Late Quaternary sea level, isostatic response, and sediment dispersal along the Queen Charlotte fault Open Access

The western shores of the Haida Gwaii archipelago (formally the Queen Charlotte Islands) off the northwestern coast of British Columbia (Canada) and the Alexander Archipelago of southeast Alaska are distinctly rugged with little refuge from the North Pacific Ocean (Fig. 1). George Dawson described this coast from his 1878 expedition as having steep rocky sides with little or no beach and bold water (Dawson, 1878). Indeed, western Haida Gwaii and the Alexander Archipelago are exposed to an extreme wave regime (Thomson, 1981, 1989). In addition to being an exposed high-wave-energy environment, the area has also undergone dramatic sea-level fluctuations and is the most seismically active area in Canada. With limited access and the energetic shoreline, these shores have been, and are, relatively uninhabited, and some marine areas are not yet charted.

Just offshore is the Queen Charlotte–Fairweather fault system, a major structural feature that extends from the Explorer triple junction, south of the islands, to well into the bight of the Gulf of Alaska (Fig. 1). The transform boundary is split into two primary faults: the northern 300 km section is defined by the transpressional Fairweather fault, which extends southward from Yakutat along the western front of the Fairweather Range to Icy Point; the fault then steps offshore at Icy Point, takes an ∼25° clockwise bend (∼340°), and becomes the Queen Charlotte fault. This system represents a major transform boundary that separates the Pacific plate from the North American plate, similar to the San Andreas fault system of California (Atwater, 1970; Plafker et al., 1978). The length of the Queen Charlotte–Fairweather fault system is 1330 km, slightly longer than the San Andreas fault, with a reported width of 1–5 km, and ∼75% of the length is located offshore (Carlson et al., 1985). Recently, most of the offshore fault zone from south of Haida Gwaii through southeastern Alaska has been imaged in detail using multibeam echosounder (MBES) data and other geophysical techniques that have documented the fault morphology and identified features associated with localized deformation along the fault (step-overs), submarine canyons, gullies, and submarine slides adjacent to the fault (Barrie et al., 2013; Brothers et al., 2019; Greene et al., 2019).

Based on this high-resolution data, a better understanding of plate tectonics and Quaternary sedimentary processes can be realized for this region of the Pacific Northwest. Because the physiography of this continental margin is shaped by the complex interplay between tectonic and sedimentary processes, which often alternate between periods dominated by constructional (sediment delivery and progradation) or destructional (erosion, slope failure, canyon incision, retrogression, and fault displacement) geomorphic processes, it is now possible to provide an interpretive chronology through the late Quaternary. Our objective here is to document how the last glaciation and subsequent sea-level changes have impacted the present morphology of the central and southern Queen Charlotte fault, and in turn, how the Queen Charlotte fault has impacted the sea-level history and Quaternary sedimentary processes along this portion of the Queen Charlotte fault margin, subsequent to the last glaciation. The exceptional preservation of faulted geomorphic features along the plate boundary fault provides an unprecedented opportunity to study the fault behavior over many earthquake cycles in a high-latitude Quaternary glacial marine setting. In addition, knowledge of the late Quaternary Pacific Northwest coastal environment provides insight into the viable pathway for early humans as they colonized the Americas (Lesnek et al., 2018).

The Queen Charlotte fault is a near-vertical fault zone that is seismically active down to ∼21 km (Hyndman and Ellis, 1981) with a mainly right-lateral transform motion of ∼50–60 mm/yr (Prims et al., 1997; Rohr et al., 2000). Recently, Brothers et al. (2020) analyzed submarine tectonic geomorphology and suggested that the Queen Charlotte fault itself accommodates the majority of relative plate motion (48–55 mm/yr). In contrast to the predominately strike-slip motion along the central and northern portions of the Queen Charlotte fault zone, plate motion along the southern portion is more oblique, with up to 20° of convergence up to central Haida Gwaii (Hyndman and Hamilton, 1993). Because of the high slip rates along the fault, the Queen Charlotte fault tends to rupture frequently in large earthquakes. During the past 100 yr, seven earthquakes of Mw 7.0 or greater have occurred along the Queen Charlotte fault (Fig. 1), including the 1949 Mw 8.1 earthquake off northern Haida Gwaii, Canada’s largest recorded earthquake (Bostwick, 1984). More recently, a Mw 7.8 thrust event near southern Haida Gwaii in 2012 (Lay et al., 2013) and a Mw 7.5 strike-slip event west of Craig (Fig. 1) suggested there are dramatic differences in plate boundary mechanics due to an increasing component of convergence to the south (Lay et al., 2013; Hyndman, 2015; Tréhu et al., 2015; Brothers et al., 2020).

In the late Quaternary, a glacier from the massive Cordilleran ice sheet extended westward across northern Hecate Strait and through Dixon Entrance and coalesced with ice from Haida Gwaii, deflecting it westward within Dixon Entrance (Sutherland-Brown, 1968; Barrie and Conway, 1999; Mathewes and Clague, 2017). This coalescence was probably short-lived (Clague, 1989). Ice also moved south down the central trough in Hecate Strait (Barrie and Bornhold, 1989; Shaw et al., 2019) and coalesced with ice flowing through the troughs of Queen Charlotte Sound south of Haida Gwaii to the shelf break (Luternauer and Murray, 1983; Luternauer et al., 1989; Hicock and Fuller, 1995; Josenhans et al., 1995, 1997). The Hecate Glacier was 20 km wide and flowed south parallel to the coast (Shaw et al., 2019). A minimum ice thickness of 400 m is suggested for some shelf areas (Josenhans et al., 1995; Barrie and Conway, 1999), and an ∼690 m thickness is suggested in the northern Hecate Strait and Dixon Entrance (Hetherington et al., 2004). Off southeastern Alaska, ice is considered to have reached the shelf break, particularly down the sea valleys such as Chatham Strait, though many areas off the Alexander Archipelago (Prince of Wales and Baranof Islands) are considered to never been glaciated, similar to Haida Gwaii (Kaufman and Manley, 2004; Carrara et al., 2007; Lesnek et al., 2018, 2020; Brothers et al., 2020). Glaciation reached its maximum extent sometime after 23.0 ka (Blaise et al., 1990).

On Haida Gwaii, small ice caps and piedmont glaciers, up to 500 m thick, developed that were independent of the Cordilleran ice sheet (Clague et al., 1982a; Clague, 1983). There may have been ice-free areas on the islands and on the coastal lowlands of northern Haida Gwaii where glaciation was minimal and of short duration (Clague et al., 1982a; Clague, 1989; Mathewes and Clague, 2017). The limited size and extent of the Queen Charlotte Mountain source areas and the proximity of deep water of the open Pacific Ocean, Dixon Entrance, and Queen Charlotte Sound limited expansion of ice on the Haida Gwaii islands (Clague, 1981, 1989; Warner et al., 1982; Barrie et al., 1993; Barrie and Conway, 1999).

Based on evidence of low-level cirques along the coast as well as glacial striae, Sutherland-Brown (1968) suggested that ice moved out onto the shelf, off western Haida Gwaii, and formed a small ice shelf. However, Clague (1989) implied that the ice extent was limited and that some areas of the shelf could have been free of ice during the last glaciation. Further, based on recent data, no evidence exists for diamicton deposition, and no identifiable glacial or deglaciation features exist, such as iceberg scours and boulders, which are usually common in glaciated areas. It is quite probable, therefore, that the west coast of Haida Gwaii had little to no ice cover during the last glaciation.

The glaciation along the Pacific North Coast terminated with rapid climatic amelioration, resulting in rapid retreat and melting of the ice. Glaciers had retreated from the lowland areas of Haida Gwaii beginning around 17.0 ka (Warner et al., 1982), but mountain valleys and cirques probably supported remnant ice masses until much later (Clague et al., 1982a; Clague, 1989; Mathewes and Clague, 2017). Offshore, glacial retreat began off the western margin sometime after 17.0 ka and possibly as early as 18.1 ka (Darvill et al., 2018). Ice had largely left the lowlands and offshore region by 14.5–13.0 ka (Barrie and Conway, 1999; Hetherington et al., 2004). Off southeast Alaska, the initial retreat of marine-terminating ice margins from their maximum extent was driven by factors acting on the ice-ocean interface, including sea-level rise and ocean warming, leading to intense ice loss via calving in the early stages of deglaciation (Lesnek et al., 2018, 2020).

Relative sea-level curves developed for specific locations along the Pacific margin differ from each other, reflecting complex glacially induced crustal displacement. Using geospatial interpolation combined with site-specific relative sea-level data, Hetherington et al. (2003) generated a model of glacially induced crustal displacement for 500 yr time intervals between 14.2 and 8.7 ka along the northern section of Canada’s Pacific margin. On the eastern side of the Haida Gwaii, sea level fell by more than 150 m between ca. 15.0 and 13.0 ka (Hetherington et al., 2004; Barrie and Conway, 2002a), with large areas adjacent to eastern Haida Gwaii being subaerially exposed (Fedje and Josenhans, 2000; Josenhans et al., 1995). Eustatic sea-level rise, coupled with subsidence of a glacio-isostatic forebulge (Clague, 1983; Luternauer et al., 1989), resulted in sea levels rising very rapidly, reaching the present shoreline on Haida Gwaii by ca. 10.5–9.6 ka (Clague et al., 1982b; Fedje et al., 2005). Sea levels reached a plateau at 15–13 m above current levels 8.2–4.8 ka and have been falling up until the present (Clague et al., 1982b; Josenhans et al., 1997; Fedje et al., 2005).

Sea-level fluctuations on the outer coast of the Alexander Archipelago of southeast Alaska resemble the pattern at Haida Gwaii to the south, although marine dates are few (Shugar et al., 2014). A wave-cut terrace at 165 m water depth off the west coast of the Prince of Wales and Baranof Islands (Carlson, 2007) may correspond to the lowstand seen off Haida Gwaii. Baichtal et al. (2017a, 2017b) suggested that a peripheral forebulge developed west of the ice front, similar to Haida Gwaii, starting at 16.9 ka, with most of the marine areas being deglaciated by 14.5 ka. Terraces mapped from multibeam bathymetry can be seen at 165–180 m water depth (Baichtal et al., 2017b). In addition, pahoehoe (subaerial) lava flows occur at water depths of 160 m offshore (Baichtal et al., 2017b). Within the island archipelago, sea level reached the highest levels of between 65 and 190 m sometime after 13.5 ka (Carlson and Baichtal, 2015; Baichtal et al., 2017a, 2017b).

Multibeam swath bathymetry was acquired off the west coast of Haida Gwaii between 2009 and 2010, and in the Dixon Entrance area in 2017 (Barrie et al., 2018) using a hull-mounted, Kongsberg 0.5° × 1.0° EM710™, dual-swath, multisector stabilization, chirp-pulse, high-definition beam-forming system, which operated at a frequency of 70–100 kHz. The surveys were carried out from the Canadian Coast Guard Ship (CCGS) Vector at a survey speed of 10 knots by the Canadian Hydrographic Service, in cooperation with the Geological Survey of Canada. The tracks were positioned so as to insonify 100% of the seafloor with 50%–100% overlap. Positioning was accomplished with a broadcast differential global positioning system (GPS), and the multibeam data were corrected for sound velocity variations in the stratified water column using sound speed casts. The data were edited for spurious bathymetric and navigational points and subsequently processed using CARIS^® software. Similarly, multibeam swath bathymetry was collected off southeastern Alaska between 2009 and 2018; data acquisition methods were summarized by Brothers et al. (2020). All multibeam data were gridded at 5 m resolution, exported as ASCII files, and imported into ArcInfo^® software for analysis and image production. Hillshade relief surfaces were created with a solar azimuth of 315°, solar zenith of 45°, and vertical exaggeration of 2×. The beam-forming feature of the EM710 multibeam system reduces the footprint at nadir to around 5 m at 400 m water depth and ∼13 m at 1000 m water depth. Thus, by gridding the data at 5 m, the data are undersampled above 400 m and oversampled at depths greater than 400 m water depth.

An initial geophysical and sampling survey was undertaken using the CCGS John P. Tully in 1995, during which 600 km of 500 Joule Huntec DTS™ high-resolution boomer seismic-reflection and 120 kHz Simrad side-scan sonar profiles were collected along the narrow continental shelf off western Haida Gwaii and into Dixon Entrance, which allowed for penetration depths up to 100 m (Barrie and Conway, 1996). In addition, a Benthos™ piston corer was used to collect seafloor sediment (13 cores) in two inlets in central Haida Gwaii and within small nearshore basins off the southern Haida Gwaii archipelago. Initial interpretations of the survey data and cores collected in Rennell Sound were reported in Barrie and Conway (1996), while results of findings from other cores and radiocarbon dates from Rennell Sound are presented here.

A full investigation of the southern and central parts of the Queen Charlotte transform fault system was undertaken using the CCGS John P. Tully in September 2015 and in 2017 along the full length of the Queen Charlotte fault to determine fault geometry and activity. A Knudson™ low-power, 12 element, 3.5 kHz, high-resolution CHIRP seismic-reflection profiling system was used to image the subsurface stratigraphy and to select sampling sites in both years. In 2017, a multichannel seismic-reflection survey collected 250 km of data with a U.S. Geological Survey (USGS) Sig2mille™ 1 kJ, 400–1000 Hz sparker system using a 48 channel, solid-core Geometrics Geoeel™ six-section hydrophone streamer with 16 hydrophones spaced 1.6525 m in the first two sections and 32 hydrophones spaced 3.125 m in the trailing sections (four sections).

A large (0.75 m³) IKU grab sampler and a Benthos™ piston corer were used to collect seafloor sediment, and a GSC 4K underwater drop camera system was used to photograph the seafloor. All cores were run through a Geotek™ Multi-Scan Core Logger (MSCL), which measured density, velocity, and magnetic susceptibility. The cores were split, examined, described, and subsampled at the Geological Survey of Canada (Pacific) laboratory in Sidney, British Columbia. Split cores were run through the Royal Roads University MSCL-XZ to collect high-resolution magnetic susceptibility and image data, including bulk density (by gamma-ray attenuation measurements) and P-wave velocities. All shell material identified from visual and multiscan images considered large enough for radiocarbon dating were collected. Interpretation of the seismic-reflection data was undertaken using both the Knudson Post Survey and the Kingdom Suite^® software packages.

Multibeam bathymetry from the central coast of Haida Gwaii and off Dixon Entrance was presented in Greene et al. (2019), along with interpretation of cores collected within these blocks. All other data collected during these surveys are presented here. In addition, detailed multibeam bathymetry images that are not presented here were published in Brothers et al. (2019, 2020).

The continental shelf off Haida Gwaii north to Dixon Entrance is narrow, extending only 4 km off Cape St. James in the south to 25 km near the Canada–U.S. boundary in the north. The striking observation from the MBES and all seismic-reflection survey lines collected from the continental shelf off western Haida Gwaii is that there is little to no Quaternary sediment, only bedrock (Figs. 2, 3, 4, and 5). Even the inlets that are open to the Pacific Ocean with a muted or no sill at the entrance are predominantly floored by bedrock. The bedrock geology of the onshore areas adjacent to the shelf and enclosing the inlets of Haida Gwaii is a diverse assemblage of Jurassic to Tertiary rocks (Anderson and Reichenbach, 1991). On northern Haida Gwaii, Masset Formation basalt predominates (Hickson, 1991), and in the south, plutonic Jurassic and Tertiary volcanic rocks occur (Anderson and Reichenbach, 1991). Sparse mobile coarse sediment occurs in the small depressions formed by the rough igneous bedrock offshore (Fig. 2). The pattern of exposed bedrock is characteristic of all areas of the shelf surveyed, except the shelf off Dixon Entrance and within protected fjords and inlets, where mixed deposition predominates.

Tréhu et al. (2015) and Brothers et al. (2020) broadly divided the Queen Charlotte fault into three sections based on predicted magnitude of oblique convergence, crustal-scale seismic-reflection profiles, and the MBES data. For the central and southern sections of the fault, the Queen Charlotte fault is located below the shelf break in water depths of between 500 and 2200 m. The slope above the fault is dominated by small canyons and gullies that mostly terminate at the fault (Harris et al., 2014). South of 53°8′N, talus fans have formed at the mouths of the largest canyons in a trough that occurs between the Queen Charlotte terrace and the fault zone (Harris et al., 2014; Hyndman, 2015; Brothers et al., 2020). North of this, the canyons and gullies enter the northward-shallowing fault valley from both sides, returning to a gentler slope with displaced gullies and canyons crossing the fault north of 54°N.

Based on the MBES data, the boundary between the central and southern sections happens at 53°5′N, where the Queen Charlotte fault bends to the east. South of this, the linear fault extends parallel to the Haida Gwaii coastline along the upper slope just below the shelf break less than 10–15 km off southern Haida Gwaii. South of ∼52°23′N, the deformation is distributed, and mapping of seafloor faulting is more difficult (e.g., Fig. 2; Brothers et al., 2020). A 4 km transpressional step-over wraps around the eastern flank of a seamount (Rohr, 2015) and then immediately takes a 3 km transtensional step that extends south almost to the southern end of Haida Gwaii (Brothers et al., 2020). Just offshore, two distinctive cones occur (Fig. 2), both of which have significant plumes extending 500 m up into the water column. These plumes have been identified on three surveys in 2011, 2015, and 2017 with little change. Samples collected from the small craters at the top of the cones consist of carbonate crusts near the multiple vents and large boulders, primarily vesicular basalts. Radiocarbon dates from shells taken from the carbonate crust fragments (201702–65; Fig. 2) varied between 23,080 and 21,880 ¹⁴C yr B.P. (25.0 ka).

The multibeam sonar bathymetry data revealed evidence of a fault valley with small depressions on the upper slope primarily within the central section of the fault (Figs. 3 and 4). The depressions form where strike-slip right-step offsets have realigned the fault due to oblique convergence (Barrie et al., 2013). The en echelon right-step offsets or pull-aparts result in subsiding basins (grabens) up to 700 m in length (Fig. 3). Multibeam mapping completed in 2017 (Barrie et al., 2018) revealed a larger basin (sag pond), 1.0 km in width and 4.0 km in length, with a right-step of 1.0 km (Fig. 3). This shift of the fault occurs 14 km south of the predicted epicenter of the 1949 Mw 8.1 earthquake. Just north of the epicenter (25 km), the Queen Charlotte fault steps to the right (30 m) within another basin, and beyond its northern terminus, it is bounded by a very rough seafloor surrounded with numerous pockmarks (Fig. 4). The rough surface is a result of significant gas expulsion along and immediately adjacent to the Queen Charlotte fault, based on the seismic-reflection data. The disturbed area occurs at a high point (300 m water depth) along the central portion of the fault and extends for 10 km with a disturbed zone width of 1.4 km adjacent to the fault on either side (Fig. 4).

Sediment moving off the shelf onto the upper slope is trapped within the fault valley for the greater part of the 1200 km of the offshore portion of the Queen Charlotte fault system. As a result, there are few outlets past the fault valley further down the slope. For example, off central Haida Gwaii, there are only four primary outlets that cross the fault, usually at one of the step-over basins. Small fans have developed downslope of each these outlets (Figs. 3 and 4). Where the surface trace of the fault reaches 600 m water depth, there is a sediment transport divide, with sediments south of this moving south down the fault valley and sediments north of this moving in the opposite direction (Fig. 3). This pattern is repeated again in the area of the highly disturbed gas expulsion area, at the shallowest point of the fault (300 m water depth) off Haida Gwaii (Fig. 4).

Submarine channels mapped from the coast of Haida Gwaii to the shelf break and into the canyon system often align with these sediment transfer outlets out of the fault valley. For example, at the narrow entrance of Tasu Inlet off southern Haida Gwaii, a channel imaged in the MBES bathymetry can be seen crossing the narrow shelf until it reaches the Queen Charlotte fault (Fig. 5). A dominant feature of the continental shelf off western Haida Gwaii is the occurrence of channels (10–25 m deep) seaward of all the major inlets (Barrie and Conway, 1996), with the Rennell Sound channel being the largest (Fig. 3). These channels are cut into bedrock and are characteristic of fluvial channels in dimension and shape (widths of 300–800 m), but they are presently devoid of any sediment other than a thin layer of mobile sand. Our interpretation is that all these channels represent sea-level lowstand fluvial channels discharging glacial meltwater to the shelf break across a narrow and relatively flat coastal plain. On the very narrow continental shelf off southern Haida Gwaii, a coastal plain is evident in MBES bathymetry, with channels that bifurcate and drain both into Hecate Strait to the east and into the Pacific Ocean, directly into a submarine canyon system (Fig. 2). Here, the sea-level lowstand is evident by the wave-cut terrace along Hecate Strait that continues around the southern end of the Haida Gwaii archipelago to the Pacific margin, where the lowstand terrace forms the shelf break. This base of the wave-cut lowstand terrace is at ∼175 m water depth.

At the southern extreme of the fault zone, just south of Haida Gwaii, three cores were collected adjacent to the fault (Fig. 2). Cores 201504–01 and 201504–02 were collected just seaward of the fault. The upper unit in both cores is olive-gray silty fine bioturbated sand (62 cm in core 201504–01 and 172 cm in core 201504–02), which overlies olive-gray laminated silt (Fig. 6). A radiocarbon date within the lower unit of core 2 is 15,981 k.y. B.P. (Table 1). Like the previous two cores, core 201504–03, on the landward side of the fault (Fig. 6), contains the same two units with a radiocarbon date of 13,090 k.y. B.P. (Table 1) in the laminated silt unit. Below this, a 23 cm gravel unit is underlain by a gray mud with highly disturbed laminations to the base of the core.

Over the central portion of the fault zone, some 12 cores were collected. Greene et al. (20189) described seven cores off central Haida Gwaii near Cartwright Canyon, including five cores collected within a 6-km-long and 3-km-wide rift ridge submarine slide (Fig. 3). Cores collected near the fault are dominated by dense sand and mud units that may have been compacted by seismic shaking throughout the late Pleistocene and early Holocene (Greene et al., 2019). One core just beyond the slide (core 201504–21; Fig. 3) consists of a thin fine sand unit with an erosional lower boundary that overlies dark-gray laminated clay with minor gravel (Fig. 6). This lower unit is interpreted to be glaciomarine. The ages of the sediments from all these cores range from 50,377 k.y. B.P. to 15,034 k.y. B.P. Greene et al. (2019) suggested, based on these dates, that sediments younger than 15.0 ka are rare due to significant sedimentation shutdown after this early stage of deglaciation. Two cores collected within the fault in 2011 showed similar results (Barrie et al., 2013). Core 201102–32 and core 201504–15, both located within the right-step basin (sag pond), have turbidites with ages of 24,698 k.y. B.P. in 201504–15 (Greene et al., 2019) and 42,000 k.y. B.P. in 201102–32 (Barrie et al., 2013). Further north along the fault valley, core 201102–31 has dates of 14,700 and 14,000 ¹⁴C yr B.P. (16.0 and 15.3 ka) in sandy gravels (Barrie et al., 2013).

Within the 4-km-long right-step basin off northern Haida Gwaii, four cores (201504–23, 201504–25, 201504–26, and 201504–27) were collected on either side of the fault (Fig. 4). These cores consist of repetitive interbedded sand and muds interpreted to be turbidites (e.g., 201504–26; Fig. 6). The cores were barren of dateable material, except in core 2015004–23 (Fig. 6), where a date of 15,052 k.y. B.P. (Table 1) was obtained at 15 cm. Further north, off Dixon Entrance, three cores (201504–28, 201504–29, and 201504–30) were collected near the fault and were described by Greene et al. (2019). No dateable material was found within these cores.

North of Dixon Entrance, five cores were collected 12 km off northern Dall Island near Craig, Alaska, on the shelf inshore of the fault. Here, large exposures of granite basement rock crop out on the seafloor and appear to have been differentially eroded along fractures and faults well exhibited in the MBES bathymetry (Fig. 7). Cores 2015004–38 and 2015004–40 are entirely olive-colored bioturbated Holocene muds with occasional shells. Core 2015004–39, collected in 174 m water depth, has 385 cm of Holocene mud that overlies a 28 cm laminated stiff gray mud and finally a 38 cm sandy gravel unit with shells and rounded clasts up to 3 cm (Fig. 7). Another core (2015004–41) in shallower water (137 m water depth) has an olive-colored bioturbated mud that overlies a coarse sandy gravel unit. An erosional boundary occurs at the base of the gravel unit (153 cm), and below this boundary, there is a dark-gray mud with distorted laminations to the base of the core at 370 cm (Fig. 7). The fifth core (2015004–42) was taken 15 km south of cores 38–41, along a north-south–trending gully at 211 m water depth. An olive-colored bioturbated sandy mud is again underlain by a coarse sand unit with shells at 132–172 cm. At the base of the coarse unit, there is an erosional boundary, below which there is a dark-gray mud with some laminations and occasional shell material to 390 cm.

The Holocene muds have radiocarbon dates ranging from 7741 k.y. B.P. to 2976 k.y. B.P. (four dates; Table 1). The underlying coarse gravel and sand unit found in cores 39, 41, and 42 all date narrowly to between 11,929 k.y. B.P. and 11,286 k.y. B.P. (five dates; Table 1). The lower mud unit in core 2015004–42 has three dates progressively getting older down core from 12,629 k.y. B.P. to 13,608 k.y. B.P. (Table 1).

Inshore, four cores were collected in two inlets along the west coast of Haida Gwaii, and nine cores were collected along the inner shelf at the extreme southern end of the archipelago in 1995. Cores collected along the inner shelf of southern Haida Gwaii consist of thin deposits of carbonate sand. Wass et al. (1970) first identified bryozoan carbonate sands in the Great Australian Bight, and later research identified the extensive southern Australian calcareous skeletal sediment deposits as a “shaved shelf,” a relict shelf with active winnowing (James et al., 1994, 2001). Similar relict shelf carbonate sands are found in southern Queen Charlotte Sound north of Vancouver Island (Nelson and Bornhold, 1983) and adjacent to eastern Haida Gwaii (Carey et al., 1995). The western Haida Gwaii deposits (cores 95–26–35; Fig. 2), however, contain a greater percentage of very fine carbonate sand. The carbonate fraction of the sediments is entirely of skeletal origin, primarily barnacles, bryozoans, and assorted bivalves and gastropods, and averages 80%–85% of the total sediment content (Fig. 2). The average thickness of the carbonate deposit is 1–5 m, and it has been interpreted to be underlain by lacustrine deposits (Josenhans and Zevenhuizen, 1996). Temperate carbonate deposits normally occur in areas where sediment input is low, the seafloor is primarily bedrock or gravel, hydrodynamic energy is high at the seabed, and ample nutrients are available (Nelson and Bornhold, 1983; Nelson et al., 1988; Scoffin, 1988; Carey et al., 1995). There is extensive exposure of a rough bedrock surface, and vertical mixing occurs off the southern cape of Haida Gwaii (Crawford et al., 1995), providing both the substrate and nutrient-rich water required for carbonate production.

Sedimentation within the fjords of western Haida Gwaii, which are constrained by a bedrock sill at the entrance, such as Tasu Inlet (Fig. 5), is composed of fine-grained, organic-rich, bottom sediments, i.e., typical sedimentation for Canadian fjords (Syvitski et al., 1987). The seaward entrances to these fjords are usually very narrow and shallow, and, consequently, Pacific wave energy does not penetrate to any appreciable degree. Inlets that are oriented such that they are not open to the Pacific oceanic conditions, such as Seal Inlet (Fig. 3), also contain fine-grained Holocene sediments overlying bedrock.

Within the largest inlet, Rennell Sound, seismic-reflection data show stratified sand overlying a coarse-grained unit with a complex cut-and-fill stratigraphy normally associated with drowned fluvial sediments (Barrie and Conway, 1996). Channels leading out of the inlet can be seen on both the seismic-reflection data and regional bathymetry (Fig. 3). In core TUL95B-05, collected in a water depth of 152 m, a bioturbated, muddy-sand unit with organic and shell material overlies well-sorted, massive fine sand with a sharp erosional contact between units (Barrie and Conway, 1996). Four radiocarbon dates (Table 1) taken from wood and one shell within this upper unit indicate a rapid sedimentation rate between 12,380 and 11,290 C¹⁴ yr B.P. (5.5–6.0 cm/yr). No dateable material was collected in the massive sand unit. Below this, there are interbedded silt and sand units with minor gravel and wood fragments. Wood fragments selected from the base of this unit at 4.40 m suggest an age of 12,340 C¹⁴ yr B.P. (12.6 ka; Table 1). The lowermost unit in this core is a black massive sand unit that grades into fine gravel that is devoid of any marine indicators or dateable fragments. This is interpreted to be an alluvial facies.

Emergence of the continental shelf off western Haida Gwaii by greater than 150 m at the beginning of the Holocene has been postulated (Barrie and Conway, 1999; Hetherington and Barrie, 2004). This is thought to have been a consequence of a migrating glacial forebulge with a very thick (2500 m) ice load on the mainland of British Columbia (Clague, 1983) and a limited ice load on Haida Gwaii and in Hecate Strait (Clague, 1989; Barrie et al., 1991, 1993), during a period of eustatic sea-level lowering. Within the largest inlet on western Haida Gwaii, Rennell Sound, sea-level lowstand occurred just prior to 12.6 ka accompanied by significant deposition. The rapid submergence of this site fits closely with the sea-level curves for eastern Haida Gwaii (Barrie and Conway, 1999, 2002b; Fedje and Josenhans, 2000; Hetherington et al., 2004; Hetherington and Barrie, 2004; Shugar et al., 2014). Isostatic uplift (70+ m) resulted in a relative sea-level drop of at least 152 m (Hetherington and Barrie, 2004).

Cores collected north of Dixon Entrance off southeast Alaska suggest a sea-level lowering by greater than 175 m between 11.3 ka and 12.0 ka. In cores 201504–41 and 201504–42, below the erosional boundary that underlies the coarse sand unit, there are dark-gray muds that progressively get older to the base of the cores. The implication is that these cores were collected near the sea-level lowstand at ca. 12.0 ka and after a steady regression were quickly drowned by a transgressing sea. This would suggest a possible sea-level lowering by up to 175 m just inshore of the fault, which occurred just after the lowstand off Haida Gwaii. As proposed by Baichtal et al. (2017a, 2017b), the development of a forebulge that resulted in a maximum sea-level lowering of ∼175 m would suggest persistent ice to the east within the mainland until ca. 11.5 ka. Upon collapse of the forebulge, and coupled with eustatic sea-level rise, sea levels reached a maximum transgression on the outer islands between 11.5 and 10.6 ka (Carlson and Baichtal, 2015; Baichtal et al., 2017a, 2017b). Baichtal et al. (2017b) presented a proposed sea-level curve for the region that complements the findings presented here.

During the glacial maximum, ice exited Dixon Entrance (Barrie and Conway, 1999), reaching the shelf edge in the vicinity of the Mukluk Fan (Dobson et al., 1998; Shaw et al., 2019). Evidence described by Lyles et al. (2017) and Lesnek et al. (2020) shows that topography exerted a strong control on ice movement in Dixon Entrance. A major ice stream flowed northward along Clarence Strait, and farther west, a second stream flowed norward up Cordova Bay and Tievak Strait (Fig. 8). Ice that remained in Dixon Entrance and approached the Pacific Ocean was likely steered by local topography, and in particular, Learmonth Bank (Fig. 8). Grounded ice moved to either side because Learmonth Bank is shallower than 50 m and the channels to the north and south are over 400 m deep. The southern channel past Learmonth Bank is less than 10 km in width, with the mountainous shores of Haida Gwaii forming the southern boundary, whereas the channel to the north is deep and over 22 km in width, suggesting most of the ice went to the north-northwest. Evidence from core data just south of Dixon Entrance, along the Queen Charlotte fault on the upper slope, where the southern channel past Learmonth Bank leads, suggests virtually no glacial sedimentation and sediment ages from radiocarbon dating generally older than 15.0 ka (mostly between 50.4 and 15.0 ka; Greene et al., 2019). Conversely, cores collected in 2017 just north of Dixon Entrance are rich in glaciomarine sediments down the continental slope of southern Alaska (Barrie et al., 2018). To the south, ice moved out of Queen Charlotte Sound and became a floating ice shelf that moved out of Moresby Trough south of Haida Gwaii (Shaw et al., 2019). West of Haida Gwaii, little to no ice extended onto the narrow shelf. Consequently, Haida Gwaii acted as an ice shadow to glacial sediment transport to the Pacific, limiting deposition to the continental shelf and upper slope to the west of the islands (Greene et al., 2019).

Deglaciation began after 17.0 ka and before 15.6 ka, based on the oldest dates in Dixon Entrance and Hecate Strait (Barrie and Conway, 1999). Evidence from Haida Gwaii shows that ice still existed in the western mountain range at 14.5 ka and possibly persisted longer (Mathewes and Clague, 2017; Shaw et al., 2019). At this same time, sea level began to drop rapidly, reaching a lowstand at ca. 12.7 ka off Haida Gwaii and 12.0 ka off southeast Alaska. Outwash from Haida Gwaii was transported across the short subaerial shelf by several channel systems to the shelf break. Sediment discharge continued until ca. 10.0 ka and abruptly terminated once most of the ice had melted out of the western Haida Gwaii mountain chain. In addition, rapid sea-level rise quickly resulted in an erosive transgression of the narrow shelf with extreme oceanic energy that continues to the present day. Consequently, sediment delivery from the west coast of Haida Gwaii was very short, starting at ca. 14.5 ka and ending just after 12.7 ka. Most of this delivery would have been earlier than 12.7 ka, as the western Haida Gwaii mountains initially deglaciated.

During deglaciation, icebergs and ice would have moved out of Dixon Entrance, mostly to the northwest and north, as sea-level lowering and the subaerial Learmonth Bank would have restricted ice exiting to the southwest (Fig. 8). At the same time, icebergs were transporting sediment south of Haida Gwaii onto the continental slope out from Queen Charlotte Sound (Fig. 8; Barrie and Bornhold, 1989; Shaw et al., 2019). Ice sediment transport north and south of Haida Gwaii would have ended by 14.0 ka (Barrie and Conway, 1999; Shaw et al., 2019), as ice retreated into the mainland fjords. At present, little to no sediment derived from Haida Gwaii and the Alexander Archipelago reaches the narrow shelf and upper slope to the Queen Charlotte fault. Reworked sediment does move along the fault from the bathymetric highs and exits the fault valley at the few eroded exit points (Figs. 3 and 4), usually associated with the step-over basins or sag ponds. Deep-water submarine dunes (>1000 m water depth) occur where the sediments exit the fault valley system onto the lower slope (Barrie et al., 2013).

Shoreline tilt during the development and collapse of the forebulge across Haida Gwaii reached 2.1 m km^–1, and Hetherington and Barrie (2004) suggested that this occurred in response to a warm, relatively thin lithosphere, combined with the tectonic influence of the decoupled Pacific and North American plates along the Queen Charlotte fault. They further suggested that the Queen Charlotte fault acted as a hinged flap, taking up the isostatic forebulge uplift and collapse along the fault (Fig. 9). The apparent near-vertical and singular knife-edge character of the Queen Charlotte fault north of 52°23′N supports this possible interpretation. However, there is no conclusive evidence that can be taken from the detailed morphology to further support this hypothesis.

Assuming that the fault did take up to 70+ m of motion for the period of 15.0–13.0 ka, there would have likely been greater earthquake activity. Evidence of enhanced volcanic frequency during this same time interval is apparent off southeast Alaska, adjacent to the Queen Charlotte fault (Praetorius et al., 2016). The increased earthquake and volcanic activity would then have resulted in increased development of submarine failures and turbidites adjacent to the fault, but there is little evidence of this within the cores collected along the fault. However, as suggested by Greene et al. (2019), sediments younger than 15.0 ka are rare due to significant sedimentation shutdown after this early stage of deglaciation, and consequently little evidence would be obtained from core data.

A distinctive fault valley occurs along most of the fault (Fig. 9) north of the southern convergence zone, following the upper continental slope until it reaches the slope break at 55°54′N. Though the fault valley occurs parallel to the slope, it has a very distinctive V-shaped valley character, usually 100–300 m wide with side slopes averaging 25–75 m in height. The 70+ m of drop and rise of the North American plate adjacent to the Pacific plate during the development and collapse of the glacio-isostatic forebulge over 2000 yr could have resulted in the development of this unique geomorphic character of the Queen Charlotte fault. The subsequent 500 m of strike-slip movement and erosion within the valley by sediment transport processes certainly would have modified the feature, but these processes may not fully explain the consistent fault valley character that is a primary feature of the present Queen Charlotte fault.

The collection of multibeam bathymetry was undertaken by the Canadian Hydrographic Service (Department of Fisheries and Oceans) in cooperation with the Geological Survey of Canada in Canadian waters, and by the U.S. Geological Survey Coastal and Marine Geology Program and the Alaska Department of Fish and Game in U.S. waters. The officers and crew of the CCGS John P. Tully are acknowledged for able seamanship during collection of the geophysical and sediment sample data in poorly charted waters. Peter Neelands, Robert Kung, Greg Middleton, and Bob Murphy are thanked for invaluable assistance at sea and in the laboratory. We thank Royal Roads University for the use of the Geotek split-core multisensor core logger.