Late Quaternary evolution of the San Antonio Submarine Canyon in the central Chile forearc (∼33°S)

Marine Geology 188 (2002) 365^390 www.elsevier.com/locate/margeo

Late Quaternary evolution of the San Antonio Submarine Canyon in the central Chile forearc (V33‡S) Jane Laursen a; , William R. Normark b a

Department of Earth Sciences, University of Aarhus, 8000 Aarhus C, Denmark b U.S. Geological Survey, Western Region, Menlo Park, CA 94025, USA Received 10 January 2001; accepted 7 June 2002

Abstract Hydrosweep swath-bathymetry and seismic-reflection data reveal the morphology, sedimentary processes, and structural controls on the submarine San Antonio Canyon. The canyon crosses the forearc slope of the central Chile margin for more than 150 km before it empties into the Chile Trench near 33‡S latitude. In its upper reaches, the nearly orthogonal segments of the San Antonio Canyon incise V1 km into thick sediment following underlying margin-perpendicular basement faults and along the landward side of a prominent margin-parallel thrust ridge on the outer mid-slope. At a breach in the outer ridge, the canyon makes a sharp turn into the San Antonio Reentrant. Resistance to erosion of outcropping basement at the head of the reentrant has prevented the development of a uniformly sloping thalweg, leaving gentle gradients ( 6 2‡) up-canyon and steep gradients ( s 6‡) across the lower slope. Emergence of an obstruction across the head of the San Antonio Reentrant has trapped sediment in the midslope segments of the canyon. Presently, little sediment appears to reach the Chile Trench through the San Antonio Canyon. The development of the San Antonio Canyon was controlled by the impact of a subducted seamount, which formed the San Antonio Reentrant and warped the middle slope along its landward advancing path. Incision of the canyon landward of the outer mid-slope ridge may be ascribed to a combination of headward erosion and entrenchment by captured unconfined turbidity currents. Flushing of the canyon was likely enhanced during the lowered sea level of the last glaciation. Where the canyon occupies the triangular embayment of the reentrant at the base of the slope, sediment has ponded behind a small accretionary ridge. On the trench floor opposite the San Antonio Canyon mouth, a 200-m-thick levee^overbank complex formed on the left side of a distributary channel emanating from a breach in the accretionary ridge. Axial transfer of sediment was inhibited to the north of the San Antonio Canyon mouth, which left the trench to the north sediment starved. Between V32‡40PS and 33‡40PS, the Chile Trench axial turbidite channel deeply incises the San Antonio distributary complex. This entrenchment may have been initiated when the barrier to northward transport was eliminated. 9 2002 Elsevier Science B.V. All rights reserved.

1. Introduction * Corresponding author. Present address: Department of Earth Sciences and Engineering, Imperial College, RSM Building, Prince Consort Road, London SW7 2BP, UK. E-mail addresses: [email protected] (J. Laursen), [email protected] (W.R. Normark).

Submarine canyons are recognised in a variety of settings on both continental passive and active margins (e.g., Farre et al., 1983; Hagen et al., 1994; Normark and Piper, 1969), as well as on

0025-3227 / 02 / $ ^ see front matter 9 2002 Elsevier Science B.V. All rights reserved. PII: S 0 0 2 5 - 3 2 2 7 ( 0 2 ) 0 0 4 2 1 - 8

MARGO 3191 19-8-02

366

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

Fig. 1. Regional setting of the South American convergence zone between 22‡S and 47‡S. Topography and bathymetry is from ETOPO2 (Wessel and Smith, 1991; Smith and Sandwell, 1997). The study area of the CONDOR investigation is shown with the square. Active volcanoes (black triangles) shown are from Smithsonian Institution, Global Volcanism Program. Relative direction of plate convergence (DeMets et al., 1994) indicated by thick arrow.

MARGO 3191 19-8-02

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

many island arcs (e.g., Klaus and Taylor, 1991), and signi¢cantly in£uence depositional patterns and processes on continental margin slopes, rises and abyssal plains, as well as within deep-sea trenches. The position and con¢guration of submarine canyons are controlled by multiple factors, including structural fabric, tectonism, sea-level variations, and sediment supply. Models for submarine canyon formation were developed from studies of the canyon systems crossing the U.S. Atlantic margin that suggested localised slope failure initiates the process of canyon evolution (Farre et al., 1983). By headward erosion, the depressions extend up-slope to form linear sediment chutes. Where the canyon heads breach the shelf break, entrenchment is augmented by the capture of shelf-derived material, and this sediment bypassing tends to modify canyon planform. Continued indentation into the shelf is intensi¢ed where £uvial drainage systems may be captured during sea-level lowstands (Farre et al., 1983). On the eastern continental margin of the Bering Sea, canyon evolution follows a similar model in which pervasive mass wasting and density £ows generated by submarine slides are responsible for the greatest amount of excavation (Carlson and Karl, 1988). The position and con¢guration of the Bering margin canyons are controlled by underlying rock type and structural fabric, some of which may be inherited from a time of subduction and tend to re£ect natural cycles in sea-level oscillations and sediment supply. Along active margins, submarine canyon con¢guration and evolution may be governed by the same processes, although tectonism tends to exert a predominant control (e.g., Normark and Curray, 1968; Greene et al., 1991; Klaus and Taylor, 1991; Hagen et al., 1994; Williams, 1997; Lewis et al., 1998). Canyon planforms typically exhibit sharp bends, which may be controlled by strikeslip faults, transverse accretionary ridges, or outer-arc basement highs or mounds. Where the subducting oceanic plate is adorned with high-standing edi¢ces, such as ridges and seamounts, the overriding plate can be deformed by intense and pervasive tectonism (e.g., Collot and Fisher, 1989; Lallemand et al., 1989; von Huene et al., 2000);

367

such deformation can have profound in£uence on the development and adjustment of slope sediment-transport systems. Compressional deformation of the margin wedge caused by underthrusting volcanic edi¢ces on the oceanic plate may form transverse ridges and mounds on the continental slope that alter slope gradients, and may divert and capture uncon¢ned turbidity currents. Moreover, underthrusting seamounts can form long and deep reentrants in the lower continental slope that may promote development of submarine canyons and cause sediment bypass to the trench (e.g., Lewis et al., 1998). On the active convergent margin of central Chile, the submarine San Antonio Canyon (SAC) excavates the forearc slope along a number of nearly right-angular segments before it empties into the Chile Trench via the San Antonio Reentrant near 33‡S (Figs. 1 and 2 ; Hagen et al., 1996). The San Antonio Reentrant and course of the lower-slope canyon segment coincides with the southern £ank of the underthrusting Juan Ferna¤ndez Ridge (JFR), a hot-spot-generated aseismic swell crested with seamounts on the Nazca plate. Along the landward projection of the reentrant, in the direction of interplate convergence, the middle slope is warped above the sub-circular Topocalma Knoll (Fig. 3). This paper examines the notion that the development of the SAC was promoted by the subduction of a seamount, which formed a large indentation in the lower slope and presently resides beneath the mid-slope Topocalma Knoll. Our study is based on the analysis and interpretation of hydrosweep swath-bathymetric and high-resolution seismic-re£ection data (40^180 Hz; penetration 1^3 s two-way traveltime (TWT) below sea £oor) acquired as part of the multi-disciplinary CONDOR (Chilean O¡shore Natural Disaster and Ocean Environmental Research) investigation (von Huene et al., 1997). The central Chile margin has hitherto received only limited scienti¢c study. While a previous study of the SAC comprised a morphologic description based on a multibeam bathymetric survey (Hagen et al., 1996), the more extensive and multidisciplinary data provided by the CONDOR cruises allow a more detailed study of the central

MARGO 3191 19-8-02

368

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

Fig. 2. Multibeam hydrosweep bathymetry of the CONDOR study area plotted as a shaded relief image on which locations of CONDOR seismic-re£ection pro¢les (white lines) are superimposed. Line A locates seismic-re£ection pro¢le across the Navidad^ Algarrobo shelf basin (Fig. 4). The Juan Ferna¤ndez Ridge swell is enclosed between dashed black lines. Contour interval 1000 m.

Chile forearc margin and trench. The principal objectives of the present study are to (1) examine the evolution of the SAC and its in£uence on continental-slope and trench sediment-transport systems; and (2) present some new insights into forearc-margin morphology and its interrelation with major geodynamic changes near 33‡S.

2. The central Chile forearc ^ regional setting The Nazca and South American plates converge at a rate of V85 km/Myr along N78.4‡E (Fig. 1 ; DeMets et al., 1994). Based on the coastal occurrence of intermediate volcanic and plutonic

rocks and volcaniclastic sediment of Mid-Triassic to Cretaceous age, subduction beneath the Chile margin is thought to have been continuous since the early Mesozoic (e.g., Mpodozis and Ramos, 1989). The Mesozoic magmatic arcs are intruded into, and superimposed upon, Palaeozoic intrusive rocks and metamorphosed remnants of a late Palaeozoic accretionary prism, together constituting the primary geologic basement unit of the central Chilean margin. For most of its history, the central Chile margin is inferred to have undergone tectonic erosion, causing margin retreat and thinning of the upper plate (e.g., von Huene and Scholl, 1991; Laursen et al., 2002). Within the study area, wide-angle seismic-re-

Fig. 3. Multibeam hydrosweep bathymetric views of the SAC area. Thick numbered lines are seismic-re£ection pro¢les, illustrated in Figs. 5^7. Seismic-re£ection pro¢les along thin lines are not presented in this paper. (a) Structural interpretations superimposed on planview image of margin. Contour interval 100 m. (b) Perspective view of the SAC in the direction of interplate convergence.

MARGO 3191 19-8-02

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

MARGO 3191 19-8-02

369

370

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

£ection transects across the margin show high velocities ( s 5.5 km/s), indicative of continental metamorphic and granitoid rocks, extending seaward to at least the lower-to-mid slope break (Flueh et al., 1998). Magnetically, continental basement is indicated beneath the upper and mid-slope regions as E^W-aligning anomalies interpreted as the seaward continuation of Palaeozoic plutons recognised on the coast (Yan‹ez et al., 2001). Moreover, a weak N^S magnetic pattern detected beneath the middle slope is thought to represent the root of a pre-Jurassic arc. At the latitude of the SAC mouth (V33‡S), prominent tectonic, geomorphologic and climatic changes occur along the western border of South America in both the o¡shore and onshore provinces (e.g., Jordan et al., 1983; Corvala¤n, 1989; Miller, 1976). The ENE landward projection of the JFR (33‡S) coincides with the southern border of the central Chile segment of ‘shallow’ subduction (the £at-slab segment), which is characterised by a subhorizontal Wadati^Benio¡ zone and an absence of Quaternary arc volcanism (e.g., Cahill and Isacks, 1992; Kay and Abruzzi, 1996). South of 33‡S, an active volcanic arc erupts above a steeper-dipping subduction thrust (Fig. 1). Collision and subduction of the JFR near 33‡S moreover coincides with distinct changes in the con¢guration of the margin morphology. South of the JFR intersection, the shelf is relatively wide (averaging 20^30 km) and the continental slope is broad and dipping gently seaward (Figs. 1 and 2). Extensive ponding of sediment south of the JFR intersection currently promotes active accretion. In contrast, above the underthrusting JFR (V32‡30PS^33‡00PS), the upper slope descends steeply from a narrow (5^10-km-wide) shelf towards the mid-slope Valparaiso Basin, and the toe of the margin is truncated at the subduction thrust (Fig. 2 ; Laursen et al., 2002). Where seamounts recently underthrust the margin, the lower continental slope is indented and scalloped from mass-wasting processes. Near the northern border of the JFR collision, the trail of the Papudo Seamount, magnetically imaged V20 km landward of the trench, is seen as the relatively insigni¢cant Papudo Reentrant (von Huene et al., 1997). On the southern £ank of the JFR,

the San Antonio Reentrant constitutes a £at£oored embayment, 15^20 km on a side, that is bounded up-slope by steep escarpments and separated from the trench by a V12-km-wide accretionary prism (Fig. 3). An extreme climatic gradient with increasing onshore precipitation and erosional denudation toward higher latitudes results in latitudinal variations in £uvial drainage and sediment supply to the o¡shore margin (Miller, 1976; Lamy et al., 1998). Adjacent to onshore regions of high precipitation, modern continental-slope sediment is supplied primarily by rivers without signi¢cant aeolian input (Lamy et al., 1998). In this region, the continental slope is extensively dissected by major submarine canyon systems, of which the SAC is the northernmost one to reach the Chile Trench (Thornburg and Kulm, 1987). The arid climate and associated low denudation rates of the northern Chilean mainland (17‡S^31‡S) cause an essentially sediment-starved margin devoid of major submarine canyons. The source area of the continental hinterland is dominated by the Andean Cordilleras and the Coastal Range, which are composed of a range of igneous and metamorphic rocks of Palaeozoic to Cenozoic age (Mapa Geologico de Chile, 1982). From the Chile Rise, which intersects the margin near latitude 46‡S, the age of the oceanic crust increases in age in the northward direction contributing to the development of a north-sloping trench-axial gradient along the surface of the trench ¢ll. Consequently, trench fans deposited at the mouth of major submarine canyons are diverted northward and coalesce to form an axial sediment-transport system (Thornburg and Kulm, 1987). The Chile Trench axial turbidite channel follows the northward slope from the mouth of the Chacao Canyon near 41‡S to beyond the SAC mouth at 33‡S, and drains the distributary network of trench fans along the base of the slope (Fig. 1). The study area has a semiarid Mediterraneantype climate controlled by the seasonal in£uence of the Southern Westerly Wind Belt (Miller, 1976; Heusser, 1989). During the global climatic cooling episodes of the Pleistocene, however, the wind belt was forced northward, bringing with it higher

MARGO 3191 19-8-02

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

precipitation (Heusser, 1989, 1990), and glaciers repeatedly expanded northward from the Patagonian Andes into the Aconcagua Valley region (33‡S lat) (Rabassa and Clapperton, 1990).

3. Forearc description 3.1. SAC morphology The swath-bathymetric coverage of the CONDOR data extends from about 1000 m water depth on the upper slope, seaward across the Chile Trench at more than 6 km water depth (Fig. 2). Depth contours of the uppermost slope and shelf are extracted from the publicly available ETOPO2 dataset (Fig. 3a; Smith and Sandwell, 1997; Wessel and Smith, 1991). Landward of V1000 m water depth, the SAC incises the Navidad^Algarrobo shelf basin and extends all the way to the coast where it heads in the small port of San Antonio (V33‡40PS) just south of the city of Valparaiso (Figs. 2 and 3a). Although small seasonal drainage gullies empty into San Antonio harbour, most sediment feeding the SAC is thought to derive from the Rio Maipo V5 km south of San Antonio that is being swept northward into the canyon head by strong longshore currents (Hagen et al., 1996). Rio Maipo originates in the Main Cordillera of the Andes at the base of active volcanoes, and crosses the Central Valley and Coastal Range in its lower course (Figs. 1 and 2 ; Mapa Geologico de Chile, 1982). From the upper slope, the SAC incises the forearc margin along a number of margin-perpendicular and margin-subparallel segments on its V130-km-long, overall north-westerly route to the Chile Trench. Immediately west of 1000 km water depth, the margin-perpendicular canyon segment ‘A’ extends seaward for about 20 km, before it turns sharply northward into the V10km-long canyon segment ‘B’, only to continue seaward for another 20 km along canyon segment ‘C’ (Fig. 3a). On the middle slope, the canyon is diverted northward for about 30 km of its length (segment ‘D’) along the landward side of a prominent margin-parallel outer mid-slope ridge. Near

371

latitude 33‡S, coinciding with the southern border of the JFR collision zone, the SAC breaches the outer mid-slope ridge and makes an abrupt turn to the WSW (V260‡) into the San Antonio Reentrant. The upper reaches of the canyon receive a pair of tributaries, which head at the upper slope and join at acute angles at their orthogonal intersection with the main canyon (Fig. 3). Although the canyon is without tributaries farther down-slope, its walls are locally cut by crescent-shaped niches concave towards the thalweg. On the landward side of segment ‘B’, the landward canyon wall forms an amphitheatre-shaped indentation, which opens down-slope into converging chutes and ridges, suggesting an internal drainage system. Minor slope-perpendicular, curvilinear gullies heading on the seaward £ank of Topocalma Knoll appear to feed into the canyon. Landward of the head of the San Antonio Reentrant, on the middle and upper slope, the SAC exhibits a U-shaped cross-section, 4^8 km wide and up to 1 km deep, and an almost constant thalweg gradient of V2‡ (Fig. 3). At the head of the reentrant, the walls of the canyon open out to form a broader (15-km-wide) V shape. From this point, the canyon £oor descends steeply (6.3‡) across the lower slope along a straight furrow to the £at-£oored ( 6 2‡) triangular embayment occupied by the canyon mouth (Fig. 3). 3.2. Seismic-stratigraphic interpretation and facies analysis Published information on the subsurface stratigraphy of the central Chile margin is limited to a seismic-re£ection pro¢le across the Navidad^Algarrobo shelf basin (Fig. 4). Seismic-stratigraphic units interpreted from this pro¢le are dated based on sparsely distributed exploratory boreholes near latitude 35‡S as well as radiometrically dated outcrops on the coast (Gonza¤lez, 1989). The CONDOR seismic-re£ection data image three major seismic-stratigraphic units beneath the central Chile margin that provisionally have been correlated with the dated Navidad^Algarrobo basin. These are: (1) the lower unit (LU) of Palaeozoic to Mesozoic metamorphic and granitoid continen-

MARGO 3191 19-8-02

372

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

facies, as de¢ned by, for instance, Nardin et al. (1979) and Mitchum (1985), are recognised beneath or in the immediate vicinity of the SAC that provide information on sediment pathways and depositional processes.

Fig. 4. Line drawing of seismic-re£ection pro¢le across the Navidad^Algarrobo shelf basin (line ‘A’ on Fig. 2) from Gonza¤lez (1989) to which CONDOR seismic-re£ection pro¢les are correlated and provisionally dated.

tal basement, (2) the middle unit (MU), consisting of Late Cretaceous to Miocene marine deposits, and (3) the upper unit (UU), overlying the Valparaiso Unconformity, consisting of inferred postMiocene marine sediment (Laursen et al., 2002). The Valparaiso Unconformity has been de¢ned in the Valparaiso Basin and traced southward across the Topocalma Knoll into the slope basin incised by the SAC where it typically constitutes the acoustic basement as well as the approximate base of the deepest canyon excavation (Figs. 5 and 6). Above the Valparaiso Unconformity, the upper-unit seismic-stratigraphic sequences are named after their correlative sequences de¢ned within the Valparaiso Basin, namely sq1 (oldest), sq3, sq4, and sq5. In the Valparaiso Basin, sq2 constitutes a local channel-overbank wedge, which in our study area presumably corresponds to a condensed section not resolvable with the available data. The dominant acoustic facies of the upper-unit sequences exhibit laterally extensive, even-bedded continuous re£ections of variable amplitude, interpreted as the equivalent of ponded lower-fan or distal basin-plain turbidite deposits alternating with pelagic and/or hemipelagic sediment (Fig. 5). Because original depositional slopes associated with basin-plain turbidite deposition rarely exceed 1‡ (Normark et al., 1993), palaeo-depositional surfaces of these facies are approximated to be nearly horizontal. Internal geometry, such as convergence/divergence, and inclinations of parallel beds are thus chie£y the result of syn- and postdepositional vertical movements of the underlying basement. More complex process-related acoustic

3.2.1. Upper to mid-slope province On the upper continental slope, the margin-perpendicular canyon segments ‘A’ and ‘C’ are located on down-thrown blocks of what appear to be roughly E^W-trending normal faults pre-dating the upper-unit seismic-stratigraphic sequences (Figs. 5a,b and 6a,b). Across canyon segment ‘A’, the Valparaiso Unconformity drops nearly 500 m to the north. At the base of the inferred fault, overlying the unconformity, lies a 100^300-mthick body with an irregularly mounded shape and a hummocky surface that contains disrupted and contorted internal re£ections indicative of a submarine slide deposit (Fig. 5a). At the toe of the slide, a 100-m-thick, 600-m-wide depression is ¢lled with parallel, discontinuous, high-amplitude re£ections interpreted as coarse-grained channel£oor deposits. Beneath canyon segment ‘C’, seismic-re£ection resolution is relatively poor and prohibits imaging of deep-seated basement structures (Fig. 5b). Nevertheless, the vertical o¡set of the Valparaiso Unconformity across canyon segment ‘C’ (V1 s TWT) implies a down-to-the-north displacement similar to what is observed beneath segment ‘A’. Similar to the up-canyon segment, a buried hummocky surface may be the top of slumped debris from the inferred fault scarp. This deposit is juxtaposed with a 2.5-km-wide sequence of stacked high-amplitude re£ections (Fig. 5b) interpreted as the down-canyon continuation of the channel thalweg deposits indicated beneath segment ‘A’ (Fig. 5a). Thick accumulations of sq1 basin-plain turbidites de¢ned within Valparaiso Basin and draping Topocalma Knoll (Laursen et al., 2002) extend southward to beneath the SAC incision where strata terminate against the underlying fault scarp (Figs. 5a and 6a). South of Topocalma Knoll, sq1 is most clearly imaged along the upper slope where the sequence exhibits a V200^400 m-thick section of chie£y low-amplitude, semi-continuous

MARGO 3191 19-8-02

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

373

S

Distance [km]

+

28

Fig. 5. Examples of CONDOR seismic-re£ection pro¢les across SAC (see interpretations Fig. 6; Fig. 3 for locations of pro¢les). Circled cross symbolises end of vector pointing down-canyon. (a) Part of pro¢le 2. (b) Part of pro¢le 28. (c) Part of pro¢le 26.

re£ections (Fig. 5a). Beneath the SAC, stacked, but laterally shifted segments of higher-amplitude re£ections suggest a local occurrence of relatively coarser sediment (Fig. 5a). On the northern side of the Topocalma Knoll, above the underthrusting JFR, post-depositional buckling has severely deformed sq1, which in parts forms structural highs onlapped by the subsequent (inferred) basin-plain turbidites of sq3^ sq4 (Fig. 6b,d). Beneath canyon segment ‘C’, the

upper V300-ms-thick section of sq3 terminates abruptly against what appears to be a buried incised canyon (Figs. 5b and 6b). Along dip pro¢le 26 (Fig. 3), sq3 and sq4 constitute a wedge-shaped unit of continuous moderate-amplitude re£ections, exhibiting landwardstepping onlap onto an angular unconformity truncating seaward-tilted sq1 and middle-unit strata, as well as seaward divergence toward the central part of the middle slope (Figs. 5c and 6c).

MARGO 3191 19-8-02

374

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

.

Profile 2

.

Profile 28

+

Profile 26

slump left-bank levee

+

axial tubidite channel

.

1

Profile 22

Fig. 6. Line drawings of CONDOR (SONNE-101) seismic-re£ection pro¢les crossing SAC. See Fig. 3 for location. Circled cross symbolises end of vector pointing down-canyon. Circled dot symbolises front of vector pointing down-canyon.

On the trenchward side of the SAC incision, beds converge toward the lower to mid-slope break. Syn- to post-sq4 formation of the prominent margin-parallel ridge £anking this segment of the canyon has elevated and rotated the outer middle slope landward. The margin-parallel ridge is interpreted as the hanging wall of a prominent seaward-vergent thrust fault (Fig. 6c), which appears to emanate from the head of the San Antonio Reentrant (Fig. 3). On the immediate northern side of the head of the reentrant, sq3 is uplifted and tilted toward the Valparaiso Basin. This ro-

tation occurs simultaneously with deposition of sq4, which converges and exhibits o¥apping re£ector con¢gurations toward the head of the reentrant (Fig. 6d). Succeeding deposition of sq4, the middle slope was warped along the landward projection of the San Antonio Reentrant, whereby the present con¢guration of the Topocalma Knoll came into place (Fig. 6b). By bringing sq4 back to its inferred depositional level, uplift of Topocalma Knoll approximates 750 m. Steepening of the slopes related to elevation of the knoll appears

MARGO 3191 19-8-02

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

to have been enhanced by sagging along its seaward periphery (Fig. 6a^c). Sq5 holds the greatest thickness ( s 400 m) on the upper middle slope south of the SAC (Fig. 6a,b). On the northern side of the canyon, sq5 wedges out against the Topocalma Knoll (Fig. 6a,b) and pinches out trenchward down its southern £ank (Fig. 6c). Along the upper middle slope, sq5 exhibits a complex pattern of moderate-amplitude, strati¢ed re£ections and low-amplitude sigmoidal re£ections interpreted as a broad system of channel cut-and-¢ll and interrelated overbank deposits (Figs. 5a,b and 6a,b). The buried channels are 0.5^1 km wide and less than 100 m deep. Distinct wedge-shaped units interpreted as channel overbank deposits (Figs. 5a and 6a) are truncated by the northern wall of canyon segment ‘A’. Along the steep incised walls of the SAC, the slope-depositional sequences are sharply truncated. Up-canyon from the head of the reentrant, a horizontally strati¢ed sequence of semi-continuous, moderate-amplitude re£ections that is as much as V250 m thick ¢lls the canyon (e.g. Figs. 5c and 6c). SeaBeam 2000 side-scan sonar data show a high backscatter signal from the surface of the canyon ¢ll, which is interpreted as coarse-grained turbidite deposits (Hagen et al., 1996). Locally, slides drape the canyon walls (Fig. 6). 3.2.2. Lower slope province At the head of the San Antonio Reentrant, a distinct change in the canyon morphology occurs. The relatively narrow, steep-walled canyon con¢guration across the middle slope is transformed into a broad canyon cross-pro¢le £anked by less steep, irregular walls exhibiting retrogressive slumping into the adjacent slope (Fig. 6d). The canyon appears to cut through upper- and middle-unit sediment, possibly into lower-unit continental basement. In the centre of the canyon, a more than 200-m-high, roughly N^S-trending obstruction remains that is incised along the canyon walls by a bifurcated channel heading in the ponded canyon ¢ll (Figs. 3a and 6d). The obstruction aligns with the margin-parallel ridge £anking canyon segment ‘D’, as well as with outer forearc

375

highs bordering the Valparaiso Basin to the north (Fig. 3). The small accretionary prism bridging the canyon mouth consists of folded sediment, which in the lower and inner parts exhibits parallel, continuous beds indicative of ponded trench turbidites, but in the upper part shows a more contorted disrupted re£ection pattern, suggesting the incorporation of slumped debris (Fig. 7d). The two channel branches heading up-canyon from the head of the reentrant coalesce into a single channel down the lower slope that breaches the accretionary ridge at the centre of the canyon mouth (Fig. 3a). Ponded sediment behind the accretionary prism forms a £at £oor in the canyon mouth. An apparently abandoned channel breaches the accretionary ridge along the southern transverse wall of the reentrant (Fig. 3b). 3.2.3. Trench basin Seaward of the main subduction thrust, the oceanic basement carries a 100^150-m-thick conformable drape of pelagic and hemipelagic sediment. Along the trench axis, the oceanic plate descends into the subduction zone at an angle of V5‡ (Fig. 7) creating a longitudinal depression in which subhorizontal strata are ponded to form a seaward-thinning wedge-shaped deposit. South of the intersecting JFR, the trench wedge is up to V2.5 km thick ; above the swell of the ridge, however, less than 0.5 km sediment has accumulated (Fig. 7). South of the SAC mouth, strata are parallel and continuous across the entire width of the trench wedge (Fig. 7e) indicative of sheet turbidite deposits (see also Thornburg and Kulm, 1987). A similar re£ection con¢guration is observed in the lower part of the wedge opposite the SAC mouth (Fig. 7c,d). Immediately seaward of the deformation front, however, these beds are overlain by a thin sequence of sub-horizontal, high-amplitude, discontinuous re£ections juxtaposed on their seaward side by a 100^200-m-thick wedge-shaped sequence exhibiting continuous moderate-amplitude re£ections converging in the seaward direction. In map view, the landward edge of this sequence is crescent-shaped concave towards the SAC mouth, beginning at the southern escarpment of the reen-

MARGO 3191 19-8-02

376

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

Fig. 7. CONDOR seismic-re£ection pro¢les crossing Chile Trench (a^e). See Figs. 3 and 8 for location. Circled cross symbolises end of vector pointing down trench slope. The sea£oor emergence of the interplate de¤collement is shown with black triangle on vertical line.

MARGO 3191 19-8-02

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

MARGO 3191 19-8-02

377

378

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

MARGO 3191 19-8-02

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

trant (Fig. 3). A piston core obtained on the positive relief immediately south of the canyon mouth contains graded sand beds, 75 cm thick, interpreted as levee facies (Thornburg and Kulm, 1987). The crescent-shaped feature is thus interpreted as a broad constructional levee formed on the left bank of a distributary channel, which emanated from the southern escarpment of the SAC mouth. The path of this channel remains as a small incision through the accretionary ridge but may have been abandoned as accretion proceeded to uplift the ridge (Fig. 3b). From the canyon mouth, the channel was diverted northward down the trench gradient, and built a prominent levee on the left channel bank. The smooth surface of the trench wedge south of the SAC mouth shows that the sediment sources from further south were integrated with the San Antonio distributary system (Fig. 8). Despite shallowing of the oceanic basement across the JFR crest, the surface of the trench wedge maintains an axial gradient beyond the intersection of the ridge. The Chile Trench axial turbidite channel skirts the seaward perimeter of the trench 1^3 km from the outer trench ‘wall’ (Figs. 7 and 8). South of the SAC mouth, near 34‡S, the channel is V1 km wide and less than 50 m deep, and appears to be con¢ned by broad constructional overbank deposits (Fig. 7e). Where the channel intercepts the San Antonio overbank deposits, it becomes erosional and incises the trench ¢ll as much as 300 m deep and 2 km wide (Fig. 7c,d). Across the JFR crest, the San Antonio distributary channel joins the axial channel, which to the north is con¢ned between the truncated toe of the continental margin and the stair-step relief of the down-£exed oceanic plate (Figs. 7a,b and 8). The ponded trench ¢ll can be traced along the outer trench ‘wall’ to near the northern £ank of the JFR where the trench £oor suddenly deepens more than 300 m (Fig. 8). At

379

the base of this steep slope, the trench gradient resumes that prevailing south of the SAC mouth (V0.1‡). About 20 km north of the slope change, the axial channel is £anked by broad overbank deposits (Fig. 7a), which Thornburg and Kulm (1987) denoted as an axial sediment lobe.

4. SAC evolution ^ discussion 4.1. E¡ects of seamount subduction on forearc deformation An array of forearc structures in the immediate vicinity of the SAC can be interpreted in accordance with the general consequences of seamount subduction. The diagnostic structures include: (1) the triangular indentation of the SAC mouth bounded up slope by steep escarpments and (2) a landward transition to a narrow furrow (lowerslope canyon segment) that ends on the seaward side of a raised mound (the Topocalma Knoll) (e.g., Lallemand and Le Pichon, 1987; von Huene et al., 2000). Progressive stages of deformation resulting from seamount subduction are well documented by bathymetric and seismic-re£ection studies of, for instance, the margin o¡ Costa Rica (von Huene et al., 2000) and along the Japan Trench (Lallemand and Le Pichon, 1987; Lallemand et al., 1989). The use of laboratory sandbox modelling to simulate seamount subduction has moreover provided a good analogy for geological observations and o¡ered insight into the processes governing seamount^forearc interactions (e.g., Lallemand et al., 1994; Dominguez et al., 1998, 2000). Where a seamount underthrusts the upper plate, a zone of compression develops in the overriding margin wedge in front of and above the seamount’s landward £ank (e.g., Lallemand and

Fig. 8. (a) Trench-axial depth pro¢le of the CONDOR survey area. Individual, measured trench-axial pro¢les have been projected onto vertical section along the axis of the trench. Annotated angles are true dips measured along each pro¢le. Note steep scarp in the trench ¢ll above the northern £ank of the JFR and accumulated trench sediment to the south that is incised by axial channel. (b) CONDOR swath-bathymetry map of the trench area. (c) Contour map (contour interval 100 m) of same area as b with superimposed structural interpretation.

MARGO 3191 19-8-02

380 J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

MARGO 3191 19-8-02 Fig. 9. Chronological summary of events. Note, non-linear time scale. See text for explanation.

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

381

Fig. 10. Progressive positions of the underthrusting San Antonio Seamount (SAS) beneath the central Chile margin shown in intervals of V150 ka.

Le Pichon, 1987). This compression is accommodated by folding and thrusting of the margin toe predominantly with a vergence sub-parallel to the advancing or leading £ank of the seamount (e.g., Dominguez et al., 2000). As the seamount is completely underthrust beneath the toe of the margin, the uplifted and shortened wedge passes over the top of the seamount. Consequent oversteepening of the lower slope above the trailing (seaward) £ank of the seamount induces gravitationally driven failure and mass-wasting of the front of the margin. Analogue models suggest that a low-stress ‘shadow zone’ develops in the wake of the underthrusting seamount because the edi¢ce de£ects the interplate de¤collement upward into the trailing wedge and debris of the overriding plate before descending back to near the top of the oceanic plate (e.g., Lallemand et al., 1994; Dominguez et al., 1998, 2000). Consequently, material from the upper plate, as well as the trench sediment, is attached to the down-going plate and passively dragged into the subduction zone leaving a reentrant in the wake of the subducted seamount (e.g., Ballance et al., 1989; Lallemand et al., 1994). Ac-

cretion may resume along the deformation front when the seamount has advanced beneath the margin and the interplate de¤collement has returned to a deeper level within the trench ¢ll. The wide embayment of the San Antonio Reentrant bordered by transverse scarps and thrust ridges that curve in the direction of interplate convergence (Fig. 3) shows that compressive stresses induced by the inferred seamount were centred along the direction of interplate convergence. Compression in the margin wedge above the leading £ank of the advancing seamount likely caused the formation of the prominent thrust ridge constituting the seaward £ank of canyon segment ‘D’ (Fig. 3). As illustrated in laboratory sandbox modelling (Dominguez et al., 2000), the bounding thrust fault may have developed from the summit of the seamount and propagated seaward and laterally, possibly connecting with pre-existing structural fabric in the margin wedge. The irregular scarps con¢ning the triangular indentation as well as the long deep furrow into the lower slope attest to the importance of mass-wasting processes and probably subduction erosion of

MARGO 3191 19-8-02

382

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

the trailing debris in the wake of the seamount. The present occupation of the reentrant by the SAC has presumably modi¢ed the morphology of the indentation through continued excavation and undercutting of its escarpments. The narrow ridge of up-folded, accreted sediment bridging the reentrant (Fig. 3) illustrates that the de¤collement has returned to a base level within the trench wedge, and thus enabled accretion to resume. A similar pattern of deformation is observed in the lower slope of the New Hebrides Island Arc above the southern edge of the subducting d’Entrecasteaux zone, a chain of linear ridges and seamounts. Here, the Malekula Reentrant aligns, along the direction of interplate convergence, with an area of maximum uplift (Malekula Island), and is bridged along the main subduction thrust by a small accretionary ridge. Collot and Fisher (1989) suggest, in line with the present study, that the reentrant was formed by a seamount subducted beneath the Malekula Island, and that accretion has resumed to heal the scar. 4.2. Timing of events: seamount advance and syn-tectonic slope sedimentation In the landward projection of the San Antonio Reentrant, the seismic-stratigraphic architecture of the continental-slope deposits records a progression of landward-migrating deformation that we ascribe to the advance of the informally named San Antonio Seamount. The intimate association between the advance of the seamount and syn-depositional deformation of sq4 and sq5 places age restrictions on the sequences as well on the time involved for SAC formation. The chronological order of events summarised in Fig. 9 shows an attempt at correlating the various e¡ects of the San Antonio Seamount subduction, and at depicting the interrelations between the SAC evolution and trench and continental-slope sedimentation. For the most recent events, a correlation to Oxygen Isotope Stages (Imbrie et al., 1984; Chappel and Shackleton, 1986) is cautiously suggested. If the raised subcircular mound of the Topocalma Knoll, V55 km inboard of the Chile Trench,

records the location of the underthrust San Antonio Seamount, this seamount, according to the present rate of interplate convergence, collided with the margin 600^650 ka ago (Fig. 10). Progression of the seamount beneath the lower to mid-slope break by V250^350 ka (Fig. 10) is depicted by the uplift of sq3 and convergence of sq4 strata toward the head of the reentrant (Fig. 6d), as well as by post-depositional thrusting and uplift of the hanging-wall ridge bordering canyon segment ‘D’ (Fig. 6c). Within the past V150^ 200 ka, the landward advance of the seamount beneath Topocalma Knoll is exhibited by onlap of sq5 strata onto the progressively warping knoll (Fig. 6a,b). Deposition of the sq4 and sq5 may thus be ascribed to the past V350 ka, which encompasses an interval of the Quaternary during which high-amplitude £uctuations in sea level controlled by glaciation and deglaciation occurred (Imbrie et al., 1984; Chappel and Shackleton, 1986). As the Topocalma Knoll grew, simultaneous sagging along the seaward perimeter of the knoll (Fig. 6c) might be ascribed to basal erosion of the margin wedge by processes related to subduction of the seamount. An example of this is inferred in Costa Rica, where the margin wedge is thinned above the summit of an underthrust seamount (Ranero and von Huene, 2000). A simple volume calculation of the accretionary prism bridging the reentrant illustrates the rate with which the prism could have formed given the current rate of accretion and plate convergence. Applying an acoustic velocity of 2.5 km/s, the average cross-sectional area of the prism (Fig. 7c,d) yields an accreted volume of 20 km3 /km length of the margin. With the present detachment of the upper V1.5 km of the trench ¢ll, the accretionary prism could have formed after only V14 km of convergence, i.e., within a period of V165 kyr. Accretion may hence have resumed when the seaward £ank of the seamount was located beneath the head of the reentrant, V40 km inboard of the trench (Fig. 10). When the de¤collement returned to a base level within the trench turbidites, material brought into the subduction zone in the wake of the seamount might have been underplated beneath and adjacent to the

MARGO 3191 19-8-02

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

head of the reentrant, possibly explaining, at least in part, the pronounced elevation of its £anking thrust ridge (Figs. 3 and 6c). 4.3. Continental-slope sediment distribution prior to SAC excavation During sq1 to sq4 time, sediment distribution on the continental slope was partly controlled by the inferred E^W-trending fault scarps underlying SAC segments ‘A’ and ‘C’ that captured sediment on their northern down-thrown side (Fig. 6a,b). Ponding of sediment on the hanging-wall side of the fault gradually eliminated the relief of the footwall block, which eventually was covered by sq5 sediment. The inferred basin-plain turbidites of sq3^sq4 ponded in a broad, shallow, mid-slope basin (Fig. 6c). This basin was separated from the Valparaiso Basin only by a small emerging structural high of deformed sq1 deposits near the present northern £ank of the Topocalma Knoll (Fig. 6b). The landward-stepping onlap pattern of sq3 (Fig. 6c) as well as the longitudinal continuation of strata into Valparaiso Basin (Fig. 6d) suggest deposition from uncon¢ned turbidity currents roughly following the present mid-slope course of the SAC. The barrier formed by post-depositional uplift of sq3 inboard of the head of the reentrant, however, prohibited the continuation of northward sediment transfer into the Valparaiso Basin (Figs. 6d and 10). The change from landward onlap (sq3^sq4) to down-slope pinchout (sq5) shows that the depositional systems during sq5 stepped landward (Fig. 6c). On the south side of Topocalma Knoll, sq5 constituted a broad fan system, which may have been fed from Rio Maipo and/or nearby river outlets to the south. A number of observations, such as the local occurrence beneath canyon segments ‘A’ and ‘C’ of inferred coarse-grained channel ¢ll (base sq1), indicates that the locus of the present up-slope segments of the SAC in the past constituted pathways for coarse-grained, possibly erosive turbidite transport. The buried canyon beneath segment ‘C’ does not appear to connect to any fossil incision up-canyon, and may be an isolated feature that began at mid-slope depths.

383

4.4. Controls on SAC formation The right-angular path of the SAC incision paralleling basement fabric and structural morphology of the continental margin suggests that tectonism controlled or in£uenced the course of the canyon. Initiation of the canyon-cutting processes may be ascribed to a number of interplaying mechanisms related to both tectonism and sediment-transport processes. Below, we address the factors that may have in£uenced the con¢guration and evolution of the SAC, and discuss the timing of events in relation to tectonism and sea-level oscillations. As the San Antonio Seamount advanced beneath the thicker, more cohesive part of the margin, we speculate that the outer mid-slope thrust ridge collapsed in its wake above the head of the reentrant and thus formed a breach connecting the middle and lower slope provinces. From the head of the reentrant, retrogressive failure into the soft deposits on the middle slope may have captured turbidity currents de£ected northward along the landward side of the thrust ridge. Canyon growth along segment ‘D’ is di⁄cult to reconcile with retrogressive failure alone as this process tends to form channels along the steepest local slope. An example of this can be seen V25 km south of the SAC where a linear drainage pattern has developed directly up-dip from a large crescent-shaped scarp on the outer middle slope (Fig. 3). A combination of headward erosion and downcutting by captured turbidity currents from an existing uncon¢ned, longitudinal system is thus probably required to explain this slope-parallel excavation. In the upper slope region, the coincidence of canyon segments ‘A’ and ‘C’ with underlying basement faults suggests that the fabric of the substratum in£uenced the location of the canyon. The continental basement beneath Topocalma Knoll is thought to be the seaward continuation of a Palaeozoic pluton intruded along VE^W-trending strike-slip faults recognised on land (Gana et al., 1994; Yan‹ez et al., 1998). Possibly, the deep-seated faults observed beneath the SAC relate to this fault system. Minor fault reactivation or adjustment would shear the sediment cover, and canyon excavation may thus have been

MARGO 3191 19-8-02

384

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

facilitated by fracture patterns and/or slope failure developed above these structures. The establishment of a sediment conduit to the Chile Trench greatly changed the depositional gradient of the continental slope. Nevertheless, resistance to erosion of the basement underlying the head of the reentrant appears to have prevented the development of a uniformly sloping thalweg. A close analogue to the SAC con¢guration and nucleation exists on the Izu Bonin forearc, where submarine canyons deeply incise (1200^1700 m) the thickly sedimented forearc basin (Bonin Trough) on gentle (1^2‡) gradients upslope of the outer arc high (Bonin Ridge). On the steep (6^18‡) inner trench slopes, basement rocks have inhibited a signi¢cant bathymetric expression (Klaus and Taylor, 1991). Because canyons have not formed across the Bonin Trough where the Bonin Ridge remains unbreached, the canyons are inferred to nucleate by headward erosion when forearc basin sediment reaches the spill point at structural low points. The deep incision of the SAC and Izu Bonin canyons along the gentle gradients of their forearc basins is in marked contrast to the canyons on non-subduction margins. For example, the California Borderland and the U.S. Atlantic margin have canyons that are best developed on the steeper upper slope and show decreasing relief on the gently sloping continental rise (e.g., Normark and Piper, 1969; Farre et al., 1983; Greene et al., 1991). The con¢guration of the SAC also shares some common features with the Poverty Canyon on the Hikurangi margin (o¡ eastern New Zealand) that incises a V-shaped reentrant in the lower slope thought to have formed by a subducted seamount on the Hikurangi Plateau (Lewis et al., 1998). The Poverty Canyon is the only canyon to reach the trench along nearly 300 km of the margin ; the adjacent canyons are trapped on the forearc slope behind outer arc highs. In contrast to the SAC and canyons on the Izu Bonin margin, canyons on the Hikurangi margin were apparently welldeveloped on the upper forearc slope prior to breaching of the outer arc high. The reentrant left by the subducted seamount merely acted as a conduit allowing the Poverty Canyon to extend seaward into the trench.

4.4.1. Age implications for SAC evolution As syn-depositional warping of sq5 across the Topocalma Knoll is inferred to record subduction of the San Antonio Seamount within the past 150 ka, age constraints are imposed on sq5 as well as on the excavation of the SAC across the upper slope and shelf. Because the SAC intercepts the entire width of the shelf, we are inclined to ascribe the landward evolution of the SAC, and thus the establishment of a direct sediment conduit from the coast to the Chile Trench, to the sea-level lowstand of the latest Pleistocene glaciation. The shoreline was then at the present-day outer shelf and rivers likely extended across the emergent shelf. Hyperpycnal river discharge may have fed directly into the SAC head on the upper slope, thus enhancing deep-water erosive turbidite transport along the canyon £oor as well as back-cutting of the incipient canyon head into the shelf. Enhanced denudation in the hinterland related to a more humid climate and northward expansion of glaciers into the Aconcagua Valley (Heusser, 1990; Rabassa and Clapperton, 1990) moreover provided increased, coarse-grained £uvial input to the central Chile margin and trench (Thornburg and Kulm, 1987; Lamy et al., 1999). Because accretion is inferred to have resumed when the trailing £ank of the San Antonio Seamount underthrust the head of the reentrant, sediment provided by headward erosion from the mid-slope canyon segments was initially trapped in the canyon mouth. After ponding to the spill point (lowest saddle) of the accretionary ridge, headward erosion by over£owing turbidity currents presumably resulted in breaching of the distributary channel feeding the levee on the trench £oor. The Poverty Reentrant on the Hikurangi margin is, similar to the San Antonio Reentrant, separated from the Hikurangi Trough by a low rise aligned along the deformation front (Lewis et al., 1998). Canyon-distributary channels emanating from the reentrant were previously tributaries to the Hikurangi Channel along the trench axis, but are presently trapped within the embayment of the reentrant, presumably as a consequence of ongoing accretion along the deformation front. Ponding of turbidites in the mid-slope segments of the SAC suggests that the obstruction crossing

MARGO 3191 19-8-02

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

the head of the reentrant temporarily formed a barrier for much of the down-canyon sediment transport. Possibly, thrust-block rotation or reactivation of deep-seated faults caused the obstruction to emerge a few hundred metres, enough to trap sediment up-canyon. Turbidite deposits are presently ponded to the spill point of the obstruction, and headward erosion along the sides of the canyon has formed the bifurcated channel (Fig. 3). As sea level rose during the latest Pleistocene deglaciation, sediment deposition was presumably mainly restricted to deltas and shelf sediment prisms. However, as the SAC cut across the shelf, the canyon intercepted the coastal Chilean shoreline-parallel current. The sediment from longshore drift may thus have been trapped in the SAC head and diverted down-slope, as occurs in the southern California Borderland (Shepard, 1973) and along the Bering shelf (Carlson and Karl, 1988). While lowered sea level typically promotes £ushing of the canyon £oor, ¢lling of the mid-slope segments of the SAC may, in addition to the effect of structural damming, be attributed to the latest Pleistocene sea-level rise. The currently small £uid discharge of Rio Maipo (3.2 km3 /yr; Milliman et al., 1995) presumably provides only a small sediment discharge to the shelf. The insigni¢cant channel incisions into the ponded turbidites in the SAC also suggest, in line with the above, that little sediment currently reaches the Chile Trench through the SAC. 4.5. The history of the trench basin In order to investigate the temporal evolution of the Chile Trench basin in the vicinity of the SAC mouth, we assume, as an approximation, that the trench wedge is in a steady-state equilibrium. If the average sedimentation rate (dy/dt) and interplate convergence rate (dx/dt) are constant, steady-state conditions are achieved. As a result, the trench wedge will approach an equilibrium con¢guration in which there is no net volume £ux of material into or out of the depositional system (see also Thornburg and Kulm, 1987). The longitudinal transport processes in the trench supply sediment from the higher source-input re-

385

gions of southern Chile to the northern trench provinces. The study area has thus presumably received substantial sediment from the south throughout the Pleistocene, and the steady-state approximation may hence be reasonable at least within time frames involving the last few sea-level cycles. From the relation tan 3 = (dy/dt)/(dx/dt), the steady-state approximation allows the average sedimentation rate and thus the maximum age of the trench wedge at a speci¢c location to be estimated, given the dip angle (3W5‡) of the subducting oceanic plate and the interplate convergence rate (8.5 cm/yr). Although the width of the trench wedge, and hence the volume accumulation, varies along the length of the margin, the trench wedge maintains an identical sedimentation rate of V7.5 m/103 yr. Applying this rate, the maximum age of the trench turbidite ¢ll (1.732.2 km thick) opposite the SAC mouth (Fig. 7c,d) approximates V230^300 ka. By analogy, the larger trench basin to the south (Fig. 7e) may contain sediment as old as 400 ka, whereas the smaller trench wedge above the JFR crest and to the north (Fig. 7a,b) could have accumulated within the past 60^100 ka (Fig. 9). Another approach to the same problem is to consider the length of time that the oceanic plate underlying the trench wedge has formed the base for trench turbidite accumulation. The segment underlying, for instance, the 25-km-wide trench wedge opposite the SAC mouth would be consumed by subduction within V300 ka. Assuming that the trench wedge has been at steady-state equilibrium since 300 ka, this number equals the maximum age of the trench ¢ll accumulated at this location. If the trench wedge within the study area is no older than V400 ka, sediment deposited in the trench during the intersection of the San Antonio Seamount V700^600 ka ago has now been consumed by subduction processes. Debris slumped from the continental slope in the wake of the San Antonio Seamount may also, to a large extent, have been subducted or partly incorporated into the accretionary prism bridging the reentrant. The Papudo Seamount on the northern £ank of the JFR (Fig. 8) must have blocked the trench for a period of V200 ka until completely subducted at

MARGO 3191 19-8-02

386

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

V120 ka (assuming a seamount diameter of V20 km; Yan‹ez et al., 2001). During this period, the oldest trench ¢ll south of the JFR was deposited (Fig. 9). Damming of the northward longitudinal sediment transport by trench-intersecting edi¢ces on the JFR swell would temporarily elevate and lower sedimentation rates to the south and north of the obstruction, respectively. Moreover, slumping and the establishment of new sediment-input points feeding the trench may locally have altered the equilibrium con¢guration of the trench wedge. The sudden deepening of the trench £oor near 32‡40PS suggests that down-gradient axial sediment transport was inhibited by a trench-intersecting and subsequently subducted obstruction. This hypothesised barrier would have allowed sediment to accumulate to the south opposite the SAC mouth, and have left the trench north of the JFR sediment starved. Consequently, the trench wedge to the north would diminish as underthrusting proceeded and no additional sediment was supplied. A similar scenario is observed along the base of the Hikurangi margin, where the sediment-¢lled Hikurangi Trough abruptly drops nearly 1 km into the southern sedimentstarved Kermadec Trench (Lewis et al., 1998). The cause of this discontinuity has been attributed to a major structural break, but several generations of seamount reentrants and debris-£ow deposits on the trench £oor near the northern border of the Hikurangi Trough suggest that sediment transfer into the Kermadec Trench was inhibited by a succession of trench-intersecting seamounts and their wake deposits (Lewis et al., 1998). In a sediment-starved situation, decreasing the thickness of the Chile Trench wedge by V300 m requires V3.2 km of interplate convergence, i.e., less than 40 ka of subduction consumption. A relatively small edi¢ce, such as an oceanic-crust horst block or a major slump, would be su⁄cient to block axial sediment transport in the trench for this length of time. Slumping might have resulted from continental-slope failure in the wake of the Papudo Seamount, or from oversteepening of the truncated inner trench ‘wall’ facing the JFR crest. The discontinuity of the trench-axial gradient

near 32‡40PS has previously been attributed to a tectonic escarpment coinciding with what has been interpreted as a major break or fault scarp in the underlying oceanic basement (Thornburg and Kulm, 1987; their ¢g. 2). This inferred fault, however, coincides with the northern border of the JFR, and the basement ‘scarp’ may in fact be the northern £ank of the JFR itself (Fig. 8). A trench-parallel seismic-re£ection pro¢le or a correlation of the trench deposits north and south of the scarp would likely reveal the origin of this feature. The absence of buried axial or distributary channels underlying the San Antonio distributary complex shows that deposition from sheet-like turbidity currents prevailed until a sediment conduit to the trench was established through the SAC. Applying the estimated average sedimentation rate for the trench wedge, the 175^200-mhigh levees opposite the canyon mouth may have accumulated within V23^26 kyr. This supports the notion that the margin-wide excavation of the SAC was established no earlier than the sea-level lowstand of the latest Pleistocene. Aggradation of thick sand beds on the levee shows that the canyon formed a conduit for coarse-grained turbidity currents, which may have encountered a large hydraulic jump as they breached the accretionary ridge and entered the Chile Trench. After the obstruction near 32‡40PS was subducted beneath the toe of the margin (or in the event of a mass-transport deposit was entrenched by overspilling sediment gravity £ows) retrogressive failure or headward erosion to the south into the axial trench ¢ll ensued. Downcutting of the axial channel into the San Antonio distributary complex relates the incision to the latest Pleistocene sea-level rise or Holocene highstand. In order to re-establish the regional northward gradient of the trench £oor, incision along the axial channel will likely continue to progress headward and hence deepen the channel to the south. Fan lobes developed opposite large submarine canyon systems in southern Chile similarly exhibit dissection by erosional processes, which leads to progradation of the axial dispersal system into more northern distal environments (Thornburg et al., 1990). Excavated material from the San

MARGO 3191 19-8-02

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

Antonio distributary complex probably fed the axial sediment lobe observed north of the JFR (Fig. 7a).

5. Summary and conclusions The submarine SAC excavates the forearc slope of the central Chile margin for more than 150 km along a number of nearly right-angular segments before it empties into the Chile Trench via the San Antonio Reentrant near 33‡S. Along the landward projection of the reentrant, the middle slope is warped above the subcircular Topocalma Knoll, which we speculate marks the underthrust position of a seamount that collided with the margin V650 ka ago and formed the reentrant. The upper and mid-slope reaches of the SAC are U-shaped and cut as much as 1 km into thick, chie£y basin-plain turbidite deposits ponded in a broad and shallow basin. From the head of the San Antonio Reentrant, the SAC descends steeply ( s 6‡) through a V-shaped incision until it reaches the £at-£oored ( 6 2‡) triangular embayment of the reentrant, which is occupied by the canyon mouth. Presumably, the resistance to erosion of outcropping basement at the head of the reentrant has prevented the development of a uniform thalweg gradient. Prior to subduction of the seamount, the sediment-transport system on the continental margin travelled sub-parallel to the slope feeding into the mid-slope Valparaiso Basin to the north of 33‡S. As the seamount advanced beneath the margin, progressive landward uplift of the middle slope inhibited northward transfer of sediment. Depositional systems deriving from the south became con¢ned to the slope south of 33‡S. Within the past 150^200 ka, an extensive upper to mid-slope fan system, up to 400 m thick, was deposited against the progressively warping Topocalma Knoll. The development of the SAC was controlled or strongly in£uenced by the impact of the San Antonio Seamount. We speculate that the SAC nucleated at the head of the reentrant, which formed a breach through the outer arc highs and thus promoted retrogressive failure into the

387

soft deposits up-slope. On the middle slope, the SAC trends parallel to the regional slope along the landward side of a prominent ridge constituting the hanging wall of a major seaward-vergent thrust fault thought to have formed above the leading £ank of the underthrusting seamount. Turbidity currents generated along the slope were probably de£ected northward along the landward side of the thrust ridge and captured by the developing depression at the head of the reentrant, whereby longitudinal downcutting along the middle slope was enhanced. The upper mid-slope canyon segments coincide with underlying margin-perpendicular basement faults, and excavation along these trends may have been facilitated by fracture patterns and/or slope failure caused by deep-seated fault reactivation or adjustment. During the sea-level lowstand of the last glaciation, the SAC cut into the emergent shelf. Direct in£ow of hyperpycnal turbid £uvial water from glacial outwash probably enhanced £ushing of the canyon £oor. Mass-transport £ows through the SAC were initially ponded within the embayment of the reentrant behind a newly evolving accretionary ridge, which began forming V165 ka ago when the seamount had subducted to beneath the head of the reentrant. A distributary channel from the SAC breached the accretionary ridge along the southern escarpment of the canyon mouth, presumably during the Last Glacial Maximum, from where it was de£ected northward down the longitudinal trench-axial gradient and, in the process, formed a prominent levee on its left bank. The transverse sediment-input source from the SAC was integrated with the longitudinal transport system of trench-wide turbidites. A sudden deepening of the trench £oor at the northern border of the JFR suggests that down-gradient sediment transfer was inhibited by a trench obstruction, which left the trench to the north sediment starved for an estimated period of V40 kyr. Enhanced incision of the Chile Trench axial turbidite channel between V32‡40PS and V33‡40PS is ascribed to a combination of downcutting and headward erosion from the northern edge of the ponded trench ¢ll as the obstruction in the trench was elimi-

MARGO 3191 19-8-02

388

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

nated. At present, the rather insigni¢cant San Antonio distributary channel is tributary to the axial channel, which continues undisturbed across the JFR crest and is con¢ned between the truncated toe of the continental margin and the stair-step relief of the down-£exed oceanic plate. Emergence of an obstruction at the head of the San Antonio Reentrant has trapped sediment in the mid-slope canyon segments, probably during the latest Pleistocene sea-level rise and Holocene highstand. Although the mid-slope segments of the SAC are without side tributaries, local failure of the steep canyon walls, as evidenced by crescentshaped slump scarps, may provide a mechanism for canyon branching. After indenting the shelf, the SAC could capture shelf sands and land-based drainage systems by the canyon head. This has provided a new mechanism for canyon erosion and ¢lling, and the SAC may, as is observed for the canyons on the U.S. Atlantic margin (Farre et al., 1983), develop into a more irregular planform through continued smoothing and undercutting of the canyon walls.

Acknowledgements The R/V Sonne, supported by the Bundesministerium fu«r Forschung und Technologie (German Federal Research and Technology Agency), was the platform for data acquisition. Funding for the CONDOR project included Grants 03G0101A and 03G0103. J.L. acknowledges the provision W rhus. of PhD funding from the University of A The U.S. Geological Survey’s Volunteer Program and the Exchange Student Program at Stanford University allowed J.L. to spend three years of her PhD studies at these institutions. J.L. acknowledges S. Dominguez for discussions and exchange of ideas concerning structures observed in the hydrosweep bathymetry. An earlier version of the manuscript bene¢ted from review by T.L. Vallier and M.A. Fisher for the USGS, Menlo Park. Journal reviews by S. Lallemand and J. Woodside greatly improved the paper.

References Ballance, P.F., Scholl, D.W., Vallier, T.L., Herzer, R.H., 1989. Subduction of a late Cretaceous seamount of the Louisville Ridge at the Tonga Trench: A model of normal and accelerated tectonic erosion. Tectonics 8, 853^962. Cahill, T., Isacks, B.L., 1992. Seismicity and shape of the subducted Nazca plate. J. Geophys. Res. 97, 17503^17529. Carlson, P.R., Karl, H.A., 1988. Development of large submarine canyons in the Bering Sea, indicated by morphologic, seismic, and sedimentologic characteristics. Geol. Soc. Am. Bull. 100, 1594^1615. Chappel, J., Shackleton, N.J., 1986. Oxygen isotopes and sea level. Nature 324, 137^149. Collot, J.-Y., Fisher, M.A., 1989. Formation of forearc basins by collision between seamounts and accretionary wedges: An example from the New Hebrides subduction zone. Geology 17, 930^933. Corvala¤n, J., 1989. Geologic-tectonic framework of the Andean region. In: Ericksen, G.E., Can‹as Pinochet, M.T., Reinemund, J.A. (Eds.), Geology of the Andes and its Relations to Hydrocarbon and Mineral Resources, Earth Sci. Ser., vol. 11: Circum-Pac. Counc. for Energy and Miner. Resour., Houston, TX, pp. 1^111. DeMets, C., Gordon, R.G., Argus, D.F., Stein, S., 1994. E¡ect of recent revisions to the geomagnetic reversal time scale on estimates of current plate motions. Geophys. Res. Lett. 21, 2191^2194. Dominguez, S., Lallemand, S.E., Malavieille, J., von Huene, R., 1998. Upper plate deformation associated with seamount subduction. Tectonophysics 293, 207^224. Dominguez, S., Malavieille, J., Lallemand, S.E., 2000. Deformation of accretionary wedges in response to seamount subduction: Insight from sandbox experiments. Tectonics 19, 182^196. Farre, J.A., McGregor, B.A., Ryan, W.G.F., Robb, J.M., 1983. Breaching the shelfbreak: Passage from youthful to mature phase in submarine canyon evolution. In: Stanley, D.J., Moore, J.C. (Eds.), The Shelfbreak. SEPM Spec. Publ. 33, 25^39. Flueh, E.R., Vidal, N., Ranero, C.R., Hojka, A., von Huene, R., Bialas, J., Hinz, K., Cordoba, D., Danobeitia, J.J., Zelt, C., 1998. Seismic investigation of the continental margin o¡and onshore Valparaiso, Chile. Tectonophysics 288, 251^ 263. Gana, P., Yan‹ez, G.A., Wall, R., 1994. Evolucio¤n geotecto¤nica de la Cordillera de la Costa del Chile Central (33^34‡S): Control Geolo¤gico y Geo¢sico. Actas VII Congreso Geolo¤gico Chileno, Concepcio¤n. Gonza¤lez, E., 1989. Hydrocarbon resources in the coastal zone of Chile. In: Ericksen, G.E., Can‹as Pinochet, M.T., Reinemund, J.A. (Eds.), Geology of the Andes and its Relations to Hydrocarbon and Mineral Resources, Earth Sci. Ser., vol. 11: Circum-Pac. Counc. for Energy and Miner. Resour., Houston, TX, pp. 383^404. Greene, H.G., Clarke, S.H., Jr., Kennedy, M.P., 1991. Tectonic evolution of submarine canyons along the California con-

MARGO 3191 19-8-02

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390 tinental margin. In: Osborne, R.H. (Eds.), From Shoreline to Abyss. SEPM Spec. Publ. 46, 231^248. Hagen, R.A., Bergersen, D.D., Moberly, R., Coulbourn, W.T., 1994. Morphology of a large meandering submarine canyon system on the Peru-Chile forearc. Mar. Geol. 119, 7^37. Hagen, R., Vergara, H., Naar, D., 1996. Morphology of San Antonio submarine canyon on the central Chile forearc. Mar. Geol. 129, 197^205. Heusser, C.J., 1989. Southern westerlies during the Last Glacial Maximum. Quat. Res. 31, 423^425. Heusser, C.J., 1990. Ice age vegetation and climate of subtropical Chile. Palaeogeogr. Palaeoclimatol. Palaeoecol. 80, 107^ 127. Imbrie, J., Hays, J.D., Martinson, D.G., McIntyre, A., Mix, A.C., Morley, J.J., Pisias, N.G., Prell, W.L., Shackleton, N.J., 1984. The orbital theory of Pleistocene climate: Support from revised chronology of the marine v18O record. In: Berger, A.L., Imbrie, J., Hays, J., Kukla, G., Saltzman, B. (Eds.), Milankovich and Climate (Pt. 1). NATO ASI Ser. C., Math. Phys. Sci. 126, 269^305. Jordan, T.E., Isacks, B., Allmendinger, R., Brewer, J., Ramos, V.A., Ando, C., 1983. Andean tectonics related to the geometry of subducted plates. Geol. Soc. Am. Bull. 94, 341^ 361. Kay, S.M., Abbruzzi, J.M., 1996. Magmatic evidence for Neogene lithospheric evolution of the central Andean ‘£at-slab’ between 30‡S and 32‡S. Tectonophysics 259, 15^28. Klaus, A., Taylor, B., 1991. Submarine canyon development in the Izu-Bonin forearc: A SeaMARC II and seismic survey of Aoga Shima Canyon. Mar. Geophys. Res 13, 131^152. Lallemand, S., Le Pichon, X., 1987. Coulomb wedge model applied to subduction of seamounts in the Japan Trench. Geology 15, 1065^1069. Lallemand, S., Culotta, R., von Huene, R., 1989. Subduction of the Daiichi Kashima Seamount in the Japan Trench. Tectonophysics 160, 231^247. Lallemand, S., Schnurle, P., Malavieille, J., 1994. Coulomb theory applied to accretionary and nonaccretionary wedges: Possible causes for tectonic erosion and/or frontal accretion. J. Geophys. Res. 99, 12033^12055. Lamy, F., Hebbeln, D., Wefer, G., 1998. Terrigenous sediment supply along the Chilean continental margin: Modern regional patterns of texture and composition. Geol. Rundsch. 87, 477^494. Lamy, F., Hebbeln, D., Wefer, G., 1999. High-Resolution Marine Record of Climatic Change in Mid-latitude Chile during the Last 28, 000 Years Based on Terrigenous Sediment Parameters. Quat. Res. 51, 83^93. Laursen, J., Scholl, D.W., von Huene, R., 2002. Evolution of the Late Cenozoic Valparaiso Forearc Basin in Central Chile: Forearc Basin Response to Ridge and Seamount Subduction. Tectonics, in press. Lewis, K.B., Collot, J.-Y., Lallemand, S.E., 1998. The dammed Hikurangi Trough: a channel-fed trench blocked by subducting seamounts and their wake avalanches (New Zealand-France GeodyNZ Project). Basin Res. 10, 441^ 468.

389

Mapa Geologico de Chile, 1982. Servicio Nacional de Geologia y Mineria, Chile. Miller, A., 1976. The climate of Chile. In: Schwerdtfeger, W. (Ed.), World Survey of Climatology, vol. 12. Elsevier, Amsterdam, pp. 113^145. Milliman, J.D., Rutkowski, C., Meybeck, M., 1995. River Discharge to the Sea: A Global River Index (GLORI). NIOZ, Texel. Mitchum, R.M., Jr., 1985. Seismic stratigraphic expression of submarine fans. In: Berg, O.R., Woolverton, D.G. (Eds.), Seismic Stratigraphy II: An Integrated Approach To Hydrocarbon Exploration. AAPG Mem. 39, 117^136. Mpodozis, C., Ramos, V., 1989. The Andes of Chile and Argentina. In: Ericksen, G.E., Can‹as Pinochet, M.T., Reinemund, J.A. (Eds.), Geology of the Andes and its Relations to Hydrocarbon and Mineral Resources. Earth Sci. Ser., vol. 11, Circum-Pac. Counc. for Energy and Miner. Resour., Houston, TX, pp. 59^90. Nardin, T.R., Hein, F.J., Gorsline, D.S., Edwards, B.D., 1979. A review of mass movement processes, sediment and acoustic characteristics, and contrasts in slope and base-of-slope systems versus canyon-fan-basin £oor systems. In: Doyle, L.J., Pilkey, O.H. (Eds.), Geology of Continental Slopes. SEPM Spec. Pub. 27, 61^73. Normark, W.R., Curray, J.R., 1968. Geology and Structure of the Tip of Baja California, Mexico. Geol. Soc. Am. Bull. 79, 1589^1600. Normark, W.R., Piper, D.J.W., 1969. Deep-sea fan valleys, past and present. Geol. Soc. Am. Bull. 80, 1859^1866. Normark, W.R., Posamentier, H., Mutti, E., 1993. Turbidite systems: State of the art and future directions. Rev. Geophys. 31, 91^116. Rabassa, J., Clapperton, C.M., 1990. Quaternary glaciations of the southern Andes. Quat. Sci. Rev. 9, 153^174. Ranero, C.R., von Huene, R., 2000. Subduction erosion along the Middle America convergent margin. Nature 404, 748^ 752. Shepard, F.P., 1973. Submarine canyons and other marine valleys. In: Submarine Geology. New York, pp. 304^341. Smith, W.H.F., Sandwell, D.T., 1997. Global sea£oor topography from satellite altimetry and ship depth soundings. Science 277, 1956^1962. Smithsonian Institution, Washington, Global Volcanism Program, http://www.volcano.si.edu/gvp/volcano/vbd_geog. htm. Thornburg, T.M., Kulm, L.D., 1987. Sedimentation in the Chile Trench: Depositional morphologies, lithofacies, and stratigraphy. Geol. Soc. Am. Bull. 98, 33^52. Thornburg, T.M., Kulm, L.D., Hussong, D.M., 1990. Submarine-fan development in the southern Chile Trench: A dynamic interplay of tectonics and sedimentation. Geol. Soc. Am Bull. 102, 1658^1680. von Huene, R., Scholl, D.W., 1991. Observations at convergent margins concerning sediment subduction, subduction erosion, and the growth of continental crust. Rev. Geophys. 29, 279^316. von Huene, R., Corvala¤n, J., Flueh, E.R., Hinz, K., Korst-

MARGO 3191 19-8-02

390

J. Laursen, W.R. Normark / Marine Geology 188 (2002) 365^390

gard, J.A., Ranero, C.R., Weinrebe, W., the CONDOR Scientists, 1997. Tectonic control of the subducting Juan Ferna¤ndez Ridge on the Andean margin near Valparaiso, Chile. Tectonics 16, 474^488. von Huene, R., Ranero, C.R., Weinrebe, W., 2000. Quaternary convergent margin tectonics of Costa Rica, segmentation of the Cocos plate, and Central American volcanism. Tectonics 19, 314^334. Wessel, P., Smith, W.H.F., 1991. Free softward helps map and display data. EOS 72, 441. Williams, T.A., 1997. Basin-¢ll Architecture and Forearc Tec-

tonics, Cretaceous Great Valley Group, Secraments Basin, Northern California, Ph.D. Thesis. Stanford University, Stanford, CA. Yan‹ez, G.A., Gana, P., Fena¤ndez, R., 1998. Origin y signi¢cado geolo¤gico de la Anomal|¤a Melipilla, Chile central. Rev. Geol. Chile 25, 175^198. Yan‹ez, G.A., Ranero, C.R., von Huene, R., D|¤, J., 2001. Magnetic anomaly interpretation across the southern central Andes (32‡^34‡S): The role of the Juan Ferna¤ndez Ridge in the late Tertiary evolution of the margin. J. Geophys. Res. 106 (B4), 6325^6345.

MARGO 3191 19-8-02