Mountain ridges in the western Yakutat microplate are riddled with swarms of antislope scarps and troughs. These landforms were previously interpreted as gravity failures (sackungen), but if partly or wholly tectonic (flexural slip or bending moment faults), they represent part of the strain budget from ongoing plate collision. To determine scarp origin, we mapped landforms, bedrock structure, and trenched scarps at Kushtaka Mountain and south of Martin Lake. The Kushtaka scarps paralleled west-dipping coal and sandstone beds in the Tertiary Kultieth Formation, and were spaced 35–60 m apart with heights of 1–4 m. Structural and paleosol relationships in a 6-m-long, 1.3-m-deep trench indicate the scarp was produced by Holocene creep on a normal fault underlying the scarp. The Kushtaka scarps thus represent toppling-style slip on bedding-plane faults in the eastern limb of a syncline, as the fold “unfolds” due to gravitational spreading. The Martin Lake scarps are more complex and include downslope-facing landslide scarps, upslope-facing flexural toppling scarps, and an oblique-slip tectonic scarp. A 2-m-deep trench across the WNW-trending tectonic scarp exposed the underlying bedrock fault plane with slickensides raking only 17°–20°, indicating mainly left-lateral slip on a sinistral-normal fault. In contrast, swarms of ENE-trending antislope scarps showed normal-oblique (dextral) slip. The two scarp sets may form a conjugate pair that simultaneously accommodates left-lateral tectonic slip and NNE-directed gravitational spreading. Our results, plus those recently published by Li et al. (2010), show that most antislope scarps in the western Yakutat microplate are formed by normal slip on bedding-plane faults that dip into the mountain, and represent gravitational spreading expressed as toppling-style failure or “unfolding” of strata on fold limbs. Conversely, the sinistral-oblique slip fault at Martin Lake is one in a family of east-trending faults accommodating accretion of the Yakutat microplate into the cuspate syntaxis of the Alaskan plate margin.

Uphill-facing fault scarps are exceptionally abundant in the Saint Elias orogen of southern Alaska, where the Yakutat microplate is colliding into the North American plate margin (Plafker, 1987; Plafker et al., 1994; Bruhn et al., 2004; Plafker and Thatcher, 2008). The origin of the scarps has major implications for evaluating the active tectonics and earthquake potential of the orogen (Bruhn et al., 2004; Plafker and Thatcher, 2008). Individual scarps are usually less than one kilometer in length, occur either along or subparallel to sedimentary layering that dips steeply into the mountainside, and have cumulative displacements that range from a few centimeters to ten meters. Most scarps occur in groups several tens to hundreds of meters in width and up to several kilometers long. Proposed mechanisms for their origin include lateral spreading of hillslopes by gravitational forces and ground shaking during earthquakes (Plafker and Thatcher, 2008; Li et al., 2010), disruption of bedding in landslides, and active folding that concentrates stresses by bending and shearing of bedding within folds (Carver and McCalpin, 1996; Bruhn et al., 2004).

Resolving the origin of the surface ruptures is important to understanding the neotectonics because gravitational collapse scarps, or “sackungen,” are not tectonic in origin, although they may reflect slope deformation caused by strong ground motion during earthquakes. Alternatively, bending moment and flexural-slip scarps caused by folding provide important constraints on the nature and location of active tectonic structures. Here we present the first detailed trenching and geochronology study of Quaternary scarps in the western part of the Saint Elias orogen, using the results to test regional tectonic models of plate interactions. The results reveal the styles and timing of faulting, evidence for episodic versus creep styles of deformation, and the origins of the scarps based upon detailed geological observations. We close by pointing out the implications of the study for understanding the tectonics and surficial processes that are active during microplate collision and accretion.

The scarps investigated in this study are located in the western segment of the Saint Elias orogen, where the Yakutat microplate is colliding into Alaska along the northeastern part of the Aleutian megathrust (Fig. 1; Plafker, 1987; Plafker et al., 1994). In this region collision of the microplate created a structural syntaxis, deforming the suture fault from a roughly E-W trend along the Chugach–Saint Elias fault to NNE trend along the Ragged Mountain fault (Fig. 1; Bruhn et al., 2004). Tectonic indentation of the plate margin also reoriented and partly refolded earlier developed folds and faults within the sedimentary rocks of the Yakutat microplate, as shown schematically by the structural form lines on the map of Figure 2 (Winkler and Plafker, 1993; Bruhn et al., 2004). Glacial erosion and sculpturing of Tertiary sedimentary rocks deformed by thrust- to oblique-slip faulting and folding creates elongated mountain blocks that are surrounded either by flat-floored valleys filled with glaciofluvial and marine sediments, steep canyons, or in some cases glaciers. The mountains are susceptible to mass wasting because of steep glacial-carved mountain slopes, rapid retreat of many glaciers from mountain walls, unusually heavy precipitation, and strong ground motion triggered by earthquakes (Plafker, 1987; Meigs and Sauber, 2000; Jaeger et al., 2001; Bruhn et al., 2004).

The Yakutat microplate is moving NW to NNW at »45–49 mm/yr with respect to interior Alaska (Fletcher and Freymueller, 1999), creating considerable seismicity and the generation of some of the world's largest earthquakes along the Aleutian megathrust (Plafker, 1969; Shennan et al., 2009). Earthquake focal mechanism solutions indicate a predominantly mixed strike-slip to thrust faulting stress regime in the western Saint Elias orogen (Doser et al., 2007; Ruppert, 2008). Rupturing of the Aleutian megathrust during the M9.2 earthquake of March 1964 produced coseismic uplift of the coast line, and also triggered many landslides, snow avalanches, and widespread surficial deformation of Quaternary deposits (Plafker, 1969; Tuthill and Laird, 1966). Recently, geodetic surveying has revealed velocity gradients within the interior of the microplate caused by strain accumulation on buried faults that also represent potential seismic sources (Elliott et al., 2007).

Given the abundant evidence for active deformation, there is a remarkable paucity of Quaternary scarps that are obviously associated with major faults. Tysdal et al. (1976) mapped an »30-km-long “normal fault” scarp along the eastern flank of Ragged Mountain to the west of our study area. They inferred the scarps represented backsliding by reactivation of the tectonic suture between the Yakutat microplate and North America (Fig. 1). No other fault scarp of similar length has been identified, even though regionally stresses are favorable for reverse and strike-slip faulting (Ruppert, 2008). This focuses attention on uphill-facing or “antislope” scarps that occur within mountains throughout the region. Most of these scarps parallel bedding planes in Tertiary strata, and offset Quaternary deposits. Are these scarps caused by superficial deformation, or are they manifestations of tectonic processes related to active folding above buried faults (e.g., Carver and McCalpin, 1996)? This question motivated neotectonic research as part of the Saint Elias Erosion and Tectonics Project (STEEP), a five-year multidisciplinary program to study the Saint Elias orogen funded by the U.S. National Science Foundation (NSF).

Glacial Geology and Geomorphology

The geomorphology of the western Yakutat microplate is dominated by the effects of glaciation in the Last Glacial Maximum (LGM) and in a subsequent post-LGM advance (Holocene?). Based on our reconnaissance observations and the work of Fleisher (1999), LGM ice from the Martin River glacier flowed south down three valleys between Ragged Mountain (on the west) and Kushtaka Lake (on the east; Fig. 3). The westernmost of these valleys contains modern Martin Lake, and its LGM ice flowed south over the present divide into the Katalla River valley. The ice surface was ∼300 m elevation at Martin Lake, but LGM (?) ice also flowed over a saddle at ∼400 m between Little Martin Lake and Lake Tokun, which added to this valley glacier. The Martin River–Katalla paleoglacier descended gently toward the Katalla River valley and continued south beyond the present coastline. Last Glacial Maximum ice also flowed over a divide at ∼365 m elevation south of Lake Tokun, and down the valley containing the present Lake Charlotte, as well as down the valley containing Kushtaka Lake. All these ice streams coalesced in the wide plain north of the Don Miller Hills.

In this study we performed detailed mapping and trenching studies of antislope scarps in two study areas in the western part of the Yakutat microplate. The Kushtaka site lies on Kushtaka Mountain, 39 km east of the east bank of the Copper River, and 88 km east of Cordova, Alaska (Figs. 2 and 3). It lies 23 km ENE of the Ragged Mountain fault, which is commonly accepted as the on-land western margin of the accreted Yakutat microplate. Kushtaka Mountain is bounded on the north by the Martin River Glacier, which flows WSW from a high ridge in the Chugach Range dominated by Mount Tom White (elevation 3411 m).

The Martin Lake site lies on the north flank of an unnamed 750-m-high mountain ridge that separates Martin Lake and Bering Lake (Figs. 2 and 3). We informally call this “Tok Ridge” after the Tok bench mark at its summit. Our study site lies on the western flank of the ridge at an elevation of ∼400 m, near the upper tree line. Tok Ridge was a minor ice source during the LGM, based on the existence of cirques on its northern side with floors at 600–800 m elevation. However, the local landscape was dominantly shaped by large LGM valley glaciers that flowed south from an expanded Martin Lake glacier.

Kushtaka Mountain lies between Lake Charlotte (to the west) and Kushtaka Lake (to the east), both of which are fed by meltwater from southern lobes of the Martin River Glacier (Fig. 2). The summit of the ridge rises to an elevation of 709 m (60.42894°N, 144.136°W), compared to the elevation of 25 m for Kushtaka Lake (relative relief of 684 m). The southwestern end of the ridge is split into two ridges by Carbon Creek, with Carbon Ridge lying to the west and Kushtaka Ridge to the east. These lower ridges have less local relief above valley bottoms, amounting to 183 m for Carbon Ridge and 213 m for Kushtaka Ridge.

Stratigraphy and Structure

The bedrock geology of Kushtaka Mountain is complex due to a combination of stratigraphic facies changes and multiple generations of structure. Stratigraphically, the Kushtaka Mountain area lies in a facies transition between Tertiary fluvial to fluvial-deltaic strata to the east and deltaic to delta-front–facies rocks to the west, complicating lithostratigraphic unit definitions due to intertonguing of lithologic packages within the transition. Thus, at Kushtaka Mountain we recognize four distinct lithostratigraphic units (Fig. 4), from oldest to youngest (Plafker, 1987 and this study): (1) the Eocene Stillwater Formation, defined by Miller (1961) is poorly exposed shelf to delta front–facies shales; (2) the Eocene Kultieth Formation, comprised of nonmarine to marine sandstone, shale, and siltstone with coal representing the fluvial-deltaic facies; (3) the Eocene–Oligocene Tokun Formation composed of thin to medium, rhythmically bedded sandstones, siltstones, and shales that are interpreted as delta-front deposits; and (4) a unit we refer to here as the Kushtaka sandstone, a distinct sandstone unit that is similar to the Kultieth facies, but apparently represents a sandstone tongue within the Tokun Formation.

The structure of Kushtaka Mountain is relatively simple overall, but is complex in detail. The general structure from Carbon Ridge to Charlotte Ridge (Fig. 4) is a steeply northwest-dipping homocline in the Kultieth and Tokun Formations. This homocline lies in the hanging wall of a thrust that is approximately coincident with Carbon Creek and duplicates the section on Kushtaka Ridge where a footwall syncline is also developed along the thrust. To the northwest, a second syncline is formed in Tokun Formation and Kushtaka sandstone, directly beneath a structurally higher thrust mapped by Miller (1961) that places broadly homoclinal Stillwater Formation and overlying units atop the Kushtaka section.

In detail, the structure at Kushtaka Mountain is quite complex because of: (1) numerous small-scale detachment folds with wavelengths of 100–200 m, that are typically limited to distinct stratigraphic levels, and (2) a series of younger faults that cut the older fold-thrust structures, and form a complicated fault network. Some of these faults are marked by scarps, indicating they are either tectonically active faults or older faults that are partly reactivated by gravity and perhaps strong ground motion during earthquakes.


Above an elevation of ∼520 m (1700 ft), Kushtaka Mountain is sculpted into a series of broad local cirques. Below that elevation, including our detailed study site (400–460 m elevation), ridges such as Carbon Ridge and Kushtaka Ridge do not show landforms of cirque glaciation. Each ridge has a relatively broad, rounded crest with a steeper SE-facing slope and a gentler NW-facing slope. This asymmetry is probably caused by the NW dip of strata on both ridges, such that the SE-facing slope is the scarp slope and the NW-facing slope is the dip slope. Superimposed on these asymmetric ridges are many small (1– 4-m-high) contour-parallel scarps and troughs (Fig. 5).

Quaternary Scarps and Troughs

The flanks of Carbon Ridge contain numerous antislope scarps and associated troughs that generally parallel contours, spaced 35–60 m apart. In our detailed study area (Figs. 5 and 6), there are nine such landforms, the highest and longest of which is the uppermost one. That scarp is 460 m long, up to 4 m high, and impounds a lake. In our pre-field logistical planning, we had chosen this scarp as our primary trenching target, but it turned out to be too high for hand-trenching, plus partly snow covered and partly submerged by the lake. The remaining antislope scarps range from 40 to 170 m long and 1 to 2 m high. Scarp slope angles are low (<25°), and none of the scarps displays free faces. The scarp face is covered with the same mat of peaty turf (tundra) that covers the trough surface and the interscarp slopes.

There are two other linear, structurally controlled (?) topographic elements in the study site. The first is a pair of linear incised gullies (green in Fig. 6) that do not flow straight downslope to the SE, but instead flow S diagonally across the slope. These “diagonal gullies” coincide with the termination of two sackung scarps including the second-longest one at 250 m, and bound an area of higher density of scarps. We infer that these gullies are eroded along cross-strike faults in the Kultieth Formation and that the faults bound domains of variable amounts of gravitational spreading. The second element is a series of three ENE-trending, downslope-facing scarps (purple in Fig. 5). The northern of these forms the northern boundary of a downdropped block (graben?). The fact that these scarps are parallel to each other, and two of them terminate against either sackung scarps or the diagonal gullies, suggest they too are structurally controlled. From crosscutting relationships, they are older than the antislope sackung scarps. We infer that the scarps are underlain by south-dipping faults in bedrock, and may represent incipient gravitational spreading that is older than or partly contemporaneous with movement on the antislope scarps.


At the trench site the sackung scarp is ∼1 m high and is asymmetric in profile (Supplemental Fig. 11). The steepest part of the scarp is the lower half, which is a relatively planar face sloping 22°N. In contrast, the upper half of the scarp is gentler and more rounded into a convex-upward shape. Such asymmetric profiles are commonly observed on “compound fault scarps” created by multiple faulting events rejuvenating the lower part of a normal fault scarp (e.g., McCalpin, 2009, p. 215–216). The trough floor at the base of the scarp is flat and undissected. We chose this location to trench because it appeared to have been occupied in former times by a closed depression, so that the trench might expose lacustrine or marsh stratigraphy with associated datable organic material.

Trench Stratigraphy and Structure

The trench exposes Kultieth Formation bedrock in the deepest part beneath the scarp, and up to 1.1 m thickness of overlying Quaternary sediments (Fig. 7). The Kultieth Formation exposed beneath the scarp on the fault footwall is mainly coal, but contains interbeds of semiconsolidated sandstone ∼30–45 cm thick. Beds dip steeply to the NW. We subdivided the Quaternary deposits into three major units and 12 subunits, not including the four pedogenic horizons of the modern soil and the modern peat ground cover (Fig. 8).

The fault zone is exposed beneath the scarp and consists of four fault strands that displace the top of Kultieth Formation bedrock down to the north a total of ∼75–100 cm. All four faults converge at the bottom of the trench, ∼1.3 m below the ground surface. Although this pattern superficially resembles the flower structures seen on strike-skip faults, it matches neither the standard geometry of a positive or negative flower structure (e.g., Twiss and Moores, 1992). Instead, it appears to be a simple case of refraction of the upper part of a normal fault due to decrease of confining pressure near the ground surface (McCalpin, 2009, p. 217).

Bedrock in the footwall strikes between N23°E and N38°E and dips 80°–85°NW, compared to strikes and dips measured elsewhere in the middle of the trench (N34°E to N50°E, dips 58°–88°NW). These strikes and dips encompass the range of strikes and dips for the faults exposed in the trench (omitting the refracted parts). Therefore it appears that the fault zone is mainly a bedding-plane fault with a normal sense of slip.

Trench Geochronology and Deformation History

Units 2c and 3d contained enough disseminated organic material to yield accelerator mass spectrometry (AMS) dates of 9920–10,200 cal yr B.P. and 6670–6800 cal yr B.P., respectively. These postglacial ages are in correct stratigraphic order and seem reasonable in light of the geomorphic freshness of the scarp and trough landforms. Unit 2c yielded an optically stimulated luminescence (OSL) age of 46.2 ± 13.7 ka, compared to its C-14 age of ca. 10 ka. The OSL age considerably predates the latest glacial maximum, which contradicts the youthful (post-LGM) appearance of the landform, as well as the C-14 age of ca. 10 ka from the same unit. We suspect this sample contains an inherited, predepositional luminescence signal that makes its apparent OSL age too old. Likewise, unit 2b directly beneath unit 2c yielded an OSL age of 170 ± 74 ka. This age is extremely unlikely, because the weak development of the unit 2c paleosol does not allow for >100 ka of soil formation, and it is unlikely that 170 ka deposits could have survived the glaciations in marine isotope stages 2 and 6, both of which presumably overrode this elevation on Kushtaka Mountain. Apparently this poorly sorted diamicton contained even more inherited luminescence than did unit 2c.

We infer the deformation history relying on principles elucidated from tectonic normal faults that have experienced sudden meter-scale displacement events (e.g., McCalpin, 2009). Such episodic displacement in a trench exposure is indicated by: (1) existence of open fissures or fractures in the fault zone, (2) coarse fissure fills and colluvial wedges, indicating rapid scarp-derived deposition, and (3) differential displacements of stratigraphic markers. If we apply this standard paradigm to the trench log (Fig. 7), we can identify three displacement events based mainly on crosscutting relationships and differential fault displacements.

Displacement event 1 (oldest)—formed the initial topographic depression, into which a small stream deposited units 1a (axial channel fill) and 1b (slopewash-eolian facies); the axis of the trough lay north of the present axis. This event must predate 10.2 cal ka but postdate the retreat of glacial ice from this elevation on Kushtaka Mountain.

Displacement event 2—deformed units 1a and 2a at the north end of the trench, but did not deform overlying unit 2b. Unit 1b was faulted up on the footwall and then eroded away. The sum of displacements in events 1 and 2 is 35 cm; event 2 is constrained to the same time interval as cited above.

Displacement event 3 (youngest)—displaced unit 2c by 40 cm, and was immediately followed by deposition of unit 3. This event thus occurred soon after 9920–10,200 cal yr B.P., but before 6670–6800 cal ka (age of unit 3d).

However, the total displacement on our trenched sackung failure plane was only 1 m in multiple events and could arguably have included a large component of creep movement (sagging). Therefore, we examine the alternative hypothesis that the observed geometry was created by creep movement rather than by episodic displacement. In general, creep displacement is typified in section by a lack of fissures and debris-facies colluvium (classification of Nelson, 1992), a gradual increase in displacement downward in the stratigraphic section, and a degree of plastic deformation (McCalpin, 2009, p. 242). The trench does not expose any fissures or debris-facies colluvium, and there is evidence for both large-scale and small-scale plastic deformation, so those aspects meet two of the three criteria for creep deformation. Note that differential displacement of stratigraphic markers, which is often taken as definitive evidence for episodic displacement, can also be caused by creep movement if deposition in the trough is episodic rather than continuous. Therefore, we explore this latter possibility below.

In general, there are four possible subsurface geometries in a scarp-trough system, depending on whether displacement beneath the scarp is episodic or continuous, and whether deposition in the trough is episodic or continuous (Fig. 8). The distinguishing features are angular unconformities in the trough stratigraphic section and the presence or absence of paleosols. Note that all cases except (A) contain packages of sediment that are deformed by clearly different amounts. Figure 8 shows that a large amount of differential displacement occurs between paleosol unit 2c and underlying units. The trench geometry thus best matches geometry “c” in Figure 8, created by episodic deposition and continuous creep. Thus the structural and stratigraphic indications in the Kushtaka trench suggest that the sackung scarp was formed mainly by creep movement accompanied by episodic deposition and soil formation in the adjacent trough.


Geologic mapping and trenching indicate that the trenched sackung scarp is underlain by a bedding-plane fault dominated by a creep style of movement. Located on the eastern limb of a syncline where beds dip steeply into the hillslope, this style of deformation can be explained by downslope toppling of beds accompanied by normal slip on bedding planes. Such a mode of deformation was interpreted elsewhere in the Yakutat microplate by Li et al. (2010), based on a combination of field observations and stress modeling. Another way to view this style of deformation is that the syncline is “unfolding” on its eastern limb due to gravitational relaxation and spreading. This unfolding results in domino-style (or bookshelf-style) normal fault displacements on many bedding planes.

The Martin Lake trenching site is located at an elevation between 350 m and 450 m on the northern side of “Tok Ridge” at 60.3322°N, -144.4377°W (Figs. 2 and 3). We chose this site for detailed study based on brief field and aerial reconnaissance and examination of 1:20,000 stereo airphotos, which showed well-preserved antislope scarps and anomalously wide flat areas (sag ponds) that might contain deposits with datable organics. However, we did not originally consider why these sag ponds were much wider than typical sag ponds on the upslope side of antislope scarps. We assumed that all the scarps were sackungen, all of the same type, and all formed by downslope gravitational spreading. The actual situation turned out to be more complex.

Stratigraphy and Structure

Bedrock at the Martin Lake site is fine-grained siltstone and sandstone of the Tokun Formation that is deformed into a syncline, with steeply north- and south-dipping limbs (Fig. 9). Antislope scarps mark several fault strands in the western limb of the fold, where bedding dips 50° to >85° south. The syncline is truncated below by a moderately north-dipping décollement, beneath which bedding dips at moderate angle to the north and northeast. The décollement presumably formed during an earlier episode of thrust faulting within the Tokun Formation.

The syncline is cut off to the north by a steeply south-dipping, west-northwest–striking fault that extends downwards and also truncates the underlying décollement. This fault strikes WNW and extends across the canyon and onto the mountainside to the east of the trenching site. Quaternary deposits are faulted against the bedrock fault surface, which is marked by east-plunging corrugations formed of smeared-out unconsolidated deposits and grooves along the bedrock fault surface. The corrugations plunge ∼17°–20° east, requiring left-oblique slip, with the south side down, to form this predominant antislope scarp.

Several smaller antislope scarps are located on the mountainside uphill from the WNW-trending fault. These scarps occur along small faults that parallel the south-dipping bedding surfaces and are truncated by the larger WNW-trending fault at an oblique angle. Computer visualization of 1 m ground-resolution light detection and ranging (LIDAR) elevation data as hillshade images indicates that these smaller scarps occur along faults that are truncated against the north- to northeast-dipping décollement that underlies the syncline. This implies partial reactivation of the older thrust fault décollement by gravitational forces (e.g., see fig. 9 in Pavlis and Bruhn, 2010).

Quaternary Scarps and Troughs

The Martin Lake study area contains two types of Quaternary scarps. Six long scarps are mapped as the expression of tectonic faults (red on Fig. 9). These scarps range from 0.8 to 1.8 km long, trend E-W to ESE-WNW, generally face upslope, but cross valleys and ridges indiscriminately. Thus, they are not primarily controlled by local topography, and are also associated with mappable offsets in the underlying stratigraphy. However, the more common scarps are a second type of shorter scarp that is confined to a single ridge and does not extend across valleys (green on Fig. 9). We show ∼60 of the more prominent scarps on Figure 9, as interpreted from the 2 m LIDAR digital elevation model (DEM) acquired during the STEEP research project. The scarps have one of three orientations: (1) parallel to the tectonic faults, (2) diagonal to tectonic faults (i.e., NE-SW) and truncated by them, as in our detailed study area, and (3) in the SW part of Figure 9, NW-SE.

Our detailed study area contains a part of one long scarp (F) and eight gravitational scarps (A–E, G, and I; see Fig. 10). Tectonic scarp F trends about N80W, averages 2–4 m high, and is the steepest of the scarps (Supplemental Fig. 22). The microtopography north of scarp F is fundamentally different from that south of scarp F (Supplemental Fig. 33). North of scarp F are three NW-trending, downslope-facing (G and I) or upslope-facing scarps with rotated bench landforms below. These scarps lie at the head of a very high and steep bedrock slope that descends down to the LGM trim line at ∼275 m elevation. In addition, scarps F, G, and I have nine fresh sinkholes lined up along them. These sinkholes resemble those created by piping into subsurface fissures, such as commonly observed in highly extended terrain at the heads (pull-away zones) of landslides. This coincidence suggests that the entire hillslope north of scarp F is pulling away from the mountain and opening up fault-related voids into which surface water is leaking rapidly downward. Vertical movement on the scarps north of scarp F is sufficiently large that we cannot visually reconstruct the pre-faulting topography, to determine if there is a lateral component of slip on the scarps.

In contrast, the five antislope scarps that lie south of scarp F (scarps A–E) are smaller, are not associated with sinkholes, and the network of erosional hillslope ridge crests and gullies can be mapped and correlated across them (Fig. 10). These antislope scarps are linear, trend N75E, and range from 0.5 to 3 m high (Supplemental Fig. 3 [see footnote 3]). We map two scarp types: (1) small (0.5–2 m) steep scarps that are clearly Holocene (or latest Pleistocene) in age; and (2) taller, more degraded scarps that reach up to 2.5–3 m. Many of the latter scarps show a compound scarp profile, with a narrow band of steeper slope crossing the larger scarp face (e.g., Supplemental Fig. 3D [see footnote 3]). On tectonic normal faults this morphology is typically caused by a recent displacement event, which rejuvenates the scarp (e.g., McCalpin, 2009, p. 215–216). In some cases (e.g., western part of scarp E), small young scarps can be traced onto the face of a larger, more degraded scarp, where they merge with the rejuvenated part of the scarp face, indicating a recent reactivation. All of these scarps parallel bedding in the Tokun Formation (Fig. 11; Supplemental Fig. 44).

Offsets of erosional ridge crests clearly show that scarps A–E have a right-lateral component of movement (Fig. 10), as well as a component down toward the mountain. The oblique component of movement in the overall scarp system has created three large, flat sag pond areas by shifting ridges laterally to block gully mouths. Sag pond 3 may also be a graben formed by eastward escape of the structural block bounded by a left-lateral (F) and a right-lateral (E) fault.

The west trench was located on antislope scarp D on the northern margin of sag pond 1, in the central western part of the detailed map area (Fig. 10; Supplemental Figures 2 [see footnote 2] and 55). The hand-dug trench was 6 m long, 0.8 m wide, and ∼1 m deep. This trench was dug to characterize the family of ENE-trending, reactivated antislope scarps (scarps A–E).

Morphology of the Scarp

At the trench site, scarp D faces sag pond 1, is ∼2.5 m high, and is asymmetric in profile (Supplemental Fig. 5 [see footnote 5]). Although most of the scarp face is a planar slope of ∼15°, the lower face is traversed by a 25–30-cm-high reactivation scarp with slopes up to 45°. This reactivation scarp is covered with the same turf mat as the higher scarp face, even though its slope is greater than the angle of repose. This combination suggests that the scarp did not ever possess an unvegetated free face, but instead deformed the turf mat in situ by folding. Below the reactivation scarp is a gently sloping, turf-covered apron sloping down to the floor of sag pond 1. This apron has relief created by two vague scarplets that are too small to show on Figure 10. These scarplets trend S65E, or ∼45° more southerly than the primary scarp.

We chose this location for our trench: (1) to see if there was a series of colluvial wedges under the reactivation scarp, from which a multi-event displacement history could be deduced, and (2) because the scarp toe extended into a closed depression, where lacustrine or marsh stratigraphy exposed in the trench might be associated with datable organic material.

Trench Stratigraphy and Structure

The trench exposes six Quaternary deposits and an overlying soil (A horizon); no Tertiary bedrock is exposed (Supplemental Fig. 66). There are two fault zones, both of which dip steeply to the south toward sag pond 1. The upper fault underlies the reactivated scarplet and is responsible for a 25-cm-high, down-to-the-south warp in the A horizon and underlying unit 3 (loess). Three lines of evidence suggest there was a significant right-lateral component of movement on this fault during its only recognized displacement event. First, loess unit 3 is missing below the scarplet, which is hard to explain by any erosional process on the scarp face. Second, the scarplet can be traced westward 25 m to where a terrace on the north side has been translated right laterally to partially block the outlet of sag pond 1, creating a swamp against the scarplet. Third, the fault plane is very difficult to see in massive unit 2b, something more typical of strike-slip than of normal faults.

The second fault zone is a 10–20-cm-wide zone of sheared material recognized by rotated clasts (shear zone on Supplemental Fig. 6 [see footnote 6]) and by the cm-scale, down-to-the-south normal displacement of units 2b, 2c, and 3. At the ground surface this fault zone barely has any vertical relief, despite the fact that the top of unit 2 is displaced ∼10 cm vertically.

Trench Geochronology and Deformation History

Aside from the surface A horizon, we did not observe any datable organic material in the west trench, so our geochronology relies on OSL age estimates (Table 1). We dated samples from units 2b (bottom), 3, and 4. Diamicton unit 2b yielded a basal OSL age of 74.2 ± 5.1 ka, but displayed multiple OSL age populations (Supplemental Fig. 77). If we exclude the De values of 170–190 Grays as representing unbleached grains in the diamicton, the younger secondary peak at De = 140 yields an age of ∼59 ka, which is probably closer to the true age of the sample. This pre-LGM age is consistent with the low stratigraphic position of the sample, which could arguably be marine oxygen isotope stage (MIS) 4 till or colluvium.

The thin loess (unit 3) yielded an OSL age of 14.6 ± 1.3 ka, based on good luminescence systematics (nearly a single age population) and a low standard deviation. This age falls between the 16 and 27 ka age of the LGM cited in Briner et al. (2005), and the local deglaciation age of 11–12 ka cited by Fleisher (1999). Unit 4 yielded an age estimate of 7.9 ± 0.6 ka, consistent with its position above unit 3 and its low degree of soil profile development.

The trench exposes evidence for only the two most recent displacements beneath scarp D. The most recent movement occurred beneath the reactivated scarplet and warped the A horizon of the modern soil and underlying unit 3 by ∼25 cm. This event may have been historic, possibly associated with ground shaking during the 1964 Mw 9.2 Alaskan earthquake. In contrast the latest displacement on the “shear zone” does not displace the A horizon, but does displace unit 3 (14.6 ka) and was followed by deposition of unit 4 (7.9 ka), bracketing it between 7.9 and 14.6 ka. Due to the poor stratification in the lower trench units, it is unclear which of the two exposed faults is responsible for most of the displacement beneath the larger, degraded scarp D.

The east trench was located on scarp F on the northern margin of sag pond 3, in the northeastern quadrant of the detailed map area (Fig. 10; Supplemental Fig. 2 [see footnote 2]). The hand-dug trench was 6 m long, 0.8 m wide, and up to 2 m deep at the fault plane (Fig. 12). The east trench was located near the center of scarp F where it was 3 m high and very steep (Fig. 10). We expected that any slope this steep would be underlain at very shallow depth by Tertiary bedrock. However, excavation revealed that the steepest slope segments were underlain by unconsolidated silt (loess).

Trench Stratigraphy and Structure

All of the trench except for the northern 0.5 m exposed unconsolidated Quaternary deposits, which we subdivided into five units (Fig. 12) The main fault in the east trench was a surprise for three reasons. First, it was considerably closer to the scarp crest than to the toe. Second, the fault plane was composed of relatively intact sandstone of the Tokun Formation, which rose nearly to the ground surface on the footwall. Third, the fault plane contained numerous inclined mullions and slickenlines indicative of oblique movement (Supplemental Fig. 88).

In hindsight, we realized that the fault plane coincided with the crest of the steepest scarp profile segment. The fault plane juxtaposed hard sandstone against Quaternary units 1, 3, 4, and 5. After exposing the plane, we measured an average rake of 17°–20°E on the pronounced slickenlines, indicating a left-lateral slip component ∼2.7 times larger than the normal slip component. In addition, two fault splays diverge from the main fault and propagate through hanging-wall Quaternary units, displacing units 1, 3a, and 3, and dying out in unit 4.

Trench Geochronology and Deformation History

Aside from the surface A horizon, we did not observe any datable organic material in the east trench, so our geochronology relies on OSL age estimates (Table 1). A thin “finger” of loess (unit 4) underlies the unit 5 colluvial wedge and yielded a basal OSL age of 32.9 ± 3.6 ka. This age was a surprise, because it is older than the LGM rather than younger, and is over twice as old as the loess exposed in the west trench, with which we had correlated it. However, the unit 4 sample displayed a bimodal distribution of De, unlike the sample from the west trench. This bimodality may have been caused by accidentally incorporating some unzeroed sand grains from the base of the overlying scarp-derived wedge (unit 5) into the sample. If this occurred, then the younger modal peak of De = 28 Grays may be a closer approximation of true sample age, which equates to 20 ka at a dose rate of 1.4 Gy/kyr (Table 1). This age does fall within the age range of the LGM cited by Briner et al. (2005), and is more consistent with the low degree of soil development above the sample.

The base of the modern A horizon yielded an OSL age of 57.3 ± 7.4 ka, older than the underlying sample, and even older than the LGM. However, the De distribution contained four modes, the highest number of any sample, and indicative of a heterogeneous mixture of different-aged materials. Even though this sample was from a soil A horizon, in which mineral grains are typically bioturbated and churned to the surface and reset for luminescence, this did not occur here. Instead, the sample was evidently collected from the organic-stained upper part of the scarp-derived colluvial wedge that contained at least four depositional components of different age, three of which had not been reset. Therefore, our best age estimate for this basal A horizon is that it is younger than the loess (ca. 20 ka here and 14.6 ka in the west trench), but older than historic.

Sedimentology and crosscutting relationships indicate four late Quaternary displacement events. Evidence for the earliest event (1) is indirect, and consists of the E-W–trending channel of unit 2. Event (1) must have created a scarp, a trough, or both, that deflected northward hillslope drainage to flow east, parallel to contours. A second event (2) is suggested by the presence of unit 3, a tapering wedge of poorly sorted gravel that pinches out away from the fault. The third event (3) is the least ambiguous, and was responsible for faulting of unit 4, and the deposition of unit 5, a colluvial wedge. Units 1–3 and the lower half of unit 4 are clearly displaced by the main fault and by its secondary splays in the hanging wall, making this event younger than the middle of unit 4. The same age range is indicated by the stratigraphic position of the colluvial wedge (unit 5), which overlies the lower half of the loess but underlies the upper half. This would tend to place event 3 in the middle of loess deposition, roughly 15–20 ka. The most recent event (4) is responsible for the main fault truncating the uphill edge of the earlier colluvial wedge (unit 5).

Origin and Significance of Quaternary Scarps

During the past several million years, the Aleutian megathrust has stepped eastward within the Yakutat microplate, from the western segment of the orogen where the Kushtaka and Martin Lake trenching sites are located, to the Pamplona zone and its onshore continuation, the Malaspina thrust system (Plafker, 1987; Chapman et al., 2008; Enkelmann et al., 2010; L.L. Worthington, 2010, personal commun.; Elliott et al., 2009, 2010). However, deformation continues throughout the orogen and is reflected in widely distributed seismicity and gradients in crustal displacement detected by GPS surveying (Bruhn et al., 2004; Ruppert, 2008; Elliott et al., 2009). Geologic evidence for active deformation includes numerous Quaternary fault scarps (Tysdal et al., 1976; Bruhn et al., 2004; Pavlis and Bruhn, 2010), and vertical deformation of the coast during megathrust earthquakes (Plafker, 1969; Shennan et al., 2009). The origin of the scarps is a contentious issue and difficult to resolve, but remains central to understanding processes of microplate accretion into the syntaxis at the western end of the Saint Elias orogen, and the seismic potential of the region. Displacement on the scarps at Kushtaka Mountain is clearly driven by gravity and caused by flexural toppling of the strata outward from the mountainside. The situation at Martin Lake merits further evaluation; we propose that the largest scarp (F), which occurs on a steeply dipping fault zone, is partly driven by tectonic shearing associated with accretion of the microplate strata, while the smaller scarps located uphill, reflect the onset of hill slope failure under gravity.

During previous research we identified two common origins for Quaternary scarps throughout the region—flexural toppling of steeply dipping beds in the middle to upper parts of mountain blocks and internal deformation of bedded strata within landslide blocks (Li et al., 2010; Pavlis and Bruhn, 2010). Flexural toppling scarps are less than a kilometer in length, occur in swarms of subparallel scarps on the upper slopes of mountain blocks, do not extend downslope into the valley, and parallel bedding where it dips steeply into the mountain. Classical examples are the scarps at the Kushtaka trenching site, which are discussed herein, and those studied by Li et al. (2010) in a mountain on the west side of the Katalla River Valley. Those scarps are not tectonic in origin, although flexural toppling may be enhanced by strong ground motion during earthquakes. There are also scarps located in the interiors of landslides. The scarps occur in coherent blocks of sedimentary rock that are embedded in hummocky topography on steep slopes below concave, downslope-facing head scarps. The scarps occur where strata are broken into a jumble of blocks and transported downslope in the slide mass. Unlike flexural toppling scarps, these include remnant landslide blocks with preserved flexural toppling scarps that formed at an earlier time (Li et al., 2010). Other scarps formed where bedding dips downhill, and the beds slide out into free space as though they are a deck of cards. The large landslide mapped by Pavlis and Bruhn (2010) on the western side of the mountain that hosts the Martin Lake trenching site is a spectacular example with scarps in jumbled blocks of transported strata.

The Martin Lake trenching site has similarities to localities where we have documented flexural toppling deformation but it is also unique because: (1) the largest scarp (F) is located along a steeply dipping fault that cuts across the mountain ridge, and (2) that fault dismembers a north-sloping thrust fault décollement, and extends across the canyon floor to the east. The fault extends to depth, unlike the smaller bedding parallel faults that are located upslope, and truncate against the older thrust fault décollement that is exposed on the western side of the mountain. Clearly, fault F originated by tectonic rather than purely gravitational processes, and the question is whether Quaternary motion is still driven, at least in part, by tectonics? The latest motion on the fault is dominantly left lateral, with a minor component of vertical slip. Lateral shearing may of course occur along the margins of landslide blocks, and Li et al. (2010) noted that complex mountain topography may also rotate principal stresses and cause oblique instead of purely downdip slip during flexural toppling. However, geologic mapping provides no evidence that the large left-oblique slip fault scarp marks the lateral edge of a landslide block (Fig. 9), and the high ratio of strike-slip to vertical displacement is larger than predicted by rotation of principal stresses in complex topography (e.g., Li et al., 2010). This opens the possibility that the scarp is generated by sinistral tectonic shearing, and represents but one of a number of late-stage, east-trending faults that we have mapped in the region (Pavlis and Bruhn, 2010). Arguably, the scarp is hybrid in origin, that is, outward spreading of the mountain slope under gravity is occurring as proven by flexural topping and downslope sliding above the older thrust fault décollement, but such spreading does not account for the dominant sinistral component of faulting where fault (F) extends through and offsets the shallow décollement preserved in the mountain block.

Regional Structure and Tectonics

Detailed mapping in the western segment of the Saint Elias orogen reveals that three styles of superimposed structures occur within the sedimentary rocks of the Yakutat microplate (Bruhn et al., 2004; Pavlis and Bruhn, 2010), providing a regional context in which to place the Kushtaka and Martin Lake sites. The first and earliest structures are concentric-style folds with horizontal to gentle plunging hinge lines and associated thrust faults that form fault ramps and décollement; the second style is characterized by plunging folds that warp and rotate the earlier fold-and-thrust structures, and the third is steeply dipping faults, some of which have postglacial scarps (e.g., see Figure 7 of Pavlis and Bruhn, 2010). Of these latter faults, the most dominant in terms of number and exposure are east-trending faults similar in orientation to the left-oblique slip fault we trenched at the Martin Lake site. The faults are up to several kilometers in length, cut across instead of parallel mountain ridges, and in several cases dismember fold-and-thrust structures by cutting at high angle across the folded bedding and remnant thrust faults. These steeply dipping faults are tectonic in origin, and formed late in the history of deformation. We propose that they are associated with sinistral shearing during second-phase folding and late-stage accretion of the previously deformed strata into the structural syntaxis in the plate margin.

Deformation at the western end of the Saint Elias orogen is marked by second-phase folding of earlier foreland fold-and-thrust faults about moderately to steeply plunging fold axes, which is either accompanied by, or postdated by, faulting (Bruhn et al., 2004). This sequence of events is consistent with two regional, but not necessarily mutually exclusive, models for accretion of the sedimentary cover of the Yakutat microplate into the syntaxis in the southern Alaska plate boundary: (1) a “train wreck” model, in which the sedimentary cover rocks are detached from the underlying crystalline basement of the microplate, rotated counterclockwise, and extruded offshore toward the Aleutian trench; or (2) an “indentor” model, in which the cover rocks of the microplate are forced into the syntaxis while being carried northwestward coupled to the underlying megathrust (Pavlis et al., 2004). Recent results from GPS geodetic measurements support model (2), wherein the Kushtaka–Martin Lake area is currently caught in a zone of left-lateral shearing (Elliott et al., 2010). In both tectonic scenarios, flexural-slip reactivation of bedding surfaces during second-phase folding generates sinistral shear along east-trending and steeply dipping faults. Evidence for flexural-slip folding is important for evaluating the seismic potential of an area, because the folds may reflect the presence of larger blind faults at depth (Yeats, 1986). If tectonic indentation is now dominant over folding (which one might suspect given the cuspate nature of the plate margin), then forcing the sedimentary strata into the syntaxis between the Chugach Mountains and Ragged Mountain requires lateral slip on east-trending faults. Focal mechanisms to earthquakes indicate sinistral slip on east-trending fault planes, and dextral slip on north-trending planes (Ruppert, 2008).

Is there dextral slip on north-trending limbs of the second-phase folds? One site that merits investigation in the future is located on the eastern ridge of the Don Miller Hills, above the Bering River. The site is marked by swarms of sackungen, which we have previously assumed to be associated with dip-slip motion on bedding, but the scarps have not been studied other than by reconnaissance overflights and a brief inspection on the ground. It is also worth noting that dextral slip in the north-trending fold limbs may be inhibited as the limbs are rotated counterclockwise and misoriented in the contemporary stress field. There is a 30-km-long scarp system on the eastern flank of Ragged Mountain, where the Chugach–Saint Elias suture is rotated more than 90° counterclockwise in the limb of a large second-phase fold (Bruhn et al., 2004), but the scarps are too long to be associated with lateral shearing by folding. Tysdal et al. (1976) proposed that the upper part of Ragged Mountain is sliding westward toward the Copper River lowlands by normal-slip reactivation along the old Chugach–Saint Elias fault surface. An alternative explanation is that the fault scarps reflect local extension and dip slip along bedding surfaces in the crest and forelimb of a large fault propagation fold above a blind imbricate thrust fault that rises from the Aleutian megathrust (Bruhn et al., 2006). Ragged Mountain is asymmetric in profile, with a gentle sloping western side, and more steeply sloping eastern side, perhaps reflecting the overall geometry of the late-stage structure, if the interpretation of Bruhn et al. (2006) is correct.

This work was done as part of the Saint Elias Erosion and Tectonics Project (STEEP) supported through the NSF Continental Dynamics Program and Office of Polar Programs, with specific support from NSF grants EAR0409009 and EAR0735402 to Pavlis and EAR0408959 to Bruhn. Lindsay Lowe-Worthington assisted with the Kushtaka site study. We acknowledge constructive reviews from the Guest Editor and one anonymous reviewer.

1Supplemental Figure 1. Jpg file. Photographs of the Kushtaka trench site: (A) overview of scarp and trough; (B) close-up of trench; string grid is on 0.5 m spacing. Ron Bruhn is photographing the faulted coal of the Kultieth Formation beneath the scarp. If you are viewing the PDF of this paper or reading it offline, please visit or the full-text article on to view Supplemental Figure 1.
2Supplemental Figure 2. Jpg file. Upper: overview of the Martin Lake detailed study site, looking NNW toward an unnamed ridge (near background) east of Martin Lake. Red dashed line shows tectonic scarp F. Orange lines show gravitational scarps B, D, and E. Our two trenches are shown in yellow. Last Glacial Maximum (LGM) ice flowed over a divide at far right. Lower: telephoto view of the unnamed ridge, showing typical anomalous ridge-and-trough topography. If you are viewing the PDF of this paper or reading it offline, please visit or the full-text article on to view Supplemental Figure 2.
3Supplemental Figure 3. Jpg file. Photographs of reactivated antislope scarps. (A) Multiple scarps east of the detailed study site, including scarps F and B, plus two higher unlabeled scarps; (B) looking west along scarp B from the eastern margin of Figure 10; snow gully at center is smaller one shown in Figure 10; (C) scarp B just east of Figure 10; view is to the NE; (D) rejuvenated scarp E just west of sag pond 2, looking west toward Ragged Mountain. If you are viewing the PDF of this paper or reading it offline, please visit or the full-text article on to view Supplemental Figure 3.
4Supplemental Figure 4. Jpg file. Bedrock exposure showing antislope fault scarp is located along a fault that parallels bedding in the south-dipping limb of the syncline at the Martin Lake trench area. The bedrock is thin-bedded siltstone of the Tokun Formation (Eocene). Unconsolidated colluvium, sag pond deposits, and fluvial sediments deposited by snowmelt runoff are faulted against bedrock on these types of structures. If you are viewing the PDF of this paper or reading it offline, please visit or the full-text article on to view Supplemental Figure 4.
5Supplemental Figure 5. Jpg file. Scarp D at the site of the west trench; view is to the west toward the northern end of Ragged Mountain (skyline). Rejuvenated part of the scarp is outlined in red, and is interrupted by two small slumps (yellow). Green dashed lines show inferred diagonal scarps on the hanging wall. If you are viewing the PDF of this paper or reading it offline, please visit or the full-text article on to view Supplemental Figure 5.
6Supplemental Figure 6. Jpg file. Log of the west trench at Martin Lake. If you are viewing the PDF of this paper or reading it offline, please visit or the full-text article on to view Supplemental Figure 6.
7Supplemental Figure 7. Jpg file. Probability density functions of the equivalent dose (De) for optically stimulated luminescence (OSL) samples from the west trench at the Martin Lake site. (A) Unit 3, loess. This sample contains the closest to a single OSL age population, consistent with its origin as a well-zeroed air-fall loess. (B) Unit 2b, colluvium. This sample contains multiple OSL age populations, including a large higher-dose group (190–210 Gy) that probably represents inherited luminescence from unzeroed sand grains. This is consistent with the variable light exposure expected during colluvial transport. If you are viewing the PDF of this paper or reading it offline, please visit or the full-text article on to view Supplemental Figure 7.
8Supplemental Figure 8. Jpg file. View of the fault plane at the head of the east trench, where Quaternary deposits abut bedrock of the Tokun Formation. Note corrugations (arrows) formed by left-oblique fault slip. The trench is about 1 m wide. The distance between gridded strings is 0.5 m on the right wall of the trench. If you are viewing the PDF of this paper or reading it offline, please visit or the full-text article on to view Supplemental Figure 8.