Controls on submarine canyon activity during sea-level highstands: The Biobío canyon system offshore Chile

Submarine canyons serve as the most important conduits for terrestrial sediments, including their associated pollutants, nutrients, and organic carbon, from the continental shelf to the abyssal ocean sink, bridging the sediment trap formed by the continental shelf and any intraslope accommodation spaces (Shepard and Dill, 1966; Normark, 1974; Normark and Carlson, 2003; Normark et al., 2009; Hung et al., 2012). As with rivers, submarine canyons are dynamic systems that adapt to changes in sediment supply, sea-level change, and tectonic forcing, by altering their courses and/or profiles, by becoming more or less active, and filling up with sediment or becoming more deeply incised. Although the latest generation of multibeam technology has recently enabled considerable advances in imaging the morphology of submarine canyons (e.g., Greene et al., 2002; Lastras et al., 2007, 2009; Mountjoy et al., 2009; Paull et al., 2010, 2011, 2013; Babonneau et al., 2013), a significant gap remains between the spatial resolution of most bathymetric maps and the level of detail and resolution required to understand the processes that shape submarine canyon systems, and how they respond to external influences.

Most of the submarine canyons identified in the global compilation of Harris and Whiteway (2011) were interpreted to have been established during periods of sea-level lowstands, and now constitute low-activity relict features on continental slopes that were cut off from any direct supply of fluvial sediments by the rapid Holocene sea-level rise. Most deep-sea terrigenous deposits have therefore formed during sea-level fall, lowstands, and periods of transgression, but specific tectonic and climatic circumstances can also promote deposition of terrigenous sediments on the deep seafloor, regardless of sea level (Covault and Graham, 2010). Canyons that extend across the shelf and act as submarine continuations of terrestrial sediment sources may be able to maintain sediment-gravity flow during sea-level highstands (e.g., Walsh and Nittrouer, 2003; Covault and Graham, 2010). Shelf-incising canyons commonly develop across tectonically active continental margins and are most abundant along the western margins of both South America and North America (Harris and Whiteway, 2011), where active faulting has formed narrow shelves and controls the location of submarine canyons (e.g., Covault and Graham, 2010). Two key controls have been proposed for terrigenous sediment delivery to the deep seafloor: (1) the tectono-morphologic character of the continental margin (e.g., the width of the continental shelf), and (2) climatic factors, for example, inflow of subglacial meltwater, intensified monsoons, and variations in the magnitude and frequency of the El Niño–Southern Oscillation (ENSO) (Walsh and Nittrouer, 2003; Romans et al., 2009; Covault and Graham, 2010; Covault et al., 2010; Puig et al., 2014) .

Sediment-gravity flows within a submarine canyon, however, can be sustained during sea-level highstands if a connection is maintained between a river mouth and the canyon head. This phenomenon has been observed in the Var Canyon off the French coast (Khripounoff et al., 2009), the Gaoping Canyon off Taiwan (C. Liu et al., 1993; J. Liu et al., 2002), and the Congo (Zaire) Canyon (Heezen et al., 1964; Babonneau et al., 2002; Khripounoff et al., 2003; Vangriesheim et al., 2009). However, river connection is not the only way of maintaining canyon activity during sea-level highstands. For example, the Monterey Canyon offshore central California, several canyons along the southern California Borderland (e.g., the La Jolla, Hueneme, and Mugu canyons), and the Nazaré Canyon off Portugal are not primarily dependent on fluvially transported detritus, but instead act as traps for longshore-transported sediment, or for shelf sediments resuspended by wave action (Covault et al., 2007; de Stigter et al., 2007; Greene et al., 2002; Lastras et al., 2009; Oliveira et al., 2007; Paull et al., 2003, 2005, 2011; Xu et al., 2010). Sediment-gravity flows can also be maintained during sea-level highstands by the funneling of dense shelf water, as in the Cap de Creus Canyon in the northwestern Mediterranean, other canyons in that vicinity, and the Halibut Canyon off Newfoundland (Canals et al., 2006; Lastras et al., 2011; Puig et al., 2013), by capturing deep-sea currents such as in the Portimão Canyon off Portugal (Marchès et al., 2007), or through the liquefaction of canyon-head sediments by storm events, as observed in the Eel Canyon off northern California (Puig et al., 2004; see also Puig et al., 2014, for a thorough review of contemporary sediment transport processes in submarine canyons).

Our study makes use of 1680 km² newly acquired high-resolution bathymetric data (with 5 m and 2 m grid sizes) and seismic reflection profiles to decipher the evolution of the shelfal extent of one of Chile’s largest submarine canyon systems, the Biobío Canyon (BbC) system. The canyon is connected to the Biobío River, which has the greatest mean annual water discharge and the third-largest catchment area in Chile (Milliman and Farnsworth, 2011). These new data permit the identification of submarine geomorphic features at the level required to identify small-scale erosional and depositional processes (e.g., Maier et al., 2011; Paull et al., 2013). The BbC system comprises two separate arms: the main stem (the BbC) is deeply incised into the shelf and connected to a major fluvial system, while the other arm (the tributary canyon) is now cut off from any source of terrestrial sediments (Fig. 1). Both canyon arms are systematically described herein to determine their different histories and to establish the relationship between their divergences and external factors, such as shelf morphology, the bottom-current regime, and the tectonic framework. Moreover, we have examined subtle details of shelf morphology that are only recognizable in very high resolution bathymetric data sets and provide unique insight into the external factors that have influenced the BbC history. The relationships described reveal detailed information on factors affecting activity levels within the two arms during the current sea-level highstand, which would have been difficult or even impossible to recognize and analyze without the new data.

Submarine canyons sensu stricto (Shepard, 1963) are defined as deep, steep-sided and relatively narrow submarine valleys. Canyons are cut into the bedrock or partially indurated sediments of continental shelves and/or slopes. They are characterized by V-shaped cross sections with occasional narrow flats at the base of the V, and may extend all the way down a continental slope to the basin plain (Shepard, 1963, 1972; Normark et al., 1993). Canyons are formed by erosive processes and are devoid of levees (Normark et al., 1993). A canyon thalweg is the line that connects the deepest points along the length of the canyon floor (e.g., Baztan et al., 2005; for a full review of the relevant terminology, see Normark et al., 1993).

Submarine canyons are considered to be active when gravity flows transport sediment along the conduit and modify canyon morphology by erosion and deposition (e.g., Weber et al., 1997; Normark and Carlson, 2003; Paull et al., 2003; Covault et al., 2007; Khripounoff et al., 2009; Romans et al., 2009; Mountjoy et al., 2014). Canyons with no sediment-gravity flows and the prevalent occurrence of general background sedimentation are considered to be inactive (Normark et al., 2006; Mountjoy et al., 2014). (A full review of the terminology herein is given in the Supplemental File¹.)

Along the active continental margin of Chile, where the oceanic Nazca plate is subducted below the South American plate, the present-day pattern of coarse-grained deep-sea sedimentation is thought to be largely controlled by 14 submarine canyon systems that are deeply incised into the continental shelf (Völker et al., 2014). The BbC system is located offshore from the city of Concepción; the canyon traverses the Chile margin from the continental shelf, crossing an accretionary prism and extending into the Peru-Chile Trench (Fig. 1; Thornburg and Kulm, 1987a; Thornburg et al., 1990; Völker et al., 2006). The BbC system was initially inferred from depth soundings collected by the Chilean Navy during the late 1960s (Galli-Olivier, 1968, 1969). New bathymetric and seismic reflection data that became available after the late 1980s allowed it to be identified as a submarine canyon sensu stricto, and its topographic and sedimentological characteristics could be described (Thornburg and Kulm, 1987a; Thornburg et al., 1990; Pineda, 1999; Völker et al., 2006). Approximately 70 km off the coast and 35 km trenchward of the shelf break, the BbC system merges with the Santa María Canyon (Fig. 1; Pineda, 1999; Rodrigo, 2010).

On the continental shelf the canyon system cuts into late Pliocene to Pleistocene bedrock comprising shallow-marine sedimentary deposits of the Tubul Formation (e.g., Melnick et al., 2006) that have been subjected to syndepositional tectonic shortening, which is still continuing today. The study area straddles the north-northeast–striking Santa María fault system (e.g., Melnick et al., 2006), a ∼100-km-long backthrust to the plate boundary that is rooted in the megathrust (Fig. 1B). The Santa María fault is an integral part of a splay-fault system with predominant blind thrusts that is characterized by a dextral shear component and associated fault-propagation folds (Fig. 1).

The location of the BbC is transitional between a semiarid Mediterranean climate with abundant winter rain and dry summers, and a temperate humid climate, with precipitation increasing toward the Andean orogen from 2000 mm/yr in coastal areas (Department of Geophysics of the Universidad de Chile, Santiago de Chile, http://www.atmosfera.cl). While most of the terrigenous sediment offshore northern Chile is eolian in origin (Lamy et al., 1998; Klump et al., 2000; Stuut et al., 2007), the bulk of the marine sediment off central Chile has reached the ocean through fluvial processes (Lamy et al., 1998, 1999). The present-day mean annual water discharge from the Biobío River is 33 km³/yr (Milliman and Farnsworth, 2011) and mean suspended sediment discharge is 1 Mt/yr (1985–1995, before dam construction in 1996; Tolorza et al., 2014). The Biobío River drains terrains in the high Andes with early Pleistocene lahar deposits derived from the Antuco Volcano (e.g., Pineda, 1999). These retransported deposits constitute basaltic black sands that are widely distributed along the coast of the Arauco Bay, in the vicinity of Concepción, and on Isla Santa María, and have been identified on uplifted marine terraces dating back to Marine Isotope Stage (MIS) 5e and MIS 3 (∼ 125 and ∼ 50 k.y., respectively; Melnick, et al., 2009; Jara and Melnick, 2015).

The coastline in the study area is characterized by a pronounced embayment, the Gulf of Arauco, and a continental shelf of up to 45 km width (Fig. 2). Downslope transport of terrestrial clastic detritus appears to be efficient due to the presence of deeply incised canyons on the continental shelf and strong bottom currents (Raitzsch et al., 2007; Völker et al., 2014). Raitzsch et al. (2007) suggested that there are no significant accumulations of unconsolidated sediment present on the shelf. In contrast, high short-term sedimentation rates of as much as 0.15 cm/yr (based on ²¹⁰Pb profiles) have been determined for the shelf (Muñoz et al., 2004). These high sedimentation rates have been attributed to localized zones of either continuous or seasonal upwelling and the large fluvial input of terrestrial detritus (Muñoz et al., 2004; Völker et al., 2014). However, the residence time of the loose sediment on the shelf seems to be low. (For a comprehensive review of the current state of knowledge regarding the marine morphology and geology of the central Chilean forearc region, see Völker et al., 2014.)

The central Chilean convergent margin is characterized by frequent seismic activity resulting in recurrent megathrust earthquakes, such as the 1960 Valdivia earthquake (moment magnitude, M_w 9.5) and the 2010 Maule earthquake (M_w 8.8) (Lomnitz, 1970, 2004; Plafker and Savage, 1970; Farías et al., 2010), together with local folding, faulting, and tectonic uplift and superposed variations in geomorphic processes (Bookhagen et al., 2006; Melnick et al., 2006; 2012; Farías et al., 2010; Rehak et al., 2010; Stefer et al., 2010; Vargas et al., 2011). The BbC system is therefore an ideal location for investigations of Holocene submarine canyon activity along a tectonically active plate margin.

In 2011 the Hydrographic and Oceanographic Service of the Chilean Navy in cooperation with the U.S. Naval Oceanographic Office acquired high-resolution bathymetric data covering the upper section of the BbC, from the coastline to the shelf edge (Fig. 2). The survey was conducted by the T-AGS 60 Pathfinder oceanographic survey vessel using Simrad EM710 and EM122 multibeam echosounders. The bathymetric data were processed using CARIS HIPS and SIPS 7.1 software (http://www.caris.com/products/hips-sips/) and were not filtered. The hydrographic survey utilized the precise point positioning (PPP) three-dimensional global positioning system (GPS) referenced to the World Geodetic System 1984 ellipsoid and traditional sounding reduction methods (González-Acuña and Arroyo-Suarez, 2013). The bathymetric data were reduced to a vertical reference plane using standard techniques and tidal records from four tide gauges along the coast. The total vertical uncertainty according to the table of standards for hydrographic surveys (International Hydrographic Organization, 2008) depends on water depth, and is between 0.72 cm in 40 m and 19.5 m in 1500 m water depth. Processed data are available in a 5 m grid and locally 2 m grid cell size (Fig. 2). A large high-resolution image of the 5 m grid bathymetry is included in Figure S1 [see ^{footnote 1}]).

Multichannel seismic reflection lines (n = 24) with penetrations of as much as 11 s two-way travel time (TWTT) were recorded along the Chile convergent margin by the RV Sonne (Cruise 161–3), as part of the Subduction Processes Off Chile (SPOC) research program (financed by Germany’s Ministry of Education and Research; Reichert, 2005). Seismic signals were generated by two tuned linear arrays, each consisting of 10 airguns, with a total chamber volume of 51.2 L. Reflected energy was recorded using a 3000-m-long digital streamer with 132 channels and a geophone group spacing of 25 m. Processing of the seismic data involved geometry definition, velocity analysis, normal moveout corrections, filtering, multiple attenuation, stacking, and post-stack time migration. A common mid-point (CMP) spacing of 12.5 m was applied and peak frequencies were centered around 25 Hz. This resulted in a maximum vertical resolution of 15 m for the near-seafloor sediments, decreasing with depth. We used parts of SPOC seismic reflection lines 24 and 25 for our interpretations.

Sediment profiles were recorded using a Parasound hull-mounted, parametric narrow-beam sediment echosounder with a footprint diameter of only 7% of the water depth, providing excellent lateral resolution. The recorded data were digitized and stored using the ParaDigMA software for subsequent digital signal processing and display (Spieß, 1993). The Parasound profiles were converted from TWTT to depth using a constant sound velocity of 1500 m/s. Peak frequencies of the Parasound system ranged between 3.5 and 2.5 kHz, resulting in a vertical resolution of 0.2–0.3 m.

Two gravity cores were recovered by the RV Sonne, one from within the BbC (GC14; Linke et al., 2011) and the second (GeoB 9802; Flüh and Grevemeyer, 2005) from the Biobío Fan in the Peru-Chile Trench (Fig. 1). Hemipelagic (nonturbiditic) sections of the sediment cores were sampled to establish depositional ages within the cores by radiocarbon dating of planktonic foraminifera. Two radiocarbon dates, one from each core, were obtained (additional information on radiocarbon dating is in the Supplemental File [see ^{footnote 1}]). Two dredge samples (3 KD and 4 KD; Raitzsch et al., 2007) and one box core (2 KG) were recovered from the BbC wall and the adjacent shelf during the RV Sonne Cruise 161 (see Fig. 2 for locations).

Open-source Matlab and ArcMap codes available on http://www.geomorphtools.org were used to automatically map canyon and gully thalweg profiles for quantitative geomorphic analyses of the canyon system.

The high-resolution bathymetric data set covers the portion of the BbC between the coastline and the shelf edge, which is at a water depth of ∼170 m (Figs. 1A and 2). The BbC meanders westward downslope from the coast for ∼33 km from the head of the main canyon, after which its orientation changes toward the northwest (Fig. 2). A tributary flows northward into the BbC ∼30 km from the canyon head (Fig. 2). Cross-sectional profiles indicate that the BbC has steep outer canyon walls and localized areas of flat canyon floor due to secondary sediment infilling (Figs. 3A, 3B). In contrast, the tributary defines a broad, relatively shallow cross section (Figs. 3A, 3C).

The flanks of the main canyon and the tributary are dissected by gullies (Fig. 4). Automated mapping of the canyon-wall gullies (with individual catchment areas >1 km²) within the BbC and its tributary reveals the distinct morphology of the tributary when compared to the main BbC stem, as gully thalwegs mirror the overall shape of the canyon walls. The tributary gullies are steep in their upper reaches just below the top of the tributary walls, while their lower reaches have a distinct concave, asymptotically shallowing shape (Fig. 4). In contrast, most of the main BbC gullies maintain a steep slope throughout and do not have a marked asymptotic shape in their lower reaches (Fig. 4). This morphological difference between the two arms is reflected in an analysis of their wall slopes (Fig. 5). Slope-area plots of the canyon and tributary walls (Fig. 5D) reveal that the walls of the tributary are generally shallower (mean slope ∼11.9°–13.4°) than the walls of the BbC (mean slope 16.2°–21.2°). The tributary forms a hanging valley in the wall of the main stem of the BbC (Fig. 4).

Detailed descriptions of the distinct present-day morphologies of the BbC and its tributary are provided in the following sections, and are then used to interpret the causes of their different evolution.

Observations. The BbC traverses the Chilean continental shelf and slope toward the west for a distance of ∼105 km, eventually reaching the Peru-Chile Trench, where it forms the Biobío submarine fan (Fig. 1; Thornburg et al., 1990; Völker et al., 2006). The BbC deflection ratio is 1.3; this ratio is the canyon length divided by the shortest distance from the canyon head to the Peru-Chile Trench (Ratzov et al., 2012). Our study focuses on that part of the BbC system that cuts into the continental shelf, from the coastline to the shelf edge. However, the high-resolution bathymetric data set does not cover the entire eastern extent of the canyon head; it extends only to 650 m from the coast where the water depth is 60 m (Figs. 2 and 6A). Pineda (1999) reported that the canyon head extends farther east to within 300 m of the coast, where the water depth is 15–20 m. The BbC reaches a maximum depth of 1100 m close to the shelf break (Figs. 3A, 3B). The canyon walls are generally steep (Figs. 3B, 4, 5A, and 5D) and slope-area plots are slightly skewed toward high slope values (Fig. 5D). The cross sections and the slope-gradient map reveal abrupt changes in gradient between the canyon walls and the canyon floor (Fig. 3, profiles D–D′–G–G′, I–I′, J–J′). Smooth, flat BbC floor areas were mapped manually (Fig. 5A) to identify areas of sediment fill.

The BbC floor is 2 km wide at its head, but quickly narrows to 200 m ∼2 km downslope (Fig. 3B, profiles A–A′, B–B′; Figs. 6A, 6B). The canyon head is crescent shaped and exposes a series of step-like features that are not evident on the canyon floor farther downslope from the canyon head (Fig. 6A). Sediment-distribution maps based on dredge samples show that the area of the canyon head is covered by unlithified sediments, including sand, sandy mud, and mud (Pineda, 1999, fig. 13 therein). Steeply incised banks in the upper reaches of the BbC have a vertical relief in excess of 50 m (Figs. 6A, 6B). An arcuate headscarp and the presence of displaced material with a hummocky upper surface indicate mass failure on the northern canyon wall (Figs. 6A, 6C). The distal (south-southeastern) part of the displaced material appears to have been subsequently eroded along the canyon thalweg (Figs. 6A, 6C).

The section of the canyon floor closest to the coast is narrow and rugged (Figs. 5 and 6), but ∼8 km downslope of the canyon head, just past a distinct tight meander, the canyon floor becomes broader, flatter, and smoother (Fig. 3, B–B′, C–C′; Fig. 5C). Farther offshore, the submarine canyon system on the outer shelf is characterized by a flat floor (Figs. 3 and 5). Down the canyon thalweg (46 km; 36 km linear distance, Fig. 5C) from the BbC head and at a water depth of 1100 m, a sudden change occurs in the gradient of the canyon thalweg (Figs. 7A, 7D), to the west of which maximum slope values can reach 30° (Fig. 5B) and the width of the canyon floor decreases abruptly to <200 m (Fig. 5C). The shape of the canyon thalweg profile changes from concave-up to convex-up toward the west (Fig. 5B). Adjacent to this knickpoint between the two thalweg sectors in the long-canyon profile, the SPOC 25 seismic reflection profile (Fig. 7C) and the corresponding Parasound line (Fig. 7E) both reveal a V-shaped cross section that is incised 1000 m into the strata. A high-angle detachment surface is present in the northeastern canyon wall (Fig. 7C). Overlying reflectors are truncated on the detachment surface and are more steeply inclined than those below and to the northeast of the surface, indicating canyon-wall mass wasting. The detachment surface is also present farther below toward the canyon floor, but the reflectors above the surface are more chaotic and deformed (Fig. 7C). Dredge samples 3 KD and 4 KD were collected from 1.8 km and 2.3 km northwest of the SPOC 25 seismic reflection profile, respectively, and within the mass-wasting feature mentioned herein (Fig. 7A). These dredge samples contained lithified bedrock, including numerous hardground blocks of gray, silty, laminated mudstone embedded in cohesive, gray, fine-grained sediment with abundant boreholes and fossil brachiopods. Nannofossil analysis has indicated that these mudstones are of Neogene age (Wiedicke-Hombach and Shipboard Scientific Party, 2002).

Interpretation. Dredge samples from the canyon head (Pineda, 1999) suggest that the step-like features (Fig. 6A) consist of unlithified sand and mud. These features are therefore unlikely to represent laterally continuous bedrock layers, which are responsible for similar step-like features along the canyon wall farther west (Fig. 13B). The step-like features are more likely to reflect sedimentary bedforms in loose sediment, resembling the crescent-shaped bedforms (CSBs) of mobile canyon-fill sediment that results either from mass-failure events or from erosion associated with cyclic steps formed by turbidity currents undergoing a hydraulic jump (Paull et al., 2010, 2011, 2013; Covault et al., 2014).

The proximal part of the BbC, immediately down-system from the canyon-head area, is characterized by sharp-edged erosional features including steep banks and reworked submarine landslide material, suggesting that erosional processes have shaped this area and transported mass-wasting material farther down canyon (e.g., Greene et al., 2002; Figs. 5A and 6A). The SPOC 25 seismic reflection line and an adjacent Parasound line (Figs. 7C, 7E) reveal that where the canyon floor is rugged there is little or no sediment fill, whereas if the canyon floor is smooth it is due to sedimentary infill (seismic reflection line SPOC 24, Fig. 8D). All areas with a smooth, low-gradient canyon floor are therefore interpreted to be areas of partial sediment fill (Fig. 5). The steepest slopes of the canyon walls revealed by the skewed slope-area curves (Figs. 5A, 5D) are interpreted to represent erosional escarpments and landslide scars in the canyon walls.

Adjacent to the knickpoint at 46 km from the canyon head (Figs. 7A, 7B), the northeastern canyon wall is interpreted to consist of a voluminous slump (Fig. 7C) that caused narrowing of the canyon floor (Fig. 5C). A secondary submarine mass movement occurs at the toe of the original slump block, showing a much higher degree of internal deformation (Figs. 7A, 7C). Parts of the toe of the secondary mass movement were subsequently removed from the canyon thalweg by erosional processes (Fig. 7). No significant smooth sediment fill (other than the lateral mass-transport deposits) can be identified along the canyon floor at this location (Figs. 5A, 7A–7C, and 7E), indicating an erosional rather than a depositional environment.

Observations. The 4–6-km-wide tributary originates ∼4.5 km north of the northern shoreline of Isla Santa María and extends northward for 10 km before merging with the main stem of the BbC (Fig. 2). The bifurcated tributary head is aligned with the –120 m isobath.

The tributary walls are steep in their upper reaches just below the margin, while the lower part of the walls have a concave, asymptotically shallowing shape (Fig. 4). The SPOC 24 seismic reflection profile reveals that the floor of the tributary is characterized by a seismic reflector package with a 140 ms TWTT thickness that onlaps onto the surrounding reflections of the host strata (Fig. 8D). If this package of onlapping, subhorizontal seismic reflectors at the bottom of the canyon is disregarded, the tributary has a more V-shaped asymmetric cross section (Fig. 8D). The asymmetry is expressed by a gently sloping margin (∼5°) on the western side and a steep margin (∼19°) on the eastern side (Fig. 8D).

High-resolution bathymetry reveals that an axial incision as much as 250 m wide extends from the western branch of the bifurcate canyon head for 10 km along the axis of the tributary (Figs. 8A, 8B). Its morphologic expression is smooth and subtle; it is barely resolved in the Parasound profile (Fig. 8E) and is too shallow to be resolved by seismic reflection data (Fig. 8C). The incision disappears ∼2.5 km upstream of the confluence with the main BbC (Fig. 8A). No signs of natural levee development along the incision can be observed (Figs. 8A, 8E). The outlet of the tributary forms a hanging valley that is located as much as ∼80 m above the BbC thalweg (Figs. 8B, 8C).

Interpretation. The seismic reflection profile shows that the tributary feature had an asymmetric V-shaped cross section before it was partially infilled with sediment (Fig. 8C). The geomorphology of the tributary feature differs from that of the BbC due to its lack of sinuosity, the shallow incision, the generally more gently dipping walls, and its limited length. The tributary feature meets the criteria for a submarine canyon (sensu stricto), because the original feature is an elongate submarine valley cut >500 m into the shelf bedrock and its walls are not bounded by any levees (Fig. 8D). Because the tributary runs into the BbC and does not extend all the way down the shelf, it is referred to as the BbC tributary canyon. The bifurcate head of this feature, its deeply incised original V-shaped profile, its elongate shape, and its layered sediment fill distinguish this feature from a simple ephemeral mass-wasting scar. It may, however, have been initiated by mass-wasting processes and then evolved under the influence of sediment-gravity flows, although at this stage it is not as well developed as the BbC.

The package of onlapping reflectors along the tributary canyon’s axis is interpreted to represent sediment fill (Fig. 8D); the sediments are the reason for the shallowing gradient of the tributary walls toward their lower reaches (Fig. 4). The original relief of the tributary canyon was subdued, but has not been entirely obliterated by the sediment fill. This sediment fill corresponds to 140 ms TWTT (Fig. 8C). Assuming P-wave velocities of between 1500 and 2000 m/s, this corresponds to an approximate canyon-fill thickness of 106–140 m.

The axial incision into the sediment fill, with no natural levees, is interpreted to represent an erosional channel. This incised thalweg is similar to, but not as pronounced as, incised canyon thalwegs described from the Mediterranean Sea (e.g., Baztan et al., 2005). The smooth, subtle morphology of the channel and the lack of sharp erosional edges suggest that the channel may be draped with hemipelagic sediments (e.g., Paull et al., 2013; Babonneau et al., 2013). However, because no acoustically transparent draping reflection package that could be interpreted as representing hemipelagic drapes can be clearly resolved in either the seismic reflection profile or the Parasound line (Fig. 8E), any such drape must be no more than a few decimeters thick.

Observations. The inner shelf of the study area is generally smooth (Fig. 5A), while the outer shelf is characterized by rugged topography with scattered smooth, mound-like features between (Figs. 5A, 9, and 10). Areas exhibiting elongate, mound-like depositional features on the outer shelf have been identified to the north (Fig. 9) and to the south (Fig. 10) of the BbC system.

Approximately 3.5 km north of the BbC, the mound-like feature is ∼4.8 km long, has a northeast-southwest orientation, and is slightly sinuous in plan view (Fig. 9A). The width of the feature varies between 200 and 750 m and its upper surface appears smooth (Figs. 9A, 9B). An extension with a maximum width of 400 m and a maximum height of 8 m extends for 1 km southeastward from the feature’s southwestern margin (Fig. 9A). The eastern side of the elongate mound is lined by a moat of ∼1 m depth (Figs. 9A, 9B). A second, smaller, irregularly shaped depositional mound occurs to the northeast of the elongate mound, extending for a maximum length of 1.2 km, with a maximum height of 3 m (Figs. 9A, 9C).

To the south of the BbC, the shelf shows a more complex pattern (Fig. 10). One elongate mound-like feature (0.2 × 1.1 km, maximum height 6 m; Fig. 10) has a northeast-southwest orientation similar to that of the feature described to the north of the BbC (Figs. 9A, 9B). The mound is located on the eastern edge of a morphologically complex area in which both rugged and smooth topographic features are present (Fig. 10). Elongate depressions with a southwest orientation are also commonly present to the southwest of the morphologic high. These depressions have a maximum depth of 17 m and maximum dimensions of 600 m length and 330 m width. Most of the southwest-northeast elongate depressions are to the west of rugged linear ridges or southwest of blocky rugged features (Fig. 10).

Interpretation. The smooth seafloor morphology of the inner shelf (Figs. 5A and 8A) is interpreted to represent a thin veneer of young, unconsolidated sediment mantling the underlying host strata (e.g., Paull et al., 2011, 2013). The rugged topography of the outer shelf is interpreted to represent bedrock outcrops, while the smooth mound-like features between appear to drape rugged outcrops (Figs. 9A and 10). The smooth parts of the outer shelf are therefore interpreted to represent piles of young unconsolidated sediment (Figs. 5A, 9, and 10) deposited between, and partly onto, bedrock. Because of their typical mound-like morphology and their elongation parallel to the continental margins, such sediment mounds have been interpreted to be sediment drifts of young, unconsolidated sediment composed of material that has been transported and shaped by near-bottom currents (e.g., Faugères and Mulder, 2011). The elongate depressions on the outer shelf are interpreted as areas in which sediment has been winnowed out from the down-current side of morphologic obstacles, including linear bedrock ridges and boulders. These scours essentially form modern flutes that indicate the current directions, and therefore currents in this area are shown to be directed obliquely off the shelf edge toward the southwest (Fig. 10).

Observations. Gravity core GC14 was recovered on top of a terrace, 40 m above the BbC thalweg and 473 m below the headscarp of the Biobío slide (Fig. 11B; Völker et al., 2011, fig. 3 therein) below a water depth of 1822 m. The core recovered 5.77 m of sediment; only the uppermost 2 m were examined in this study. The section below 1.70 m core depth is characterized by angular clasts of silty mudstone of various colors, with homogeneous silty clay interbeds (Fig. 11A; Völker et al., 2011). The chaotic mud-clast rich section is overlain by 1.70 m of silty clay comprising 6 upward-fining silt layers, each a few millimeters thick, and several silt lenses. The hemipelagic clay sampled at 0.81–0.93 m core depth yielded sufficient bulk planktonic foraminifera for this section to be dated as 5486 ± 250 calendar (cal) yr B.P. (2σ range).

A 1.70 m sedimentary section was recovered in the GeoB 9802 core from the distal northwestern fringe of the Biobío trench fan, 29 km west-northwest of the canyon mouth below a water depth of 4822 m (Figs. 1 and 11C; Heberer et al., 2010). The core was taken from within the trench, 3.5 km east of and 41 m higher than the trench’s axial channel. The sediment is composed of silty clay and clayey silt of dark olive-green to olive-gray intercalated with up to medium-grained sandy layers (Fig. 11A). Burrows filled with fine sand are present throughout the core. From 14 to 17 distinct turbidite events are expressed as medium- to fine-grained, 1–6-cm-thick, dark gray sand layers with erosional lower contacts and irregular upper contacts. Turbidite bases are often structureless and planar lamination is present in some of the sand layers (Fig. 11A). Turbidites at ∼0.50 m and 1.50 m core depth have been disturbed by bioturbation and soft-sediment deformation, making it difficult to determine the exact number of turbidites. The upper contacts of turbidites are mostly sharp, and fining-upward trends are both rare and subtle. The hemipelagic clay sampled at 0.88–0.92 m core depth yielded an age of 5081 ± 254 cal yr B.P. (2σ range).

Interpretation. The sedimentary section in the GC14 core below 1.70 m core depth was interpreted to have been affected by slumping and debris-flow depositional processes (Fig. 11A), possibly related to mass wasting associated with the large Biobío slide upslope (Völker et al., 2011; Fig. 11B). The thin, upward-fining silt and mud layers are interpreted to have been diluted silty turbidites representing T_d and T_e divisions (Bouma, 1962). There are 4 or 5 turbidites present above the level dated as 5486 ± 250 cal yr B.P. (2σ range), indicating a turbidite- recurrence rate of ∼1100–1370 yr (0.73–0.91 turbidites/k.y.) and a sedimentation rate of ∼15 cm/k.y. for the second half of the Holocene. Pore-water analysis in the GC14 core indicates that a few centimeters of surface sediment were lost during core recovery, because the total alkalinity is slightly higher than that of normal seawater (Völker et al., 2011). The sulfate concentration at the top of the core is, however, approximately equivalent to that of seawater (Völker et al., 2011), suggesting that sediment removal during coring was probably not extensive. The sedimentation rate presented here therefore represents a minimum rate and is not likely to have been significantly underestimated.

The turbidity-current events recorded in the GeoB 9802 core from the distal fan deposited beds that are significantly coarser grained and thicker than those in the GC14 core from an in-canyon terrace, mostly representing T_a and T_b turbidite divisions (Bouma, 1962). At least 8 and possibly 10 distinct turbidite events are present above the level dated as 5081 ± 254 cal yr B.P. (2σ error). The turbidite-recurrence rate is therefore ∼510–640 yr (1.57–1.97 turbidites/k.y.) and the sedimentation rate is 17 cm/k.y., integrated over the second half of the Holocene.

Heberer et al. (2010) reported that sediment samples in this core contained black basaltic sands composed of fresh, glassy, angular, volcanic fragments that are commonly highly vesicular. The sediment composition within the BbC suggests that it has not undergone extensive transport and sorting, or prolonged periods of subaerial weathering (Heberer et al., 2010). The Biobío Fan sediments are very low in quartz compared to other submarine fan sediments found along the Peru-Chile Trench between 36°S and 47°S, and have the strongest volcanic provenance signal (Heberer et al., 2010). These data are consistent with results obtained by Lamy et al. (1998) from marine surface sediments on a transect along 36°S and by Thornburg and Kulm (1987b).

Observations. Immediately north of the BbC and ∼23 km west of the coastline the continental shelf is characterized by a high degree of surface roughness with areas of rugged relief surrounded by smoother areas of lower relief (Figs. 12 and 13A). The rugged areas are dissected by linear asymmetric ridges of several kilometers length (Figs. 12 and 13A). These ridges have an approximate north-northeast orientation and an elevation that is as much as 14 m above the rugged areas (Fig. 13C). To the west, the ridges form mostly linear features while to the east they are rhombohedral (Fig. 13A). Maximum elevations from a swath profile reveal that the ridges are located in an area of enhanced seafloor elevation, with a maximum relief of 17 m (Fig. 13D). This area of upwarped seafloor and linear ridge topography is located directly adjacent to the large slump on the BbC wall documented in the SPOC 25 seismic reflection profile, the BbC thalweg knickpoint, and an abrupt reduction in canyon width (Figs. 5C, 7C, and 12).

Interpretation. Our analysis of the seafloor morphology has been combined with previous interpretations of industry seismic reflection lines (Empresa Nacional del Petróleo, Chile, ENAP 17, D4–13; Melnick et al., 2006) to reveal the presence of several northeast-striking blind reverse faults and associated fault-propagation folds (Fig. 12). More specifically, seismic reflection line ENAP 17 (Melnick et al., 2006, fig. 10 therein) reveals an asymmetric anticline ∼7.5 km east-southeast of the tributary axis (Fig. 12). The discontinuous offset reflectors beneath the anticline have been interpreted to be related to a west-dipping reverse fault that was responsible for the formation of the anticline (Melnick et al., 2006). The crest of the anticline is obscured by horizontal seafloor multiples at depths of ∼300 and ∼400 ms (TWTT), and is not expressed in the present-day seafloor morphology (Fig. 12). However, the Isla Santa María, located ∼10 km south of the ENAP 17 seismic reflection line (Melnick et al., 2006, fig. 2b therein), has been interpreted to be associated with this fold. Aligned microseismicity suggests that the reverse fault is rooted in the interface between the oceanic Nazca plate and the South American plate at ∼13 km depth (Melnick et al., 2006).

Because of their rugged topography, the high-roughness areas are interpreted as outcrops of bare bedrock. The linear ridges that cut across these bare bedrock areas are interpreted to represent normal faults along anticline crests (e.g., Morley, 2007). The presence of anticlines in this area is supported by localized seafloor warping (as revealed in the swath profile; Fig. 13D) due to ongoing shortening. The presence of two anticlines in this area can be inferred from the shape and orientation of the normal fault ridges, one to the west with linear normal faults along its crest, and the other to the east with rhombohedral normal fault patterns (Fig. 13A), indicating a component of strike-slip kinematics. The ENAP 17 seismic reflection line (Melnick et al., 2006) suggests that these two anticlines may be linked at depth by a shallow ramp-flat structure. During the Maule megathrust earthquake in February 2010 (Mw = 8.8), movement along the blind reverse thrusts caused the growth of a fault-propagation anticline, resulting in steep, margin-parallel tilting and newly formed normal faults on Isla Santa María and the adjacent seafloor (Melnick et al., 2012), analogous to the normal faulting on the seafloor to the north of the BbC inferred from our study.

Several lines of evidence indicate that the main stem of the BbC maintained moderate levels of activity during the Holocene sea-level highstand. Sediment cores record Holocene turbidite deposition within the canyon and on the submarine fan. Furthermore, within the BbC head, step-like bedforms made up of loose sand and mud are interpreted as crescent-shaped bedforms (CSBs; sensu Paull et al., 2010). The mobile sediment within the canyon is subsequently reworked into CSBs, either by mass-failure processes or by erosion associated with cyclic steps formed by turbidity currents undergoing a hydraulic jump (Paull et al., 2010, 2011, 2013; Covault et al., 2014). CSBs in the BbC head appear particularly similar to the CSBs developed in the La Jolla Canyon (offshore California; Paull et al., 2013, fig. 6 therein), and similar bedforms have also been observed in several other active California canyons, including the Monterey, Hueneme, Mugu, and Redondo canyons (Paull et al., 2008, 2010, 2011, 2013), in the active Saint-Etienne Canyon system off La Réunion (Babonneau et al., 2013), and in four British Columbia fjords (Conway et al., 2012; Hughes Clarke et al., 2014). It has been suggested that CSBs are typical of active submarine canyons and are absent from inactive ones (Babonneau et al., 2013; Paull et al., 2011).

The present-day discharge of the Biobío River is ∼33 km³ of water per year (Milliman and Farnsworth, 2011). Because of this high discharge, the Biobío River is unable to produce hyperpycnal currents, according to the equations of, and using estimated sediment concentrations from flooding events of Mulder and Syvitski (1995), together with measured sediment concentrations (corrected for damming) from Fernández Valenzuela (2002). Initiation of down-canyon transport of the Biobío River sediment may be due to sediment failure of oversteepened sediment piles (e.g., Hughes Clarke et al., 2014), possibly combined with seismic shaking associated with megathrust earthquakes that liquefied sediments in the vicinity of and within the canyon head. However, because there is no paleosediment or paleowater discharge information available for the Biobío River, it remains unclear whether hyperpycnal flow occurred during the Holocene and late Pleistocene—the time scale relevant for the present-day geomorphology of the BbC.

In-canyon mass wasting is responsible for knickpoint formation and canyon widening, but most of the displaced material has been subsequently evacuated from the canyon thalweg. The described knickpoint (Figs. 7A, 7B) is interpreted to have been generated by canyon-wall mass wasting (e.g., Greene et al., 2002; Paull et al., 2011) that probably deposited part of the transported material along the canyon thalweg, creating a barrier. The landslide in the northern wall of the BbC (Figs. 7A, 7C) is inferred to be related to anticlinal fold growth directly to the northeast (e.g., McAdoo et al., 2000), as evidenced by crestal normal faults and seafloor upwarping (Figs. 12 and 13). The secondary mass-wasting feature at the toe of the slumped material (Figs. 7A, 7C) may be due to instabilities in the canyon wall induced by axial incision and erosional undercutting (e.g., Pratson and Coakley, 1996; Baztan et al., 2005). The slump deposits along the BbC thalweg have now been re-eroded by sediment-gravity flows, but these erosive flows have not yet erased the knickpoint. Removal of mass-wasting material from the canyon floor is also suggested farther down-canyon in the vicinity of the large, multiphase Biobío submarine landslide 16 km west of the shelf edge (Figs. 1 and 11; Völker et al., 2011). There appears to be ∼1.75–2.0 km³ of sediment missing, inferred to have been transported away through the BbC (Völker et al., 2011). While the timing of the onset of landsliding in this area is unknown, the latest retrogressive failure is reflected in core sediments recovered from the canyon terrace (Fig. 11A; Völker et al., 2011). The failure has previously been interpreted to have occurred ∼1–0.7 k.y. ago (Völker et al., 2011); however, our new radiocarbon dating of the sediment above the mass-transport deposits suggests that the event predates 5486 ± 250 cal yr B.P. (GC14; Fig. 10A). The event presumably either deposited material within the canyon that was later removed by erosional processes or resulted in material being transported directly along the canyon. The lack of accumulated mass-wasting material in the BbC thalweg hints to erosional sediment-gravity flows that have been active in the BbC. The exact timing of material evacuation remains elusive, but probably predates 5.5 k.y. ago in the case of the latest retrogressive failure of the Biobío landslide.

Hemipelagic drape is an indicator of canyon inactivity. Neither the SPOC 25 seismic reflection line nor the corresponding Parasound line (Figs. 7C, 7E) shows any acoustically transparent reflection package that could be interpreted as a hemipelagic sediment drape (e.g., Maier et al., 2011; Walsh et al., 2007).

The provenance of the fan sediments in the Peru-Chile Trench also provides valuable clues concerning the transport processes operating within the BbC. The freshness of the volcanic, vesicle-rich particles in the turbidites of the GeoB 9802 core from the canyon fan (Heberer et al., 2010) suggests direct rapid transport of the Andean basaltic black sands, passing through the mouth of the Biobío River into the BbC, even during the Holocene sea-level highstand.

A number of observations from previous analyses as well as from our study suggest that moderate activity within the BbC has been maintained throughout the Holocene sea-level rise and highstand. These observations include (1) terrigenous detritus in gravity cores that has been deposited by turbidity currents and hemipelagic settling during the Holocene; (2) largely unaltered vesicle-rich volcanic particles in turbidites found within the canyon fan (Heberer et al., 2010); (3) the proximity of the canyon head to terrestrial sediment sources; (4) steep erosional features having a young geomorphic expression within the canyon; (5) numerous canyon-wall mass failures on a variety of scales (as would be expected in a region of high tectonic activity), but low accumulation of slide debris along the canyon’s axis; (6) the development of CSBs in the canyon head, which have been interpreted to occur only in active canyons; and (7) the absence of any acoustically transparent reflection package in seismic or Parasound profiles that could be interpreted to represent hemipelagic drape.

The tributary canyon is interpreted to have undergone four distinct evolutionary stages, involving (1) erosive formation of the main valley; (2) partial infill of the canyon with sediment; (3) incision of an axial channel along the thalweg, cutting through the canyon fill; and (4) possible formation of a thin drape of hemipelagic sediments, covering part of the canyon floor and smoothing the morphologic expression of the axial channel. The tributary initially formed by asymmetric incision into the continental shelf. Differential uplift of the seafloor to the west of the tributary due to tectonic shortening (Melnick et al., 2006) may have induced the asymmetry of the tributary’s pre-fill topographic profile (Fig. 8C). The second phase of the tributary evolution involved its partial infilling with ∼90–120 m of sediment. Because of the absence of core data it remains unclear whether this sediment accumulation is of hemipelagic or turbiditic origin. The disturbed reflectors that can be observed in the deepest part of the tributary may reflect mass-failure events (Fig. 8D).

In the third phase, a minor axial channel was cut into the sediment fill during its final erosive episode as a result of canyon rejuvenation. Several submarine canyons in the Gulf of Lyon (Mediterranean) are characterized by an axially incised channel within the main part of the canyon and the heads of these axially incised canyons correspond to the mouths of fluvial systems (Baztan et al., 2005). Baztan et al. (2005) suggested that the most probable mechanism for rejuvenated canyon incision is erosion by turbidity currents, associated with a direct connection with a fluvial system, or the downslope transport of mass-wasting material. Following the ∼120 m lowering of sea level during the Last Glacial Maximum (LGM, ∼19 k.y. ago; Siddall et al., 2003), parts of the present-day continental shelf were subaerially exposed and formed a coastal plain with part of the coastline close to the head of the tributary. It is not clear if, or where, the rivers of the Arauco Peninsula traversed the shelf during this lowstand; no high-resolution bathymetry data are available for this part of Arauco Bay and it is therefore not possible to identify any paleovalleys that may have traversed the shelf and were associated with processes that deposited sediment directly into the canyon system. Hebbeln et al. (2007) used radiocarbon dating of marine sediments in cores from off the coast of Chile to detect an increase in terrigenous sediment accumulation rates from the late Pleistocene to the Holocene; they showed that the delivery of terrigenous sediment into the region increased during the LGM, with glacial submarine sediment accumulation rates being consistently higher by a factor of ∼1.6 than during the Holocene (Hebbeln et al., 2007). The amount of terrigenous sediment delivered to the heads of the BbC and its tributary was therefore probably greater by a similar factor during the late Pleistocene compared to the Holocene. For the tributary canyon, the combination of readily available unconsolidated sediment on the exposed shelf and possible increased fluvial transport of this sediment to the vicinity of the canyon head may have resulted in sediment-gravity flows, initiated by mass failure, that were sufficiently erosive to cause the incision along the canyon thalweg.

Because the axial incision along the tributary canyon was discontinued, the mouth of the tributary forms a hanging tributary to the main stem of the BbC, with a distinct knickpoint at the confluence (Figs. 4 and 8A–8C). The knickpoint appears to be stationary. This morphology suggests that greatly reduced or terminated sediment gravity-flow activity is responsible for preserving the knickpoint as a hanging valley, while the main stem of the BbC has continued cutting into the host strata. Similar relationships have been described from canyons in the Mediterranean Sea offshore southern France and northeastern Spain: the Hérault Canyon forms a 400-m-high hanging valley at its confluence with the Séte Canyon (Baztan et al., 2005), the Hirta Canyon is located 60 m above the Valencia Channel (Amblas et al., 2011), the western branch of the Foix Canyon is 220 m above the thalweg of the eastern branch, the Cunit Canyon mouth is 104 m above the thalweg of the Foix Canyon, and the outlet of the Valldepins tributary canyon is 150 m above the floor of the Foix Canyon (Tubau et al., 2013). In the California Monterey Canyon system, all tributary canyons form hanging valleys, indicating that the Monterey Canyon is currently the dominant conduit for terrigenous sediment transfer (Greene et al., 2002).

The geomorphic differences between the main stem of the BbC and its tributary are interpreted to be mainly due to differences in their present and past levels of activity with regard to sediment transport by gravity flows, gravity-flow transport efficiency, erosion, and the timing of canyon evolution. The tributary has been partially infilled with sediment, while the main BbC is still being excavated. Tributary walls are relatively shallow and do not seem to foster mass failure that may transform into turbidity currents. Because of the absence of any sharp erosional features, the smooth appearance of the canyon floor, and the axial incision, all of which are suggestive of a thin hemipelagic drape, and its hanging outlet above the BbC floor, the tributary is interpreted as being much less subjected to erosive sediment-gravity flows than the main stem of the BbC; the tributary is thus considered to be inactive.

Bottom currents can act as important sediment redistributors on the seafloor and as additional sources of terrigenous sediment input to submarine canyons. The poleward-flowing Gunther Undercurrent (Fonseca, 1989; Strub et al., 1998), which flows toward the south offshore coastal central Chile, is strongest at 150–300 m water depth. It therefore flows mainly over the outer shelf, the shelf edge, and the upper continental slope and can induce alongshore, southward-directed sediment transport. The undercurrent is characterized by significant seasonal and interannual variability, including reversals, but it generally flows southward with a mean velocity of 0.128 m/s and a maximum velocity of 0.689 m/s (measured over a 6 yr period at 30°S ∼750 km north of the study area) at a water depth of 220 m (Shaffer et al., 1995, 1999; Shaffer and Pizarro, 1997; Pizarro and Shaffer, 2002). The kinetic energy of north-south–directed bottom currents on the northern margin of the canyon is at least one order of magnitude greater than on the southern margin (80 m of water depth; Sobarzo et al., 2001) or close to the canyon head, indicating an important discontinuity in the longshore flow. Current moorings in 200 m of water just a few kilometers north of the BbC revealed southward-directed bottom currents that intensified southward toward the BbC (Sobarzo and Djurfeldt, 2004). These currents are probably responsible for the nondeposition and/or erosion of a Holocene sediment drape in the vicinity of the normal-faulted bedrock platforms and on morphologic highs, especially on the outer shelf where the currents are strongest (e.g., Raitzsch et al., 2007; Völker et al., 2014).

The new high-resolution data set provides new insights that allow further refinement of the previously proposed sediment dispersal patterns on the Chilean shelf in the Concepción area, allowing for greater detail in describing specific modes of sediment removal from the continental shelf. The inner shelf appears smooth and mantled by a veneer of young, unconsolidated sediment (Figs. 5A and 8A). In the outer shelf region, bedrock exposure is increasingly common (Figs. 5A, 10, and 13), attesting to the increased importance of bottom-current erosion and sediment removal. This observation is supported by the results from previous studies ∼90 km north of our study area along a 36°S transect (Lamy et al., 1998), where the outer shelf and the uppermost slope sediments are coarser grained than in areas more proximal to the coast, probably due to winnowing by bottom currents and/or to resedimentation processes. Bedrock exposure due to the influence of the coast-parallel, southward-directed undercurrent has been reported from the Peruvian shelf at a slightly deeper water depth (180 m), where sediment accumulation only occurred in niches protected from the undercurrent (Reinhardt et al., 2002).

Along the outer shelf in our study area, unconsolidated sediment was redeposited by bottom currents to form sediment drifts (Figs. 9 and 10). These sediment-drift bodies are subject to constant relocation and reshaping by near-bottom currents. Sediment drifts formed by bottom-contour currents are considered to range from small patch drifts with an areal extent of ∼100 km², to giant elongate drifts covering more than 100,000 km² that are as much as 500 km long and 100 km wide, with a positive relief of as much as 2000 m (Faugères and Mulder, 2011, and references therein). The sediment drifts on the Concepción shelf are about two orders of magnitude smaller than those that have been described from other continental margin settings (e.g., the Argentine Basin or the eastern New Zealand margin; Faugères and Mulder, 2011). However, the small drifts may be more common than previously recognized and may have been largely unnoticed due to low-resolution bathymetric and seismic reflection data or due to the obliteration of these features in the geologic record by compaction processes.

The south-directed currents may also have had a distinct influence on the canyon evolution. These currents carry sediment into the BbC, sediment that is then evacuated westward toward the trench. The tributary is located downcurrent from the main-stem canyon, with regard to bottom-current direction, which means that it cannot act as an active conduit for sediment carried by bottom currents. Southwestward-flowing bottom currents at the edge of the continental shelf carry the Holocene material over the edge and onto the continental slope.

Turbidite recurrence and sedimentation rates vary between sediment core sites, and the depositional environment of the sediment sample needs to be taken into account when assessing sedimentary processes in canyon systems and adjacent areas. Sedimentation rates obtained from both the in-canyon terrace and the distal sectors of the Biobío Fan are similar, ranging between 15 and 17 cm/k.y., with mean recurrence rates of 0.7–0.9 turbidites/k.y. and 1.6–2 turbidites/k.y., respectively. The lower turbidite-recurrence rate and the more distal and fine-grained sedimentary character of the turbidite layers in the GC14 core are interpreted to be due to the core location. The terrace from which the GC14 core was obtained is more than 40 m above the present BbC thalweg (Fig. 11B), and turbidity-current clouds that are <40 m thick are thus not expected to leave any deposit on the surface of the terrace. The turbidite-recurrence rate of ∼1.1–1.4 k.y. therefore only represents the largest flows and is a minimum for the BbC, only representing the recurrence rate for large turbidity currents with a vertical extent >40 m. In addition, the grain-size distribution within a moving turbidity current is such that the coarsest particles are carried at the base of the flow and within its turbulent head, while the upper (dilute) turbidity cloud carries the finest grains (Stacey and Bowen, 1988; Kneller and McCaffrey, 1999). Therefore, only the upper turbidity cloud of a gravity flow is likely to pass over the terrace, depositing thin, fine-grained turbidites, while the coarser grained, higher density turbidity clouds remain confined to the deepest parts of the canyon.

Turbidite layers in the GeoB 9802 core are relatively coarse, thick, and have sharp tops in the sand layers, with no current ripples (T_c turbidite division; Bouma, 1962; Fig. 11). These sedimentary characteristics suggest that the sand was deposited rapidly, either by suspension fall-out (T_a) or as high-velocity planar beds (T_b). The finer grained sandy and silty material within this flow probably bypassed this site, and flows continued waning farther down the depositional system, depositing sediments from T_c, T_d, and T_e turbidite divisions. The GeoB 9802 core was recovered from the outer fringe of the submarine fan, close to the axial channel of the trench. The thinning pattern of the fan from the mouth of the BbC toward the west and northwest suggests that most of the sediment accumulation in the fan was sourced from the BbC, but a minor quantity of the sediment in the core may have been derived from the trench-axial channel.

Due to their core location, the in-canyon and fan-sedimentation rates are interpreted to represent minimum rates because they were measured on an elevated in-canyon terrace and on the outer fringe of the fan. When these sedimentation rates are compared to those obtained from outside of the canyon-fan system on the continental slope, the differences are minimal (Fig. 14). The GeoB 7165–1 core was recovered from the continental slope, 10 km west of the shelf edge and 13 km northeast of the GC14 core, outside the canyon and at a water depth of 787 m (Figs. 1 and 11B), and yielded a sedimentation rate of 15 cm/k.y., averaged over the past 11 k.y. (Mohtadi et al., 2008). The 22SL gravity core, taken from the continental slope at a water depth of 1000 m, and 41 km north of the GC14 core, revealed higher out-of-canyon sedimentation rates of 25 cm/k.y. averaged over most of the Holocene (10 k.y. ago to present; De Pol-Holz et al., 2010; ages have been recalibrated using Calib 7.0 and the Marine 13 calibration curve [Reimer et al., 2013; http://calib.qub.ac.uk/calib/]).

To put the turbidite-recurrence rates and sedimentation rates (and therefore activity within the BbC) into a broader perspective, we compared the obtained rates to those reported from other, currently active canyons and fans around the world (Fig. 14). The systematic analysis of these sedimentation rates averaged over irregular time intervals has been shown to introduce a bias resulting in higher sedimentation rates for shorter measured time intervals, and lower rates for longer intervals (Sadler, 1981). To avoid this bias, we compared sedimentation rates averaged over similar time intervals and the time interval over which the rates were averaged has been reported in each case (Fig. 14).

A wide range of sedimentation rates has been recorded for submarine fans during the Holocene (Fig. 14). The Swatch of No Ground Canyon in the Bay of Bengal has served as an active sediment conduit to the Bengal fan during the Holocene sea-level rise, and is connected to a channel-levee system downslope (Weber et al., 1997). Sedimentation rates within the inner levees were reported to be >4 times higher than on the Biobío fan, but outer-levee sedimentation rates are similar to rates on the Biobío Fan (Fig. 14; Weber et al., 1997). In contrast, the Rhône Fan is growing more slowly than the Biobío Fan (Dennielou et al., 2009). Sedimentation rates in the overbank areas of the Nile Fan (Ducassou et al., 2009) are lower than in the Biobío Fan, but accumulation occurs more rapidly within a fan channel (Fig. 14). Every core site introduces a certain bias to estimations of sedimentation rates, the amount of activity within a canyon, and/ or the rate of fan growth. The sedimentation rates summarized here should therefore not be interpreted as being representative of bulk canyon sedimentation or fan growth, but they do provide a framework that can be used to put the level of activity in the BbC during the Holocene into perspective. A comparison of the minimum sedimentation rates obtained from the BbC and its fan system with several other canyon and fan systems suggests that the BbC system has been the site of moderate activity since the mid-Holocene (Fig. 14).

The Capbreton Canyon in the Bay of Biscay provides an example of a submarine canyon that was highly active during the Holocene with canyon head and in-canyon terrace-sedimentation rates of 164–168 cm/k.y. (averaged over 4 and 7 k.y. ago to present) (Fig. 14; Brocheray et al., 2014). In contrast, out-of-canyon sedimentation rates were ∼24 times lower (Fig. 14; Brocheray et al., 2014). Annual turbidite frequencies recorded from a 125-m-high terrace above the canyon thalweg are interpreted to be related to storm events (Mulder et al., 2012; Brocheray et al., 2014). The Kushiro submarine canyon offshore Hokkaido records similarly high in-canyon sedimentation rates of 93–126 cm/k.y. (3.8–2.8 k.y. cal B.P. to present), respectively, but these reduce to 27–30 cm/k.y. farther down the canyon (11–10 k.y. cal B.P. to present; Noda et al., 2008). Out-of-canyon sedimentation rates obtained from a continental-slope terrace located 10 km east of the Kushiro Canyon were 54 and 50 cm/k.y. when averaged from 8 and 11 k.y. cal B.P. to present, respectively (Fig. 14; Noda et al., 2008).

The Capbreton Canyon is the main sediment conduit and sediment sink in the southern Bay of Biscay. In contrast, the out-of-canyon slope terrace off the Japanese coast shows sedimentation rates similar to those recorded along the Kushiro Canyon thalweg. Similarly, slope-sedimentation rates in the BbC area do not differ greatly from the minimum in-canyon rate (Fig. 14). This sedimentation pattern suggests that, during the Holocene, the forearc basins and plateaus, the BbC, and the Biobío Fan in the Peru-Chile Trench formed equally important sediment sinks (Fig. 14). The combination of two conditions is interpreted to be responsible for this sedimentation pattern: first, the high fluvial sediment input from the Biobío River and second, the subsequent redistribution of sediments by bottom currents that are strong at the depth of the shelf edge and sweep material off the shelf edge and onto the slope (Fig. 10).

The results of our study suggest that the ability of a canyon to act as an active conduit for sediment to the deep ocean during sea-level highstands can be simultaneously controlled by several different factors, and may vary within an individual canyon system (Fig. 15).

In the case of the BbC, there are three main factors.

1. The BbC axis is oriented perpendicular to the area with the highest shelf gradient. Such a morphologic situation causes minimal coastal retreat during transgression and facilitates headward erosion of a canyon. The canyon-axis orientation is partly controlled by the underlying regional geologic structures. Within the study area, the proximity and parallelism of the tributary axis to an anticline suggest that the orientation of the tributary axis may be controlled by the underlying northeast-striking blind reverse faults and associated fault-propagation folds (Figs. 12 and 15). In contrast, the BbC main stem is oriented perpendicular to the orientations of the main tectonic structures; however, the position of the main meander bends on the shelf seems to be controlled by tectonic structures (Fig. 12). The initial development of the BbC system, including its tributary, on the continental shelf seems to have been associated with orientations of major fault-associated folds following structurally generated seafloor relief, very similar to the Cook Strait Canyon off the coast of New Zealand, where a tributary canyon is aligned with a major thrust fault (Mountjoy et al., 2009).

A narrow shelf width facilitates the maintenance of a canyon head-to-shore connection (Covault et al., 2007; Covault and Graham, 2010). Harris and Whiteway (2011) used the ETOPO1 bathymetric grid (www.ngdc.noaa.gov/mgg/global/global.html) to show that 4.6% of the canyons on tectonically active margins (where shelves are generally narrower than along passive continental margins) cut into the continental shelf and connect to a river system, but only 1.5% of the canyons on passive margins show this relationship. On a regional scale, the effect that shelf width and gradient have on headward erosion and activity in a canyon can vary within a single canyon system. In our study area, where the seafloor gradient from the coast to the –120 m contour is steep, eastward coastal retreat during Holocene sea-level rise was minimal (5–8 km; Figs. 2 and 15). Therefore, the connection of the Biobío River mouth and the BbC head could be maintained, connecting the fluvial sediment source of the Biobío River to the BbC. The opposite is true for the tributary canyon and the Arauco Bay area; here, the ∼120 m Holocene sea-level rise has shifted the shoreline at least 24 km southward from the head of the tributary canyon and away from fluvial sediment sources (Figs. 2 and 15). The abandonment of the tributary during the Holocene sea-level rise is interpreted to have been particularly effective because of the large distance (24–38 km) of shoreline retreat. The location of an anticline and related structures on Isla Santa María between the head of the tributary and the present-day coastline may have obstructed further incision of the canyon during sea-level rise, and the northeast-oriented fold axes bounding the tributary may also have formed an additional obstacle to its headward erosion (Fig. 12).

Apart from the issue of canyon-head connectivity to a fluvial system, sustained connections to additional sediment sources can also maintain canyon activity during marine transgressions and sea-level highstands. This is the case for La Jolla Canyon, California, which is not connected to any fluvial system but intercepts a littoral cell that allowed the submarine fan to grow during the Holocene sea-level rise and ensuing highstand (Covault et al., 2007). The BbC head location close to the coast allows for the interception of sediment transported northward within the littoral cell. There are no quantitative data available for the BbC area on the transport capacity of the longshore littoral cell, but Pineda (1999) highlighted its significance for sand distribution along the coast off Concepción and into the BbC head. The tributary head is located too far off the coast to receive sediment from northward-drifted sediment of the littoral cell.

2. The BbC axis is oriented perpendicular to bottom currents, facilitating the capture of sediment transported by these currents. High-resolution bathymetric data reveal that sediment transported by bottom currents serves as a third source of terrestrial material (in addition to direct input of Biobìo River fluvial sediment and the littoral cell) (Fig. 15). A canyon axis oriented perpendicular to the bottom-current direction ensures maximal sediment capture, as in the case of the BbC. The location of the tributary is downcurrent of the BbC main stem with respect to bottom currents, so that sediment transported by these currents cannot contribute to its activity.

3. The steep BbC canyon walls, which are located along zones of active tectonic deformation, foster mass-failure events that feed sediment into the canyon, which may subsequently transition into turbidity currents. In contrast, sediment infilling has lowered canyon-wall gradients in the tributary, making them less vulnerable to failure.

The ability of a submarine canyon to maintain activity during sea-level highstands can be controlled by several factors simultaneously, and these factors may vary within a single canyon system. Offshore the city of Concepción in south-central Chile, the BbC main stem receives terrestrial sediment from three sources: (1) direct input from the Biobío River through the river-mouth to canyon-head connection; (2) reworked terrestrial sediment from southward-directed bottom currents; and (3) sediment from the longshore littoral cell. The maintenance of the river-mouth to canyon-head connection and connectivity with the littoral cell has in turn been facilitated by minimal coastal retreat during the Holocene sea-level rise due to the steep shelf gradient. In addition, the BbC axis is oriented perpendicular to the prevailing bottom currents and aligned in an optimal way to receive sediments transported by bottom currents. Moreover, sediment-gravity flows may be generated from the collapse of steep canyon walls that may be oversteepened by tectonic deformation and/ or lateral undercutting of canyon walls by erosional sediment-gravity flows. As a result of this multisourcing, the BbC main stem has remained moderately active (from the mid-Holocene to the present) when compared to other examples worldwide.

In contrast, the BbC tributary no longer appears to serve as an active conduit for sediment-gravity flows, despite being characterized by a deep channel incised into the continental shelf. The tributary is disconnected from fluvial sources and the littoral cell as a result of a gentle shelf gradient of the Arauco embayment and associated long-distance shoreline retreat during the Holocene sea-level rise. Moreover, the BbC is located upcurrent with regard to bottom currents. Therefore, the tributary is isolated from several potential sediment sources and remains inactive.

Sediment delivery to the continental slope seems just as efficient as to the BbC and the submarine fan, which is interpreted to be related to the redistribution of terrestrial sediment by shelfal bottom currents. Therefore, slope-depositional environments may constitute equally important sediment sinks along the Concepción continental margin during the Holocene.

The submarine canyon example of the BbC demonstrates that activity in submarine canyons along convergent margins, and their role in transporting large quantities of sediment and associated pollutants, nutrients, and organic carbon to the deep seafloor during sea-level highstands, is controlled by several local variables, including bottom-current direction, structural deformation of the seafloor affecting canyon location and canyon-wall stability, shelf gradient, and fluvial networks. Taken together, these factors may produce variations in activity levels of submarine canyons, not only between canyons on different continental margins, but also within individual canyon systems.

We thank the Servicio Hidrográfico y Oceanográfico de la Armada de Chile for acquiring, processing, and providing the high-resolution bathymetric dataset. The German Federal Institute for Geosciences and Natural Resources (BGR, Hannover) kindly provided the SPOC (Subduction Processes Off Chile) seismic reflection lines. We are indebted to D. Völker, B. Heberer, and A. Kopf for providing information on sediment cores and sediment-core samples from the core-storage facilities of the Helmholtz Centre for Ocean Research (GEOMAR) Kiel and the MARUM Research Center Bremen. V. Viert was of tremendous help during the preparation of radiocarbon samples. We thank C. Paull for invaluable insight to crescent-shaped bedforms around the globe. The review of an earlier version of the manuscript by D. Völker greatly improved the focus of this contribution. We thank Editor T. Wawrzyniec, Associate Editor A.B. Rodriguez, J. Covault, A. Billi, and two anonymous reviewers for their constructive comments and criticism that helped to improve and focus the manuscript. A. Bernhardt was funded by the DFG (Deutsche Forschungsgemeinschaft) Leibniz Center for Surface Process and Climate Studies to M. Strecker (DFG grant STR 373/16-1) and by DFG grant BE 5070/1-1. D. Melnick was funded by DFG grant ME 3157/1-2/2-2. J. Jara was supported by DFG grant STR 373/30-1 to M. Strecker.