Abstract
The Lawn Lake site is a stratified hunting camp situated on a glacial lake outlet river terrace in Rocky Mountain National Park’s upper subalpine forest zone. Its archaeological assemblage represents 9,000 years of hunter-gatherer use as a summer game and plant processing camp for subalpine forest and nearby alpine tundra resource areas. This article’s focus is on the site’s earliest camp levels which contain artifacts and AMS radiocarbon dated hearth charcoal between 8,900 and 7,900 cal yr BP, placing them among the region’s earliest high montane (3,353 m ASL) Paleoindian hunting camps, once part of a network of such sites designed to support systematic high altitude procurement of summer migratory game animals and plant foods in Southern Rocky Mountain subalpine forest and tundra ecosystems. Lawn Lake paleoclimate and paleoecology studies produced long-term pollen records and climate-proxy sediment data for modeling the site’s prehistoric climate and ecology history, useful for interpreting its high-altitude Late Paleoindian hunter-gatherer adaptations.
Keywords
Introduction
The Lawn Lake site (5LR318), located in Rocky Mountain National Park and Colorado’s Southern Rocky Mountains (USA), was investigated in 2000 as an archaeological mitigation project by the University of Northern Colorado (Figure 1).
Location map of the Lawn Lake archaeological site in Rocky Mountain National Park and Colorado’s Southern Rocky Mountains (USA).
Excavations of the site’s prehistoric through early historic cultural deposits recorded 50-cms of vertical stream terrace micro-strata within 5 m2 (cf., Brunswig, 2001a). Its cultural deposits represented thinly layered short-term hunting camps situated on a glacial lake outlet-stream terrace on the margins of subalpine forest. A simultaneous sediment coring program to the archaeology was conducted in a peat fen at the lake’s upper north end, resulting in a continuous record of paleoecological and paleoclimate change from 8,190 cal yr BP to the present (radiocarbon date calendar-age conversions in this article used the internet-based Oxcal 4.3 software program at https:c14.arch.ox.ac.uk, cf. Ramsay, 2000; Ramsey and Lee, 2013). The following article describes results and interpretations of archaeological and paleoenvironmental research focused on the site’s earliest Late Paleoindian cultural occupations, placing them within a backdrop of park and Southern Rocky Mountain landscapes of the Early Holocene.
Lawn Lake’s environmental and geological context
Physiography
Lawn Lake is located at the southern edge of a high altitude (elevation 3,353 m ASL) glacial col valley, once occupied by the upper Roaring River Glacier from the region’s Late Glacial Maximum (LGM) through the end of the late Pinedale (Terminal Pleistocene) Stadial (Madole et al., 1998). The site’s adjacent glacial tarn lake, situated at the lower end of a small confined cirque valley, is upslope from Rocky Mountain National Park's Horseshoe Park, east of the continental divide (Figure 2).
GoogleEarth™ satellite image of Lawn Lake, Roaring River, and Horseshoe Park Section of Rocky Mountain National Park.
Geology and hydrology
A second upland glacial cirque, the Crystal Lake tarn (glacial lake), located 1.15 km northwest and up-valley from Lawn Lake, was the initial origin point of the Roaring River Glacier (Figure 3). Lawn Lake Valley’s northern margins are comprised of pre-Cambrian granite bedrock (Silver Plume and Hague Creek granites) dating to 1.4 ga (Braddock and Cole, 1990) while its lower eastern, western, and southern slopes contain lateral and terminal moraine till deposits dating from final recession of the Roaring River Glacier between ca. 13 and 11 ka (see the paleoenvironment section below). Lawn Lake itself is a terminal-moraine dammed lake whose surface area, prior to construction (AD 1903) of a reservoir dam at its lower south end, covered ∼0.0066 km2 (Jarrett and Costa, 1993:5). Lawn Lake and its upslope glacier’s originating Crystal Lake tarn together form upper headwaters of Roaring River, a rapidly flowing stream that descends several km from Lawn Lake to join Fall River at the base of Bighorn Mountain in the park’s upper Horseshoe Park. Several millennia before establishment of the first Lawn Lake hunting camp, the Roaring River Glacier fed into the larger downslope Fall River Glacier, during the region’s LGM, ca. 21–16 ka (Figure 3).
Overview map of the Crystal Lake and Lawn Lake tarns, the Lawn Lake site, and outline of the Roaring River Glacier to where it joined Fall River Glacier during the Late Glacial Maximum.
The modern Lawn Lake ecosystem
Lawn Lake, at 3,352 m ASL, is situated just below the interface of upper margins of subalpine forest and alpine/sub-alpine (krummholz) ecotone. At the lake and site location, local sub-alpine forest plants are subalpine fir (Abies lasiocarpa), Engelmann spruce (Picea engelmannii), cinquefoil (Pentaphylloides sp.), blue gentian (Pneumonanthe affinis), elk sedge (Carex geyeri), mutton-grass (Poa fendleriana) and needle-grass (Stipa lettermanii) (cf., Jarrett and Costa, 1993; Mansfield, 1993).
Sub-alpine dwarf Engelmann spruce (Picea engelmannii) and dwarf fir (Abies lasiocarpa) patch tree islands, referred to as krummholz, are the primary tree species in the alpine-subalpine ecotone (cf., Baker and Weisberg, 1995; Peet, 1981: 29–30), which grows on upper Roaring River Valley mountain slopes and eroded benches northeast, east, and southeast of Lawn Lake. Ecotone ground species often consist, among others, of poleminium (Polemonium delicatum), artemisia (Artemisia arctica), alpine sandwort (Arenaria obtusiloba), bistort (Bistorta bistortoides), alpine cinquefoil (Potentilla diversifolia) and alpine grasses such as wheatgrass (Elymus scribneri).
The park’s alpine tundra has, for millennia, provided rich foraging territory for summer migrating elk and bighorn sheep. In the upper Roaring River headwaters, largely open tundra emerges from scattered krummholz tree island ecotone patches above Lawn Lake and at near the Crystal Lake tarn between 3400 and 3500 m ASL. Dominant tundra species are arctic willow (Salix arctica), dwarf clover (Trifolium nanum), arctic gentian (Gentianodes algida), sedge (Carex pyrenaica) and spreading wheatgrass (Elymus scribneri).
Some modern mammal species, including those known to have been prey of Native American hunters during the latest Pleistocene and Early Holocene, are known from archaeological and/or paleontological records to have seasonally (late spring-early fall) migrated into Lawn Lake’s subalpine forest and alpine tundra eco-zones (Crystal Lake). Among those were mule deer (Odocoileus hemionus), elk (Cervus elaphus), bighorn sheep (Ovis canadensis), and red fox (Vulpes vulpes) (cf. Armstrong, 2008). Late Pleistocene-Early Holocene bison (Bison antiquus and Bison occidentalis), while not believed to have been a regular high-altitude summer resident in high-elevation eco-zones, were present as individuals or in small herds during warmer Early Holocene climatic episodes when Lawn Lake Late Paleoindian hunting camps were occupied (Brunswig, 2015a). Historical records of bison in the park and evidence from high mountain pass glaciers and millennia-old ice patches have produced bison fossil skeletal remains dating from ca. 2,000 BP through the late 19th century, after which time bison were hunted from the region (Brunswig, 2014b: 100, 2015a: 15–25; LaBelle, 2012; LaBelle and Whittenburg, 2015; Lee and Benedict, 2012: 43–44, Table 1; Lee et al., 2006). Smaller fauna found in the area today (and throughout earlier Holocene periods) include yellow-bellied marmot (Marmota flaviventris) and pika (Ochotona princeps), mainly restricted to alpine areas above Lawn Lake valley, while gold-mantled squirrel (Spermophilus lateralis), black bear (Ursus americanus) and mountain lions (Felis concolor) more frequently are found in its sub-alpine forest zone (Armstrong, 2008).
AMS Radiocarbon Date Data for Lawn Lake Fen Sediment Cores.
aConventional AMS radiocarbon age with fractionation adjustment as reported by Beta Analytic Laboratory. Dates rounded to nearest 10th whole numbers. Standard deviation is one sigma (1 σ).
bCalendar year (calibrated) age was calculated using the Oxcal 4.3 software program on-line at https:c14.arch.ox.ac.uk (cf., Ramsay, 2000; Ramsey and Lee, 2013). Standard deviation is two sigmas (2σ) and dates are rounded to the nearest tenth whole numbers.
Modern climate
Direct modern climate data for the immediate Lawn Lake area are unavailable. However, analogous data from the Berthoud Pass (3447 m ASL) meteorological station, 93 km to the south-southwest, provide a 35-year record (1950–1985) of temperature and precipitation that broadly parallel today’s Lawn Lake conditions (Western Regional Climate Center, 2020). Berthoud Pass autumn (September-November) temperatures average a maximum of 5.5 °C and a minimum of −6.3 °C. Winter (December-February) averages were −5.05°C (maximum) and −16.6°C, respectively. The spring (March-May) maximum average was 2.6°C and its minimum average was −6.4°C. The summer (June-August) maximum average was 14.8°C. and the minimum average was 2.4°C. Total modern precipitation values for each of the four seasons were: 1) autumn 197 mm, 2) winter 259 mm 3) spring 315 mm, and 4) summer 186 mm. The highest precipitation months (60%) are in winter and autumn, mainly falling as snow. Berthoud Pass records provide snowfall amounts, believed analogous to those at Lawn Lake, between 1950 and 1985 as: 1) autumn 2180 mm, 2) winter 3650 mm, 3) spring 3800 mm, and 4) summer 307 mm. Other Lawn Lake Valley climatic variables, such as wind patterns, are poorly known, but based on data from similar locales in the region (Hansen et al., 1978), prevailing modern (and presumably pre-modern) wind directions are north to northwest in all seasons, subject to minor localized air current “channeling” due to topographic confinement variables of the Crystal Lake-Lawn Lake valley and seasonal convection conditions. Storms and weather fronts normally enter the valley from the northwest where they descend onto the Lawn Lake site through the higher Crystal Lake basin, frequently bringing intense winds. Due to its south-southeast aspect, the valley tends to be mainly snow-free by late spring through early summer when most other high-altitude valleys and pass areas in the park retain snowfields into late summer months.
Lawn Lake and RMNP Late Pleistocene glacial history
Since this article focuses on the Late Pleistocene/Early Holocene cultural and environmental transition, this section describes paleoenvironment/paleoclimate data confined to that period. Data and resulting interpretive models of RMNP Late Pleistocene/Early Holocene paleoclimates and paleoenvironments have been generated from park fens, ponds, and lakes at fifteen sampling locations studied over the past four decades by the authors and other researchers. In the late 1970s and 1980s, La Poudre Pass (3,110 m ASL, 27.4 km west-northwest of Lawn Lake) was an important location for early park paleoclimate projects by Madole (1976, 1980), Elias (1983, 1985, 1996, 2015), and Short(1985). From the early 2000s to the present, fifteen paleoclimate study sites (fens, alpine ponds, and lakes) were sampled by sediment coring and subjected to radiocarbon-dating and climate proxy analysis (pollen, sediment lithostratigraphy, and magnetic susceptibility data) by the authors (cf., Brunswig, 2014b, 2015b; Brunswig et al., 2009, 2014a, 2014b; Doerner, 2004, 2005, 2008, 2009, 2014; Doerner and Brunswig, 2001, 2008) (Figure 4).
Location map of Lawn Lake in relation to other park paleoenvironmental study sites discussed in the text. The boundary of Rocky Mountain National Park is shown as a black dashed line and the continental divide is marked as a black-and-white dashed line.
Geological studies of La Poudre Pass (3,103 m ASL) (Madole, 1976, 1980; Madole et al., 1998), west-northwest of Lawn Lake (Figure 4), show it once was the originating point (col) of a large double-lobe Late Pleistocene glacier that flowed in opposing directions from the Continental Divide into modern-day Colorado River headwaters (southwest) and into a headwaters branch of the Cache La Poudre River (northwest). Pollen and sedimentological evidence from a Poudre Pass peat fen were interpreted to show the pass’s glacier initiated post-LGM recession at ca. 16,570 cal yr BP, with final de-glaciation occurring at the end of the Younger Dryas climatic episode at ca. 11,460 cal yr BP (Madole, 1976: 163–165, 167, 1980; Short (1985: 11–13). A La Poudre Pass Fen core recovered by one of the authors, Doerner (2002) yielded a basal organic sediment age of 11,030 cal yr BP, several centuries after the Younger Dryas and Late Pleistocene (Doerner, 2004, 2005, 2008, 2009, 2014), and dated peat formation at ca. 9,930 cal yr BP. Both La Poudre Pass core studies provided evidence for onset of earliest Early Holocene warming between ca. 11,460 and 9,930 cal yr BP.
Geomorphic studies of post-glaciation landslide and lateral-moraine dammed lacustrine deposits in the Roaring River Valley, down-slope from Lawn Lake, produced post-glacial basal lacustrine sediments radiocarbon-dated at 13,620 cal yr BP, 12,830 cal yr BP, and 11,420 cal yr BP (Madole, 2012: Table 1, Figure 3, 199–202, 204; Summer, 1993:47–48), suggesting its Late Pleistocene glacier had fully recessed into its originating Crystal Lake cirque within that time range. The Roaring River Glacier, a tributary to the larger Fall River Glacier (Figure 3), completed its recession into its originating cirques by the time the Fall River Glacier retreated well up-valley past the junction of the two glaciers. A 1.5 km down-valley from the two glaciers’ junction, basal sediment core samples from a kettle lake pond (Sheep Lakes, Figure 3), formed when detached glacier ice blocks melted and created the pond basin, were radiocarbon dated between 14,370 cal yr BP and 14,430 cal yr BP (Rainey, 1987: 48–49, Table 2; note: Rainey dates converted to Oxcal 4.3 median calendar age range). Evidence that the larger Fall River Glacier was actively receding by 12,410 cal yr BP is documented by organic fen sediment radiocarbon dates in Hidden Valley, a small basin trapped behind a Fall River lateral moraine, 1.5 km half up-valley from the Sheep Lake kettle ponds.
Lithic tool debitage data from Lawn Lake’s Late Paleoindian occupations.
aLate Paleoindian stratigraphic unit 3 is subdivided into upper and lower sub-units, broadly radiocarbon-dated to early (ca. 8,926 cal BP) and late (7,960 cal BP) occupation periods.
bRock (geological) material types defined according to technical and common archaeological terminologies in Colorado’s Southern Rocky Mountains, Front Range foothills, and eastern High Plains. In some cases, (e.g. Kremmling chert (Black, 2000; Metcalf et al., 1991), material names include terms used to reference a well-known material type associated with a specific geological formation.
cLithic source refers to the geographic region or sub-region from which a material type has been geologically and archaeologically documented. Lawn Lake artifact materials have been identified through technical comparisons, including microscopic and ultraviolet light analysis, with an extensive University of Northern Colorado lithic sample collection from geological and archaeological sites in the western United States.
dIdentification of lithic materials and their source locations/areas is based on a maximum foot-travel radius of 3–4 days outward/inward from or to an archaeological site (in this case, Lawn Lake), e.g. the nearest lithic material source. In the Southern Rocky Mountain region, which includes Lawn Lake, a local lithic source radius zone is broadly defined as occurring within 75-km while non-local sources are those which occur beyond a 75-km radius.
Twenty-one kilometers south-southwest of Lawn Lake, geological studies in the park's Loch Vale (Nash, 2000) and nearby alpine tundra Sky Pond areas (3,316 m ASL) (Menounos and Reasoner, 1997; Reasoner and Jodry, 2000:41–44, Table 1; Figure 4) provide evidence of their local glacier’s full recession by 14,070 cal yr BP (Sky Pond), followed by modest Younger Dryas cirque-restricted re-glaciation around 12,240 cal yr BP. Sky Pond sediment and pollen data also indicated the presence of local year-round snowfields with limited or no local ground vegetation and lowering of area tree lines during the Younger Dryas (cf. Menounos and Reasoner, 1997). Aside from an undergraduate thesis (Routt, 1990: 92–97) coring project, no glacial geology studies exist for the Lawn Lake and Crystal Lake tarns (glacial lakes) and cirque basins, so it is not known if either cirque experienced Younger Dryas glacial re-advances. However, it is plausible minimal re-glaciation occurred at Crystal Lake and Lawn Lake but remained confined within their cirques. Routt (1990: 94), in his study of Late Pinedale park glaciation, suggested stagnant Late Pleistocene glacier ice persisted in the Lawn Lake and Crystal Lake cirque basins as late as ca. 13,000 cal yr BP, followed by later minor Younger Dryas re-glaciation. Above cited evidence does indicate that Late Pleistocene de-glaciation (including the Younger Dryas) at the adjoining Lawn Lake and Crystal Lake cirques was complete by ca. 13,000 cal yr BP. Pollen, sediment, and fossil insect data from La Poudre Pass were cited by Short (1985) and Elias (1983) to suggest a minimum date of 12,410 cal yr BP for that pass’s final Late Pleistocene de-glaciation. Coincident with La Poudre Pass, rapid Early Holocene warming led establishment of modern to above-modern tree lines at that pass, the park in general, and the surrounding region no later than ca. 11,460 cal yr BP (Brunswig, 2014b: 104–105, 2015b: 50; Brunswig et al., 2014b: 104–105; Short, 1985: 11–15, Figures 3 and 4).
Lawn Lake and RMNP Early Holocene paleoclimate
Recent paleoclimate research by the authors in an extensive Forest Canyon Pass-Mount Ida Ridge research area, several kilometers southwest of Lawn Lake, utilized sediment core data from alpine and sub-alpine fens and ponds and radiocarbon-dated above-modern tree line ice patch fossil tree remains to generate a Geographic Information System (GIS) model of latest Pleistocene and post-Pleistocene climate and tree line boundary changes (Brunswig, 2014b, 2015b: 48–51, 68, 76, 83; Brunswig et al., 2014a, 2014b). The model drew on two decades of paleoclimate research by the authors in Rocky Mountain National Park and the adjacent interior montane basin valley of North Park (cf. Brunswig, 2005: 42–63; Brunswig et al., 2009; Doerner and Brunswig, 2001, 2007, 2008; Doerner, 2004, 2005, 2008, 2009, 2014). It further refined pre-existing La Poudre Pass Post-Younger Dryas paleoclimate data, showing that, integrated with our more recent Forest Canyon Pass-Mount Ida Ridge fen and pond sediment data (Figure 3), Mount Ida continental divide alpine ridgeline and mountain slope tree line reached ∼130 m higher-than-modern levels between 10,140–9,330 cal yr BP, within a millennium of the end of Younger Dryas cold climate (Brunswig, 2014a: 103). Lawn Lake paleoclimate data (see below), correlated with La Poudre Pass, Sky Pond, and Loch Vail pollen and fossil insect studies and those applied to the Forest Canyon Pass-Mount Ida GIS paleoclimate model, support existence of an Early Holocene (post-Younger Dryas) climatic optimum and tree lines rising to or above modern-day levels between 10,300 and 9,700 cal BP (cf. Brunswig, 2014b: 103, 2015b: 50; Elias, 1985: 33, 35–36, 43–45). In 1984, a sediment core from Mount Ida Ridge pond at 130 m above modern tree line recovered fossil insect remains and fossil tree evidence, radiocarbon-dated between 10,140 and 9,330 cal yr BP, of an Early Holocene tree line significantly higher than today. Elevated tree line, despite small, alternating late Early Holocene cooler-warmer climate cycles (see below), remained in place with minor fluctuations through subsequent Middle Holocene warming until regional cooling (neoglaciation) depressed park tree lines after ca. 5,650 cal yr BP (Brunswig, 2014b: 104; Elias, 1985: 46).
Lawn Lake paleoenvironment study methods
In 2000, two adjacent 120–cm long sediment cores were collected by the authors from a fen 850 m. northwest of the Lawn Lake archaeological site (Doerner and Brunswig, 2001) using a modified Livingstone square-rod piston corer (Wright, 1967). The core, recovered in segments, was wrapped in plastic film and aluminum foil, and transported to the University of Northern Colorado’s Paleoenvironmental Laboratory where the cores were split lengthwise and their sediments described.
Sub-samples of the sediments were collected for radiocarbon dating, sediment analysis (loss-on-ignition, bulk density, and magnetic susceptibility), and pollen analysis. Several 15–20–cm increment samples were extracted from one of the cores for AMS radiocarbon dating, and 1–cm3 samples were taken at 2.5–cm increments from the identical second core for bulk density, organic content, and magnetic susceptibility analysis. Loss-on-ignition (LOI) and bulk density (BD) were measured using techniques modified from those described by Dean (1974). Magnetic susceptibility (ĸ) measurements followed the standard techniques using a Barrington meter (Thompson and Oldfield, 1986). Samples for pollen analysis (1 cm3) were taken at 10-cm intervals in the core. Pollen samples were prepared using standard pollen extraction techniques (Faegri et al., 1989). Known quantities of Lycopodium spores were added to each sample to calculate pollen concentrations (grains/cm3). Pollen residues for each sample were stained and stored in glass vials containing silicone oil.
Pollen was identified by Doerner to the lowest possible taxonomic level under a light microscope (400×magnification) and compared with the university’s pollen reference collection and published keys (Kapp, 1969; McAndrews et al., 1973; Moore et al., 1991). Grains were categorized as “indeterminate” if they were corroded, degraded, broken, or hidden. If pollen grains could not be identified with available reference material, they were classified as “unknown.” A minimum of 300 terrestrial pollen grains was counted for each level sampled in the core. Birks and Birks (1980) note that numbers of grains counted should be large enough to achieve a random sample of the pollen grains present and suggest counts should be in the range of 300–500 grains per sample.
In construction of the Lawn Lake Fen percentage diagram (Figure 5), its pollen sum is only based on terrestrial taxa (trees, shrubs and herbs, and riparian types). Percentage values for aquatic-types (Cyperaceae) were based on the total pollen sum. Traditionally, palynologists exclude aquatic pollen from the pollen sum because they are produced locally and therefore tend to reflect only localized conditions. Pollen grains classified as “aquatic,” “indeterminate,” and “unknown” were excluded from our pollen sum. Pollen zones in the diagram were based on interpreted significant changes in pollen frequencies, influx rates, and using Tilia‘s constrained cluster analysis software program. ([CONISS] Grimm, 1988).
Lawn Lake pollen diagram with labeled pollen zones and associated AMS (cal yr BP, Oxcal 4.3 calendar-age corrected) radiocarbon dates (left figure) and sediment cores’ lithology diagram with AMS sample locations and associated calendar-age dates (right figure).
Pollen spectra from mountain environments can be challenging to interpret due to long-distance transport of pollen grains (Maher, 1963; Wright et al., 1973). Markgraf (1980) found that strong gradient winds transport substantial amounts of pollen upslope from lower elevations such that the pollen rain at higher elevations contains a mixture of both high elevation and lowland taxa. Conversely, she also found that the pollen rain at lower elevations contained few high elevation taxa. Pollen diagrams from the Rocky Mountains present additional problems of interpretation because many species do not produce significant amounts of pollen to signal changes in their relative abundance (Beiswenger, 1991). As a result, pollen reconstructions rely heavily on limited taxa which typically produce large amounts of wind-transported pollen (i.e., Picea, Pinus, Artemisia, Amaranthaceae, and Poaceae) or on indicator taxa, often present in small amounts.
Loss-on-ignition (LOI), bulk density (BD), and magnetic susceptibility (ĸ) data were analyzed to provide additional paleoclimate and environmental change proxy evidence. LOI is a measure of the build-up of organic matter reflecting biological productivity. When environmental conditions are more favorable (warmer and/or wetter), there is an increase in organic productivity (Andrews et al., 1975). When conditions are more limiting (cooler or drier), there is a corresponding decrease in organic productivity. Variations in bulk density values reflect changing inputs of mineral material into a depositional location. Bulk density increases coincide with periods of greater influx of fine-grained sediments by aeolian and/or fluvial processes. Typically, paired loss-on-ignition and bulk density measurements are inversely related since, as one variable increases, the other decreases. Magnetic susceptibility curves tend to mirror changes in the bulk density curve but are more variable in their overall range.
The Lawn Lake Fen cores provided “mirrored” stratigraphic sequences for systematic sampling and analysis. AMS radiocarbon-dated organic sediment from the base (–120 cm) of core 1 yielded a basal mid-late Early Holocene median date of 8,125 cal yr BP. The upper 86 cm of both cores was a fine very dark grayish-brown (10YR 3/2 moist) organic peat, with the next lower 8 cm (–86 to –94 cm) being a coarser brown (10YR 4/3 moist), woody peat. Non-woody peat (10YR 3/2 moist) occurred in both cores between –107 and –116 cm and organic sediments (gyttja) very dark gray (10 YR 3/1 moist) constituted the remaining core material –116 cm to –120 cm at its base. The gyttja basal core section rested on coarse gravel.
Eight accelerator mass spectrometer (AMS) dates, with high-resolution standard deviations (δ) of 40 and 50 years, were obtained from the Lawn Lake Fen core (Table 1; Figure 5). The AMS samples were processed by Beta Analytic Inc. in Miami, Florida, USA. The AMS dates were input into the R-based statistical program BACON v2.2 (Blaauw and Christen, 2011) to develop an initial age model using the IntCal13 calibration curve (Reimer et al., 2013). BACON uses a Bayesian approach to accumulation histories by separating the core into many sections and then estimating accumulation rate for each through millions of Markov Chain Monte Carlo iterations.
One potential, but of limited impact, source of dating error was reliance on organic sediments in the lowest (basal) level between –116 to –120 cm. As noted by Grimm et al. (2009), organic sediments may produce older than true ages due to uptake of more ancient bedrock carbon through groundwater percolation. However, that potential age error was not a factor above the cores’ basal section, since all subsequent AMS dates were derived from plant material (peat) and unlikely to have been subject to bedrock carbon contamination. Furthermore, the close together physical proximity and age dates of the basal (organic sediment) and next higher peat core sample indicate that contamination by ancient groundwater carbon from adjacent geological formations is unlikely.
The LLF pollen diagram was divided into three zones. Only the most abundant and ecologically important taxa are presented on the pollen diagram (Figure 5).
The pollen spectra across all three zones is dominated by high percentages Pinus (pine) pollen types with lesser representation of Picea (spruce), and Abies (fir). Abies (fir) pollen percentages are relatively low compared to today’s abundance of fir trees at the site. This is likely due to over-representation of Pinus pollen. High percentages of pine can be partly explained by the high productivity-dispersibility of pine pollen (Fall, 1992) as well as upslope transport of pine pollen from the Douglas fir-ponderosa pine dominant forests of the upper montane zone (Markgraf, 1980).
Pollen zone 1 extends from the top of the core to –40 cm and dates from the present. to ca. 1800 cal yr BP. Average pollen concentrations increase to 186,500 grains/cm3 and average accumulations rates increase to about 34,250 grains/cm2/yr. The lower part zone is characterized by declining arboreal (tree) pollen percentages and increases in nonarboreal pollen (∼30%). In the upper portion of this zone, Pinus values peak at an average of 54%. The increase in Pinus along with increases in Amaranthaceae (Amaranth family, formerly Cheno/Am; Judd et al., 2002) and Sarcobatus (greasewood) suggest that conditions were warmer and/or drier.
Pollen zone 2 extends from -40 to -90 cm and dates from ca. 1800 to 5400 cal yr BP. The pollen concentration averages about 115,500 grains/cm3 and accumulation rates average of 3050 grains/cm2/yr. The frequencies of pine and spruce declined as nonarboreal pollen percentages increased to an average of about 18%. Artemisia, Asteraceae, Ambrosia (ragweed), Amaranthaceae and Polygonaceae (buck-wheat family) are the most important nonarboreal taxa in this zone. The increase suggests that conditions became cooler as those taxa are shade intolerant. Salix percentages declined and Cyperaceae (sedge) achieved its greatest representation.
Only the lowest pollen zone 3 is contemporary with the Lawn Lake site’s Late Paleoindian occupations. Evidence for that zone was identified in –90 and –120 cm below surface core sections and AMS-dated from the later Early Holocene to the Mid Holocene, ca. 8,200–6,150 cal yr BP (Figure 5).
Pollen zone 3 correlates with upper (most recent) Late Paleoindian cultural layers of the Lawn Lake site while lower, older cultural deposits pre-date the core’s earliest pollen record. Pollen concentrations averaged nearly 200,000 grains/cm3 and pollen accumulation averaged about 2400 grains/cm2/yr. Zone 3 was dominated by arboreal pollen (AP) taxa, especially Pinus and Picea (pine and spruce) (Figure 5). Spruce pollen frequency reached its maximum representation (27%) at –100 cm., dated to ca. 6,650 cal yr BP. Non-arboreal (NAP) pollen types averaged only about 13.5%. Important NAP taxa were Asteraceae (sunflower family), Amaranthaceae, and Bistorta (bistort). Salix spp. (willow) pollen is minimally present, averaging less than 1%.
The earliest pollen zone 3 (see Figure 5) supports a scenario of significantly warmer-than-modern late Early to Middle Holocene period temperatures (∼ 8,190–5,620 cal yr BP), as inferred from frequencies of arboreal (AP), non-arboreal (NAP) and riparian aquatic species’ pollen. Forest Canyon-Mount Ida Ridge fen and pond pollen and pollen data from study sites a few kilometers south, as well as the Lawn Lake Fen cores, support a cycle of mild annual temperature decreases, still significantly warmer than today, between ca. 7,600 and 7,200 cal yr BP (cf., Brunswig, 2015b: 50; Brunswig et al., 2014b: 279). Those mild cooling trends are documented in the Lawn Lake pollen record by decreases in pine (Pinus sp.) and increases in spruce (Picea sp.) pollen (Figure 5). Based on paleoclimate data from other park core study sites with earlier recovered sediments discussed above, it appears the Lawn Lake site experienced warmer than modern summers and longer growing seasons, resulting in consistently robust peat growth not appreciably slowed by late Early Holocene intermittent cooling cycles between ca. 8,500 and 7,500 cal yr BP. Within a millennium of Lawn Lake’s final Late Paleoindian occupation at ca. 7,900 cal yr BP, annualized temperatures had risen to even higher than previous Early Holocene warm conditions with on-set of a warmer climatic episode (Benedict and Olson, 1978; Brunswig, 2015b: 47, 50; Doerner, 2014: 24–25). After ca. 5,620 cal yr BP, Lawn Lake Fen core data support the development of terminal Middle Holocene to early Late Holocene climate conditions consistent with late Holocene cooling documented elsewhere in the region (cf., also Benedict, 1968, 1973, 1985: 11–83).
After analysis and comparison of all core climate proxy data, magnetic susceptibility (ĸ) and bulk density (BD) data were judged less useful for paleoclimate modeling than organic content (LOI), which was then selected to supplement the above described pollen-based paleoclimate modeling interpretation. LOI values were converted to standard deviations for analysis and interpretation where positive standard deviation scores signaled increased botanic productivity and indicated warmer conditions. In contrast, negative values suggest lower productivity and cooler temperatures. LOI results plotted over more than eight millennia throughout the 120 cm long Lawn Lake Fen core sequence show closely similar warmer-cooler climate cycle patterning to the fen core’s unit 3 pollen diagram (Figure 6). LOI results also reinforce unit 3 pollen data in illustrating the region’s late Early Holocene half millennium of very modest cooling which began near the end of Lawn Lake’s AMS-dated Late Paleoindian Occupations (Figure 7), but ameliorated with resurgent warming after ca. 7,000 cal yr BP.
Lawn Lake fen record of paleoclimate cycles associated with inferred cooler or warmer than modern era temperature regimes as reconstructed from loss-by-ignition (LOI) analysis used as a primary climate proxy.
Composite graphs of three high altitude UNC paleoclimate study sites in Rocky Mountain National Park comparing LOI proxy paleotemperature patterns from ca.13,000 cal yr BP to the present. Lawn Lake’s Late Paleoindian occupation phase paleoclimate conditions are compared with contemporary and earlier records from La Poudre Pass (upper subalpine) and Forest Canyon Bench (subalpine-alpine ecotone) fens (see citations in text).
Comparison of Lawn Lake LOI temperature proxy data with LOI data from two other UNC paleoclimate fen and pond coring sites (La Poudre Pass and Forest Canyon Bench) are illustrated in Figure 7. They clearly show generally warmer than present day annual climate conditions in the late Early Holocene when the site’s Late Paleoindian occupations were occurring. Viewed together with the Lawn Lake Fen pollen record, LOI temperature patterns for the three park study sites support the scenario that Lawn Lake’s Late Paleoindian hunting camps were, like today, situated within the upper spruce-fir subalpine forest zone.
The effective end of Terminal Pleistocene glaciation in the Lawn Lake area was followed by rapid Holocene warming, which opened it to increased plant and animal colonization and economic exploitation by Native American hunter-gatherers who had been present in the park and its region since Clovis times (see Brunswig, 2003, 2004, 2014a, 2014a, 2015b).
Lawn Lake’s archaeological record
Artifacts recovered by Lawn Lake surface collections prior to and during University of Northern Colorado 2000 investigations show virtually unbroken use of the site from Late Paleoindian through historic Euro American periods (Figure 8 for the Rocky Mountain National Park and region cultural/chronological sequence below) (Butler, 1996, 1999; Brunswig, 2001a: 13–21; Nykamp, 1996).
Graph illustrating cultural periods and associated conventional and calendar-age (calibrated) chronologies for Rocky Mountain National Park and the Southern Rocky Mountains. The AMS date range of the Lawn Lake site’s Late Paleoindian camp occupations is indicated by a box to the left of the Late Paleoindian timeline.
Evidence from the 2000 research program significantly expanded understanding of Lawn Lake cultural history with more precise knowledge of its absolute chronologies, based on AMS radiocarbon dating and diagnostic-artifact correlations. Chronological and/or cultural component documentation, in addition to this article’s Late Paleoindian emphasis, included subsequent regional cultural period occupations of the Early Archaic (Mount Albion Complex projectile points, ca. 8,290–5,740 cal yr BP), Middle Archaic period (McKean Complex projectile points, ca. 5,740–3,170 cal yr BP), Late Archaic (projectile points, ca. 3,170–1,740 cal yr BP), the Early Ceramic (projectile points and two stratified AMS-dates-2,000-1,420 cal yr [1ϭ] date range and, 1,690–1,530 cal yr BP [1ϭ] date range), and the Protohistoric-Early Historic (based on Apachean and Ute projectile point types and an Apache potsherd organic-sherd residue AMS radiocarbon date of 580 cal yr BP, median 640–85 cal yr BP (2ϭ) date range (Brunswig, 2012: 22–25).
Surface evidence of the site’s earliest Late Paleoindian occupations
Existence of buried Late Paleoindian occupations at Lawn Lake prior to the 2000 excavations was previously documented with an early graduate thesis (Yelm, 1935: 42, 61, Plate VIII, No. 26) field survey which recovered what was identified, at the time, as a “Yuma” (Late Paleoindian) projectile point. An inventory of the park museum’s artifact collection by one of the authors (Brunswig) located the point in question, deposited by Yelm in 1931, and identified it as the lower section of a Late Paleoindian James Allen parallel-oblique flaked projectile point. That point, shown in Figure 9, is a lanceolate, deeply indented, almost “eared”, base made of quartzite (cf., also Brunswig, 2001b; Husted, 1962:104 as well as Yelm, 1935, cited above; and Pitblado, 2014: 18–327).
Late Paleoindian (James Allen) projectile point base and lower shaft recovered from the Lawn Lake site surface in 1931.
Description and interpretation of Lawn Lake test excavations
Lawn Lake surveys and excavations were a mitigation project in advance of the removal of an early 1900s-era reservoir dam at the lake’s south end. The dam failed catastrophically in 1982, resulting in severe down-stream flooding that reached as far as the town of Estes Park, 12 km to the southeast (cf., McCutchen et al., 1993). Excavation was preceded by pedestrian survey which recorded surface artifact scatter concentrated along a narrow terrace “corridor” between the Roaring River (west) and the dam spillway’s western erosion scarp (see Figure 10).
Map of the Lawn Lake site showing locations of the 2000 excavation units and areas with surface artifacts.
Lithic tools and flaking debris were found concentrated within a 30–m radius of the excavation area on the dam spillway (and Roaring River left bank) terrace. Observation of exposed charcoal staining and artifacts eroding from the spillway’s east-facing erosion scarp determined placement of three excavation unit locations (Figure 10). One of the excavation locations (no. 1) was eventually expanded from a single 1m2 unit to three, resulting in the final excavation of 5m2. Excavations extended from the exposed scarp into the terrace bank and uncovered three well-defined and stratified cultural units each with multiple, thinly layered camp occupation levels, representing nearly nine millennia of short-term visits. The lowest cultural unit (3), subject of this article, was 15-cm thick and contained the site's earliest (Late Paleoindian) occupation levels. Its moderately developed A soil horizon was a dark yellowish-brown (Munsell 10YR 4/4) sandy loam paleosol with a reddish tint, its red coloration due to higher soil iron oxidization than observed in upper unit and local surface soils. Unit 3 soil reddening was interpreted as due to higher-than-modern day annualized temperatures (cf., Taylor, 1982).
Unit 3 produced a Late Paleoindian projectile point base, 3 flake tools, and 181 waste flakes, all inter-mixed with cultural charcoal. Its basal contact with underlying unit 4 non-cultural deposits was gradational for ∼2 cm and showed initial unit 3 cultural deposition was directly on a once-exposed natural surface (unit 4). That former surface was identified as ground moraine formed from eroded Terminal Pleistocene (Pinedale IV) lateral moraine material left after the Roaring River Glacier receded into the Lawn Lake tarn. Its matrix was fine glacial silts mixed with sand, gravel, and rounded cobbles and small boulders.
Late Paleoindian occupation chronology
Two unit 3 AMS radiocarbon dates were derived from hearth charcoal and are among the oldest recorded for high-altitude archaeological sites in the region. The earliest date, 8,930 cal yr BP (Beta-144867), came from charcoal in a small basin hearth (Feature 1, Figure 11(a)). The date (and that of its associated archaeological assemblage) was supported by a Late Paleoindian Angostura projectile point base. Charcoal from an upper unit 3 level was recovered from a simple hearth (Feature 2, Figure 11(b)) positioned between two small glacial boulders and produced an AMS date of 7,960 cal yr BP (Beta-144869). Both upper and lower unit 3 dates fit within the region’s radiocarbon-based time range for the Late Paleoindian Period, i.e. ca. 10,700–7,800 cal yr BP (cf., Figure 8 above; Brunswig, 2014a, 2015b: Table 1; Chenault, 1999: 80–82; Pitblado, 1999a: 451–452; Pitblado, 2014; Figure 8 above). It is unlikely the dated hearths and their associated camp levels represent the only Late Paleoindian visits to the site since the vertical distribution of excavated artifacts throughout the unit 3 soil matrix suggests the presence of more than two short-term hearth-dated camps over their millennium-long stratigraphic and time separation.
(a) Profile illustration and labeled photo showing the stratigraphic context of the site’s older Late Paleoindian feature 1 hearth, sampling locations for its associated radiocarbon dates (RS), and pollen/soil sampling locations (PS); (b) illustration and profile photo of the later feature 2 hearth, its stratigraphic positioning between two glacial till boulders, and its radiocarbon date sampling location (RS).
Paleoindian cultural features
Feature 1 was initially visible in the dam spillway’s erosion wall after profile cleaning and was further exposed during excavation (Figure 11(a) and its accompanying profile photo). Feature 1 hearth morphology was clearly outlined in the erosion-cut terrace scarp profile as a 5-10 cm thick, 60 cm diameter, lenticular outline of dark charcoal-rich sediment, as shown in Figure 11(a). As excavation advanced, the profiled 7,960 cal yr BP hearth was exposed as an oval horizontal concentration of fire-burnt rock and diffuse charcoal. The earlier Late Paleoindian-age feature 2, dated at 8,930 cal BP, was a small 20–cm diameter hearth, placed between two small (30–40–cm diameter) boulders in mid-lower unit 3 (see Figure 11(b)).
Late Paleoindian lithic tools
The most significant (culturally diagnostic) artifact recovered from the Late Paleoindian deposits was an Angostura projectile point base (Figure 12), excavated 26 cm west and on the same level as the earliest Feature 2 hearth (Brunswig, 2001b; Pitblado, 1999b, 2014: 315–318).
Late Paleoindian Angostura projectile point base from lower unit 3.
The projectile point’s associated hearth date of 8,930 cal BP is consistent with its regional Late Paleoindian type time range of ca. 10,690 to 7,840 cal yr BP (cf., Pitblado, 1999a:Table 4.12, 1999b: 65–66, 2014). As shown in Figure 12, the point base has a lenticular cross-section, a lightly indented base, and lightly ground base and lower blade sections. General flaking traits and its surviving shape suggests the base came from an Angostura projectile point for which examples are known from other surface and excavated contexts in Colorado's Southern Rocky Mountains (Kornfeld and Frison, 2000: Figure 6; Pitblado, 1999a: Figures 4.32a, e, 6.14a, 1999b:Figure 1, 65–66, 2000: Figures 4.12a, b, 141–145, 2000: Figures 4.12a, b, 2014: Figures 10.1, 10.2). Macroscopic and microscopic inspection of the point’s quartzite material, its comparison with lithic reference collection samples, and the use of short and long wave ultraviolet analysis showed its parent material was Dakota Orthoquartzite. The closest documented prehistoric quarry sources for Dakota Orthoquartzite are Windy Ridge (83 km west) and Williams Fork Reservoir (69 km southwest), both located in the adjacent interior mountain basin valley of Middle Park west of the park (Bamforth, 2006; Benedict, 2000: 50–51; Black, 2000: 136, 144).
The only other stone tools recovered from Late Paleoindian occupation levels were three informal (expedient) tools made from secondary flakes. One was a brownish gray Kremmling chert flake knife with edge “nibbling” and polish, its material widely available from primary and secondary lithic sources within 40–km of Lawn Lake in the Middle Park and North Park valleys west and northwest of the park (Metcalf et al., 1991; Miller, 2010: 593). The knife was recovered mid-way in the Late Paleoindian deposits. The remaining tools, recovered from lower Late Paleoindian levels, were a large oval informal tool flake knife-scraper with edge-wear smoothing and microscopic step (hard cutting) fractures and a small flake knife with fine edge flake scarring and edge wear polish. Both were made of yellowish-brown Morrison quartzite, documented from a prehistoric quarry site in the Upper Laramie River Valley 67-km northwest of Lawn Lake (Brechtel, 2010).
Stone tool debitage and past hunter-gatherer behavior
Analysis of stone tool flaking debris (i.e. debitage) potentially informs us about interrelationships of stone tool assemblages and hunter-gatherer behavioral patterns and subsistence systems (cf., Andrefsky, 2001, 2005, 2009; Brunswig and Diggs, 2014: 77–78; Cotterel and Kamminga, 1987; Clark and Barton, 2017; Shott, 1986, 1994; Stahle and Dunn, 1982). Inferences about past human economic and social systems, embedded in recent decades of theoretical, ethnographic analogy, and experimental lithic artifacts research literature, are frequently derived from statistical, morphological, and material analysis of site assemblage flaked tool functional types and lithic waste (debitage) which can then be proposed to reveal insights into: 1) forager-collector oriented subsistence strategies based on the presence, absence, or percentages of tool debris categories presumed associated to past populations’ engagement in primary (direct manufacture) through tertiary (re-tooling) stone tool manufacturing and refurbishment activities, and 2) evidence of once-existing seasonal migration patterns based on geographic source locations and lithic tool materials transport.
Over the past few decades, archaeologists have frequently engaged in identifying archaeological clues believed related to hunter-gatherer subsistence strategies labeled as forager (characterized by frequent residential mobility with large numbers of shorter-term procurement camps), collector (less-frequent residential mobility centered in fewer long-term residency [base] camps centered on a network of logistically planned outlier procurement and processing camps), or, often identified recently, seasonal combinations of the two systems designed to maximize economic returns in variably productive (e.g., patchy) natural resource environments (Binford, 1977, 1980; Bicho and Cascalheira 2020; Brunswig, 2015b: 91–95; Clark and Barton, 2017; Kelly, 1992, 1995, 1998, 2013; Kuhn, 1989, 1994).
An often-utilized means of inferring forager- or collector- oriented subsistence behavior involves the analysis of stone tools and their lithic waste (debitage), produced by a broadly common and well-documented manufacturing process (cf., Ahler, 1989; Andrefsky, 2009; Clark and Barton, 2017; Pecora, 2001). Although there are multiple models which describe progressive sequences of stone tool manufacture, a simplified linear model was chosen for use here to categorize recovered Lawn Lake Late Paleoindian debitage, involving primary, secondary and tertiary stages (or subsystems) tool production and refurbishment (cf., Brunswig, 2020: 125–127; Flenniken, 1984; Flenniken and White, 1985; Magne, 1989; Magne and Pokotylo, 1981; Yerkes and Kardulias, 1993: 92–99, Figure 1). Broadly defined, the primary manufacturing stage consists of initial raw material selection and pre-tool core preparation which usually produces four categories of waste flakes, primary flakes with cortex, secondary flakes with cortex, and shatter flakes. Primary cortex flakes have visible sections, usually on one surface, of original weathered rock (cortex) along with an interior freshly decorticated surface with percussion flake depressions, usually visible with circular concentric lines (“bulbs” of percussion). The secondary stage of tool production involves further tool core (or large primary or secondary flake) reduction to shape desired functional tool types, e.g., projectile points, scrapers, knives, etc. The process may produce secondary flakes without cortex, small often petraloid (oval) tertiary flakes detached by precise soft-hammer or pressure flaking, and, usually involved in work edge-sharpening, and often jagged, splintery shatter flakes. Tertiary stage tool reduction is most usually involved in edge sharpening original formal tools or re-sharpening or re-working second or third generation formal or informal tools once they’ve been dulled by use or breakage. Small secondary, tertiary, and shatter flakes are often associated with tertiary stage tool manufacturing. Multiple methods have been developed to identify and quantitatively analyze waste flakes belonging to experimentally, ethnographically, and field-laboratory tested tool manufacturing stages and hypothesize their results to infer the existence of past forager, collector, or seasonally alternating (mixed) forager-collector subsistence systems (Carr and Bradbury, 2011; Kelly, 1994).
A simplified debitage analysis approach was used for this article to generally characterize the Lawn Lake’s Late Paleoindian debitage sub-assemblage. That approach intentionally doesn’t involve complex tool and individual flake trait analyses but specifically focuses on defining and enumerating general flake types and material origin sources. The overall goal was to only reconstruct generalized tool manufacturing and source materials patterns through time which were most likely to mirror task activities and occupation durations. Recovered flakes were analyzed with visual (macroscopic and microscopic) analysis and mechanical screening debitage into size (aggregate size grade) and shape (flake type morphology) classes broadly produced within one or more of the three tool manufacturing stages (cf., Ahler, 1989; Magne and Pokotylo, 1981; Patterson and Sollberger, 1978; Rinehart 2008; Shott, 1994; Stahle and Dunn, 1982). For the purposes of this article, identification and tabulation of flake types from the relatively small Lawn Lake Late Paleoindian debitage sub-assemblage were intended to identify a generalized pattern of tool use and manufacturing or refurbishment for the inference of overall site activity function and relative length of residence. For instance, if the debitage pattern included a minimal or complete absence of primary flakes, then early stage tool production from local lithic materials could be eliminated as a site activity and not part of its intended function. The absence of locally available tool lithic materials in the site’s lithic assemblage would also help explain the absence of primary stage tool production.
Table 2 summarizes the results of Lawn Lake’s Paleoindian debitage flake class, material type groups, and geographic source analysis. In all, 182 flakes were recovered in situ within 5–cm increments of the site’s excavated Late Paleoindian deposits. Although flake quantities were statistically small, they also were extracted from a limited volume, 3.75 m3, of excavated Late Paleoindian deposits.
The first three columns in Table 2 show numbers and percentages of secondary, tertiary, and shatter flakes for upper (∼top 8 cm) and lower (∼bottom 7 cm) levels (of Lawn Lake’s Late Paleoindian stratigraphic unit (15 cm). Individually and collectively, those levels are virtually identical, with secondary flakes at 43.9% (upper) and 41.9% (lower), tertiary flakes at 47.5% (upper) and 44.2% (lower), and shatter flakes at 8.6% (upper) and 11.6% (lower). The complete absence of primary and secondary flakes with cortex suggest secondary and tertiary (tool reworking and refurbishment) stage repair, remanufacturing, and tool edge sharpening (micro pressure flaking) were primary lithic reduction activities. It is also possible that some new tools were being produced from use-exhausted or broken tools.
Table 2 columns 4, 5, and 6 list data on tool debitage rock types and material (geographic-geologic) sources. The predominant material type was fine-quality chert (87.2%), followed by petrified wood (11.1%), fine-quartzite (1.7%), and white quartz (1.0%). The high percentage of fine-quality chert use is consistent with a mobile, summer-based forager (frequent mobility) oriented Paleoindian lifestyle in Colorado’s high mountains, with finished tools and pre-forms brought to the site from more distant geological sources. Two remaining rock types, petrified wood and quartzite, presented fine-grained alternatives to chert with excellent flaking qualities for tool refurbishment and remanufacturing, also suitable for producing second and third order tool types from broken or heavily worn first-order tools. The limited presence of quartz (1%), a less easily flaked but durable tool material, available in the site vicinity, presented an easily acquired resource for heavier-duty informal task tools, such as hide-scrapers.
All Late Paleoindian unit debitage was identified to known geographic-geologic origin sources (Table 2). Lawn Lake outward radii travel distances were calculated to known primary (in situ formations) and secondary (e.g., outwash and stream borne lag deposits) sources. All Lawn Lake lithic artifacts, including debitage materials, were analyzed using macroscopic, microscopic, and ultraviolet florescence methods, and source-identified using a reference collection of raw lithic materials from more than thirty quarries and geologic formations in the western and west-central United States. With the exception of local park obsidian (not recovered from the Lawn Lake assemblage) and volcanically derived jasper, rhyolites, and basalts, regionally sourced high-quality tool materials such as chert, petrified wood, and quartzite from the site’s Late Paleoindian deposits are accessible outside a 20–km walking radius west and north of the site (Figure 13). While Colorado Front Range foothills and eastern plains east of the park contain a variety of (mainly ancient mountain outwash river deposits) chert, petrified wood, and quartzites, none of the site’s debitage was identified as coming from those eastern sources.
Map of local (within 75-km) and non-local (beyond 75-km) Late Paleoindian debitage sources with probable transport routes of those materials to Lawn Lake shown by arrows.
Lithic sources for Lawn Lake's Late Paleoindian deposit debitage assemblage (Figure 13) came from both local (defined here as occurring within a 3 to 4-day walking radius of 75–km, Brunswig and Diggs, 2014: 78–79) and non-local (beyond a 75–km site-centered radius) in park-adjacent, interior mountain Middle Park, North Park, and upper Laramie River valleys (local) and south-eastern Wyoming geological formations (non-local) (Figure 13). There were differences between local and non-local material source frequencies in earlier and later occupation levels. In the early level, dated ca. 8,930 cal yr BP, use of local source materials was 62.8% versus 37.2% for non-local southeastern Wyoming Hartville chert (cf., Miller, 2010: 581–584) while in the later ca. 7,960 cal yr BP level, locally sourced debitage increased to 86.3% to a much lower representation, 13.7%, of non-locally acquired southeastern Wyoming chert. In addition to stone tools and debitage, a single sandstone metate (grinding stone) fragment was excavated near the lower unit hearth (Feature 2), documenting the presence of plant collecting and processing. The metate, made of Front Range Lyons sandstone, provides the only eastern foothills link to the site’s Late Paleoindian occupations (cf., Brunswig, 2015b: 90–91). Plant foods gathered and ground-stone processed during those occupations likely consisted of a variety of documented Native American food plants utilized in the region (cf., Benedict, 2007: 13–32). Based on their pollen presence in earlier discussed Lawn Lake Fen cores, available food products during Late Paleoindian occupations would have included sunflower (Helianthus sp.) seeds and species of Amaranthaceae, along with bistort (Bistort sp.) tubers.
Conclusion
The Lawn Lake archaeological site and its nearby glacial lake fen provide valuable insight into 9,000 years of high elevation, seasonal hunting camp adaptations, and evolving paleo-landscapes and paleoclimate histories within Rocky Mountain National Park and its region. Important (and still rare) clues to the site’s role in local and regional hunter-gatherer adaptations during its earliest Late Paleoindian period occupations, radiocarbon-dated between ca. 8,930 and 7,960 cal yr BP, are particularly significant. Lawn Lake, set in a protected stream terrace below a high subalpine forest lake, assured access to water, local woodland game species, summer food plants, firewood, and temporary shelter building materials. Evidence from the site highlights its role as an ephemeral staging and processing camp for hunting large seasonally migrating herbivores, such as elk and bighorn sheep, which grazed in up-valley alpine tundra and alpine-subalpine ecotone grasslands in summer. Its Paleoindian camp occupations, as well as succeeding cultural period occupations, extending to early Historic times, produced meat and hide processing tools, a seed-grinding stone fragment, and stone waste flakes with evidence of cutting and scraping tool sharpening and reworking. Geologic source analysis of tools and debitage shows that most Lawn Lake Late Paleoindian tool materials (80.8%) came from local (within 75–km) interior mountain valleys to the west and north and far fewer numbers of non-local (farther than 75–km) materials (19.2%) were transported from southeastern Wyoming to the north. Locations of local lithic sources and their most probable transport routes from the North Park, Middle Park, and upper Laramie valleys shown in Figure 13 illustrate likely pathways of seasonal transhumant migrations taken by Late Paleoindian hunting parties that used the camp as a short-term procurement base for summer hunting and gathering.
Lawn Lake’s stratified camps represent one of many park and adjacent mountain region sites believed to have contributed elements of an evolving seasonally-migratory subsistence system practiced in the Southern Rocky Mountains since onset of the Early Holocene (Benedict, 1990: 68–71, 1992: 1–14; Brunswig, 2004, 2005: 229–293, 2014a, 2015b; Brunswig et al., 2014b; Doerner and Brunswig, 2008; LaBelle, 2012; LaBelle and Pelton, 2013; Pitblado, 2000). The system, or iterations of it over time, is believed to have emphasized late fall-winter-early spring residential and foraging camps in the large interior valleys west of the park and late spring through summer transhumant migrations to adjacent high mountain hunting territories, returning to valley winter camps in the autumn. High mountain summer subsistence took advantage of game species, such as elk, bighorn sheep and limited numbers of bison, which migrated from interior valley winter ranges in the spring and, by mid-summer arrived in annually frequented tundra grazing ranges where they remained until returning to lower elevation valley winter ranges in early autumn. Native American high mountain hunting systems concentrated on tundra game in mid and late summer months, adjusting hunt strategies to maximize success within differing topographic settings. In several park locations, game drives in the form of low stone walls, cairn lines, and hunting blinds, were built and re-used over millennia to maneuver game animals into hunting ambushes. Planning and use of game drives were logistically organized and frequently involved systematic use of specialized support camps: hunting base camps at lower elevations on or near subalpine tree lines, the drives themselves usually situated on steep, often rugged, tundra (usually) mountain slopes, while early stage butchering and processing areas were situated nearby. Dismembered meat sections and detached hides were then taken from game drive processing areas for final processing (meat drying and smoking, hide finishing…) at lower level base camps. Grinding stone tools, often found at both game drives and base camps, show plant collection and processing often occurred in parallel with hunting. Other prehistoric hunting camps not directly associated with game drives, such as Lawn Lake, were used as procurement and processing camps to hunt and gather local resources and from which summer hunts in higher elevation tundra areas could be staged but which didn’t always require or facilitate the use of formal game drives. Open tundra and scattered stands of alpine shrubs and krummholz (dwarf fir and spruce thickets) in the upland Crystal Lake area would have been ideal hunting territory for Lawn Lake based hunter-gatherers. The presence of a grinding stone artifact in the site’s Late Paleoindian levels suggests that both men and women may have been migratory hunting-gathering party members.
Over nine millennia, the Lawn Lake site was periodically occupied as a summer hunter-gatherer camp. Its long-term persistent use underlines its logistical and strategic importance at an interface of high montane economic resources between subalpine forest (at the Lawn Lake site) and upland tundra (up-valley from Lawn Lake). Alpine grazing pastures for elk and bighorn sheep today are located less than an hour’s walk up-valley to Crystal Lake, making the site ideally suited as a forager procurement and processing camp, possibly based out of a longer-term base camp located at lower elevation (cf., Brunswig, 2015b: 49–52). Paleoenvironment conditions in the site area, based on sediment core data from a nearby fen and from similar studies at other park coring locations, provide a detailed record of local and regional climate patterns and ecological conditions during its Late Paleoindian occupation history.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Research at Lawn Lake and other localities in Rocky Mountain Park cited in this article were funded under the University of Northern Colorado’s Archaeological Surveys and Cultural Resource Inventory of Rocky Mountain National Park Project” cooperative agreement with the National Park Service, U.S. Department of the Interior (1998–2003). The project’s paleo-environment analyses, including those accomplished on Lawn Lake sediment cores, were enhanced with a magnetic susceptibility instrument grant from the University of Northern Colorado’s Office of Sponsored Programs.
