Pandanus nutshell generates a palaeoprecipitation record for human occupation at Madjedbebe, northern Australia

Little is known about the Pleistocene climatic context of northern Australia at the time of early human settlement. Here we generate a palaeoprecipitation proxy using stable carbon isotope analysis of modern and archaeological pandanus nutshell from Madjedbebe, Australia’s oldest known archaeological site. We document fluctuations in precipitation over the last 65,000 years and identify periods of lower precipitation during the penultimate and last glacial stages, Marine Isotope Stages 4 and 2. However, the lowest effective annual precipitation is recorded at the present time. Periods of lower precipitation, including the earliest phase of occupation, correspond with peaks in exotic stone raw materials and artefact discard at the site. This pattern is interpreted as suggesting increased group mobility and intensified use of the region during drier periods.


SI Section 5: Isotope standards
The isotope samples reported on in this paper were analysed in 5 separate batches and 37 runs. The below tables provide the Isotope standards for each of these batches.

SI Section 8: Comparison of the Madjedbebe pandanus palaeoprecipitation proxy to other Australian environmental records
Modern moisture in northern Australia is closely related to the strength of the summer monsoon. This pattern is maintained through the whole of the last glacial cycle, with the monsoon waxing and waning in tandem with precessional forcing 14 .
Phase 2: 65-52.7 kya. The highest lake levels in Lake Eyre in the last glacial cycle occurred between 65 and 60 kya 15 . Since Lake Eyre is mostly fed by flows from tropical northern Australia this suggests an effective northern Australian monsoon at this time. However, this does not correspond to a period of strong summer insolation in northern Australia and more recent work on mega-lake Frome 16 now indicates that westerly derived (temperate) moisture may have reached Lake Eyre at this time.
There is little direct evidence for rainfall in northern Australia at this time, but the suggestion is that the period is relatively dry.
Phase 3: 51.6-28.1 kya. The period between 50-40 kya (Phase 3a) is represented by strong river flows from the north into the Lake Eyre Basin 17 and along the tropical Queensland coast 18 . Both these records suggest that the tropical monsoon is strong at this time. 40 kya also represents a high stand at Lake Eyre 15 and in this case there is no suggestion of temperate waters from Lake Frome reaching Lake Eyre. Phase 3a appears to be relatively moist in the north. An interstadial moderately high sea-level occurred at c. 50 kya and then sea-levels declined. This would have favoured higher effective moisture during the 50-40 kya period with a declining trend thereafter. The evidence relating to moisture availability between 40 and 30kya is limited but at least some records indicate that the monsoon was active 19 .
Phase 4: 26.7-13.2 kya. Orbital conditions favour a strong monsoon at this time but sea-levels at -125 m mean that the coastline is far removed across the northern Australia region, with a brackish lake occupying part of the Gulf of Carpentaria and the shoreline extended hundreds of kilometres to the NW in the Arafura Sea region. Long assumed to be a period of great aridity, recent climate modelling results 20 and speleothem data 21,22 suggest that the northern Australian monsoon was active during the Last Glacial Maximum, 24-20 kya. Recent work from Girraween Lagoon near Darwin indicates that vegetation transitioned to a more open grassy savannah during the Last Glacial Maximum and this was interpreted as reflecting 700-1000 mm of precipitation 23 . These authors interpret the monsoon as active at this time and attribute the changes to increased distance to the coast and reduced CO2. River flows in the Australian tropics increased after 30 kya and declined after the LGM between 17 and 10 kya 24 . Waterfall plunge pools may have been active during the Last Glacial Maximum 25 , although some of this data has been reinterpreted to be younger based on newer OSL ages 26 . Despite all this evidence for an active monsoon, because of the expanded Australian landmass, precipitation gradients were based on a coastline hundreds of kilometres seaward of the present position. This means that at many sites, precipitation was lower than during the Holocene during the Last Glacial Maximum period.
Phase 5: 9.7-8 kya. The modern Northern Australian monsoon is inferred to have restarted at about 14 kya 27,28 . Across most of Australia the early to mid-Holocene had higher effective precipitation than the present day with several records indicating a maximum between c. 10 kya and 6 kya 29,30 . This pattern is confirmed from the Kimberley 27 .
Phase 7: 602 cal BP to present. Overall, the late Holocene in northern Australia is a period of less reliable rainfall and a weaker monsoon than the preceding mid-to early Holocene 29 . The direst phase in the Kimberley precedes Phase 7 at between 2.6 and 1.3 kya 27 . The last few thousand years are not monotonically dry, however. For example, increased fluvial activity in arid central Australia has been recognised to co-eval with the early part of the medieval climatic anomaly in Europe about 900 ya 31 . Records are quite confusing. Yan et al. 20 reviewed monsoon paleo-records from northern Australia and suggested that conditions were arid during the 'Little Ice Age' (defined by them as 1400-1850 AD) and related this to a withdrawal of the Intertropical Convergence northward at this time. This is roughly co-eval with a dry phase interpreted from the Kimberley in NW Australia 27 . In contrast, Rouillard et al. 32 recorded evidence for extreme floods in NW Australia during the same period. These records are not incompatible with modern rainfall at high levels in NW Australia due to favourable Indian Ocean Dipole conditions, while NE Australia has suffered repeated poor rainfall years in the last few decades due to ENSO and Pacific Decadal Oscillation Effects. One record appears to align well with our findings, Hendy et al. 33 noted a sharp decrease in salinity at the end of the Little Ice Age in waters off the barrier reef. This is consistent with an uptick in the monsoon roughly coincident with our Phase 7b.
Our overall conclusion is that, given the relatively poor understanding of long-term moisture balances in tropical Australia, the δ13C record presented here is consistent with the current observations and provides a powerful new tool for palaeoprecipitation reconstruction.