. Vegetation is characterized by a transition from mixed short grass prairie at lower elevations and on south facing slopes, to Populus tremuloides (trembling aspen) and Picea glauca (white spruce) along the north facing escarpment, upland valleys, and moist wetlands, to Pinus contorta (lodgepole pine) and Festuca scabrella (fescue) on the comparatively drier upland plateau (Breitung, 1954; DeVries and Bird, 1968).

Holocene Geomorphology and Climate

Although abundant evidence of both fluvial erosion and mass wasting exists throughout the Cypress Hills, the dominant geomorphic process operating through much of the Holocene has been mass wasting via landsliding (Goulden and Sauchyn, 1986; Sauchyn and Lemmen, 1996). Sauchyn (1999) concluded that the frequency of landsliding is correlated to periods of cooler, wetter conditions. Goulden and Sauchyn (1986) suggest that this is the result of a combination of unique physiographic, hydrogeologic and lithologic conditions that favour slope instability during periods of persistent groundwater recharge as a result of excessive porewater pressure and low shear strength of bentonitic clays.

A reconstruction of Holocene climate and geomorphology based on pollen records and sedimentation rates (Sauchyn, 1990; Sauchyn and Sauchyn, 1991) suggests a period of slope stability occurred from 5000 to 7700 BP, during which fluvial and eolian processes dominated. Conversely, cooler and wetter conditions in the late Holocene resulted in increased slope instability and frequency of mass wasting (Goulden and Sauchyn, 1986).

With respect to the Elkwater Channel, this implies that conditions most favourable to frequent massive landslides occurred during the Middle Creek Advance ca. 15 000 to 13 500 BP as a result of slope adjustment to channel formation, and again during the early and late Holocene as a result of slope instability during cooler and wetter periods.

The Stampede Site

The Stampede site is located on the proximal end of an alluvial fan debouching into the Elkwater Channel, east of Elkwater Lake. The site originally excavated by Gryba in 1971 (Gryba, 1972) consists of 24 buried A-horizons and 14 cultural layers in the upper 3.85 m (Gryba, 1975). Mazama ash, dated at ca. 6700 BP (Hallet et al., 1997), occurs at a depth of 3 m (1 234.5 m asl); a charred bison bone at 3.5 m depth (1 234 m asl) was dated at 7245 ± 255 BP (NMC-571).

This site was re-opened in 2000. Subsequent stratigraphic analyses have revealed no lacustrine or nearshore deposits above Mazama tephra suggesting that the lake has not risen above 1 234.5 m asl since 6700 BP. However, the C/N ratios of paleosols below this marker suggest near-saturated soils consistent with a nearshore wetland environment (Klassen, 2003).

Sedimentation rates based on dates from the aforementioned bison bone, Mazama tephra, and the estimated age of projectile points (G. Oetelaar, University of Calgary, personal communication, 2003) imply that a period of rapid sediment influx (ca. 12 mm a-1) occurred from 7245 to 5800 BP (Klassen, 2003). This is consistent with increased sedimentation rates at nearby Harris Lake from 6100 to 5300 BP (Sauchyn, 1990). Since 5800 BP sedimentation rates have decreased significantly, averaging 3 mm a-1, with a period of very low sediment influx between 5000 and 3500 BP, increasing gradually to the present day (Klassen, 2003). These observations fit well with Sauchyn’s (1990) model in which higher sedimentation rates are expected during the height of the Altithermal when reduced vegetative cover accelerates fluvial erosion. During these periods of increased aridity, extreme precipitation events resulted in an overall increase in erosion due to the lack of vegetative cover (Katz and Brown, 1992). Conversely, the sudden decrease in sediment influx at the site between 5000 and 3500 BP can be inferred to represent a return to cooler and wetter conditions, increased vegetative cover, and decreased erosion (Sauchyn, 1990).

Elkwater Lake

General Description

Elkwater Lake is fed by a number of intermittent streams flowing off the northern flank of the Cypress Hills and numerous underground springs. The lake has an area of approximately 2 km2, a perimeter of 11 km, and a volume of 8.0 M m3 (Bradford, 1990). The east and west arms of the lake parallel the Elkwater Channel, with the north arm extending toward the current outlet of Ross Creek (Fig. 2). Vreeken (1990) has suggested that this arm represents a preglacial valley deepened by a meltwater channel.

Lake levels have been controlled at the Ross Creek outlet since 1908. A drop inlet spillway, constructed in 1978, maintains current water levels at 1 226.5 m asl to a maximum depth of 8.4 m. The weir adjacent to the spillway is at 1 227.5 m asl, a height of 3.65 m above the original ground line (Bradford, 1990). However, Bradford (1990) notes that the Canadian Pacific Railway deepened the outlet prior to the installation of the original control structure to increase flow to their reservoir at Irvine, Alberta. Consequently, it is estimated that the natural uncontrolled elevation of this outlet (i.e. the minimum lake level required for it to convey water) prior to modification would have been roughly 1 225 m asl; assuming that the outlet was deepened by approximately 1 m. The elevation of the adjacent upland is up to 9 m above the current channel bed, suggesting that the outlet may have been as high as 1 234 m asl at the time it began to drain Elkwater Lake.

Formation of Elkwater Lake

Currently, very little is known about the formation and Holocene history of Elkwater Lake. It is generally accepted that the lake formed after a landslide on the north-facing escarpment of the plateau blocked drainage through the Elkwater channel to the west (Westgate et al., 1972; Vreeken, 1986, 1990). Figure 2 shows the location of two large slump blocks at the west end of Elkwater Lake that extend across the channel bottom and abut the opposite valley wall. These are referred to as the east slump block (ESB) and west slump block (WSB). Visual interpretation of stereo-photography and field reconnaissance suggests that the ESB is most likely older than the WSB based on the relative degree of weathering on the scarp wall, debris slope morphometry, and channel incision. However, assuming the slump-dammed lake hypothesis is correct, determination of the absolute age of the landslide creating each slump block is necessary in order to establish a maximum age of lake formation (i.e. the lake is no older than the landslide that impounded it).

Existing data suggest that the lake formed in the mid- to late Holocene. Terasmae and Mott (in Lowden et al., 1971) cored the lake to a depth of 3.8 m in 1968 and obtained a basal date of 5100 ± 280 BP (GSC-1101). However, Vreeken (1986) cored the toe of the alluvial fan on which the Stampede site is located (Fig. 2) and, based on the presence of Mazama tephra within a lacustrine sequence in this core, concluded that the lake formed prior to 6700 BP. In addition, he points out that the elevation of the lake at this time must have been somewhere between the subaqueous tephra in this core and the subaerially deposited tephra at the Stampede site. The elevations of stratigraphic units reported by Vreeken were based on a surface elevation interpreted from a 1:50 000 NTS map sheet (72E/09). These elevations have been recalculated with reference to mapping grade differential GPS (dGPS) survey that establishes the surface elevation of this core at 1 233 m asl. The adjusted elevations place the lake level between 1 226 m asl and 1 234.5 m asl at 6700 BP. Furthermore, Vreeken concludes that at the close of uninterrupted lacustrine sedimentation at this location, Elkwater Lake had no outlet below an elevation of 1 230.5 m asl (i.e. the top of the lacustrine sequence in this core).

North (1992) has conducted palynological investigations on several cores from Elkwater Lake with a maximum basal date of 4645 ± 35 BP (no lab number provided). Vance and Last (1994) extracted four cores from the lake. A core located in the deepest basin penetrated the lakebed to 2.8 m and yielded the oldest basal date of 4940 ± 70 BP (CAMS-3988). Analysis of three of these cores indicated a period of increased salinity between approximately 5000 and 4000 BP suggesting that the lake may have had no outlet at that time and that this interval was followed by a prolonged, high freshwater stand.

Geomorphic Investigations

All archeological and physiographic data sets used in these analyses were assembled in a Geographic Information System (GIS). The analytical and modeling capabilities of the GIS were used to reconstruct pre-slump conditions, estimate the volume of material displaced as a result of each landslide, integrate the stratigraphic data collected across the study area, and recreate paleolake levels in order to evaluate the viability of alternative outlets.

A variety of existing and derived geospatial datasets were used including topography, hydrology, hypsography, and road networks from the National Topographic Database (NTDB), a digital elevation model (DEM) constructed for the study area, and large-scale orthorectified colour photography. Spatial data were converted to a common coordinate system (UTM, Zone 12, WGS84) and all elevations reported herein are based on height above mean sea level with reference to the EGM96 earth geopotential model.

Dem Construction

Much of this work necessitated the development of an accurate topographic model of the study area. A DEM was created by integrating data collected in the field with existing elevation data from the NTDB. Two Trimble ProXRS GPS receivers with real-time differential correction and a Topcon GTS 210 total station were used to collect over 30 000 control points. The majority of these coordinates were collected within a 500 m radius of the Stampede site and in a series of transects across the Elkwater Channel and adjacent slump blocks at the west end of Elkwater Lake. The estimated positional accuracy of the dGPS and total station survey was approximately ±25 cm horizontal and ±75 cm vertical.

These data were integrated with the NTDB hypsographic data, which is equivalent in form and positional accuracy to contour lines on the corresponding 1:50 000 NTS map sheet (72E/09). The estimated positional accuracy of the NTDB data reported in the accompanying metadata file was ±50 m horizontal and ±10 m vertical.

Due to the substantial difference in positional accuracy, a method was devised to adjust the elevation of NTDB contours based on their mean deviation from corresponding dGPS or total station coordinates collected at 36 sample sites. The resulting model was used to calculate adjusted heights for the NTDB contours. The height-adjusted contours were then integrated with the dGPS and total station survey and hydrologic features from the NTDB to create a DEM of the area. The resulting DEM (Fig. 3) was interpolated at a 5 m spatial resolution.

Digital Orthophotography

A digital orthorectified photo-mosaic (background image Figs. 2, 4, 5, 6, and 7) of the area was created by scanning conventional 1:15 000 colour stereo-photography (courtesy of Alberta Environment, Airphoto Services) and orthorectifying the resulting digital images to the road network and surveyed ground control points. The spatial resolution of the resulting images was approximately 1 m. The orthorectified and original stereo-photography were used to delineate the extent of the ESB and WSB and reconstruct pre-slump conditions as described below.

Figure 3

Digital Elevation Model (DEM) of Elkwater Lake study area and location of alternative outlets A, B, C, and D.

Modélisation numérique des altitudes dans la région de l’Elkwater Lake et localisation d’autres exutoires en A, B, C et D.

-> See the list of figures

Reconstruction of Slump Blocks

A topographic reconstruction of pre-slump conditions was created by manipulating the existing DEM. The ArcGIS Spatial Analyst extension was used to generate contour lines from the DEM. An admittedly subjective process was then used to manually edit the contour lines, removing the debris surface from the channel floor and replacing this material on the scarp wall to recreate pre-slump conditions. This process relied extensively on the digital orthophotography and stereo-photography described above to guide the reconstruction process and recreate the profile of the original valley wall as accurately as possible. The Spatial Analyst extension was then used to perform a cut and fill operation to determine the volumetric change between the post- and pre-slump DEMs and estimate the area and total volume of material displaced as a result of each landslide.

Results suggest that approximately 9 M m3 of material was displaced over an area of 1.28 M m2 as a result of the landslide depositing the ESB and that the debris would have been able to impound the lake to a maximum elevation of 1 242 m asl, the current minimum elevation of the debris surface. The adjacent WSB event involved approximately 12 M m3 of material displaced over an area of 1.95 M m2, but would have been able to impound the lake to a maximum elevation of only 1 230 m asl.

For comparison, the Police Point landslide, the largest historical mass wasting event in the study area, involved and estimated 1.5 M m3 of material displaced over and area of approximately 458 000 m2 (Sauchyn and Lemmen, 1996). The landslides resulting in the deposition of the ESB and WSB are in the order of 6 to 8 times larger than this event. However, other late Holocene mass wasting events in the Cypress Hills are of similar magnitude as the Police Point landslide, suggesting that the ESB and WSB events are not typical late Holocene landslides and, therefore, most likely of early to mid-Holocene origin when initial readjustment of the fluvial geomorphic system resulted in larger magnitude slides (Sauchyn, 1999).

Figure 4

Phase I (ca. 13 500 to 9440 BP) showing drainage to the west through the Elkwater Channel.

Phase I (ca 13 500-9440 BP) illustrant l’écoulement vers l’ouest par l’Elkwater Channel.

-> See the list of figures

Dating Techniques

Previous research has indicated that Elkwater Lake predates the deposition of Mazama tephra (Vreeken, 1986). Therefore, it is unlikely that relative dating techniques, such as those employed by Goulden and Sauchyn (1986), can provide an accurate estimate of the age of these events since the processes upon which they rely typically reach an equilibrium state within 3000 to 5000 years (D. Sauchyn, University of Regina, personal communication, 2002). Consequently, an absolute date is required to estimate the age of this event and establish a maximum age of lake formation since the lake can not be older than the landslide that impounded it. Thus, a series of cores were taken from sites within semi-permanent wetland areas on the debris surface of the slump blocks, where deposition and soil formation processes were less likely to have been interrupted and a continuous record of Holocene sedimentation and soil formation since deposition of the debris surface would most likely occur.

A truck-mounted, direct-push hydraulic coring rig (GeoProbe ®) was used to retrieve five cores from sites on the east and west slump blocks. Three cores were taken from a circular depression approximately 70 m in diameter located on the ESB. These cores were driven to a depth of approximately 8 m and appear to represent a relatively complete record of Holocene deposition. Unconsolidated fragments of weathered sandstone and quartzite cobbles encountered at approximately 6.5 m were interpreted as the unaltered debris surface of the slump block. Several organic layers exhibiting varying degrees of pedogenesis and containing charcoal and numerous gastropod shells were identified above the debris surface and Mazama ash (ca. 6700 BP) was encountered at a depth of 2.69 m.

Figure 5

Phase II (ca. 9440 to 7245 BP) showing conditions following the landslide that resulted in the deposition of the ESB. Subsequent to impoundment lake levels rose to between 1 230.5 and 1 234 m asl prior to 7245 BP when the Feleski Creek outlet was established. Early occupants of the Stampede site inhabited a nearshore, wetland environment approximately 50 m from the lake.

Phase II (ca 9400-7245 BP) illustrant l’état des lieux après le glissement du Eastern Slump Block (ESB). Par suite de l’endiguement, le niveau lacustre s’est élevé entre 1 230,5 et 1 234 m avant 7245 BP lorsque s’est établi l’exutoire de Feleski Creek. Les premiers occupants du site Stampede habitaient un milieu humide à environ 50 m du rivage.

-> See the list of figures

Unidentified chitin was recovered from a paleosol just above the debris surface at a depth of 6.13 m. A standard acid-base-acid pretreatment was performed on the sample and an AMS date of 9440 ± 40 BP (BETA 169431) was obtained. Chitin was selected for dating since other soil organics were more likely to represent multiple ages. Since the upper boundary of this paleosol was very abrupt and appeared undisturbed it is believed that the chitin was introduced during soil formation and not as a result of post-depositional processes.

Two similar although considerably smaller semipermanent wetland locations were cored on the adjacent WSB. Slump debris was encountered at approximately 1.5 m depth but Mazama ash was not present at either location. Only one organic layer was encountered in the first core between 0.5 and 1.0 m depth.

While an absolute age for the WSB is not currently available, the comparative lack of pedologic development and absence of Mazama ash in the stratigraphic record appears to support the hypothesis that the WSB is considerably younger than the ESB, most likely no older than 6700 BP.

Alternative Outlets of Elkwater Lake

Four alternative outlets for Elkwater Lake were evaluated (Fig. 3). All flow northward through the Green Lake moraine and into the Ross Creek system. Outlets A and C were proposed by Vreeken (1986) and outlets B and D were identified through an analysis of the DEM and accompanying stereo-photography. The viability of each outlet was assessed with reference to the lake level required to produce flow through each outlet and the occurrence of corroborating geomorphic or stratigraphic evidence.

Figure 6

Phase III (ca. 7245 to 3500 BP) showing development of the alluvial fan that extends completely across the Elkwater Channel, blocking drainage to the east.

Phase III (ca 7245-3500 BP) illustrant le matériel glissé qui ensevelit l’Elkwater Channel, empêchant l’écoulement vers l’ouest.

-> See the list of figures

Outlets B, C, and D are all located well above the estimated uncontrolled elevation of Elkwater Lake. Outlet B is located at 1 242 m asl, coinciding with the maximum water level that could have been retained behind the ESB. There is currently no channelized flow through this outlet. Outlets C and D are located at approximately 1 249 m asl, 7 m above the maximum water level that could be retained behind the ESB. There is currently no channelized flow through either of these outlets and no significant stratigraphic or geomorphic evidence to support the hypothesis that either of these outlets once drained Elkwater Lake.

Pending further investigation, outlets B, C, and D are not considered viable outlets at this time. It is likely that these low points along the edge of the spillway represent preglacial and/or meltwater channels or tunnel valleys and at no time actively drained Elkwater Lake.

Outlet A is 3.5 km from the eastern shore of Elkwater Lake at an elevation of 1 224 m asl; approximately 2.5 m below the current controlled lake level and 1 m below the estimated uncontrolled lake level. The adjacent upland rises 18 m above the channel bed so it is likely that this too represents a preglacial channel that was reoccupied by a meltwater channel and, consequently, may have represented the lowest point of overflow at the time it captured Elkwater Lake.

An active channel up to 2 m wide, informally referred to as Feleski Creek (after the current landowner), intermittently drains two small semipermanent wetlands through a series of drainage ditches. The channel bed is armoured with large cobbles and boulders ranging from 5 to 15 cm in diameter (b-axis) suggesting that the current stream is misfit. Discharge in July 2002 following an abnormally wet spring was less than 1 m3 s-1.

Figure 7

Phase IV (ca. 3500 BP to present) showing the alluvial fan, which cut off the Feleski Creek outlet. Drainage is to the north through Ross Creek. The modern uncontrolled lake level is approximately 1 225 m asl.

Phase IV (de ca 3500 BP à aujourd’hui) montrant la formation du cône de déjection qui obstrue l’exutoire du Feleski Creek. L’écoulement se fait vers le nord par le Ross Creek. Le niveau lacustre actuel se situe à environ 1 225 m.

-> See the list of figures

A Reconstruction of Elkwater Lake

The investigations described above, in conjunction with the results of previous research, provide several key pieces of information for reconstructing the events responsible for the formation of Elkwater Lake and subsequent changes in lake level and extent. The following reconstruction consists of four phases describing conditions and events at specific periods in time; these are illustrated in Figures 4 through 7 and summarized in Figure 8.

Phase I (Fig. 4) depicts early Holocene conditions following abandonment of the Elkwater Channel (ca. 15 000 BP) to approximately 9440 BP. During this period surface runoff and subsurface discharge were conveyed through the Elkwater Channel westward. Analysis of the DEM suggests that drainage prior to impoundment was to southwest through Medicine Lodge coulee or Bullshead Creek, but may have later been diverted to the northwest through Gros Ventre Creek as a result of channel infill.

Phase II depicts conditions starting at approximately 9440 BP following impoundment of the lake by a massive landslide that resulted in the deposition of the ESB (Fig. 5). Subsequent to this event, lake levels rose throughout the comparatively cooler and wetter early Holocene. Elkwater Lake attained its highest early Holocene elevation prior to 7245 BP when it spilled through outlet A into Feleski Creek. At this time, the lake would have had no other outlet (including Ross Creek) lower than this point of overflow. Based on the estimated mid-Holocene elevation of the Ross Creek outlet and elevation of lacustrine sediments in Vreeken’s core, the lake level at this time would have been greater than 1 230.5 m but less than 1 234 m asl.

Phase III (Fig. 6) is a period of increasing aridity beginning approximately 7245 BP with a prolonged low-water stand ending at approximately 3500 BP. Stratigraphic evidence from the Stampede site suggests a nearshore environment between 7245 and 6700 BP. At 6700 BP the lake stood between 1 226 and 1 234.5 m asl, between the elevation of the subaqueously deposited Mazama ash and the subaerially deposited ash at the Stampede site. Increased salinity of the lake between 5000 and 4000 BP (Vance and Last, 1994) suggests that the lake had no outlet at this time and, consequently, would have dropped to less than 1 224 m asl, the estimated height of the Feleski Creek outlet. Concurrently, decreasing vegetative cover and/or a transition from a primarily forest to grassland environment resulted in increasing erosion and sedimentation rates due to the increasing erosive ability of extreme precipitation events. Consequently, the interval between 7245 and 5800 BP represents a period of peak aggradation of the alluvial fan as it prograded across the Elkwater Channel. There is no evidence indicating the lowest lake levels occurring during this period, but it is reasonable to suggest that only the deepest basins in the lake may have contained open water.

Figure 8

Proposed phases depicting conditions and events at specific periods in time. Grey boxes and dimension bars indicate estimated range in elevation and/or time.

Particularités liées aux différentes phases proposées. Les boîtes grises et les barres donnent l’estimation des altitudes dans le temps et de la durée des différentes étapes.

-> See the list of figures

Phase IV depicts present conditions reflecting geomorphic processes operating since 3500 BP (Fig. 7). With a return to cooler and wetter conditions during this period lake levels began to rise. With the former outlet through Feleski Creek now cut off by the alluvial fan, the lake attained its highest Holocene level of approximately 1 234 m asl and spilled over its new outlet through Ross Creek. Subsequent incision of the channel has since lowered the outlet to its estimated uncontrolled elevation of 1 225 m asl.

Conclusions

Investigations in the Elkwater Lake area have contributed new information regarding the paleoenvironmental history of Elkwater Lake. In summary, these are: 1) Elkwater Lake formed no earlier than 9440 BP; 2) impoundment was a result of a large landslide resulting in the deposition of the ESB; 3) Elkwater Lake attained its highest early Holocene level (ca. 1 230.5 to 1 234 m asl) prior to 7245 BP; 4) during this high-water stand an outlet was established through Feleski Creek, which at that time represented the lowest point of overflow; 5) this outlet was later abandoned as lake levels dropped and later cut-off by deposition of the alluvial fan on which the Stampede site is located (ca. 7245 to 5800 BP); and 6) the current outlet through Ross Creek was established during a second high-water stand (ca. 1 234 m asl) during the late Holocene (Fig. 8).

This evidence supports the hypotheses that early occupants (i.e. ca. 9440 to 7245 BP) of the Stampede site may indeed have selected this location as a result of its proximity to the lakeshore, which was estimated to be within 50 m of the site. However, this is not likely the case for subsequent occupations during the mid- to late Holocene since the site at this time would have been several hundred metres from the lakeshore. The decrease in lake level and rapid aggradation of the alluvial fan during the comparatively warmer and drier mid-Holocene resulted in a change in the position of the site relative to both the lakeshore and on the fan itself, as the site migrated from the proximal to distal end of the fan. As a result, it is likely that variations in sedimentation rates and the textural composition of sediments at the site are a function of both changing climatic conditions and site-specific geomorphic processes.