A quantitative Late Quaternary temperature reconstruction from western Tasmania, Australia

Abstract

Late Quaternary temperature estimates from the mid latitudes of the Australian region suggest a breakdown in the tight coupling observed between oceanic and atmospheric temperatures over the recent past that has significant implications for our understanding of the response of the Earth’s climate system to global climate change and orbital forcing. Here, we present a pollen-based quantitative temperature reconstruction from the mid latitudes of Australia that spans the last 135 000 years, enabling us to address this critical issue. Gradient analysis of a pollen dataset inclusive of over 1100 Quaternary and modern pollen spectra demonstrates the dominant influence of temperature over Quaternary pollen composition and vegetation change in western Tasmania, Australia. We develop and apply a transfer function for average annual temperature that performs excellently under cross-validation (r² = 0.76; RMSEP 1 °C), is not influenced by spatial autocorrelation and that reveals a remarkably close correlation between oceanic and atmospheric temperature change over the last 135 000 years. Significantly, we report a substantially lower degree of cooling during the LGM/MIS 2 (3.7–4.2 °C below present) than previously estimated; a similar degree of cooling during MIS 4 as the LGM (ca 4 °C); and a 1 °C warming during the Last Interglacial relative to today. We conclude that atmospheric and oceanic temperature changes in this region have remained coupled throughout the substantial climatic shifts associated with glacial–interglacial cycles over the last 135 000 years. Western Tasmanian pollen records have great potential as a Southern Hemisphere terrestrial palaeothermometer and are critically located to provide significant input in to debates over the occurrence and influence of regional and global climatic episodes in the Southern Hemisphere.

Introduction

The development of quantitative palaeoenvironmental reconstructions is a crucial step in attempting to understand past, present and future Earth system dynamics, as they permit a unique opportunity for a quantitative comparison between often disparate and complex palaeoenvironmental proxies. There is a general lack of land based continuous and quantitative temperature reconstructions from the Southern Hemisphere that span the full period between the Last Interglacial to the present that precludes a direct comparison between terrestrial temperature change and temperature change in the oceans and polar regions through this time. Advances in mathematical modelling and computational power have enabled a proliferation of quantitative niche-based palaeoenvironmental reconstructions and some of these are now being applied to Southern Hemisphere Late Quaternary palaeoenvironmental data (e.g. Marra et al., 2004, Barrows et al., 2007a, Wilmshurst et al., 2007, Woodward and Shulmeister, 2007, Massaferro et al., 2009, Rees and Cwynar, 2009, Tonello et al., 2009), often without regard to important philosophical caveats, not the least of which include critical assumptions about the representativeness of modern bioclimatic envelopes for past species/taxa distributions and the degree of spatial autocorrelation in species datasets (Birks, 1998, Telford and Birks, 2005, Telford and Birks, 2009, Belyea, 2007). In an attempt to address the gap between the philosophical underpinnings and the application of the niche-based approach, we (i) interrogate the relationship between pollen composition and temperature through the Quaternary in western Tasmania, Australia, using a novel application of gradient analysis, (ii) develop and test a pollen-based transfer function for temperature for the effects of spatial autocorrelation and (iii) finally apply the model to continuous Late Quaternary pollen records from this part of the mid latitudes of the Southern Hemisphere.

Millennial scale changes in solar insolation resulting from variations in the Earth’s solar orbit (sensu Milankovitch, 1941, Berger, 1978) are considered the main driver of atmospheric and oceanic temperature change over glacial–interglacial time-scales and there is a close correlation between insolation, terrestrial vegetation and sea surface temperature (SST) change in the mid latitudes of the Southern Hemisphere during the Late Quaternary (Vandergoes et al., 2005). Attempts at quantitatively reconstructing temperature change in western Tasmania, Australia, a mountainous mid-latitude region in the Southern Hemisphere with a pronounced maritime climate, have revealed a significant correlation between decadal-scale changes in warm-season temperature on land and in adjacent Indian Ocean SST over the most recent millennia (Cook et al., 2000) that contrasts with a marked divergence between Indian Ocean SST and terrestrial temperature estimates during the Late Quaternary (Colhoun, 1985b, Colhoun, 2000, Colhoun et al., 1999, Barrows and Juggins, 2005, Mackintosh et al., 2006, Barrows et al., 2007a, Williams et al., 2009). A breakdown in the tight coupling between oceanic and atmospheric temperatures through the Late Quaternary, if real, has significant implications for our understanding of the response of the Earth’s climate system to global climate change and orbital forcing.

Estimates of terrestrial temperature change during the Last Glacial Stage (Marine Oxygen Isotope Stage 2 – MIS 2) in western Tasmania range considerably in magnitude and vary in regards to what component of the temperature climate is being inferred: estimates based on western Tasmanian pollen records, probably reflecting the influence of temperature minima (cold-season) on vegetation (Kirkpatrick and Brown, 1987, Read and Hill, 1989, Read and Busby, 1990), report a 5–6 °C cooling during MIS 2 (Colhoun, 1985b, Colhoun and van der Geer, 1986, Colhoun et al., 1999); estimates based on snowline depression, possibly reflecting warm-season temperatures (Seltzer, 1994), depict a 6.5 °C cooling (Colhoun, 1985a); while estimates based on equilibrium line altitude estimates of glacial limits probably reflect warm-season freezing temperatures (Ohmura et al., 1992) and are between 7 and 8 °C cooler than present for MIS 2 (Mackintosh et al., 2006). Niche-based quantitative estimates of Indian and Southern Ocean MIS 2 average annual SST, conversely, report a substantially lower degree of cooling (3–4 °C) of the ocean’s surface in the mid-latitudes relative to modern values (Barrows and Juggins, 2005) that is supported by recent beetle-based terrestrial temperature reconstructions from the maritime mid latitudes of the west coast of South Island New Zealand (Marra et al., 2004). There is a clear discrepancy, then, between estimates of MIS 2 SST and terrestrial temperature change in western Tasmania that implies either a substantial weakening of the oceanic influence on western Tasmanian climate during the Last Glacial Stage or inaccurate estimations of temperature change. In this paper we present a niche-based average annual temperature (AAT) reconstruction for western Tasmania that is derived from modern pollen – temperature relationships and that spans the entire period between the Last Interglacial to the present (ca 135 000 years (ka)), enabling a direct comparison of atmospheric and oceanic temperature change in this region.

Tasmania is a mountainous island with an oceanic climate located between 41 and 43°S that shares remarkable climatic, geographic and floral affinities with the mid latitudes of South Island New Zealand and southern South America (Fig. 1). Orographic uplift of the moisture laden Southern Hemisphere Westerly Winds (SWW) as they are advected over the north–south trending central ranges that bisect these landmasses results in distinct west (superhumid) and east (subhumid–semiarid) climatic and biogeographic zonation (Gentilli, 1972, Sturman and Tapper, 2006, Garreaud et al., 2009). The continental shelf to the west of each of these regions is narrow, with less than 20 km exposed west of Tasmania during the MIS 2 glacial stage (Lambeck and Chappell, 2001). Unlike New Zealand and southern South America, the present day landscape of western Tasmania is free from glacial ice, yet the steep glacially moulded topography reveals a dynamic Tertiary and Quaternary glacial history (Colhoun, 2004).

The complex landscape of western Tasmania has allowed the in situ persistence of a relictual mesophytic Gondwanan flora throughout the climatic vicissitudes of the Tertiary and Quaternary (Hill, 2004). The distribution of modern vegetation types is governed primarily by temperature and its relation to altitude, with evidence that temperature minima exert a substantial influence over plant species distributions (Kirkpatrick and Brown, 1987, Read and Hill, 1989, Read and Busby, 1990). The relationship between temperature and vegetation is faithfully reflected in the modern pollen rain, with average annual temperature (AAT) significantly correlated to modern pollen composition (Fletcher and Thomas, 2007a). There is a close match between changes in western Tasmanian Late Quaternary pollen records and regional temperature reconstructions from Antarctica and the surrounding oceans (Colhoun and van der Geer, 1986, Colhoun and van der Geer, 1998, Colhoun et al., 1999, Colhoun, 2000) that, coupled with the close relationship between the modern pollen rain and temperature, indicates this region is ideal for using a modern pollen training set to derive quantitative estimates of temperature change from Late Quaternary pollen records using niche-based quantitative approaches.

In this paper, we use gradient analysis of a western Tasmanian modern and Quaternary fossil pollen dataset to asses the relationship between Quaternary pollen compositions and modern pollen–climate relationships. We then develop a pollen-based transfer function for AAT and apply it to two continuous Late Quaternary pollen records from this region. We aim to specifically address the following questions: (i) has temperature been the main determinant of pollen composition (read: vegetation) through the Quaternary; (ii) can modern pollen–temperature relationships be used to generate accurate temperature estimates in this region; (iii) what was the nature and magnitude terrestrial temperature change through the Late Quaternary in western Tasmania; and (iv) was there a divergence between oceanic and terrestrial temperatures during the Late Quaternary.