Journal of the Geological Society, London, Vol. 158, 2001, pp. 709–724. Printed in Great Britain.

Biodiversity and terrestrial ecology of a mid-Cretaceous, high-latitude floodplain, Alexander Island, Antarctica H. J. FALCON-LANG 1,3 , D. J. CANTRILL 1 & G. J. NICHOLS 2 British Antarctic Survey, High Cross, Madingley Rd, Cambridge CB3 0ET, UK 2 Department of Geology, Royal Holloway, University of London, Egham, Surrey, TW20 0EX, UK 3 Present address: Department of Earth Sciences, Dalhousie University, Halifax, Nova Scotia B3H 3J5, Canada (e-mail: [email protected]) 1

Abstract: The biodiversity and terrestrial ecology of the Late Albian Triton Point Formation (Fossil Bluff Group), Alexander Island, Antarctica is analysed to improve our understanding of polar biomes during the mid-Cretaceous thermal optimum. This formation was deposited on a high-latitude (75S) floodplain and consists of two facies associations, a lower braided alluvial plain unit and an upper coastal meander-belt unit. Analysis of fossil plants in well exposed palaeosols reveals the existence of spatially complex plant communities. Braidplains supported patchy, low-density (91 trees/ha) stands of podocarp and taxodioid conifers on floodbasin substrates, and conifer–cycadophyte–fern–angiosperm thickets in riparian settings. Coastal meander-belts supported medium density (568 trees/ha) podocarp–araucarian conifer forests on mature floodbasin soils, and fern–angiosperm–ginkgo thickets in riparian settings. Growth-ring analysis indicates plants experienced stressful growing conditions on the braidplain characterized by high-frequency flood events, but more favourable growing conditions on the coastal plain. Additional vegetation disturbances were caused by arthropod–fungal attack, frost and wildfire. In terms of structure, composition, ecology and productivity these predominantly evergreen, broad-leafed conifer forests bear similarities to the extant temperate rainforests of New Zealand. Keywords: Albian, Antarctica, polar regions, conifers, tree rings.

The recognition that polar regions were once covered by forest vegetation ranks as one of the most important palaeontological discoveries of the past 20 years (Axelrod 1984). The earliest polar forests appeared in Late Permian times, and were composed of glossopterid trees which grew at palaeolatitudes of up to 80S (Taylor et al. 1992). Forest vegetation remained established at both poles throughout the Mesozoic and Early Tertiary surviving at palaeolatitudes as high as 85 (Spicer & Parrish 1986), before finally contracting equatorward during the Late Tertiary in response to global cooling (Spicer & Chapman 1990). An extraordinary feature of these forests was their ability to tolerate extremes of light seasonality. Assuming that the obliquity of the Earth’s rotational axis has remained constant over geological time (Barron 1984), forests growing at 70 of latitude would have experienced up to 70 days of unbroken darkness each year, whilst those at 85 would have been deprived of light for nearly 160 days per year (Read & Francis 1992). The poleward limit of present-day forest vegetation is controlled by a combination of light seasonality, mean annual temperature and annual temperature range (Woodward 1987). These factors constrain growing season length, rate of photosynthesis and respiration, and seed germination potential, and determine whether the long-term positive carbon balance required for tree survival is achieved (Gower & Richards 1990). In the Northern Hemisphere the tree-line ranges from 59N to 72N whilst in the Southern Hemisphere forest vegetation only extends to 55S because no ice-free continental landmasses exist in the high southern latitudes (Creber & Chaloner 1985; Read & Francis 1992). The massive poleward advance in the tree-line envisaged for Late Palaeozoic–Early Tertiary times has been attributed to the existence of a substantially warmer polar environment with only small

permanent glaciers existing at altitude (Frakes & Francis 1988; Spicer & Parrish 1990). Maximum global warmth occurred during the midCretaceous (Albian–Turonian; 88–112 Ma) (Clarke & Jenkyns 1999) and abundant polar forest localities have been described from this time interval. Fossil forests dominated by taxodioid and pinoid conifers are known from northern Alaska (Spicer & Parrish 1986), Kamchatka and NE Russia (Herman & Spicer 1996) at palaeolatitudes as high as 82N. In the Southern Hemisphere, broad-leafed podocarp and araucarian conifer forests are known from SE Australia (Dettman et al. 1992), southern South America (Archangelsky 1963), New Zealand (Parrish et al. 1998) and Antarctica (Jefferson 1982) at palaeolatitudes of up to 75S. Forests were inhabited by a diverse terrestrial fauna which included small, large-eyed, largebrained dinosaurs which may have been adapted to the cool, dark winters or seasonally migrated with the sun-line (Parrish et al. 1987); birds, mammals, amphibians and arthropods were also present (Rich et al. 1988). None of these important Cretaceous fossil forest localities have yet received rigorous palaeoecological analysis. Little is known of tree density and community structure, nor is there much data concerning ecophysiological response to climate and growing environment, and animal–plant interactions. This study presents new data concerning the biodiversity, palaeoenvironments and palaeoecology of the classic fossil forests of SE Alexander Island, Antarctic Peninsula (Jefferson 1982), which are the most southerly (palaeolatitude 75S; Smith et al. 1994) and well-exposed Cretaceous polar forests so far discovered (Fig. 1a). These data are used to test and refine numerical models of the Cretaceous polar environment (e.g. Otto-Bliesner & Upchurch 1997; Beerling 2000). 709

710

H. J. FALCON-LANG ET AL.

Study area

a

Ä

Ä

Alexander Island, Antarctica

Ä

Ä

Ä

Ä Ä

Offset Ridge

UP

b

Antarctic Peninsula

AY

G RO

60°

M

Ä

LE

KG. 2813

KG. 4710

Fossil forests

° 70

Triton Point

71 40'S

500

0

°S 80

1000

km

Triton Point Member Other rock outcrop

71 50'S

Alexander Island on the western side of the Antarctic Peninsula represents the uplifted fore-arc region of a midPalaeozoic to Tertiary calc-alkaline volcanic arc (McCarron & Larter 1998). The Fossil Bluff Group exposed in eastern Alexander Island was deposited in a narrow (<60 km wide), elongate fore-arc basin, fault-bounded to the west by the accretionary complex of the Le May Group and to the east by the magmatic arc (McCarron & Millar 1997). The fossil forests described in this paper occur in the Triton Point Formation of the Fossil Bluff Group, SE Alexander Island between 7140 S and 7208 S (Moncrieff & Kelly 1993; Nichols & Cantrill 2001) (Fig. 1b) and have been assigned a Late Albian age on the basis of a bracketing molluscan fauna (105 Ma; Kelly & Moncrieff 1992). This stratigraphic unit consists of a wedge-like package of fluvial sedimentary rocks which thickens southward from 200 m at Triton Point to c. 950 m at Citadel Bastion, Titan Nunataks and Coal Nunatak, and has been divided into two facies associations (Cantrill & Nichols 1996). The lower facies association crops out at all localities and is interpreted as the product of a braided alluvial plain environment. The overlying facies association is geographically restricted to the upper part of Coal Nunatak, and has been interpreted as a coastal meander belt deposit (Cantrill & Nichols 1996). In the course of recent stratigraphic revision by Nichols & Cantrill (2001) these two sedimentary packages have been formally assigned member status, the lower unit being named the Citadel Bastion Member (c. 820 m thick), and the upper unit named the Coal Nunatak Member (c. 130 m thick). Petrographic and palaeocurrent analysis indicates that most sediments were derived from the magmatic arc with only minor input from the accretionary complex (Browne 1996). Evidence from general circulation models, carbonaceous palaeosols and fossil plants suggest that climate was temperate with a high mean annual rainfall (Parrish et al. 1982, 1998; Spicer & Chapman 1990). KG. numbers mentioned in the following text refer to British Antarctic Survey field stations in SE Alexander Island. Precise location details of these sites are given in Figure 1b.

Floral biodiversity Taxonomic studies have shown than the vegetation of the Triton Point Formation was diverse compared with other Early Cretaceous floras; so far 42 form genera have been recorded containing approximately 69 species (Table 1). Conifers are the most abundant plant fossils, represented by silicified woods and compressed foliage belonging to the Araucariaceae, Podocarpaceae and Taxodiaceae families (Falcon-Lang & Cantrill 2000; Cantrill & Falcon-Lang 2001).

KG. 2821

Citadel Bastion

SS

IL

BL UF F

G RO

UP

Pagoda Ridge

FO

Hyperion Nunataks

George VI Sound

KG. 4725

KG. 4660 KG. 4737

Coal Nunatak KG. 4702 loose wood

KG. 1702

Titan Nunataks KG. 4717 loose wood

KG. 2817 KG. 4657

KG. 2814, 4740 KG. 4741, 4745, 4746, 4747 KG. 2815 KG. 4688, 4697, 4699

KG. 2816, 4719 KG. 1704

68 30'W

0

10 km

20

Fig. 1. (a) South Pole centred global palaeogeographic map (Late Albian, 100 Ma). Light grey: open ocean, medium grey: continental shelf, and dark grey: land (after Smith et al. 1994). Dotted line marks polar circle at 66.6S. Crossed circles mark mid-Cretaceous localities where podocarp–araucarian conifer woods and foliage are dominant. Data from Antarctic Peninsula (Francis 1986), SE Australia (Frakes & Francis 1988), Keguelen Plateau (Francis & Coffin 1992), New Zealand (Parrish et al. 1998), South Africa (Bamford & Corbett 1994) and South America (Torres & Biro-Bagoczky 1986). Squares mark mid-Cretaceous localities with podocarp–araucarian pollen close to the polar circle (Truswell 1990). (b) Inset: position of SE Alexander Island fossil forests on Antarctic Peninsula. Outcrop map of SE Alexander Island showing main fossil forest localities in the Triton Point Formation.

CRETACEOUS HIGH-LATITUDE FLOODPLAIN, ANTARCTICA

711

Table 1. Floral biodiversity of Triton Point Formation Hepatopsida (liverworts) Marchantiales Marchantites (5 species) Hepaticites (3 species) Thallites (2 species) Metzgeriales 1 species Lycopsida (lycopods) Selaginellales 1 species Sphenopsida (horse-tails) Equisetales 1 species Filicopsida (ferns) Dipteridaceae Hausmania (1 species) Matoniaceae Matonia (1 species) Osmundaceae Cladophlebis (6 species) Phyllopteroides (2 species) Gleicheniaceae Gleichenites (1 species) ?Lophosoriaceae Microphyllopteris (2 species) Incertae sedis Aculea (1 species) Adiantites (1 species) Alamatus (1 species) Sphenopteris (4 species) (*In total, fern component comprises approximately 14 genera and 26 species, many awaiting formal description).

Coniferopsida Coniferales (conifers) Araucariaceae Araucaria (2 species) Araucarites (2 species) Araucarioxylon (1 species) Araucariopitys (1 species) Podocarpaceae Podocarpites (1 species) Podocarpoxylon (2 species) Taxodiaceae Athrotaxites (1 species) Taxodioxylon (1 species) Incertae sedis Brachyphyllum (1 species) Elatocladus (2 species) Pagiophyllum (1 species) Podozamites (1 species) Other minor gymnosperms Bennettitales (extinct) Ptilophyllum (2 species) Cycadales (cycads) 1 species Ginkgoales (ginkgos) Ginkgoites (1 species) *Pentoxylales (extinct) Taeniopteris (3 species) Angiospermopsida (flowering plants) Araliaephyllum (1 species) Dicotylophyllum (1 species) Ficophyllum (1 species) Gnafalea (2 species) Hydrocotylophyllum (1 species) Timothyia (1 species)

For details of systematic taxonomy see: liverworts (Cantrill 1997), ferns (Jefferson 1981; Cantrill 1995, 1997), conifers (Falcon-Lang & Cantrill 2000; Cantrill & Falcon-Lang 2001), and angiosperms (Cantrill & Nichols 1996). Other seed plants currently undescribed. *Systematic taxonomy of ferns currently being revised by N. Nagalingum (Univ. Melbourne, Australia), and Pentoxylales by J. Howe (Univ. Leeds, UK).

All the conifers were large trees except for one podocarp species which was probably a shrub (Cantrill & Falcon-Lang 2001). However, although the conifers are abundant, they make up only a small percentage of the total species diversity in the Triton Point Formation (c. 24%). The most diverse component of the flora are the ferns; the systematic taxonomy of this group is currently being revised by N. Nagalingum (Univ. Melbourne, Australia), but conservative estimates are that as many as 26 species in 14 genera occur (c. 39% species diversity). Some fern foliage can be allied to extant families such as the Dipteridaceae, Gleicheniaceae, ?Lophosoriaceae, Matoniaceae, and Osmundaceae (Jefferson 1981, 1982; Cantrill 1995), whilst the majority of forms cannot at this stage be attributed with confidence to family level (Cantrill 1996) (Table 1). Angiosperms are represented by a moderately diverse flora (c. 12% species diversity) consisting mainly of large arborescent plants and one or two small herbaceous forms (Cantrill & Nichols 1996). Other seed bearing plants include members of the Bennettitales, Cycadales, Ginkgoales and Pentoxylales; collectively they make up only a minor component of the flora (c. 9% species diversity). Of the lower plants, lycopods and horse-tails are rare whilst liverworts are abundant, particularly when com-

pared to Cretaceous floras from lower palaeolatitudes (Cantrill 1997). Liverworts (c. 16% species diversity) include forms allied to the Marchantiales with minor representatives of the Metzgeriales. All these plant remains occur as assemblages in particular sedimentary facies, indicating the existence of spatially complex vegetation communities.

Facies and plant assemblages Braided alluvial plain facies association (Citadel Bastion Member) This member is characterized by erosive-based units of coarsegrained sandstone and gravel; they are laterally continuous for several kilometres and contain channelized incisions up to 14.5 m deep and hundreds of metres wide. Channels are infilled by a basal layer of sandstone and conglomerate containing up to 3 m long conifer logs (93% Podocarpoxylon and 7% Taxiodioxylon, n=14, Table 2). Upper channel-fill units consist of 1–4.5 m thick sandstone packages that exhibit decimetrescale trough cross-bedding or metre-scale low angle crossstratification and contain Brachyphyllum and Ptilophyllum

Braided alluvial plain association KG. 1702.2 (Hyperion N’taks) KG. 1702.3 (Hyperion N’taks) KG. 1702.6 (Hyperion N’taks) KG. 1704.10 (Titan Nunatak) KG. 1704.11 (Titan Nunatak) KG. 2813.1 (Triton Point) KG. 2813.2 (Triton Point) KG. 2813.15 (Triton Point) KG. 2816.36 (Titan Nunataks) KG. 2816.39 (Titan Nunataks) KG. 2817.15 (Titan Nunataks) KG. 2817.16 (Titan Nunataks) KG. 2817.17 (Titan Nunataks) KG. 2817.18 (Titan Nunataks) KG. 2817.22 (Titan Nunataks) KG. 2821.97 (Pagoda Ridge) KG. 2821.98 (Pagoda Ridge) KG. 4586.5 (Hyperion N’taks) KG. 4626.1 (Hyperion N’taks) KG. 4645.1 (Hyperion N’taks) KG. 4657.9 (Titan Nunataks) KG. 4660.1 (Citadel Bastion) KG .4660.4 (Citadel Bastion) KG. 4660.7 (Citadel Bastion) KG. 4710.1 (Offset Ridge) KG. 4710.2 (Offset Ridge) KG. 4710.3 (Offset Ridge) KG. 4710.19 (Offset Ridge) KG. 4712.1 (Titan Nunataks) KG. 4717.42 (Titan Nunataks) KG. 4717.43 (Titan Nunataks) KG. 4717.44 (Titan Nunataks) KG. 4717.45 (Titan Nunataks) KG. 4717.46 (Titan Nunataks) KG. 4717.47 (Titan Nunataks) KG. 4717.50 (Titan Nunataks) KG. 4717.51 (Titan Nunataks) KG. 4719.3 (Titan Nunataks) KG. 4719.4 (Titan Nunataks) KG. 4719.8 (Titan Nunataks) KG. 4719.12 (Titan Nunataks) KG. 4737.152 (Citadel Bastion)

Specimen no. (locality)

Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Taxodioxylon Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp.

Form genus of conifer wood

1 1 2 1 1 2 1 2 2 1 1 2 1 1 1 2 1 1 1 1 1 2 2

1 1 1 1 1 1 1 1 1 2 1 1 1 1 1 1 1 1

Wood fragment Wood fragment Wood fragment Large drifted log Large drifted log in situ tree stump in situ tree stump in situ tree stump in situ tree trunk in situ tree trunk in situ tree stump in situ tree stump in situ tree stump in situ tree stump in situ tree stump in situ tree stump in situ tree stump Large drifted log Large drifted log Large drifted log Large drifted log Wood fragment Large drifted log Large drifted log in situ tree stump in situ tree stump in situ tree stump in situ tree stump Wood fragment Wood fragment Wood fragment Wood fragment Wood fragment Wood fragment Wood fragment Wood fragment Wood fragment Large drifted log Large drifted log Large drifted log Large drifted log Large drifted log

Type of botanical material

Table 2. Inventory of anatomical preserved woods from the Triton Point Formation

Smg Smg Smg ZS ZS MSf MSf MSf Md Md Md Md Md Md Md MSf MSf Sxb Sxb Sxb Sxb Smg ZS Sxb MSf MSf MSf MSf Sxb talus talus talus talus talus talus talus talus Sxb Sxb Sxb Sxb ZS Mean

Facies code

36 53* 48 58* 45* 52* 28 33* 15 44* 96* 68* No data 19 17 16 21* No data 49 No data 68 4‡ 36 22 27 29† 13 13 No data 97† 62 19 No data 11 55 No data 81† 5‡ 90 No data No data No data 1330

No. of growth rings

1.45 1.34* 1.26 2.52* 2.94* 1.50* 1.21 1.26* 0.98 0.58* 1.08* 0.85* No data 2.44 0.65 1.54 1.93* No data 2.59 No data 0.86 0.51‡ 1.54 2.38 1.37 1.17† 0.95 1.79 No data 1.33† 0.45 2.69 No data 2.74 1.73 No data 1.69† 7.32‡ 0.45 No data No data No data 1.66 (1.42)

Mean ring width (mm)

2.69 2.77* 2.75 6.85* 5.85* 3.02* 2.88 3.90* 2.55 1.83* 3.12* 2.21* No data 4.28 1.53 3.53 3.96* No data 4.80 No data 2.20 0.78 2.53 4.05 2.78 4.33 1.65 4.08 No data 5.28 1.18 5.35 No data 3.93 4.05 No data 8.67 9.10 2.23 No data No data No data 3.66

Maximum ring width (mm)

0.492 0.528* 0.573 0.468* 0.421* 0.422* 0.480 0.415* 0.390 0.408* 0.443* 0.492* No data 0.405 0.557 0.499 0.403* No data 0.270 No data 0.425 0.341‡ 0.400 0.252 0.383 0.391† 0.310 0.320 No data 0.445† 0.436 0.517 No data 0.187 0.456 No data 0.501† 0.227‡ 0.440 No data No data No data 0.42 (0.44)

Mean sensitivity (MS)

712 H. J. FALCON-LANG ET AL.

Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Podocarpoxylon sp. Araucariopitys Podocarpoxylon sp. Podocarpoxylon sp. Araucariopitys Podocarpoxylon sp. Podocarpoxylon sp. Araucarioxylon Araucariopitys Araucariopitys Araucariopitys Araucariopitys Araucariopitys Araucariopitys Podocarpoxylon sp. Podocarpoxylon sp.

Form genus of conifer wood

2 2

1 2

1 1

1 2 2 2 2 1 1 1 1 1 1

Large drifted log in situ tree stump in situ tree stump in situ tree stump in situ tree stump in situ tree stump in situ tree stump in situ tree stump in situ tree stump in situ tree stump in situ tree stump Large drifted log Large drifted log in situ tree stump in situ tree stump in situ tree stump Wood fragment Wood fragment Wood fragment Wood fragment in situ tree stump in situ tree stump in situ tree stump in situ tree stump Large drifted log Large drifted log

Type of botanical material

ZS Md Md Md Md Md Md Md Md Md Md Sxl Sxl Md Md Md talus talus talus talus Md Md Md Md Sxb Sxb Mean

Facies code

29 37 6‡ 45 60† 5‡ No data No data No data 34* 5‡ No data No data No data No data No data No data 11 17 6‡ 3‡ 4‡ No data 28 18 10 318

No. of growth rings

1.15 0.56 1.35‡ 1.04 1.68† 2.75‡ No data No data No data 1.16* 4.09‡ No data No data No data No data No data No data 2.17 1.46 3.71‡ 12.98‡ 7.42‡ No data 2.95 2.64 1.56 3.04 (1.79)

Mean ring width (mm)

2.15 1.10 1.88 4.80 4.08 3.05 No data No data No data 4.24* 4.98 No data No data No data No data No data No data 3.28 2.58 4.30 14.70 10.95 No data 5.78 4.30 2.78 4.68

Maximum ring width (mm)

0.312 0.280 0.289‡ 0.385 0.376† 0.094‡ No data No data No data 0.326* 0.175‡ No data No data No data No data No data No data 0.289 0.378 0.114‡ 0.155‡ 0.385‡ No data 0.298 0.376 0.281 0.28 (0.33)

Mean sensitivity (MS)

Facies code (after Nichols & Cantrill 2001); Channel fill facies: Smg, medium- to coarse-grained sandstone; Sxb, cross-bedded sandstone. Overbank facies: Sxl, ripple cross-laminated sandstone; ZS, thinnly-bedded sandstone and siltstone; MSf, fine sandstone and mudstone; Md, dark grey or black mudstone; talus, loose wood fragment on talus slope. *Data taken from Jefferson (1982). †Composite data taken from more than one sample. ‡Less than 10 ring increments measured.

Coastal plain association KG. 2814.2 (Coal Nunatak) KG. 2814.252 (Coal Nunatak) KG. 2814.253 (Coal Nunatak) KG. 2814.254 (Coal Nunatak) KG. 2814.256 (Coal Nunatak) KG. 2814.257 (Coal Nunatak) KG. 2815.52 (Coal Nunatak) KG. 2815.67 (Coal Nunatak) KG. 2815.70 (Coal Nunatak) KG .2815.71 (Coal Nunatak) KG. 2815.77 (Coal Nunatak) KG. 4669.1 (Coal Nunatak) KG. 4669.2 (Coal Nunatak) KG. 4688.44 (Coal Nunatak) KG. 4699.1 (Coal Nunatak) KG. 4699.2 (Coal Nunatak) KG. 4702.1 (Coal Nunatak) KG. 4702.4 (Coal Nunatak) KG. 4702.17 (Coal Nunatak) KG. 4702.28 (Coal Nunatak) KG. 4740.1 (Coal Nunatak) KG. 4740.3 (Coal Nunatak) KG. 4740.5 (Coal Nunatak) KG. 4740.11 (Coal Nunatak) KG. 4747.133 (Coal Nunatak) KG. 4747.138 (Coal Nunatak)

Specimen no. (locality)

Table 2. Continued

CRETACEOUS HIGH-LATITUDE FLOODPLAIN, ANTARCTICA 713

714

H. J. FALCON-LANG ET AL.

foliage. At Titan Nunataks (KG. 4725) the rippled tops of the channel-fill units are patchily covered by Thallites mats (Cantrill 1997). Interbedded with the large channel bodies lie an overbank facies association which includes sheet sandstone beds (2–7 m thick) with flat, non-erosive bases. They are composed of 0.2–2 m thick packages of medium- to coarse-grained sandstone and siltstone that variously exhibit inverse-grading, trough cross-bedding, ripple cross-lamination or plane bedding with primary current lineation. Adjacent to a channel body at KG. 2821 (Pagoda Ridge), immature palaeosols occur on top of a sheet sandstone unit and bear abundant Hausmania ferns in growth position, attached to the bed by stipes (Cantrill 1995). In addition upright podocarp conifer trunks and stumps occur rooted in mature palaeosols beneath the sandstone sheets and extend up through the coarse-grained units to a height of 7 m. Centroclinal cross-stratification is associated with many of these upright trunks. Thinly bedded carbonaceous mudstone, siltstone, finegrained sandstone units together with intermittent 1 m thick medium-grained sandstone beds comprise only a small part of the Citadel Bastion Member, but contain numerous palaeosol horizons. At Titan Nunataks (KG. 4718), an immature palaeosol is dominated by the shrubby angiosperm Gnafelea, together with ferns (Cladophlebis, Aculea), ginkgos, Bennetittes (Ptilophyllum), and rare conifers (Brachyphyllum). Another palaeosol of greater maturity at KG. 4737 is dominated by the foliage of Pentoxylales (Taeniopteris, up to 95% of plant remains) with rare fragments of ferns (Cladophlebis and Hausmania), angiosperms (Araliaephyllum, Timothyia and Dicotylophyllum), and occasional liverworts (Marchantites) (Cantrill & Nichols 1996; Cantrill 1997). A small number of drifted podocarp conifer trunks (100% Podocarpoxylon, n=4, Table 2) and araucarian and podocarp foliage occur in the sandstone beds between the palaeosols. At Offset Ridge (KG. 4710) mature palaeosols occur in fine-grained units and bear conifer stumps in growth position; these are described further in a later section. This facies association is interpreted as being deposited on a braided alluvial plain. Large channelized incisions were cut by braided river channels c. 15 m deep and hundreds of metres wide; lower channel-fill units represent the coarse-grained lag deposits, and upper cross-stratified channel fill units are interpreted as the deposits of mid-channel bars (Nichols & Cantrill 2001). Thick sheet sandstone units containing upright trees represent the product of single large-scale flood events that deposited coarse-grained sediment across the proximal interchannel region, locally burying vegetation (Moncrieff 1989). Centroclinal stratification around some standing trees was formed as the result of current scouring around the upright trunks (Underwood & Lambert 1974). Fine-grained units represent deposition from suspension on the floodplain distant from the braided channel, although the presence of numerous thin sandstone beds indicate that even this distal environment was subject to regular flooding (Jefferson 1981).

Meandering coastal plain facies association (Coal Nunatak Member) This member is only exposed on Coal Nunatak. It contains much thicker successions of fine-grained units than the underlying braidplain association and is characterized by gravelly to coarse-grained, trough cross-bedded sandstone bodies with a distinct channelized geometry. Individual sandstone bodies are

lens-shaped with a width: depth ratio of <15:1 (sandstone ribbons), and can only be laterally traced for a few tens of metres. Where channel margins are exposed they are inclined at a low-angle and within the channel bodies themselves mud draped epsilon cross-stratification is locally present. Plant material in the channels is usually highly macerated, but a few drifted conifer logs occur (100% Podocarpoxylon, n=3, Table 2). Coarse-grained sheet sandstone units (0.1–2 m thick) adjacent to these channel bodies contain ripple cross-lamination, trough cross-bedding or plane bedding and possess non-erosive bases. At KG. 4697, one sheet sandstone unit passes laterally into a channelized sandstone ribbon, and contains allochthonous foliage remains dominated by Cladophlebis (35%), Ginkgoites (32%) and Taeniopteris (9%), together with angiosperms (Hydrocotylophyllum) and conifers (Brachyphyllum) (Cantrill & Nichols 1996). Elsewhere at KG. 4669, proximal deposits contain leafy coniferous branches of Araucaria and drifted conifer logs (50% Araucariopitys and 50% Podocarpoxylon, n=2). Other similar units are covered by laterally extensive liverwort mats of Marchantites type (KG. 4746; Cantrill 1997). The sheet sandstone units also contain common palaeosols in regions distal to the channel bodies. The leaf litter of these palaeosols is typically dominated by fern foliage (Aculea and Alamatus) with subordinate ginkgos, araucarian and podocarp conifers (Araucarites, Pagiophyllum, Podocarpites), and liverworts (KG. 4741 and KG. 4745; Cantrill & Nichols 1996; Cantrill 1997; Cantrill & Falcon-Lang 2001). Fine-grained sequences up to 30 m thick are common but poorly exposed in the Coal Nunatak Member. These consist of thinly bedded mudstone, siltstone and fine-grained, ripple cross-laminated sandstone. Palaeosols with well-developed leaf litter layers occur at multiple horizons and contain abundant silicified conifer stumps in growth position; these are described in greater detail below. This facies association was deposited in a fluvial environment, and sandstone ribbons exhibiting epsilon crossstratification are interpreted as the lateral accretion deposits of meandering river channels (Nichols & Cantrill 2001). The occurrence of mud drapes in one channel body implies tidal influence, and suggests that the meander-belt developed in a coastal setting. Coarse-grained sheet sandstone units are interpreted as channel leve´ es, proximal flood deposits or crevasse splays (Nichols & Cantrill 2001). Fine-grained units represent the product of suspension deposition in distal inter-channel areas.

Mature palaeosols with rooted conifer stumps As noted above mature palaeosols occur in the distal floodplains of both facies associations. Where fully developed they consist of a complex tripartite palaeo-weathered zone composed of an upper leaf litter layer (O horizon), a middle, medium brown, carbonaceous mudstone layer exhibiting closely spaced vertical fractures (c. 8 cm thick; A horizon) and a lower bleached sandstone or siltstone layer (<60 cm thick; E/C horizon). Conifer stumps with basal diameters of 8–50 cm commonly occur in growth position. Roots depart downward from the stumps at angles of 35–90 to the horizontal; they are most abundant in the A horizon where they form a dense mat, but also penetrate throughout the weathered E/C zone (Cantrill & Falcon-Lang 2001). Where they penetrate sandy soils rooting systems exhibit distinctive mycorrhizal nodules characteristic of podocarp conifers (cf. Cantrill & Douglas 1988). Significant differences in the composition and spatial

CRETACEOUS HIGH-LATITUDE FLOODPLAIN, ANTARCTICA A

715

BRAIDPLAIN Cycadophyte-conifer scrub

Broken podocarp woodland (c. 91 trees per hectare)

Vegetation height (m)

30

KEY: Gymnosperm-fern-angiosperm thickets

15 Liverwort mats

Araucarian conifer

Podocarp conifer

Taxodioid conifer

0

Locality:

KG. 4710

KG. 4737

KG. 2821

KG. 4718

B

Bennettite

Arborescent angiosperm

Ginkgo-conifer scrub Fern thickets/conifer woodland

Fern

30

Vegetation height (m)

Gingko

Herbaceous angiosperm

MEANDER-BELT Podocarp-araucarian climax forest (c. 568 trees per hectare)

Taeniopterid KG. 4725

Liverwort

15

0

Locality:

KG. 2814

KG. 4747

KG.4741

KG. 4745

KG. 4697

arrangement of conifer stumps, and the composition of leaf litter layers exist between braidplain and coastal plain settings. In the braided alluvial plain association, 13 palaeosol horizons bearing conifer stumps and trunks have been identified at Pagoda Ridge, with many others occurring at Triton Point, Adonis Ridge, Offset Ridge and Phobos Ridge (Moncrieff 1989). At KG. 4710 on Offset Ridge, thirteen stumps occur over 1540 m2 of palaeosol exposure, where they are grouped into clumps of four to five individuals with a mean density of 91 stumps per hectare. All the stumps in the braidplain association belong to podocarp conifers (100% Podocarpoxylon, n=16, Table 2), but nothing is known of the leaf litter composition of these mature palaeosols. In contrast, the mature palaeosols of the coastal plain meander-belt contain a much higher density of conifer stumps in growth position. At KG. 2815 on Coal Nunatak 54 stumps occur on an exposed area of palaeosol, 25 m by 38 m (950 m2); this equates to a mean density of 568 stumps per hectare (Jefferson 1981). Furthermore coastal plain stumps have a greater compositional diversity than those of the braidplain; both araucarian and podocarp conifers occur (29% Araucariopitys and 71% Podocarpoxylon, n=17, Table 2). At KG. 4747 the litter layer of these palaeosols is dominated by conifers (46% Podozamites and 41% Elatocladus with rare Pagiophyllum and Brachyphyllum) together with a minor component of ferns (Cladophlebis and Sphenopteris), angiosperms (Ficophyllum) and liverworts (Hepaticites). The age of the palaeosols in both facies associations is indicated by growth ring counts in the largest of the conifer stumps; these show that many of the trees lived for more than 100–200 years (Chapman 1994). In terms of general structure and composition, the palaeosols most closely resemble the leached podzolic soils of New Zealand which form on acidic volcanic terrains colonized by podocarp–araucarian conifer forest under a humid, warm temperate climate (Wardle 1991). Similar conditions were probably responsible for the development of the Triton Point Formation palaeosols.

Fig. 2. Summary diagram of Triton Point Formation ecosystems based on facies analysis of plant assemblages, (a) braided alluvial plain facies association and (b) coastal meander-belt association.

Interpretation of plant community composition and structure Analysis of plant taxa distribution within the above palaeoenvironmental context reveals the existence of spatially heterogeneous plant communities in the Triton Point Formation (Fig. 2). In both floodplain settings, newly emergent sand sheets were first colonized by diverse and continuous liverwort mats, plants which may have played an important role in initial sediment stabilization and soil formation. Both floodplain settings were also characterized by a gradient of increasing vegetation complexity moving from frequently disturbed sandy riparian sites to more distal floodbasin localities where flood disturbances were rarer and soils were of greater maturity. The size of the conifer stumps preserved in distal inter-channel areas indicate these were large trees with probable heights in the range 14–29 m (Falcon-Lang & Cantrill 2000). However, beyond these initial similarities, significant differences in the vegetation ecology of the two floodplain settings existed. This variability in ecosystem structure probably resulted from differences in floodplain hydrology, rather than changing climate between the lower and the upper unit. Evidence from palaeosols indicate that climate was similar during the deposition of both units, however, in terms of hydrology there were much greater differences. The braided alluvial plain setting consisted of mobile flood-prone channel belts in which vegetation grew under considerable ecological stress, whilst the coastal plain was characterized by confined meander-belts where floods were less common and climax vegetation could develop. In the braided alluvial plain environment (Fig. 2a), immature, sandy, riparian soils appear to have locally supported patchy, monotypic communities of herbaceous Hausmania ferns. However, the presence of abundant Podocarpoxylon and Taxodioxylon conifer logs and Ptilophyllum and Brachyphyllum foliage within the braided channel deposits, together with in situ podocarp stumps on near-channel sites implies that

716

H. J. FALCON-LANG ET AL.

mature riparian vegetation was elsewhere composed of large podocarp and taxodiod conifers interspersed with bennettitalean trees. Rooting systems with mycorrhizal nodules suggest these sandy soils were well-drained and nutrient-poor. Slightly further from channel influence, sandy soils supported a variety of twotier plant communities. At one site, thickets of shrubby angiosperms and ferns occurred between widely spaced ginkgos and Bennettites. At another site vegetation was dominated by an overstorey of Pentoxylales (Taeniopteris) interspersed with occasional angiosperms and conifers, and a diverse understorey vegetation of ferns (Cladophlebis and Hausmania), rare angiosperm shrubs (Timothyia and Dicotylophyllum) and liverworts (Marchantites). Podzolic soils in inter-channel areas were frequently subjected to flooding, preventing the development of forest climax vegetation on the distal parts of the braided alluvial plain. Instead disturbed inter-channel sites supported scattered low-density (91 trees/ha) clumps of large podocarp conifers as indicated by the distribution of rooted stumps; nothing is known of associated understorey plants. In the coastal plain meander-belt environment (Fig. 2b), where river channels were more confined, extensive immature sandy substrates covered by dense liverwort mats (Marchantites) were rarer. Vegetation on channel leve´ es consisted of a podocarp– araucarian conifer canopy, a subcanopy of ginkgos and cycadophytes and a monotypic fern understorey (Cladophlebis). Further away from channel influence, sandy substrates were colonized by continuous fern-dominated thickets (Aculea and Alamatus) interspersed with shrubby podocarp conifers and scattered stands of ginkgos and araucarian conifers. Rare marchantiod liverworts formed a ground layer (Cantrill 1996). Mature podzolic soils in distal floodbasin regions where flood disturbance was of low frequency, supported medium density (568 trees/ha) podocarp–araucarian climax forests, a similar tree density to that seen in present-day New Zealand podocarp–araucarian rainforests (450–1400 trees/ha) (Duncan 1993). Understorey vegetation was dominated by ferns (Cladophlebis and Sphenopteris), with intermittent ginkgos, angiosperms (Ficophyllum) and liverworts (Hepaticites).

Growth rings in conifer woods Further insight into the dynamic nature of the Triton Point Formation ecosystems may be gained from an examination of growth ring sequences in the silicified conifer stumps and trunks. These represent a continuous record of what growing conditions were like in both facies associations over periods of tens to hundreds of years. In particular two growth ring parameters are useful in analysing growing conditions, mean ring width and mean sensitivity (Creber 1977; Creber & Chaloner 1984). The first parameter, mean ring width, is a measure of the annual productivity of the tree and therefore the favourability of the growing environment (e.g. climate, soil quality, degree of leaf shading). The second parameter, mean sensitivity (MS), is a numerical expression of the year-to-year variability in ring width and is given by the formula:

where x is ring width, n is the number of rings in the sequence analysed, and t is the year number of each ring. Values of mean sensitivity range from 0 where there is no year-to-year variability, to a maximum approaching 2 representing the

greatest possible variability. Under environmentally favourable conditions, ring increments have a relatively constant year-to-year width that reflects the genetic potential of the tree. If a ring sequence contains a high degree of width variability then this implies that environmental conditions were sufficiently stressful to limit growth so that the genetic potential was not attained. An arbitrary MS value of 0.3 is used to distinguish ‘sensitive’ ring sequences with a high degree of incremental width variability (MS>0.3) formed under stressful conditions, from ‘complacent’ sequences with little incremental variability (MS<0.3) formed under favourable conditions (Fritts 1976).

Quantitative ring width and mean sensitivity data The ring width data-set presented here for the Triton Point Formation is the largest yet published for Mesozoic woods (cf. Jefferson 1982; Francis 1984, 1986; Keller & Hendrix 1997; Morgans et al. 1999); a total of 1648 ring increments were measured from undeformed wood samples with the aid of 6 cm by 3 cm thin sections, 1330 from 33 wood samples in the braidplain association and 318 from 16 wood samples in the coastal plain association (Table 2). An important feature to note about the data-set, is that data from short ring sequences (n=<10) are usually characterized by relatively high ring widths and relatively low mean sensitivities. This does not mean that these data are anomalous and should be excluded from the overall data-set on the basis of their statistically questionable short ring sequences, because of the simple fact that wider growth ring increments will inevitably occur in fewer number in a 6 cm by 3 cm thin section. However, in order to emphasize this problematic feature of the data, the mean ring width and mean sensitivity values for the whole of each two facies associations are expressed in two ways in Table 2 (in bold). The first figure is a simple arithmetic mean of data from each wood sample (SM) whilst in the second (in parenthesis) the mean is weighted in proportion to the number of ring increments in each wood sample (WM). Although there is a high degree of variability amongst individual samples, there are clearly discernible differences between the ring characteristics of the two facies associations (Table 2); growth rings are consistently wider (SM: 3.04 mm, WM: 1.79 mm) with a lower mean sensitivity (SM: 0.282, WM: 0.326) in the coastal plain meander-belt association compared with those of the braidplain (ring width, SM: 1.66 mm, WM: 1.42 mm; mean sensitivity, SM: 0.415, WM: 0.439) (Figs 3a–c, 4 & 5). Mean maximum ring width is also wider in the coastal plain association (4.68 mm) compared with the braidplain (3.66 mm). Mean ring width for all trees in both facies associations is 1.92 mm (SM) or 1.49 mm (WM).

Qualitative growth ring data A further qualitative feature of the Triton Point Formation growth rings is the occurrence in eight wood specimens of zones composed of extremely narrow growth rings (Figs 3a–b & 4b). These zones are typically less than 2 mm wide, and may occur at several randomly distributed sites with a single trunk crosssection. They are composed of rings which range from two cells across (the minimum identifiable growth ring) to about nine cells across, which equates to absolute ring widths in the order of 0.05–0.26 mm. Single narrow rings are never observed, but they typically occur grouped together into sequences of three to fourteen. The ring boundaries of these narrow growth rings are qualitatively different from those of the normal growth rings in

CRETACEOUS HIGH-LATITUDE FLOODPLAIN, ANTARCTICA

717

Fig. 3. Conifer woods and barks. (a) Narrow, growth rings of variable width in Podocarpoxylon sp. 1 wood, KG. 4719.4, scale-bar is 1 mm, braided fluvial association (MS=0.44); some very narrow flood rings occur on left side (arrowed). (b) Flood rings in P. sp. 1, KG. 4719.4, scale-bar is 1 mm, braided association (MS=0.28); flood rings (arrowed). (c) Wide growth rings of regular width in P. sp. 2 wood, KG. 2814.252, scale-bar is 1 mm, meandering fluvial association (MS=0.28). (d) Thick bark in P. sp. 2, KG. 4710.3, scale-bar is 3 mm, braided association. (e) Root trace in bark of P. sp. 2, KG. 4710.3, scale-bar is 1 mm, braided association.

that they are usually subtly developed, being marked by latewood cells of fewer number and larger diameter. In addition, the narrow growth rings are locally symmetrical across the ring boundary, and do not persist around the whole trunk circumference. Furthermore individual tracheids within these anomalous wood zones often possess a rounded cross-sectional shape, such that they do not perfectly tessellate resulting in the occurrence of abundant intercellular spaces. Good examples of the ‘narrow ring’ phenomenon are restricted to seven podocarp wood specimens from the braided river facies association (particularly KG. 4717.51 and KG. 4719.3), and one poorly developed example from in the coastal plain facies association. These narrow ring sequences were excluded from the data used to calculate the above MS values.

Environmental and ecological interpretation In the mid-Cretaceous polar circle, climate possessed a distinctly seasonal cycle characterized by warm, light summers and cool, dark winters (Read & Francis 1992), and the growth rings of the Triton Point Formation woods almost certainly reflect this rhythm, with one ring forming annually (Francis 1986). Narrower growth rings with greater year-to-year variability in the braided river association on Alexander Island imply that growing conditions were more stressful than in the coastal plain meander-belt environment. Two factors may have been important in generating this pattern. First, mean sensitivity is generally

greater for trees growing in discontinuous woodlands compared with those in closed forests because the growing environment is generally more stable in the latter situation (Creber 1977). Maximum conifer density on the Triton Point Formation braidplain was only one sixth of that on the coastal plain, consequently braidplain trees were probably more exposed to adverse environmental impacts such as wind damage which may have resulted in more irregular growth patterns (higher MS). However, if treespacing was the primary factor controlling the nature of the growth ring sequences, one might have expected the braidplain trees to have formed wider rings than those of the meander-belt, because in the former situation there would have been less competition for light and soil nutrients (Koga et al. 1997). In fact the opposite is the case, and rings are widest on the densely forested meander-belt. A second possible factor responsible for the observed growth ring patterns may have been the differing hydrological regimes of the two floodplains. Braided river systems are much more prone to bank bursting and overbank flooding than meandering systems because energy conditions are much greater in the former (Reading 1996). The impact of flooding on vascular cambial growth is complex, and results in an initial short-term increase in growth rate followed by a longer term decline (Kozlowski 1984), such that the growth rings of flooded trees may be only 60% of the width of unflooded neighbours (Yamamato et al. 1987). The frequent inundation of the braidplain woodlands of Alexander Island would therefore have given rise to a much more irregular tree growth rate and

718

H. J. FALCON-LANG ET AL.

0.6

a KG. 1702.6

2

0.3 0.2 0.1 0

10

0

20

30

0

40

Ring number

b 3

Ring width (mm)

0.4

1

0

MS=0.501

KG. 4717.51

2

Flood rings

Flood rings

1

0

10

20

30

40

50

60

Ring number

c KG. 2814.2

MS=0.312

2

1

0 0

10

2

4

6

8

10

12

Growth ring width (mm)

0

Ring width (mm)

0.5

MS=0.573

Mean sensitivity

Ring width (mm)

3

20

Ring number

Fig. 4. Examples of ring width sequences (a) braidplain wood with high mean sensitivity (MS), (b) braidplain wood with high MS and narrows of very narrow flood rings (shaded), (c) coastal meander-belt wood with moderate MS.

Fig. 5. Growth ring data showing mean sensitivity versus ring width; black diamonds denote braidplain woods, white squares denote coastal meander-belt woods.

therefore higher mean sensitivity than on the meander-belt. In addition, the long-term suppression of growth due to flooding also explains why the braidplain trees generally possess narrower annual rings. Similar ring patterns are observed in modern floodplain forests. For example, Martens (1993) showed that Salix growing on flood-prone, in-channel benches possessed narrower rings with greater year-to-year variability than those growing on adjacent, permanently drained leve´ es and terraces. Zones of very narrow rings in Triton Point Formation woods provide additional information about growing conditions on the two floodplain systems. It is unlikely that these very narrow rings (0.05–0.26 mm) represent annual growth increments because their subtle, discontinuous ring boundaries imply that they were formed by a short-term reduction in cambial activity as opposed to a long-term cessation associated with the onset of winter. Instead we interpret them as intraannual rings (‘false rings’ in the terminology of Fritts 1976). Several environmental disturbances may induce a short-term reduction in cambial activity during the growing season, and give rise to multiple rings in each year’s growth increment; these include frost (Glock et al. 1960), drought (Ash 1983), fire (Dechamps 1984) and flooding (Young et al. 1993). In the mild and humid climatic setting envisaged for the Triton Point Formation, the only one of these disturbances which is likely to have occurred regularly enough to produce the observed false ring sequences is flooding. This conclusion is supported by the presence of rounded tracheids and intercellular spaces in the podocarp woods. This anatomical feature has been associated with greatly stimulated ethylene production in the submerged portions of conifer boles during floods (Yamamoto 1992). The development of such intercellular spaces in flooded conifers is believed to improve oxygen diffusion rate into the root system (Hook 1984). Woods containing flood-rings are almost entirely restricted to the braidplain environment on Alexander Island, further supporting the hypothesis that vegetation in this setting was prone to regular flooding. Up to 14 adjacent flood rings in one wood sample implies that this particular tree was inundated many times during a single growing season.

CRETACEOUS HIGH-LATITUDE FLOODPLAIN, ANTARCTICA

Stumps with attached barks and adventitious roots Description Three tree stumps of Podocarpoxylon sp. 1 and sp. 2 rooted in palaeosols in the braidplain facies association (KG. 4660.1, KG. 4710.2 and KG. 4710.3) exhibit extremely thick (up to 43 mm wide) bark. The periderm is largely composed of alternate layers of thin and thick walled cells arranged in approximately radial files. Scattered thoughout the periderm are small diameter (0.3–3.0 mm) roots (Fig. 3d–e). The roots comprise a central diarch or rarely triarch vascular bundle surrounded by a prominent endodermis. Outside the endodermis are thin walled parenchymatous cells of the cortex. In many instances this cortical material is crushed particularly around the edges. Root traces pass vertically or slightly radially through the bark towards the outside of the trunk. Rooting systems associated with the podocarp stumps and those elsewhere in the braidplain unit are shallowly penetrative (typically <20–30 cm), horizontally orientated, and possess mycorrhizae (cf. Cantrill & Douglas 1988).

Ecological and environmental interpretation Growth ring data indicates that the braidplain trees grew in a flood-prone environment. Further support for this interpretation is provided by the podocarp stumps with thick barks containing root traces. Some ecological studies interpret the occurrence of thick bark as a fire adaptation, which acts to insulate the vascular cambium from catastrophic heating (Burns 1993). However, fires appear to have been uncommon in the humid temperate ecosystems of Alexander Island (see below), and are unlikely to have been a sufficient ecological pressure to influence stem anatomy. Other studies suggest that the thick bark represents a response to flooding because it facilitates greater aeration of the submerged portion of the trunk (Hook 1984). For example, 150–230% increases in bark thickness have been reported in some conifers following flooding (Yamamoto & Kozlowski 1987; Yamamato et al. 1987). Some support for this is seen in entombed trunks where individual trees have been partially buried and the trunk forms an expanded bole at the top of the new sediment surface (Jefferson 1982, pl. 66, fig. 4). In flooded coniferous trees from western Canada new roots developed at the top of the new soil surface (Stone & Vasey 1968). Given the braidplain setting of the Triton Point Formation podocarp trunks, thick barks may represent a flood adaptation. Rootlets within the podocarp barks must have grown late in the tree’s life, and are interpreted as being adventitious. Adventitious root growth is stimulated by flooding in many present-day conifers and appears to aid nutrient absorption from remaining aerated soil zones (Kozlowski 1984; Yamamoto 1992). In addition, the shallow, horizontally orientated nature of podocarp rooting systems in the braidplain palaeosols provide yet further evidence that these conifer communities were flood-prone; such root gross-morphology is characteristic of periodically waterlogged soils because roots preferentially grow in the most aerated upper soil zone (Kozlowski 1984).

Tissue damage in conifer woods Description Three types of tissue damage occur in the conifer woods. First, traumatic parenchyma is present in <2% of the specimens. One juvenile specimen of Araucariopitys (KG. 4702.28) from the coastal plain facies association, exhibited two discontinuous

719

rings of traumatic parenchyma which occurred within the first three growth increments (Fig. 6a–b). Wood immediately following these traumatic zones was characterized by cells with resinous contents. Chapman (1994) also noted similar traumatic parenchyma scattered throughout the innermost nine rings of an unidentified conifer from the braidplain facies association. Second, fungal material is common. For example, in 4% of the specimens studied large spindle-shaped cavities (up to 3 mm in diameter and up to 5 cm high) occur in the latewood (Fig. 6d). They contain oval fungal organs (9–12 µm by 3–15 µm) comparable to modern basidospores or teliospores produced by Basidiomycetes, and aggregates of circular bodies which may represent fruiting structures, along their inner margin. Third, 7% of the wood samples contain large, complex, branching chambers several centimetres across (Fig. 6e). These are filled with abundant, closely packed, 1–3 mm long, oval, dark-coloured bodies containing highly digested xylem (frass), interpreted as arthropod coprolites (Fig. 6c). One slender stem of Podocarpoxylon sp. 2 (KG. 4717.43) contained two such chambers. The first was emplaced following the growth of the eleventh ring increment (stem diameter at 13 mm); the formation of seven further ring increments were required to entirely cover this scar. The second chamber was emplaced following the twentieth ring increment (stem diameter at 24 mm) (Fig. 6f). This resulted in cambial damage over 17% of the stem circumference and even following the growth of a further 42 ring increments this scar had not been entirely covered. Growth ring width decreased markedly associated with each chamber emplacement event (mean ring width is only 0.45 mm) (Fig. 7).

Ecological and environmental interpretation Rings of traumatic tissue similar to those observed in the Araucariopitys specimen (KG. 4702.28) are common in some modern conifer woods and usually represent cambial response to canopy damage by fire (Dechamps 1984), frost (Glock et al. 1960) or arthropod defoliation (Fritts 1976). Given the rarity of these traumatic features it is impossible to attribute them with certainty to any of these causal agents. However, it is known that frost rings are commonly formed immediately after volcanic events, when temperatures drop unusually low (LaMarche & Hirschboek 1984). The forest bearing the Araucariopitys specimen is buried by a volcanic ash layer on Coal Nunatak and it is possible that the traumatic tissue may represent a volcanically induced frost ring. The presence of large spindle-shaped cavities in the latewood which contain fungal bodies represent the seasonal attack of conifers by fungi. In their gross-morphology these structures are very similar to some white pocket rots (Blanchette 1992). Similar features have been seen in Permian and Triassic conifer trunks from Antarctica (Stubblefield & Taylor 1986) and North America (Creber & Ash 1990). The position of the white rots within latewood component of the ring increment may indicate that fungal attack occurred during the dusky late autumn or dark winter when the trees’ defense mechanism would have been particularly vulnerable. Another biological disturbance to tree growth was the result of arthropod attack. The emplacement of frass-filled chambers following the production of the latewood, again suggests that arthropod attack occurred to the trees’ dormant phase during the late autumn/winter months. The Podocarpoxylon sp. 2 trunk described (KG. 4717.43) appears to have been badly damaged by arthropod attack on two occasions, when the tree

720

H. J. FALCON-LANG ET AL.

Fig. 6. Traumatic tissue in trunks. All transverse section. (a–b) Frost rings in young trunk of Araucariopitys, KG. 4702.28, (a) scale-bar is 100 µm (b) scale-bar is 50 µm. (c) Oval arthropod coprolites in excavated chamber, Podocarpoxylon sp. 2, KG. 4717.43, scale-bar is 500 µm. (d) Fungal white rots concentrated in latewood region, P. sp. 2, KG. 4633.46, scale-bar is 2 mm. (e) Complex chamber filled with arthropod coprolites, P. sp. 2, KG. 4717.43, scale-bar is 1 mm. (f) Cambial regrowth over wound caused by emplacement of arthropod chamber, P. sp. 2, KG. 4717.43, scale-bar is 2 mm.

was 11 and 20 years old. Attacks resulted in a marked decrease in growth rate and massive cambial damage from which the tree never fully recovered. A similar reduction in growth ring width has been observed in Douglas fir following arthropod attack; reduction in ring width was strongly correlated with degree of canopy defoliation (Alfaro & Shepherd 1991).

Charcoal deposits Abundance and taxonomic composition At some forest horizons plant material is anatomically preserved as charcoal (fusain), the product of incomplete combustion in vegetation fires (Jones & Chaloner 1991). Charcoal occurs in association with pale water-lain tuff layers, and is present in low abundance throughout the Triton Point Formation. In addition wood reflectance studies show rare bimodal peaks throughout the Fossil Bluff Group implying that both vitrinite and fusinite were present in measured samples (Doubleday 1994).

In one tuff sample (KG. 2818.1) collected from an overbank sequence within the meander-belt association, charcoal occurs as an unsorted accumulation consisting of fragments ranging in size from 50 µm to 14 mm. Given the fragmented nature of the material only a small proportion was identifiable. Coniferous woods dominate the assemblage and include Araucariopitys and both Podocarpoxylon species in approximately equal proportion (Fig. 8a–b). Small leafless twigs (0.5 mm diameter) with parenchymatous piths were also common, and on the basis of juvenile wood anatomy were probably derived from podocarp conifers. Other probable coniferous remains included young primary shoots, isolated diamond-shaped scale leaves similar to those of Brachyphyllum or Pagiophyllum, and numerous seed-coat fragments (Fig. 8c–d).

Interpretation of fire ecology Charcoal abundance in Triton Point Formation is very low when compared with the deposits of well-studied, densely vegetated floodplains elsewhere in the geological record (e.g.

CRETACEOUS HIGH-LATITUDE FLOODPLAIN, ANTARCTICA

Growth ring width (mm)

1.4

KG. 4717.43 Second injury

1.2 1.0

First injury

0.8

R

R

0.6 0.4 0.2 0.0

0

10

30 40 20 Ring number

50

60

Fig. 7. Ring width sequence in trunk attacked twice by wood-boring arthropods. Each injury caused large decrease in radial growth rate. R denotes the period during which the tree recovered from the injury.

721

Cope 1993; Falcon-Lang 1999, 2000). Charcoal has been rarely documented at other mid-Cretaceous southern high-latitude forest sites, the only record being that of Francis & Coffin (1992) who described some Albian podocarp wood charcoal from the Kerguelan Plateau (ODP Site 750) at a palaeolatitude of 56S. The frequency with which fires occur in an ecosystem is closely related to the moisture content of the vegetation which is itself linked to atmospheric humidity (Uhl et al. 1988). In modern humid tropical and temperate rainforests high vegetation moisture content is maintained year-round and consequently these ecosystems have very low fire-frequencies with fire events spaced several hundred or thousand years apart (Veblen 1982; Sandford et al. 1985; Burns 1993). Independent evidence suggests that the Triton Point Formation was probably deposited under a humid temperate palaeoclimate (Parrish et al. 1982, 1998) and we interpret the rarity of charcoal in this unit as indicating that fires were rare events in the mid-Cretaceous polar biome. The association of charcoal with tuff layers implies that volcanism may have initiated some fires as occurs in the podocarp temperate rainforests on the flanks of Mount Taupo, New Zealand (Wilmshurst & McGlone 1996).

Fig. 8. Scanning electron micrographs of charcoal from coastal plain meander-belt facies association at KG. 2818.1. (a) Cross-section of podocarp conifer twig, scale-bar is 200 µm. (b) Radial view of Araucariopitys conifer wood, scale-bar is 20 µm. (c) Conifer seed coat, scale-bar is 500 µm. (d) Conifer scale-leaf of Brachyphyllum- or Pagiophyllum-type, scale-bar is 500 µm.

722

H. J. FALCON-LANG ET AL.

Discussion Palaeoecological synthesis A kaleidoscopic variety of intergrading conifer–ferndominated plant communities existed on the floodplains of Alexander Island, Antarctica during the Late Albian. These grew under a seasonal climate characterized by cool, dark winters and warm, light summers; and high year-round rainfall (Spicer & Chapman 1990). Vegetation dominated by low density (91 trees/ha) stands of podocarp and taxodiod conifers grew on mobile braided alluvial plains where it was subject to regular catastrophic flooding events. On coastal meander-belts conditions were more stable and floods less frequent, so that medium density (568 trees/ha) araucarian–podocarp rainforests were able to establish. This climax vegetation possessed a four-tier structure consisting of a c. 30 m high conifer canopy, a sub-canopy of tree ferns, ginkgos, Bennettitales, Pentoxylales and angiosperms, a herbaceous layer dominated by ferns and minor angiosperms, and a liverwort ground layer. Almost all of the arborescent vegetation on both floodplain types possessed a broad-leafed evergreen canopy, although some elements of the sub-canopy and herbaceous layer were probably deciduous (Falcon-Lang & Cantrill 2001). Vegetation disturbances due to unusually hard frosts, wildfires and attack by wood-boring arthropods and fungi periodically occurred, but with very low frequency.

Comparison with warm temperate rainforests of New Zealand The composition, structure and ecology of the Alexander Island vegetation bears some resemblance to the extant rainforests of New Zealand which grow under a humid (1500 mm a
1) and warm (MMST 16–22C, MMWT 3–8C) temperate climate (Wardle 1991). Like the vegetation described here from Alexander Island, these forests consist of a podocarp–araucarian conifer canopy (20–30 m high) and several understorey layers dominated by angiosperms, tree-ferns, palms, ferns, mosses and liverworts (Wardle 1991). The largest difference between the Cretaceous forests and their putative closest modern analogue is the greater abundance of angiosperms in the latter (locally up to 52% of the woody vegetation; Duncan 1993), a plant group which only evolved and migrated to polar regions immediately prior to the deposition of the Triton Point Formation (Hill & Scriven 1995). Whilst angiosperm radiation has greatly modified the ecology of the Southern Hemisphere temperate rainforests, a few features of these extant ecosystems provide important clues for interpreting the terrestrial palaeoecology of Alexander Island. For example, the araucarian and podocarp conifers of New Zealand exhibit a degree of ecological partitioning; araucarians prefer well-drained, ultra-infertile podzolic soils (Ecroyd 1982) whilst podocarps prefer flood-prone, semiinfertile alluvial substrates where periodic flooding plays an important role in forest regeneration (Duncan 1993; Odgen & Stewart 1995). These present-day ecological preferences explain the almost complete restriction of araucarian conifers to the stable meander-belts of Alexander Island, and the dominance of podocarps on the disturbed flood-prone braidplains. Furthermore, modern araucarian conifers are frost hardy down to
11C and podocarps down to
23C; below such temperatures frost rings are produced (Sakai & Larcher 1987). Assuming that Cretaceous conifers had similar thermal tolerances, the rare occurrence of frost rings in araucarian woods from Alexander Island imply that temperatures only occasion-

ally fell below the
11C, whilst their absence in the podocarp woods indicates that the
23C threshold was never reached. Mean growth ring width in extant New Zealand conifers and those of Cretaceous Alexander Island are also very similar. Mean annual ring widths for mature podocarps range up to 1 mm in South Island (42–46S), and for araucarians and podocarps in North Island (35–42S), up to 2.2 mm (Dunwiddie 1979; Ahmed & Ogden 1987; Norton et al. 1988). Annual ring widths in the order of 1–2.5 mm have also been recorded in araucarian conifers in warm temperate Chile (Burns 1993). On Alexander Island conifers have a total mean ring width of 1.92 mm. However, these similarities in ring width do not necessarily imply that growing climate was the same on Alexander Island as it is in present-day warm temperate Chile and New Zealand. Studies by Downs (1962) rediscovered by G.T. Creber (Univ. London, pers. comm. 2000) have shown that conifer seedlings grow at a very high rate under conditions of permanent sunlight, as would have existed during the summer months on Alexander Island. Elevated growth rates under polar light conditions may be caused by three factors. First, plants can make use of the whole year’s 4380 hours of daylight (i.e. 36512 hours which every point on the planet receives annually) which is delivered entirely during the warm summer growing season. Second, growth inhibitors build up during dark periods which need to be broken down in the subsequent light period before growth can recommence; such growth inhibitors do not form during a regime of continuous daylight. Third, sucrose needed to translocate photosynthate from the leaves to the trunk is only produced during daylight. During the dark periods of lower latitude regions, vascular cambium activity is greatly reduced because translocation slows. However, under the continuous light regime of the polar biome growth may occur continuously (G.T. Creber pers. comm. 2000). As a consequence of this elevated growth rate under polar conditions, it is impossible to relate ring width directly to climatic favourability. Nevertheless, ring width data may be used to calculate the annual productivity of the Alexander Island forests. Using mean ring increment, mean stump diameter and forest density data, Creber & Francis (1999) estimated total wood productivity for the Alexander Island climax rainforest to be 17.65 m3 ha
1 a
1 based on Jefferson’s (1981) data. Using our revised data (presented above and based on additional fieldwork), we estimate that the wood productivity of the Alexander Island forests was probably closer to 5.62–7.33 m3 ha
1 a
1. These values are similar to those of compositionally similar araucarian–podocarp stands in warm temperate North Island, New Zealand, which have productivities in the order of 4.5–7.5 m3 ha
1 a
1 (Wardle 1991).

Polar rainforests and mid-Cretaceous climate-vegetation models Warm temperate broad-leafed, evergreen araucarian– podocarp rainforests similar to those of Alexander Island appear to have dominated much of Gondwana above a palaeolatitude of 60S (Fig. 1a; Spicer & Chapman 1990). Various attempts have been made to understand the climate and vegetation of this region using sophisticated numerical models. For example, Valdes et al. (1996) applied the UGAMP model to Smith et al.’s (1994) representation of midCretaceous palaeogeography to analyse summer and winter surface air temperatures. They estimated that SE Alexander Island, and much of coastal Antarctica, central Australia and New Zealand was characterized a mild climate, with winter

CRETACEOUS HIGH-LATITUDE FLOODPLAIN, ANTARCTICA

temperatures of 0C to
4C and summer temperatures of 20–24C. These values are very similar to those estimated for SE Alexander in this paper on palaeobotanical grounds. In another study, Beerling (2000) modelled mid-Cretaceous (Albian) net primary productivity for the South Hemisphere. His model indicated that the whole of Antarctica, and surrounding high-latitude landmasses had a terrestrial net primary productivity similar to modern Southern Hemisphere warm temperate evergreen forests. Calculations of productivity above based on palaeobotanical data have verified Beerling’s (2000) model estimates, indicating that the SE Alexander Island forests had a similar productivity to those of present warm temperate New Zealand.

Conclusions (1) Humid temperate rainforests grew on mid-Cretaceous (Late Albian), high-latitude (75S) braided alluvial plains and coastal meander-belts in SE Alexander Island, Antarctica, and comprised a kaleidoscopic variety of inter-grading conifer–fern dominated communities. (2) Floodplain hydrology exerted the greatest influence on community structure. Only scattered woodlands and thickets grew on braidplains where flood disturbance was frequent, but on more stable meander-belts dense climax rainforest established. (3) In terms of composition, structure, ecology and productivity, vegetation resembled the warm temperate rainforests of present-day New Zealand. The presence of warm temperate forests close to the mid-Cretaceous South Pole verifies the results of recent numerical computer models of midCretaceous climates and biomes. Fieldwork on Alexander Island by two of us (G.J.N. and D.J.C.) was supported by the British Antarctic Survey, and was undertaken during the 1992–93 austral summer season. We thank Mike Tabecki and Jeanette Ringman for preparing petrographic thin sections, Chris Gilbert and Pete Bucktrout for preparing photographic plates, and Ken Robinson for assistance with Scanning Electron Microscopy. The reviews of John Howell, Geoff Creber and Helen Morgans-Bell greatly improved this paper.

References A, M. & O, J. 1987. Population dynamics of the emergent conifer Agathis australis (D.Don) Lindl. (Kauri) in New Zealand. I. Population structures and tree growth rates in mature stands. New Zealand Journal of Botany, 25, 217–229. A, R.I. & S, R.F. 1991. Tree-growth of interior Douglas-fir after one year’s defoliation by Douglas-fir tussock moth. Forest Science, 37, 959–964. A, S. 1963. A new Mesozoic flora from Tico, Santa Cruz Province, Argentina. Bulletin of the British Museum (Natural History), Geology, 8, 47–92. A, J. 1983. Growth rings in Agathis robusta and Araucaria cunninghamii from tropical Australia. Australian Journal of Botany, 31, 269–275. A, D.I. 1984. An interpretation of Cretaceous and Tertiary biota in polar regions. Palaeogeography, Palaeoclimatology, Palaeoecology, 45, 105–147. B, M.K. & C, I.B. 1994. Fossil wood of Cretaceous age from the Namaqualand continental shelf, South Africa. Palaeontologia africana, 31, 83–95. B, E.J. 1984. Climatic implications of the variable obliquity explanation of Cretaceous-Palaeogene high-latitude floras. Geology, 12, 595–598. B, D.J. 2000. The influence of vegetation cover on soil organic matter preservation in Antarctica during the Mesozoic. Geophysical Research Letters, 27, 253–256. B, R. 1992. Anatomical responses of xylem to injury and invasion by fungi. In: B, R.A. & B, A.R. (eds) Defense mechanisms of woody plants against fungi. Springer, New York, 76–95. B, J.R. 1996. Sandstone provenance and diagenesis of arc-related basins; James Ross Island and Alexander Island, Antarctica. PhD thesis. University of Exeter.

723

B, B.R. 1993. Fire-induced dynamics of Araucaria araucana-Nothofagus antarctica forest in the southern Andes. Journal of Biogeography, 20, 669–685. C, D.J. 1995. The occurrence of the fern Hausmannia Dunker (Dipteridaceae) in the Cretaceous of Alexander Island, Antarctica. Alcheringa, 19, 243–254. C, D.J. 1996. Fern thickets from the Cretaceous of Alexander Island, Antarctica containing Alamatus bifarius Douglas and Aculea acicularis sp. nov. Cretaceous Research, 17, 169–182. C, D.J. 1997. Hepatophytes from the Early Cretaceous of Alexander Island, Antarctica: systematics and palaeoecology. International Journal of Plant Sciences, 158, 476–488. C, D.J. & D, J.G. 1988. Mycorrhizal conifer roots from the Lower Cretaceous of the Otway Basin, Victoria. Australian Journal of Botany, 36, 257–272. C, D.J. & F-L, H.J. 2001. Cretaceous (Late Albian) Coniferales of Alexander Island, Antarctica. 2: Foliage, reproductive structures and roots. Review of Palaeobotany and Palynology, in press. C, D.J. & N, G.J. 1996. Taxonomy and palaeoecology of Early Cretaceous (Late Albian) angiosperm leaves from Alexander Island, Antarctica. Review of Palaeobotany and Palynology, 92, 1–28. C, J.L. 1994. Distinguishing internal developmental characteristics from external palaeoenvironmental effects in fossil wood. Review of Palaeobotany and Palynology, 81, 19–32. C, L.J. & J, H.C. 1999. New oxygen isotope evidence for long-term Cretaceous climatic change in the Southern Hemisphere. Geology, 27, 699–702. C, M.J. 1993. A preliminary study of charcoalified plant fossils from the Middle Jurassic Scalby Formation of North Yorkshire. Special Papers in Palaeontology, 49, 101–111. C, G.T. 1977. Tree rings: a natural data-storage system. Biological Reviews, 52, 349–383. C, G.T. & A, S.R. 1990. Evidence of widespread fungal attack on Upper Triassic trees in the southwestern USA. Review of Palaeobotany and Palynology, 63, 189–195. C, G.T. & C, W.G. 1984. Influence of environmental factors on the wood structure of living and fossil trees. The Botanical Review, 50, 357–448. C, G.T. & C, W.G. 1985. Tree growth in the Mesozoic and early Tertiary and the reconstruction of palaeoclimates. Palaeogeography, Palaeoclimatology, Palaeoecology, 52, 35–50. C, G.T. & F, J.E. 1999. Fossil tree-analysis: palaeodendrology. In: J, T.P. & R, N.P. (eds) Fossil plants and spores: modern techniques. Geological Society, London, 245–250. D, R. 1984. Evidence of bush fires during the Plio-Pliestocene in Africa (Omo and Sahabi) with aid of fossil woods. In: C, J.A. & V Z B, E.M. (eds) Paleoecology of Africa and the surrounding islands 16. Balkeema, Cape Town, 231–239. D, M.E., M, R.E., D, J.G., B, D., F, C., C, H.T., F, J.E., J, P., R, T., W, W., R, P.V., P, N., K, A. & R, A. 1992. Australian Cretaceous terrestrial faunas and floras: biostratigraphic and biogeographic implications. Cretaceous Research, 13, 207–262. D, P.A.D. 1994. Structural and tectonic evolution of parts of the Mesozoic fore-arc of Alexander Island, Antarctica. PhD Thesis. University of Leeds. D, R.J. 1962. Photocontrol of growth and dormancy in woody plants. In: K, T.T. (ed.) Tree growth. Ronald Press, New York, 133–148. D, R.P. 1993. Flood disturbance and the coexistence of species in a lowland podocarp forest, south Westland, New Zealand. Journal of Ecology, 81, 403–416. D, P.W. 1979. Dendrochronological studies of indigenous New Zealand trees. New Zealand Journal of Botany, 17, 251–266. E, C.E. 1982. Biological flora of New Zealand 8. Agathis australis (D. Don) Lindl. (Araucariaceae) Kauri. New Zealand Journal of Botany, 20, 17–36. F-L, H.J. 1999. Fire ecology of a Late Carboniferous floodplain, Joggins, Nova Scotia. Journal of the Geological Society, London, 156, 137–148. F-L, H.J. 2000. Fire ecology of the Carboniferous tropical zone. Palaeogeography, Palaeoclimatology, Palaeoecology, 164, 355–371. F-L, H.J. & C, D.J. 2000. Cretaceous (Late Albian) Coniferales of Alexander Island, Antarctica. 1: Wood taxonomy, a quantitative approach. Review of Palaeobotany and Palynology, 111, 1–17.

724

H. J. FALCON-LANG ET AL.

F-L, H.J. & C, D.J. 2001. Leaf phenology of some mid-Cretaceous polar forests, Alexander Island, Antarctica. Geological Magazine, 138, 39–52. F, L.A. & F, J.E. 1988. A guide to Phanerozoic cold polar climates from high latitude ice-rafting in the Cretaceous. Nature, 333, 547–549. F, J.E. 1984. The seasonal environment of the Purbeck (Upper Jurassic) fossil forests. Palaeogeography, Palaeoclimatology, Palaeoecology, 48, 285–307. F, J.E. 1986. Growth rings in Cretaceous and Tertiary wood from Antarctica and their palaeoclimatic implications. Palaeontology, 29, 665–684. F, J.E. & C, M. 1992. Cretaceous fossil wood from the Raggatt Basin, southern Kerguelen Plateau (Site 750). Proceeding of the Ocean Drilling Program, Scientific Results, 120, 273–280. F, H.C. 1976. Tree Rings and Climate. Academic Press, London. G, W.S., S, R.A. & A, S.R. 1960. Classification and multiplicity of growth layers in the branches of trees. Smithsonian Miscellaneous Collections, 140. Publication No. 4421, Washington. G, S.T. & R, J.H. 1990. Larches: deciduous conifers in an evergreen world. BioScience, 40, 819–826. H, A.B. & S, R.A. 1996. Palaeobotanical evidence for a warm Cretaceous Arctic Ocean. Nature, 380, 330–333. H, R.S. & S, L.J. 1995. The angiosperm-dominated woody vegetation of Antarctica: a review. Review of Palaeobotany and Palynology, 86, 175–198. H, D.D. 1984. Adaptations to flooding with fresh water. In: K, T.T. (ed.) Flooding and plant growth. Academic Press, London, 265–293. J, T.H. 1981. Palaeobotanical contributions to the geology of Alexander Island, Antarctica. PhD Thesis. University of Cambridge. J, T.H. 1982. Fossil forests from the Lower Cretaceous of Alexander Island, Antarctica. Palaeontology, 25, 681–708. J, T.P. & C, W.G. 1991. Fossil charcoal, its recognition and palaeoatmospheric significance. Palaeogeography, Palaeoclimatology, Palaeoecology (Global Change Section), 97, 39–50. K, S.R.A. & M, A.C.M. 1992. Marine molluscan constraints on the age of Cretaceous fossil forests of Alexander Island, Antarctica. Geological Magazine, 129, 771–778. K, A.M. & H, M.S. 1997. Paleoclimatologic analysis of a Late Jurassic petrified forest, southeastern Mongolia. Palaios, 12, 282–291. K, S., O, K., T, J. & F, T. 1997. Effect of thinning on the wood structure in annual growth rings of Japanese larch (Larix leptolepis). IAWA Journal, 18, 281–290. K, T.T. 1984. Responses of woody plants to flooding. In: K, T.T. (ed.) Flooding and plant growth. Academic Press, London, 129–163. LM, V.C. & H, K.K. 1984. Frost rings in trees as records of major volcanic eruptions. Nature, 307, 121–126. M, D.M. 1993. Hydrologic inferences from tree-ring studies on the Hawkesbury River, Sydney, Australia. Geomorphology, 8, 147–164. MC, J.J. & L, R.D. 1998. Late Cretaceous to early Tertiary subduction history of the Antarctic Peninsula. Journal of the Geological Society, London, 155, 255–268. MC, J.J. & M, I.L. 1997. The age and stratigraphy of fore-arc magmatism on Alexander Island, Antarctica. Geological Magazine, 134, 507–522. M, A.C.M. 1989. Field report of sedimentological studies in southeastern Alexander Island, Antarctica 1988/89. British Antarctic Survey Field Report, AD6/2R/1988/G4. M, A.C.M. & K, S.R.A. 1993. Lithostratigraphy of the uppermost Fossil Bluff Group (Early Cretaceous) of Alexander Island, Antarctica: history of an Albian regression. Cretaceous Research, 14, 1–15. M, H.S., H, S.P. & S, R.A. 1999. The seasonal climate of the Early-Middle Jurassic, Cleveland Basin, England. Palaios, 14, 261–272. N, G.J. & C, D.J. 2001. Fluvial sedimentology and revised stratigraphy of the Triton Point Formation, Alexander Island, Antarctic Peninsula. Geological Magazine, 138, in press. N, D.A., H, J.W. & B, A.E. 1988. The ecology of Dacrydium cupressinum: a review. New Zealand Journal of Botany, 26, 37–62. O, J. & S, G.H. 1995. Community dynamics of the New Zealand conifers. In: E, N.J. & H, R.S. (eds) Ecology of the Southern Conifers. Melbourne University Press, Australia, 81–119. O-B, B.L. & U, G.R. 1997. Vegetation-induced warming of high-latitude regions during the Late Cretaceous period. Nature, 385, 804–807.

P, J.M., P, J.T., H, J.H. & S, R.A. 1987. Late Cretaceous vertebrate fossils from the North Slope of Alaska and implications for dinosaur ecology. Palaios, 2, 377–389. P, J.T., D, I.L., K, E.M. & S, R.A. 1998. Palaeoclimatic significance of mid-Cretaceous floras from the Middle Clarence Valley, New Zealand. Palaios, 13, 149–159. P, J.T., Z, A.M. & S, C.R. 1982. Rainfall patterns and the distribution of coals and evaporites in the Mesozoic and Cenozoic. Palaeogeography, Palaeoclimatology, Palaeoecology, 40, 67–101. R, J. & F, J. 1992. Responses of some Southern Hemisphere tree species to a prolonged dark period and their implications for high-latitude Cretaceous and Tertiary floras. Palaeogeography, Palaeoclimatology, Palaeoecology, 99, 271–290. R, H.G. 1996. Sedimentary Environments: Processes, facies and stratigraphy, 3rd Ed. Blackwell Science, Oxford. R, P.V., R, T.H., W, B.E., ME M, M.J., D, C.B., G, R.T. & F, E.A. 1988. Evidence for low temperatures and biological diversity in Cretaceous high latitudes of Australia. Science, 242, 1403–1406. S, A. & L, W. 1987. Frost survival in plants. Eco-studies 62. Springer-Verlag, Berlin, 221. S, R.L., S, J., C, K.E., U, C. & H, R. 1985. Amazon rainforest fires. Science, 227, 53–55. S, A.G., S, D.G. & F, B.M. 1994. Atlas of Mesozoic and Cenozoic coastlines. Cambridge University Press, Cambridge. S, R.A. & C, J.L. 1990. Climate change and the evolution of high-latitude terrestrial vegetation and floras. Trends in Ecology and Evolution, 5, 279–284. S, R.A. & P, J.T. 1986. Palaeobotanical evidence for cool north polar climates in middle Cretaceous (Albian-Cenomanian) time. Geology, 14, 703–706. S, R.A. & P, J.T. 1990. Latest Cretaceous woods of the central North Slope, Alaska. Palaeontology, 33, 225–242. S, E.C. & V, R.B. 1968. Preservation of coast redwood on alluvial flats. Science, 159, 157–161. S, S.P. & T, T.N. 1986. Wood decay in silicified gymnosperms from Antarctica. Botanical Gazette, 146, 116–125. T, E.L., T, T.N. & C, N.R. 1992. The present is not the key to the past: a polar forest from the Permian of Antarctica. Science, 257, 1675–1677. T, T. & B-B, Y.L. 1986. Xylotomy of fossil conifers from the Quiriquina Island, Chile. Comunicaciones, 37, 65–80. T, E.M. 1990. Cretaceous and Tertiary vegetation of Antarctica: a palynological perspective. In: T, T.N. & T, E.L. (eds) Antarctic Palaeobiology: its role in the reconstruction of Gondwana. Springer, New York, 71–88. U, C., K, J.B. & C, D.L. 1988. Fire in the Venezuelan Amazon 2: Environmental conditions necessary for forest fires in the evergreen rainforest of Venezuela. Oikis, 53, 176–184. U, J.R. & L, W. 1974. Centroclinal cross strata, a distinctive sedimentary structure. Journal of Sedimentary Petrology, 44, 1111–1113. V, P.J., S, B.W. & P, G.D. 1996. Evaluating concepts of Cretaceous equability. Palaeoclimates, 2, 139–158. V, T.T. 1982. Regeneration patterns in Araucaria araucana forests in Chile. Journal of Biogeography, 9, 11–28. W, P. 1991. Vegetation of New Zealand. Cambridge University Press, Cambridge. W, J.M. & MG, M.S. 1996. Forest disturbance in the central North Island, New Zealand, following the 1850 BP Taupo eruption. The Holocene, 6, 399–411. W, F.I. 1987. Climate and plant distribution. Cambridge University Press, Cambridge. Y, F. 1992. Effects of depth of flooding on growth and anatomy of stems and knee roots of Taxodium distichum. IAWA Bulletin, 13, 93–104. Y, F. & K, T.T. 1987. Effects of flooding, tilting of stems, and ethrel application on growth, stem anatomy and ethylene production of Pinus densiflora seedlings. Journal of Experimental Botany, 38, 293–310. Y, F., K, T.T. & W, K.E. 1987. Effect of flooding on growth, stem anatomy, and ethylene production of Pinus halepensis seedlings. Canadian Journal of Forestry Research, 17, 69–79. Y, P.J., M, J.P., S, R.R. & D, F.P. 1993. False ring formation in Baldcypress (Taxodium distichum) saplings under two flood regimes. Wetlands, 13, 293–298.

Received 28 September 2000; revised typescript accepted 10 March 2001. Scientific editing by John Howell.