Introduction

The A-AuMS is the most typical cross-equatorial coupled monsoon system in the world. At the seasonal timescale, the summer monsoon in one hemisphere is usually linked with the winter monsoon of the other hemisphere via outflows1,2,3,4. Meanwhile, the A-AuMS passes through the Indo-Pacific Warm Pool, and can transfer the meridional (e.g., average position of the ITCZ) and zonal (e.g., Walker Circulation, ENSO activity, and even Indian Ocean Dipole) climate variations in the tropics to the extra-tropics, and hence has an important impact on global climatic changes2,3,5,6. Thus, unraveling the A-AuMS history is crucial for understanding global climate variability.

Due to the close linkage between the two hemispheres, the combination of monsoon records from both hemispheres is essential for understanding the A-AuMS dynamics. However, current knowledge is not balanced. Although a large number of high-quality EASM records have been obtained from multiple archives such as stalagmites, lacustrine sediments, and loess depositions in the past few decades7,8,9,10,11, there exist many fewer high-resolution AuSM records in the northern Australia monsoon region, which lies in the southernmost margin of the A-AuMS and could offer a unique vantage point for capturing Southern Hemisphere monsoon variability. Meanwhile, the lack of records in northern Australia also results in greater uncertainty in understanding the AuSM dynamics on multiple timescales.

On a millennial timescale, variations of AuSM during the YD and Heinrich (H) events have been relatively well studied. Some hydrological records derived from stalagmites and lacustrine sediments indicate that the monsoon precipitation increased in northern Australia during the YD and H events12,13,14,15. This interpretation is also supported by many hydrological records from the Maritime Continent area, indicating that the Southern Hemisphere Indonesian-Australian summer monsoon may have experienced systematic strengthening during the YD and H events16,17,18. The strengthening of the monsoon in the Southern Hemisphere during these millennial-scale abrupt events is consistent with the hypothesis of ITCZ southward movement caused by the heat imbalance between the Northern and Southern Hemispheres triggered by freshwater input into the North Atlantic Ocean. However, the drought conditions or a lack of significant hydrological changes during the YD period are also observed in the western Pacific region, indicating that the scope, magnitude, and spatial pattern of hydrological changes in the Asian-Australian monsoon region during the YD period still need further clarification19,20,21. Meanwhile, the millennial-centennial scale seesaw pattern between EASM and AuSM has also been detected in stalagmite records during the Holocene, and solar activity is proposed as the potential driving force22.

Compared to millennial-scale variations, the change of AuSM on the orbital scale remains more uncertain, both in terms of records and driving mechanisms. Only a few records describe the orbital-scale hydroclimate changes in the monsoon domain of northern Australia. The major one is a 45 kyr humification record from lake sediment in Lynch’s Crater, which suggests that the precipitation changes in northeastern Australia followed a semi-precessional cycle associated with long-term changes in ENSO variability, rather than with direct monsoon evolution23. As a complementary approach to enriching AuSM study, records from nearby oceans and islands have been employed to reconstruct the orbital-scale Indonesian-Australian summer monsoon changes, and are generally influenced by the changes in Southern Hemisphere insolation, sea level change, Northern Hemisphere ice sheet dynamics, and the migration of ITCZ3,20,21,24,25. However, such current recognitions remain inconsistent with significant contradictions among different studies on one hand. On the other hand, existing studies are mostly concentrated in the western Pacific region, while records from the Australian monsoon front are still scarce5,18,24,26,27.

The sub-orbital-scale monsoon variations during the Holocene are also very important for understanding the monsoon dynamics, as they represent a recent interglacial period with relatively unambiguous climate boundary conditions9,10,28. However, the changes of the Holocene AuSM remain unclear. A Holocene stalagmite oxygen isotopes (δ18O) record from northwestern Australia seems to be decoupled with the Southern Hemisphere summer insolation, which couldn’t be reconciled with the classical hypothesis of tropical insolation control deduced by comparison of Asian and South American stalagmite δ18O records6,7,10,27,29,30,31. However, the stalagmite δ18O record from western Australia is just a single record, lacking cross-validation with other records in northern Australia. There also exist some Holocene stalagmite δ18O records in the Maritime Continent area, but most of them were influenced by sea level rise during the early Holocene. In addition, stalagmite δ18O records at different locations also have obvious diversities16,18,21,24,32. Stalagmite δ18O is suggested to be influenced by circulation or source effects in recent years, and thus some studies have attempted to develop new hydrological proxies from stalagmites, such as carbon isotopes (δ13C) and trace elements33,34,35,36,37. These new proxies are related to prior calcite precipitation and have been proposed to be associated with local hydrological conditions. However, these records from different study sites in the Maritime Continent also show inconsistency with each other. Some studies suggest that these proxies have thresholds when responding to hydrological changes in this region5,33,34. Moreover, most of these records exist in Southeast Asia or the western Pacific region; similar records in northern Australia are still absent until now. In addition to stalagmite records, some marine sediment records in the western Pacific are also used to infer the Holocene Indonesian-Australian monsoon variations. However, these studies also fail to reach consistency2,38,39,40,41. Overall, the dynamics of the Holocene Australian summer monsoon currently remain unclear, probably due to the scarcity of high-resolution records from the monsoon front region of northern Australia, with current records mainly coming from the Maritime Continent area. However, due to the distance from the front edge of the AuSM, complex terrain, and the influence of multiple meridional and latitudinal climate systems, hydroclimatic records from the Maritime Continent area may be difficult to accurately capture the evolution of the Holocene AuSM. In this case, it can be of great interest to obtain new Holocene hydrological records from northern Australia beyond stalagmites to understand Holocene Australian monsoon dynamics.

In this study, a 13.5 kyr AuSM record is reconstructed using lacustrine sediments from the Bromfield Swamp (17.37°S, 145.54°E) in the northeastern Australian monsoon frontal region (Fig. 1 and Supplementary Text 1, 2)42,43. Five proxies related to hydroclimatic changes, including Rb/Sr, Al/Ca, Ti/Ca, mean grain size (MGS), and organic content (OC), are determined, and all of them show similar variations, proving that they may be simultaneously affected by monsoon precipitation (Fig. 2, Method 1 and Supplementary Text 3, 4). Therefore, principal component analysis (PCA) is performed on these five proxies, and PC1 is used in this study to reconstruct the AuSM index (AuSMI) over the last 13.5 kyr (Fig. 2e, Method 2, Fig. S2). Furthermore, the dynamics of Holocene sub-orbital-scale A-AuMS and western Pacific ITCZ are investigated through combining the paleoclimate records of the two hemispheres. The spatial pattern of A-AuMS during the YD event is also discussed.

Fig. 1: 850 hPa seasonal wind field and precipitation during 1979–2020 in the Asian-Australian monsoon region.
figure 1

The left is the average wind field and precipitation from June to August (JJA), the right is the average wind field and precipitation from December to February (DJF). Wind data comes from the National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) Reanalysis, and precipitation data from the Global Precipitation Climatology Project (GPCP). The star indicates the location of the Australian summer monsoon (AuSM) record reconstructed from the Bromfield Swamp in this study. This figure is generated using the MATLAB software.

Full size image
Fig. 2: Multiple hydrological proxies from the Bromfield Swamp.
figure 2

a Organic content (OC); b three-point smoothed mean grain size (MGS); c Rb/Sr ratio; d Ti/Ca and Al/Ca ratios; e reconstructed Australian summer monsoon index (AuSMI) through the principal component analysis on the above five proxy records. The error bar shows standard deviation (SD) values among the data of the five proxies. The black crosses indicate AMS 14C dating points. The blue bar indicates the period of the Younger Dryas (YD) event, which is determined by referring to the nearby and well-dated Liang Luar Cave stalagmite record16.

Full size image

Results and discussion

Decoupling of early Holocene AuSM and Southern Hemisphere tropical summer insolation

The reconstructed AuSM index presents a higher value during the YD event, probably indicating a wet condition and a strong monsoon. After the YD event, a gradual decline of AuSM is observed from the early to middle Holocene. The turning point occurred at 7.8 cal kyr BP, after which AuSM gradually increased and remained at the highest value in the late Holocene.

The AuSM record reconstructed in this study is similar to some other hydrological records derived from lake/marine sediments in the Indonesian-Australian monsoon region, indicating that the Indonesian-Australian summer monsoon may have undergone similar changes on a larger region2,12,20,40 (Fig. S4). However, compared with the record we derived from the Bromfield Swamp, these records may not catch all details of the monsoon changes in the YD and early Holocene, which is possibly because they are either of relatively low resolution or not located in the monsoon front.

Tropical summer insolation has been proposed to be one of the most important drivers of the global summer monsoons10. In early studies, stalagmite records from China and South America suggested that the Holocene Asian and South American summer monsoon were anti-phased, which was synchronous with summer insolation changes in both hemispheres7,10,27,29,30,31. An important implication of this viewpoint is that the tropical monsoon evolution over the Holocene was mainly controlled by low-latitude insolation (here we simply call it “tropical insolation control hypothesis”), and it has little connection with the high-latitude processes. It is worth noting that the stalagmite records used to support this viewpoint are mainly derived from Asia and South America, regions that are not included in a coupled monsoon system. In recent years, this hypothesis has been challenged. First, some studies suggest that stalagmite δ18O changes might be affected by source effect and/or large-scale circulation, which could result in biases when inferring local monsoon precipitation37,44. Second, some stalagmite records from the tropics of the western Pacific region revealed asynchronous changes in δ18O and local solar insolation6,16,45. For example, a stalagmite δ18O record from northwestern Australia doesn’t show negative correlation with the Chinese stalagmite δ18O and is not in parallel with Southern Hemisphere tropical summer insolation changes, implying that the “tropical insolation control hypothesis” may not be universally applicable6,14. However, the stalagmite record in western Australia is a single record and has not been cross-validated by other records in northern Australia, nor can it independently clarify this controversy. Therefore, Holocene monsoon changes in the Australian monsoon front are still unclear.

In this study, the reconstructed AuSM index is higher during the late Holocene than early and middle Holocene (Fig. 3a), which is consistent with higher 20°S January insolation during the late Holocene, indicating that the stronger summer insolation probably had a direct effect on the AuSM enhancement. However, AuSM did not continue to strengthen from the early to late Holocene, which did not entirely follow the summer insolation changes. In contrast, AuSM weakened gradually by about 7.8 cal kyr BP, and then rapidly strengthened from the middle to late Holocene (Fig. 3a). The mismatch between the early Holocene AuSM weakening and summer insolation increase indicates that the AuSM evolution during this period was also modulated by other factors. Some studies suggest that sea level changes also had an important influence on hydrological variations during the last deglaciation-early Holocene in the tropical western Pacific and surrounding regions. The rapid sea level rise during the early Holocene could promote local vapor supply and bring more precipitation to surrounding lands13,16,25. However, the reconstructed AuSM index in this study reflects decreased monsoon precipitation during the early Holocene in northeastern Australia, which suggests that sea level rise was not mostly responsible for the monsoon changes during this period (Fig. 3a). ENSO activities are also considered to have important impacts on the AuSM, with a significant decrease in monsoon precipitation during El Niño events in modern observation6,46. However, both paleoclimatic reconstructions and model simulations suggest that ENSO activity was weaker during the early Holocene or the middle Holocene, which cannot explain the weaker AuSM during the early to middle Holocene than during the late Holocene47,48,49. These results indicate that although the higher summer insolation could explain the strengthening of the mid-late Holocene AuSM, local insolation, as well as sea level change and ENSO activity, are difficult to reconcile with the weakening AuSM from the early to middle Holocene.

Fig. 3: Comparison of the Australian summer monsoon record with other paleoclimate records.
figure 3

a Reconstructed Australian summer monsoon index (AuSMI) in this study and the January insolation at 20°S52; b synthesized East Asian winter monsoon (EAWM) index based on the records on the Chinese Loess Plateau; c East Asian summer monsoon (EASM) records, including: pollen-based reconstructed precipitation record from Lake Gonghai, synthesized moisture index at monsoonal margin and integrated EASM index over the monsoon region9,67,68; d Southern Hemisphere (SH)-Northern Hemisphere (NH) temperature gradients. The curve covering the Holocene is the gradient between 90°−30°S and 90°−30°N, the other curve is the gradient between the whole Southern and Northern Hemispheres55,56; e Intertropical Convergence Zone (ITCZ) record derived from bulk Ti content in the Cariaco Basin69; f Holocene Northern Hemisphere ice sheet content57. The arrows on both sides of the figure indicate the intensity of AuSM, EAWM, EASM, and the magnitude of the inter-hemispheric temperature gradient, respectively.

Full size image

Impact of Northern Hemisphere high-latitude ice sheets on early Holocene AuSM changes

The A-AuMS is a cross-equatorial coupled monsoon system, and the summer monsoon in one hemisphere is usually linked with the winter monsoon of the other hemisphere via outflows at a seasonal timescale1,2,3. On orbital and millennial timescales, there have also been some studies that suggest Indonesian-Australian summer monsoon changes were influenced by Northern Hemisphere high-latitude climates through the cross-equatorial airflow, in which the Asian winter monsoon may play an important role1,2,3,50. However, the links between Holocene AuSM and the Northern Hemisphere high-latitude climates, as well as the East Asian winter monsoon (EAWM), have not been discussed in detail up to now. Here, we collect the Holocene EAWM records reconstructed by loess-paleosol sediments on the Chinese Loess Plateau (CLP) (Fig. 3b, Method 3, Fig. S5, Table S2). The composited EAWM index shows a substantial decrease in intensity during the early to middle Holocene, which is consistent with the decline of AuSM. It has been suggested that during the early Holocene, the weakened EAWM was probably controlled by the retreat of Northern Hemisphere remanent ice sheets51. Therefore, early Holocene AuSM changes may also be associated with the retreat of Northern Hemisphere ice sheets. The retreat of ice sheets during the early Holocene warmed up the high latitudes of the Northern Hemisphere, leading to the gradual weakening of EAWM51. A weakened EAWM thus reduced the cross-equatorial pressure gradient and airflow in the western Pacific, probably resulting in a gradual decrease of AuSM from the early to middle Holocene2. The Northern Hemisphere remanent ice sheets almost disappeared at around 7–8 ka, and then AuSM strengthened from the middle to late Holocene, driven by increased Southern Hemisphere summer insolation (Fig. 3). This would suggest that the evolution of Holocene AuSM was not only controlled by tropical summer insolation, but also modulated by climatic changes in the Northern Hemisphere high latitudes.

We also compare the Holocene AuSM record in this study with EASM records from the Northern Hemisphere (Fig. 3a, c, Fig. S6). Summer monsoons in the two hemispheres had an anti-phase pattern during the Holocene. EASM strengthened from the early to middle Holocene, which was opposite to AuSM variations, indicating that the retreat of high-latitude ice sheets in the Northern Hemisphere probably resulted in a northward migration of the ITCZ, and thus an enhanced EASM and a weakened AuSM (Fig. 3, Fig. S6). From the middle to late Holocene, after the influence of ice sheets abated, EASM became weakened as driven by the decreased Northern Hemisphere summer insolation, while AuSM strengthened in response to increased summer insolation in the Southern Hemisphere52.

Modulation of inter-hemispheric temperature gradients on Holocene A-AuMS and western Pacific ITCZ

The A-AuMS evolution is usually coupled with the position of the western Pacific ITCZ. Summer monsoons in the two hemispheres usually present opposite variations in response to the ITCZ north-south migration53,54. In this study, we find that AuSM and EASM showed opposite changes during the Holocene, indicating that changes in the ITCZ mean position played an important role in modulating Holocene A-AuMS. The mean position of the ITCZ is essentially determined by heat distribution between the two hemispheres53,54. Therefore, in this study, we use the temperature gradient between the Southern and Northern Hemispheres to indicate the relative mean position of the ITCZ from the perspective of energy balance55,56 (Fig. 3d, Method 4).

The south-north temperature gradient decreased from the early to middle Holocene and then increased from the middle to late Holocene, with the smallest value in the middle Holocene (Fig. 3d). The temperature gradient record is positively correlated with AuSM and is negatively correlated with EASM, indicating that the south-north temperature gradient, which involved various climate forces, such as solar insolation and Northern Hemisphere ice sheets, probably directly controlled the western Pacific ITCZ position and thus the A-AuMS evolution during the Holocene (Fig. 3, Fig. S7, Supplementary Text 5). This interpretation does not contradict the above discussion. For example, from the early to middle Holocene, the retreat of Northern Hemisphere ice sheets may have contributed to the gradual warming of the Northern Hemisphere and consequent decrease of south-north temperature gradient, driving the mean position of the ITCZ northward, strengthening EASM and weakening AuSM1,2,3,57. From the middle to late Holocene, an increased temperature gradient may be associated with the reverse variation of summer insolation between the Southern and Northern Hemispheres caused by Earth’s precessional changes, resulting in a southward migration of the ITCZ, as well as strengthening of AuSM and weakening of EASM. This “thermal control hypothesis” is quite different from the previous “tropical insolation control hypothesis”, with the former indicating that, in addition to low-latitude processes, climate conditions in Northern Hemisphere high latitudes could also modulate Holocene AuSM changes.

Spatial pattern of A-AuMS during the YD event

Beyond the Holocene, strengthened AuSM with relatively humid conditions in northeastern Australia during the YD event is also observed in this study, which is consistent with some hydrological records derived from stalagmites and lacustrine sediments in northern Australia13,14,15. (Fig. 3a). In contrast, most hydrological records from the Asian monsoon region show that the climate was relatively dry and the EASM was weakened during the YD event7,9,11 (Fig. 3c).

In order to better understand the detailed spatial pattern of hydroclimatic conditions in the Asian-Australian monsoon region during the YD event, we integrate hydrological records in the EASM and AuSM regions, and define the hydrological index (HI) to reflect monsoon conditions during the YD (Fig. 4, Method 5, Fig. S8, Table S3). Our integration shows that a general dry condition was presented in the EASM region, while the AuSM region was generally wet. In the western Pacific region, the transition zone between the Asian and Australian monsoon, the hydrological conditions exhibited diversity. The northwestern part of the western Pacific was mainly characterized by drought, while the southeast was mostly characterized by no obvious changes. The diversity indicates that although the ITCZ experienced a southward migration during the YD event, there may be regional differences in hydrological response in the western Pacific region due to its location in a transitional zone and the complex terrain. In addition, most of the records in the western Pacific region are derived from marine sediments, and the larger thermal capacity of the western Pacific may also result in muted responses for the millennial-scale abrupt events58.

Fig. 4: Distribution of hydrological records in the Asian-Australian monsoon region during the Younger Dryas event.
figure 4

The base map is generated using the MATLAB software, which shows the annual mean precipitation from 1979 to 2020. The precipitation data comes from the Global Precipitation Climatology Project (GPCP).

Full size image

It has been proposed that during the YD event, a large volume of freshwater was injected into the North Atlantic Ocean, which suppressed the thermohaline circulation and caused southward migration of the ITCZ. It is also consistent with the changes in inter-hemispheric temperature gradient, which also conform to the “thermal control hypothesis” (Fig. 3). The weakened Atlantic Meridional Overturning Circulation (AMOC) during the YD event led to cooling of the Northern Hemisphere, which resulted in a greater thermal imbalance between the two hemispheres. This ultimately led to the southward shift of the ITCZ and resulted in weakened EASM and strengthened AuSM59,60. It is worth noting that some other millennial-scale climate fluctuations before the YD event and during the Holocene are also observed in our AuSM record. In fact, the instability of the climate during the last deglaciation and Holocene has been found in many records of high latitudes and tropics in the Northern Hemisphere7,61,62,63. However, the presentation of these events in the Australian monsoon region has not been fully clarified, which also deserves study subsequently.

In summary, a new high-resolution Holocene hydroclimatic record obtained from the northern Australian monsoon region in this study provides important evidence for investigating A-AuMS dynamics over the past 13.5 kyr. Our results suggest that AuSM in the southern tropics was not only controlled by Southern Hemisphere tropical summer insolation, but also modulated by Northern Hemisphere high-latitude climatic changes through cross-equatorial airflow, which may have occurred during both the YD event and the early Holocene. Further, we find that Holocene A-AuMS dynamics could be unified from the perspective of energy distribution. The inter-hemispheric energy imbalance was probably the direct force driving the western Pacific ITCZ position and A-AuMS. The thermal imbalance was mainly related to the retreat of the Northern Hemisphere remanent ice sheets (early Holocene) and summer insolation imbalance between the two hemispheres (mid-late Holocene). During the YD event, the thermal imbalance was mainly caused by the freshwater injection into the North Atlantic Ocean and subsequent AMOC weakening.

Methods

Proxy analysis

Elemental analysis

254 evenly spaced samples were selected for elemental analysis. After drying at 45 °C, they were ground in the glass mortar and sieved through a 150 μm mesh. About 0.1 g of each sample was digested in HF-HNO3-HClO4, then measured using the Inductively Coupled Plasma-Atomic Emission Spectrometry (ICP-AES). The detection limits of the elements used in this study were as follows: Al (10 mg/kg), Ti (0.5 mg/kg), Ca (2 mg/kg), Rb (0.005 mg/kg), and Sr (0.1 mg/kg). These analyses were conducted at the Lake Sediment and Environment Laboratory, Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences.

Grain size analysis

117 samples were used for grain size analysis. For each 0.5 g sample, 10 mL 10% H2O2 was added with gentle heating to remove organic content. Then 10 mL 30% HCl was added to remove carbonates. When the reaction was complete, samples were placed in distilled water and left to stand for 24 h. The distilled water was then removed, and 10 ml 10% (NaPO3)6 liquid was added. The samples were then shaken for 10 min using an ultrasonic cleaning machine to form a highly dispersed particle suspension. The analysis was conducted on a Mastersizer 3000 Laser Particle Size Analyzer at the Sedimentology Laboratory, Institute of Earth Environment, Chinese Academy of Sciences (IEECAS). In this study, the MGS data was smoothed using a three-point moving average.

Organic content analysis

The organic content was estimated using a loss on ignition analysis (LOI). 128 samples were used for the analysis. The samples were weighed to 0.35–0.4 g each and placed in a muffle furnace set at 550 °C for 4 h. When the ashing was completed and the samples had cooled to room temperature, they were reweighed to calculate the mass loss. This analysis was carried out in the Lake Sediment Laboratory, IEECAS.

Reconstruction of the Australian summer monsoon index

The AuSM index was reconstructed using a principal component analysis64. Here, five proxies, Rb/Sr, Ti/Ca, Al/Ca, MGS, and OC, were input into the PCA. The data was first standardized through the Z-score method (see Eq. (1), where Z is the normalized value, X is the original value, AVG and SD are the average value and standard deviation of each time series). All the data was then interpolated to a 50-year time window for the principal component analysis.

$${{\rm{Z}}}=({{\rm{X}}}-{{\rm{AVE}}})/{{\rm{SD}}}$$
(1)

Five main components were identified, with the variances of 82.64% (PC1), 9.14% (PC2), 5.44% (PC3), 2.08% (PC4), and 0.70% (PC5), respectively (Fig. S2). The PC1 was dominated by high positive loadings of all proxies with the highest variances. We interpreted it to reflect AuSM changes and used it as the Australian summer monsoon index (AuSMI). The SD values of the five proxies were also performed as errors (Fig. 2e).

Although our OC and grain size records share an overall structure of changes similar to the elemental records, they differ slightly in detail. In order to test the robustness of the PC1 signal, we repeated PCA analysis by (1) removing OC, (2) removing grain size, and (3) using only one composited elemental record. PC1 in these experiments is essentially the same as before (the variance percentage of these three tests is 85.09%, 87.49% and 83.28%, respectively), indicating the robustness of the PC1 signal (Fig. S3).

Reconstruction of the EAWM index

In this study, we collected published EAWM records over the past 14 kyr. Among 20 collected records, 8 were loess or eolian records from 10 sections on the Chinese Loess Plateau (CLP), and 12 were other records from 11 sites on land or ocean (Table S2). These records were interpolated to 100 yr/point and were standardized using the Z-score method. Considering that loess and eolian sediments on the CLP mainly come from dust deposition of the winter monsoon and are a direct indicator of EAWM, we separately obtained the composite data of the records on the CLP and other regions by calculating their mean values65,66. The two composite data sets show similar changes. However, it could be observed that the variations among the records from the CLP were relatively consistent. The records from other regions showed more variability in comparison (Fig. S5). Thus, in this study, we used the composite record from the CLP to reflect EAWM changes.

Temperature gradient calculation

Two sets of cross-hemisphere temperature gradient data were used in this study. One was the Southern-Northern Hemisphere temperature gradient during the past 11 kyr, which was calculated using the stacked annual mean temperature of 90°–30°S and 90°–30°N56. The other was the Southern-Northern Hemisphere temperature gradient from the last deglaciation to the middle Holocene, which was calculated using the stacked annual mean temperature records of the whole Southern and Northern Hemisphere55.

Reconstruction of the hydrological spatial pattern during the YD event

To better understand the spatial pattern of hydrological changes in the Asian-Australian monsoon region during the YD event, we collected records from the A-AuMS region (Fig. 4, Table S3). Records with resolution better than 100 yr/point were selected. A total of 36 records met this criterion. The data was interpolated to 50 yr/point. We chose 12.3−11.7 kyr as the main period of the YD event, and selected two adjacent periods, 11.3−10.7 kyr and 13.3−12.7 kyr, for comparison. We defined the relative hydrological index (HI) of the YD compared to the two adjacent periods, in which AVE and SD mean the average value and standard deviation of corresponding periods, respectively:

$${{\rm{HI}}}1=({{{\rm{AVE}}}}_{12.3-11.7{{\rm{kyr}}}}-{{{\rm{AVE}}}}_{11.3-10.7{{\rm{kyr}}}})/{{{\rm{SD}}}}_{12.3-11.7,11.3-10.7{{\rm{kyr}}}}*100$$
(2)
$${{\rm{HI}}}2=({{{\rm{AVE}}}}_{12.3-11.7{{\rm{kyr}}}}-{{{\rm{AVE}}}}_{13.3-12.7{{\rm{kyr}}}})/{{{\rm{SD}}}}_{12.3-11.7,13.3-12.7{{\rm{kyr}}}}*100$$
(3)

When both HI1 and HI2 values were positive, the hydrological conditions of the YD were considered as relatively “wet”. When both HI1 and HI2 were negative, the hydrological conditions of YD were considered as relatively “dry”. When they were one positive, and the other negative, the hydrological conditions of YD were considered as “no obvious change”.