Including the top and bottom ages (14.5 kyr BP at 7 mm and 33.3 kyr BP at 155 mm) reported by Chen et al. (2020), 18 U-Th high-precision ages were obtained (Table 1). The uranium concentration of the DDH-B15 stalagmite ranges from 757.4 to 1,245.8 parts per billion (ppb), with an average of 1,007.8 ppb, which is a relatively high uranium content for calcite stalagmites. The content of Th depends mainly on the ²³²Th content. The range of ²³²Th is 74-3917 parts per trillion (ppt), and the average value is 514ppt. The range of ²³⁰Th/²³²Th is 3302-258162 (atomic ratio), and its average value is 111828. Compared with the initial ²³⁰Th/²³²Th value of 4.4 ± 2.2 (atomic ratio), the correction error caused by the initial thorium does not exceed 1‰, which can be ignored. δ²³⁴U_Initial varies between 4,434‰ and 4,744‰, with an average value of 4,575‰. The obtained 18 U-Th ages are between 33.4 kyr BP and 13.4 kyr BP, with errors between 26 and 142 years, and the average error being 61 years (the average relative error is 2.76‰). All ages are in stratigraphic order, and linear interpolation was used to establish a chronological model. The age model results show that the growth period of the stalagmite is from 33.97 to 13.12 kyr BP (Figure 2B). The growth rate of DDH-B15 is relatively stable (2.3–14.5 mm/kyr), and the average growth rate is 7.89 mm/kyr. The stalagmite grew fast during 34–29 kyr BP and 22–16 kyr BP but slowly in other periods (Figure 2B).

Using the chronological scale established earlier, a δ¹⁸O time series was constructed for the DDH-B15 stalagmite (Figure 3). The average temporal resolution of δ¹⁸O series is 64.8 years during 33.97–13.12 kyr BP, which can be used to study climate variability on centennial–millennial–suborbital timescales. DDH-B15 δ¹⁸O values vary from −11‰ to −5.6‰, with an overall increasing trend from 34 kyr BP to 16 kyr BP, and rapidly shift to negative values after 15 kyr BP. The entire δ¹⁸O sequence is interrupted by three significant positive events, roughly at 16 kyr BP, 24 kyr BP, and 29.5 kyr BP. In addition, among several significant positive events, the δ¹⁸O values of 34–30 kyr BP, 29–26 kyr BP, and 21–17 kyr BP were relatively negative. Among them, the δ¹⁸O values during the 34–30 kyr BP period is the most negative, which are ∼5‰ more negative than the most positive value observed at 16 kyr BP. Furthermore, during 21–17 kyr BP, δ¹⁸O series values grasp an obvious negative event, which is also reflected in records from other stalagmites in the ASM region (Wang et al., 2001b; Cai et al., 2010; Jiang et al., 2014).

DDH-B15 δ¹³C values fluctuate between −7.73‰ and −1.16‰. While we observe a significant long-term increasing trend of δ¹⁸O, the δ¹³C values show only a slight positive trend (Figure 3) on the orbital timescale. On the millennial timescale, δ¹³C and δ¹⁸O show similar characteristics at some excursions of enriched isotope events but also with some differences. For example, during 21–17 kyr BP, δ¹³C does not have a significant depleted value as seen in δ¹⁸O; while during 17–15 kyr BP, δ¹³C exhibits large fluctuations and tends to be negative but not positive. Subsequent to 15 kyr BP, the positive trend of δ¹³C is gradually reestablished. However, during the period of 34–31 kyr BP, δ¹⁸O only shows a modest (∼1‰) oscillation, while δ¹³C shifts up to 4.5‰. δ¹⁸O has a long-term trend, with an overall variation range of about 5.5‰, while δ¹³C has no obvious trend on the suborbital timescale, and its millennial signal varies by as much as 6.5‰. This suggests that δ¹³C is more sensitive to climate change than δ¹⁸O on the millennial timescale.

Before interpreting the climatic significance of stalagmite δ¹⁸O of the DDH Cave, it is necessary to judge whether the stalagmite calcite was deposited under an isotopic equilibrium. Recently, the most widely used equilibrium fractionation test is the replication of δ¹⁸O records from different stalagmites in the same cave or from different caves in the same climate region (Dorale et al., 1998; Wang et al., 2001b; Dorale and Liu 2009; Li et al., 2019). Comparing the δ¹⁸O records of the DDH stalagmite with other contemporaneous stalagmite records in the ASM region, it is observed that the DDH stalagmite δ¹⁸O record is consistent with that of the Hulu Cave (Wang et al., 2001b) and Yangzi Cave (Wu et al., 2020) and that major millennial timescale events are well recorded, such as HS1 at 16 kyr BP, HS2 at 24 kyr BP, HS3 at 29.5 kyr BP, and seven DO events (Figure 4). The correlation coefficient between the DDH stalagmite δ¹⁸O and Hulu stalagmite δ¹⁸O is 0.76 (N = 349 and p < 0.001), and the correlation coefficient of the Yangzi stalagmite δ¹⁸O is 0.83 (N = 349 and p < 0.001). Therefore, we have reason to assume that the DDH-B15 stalagmite was deposited under (near-)equilibrium fractionation conditions, and the influence of kinetic fractionation is negligible.

The stalagmite δ¹⁸O is mainly controlled by the cave temperature and drip δ¹⁸O changes (Lachniet, 2009). The DDH-B15 δ¹⁸O variation is 5.5‰. The δ¹⁸O fractionation coefficient (−0.23‰/°C, O'Neil et al., 1969) caused by the temperature is very small. The 34–13 kyr BP belongs to the MIS3-2 stage, when the global average temperature did not change in the region by more than 4°C (Peterse et al., 2014). Indeed, an 8°C change between glacial and interglacial cycles in the Loess Plateau (Peterse et al., 2014) only implies a maximum temperature contribution around 1.8‰. Therefore, we can interpret the DDH-B15 stalagmite δ¹⁸O values as indicators of precipitation δ¹⁸O.

The climatic significance of the Chinese stalagmite δ¹⁸O is controversial (Maher, 2008; Pausata et al., 2011; Liu et al., 2020). In the ASM region, especially in eastern China, the stalagmite δ¹⁸O values are variably affected by many factors. Early studies had shown that the amount of rainfall was not significantly negatively correlated with δ¹⁸O values (Wang et al., 2001b). At the same time, Cai et al. (2001) had proposed that different moisture sources may affect stalagmite δ¹⁸O values. Wang et al. (2001a) used the ratio of summer/winter rainfall to explain the δ¹⁸O of stalagmites and introduced the concept of “monsoon intensity.” Cheng et al. (2009, 2016, 2021) followed and extended the concept of “monsoon intensity” and believed that the lower stalagmite δ¹⁸O mainly represented the higher spatially integrated monsoon rainfall between the tropical monsoon sources and reflected “wind” characteristics of the monsoon. Yuan et al. (2004) believed that the stalagmite δ¹⁸O record was dominated by Rayleigh distillation processes associated with the water vapor transport, that is, the “rainout effect” or “continental effect”. Subsequent work on paleoclimate records and simulations also basically supported these two interpretations (Hu et al., 2008; Pausata et al., 2011; Liu et al., 2014; Hu et al., 2019). The contribution of the rainout effect and the proportion of summer rainfall remain uncertain (Orland et al., 2009; Liang et al., 2020). Chinese stalagmite δ¹⁸O records are very consistent on precessional-millennial timescales (Wang et al., 2001b; Cheng et al., 2009). At present, δ¹⁸O is still the most effective stalagmite paleoclimate proxy. Most researchers approve that δ¹⁸O can be an indicator of monsoon rainfall in northern China or the Asian continent.

The DDH cave is located on the southern slope of the Qinling Mountains, about 500 km away from the Sanbao Cave and Linzhu Cave (Cheng et al., 2016). The DDH stalagmite δ¹⁸O is very similar to the compositive stalagmite δ¹⁸O-based records from the Sanbao–Dongge–Hulu–Linzhu Cave (Figure 4), and the long-scale changes of the DDH stalagmite δ¹⁸O are consistent with summer insolation changes in the northern hemisphere. Recent EASM reconstruction by the loess microcodium δ¹⁸O presented an opportunity to reconcile the paradox between speleothems and loess, which showed that the dominated signal of the EASM was precession but the ice volume influenced the amplitude. The loess microcodium δ¹⁸O consisted of speleothem δ¹⁸O on precessional and millennial scales (Zhang et al., 2022). This also reveals that the speleothem δ¹⁸O can be used as a monsoon tensity index. In addition, the stalagmite δ¹⁸O in the Xianglong Cave, 20 km away from the DDH Cave, was consistent with the local rainfall on the decadal timescale (Tan et al., 2015, 2018). The DDH stalagmite δ¹⁸O should show similar disciplinary. Considering the aforementioned data, the “monsoon intensity” concept is followed (Cheng et al., 2016); we argue that the δ¹⁸O of the DDH stalagmite reflects monsoon intensity changes and integrated rainfall changes on a large spatial scale. When the stalagmite δ¹⁸O is negative, the ASM rain belt is pushed northward, the rainfall in north China increases, and the rainfall in the south may be different (Cheng et al., 2016, 2019). On a shorter timescale, it is affected by local hydrology and climate. There may be some differences between the stalagmite δ¹⁸O and rainfall index. δ¹³C may represent changes of the local hydroclimate because when precipitation increases, the proportion of C₃ vegetation increases, vegetation coverage becomes larger, and the water–rock interaction (WRI) time becomes shorter, which will make δ¹³C more negative (Dorale et al., 1992; Baker et al., 1997). The relationship between δ¹³C and δ¹⁸O on different timescales may represent the relationship between local precipitation and Asian monsoon intensity. They are not identical on the orbital scale but have certain similarities on the millennial scale (Figure 4). But stalagmite δ¹³C is also affected by temperature, cave ventilation, prior calcite precipitation (PCP) process, atmospheric CO₂, and other factors (Fairchild et al., 2006; Genty et al., 2006). Therefore, it is necessary to further study the DDH stalagmite δ¹³C combined with other hydrological proxies, but it is not discussed here. In the following sections, we focused on the changes of our DDH stalagmite δ¹⁸O on multiple timescales and regional differences.

On the orbital-suborbital timescales, the δ¹⁸O of the DDH stalagmite is consistent with other caves and is mainly controlled by summer insolation in the northern hemisphere (Figure 4). The overall trend of the DDH stalagmite δ¹⁸O is similar to the Yangzi Cave (Wu et al., 2020) and the Hulu Cave (Wang et al., 2001b) stalagmite δ¹⁸O; however, during the period of 34–30 kyr BP, the Chinese cave records also showed some differences. The DDH δ¹⁸O changes from a negative value to a positive value, while the Hulu Cave stalagmite δ¹⁸O is relatively stable, with no significantly persistent positive trends during 34–18 kyr BP. In addition, the Hulu cave δ¹⁸O values during the 19–17 kyr BP period are nearly as negative as those observed in the 34–30 kyr BP period (Figure 4). The stalagmite δ¹⁸O values from the Xiaobailong Cave in Yunnan (Cai et al., 2015) were relatively stable or even gradually negative during the 34–20 kyr BP period. Mawmluh Cave (Dutt et al., 2015) and Bitto Cave (Kathayat et al., 2016) from the ISM region also share similar characteristics, which may be influenced by the low time resolution of the record, or be a characteristic in the ISM region. Therefore, further research is needed for verification. The significant negative δ¹⁸O values of the DDH Cave and the Yangzi Cave during the 34–30 kyr BP period may be related to a stronger summer insolation. However, the Hulu Cave stalagmite δ¹⁸O values are not closely related to insolation during this period.

On the millennial timescale, the DDH stalagmite δ¹⁸O record shows seven warm DO cycles (Figure 4A). The DO1 event, which is the Bolling event, occurred at 15–14 kyr BP, DO2 occurred at 23.5–23 kyr BP, DO3 occurred at 28–27.5 kyr BP, DO4 occurred at 29.2–28.6 kyr BP, DO5.1 occurred at 31.5–30.5 kyr BP, DO5.2 occurred at 32.5–32 kyr BP, and DO6 occurred at 34–33.5 kyr BP. The seven DO events in our records are temporally consistent with the Greenland ice core, but there are some differences in the structure. The onsets of these DO warm events are not very rapid, which are similar to other stalagmites in the Asian monsoon region, but not as rapid as the warming shift recorded by the arctic ice core, which may be influenced by the climate of the southern hemisphere (Cai et al., 2006; Rohling et al., 2009). In addition, the DDH stalagmite δ¹⁸O and other stalagmites δ¹⁸O records from the ASM region also record three significant HS cold events on the millennial timescale (Figure 4). HS1 occurred at 17–15 kyr BP, HS2 occurred at 24.5–23.5 kyr BP, and HS3 occurred at 30.5–29.2 kyr BP. As described by Li et al. (2019), two phases of weak monsoon occurred in the Central China at the onset of the HS1 cold event. The stalagmite records in the ISM area also exhibit a two-phase structure (Figure 4). In addition, at the end of HS1, the DDH record and the Hulu Cave, Xiaobailong Cave, Sanxing Cave, and others also showed a two-stage process of monsoon enhancement. In general, this two-stage enhancement process can be regarded as a multi-cycle slow enhancement process. Four internal turbulent events at the onset of HS1 can be found in the Greenland ice core, Hulu Cave, and DDH records. The onset and termination of HS2 and HS3 events from high latitudes may also have multi-stage structures, but the performance is not obvious in DDH and other stalagmite records, which may be attributed to duration or time resolution limitations (Figure 4).

Comparing the DDH stalagmite δ¹⁸O record with other stalagmite δ¹⁸O records, it is found that the three HS events do not have exactly the same structural characteristics. Generally speaking, HS1 and HS3 events are characterized by slow entry and rapid termination. This mode has similar characteristics to the glacial–interglacial cycle, which may be related to the “binge-purge” dynamic mechanism of ice sheet evolution (MacAyeal, 1993), while HS2 and Holocene millennial events (such as 4.2 and 8.2 kyr) mainly enter quickly and terminate slowly (Liu et al., 2018). This implies that the HS1 and HS3 events of the glacial period are different from the Bond events of the interglacial period (Liu et al., 2018), and the structure of the HS2 event is different from the other two HS events.

Considering the aforementioned characteristics together, we found that precessional and millennial signals are the main portrait of the ASM, and low latitude summer insolation is the driving force of the ASM which is coupled with high-latitude climate changes on the millennial scale. However, there are some differences in the stalagmite δ¹⁸O of different caves on the suborbital scale, and there are also some differences in the structure of different HS events.

In order to assess the dominant period of the DDH stalagmite δ¹⁸O, we first used REDFIT software (Schulz and Mudelsee, 2002) to analyze the δ¹⁸O signal. It is found that the DDH stalagmite δ¹⁸O has significant periods of 555, 206, and 142 years (Figure 5), consistent with the main periods observed in the Holocene Dongge Cave δ¹⁸O record (Wang et al., 2005; Dykoski et al., 2005). These periodicities are close to significant periods (512, 206, and 148 years) of the △14C record (Stuiver and Braziunas, 1993). Therefore, 555, 206, and 142 years are the main periods associated with solar activity. It is observed that 206 years is the Suess period of solar activity and 555 years may be the significant 500-year period of the ASM (Xu et al., 2019), Greenland temperature (Stuiver et al., 1995), and ENSO variability (Zhu et al., 2017). These periods support the idea that solar activities are responsible for changes of glacial ASM intensity.

For distinguishing the variability of the DDH stalagmite δ¹⁸O on different timescales, the EEMD method (Wu and Huang, 2009) is employed to decompose it. At the same time, we also compared the variability of the stalagmite δ¹⁸O from the ISM region and the EASM region on different timescales. The Hulu Cave and XBL Cave with high resolution δ¹⁸O and an accurate age model are selected as representatives of the EASM and ISM. Interpolating the three records to a resolution of 30 years, we used the EEMD method to decompose the DDH stalagmite δ¹⁸O into nine components, including the long trend. These nine components extend the period from orbital to suborbital, millennial, and centennial time scales (Figure 6). IMF C7-C9 (Figure 6B) are reconstructed to form a new component of the orbital-suborbital timescale, which should be dominated by precessional signals (Yuan et al., 2004; Wang et al., 2008); IMF C4-6 (Figure 6C) are reconstructed to form a millennial time scale component, including cycles of 5,000, 2000, and 1,000 years, which are consistent with the northern high latitude temperature changes (Bond et al., 1999; Wang et al., 2008); IMF C1-3 (Figure 6D) are reconstructed to obtain cycles below the millennial time scale, which are mainly on the centennial time scale, corresponding to 200- and 500-year cycles of solar activity.

On the three different timescales, the stalagmite δ¹⁸O records of the three regions have similarities and differences. On the centennial (Figure 6D) and millennial (Figure 6C) timescales, stalagmite δ¹⁸O changes in the three regions are almost the same, and their amplitude and frequency are similar. On the millennial scale, three HS events can be clearly identified. All these show that the climate of the ASM region (EASM, ISM, and Central China) changes contemporaneously, within the age error range on the centennial and the millennial timescale. In addition, millennial and centennial variability both show an increasing trend since 19 kyr BP. This may be associated with the abnormally dry and cold HS1 event or deglaciation on the millennial–centennial time scales, which could be verified by long high-resolution records. On the suborbital time scale (Figure 6B), stalagmite δ¹⁸O records in the three regions show three trends during the 24–16 kyr BP period. The Hulu Cave stalagmite δ¹⁸O values in eastern China become gradually negative, consistent with the more positive nature of HS2, as compared to HS1. The Xiaobailong Cave δ¹⁸O values from the ISM region gradually become more positive, while the DDH stalagmite δ¹⁸O in Central China adopts an intermediate state between the EASM and ISM, further supporting the idea that the precipitation in Central China during the 24–19 kyr BP period was a mixture of water vapor from the Indian Ocean and the Pacific Ocean. This is probably caused by a decline in the sea level during the glacial period, while the ISM region was mainly influenced by ice volume (Cai et al., 2015; Xue et al., 2019). This difference on the orbital to sub-orbital timescale indicates that ASM rainfall isotope data also have regional differences, although precession is the main driver.