Palaeoenvironments of insular Southeast Asia during the Last Glacial Period: a savanna corridor in Sundaland?
Introduction
The shallow epicontinental seas surrounding the islands of the Indonesian archipelago, often called the ‘Maritime Continent’, is currently a region of significance to global climate and ocean circulation. These seas comprise part of the Indo-Pacific Warm Pool (De Deckker et al., 2002) and are the warmest on earth, with temperatures averaging 28 °C or more (Yan et al., 1992). The region is a major source of latent heat to the atmosphere and acts as a major driver of both the Hadley circulation and ENSO oscillations associated with the Walker circulation. The major heat source to the atmosphere migrates seasonally from the Tibetan plateau in July, through the Sundaland region to the West Pacific in January (McBride, 1998).
Rainfall is high (generally >2000 mm) and modulated by the seasonal reversal of winds associated with the East Asian (northeast) and Australasian (southwest) monsoons. Thus the surface ocean waters in the region are not only warm, but also generally of low salinity (De Deckker et al., 2002). These warm, low-salinity waters are transferred from the South China Sea and Pacific Ocean to the Indian Ocean via the Indonesian throughflow, a number of narrow channels between the southern- and easternmost islands of Indonesia (Schneider and Barnett, 1997).
The hot and humid conditions that pertain throughout the region mean that the islands of the maritime continent were largely covered by closed lowland rainforest before the considerable deforestation that has occurred in recent times. These forests are gradually replaced to the north and south by deciduous forest types and savanna woodlands (Wikramanayake et al., 2001).
There is no other area in the tropics where the contrast between the modern distribution of land and sea with their distributions during the Last Glacial Period (LGP; Oxygen Isotope Stages 2–4), and in particular the Last Glacial Maximum (LGM; 20,000 years ago) is so marked, or where these differences could potentially have had such large impacts on global climate (De Deckker et al., 2002). Glacio-eustatic depression of sea level by ∼120 m at the LGM fully exposed the Sunda shelf joining mainland Southeast Asia to Sumatra, Java, Borneo and (possibly) Palawan. The Gulf of Thailand was exposed, as was the very broad continental shelf east of Malaysia and north of Borneo, substantially reducing the size of the South China Sea. This exposed continent has been called ‘Sundaland’ (Molengraaff, 1921).
The magnitude of the changes in palaeogeography that have occurred in the region since the LGM present a challenge to models that aim to deduce the climate of the LGM and thereby the vegetation of the LGM. The emergence of Sundaland meant that the surface area of ocean water available for evaporation in the Indo-Pacific Warm Pool was substantially reduced and the flow of water between the Pacific and Indian Oceans restricted to the deep-water channels east of Borneo and Bali.
There are currently two general alternatives for the vegetation (and therefore climate) of Sundaland at the LGM. The first has been most clearly articulated by Heaney (1991) who postulated a wide ‘savanna corridor’ extending down the Malaysian Peninsula and across the now flooded region between Borneo and Java, flanked east and west by tropical forest. Palawan and the western Philippines are also considered to have been savanna covered in this scenario. The second possibility is that a belt of tropical rain forest extended right across Sundaland from east to west, possibly diminished in north–south extent over the modern latitudinal range of tropical forest. This scenario has been advocated on the basis of some pollen records (e.g. Sun et al., 2000; Hope et al., 2004) and predicted by a range of vegetation models for the LGM (e.g. Prentice et al., 1993; Crowley and Baum, 1997; Otto et al., 2002).
Determining which of these divergent possibilities is the more correct is important for two reasons. The first of these bears upon the causes, development and maintenance of modern biogeographic patterns, and also relates to the possible routes available for early human dispersal through the region and on into Australasia during the LGP (Stringer, 2000; Barker et al., 2001; Turney et al., 2001; Bird et al., 2004).
The second reason is that the sheer size of Sundaland at the LGM (similar to Europe) means that it potentially stored a significantly greater amount of terrestrial carbon in soil and vegetation than it does today. The type and distribution of vegetation inferred to have been growing in the region at the LGM will therefore have an impact on estimates of global terrestrial storage of carbon and carbon-isotopes at the LGM (e.g. Bird et al., 1994; Otto et al., 2002).
The purpose of this paper is to review and assess the available evidence for the distribution of terrestrial environments across Sundaland through the LGP, with emphasis on the LGM, to determine whether it is possible to distinguish between the main competing scenarios described above.
Section snippets
The extent of Sundaland
The boundary of ice-age Sundaland is approximated by the −120 m isobath (Fig. 1). It is easily defined to the south and west by the deep waters of the Indian Ocean, and included what are now small island chains west of Sumatra such as the Mentawai Islands. To the east, Sundaland is separated from the biogeographically distinct region of Wallacea by deep-water channels that have ensured that no land bridge has ever existed between the two. This boundary corresponds with Huxley's Line, running
Early Sundaland
Though it is not the major focus of this paper, the Neogene geological and biogeographic evolution of the region does provide information of relevance to assessing the trajectory of climate and vegetation of Sundaland in more recent times. Detailed reviews of the subject have been published elsewhere (e.g. Sartono, 1973; Metcalfe, 1998; Hall, 1998, Hall, 2001; van den Bergh et al., 2001).
The relative areas of land and sea on Sundaland have changed dramatically during the Neogene in response to
Paleogeography of Sundaland since the last interglacial
The extent of Sundaland since the last interglacial period is tied closely to the global changes in sea-level that have occurred during this period (e.g. Hanebuth et al., 2000; Voris, 2000; Fig. 1). The relationship between relative sea-level in Sundaland and glacio-eustatic changes in ocean volume is not straightforward, because the landmass of Sundaland itself has changed its elevation in response to diachronous changes in water loading as the land was progressively flooded by sea-level rose
The climate and vegetation of Sundaland during the LGP
Evidence for the terrestrial environments likely to pertain in Sundaland over the LGP comes from a variety of sources. The available evidence for terrestrial environments is summarized in Table 1 and Fig. 2, and is discussed below:
Terrestrial environments in Sundaland and vegetation modelling
There is reasonable evidence for at least parts of Sundaland as to the general nature of the terrestrial environment during the LGP and the LGM in particular. The climate was generally 2–3 °C cooler and rainfall was reduced at least regionally. Kershaw et al. (2001) concluded that rainfall was reduced by 30–50% over much of Sundaland, though this does not provide any indication regarding the regional distribution of rainfall, and the same authors also conclude that the rainfall reduction was not
Conclusion: a savanna corridor in Sundaland
An assessment of the evidence available from geomorphology, biogeography, palynology and vegetation modelling for insular Southeast Asia over the LGP suggests that there is relatively strong and consistent evidence supporting a northward expansion of open vegetation types from southern Sundaland towards the equator during the LGM. In contrast, evidence for the nature of palaeo-environments in the core of Sundaland and areas north of the equator is sparse and conflicting. However, consideration
Acknowledgements
Colin Prentice and Louis Francois kindly provided access to the most recent output of their vegetation models for this study. Peter White provided a thoughtful review of an earlier draft of this manuscript.
References (82)
- et al.
A late Pleistocene and Holocene pollen and charcoal record from peat swamp forest, Lake Sentarum Wildlife Reserve, West Kalimantan, Indonesia
Palaeogeography, Palaeoclimatology, Palaeoecology
(2001) Dating of Malaysian fluvial tin placers
Journal of Southeast Asian Earth Sciences
(1988)- et al.
Landscape development preceding Homo erectus immigration into Central Java, Indonesia: the Sangiran Formation Lower Lahar
Palaeogeography, Palaeoclimatology, Palaeoecology
(2004) - et al.
Populating PEP II: dispersal of humans and agriculture through Austral-Asia
Quaternary International
(2004) The Asian Colobinae (Mammalia: Cercopithecidae) as indicators of quaternary climatic change
Biological Journal of the Linnean Society
(1996)- et al.
Palaeogeographical evolution of the Mahakam Delta in Kalimantan, Indonesia during the Quaternary and late Pliocene
Review of Palaeobotany and Palynology
(1988) - et al.
Pre-industrial-potential and Last Glacial Maximum global vegetation simulated with a coupled climate-biosphere model: diagnosis of bioclimatic relationships
Global and Planetary Change
(2005) - et al.
pelaeoenvironmental developments in the Lake Tondano area (N. Sulawesi, Indonesia) since 33,000 yr BP
Palaeogeography, Palaeoclimatology, and Palaeoecology
(2001) - et al.
Carbon stocks and isotopic budgets of the terrestrial biosphere at mid-Holocene and last glacial maximum times
Chemical Geology
(1999) - et al.
Biomass burning in Indonesia and Papua New Guinea: Natural and human induced fire events in the fossil record
Palaeogeography, Palaeoclimatology, Palaeoecology
(2001)