Introduction

Earthquakes are known to trigger submarine landslides in subduction zones, and these landslides can potentially amplify the generated tsunami wave (Kawamura et al. 2012; Strasser et al. 2013; Tappin et al. 2014), break strategically important undersea cable networks, and undercut structural foundations for oil and gas pipelines (Masson et al. 2006). However, not all recent megathrust earthquakes have produced large landslides (Henstock et al. 2006; Völker et al. 2011). In such regions, the recurrence of submarine landslides is lower than the earthquake frequency (Behrmann et al. 2014; Strasser et al. 2011), suggesting that earthquakes might represent the final trigger but not the main forcing mechanism for the generation of landslides in all regions. Nevertheless, in terms of mitigating hazards related to submarine landslides, it is important to understand their frequency and causative factors.

Drill site C0018 of the Integrated Ocean Drilling Program (IODP) in the Nankai Trough was chosen to provide unprecedented long-term records of submarine landslides in a convergent margin setting (Fig. 1). Although this margin is characterized by magnitude 8+ earthquakes occurring every ~150 years on average (Ando 1975), only six large landslide deposits (herein referred to as mass-transport deposits, MTDs) are recognized in the sedimentary record of drill site C0018 during the past 1 Ma (Henry et al. 2012). In this region, repeated periods of intensified mass wasting lasting a few hundreds of thousands of years have coincided with pulses of enhanced tectonic activity along a margin-dominating out-of-sequence thrust (megasplay) (Strasser et al. 2009), where slope oversteepening, an extensional stress regime, and the lateral transmission of fluid overpressures may have preconditioned the slope sediments to fail (Ikari et al. 2011; Strasser et al. 2011). However, the age model of the sedimentary record is constrained solely by bio- and magnetostratigraphy and two tephra ages (Ss-Az and Ss-Pnk) that date the emplacement of the oldest and largest MTD 6 to between 0.85 and 1.04 Ma (Henry et al. 2012). Thus, with the current age resolution, it is not possible to constrain the precise timing of individual MTDs, and we cannot resolve the complex interplay between tectonic and submarine landslides vs. other driving mechanisms such as climatic and oceanographic conditions that control sea-level change, sedimentation patterns, and gas hydrate stability. To better understand the preconditioning factors and triggers of submarine landslides in this highly active tectonic region, here, we present a long-term age model based on the δ18O isotope stratigraphy of planktonic foraminifera in combination with tephrochronology to date MTDs in the Nankai Trough.

Fig. 1
figure 1

Setting of the coring location in the Nankai accretionary wedge. a Overview map showing the subduction of the Philippine Sea Plate beneath the Eurasian Plate. The path of the Kuroshio Current is marked in blue. b Enlarged bathymetric map of the study site. C0018 (gray dot) is located in a basin downslope of the megasplay fault zone

Full size image

Regional setting and MTDs

The Nankai Trough is a convergent plate boundary where the Philippine Sea Plate is subducting beneath the Eurasian Plate, southwest from the Japan coast, at a rate of 4.1–6.5 cm/year (Seno et al. 1993). A SE to NW transect (Kumano transect) across the region can be divided into six morphotectonic zones: trench, frontal thrust zone, imbricate thrust zone, megasplay fault zone, Kumano Basin edge fault zone, and the forearc (Kumano) basin (Moore et al. 2009). The studied sediment sequence was retrieved from a slope basin located ~3 km seaward of the outer forearc high, which constitutes a natural boundary between the Kumano forearc basin and the outer Nankai accretionary prism (Fig. 1; Henry et al. 2012; Strasser et al. 2011).

The basin is isolated from the shore, shelf, and forearc basin, and water depth reaches ~3000 m. This basin represents the depocenter of downslope mass transport from the forearc high and megasplay fault zone (Figs. 1 and 2), as imaged by 3D seismic reflection data (Henry et al. 2012; Strasser et al. 2011). Acoustically transparent to chaotic facies with truncated reflections at the base are interpreted as MTDs that occur within the upper seismic unit 1a (SU1a; Fig. 2).

Fig. 2
figure 2

Seismic reflection profile showing the stratigraphic succession of the lower slope basin downslope of the forearc high and the megasplay fault zone (MSF). The multiple stacked MTD features (dashed black lines represent the bases of MTDs, continuous black lines the tops of the MTDs) can be identified on the seismic reflection profile. Bottom-simulating reflector (BSR; white dashed line) is indicative of the occurrence of gas hydrate-laden sediments. Site C0018 (black bar) is located within the slope basin. SU1a seismic unit 1a, SU1b seismic unit b (white transparent area marks the limits of SU1b)

Full size image

Within seismic unit 1b (SU1b), low-amplitude wedges and layers thinning out toward the basin flanks are also interpreted as MTDs. The MTDs imaged in the seismic reflection data show thicknesses of between 20 and 50 m and extend laterally over several kilometers along strike (Strasser et al. 2011). Site C0018 was drilled where the MTDs wedge out, and basal erosion by the MTDs is minimal (Henry et al. 2012). The ~300 m thick sediment sequence drilled at C0018 is subdivided into two lithological units (U1a and U1b) that correlate with seismic units SU1a and SU1b, respectively, (Figs. 2 and 3; Strasser et al. 2012). U1a corresponds to the upper Pleistocene sediments that are ~200 m thick (Fig. 3) and is composed of bioturbated hemipelagic mud with subhorizontal and SE-dipping volcanic ash layers that are a few centimeters thick, and is intercalated with a succession of six MTDs (MTD 1 to MTD 6; Fig. 3; Henry et al. 2012; Strasser et al. 2012). The MTDs are recognized based on their chaotic pebbly and/or mixed lithologies with inclined layers and shear zones (Henry et al. 2012; Strasser et al. 2012). The tops of MTDs 1, 2, and 6 are well defined and, in the cases of MTDs 2 and 6, marked by an overlying turbidite (Fig. 4; Strasser et al. 2012). X-ray computed tomography (X-CT) images were used to identify the bases of the MTDs (Strasser et al. 2012). Lineations formed by pyrite mineralization/concretions and low discontinuities associated with increased X-CT values are interpreted as shear zones. These shear zones represent the bases of MTDs 2, 3, and 5 (Strasser et al. 2012). The thicknesses of the MTDs within the U1a section vary from 0.5 to 61 m, with a cumulative thickness of 98 m (Table 1).

Fig. 3
figure 3

Correlation of seismic data with core C0018. Data from previous studies are shown, including lithology and dating of C0018 based on magnetostratigraphy, nannofossil biostratigraphy, and tephrochronology (Ss-Az and As-Pnk) from lithological unit U1a (Henry et al. 2012). Data from this study include δ18O data and tephrochronology (Ata-Th, Aso-1, BT72, and Hkd-Ku)

Full size image
Fig. 4
figure 4

Examples of sedimentological features of the hemipelagic sediment (a) and of various parts of the MTDs (such as mudclasts in b, d and e; turbidites in c and mixed sediments in f)

Full size image
Table 1 Summary of lithological features characterizing MTDs at site C0018
Full size table

The lowermost 100 m of lithological unit U1b is composed of hemipelagic sediment intercalated with turbidites and tephra layers of millimeter to centimeter scale. No MTD-features are recognized in U1b. Our study focusses on the upper 200 m of lithological unit U1a.

Methods/Experimental

Radiocarbon dating

Five planktonic foraminifera samples were extracted from the sediments of C0018 (between 0.2 and 1.4 m depth; Table 2) for radiocarbon dating. Four samples were taken from the hemipelagic sediments overlying MTD 1 while sample ETH-59792 was taken from sediments of MTD 1 (Table 2; Fig. 5). As the sampled material from MTD 1 is interpreted as redeposited sediments, the age of sample ETH-59792 is assumed to be too old and was thus excluded from the age–depth model.

Table 2 Radiocarbon ages obtained for foraminiferas (Globigerina inflata). Calibration was performed with the “calibrate” function of the clam code (Blaauw 2010) from the open-source statistical software “R” using the marine calibration curve Marine13.14C (Reimer et al. 2013). A reservoir age of 98 years was used
Full size table
Fig. 5
figure 5

Age model of MTD 1. Calibrated age intervals are indicated with 95% probability (Table 1). The gray envelope shows the 95% confidence interval, giving an extrapolated age of 13,000–14,200 year cal BP for MTD 1. The inset represents the age probability of MTD 1. The gray curve refers to the δ18O values of benthic foraminifera of the stacked curve, representative of the intermediate Pacific (Stern and Lisiecki 2014)

Full size image

The 14C ages were calibrated using the calibration curve Marine13.14C (Reimer et al. 2013) and a 14C reservoir age of 109 (standard deviation of 84) years (average value calculated from the nearest reference sites of Shishikura et al. 2007; Yoneda et al. 2007).

Tephrochronology

We sampled and analyzed tephra layers in thin sections to determine the composition and shape of glass shards. Major element compositions of volcanic glass shards were determined using an energy dispersive spectroscopy (EDS)-type EPMA at Fukushima University, Japan. An INCA x-act EDS system was attached to a JSM-5800LV scanning electron microscope. Operating conditions were 15 kV accelerating voltage, 0.3 nA electric current on a faraday cup, and 5 μm × 5 μm beam scanning area (Nagahashi et al. 2003). A geographically representative spread of volcanic glass chemistry for the past 5 Ma is reported in numerous land-based tephra studies (e.g., Kimura et al. 2015; Nagahashi et al. 2004, 2015). Using these correlations, we obtained the relative ages of tephra layers from site C0018 (Table 3; Fig. 6). The tephrochronology is used to establish an initial age–depth model of lithological unit U1a in C0018A.

Table 3 Tephra ages used as time references for δ18O isotope stratigraphy. The estimated age, method of age determination, and references are given for each tephra age
Full size table
Fig. 6
figure 6

Major oxide compositions of on-land reference tephras and tephras in C0018. The data of the on-land tephras (Table 3; Hayashida et al. 1996; Kuwae et al. 2002; Nagahashi et al. 2004, 2015; Suzuki et al. 2005; Yoshikawa and Mitamura 1999) are marked by triangles, and data of the C0018-tephras are marked by circles. Matching tephras are highlighted by the same color. Error bars represent one standard deviation

Full size image

Oxygen isotope stratigraphy

δ18O oxygen isotope compositions were measured on planktonic foraminifera (Globigerina inflata) hand-picked from hemipelagic sediments of C0018 at sampling steps of around 50 cm and considering the fraction between 125 and 350 μm. The isotopic composition was measured according to the method described by Breitenbach and Bernasconi (2011). Briefly, 100–200 μg of foraminiferal calcite tests was placed in 12 ml Exetainers (Labco, High Wycombe, UK) and flushed with pure helium. The samples were reacted with 3–5 drops of 100% phosphoric acid at 70 °C with a ThermoFisher GasBench II device connected to a ThermoFisher Delta V mass spectrometer. The average long-term reproducibility of the measurements, based on repeated analyses of standards, was ±0.05‰ for δ13C and ±0.06‰ for δ18O. The instrument was calibrated with the international standards NBS19 (δ13C = 1.95 and δ18O = –2.2‰) and NBS18 (δ13C = –5.01 and δ18O = –23.01‰). The isotope values are reported in conventional delta notation with respect to VPDB (Vienna Pee Dee Belemnite). The generated δ18O curve was then correlated to the LR04 stack isotope reference curve (Lisiecki and Raymo 2005) and the dataset of Bassinot et al. (1994).

Age–depth model

The age-depth model was constructed by linear interpolation of neighboring samples using the clam code (Blaauw 2010) from the open source statistical software “R.” The 14C dates and tephrochronology were used to establish an initial age–depth model in order to express the δ18O curve of C0018A as a function of time and to compare the isotope data with the LR04 stack reference curve of Lisiecki and Raymo (2005) and the stack isotope curve of Bassinot et al. (1994). The Bassinot et al. (1994) dataset has been used in previous studies to date these tephras (see references in Table 3). This comparison enables the definition of four matching points (X1–X4; Fig. 7), while the ages of the matching points are subsequently taken from the LR04 reference curve and attributed to peaks at C0018A. The input data used for this age model are summarized in Table 4.

Fig. 7
figure 7

Core C0018 δ18O profile as a function of age based on tephrochronology and the LR04 stack (Lisiecki and Raymo 2005) before and after matching. The dataset from Bassinot et al. (1994) is shown in black. The gray arrows between the two curves indicate the timing of the matching points (X1–X4). The pink dotted lines indicate the dated tephras used as time markers for the correlation. Green arrows indicate MTDs. MIS stages are from Bassinot et al. (1994)

Full size image
Table 4 Summary of input data for the age–depth model shown in Fig. 8
Full size table

To establish the age–depth model, we expressed the depth as MTD-corrected depth. The MTDs are considered to be redeposited sediments from the slopes and therefore might repeat parts of the δ18O sequence. This assumption might be incorrect for the very large MTDs 5 and 6. The bases of these MTDs are characterized on the seismic reflection profile by truncated upramping reflections, indicating that the landslide partly eroded the sedimentary succession of the lower basin within its depositional area (Strasser et al. 2011), and incorporated it into the landslide deposit. Thus, the thicknesses of MTDs 5 and 6 might incorporate “in situ” sediments at their base. Consequently, sediment sections may be missing in the continuous sedimentary record at the base of MTDs 5 and 6. However, based on lithology, the thickness of these in situ remobilized sediments cannot be defined. As the dating of MTD 6 is based on the two tephra layers Hdk-Ku and Ss-Az, this remobilization does not affect the estimated age of MTD 6 (Fig. 7).

Results and discussion

Tephrochronology

Based on the geochemical composition of the sampled tephras at C0018 and comparison with on-land tephras, the ages of the tephras were determined (Fig. 6; Table 3). Tephras Ata-Th (0.238 Ma; Kuwae et al. 2002; Nagahashi et al. 2004), Aso-1 (0.249 Ma; Kuwae et al. 2002; Nagahashi et al. 2004), BT72 (0.349 Ma; Kuwae et al. 2002; Nagahashi et al. 2004), and Hkd-Ku (0.76 Ma; Nagahashi et al. 2015; Suzuki et al. 2005) occur at depths of 24.9, 29.2, 57.1, and 114.3 m, respectively, (22.02, 26.4, 46.9, and 77.9 mf MTD corrected depth, respectively).

δ18O isotope stratigraphy

Oxygen isotope values of C0018 for the studied sequence (upper 200 m of C0018A) range between −0.31 and 2.39‰. While the uppermost 75 m of C0018A shows high variability in δ18O data, with values between minima of around 1‰ and maxima of 2.5‰, the lowermost 100 m display a more regular pattern (Fig. 2). For the lower part, a comparison of the C0018A δ18O dataset with the stack reference curves is not straightforward because of the difference in δ18O values between the planktonic and benthic foraminifera datasets. However, we observe broadly similar patterns between the two curves and therefore compare the trends between warm and cold phases.

Age–depth model and dating of MTDs

The final age–depth model of C0018A is based on 14C age data, tephrochronology, and isotope stratigraphy. The matching points (X1–X4) are dated at 125, 135, 328, and 400 ka on the LR04 reference curve, respectively (Fig. 7, before matching). The matching between peaks of C0018A and LR04 is based on visual correlation with uncertainties of ~5000 years due primarily to the sampling interval in C0018A. The uncertainties related to the tephrochronology can be up to several tens of thousands of years (e.g., Hkd-Ku; Suzuki et al. 2005). To take these uncertainties into account, we introduced an average error of ~2000 years for the tephras (Table 4). A comparison between the reference datasets of Lisiecki and Raymo (2005) and Bassinot et al. (1994) shows several offsets in the isotope stratigraphy.

The ages of the MTDs can be extracted from the age–depth model. MTD 1 is essentially based on the results of radiocarbon dating and is dated to 13.0–14.2 cal kyr BP (Fig. 5).

The ages of MTDs 2, 3, 4, and 5 are based on tephrochronology and isotope stratigraphy and are dated at 323–339, 372–384, 394–413, and 508–521 kyr, respectively (Fig. 8). The age of MTD 6 is based on sedimentation rates calculated between Hkd-Ku and Ss-Az and yield an age of 857–867 kyr.

Fig. 8
figure 8

Age–depth model and profile of sedimentation rates at C0018A based on 14C dates from the top of the drill core, tephrochronology, and matching points (X1–X4) used for correlation of the δ18O values of C0018A and the stack reference curve LR04. The depth of the MTD is also shown

Full size image

Furthermore, MTD 1, which is dated at 13.0–14.2 kyr cal BP, can be placed within a phase of decreasing δ18O values (Fig. 7). The onset of the termination of the last glacial cycle has been dated at ~16.5 kyr BP in the Intermediate Pacific (Stern and Lisiecki 2014). Thus, MTD 1 can be placed within Marine Isotope Stage 1 (MIS 1) (Fig. 7). The decreasing trend in δ18O values above MTD 1 (from 4.7 to 3.2‰) correlates with the trend in δ18O values of the Intermediate Pacific (Stern and Lisiecki 2014). Thus, MTD 1 can be placed within the warming trend of MIS 1 (increases in temperature and sea level), and it occurred ~2000–3000 years after the onset of MIS 1. Similarly, MTDs 2 and 6 can be placed within the warming trends of MIS 9 and 21, respectively. MTD 5 can be placed within the phase of decreasing δ18O values of MIS 13, while MTDs 3 and 4 are also placed within an interglacial (MIS 11), albeit not within the transition to heavier δ18O values. Although uncertainties of around 10,000–20,000 years exist, the dating of C0018A provides a better constraint on the timing of the MTD. Furthermore, the use of the isotope stratigraphy illustrates that the MTDs are all placed within odd-numbered marine isotope stages, thus raising the possibility of a climatic influence on the preconditioning of MTD-related slope failures.

Climate as a preconditioning factor?

The data of the present study show that the MTDs recorded at site C0018 in the slope basin of the Nankai accretionary prism occur within interglacials, suggesting that the MTDs were emplaced during phases of relative high sea level and temperatures. This implicates the role of climate as a primary preconditioning factor for slope failures in this highly active tectonic region. If we assume the result to be representative, it is necessary to discuss this possible relationship. As numerous articles have discussed the relationship between climate and submarine landslides (e.g., Owen et al. 2007; Maslin et al. 2010), here, we consider the climate-influenced factors (e.g., sea level, water temperature, and related factors such as gas hydrate destabilization and sedimentation rates) that could precondition slope failures and only briefly discuss how these factors may be applicable in the studied setting. The discussion will mainly focus on the example of MTD 1 in the context of the last interglacial cycle.

Sea level changes

Rising sea level during interglacial periods results in enhanced water loading on continental margins and overpressure in sediments. This has been suggested as a trigger for the Holocene Storegga Slide and tsunami (e.g., Smith et al. 2013). The timing of MTD 1 coincides with a concentration of passive margin deposits aged between 8 and 15 kyr (Brothers et al. 2013). Following Brothers et al. (2013), the eustatic sea level rise during this phase had the potential to cause bending stresses that could lead to enhanced seismicity, thereby establishing a direct relationship between sea level rise and slope failure on passive margins. However, Urlaub et al. (2013) instead showed that the ages of landslides on passive margins can be described by a temporally random Poisson distribution and therefore do not show a positive correlation between the frequency of major slope failures on passive margins and sea level during the past 180,000 years. Thus, it would be speculative to state that glacial-to-interglacial sea-level variations caused the slope failures of the present study.

Temperature and gas hydrates

Changes in water temperature may alter the stability of gas hydrates. Furthermore, large amounts of gas hydrates have been detected in the present study region (Bangs et al. 2010) and at drilling site C0008, located directly in the upslope region of our studied MTD site (Kinoshita et al. 2011; Fig. 1), as indicated by the presence of a bottom-simulating reflector (BSR) in seismic data (Fig. 2) and chemical anomalies in the borehole (Kinoshita et al. 2011), respectively. Most of the 250–350 m thick sedimentary succession overlying the accretionary prism is currently located within the stability zone of gas hydrate. However, during glacial terminations, bottom-water temperature can increase by as much as 2 °C, as documented in the deep waters of the western Pacific (Siddall et al. 2010). Indeed, a theoretical methane hydrate stability calculation (Sloan and Koh 2008), assuming a 2 °C increase in bottom-water temperature, diffusive heat transport through sediments, and an average heat flow 60 mW/m2 (Yamano et al. 2014), indicates that changes in bottom-water temperatures can result in an upward shift of the gas hydrate stability zone by up to 11 m over 10,000 years. This may result in hydrate dissociation, producing weak zones within sediments and therefore initiating or preconditioning slope failure (Sultan et al. 2004). In contrast, increasing pressure at the seafloor due to rising sea level would shift the base of the hydrate stability zone downward.

The Kuroshio Current off Japan is one of the strongest warm currents in the northwest Pacific Ocean (Iwatani et al. 2016; Fig. 1), as it transports heat northward from the tropics. The Kuroshio Current might therefore play an important role in increasing bottom-water temperatures in the study region. However, sea surface temperature (SST) reconstructions have shown that the area south of Japan was occupied by cold water masses during the period between the LGM and the start of the Holocene (Ikehara et al. 2011 and references therein), associated principally with a southward shift of the Kuroshio path during the last glaciation–deglaciation (Sawada and Handa 1998). Thus, sea surface temperatures were lower at the time of MTD 1.

Sedimentation rates

During glaciations and associated relative sea-level lowstands, sediment delivery to slopes is expected to be higher, thus possibly causing unstable slopes (Brothers et al. 2013). IODP drill site C0018 is separated from the shore, shelf, and upper slope areas by the morphologic high of the megasplay fault and only receives remobilized sediments from slopes that are susceptible to fail at water depths of 2000–2500 m (Figs. 1 and 2). There is no direct sediment source into the basin and the main depocenters nearby are located in the Kumano basin and downslope in the Nankai Trough (Buchs et al. 2015; Usman et al. 2014). There is possibly sediment input from the NNE, from the Tenyu or Fuji rivers (Usman et al. 2014), but as the sediments deposited in the studied basin are very fine grained, this represents only the very distal to pelagic part of the supplied sediment. The sedimentation rates at C0018A determined in this study vary by ~100 cm/kyr (which is the same order of magnitude as estimated by Henry et al. 2012). A significant increase in sedimentation rate is only observed before MTDs 3 and 4 of up to 300 cm/kyr. Thus, although the basin is isolated, we cannot exclude the possibility that sediment availability does not influence MTD occurrence. However, the exact timing of this change is not well constrained by the age model, as the sedimentation rate is based on linear interpolation between neighboring dating points.

Overall, this discussion has shown that the data are not sufficient to explicitly determine how climate is affecting slope stability. In particular, MTDs at C0018A do not coincide with every interglacial during the past 1 Ma. Thus, further studies are needed to better understand the preconditioning of mass movements.

Climate vs. local effects vs. tectonic factors

As mentioned above, MTDs at C0018 do not coincide with every interglacial during the past 1 Ma. However, we have to consider that C0018 does not record all MTDs of the slope basin, as the source and deposition of MTDs might vary through time (e.g., Yamada et al. 2010). Thus, C0018 may not be fully representative of the entire basin. However, as the 3D reflection seismic data in the region around C0018 do not image any other thick MTDs (Strasser et al. 2011), C0018 might therefore be representative in terms of its ability to capture the largest MTD events, although we cannot exclude the potential for lower-volume, higher-frequency MTDs that cannot be resolved in the seismic data. If we assume that, at least for the large landslides, C0018 is representative of the entire basin, the data suggest that MTDs within odd-numbered marine isotopic stages occur during three phases (MTD 1 at ~13–14.2 cal kyr, MTDs 2–5 between 0.3 and 0.5 Ma, and MTD 6 at 0.86 Ma). The first MTD (MTD 6) occurs close to the end of a period of major tectonic uplift of the forearc high between 0.9 and 1.3 Ma (Gulick et al. 2010). This uplift resulted in slope steepening, thereby creating the slopes necessary for the initiation of slope failure and perhaps explaining the onset of mass transport deposition at ~0.86 Ma with MTD 6. For the period around 0.5 Ma, as inferred from the reconstructed flow directions of MTDs and bathymetric analyses (Kanamatsu et al. 2014), seamount subduction could have affected the study area (Fig. 1; Kimura et al. 2011) and resulted in a change in slope orientation from NE–SW to NW–SE (the present slope). This seamount subduction might also have been responsible for the occurrence of MTDs between 0.3 and 0.5 Ma. For the youngest MTD phase (MTD 1), which begun with the end of the last glacial, no plausible tectonic activity has yet been described, but we speculate that it may relate to the current phase of recent megasplay fault activity (Moore et al. 2007). Importantly, the reconstructed tectonic uplift phases (before 0.9 Ma and at ~0.5 Ma) would also have affected the gas hydrate stability zone and may therefore have preconditioned the slopes, making them more prone to submarine landsliding.

Conclusions

Using a combination of isotope stratigraphy and tephrochronology, we present a 1 Ma record of MTDs in the Nankai Trough, offshore Japan. All of the MTDs occurred within interglacial periods, which might indicate climate is a key preconditioning agent for landslide generation. The occurrence of MTDs also coincides with changes in the tectonic framework, which therefore might also precondition slope failures, but whether an earthquake constitutes the final trigger cannot be resolved from the resolution of the data. Our conclusion is that within the investigated 1 Ma period, the relative contributions of tectonic vs. climate preconditioning factors cannot be simply resolved. Therefore, based on our single drill site, our discussion on various forcing mechanisms, including both tectonic and climatic, remains speculative. Further studies are needed to establish long-term, continuous MTD records to better elucidate the preconditioning factors that govern slope failures.