Shaken and Stirred: A Comparative Study of Earthquake‐Triggered Soft‐Sediment Deformation Structures in Lake Sediments

Subaqueous paleoseismic studies used soft sediment deformation structures (SSDS) to discern the shaking strength of past earthquakes, as the deformation degree of SSDS related to Kelvin Helmholtz Instability evolves from disturbed lamination and folds to intraclast breccia with higher peak ground accelerations (PGA). We lack comparative studies of different sediment types with SSDS related to earthquakes from different seismogenic sources to comprehend how these factors modulate earthquake‐induced deformation. Here, we compile sediment records with seven earthquake‐triggered SSDS from 10 lakes with organic‐, carbonate‐, siliciclastic‐, and diatom‐rich sediment from three subduction zones and one collisional setting. We target basin sequences with slope angles <0.65° to reduce the influence of gravitational downslope stress. We find that even minimal increases in slope angle, maximal 1°, lead to higher deformation degrees and, for some earthquakes, SSDS are only present at >0.65°. Fine‐grained clastics enhance sediment susceptibility to deformation, whereas abundant diatoms reduce it, demonstrating the influence of composition. Deformation correlates best with PGA and the vicinity of the earthquakes, suggesting that high frequency shaking promotes deformation. In addition, deformation only occurs above a minimum magnitude dependent on sediment composition, and higher deformation degrees in our studied basin sedimentary sequences only above Mw 4.9 for all sediment types, suggesting that sufficient duration of shaking—magnitude correlates with duration—is essential for SSDS development. We advise taking multiple cores on gentle slopes to study SSDS—additional to basin cores—to resolve small magnitude local earthquakes and relative differences in frequency content of past events.


Introduction
Seismic hazard assessment is essential for the safety of societies in areas of high seismicity as well as in those with low seismic activity where strong but unexpected earthquakes can occur.However, it is impossible to evaluate the recurrence rate of earthquakes and maximum possible magnitude based on instrumental and historical records alone, as the time between events often exceeds the timespan of these records.Subaqueous paleoseismology showed its potential to extend the observation window and, in some cases, provide information on recurrence rate, rupture extent and/or magnitude using sedimentary records of earthquake-triggered turbidites, mass-transport deposits, catchment response and soft-sediment deformation structures (SSDS) (Goldfinger et al., 2012;Howarth et al., 2016;Kremer et al., 2017;Lu et al., 2020;Moernaut et al., 2018;Praet et al., 2022;Usami et al., 2018;Vanneste et al., 2018).Most of these earthquake-induced event deposits are studied in the depocenters and not at their source areas.However, transportation and deposition can alter the volume and composition of the remobilized sediment, possibly leading to loss of essential information on the causative earthquakes (Schwestermann et al., 2020).On the other hand, earthquake-triggered SSDS form and are studied in situ, potentially preserving information on the seismic shaking that created them.
Seismic shaking can deform sediment by inducing shear stress between sediment layers, leading to Kelvin-Helmholtz Instabilities at their boundary (KHI; Heifetz et al., 2005;Wetzler et al., 2010).Gravitational downslope stress then modulates how the instable sequence deforms and moves downslope, even on slopes of less than 1° (Alsop & Marco, 2013).Numerical modeling and field studies of different sediment types showed the potential of KHI-related deformation for quantitative paleoseismology as stronger shaking increases the deformation degree from disturbed lamination over folds to intraclast breccia (e.g., Heifetz et al., 2005;Lu et al., 2020;Wetzler et al., 2010).However, most paleoseismic studies using SSDS records focus on a single site (e.g., Avşar et al., 2016;Lauterbach et al., 2019;Migowski et al., 2004;Monecke et al., 2006;Oswald et al., 2021).Until now, no studies have compared KHI-related SSDS in different sedimentary and seismotectonic settings, thereby inhibiting thorough evaluation, which ground motion characteristics drive and how lithology modulates earthquake-induced deformation.
Sediment characteristics alter sediment sensitivity for earthquake-triggered deformation through their influence on geotechnical and rheological properties (e.g., Alsop & Marco, 2014;Balaban-Fradkin et al., 2022;Molenaar et al., 2022;Wetzler et al., 2010).Therefore, thorough knowledge of the effect of lithology on earthquaketriggered deformation is essential to derive quantitative shaking constraints from SSDS records.In addition, it remains uncertain which ground motion parameters would be most suitable to quantify past shaking based on SSDS.Previous studies compared either peak ground acceleration (PGA) or seismic intensity with SSDS development (e.g., Avşar et al., 2016;Fan et al., 2022;Lu et al., 2020;Monecke et al., 2004;Oswald et al., 2021) and also investigated the influence of shaking duration (Wetzler et al., 2010).We lack a comparative study in different lithological environments that pinpoints the importance of all these ground motion parameters on seismic shaking, and elucidates how seismic waves interact with the surficial sediment.
Here, we compare SSDS in different sediment types with ground motion values from known historical earthquakes.Our objectives are twofold: firstly, to evaluate the role of sediment type on seismically induced deformation, and secondly, to assess which ground motion characteristics control deformation.To achieve these goals, we choose 10 lakes with well-studied sedimentary archives at locations with different seismogenic sources (i.e., subduction megathrust, crustal and intraslab earthquakes).We describe the sediment composition for each of these lakes using microscopic-scale image analysis, chemical element composition and radiodensity, and compare these data with ground motion and rupture characteristics of historical earthquakes.Using 3-dimensional X-ray computed tomography data (CT), we track both positive and negative evidence of earthquake-triggered deformation as well as the degree of deformation of SSDS.Using this strategy, our study is the first to evaluate SSDS records from different lithologies and seismotectonic settings, allowing for better comparison between SSDS records from different settings for future paleoseismological studies.

Setting
Our study focuses solely on basin-floor sedimentary sequences to minimize the influence of gravitational downslope stress on deformation, which has already become relevant at slope angles of ∼1°(e.g., Alsop & Marco, 2013;Molenaar et al., 2021).Therefore, we choose sediment cores from sites where slope angles are <0.65°toevaluate lithology and ground motion characteristics and subsequently compare them with sedimentary sequences from slopes with angles 0.65°-1°to investigate the effect of small-scale variations of slope angle on the basin floor.The 10 selected lakes (Figure 1a) are located either at a collisional convergent setting (i.e., Faakersee, Klopeinersee, Baldeggersee, Plansee and Piburgersee in the European Alps) or along a subduction zone (i.e., Eklutna Lake and Skilak Lake on the Kenai Peninsula, Alaska, Riñihue and Calafquén in Chile, and Maninjau in Sumatra, Indonesia).The collisional setting is solely affected by crustal earthquakes, while the subduction settings see megathrust, intraslab, as well as crustal earthquakes (Figure 1b).
Previous studies extensively investigated the sedimentary archive and earthquake-related deposits and SSDS of these 10 lakes reporting a wide variety of lithologies (see Text S2 in Supporting Information S1).Klopeinersee and Piburgersee are highly enriched in organic matter (Daxer et al., 2024;Oswald et al., 2021), while lakes Baldeggersee (Monecke et al., 2004), Faakersee (Grießer, 2022) and Plansee (Kiefer et al., 2021;Oswald et al., 2021) contain abundant carbonate material due to their carbonate-rich catchment and authigenic carbonate precipitation.Sediments of lakes Skilak and Eklutna are rich in fine-grained clastic material as sediment is brought in by glacier meltwater (Praet et al., 2022;Van Daele et al., 2020).Background sediment in lakes Riñihue and Calafquén consists of diatomaceous ooze with small amounts of terrestrial input intercalated with thin tephrafall deposits and occasional lahar deposits (Moernaut et al., 2014;Molenaar et al., 2021;Van Daele et al., 2015).Lake Maninjau's background sediment consists of diatoms along with other algae and almost no visible mineral components (Wils et al., 2021).Except for Eklutna Lake and Plansee, previous research has already described some earthquake-triggered SSDS within sedimentary sequences either from the basin or the slopes of all these lakes.

Methods
We reevaluate the published data for each lake in terms of deformation according to our research strategy consisting of three steps, as outlined below (Figure 2).More details on the performed analyses, ground motion and seismic intensity calculations, exact lake location, sediment core correlation and age-depth models can be found in Supporting Information S1.

Identifying SSDS and Determining Sediment Characteristics
In this study, we focus on three types of in situ SSDS related to KHI with increasing degree of deformation: (a) disturbed lamination, (b) folds and (c) intraclast breccia (Heifetz et al., 2005;Lu et al., 2020;Wetzler et al., 2010).For disturbed lamination, individual laminations are intact but undulating, while for folds, laminations fold with a clear vergence.For intraclast breccias, the sediment is mixed, potentially homogenized, and comprises clasts of laminations although sedimentary sections can still be correlated with other sediment cores, attesting that deformation occurred in situ.Previous studies demonstrated that deformation develops from disturbed lamination to finally intraclast breccia with increasing PGA (e.g., Lu et al., 2020;Wetzler et al., 2010).We used 3dimensional CT data to identify SSDS and the corresponding degree of deformation.It should be noted that a single SSDS interval could comprise several degrees of deformation.In this case, we always consider the highest degree of deformation for further analysis.
We determined the sediment characteristics using (a) the radiodensity derived from the CT data, (b) XRF chemical element profiles in combination with principal component analysis (PCA) and (c) smear slide analyses.We use the CT-derived radiodensity as a proxy for bulk density: both are strongly correlated (Figure S2 in Supporting Information S1), because radiodensity depends on both bulk density and atomic number (e.g., Reilly et al., 2017).All cores were taken using a hammer or gravity short coring system.Therefore, we consider the compaction by coring comparable for all cores.

Determining Seismic Shaking Strength at the Lake
Firstly, we compile a complete database of historical earthquakes, for which we only consider those events that are younger than 1900 CE or that have multiple reliable intensity data points in the vicinity of the lake to ensure higher-quality magnitude and location estimates.Based on this consideration, the only pre-1900 CE earthquakes considered in this study are the 1855 CE Swiss earthquakes, the 1690 CE and 1348 CE earthquakes in and around Carinthia, Austria, and the 1837 CE, 1737 CE and 1575 CE Chilean megathrust earthquakes.Then, we calculate both PGA and seismic intensity (Modified Mercalli Intensity, MMI) for each of the chosen events by considering their magnitude and rupture location and applying ground-motion prediction equations (GMPEs) and intensity prediction equations (IPEs), respectively.For each lake, we use GMPEs (Atkinson & Boore, 2003;Edwards & Fäh, 2013;Montalva et al., 2017;Zhao et al., 2006) and IPEs (Allen et al., 2012;Atkinson & Wald, 2007;Bakun & Wentworth, 1997;Barrientos, 1980;Doser, 2009;Dowrick & Rhoades, 2005;Fäh et al., 2011) that are empirically validated for the particular region, tectonic setting and/or source mechanism.Given the uncertainties of these calculations, the calculated shaking strengths should be considered as approximations.Lastly, we only consider earthquakes with a maximum MMI of at least IV ½ at the studied lakes for further analysis.This value is ½ intensity unit lower than the lowest reported threshold value of V for triggering SSDS (Agnon et al., 2006).In this way, we ensure that all earthquakes that could potentially deform sediment, even in sediment types that are very susceptible to deformation, are considered in our study.

Linking SSDS to Earthquakes: Positive and Negative Evidence of Earthquake-Triggered Deformation
To link the previously identified SSDS to the established earthquake record, we first determine which stratigraphic depth in our sedimentary sequences temporally corresponds to the timing (i.e., minimum, mean and maximum age) of the considered earthquakes using the age-depth models of the sedimentary sequences.These age-depth models are based on either varve counting, radiocarbon measurements, or short-lived radionuclide activity and mostly taken from previous studies unless otherwise specified in Text S2 of Supporting Information S1.
We subsequently compared these stratigraphic depths to the identified SSDS, thus, allowing identification of positive and negative evidence for earthquake-induced deformation (Figure 2).If the top of an SSDS falls within a stratigraphic depth range in which an earthquake affected the lake, we link the SSDS to the respective earthquake and note it as positive evidence for seismically induced deformation.Conversely, if no SSDS falls within the stratigraphic depth range, we classify this absence as negative evidence for earthquake-induced deformation.Note that negative evidence in our studied sedimentary sequences does not equal evidence of absence within the entire lake, as SSDS could still be present at other unsampled sites.For our study, we assume that earthquakes deform the uppermost sediment through KHI, as previous research (e.g., Avşar et al., 2016;Marco et al., 1996;Monecke et al., 2004) found that KHI-related SSDS occur at the sediment-water interface during seismic shaking (see discussion in Section 5.4).When multiple earthquake stratigraphic depth ranges correlate to a single SSDS, we assume that the earthquake with the highest local seismic intensity caused the deformation and note the other earthquakes as negative evidence.

Evaluating the Influence of Lithology and Seismic Shaking Strength on Deformation
Finally, we compare the positive and negative evidence to the sedimentological characteristics of the respective stratigraphic interval (i.e., chemical and sediment composition and radiodensity), and to the magnitude and local shaking strength (i.e., seismic intensity and PGA) for the considered earthquake.For positive evidence, we determined the XRF-derived chemical composition and radiodensity by considering the data that fall within the depth range of the entire SSDS.For negative evidence, we consider the data that correspond to the interval between the stratigraphic depth corresponding to the mean earthquake age and 2 cm below this depth to extract lithological information of the uppermost sediment where KHI would initiate (Lu et al., 2020;Wetzler et al., 2010).For the radiodensity, we determine the median for each CT slice within the targeted depth intervals and average these values to obtain the average median radiodensity (from here onwards called average radiodensity) for each instance of positive and negative evidence.

Description of the Four Sediment Types
We define four sediment types based on the sediment composition of our studied sedimentary sequences as observed by (a) smear slide images, (b) PCA of XRF elemental scanning data and (c) average radiodensity of positive and negative evidence.The four sediment types are (a) diatom-rich, (b) organic-rich, (c) carbonate-rich, and (d) siliciclastic-rich (Figure 3 and Table 1).The background sediment of the diatom-rich lakes-Maninjau, Riñihue, and Calafquén-consists mainly of diatoms in different shapes (e.g., pennate or centric) along with some amorphous organic matter and few siliciclastics and volcanoclastics.The XRF data show a high loading of Si, which is in line with previous research that linked higher loading of Si to increased biogenic silica content (i.e., diatoms; Molenaar et al., 2022).The radiodensity of the diatom-rich is low with a maximum of ∼500 HU, which is due to the high void space within each diatom frustule (i.e., 60%-70%; Losic et al., 2007).The organic-rich lakes-Piburgersee and Klopeinersee-contain abundant amorphous organic matter but also diatoms are common.The XRF data have high loading of both incoherent (Cr inc) and coherent scattering (Cr coh) as well as Fe.The incoherent and coherent scattering relate to light and heavy elements, respectively.Former studies showed that the incoherent to coherent scattering ratio positively correlates with organic matter and/or sediment water content (e.g., Croudace et al., 2006;Huang et al., 2016).The decay of organic matter consumes oxygen, thereby promoting redox conditions in the lake.Higher Fe counts can be an indication of a more reducing environment potentially caused by abundant organic matter and resulting decay (e.g., Davison, 1993;Kylander et al., 2011).The radiodensity of the organic-rich lake sediment is lowest with a maximum of ∼200 HU resulting from the low density of organic matter (e.g., Avnimelech et al., 2001), barely higher than water, which has a radiodensity of 0 HU.The carbonate-rich lakes-Baldeggersee, Plansee and Faakersee-contain very few diatoms and organic-matter, but abundant carbonate minerals in the form of detritus, chemically precipitated grains, or fine-grained mineralization on algae, all recognizable by very high birefringence colors.High positive loading of Ca and a radiodensity ranging from ∼400 to 800 HU are other characteristics.Abundant carbonates lead to high Ca counts in chemical element profiles.Finally, the background sediment of the glaciogenic siliciclastic -rich lakes-Eklutna Lake and Skilak Lake -have the smallest mean grainsize (i.e., clay-to silt-sized; Boes et al. (2017) and consist mainly of siliciclastic material with characteristic low to moderate birefringence.Strong loading of Ti characterizes the XRF data of this sediment type.Previous studies related higher counts of Ti to increased terrigenous input (e.g., Gregory et al., 2015), which agrees with the sedimentary setting of glacial-fed Skilak Lake and Eklutna Lake.The siliciclastic-rich sediment has the highest radiodensities ranging from ∼650 to 1000 HU and contains the least diatoms and organic matter of all four types.The bulk density of glacigenic lacustrine sediment is typically high and positively correlates with the amount of fine-grained sediment (Bakke et al., 2005).

Comparing Lithology With Deformation
When considering the basin sedimentary sequences (slope angle <0.65°), we observed seven SSDS related to seismic shaking in six of the 10 lakes and deformation took place in all four sediment types.We can trace all but one of the earthquake-induced SSDS over multiple sediment cores.Most of these SSDS occur in carbonate-rich sediment (two SSDS in Faakersee with and one in Baldeggersee).
Only for four of the 10 lakes, we did not observe SSDS in any of the basin-floor sedimentary sequences: Plansee (carbonate-rich), Eklutna (siliciclastic-rich), Calafquén (diatom-rich) and Piburgersee (organic-rich).The sediments of Plansee and Eklutna have distinct differences compared to those of lakes of the same group.Smear slides show that the sediment of Plansee consists of coarser grains and more fine-grained calcite precipitation mixed Note.In addition, we noted if the SSDS occurrence correlates to either magnitude, shaking intensity, PGA or vicinity of the earthquake (yes: y and no: n).
Geochemistry, Geophysics, Geosystems 10.1029/2023GC011402 with organic matter compared to the other two carbonate-rich lakes (see smear slide images in Figures S4, S7, and S10 of Supporting Information S1).The sediment of Eklutna Lake is slightly bioturbated (i.e., the varves are still clearly visible), whereas the sediment of Skilak Lake is not bioturbated.The sediment of Calafquén contains more diatoms than that of Riñihue, as shown by the smear slide and higher loading of Si (Figures S19 and S22 in Supporting Information S1; Figure 3b).The sediment of Piburgersee is highly similar to that of Klopeinersee, the other organic-rich lake, and more results in terms of shaking strengths can be found in Section 5.2.
The only intraclast breccia is observed in Riñihue in a sediment interval with the highest average radiodensity of the lake (473 HU) caused by abundant volcanogenic deposits at this stratigraphic depth (i.e., for the 1575 CE M w 9.5 megathrust earthquake; Figures S20 and S22 in Supporting Information S1).However, an earthquake with highly similar rupture extent and magnitude that affected lower radiodensity sediment with higher diatom content in the same lake did not cause deformation (i.e., the 1960 CE M w 9.5 megathrust earthquake; Figures S20 and S22 in Supporting Information S1).

Comparing Seismic Shaking Strength to Deformation
We observe deformation for earthquakes from magnitude M w 4.7 (i.e., the 1998 CE earthquake at Faakersee), a maximum rupture distance of 77 km (i.e., the 1943 CE earthquake at Maninjau), a minimum calculated average local PGA and seismic intensity range of 0.01 g and III½-IV½, respectively (i.e., the 1987 CE earthquake at Skilak Lake).More details on these three earthquakes can be found in Data Set S1 and Figures S4, S25, and S27 in Supporting Information S1.As mentioned in the previous section, we observed most SSDS in carbonate-rich sediment (i.e., 43%; three out of a total of seven earthquake-triggered SSDS).However, the average seismic shaking strength of the 21 earthquakes affecting the carbonate-rich sediment is the second lowest compared to those calculated for the earthquakes affecting the other three sediment types (local PGA and seismic intensity of 0.03 g and V¼ MMI, respectively) and also the average shaking strength related to the positive evidence is among the lowest of all positive evidence (Table 1).Conversely, we calculated the highest average seismic shaking strength for the 69 and 30 earthquakes affecting the diatom-rich and siliciclastic-rich sediment (i.e., local PGA of 0.09 and 0.06 g, and seismic intensity of V½ MMI and V MMI, respectively; Table 1).For organic-rich Piburgersee, for which we did not observe any SSDS, the local PGAs of all studied earthquakes were lower (i.e., maximum 0.04 g) than the local PGA of the earthquake that created deformation in organic-rich Klopeinersee (i.e., 0.12 g; Table 1).
We evaluated the effect of magnitude and local shaking strength (PGA and seismic intensity) versus rupture distance on deformation (Figure 4).Higher deformation degrees (i.e., folds and intraclast breccia) occur in our studied sedimentary sequences only from magnitudes of >M w 4.9 (Figure 4a).Below M w 4.9, all earthquakerelated SSDS are classified as disturbed laminations.We cannot determine a correlation between seismic intensity and deformation occurrence, as for four out of six lakes, we did not observe SSDS correlated to earthquakes with the highest seismic intensity, whereas we did for earthquakes with lower seismic intensity (Figure 4b and Table 1).For half of the lakes with at least one instance of deformation (i.e., Baldeggersee, Riñihue and Klopeinersee), the deformation does link to the earthquakes with the highest PGA at the lake (Figure 4c and Table 1).For another half of the six lakes (i.e., Baldeggersee, Skilak Lake, and Klopeinersee), we observe that only the earthquakes that occurred closest to the lakes deformed the studied basin floor sediment (Figure 4 and Table 1) and the furthest earthquake associated to an SSDS occurred at 77 km from the lake.For carbonate-rich Faakersee and diatom-rich Maninjau, the deformation occurrence did not correlate with either the highest local PGA values or the vicinity of the earthquake.For Faakersee, the closest earthquake with the highest local PGA (i.e., in 1976 CE with local PGA ∼0.04 g and rupture distance of 9 km; Figure S3 in Supporting Information S1) also had the lowest magnitude of all studied earthquakes with M w 4.5.This earthquake affected sediment with highly similar composition and radiodensity as the two instances of positive evidence in Faakersee, but did not create and SSDS.For Maninjau, the SSDS corresponds to the positive evidence with the furthest rupture distance (77 km) and a magnitude of M w 7.6 (i.e., in 1943 CE) All earthquakes rupturing closer to the lake had lower magnitudes and did not create deformation on the studied basin sedimentary sequences.

The Influence of Slope Angle on Deformation
In the previous sections, we focused on basin-floor sediment from sites with a slope angle lower than 0.65°.When considering higher slope angles (i.e., up to 1°), we find that the SSDS evidence is more prominent for five earthquakes in four different lakes-Maninjau, Faakersee, Klopeinersee and Skilak Lake (Figure 5).For three of the five earthquakes, the deformation degree is higher at the core sites with a higher slope angle (i.e., the 1943 CE event in Maninjau, 1998 CE in Faakersee and 1899 CE in Klopeinersee).For the other two earthquakes, we did not observe SSDS at core sites with slope angles below 0.65°, whereas at higher slope angles the earthquakes did deform the surface sediment (i.e., the 1998 CE in Maninjau and 2016 CE in Skilak Lake).The PGA plot also has a zoom-in.Plot (d) also considers data of the three lakes without any earthquake-triggered SSDS (within dashed circle) and shows zooms of magnitude versus rupture distance plots per sediment type.For all plots, the negative evidence is shown with a lighter hue than positive evidence.+ evidence and evidence in the legend mean positive and negative evidence, respectively.The disturbed lamination symbol (triangle) is chosen as an example in the legend to show the color of positive evidence in each lake.For the lakes that contain no earthquake-triggered SSDS, a cross is shown in the legend instead of a triangle.

Lithology Modulates Sediment Sensitivity to Earthquake-Triggered Deformation
We identify most of the earthquake-triggered SSDS within carbonate-rich basin-floor sediment (i.e., Faakersee with two SSDS and Baldeggersee with one SSDS), while the seismic shaking strengths causing these SSDS are among the lowest (i.e., PGA ranging from 0.02 to 0.06 g and MMI from ∼V-VI½) compared to the other instances of positive evidence (Data Set S1).Nevertheless, in the siliciclastic-rich sediment of Skilak, we found that the SSDS was created by the earthquake with the overall lowest calculated shaking strength of all earthquakes associated with positive evidence (i.e., PGA ∼0.01 g and MMI III-IV½).Both the carbonate-rich and siliciclasticrich lithologies mainly consist of mineral grains: either carbonates or siliciclastics with only minor diatom and organic matter content.Previous research showed that deformation mainly concentrates above fine-grained detrital-rich layers in both diatom-rich sediment and evaporitic varves (Alsop & Marco, 2014;Molenaar et al., 2022), demonstrating that fine-grained clastics can enhance the sediment's sensitivity to deformation.This observation could be explained by strain softening of the overlying sediment during seismic shaking (e.g., Gatter et al., 2021) or liquefaction (e.g., Youd, 1973).Also, previous research on evaporitic varves showed that finegrained clastics determine the physical properties of deformation detachment surfaces, even if these surfaces consist of an even mixture of clastics and aragonite (Balaban-Fradkin et al., 2022).This again demonstrates the dominant role of fine-grained clastics for deformation.Furthermore, both siliciclastics and carbonates grains observed in the studied smear slides have relatively high sphericity (Figures S4, S7, and S27 in Supporting Information S1), which makes their shear strength lower compared to less spherical particles such as diatoms (e.g., Daghistani & Abuel-Naga, 2023; Peng et al., 2021).We did not observe any SSDS in the studied sedimentary sequences of carbonate-rich Plansee and siliciclastic-rich Eklutna, despite strong seismic shaking and composition comparable to other lakes of the same sediment types.However, the Plansee sediment is coarser than that of the other two carbonate-rich lakes (Figure S10 in Supporting Information S1).A coarser grain size distribution increases the friction angle and thereby the shear strength compared to finer sediment (e.g., Deganutti et al., 2019;Sezer et al., 2011).For loose soils, a higher content of fines could lead to a lower friction angle and higher cohesion values, and thus a higher cyclic shear strength (Noda & Hyodo, 2013;Wiemer et al., 2012).This can be explained by the high specific surface area and electrostatic attraction of clay minerals.However, fine sediments in distal proglacial lakes (such as Skilak Lake) are dominated by clay-sized pulverized rock fragments (Torrance, 1983) instead of clay minerals, and thus geotechnical knowledge on soils rich in clay minerals cannot be directly translated to our comparative lake sediment study.Unlike the sediment of Skilak Lake, that of Eklutna Lake is bioturbated, which enhances shear strength due to the formation of aggregates by burrowing or the use of organic cement along worm tubes (e.g., Locat et al., 2002).We propose that the enhanced shear strength for both Plansee and Eklutna Lake reduces the sediment's sensitivity to deformation, thereby explaining the lack of SSDS despite strong seismic shaking.Based on our findings and previous field studies, we suggest that fine-grained clastic material enhances a sediment's sensitivity to deformation regardless of mineralogy, whereas abundant coarse grains or bioturbation burrows reduce it.
We identified only two earthquake-related SSDS in diatom-rich sediment, even though the earthquakes that affected the three diatom-rich lakes had the highest average shaking strength compared to those of the other three sediment types.Furthermore, the only earthquake-triggered SSDS in Lake Riñihue occurs in the densest sediment of the lake, corresponding to lower diatom content and abundant volcanogenic clastics (Figure 3c and Figure S22 in Supporting Information S1).Conversely, another earthquake that affected sediment with higher diatom content (Figure 4 and Figure S22 in Supporting Information S1) at Lake Riñihue did not deform the sediment, despite having a highly similar rupture extent as the earthquake that created the SSDS.Moreover, the sediment of Calafquén has higher diatom content than that of Lake Riñihue (Figures S19 and S22 in Supporting Information S1).In the studied sequences of Calafquén, we did not observe any SSDS, even though the seismic shaking strengths at lakes Riñihue and Calafquén were highly comparable as they are only ∼30 km apart.Based on all these observations, we suggest that a higher diatom content reduces sediment sensitivity to deformation.This is in line with previous studies showing that diatoms increase shear strength due to particle interlocking and high surface roughness (Wiemer & Kopf, 2017).
We did not trace any SSDS in the organic-rich sediment of Piburgersee.For organic-rich Klopeinersee, only the closest earthquake with the second highest seismic shaking strength of all earthquakes associated with positive evidence deformed the sediment.Based on our data set, we cannot explain this lack of deformation in Piburgersee by reduced sensitivity to deformation, as the shaking strength of the studied earthquakes of this lake did not reach those of the earthquake that deformed the sediment of Klopeinersee.Previous research described SSDS of all three deformation types in Piburgersee related to strong prehistorical earthquakes (Oswald et al., 2021), showing that SSDS formation is generally possible in the basin of Piburgersee.Based on the seismic shaking strengths related to positive and negative evidence in both organic-rich lakes, we propose that the sensitivity of organic-rich sediments to earthquake-triggered deformation is higher than for the diatom-rich sediment, but lower than for the carbonate-rich and siliciclastic-rich sediment types.This could be explained by the fact that diatoms are still common in these organic-rich sediments (Figure 3a; Figures S13 and S16 in Supporting Information S1).
We observe at least one instance of deformation in each of our four sediment types, showing that SSDS can form in near-horizontal surface sediment of all four lithological types provided that seismic shaking is strong enough.The minimum seismic shaking strength for deformation to occur on these flat basin floors depends on the sediment's sensitivity to earthquake-triggered deformation.Based on our observations, we suggest that increased fine-grained clastics enhance sensitivity to deformation, whereas diatoms reduce sensitivity to deformation.

High-Frequency Content and Long Shaking Duration Favor Deformation
For half of the lakes with at least one instance of deformation, the SSDS link to the earthquakes with the highest local PGA (i.e., Baldeggersee, Riñihue and Klopeinersee).Also, for another half of these lakes, only the earthquakes that occurred closest to the lakes created an SSDS (i.e., Baldeggersee, Skilak and Klopeinersee).For Skilak, this closest earthquake is considered an intraslab earthquake that still occurred quite far from the lake (i.e.,

Geochemistry, Geophysics, Geosystems
10.1029/2023GC011402 62 km) and caused low local shaking strength according to our calculations (i.e., 0.01 g and MMI III ½-IV ½).Previous research also found a positive correlation between deformation occurrence and degree with PGA based on both numerical models (Heifetz et al., 2005;Wetzler et al., 2010) and field studies (Lu et al., 2020;Molenaar et al., 2021).The PGA of seismic shaking is mainly controlled by high-frequency ground motion (e.g., Kramer, 1996).Moreover, the rupture distance strongly influences the frequency content of seismic waves arriving at the lake as high-frequency content attenuates strongest with distance from the rupture (e.g., Anderson & Hough, 1984).This leads to more dominant higher frequency components closest to the rupture as opposed to more lower frequency content at further travel distances.Based on our observations and the well-established relationships between PGA, rupture distance and frequency content, we propose that high frequency shaking favors deformation.The seismic source also influences the frequency content of ground motion, as crustal and intraslab earthquakes have a higher frequency source spectrum for a given magnitude compared to megathrust earthquakes (e.g., Abercrombie et al., 2021;Allmann & Shearer, 2009;Herrera et al., 2020;Van Daele et al., 2019).We observe an SSDS in Skilak Lake for an intraslab earthquake despite low corresponding local shaking strength.In addition, we observed another SSDS in Skilak Lake (only at a site with slope angle higher than 0.65°), corresponding to an intraslab earthquake with lower local PGA and further rupture distance than a megathrust earthquake that affected the lake (Figure 5 and Data Set S2).These observations also support the hypothesis that high frequency content favors deformation, as the high-frequency ground motion induced by the intraslab earthquakes deformed sediment despite low calculated PGA and seismic intensity.Previous research showed that the build-up of excess pore pressure during seismic shaking mainly relies on the maximum shear stress induced by the earthquake, which relates to PGA, and the number of shearing cycles, which is controlled by the frequency content as well as shaking duration (e.g., Tuttle et al., 2019).Even if the seismically induced excess pore pressure would not suffice to liquefy the sediment, it would still reduce the sediment's shear strength.Therefore, we propose that excess pore pressure induced by rapid shaking makes sediment more susceptible to deformation related to shear stress exerted by the same earthquake.
Using our data set, we cannot evaluate the correlation between PGA and deformation degree attested in previous research (Lu et al., 2020;Molenaar et al., 2021;Wetzler et al., 2010) as this would require multiple SSDS in one lake created by earthquakes with different local PGAs.However, we only identified single earthquake-related SSDS in all but one of the studied lakes (i.e., Faakersee: two SSDS).Also, the thickness of the deforming sequence can influence the deformation degree, as numerical modeling and field studies showed that shear energy acts more effectively on thinner deforming sequences, thereby increasing the deformation degree for thinner deforming sequences at a given PGA (Lu et al., 2020;Molenaar et al., 2022;Wetzler et al., 2010).We cannot evaluate this effect either, as it would also require multiple SSDS per lake.Nevertheless, if the relation between deformation degree and PGA holds true, we would expect that earthquakes of the same magnitude would create SSDS with a subsequently lower deformation degree the further they rupture from the lake as the PGA of the seismic waves reduces with travel distance.
Our data also show that other ground motion characteristics, aside from PGA or frequency content, influence earthquake-triggered deformation as SSDS occurrence does not correlate with local PGA or rupture distance for Faakersee, Maninjau and Riñihue.Numerical modeling also showed that the duration of the ground motion affects the geometry of SSDS (Wetzler et al., 2010) as the degree of deformation increases with a longer duration of strong seismic shaking.This relationship between deformation degree and shaking duration is comparable to that between local PGA and deformation degree, as mentioned before.Studies have shown that the duration of strong shaking positively correlates with the magnitude of earthquakes (e.g., Lee, 2002).We propose that seismic shaking duration could explain why we only identified higher deformation degrees for earthquakes with magnitudes higher than M w 4.9: Below this magnitude, the earthquake duration was too short for SSDS to develop to higher deformation degrees, independent of PGA or frequency content.Indeed, for Faakersee, we suggest that the duration of the M w 4.5 earthquake was too short to deform sediment, despite high PGA and frequency content due to its vicinity to the lake.Therefore, our data suggest that a minimum duration (or magnitude) is necessary to deform sediment.In line with previous studies (e.g., Avşar et al., 2016), our data suggest that higher magnitude earthquakes can deform at further distances, as they correspond to higher epicentral PGA values, thus affecting larger areas with intense shaking.Based on the SSDS in diatom-rich Maninjau and Riñihue, linked to high magnitude earthquakes with some of the largest rupture distances associated with all positive evidence (Data Set S1), we propose that this magnitude threshold is higher for sediment that is less sensitive to deformation, as both the local PGA and shaking duration need to be sufficient.

The Influence of Slope Angle
Our data show that even minimal changes in the slope angle (i.e., between 0.65°and 1°) can alter a site's susceptibility to deformation.In sedimentary sequences from higher slope angles (i.e., >0.65°), the deformation degree increases for some earthquake-related SSDS in lakes Maninjau, Faakersee and Klopeinersee, and, for lakes Maninjau and Skilak, earthquake-related SSDS occur that were not present in the studied cores from slope angles below 0.65°(Figure 5).Previous research showed that steeper slope angles, and corresponding higher gravitational downslope stress ease the deformation of the uppermost sediment (e.g., Biscontin & Pestana, 2006;Molenaar et al., 2021).For our study, we did not observe any SSDS on the flat basin floor of Lake Calafquén and only one SSDS for both lakes Riñihue and Skilak.However, previous studies reported abundant SSDS on the slopes of both lakes Calafquén and Riñihue (Molenaar et al., 2021) as well as in Skilak (Rymer & Sims, 1976).This contrasting observation on the slopes and in the basin of our studied lakes underlines the importance of slope angle on deformation.Furthermore, the total deformation thickness per site was already found to increase with slope angle on the same slope sections of Calafquén and Riñihue (Molenaar et al., 2021).Extensive outcrops around the Dead Sea demonstrated the influence of gravitational downslope stress at slope angles as low as ∼1°as it controlled the vergence of folds (Alsop & Marco, 2012).Our data show that SSDS are rare at sites with a slope angle below 1°, but become more abundant at higher slope angles even for lithologies that are less sensitive to deformation.However, at steeper slopes also earthquake-driven surficial remobilization becomes more prevalent (Molenaar et al., 2021), thereby potentially eroding SSDS.Previous research showed that remobilization is not necessarily a problem for SSDS-based paleoseismology provided that the SSDS is not entirely eroded and the deformation degree can still be recognized (Molenaar et al., 2021).Furthermore, above a certain threshold, the steepness of the slopes prevents sediment accumulation (e.g., Wilhelm et al., 2016), which logically also hampers SSDS formation.The upper limit of slope angle must be evaluated per lake setting as it can change with lithology due to varying geotechnical properties.Based on our findings and these previous studies, we suggest that gentle slopes with a slope angle higher than 1°are ideal for tracing SSDS in sedimentary records, but slope angles must not be too steep to avoid complete erosion of SSDS.

Implications and Recommendations for Quantitative Paleoseismology
We propose that crustal and intraslab earthquakes are especially likely to deform sediment due to their highfrequency content and the importance of high frequency shaking for deformation.Also, deeper megathrust earthquakes would then leave more SSDS than more shallow megathrust earthquakes, as the high frequency shaking increases with rupture depth (e.g., Herrera et al., 2020).In the flat basin, we observe SSDS for seismic intensities as low as IV½ at lithologies that are highly sensitive to deformation, whereas other coseismic records -of earthquake-induced turbidites, MTDs, or increased postseismic erosion rates-require minimum shaking intensities of V½ (i.e., turbidites due to deltaic slope failure: Moernaut et al., 2014;Van Daele et al., 2020).Therefore, we state that SSDS are a valuable addition to other paleoseismic records as they can reveal relative differences in the frequency content of the source spectrum of past earthquakes and resolve the occurrence of lower magnitude earthquakes provided they occur close enough to the lake.However, some factors must be considered when studying earthquake-triggered SSDS records.
Before using earthquake-related SSDS records for quantitative paleoseismology, we recommend evaluating the sediment lithology.The deformation sensitivity can be tested by (a) determining if fine-grained clastics or diatoms are dominant in the sediment composition (see Section 5.1); and by (b) comparing the SSDS record to ground motion characteristics and rupture distance of historical earthquakes.Note that lithology changes within a single sedimentary sequence also affect the sensitivity to earthquake-triggered deformation through time.Our data and previous research show that SSDS become more abundant at slope angles above 1°even for lithologies that are less susceptible to deformation due to higher diatom content.Additionally, low radiodensity contrast, gas cracks and bioturbation may hamper the recognition of SSDS.Low contrast in radiodensity can make deformation hard to recognize in CT and optical images, for example, in homogenous organic-rich sediment.In case of gas cracks and bioturbation, folds and intraclast breccia could still be visible, but disturbed laminations become successively harder to recognize when gas cracks and burrows become bigger and/or more abundant.Gas cracks could also lead to deformation resembling SSDS.Furthermore, the spatial extent of SSDS can vary strongly throughout the basin (Molenaar et al., 2021;Monecke et al., 2004), potentially leading to oversight of earthquakerelated SSDS in sediment cores.However, field studies showed that the spatial extent of SSDS (represented by abundance in sediment cores) increases with stronger seismic shaking (Molenaar et al., 2022), thereby increasing the chance of extracting an SSDS with sediment coring for these earthquakes.Therefore, we studied multiple cores per lacustrine basin to tackle this problem of underdetection of SSDS.In some cases, SSDS structures do not develop solely at the sediment-water interface but also deeper in the stratigraphy ("intrastratal deformation," Alsop et al., 2022) and thus the SSDS stratigraphic levels cannot be used to infer past earthquakes.As the SSDS in our study are in rather recent sediments and can be linked to well-documented recent earthquakes, it is plausible that sub-surface deformation is not a dominant mechanism behind our records.This mechanism can even be ruled out when the studied SSDS are located on the same stratigraphic level as seismo-turbidites (e.g., in Lake Riñihue), indicating that this level was located at the lake bottom at the time of SSDS formation.Sub-surface bed-parallel slip and translational sliding typically take place at lithological boundaries with contrasting mechanical properties (Gatter et al., 2021).For example, the detailed case studies on Lake Lisan (Dead Sea margin) show SSDS within alternations of aragonite and detrital beds (Alsop et al., 2022).Several SSDS in our study developed within overall poorly contrasting organic (Faakersee and Klopeinersee) and diatomaceous sediments (Maninjau) and thus the development of deep bed-parallel slip seems unlikely.A more detailed evaluation of surface versus subsurface deformation requires analysis of other criteria, such as folding geometries, growth sequences, or erosional surfaces (Alsop et al., 2022), which is not possible due to the small observational window formed by sediment cores.Finally, younger earthquakes may (partly) overprint the SSDS of older earthquakes depending on the thickness of the younger SSDS and the time between two events.Potentially, this could lead to underestimation or even overlooking the imprint of the younger event (Agnon et al., 2006).Contrarily, turbidites can be used to discern different earthquakes even if they occur only hours apart (Wils et al., 2020).Thus, based on all these considerations, we propose that the optimal strategy for a holistic lacustrine paleoseismological study is to sample the flat basin to map the distribution of earthquake-induced turbidites, MTDs and postseismic sediment flux, and to take multiple cores on the gentle slopes to target SSDS.

Conclusion
We present the first systematic study of earthquake-related SSDS records for different lithologies in different seismogenic settings.Here, we summarize our findings on the influence of (a) lithology on earthquake-triggered deformation and (b) magnitude, rupture distance as well as ground motion characteristics of the causative earthquakes: • Deformation can occur in basin sedimentary sequences (i.e., <0.65°) of carbonate-rich, diatom-rich, siliciclastic-rich and organic-rich sediments.However, the sensitivity to seismically induced deformation differs per sediment type.We suggest that fine-grained clastics increase the sediment's sensitivity to deformation, whereas higher diatom content reduces it (Figure 6a).Bioturbation or coarse sediment grains strongly reduce a sediment's sensitivity to deformation, regardless of the sediment type.• We find that an earthquake needs to be above a certain magnitude to deform sediment independent of the rupture distance or local shaking strength.This magnitude threshold is higher for sediments that are less sensitive to deformation (Figure 6b).Earthquake magnitude also modulates deformation degree as we only found higher deformation degrees (i.e., folds and intraclast breccia) above M w 4.9.As magnitude positively correlates to the duration of ground motion and numerical modeling showed that longer duration leads to higher deformation degrees, we propose that duration of strong ground motion explains the observed magnitude threshold and correlation between higher magnitudes and higher deformation degrees.• High frequency content of ground motion and high PGA favor deformation.We attribute many SSDS to the closest and highest PGA earthquakes.High frequency components of the source spectrum dominate the PGA and are more susceptible to attenuation than low frequency shaking.Our data show that higher magnitude earthquakes are generally capable of causing deformation at further distances than lower magnitude earthquakes, possibly due to their higher epicentral PGA and thus higher remaining PGA at further distances (Figures 6b and 6c).• Minimal changes in slope angle ease deformation as the gravitational downslope stress becomes larger.We see (a) SSDS for more earthquakes and (b) SSDS with higher deformation degrees when also looking at sedimentary sequences at slope angles from 0.65°to 1°.• Deformation can occur at a shaking strength lower than is necessary for turbidites, MTDs, or catchment response.In addition, earthquake-triggered SSDS records can reveal relative differences in the frequency content of the source spectrum of past earthquakes.However, a sediment's sensitivity to deformation fluctuates highly with both lithology and slope angle and younger earthquakes might overprint older earthquake-triggered SSDS.Our data show that SSDS are scarce in the flat basin floor but become more abundant at slope angles above 1°even for lithologies that are less susceptible to deformation.One can reduce the risk of "missing" SSDS by taking multiple cores.We propose that the most effective strategy for quantitative paleoseismology is to target the flat basin to study the distribution of turbidites, MTDs and postseismic sediment flux and to take multiple sediment cores on gentle slopes (i.e., >1°) to resolve the earthquaketriggered SSDS record. for both sensitive and less sensitive sediments and the rupture distance ranges (blue area) for both higher and lower magnitude earthquakes.Both magnitude and rupture distance of an event need to be within range to deform sediment.Subfigure (c) shows a conceptual map with the locations of hypothetical lakes A and B with the same lithologies as well as epicenters of three earthquakes with the same seismogenic source type to ensure similar frequency content: two high magnitude earthquakes (EQ1, EQ3) and a lower magnitude earthquake (EQ2).Below the map, the earthquake-related SSDS evidence in conceptual sediment cores of lakes A and B is depicted.

Figure 1 .
Figure 1.(a) World map with the locations of the 10 studied lakes.(b) Conceptual profiles for the two types of convergent settings: the subduction setting (modified from Van Daele et al. (2019)) and collisional setting (modified from Reiter et al. (2018)) along with exemplary hypocenters of the prevalent types of earthquake source mechanisms at these settings.

Figure 2 .
Figure 2. Strategy outline of this study along with examples of the three different deformation degrees (disturbed lamination: Skilak, folds: Baldeggersee, intraclast breccia: Riñihue) and an example of Faakersee with core correlation with CT and optical images, soft sediment deformation structures identification and age model along with stratigraphic depth ranges according to the age model.The map shows epicenters (diamonds) of the studied earthquakes for Faakersee.

Figure 3 .
Figure 3. Sedimentary characterization of depth intervals related to earthquakes without and without soft sediment deformation structures (SSDS) evidence of the 10 studied lakes.(a) Microscope images of smear slides representing the four sediment types.Characteristic components of each sediment type are marked by annotations and arrows or dashed areas.(b) Three dimensional principal component analysis loading-score plot of PC1, PC2, and PC3 of XRF chemical element data for sediment affected by seismic shaking of the 10 different lakes, and grouping of the lakes in the four sediment types.A rotating video of this subfigure can be found in Movie S1.(c) CT data-derived average radiodensity (HU) versus the PGA of the corresponding earthquake.Figure symbols represent deformation degree increasing from negative evidence/no deformation (circle) to disturbed lamination (triangle) to folds (cross) and intraclast breccia (square).The latter three deformation degrees are positive evidence of earthquake-triggered deformation.The dark gray arrow marks the radiodensity range where the most positive evidence of SSDS is observed.

Figure 4 .
Figure 4.The data in plots (a), (b) and (c) only show lakes with at least one instance of earthquake-triggered soft sediment deformation structures (SSDS) in the studied basin sedimentary sequences with slope angle <0.65°.The plots show: (a) magnitude, (b) seismic intensity, and (c) PGA vs. rupture distance.The PGA plot also has a zoom-in.Plot (d) also considers data of the three lakes without any earthquake-triggered SSDS (within dashed circle) and shows zooms of magnitude versus rupture distance plots per sediment type.For all plots, the negative evidence is shown with a lighter hue than positive evidence.+ evidence and evidence in the legend mean positive and negative evidence, respectively.The disturbed lamination symbol (triangle) is chosen as an example in the legend to show the color of positive evidence in each lake.For the lakes that contain no earthquake-triggered SSDS, a cross is shown in the legend instead of a triangle.

Figure 5 .
Figure5.Positive and negative soft sediment deformation structures evidence of five earthquakes in four different lakes (Maninjau, Faakersee, Klopeinersee, and Skilak Lake) plotted against slope angle.In contrast, to all other figures, we also incorporate cores from sites with slope angles >0.65°.The gray area depicts core sites with a slope angle of >0.65°.

Figure 6 .
Figure 6.Summary of the findings of this study showing the influence (a) of lithology and (b) magnitude as well as rupture distance, while (c) shows the resulting soft sediment deformation structures (SSDS) in two lakes associated with earthquakes with different magnitudes and rupture distances.Subfigure (a) shows the influence of lithology.The dominance of diatoms and fine-grained clastics in a sediment is shown above the arrow and the corresponding influence on the sensitivity to deformation is shown below the arrow.Subfigure (b) shows the magnitude ranges in which deformation can occur (red area)for both sensitive and less sensitive sediments and the rupture distance ranges (blue area) for both higher and lower magnitude earthquakes.Both magnitude and rupture distance of an event need to be within range to deform sediment.Subfigure (c) shows a conceptual map with the locations of hypothetical lakes A and B with the same lithologies as well as epicenters of three earthquakes with the same seismogenic source type to ensure similar frequency content: two high magnitude earthquakes (EQ1, EQ3) and a lower magnitude earthquake (EQ2).Below the map, the earthquake-related SSDS evidence in conceptual sediment cores of lakes A and B is depicted.

Table 1
Summary Table of Positive and Negative Evidence as Found in the Ten Studied Lakes Showing per Lake the Prevalent Sediment Type, Number of Soft Sediment Deformation Structures (SSDS)/Positive Evidence, Average Local PGA and Average Rupture Distance (Rrup) of Positive Evidence Only, and Average and Maximum Local PGA As Well As Average Rupture Distance (Rrup) Considering Positive and Negative Evidence Together