In July 2019, a 3.9-m-long sediment core (MGC-2) from the centre of Lake Mugeco was recovered by an Uwitec drilling platform at a depth of 30.5 m (Figure 1C). Then, in May 2021, four short sediment cores were obtained using a piston gravity corer at different depths varying from 23 to 30.5 m. These short cores were respectively named of MGC21A (31 cm-long), MGC21B (27 cm-long), MGC21C (25 cm-long), and MGC21D (32 cm-long) (Figure 1C).

All short sediment cores were split along the central axis with a core-cutting machine in our laboratory. Halved cores were photographed, and the lithology was described. This study focuses mainly on short-core MGC21A for dating, XRF core scanning, and analyses of grain size, total organic carbon (TOC), and total nitrogen (TN). For MGC21B, MGC21C, and MGC21D, XRF core scanning was used to get X-radiographic images and element variations.

The deposits of the MGC21A core mainly consist of greyish-brown and light greyish-black clayey silt. Percentages of silt and clay range from 75.15 to 85.13% and 13.37–23.15%, with an average of 82.34 and 16.09%, respectively (Figure 2). The proportion of sand is less than 2.96% on average. It is found that the variation of clay is the opposite to that of silt. Also, the median diameter ranges from 9.33–11.40 μm with an average of 10.37 μm, in accordance with the variation of silt. Besides, Rad values are associated with grain size variations, which means that overall low Rad values respond to high content of clay fraction, and vice versa (Figure 2). There are three whitish layers presented as dark bands in the X-radiographic image at the depths of 6.5, 17, and 22 cm in the MGC21A core (shown as E1, E3, and E4 in Figure 2). They are also characterized by relatively high clay content and low silt content, and the median diameter and Rad values decrease correspondingly as well. At a depth of 9.5 to 8.5 cm, there are two whitish layers characterized by dark bands in the X-radiographic image and low Rad values. Due to the closeness of these two layers, we named them E2 (Figure 2). Given that the resolution of grainsize is low, clay and silt content do not have variations, but X-radiographic images present dark and Rad values decrease in the E0 layer.

In the core MGC21A, the contents of TOC and TN vary from 1.88 to 7.22% and from 0.29 to 0.78%, respectively, and their variations are consistent with each other. Carbon/nitrogen atomic (C/N) ratios are generally less than 10 (Figure 2). The incoherent/coherent scattering intensity ratios (inc/coh ratios) display a similar trend with TOC and TN as well (Figure 2). Contents of TOC, TN, and inc/coh ratios rapidly decrease in E2–E4 layers while they do not show obvious decreasing trend in E0 and E1 due to their thin layer or low resolution of TOC and TN (Figure 2). E2 also shows a double wiggle on variations of inc/coh ratios.

Variations of elements transformed by CLR in the core MGC21A are shown in Figure 3. PCA results show that the first two principal components together capture 65.29% of the variance (PC 1: 55.64%, PC 2: 9.65%) (Figure 4A). PCA 1 has positive loadings for elements of K, Rb, Ca, Sr, Ti, and Si, most of which belong to detrital elements. PCA 2 has positive loadings for Fe and negative loadings for Mn (Figure 4A), possibly reflecting the redox status of Mugeco Lake. It is discovered that the contents of K, Rb, Ca, Sr, Ti, Si, and PCA 1 values increase sharply in the base of four whitish layers and decrease upward, labeled as E0, E1, E3, and E4, especially in the E1, E3, and E4 layers (Figure 3). E2 also shows a double wiggle on the content of detrital elements and PC 1 values (Figure 3).

Generally, there are the same sedimentary characteristics in the E1, E2, E3, and E4 layers. The sediment colors in these layers turn white, and their dark X-radiographic images suggest increasing clay content (Croudace et al., 2006), decreasing median diameter, a higher density, and lower Rad values than other parts (Figure 2). As the sediment accumulation rate of the core is very low (0.91 mm/yr) and the highest resolution we can achieve for grainsize analysis is at a 0.5 cm interval, which spans ca. 6 years, it is impossible for grainsize data to show the graded bed with coarser sediment at the base finning upward in E0–E4 layers. However, the coarse sediment fractions are often rich in detrital elements (K, Rb, Ca, Sr, Ti, and Si), which can reflect the mm-scale changes in grain size (Avşar et al., 2014; Gastineau et al., 2021; Wilhelm et al., 2022). Sharp increasing of these detrital elements and PCA 1 values in the base of E1–E4 layers and decreasing upward well indicate coarser sediment at the base finning upward, which suggests that E1–E4 layers are characterized by turbidites. In order to avoid the one-core occasionality and to validate the universality of E0–E4 layers, three short cores (i.e., MGC21B, MGC21C, and MGC21D) in different parts of Lake Mugeco were analyzed and compared with the core MGC21A (Figure 2). It is found that all short cores contain five layers characterized by darker X-radiographic images, decreased Rad values, and PC 1 representing detrital element content (Figures 4B–D), and decreased inc/coh ratios, which can represent organic matter content (Figure 5).

A logarithmic plot of ²¹⁰Pb_ex activity shows a general linear trend with depth (Figure 6A). The exponential decay pattern of ²¹⁰Pb_ex activity along the depth is employed to develop a chronology by applying the CFCS model (Bruel and Sabatier, 2020), indicating a low sediment accumulation rate (SAR) of 0.91 mm/yr (Figure 6A). The artificial isotope ¹³⁷Cs has been mainly released from atomic bomb tests, nuclear industrial emissions, or accidents (Foucher et al., 2021). After release, it can enter the atmosphere, and then be precipitated and preserved on land or water. The ¹³⁷Cs activity increases upward from the depth of 7 cm in the core of MGC21A and reaches its highest values (∼160 Bq/kg) at 5 cm (Figure 6A), which could be related to the most intense period of nuclear testfallout. And the values gradually decrease upward from a depth of 5 cm. Nuclear tests started in 1955 C.E. and reached their climax around 1963 C.E. (Foucher et al., 2021), thus the peak at 5 cm-depth is fixed at 1963 C.E. (Figure 6A) (Norris et al., 1994). The ¹³⁷Cs age of 1963 C.E. at 5 cm is consistent with the CFCS model result (Figure 6A). The ¹⁴C age dated on the plant remains at 50 cm in the long core MGC-2 parallel to the short core MGC21A, which is 1230 ± 30 BP. Combined with the age model derived from ²¹⁰Pb/¹³⁷Cs and AMS¹⁴C dating, the chronology is established by the Bacon 2.5.7 procedure in R software using the Bayesian method (Blaauw and Christen, 2011; Reimer et al., 2020) for the upper 50 cm sediment of Lake Mugeco (Figure 6B).

Previous studies indicate that turbidites can be produced by flood, landslide, deltaic collapse, or earthquake (Brocard et al., 2014; Ghazoui et al., 2019; Vandekerkhove et al., 2019; Wils et al., 2021). Flood-induced turbidities usually contain multitudes of allochthonous detrital materials and present a fining-upward unit and a better sorting in grain-size (Vandekerkhove et al., 2019; Zhang et al., 2015). Furthermore, floods and landslides have a short recurrence interval (Archer et al., 2019). It was documented that lasting heavy precipitation in June and July 1995 caused rising water levels in the Yala River and Zheduo River, which originates from Lake Mugeco and Gongga Mountain, respectively. This heavy precipitation induced a flood occurring in 1995 in Kangding city that has encountered the most serious flood since 1776 (Xie et al., 1997). E0 at a depth of ∼2.0 cm in the core of MGC21A is characterized by a dark X-radiographic image, a slight decrease in Rad and inc/coh ratios, and an increase in PC1 and elements of K, Rb, Ca, Sr, Ti, and Si. The E0 layer in other short cores also has the same characteristics. The ¹³⁷Cs/²¹⁰Pb dating suggests that the E0 layer is dated to 1995–1998 C.E. and was presumably triggered by the flood in 1995 (Figure 6A). The proxies in the E0 layer are similar to those in the E1–E4 layers (Figure 5), but have much little variations, which indicate that the E1–E4 layers are impossible to trigger by a flood event. Landslide-induced subaqueous deposits are generally rich in terrestrial organic matter (Avşar et al., 2014; Bussmann and Anselmetti, 2010), but with low C/N ratios in E1–E4 layers, that organic matter is endogenous (Figure 3). So, the E1–E4 layers are not landslide-induced deposits in Lake Mugeco. Turbidites induced by slope failure are either similar to or coarser than the background sediment as they originate in the slope areas (Moernaut et al., 2014; Van Daele et al., 2015; Wilhelm et al., 2016). E1–E4 layers with higher clay content indicate that they are unlikely to be induced by slope failure (Figure 2).

At the bottom of turbidite layers, there is a sharp increase in detrital element content at the base and fine-grained matter in these layers (Figures 2, 3, 5). In terms of these observations, we suggest that E1–E4 turbidite layers in Lake Mugeco are considered the result of earthquake-induced seiches (Chapron et al., 1999; Rapuc et al., 2018; Schwab et al., 2009). Beck (2009) and Aurelia et al. (2013) attributed fine-grained whitish muddy turbidites to earthquake-induced subaqueous deposits (Aurelia et al., 2013; Beck, 2009). Due to the small and narrow surface of Lake Mugeco, the seiches are prone to being triggered when an earthquake takes place (Avşar et al., 2014; Barberopoulou, 2006). Seiche waves erode surface materials from the lakeshore, and coarse grains are moved and then deposited in shoal water while separated and suspended fine grains are transported to the central lake. Meanwhile, the surficial soft sediments on the lakebed are remobilised and resuspended. Under the effect of gravity separation, fine grains are precipitated more slowly than coarse ones, and transformed into a whitish silt homogenites layer that is rich in detrital elements. From XRF core scanning, inc/coh ratios usually increase in agreement with concentrated organic matter (Woodward and Gadd, 2019), and thus Woodward et al. (2018) used inc/coh ratio as a proxy index of organic matter to reveal seismic events recorded in Lake Chappa’ai in the Southern Alps, New Zealand (Woodward et al., 2018). However, the MGC21A core shows a decrease in the inc/coh ratios, TOC and TN contents in E1–E4 layers, which means that during the formation of homogenites, suspending organic matter is oxidised and decomposed in the water column before precipitation (Schwestermann et al., 2020). Moreover, the lower C/N ratio in E1–E4 layers could indicate a lacustrine origin through mass reworked deposits related to earthquake-trigged seiches rather than a terrestrial origin characterized by higher C/N attributed to floods (Figure 2) (Howarth et al., 2012).

To further validate that the deposits in the E1, E2, E3, and E4 event layers of MGC21A were generated by earthquakes, their ages based on our chronology model were compared with the records of past earthquakes occurring in the study area. Four turbidite layers (E1-E3) are dated to 1944–1956 C.E., 1919–1932 C.E., 1673–1837C.E., and 1507–1739 C.E. respectively (Figure 6A). On April 14th, 1955,a M _s 7.5 earthquake happened in the Zheduotang segment (101.8°E, 30°N) to the 17 km southeast of Lake Mugeco (Figure 1B), its shaking intensity was measured as X at the epicentre and VII in the study area referring to the Chinese Seismic Intensity Scale (1980) (Wang et al., 1996; Yan et al., 2019). This earthquake produced a 43-km-long surface rupture to the east of the epicentre, spreading from Kangding to Wasigou, where the seismic intensity was also measured as VII. Such natural hazards as collapse, landslides, and rockfall were also triggered at Nanmenguan and Angzhou, in Luding County to the southeast (Wang et al., 1996). Based on the ¹³⁷Cs and ²¹⁰Pb chronology, the E1 layer may coincide with the M _s 7.5 earthquake in 1955 (Figure 6A). According to data from the National Earthquake Data Center, there was another M _s 6.0 earthquake in southwest Ganzi County, Sichuan Province (101.8°E, 30.1°N) on 3rd May, 1932. The epicentre was only about 7 km from Lake Mugeco, and the earthquake intensity exceeds VII. As there were no other earthquakes from 1919 to 1932 C.E., the E2 layer, which consists of two whitish layers, may be an amalgamated turbidite resulting from a synchronous trigger by an earthquake in 1932. There were M _s 7.8 and M _s 7.0 earthquakes with the earthquake intensity of Ⅸ at the epicentre in Dajianlu (nowadays known as Kangding) on 1st June 1786, and on 1st August, 1725, respectively (Writing Group of “Compilation of Sichuan Earthquake Data”, 1980) (Figure 1B). Based on the chronology model (Figure 6B), the E3 and E4 layers are possibly related to the M _s 7.8 and M _s 7.0 earthquakes in 1786 and 1725, respectively. Generally, within dating errors, E1–E4 layers coincide with the M _s 7.5 earthquake in 1955, the M _s 6.0 earthquake of 1932, the M _s 7.8 earthquake in 1786, and the M _s 7.0 earthquake in 1725, respectively (Figure 6).

The sensitivity of the lacustrine sediment to seismic activity depends on the magnitude of the earthquake and the distance between the position of the lake and the epicentre. Chassiot et al. (2016b) found that maar Lake Pavin was imprinted by earthquakes with a seismic intensity of Ⅴwithin a 15-km radius. Ghazoui et al. (2019) discovered that a minimum M _s 5.6 earthquake within a 15-km radius or more than a M _s 6.5 earthquake within an 80-km radius could be recorded by the sediment of Lake Rara (Ghazoui et al., 2019). It was shown that earthquake-triggered water oscillation (seiche) occurs when the earthquake magnitude exceeds 7.0 (Alsop & Marco, 2012; Avşar et al., 2014). Although there are lots of earthquakes with different magnitudes around Lake Mugeco, only four earthquakes of M _s 6.0–7.5 within a 40-km radius have been well-preserved in the sediment of Lake Mugeco, and the earthquake intensity has been greater than Ⅶ in Lake Mugeco during the last 300 years. It is clear that Lake Mugeco can only record strong earthquakes with a short epicentre distance.