holocene paleoclimate change in the antarctic peninsula ... · from the diatom, sedimentary and...

19
Holocene paleoclimate change in the Antarctic Peninsula: evidence from the diatom, sedimentary and geochemical record F. Taylor a, * , J. Whitehead b , E. Domack a a Department of Geology, Hamilton College, Clinton, NY 13323, USA b Antarctic Co-operative Research Centre/Institute of Antarctic and Southern Ocean Studies, GPO Box 252-80, Hobart 7001, Tasmania, Australia Received 7 December 1999; revised 2 July 2000; accepted 10 July 2000 Abstract Holocene, marine deposition in Lallemand Fjord, Antarctic Peninsula, is reinterpreted using statistical analyses (cluster analysis, analysis of variance, nonmetric multidimensional scaling and multiple regression) to compare diatom assemblages and the primary sedimentological proxies. The assemblages have been deposited in a variable sea ice zone over the last ca. 10,500 yr BP in response to a climate change. In the Late Pleistocene/early Holocene (10,580–7890 yr BP), a sea ice diatom assemblage was deposited in the presence of a retreating ice shelf at the head of the fjord. In the mid Holocene (7890–3850 yr BP), an open water assemblage was deposited and sea ice cover was at a minimum. We associate the assemblage with climatic warming, which characterizes much of the Antarctic Peninsula during this time. A second sea ice assemblage, different from that deposited in the early Holocene, has been deposited in Lallemand Fjord since the late Holocene (,3850 yr BP). The assemblage reflects Neoglacial cooling, an increase in sea ice extent and/or an advance of the Mu ¨ ller Ice Shelf. q 2001 Elsevier Science B.V. All rights reserved. Keywords: Holocene; Antarctica; paleoclimate; diatoms; sedimentology; statistical analysis 1. Introduction Sedimentary diatom assemblages have been used successfully in numerous studies as a proxy for Antarctic marine paleo-reconstructions (e.g. Truesdale and Kellogg, 1979; Pichon et al., 1987; Leventer et al., 1996; Cunningham et al., 1999). Many of these studies include statistical techniques to reconstruct Quaternary glacial history, but few incorporate a subjective, multi-disciplinary approach. In the present study, we incorporate diatom, sedimentological and geochemical data with classification and indirect ordination analyses to interpret the Holocene paleo- environment of Lallemand Fjord on the western Antarctic Peninsula. Climate records from both Hemispheres demonstrate increasingly that the Holocene (,11,500 yr BP, after Roberts, 1998) has been a period of rapid and variable climate change (Domack and Mayewski, 1999; Rosqvist et al., 1999). Marine sediment cores from the Antarctic Peninsula revealed multi-century and millennial-scale variations in primary production (Leventer et al., 1996; Shevenell et al., 1996; Marine Micropaleontology 41 (2001) 25–43 0377-8398/01/$ - see front matter q 2001 Elsevier Science B.V. All rights reserved. PII: S0377-8398(00)00049-9 www.elsevier.nl/locate/marmicro * Corresponding address. Antarctic Co-operative Research Centre/Institute of Antarctic and Southern Ocean Studies, GPO Box 252-80, Hobart 7001, Tasmania, Australia. Tel.: 161-3- 6226-7888; fax: 161-3-6226-2973. E-mail addresses: [email protected] (F. Taylor), [email protected] (J. Whitehead), [email protected] (E. Domack).

Upload: phunghanh

Post on 05-Jun-2018

214 views

Category:

Documents


0 download

TRANSCRIPT

Holocene paleoclimate change in the Antarctic Peninsula: evidencefrom the diatom, sedimentary and geochemical record

F. Taylora,*, J. Whiteheadb, E. Domacka

aDepartment of Geology, Hamilton College, Clinton, NY 13323, USAbAntarctic Co-operative Research Centre/Institute of Antarctic and Southern Ocean Studies, GPO Box 252-80, Hobart 7001,

Tasmania, Australia

Received 7 December 1999; revised 2 July 2000; accepted 10 July 2000

Abstract

Holocene, marine deposition in Lallemand Fjord, Antarctic Peninsula, is reinterpreted using statistical analyses (cluster

analysis, analysis of variance, nonmetric multidimensional scaling and multiple regression) to compare diatom assemblages

and the primary sedimentological proxies. The assemblages have been deposited in a variable sea ice zone over the last ca.

10,500 yr BP in response to a climate change. In the Late Pleistocene/early Holocene (10,580±7890 yr BP), a sea ice diatom

assemblage was deposited in the presence of a retreating ice shelf at the head of the fjord. In the mid Holocene (7890±3850 yr

BP), an open water assemblage was deposited and sea ice cover was at a minimum. We associate the assemblage with climatic

warming, which characterizes much of the Antarctic Peninsula during this time. A second sea ice assemblage, different from

that deposited in the early Holocene, has been deposited in Lallemand Fjord since the late Holocene (,3850 yr BP). The

assemblage re¯ects Neoglacial cooling, an increase in sea ice extent and/or an advance of the MuÈller Ice Shelf. q 2001 Elsevier

Science B.V. All rights reserved.

Keywords: Holocene; Antarctica; paleoclimate; diatoms; sedimentology; statistical analysis

1. Introduction

Sedimentary diatom assemblages have been used

successfully in numerous studies as a proxy for

Antarctic marine paleo-reconstructions (e.g. Truesdale

and Kellogg, 1979; Pichon et al., 1987; Leventer et al.,

1996; Cunningham et al., 1999). Many of these

studies include statistical techniques to reconstruct

Quaternary glacial history, but few incorporate a

subjective, multi-disciplinary approach. In the present

study, we incorporate diatom, sedimentological and

geochemical data with classi®cation and indirect

ordination analyses to interpret the Holocene paleo-

environment of Lallemand Fjord on the western

Antarctic Peninsula.

Climate records from both Hemispheres demonstrate

increasingly that the Holocene (,11,500 yr BP, after

Roberts, 1998) has been a period of rapid and variable

climate change (Domack and Mayewski, 1999;

Rosqvist et al., 1999). Marine sediment cores from

the Antarctic Peninsula revealed multi-century and

millennial-scale variations in primary production

(Leventer et al., 1996; Shevenell et al., 1996;

Marine Micropaleontology 41 (2001) 25±43

0377-8398/01/$ - see front matter q 2001 Elsevier Science B.V. All rights reserved.

PII: S0377-8398(00)00049-9

www.elsevier.nl/locate/marmicro

* Corresponding address. Antarctic Co-operative Research

Centre/Institute of Antarctic and Southern Ocean Studies, GPO

Box 252-80, Hobart 7001, Tasmania, Australia. Tel.: 161-3-

6226-7888; fax: 161-3-6226-2973.

E-mail addresses: ®[email protected] (F. Taylor),

[email protected] (J. Whitehead),

[email protected] (E. Domack).

Rosqvist et al., 1999; Domack et al., 2000), and

similar patterns have been observed in Prydz Bay,

East Antarctica (e.g. Taylor, 1999). Yet the circum-

Antarctic marine paleoenvironmental record, as a

whole, remains poorly understood with regard to

rapid climate change (Domack and Mayewski,

1999). Given that the Antarctic Peninsula has under-

gone one of the most dramatic and rapid periods of

climatic warming in recent history (Smith et al.,

1999), the identi®cation and interpretation of paleo-

climatic events in this region is integral in under-

standing its response to predicted global climate

change. To investigate this, we provide a new,

detailed analysis of down-core diatom assemblages

from Lallemand Fjord, and compare the assem-

blages statistically to sedimentary and geochemical

variables. This builds upon the work commenced by

Shevenell et al. (1996).

2. Physical setting

The Lallemand Fjord is located on the Antarctic

Peninsula's west coast (Fig. 1), between 66850 0±

67820 0S and 66830 0W. It is the largest embayment

on that side of the peninsula and it lies close to the

northern limit of the polar climatic regime (Domack

and McClennen, 1996). Mean annual temperatures

range from 25.08 to 26.08C (Reynolds, 1981). The

fjord is within the Antarctic sea ice zone (SIZ),

which consists mostly of ®rst-year ice that melts

each season. The SIZ is a complicated, mobile

mixture of open water and ice of different types and

thickness (Allison and Worby, 1994) in a constant

state of ¯ux and rapid change (Foster, 1984). The

structure of the sea ice margin varies with wind

strength and direction. It may be a compact zone

that has a distinct boundary with the open water;

there may be a transitional zone between close pack

and open water that is several hundreds of meters

wide, or an even broader, loose pack zone kilometers

wide (Foster, 1984). In Lallemand Fjord, sea ice inter-

annual variability differs greatly and land-fast ice

often persists until late summer (Shevenell et al.,

1996).

Ocean circulation patterns along the western side of

the Antarctic Peninsula are poorly known (Domack and

Ishman, 1993). In Lallemand Fjord, oceanographic

F. Taylor et al. / Marine Micropaleontology 41 (2001) 25±4326

PalmerDeep

LarsenIce

Shelf

b

GC1

66û30'

67û00'

67û10'

67û20'

c

67û45'

Antar

ctic

Peninsula

Müller IceShelf

LallemandFjord

a

Fig. 1. Location of Lallemand Fjord and MuÈller Ice Shelf in relation to the Antarctic continent.

conditions appear generally uniform and bottom

waters are dominated by Circumpolar Deep Water

(Ishman and Domack, 1994). Water column data

suggest that this warm (.18C) water mass intrudes

into the Lallemand Fjord to cause sub-ice shelf melt-

ing and in¯uence the distribution of suspended

organic matter (Brandon, 1998).

At the head of the fjord, two low-pro®le valley

glaciers feed the MuÈller Ice Shelf. At least three

other major tidewater glaciers also empty into the

fjord and, combined, provide a total drainage

surface area of ca. 1290 km2 (Domack and

McClennen, 1996). Several deep, steep-sided basins

are present within the Lallemand Fjord. The largest

and deepest (.1500 m) occurs in the outer reaches

of the fjord system (Domack and McClennen,

1996). Sediment characteristics are controlled by

proximity to the glaciers at the head of the fjord

(Frederick et al., 1991). Sand and magnetic suscept-

ibility (MS) decrease with distance from its head; total

organic carbon (TOC) content increases with distance

from terrigenous sources (Domack and McClennen,

1996).

3. Materials and methods

3.1. Core description

PD92-II-01 GC1 (GC1) is a 550 cm-long gravity

core recovered from near the head of the Lallemand

Fjord (67810.765 0S, 66847.822 0W, Fig. 1) in a water

depth of 630 m. It consists of olive gray (5Y 5/2) clay

at the top, grading into olive gray (5Y 4/2), sandy clay

at the base. Dark yellowish brown (10Y 4/4) staining

occurs from 0 to 9 cm. Moderate disturbance occurs

from 0 to 12, 50 to 91, 244 to 276 and 328 to 335 cm,

and slight disturbance over 107±110 and 276±

328 cm. At 535 cm, one olive gray (5Y 4/2), sandy

lamination is present. Clasts of irregular-shaped,

sandy mud occur at 541, 542 and 545 cm.

Subrounded, coarse, granitic pebbles are scattered

throughout the core, at 37±39, 354±355, 357±360,

390±391 and 441±442 cm. Angular, ®ne basaltic

pebbles occur at 132±132.5 and 461±461.5 cm. A

whole scaphopod (Class: Mollusca) is present at

467 cm. Shell orientation, lack of coarse in®ll sedi-

ment in the shell and presence of a burrowing ®lter

strongly suggest that it is in situ (Shevenell et al.,

1996). Other sedimentological characteristics are

summarized in Shevenell et al. (1996).

3.2. Radiocarbon dates

Radiocarbon dating was conducted at the Univer-

sity of Arizona's Accelerator Mass Spectrometer

laboratory. A modern marine reservoir age of

1460 ^ 60 yr (Lab. #AA29182) was obtained from a

living scaphopod recovered from a surface sediment

grab sample (LMG98-02 G8) at the head of the MuÈller

Ice Shelf. The in situ scaphopod removed from GC1 at

456±460 cm has been radiocarbon dated previously at

9360 ^ 70 yr (Shevenell et al., 1996).

3.3. Sample preparation for diatom analysis

The core was sub-sampled at 10 cm intervals, and

samples stood for 24 h in distilled water to which

10 ml of 30% H2O2 had been added. Two drops of

HCl (2 N) were then added to remove organic carbo-

nate, and samples allowed to stand for another 24 h.

They were then centrifuged three times at 2500 revo-

lutions21 for 5 min; samples were washed in distilled

water to remove chemical residue and salt crystals

between centrifuging. Washed samples were diluted

in 10 ml distilled water and ,1±3 drops of the diluted

solution pipetted onto a glass cover-slip and dried on a

hotplate at 508C. Permanent, qualitative slides were

mounted in Norland Optical Adhesive 61 (refractive

index 1.56) and cured under a UV light for at least

5 min.

Diatoms were identi®ed and counted at 1000 £magni®cation using a phase contrast Zeiss Standard

25 phase contrast light microscope. Each slide was

traversed horizontally until at least 400 valves were

counted. To avoid counting the same specimen

twice, only valves that were .50% intact were

counted. For elongate species that are rarely preserved

intact, such as Trichotoxon, Thalassiothrix and

Pseudonitzschia, only end pieces were counted and

divided by two. Valves that could not be identi®ed,

due to orientation or obscuring debris, were class-

i®ed as ªmiscellaneous centricsº or ªmiscellaneous

pennatesº. Due to the high abundance of Chaetoceros

resting spores (spores, hereafter) in all samples

(.50%, Fig. 2), slides were re-counted, excluding

Chaetoceros. This allowed the background abundance

F. Taylor et al. / Marine Micropaleontology 41 (2001) 25±43 27

of ecologically important species, which may have

been otherwise masked by Chaetoceros spores, to be

determined.

We have chosen to split Thalassiosira antarctica

var. antarctica resting spores into separate varieties

as two morphological types were observed (Plate 1).

Resting spore type 1 (T. antarctica T1) has a ®ner

areole structure than resting spore type 2 (T. antarc-

tica T2), and the valves were often smaller in

diameter. Thalassiosira antarctica T2 appears similar

to that described by Fryxell et al. (1981), with distinct

ªshoe-likeº projections around the valve margin.

Large occluded processes in the margin are some-

times visible also in T. antarctica T2. Neither of

these characteristics was obvious in T. antarctica

T1. Similar morphological variation between T.

antarctica spores has been observed in surface

sediment samples from the Ross Sea and Antarctic

Peninsula (Leventer, pers. comm.) and Palmer Deep

(Sjunneskog and Taylor, submitted). Detailed taxo-

nomic work beyond the scope of this study is required

to de®ne the two varieties.

In addition to diatoms, the Parmales species Penta-

lamina corona was counted. Parmales is a group of

Chrysophyte with siliceous cell wall plates. They have

been observed in abundance in surface waters surface

sediment of Prydz Bay, East Antarctica, (Marchant,

1993; Franklin and Marchant, 1995; Taylor et al.,

1997) and the Weddell Sea (Zielinski, 1997). All

suggest that Parmales are useful indicators of sea ice

environments.

3.4. Statistical analyses

Diatoms and Parmales were expressed as a percen-

tage of the total number of cells counted per sample

(excluding Chaetoceros spores). Rare species (rela-

tive abundance ,2%) were removed prior to analysis

as they are not present in suf®cient abundance to be

signi®cant statistically (Katoh, 1993). The remaining

data were logarithmically transformed �log10x 1 1� to

reduce the score and bias of other abundant species

that may have otherwise masked the effect of less-

abundant species (Field et al., 1982). Data transforma-

tion does not alter zero values.

Data were analyzed using Bray±Curtis cluster

analysis, the student Newman±Keuls multiple range

test (SNK), non-metric multidimensional scaling

(NMDS), and multiple regression, following the

method outlined in Taylor et al. (1997). Down-core

cluster analysis was conducted similar to the method

of Whitehead (1996). Eight core variables were

compared to the diatom data by multiple regression:

MS, d 13C, percent TOC, clay, ®ne-medium silt and

coarse silt, sand and mean grain size. Methods for data

collection of individual variables (except d 13C) are

described in Shevenell et al. (1996).

3.5. d 13C analysis

Sediment samples for analysis of total organic

carbon 13C/12C ratios were dried (608C), powdered

and 10±20 mg weighed into silver foil envelopes.

Weighed samples were acidi®ed in situ with 6%

sulfurous acid to remove calcium carbonate. Carbon

stable isotopic composition was determined using a

F. Taylor et al. / Marine Micropaleontology 41 (2001) 25±4328

50.0 60.0 70.0 80.0 90.0 100.00

50

100

150

200

250

300

350

400

450

500

0

1000

3000

5000

7000

8000

10000

9000

3000

4000

6000

% Chaetoceros Spores

Dep

th(cm

)

Ag

e(yr

BP

)

Fig. 2. Relative abundance (%) of Chaetoceros spores from GC1.

Carlo Erba NA1500 elemental analyzer/Con¯o II

device and a Finnigan Delta Plus mass spectrometer

at Stanford University. Carbon isotopic reproducibil-

ity, as determined by replicate analyses of NBS-21,

is 0.08½.

4. Results

4.1. Chronology

Using a marine-reservoir correction factor of

F. Taylor et al. / Marine Micropaleontology 41 (2001) 25±43 29

Plate 1. Key indicator taxa in GC1; 1 Ð Fragilariopsis curta; 2 Ð Fragilariopsis kerguelensis; 3 Ð Eucampia antarctica (a) intercalary

winter valve, (b) terminal winter valve; 4 Ð Thalassiosira gracilis var. expecta; 5 Ð T. gracilis var. gracilis; 6 Ð Thalassiosira antarctica T2;

7 Ð T. antarctica T1; 8 Ð Thalassiosira lentiginosa.

1460 ^ 60 yr, based on the living scaphopod, the

uncorrected radiocarbon age of 9360 ^ 70 from

the in situ shell at 456±461 cm was calibrated to

8844 calendar years (yr BP) after Stuiver and

Reimer (1993). Based on this, and assuming a

constant deposition rate, a sedimentation rate of

0.052 cm yr21 is calculated over the entire length of

the core.

4.2. Statistical analyses

Twenty-eight taxa with an abundance of .2%

(excluding Chaetoceros spores) were observed in

the 55 core samples (Table 1). One outlier sample

was identi®ed in the preliminary cluster analysis and

removed from further analysis. The sample (from

440 cm) was observed to have an unusually high

abundance of Thalassiosira gracilis var. gracilis and

Thalassiosira gracilis var. expecta. Three cluster

groups were identi®ed in the subsequent analysis

(Fig. 3), at 34.4% dissimilarity and with a cophenetic

correlation of 0.65 (0.00 indicating no match with the

original data; 1.00 indicating a perfect match). Signif-

icant differences in species abundance between cluster

groups were identi®ed by SNK (Table 1).

Cluster group 1 characterizes the upper 160 cm

(,3080 yr BP). Between 190 and 220 cm (3650±

4230 yr BP), this group alternates with cluster group

2. One sample from cluster group 1 occurs at 340 cm

(6540 yr BP). Thalassiosira antarctica T1 is dominant

(28.3%) in cluster group 1; T. antarctica T2 (19.9%)

and Fragilariopsis curta (15.7%) are subdominant.

Also common (2±10% abundance) are Fragilariopsis

cylindrus, Navicula spp., T. gracilis var. gracilis, and

the Chrysophyte Pentalamina corona. Five taxa are

identi®ed by the SNK test as indicators of the assem-

blage: F. rhombica, Pseudonitzschia turgiduloides,

Rhizosolenia spp., ªmiscellaneous centricsº and P.

corona.

Cluster group 2 characterizes the mid-section of

the core. Between 170 and 210 cm (4420±7880 yr

BP), the cluster group alternates with group 1.

Between 230 and 410 cm (4420±7880 yr BP),

cluster group 2 forms a continuous unit (with an

exception at 340 cm). Cluster group 2 is also

present at 460 cm (8850 yr BP). Thalassiosira

antarctica T2 (35.0%) dominates the assemblage;

T. antarctica T1 (19.7%) and Eucampia antarctica

(17.5%) are subdominant. Common species are

Fragilariopsis curta, Fragilariopsis kerguelensis

and Fragilariopsis cylindrus. Unique abundance

indicator taxa are Achnanthes spp., F. kerguelensis,

Stellarima microtrias and Thalassiosira lentiginosa.

Cluster group 2 can be distinguished further from

cluster group 1 by a signi®cantly greater abundance

of T. antarctica T2 and a signi®cantly smaller

abundance of T. antarctica T1.

Cluster group 3 occurs below 420 cm (.8080 yr

F. Taylor et al. / Marine Micropaleontology 41 (2001) 25±4330

Table 1

Mean arithmetic abundance (%), analysis of variance (F) and SNK

multiple range tests of dominant (.2%) species in cluster groups.

Analyses were carried out on log10�x 1 1� transformed abundance.

Degrees of freedom� 2.54. ANOVA P values: n.s. not signi®cant,

* , 0.05, ** , 0.005, *** , 0.0005. Bold: species with signi®cant

differences in mean abundance. Underlined: species with signi®-

cantly higher abundance in a cluster group

Species Mean abundance

(%) per cluster

group

F P

1 2 3

Achnanthes spp. 0.9 1.3 0.6 4.33 *

Actinocyclus actinochilus 0.3 0.4 0.8 5.37 *

Cocceneis spp. 0.3 0.9 0.7 8.34 **

Eucampia antarctica 1.1 17.5 20.6 43.63 ***

Fragilaria spp. 0.2 0.0 0.4 3.74 **

Fragilariopsis curta 15.7 10.1 25.5 9.91 ***

Fragilariopsis cylindrus 10.4 2.2 9.8 71.86 ***

Fragilariopsis kerguelensis 1.8 3.8 2.3 13.43 ***

Fragilariopsis obliquecostata 0.9 1.6 2.0 6.62 ***

Fragilariopsis rhombica 1.3 0.6 1.0 6.81 **

Fragilariopsis sublinearis 0.2 0.1 0.0 2.85 n.s.

Fragilariopsis vanheurckii 0.9 0.7 3.7 20.35 ***

Navicula spp. 2.5 0.4 1.6 42.74 ***

Odontella spp. 0.3 0.9 2.9 21.11 ***

Pentalamina corona 2.3 0.3 1.8 27.73 ***

Pseudonitzschia turgiduloides 1.2 0.1 0.2 29.02 ***

Rhizosolenia spp. 1.6 0.7 0.9 10.09 ***

Stellarima microtrias 0.2 0.7 0.3 6.09 **

Synedra spp. 1.6 0.7 0.5 11.58 ***

Thalassiosira antarctica T1 28.3 19.7 1.5 62.62 ***

T. antarctica T2 19.9 35.0 13.7 21.92 ***

Thalassiosira gracilis var.

gracilis

2.2 1.5 2.0 1.33 n.s.

T. gracilis var. expecta 1.1 0.6 2.6 13.09 ***

Thalassiosira lentiginosa 0.2 0.9 0.4 13.38 ***

Thalassiosira tumida 0.1 0.1 0.7 3.12 n.s.

Thalassiosira sp. A 0.0 0.8 0.6 5.62 *

Miscellaneous pennates 0.6 0.5 0.9 3.34 *

Miscellaneous centrics 0.7 0.2 0.0 11.01 ***

BP). The diatom assemblage is dominated by Fragi-

lariopsis curta (25.5%) and Eucampia antarctica

(20.6%). Thalassiosira antarctica T2 is subdominant

(13.7%), but signi®cantly less compared to that in

cluster groups 1 and 2. Common taxa are F. cylindrus,

F. kerguelensis, F. vanheurckii, Fragilariopsis obli-

quecostata, Odontella spp., T. gracilis var. gracilis

and T. gracilis var. expecta. Three unique indicator

F. Taylor et al. / Marine Micropaleontology 41 (2001) 25±43 31

0cm20cm40cm

80cm140cm120cm30cm10cm

90cm150cm160cm100cm180cm220cm70cm60cm50cm

190cm130cm110cm

310cm300cm260cm250cm270cm240cm230cm210cm200cm170cm340cm

410cm390cm290cm330cm320cm280cm360cm380cm350cm370cm

430cm540cm470cm490cm500cm510cm480cm520cm530cm450cm

460cm400cm

420cm

Gro

up1

Gro

up2

Gro

up3

% Dissimilarity

20.010.0 30.0 40.00.0

Fig. 3. Dendrogram illustrating sample af®nities, based on diatom abundance. Cluster group 1� dense sea ice assemblage; cluster group 2�seasonally open water assemblage; cluster group 3� loose sea ice assemblage.

00.

10.

20.

30.

4

Axis

Str

ess

1 432

Fig. 4. Nonmetric multidimensional scaling (NMDS) ordination axis versus stress. Two axes were selected as best ®tting the data, based on the

point of maximum change in direction of the curve (from Kruskal and Wish, 1978).

taxa are present: Actinocyclus actinochilus, Fragi-

laria spp. and F. vanheurckii.

Two NMDS ordination axes were chosen as best

summarizing the data (Fig. 4). Stress values

converged after 10 iterations at a value of 0.1202,

indicating a good ®t with the original data (Hosie,

1994). The NMDS results are illustrated in Fig. 5.

There is good agreement between NMDS and cluster

analysis, visible when the cluster groups are circled on

the NMDS plot (Fig. 5). Using the ordination scores,

multiple regression analysis compared the diatom data

with the eight core variables. Three variables are

signi®cantly correlated with the data (Table 2), and

the direction of maximum correlation for each (Table

3) is illustrated in Fig. 5. (In Fig. 5, arrow length

represents the signi®cance of the correlation between

the diatom data and core variable, i.e. the longer the

arrow the greater the correlation with that variable,

and arrow direction indicates the direction in which

the variable is most correlated to the data.) Percent

TOC accounts for 54.9% of the variance observed in

the data; MS for 38.6% of the variance, and percent

®ne-medium silt for 30.7% of the variance. d 13C is

slightly less signi®cant, accounting for 22.3% of the

variance. The signi®cant variables and cluster groups

2 and 3 are clearly separated. Arrow length and

direction (Fig. 5) indicate the direction of maximum

correlation between the core variables and cluster

groups. Cluster group 2 is closely associated with

high TOC values, a high abundance of ®ne-medium

F. Taylor et al. / Marine Micropaleontology 41 (2001) 25±4332

Fig. 5. Ordination plot of samples, based on diatom species abundance (%). Arrows indicate direction of maximum correlation for signi®cant

multiple regressions between ordination scores and core variables. Cluster groups identi®ed in Fig. 3 are superimposed. Cr silt� coarse silt;

FM silt� ®ne-medium silt.

Table 2

Multiple regression analysis between dependent variables (physical

properties) and NMDS scores for a two-dimensional ordination.

Degrees of freedom� 2.48. ANOVA P values as for Table 1. Radj2 �

adjusted coef®cient of determination, which gives the fraction of

variance accounted for by the explanatory variable (Jongman et al.,

1987)

Dependent

variable

Direction cosine Radj2 F P

x y

d13C 1.778 21.213 0.223 8.190 **

MS 31.048 390.505 0.386 17.331 ***

TOC 0.740 20.420 0.549 33.247 ***

Clay 5.480 7.633 0.059 2.655 *

FM Silt 21.230 214.868 0.307 12.719 ***

Coarse silt 1.603 5.389 0.058 2.644 *

Sand 25.853 1.846 0.003 1.072 n.s.

Mean Grain Size 28.017 0.749 0.006 1.153 n.s.

silt and high d 13C; group 3 is closely associated with

high MS (Fig. 5). There is little or no signi®cant

association between the cluster groups and percent

clay, coarse silt, sand or mean grain size (Table 2).

5. Discussion

5.1. Chaetoceros abundance

Chaetoceros spores are the dominant taxa in GC1,

forming .50% of frustules counted in all samples

(Fig. 2). Abundance is high, but variable, in the

upper 520 cm of the core, and ranges from 66.8 to

95.1%. High concentrations of Chaetoceros spores

in Antarctic sediment are considered to be indicative

of high primary production in the water column

(Donegan and Schrader, 1982; Leventer, 1992;

Leventer et al., 1996). During spring diatom blooms,

surface waters can become so nutrient-depleted that

diatom growth is limited (Nelson and Smith, 1986;

McMinn et al., 1995) and spore formation is induced

(Davis et al., 1980). They remain dormant at the

sediment±water interface until favorable conditions

induce germination. Chaetoceros spore abundance

decreases relative to other diatom taxa below

520 cm, reaching a minimum of 53.1%. The decrease

suggests reduced primary production. The abundance

of Chaetoceros spores may also have been diluted by

higher siliciclastic deposition, although this is less

likely as there is no increase in gravel abundance in

this section of the core (Shevenell et al., 1996).

5.2. Diatom assemblages

Excluding Chaetoceros spores, cluster analysis and

NMDS identify three cluster groups (representing

diatom assemblages). The assemblages are interpreted

to have been deposited within the complex SIZ, where

studies have indicated that the different sea ice types

contain different algal assemblages (e.g. Garrison et

al., 1986; Scott et al., 1994; Leventer and Dunbar,

1996). Based on the dominant and indicator taxa

with known ecological af®nities in the present study

(Fig. 6 and Plate 1), the diatom assemblages discussed

are interpreted to represent different sub-environ-

ments within the SIZ.

5.2.1. Cluster group 3 (sea ice associated)

Cluster group 3 dominates the lower third of GC1

(550±420 cm; Fig. 7). Fragilariopsis curta and

Eucampia antarctica are the most abundant species;

Thalassiosira antarctica T2 is subdominant (Table 1).

Based on the known ecology of the abundant and

indicator taxa in cluster group 3, the diatom assem-

blage is described as sea ice associated.

Fragilariopsis curta occurs commonly in ice edge

and within-ice algal assemblages (Scott et al., 1994;

Leventer and Dunbar, 1996) and in meltwater-

strati®ed surface water layers associated with retreat-

ing sea ice (Garrison et al., 1987). Leventer and

Dunbar (1996) hypothesized that the high abundance

of F. curta within the water column and surface

sediment of the Ross Sea is due to its being seeded

into the water column from fast ice during the spring

ice recession. In surface sediment diatom assemblages

from Prydz Bay and the Ross Sea, F. curta occurs

in high abundance where sea ice often persists

throughout summer (Taylor et al., 1997; Cunningham

and Leventer, 1998).

Eucampia antarctica is also widely considered to

F. Taylor et al. / Marine Micropaleontology 41 (2001) 25±43 33

Table 3

Cosine and acosine values (using NMDS axes 1 and 2 coef®cient values) used to determine direction of maximum correlation in Fig. 5

Variable Coeff1 Coeff2 Cos1 Cos2 Acos1 Acos2

d13C 1.778 21.213 0.826 20.564 34.303 124.303

MS 31.048 390.505 0.079 0.997 85.454 4.546

TOC 0.740 20.420 0.870 20.494 29.578 119.578

Clay 5.480 7.633 0.583 0.812 54.324 35.676

FM Silt 21.230 214.868 20.082 20.997 94.729 175.271

Coarse Silt 1.603 5.389 0.285 0.958 73.434 16.566

Sand 25.853 1.846 20.954 0.301 162.495 72.495

Mean Grain Size 28.017 0.749 20.996 0.093 174.663 84.663

F. Taylor et al. / Marine Micropaleontology 41 (2001) 25±4334

Fig. 6. Relative abundance (%) of key indicator taxa in GC1.

F.

Ta

ylor

eta

l./

Ma

rine

Micro

paleo

nto

logy

41

(2001)

25

±43

35

Fig. 7. Core log, core variables, sedimentary units (dashed horizontal line) and diatom assemblages from GC1. Diatom assemblages in relation to cluster groups are: dense sea ice

(cluster group 1); seasonally open water (cluster group 2); loose sea ice (cluster group 3). Sedimentary units from Shevenell et al. (1996). December solar insolation from Berger

(1978).

be a sea ice diatom (Burckle et al., 1990), although

Zielinski and Gersonde (1997) suggest that it should

not be de®ned as such. They have noted that E.

antarctica is most abundant where surface waters

are 22±08C and 2.5±5.58C, indicating that it is

related to surface waters in the Antarctic and the

Polar Front Zone of the Southern Ocean (Zielinski

and Gersonde, 1997). This discrepancy in distribution

may be attributable to the two varieties of Eucampia

that Fryxell and Prasad (1990) identify. One variety is

a truly Antarctic, ice edge organism (E. antarctica

var. recta), and the other is subpolar (E. antarctica

var. antarctica). Abundance of E. antarctica in

surface sediment assemblages has been interpreted

to indicate reworking and/or current winnowing

(Truesdale and Kellogg, 1979; Taylor et al., 1997;

Cunningham and Leventer, 1998). The high abun-

dance (up to 46%) of E. antarctica in cluster group

3 is not attributed to reworking mechanisms, however.

Reworked and current winnowed assemblages typi-

cally contain a high abundance of other heavily

silici®ed, robust, often extinct, taxa, and lack small

and fragile taxa that have been removed by dissolution

or strong water currents. In contrast, the assemblage in

GC1 contains fragile taxa, such as Fragilariopsis

cylindrus and the Chrysophyte Pentalamina corona,

in comparable, if not greater, abundance to the other

diatom assemblages observed in the core.

Less abundant, but statistically signi®cant, indica-

tor taxa in cluster group 3 are underlined in Table 1.

Planktic pennates, such as Fragilariopsis obliquecos-

tata and F. vanheurckii, are considered to be ice asso-

ciated (Garrison and Buck, 1985; Medlin and Priddle,

1990). F. obliquecostata has been observed in sub-ice

microalgal strands under coastal fast ice (Watanabe,

1988), but Cunningham et al. (1999) report it to be

open water associated. The planktic, centric Actino-

cyclus actinochilus is considered a typical Antarctic,

neritic species (Kozlova, 1966) that occurs in the ice

edge zone (Medlin and Priddle, 1990). It is one of the

characteristic species found in the ªSouth Weddellº

diatom assemblage described in surface sediment by

Pichon et al. (1987), which is associated with an area

where sea ice is absent for only ,2 months.

Odontella spp. are also important statistically in

cluster group 3 and reach a maximum abundance of

10.2%. There is little documentation of the ecology

of this spore-forming genus, although Odontella

weis¯oggii (Janisch) Grunow is considered endemic

to the Southern Ocean and occurs in Antarctic near-

shore regions where water temperatures are between

22 and 58C (Zielinski and Gersonde, 1997).

Froneman et al. (1997) report it to be a temperate,

neritic species that probably gets transported into

Antarctic waters by unusual, southern intrusions of

subantarctic surface waters. The ecology of Thalas-

siosira gracilis var. expecta and the benthic taxa

(Cocconeis, Fragilaria, Navicula) are also amongst

the less-well-documented taxa. Zielinski and

Gersonde (1997) observe that T. gracilis (no variety

speci®ed) reaches maximum abundance in Antarctic

surface sediment that occurs below relatively warm

waters with a temperature 20.5±28C, but should be

considered a taxon with no de®nitive relation to envir-

onmental parameters. Due to the generally low abun-

dance of the above taxa and ecological uncertainties,

these species have not been used to interpret the

assemblage's paleoecology.

5.2.2. Cluster group 2 (seasonal open water)

Cluster group 2 characterizes the mid-section of the

core (Fig. 7). The most abundant taxon is Thalassio-

sira antarctica T2 (54.4%). T. antarctica T1 and

Eucampia antarctica are subdominant. Fragilariopsis

curta is relatively common, but it is signi®cantly less

abundant compared to cluster groups 1 and 3. The

presence of species such as Fragilariopsis kergue-

lensis and Thalassiosira lentiginosa suggests that the

diatom assemblage was deposited in a seasonally

open water environment (discussed below). F.

kerguelensis attains a maximum abundance of 6.5%

in cluster group 2. Whilst this does not rank it amongst

the most common species in the group, the abundance

is signi®cantly higher compared to groups 1 and 3,

and it forms a unique indicator of the assemblage.

Fragilariopsis kerguelensis is a valuable paleo-

indicator, used to identify open marine deposition.

Today it is dominant between 52 and 638S (Burckle

et al., 1987), where summer surface water tempera-

tures are .08C (Krebs et al., 1987). Abundance is also

known to be negatively correlated with sea ice

(Burckle et al., 1987), and to increase with distance

from the Antarctic continent in both surface water

(Kozlova, 1966) and sedimentary assemblages

(Leventer, 1992; Harris et al., 1998). Similar observa-

tions have been made of Thalassiosira lentiginosa,

F. Taylor et al. / Marine Micropaleontology 41 (2001) 25±4336

suggesting that it is an equally valuable indicator of

open water (Zielinski and Gersonde, 1997). In cluster

group 2, T. lentiginosa is numerically rare (maximum

abundance 2.5%), but its abundance is signi®cantly

high and it forms a unique indicator in the assemblage.

The high abundance of Eucampia antarctica in

cluster group 2 also supports the hypothesis that this

species is not solely a sea ice indicator (Zielinski and

Gersonde, 1997). We have not distinguished between

the two varieties of Eucampia identi®ed by Fryxell

and Prasad (1990), and demonstrated to have different

geographical distributions and to inhabit different

environments (Fryxell and Prasad, 1990; Kaczmarska

et al., 1993), but suggest this in future analyses. E.

antarctica var. antarctica has a subpolar distribution,

whilst E. antarctica var. recta is restricted to cold,

polar waters associated with sea ice cover (Fryxell

and Prasad, 1990; Kaczmarska et al., 1993).

The genus Thalassiosira is widespread in Antarctic

waters, where it generally occurs in open water. It is

uncommon in sea ice (Fryxell and Kendrick, 1988;

Leventer and Dunbar, 1996; Zielinski and Gersonde,

1997), which Fryxell et al. (1987) attribute to its

inability to survive the low light intensity beneath,

and within, sea ice. The observation that Thalassiosira

antarctica is a member of some sea ice samples (e.g.

Villareal and Fryxell, 1983; Leventer and Dunbar,

1996), however, has led to the suggestion that it

may be associated with coastal sea ice and zones of

lose platelet ice (Cunningham and Leventer, 1998).

Indeed, some Thalassiosira species are sea ice related.

A bloom of Thalassiosira tumida, for example, is

reported in slush ice forming near the Ronne Ice

Shelf (El-Sayed, 1971), and Thalassiosira australis

Peragallo 1921 is observed amongst the dominant

species beneath snow-free fast ice in Ellis Fjord

(Vestfold Hills, East Antarctica), and McMurdo

Sound (McMinn, 1996; McMinn, 1999). Taylor

(1999) suggests that the formation of T. antarctica

spores could be triggered by the low light intensities

that occur beneath developing pack and platelet ice.

Reduced wind mixing below the sea ice may also

induce spore formation.

As in cluster group 3, the ecology of many of the

rarer, but statistically signi®cant diatoms is ambigu-

ous. Stellarima microtrias, for example, is reported as

being restricted to the Antarctic Zone south of the

Polar Front, in waters 22±18C (Zielinski and

Gersonde, 1997). Hasle et al. (1988) ®nd S. microtrias

benthic on or in sea ice and planktic in waters in¯u-

enced by sea ice Ð a paradox that they suggest may

be explained by its ability to produce spores. Along

with the benthic taxa (Achnanthes and Cocconeis), the

species whose ecology is poorly documented and

whose ecological af®nity is uncertain are not used

for paleoenvironmental interpretation.

5.2.3. Cluster group 1 (sea ice associated)

Cluster group 1 is present in the upper 200 cm

of GC1 (Fig. 7). Thalassiosira antarctica T1 is

signi®cantly more abundant (up to 47.6%)

compared to that in cluster groups 2 or 3. T.

antarctica T2 and Fragialriopsis curta are subdo-

minant members of the assemblage (with an aver-

age of 28.3 and 15.7%, respectively), but both are

signi®cantly less abundant compared to abundance

in cluster group 2. The diatom assemblage of

cluster group 1 is interpreted to represent deposi-

tion in a sea ice-associated environment, but is

statistically different from the sea ice diatom

assemblage of cluster group 3. Each probably

represents deposition within a different zone of

the seasonal sea ice zone, but we ®nd it dif®cult

to distinguish these differences based on diatoms

alone.

The subdominant and common taxa in cluster

group 1 are of mixed ecological preference. As

discussed previously, the genus Thalassiosira tends

to be associated with open water deposition, but rest-

ing spore formation may be induced by sea ice. Fragi-

lariopsis curta is a member of sea ice assemblages

where ice retreat has created a melt-water, strati®ed

surface water layer. Fragilariopsis cylindrus has been

observed amongst the dominant taxa in pack and fast

ice (Garrison and Buck, 1989; Scott et al., 1994) and

ice edge blooms (Kang and Fryxell, 1992), and has

been also found in open water (Garrison et al., 1987;

Leventer et al., 1993). The ecology of the Chrysophyte,

Pentalamina corona, is not well known, although

evidence suggests that Parmales inhabit ice edge and

pack ice environments. Silver et al. (1980) report

ªsiliceous cystsº (now known to be Parmales) in low

abundance in sea ice samples from the Weddell Sea, and

Brandon (1998) has found one species in water column

samples from in front of the MuÈller Ice Shelf. Three

of the ®ve indicator species in the cluster group

F. Taylor et al. / Marine Micropaleontology 41 (2001) 25±43 37

(Fragilariopsis rhombica, Pseudonitzschia turgidu-

loides and P. corona) are indicators of ice-associated

and ice-edge diatom assemblages in the surface sedi-

ments of Prydz Bay where multi-year ice frequently

persists (Taylor et al., 1997). Grouped genera such as

Navicula, Rhizosolenia and Synedra are statistically

signi®cant but numerically rare in the cluster group.

5.3. Multiple regression

The direction of maximum correlation for TOC,

®ne-medium silt and d 13C corresponds with cluster

group 2 (Fig. 5). High TOC and d 13C values in

sediment samples containing this diatom assemblage

indicate that it is associated with high primary produc-

tion. The abundance of ®ne-medium sized silt

particles (4±31 mm), which are within the size range

of most diatom frustules, could be used to indicate that

diatoms were the main primary producers. The direc-

tion of maximum correlation for MS corresponds with

cluster group 3 (Fig. 5). This suggests that terrigenous

input, relative to biogenic particle ¯ux, during deposi-

tion of the diatom assemblage was higher. Simulta-

neously, TOC and ®ne-medium silt deposition are

low.

5.4. Paleoecological interpretation

5.4.1. Late Pleistocene±Early Holocene (10,580±

7890 yr BP)

The ice-associated diatom assemblage from cluster

group 3 corresponds with the sedimentological unit III

of Shevenell et al. (1996) in GC1 (Fig. 7). The unit

contains laminated, gray, silty mud, overlain by struc-

tureless, gray, silty mud. It is characterized by an

upward increase in TOC, and a decrease in MS and

mean grain size (Fig. 7). The sedimentological struc-

ture is inferred to represent a transition from ice

proximal to open marine deposition as Late Pleisto-

cene±Early Holocene deglaciation occurred in the

fjord (Shevenell et al., 1996). This is supported by

the presence of the ice-associated diatom assemblage

of cluster group 3, which was deposited from at least

10,580 to 8080 yr BP (assuming a constant sedimen-

tation rate based on 0.052 cm yr21). The increasing

TOC and d 13C values suggest increasing primary

production, although TOC remains relatively low

compared to mid Holocene values (Fig. 7) and may

re¯ect dilution of this signal due to increased sili-

clastic sedimentation (as would occur in close

proximity to a receding glacier). Based on the diatom

taxa, their preferred ecological habitat, and sedimen-

tology, sea ice cover is thought to have been in the

form of loose sea ice that expanded and contracted

seasonally over water exposed by the retreating

MuÈller Ice Shelf.

Deposition of the loose sea ice assemblage was

interrupted brie¯y at 8850 and 8460 yr BP. During

both times, seasonally open water and unusual

Thalassiosira gracilis-dominated assemblages were

deposited (Figs. 6 and 7). The latter assemblage

formed an outlier in cluster analysis and was excluded

from further analysis. We suggest that both assem-

blages represent deposition in an environment asso-

ciated with windy, cool climatic episodes, which

interrupted the deglaciation phase, resulting in more

open water conditions in comparison to that during

deposition of the ice-associated assemblage. The

open water conditions may have been associated

with a polynya that may have formed due to high

winds pushing sea ice offshore. This follows the

hypothesis of Leventer and Dunbar (1988), who

suggest that a relative increase in abundance of

Thalassiosira spp., compared to Fragilariopsis

curta, approximately 350 yr BP in McMurdo Sound

is the result of more persistent and/or strong katabatic

and synoptic winds. Increased winds would reduce the

amount of near-shore sea ice cover and allow for

higher primary production.

A century-scale, globally distributed, cooling event

to support our hypothesis has been identi®ed in the

Taylor Dome (Antarctica) and GISP2 (Greenland) ice

core records between 8400 and 8200 yr BP (Alley et

al., 1997; Stager and Mayewski, 1997). It is inferred to

have been synchronous with global cooling of

approximately half the amplitude of the Younger

Dryas. A cooling event has also been identi®ed in

sediments from the nearby Palmer Deep (Fig. 1),

following the Last Glacial Maximum's deglacial

episode (Domack et al., 2000). On South Georgia, in

the sub-Antarctic Southern Ocean, a short-lived cold

event is recorded in age-calibrated lake sediments

between 7800 and 7400 yr BP (Rosqvist et al.,

1999). If these events are correlated with the age

calibrated data from GC1, increased wind associated

with cooling in Lallemand Fjord 8800 yr BP could

have reduced sea ice cover, allowing the deposition

F. Taylor et al. / Marine Micropaleontology 41 (2001) 25±4338

of the Thalassiosira antarctica T2-dominated assem-

blage. Increased primary production 8500 yr BP could

be inferred in GC1 where there is a peak in TOC

(discussed below) at 440 cm, correlating with the

Thalassiosira gracilis-dominated assemblage.

5.4.2. Mid Holocene (7890±3850 yr BP)

The cluster group 2 diatom assemblage, character-

ized by open water taxa, dominates the mid Holocene

in Lallemand Fjord between 7890 and 3850 yr BP

(Fig. 7). It corresponds with sedimentological unit II

that is characterized by silty mud. Scattered ice rafted

debris are present and decrease up-core. Total organic

content is generally high. It is suggested that climatic

warming, which caused deglaciation of the fjord in the

Early Holocene, created a more open marine, ice

distal, depositional environment with high biogenic

input and elevated primary production (Shevenell et

al., 1996). The observed maxima in d 13C over this

period (Fig. 7) further suggests high primary produc-

tion in which CO2 may have been limited in the photic

zone. Our results, combined with Shevenell et al.

(1996), support the hypotheses that the Antarctic

Peninsula experienced a climatic optimum during

the mid Holocene (e.g. Leventer et al., 1996; Hjort

et al., 1998; Rosqvist et al., 1999).

Lake sediment data from South Georgia (Birnie,

1990; Rosqvist et al., 1999) and James Ross Island

(BjoÈrck et al., 1996), and marine records from the

Ross Sea (Cunningham et al., 1999), suggest that

during the mid Holocene sea surface temperatures

were warmer and atmospheric temperatures were

warmer and/or more humid than present. Cunningham

et al. (1999) also show that, in contrast to our ®ndings,

mid Holocene diatom assemblages from the Ross Sea

contained a higher abundance of Fragilariopsis curta,

and fewer open water species, compared to modern

Ross Sea sediments. They suggest that sea ice melt

and water column strati®cation probably occurred

earlier than today in the Ross Sea, as a result of earlier

spring warmth. They speculate that earlier ice melt

allowed the spring bloom to be dominated by F.

curta seeded from the melting sea ice, rather than

for the bloom to be dominated by open water taxa.

Our data indicate that in the mid Holocene diatom

assemblage was characterized by signi®cantly fewer

F. curta than today, and more open water taxa. This

may indicate that less winter sea ice was present

during the mid Holocene and that during the spring

melt a signi®cant abundance of ice-associated diatoms

were not being released to seed the water column.

The variable TOC levels observed in Lallemand

Fjord during the mid Holocene probably re¯ect varia-

tions in primary productivity, which, in turn, re¯ects

variation in sea ice extent (Shevenell et al., 1996).

Low TOC values correspond with deposition of the

sea ice-associated assemblage 6730 yr BP. This event

occurs within the prolonged period of TOC minima,

between 7310 and 5390 yr BP, which was interpreted

by Shevenell et al. (1996) to indicate extensive sea ice

cover in the fjord. Low TOC and the sea ice assem-

blage may also be correlatable with a climatically cold

triple event (minima in solar irradiance; Stuiver and

Braziunus, 1989) documented in the GISP2 ice core

6200±5000 yr BP (O'Brien, 1995). The most variable

period of TOC accumulation in Lallemand Fjord

occurs between 4810 and 2880 yr BP, as indicated

by the ¯uctuating values (Fig. 7). During this time,

the diatom assemblages also undergo alternating

deposition of the seasonally open water assemblage

of cluster group 2, and the ice assemblage of cluster

group 1. The latter is deposited during periods of TOC

minima at the top of the core, supporting the hypoth-

esis that low TOC re¯ects increased ice cover.

5.4.3. Late Holocene (,3850 yr BP)

The cluster group 1 diatom assemblage has been

deposited from about 3850 yr BP. It was interrupted

brie¯y 3270 yr BP by redeposition of the seasonally

open water assemblage. From 2880 yr BP, cluster

group 1 corresponds with the core's sedimentary

unit I (Fig. 7). Cluster group 1 contains an ice-

associated diatom assemblage, and the onset of its

deposition is hypothesized to represent late Holocene

climatic cooling (the Neoglacial). Before discussing

this in detail, however, it is important to distinguish

how the ice-associated diatom assemblage of cluster

group 1 differs from the ice-associated assemblage of

cluster group 3. We ®nd it dif®cult to determine the

speci®c sea ice environments they each represent,

based on diatom abundance alone.

Cluster groups 1 and 3 both contain abundant

ice-associated diatoms, but the abundance of these

diatoms is signi®cantly different (Table 1). We

suggest that each cluster group represents a different

sub-environment within the SIZ, which is more

F. Taylor et al. / Marine Micropaleontology 41 (2001) 25±43 39

readily identi®ed by considering other data. Combin-

ing the diatom, sedimentary and geochemical data we

hypothesize that cluster group 1 (,3850 yr BP)

represents a period of more dense sea ice than that

associated with the loose sea ice-associated assem-

blage of cluster group 3 (deposited .7890 yr BP).

In sedimentary unit I, there is an overall decrease in

TOC, hypothesized to re¯ect decreased primary

production in response to the change in depositional

regime from an open marine environment with mini-

mal sea ice coverage to one with increased fast or

shelf ice (Shevenell et al., 1996). The tandem decrease

in d 13C also suggests that primary production

decreased in comparison to the mid Holocene.

Shevenell et al. (1996) also note an increase in the

siliciclastic (relative to biogenic) component of the

sediment, an increase in coarse silt and mean grain

size, and minimal ice-rafted debris and coarse-grained

terrigenous input, indicative of proximal ice shelf

deposition. These observations assist in differentiating

the depositional environments that the two signi®-

cantly different, but ice-associated, assemblages

were formed from. Based on these results, we also

suggest that the two varieties of Thalassiosira

antarctica resting spores have the potential to be

good paleo-indicators. Modern studies are required

to con®rm our hypothesis that T. antarctica T1 is

associated with dense sea ice, and T. antarctica T2

with less sea ice cover and more open water.

Domack and McClennen (1996) suggest that

climatic cooling in Lallemand Fjord, associated with

the Neoglacial, commenced sometime prior to

2000 yr BP. Based on the diatom record herein, it is

suggested that the transition from a seasonally open

water environment to one in¯uenced by more dense

sea ice commenced as early as 4420 yr BP. The

increased sea ice cover may have even been in the

form of compact pack ice, multi-year ice, or fast ice.

The latter dominates in the fjord today until late

summer. It is also possible that the ice shelf at the

head of Lallemand Fjord came into closer proximity.

The transition from a seasonally open water environ-

ment to an ice-associated environment is similar to the

®ndings of Sjunneskog and Taylor (submitted). They

mark the transition from the mid Holocene climatic

optimum to cooler Neoglacial conditions in the

Palmer Deep at 4420 yr BP, based on a decline in

total diatom abundance. Both results suggest that the

Neoglacial transition commenced approximately

1000 yr earlier than sedimentary data from the Palmer

Deep implies (Domack et al., 2000). Deposition of the

dense sea ice diatom assemblage of cluster group 1 in

Lallemand Fjord did not become well established

until 3080 yr BP. Prior to this, deposition oscillated

with the seasonally open water assemblage. This may

indicate that the environment ¯uctuated during the

transition from the mid Holocene climatic optimum

before stabilizing. Ciais et al. (1994) also suggest that

a stable average temperature has characterized the

Late Holocene with short-lived ¯uctuations from

4420 yr BP. From 3080 yr BP, the cluster group 1

assemblage has been deposited without major inter-

ruption.

Before concluding, we address the question: what is

the driving force behind Holocene variation in the

Antarctic Peninsula primary production? Variation

in solar insolation is regarded as one primary climatic

forcing mechanism (Nesje and Johannessen, 1992),

but we hypothesize it is not the mechanism in place

here. At polar latitudes (608S), summer insolation

reached a Holocene maximum ,2000 yr BP (Berger,

1978) (Fig. 7). This is well after the mid Holocene

climatic optimum, de®ned as terminating in Lalle-

mand Fjord 3850 yr BP based on the transition from

the seasonally open water to ice-associated assem-

blages. It seems likely that the changes in primary

production that we observe in Lallemand Fjord are

related more closely to shifts in water mass distribu-

tion (namely Circumpolar Deep Water) and sea

surface temperature.

6. Conclusion

This study indicates that detailed diatom analysis

highlights the variability of the transitions in paleo-

environmental reconstruction. Three diatom assem-

blages (two sea ice-associated and one seasonally

open water associated) are identi®ed in Lallemand

Fjord by cluster analysis and NMDS. The assem-

blages' association with down-core sedimentary and

geochemical variables are determined by multiple

regression. The assemblages have been deposited

within the SIZ since the Early Holocene, but each

re¯ects variations within this zone. In the Early

Holocene, Lallemand Fjord underwent deglaciation

F. Taylor et al. / Marine Micropaleontology 41 (2001) 25±4340

and we hypothesize that seasonal, loose sea ice was

present. The mid Holocene is characterized by more

open water and higher primary production. The Late

Holocene has undergone climatic cooling and

increased sea ice cover, in the form of compact pack

ice, multi-year ice, fast ice and/or closer proximity of

the MuÈller Ice Shelf. There is no association between

primary production and insolation in this sedimentary

record.

Acknowledgements

This work was supported by a grant from the

National Science Foundation's Of®ce of Polar

Programs (Grant OPP-9615053 to Hamilton College).

We wish to thank R. Dunbar for d 13C data, and

Leanne Armand, Amy Leventer and an anonymous

reviewer for critically reviewing the manuscript.

Appendix A. Taxonomic appendix

List of diatoms (Bacillariophyceae) cited in text.

Taxonomy is based on Johansen and Fryxell (1985);

Priddle and Fryxell (1985); Medlin and Priddle (1990)

and Roberts and McMinn (1999). Non-diatom

taxa (Chrysophyceae) are listed beneath the diatom

taxonomy. Chrysophyte taxonomy is based on

Booth and Marchant (1987).

Bacillariophyceae

Achnanthes Bory 1822

Actinocyclus actinochilus (Ehrenberg) Simonsen

1982

Chaetoceros (Ehrenberg) 1844 (spores)

Cocconeis Ehrenberg 1838

Eucampia antarctica (Castracane) Mangin 1915

Fragilaria Lyngbye 1819

Fragilariopsis curta (Van Heurck) Hustedt 1958

Fragilariopsis cylindrus (Grunow ex Cleve)

Helmcje & Krieger 1954

Fragilariopsis kerguelensis (O'Meara) Hustedt

1952

Fragilariopsis obliquecostata (Van Heurck)

Heiden in Heiden & Kolbe 1928

Fragilariopsis rhombica (O'Meara) Hustedt

1952

Fragilariopsis sublinearis (Van Heurck) Heiden

in Heiden & Kolbe 1928

Fragilariopsis vanheurckii (Peragallo) Hustedt

1958

Navicula Bory 1822

Pseudonitzschia turgiduloides (Hasle) Hasle

1995

Rhizosolenia Brightwell 1858

Stellarima microtrias (Ehrenberg) Hasle et Sims

1986

Synedra Ehrenberg 1832

Thalassiosira antarctica var. antarctica Comber

1896 (spores)

Thalassiosira gracilis var. gracilis (Karsten)

Hustedt 1858

Thalassiosira gracilis var. expecta (Van Land-

ingham) Fryxell et Hasle 1979

Thalassiosira lentiginosa (Janisch) Fryxell 1977

Thalassiosira tumida (Janisch) Hasle 1971

Thalassiosira sp. A

Chrysophyceae

Pentalamina corona Marchant 1987

References

Alley, R.B., Mayewski, P.A., Sowers, T., Stuiver, M., Clark, P.U.,

1997. Holocene climatic instability: a prominent, widespread

event 8200 yr ago. Geology 25 (6), 483±486.

Allison, I., Worby, A., 1994. Seasonal changes of sea-ice character-

istics off East Antarctica. Ann. Glaciol. 20, 195±201.

Berger, A.L., 1978. Long-term variations of caloric insolation

resulting from the Earth's orbital elements. Quat. Res. 9, 139±167.

Birnie, J., 1990. Holocene environmental change in South Georgia:

evidence from lake sediments. J. Quat. Sci. 5 (3), 171±187.

BjoÈrck, S., Olsson, S., Ellis-Evans, C., HaÊkansson, H., Humlum, O.,

de Lirio, J.M., 1996. Late Holocene palaeoclimatic records from

lake sediments on James Ross Island, Antarctica. Palaeogeogr.,

Palaeoclimatol., Palaeoecol. 121, 195±220.

Booth, B.C., Marchant, H.J., 1987. Parmales, a new order of marine

Chrysophytes, with descriptions of three new genera and seven

new species. Phycologia 25, 245±260.

Brandon, J., 1998. The relationship between warm Circumpolar

Deep Water and the composition and distribution of suspended

particulate matter within Lallemand Fjord, Antarctica. BA

thesis, Hamilton College, NY.

Burckle, L.H., Jacobs, S.S., McLaughlin, R.B., 1987. Late austral

spring diatom distribution between New Zealand and the Ross

Ice Shelf, Antarctica: hydrographic and sediment correlations.

Micropaleontology 33 (1), 74±81.

Burckle, L.H., Gersonde, R., Abrams, N., 1990. Late Pliocene±

Pleistocene paleoclimate in the Jane Basin region: ODP Site

679. Proc. ODP Sci. Res. 113, 803±809.

F. Taylor et al. / Marine Micropaleontology 41 (2001) 25±43 41

Ciais, P., Jouzel, J., Petit, J.R., Lipenkov, V., White, J.W.C., 1994.

Holocene temperature variations inferred from six Antarctic ice

cores. Ann. Glaciol. 20, 427±436.

Cunningham, W.L., Leventer, A., 1998. Diatom assemblages in

surface sediments of the Ross Sea: relationship to present

oceanographic conditions. Antarct. Sci. 10 (2), 134±146.

Cunningham, W.L., Leventer, A., Andrews, J.T., Jennings, A.E.,

Licht, K.J., 1999. Late Pleistocene±Holocene marine conditions

in the Ross Sea, Antarctica: evidence from the diatom record.

Holocene 9 (2), 129±139.

Davis, C.O., Hollibaugh, J.T., Seibert, D.L.R., Thomas, W.H.,

Harrison, P.J., 1980. Formation of resting spores by Leptocylin-

drus danicus (Bacillariophyceae) in a controlled experimental

ecosystem. J. Phycol. 16, 296±302.

Domack, E.W., Ishman, S.E., 1993. Oceanographic and physio-

graphic controls on modern sedimentation within Antarctic

fjords. Geol. Soc. Am. Bull. 105, 1175±1189.

Domack, E.W., McClennen, C.E., 1996. Accumulation of glacial

marine sediments in fjords of the Antarctic Peninsula and their

use as Late Holocene palaeoenvironmental indicators. Antarct.

Res. Ser. 70, 135±154.

Domack, E.W., Mayewski, P.A., 1999. Bi-polar ocean linkages:

evidence from late-Holocene Antarctic marine and Greenland

ice-core records. Holocene 9 (2), 237±241.

Domack, E.W., Leventer, A., Dunbar, R., Taylor, F., Brachfeld, S.,

2000. ODP Leg 178 Scienti®c Party. Chronology of the Palmer

Deep Site, Antarctic Peninsula: a Holocene paleoenvironmental

reference for the circum-Antarctic. Holocene (in press).

Donegan, D., Schrader, H., 1982. Biogenic and abiogenic compo-

nents of laminated hemipelagic sediments in the central Gulf of

California. Mar. Geol. 48, 215±237.

El-Sayed, S.Z., 1971. Observations on phytoplankton blooms in the

Weddell Sea. Antarct. Res. Ser. 17, 301±312.

Field, J.G., Clarke, R.K., Warwick, R.M., 1982. A practical strategy

for analysing multispecies distribution patterns. Mar. Ecol.

Prog. Ser. 8, 37±52.

Foster, T.D., 1984. The marine environment. In: Laws, R.M. (Ed.),

Antarctic Ecology, Academic Press, New York, pp. 345±372.

Franklin, D.C., Marchant, H., 1995. Parmales in sediments of Prydz

Bay, East Antarctica: a new biofacies and paleoenvironmental

indicator of cold water deposition? Micropaleontology 41 (1),

89±94.

Frederick, B.C., Domack, E.W., McClennen, C.E., 1991. Magnetic

susceptibility measurements in Antarctic glacial-marine sedi-

ment from in front of the MuÈller Ice Shelf, Lallemand Fjord.

Antarct. J. U.S. 26 (5), 126±128.

Froneman, P.W., Pakhomov, E.A., Laubscher, R.K., 1997. Micro-

phytoplankton assemblages in the waters surrounding South

Georgia, Antarctica, during austral summer 1994. Polar Biol.

17 (6), 515±522.

Fryxell, G.A., Doucette, G.J., Hubbard, G.F., 1981. The genus

Thalassiosira: the bipolar diatom T. antarctica Comber. Bot.

Mar. 24, 321±335.

Fryxell, G.A., Kang, S., Reap, M., 1987. AMERIEZ 1986: phyto-

plankton at the Weddell Sea ice edge. Antarct. J. U.S. 22,

173±175.

Fryxell, G.A., Kendrick, G.A., 1988. Austral spring microalgae

across the Weddell Sea ice edge: spatial relationships found

along a northward transect during AMERIEZ 83. Deep-Sea

Res. 35, 1±20.

Fryxell, G.A., Prasad, K.S.K., 1990. Eucampia antarctica var. recta

(Mangin) stat. nov. (Biddulphiaceae, Bacillariophyceae): life

stages at the Weddell Sea ice edge. Phycologia 29, 27±38.

Garrison, D.L., Buck, K.R., 1985. Sea-ice algal communities in the

Weddell Sea: species composition in the ice and plankton

assemblages. In: Gray, J.S., Christiansen, M.E. (Eds.), Marine

Biology of Polar Regions and Effects of Stress on Marine

Organisms, Wiley, New York, pp. 103±121.

Garrison, D.L., Sullivan, C.S., Ackley, S.F., 1986. Sea ice microbial

communities in Antarctica. BioScience 36, 243±250.

Garrison, D.L., Buck, K.R., Fryxell, G.A., 1987. Algal assemblages

in Antarctic pack ice and ice-edge plankton. J. Phycol. 23,

564±572.

Garrison, D.L., Buck, K.R., 1989. The biota of Antarctic pack ice in

the Weddell Sea and Antarctic Peninsula regions. Polar Biol. 10,

211±219.

Harris, P.T., Taylor, F., Pushina, Z., Leitchenkov, G., O'Brien, P.E.,

Smirnov, V., 1998. Lithofacies distribution in relation to the

geomorphic provinces of Prydz Bay, East Antarctica. Antarct.

Sci. 10 (3), 227±235.

Hasle, G.R., Sims, P.A., Syvertsen, E.E., 1988. Two recent

Stellarima species: S. microtrias and S. stellaris (Bacillario-

phyceae). Bot. Mar. 31, 195±206.

Hjort, C., BjoÈrck, S., IngoÂlfsson, OÂ ., MoÈller, P., 1998. Holocene

deglaciation and climate history of the northern Antarctic

Peninsula region: a discussion of correlations between the

Southern and Northern Hemispheres. Ann. Glaciol. 27,

110±112.

Hosie, G.W., 1994. Multivariate analyses of the macrozooplankton

community and euphausiid larval ecology in the Prydz Bay

region, Antarctica. ANARE Reports, 137, 209 pp.

Ishman, S.E., Domack, E.W., 1994. Oceanographic controls on

benthic foraminifers from the Bellingshausen margin of the

Antarctic Peninsula. Mar. Mircopaleontol. 24, 119±155.

Johansen, J.R., Fryxell, G.A., 1985. The genus Thalassiosira

(Bacillariophyceae): studies on species occurring south of the

Antarctic Convergence Zone. Phycologia 24 (2), 155±179.

Jongman, R., ter Braak, C., van Tongeren, O., 1987. Data Analysis

in Community and Ecology Landscape. Pudoc Wageningen.

Kaczmarska, I., Barbrick, N.E., Ehrman, J.M., Cant, G.P., 1993.

Eucampia index as an indicator of the Late Pleistocene oscilla-

tions of the winter sea-ice extent at the ODP Leg 119 Site 745B

at the Kerguelen Plateau. Hydrobiology 269/270, 103±112.

Kang, S., Fryxell, G., 1992. Fragilariopsis cylindrus (Grunow)

Krieger: the most abundant diatom in water column assem-

blages of Antarctic marginal ice-edge zones. Polar Biol. 12,

609±627.

Katoh, K., 1993. Deletion of less-abundant species from ecological

data. Diatom 8, 1±5.

Kozlova, A.P., 1966. Diatom algae of the Indian and Paci®c Sectors

of Antarctica. Academy of Sciences of the USSR Institute of

Oceanology, Moscow, 191 pp.

Krebs, W.N., Lipps, J.H., Burckle, L.H., 1987. Ice diatom ¯ora,

Arthur Harbor, Antarctica. Polar Biol. 7, 163±171.

F. Taylor et al. / Marine Micropaleontology 41 (2001) 25±4342

Kruskal, J.B., Wish, M., 1978. Multidimensional Scaling. In:

Lewis-Beck, M.S. (Ed.), Quantitative Applications in the Social

Sciences, Sage Publications, Newbury Park.

Leventer, A., Dunbar, R., 1988. Recent diatom record of McMurdo

Sound, Antarctica: implications for history of sea ice extent.

Paleoceanography 3 (3), 259±274.

Leventer, A., 1992. Modern distribution of diatoms in sediments

from the George V Coast, Antarctica. Mar. Micropaleontol. 19,

315±332.

Leventer, A., Dunbar, R., DeMaster, D.J., 1993. Diatom evidence

for Late Holocene climatic events in Granite Harbor, Antarctica.

Paleoceanography 8 (3), 373±386.

Leventer, A., Domack, E.W., Ishman, S.E., Brachfeld, S., McClennen,

C.E., Manley, P., 1996. Productivity cycles of 200±300 years in

the Antarctic Peninsula region: understanding linkages among

the sun, atmosphere, oceans, sea ice, and biota. Geo. Soc. Am.

Bull. 108 (12), 1626±1644.

Leventer, A., Dunbar, R., 1996. Factors in¯uencing the distribution

of diatoms and other algae in the Ross Sea. J. Geophys. Res. 101

(C8), 18489±18500.

Marchant, H., 1993. Antarctic marine nanoplankton. Curr. Top. Bot.

Res. 1, 189±201.

McMinn, A., Gibson, J., Hodgson, D., Aschman, J., 1995. Nutrient

limitation in Ellis Fjord, eastern Antarctica. Polar Biol. 15,

269±276.

McMinn, A., 1996. Preliminary investigation of the contribution of

fast-ice algae to the spring phytoplankton bloom in Ellis Fjord,

eastern Antarctica. Polar Biol. 16, 301±307.

McMinn, A., 1999. Species succession in fast ice algal commu-

nities: a response to UV-B radiation. Kor. J. Polar Res. 8,

47±52.

Medlin, L., Priddle, J., 1990. Polar Marine Diatoms. British

Antarctic Survey, Cambridge.

Nelson, D.M., Smith, W.O., 1986. Phytoplankton bloom dynamics

of the western Ross Sea ice edge Ð II. Mesoscale cycling of

nitrogen and silicon. Deep-Sea Res. 33, 1389±1412.

Nesje, A., Johannessen, T., 1992. What were the primary forcing

mechanisms of high-frequency Holocene climate and glacier

variations?. Holocene 2 (1), 79±84.

O'Brien, S.R., Mayewski, P.A., Meeker, L.D., Meese, D.A.,

Twickler, M.S., Whitlow, S.I., 1995. Complexity of Holocene

climate as reconstructed from a Greenland ice core. Nature 270,

1962±1964.

Pichon, J.J., Labracherie, M., Labeyrie, L., Duprat, J., 1987.

Transfer functions between diatom assemblages and surface

hydrology in the Southern Ocean. Palaeogeogr., Palaeo-

climatol., Palaeoecol. 71, 79±95.

Priddle, J., Fryxell, G., 1985. Handbook of the Common Plankton

Diatoms of the Southern Ocean: Centrales Except the Genus

Thalassiosira. British Antarctic Survey, Cambridge.

Reynolds, J.M., 1981. Distribution of mean annual air temperatures

in Antarctic Peninsula. Br. Antarct. Surv. Bull. 54, 123±133.

Roberts, D., McMinn, A., 1999. Diatoms of the saline lakes of the

Vestfold Hills, Antarctica, Bibliotheca Diatomologica, vol. 44.

Cramer, Berlin.

Roberts, N., 1998. The Holocene. An Environmental History.

Blackwell, Oxford.

Rosqvist, G.C., Rieti-Shati, M., Shemesh, A., 1999. Late glacial to

middle Holocene climatic record of lacustrine biogenic silica

oxygen isotopes from a Southern Ocean island. Geology 27

(11), 967±970.

Scott, P., McMinn, A., Hosie, G., 1994. Physical parameters

in¯uencing diatom community structure in eastern Antarctic

sea ice. Polar Biol. 14, 507±517.

Shevenell, A., Domack, E.W., Kernan, G.M., 1996. Record of

Holocene palaeoclimate change along the Antarctic Peninsula:

evidence from glacial marine sediments, Lallemand Fjord. Pap.

Proc. R. Soc. Tas. 130 (2), 55±64.

Silver, M., Mitchell, G.J., Ringo, D., 1980. Siliceous nanoplankton

II. Newly discovered cysts and abundant choano¯agellates from

the Weddell Sea, Antarctica. Mar. Biol. 58, 211±217.

Sjunneskog, C., Taylor, F. Postglacial marine diatom records of the

Palmer Deep, Antarctic Peninsula (ODP Leg 178, Site 1098) I:

total diatom abundance. Submitted for publication.

Smith, R.C., Ainley, D., Baker, K., Domack, E., Emslie, S.,

Fraser, B., Kennett, J., Leventer, A., Mosley-Thompson, E.,

Stammerjohn, S., Vernet, M., 1999. Marine ecosystem

sensitivity to climate change. BioScience 49, 393±404.

Stager, J.C., Mayewski, P.A., 1997. Abrupt mid-Holocene climatic

transitions registered at the equator and poles. Science 276,

1834±1836.

Stuiver, M., Braziunus, T.F., 1989. Atmospheric 14C and century-

scale solar oscillations. Nature 338, 405±408.

Stuiver, M., Reimer, P.J., 1993. Extended 14C database and revised

CALIB radiocarbon calibration program. Radiocarbon 35,

215±230.

Taylor, F., McMinn, A., Franklin, D., 1997. Distribution of

diatoms in surface sediments of Prydz Bay, Antarctica. Mar.

Micropaleontol. 32 (3/4), 209±230.

Taylor, F., 1999. Sedimentary diatom assemblages of Prydz Bay

and Mac.Robertson Shelf, East Antarctica, and their use as

paleoecological indicators. PhD thesis, University of Tasmania,

Hobart.

Truesdale, R., Kellogg, T., 1979. Ross Sea diatoms: modern assem-

blage distributions and their relationship to ecologic, oceano-

graphic and sedimentary conditions. Mar. Micropaleontol. 4,

13±31.

Villareal, T.A., Fryxell, G.A., 1983. The diatom genus Porosira

Jorg: cingulum patterns and resting spore morphology. Bot.

Mar. 33, 415±422.

Watanabe, K., 1988. Sub-ice microalgal strands in the Antarctic

coastal fast ice near Syowa Station. Jpn J. Phycol. 36, 221±229.

Whitehead, J.M., 1996. A method for correlating recent to upper

Pliocene sediment cores from the Kerguelen Plateau, Southern

Ocean. In: Ritz, D. (Ed.), AMSA `96, Abstracts. Australian

Marine Science Association, Hobart.

Zielinski, U., 1997. Parmales species (siliceous marine nano-

plankton) in surface sediments of the Weddell Sea, Southern

Ocean; indicators for sea ice environment? Mar. Micro-

paleontol. 32 (3/4), 387±395.

Zielinski, U., Gersonde, R., 1997. Diatom distribution in Southern

Ocean surface sediments (Atlantic sector): implications for

paleoenvironmental reconstructions. Palaeogeogr., Palaeo-

climatol., Palaeoecol. 129, 213±250.

F. Taylor et al. / Marine Micropaleontology 41 (2001) 25±43 43