Volume 49, Issue 21 e2022GL100529
Research Letter
Open Access

How Clay Mineral Assemblages Affect Instability on the Upper Slope of the Hikurangi Subduction Zone, New Zealand

Michael B. Underwood

Corresponding Author

Michael B. Underwood

Department of Earth and Environmental Science, New Mexico Institute of Mining and Technology, Socorro, NM, USA

Correspondence to:

M. B. Underwood,

[email protected]

Search for more papers by this author
Brandon Dugan

Brandon Dugan

Department of Geophysics and Hydrologic Science and Engineering Program, Colorado School of Mines, Golden, CO, USA

Contribution: ​Investigation, Writing - original draft, Writing - review & editing, Funding acquisition

Search for more papers by this author
Sebastian Cardona

Sebastian Cardona

Department of Geology and Geophysics, Texas A&M University, College Station, TX, USA

Contribution: ​Investigation, Writing - original draft, Writing - review & editing

Search for more papers by this author
First published: 31 October 2022

Abstract

The International Ocean Discovery Program cored Sites U1517 (Tuaheni landslide) and U1519 (Tuaheni Basin) on the Hikurangi slope, North Island, New Zealand. We employed X-ray diffraction to document clay mineral assemblages within the muddy sediments, with the aim of testing their potential influences on slope stability. Detrital smectite dominates the clays, with average proportions of 52 wt% at Site U1517 and 53 wt% at Site U1519. Bulk sediment from Site U1517 contains up to ∼29 wt% smectite (average = 21 wt%), high enough to reduce the angle of internal friction (on average) to ∼6°. Stratigraphic sections at both sites are homogeneous; compositional excursions are not evident along inferred slip surfaces or weak layers. Smectite decreases in the “downstream” direction of the East Cape Current, and that spatial trend correlates with lower densities of slide scars. The uniformity of compositional preconditioning, however, points to other factors as determinants of slip nucleation.

Key Points

  • Hikurangi trench-slope sediments, including muds within the Tuaheni landslide complex, contain high concentrations of detrital smectite

  • Smectite is abundant enough to reduce the bulk sediment's coefficient of friction, with little stratigraphic variability

  • The homogeneous mud composition results from strong currents and does not change significantly along inferred slip surfaces or weak layers

Plain Language Summary

Expandable clay minerals (smectite) are notorious for weakening sediments and contributing to conditions that promote landslides. The Hikurangi margin offshore North Island, New Zealand, is noted for its abundance of submarine landslides. To assess their possible causes, we analyzed clay mineral assemblages from two sites on the upper trench slope that were sampled by the International Ocean Discovery Program. Cores from Sites U1517 and U1519 display unusual degrees of compositional homogeny, with smectite as the dominant clay mineral. We did not find any compositional anomalies at inferred slip surfaces or along weak layers of the Tuaheni landslide. Instead, the entire stratigraphic succession appears to be relatively weak. With inherited preconditioning, dynamic loading during large earthquakes is probably enough to trigger landslides, especially on steeper slopes.

1 Introduction

1.1 Conceptual Framework

The Hikurangi margin offshore North Island, New Zealand (Figure 1a) conforms to most of the paradigms of subduction zones with high rates of terrigenous sedimentation (e.g., Moore & Karig, 1976; Underwood & Moore, 1995). The margin's bathymetric architecture includes a narrow shelf, forearc basins, imbricate thrusts and folds within the frontal accretionary prism, scattered slope basins, and through-going submarine canyons (Barnes et al., 2002; Paquet et al., 2009; Pedley et al., 2010; Pouderoux et al., 2012). Submarine landslides are abundant (Collot et al., 2001; Crutchley et al., 2022; Mountjoy, Pecher, et al., 2014; Watson et al., 2020). Submarine landslides are typical of subduction margins with irregular, oversteepened slopes, and frequent earthquakes increase the potential for dynamic triggering (e.g., Harders et al., 2011; Hill et al., 2020; Riedel et al., 2019; Yamada et al., 2010). Given their importance as potential geo-hazards (Hampton et al., 1996; Locat & Lee, 2002; Masson et al., 2006; Talling et al., 2014), significant investments have been made to pinpoint the causes of landslides. Geotechnical tests, for example, have revealed how frictional properties change with concentrations of specific minerals (e.g., Ikari et al., 2018; Logan & Rauenzahn, 1987; Lupini et al., 1981; Shimamoto & Logan, 1981; Tembe et al., 2010; Tiwari & Marui, 2005). Many additional variables need to be considered, however, as contributors to preconditioning: seafloor topography; rates of sediment accumulation; grain size distribution, especially the proportion of clay-sized grains; particle shape and microfabric; hydration state of clay minerals; porosity, permeability, and pore-fluid pressure; consolidation and lithification; eustacy and gas-hydrate dynamics (e.g., Ai et al., 2014; Bartetzko & Kopf, 2007; Ikari et al., 20072011; Kock & Huhn, 2007; Kremer et al., 2017; Mountjoy et al., 2020; Nakamura et al., 2010; Ratzov et al., 2007; Saffer & Tobin, 2011; Sawyer et al., 2009; Silver & Dugan, 2020; Smith et al., 2013; Stegmann et al., 2007; Sultan et al., 2004; Takahashi et al., 2014; Trütner et al., 2015; Vanneste et al., 2014). Given that long list, the challenge with any individual occurrence is to isolate and quantify the impact of each variable.

Details are in the caption following the image

(a) Simplified map of onshore geology, North Island, New Zealand (modified from Jiao et al., 2014), prominent bathymetric features, ocean currents, and pathways for sediment transport by gravity flow. Giant mounded drifts and slope plastered drifts from Bailey et al. (2020). (b) Bathymetric map with locations of International Ocean Discovery Program sites (from Saffer et al. (2017)). (c) Seismic reflection profile crossing Site U1517. Interpretations from Barnes, Pecher, et al. (2019). (d) Seismic reflection profile crossing Site U1519. Interpretation of slope plastered drift from Bailey et al. (2020). Interpretations of faults and unconformity from Barnes, Wallace, et al. (2019). See panel (b) for tracklines.

The principal focus of our contribution is clay composition, which we use as a fingerprint for how sediment gets routed to sites on the trench slope. Sediment gravity flow provides one of the likely routes, as unconfined turbidity currents and/or by funneling suspensions down submarine canyons (Alexander et al., 2010; Mountjoy, Barnes, & Pettinga, 2009; Mountjoy, Micallef, et al., 2014; Pouderoux et al., 2012). Margin-parallel ocean currents furnish the main competition (Carter & Wilkin, 1999; Chiswell et al., 2015). The Wairarapa Coastal Current (WCC; Figure 1a) flows toward the northeast, carrying suspensions above the continental shelf and uppermost slope. That surface current can be traced 40–50 km from the shoreline, but only as far north as Hawke's Bay (Chiswell, 2000; Foster & Carter, 1997); its maximum effective depth is unresolved. North of Hawke's Bay (Figure 1a), the SW-directed East Cape Current (ECC) dominates the mid to upper slope (Figure 1a); that current entrains Wairarapa suspensions and penetrates the water column to a depth of ∼1,300 m (Chiswell & Roemmich, 1998). Transient counterclockwise gyres (e.g., Wairarapa Eddy) complicate surface-water circulation farther offshore (Figure 1a) (Chiswell, 2005; Chiswell & Roemmich, 1998). Abyssal circulation seaward of the trench is dominated by the Deep Western Boundary Current (Figure 1a) (Carter & McCave, 19972002; Carter et al., 2004). Mindful of those processes, we document how mineral assemblages change stratigraphically and along strike, as part of a more comprehensive regional study of trench and forearc sedimentation. The relevance to studies of landslides on the trench slope is to test whether clays have any demonstrable influence on frictional properties. Our results provide a tightly constrained case study of cause-and-effect that serves as a benchmark for comparisons with slope failures along other margins.

1.2 Hikurangi Drill Sites

The International Ocean Discovery Program drilled Site U1519 in Tuaheni Basin (Figure 1b), ∼38 km from shore at a water depth of 1,000 m. The site lies within the domain of the ECC (Figure 1a), and W. S. Bailey et al. (2020) and W. Bailey et al. (2021) recognized the nearby acoustic character as a “slope plastered” contourite drift (Figure 1c). Shipboard sedimentologists (Barnes, Wallace, et al., 2019) defined two lithostratigraphic units (Figure 2b). Unit I (0–282.66 mbsf) is composed of mud with sparse and very thin interbeds of silt and volcanic ash (Figure 2b); ages range from middle Pleistocene (205 ka) to Holocene (Crundwell & Woodhouse, 2022). Unit II (282.66–635.65 mbsf) consists of lower to middle Miocene mudstone (867–205 ka) with scattered interbeds of siltstone, sandy siltstone, and sandstone (inferred turbidites). Some intervals in unit II display convolute laminae, mesoscale folds, dismembered bedding, and clasts of mudstone in a mudstone matrix, all suggestive of mass transport. Evidence from cores for contourite deposition is inconclusive. In general, diagnostic criteria remain elusive for distinguishing among turbidites, contourites, and hemipelagites (D. A. Stow et al., 2002; D. Stow & Smillie, 2020), and definitive bed-by-bed identification requires time-consuming post-cruise analyses of microstructures, ichnofacies, grain size distributions, mineralogy, geochemistry, and/or computed tomography (e.g., Alonso et al., 2016; de Castro et al., 2020; Nishida, 2016; Rodriguez-Tovar & Hernandez-Molina, 2018; Vandorpe et al., 2019). Rigorous analyses of the cores from Site U1519 were thwarted by wide gaps between intervals of core (Figure 2b), unusually poor recovery (∼55%), and widespread (severe) drilling disturbance. Thus, the prevalence of contourites near Site U1519 is based largely on seismic-reflection interpretations (W. S. Bailey et al., 2020).

Details are in the caption following the image

(a) X-ray diffraction (XRD) results for Site U1517 (Underwood & Dugan, 2021). Proportions of total clay minerals and values of shear strength from Barnes, Pecher, et al. (2019). (b) XRD results for Site U1519 (Underwood, 2022). Proportions of total clay minerals from Barnes, Wallace, et al. (2019). Microfossil ages from Crundwell and Woodhouse (2022).

At a water depth of ∼732 m (Figure 1b), Site U1517 is also well within the influence of the ECC (Chiswell et al., 2015), and seismic-reflection data are indicative of a “slope plastered” contourite drift (Barnes et al., 2017; W. Bailey et al., 2021; W. S. Bailey et al., 2020). The primary purpose of drilling Site U1517, however, was to investigate the Tuaheni landslide complex (Mountjoy, Pecher, et al., 2014). Creep seems to have occurred on the slope where the base of gas-hydrate stability pinches out at the seafloor (Mountjoy, McKean, et al., 2009; Mountjoy, Pecher, et al., 2014), so one hypothesis is that occlusion of permeability by gas hydrate led to overpressured conditions and hydrofracturing (Crutchley et al., 2010; Ellis et al., 2010; Gross et al., 2018). According to Mountjoy, Pecher, et al. (2014), hydrate-bearing sediments also might be prone to time-dependent plastic deformation. To test those ideas, continuous coring crossed interpreted positions of the slide décollement at ∼37 mbsf, the base of landslide debris at ∼59 mbsf, and the base of gas-hydrate stability at ∼162 mbsf (Barnes, Pecher, et al., 2019). Shipboard scientists (Barnes, Pecher, et al., 2019) defined unit I (0–3.0 mbsf) as a thin Holocene mud blanket (Figure 2a), whereas unit II (3.0–40.74 mbsf) contains interbeds of mud and very fine sand. The bioturbated mud within unit III (40.74–66.38 mbsf) alternates with laminated stacked couplets of silt and clay, resembling silty contourites described elsewhere (e.g., D. A. Stow et al., 2002; D. Stow & Smillie, 2020). Unit IV (66.38–103.16 mbsf) consists mostly of structureless mud, whereas unit V (103.16–187.53 mbsf) contains interbeds of normally graded sand and mud (Figure 2a). The age range for the entire interval is middle Pleistocene (456 ka) to Holocene, and the base of unit III is between 45 and 25 ka in age (Crundwell & Woodhouse, 2022).

Barnes, Pecher, et al. (2019) assigned the upper ∼67 m of strata at Site U1517 to the Tuaheni landslide (Figure 2a), but many questions about the landslide remain unanswered. Gas hydrate was detected during drilling, for example, but only from intervals deeper than 100 mbsf (Barnes, Pecher, et al., 2019); that observation, together with the results of numerical modeling, indicate that gas-hydrate dynamics are unlikely as the main destabilizing variable (Screaton et al., 2019). Couvin et al. (2020) identified two chaotic acoustic intervals in nearby seismic-reflection profiles, but the correlative intervals of core (units II and III) are perplexing because they display almost no internal deformation; moreover, there are no mesoscopic indicators of concentrated slip at the base of the landslide (Barnes, Pecher, et al., 2019). The only core-scale indicators of gravity-driven, soft-sediment deformation are scattered patches of convolute bedding and intraformational mud clasts (Barnes, Pecher, et al., 2019). Couvin et al. (2020) suggested that Hole U1517C intersected an intact block within an otherwise chaotic landslide. Adding to the conundrum, Luo et al. (2020) used numerical modeling of porewater-chemistry data to support their contention of two slip events. The more-recent failure evidently occurred as a coherent 40-m-thick block, with the slide décollement at the base of unit II. Shipboard measurements of shear strength (Barnes, Pecher, et al., 2019) revealed a weak interval centered at ∼31 mbsf, within the middle of unit II (Figure 2a). These conflicting results deserve more scrutiny. Our documentation of clay mineralogy contributes by testing whether excursions in composition are large enough to have reduced the sediment's frictional properties and shear strength preferentially along any layers in the Tuaheni stratigraphy.

2 Materials and Methods

Samples from Sites U1517 and U1519 were extracted from split cores. The intervals abut whole rounds that had already been extracted from cores for tests of interstitial water chemistry and a variety of geotechnical/frictional/hydrogeologic properties. Coarser interbeds (turbidites, ash layers) were avoided. Underwood and Dugan (2021) provided detailed descriptions of sample preparation and X-ray diffraction (XRD) methods. To reiterate briefly, mud specimens were disaggregated, and oriented aggregates of glycol-saturated clay-sized splits (<2 μm equivalent spherical settling diameter) were prepared using the filter-peel method. Scans were completed using a Panalytical X’Pert Pro diffractometer, and values of normalized relative mineral abundance were computed using a set of regression equations that relate wt% values to peak area (Underwood et al., 2020). Absolute errors of accuracy are: illite = 3 wt%, smectite = 4 wt%, and undifferentiated (chlorite + kaolinite) = 5 wt% (Underwood et al., 2020). Compositional differences among individual specimens, or between lithologic units, are not regarded as geologically significant unless they exceed those errors.

3 Data

Results from Site U1519 reveal unusually small amounts of compositional scatter, with minor (but statistically significant) differences between the two lithologic units (Figure 3b). Measurements of 76 specimens (Underwood, 2022) show smectite to be the most abundant clay mineral (site average = 53 wt%), with normalized abundances of 33–65 wt%. Proportions of illite are 32–49 wt%, and the values for undifferentiated (chlorite + kaolinite) are 0–16 wt%. The mean (μ) and standard deviation (σ) values for unit I are smectite: μ = 50 wt%, σ = 4; illite: μ = 38 wt%, σ = 3; and chlorite + kaolinite: μ = 12 wt%, σ = 1. Comparable statistics for unit II are smectite: μ = 57 wt%, σ = 5; illite: μ = 36 wt%, σ = 4; chlorite + kaolinite: μ = 7 wt%, σ = 2.

Details are in the caption following the image

(a) Locations of piston-gravity cores used in this study, with depths of specimens in cm below seafloor (Underwood, 2020). Pie diagrams depict normalized proportions of smectite, illite, and undifferentiated (chlorite + kaolinite). Background shows frequency distribution of landslide scars (from Watson et al., 2020). (b) Empirical relation between proportion of smectite in bulk sediment and residual angle of internal friction, as determined by Tiwari and Marui (2005). Average and maximum values for Site U1517 yield predictions for the Tuaheni landslide. Yellow field depicts one standard deviation around the mean.

The 99 specimens analyzed from Site U1517 (Underwood & Dugan, 2021) likewise reveal small variations (Figure 2a). Smectite is the most abundant clay mineral (site average = 52 wt%), with a range of 30–63 wt% (Figure 2a). Percentages of illite are 31–54 wt%, and the range for undifferentiated (chlorite + kaolinite) is 5–16 wt%. The mean and standard deviation statistics for unit II are smectite: μ = 55 wt%, σ = 3; illite: μ = 34 wt%, σ = 2; chlorite + kaolinite: μ = 11 wt%, σ = 1. Values for unit III are smectite: μ = 48 wt%, σ = 2; illite: μ = 40 wt%, σ = 2; chlorite + kaolinite: μ = 11 wt%, σ = 0.5. Statistics for unit IV are smectite: μ = 51 wt%, σ = 5; illite: μ = 39 wt%, σ = 4; chlorite + kaolinite: μ = 10 wt%, σ = 1. In unit V, the statistics are smectite: μ = 53 wt%, σ = 5; illite: μ = 38 wt%, σ = 4; chlorite + kaolinite: μ = 9 wt%, σ = 2. Unit V displays the greatest amount of statistical scatter, with subtle gradients of decreasing, then increasing proportions of illite moving down-section (Figure 2a). Compositional shifts at horizons of interest (e.g., ∼31, ∼41, and ∼66 mbsf) are trivial, however, and within the method's normal error range (Figure 2a).

We regard concentrations of individual minerals in bulk sediment to be more diagnostic for assessing how composition affects frictional properties. The average content of total clay minerals at Site U1517 is 40 wt%, with a range of 10–49 wt% (Barnes, Pecher, et al., 2019). Average values for bulk sediment are: smectite = 21 wt% (σ = 4); illite = 15 wt% (σ = 3); and chlorite + kaolinite = 4 wt% (σ = 0.9). Contrasts across unit boundaries are within the background range of scatter, and excursions are absent at the suspected slip surfaces (Figure 2a). Coeval strata at Site U1519 (unit I) are similar (Figure 2b), essentially serving as a “reference site” of sediment unaffected by landslides.

To expand spatial coverage, we also analyzed 30 specimens from piston and gravity cores (Underwood, 2020), covering a swath of the landward slope from the vicinity of the Ruatoria debris avalanche in the NE to offshore Hawke's Bay in the SW (Figure 3a). Smectite ranges from 29 to 54 wt% (μ = 45, wt% σ = 6). The range for illite is 36–57 wt% (μ = 43 wt%, σ = 5), and undifferentiated (chlorite + kaolinite) ranges from 9 to 16 wt% (μ = 12 wt%, σ = 2). Those values overlap data from Sites U1517 and U1519. Samples from the NE corner of the study area generally contain more smectite (and less illite) compared to those to the SW (Figure 3a).

4 Results

4.1 Principal Sources for Detrital Clays

Viewing our results in the context of geotechnical properties, it doesn't really matter where the clays originated, or how they arrived at sites of deposition. On the other hand, the processes responsible for sediment routing may have imparted other sediment attributes that do influence slope stability. With that possibility in mind, we note that sedimentologists usually attribute high concentrations of smectite to weathering or alteration of volcanic sources (Biscaye, 1965; Fagel, 2007; Huff, 2016). The Taupo Volcanic Zone (TVZ) fits the prerequisites for a regionally extensive source of volcanic and volcaniclastic rocks (Figure 1a) (Cole et al., 2014; Cronin et al., 1999; Downs et al., 2014; Gravley et al., 2016; Hidgson & Manville, 1999; Kohn & Topping, 1978; Lowe et al., 2013; Procter et al., 2014; Wilson & Rowland, 2016). Their chemical weathering products are demonstrably smectite-dominant, and smectite is ubiquitous in TVZ geothermal and hydrothermal systems (Heap et al., 2020; Hocking et al., 2010; Libbey et al., 2013; Simpson et al., 2019). Heavily weathered tephra is also widespread in coastal exposures around the Bay of Plenty (Iso et al., 1982). Sediment yields are large for rivers discharging into the Bay of Plenty, averaging 13.6 Mt/y (Hicks et al., 2011). Those suspensions are picked up by the East Auckland Current before it bifurcates around East Cape. The East Cape region (Figure 1a) also exposes potential parents for detrital smectite, including mafic mélange (disrupted ophiolite) and volcanic conglomerate (Brothers & DeLaloye, 1982; Cluzel et al., 2010; Marsaglia et al., 2014). The Waiapu River (Figure 1) is noteworthy for its remarkably high loads of suspended sediment, averaging 35.07 Mt/y (Hicks et al., 2011), and that discharge probably injects some smectite to the path of the ECC.

Illite is the expected weathering product of plutonic, low-grade metasedimentary, and sedimentary sources (Biscaye, 1965; Fagel, 2007). Basement rocks across New Zealand (Figure 1a) consist mostly of illite-rich metagraywackes and argillites of the Torlesse and Waipapa terranes (MacKinnon, 1983; Sporli, 1978; Warr & Cox, 2016). Younger cover sequences and forearc deposits in the East Coast Basin, and in regions immediately southeast of the Axial Range (Figure 1a), are composed of moderately indurated illite-rich sedimentary strata (e.g., Browne, 2004; Maison et al., 2018; Neef, 1999). Accordingly, suspensions in the WCC are almost certainly illite-rich. Rivers draining the Poverty Bay area (e.g., Waipaoa) erode into Miocene-Pliocene sediments (Browne, 2004; Reid, 1998). Smectite is a reported constituent of some Neogene mudstones (Claridge, 1960), but illite is the dominate clay in soil samples from steep-land fields (Officer et al., 2006). The Waipaoa River also generates high yields of suspended sediment, averaging 14.66 Mt/y (Hicks et al., 2011); its discharge should supply illitic mud to the Poverty shelf and slope (Orpin, 2004), as well as Māhia Canyon. In summary, the typical clay mineral assemblage of Tuaheni, with its high proportions of both smecite and illite, appears to be a product of blending among suspensions emanating from multiple sources, with transport by the WCC and ECC.

4.2 Implications for Slope Stability

Rapid accumulation of low-permeability, muddy sediments in environments such as Hikurangi often contributes to overpressured conditions (e.g., Micallef et al., 2015; Stoecklin et al., 2017), and fluid pressures can be exacerbated by trapping or seepage of free gas (e.g., Crutchley et al., 2022; Faure et al., 2006; Gross et al., 2018). Contourite drifts elsewhere are known to be susceptible to failure due to localized oversteepening of the seafloor, scouring by currents, and/or pronounced climate-induced changes of lithology (e.g., Gatter et al., 2020; Laberg et al., 2016; Miramontes, Garziglia, et al., 2018; Miramontes, Sultan, et al., 2018; Prieto et al., 2016). The Hikurangi margin, in addition, has a propensity for moderate to large subduction-induced earthquakes (Cubrinovski et al., 2022; Dowrick & Rhoades, 1998), so it is logical to expect dynamic loading coupled with transient increases of fluid pressure (e.g., Stegmann et al., 2007).

No matter how sedimentologists might debate detrital provenance and routing paths, the abundance of clay and the composition of clay minerals (especially smectite) play key roles in modulating the frictional strength of Tuaheni sediments. Coefficients of internal friction decrease from 0.8 to 0.6 or less once the proportion of smectite exceeds 25%–30% of the bulk mineralogy, and they degrade to 0.2 or less if smectite concentrations rise above 70% (e.g., Logan & Rauenzahn, 1987; Lupini et al., 1981; Tembe et al., 2010). We utilized the results of Tiwari and Marui (2005), which are based on tests of both natural specimens and standard mineral mixtures, to make informed predictions for Tuaheni strata. Using the empirical relation for binary quartz-smectite mixtures, those %-smectite values translate to predicted angles of internal friction that range from an average of 14.2° to a minimum of 7.8° (Figure 3b). The three-component quartz-smectite-kaolinite group (Tiwari & Marui, 2005) provides a more realistic match for Hikurangi bulk sediment. Using that curve, our calculated values for bulk %-smectite translate to predictions of 5.9° (average) and 3.8° for the minimum angle of internal friction (Figure 3b). Equivalent frictional coefficients are <0.2.

Crutchley et al. (2022) contended that the basal shear zone of the Tuaheni slide exploited a pre-existing weak layer in an otherwise monotonous interval of strata, but the specific reason remains open to question. Weak layers elsewhere are known to occur in diverse lithologies (Gatter et al., 2021). In some examples, tephra beds act as weak layers; their suggested causes include fabric rearrangement and volume reduction during shearing (Harders et al., 2010), alteration of the volcanic glass to smectite (Miramontes, Garziglia, et al., 2018; Miramontes, Sultan, et al., 2018), or the tephra's role in focusing transient perturbations of pore-fluid pressure (Kuhlmann et al., 2016; Wiemer & Kopf, 2015). Some weak layers evidently predate failure and function as the preferred glide planes, whereas others form during failure events due to realignment of grain fabric (Gatter et al., 2021; Locat et al., 2014). In other instances, contrasts of hydrogeologic properties across a pronounced lithologic boundary allow pore pressure to build up along a potential failure plane (e.g., Stegmann et al., 2007). Dutilleul et al. (2022) showed that the base of the Tuaheni landslide coincides with a zone of anomalous porosity, which they attributed to a combination of lithology and elevated pore pressure.

Surprisingly, we found no evidence at Site U1517 for pronounced compositional variations within or across any of the weak layers or inferred slip surfaces. We attribute the compositional homogeny to blending of suspensions by strong margin-parallel currents. The clays are no different across the slip surface at ∼41 mbsf, across the base of the creeping slide interval (as defined by seismic-reflection data at ∼59 mbsf), or within the weak layer documented by shipboard measurements of shear strength at ∼31 mbsf (Figure 2a). None of those intervals coincides with a bed of volcanic ash, or unusually high amounts of total clay, or unusually high concentrations of smectite (Figure 2a). Instead, we only see a subtle facies transition between the underlying facies of mostly mud (unit IV) and the overlying facies of mud with stacked couplets of silt and mud (unit III). That facies change, from a more-hemipelagic character to more contourite-like, probably resulted from climate-induced variations in sediment supplies and current strength. Temporal shifts in ECC behavior might have also created subtle variations in the sediment's microfabric. The demonstrably weak layer at ∼31 mbsf, on the other hand, occurs in the middle of unit III, without any obvious facies change or anomaly in sediment texture (Figure 2a). That weak layer probably developed during a slip event via realignment of phyllosilicate grain fabric (Crutchley et al., 2022).

We acknowledge that direct tests of frictional properties are warranted to verify our contentions, but clay composition seems to have preconditioned the Tuaheni strata to fail without layer-specific weaknesses inherited from the ambient stratigraphy. The comparative stability of coeval, and compositionally similar, deposits at Site U1519 also indicates that failures are less likely where seafloor gradients are flatter (Figure 1d). If we accept the validity of the ECC as the principal path for routing smectite toward the SW, then it follows for slope stability to improve “downstream” where more detrital illite has entered the dispersal system from the Waipaoa and kindred watersheds (Figure 1a). Our XRD data from piston/gravity cores are consistent with that interpretation (Figure 3a). The frequency of landslide scars decreases where proportions of illite and chlorite are higher (Watson et al., 2020), and in the opposite (“upstream”) direction, landslide scars are more concentrated where closer to the volcanic sources of smectite-rich clay (Figure 3a).

5 Conclusions

Submarine slopes fail in response to many interwoven variables, including mineralogy, and those variables combine in ways that are unique to each individual landslide. The homogeny of mud across the Hikurangi trench slope is a byproduct of strong contour currents blending suspensions from multiple sources. Concentrations of smectite are high enough to reduce the bulk mud's angle of internal friction to an average of ∼6° and a minimum of ∼4°. Compositional excursions are not evident along any inferred slip surfaces or weak layers within the Tuaheni landslide. Smectite abundance and slide scars both decrease toward the SW, however, in the “downstream” direction of the ECC. Our results reinforce the notion that sediment-dispersal systems can contribute, though perhaps inconspicuously, to strike-parallel changes in slope instability by imparting changes in composition. That possibility validates the wisdom, in any holistic investigation of landslides, of testing for compositional preconditioning.

Acknowledgments

This study used samples provided by the International Ocean Discovery Program and repositories at Lamont-Doherty Earth Observatory, Oregon State University (supported by NSF Grant number OCE-1558679), and the New Zealand National Institute of Water and Atmospheric Research. Funding was provided by the U.S. Science Support Program. For piston/gravity cores, the authors are indebted to Laura Wallace for sharing money from her grant (New Zealand MBIE Endeavour Research Programme contract C05X1605 awarded to GNS Science). The authors thank crew members on JOIDES Resolution, technicians, and fellow shipboard scientists for assisting with sample acquisition. Karissa Rosenberger and Mercedes Salazar helped with sample preparation. Kelsey McNamara completed XRD scans at the New Mexico Bureau of Geology and Mineral Resources.

    Data Availability Statement

    Tabulations of X-ray diffraction data, methods, and error are accessible in data reports distributed by the International Ocean Discovery Program: https://doi.org/10.14379/iodp.proc.372B375.201.2020; https://doi.org/10.14379/iodp.proc.372B375.203.2020; https://doi.org/10.14379/iodp.proc.372B375.209.2022; and https://doi.org/10.14379/iodp.proc.372A.201.2021.