permian polar forests: deciduousness and environmental ... et al 2012.pdfof polar forests and...

Permian polar forests: deciduousness and environmentalvariationE. L . GULBRANSON,1 ,* J . L . ISBELL,1 E . L . TAYLOR,2 , 3 P . E . RYBERG,2 , 3 T . N. TAYLOR2 , 3

AND P. P. FLAIG4

1Department of Geosciences, University of Wisconsin-Milwaukee, Milwaukee, WI, USA2Department of Ecology and Evolutionary Biology, Natural History Museum and Biodiversity Institute, University of Kansas,

Lawrence, KA, USA3Division of Paleobotany, Natural History Museum and Biodiversity Institute, University of Kansas, Lawrence, KS, USA4Bureau of Economic Geology, University of Texas-Austin, Austin, TX, USA

ABSTRACT

Forests are expected to expand into northern polar latitudes in the next century. However, the impact of

forests at high latitudes on climate and terrestrial biogeochemical cycling is poorly understood because such

forests cannot be studied in the modern. This study presents forestry and geochemical analyses of three

in situ fossil forests from Late Permian strata of Antarctica, which grew at polar latitudes. Stem size mea-

surements and stump spacing measurements indicate significant differences in forest density and canopy

structure that are related to the local depositional setting. For forests closest to fluvial systems, tree density

appears to decrease as the forests mature, which is the opposite trend of self-thinning observed in modern

forests. We speculate that a combination of tree mortality and high disturbance created low-density mature

forests without understory vegetation near Late Permian river systems. Stable carbon isotopes measured

from permineralized wood in these forests demonstrate two important points: (i) recently developed tech-

niques of high-resolution carbon isotope studies of wood and mummified wood can be applied to permin-

eralized wood, for which much of the original organic matter has been lost and (ii) that the fossil trees

maintained a deciduous habit at polar latitudes during the Late Permian. The combination of paleobotani-

cal, sedimentologic, and paleoforestry techniques provides an unrivaled examination of the function of

polar forests in deep time; and the carbon isotope geochemistry supplements this work with subannual

records of carbon fixation that allows for the quantitative analysis of deciduous versus evergreen habits

and environmental parameters, for example, relative humidity.

Received 22 February 2012; accepted 2 July 2012

Corresponding author: E. L. Gulbranson; Tel.: 530-601-1927; Fax: 414-229-4561; e-mail: [email protected]

*Present address: Department of Geology and Geophysics, University of Hawaii, Honolulu, HI, USA

INTRODUCTION

Well-preserved fossils of the extinct plant Glossopteris and

other plants from Antarctica demonstrate that the conti-

nent was vegetated in the late Paleozoic (e.g., Seward,

1914; Schopf, 1970; Francis et al., 1993; Taylor, 1996).

Apparent polar wander paths indicate that Antarctica was

near the South Pole during the Permian (Powell & Li,

1994; Lawver et al., 2008; Domeier et al., 2011; Isbell

et al., 2012), implying that Glossopteris and other flora

grew and thrived near or above the South Polar Circle.

Tree ring studies of permineralized Glossopteris wood in

Upper Permian strata of Antarctica reveal that tree rings

are dominated by earlywood with very minimal amounts of

latewood (1–3 cells wide), suggesting that the unique pho-

toperiod of the polar latitudes promoted rapid cessation of

photosynthesis at the onset of winter (Ryberg & Taylor,

2007; Taylor & Ryberg, 2007). The occurrence of leaf-rich

sedimentary layers has been used to invoke a seasonally

deciduous habit for Glossopteris (Plumstead, 1958; Retal-

lack, 1980; Taylor et al., 1992). Pigg & Taylor (1993)

observed that Glossopteris leaf compressions vary little in

leaf size and suggest that this may reflect periodic leaf drop

owing to environmental stresses similar to the habit of

© 2012 Blackwell Publishing Ltd 479

Geobiology (2012), 10, 479–495 DOI: 10.1111/j.1472-4669.2012.00338.x

evergreen trees and that more evidence is required to link

these leaf compression layers with seasonal conditions that

would indicate a deciduous habit for Glossopteris. Evergreen

trees and deciduous trees, however, exhibit distinct pat-

terns in carbon isotope ratios within tree rings (Schubert

& Jahren, 2011). Carbon isotope ratios in tree rings of

evergreen trees display symmetrical variation while decidu-

ous trees display pronounced asymmetrical carbon isotope

ratio variation within tree rings. These distinct patterns

open the possibility that high-resolution carbon isotope

analysis of fossil wood can determine a deciduous or ever-

green habit for fossil trees. This study provides additional

evidence from variations of carbon isotope ratios in per-

mineralized wood that confirm a seasonally deciduous

habit for Glossopteris and provides a promising avenue for

further work in quantifying the environmental effects on

plant growth at polar latitudes at unprecedented temporal

resolution. Extracting an isotopic signature over the life

history of several plants in an ecosystem is critically impor-

tant in quantitatively reconstructing past climates, as isoto-

pic variation in plant material provides a record of a first-

order response of the plant, via stomatal conductance and

photosynthetic rate, to the environment (Farquhar et al.,

1982, 1989; Leavitt & Long, 1991; Diefendorf et al.,

2010; Kohn, 2010). As forested polar regions have no

modern analogue (Taylor et al., 1992; Creber & Francis,

1999) they can only be studied from deep time case stud-

ies (e.g., Francis, 1990; Falcon-Lang et al., 2001; Falcon-

Lang, 2004; Jahren, 2007; Williams et al., 2008) or by

simulating environmental parameters unique to the polar

regions in growth chambers (e.g., Royer et al., 2005).

Therefore, quantification of plant–environment interactions

from deep time examples and experimentation with extant

plants via growth chambers will aid in understanding bio-

geochemical cycling and climate feedbacks of high-latitude

environments colonized by plants.

Paleoenvironmental reconstructions have suggested that

the Antarctic forests occupied a riparian niche in the Perm-

ian (Cuneo et al., 1993; Isbell and MacDonald, 1991)

implying that the associated soils maintained at least peri-

odic wetness and dryness. The complex morphology of

Glossopteris roots, known as Vertebraria, has led to the

concept that morphologic variation in Vertebraria was an

adaptation for growth in a semi-aquatic environment

(Gould, 1975). Sheldon (2006) compared Permian and

Triassic paleosols (cf. Krull & Retallack, 2000) and inferred

a change in soil horizonation through time not related to

landscape position, but rather related to the duration of

soil formation and specific environmental conditions of the

region. This indicates that depositional environment played

an important role in determining the type of soil in an

area, and subsequently helped control the plants adapted

to grow in these soils. This study presents paleosol descrip-

tions and interpretations from three locations hosting

in situ fossil forests to compare paleosol composition and

maturity with forest succession and structure.

This study aims to describe (i) the soils and sedimentol-

ogy of three Late Permian polar forests, (ii) the variation

of forests with paleolandscape position, and (iii) the poten-

tial to use stable carbon isotopes in permineralized wood

at subannual resolution to further study the environment

of polar forests and perhaps plant physiology at polar lati-

tudes. This work is the first comparison of multiple con-

temporaneous Permian fossil forests at polar latitudes and

provides an integration of paleobotany, sedimentology,

and isotope geochemistry to study carbon cycling in deep

time terrestrial environments at unprecedented temporal

resolution. Two recently discovered fossil forests are

described here for the first time from localities on Graphite

Peak (85.05251ºS, 172.34792ºE) and near Wahl Glacier

(84.09500ºS, 165.33167ºE), and we compare these forests

with a previously studied fossil forest on Mt. Achernar

(84.17824ºS, 160.90057ºE) supplemented with geochemi-

cal analysis of Glossopteris wood from Mt. Sirius and

Mt. Achernar (Fig. 1).

BACKGROUND

Late Permian paleogeographic reconstructions place the

Transantarctic Mountain region between 70 and 80ºS(approximately 75ºS for the study area), which is well

within the South Polar Circle (Powell & Li, 1994; Torsvik

& Cocks, 2004; Lawver et al., 2008; Domeier et al.,

2011). The stratigraphy presented here is part of the Vic-

toria Group of the Beacon Supergroup, which is earliest

Permian to Triassic in age (Kyle & Schopf, 1982; Barrett

et al., 1986; Collinson et al., 1994; Isbell et al., 2001).

The fossil forests studied herein are found in the upper

member of the Buckley Formation (Fig. 2), which is dated

to the Late Permian on the basis of palynology (Kyle &

Schopf, 1982; Farabee et al., 1990, 1991) and from detri-

tal zircons that exhibit an abundance of zircons of Late

Permian age (Elliot & Fanning, 2008). The oldest sedi-

mentary rocks in the studied depositional basin are earliest

Permian (i.e., Asselian/Sakmarian, Kyle & Schopf, 1982),

indicating that the Transantarctic Basin was actively subsid-

ing at least by the Carboniferous-Permian boundary

(approximately 299 Ma, Gradstein et al., 2004).

Fossil forests in the Permian of Antarctica are exclusive

to the Late Permian Buckley Formation and its strati-

graphic equivalents (Doumani & Minshew, 1965; Cuneo

et al., 1993; Francis et al., 1993). A published description

is only available for the forest on Mt. Achernar, whereas

Permian fossil forests are known to exist in the central

Transantarctic Mountains at Wahl Glacier, Lamping Peak,

McIntyre Promontory, and Mt. Picciotto in the upper

Buckley Formation near the Beardmore and Shackleton

glaciers; and Mt. Weaver in the Queen Maud Mountains

© 2012 Blackwell Publishing Ltd

480 E. L. GULBRANSON et al.

near the Robert Scott Glacier. The fossil forest in the

upper Buckley Formation at Mt. Achernar (Figs 1 and 2)

is notable in that the preserved tree stumps have diameters

of 9–18 cm with no more than 15 rings per trunk, imply-

ing that this was a stand of saplings or juvenile trees (Tay-

lor et al., 1992). Fossil leaf mats and permineralized peat

from the upper Buckley in the study area suggest that

these polar forests were low diversity and probably did not

have an understory flora (Cuneo et al., 1993). However,

the Mt. Achernar in situ fossil forest does contain evidence

of understory vegetation comprised of herbaceous lycops-

ids, preserved as compression/impression fossils, whereas

the permineralized tree stumps of the forest are attributed

to the Glossopteris seed plant (Schwendemann et al., 2010).

Despite the fact that Glossopteris trees grew at polar lati-

tudes, Taylor et al. (1992) found no evidence of frost rings

in the fossil forest at Mt. Achernar.

The upper Buckley Formation is primarily composed of

volcaniclastic sedimentary rocks with terrestrial facies of flu-

vial and lacustrine affinity (Isbell et al., 1997; Isbell and

MacDonald, 1998), but coal also occurs within this part of

the formation and it is notable that coal is absent in Ant-

arctica between the top of the Buckley Formation and the

upper portion of the overlying Triassic Fremouw

Formation (Retallack et al., 1996). The Buckley Formation

has local variations in grain size (Barrett et al., 1986; Isbell

et al., 1997). Some regions are predominantly sand rich,

but coeval stratigraphic intervals in other areas are com-

posed of thinner sandstone beds interbedded with thicker

beds of shale and silty sandstone containing either carbona-

ceous plant detritus or non-carbonaceous plant compres-

sions (Briggs et al., 2010).

Glacial deposits of the late Paleozoic ice age do not

occur in Antarctica after (i.e., younger than) the Early

Permian (Isbell et al., 2003, 2008, 2012); however, glacia-

tion is interpreted to have existed at lower paleolatitudes in

eastern Australia until the Late Permian (Fielding et al.,

2008). Therefore, it is likely that the Late Permian time

interval in which these forests developed represents a tran-

sitional climate period from the icehouse(s) of the late

Paleozoic to the greenhouse state of the Mesozoic.

METHODS

Measured stratigraphic sections (Fig. 2) were made at

Graphite Peak, Wahl Glacier, and Mt. Achernar noting

sedimentary structures, sediment composition, and fossils.

Three permineralized stumps and root material were

RO

SS

I

CE

SH

EL

F

B o w d e n N é v é

W a l c o t t N é v é

Ma

rsh

Gla

cie

r160ºE

170ºE

85ºS

84ºS

83ºS

Mil

ler

Ra

ng

e

Skaar Ridge

Mt. AchernarGordon Valley

Mt. Falla

Fremouw Peak

Lamping peak

Mt. Sirius

Wahl Glacier

Keltie Glacier

5 S

Graphite Peak

BE

AR

DM

OR

E

GL

AC

IE

R

4 5N

A

U

TI

CA

LM

IL

ES

CTAM

Co

mm

on

we

al

th

Ra

ng

e

Qu

ee

n

El

iz

ab

et

hR

an

ge

4 0 k m0

McMurdo

South Pole

L. Vostok

WES

T

ANTAR

CTIC

A

EAST ANTARCTICA

StudyArea

Weddell Sea

Ross Sea

0º

180º

Eº09Wº09

Qu

ee

nA

le

xa

n

dr

aR

an

ge

Supporters

R

ange

Permian fossil forest locality studied herein

Permian sections studied in 2010-2011

Triassic sections studied in 2010-2011

Law Glacier

P o l a r P l a t e a u

Permian fossil forest locality with Triassic strata

Permian and Triassic sections studied in 2010-2011

LatePermian

paleoflow

300 Ma

275 Ma

250Ma

polar wander (Domeier et al., 2011)

EastAntarctica

Africa

India

Australia

S.Amer.

studyarea

300Mapola

rcirc

le

C

B

A

Fig. 1 Map of the study area within Antarctica. (A) Present-day map of the study area, gray areas represent exposed rock of the Transantarctic Mountains.

Symbols denote studied locations during the 2010–2011 austral summer season; star symbols represent fossil forest localities studied herein. Large arrow

denotes general Late Permian paleoflow direction compiled from paleocurrent measurements throughout the Beardmore Glacier region. (B) Map of Antarctica

showing position of study area on the continent. (C) Paleogeographic reconstruction showing the position of the study area (star) relative to the rest of

Gondwana (gray areas). The most recent paleo-polar wander path for the studied time interval (Domeier et al., 2011; Isbell et al., 2012) is shown for

reference.


Permian polar forests 481

sampled for stable isotope analysis at Graphite Peak, taking

care to preserve as much of the fossil in place. Six stumps

were collected from Wahl Glacier, and four stumps were

collected from Mt. Achernar for stable isotope analysis.

Existing samples are housed in the Department of Geology

and Geophysics, University of Hawaii, but will be perma-

nently deposited in the Paleobotanical Collections, Natural

History Museum, University of Kansas (Taylor et al.,

1992) under accession numbers 17169 A, B top and bot-

tom, and C for the Mt. Achernar specimen, and accession

number 17170 for the Mt. Sirius specimen. Leaf compres-

sions and the surrounding sedimentary rocks were also

sampled for geochemical and petrographic analysis.

We analyzed fossil forests at Graphite Peak and Wahl

Glacier localities by measuring the spacing between trees

and inferring tree height from the stem diameter (Table 1)

0

10

20

30

40

50

60

70

80

90

100

110

93

203T

hic

knes

s (m

)

Triassic Fremouw Fm.

Upper Permian upper Buckley Fm.

fossil forest

Trough cross-bedding

Current ripple lamination

Horizontal bedding

Coal

In situ stumps

Leaf compression

Explanation

co.md.f.vf.sil.cl/coalGrain Size

heig

ht fr

om b

ase

of s

ectio

n (m

)

0

10

20

30

40

50

60

70

80

90

100

110

130

240Triassic Fremouw Fm.

Upper Permianupper Buckley Fm.

0

10

20

30

40

50

60

50 m

co.md.f.vf.sil.cl/coalGrain Size co.md.f.vf.sil.cl/coal

Grain Size

GRAPHITE PEAK

WAHL GLACIERMOUNT ACHERNAR

Fig. 2 Stratigraphic columns at Graphite Peak, Wahl Glacier, and Mount Achernar. Only the thickness that contains the fossil forest and the contact between

the upper Buckley and Frewmouw fms. are shown. Grain size on the x-axis increases from left to right.



to understand the canopy structure. Carbon isotopes are

measured from permineralized wood at 1.5–2-mm intervals

in the direction of wood growth (i.e., radially), to infer the

annual or seasonal uptake of CO2 from the Late Permian

atmosphere. The carbon isotope measurements are likely

indicative of seasonal carbon fixation because the sampling

intervals are much thinner than the tree ring widths, which

range from 4 to 11 mm. Sampling tree rings at higher res-

olution (e.g., 0.1-mm intervals) was completed to observe

whether the carbon isotope pattern of a tree ring changes

with increased sampling density.

Tree height estimates were determined using the transfer

functions of Niklas (1994) for woody species that correlate

stem diameter to tree height (Table 1). We calculated the

95% confidence interval for the original data set of Niklas

(1994), which indicates that stem diameters must differ

between 6 and 10 cm to produce statistically significant

tree height estimates (Fig. S1). Forestry analysis of the Mt.

Achernar fossil forest has been previously completed. This

manuscript presents new field measurements of tree posi-

tions for Graphite Peak and Wahl Glacier to quantify tree

density and basal area using the point-centered quarter

method (Cottam & Curtis, 1956; Cuneo et al., 2003).

The point-centered quarter method is designed to estimate

the mean area of plants by measuring the distances of

plants to arbitrary sampling points along a transect. Each

sampling point is divided into four quadrants and the dis-

tance between the sampling point to the closest tree in

each quadrant is used as an individual sample and the aver-

age distances of all the samples is equal toffiffiffiffiffi

Mp

, where M

is the mean area of plants (Cottam & Curtis, 1956).

Samples of tree rings were cut from permineralized

wood with a tile saw equipped with a wafer blade at the

University of Wisconsin-Milwaukee. The tree ring samples

were split in half along the tangential plane such that the

two halves represent one half that is predominantly early-

wood and one half that has earlywood and a portion of

latewood. The halved tree rings were crushed in an agate

mortar to pass through a 40-lm sieve and combusted off-

line for carbon isotope ratio measurement on a Thermo

Finnigan MAT 253 (and a Finnigan MAT 252 for small

volumes) in dual inlet mode at the Huffington Department

of Earth Sciences, Southern Methodist University. XRD

analysis of homogenized powders of the sample did not

indicate the presence of carbonate minerals, and carbonate

minerals were not observed in thin section, and therefore,

contamination from inorganic carbon sources is unlikely.

Samples were combusted in sealed Vycor tubes at 900 °Cfor two hours followed by a linear decrease in temperature

to 650 °C at 1 °C min�1, temperature was held at 650 °Cfor two hours followed by cooling to 25 °C overnight.

Carbon dioxide produced during the combustion of the

sample was cryogenically purified from water vapor, SOx,

and non-condensable gasses. Reaction with 0.5 g native

Cu was complete for the majority of samples, and there-

fore, the cryogenic distillation procedure was modified to

separate CO2 from the large quantities of SOx possibly

remaining in the Vycor tubes. A mixture of liquid nitrogen

(LN) and n-pentane (�131 °C) was made in place of the

usual dry ice methanol/ethanol slush (�72 °C) to facilitate

the separation of CO2 from SOx by lowering the tempera-

ture closer to the freezing point of sulfur oxides. The gas

was passed through a LN trap and non-condensable gasses

were evacuated. The LN trap was quickly replaced with the

LN-n-pentane trap to maintain temperatures close to

liquid nitrogen temperature during the switch. CO2

evolved rapidly from the LN-n-pentane trap and was

allowed to be collected in a second LN trap. The amount

of purified CO2 gas was measured manometrically

and transferred to borosilicate tubes for isotope ratio

measurement.

The results of the offline combustion and dual inlet iso-

tope ratio measurements indicated that sampling at higher

spatial resolution (i.e. 0.1-mm sample intervals within tree

rings) might be possible. In advance of this higher-resolu-

tion sampling strategy we analyzed replicate samples from

the dual inlet data set via an elemental analyzer (EA)

interfaced to an isotope ratio mass spectrometer in contin-

uous flow mode at the Stable Isotope Facility at the Uni-

versity of California-Davis to assess reproducibility

between these very different analytical procedures. Sec-

ondary reference materials (nylon d13C = �27.81&, peach

leaves d13C = �26.12&, and bovine liver

d13C = �21.69&), which are calibrated against primary

reference materials, were analyzed within each batch run

to correct and normalize the measured values from the

sample. All materials resulted in precision better than

0.1& (1r). A quality assurance standard (USGS-41 Glu-

tamic Acid, d13C = +37.626&) was run within each batch

and had a standard deviation (1r) <0.3& over the course

of the experiment with an average d13C value of

+37.62&.

Table 1 Permineralized stump diameters at Graphite Peak

Stump

Stem

diameter

(cm)

Tree

height

(m)*

Basal

area

(m2)† Latitude (ºS) Longitude (ºE)

1 60 31 0.28 85.05251 172.34792

2 32 23 0.080 85.05251 172.34798

3 43 26 0.15 85.05260 172.34724

4 28 21 0.062 85.05262 172.34709

5 30 22 0.071 85.05279 172.34688

*Tree height inferred from transfer function: Log(y) = 1.59 + 0.39 Log (x)

� 0.18[Log(x)]2, where y is tree height in meters, and x is stem diameter in

meters (Niklas, 1994).†Basal area (BA) calculated as BA = dbh2 9 7.854

(10�5), where dbh is diameter-at-breast-height in centimeters.Total basal

area [m2 ha�1] is calculated by summing all of the basal areas of trees in a

stand and dividing by the area of the studied.



To sample permineralized wood at a spatial resolution of

approximately 0.1 mm thickness we milled samples by

hand with a rotary tool equipped with a diamond bit and

monitored depth of sampling with calipers. Sampled sec-

tions of permineralized wood (0.050–1.0 mm thick) from

the upper Buckley Formation at Mt. Sirius were prepared

for analysis via elemental analyzer-continuous flow isotope

ratio mass spectrometry (EA-CFIRMS) at the laboratory of

Dr. A. Hope Jahren at the University of Hawaii-Manoa.

The sampling strategy is designed to make ultra-thin sam-

ples of permineralized wood sequentially along the direc-

tion of wood growth and has the potential to directly

sample latewood, which is commonly a few cells thick

(0.01–0.2 mm thick) in these samples (Ryberg & Taylor,

2007; Taylor & Ryberg, 2007). For the determination of

evergreen habit versus deciduous habit from stable isotope

trends in tree rings it is important to isolate latewood from

earlywood to conclusively determine whether a wood sam-

ple has an asymmetric carbon isotope profile in the tree

ring (deciduous habit) or a symmetrical carbon isotope

profile (evergreen habit). Samples are ground away from

the permineralized wood sample, massed, and sealed in a

tin capsule for EA-CFIRMS analysis thereby permitting rel-

atively rapid collection of isotope data from multiple sam-

ples within a single tree ring and for several rings. This

method can be envisioned as an improvement over the

more analytically precise dual inlet measurements in that a

greater portion of the total isotopic signal can be extracted

from a tree ring (Schubert & Jahren, 2011).

RESULTS

Graphite peak

During the 2010–2011 austral summer season in Antarc-

tica we discovered a Permian in situ fossil forest at Graph-

ite Peak near the Keltie Glacier (Fig. 1), which occurs 99

m below the contact of the upper Buckley and the Triassic

Fremouw fms (Fig. 2). The stumps are found upright with

intact roots in a approximately 50 cm thick, thinly bedded

(<1 cm thick), light gray siltstone (Fig. 3A–C) that con-

tains carbonaceous Glossopteris leaf compressions on bed-

ding planes. The compressions preserve leaf morphology

with both the blade and attached petiole in many speci-

mens. The siltstone, interpreted as a paleosol, is laterally

continuous for >100 m and is overlain by 6.5–7 m of par-

tially covered siltstone capped by a conspicuous, partially

silicified coal seam. Root structures were excavated from

A B

C

E

D

F

Fig. 3 Field photographs of permineralized

stumps. (A–B) Drawing overlay and

photograph of permineralized stump 1 from

Graphite Peak showing upright position, intact

roots, and two sedimentary units ‘i’ and ‘ii’.

Unit i is an organic-rich siltstone into which

the roots penetrate, unit ii corresponds to the

siltstone with leaf compression fossils that the

stump is set within and covered by. (C)

Upright permineralized stump 3 at Graphite

Peak; tree rings can be observed in the

transverse plane (top surface). (D) Upright

permineralized stump 2 at Graphite Peak.

Note that the only the high-relief feature is

permineralized wood; the depression is infilled

with blocky gray siltstone. (E) Upright

permineralized stump at Wahl Glacier. (F)

Upright stump at Wahl Glacier with attached

root trending away from the stump toward

the bottom of the image.



the in situ stumps and these roots were traced tens of cen-

timeters into the thinly bedded gray siltstone (Fig. 3A,B),

tapering downwards. In cross section the roots are com-

pletely permineralized but lack the diagnostic lacunae of

many Vertebraria roots, although this may be due to their

proximity to the root-shoot transition zone (Neish et al.,

1993; McLoughlin et al., 1997, 2005; Decombeix et al.,

2009). Vertebraria was also observed in float in the overly-

ing beds. In total, six in situ stumps were observed in the

exposed bedding plane; five stumps were completely per-

mineralized and one was permineralized only in the outer-

most portions of the stump, whereas the innermost

portions were filled with gray, blocky, siltstone (Fig. 3D).

Sandstones that crop out above and below the fossil for-

est layer are laterally extensive across the outcrop having a

sheet-like geometry. The sandstones are fine- to medium-

grained, with a subangular to subround texture; the com-

position is dominated by lithic fragments, although mud-

stone clasts and quartzite clasts are observed at the base of

sandstone beds, and the sandstones truncate underlying

strata at the centimeter-scale. Trough cross-bedding and

ripple laminations are observed within the sandstone beds.

Interpretation of Graphite Peak Stratigraphy

The sandstone lithofacies at Graphite Peak above and

below the fossil forest layer represent fluvial channel and

floodplain (splays, mires, lakes) deposits of low-sinuosity

streams (Fig. 2). Paleoflow directions for fluvial sandstones

at Graphite Peak are oriented toward the South Pole and

represent transverse flow relative to axial drainage patterns

directed toward South Victoria Land (Collinson et al.,

1994; Isbell et al., 1997; Fig. 1). Sandstones at Graphite

Peak contain mudstone rip-up clasts and quartzite pebbles

near the sandstone base, indicating erosion of underlying

strata during the entrenchment of the channel system. The

fine-grained texture and weak horizontal bedding of the

fossil forest layer suggests a low energy environment of

deposition. These sedimentary rocks are likely floodplain

deposits formed from the continued aggradation of sedi-

ment after an avulsion episode (Smith et al., 1989). This

interpretation is consistent with the low degree of pedo-

genic modification of these sedimentary rocks inferred

from the lack of field-based indications of soil formation

(e.g., horizonation, reddening, clay accumulation, redoxi-

morphic features), where pedogenesis is punctuated by the

continued addition of sediment (Kraus & Aslan, 1993).

Wahl glacier

During the 2003–2004 austral summer season we (J.L. Is-

bell) discovered the remains of a fossil forest in the upper

Buckley Formation at Wahl Glacier, near Walcott Neve

(Knepprath, 2006; Fig. 1). The discovery of the Wahl

Glacier and Graphite Peak forests brings the total number

of in situ fossil forest localities in the Beardmore Glacier

region to four, with the other two being the Permian

forest at Mt. Achernar (Taylor et al., 1992) and a Triassic

forest in the Fremouw Formation at Gordon Valley

(Cuneo et al., 2003). The fossil forest at Wahl Glacier is

approximately 100 m below the contact between the Buck-

ley and Fremouw formations (Fig. 2). Thirteen permineral-

ized stumps are found upright with intact root systems

that penetrate underlying sedimentary rocks (Fig. 3E,F).

The root systems trend horizontally away from the stumps.

The sedimentary rocks surrounding the permineralized

stumps are divided into a lower siltstone that contains the

root systems and an overlying sandstone that covers most

or all of the stumps. The lower siltstone is gray with a

white patina, thinly bedded, and contains leaf compres-

sions. The overlying sandstone is brown, fine- to medium-

grained and medium-bedded, and contains crude

cross-stratification; the basal contact of the sandstone trun-

cates underlying siltstone. The stratigraphic succession of

the upper Buckley Formation at Wahl Glacier has a general

fining upward trend in grain size, and the fossil forest

layer, including the sandstone surrounding the stumps, is

bounded by organic-rich shale.

Interpretation of Wahl Glacier stratigraphy

The sedimentary strata surrounding the fossil forest at

Wahl Glacier, like Graphite Peak, contain both thick and

thin lithic sandstones with sheet-like geometries that

contain trough cross-bedding, organic-rich shales, thin

laminated siltstones, and coal. Therefore, these sandstones

are interpreted as low-sinuosity channel, splay, lake, and

mire fluvial deposits. The mean paleocurrent vector of

these sandstones indicates an oblique direction (i.e., trans-

verse flow) relative to the basin axis, with a general orienta-

tion toward South Victoria Land. The fossil forest layer,

however, is found rooted within a thinly bedded, gray silt-

stone that is in contact with crudely laminated sandstones.

The environment of deposition at Wahl Glacier for the fos-

sil forest layer is likely a channel margin or forested levee/

splay complex of a low-sinuosity fluvial system, suggesting

a closer proximity to the fluvial system than the Graphite

Peak stumps that are surrounded by finer-grained and

thinly bedded strata representing a relatively distal portion

of a splay complex (cf., Smith et al., 1989).

Mount Achernar

We studied two Permian fossil forest levels at Mt. Achernar

in the vicinity of the Law Glacier (Fig. 1). The overlying

Fremouw Fm. does not crop out at this locality; however,

comparison with surrounding locations suggests that this

section of upper Buckley occurs approximately 50 m below



the upper Buckley Fm. and Fremouw Fm. contact (Fig. 2).

The basal 3 m of this section consists of a massive white

siltstone and fine root haloes that branch downwards but

diverge horizontally at approximately 1 cm depth. The silt-

stone has sharp upper and lower boundaries and interbed-

ded white siltstone beds with crude bedding that contain

Glossopteris leaf compressions and carbonaceous woody tis-

sue. The preservation of the fossil plant material is variable

with some fossils preserving only vein skeletons and other

beds preserving most of the blade attached to the petiole.

Glossopteris reproductive structures are present in some of

the beds as well (Ryberg et al., 2011).

Overlying these units is approximately 1 m of coal inter-

bedded with medium- to thin-bedded (5 cm to <1 cm)

white siltstone with abundant plant debris on bedding

planes. The coal contains woody tissue and leaf fragments.

Greenish brown mudstone clasts occur within the white

siltstone. Overlying the coal and siltstone is approximately

1 m of medium-bedded, interbedded siltstone with a white

patina, blue-gray interior, and rare ripple laminations. Fos-

sil leaf fragments are abundant on some bedding planes.

Two truncation surfaces were noted that cut through the

underlying siltstone; these have a lenticular geometry and

are roughly 1.5 m wide with bedding that onlaps onto the

truncation surface. The upper 2 m of the section contains

the two beds hosting the in situ tree stumps and roots

(Fig. 4A,B).

Interpretation of Mount Achernar stratigraphy

Mt. Achernar contains evidence of subaerial exposure in

the form of rooted siltstones interpreted as weakly devel-

oped paleosols (Protosols) and coals (Histosols). Unlike

Graphite Peak (Fig. 5A,B), however, the environment of

deposition at Mt. Achernar is interpreted as being lower

energy on the basis of an abundance of fine-grained sedi-

mentary rocks near the fossil forest beds (Fig. 5C,D).

Channelization of some siltstone beds indicates water

movement and transport of material (Fig. 5D), and the

horizontal bedding with abundant plant remains implies a

subaqueous environment with low energy (e.g., rare ripple

laminations) or perhaps standing water allowing for sus-

pension settling of material. Interbedding of splays, coal,

and siltstone suggest deposition along the margins of a

shallow lake. These environments are consistent with a

shallow lacustrine system or wetland with fluvial input of

sediment. Therefore, the lower energy lacustrine, marginal

lacustrine, mire, and splay environments interpreted at Mt.

A B

Fig. 4 (A) Upright permineralized stump at

Mt. Achernar in white siltstone. (B)

Permineralized root (arrow) at Mt. Achernar

trending subhorizontally in siltstone.

A B

C D

Fig. 5 Field photographs of fossil forest sites.

(A) upper Buckley Fm. and upper Buckley/

Fremouw fms. contact at the Graphite Peak

locality. Fossil forest location indicated by solid

black line, dashed where inferred laterally. (B)

Sedimentary units (i, ii) that contains the fossil

forest at Graphite Peak, note the preservation

of bedding throughout. See explanation in

Fig. 3. (C) upper Buckley Fm. at Mt.

Achernar. Fossil forest is marked (light gray

bench) as is the measured section. (D) Bedded

and laminated siltstone lithofacies at Mt.

Achernar. Fossil forest layers are in the white

colored units at the top of the photograph.

Note the small cut-and-fill structure in the

center of the photograph.



Achernar are likely part of a mosaic of landscapes that were

laterally equivalent to low-sinuosity channel and floodplain

fluvial deposits at Wahl Glacier and elsewhere in the Beard-

more Glacier region (Isbell et al., 1997).

Paleoforestry

The permineralized stumps at Graphite Peak have diame-

ters that range from 28 to 60 cm (Fig. S1), which corre-

spond to tree heights that range from 21 to 31 m

(Table 1). The mean distance between trees in the Graph-

ite Peak forest is 20.8 m (Fig. S3A), which corresponds to

a density of 23 individuals ha�1 using the method of Cot-

tam & Curtis (1956). Moreover, these measurements were

made on upright stems found with attached roots that

penetrate siltstone (Figs 3, 4 & 6), and we interpret these

fossil wood samples as being in situ. Stem diameter mea-

surements for the fossil forest at Mt. Achernar are pub-

lished as a range of 9–18 cm (Fig. S3B, Taylor et al.,

1992) for fifteen permineralized stumps, and correspond

A

B C

D E

Fig. 6 Fossil forest photographs. (A) Two

upright permineralized stumps from Graphite

Peak (stumps 4 and 5) found within light gray

siltstone (unit ‘i’ of Figs 3 & 5). (B) Upright

stump at Graphite Peak (stump 1) showing

two intact roots trending down into organic-

rich siltstone (unit ‘ii’ of Figs 3 and 5). (C)

Glossopteris leaf impressions indicated by

ovals, and the reproductive structure

Plumsteadia (Taylor et al., 2007) indicated by

the squares on a bedding plane from Mt.

Achernar. (D) Upright stump from Wahl

Glacier fossil forest with at least two attached

roots. Brown sandstones covers most stumps

and exhibits an erosional contact with the

underlying siltstone. (E) Glossopteris leaf

impression from Graphite Peak.



to tree heights ranging from 10 to 16 m, respectively (Fig.

S1).

Permineralized stumps at Wahl Glacier have a mean

diameter of 25.6 cm (Fig. S1), which corresponds to a

mean tree height of 19.2 m. The mean distance between

trees at Wahl Glacier is 6.2 m and corresponds to a tree

density of 263 individuals ha�1 (Fig. S3C), which reflects

an average obtained by using the point-centered quarter

method and by dividing the number of stumps by the

area of the exposed surface (Knepprath, 2006). It is

noteworthy that several permineralized stumps at Wahl

Glacier contain secondary mineralization of sulfur-bearing

minerals, such as pyrite or chalcopyrite, near the pith

of the stump, whereas Graphite Peak permineralized

wood appears to be homogeneous with respect to

mineralization.

Geochemistry

Preliminary geochemical analyses were made on a permin-

eralized Glossopteris stump from Mt. Achernar. XRD

analysis of powdered and mixed permineralized wood from

Mt. Achernar indicates that quartz is the only mineral pres-

ent in the sample. Offline combustion and cryogenic sepa-

ration of CO2 allowed for high-precision measurements of

the carbon yield of the samples (Fig. S2A,B). A range of 3

–23 mg of sample was combusted and corresponds linearly

to a range of CO2 quantities of between 9 and 31 lmol,

respectively (Table 2). Thus, the average weight percent of

carbon in a bulk sample of this permineralized wood is 2%

(±0.4% 1 SD).

Stable carbon isotope ratios were measured within

growth rings of permineralized wood to assess whether or

not seasonal variation observed in extant plants (e.g., Fran-

cey & Farquhar, 1982; Leavitt & Long, 1991; Helle &

Schleser, 2004; Schubert & Jahren, 2011) occurs in the

fossil plants and whether the variation in carbon isotope

ratios can elucidate whether Glossopteris maintained a

deciduous or evergreen habit. For the Mt. Achernar sam-

ple, measured d13C values range from �21.4& to

�24.9& (Table 2), with an average d13C value of �23.2&(±0.9& 1SD). Carbon isotope ratios vary within tree rings

with the beginning of a ring having the most positive d13Cvalue and the later portion of the ring having the most

negative d13C value (Fig. 7). This pattern repeated for the

majority of rings analyzed, with exception of the exterior

of the sample that showed a pronounced enrichment in13C relative to samples from the interior of the specimen.

In this outer part of the stump, the beginning of the rings

had a fairly consistent d13C value of approximately

�22.6&, whereas the values for the later portion of the

rings varied from �24& to �25&. Thus, d13C values vary

by 1–2& within tree rings in the permineralized wood

specimen from Mt. Achernar.

Replicate samples from Mt. Achernar analyzed via

EA-CFIRMS, a technique that will efficiently accommodate

an increase in the number of samples analyzed, were sys-

tematically more positive than the samples analyzed via

dual inlet by approximately 0.2& (Fig. S2C). The system-

atic and small offset between the dual inlet and elemental

analyzer methods indicates that the samples analyzed via

EA-CFIRMS will likely preserve the same isotopic signals

identified with dual inlet measurements.

Ultra-thin sections (<0.1 mm thick) of permineralized

wood from an Araucarioxylon specimen collected in the

uppermost Buckley Fm. at Mt. Sirius (Fig. 1) were ana-

lyzed for carbon isotope ratios via EA-CFIRMS to test

whether a deciduous or evergreen habit can be inferred

from high-resolution isotope geochemistry of fossil wood

(Fig. 8). Modern deciduous and evergreen trees display

pronounced differences in intra-ring carbon isotope ratios,

where deciduous trees show an asymmetric, three-phase

trend with more negative values in latewood, increasing

through the ring boundary an into the earlywood of a sub-

sequent tree ring (Helle & Schleser, 2004). On the other

hand, evergreen trees display a symmetric carbon isotope

trend throughout a tree ring, with more positive values in

the earlywood (Schubert & Jahren, 2011). The results

show clear isotopic variation within tree rings (Fig. 8). The

d13C values are more negative than the Mt. Achernar

Table 2 Carbon yield and d13C from Mt. Achernar

Sample*,†Mass

(mg)

Carbon yield

(lmol�C)d13C

(& VPDB)

2p 17.5 30.7 �22.7

2c 5.1 10.5 �23.9

3p 6 11.8 �23.7

3c 16.1 23.8 �22.6

4p 14.8 22.4 �22.9

4c-1 7.3 10.9 �23.7

4c-2 5.7 9 �24

5p 13.5 21.1 �22.6

5c 4.6 6.04 �24.9

6p-1 18.2 22.8 �22.8

6p-2 11 12.6 �23.3

6c-1 9.8 18 �23.3

6c-2 11.2 16.5 �23

7p-1 2.6 5.9 �24.9

7p-2 26.8 37.6 �22.44

7p-3 25.9 33.2 �22.5

7c 17.3 23.8 �22.6

8p 3.7 9.1 �23.9

8c 20.5 27.7 �22.5

9p 13.9 18.8 �22.3

9c 23 31.3 �21.4

*Samples are referenced by ring number ‘2’ and whether they are early-

wood dominated ‘p’ or a mixture of earlywood and latewood ‘c’. Replicate

analyses are indicated by the number, for example, 7p-2 is the second rep-

licate analysis of sample 7p.†University of Kansas accession number for the

sample is 17169B top and bottom, the sample is from Mt. Achernar, upper

Buckley Formation, Antarctica.



stump, by up to 2&, and the magnitude of d13C value var-

iation within a tree ring is approximately 1& (Table 3).

Moreover, an important trend is observed between the two

tree rings in the Mt. Sirius sample. d13C values are most

positive in the initial wood of the tree ring and then

become progressively more negative throughout the tree

ring, exhibiting a change toward more positive values near

the earlywood/latewood boundary, which continues into

the next tree ring. This three-phase variation in carbon sta-

ble isotopes is similar to stable isotope variation observed

in extant deciduous trees (Helle & Schleser, 2004). As

demonstrated by Schubert & Jahren (2011), an increase in

the number of samples per tree ring will yield an increase

in the amount of seasonal signal recorded in stable isotope

variations. Therefore, the sampling approach used on the

Mt. Sirius specimen is clearly more advantageous than the

approach used for the Mt. Achernar specimen as 60–80%

of the total isotopic signal was recovered from the tree

rings in the Mt. Sirius sample. However, potentially impor-

tant differences in d13C values and the range of d13C val-

ues within tree rings between the two localities are

apparent despite the differences in sampling strategy and

isotope analysis.

DISCUSSION

Fossil forests in the upper Buckley Formation at Graphite

Peak and Wahl Glacier occur at similar lithostratigraphic

positions approximately 100 m below the contact between

A

B

Fig. 7 Mt. Achernar wood and carbon isotope values. (A) Transverse sec-

tion of permineralized wood from Mt. Achernar used for isotope ratio mea-

surements. (B) Tree ring d13C values, gray shading indicates where two

samples were taken from the same ring, other values likely represent one

d13C value for the entire ring. Arrows point to region in the stump that the

data points correspond to. Dashed vertical lines denote tree ring bound-

aries. Note peculiar increase in d13C values toward the exterior of the sam-

ple (to the right).

A B

Fig. 8 Carbon isotope ratios for the permineralized wood specimen from Mt. Sirius. (A) Individual data points (circles) were used to calculate a 3-point run-

ning average (black line). Photograph (inset) of the section of permineralized wood used for sampling. Sample widths are indicated by the widths of the alter-

nating black and gray bars, with 10 samples for rings 1 and 2, sample numbers are shown for ring 1 for reference. Note that sample widths <0.1 mm

(Table 3) are difficult to show clearly. (B) Idealized carbon isotope trends within tree rings for a deciduous tree and an evergreen tree generalized from a glo-

bal compilation of high-resolution isotope records of modern wood (modified from Schubert & Jahren, 2011). The Mt. Sirius sample displays carbon isotope

trends that are more similar to a tree of a deciduous habit, having a three-phase carbon isotope trend, than an evergreen habit that has a prominent positive

d13C value peak yet similar d13C values at the beginning and end of a growth ring.



the Buckley and Fremouw formations. The fossil forest at

Mt. Achernar occurs within the upper Buckley Formation;

however, the Fremouw Formation does not crop out at

this locality, and so it is not possible to definitively corre-

late the fossil forest at Mt. Achernar with Graphite Peak

and Wahl Glacier other than to assign it a Late Permian

age on the basis of palynology (Farabee et al., 1990,

1991). In the following discussion, we interpret the nature

of the forests at each locality in regard to the environment

of deposition. The preliminary stable carbon isotope data

will be discussed in terms of the potential for recovering

high-resolution geochemical archives from these fossil

plants.

Retallack & Krull (1999) interpreted broad areas of Ant-

arctic landscapes in the Permian as densely forested, a

monotonous extension of vegetation observed from sedi-

mentary deposits of fluvial-floodplain affinity. Such general-

izations are problematic as the preservation of plant fossils

is biased toward riparian environments and without direct

observation from other environments, it is impossible to

describe the vegetative habit and ecology without incurring

significant errors in paleoclimate/paleoforestry reconstruc-

tion (Demko et al., 1998). The occurrence of only Glossop-

teris leaves (Fig. 6C,E) in the Graphite Peak fossil forests

implies that the stand was low diversity, which is consistent

with Late Permian fossil forests elsewhere in Antarctica

(Cuneo et al., 1993). In addition, no other fossil macrofl-

ora was observed in the fossil forest and this suggests the

absence of understory vegetation at Graphite Peak in con-

trast to the Mt. Achernar in situ forest that does contain a

herbaceous lycopsid understory (Schwendemann et al.,

2010). The field measurements of spacing between trees

indicates that the preserved trees at Graphite Peak and

Wahl Glacier are widely spaced relative to the fossil forest

at Mt. Achernar and were likely more mature trees relative

to Mt. Achernar (Fig. S3; Fig. 9A). There is overlap

between tree height estimates at the fossil forest localities,

however, taking into account the 95% confidence interval

for the regression equation used to estimate tree height

(Fig. 1), which suggests that the mean height of the fossil

forest at Wahl Glacier was likely intermediate in a grada-

tion between taller trees at Graphite Peak and shorter more

juvenile trees at Mt. Achernar.

At least three wood morphogenera have been described

from the study region: Araucarioxylon, Kaokoxylon, and

Australoxylon (Pigg & Taylor, 1993; Taylor & Ryberg,

2007; Decombeix et al., 2010, 2012); therefore, it is pos-

sible that tree height differences are related to species dif-

ferences in the region. As each wood morphogenera has

been found among Glossopteris leaf fossils, however, it is

difficult to assess the impact of these different wood mor-

phogenera on the paleoecology of the forests. Alterna-

tively, the variation of tree height and tree density likely

reflects some combination of forest maturity, environmen-

tal stresses, and/or disturbance. It is possible that forest

maturity, disturbance and environmental stress may be

linked to landscape position as pronounced variation in

tree density and plant diversity are observed in extant

plants on a variety of landscapes (Fig. 9B,C; Gregory et al.,

1991; Viereck et al., 1993; He & Duncan, 2000).

Both Wahl Glacier and Graphite Peak fossil forests occur

in similar depositional environments, proximal to low-sinu-

osity stream systems, whereas the fossil forest at Mt. Acher-

nar occurs in a lower energy environment further away

from large drainage systems in the Transantarctic basin.

Stark differences in basal area and tree density further sug-

gest an environmental control on forest development

(Fig. 9A), where low basal area and low tree density corre-

sponds to levee/splay complexes at Wahl Glacier and

Graphite Peak, and much higher basal area and tree density

are estimated for lacustrine and distal floodplain environ-

ments at Mt. Achernar and a suite of stumps from Lamping

Peak (Figs 1, 9A, Knepprath, 2006). This trend of decreas-

ing tree density and increasing tree maturity is known as

‘self-thinning’ (Cao et al., 2000; Falcon-Lang, 2004) and

is observed in modern landscapes as forests age away from

Table 3 Carbon isotopic data of fossil wood from Mt. Sirius

Sample*,†Position

(mm)

Mass

(mg)

Carbon yield

(lg C)

d13C

(VPDB)

Ring 1

1 0.500 1.82 90 �24.73

2 1.00 1.66 41 �24.90

3 1.50 1.51 17 �24.95

4 1.60 1.67 26 �25.56

5 2.55 1.49 18 �25.48

6 2.59 1.65 16 �25.25

7 2.60 1.89 21 �25.16

8 3.25 1.59 10 �24.93

9 3.35 1.82 12 �25.15

10 3.70 1.42 16 �25.14

Ring 2

1 3.80 1.83 10 �24.71

2 3.85 1.67 11 �25.08

3 3.90 1.66 10 �24.82

4 4.25 1.55 11 �25.10

5 4.40 1.22 8 �24.27

6 4.60 1.79 13 �24.86

7 4.65 1.37 9 �24.71

8 4.75 1.09 27 �24.75

9 5.00 1.22 11 �25.31

10 5.05 1.7 14 �25.07

11 5.23 1.31 35 �25.34

12 5.55 1.22 32 �24.84

13 5.70 1.36 14 �25.42

14 5.83 1.51 23 �25.03

Ring 3

1 6.23 1.85 10 �25.42

*Numbers correspond to samples within a tree ring; thus, for ring 1,

there are 10 samples individually analyzed for carbon isotope composition.

†University of Kansas accession number is 17170, the sample is from Mt.

Sirius, upper Buckley Formation, Antarctica.



zones of high and persistent disturbance (i.e., streams). A

common progression of diversity and tree densities in the

Pacific northwest trend from bare alluvium to mixed shrubs

and thin alders of high density (approximately

100 000 ind. Ha�1) on terraces adjacent to active stream

channels, whereas the oldest landscapes (>700 years) of

bogs or mires occupied by black spruce with densities rang-

ing from 500 to 2000 ind. Ha�1 (Fig. 9B,C, Viereck et al.,

1993). The observed self-thinning in the Permian, how-

ever, is opposite of the modern, where Permian tree matu-

rity increases (and density decreases) closer to the area of

high potential disturbance and is similar to younger high-

latitude fossil forests of Cretaceous age in the Antarctic

peninsula region (Falcon-Lang et al., 2001; Falcon-Lang,

2004). This trend is possibly related to intense competition

for resources in a persistently disturbed (e.g., flooding)

environment that was unsuitable for understory vegetation

in the region. Moreover, as trees age, tree mortality can be

induced by intense competition for light, water, and nutri-

ents (Cao et al., 2000; Falcon-Lang, 2004). Forest canopy

structure is an important determining factor in resource

competition and future work to assess the growth form of

glossopterid trees at different latitudinal gradients, as well

as the ecology of different late Permian wood morphogen-

era in Antarctica, may elucidate competition for light as

opposed to nutrients or water.

The sedimentary rocks that host in situ fossil forests at

Graphite Peak, Mt. Achernar, and Wahl Glacier are inter-

preted as paleosols as they meet the principle criterion for

soil, which is to support plant life (Soil Survey Staff,

2010). The paleosols at each locality are gray to whitish

gray (Figs 3B–F; 4A,B; 6B,D), contain some degree of

bedding or sedimentary fabric, including leaf compressions

on bedding planes, and have a silt texture. These paleosols

lack field-based indicators indicative of seasonal changes

in moisture content (Kraus & Aslan, 1993), that is,

60

40

20

0

80

100

0 500 1000 1500 2000 2500 3000

GP

WG

LP

LP MA

Levee/splayDistal floodplain/

lacustrine

Tree Density (trees ha–1)

Bas

al A

rea

(m2

ha–

1 )

Bas

al A

rea

(m2

ha–

1 )

A

Curio Bay

Gordon Valley

Axel Heiberg

Decreasing tree density

Increasing forest maturity/age

Mt. AchernarWahl Gl./Graphite Pk.

C

D Lamping Pk.

60

40

20

80

100

6543210Log10 Tree Density (trees ha–1)

B

Distal

Splay

Modernfloodplain

position

Dis

tal

Spla

y

river

Braided streamLake

Shr

ubs

Ald

erW

hite

sp

ruce

Bla

cksp

ruce

Bog

Late Permian forests

Decreasing tree density

Increasing forest maturity/age

Perm

ian

flood

plai

npo

sitio

n

Glossopterids

Lycopsids

Modern boreal forests

Fig. 9 Paleoforestry calculations and modern comparison. (A) Basal area and tree density estimates for Permian fossil forests (green circles): Graphite Peak

(GP), Wahl Glacier (WG), Lamping Peak (LP), and Mount Achernar (MA). Lamping Peak estimates are from Knepprath (2006). Additional polar forest locali-

ties are displayed by the blue square symbols: Gordon Valley (Middle Triassic, Cuneo et al., 2003), Curio Bay (Middle Jurassic, Pole, 1999), and Axel Heiberg

Island (Eocene, Francis, 1991; Basinger et al., 1994; Greenwood & Basinger, 1994). (B) Basal area and tree density measurements for a modern boreal forest.

Data from Viereck et al. (1993). Blue square symbols correspond to early succession forest near a river. Red circles correspond to mature forest on terraces

and within bogs. Gray symbols indicate the Late Permian forestry data for reference. (C) Illustration of ‘self-thinning’ within a modern boreal forest (modified

from Viereck et al., 1993) where tree density decreases from left to right and forest maturity increases from left to right. (D) Interpretation of Late Permian

paleoenvironments and forest structure at Wahl Glacier, Graphite Peak, Lamping Peak, and Mt. Achernar. Self-thinning is interpreted to be the exact opposite

trend from the modern, that is, toward areas of high disturbance in the Late Permian.



redoximorphic coloration, sequestration of redox-sensitive

elements such as Fe and Mn, and differentiation of clay-

sized material into argillic horizons (sensu Alfisols, Verti-

sols). Rather the studied paleosols are reflective of nearly

perennial saturation by freshwater (i.e., immature and

poorly drained). The occurrence of well-preserved, abscised

leaves in the sediment directly adjacent to permineralized

stumps in growth position further confirms prolonged

water saturation as this situation produces an increase in

acidity and contributes to the preservation of leaf litter rel-

ative to exposure at the soil-air interface (Gastaldo &

Staub, 1999). A persistent and high water table is inferred

from the gray to whitish gray color and lack of mottling as

seasonal aeration would promote the oxidation and trans-

port of Fe- and Mn-bearing compounds resulting in redox-

imorphic features (Vepraskas, 1994).

Applying modern soil classification (Soil Survey Staff,

2010) to the paleosols studied herein results in their classi-

fication as Aquents, Aqu-Aquic those forming under

perennial water saturation, and ents-Entisols, poorly devel-

oped soils on young or unstable landscapes. The term

aquic is problematic in paleosol taxonomy as it is inher-

ently interpretative, because we cannot know for certain

whether water saturation was perennial or seasonal. There-

fore, our interpretations are guided by the range of afore-

mentioned observations on soil color, mineralogy, root

morphology, and tree ring analysis. The paleosol classifica-

tion system of Mack et al. (1993) is adapted for use in

geologic materials where these paleosols (Aquents) are clas-

sified as Protosols on the basis of poor (i.e., absent) horiz-

onation. The modern equivalents of Protosols (Entisols

and Inceptisols) dominate the global landscape and occur

across all climate regimes, slopes, vegetation types, land

surface ages, and hydrology. Based on the aforementioned

criteria, we interpret these paleosols as being saturated with

liquid water for several months of the year, if not perenni-

ally. Freezing conditions were apparently short-lived given

the absence of frost rings in Mt. Achernar tree stumps

(Taylor et al., 1992); therefore, despite the polar latitude

of these fossil forests, perennial water saturation was a

plausible condition.

The isotope geochemistry of permineralized wood from

Mt. Achernar indicates that carbon isotopic records at sub-

annual resolution can be extracted from permineralized

wood (Fig. 7). However, the data we collected from the

sample from Mt. Sirius (Fig. 8) are the oldest subannual

carbon isotope study of tree growth rings, with the next

oldest samples being the ‘sub fossil’ wood of Eocene age

in the Canadian Arctic (Jahren, 2007; Jahren & Sternberg,

2008; Schubert et al., 2012), underscoring the need for

high-resolution studies such as this for other time periods

in the Phanerozoic. The results for the Mt. Achernar and

Mt. Sirius samples provide evidence for a deciduous habit

for Glossopteris trees. Measured d13C values within tree

growth rings show markedly different trends between trees

with a deciduous versus evergreen habit (Fig. 8B; Schubert

& Jahren, 2011). The permineralized stump from

Mt. Achernar contains d13C patterns that are consistent

with a deciduous habit (Fig. 7); however, the low sampling

density (i.e., two data points per ring) does not permit

trends to be established to rigorously support such a con-

clusion. On the other hand, because of the higher sam-

pling density, the permineralized wood from Mt. Sirius

displays a three-phase d13C trend within individual growth

rings that is consistent with a deciduous habit (Fig. 8);

however, the isotope geochemistry does not indicate

whether this habit was advantageous at polar latitudes. Ro-

yer et al. (2005) indicate that deciduousness under a polar

light regime (i.e., months of continuous light and months

of continuous darkness in one year) and cold winters

results in a reorganization of carbon fixation strategies

when compared to evergreen trees. While these results may

seem to cast doubt on an elemental cycling competitive

edge of deciduous trees over evergreen trees at polar lati-

tudes, the fact that deciduous trees existed at polar lati-

tudes is a fact that is as yet not fully understood. Perhaps,

the explanation lies in the dominance of deciduous plants

at polar latitudes during past times of global warmth, for

example, the Late Permian-Triassic of Antarctica and the

Late Cretaceous of Alaska.

CONCLUSIONS

A field- and laboratory-based study of Late Permian in situ

fossil forests of the central Transantarctic Mountains pro-

vides the first integrated geochemical, sedimentologic,

paleopedologic, and paleobotanical analysis of this unique

polar ecosystem. The analysis of paleosols and contempora-

neous sedimentary rocks indicates that the fossil forests

grew in two different paleoenvironments, a low energy

environment adjacent to axial drainage systems of the

Transantarctic basin, and two higher energy environments

proximal to low-sinuosity fluvial systems. Tree height, tree

density, and basal area differ markedly between the inferred

distal floodplain environment and levee or splay environ-

ments; however, tree density and basal area are similar for

localities in the same inferred paleoenvironment. We inter-

pret this tree density trend as ‘self-thinning’. A comparison

with modern forests indicates that the observed Late Perm-

ian self-thinning is opposite of the modern, which suggests

that the unique ecology of Late Permian polar forests pro-

moted the loss of understory vegetation in regions of high

disturbance (e.g., fluvial systems).

Carbon isotope ratios of permineralized wood presented

herein are, to our knowledge, the first high-resolution data

set of this kind from material older than the Cenozoic.

The stable isotope geochemistry indicates that intra-tree

ring measurement of carbon isotope ratios is possible even



in permineralized samples. Such detail can provide impor-

tant insights into the habit of arborescent plants and implies

that additional environmental variables, such as water stress,

could be quantified from such ancient ecosystems.

ACKNOWLEDGMENTS

We are especially grateful to Drs. Neil Tabor and A. Hope

Jahren for assistance and generosity in the use of their lab-

oratory facilities; and Mr. Bill Hagopian and Dr. Scott Me-

yers for assistance with stable isotope ratio measurements.

Danielle Sieger is thanked for her assistance with the mea-

sured section at Mt. Achernar. We thank Brian Staite for

his help with logistics, mountaineering, and for sharing

with neophyte Antarctic researchers his >2 decades of

experience working on ‘the ice.’ Fixed wing transport was

provided by the 62nd and 446th Airlift Wing of the U.S.

Air Force, stationed at Joint Base Lewis-McChord, WA,

the 109th Airlift Wing of the New York Air National

Guard, and Kenn Borek Air Ltd. This work was supported

by funding from NSF grants ANT-0943935 and ANT-

0944532 to the University of Wisconsin-Milwaukee, and

ANT-0943934 to the University of Kansas.

REFERENCES

Barrett PJ, Elliot DH, Lindsay JF (1986) The Beacon Supergroup

(Devonian-Triassic) and Ferrar Group (Jurassic) in the

Beardmore Glacier area, Antarctica. In Geology of the CentralTransantarctic Mountains, Antarctic Research Series, vol. 36(eds Turner MD, Splettstoesser JF). American Geophysical

Union, Washington, D.C., pp. 339–428.Basinger JM, Greenwood DR, Sweda T. (1994) Early Tertiaryvegetation of Arctic Canada and its relevance to paleoclimatic

interpretation. In: Cenozoic Plants and Climates of the Arctic.NATO-ASI Series I, v. 27 (eds Boulter M, Fisher HC). Spinger-

Verlag, Berlin, pp. 175–198.Briggs DEG, Miller MF, Isbell JL, Sidor CA (2010) Permo-

Triassic arthropod trace fossils from the Beardmore Glacier area,

central Transantarctic Mountains, Antarctica. Antarctic Science,22, 185–192.

Cao QV, Dean TJ, Baldwin VC (2000) Modeling the size-density

relationship in direct-seeded slash pine stands. Forest Science, 46,317–321.

Collinson JA, Isbell JL, Elliot DH, Miller MF, Miller JMG,

Veevers JJ (1994) Permian-Triassic Transantarctic basin. In:

Permian-Triassic Pangean Basins and Foldbelts along thePanthalassan Margin of Gondwanaland (eds Veevers JJ, PowellCMcA). Geological Society of America Memoir 184. Geological

Society of America, Boulder, CO, pp. 173–222.Cottam G, Curtis JT (1956) The use of distance measures in

phytosociological sampling. Ecology, 37, 451–460.Creber GT, Francis JE (1999) Fossil tree-ring analysis:

palaeodendrology. In: Fossil Plants and Spores: ModernTechniques (eds Jones TP, Rowe NP). Geological Society,London, pp. 245–250.

Cuneo NR, Isbell JL, Taylor EL, Taylor TN (1993) The

Glossopteris flora from Antarctica: taphonomy and

paleoecology. Compte Rendu, XII International Congress on

Carboniferous-Permian, Buenos Aires, Argentina, Stratigraphyand Geology, 2, 13–40.

Cuneo NR, Taylor EL, Taylor TN, Krings M (2003) In situ fossil

forest from the upper Fremouw Formation (Triassic) ofAntarctica: paleoenvironmental setting and paleoclimate analysis.

Palaeogeography, Palaeoclimatology, Palaeoecology, 197, 239–261.

Decombeix A-L, Taylor EL, Taylor TN (2009) Secondary growthin Vertebraria roots from the Late Permian of Antarctica: a

change in developmental timing. International Journal of PlantSciences 170, 644–656.

Decombeix A-L, Taylor EL, Taylor TN (2010) Epicormic shoots

in a Permian gymnosperm from Antarctica. InternationalJournal of Plant Science, 171, 772–782.

Decombeix A-L, Taylor EL, Taylor TN (2012) Gymnosperm treesfrom the Permian of Antarctica: an anatomically preserved trunk

of Kaokoxylon sp. Comptes Rendus Palevol, 11, 21–29.Demko TM, Dubiel RF, Parrish JT (1998) Plant taphonomy in

incised valleys: implications for interpreting paleoclimate fromfossil plants. Geology, 26, 1119–1122.

Diefendorf AF, Mueller KE, Wing SL, Koch PL, Freeman KH

(2010) Global patterns in leaf 13C discrimination and

implications for studies of past and future climate. Proceedings ofthe National Academy of Sciences of the United States ofAmerica, 107, 5738–5743.

Domeier M, Van Der Voo R, Tohver E, Tomezzoli RN, Vizan H,Kirshner J (2011) New Late Permian paleomagnetic data from

Argentina: refinement of the apparent polar wander paths of

Gondwana. Geochemistry, Geophysics, Geosystems, 12, 1–21.Doumani GA, Minshew VH (1965) General geology of theMount Weaver area, Queen Maud Mountains, Antarctica. In

Geology and Paleontology of the Antarctic, Antarctic ResearchSeries 6 (eds Hadley JB). American Geophysical Union,

Washington, D.C., pp. 127–139.Elliot DH, Fanning CM (2008) Detrital zircons from upper

Permian and lower Triassic Victoria Group sandstones,

Shackleton Glacier region, Antarctica: evidence for multiplesources along the Gondwana plate margin. Gondwana Research,13, 259–274.

Falcon-Lang HJ (2004) Early Mississippian lycopsid forests in a

delta-plain setting at Norton, near Sussex, New Brunswick,Canada. Journal of the Geological Society, London, 161, 969–981.

Falcon-Lang HJ, Cantrill DJ, Nichols GJ (2001) Biodiversity and

terrestrial ecology of a mid-Cretaceous, high-latitude floodplain,Alexander Island, Antarctica. Journal of the Geological Society,London, 158, 709–724.

Farabee MJ, Taylor EL, Taylor TN (1990) Correlation of Permianand Triassic palynomorph assemblages from the central

Transantarctic Mountains, Antarctica. Review of Palaeobotanyand Palynology, 65, 257–265.

Farabee MJ, Taylor EL, Taylor TN (1991) Late Permianpalynomorphs from the Buckley Formation, central

Transantarctic Mountains, Antarctica. Review of Palaeobotanyand Palynology, 69, 353–368.

Farquhar RJ, O’Leary MH, Berry JA (1982) On the relationshipbetween carbon isotope discrimination and the intercellular

carbon dioxide concentration in leaves. Australian Journal ofPlant Physiology, 9, 121–137.

Farquhar GD, Ehleringer JR, Hubick KT (1989) Carbon isotopediscrimination and photosynthesis. Annual Review of PlantPhysiology and Plant Molecular Biology, 40, 503–537.

Fielding CR, Frank TD, Birgenheier LP, Rygel MC, Jones AT,Roberts J (2008) Stratigraphic record and facies associations of



the late Paleozoic ice age in Eastern Australia (New South Wales

and Queensland). In: Resolving the Late Paleozoic Ice Age in Timeand Space (eds Fielding CR, Frank TD, Isbell JL). Geological

Society of America Special Paper 441. Geological Society ofAmerica, Boulder, CO, pp. 41–57.

Francey RJ, Farquhar GD (1982) An explanation of 13C/12C

variations in tree rings. Nature, 297, 28–31.Francis JE (1990) Polar forests. Geology Today, 6, 92–95.Francis JE (1991) The dynamics of polar fossil forests: Tertiary

fossil forests of Axel Heiberg Island, Canadian Arctic

Archipelago. In: Tertiary Fossil Forests of the Geodetic Hills, AxelHeiberg Island, Arctic Archipelago. Geological Survey of CanadaBulletin, vol. 403 (eds Christie RL, McMillan NJ). Geological

Survey of Canada, Ottawa, ON, Canada, pp. 29–38.Francis JE, Woolfe KJ, Arnott MJ, Barrett PJ (1993) Permianforests of Allan Hills, Antarctica: the palaeoclimate of

Gondwanan high latitudes. Special Papers in Paleontology, 49,75–83.

Gastaldo RA, Staub JR (1999) A mechanism to explain thepreservation of leaf litter lenses in coals derived from raised

mires. Palaeogeography, Palaeoclimatology, Palaeoecology, 149, 1–14.

Gould RE (1975) A preliminary report on petrified axes ofVertebraria from the Permian of eastern Australia. In GondwanaGeology (ed. Campbell KSW). Australian National University

Press, Canberra, pp. 109–115.Gradstein FM, Ogg JG, Smith AG, Agterberg FP, Bleeker W,

Cooper RA, Davydov V, Gibbard P, Hinnov LA, House MR,

Lourens L, Luterbacher H-P, McArthur J, Melchin MJ, Robb

LJ, Sadler PM, Shergold J, Villeneuve M, Wardlaw BR, Ali J,Brinkhuis H, Hilgen FJ, Hooker J, Howarth RJ, Knoll AH,

Laskar J, Monechi S, Powell J, Plumb KA, Raffi I, Rohl U,

Sanfilippo A, Schmitz B, Shackleton NJ, Shields GA, Strauss H,

Van Dam J, Veizer J, van Kolfschoten Th, Wilson D. (2004) AGeologic Time Scale 2004. Cambridge University Press,

Cambridge, 610 p.

Greenwood DR, Basinger JF (1994) The paleoecology of high-latitude Eocene swamp forests from Axel Heiberg Island,

Canadian High Arctic. Reviews of Paleobotany and Palynology,81, 83–97.

Gregory SV, Swanson FJ, McKee WA, Cummins KW (1991) Anecosystem perspective of riparian zones. BioScience, 41, 540–551.

He F, Duncan RP (2000) Density-dependent effects on tree

survival in an old-growth Douglas fir forest. Journal of Ecology,88, 676–688.

Helle G, Schleser GH (2004) Beyond CO2-fixation by Rubisco –an interpretation of 13C/12C variations in tree rings from novelintra-seasonal studies on broad-leaf trees. Plant, Cell andEnvironment, 27, 367–380.

Isbell J, MacDonald DIM (1991) Lithofacies analysis of the

Triassic Fremouw Formation at the Gordon Valley vertebratesite. Antarctic Journal of the United States, 26, 15–16.

Isbell JL, Seegers GM, Gelhar GA (1997) Upper Paleozoic glacial

and postglacial deposits, central Transantarctic Mountains,

Antarctica. In Late Glacial and Postglacial EnvironmentalChanges: Quaternary, Carboniferous-Permian, and Proterozoic(ed. Martini IP). Oxford University Press, Oxford, U.K., pp.

230–242.Isbell JL, Miller MF, Babcock LE, Hasiotis ST (2001) Ice-marginal environments and ecosystem prior to initial advance of

the late Paleozoic ice sheet in the Mount Butters area of the

central Transantarctic Mountains, Antarctica. Sedimentology, 48,953–970.

Isbell JL, Miller MF, Wolfe KL, Lenaker PA (2003) Timing of late

Paleozoic glaciation in Gondwana: was glaciation responsible for

the development of northern hemisphere cyclothems? In ExtremeDepositional Environments: Mega end Members in Geologic Time(eds Chan MA, Archer AW). Geological Society of America

Special Paper 370. Geological Society of America, Boulder, CO,

pp. 5–24.Isbell JL, Koch ZJ, Szablewski GM, Lenaker PA (2008) Permianglacigenic deposits in the Transantarctic Mountains, Antarctica.

In: Resolving the Late Paleozoic Ice Age in Time and Space (edsFielding CR, Frank TD, Isbell JL). Geological Society ofAmerica Special Paper 441, Geological Society of America,

Boulder, CO, pp. 59–70.Isbell JL, Henry LC, Gulbranson EL, Limarino CO, Fraiser ML,

Koch ZJ, Ciccioli PL, Dineen AA (2012) Evaluation of glacialparadoxes during the late Paleozoic ice age using the concept of

the equilibrium line altitude (ELA) as a control on glaciation.

Gondwana Research, 22, 1–19.Jahren AH (2007) The Arctic forest of the Middle Eocene.Annual Review of Earth and Planetary Sciences, 35, 509–540.

Jahren AH, Sternberg LSL (2008) Annual patterns within tree

rings of the Arctic middle Eocene (ca. 45 Ma): isotopic

signatures of precipitation, relative humidity, and deciduousness.Geology, 36, 99–102.

Knepprath NE (2006) Late Permian forests of the Buckley

Formation, Beardmore Glacier area, Antarctica. M.S. Thesis,Vanderbilt University, 92 pp.

Kohn MJ (2010) Carbon isotope compositions of terrestrial C3

plants as indicators of (paleo)ecology and (paleo)climate.

Proceedings of the National Academy of Sciences of the UnitedStates of America, 107, 19691–19695.

Kraus MJ, Aslan A (1993) Eocene hydromorphic paleosols –significance for interpreting ancient floodplain processes. Journalof Sedimentary Petrology, 63, 453–463.

Krull ES, Retallack GJ (2000) Delta C-13 depth profiles from

paleosols across the Permian-Triassic boundary: evidence for

methane release. GSA Bulletin, 112, 1459–1472.Kyle RA, Schopf JM (1982) Permian and Triassic

palynostratigraphy of the Victoria Group, Transantarctic

Mountains. In Antarctic Geosciences, Series B-4 (ed. Craddock

C). University of Wisconsin Press, International Union ofGeological Sciences, Madison, pp. 649–659.

Lawver LA, Dalziel IWD, Norton IO, Gahagan LM (2008) The

Plates 2007 Atlas of Plate Reconstructions (750 Ma to Present

Day), Plates Progress Report No. 305-0307, University ofTexas Technical Report No. 195, 160 pp.

Leavitt SW, Long A (1991) Seasonal stable-carbon isotope

variability in tree rings: possible paleoenvironmental signals.Chemical Geology, 87, 59–70.

Mack GH, James WC, Monger HC (1993) Classification of

paleosols. GSA Bulletin, 105, 129–136. doi: 10.1130/0016-7606(1993)<0129:COP>2.3.CO;2.

McLoughlin S, Lindstrom S, Drinnan AN (1997) Gondwanan

floristic and sedimentological trends during the Permian-Triassic

transition: new evidence from the Amery Group, northern

Prince Charles Mountains, East Antarctica. Antarctic Science, 9,281–298.

McLoughlin S, Larsson K, Lundstrom S (2005) Permian plant

macrofossils from Fossilryggen, Vestfjella, Dronning Maud

Land. Antarctic Science, 17, 73–86.Neish PG, Drinnan AN, Cantrill DJ (1993) Structure and

ontogeny of Vertebraria from silicified Permian sediments in

East Antarctica. Reviews of Palaeobotany and Palynology, 79,221–244.



Niklas KJ (1994) Predicting the height of fossil plant remains: an

allometric approach to an old problem. American Journal ofBotany, 81, 1235–1242.

Pigg KB, Taylor TN (1993) Anatomically preserved Glossopterisstems with attached leaves from the central Transantarctic

Mountains, Antarctica. American Journal of Botany, 80, 500–516.

Plumstead EP (1958) The habit of growth of Glossopteridae.Transactions and proceedings of the Geological Society of SouthAfrica, 61, 81–96.

Pole M (1999) Structure of a near-polar latitude forest from theNew Zealand Jurassic. Palaeogeography, Palaeoclimatology,Palaeoecology, 147, 121–139.

Powell CM, Li ZX (1994) Reconstruction of the Panthalassan

margin of Gondwanaland. In Permian-Triassic Transantarcticbasin, Permian-Triassic Pangean Basins and Foldbelts alongthe Panthalassan margin of Gondwanaland (eds Veevers J,

Powell C). Geological Society of America Memoir 184,

Geological Society of America, Boulder, CO, pp. 5–9.Retallack G (1980) Late Carboniferous to Middle Triassic

megafossil floras from the Sydney Basin. Bulletin of theGeological Survey of New South Wales, 26, 384–430.

Retallack GJ, Krull ES (1999) Landscape ecological shift at thePermian-Triassic boundary in Antarctica. Australian Journal ofEarth Sciences, 46, 785–812.

Retallack GJ, Veevers JJ, Morante R (1996) Global coal gapbetween Permian-Triassic extinction and Middle Triassic

recovery of peat-forming plants. GSA Bulletin, 108, 195–207.Royer DL, Osborne CP, Beerling DJ (2005) Contrasting seasonal

patterns of carbon gain in evergreen and deciduous tress ofancient polar forests. Paleobiology, 31, 141–150.

Ryberg P, Taylor EL, Taylor TN (2011) Mt. Achernar: A window

into Antarctic glossopterid diversity. Botany 2011, Abstracts,

pp. 125–126. http://2011.botanyconference.org/engine/search/index.php?func=detail&aid=51.

Schopf JM (1970) Petrified peat from a Permian coal bed in

Antarctica. Science, 169, 274–277.Schubert BA, Jahren AH (2011) Quantifying seasonal precipitation

using high-resolution carbon isotope analyses in evergreen wood.

Geochimica et Cosmochimica Acta, 75, 7291–7303.Schubert BA, Jahren AH, Eberle JJ, Sternberg LSL, Eberth DA(2012) A summertime rainy season in the Arctic forests of the

Eocene. Geology, 40, 523–526.Schwendemann AB, Decombeix A-L, Taylor EL, Taylor TN

(2010) Collinsonites schopfii gen. et sp. nov., a herbaceouslycopsid from the Upper Permian of Antarctica. Review ofPaleobotany and Palynology, 158, 291–297.

Seward AC (1914) Antarctic fossil plants. Bulletin of the BritishMuseum (Natural History). Geology, 1, 1–49.

Sheldon ND (2006) Abrupt chemical weathering increase across

the Permian-Triassic boundary. Palaeogeography,Palaeoclimatology, Palaeoecology, 231, 315–321.

Smith ND, Cross TA, Dufficy JP, Clough SR (1989) Anatomy of

an avulsion. Sedimentology, 36, 1–23.Soil Survey Staff (2010) Keys to Soil Taxonomy, 11th edn. U.S.

Department of Agriculture, Natural Resources ConservationService, Washington, D.C., 338 pp.

Taylor EL (1996) Enigmatic gymnosperms? Structurally preserved

Permian and Triassic seed ferns from Antarctica. Review ofPalaeobotany and Palynology 90, 303–318.

Taylor EL, Ryberg PE (2007) Tree growth at polar latitudesbased on fossil tree ring analysis. Palaeogeography,Palaeoclimatology, Palaeoecology 255, 246–264.

Taylor EL, Taylor TN, Cuneo NR (1992) The present is not the

key to the past: a Permian petrified forest in Antarctica. Science,257, 1675–1677.

Taylor EL, Taylor TN, Ryberg PE (2007) Ovule-bearing

reproductive organs of the glossopterid seed ferns from the LatePermian of the Beardmore Glacier Region, Antarctica. In

Antarctica: A Keystone in a Changing World – Online

Proceedings of the 10th ISAES (International Symposium on

Antarctic Earth Sciences), U.S. Geological Survey Open FileReport 2007-1047, Short Research Paper 082 (eds Cooper AK,

Raymond CR, the 10th ISAES Editorial Team). The National

Academies Press, Washington, DC.

Torsvik TH, Cocks LRM (2004) Earth geography from 400 to250 Ma: a palaeomagnetic, faunal and facies review. Journal ofthe Geological Society 161, 555–572.

Vepraskas MJ (1994) Redoximorphic features for identifying aquic

conditions: North Carolina Agricultural Research ServiceTechnical Bulletin, no. 301.

Viereck LA, Dyrness CT, Foote MJ (1993) An overview of the

vegetation and soils of the floodplain ecosystems of the TananaRiver, interior Alaska. Canadian Journal of Forest Research, 23,889–898.

Williams CJ, Mendell EK, Murphy J, Court WM, Johnson AH,

Richter SL (2008) Paleoenvironmental reconstruction of aMiddle Miocene forest from the western Canadian Arctic.

Palaeogeography, Palaeoceanography, Palaeoecology, 261,160–176.

SUPPORTING INFORMATION

Additional Supporting Information may be found in the

online version of this article:

Fig. S1 Results of tree height estimate using the Niklas (1994) regression

calculation (red line); 95% confidence interval for the regression is indicated

by the dashed lines.

Fig. S2 Carbon isotope values for fossil stumps.

Fig. S3 Permineralized stump positions along three transects.

Please note: Wiley-Blackwell are not responsible for the con-

tent or functionality of any supporting materials supplied by

the authors. Any queries (other than missing material)

should be directed to the corresponding author for the

article.



Copyright of Geobiology is the property of Wiley-Blackwell and its content may not be copied or emailed to

multiple sites or posted to a listserv without the copyright holder's express written permission. However, users

may print, download, or email articles for individual use.

permian polar forests: deciduousness and environmental ... et al 2012.pdfof polar forests and...

Documents