Late pleistocene climate and environment from speleothems on Flores, Indonesia: vegetation, volcanoes and Homo floresiensis

Abstract

The diminutive hominin Homo floresiensis, popularly known as the "Hobbit", was discovered in Liang Bua cave on the island of Flores in 2003. Homo floresiensis was the last surviving non-human member of the Homo genus, and the cause of its disappearance sometime between 17.1 and 10.2 kyr BP is not yet fully understood. To investigate the disappearance of H. floresiensis, a detailed, high-resolution, precisely dated record of the climatic and environmental history of Flores is required. Such a record must extend back in time beyond the H. floresiensis disappearance interval (17.1-10.2 kyr BP, thousands of years before the present), to create a baseline record that is long enough to characterise the climatic and environmental forcings on Flores, and their natural variability. Located in the monsoonal rain belt of the Southern Hemisphere, the island of Flores is also in an important location climatically. A long-term terrestrial palaeoclimate record of rainfall on Flores will complement established palaeomonsoon records from the Northern Hemisphere (China) and the equatorial tropics (Borneo). This new record will provide important information about the movement of the Inter-Tropical Convergence Zone (ITCZ) and associated monsoon rainfall during the late Pleistocene. Such a record has the potential to answer key scientific questions in tropical palaeoclimatology regarding hemispheric anti-phasing of rainfall on both orbital and millennial time-scales. In this thesis, the climate and vegetation of Flores over the last 92,000 years is explored using carbon- (513C) and oxygen- (5180) isotope ratios in multiple speleothems from Liang Luar cave, just 800 m south of Liang Bua. We also use stalagmites from southern Sulawesi, ~400 km north of Flores, which span the last 530,000 years and provide regional context for environmental events on Flores. Individual records from Sulawesi allow focus on particular intervals, while broader population analyses of the basal U/Th ages of stalagmites from Sulawesi is used to uncover changes in regional climate change over the last half a million years, and reveal insights into the survival of stalagmites through time. The age distribution of basal U/Th dates for the stalagmites from Sulawesi reveals an exponential decrease in the number of older stalagmites surviving to the present day. This finding indicates that the processes that act to destroy stalagmites are approximately constant through time. Such natural processes include the downward erosion of the karst, in-cave sedimentation covering stalagmites, in-cave erosional processes and cave collapse. The result highlights that random sampling of stalagmites from a cave area will result in the unnecessary collection and wastage of stalagmite material and failure to sample the full length of the available palaeoclimate record. Deviations from this idealised exponential distribution of stalagmite ages are caused by an increased number of stalagmites during wet periods, or the absence of stalagmite growth during dry periods. Periods with a high frequency of stalagmite growth initiations, 75-70 and 10-5 kyr BP, coincide with wet periods in the Borneo speleothem 6180 record in response to regional changes in palaeoclimate. Stalagmite basal ages can therefore be used to extract firstorder climate information without removing stalagmites from a cave. On Flores, 5180 in speleothems from Liang Luar is an accurate recorder of 6180 in meteoric water, and therefore records changes in ice-volume and rainfall amount through time. After correcting for changing ice-volume, the speleothem 6180 record predominantly reflects changes in the amount of precipitation on Flores over the last 92,000 years. A 20 kyr-long cycle in speleothem 6180 on Flores is anti-phase with the Hulu- Sanbao speleothem 5180 record from China, indicating that when Flores is wet, China is dry, and vice-versa. Changing local summer insolation over each hemisphere pulls the mean position of the ITCZ northwards and southwards through time due to changes in the degree of local insolation forcing, controlled by the earth's precession. Negative excursions in speleothem 6180 (wet events) occur synchronously with all six Heinrich stadials during the last glacial period, again showing an antiphase relationship with dry excursions in China. This suggests that Northern Hemisphere controlled movements of the ITCZ are responsible for changing rainfall patterns on Flores at the millennial scale. The magnitude of the 6180 shift at the onset and end of the Heinrich stadials is modulated by the prevalent precessional conditions. Carbon isotope ratios in Liang Luar speleothems record the complex interactions of climatic, hydrological and vegetative changes. Speleothem 613C is primarily a proxy for palaeovegetation productivity and serves as a record of soil respiration, atmospheric CO2 and local precipitation. There is a first-order precipitation control on vegetation productivity. As a result, speleothem S180 and 613C typically co-vary significantly throughout the record: the correlation of 0.51 (using a 6.6 kyr low pass filter and moving window) is above the 90% significance threshold (correlation of 0.44). For over two thirds of the record (62 kyr) the correlation is above the 95% test threshold (correlation of 0.53). Geochemical idiosyncrasies of individual speleothems lead to offsets in 813C signatures among coevally growing specimens. Inter-stalagmite offsets can mostly be explained by differences in mineralogy and growth rate. Coeval stalagmites with contrasting growth-rates show different responses to changes in palaeovegetation; slower growing stalagmites show larger magnitude isotope excursions during positive 513C anomalies (e.g. at 88.4 and 14.3 kyr BP). The most notable positive 513C excursion in the Liang Luar record is the "68kyr event", which starts at 68.3 kyr BP when 613C increased rapidly by 3.8%o. 813C continued to increase more slowly until 61.5 kyr BP to reach the highest values of the last 92,000 years, and a total 513C change of +6.7%o. A coeval flowstone 513C record for Liang Luar replicates the event. The range of 513C values recorded during the 68 kyr event is consistent with a switch in vegetation to plants using the C4 photosynthetic pathway (e.g. tropical grasses). However, alimited 613C-response to the 68 kyr event is registered by a speleothem from Sulawesi, indicating that the event is likely to be centred around Flores and too small to dramatically influence the vegetation on Sulawesi. A simultaneous decrease in speleothem 6180 caused a marked divergence of the two stable isotope systems and indicates a non-precipitation control of the 68 kyr event. Simultaneous analysis of 6180, 513C and sulphur concentration in the speleothems reveals details of this large event. A combination of decreasing atmospheric pCO2 and volcanic eruptions caused (and prolonged) a switch in vegetation to C4 plants that lasted for ~7000 years. A rapid recovery from the event at 61 kyr BP suggests that threshold effects may control the vegetation. In contrast to the 613C excursions during the 68 kyr event, Heinrich stadials, and other rapid environmental changes common to the last glacial period, the H. ßoresiensis disappearance interval (HDI, 17.1-10.2 kyr BP) is a relatively benign period in the Liang Luar record. The largest increase in speleothem 813C during the deglaciation occurs at 18.1 kyr BP, before the HDI, while the HDI itself encompasses a gradual decrease in ö13C, indicating relatively stable and improving environmental conditions. The Liang Luar stalagmite 613C record shows that there was little environmental impact from known millennial scale climate events such as Heinrich Stadial 1, the B0lling-Aller0d and the Younger Dryas, or local volcanic eruptions at 17.1 and 12.2 kyr BP. The largest positive 813C excursion of the HDI (+2.7%o at 14.3 kyr BP) is shown to be a 1 in 4800 year event; excursions of a similar size occur 22 times in the last 92,000 years, as recorded by four separate speleothems. The HDI event is therefore of a size that H. ßoresiensis would have encountered often during its ~80,000-year occupation history. This suggests a limited involvement of climate or volcanism in the disappearance of H. ßoresiensis, and points to the potential involvement of Homo sapiens, which occupied Liang Bua immediately after the disappearance of H. ßoresiensis. Show more