The reconstruction of moisture availability in south-eastern Australia during the Holocene

Due to the temperate and relative dry climate of south-eastern Australia, there are relatively few palaeo-environmental records in the region, and moisture-sensitive proxies are particularly rare. Any new and independent palaeo-hydrological reconstruction that offers insight to the climate system in south-eastern Australia is hence a novel contribution. Reconstruction of the palaeo-climatic conditions of south-eastern Australia is likely, however, to be not only of regional importance, but potentially significant to much wider spatial scales, because the region is influenced by several major ocean-atmospheric modes. This includes El Niño–Southern Oscillation (ENSO) and the Inter-decadal Pacific Oscillation (IPO), modes that impact both hemispheres and may be highly sensitive to future climate change. This thesis represents the first attempt to apply testate amoebae analysis to the quantitative reconstruction of moisture availability in south-eastern Australia over the Holocene. Firstly, this thesis proposed a revised laboratory method, incorporating an organic co-solvent (acetone) and a sodium-based dispersant (sodium pyrophosphate), to address issues with the identification and quantification of testate amoebae in minerogenic sediments. This revised method effectively concentrated testate amoebae and significantly improved the clarity of slides, thereby greatly reducing counting time. These methods made the counting of testate amoebae feasible, and allow for the use of testate amoebae as a (moisture-sensitive) proxy in south-eastern Australia. The methods are also applicable for the concentration of testate amoeba in other more minerogenic sediments. This thesis then considered if there was a strong ecological relationship between the community composition of testate amoebae and water-table depth (WTD). A strong, significant relationship between WTD and testate amoebae in south-eastern Australia was supported and this allowed further development of robust and reliable transfer functions. This relationship was tested using several analyses, including canonical correspondence analysis (CCA), stepwise selection, variation partitioning and the λ1/λ2 ratio (the variance explained by the first constrained to first unconstrained axis in an CCA with the environmental variable of interest). Transfer functions relating testate amoeba community composition and WTD were then developed and thoroughly tested using various statistical perspectives. Transfer functions were built using all species (minus rare species, all-species models) and with only informative taxa (species-pruned models). The performance of these transfer functions was tested using leave-one-out, traditional bootstrap, cluster-bootstrap and leave-one-site-out cross-validation, segment-wise analysis (un-even sampling bias) and spatial auto-correlation. Both the all-species and species-pruned transfer functions had statistically sound performance, with R2 values of around 0.8 and RMSEP values of ~ 7 cm (with the range of WTD: 60 cm). The all-species modern analogue technique (MAT), was found to have negligible bias due to un-even sampling and spatial autocorrelation, and this was recommended as the best transfer function for future reconstruction of water-table from testate amoebae in south-eastern Australia. The species-pruned transfer functions proved to be a useful way to consider the importance of taxa and verify the performance of different types of transfer functions. Lastly, two reconstructions of moisture availability were derived at Snowy Flat, a high-altitude Sphagnum-shrub bog. One was a reconstruction calculated using the derived transfer functions between (sub)fossil testate amoebae and WTD, covering the last ~3100 cal. yr BP. The other was derived from the principal component analysis (PCA) of geochemical elements measured by XRF, and this extended to 9600 cal. yr BP at a much higher temporal resolution than the first. The reconstruction of moisture availability from geochemical elements was supported by a PCA biplot and similar zonation and synchronous changes to the reconstructed WTD from testate amoebae during the period of their overlap. In addition, loss-on-ignition and charcoal were analyzed to better characterise past environmental conditions and/or change. It should be emphasised that uncertainties with the depth-age model at the Snowy Flat bog meant that the Holocene chronology is best constrained in the period prior to 7600 cal. yr BP and after 3600 cal. yr BP: the veracity of the chronology in the intervening period needs confirmation, and a hiatus in sediment accumulation at the coring site during this period is plausible but not certain. The reconstruction of moisture availability suggested several distinct periods during the Holocene in south-eastern Australia. Before 7600 cal. yr BP (early Holocene), south-eastern Australia saw enhanced moisture availability, followed by a decrease that remained steady afterwards until 3600 cal. yr BP (middle Holocene). From 3600 to 1800 cal. yr BP (late Holocene), the variability of moisture availability increased, and this was particularly apparent after 1800 cal. yr BP (recent past). Periods of enhanced moisture availability were identified from 3500-3300, 2400-2200, 1350-1100, 920-860 and 670 to 60 cal. yr BP. After 1800 cal. yr BP, moisture availability in south-eastern Australia was reminiscent of trends in the Northern Hemisphere, that is, moisture availability followed the trends of the ‘Roman Warm Period’, the ‘Dark Ages Cold Period’, the ‘Medieval Climatic Anomaly’ and ‘Little Ice Age’. As an example, south-eastern Australia saw enhanced moisture availability in the Little Ice Age and reduced moisture in the Medieval Climate Anomaly, with the latter interspersed by a period of higher moisture availability at 920-860 cal. yr BP. This research demonstrated that there was a significant negative relationship between the IPO and rainfall at a decadal scale during the instrumental records. Coupled with an IPO-like tripole pattern between rainfall and sea surface temperatures in the Pacific Ocean, this thesis interpreted enhanced moisture availability in the region resulted from a negative IPO. Wavelet analysis indicated significant 10-64 year cycles in the reconstruction of moisture availability after 3600 cal. yr BP when the chronology of the sediments was more secure. These cycles conform to the periodicity of the IPO. The reconstruction suggested that the modern IPO became apparent in south-eastern Australia at 3600 cal. yr BP and intensified after 1800 cal. yr BP. The reconstruction also implied that a positive IPO (El Niño-like) mean state prevailed in the MCA and a negative IPO (La Niña-like) mean state dominated the LIA. The enhanced moisture availability (negative IPO) state after 3.6 ka was apparently closely related to low total solar irradiance, based on a 10Be record in polar ice. As an example, periods of enhanced moisture availability (mean negative IPO states) around 920-860 and 670-60 ca. yr BP coincided with Oort Minima (910-870 BP) and a cluster of grand solar minimum, respectively. And the period of reduced moisture availability from 850-700 cal. yr BP coincided with Medieval Maximum. The results from this thesis support the ‘atmosphere hydrological cycle’ rather than a ‘ocean dynamical thermostat’ mechanism at decadal to centennial scales