Journal of Paleolimnology **: *** ***, ****. ***

# **** ****** ******** **********. ******* in the Netherlands.

Late Pleistocene vegetation and climate in the southern Cascade Range and

the Modoc Plateau regionw

Katharine J. Hakala1,* and David P. Adam2

1

Department of Geology and Planetary Science, University of Pittsburgh, Pittsburgh, PA 15260, USA;

2

California Academy of Sciences, San Francisco, CA, USA; *Author for correspondence (e-mail:

abpj9e@r.postjobfree.com)

Received 15 August 2002; accepted in revised form 2 July 2003

Key words: Cascade Range, Paleoclimate, Paleolimnology, Palynology, Quaternary

Abstract

Pollen and sediment from Grass Lake, California provide a history of vegetation and climate in the southern

Cascade Range from 36 to 19 cal ka, revealing climate changes that led to the glacial advances recorded at

Upper Klamath Lake (Rosenbaum and Reynolds 2004a this issue). Variations in the percentages of conifer

and Artemisia (sagebrush) pollen at Grass Lake recorded shifts in vegetation that reflect changes in

precipitation. Between 36 and 34 cal ka, a progression from steppe to open pine forest to dense pine forest

indicates that precipitation increased. After 32 cal ka, the forest became more open and by 30 cal ka

sagebrush steppe surrounded the lake, implying that precipitation decreased. The area was arid for most

of the interval between 30 and 19 cal ka. Increases in conifer pollen recorded increases in precipitation from

21 through 19 cal ka, when open pine forest colonized the lake area. Throughout the period from 36 to

19 cal ka, centennial- to millennial-scale intervals with increased conifer pollen imply that the arid interval

was interrupted by periods of increased precipitation. Pollen data also provide evidence that the major

fluctuations in sand concentration in the Grass Lake core reflect temperature shifts. Changes in sediment

particle size are closely related to variations in pollen concentration and accumulation rate, which in turn

reflect changes in plant cover, implying that sand was deposited in the lake due to deflation of clay- and silt-

sized particles from sparsely-vegetated alpine areas of the watershed. Sand deposition increased as climatic

cooling led to reductions in the elevation of upper treeline and alpine conditions affected a larger part of the

watershed. There is no evidence of glaciation in the basin, but pollen data show the area was above upper

treeline during Cold Period III (34 32 cal ka), one of several very cold intervals. Vegetation decreased at

about 28 cal ka and remained sparse for at least 9000 years, implying that the climate became cooler and

remained cool until after 19 cal ka. Cold Period II developed at about 25 cal ka and terminated by 23 cal ka.

The Grass Lake watershed was again above upper treeline with the onset of Cold Period I, soon after 19

cal ka. Comparison of the Grass Lake record with those from Upper Klamath Lake, Oregon and Tulelake,

California suggests a persistent pattern of environmental changes in this time interval throughout the Modoc

Plateau region.

Introduction

w

This is the fifth in a series of eight papers published in this

special issue, resulting from paleoenvironmental studies in the

Upper Klamath Lake Basin. These studies were conducted by Many records describe the environmental history

the US Geological Survey and its collaborators as part of a

of the western United States during the period

paleoclimate research effort called the Correlation of Marine

following the last glacial maximum about 21 cal ka

and Terrestrial Records Project. Steven M. Colman served as

(Thompson et al. 1993). In contrast, little is known

guest editor of this special issue.

190

about the changes in regional vegetation and

climate that accompanied the development of ice

sheets and glaciers leading up to the last glacial

maximum.

Lack of information about several specific cli-

mate issues limits the present understanding of

the climate system of the western United States.

(1) Paleoclimate records from Greenland and the

North Atlantic show temperature oscillations

occurring at 1500-year intervals, the Dansgaard

Oeschger (D/O) cycles (Johnsen et al. 1992;

Dansgaard et al. 1993; Grootes and Stuiver 1997).

During the past 60 000 years, correlative tempera-

ture oscillations were recorded in the Pacific Ocean

near the southern California coast (Hendy and

Kennett 1999; Hendy et al. 2002). Similarly corre-

lated changes in moisture affected south-central

Oregon between 46 and 23 cal ka (Zic et al. 2002).

More information about the expression of these

cycles in western North America would clarify the

relationship between local, regional, and global

climate controls. (2) It is not clear why alpine gla-

ciers between 35 and 50 N near the Pacific coast

advanced specifically at about 31, 25, and 19 cal ka Figure 1. Ages, sedimentary units, and age depth curve for the

upper 18 m of core GL-2 from Grass Lake. Radiocarbon ages

(Phillips et al. 1996; Thackray 2001). The largest ice

were calibrated using CALIB 4.2 (Stuiver et al. 1998) and the

advances did not coincide with the 21.5 cal ka formula given by Bard (1998). Square 1910 clearcut date; filled

minimum in summer insolation at 40 N (Berger circles radiocarbon ages from sediment with >2500 pollen

and Loutre 1991; Berger 1992), nor did they follow grains/cm3; open circles radiocarbon ages from sediment with

the gradual increase in global ice volume, which

reached a maximum about 21 cal ka (Imbrie et al.

mountains from the glacial period to the present.

1984).

First, pollen from Caledonia Marsh reveals vegeta-

Vegetation and climate changes in the region of

tion changes near Upper Klamath Lake, Oregon

the southern Cascade Range and the Modoc

(Adam et al. 1995). A wide variety of paleoenvir-

Plateau are analyzed here in order to relate these

onmental data are available from Upper Klamath

changes to the regional environmental history and

Lake (Colman et al. 2004a this issue). Second, to

to suggest explanations for the inferred changes.

supplement the summary of pollen data from a 3-

To this end, a core from Grass Lake, California,

million-year core from Tulelake, California (Adam

provided a detailed record of conditions in the

et al. 1989), we describe further details of this pol-

southern Cascade Range from 36 to 19 cal ka

len record for the interval from 27 cal ka to the

(Figure 1, Colman et al. 2004a this issue). After

present. Comparisons among the records from

establishing a chronology for this record, we

Grass Lake, the Modoc Plateau region, and other

describe the core lithology, which is directly related

areas suggest regional patterns of environmental

to climate change. A zoned pollen diagram shows

change.

the pollen data, which are interpreted in terms of

vegetation types. We analyze the variations in sedi-

ment lithology and reconstruct a history of the

vegetation and climate in the Grass Lake area. Setting

Also included are pollen data from two nearby

The Grass Lake (Lat. 41 390 N, Long. 122 100 W,

areas, which show the broad-scale environmental

history of the western Modoc Plateau and adjacent elev. 1537 m) basin formed when Pleistocene High

191

Cascades volcanic deposits blocked a small valley forest extends to the summit of Goosenest

near the crest of the southern Cascade Range Volcano (elev. 2524 m), the highest point in the

(Williams 1949; Adam et al. 1994). No evidence basin.

of glaciation has been found in the watershed. At present, Juniperus occidentalis (western juni-

Grass Lake covers 5.6 km2 and receives drainage per) and Artemisia (sagebrush) form a woodland

from an area of 99 km2 (Best 1996). Surface flow is that abruptly replaces the pine forest below 1500 m

limited, and much of the lake originates as ground- elevation on slopes southwest of the watershed.

water from springs and seepage. The basin lacks a J. occidentalis probably colonized sagebrush com-

surface outlet, but lake water drains into the por- munities near Grass Lake between 1890 and 1910

ous volcanic bedrock (Wood 1960). (Young and Evans 1981; Miller and Rose 1995).

The climate at Grass Lake is strongly seasonal: Sagebrush steppe covers the lower slopes of the

winters are cold and wet; summers are dry and range at present and probably extended up to the

warm. At Mt. Hebron Ranger Station in Butte pine forest during the nineteenth century.

Valley (elev. 1295 m), 20 km northeast of Grass The study sites in the Modoc Plateau region

Lake, the annual average maximum temperature is (Upper Klamath Lake and Tulelake) are located

about 16.7 C; the average minimum temperature at lower elevations and are slightly warmer than

is about 1.6 C (written communication, US Grass Lake (Colman et al. 2004a this issue;

Forest Service). Lapse rates for the eastern Adam et al. 1989). The coring site at Caledonia

Marsh (Lat. 42 180 N, Long. 121 53 550 W,

slope of the southern Cascade Range imply that

temperatures at the lake are about 2 C lower elev. 1263 m) is surrounded by marsh vegetation,

(Major 1995). but ponderosa pine forest covers the surrounding

Storms originating over the Pacific Ocean supply slopes. Annual maximum temperatures at Tulelake

(Lat. 41 570 N, Long. 121 290 W, 1231 m elev.)

most of the precipitation, which averages 31.6 cm

average 16.7 C; average minimum temperatures

annually in Butte Valley (Climate Normals for the

average 0.7 C (Climate Normals for the United

United States, 1983). Elevational precipitation gra-

dients measured in eastern California (Major 1995) States, 1983). Of the three study sites, the Tulelake

and the presence of the steppe-forest ecotone near area has the lowest average annual precipitation,

the lake suggest that annual precipitation at Grass 26.77 cm, and the most xeric natural vegetation,

Lake is about 35 cm (Franklin and Dyrness 1973). sagebrush scrub (Kuchler 1977; Adam et al. 1989).

About 70% of the precipitation falls between

October and March, largely as snow. As a result,

the lake fills during the winter and spring. During Methods

the summer months the Pacific Subtropical High

(PSH) deflects storm systems, reducing precipita- In 1991, the US Geological Survey collected a

tion, and the lake level declines. By July emergent 30.62-m long sediment core (GL-2) from the cen-

Juncus balticus (Baltic rush) conceals the lake tral part of Grass Lake (Adam et al. 1994). The

surface. original drilling depths given by Adam et al. have

Terrestrial vegetation near the lake has been been adjusted to reflect irregularities in the drilling

altered since 1900 by logging and other distur- process (Hakala 1999).

bances. The pre-settlement forest near Grass Lake Samples were taken at intervals of 10 50 cm

probably resembled the nineteenth-century forest between 0 and 8.56 m depth; additional samples

growing at similar elevations (1220 1680 m) on were collected from the core at intervals of 1 3 cm

Mt. Shasta (Merriam 1899). There, Pinus ponder- from 5.55 to 7.48 and 8.56 to 9.25 m depth, as

osa (ponderosa pine) dominated an open forest, preliminary pollen analysis and radiocarbon

but mesic conifers such as Abies concolor (white ages indicated these intervals contained significant

fir) grew in moist sites. This vegetation charac- amounts of glacial-age pollen. A single sample was

terizes the eastside pine phase of the Mixed obtained from bagged sediment originally located

Conifer Forest of Griffin (1967), which grows on between 6.94 and 7.01 m depth. The stratigraphic

northeastern California sites receiving 35 75 cm location of the sample examined is placed at 6.98 m,

annual precipitation (Smith 1994). A subalpine although the exact location is not known. Three

192

samples from the lower core (13.17 13.19 and were collected from Caledonia Marsh core CM2

14.16 14.21 m depth) were also examined. Most (Colman et al. 2004a this issue) at intervals of 50

samples contained 1 2 cm3 of sediment, but up to 8 cm or less (Adam et al. 1995). Values for all pollen

cm3 was used for samples from sandy sediment. types, including Cyperaceae, were calculated as a

Samples were prepared using standard palynolo- percentage of the total pollen sum. Samples for

gical techniques (Cwynar et al. 1979; Faegri and pollen analysis were collected from Tulelake core

Iversen 1989). Prior to chemical processing, a 1 at intervals of 10 20 cm between 0 and 6.30 m

known amount of Lycopodium spores was added depth. Three additional samples were collected at

to each sample to standardize the pollen concentra- irregular intervals between 6.36 and 9.17 m depth.

tion; sand-sized particles were removed from all For the Tulelake record, pollen values for the taxon

samples by swirling to minimize the formation of Cyperaceae were calculated as percentages of

mineral precipitates. In addition, samples were the total pollen sum; pollen values for other taxa

treated with HCl to eliminate precipitate that were calculated as percentages of the terrestrial

formed during the standard HF treatment. A pollen sum.

total of 199 samples were processed, mounted in

silicone oil, and examined at magnifications of The Grass Lake record

400 and 1000.

At least 500 terrestrial pollen and spores were Chronology

counted for each stratigraphic level, excluding

(1) eight samples with deteriorated pollen from The sediment chronology was determined through

the uppermost meter of the core and (2) 29 samples a combination of methods, including AMS radio-

with very low pollen concentrations ( 100 grains were (Figure 1). Anomalies in pollen and sediment

counted from one stratigraphic level. The pollen deposited above 1.22 m depth indicate that this

sum used for calculation of terrestrial pollen per- section of core GL-2 was deposited after the

centages included all terrestrial taxa. Percentages Weed Lumber Company clearcut the ponderosa

of aquatic/riparian taxa were calculated as a per- pine forest surrounding Grass Lake about 1910 AD

centage of both terrestrial and aquatic/riparian (Shoup 1981; Hakala 1999). This implies that the

pollen and spores. sediment at 1.22 m depth was deposited about 1910.

Cyperaceae (sedge) pollen can represent vegeta- AMS and conventional radiocarbon ages were

tion of marshes or cool steppes, but the Cyperaceae obtained from bulk sediment (Table 1). Ages

pollen at Grass Lake is included in the aquatic/ 1 g/cm3), which contains little total carbon

analysis.

The reference collection at the University of (0.060 0.064%), much of which is inorganic (17

California at Berkeley, several private reference 48%) (Best 1996). Such sediment does not provide

collections, and standard pollen manuals were used reliable radiocarbon ages. In contrast, inorganic

for pollen identification. In the following discus- carbon comprised only 0.2% of the total carbon in

sion, scientific names are used for pollen taxa but a sample composed of sandy mud with >52 000

pollen grains/cm3 and 2.443% total carbon.

common names refer to the plants they represent.

Adam et al. (1989) describe the palynological Two identified tephra layers provide ages for

methods used in the Caledonia Marsh and the lower sections of core GL-2 (A. Sarna-Wojcicki,

Tule Lake studies. Samples for pollen analysis written communication, 1997). The averaged age for

193

Table 1. Radiocarbon ages and tephra information, Grass Lake.

Uncalibrated 14C Lab No.d/

Calibrated age

Calibrationa Tephrae

Core depth (m) age (year BP) (calendar year BP) Type of date

AMSb

6640+/ 70 7520+/ 50

1.41 1.46 CALIB 4.2 WW353

CAMS14786

AMSb

3640+/ 70 3930+/ 70

2.91 2.94 CALIB 4.2 WW354

CAMS14787

AMSc

18 980+/ 80 22 530+/ 360

3.23 3.25 CALIB 4.2 Beta87415

AMSb

15 310+/ 90 18 300+/ 290

3.67 3.72 CALIB 4.2 WW525

CAMS22094

AMSc

23 510+/ 130 27 600+/ 130

3.82 3.85 4090 year Beta87416

AMSb

12 810+/ 110 15 450+/ 290

4.08 4.11 CALIB 4.2 WW272

CAMS12667

AMSb

16 160+/ 170 19 280+/ 350

4.84 4.87 CALIB 4.2 WW271

CAMS12666

AMSb

15 680+/ 90 18 720+/ 300

5.54 5.59 CALIB 4.2 WW524

CAMS22093

AMSc

24 720+/ 160 29 030+/ 160

5.62 5.64 4310 year Beta90220

conventionalc

15 710+/ 670 18 760+/ 820

5.70 5.74 CALIB 4.2 USGS3342

AMSc

19 840+/ 80 23 510+/ 380

6.11 6.14 CALIB 4.2 Beta87417

AMSb

22 990+/ 410 27 050+/ 410

6.29 6.31 4060 year WW355

AMSc

27 690+/ 190 32 390+/ 190

6.56 6.59 4700 year Beta87418

conventionalc

25 640+/ 580 30 080 +/ 580

6.73 6.77 4440 year USGS3329

AMSc

28 010+/ 200 32 740+/ 200

7.09 7.12 4730 year Beta87419

GL11069f

7.61 7.63 Insufficient carbon

AMSc

24 550+/ 150 28 820+/ 150

7.64 7.66 4330 year Beta90221

AMSb

29 490+/ 450 34 390+/ 450

8.74 8.76 4900 year WW527

AMSc

29 080+/ 200 33 930+/ 400

8.76 8.79 4850 year Beta90222

Olema Ashg

13.88 13.92 55 000

71 000+/ 6000

16.38 16.40 Pumice Castle

Tephrah

72 000+/ 7000

16.38 16.40 Pumice Castle

Tephrah

a

Radiocarbon age calibration using CALIB 4.2 (Stuiver et al. 1998) for ages 30 14C ka Lake lack centennial-scale accuracy, although they

have large errors and are difficult to convert to are expressed to the nearest century to furnish

calendar year ages, due to the large fluctuations in information about the sequence and duration of

atmospheric 14C between 30 and 44 14C ka (Beck events.

et al. 2001). In addition, ages from sediment below

9.00 m depth were irregular. For these reasons we

constructed the age model using linear interpola- Lithology

tion between the age at 8.79 m depth and the 55-ka

Olema ash at 13.90 m depth. This is consistent Descriptions of core lithology are based on Adam

with Bischoff et al. (1993), who concluded that et al. (1994), supplemented by laboratory data.

195

Sections of the core were classified into units based vegetation from the pollen record. Surface samples

on the dominant sediment lithology, which varied from the arid western United States show that

from 95% clay plus silt in the Silty Mud Unit to certain precipitation regimes are associated with

98% sand in Sand Unit B (Figure 1) (Best 1996). certain percentages of Artemisia and Pinus pollen

The sand fraction, obtained by swirling, consists in all but the most extreme environments (Davis

of inorganic particles >100 m in diameter. Most of 1995). Pinus pollen levels exceed 40% only in areas

these particles are composed of the dense volcanic receiving >300 mm annual precipitation. Pinus

minerals amphibole, pyroxene, and olivine, which percentages increase as precipitation increases,

have specific gravities of 3.0 3.6 (H. Prellwitz, and Pinus levels >65% are found only in areas

written communication). Samples with >1000 pollen with >500 mm annual precipitation. In contrast,

grains/cm3 also contain $15% plagioclase and Artemisia pollen is mainly deposited in areas

$5% quartz, which have specific gravities of 2.6 receiving 90% of the pol-

pollen. This is consistent with a Pinus value of 53%

len in most of the Grass Lake record (Figure 3).

for a similar lower montane forest in the Sierra

Pollen concentrations vary over a wide range (7 to

76 300 gr/cm3), and different pollen concentrations Nevada (Anderson and Davis 1988). Based on

these values, most intervals with >50% Pinus were

characterize various sections of the core. Pollen

interpreted as eastside pine forest. Higher values of

from Juncus balticus was not found in fossil samples

Pinus (to 75%) and Abies (to 3%) pollen characterize

although it dominates the present lake flora. This

several intervals. Modern eastside pine forests in

is not unexpected, as pollen from the Juncaceae

moist sites have more overstory tree cover and

normally disintegrates during sample preparation

more fir than eastside pine forests in relatively dry

(Moore et al. 1991). Long-distance pollen transport

sites (Smith 1994), suggesting that increases in Pinus

is probably responsible for much of the Abies and

and Abies represent increases in forest density due to

Tsuga pollen in the sediment (Davis 1995).

increased precipitation. In addition, the pine forest

Significant amounts of Salix (willow) pollen

(1674 and 158 gr/cm3) occur in two samples from may have expanded into lower-elevation steppe

areas as precipitation increased.

Sand Unit B (7.40 7.42 and 7.50 7.52 m depth,

We used pollen levels of 35% Artemisia and 50%

respectively) (Figure 3, where Salix is plotted as a

Pinus to approximate the threshold between sage-

percentage of the total pollen sum). Salix levels

brush steppe and pine forest, based on surface

are low in the sediment adjacent to these samples,

sample data from transitional sites (Anderson and

indicating that there was little sediment mixing.

Davis 1988; Fall 1992). Artemisia averages 39% in

The restricted distribution of the Salix pollen is

surface samples from steppe areas of the Great

probably due to deposition in sand composed of

Basin (Davis 1995), and Artemisia values between

dense volcanic minerals, which resist resuspension

35 and 42% in core GL-2 imply that sagebrush

by lake currents. The distinctness of these pollen

steppe surrounded the lake. Increased percentages

layers implies that sandy sections of the core con-

of Poaceae (12 19%) in sediment with low Pinus

tain a high-resolution record of vegetation change.

values (1.0 g sand/cm3) of the Grass tions of the core, together with high levels of Pinus

Lake core (Figure 4f, g). Several factors could (61 71%) and the lowest Artemisia percentages

explain this. First, pollen degradation; however, counted in pre-clearcut samples (16 20%). These

few pollen grains are corroded and Pinus grains samples suggest forests similar to modern eastside

are not exceptionally broken in the low- pine communities that have currants as a codomi-

pollen sediment (Figure 3), so it is unlikely that nant in a shrub understory (Smith 1994). These

degenerative processes significantly affected the modern forests contain little sagebrush, suggest-

pollen in these layers. ing that the Artemisia pollen in these fossil

Second, rapid sedimentation; however, changes samples represents low-elevation sagebrush steppe.

in pollen percentages within the sand units indicate Assuming that sagebrush steppe contributes 20%

that the vegetation changed while the sediment was of the pollen to samples representing pine forest

(PAR of 900 gr/cm2 per year), we estimate that

being deposited, implying that low pollen con-

centrations in these layers were not due to rapid low-elevation steppe vegetation contributed about

180 gr/cm2 per year to the core sediment when the

depositional events (Figure 5). Relatively rapid

lake area was forested. The PAR of 70 gr/cm2 per

deposition of inorganic particles in Sand Unit B

(Figure 1) would have diluted the pollen deposits, year for Sand Unit B implies that most of the pollen

reducing the concentration of pollen in this section in this unit originated in low-elevation areas. Very

of the core. In order to compare average annual little pollen was produced in the watershed, and

rates of pollen deposition independent of varia- pollen production was probably reduced in low-

tions in sand deposition, we calculated pollen accu- elevation steppe areas. Consistent with this, Sand B

samples with >1500 pollen gr/cm3 have high levels

mulation rates (PARs) for different sections of

the core. For example, a sedimentation rate of of Pinus (>50%), suggesting that pine trees were

200 cm/kyr was used to calculate a PAR of 70 gr/cm2 growing in the watershed; most samples with

0.4 g/cm3), sug-

the slopes surrounding the lake when the climate

was warm and moist enough for trees but cooler gesting an upper treeline was present on the slopes

than the present. A subalpine forest would have above the lake (Significance of lithological changes,

grown below upper treeline, with such conifers below). Accordingly, in the Grass Lake record

as P. albicaulis (whitebark pine), P. contorta Pinus pollen levels of 45 60% combined with

200

1500 16 000 pollen gr/cm3 were considered evi- (Figure 4d, g) (Davis 1995). Clay-silt sediment is

dence of subalpine forest. also associated with both high (30 29 cal ka) and

The Grass Lake record was divided into pollen low (31.8 30.9 cal ka) Artemisia levels. Similarly,

zones based on a constrained incremental sum of both sandy deposits and clay-silt deposits show

squares (CONISS) method using TILIA software variations in rates of accumulation (Figures 1

(Grimm 1988, 1993) (Figure 3). The data asso- and 5g). The lack of correlation between inferred

ciated with each pollen zone is described in Table 2. precipitation and sedimentation rates, or between

sediment particle size and sedimentation rates in

core GL-2 suggests that variations in the amount

of surface flow within the watershed were not

Discussion of the Grass Lake record

responsible for the major shifts in core lithology.

Weathering can affect sediment lithology, and

Significance of lithological changes

weathering rates can vary with climate. However,

No drainage channels enter the lake near the coring

the abundance of relatively unstable minerals

site (Adam et al. 1994), so it is unlikely that deltaic

(compared to deficiency of quartz and feldspar) in

shifts were responsible for variations in sediment

low-pollen sand argues against a major weathering

particle size. The rate of sand deposition varied

effect. In addition, the rapid shifts between differ-

from 4.9 cm/kyr at 30.1 cal ka (Sandy Mud Unit;

ent lithologies in core GL-2 imply that changes in

Figure 1) to 16.2 cm/kyr at 35.2 cal ka (Silty-Sandy

the proportions of sand occurred within decades or

Mud Unit) and 184.2 cm/kyr at 33.4 cal ka (Sand

centuries, whereas chemical weathering requires

Unit B) (Best 1996). These rates are independent of

thousands of years to produce visible changes in

changes in the accumulation rate for clay- or silt-

rocks (Birkeland 1999).

sized biogenic sediment, implying that variations in

Sand deposition increased when pollen con-

biogenic dilution do not explain the large varia-

centrations were low (Figure 4f, g), and sand

tions in sand concentration in core GL-2.

concentrations decreased as pollen accumulation

Sediment deposited from 36.0 to 34.1 cal ka is

increased. Sand Unit B has a PAR of 70 gr/cm2

the most variable in the core, as layers of sand

per year or less, and a median 1.27 g sand/cm3, the

alternate with muddy sand and sandy mud (0.36

Muddy Sand Unit has a PAR of 300 gr/cm2 per

to 1.28 grams sand/cm3; Figure 4g). During the

year and a median 0.55 g sand/cm3 (5.60 6.12 m

same interval conifer pollen percentages gradually

depth), and the Sandy Mud Unit has a PAR of

increased and remained at high levels. Abrupt con-

900 gr/cm2 per year and a median 0.10 g sand/cm3

ifer pollen changes coinciding with the shifts in

sand concentration (in g/cm3) would suggest breaks (6.74 7.11 m depth). These relations indicate that

sand deposition was closely related to rates of

in deposition or irregular sedimentary events, but

pollen deposition, which in turn reflected changes

the observed gradual increases in conifer pollen

in the distance of upper treeline, which varies with

suggest deposition was nearly continuous and

elevation near Grass Lake. In temperate zones of

undisturbed.

the Northern Hemisphere this elevation is mainly

Changes in pollen percentage changes within the

controlled by summer temperature (Troll 1973),

sand units show that vegetation changed as these

suggesting that rates of sand deposition increased

layers were deposited, implying that they do not

as summer temperatures decreased and vegetative

represent rapid depositional events such as slides

cover decreased.

or mass flows. The coring site was located in the

Most increases in pollen concentration coincide

central part of the lake (Adam et al. 1994), mini-

with decreases in sand concentration, and most

mizing the effect of shoreline changes on sediment

decreases in pollen concentration coincide with

lithology.

increases in sand concentration (Figure 4f, g).

Changes in precipitation and runoff affect the

However, at several levels (7.12, 8.65, and 8.85 m

amount of energy available to transport sediment.

depth) pollen concentrations increase in sandy

In core GL-2 sediment sandy sediment is associated

sediment 1.5 2.0 cm below increases in clay and

with both high (22.9 20.8 cal ka) and low (35.1

silt (Figure 5). Similarly, a decrease in pollen con-

34.1 cal ka) Artemisia levels, suggesting deposition

centration in muddy sediment preceded a shift to

during both relatively wet and dry intervals

201

Table 2. Pollen zones: description, vegetation type, and inferred climate.

Sedimentary

Pollen zone unit lithology

age (AD or cal (median sand Vegetation Inferred

ka) depth (m) concentration) Description type climate

GL-1a Silty Mud The base of the subzone provided two samples Disturbed Moderate

1910 1991 AD Unit layers of with very low Pinus percentages (32%). eastside pine temperature,

0 1.22 m (core peat, sand, and Younger sediments contain high but forest wet

break 1.22 1.27 m) silty mud fluctuating concentrations of deteriorated

(0.08 gm/cc) Pinus (55 90%), low pollen concentrations

(300 13 700 gr/cc), and abundant Cyperaceae.

GL-1b Silty Mud Unit Moderate Pinus percentages (median 55%) Eastside pine Moderate

1910 AD ? silty mud and moderately high pollen concentrations forest temperature,

1.27 3.00 m (0.00 gm /cc) (median 26 000 gr/cc) characterize this wet

(hiatus 3.00 m) subzone.

GL-2 Sand Unit A Pollen concentrations are very low (110 2400 Sparse alpine Cold, wet

? 18.9 cal ka very fine to gr/cc) in most of this zone. Artemisia herbs;

3.00 5.61 m coarse sand percentages are high (median 38%) and half of subalpine

(1.24 gm/cc) the samples contain significant amounts of forest at 5.61

pollen from Poaceae (grasses) (11 18%). Pinus 5.68 m depth

values are low (27 42%). Less Artemisia (29

33%), more Pinus (median 52%) and more

pollen (1200 6100 gr/cc) were dep

Contact this candidate