We selected organic matter samples from 6 horizons, plant residues and freshwater snail shells from 3 horizons in the LYG profile to conduct Accelerator Mass Spectrometry (AMS) ¹⁴C dating. The results are shown in Table 1. The results of AMS ¹⁴C dates were calibrated with IntCal 20 dataset in Calib 8.1 program with 2 sigma uncertainty. The AMS ¹⁴C dating of the LYG sediment profile was conducted by the BETA Laboratory and Xi’an-AMS Center.

Loss on ignition (LOI) measurements were determined at 1 cm intervals using KSL-1000X-M, weighed the appropriate amount of samples in the crucible, set 550 °C burning 3 h (Heiri et al., 2001; Santisteban et al., 2004). Elemental measurements were determined at 1 cm intervals using an Avaatech XRF scanner (Richter et al., 2006; Liang et al., 2012; Davies et al., 2015; Sun et al., 2016; Guo et al., 2021). TOC concentrations were measured at 4 cm intervals with a Costech ECS4010/4024 cube analyzer, with a measurement uncertainty of <0.1%. Total Nitrogen (TN) content was also measured simultaneously in order to calculate TOC/TN. Grain size analysis was made at 2 cm intervals with a Malvern Mastersizer 3000 laser-diffraction analyzer, which determines grain sizes of 0.01–3000 μm, with an analytical error within 2% (Sun et al., 2006; An et al., 2012). The grain size data were analyzed using the open source software QGrain (Liu et al., 2021). The above experiments were done at the Institute of Earth Environment, Chinese Academy of Sciences. All results are shown in Figure 4.

Nine ¹⁴C dating results (Table 1) are generally consistent with the stratigraphic sequence, except for LYG160. We selected two different organic materials (TOC and plant residues) for AMS ¹⁴C dating at 190 cm depth in the LYG profile, and the results showed that the ¹⁴C age of LYG190 organic matter and plant residues were 3290 ± 30 a BP and 3520 ± 45 a BP, respectively. After calibration, organic matter and plant residues reflected a consistent range of ages. Therefore, we used 8 ¹⁴C ages except LYG160, and the Bacon model (Blaauw and Christen, 2011) in R software was used for age and depth interpolation and extrapolation to obtain the age results of 0–300 cm for this profile. The deposition rate of 0–300 cm in this profile is calculated in the model, and the results are shown in Figure 3.

The sediment particle sorting coefficient varies from 2.49 to 3.97 (Figure 4), with an average value of 3.00. The LOI varies from 0.83% to 39.26% (Figure 4), with an average value of 12.34%. The TOC/TN varies from 1.22 to 12.42 (Figure 4), with an average value of 8.32. The Rb/Sr ratio varies from 0.14 to 1.04 (Figure 4), with an average value of 0.63.

Based on particle sorting coefficient, LOI, XRF results, TOC/TN and Rb/Sr ratio results, combined with the sedimentological observations in the field, the LYG profile can be roughly divided into two units from bottom to top, with the boundary at 135 cm (Figure 4):

1) 136–300 cm, which is the fluvial-lacustrine unit. This sediments unit has high particle size sorting coefficient values, with average value of 3.12. The LOI indicates low organic matter content (average value of 7.25) and good agreement with TOC content. TOC/TN ratio is low than 10. The Rb/Sr ratio is also low, with average value is 0.61.

Some studies have shown that the determination of organic matter in stratigraphic profiles can provide a basis for recovery of past environmental changes to a certain extent (Zhang, 1998). The productivity of organic matter in arid regions is regulated by the impact of precipitation, as noted by An et al. (1993). Chen et al. (2003) have observed that enhanced precipitation not only enhances productivity in lake, but also serves as a crucial factor in the preservation of organic material within the lake. In general, during the humid period, the water level of the lake rose, which became a lacks oxygen environment and conducive to the preservation of organic matter. On the contrary, the content of organic matter in the sediment decreases (Liu, 2012). Moreover, TOC/TN can be used to distinguish endogenous and exogenous organic matter in lake sediments (Meyers, 1990; Talbot and Johannessen, 1992; Meyers, 1994; Lamb et al., 2006). Aquatic phytoplankton are relatively rich in proteins, which are rich in organic nitrogen, so the endogenous C/N from organic matter is low, usually between 4 and 10. Terrestrial vascular plants are mainly lignin and cellulose, which have very low nitrogen content, thus the C/N ratio of exogenous organic matter is often greater than 20 and even as high as 100 (Meyers, 1990). Although the LYG profile has a sedimentary phase change, and the sorting coefficient can better reflect the hydrodynamic changes in the profile. The research shows that the standard deviation of sediment reflects the degree of sorting, the higher value represents the lower the degree of sorting, the stronger the hydrodynamic force (Chen and Wan, 1999). Rb appears inert during weathering and remains in the residue fraction in parent rocks as residual silicate, often accumulating in clay minerals (Koinig et al., 2003). In general, Rb and silicates are carried to basin areas (lakes) by physical weathering. Sr is one of the alkaline Earth metal elements, which can form soluble bicarbonate, chloride and sulfate into aqueous solution. Sr is often attached to the surface of carbonate or organic matter in sediments, and is eventually transferred to lake deposits as ions (Sr²⁺). Therefore, the chemical weathering intensity of the basin is strong, and the Sr content in the corresponding lake sediments will also increase. Therefore, the Rb/Sr ratio can be used as a proxy to indicate the summer monsoon intensity (Gallet et al., 1996; Chen et al., 2001). It has also been found that low Rb/Sr ratios reflect stronger transport of terrigenous detrital material under weak chemical weathering in the basin (Kalugin et al., 2005).

Here, we use the LOI of the organic matter to represent effective humidity, grainsize sorting coefficient to represent regional hydrodynamics, TOC/TN ratio to represent vegetation changes, and Rb/Sr ration to represent weathering intensity in peats and fluvial-lacustrine records to reconstruct the climate and environment changes since 9 ka in the LYG area. ( Figure 5 ). We divided the study interval into two substages for discussion.

The first stage in 9–4.4 ka BP is a wet period. During this period, the strong summer solar radiation ( Figure 5A ), strengthening the summer monsoon ( Figure 5D ) with sufficient precipitation. Average value of TOC/TN ration in LYG is 8.40, which indicates that the organic matter originated from aquatic plants. At this time, the physical properties of sedimentary profiles with high grainsize sorting coefficients reflect a stronger hydrodynamic. All of these proxies confirm the fluvial-lacustrine deposition environment ( Figure 5G ). Ma et al. (2011) integrated the multiple records on the fluctuation of the surface of the Lake Qinghai and the δ¹⁸O records of the ostracod shells of QH-2000 pore in the Lake Qinghai to show that the surface of the Lake Qinghai was relatively stable at an altitude of 3213 m from 9 to 6 ka, which was ∼20 m higher than the modern surface of the lake water. Liu (2018) reanalyzed the palaeoenvironmental records and climate model simulations of in the Lake Qinghai and found that the Holocene climatic optimum period occurred from 8 to 6 ka. The vegetation in the basin was dominated by forests and grasslands, and the water level of the lake was rising during the middle Holocene (Liu, 2018). Aeolian records located around the Lake Qinghai, for example, Zeku profile (Hu et al., 2022), eastern sandy land profile (Lu et al., 2015; Ding et al., 2019; Chen, 2021), Jiangxigou Profile (E et al., 2013; Yang et al., 2016), Xiala Reservoir in Qinghai Nanshan, and Xiangpi Mountain Loess Profile (Chang, 2016) all indicated that the aeolian activity was weak, and most dunes were stabilized.

From 4.4 ka and continuing to the present day, the second stage has been characterized as a period of low temperatures and reduced precipitation. The decreased summer solar radiation ( Figure 5A ) during this period weakened the summer monsoon, and the increased Rb/Sr ratio ( Figure 5F ) also indicates poor regional hydrothermal conditions. The sedimentation profiles and reduced sorting coefficients ( Figure 5G ) indicate that the shrinkage of watersheds until final disappearance, and LYG profiles shift from fluvial-lacustrine deposition to peat deposition. TOC/TN ( Figure 5I ) indicates that plant organic matter still originated from aquatic plants. In addition, the weak reducing environment of peat and colder climatic conditions favor the preservation of organic matter, thus the loss on ignition shows high values ( Figure 5H ). The aeolian sedimentary profile around the Lake Qinghai shows that the aeolian sand activity intensified, the climate deteriorated at ∼4.5 ka, and the whole was mainly cold and dry, with a low degree of chemical weathering (E et al., 2013; Lu et al., 2015; Chang, 2016; Yang et al., 2016; Lv, 2018; Ding et al., 2019; Chen, 2021; Hu et al., 2022).

Although the 8.2 ka cold event was the strongest cold event since 11.5 ka in the Holocene, here the LYG region did not respond strongly to it, and the organic matter only showed a weak trough (drops ∼5%). This is probably because the organic matter content is not high in the first place, and thus the variation is moderate. However, the LYG profile shows strong fluctuations in organic matter during the Late Holocene, with six rapid hydrological changes (blue stripes in Figure 5 ) recorded at ∼4.0 ka, ∼3.6 ka, ∼3.2 ka, ∼2.8 ka, ∼2.1 ka and ∼1.4 ka, respectively. These six declines in organic matter (from ∼30% to ∼10%) are highly consistent with the peak of the North Atlantic ice rafted debris (IRD) events ( Figure 5B ). Therefore, we focus on these six rapid hydroclimatic changes in the Late Holocene.

Bond et al. (1997) identified nine IRD events in the North Atlantic during the Holocene based on IRD, which were later gradually referred to as cold events. Among them, the cold events that occurred in the late Holocene were 0.4 ka, 1.4 ka, 2.8 ka and 4.2 ka ( Figure 5B ), and the corresponding numbers were 0–4 (Bond et al., 1997; Bond et al., 2001). In this study, except for 0.4 ka, the LYG profile responds to the other 3 cold events.

Among those cold events, the LYG profile has the strongest response to the 4.2 ka North Atlantic IRD event at∼4.0 ka. The organic matter ( Figure 5H ) decreased ∼30%. The summer monsoon in the Lake Qinghai Basin ( Figure 5D ) weakened, with a corresponding decrease in precipitation. Secondly, the duration of the rapid hydroclimatic change of ∼2.8 ka is longer. The organic matter ( Figure 5H ) decreases and remains at a low value (the minimum is ∼3.6%). The δ¹⁸O value of the Delingha tree ring ( Figure 5C ) was in the low value zone, with a minimum value of ∼28‰. In addition, the rapid hydroclimatic change in ∼1.4 ka period lasted much longer (∼500 years) and had the strongest impact. During this period, the low value of organic matter ( Figure 5H ) indicated vegetation coverage decreased, and it did not recover until nearly 100 years ago. Affected by the cold event, terrestrial C₃ vegetation in the Lake Qinghai Basin decreased significantly ( Figure 5E ). The δ¹⁸O of Delingha tree ring also shows a continuous decreasing trend ( Figure 5C ). Although not labelled as specific cold events, both ∼3.2 ka and ∼2.1 ka coincides with the peak of the North Atlantic IRD event ( Figure 5B ). The Lake Qinghai Basin summer monsoon ( Figure 5D ) weakened to different degrees in these two periods. Unlike all the above five cold events, there was also a rapid and prolonged hydroclimatic change at 3.6 ka, but this event is not associated with the North Atlantic IRD event ( Figure 5B ), which may be a regional feature of this region or dating uncertainty.