ARTICLE IN PRESS

Continental Shelf Research 28 (2008) 2031–2047

Sediment accumulation in the western Gulf of Lions, : The role of Cap de Creus Canyon in linking shelf and slope sediment dispersal systems

A.L. DeGeesta,1, B.L. Mullenbacha,Ã, P. Puigb, C.A. Nittrouerc, T.M. Drexlerc, X. Durrieu de Madrond, D.L. Orangee,f

aDepartment of Oceanography, Texas A&M University, College Station, TX 77843-3146, USA bInstitut de Cie`ncies del Mar (CMIMA-CSIC), Passeig Marı´tim de la Barceloneta, 37-49. E-08003 Barcelona, Spain cSchool of Oceanography, University of Washington, Seattle, WA 98195-7940, USA dCEFREM, CNRS—University of Perpignan, 66860 Perpignan, France eDepartment of Sciences, UCSC, Santa Cruz, CA 95064, USA fAOA Geophysics Inc., 7532 Sandholdt Rd., Suite 6, Moss Landing, CA 94039, USA

Received 1 June 2007; received in revised form 10 December 2007; accepted 19 February 2008 Available online 4 March 2008

Abstract

Previous work in the Gulf of Lions (western Mediterranean ) has suggested that significant amounts of sediment escape through the western part of this tectonically passive margin, despite it being far removed from the primary sediment source (the Rhone River, 160 km to the NE). The primary mechanism behind this export is hypothesized to be the interaction of a regional, southwestward sediment-transport path with a canyon deeply incising the southwestern part of the shelf, Cap de Creus Canyon. To understand the pattern of off-shelf sediment export from the western Gulf of Lions, and more specifically, the role of Cap de Creus Canyon in this transport, box cores were collected within the canyon and on the adjacent shelf during five cruises from November 2003 to April 2005. Geochronology (210Pb-derived accumulation rates), grain-size distributions, and sedimentary structures (X-radiography) were analyzed to assess temporal and spatial sedimentation patterns. Results indicate two mid-shelf depocenters (30–90 m water depth) in the northern and southern portions of the study area, separated by a zone of bypassing due to current acceleration around a headland (Cap Bear). Estimates of a sediment budget indicate that 6–8% of the sediment input to the Gulf is sequestered on the shelf . Within the Cap de Creus Canyon, there is a significant spatial asymmetry in both grain size and accumulation rates. The northern flank is a modern depocenter of fine-grained sediments, while the southern flank is primarily non-depositional for mud and includes locations of apparent erosion. This suggests the influence of multiple oceanographic processes supplying sediment to the canyon: advection of nepheloid layers from the northern rim that provide a relatively continuous sediment supply (over decadal timescales), and episodic strong currents affecting the southern rim, which can scour sediment from the southern flank. The mid-depth thalweg has an ephemeral mud layer, overlying sand and consolidated mud. The mud layer appears to be flushed down canyon periodically. The canyon head contains coarse material, suggesting reworked sands may be entering. The 100-year sediment budget, based on accumulation rates for the fine-grained fraction in the upper canyon, indicates that 1% of the total sediment input to the Gulf is accumulating in upper Cap de Creus Canyon. However, this number may significantly underestimate the total sediment entering the canyon because water-column measurements show that sediment is likely moving through the upper canyon during major dense-water cascading events from the shelf and being deposited deeper in the canyon system. The ephemeral mud layer also indicates rapid deposition and frequent flushing of sediment through the upper canyon. Overall, this study

ÃCorresponding author. E-mail address: bmullenbach@.tamu.edu (B.L. Mullenbach). 1Current address: Chevron Energy Technology Company, 1500 Louisiana Street, Houston, TX 77006, USA.

0278-4343/$ - see front matter r 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.csr.2008.02.008 ARTICLE IN PRESS 2032 A.L. DeGeest et al. / Research 28 (2008) 2031–2047 shows that Cap de Creus Canyon is an active conduit of sediment past the shelf break, despite its location distal to the primary sediment source to the Gulf. r 2008 Elsevier Ltd. All rights reserved.

Keywords: Submarine canyon; Gulf of Lions

1. Introduction studies have shown that significant quantities of sediment do not reach the shelf break adjacent to the Rhone River Submarine canyons are common bathymetric features mouth (Got and Aloisi, 1990; Arnau et al., 2004). Rather, that can strongly impact across-margin sediment dispersal hydrodynamical modeling and canyon mooring studies patterns (Shepard and Dill, 1966; May et al., 1983). Many suggest that sediment is funneled towards the narrowing, canyons incise far onto the continental shelf, effectively southwestern portion of the Gulf and the westernmost narrowing the distance between fluvial sediment sources canyon, Cap de Creus Canyon (Palanques et al., 2006; and the shelf break and slope. This property allows Heussner et al., 2006; Ulses et al., this issue). This canyon, canyons to intersect along-margin transport and serve as although far from the primary sediment source (160 km), preferential conduits of sediment past the shelf break may directly intercept the regional transport pathway and (Mullenbach et al., 2004). Off-shelf transport is particularly therefore serve as a primary conduit of sediment escape for enhanced in canyons along many modern collisional the entire GOL. margins where narrow shelves and high fluvial sediment The overall goal of this project is to understand the input co-exist (e.g. Milliman and Meade, 1983; Paull et al., processes of sediment export from the western GOL and, 2003; Walsh and Nittrouer, 2003). In contrast, canyons on more specifically, the role of Cap de Creus Canyon in this tectonically passive margins can be less effective at transport. The objectives are to: (1) identify modern exporting sediment due to greater shelf width and presence depocenters on the western shelf and within Cap de Creus of estuaries, which can trap sediment (e.g. Gardner, 1989). Canyon, (2) infer transport mechanisms creating the A necessary component for off-shelf transport via observed depositional patterns, and (3) calculate a sedi- canyons is the interception of active sediment-transport ment budget for upper Cap de Creus Canyon and the pathways. For some collisional margins, a canyon may be nearly connected to a river mouth, directly funneling the fluvial sediment to the deep sea (e.g. the Sepik and Monterey) (Kineke et al., 2000; Paull et al., 2003). On active margins, canyons also have been shown to intercept significant along-shelf fluxes associated with proximal river discharge (e.g. the Eel and Quinault) (Hickey et al., 1986; Mullenbach, 2002; Puig et al., 2003). The interception of along-margin sediment dispersal can create significant off- shelf sediment escape, which may not be directly related to the distance between a fluvial sediment source and a canyon head. Along-margin variability, such as changes in shelf width and shoreline irregularity (e.g. headland intrusion), can control the proximity of canyons to large sediment fluxes on the shelf and potentially create off-shelf steering far from a fluvial source. Oceanographic conditions, such as favorable along- and across-shelf currents, also play a role in the primary location of off-shelf transport within a system (i.e. distal vs. proximal to the source). Determining the main area of sediment export from a shelf can have implications for identifying long-term depocenters. If distal canyons are active, the specific processes allowing sediment to escape need to be further defined. The western Gulf of Lions (GOL) is a wide, passive Fig. 1. Map of the GOL, including inset to show location of the Gulf and margin in which the primary area of off-shelf sediment box to show study area. The approximate location of the Liguro- escape is hypothesized to be at a location distal to the main Provencal-Catalan Current is shown by the gray arrow, with the main sediment source (Durrieu de Madron et al., 1990; Courp canyons discussed in this paper marked. Dashed boxes include accumula- tion rate ranges for different margin areas as determined by Zuo et al. and Monaco, 1990; Got and Aloisi, 1990; Durrieu de (1997). Cap de Creus (marked by a P for promontory), the cape on the Madron, 1994; Frignani et al., 2002)(Fig. 1). Previous northern coast of Spain, critically influences the regional currents. ARTICLE IN PRESS A.L. DeGeest et al. / Continental Shelf Research 28 (2008) 2031–2047 2033 western shelf. This paper will provide insight into the 1994; Durrieu de Madron et al., 1999). Further, miner- modern processes moving sediment through a submarine alogical evidence indicates that Rhone sediments are canyon distal to a fluvial source and the importance of reaching Lacaze-Duthiers Canyon (western GOL) (Frigna- margin characteristics and regional currents in creating off- ni et al., 2002). Previous studies by Durrieu de Madron et shelf conduits. al. (1990), Heussner et al. (2006),andPalanques et al. (2006) show that suspended-sediment concentrations and 2. Background particulate fluxes within the canyons increase westward, away from the Rhone sediment source, and that the 2.1. Geologic setting western Gulf is likely the primary outlet of sediment for the entire GOL. Extending from the Pyreanean Mountains of northern Spain to the Alps of eastern France, the GOL is a 2.2. Oceanographic setting tectonically passive margin whose shelf reaches a maximum width of 72 km (near the mouth of the Rhone) and narrows The GOL is a wave-dominated, microtidal environment westward to less than 15 km near Cap de Creus Canyon (Certain et al., 2005). The general circulation pattern is (Berne et al., 2004)(Fig. 1). The shelf break is incised by at dominated by the Liguro-Provencal-Catalan (LPC) or least 12 submarine canyons that coalesce into two primary Northern current, which flows southwestward along the channels on the deep slope (Canals et al., 2004). Although slope, and creates an overall east-to-west circulation along the origin of these canyons is still debated, recent studies the outer shelf (Durrieu de Madron et al., 1990; Millot, suggest that the canyons were formed from halokinesis- 1990)(Fig. 1). The LPC enters the Gulf via the Liguran Sea derived depressions above Messinian evaporates, which to the east and moves at speeds of up to 50 cm/s towards were then enhanced by turbidity current erosion and mass the Catalan margin (Millot, 1990), transporting particles wasting (Canals et al., 2004). Some of these canyons extend westward along the shelf and slope. relatively far onto the shelf (e.g. Cap de Creus and Lacaze- Episodic Marine (southeasterly) wind can create sig- Duthiers Canyons in the western Gulf), while others nificant wave energy within the GOL and thereby initiate much farther from shore (e.g. Grand and Petit resuspend sediments. The limited fetch associated with Rhone Canyons in the eastern Gulf). Previous seabed continental winds (-northwesterly; - studies have identified a mid-shelf mud deposit (MSMD) northerly) reduces their impact on sediment resuspension on most of the shelf (Jago and Barusseau, 1981). Modern (Ferre´et al., 2005; Guille´n et al., 2006). Both Marine and sands extend from the shoreline to 30 m water depth (the Tramontane wind regimes transport coastal water and beginning of the MSMD), and reworked Pleistocene sands suspended sediment towards the southwest (Ulses et al., are exposed on the outer shelf (Courp and Monaco, 1990; this issue). Certain et al., 2005). Continental winds also are responsible for winter dense- The main source of sediment to the GOL is the Rhone water formation on the GOL shelf. Density changes River, which provides 7.4–10.1 106 tons of sediment induced by wind cooling of shallow shelf water cause the annually and accounts for approximately 94% of the total water to sink to a level of neutral buoyancy, establishing sediment input to the Gulf (Pont et al., 2002; Bourrin and unique thermohaline circulation within the Gulf. Ulses et Durrieu de Madron, 2006). Six smaller, western rivers al. (2008) reported that these waters tend to accumulate on (He´rault, , , Agly, Tech, Teˆt) supply most of the the outer shelf until they eventually cascade down the slope remainder of the Holocene sediment to the GOL (Fig. 1). as a near-bottom, cold-water density flow that could act as These rivers, which flow primarily from the Pyrenees, a possible mechanism of escape for suspended sediment. respond quickly to climatic variations and have episodic These cascades have been identified in several submarine discharge that is difficult to quantify (Certain et al., 2005). canyon heads along the entire GOL margin (Palanques Consequently, little data exist on the annual sediment and et al., 2006), being particularly intense in the Cap de Creus water discharge of these rivers. In their compilation, Canyon, and have been recognized as a mechanism moving Miralles et al. (2005) reported 210Pb-derived sedimentation sediment into the GOL canyons (Palanques et al., 2006; rates of 0.02–0.65 cm/yr in the GOL. The highest rates were Canals et al., 2006; Heussner et al., 2006). located just west of the Rhone River mouth, while the lowest were found on the deep-slope interfluves and in the 3. Methods deep sea. Previous studies have shown that sediment is escaping 3.1. Field methods the shelf via submarine canyons (Durrieu de Madron et al., 1999; Heussner et al., 2006; Palanques et al., 2006). Sediment samples were collected using a box corer Enhanced suspended-sediment concentrations within the (20 30 cm cross section, 60 cm length) during five cruises boundaries of the Grand Rhone Canyon indicate a in the western GOL from November 2003 to April 2005 funneling of sediment, suggesting canyons are preferential (Fig. 2). Upon recovery, cores were subsampled for conduits of sediment off the shelf (Durrieu de Madron, radionuclide and textural analysis using push-core tubes ARTICLE IN PRESS 2034 A.L. DeGeest et al. / Continental Shelf Research 28 (2008) 2031–2047

low accumulation rates (sediment4100 years old). Excess 210 210 Pb activity ( Pbxs, the activity particles attain while sinking through the water column) was calculated by subtracting supported from total activity. In order to account for variability in activity due to down-core grain- size variations, activities were normalized to the percent of clay in each sample interval (Fig. 3). Accumulation rates over decadal to 100-year timescales 210 were calculated using the decrease of Pbxs activity with depth in the seabed. Assuming the rate of biological mixing is negligible relative to the rate of sediment accumulation, 210 the exponential decrease of Pbxs activity with depth gives an accumulation rate: lz A ¼ , lnðCz=CoÞ where A is the accumulation rate, l is the decay constant 210 for Pb (0.693/half life), z is the depth in the seabed, Co is the excess activity of sediment accumulating at the sediment–water interface (assumed constant over 100-year timescale), and Cz is the excess activity at depth z (Nittrouer et al., 1979). These are average accumulation rates based on the assumption of steady-state accumulation Fig. 2. Map showing core locations and designated physiographic . over a 100-year timescale. Because data on biological The shelf has three distinct regions (as distinguished by sedimentary mixing rates were not collected as part of this study, the characteristics): the northwestern, middle, and southwestern shelf. The influence of bioturbation on the observed 210Pb profiles canyon has four regions: the canyon head (1), the thalweg (excluding the canyon head) (2), the northern flank (3), and the southern flank (4). Inset cannot be ascertained. It is unknown if the assumption of shows the location of across-canyon coring transect A–A0 (Fig. 8). negligible biological mixing used for the accumulation rate calculation is valid. Therefore, all references to accumula- (15 cm ID, thin-walled PVC tubing). Sediment within the tion rates and sediment budgets herein are reported as sub-cores was extruded, sampled in 1-cm depth intervals, maximum apparent rates and upper-bound budget values, and stored for laboratory analysis. Plexiglass trays were respectively. Sandy layers that exhibited little down-core used to extract vertical slabs of sediment, which were variation in 210Pb activity were simply classified as recent X-rayed shipboard for estimating density variations, (within the last 100 years) or relict (older than 100 years) 210 stratigraphy and sedimentary structures. based on the presence or absence of Pbxs activities, respectively. 3.2. Radioisotopic analysis 3.3. Sedimentological analysis Down-core changes in radioisotopic 210Pb activities (t1/2 ¼ 22.3 years) were used to examine the decadal to Sedimentological characteristics of each core were 100-year sediment records by a technique modified from assessed by grain-size analysis. Grain-size profiles for each Nittrouer et al. (1979). Dry samples were spiked with a core were created by wet sieving each depth increment at 209Po tracer (used for yield determination), sediments were 63 mm to isolate sand and coarser materials from finer 209 210 leached with 16 N HNO3 and 6 N HCl, Po and Po fractions. All material passing through the sieve was were plated onto silver planchets by spontaneous electro- analyzed using a Sedigraph 5100 particle size analyzer to deposition, and planchets were counted for relative alpha determine the distributions of silt- and clay-sized particles. decay (disintegrations per minute, dpm) over a 24-h period Profiles of % mud were created for comparison with and on a Canberra Alpha Analyst detector. Total 210Pb activity correction of radioisotopic data. was determined by assuming 210Po is in secular equilibrium with 210Pb (Nittrouer et al., 1979). Samples that had large 3.4. Ancillary data fractions of sand or shell fragments were treated similarly to muddy samples; however very large, articulated shells or High-resolution multibeam bathymetry of Cap de Creus wood fragments (anything that would not behave similarly Canyon was acquired, processed, and provided by Fugro to the surrounding sediments) were removed prior to Survey Ltd. and AOA Geophysics Inc. GOL regional analysis. Supported 210Pb activity (derived from the decay multibeam bathymetry was supplied by S. Berne (IFRE- of its effective parent, 226Ra, in the seabed) is determined MER) (Berne et al., 2002). Additionally, hydrographic by measuring 210Pb activity at the bottom of a core with time series (currents, suspended-sediment concentrations, ARTICLE IN PRESS A.L. DeGeest et al. / Continental Shelf Research 28 (2008) 2031–2047 2035

Shelf Cores temperature) from moorings and tripods located at % Sand % Sand 145–750 m water depth in the canyon (see Puig et al., this 0 20 40 60 80 100 020406080100 0 0 issue for details) have been used in this study to interpret canyon sedimentation processes. 5 10 4. Results 10 4.1. Definition of regions 20 15 Depth (cm) Depth (cm) For the purpose of this study, the western GOL shelf and 1.6 mm/yr 20 1.3 mm/yr 30 slope is divided into regions based on sedimentary characteristics, oceanographic conditions, and seafloor morphology. The shelf is separated into three regions: (1) CSD58 XX50 25 40 the northwestern shelf (from the northern limit of the study 0.1 1 10 100 0.1 1 10 100 area to Cap Bear), (2) the middle shelf (from Cap Bear to Excess Pb-210 Activity Excess Pb-210 Activity the Spanish-French border), and (3) the southwestern shelf (dpm/g of clay) (dpm/g of clay) (from the border to Cap de Creus headland) (Fig. 2). The Rim Cores portion of the shelf near Cap de Creus Canyon is classified % Sand % Sand as the canyon rim. Cap de Creus Canyon is divided into 0 20 40 60 80 100 0 20 40 60 80 100 four morphological regions: (1) canyon head, (2) thalweg 0 0 (excluding the canyon head), (3) northern flank, and (4) southern flank (Fig. 2). 5 5

10 4.2. Western Gulf of Lions continental shelf

0.7 mm/yr 10 0.9 mm/yr 15 Grain-size results from all shelf regions studied show Depth (cm) Depth (cm) patterns similar to those defined by previous studies for the

15 20 entire GOL region (Fig. 4). A MSMD, composed primarily of silty clays or clayey silts (all o40% sand), is located in 25 30–85 m water depth. Cores collected deeper on the shelf CST135 20 #20 were generally composed of coarse sands. No cores were

0.1 1 10 100 0.1 1 10 100 collected in water depths shallower than 30 m, but previous Excess Pb-210 Activity Excess Pb-210 Activity research indicates that nearshore sands exist out to (dpm/g of clay) (dpm/g of clay) 20–30 m water depth and therefore this value is used as the inner boundary for this study (Martin et al., 1981; Canyon Cores % Sand % Sand Courp and Monaco, 1990; Got and Aloisi, 1990; Certain 0 20 40 60 80 100 0 20 40 60 80 100 et al., 2005). 0 0 Where the shelf begins to narrow due to a coastal 5 promontory (Cap Bear; middle shelf), the across-shelf extent of the mud begins to narrow. Core U85 is 10 5 uncharacteristically coarse grained for its water depth 2.3 mm/yr 15 (silty sand) and therefore marks the narrowest extent of the MSMD (Fig. 4). South of this line (southwestern shelf), the 1.6 mm/yr 10 20 mud deposit again extends seaward to form a crescent Depth (cm) 25 Depth (cm) shape parallel to the coastline. This deposit tapers east-

15 30 ward, giving way to sands and gravels near the headland. In general, grain size tends to coarsen southward on the 35 shelf. The finest sample collected was a silty clay taken at CFM138 CFM440 20 40 site XX80 (the most northern core transect), and the 0.1 1 10 100 0.1 1 10 100 Excess Pb-210 Activity Excess Pb-210 Activity (dpm/g of clay) (dpm/g of clay) Fig. 3. Excess 210Pb activity and sand content from different areas of the study. Circles represent excess 210Pb values, ‘‘X’’s represent percent of sand. Shelf cores tend to have steady-state accumulation (as much as 2.5 mm/yr) of fine sediment, while cores collected near the canyon rim had much lower accumulation rates and coarser grain sizes. Cores from the canyon have significant variability, but two cores from the northern flank/ rim show some areas of fine-grained sediment accumulation. ARTICLE IN PRESS 2036 A.L. DeGeest et al. / Continental Shelf Research 28 (2008) 2031–2047

Fig. 4. Surficial grain-size distributions on the GOL shelf. All cores collected from 50 to 85 m water were primarily muddy, whereas those collected farther seaward on the shelf tend to be sandy (some with significant shell hash) or gravelly. Categories are distinguished by the predominant grain size within the sample: gravel: sample contained particles 42 mm in diameter (excluding shell hash); sand: 475% sand; sand with mud: sample composed of mixed grain sizes, but the highest proportion was sand-sized particles; silt: sample dominated by mud (o63 microns) and contained more silt than clay; and clay: sample composed of o20% sand with more clay than silt. The dashed line denotes the region where the MSMD would be expected based on previous work. The down-core grain-size profile of core XX50 is included as an example of down-core variability. Inset shows near-bottom current velocities from an ADCP survey in February 2005. Note the acceleration around headlands and the near-shore anticyclonic eddy just north of Cap de Creus headland. coarsest samples were composed of sand and gravel the northwestern region tend to be finer grained and have less collected at sites near Cap de Creus (cores J01, #3, and variability than those in the southwestern region. Accumula- NCC80, Fig. 4). The R-line (south of Cap Bear) is an tion rates are 1.0–2.0 mm/yr on the northwestern region of exception to this generality. Although still classified as the shelf, with the highest rates occurring at 80-m water muddy samples, these cores (R55 and R87) have greater depth. In the southwestern region, cores were generally amounts of sand (430%) than those collected at similar sandier and the highest accumulation rate was 2.5 mm/yr at depths elsewhere on the shelf (o30%). There does not site U50. Accumulation rates on the southwestern shelf appear to be significant down-core variability in grain size decrease southwestward, creating the crescent shaped deposit for most cores collected on the western shelf. defined previously by grain-size characteristics. Accumulation rates on the shelf reach as much as Many cores collected on the shelf at the canyon rim are 2.5 mm/yr. The highest rates are generally associated with sandy and tend to coarsen seaward, creating deposits of the MSMD (Figs. 3 and 5). However, cores collected along 470% sand at locations adjacent to the canyon head (Figs. the R-line (R55 and R87) reveal uncharacteristically low 3 and 4). Accumulation rates vary with location (Fig. 5). rates at the same location where grain size was coarser than Comparison of core characteristics from the canyon rim to other parts of the MSMD (at the middle shelf). These data the north and south of Cap de Creus Canyon reveals suggest that there are two distinct ‘‘cells’’ of accumulation distinct differences. To the north, the shelf is predomi- on the shelf: the northwestern-shelf MSMD (north of Cap nantly clayey sand (o70% sand with more clay than silt) Bear) and the southwestern shelf deposit (Fig. 5). Cores in with appreciable accumulation rates (i.e. 1.6 mm/yr at ARTICLE IN PRESS A.L. DeGeest et al. / Continental Shelf Research 28 (2008) 2031–2047 2037

Fig. 5. Accumulation rates on the western shelf. All rates are assumed to be maximum apparent accumulation rates due to uncertainty of the influence from biological mixing on 210Pb profiles. Rates reach as much as 2.5 mm/yr, with the highest rates generally occurring within the MSMD on the shelf (from 30 to 80 m water depth). Some cores were too coarse- grained to obtain valid accumulation rates and are indicated by the black circles. Three main regions were evident from the accumulation pattern: the northwestern shelf (with a mid-shelf mud deposit), southwestern shelf (with a mid-shelf mud deposit), and middle shelf (bypassing).

CFM138, Fig. 6). To the south, canyon rim cores are generally very coarse grained (sand and gravel up to 4 cm diameter at site J01) and show no evidence of fine-sediment accumulation (actual rates could not be determined due to Fig. 6. Cap de Creus Canyon grain-size and accumulation rate distribu- tion. The canyon head and rim are clearly dominated by coarse-grained the large proportion of coarse material in these samples) material and low accumulation rates. The northern flank and thalweg tend (Fig. 6). Box core attempts in this area returned only a few to be dominated by finer sediments with higher accumulation rates. The centimeters of coarse-grained and/or consolidated mud canyon also shows some erosional points evidenced by consolidated muds despite high fall rates of the corer. This indicates strong, with no excess 210Pb activity, including cores CFM230 and CTM571. resistive material not easily removed from its present setting (the southern rim of the canyon). disappears at 300–400 m. At this point, there is a distinct 4.3. Cap de Creus Canyon switch to soft, unconsolidated muds overlying coarser material in the thalweg region (Fig. 7B). This pattern 4.3.1. Canyon head and thalweg extends through portions of the thalweg to the base of the Surficial grain-size patterns and accumulation rates study area (Fig. 7C). within Cap de Creus Canyon are more complex than those The distinct mud layer (4–22 cm in thickness) is primarily on the shelf. Surface sediments in the canyon head are contained within the thalweg (Fig. 7C). It is characterized coarse grained (sand and shell hash) down to 400 m water by: (1) very fine grain size (primarily silty clays, coarsening depth (Fig. 6). The coarse-grained nature of these cores slightly down canyon), (2) low dry bulk density (average inhibit determination of accumulation rates, but the sand 0.75 g/cm3), (3) a lack of sedimentary structures, (4) 210 210 layer in the canyon head was found to have Pb activity relatively high Pbxs activities (425 dpm/g of clay), and 210 above supported levels, which indicates recent accumula- (5) little or no evidence of a down-core decrease in Pbxs tion (within the last 100 years)(Fig. 7A). This surficial sand activity within the layer. layer unconformably overlies a stiff, consolidated gray Minimum accumulation rates, determined by assuming mud with supported levels of 210Pb activity, indicating that that this layer had to be deposited within one half life (22.3 the basal muds are older than 100 years. This distinct years; based on the lack of down-core decrease in 210Pb layering of coarse sediment overlying consolidated muds activity), are between 5.0 and 10 mm/yr. Many of the becomes less recognizable down the thalweg until it thalweg coarse basal layers have 210Pb activity above ARTICLE IN PRESS 2038 A.L. DeGeest et al. / Continental Shelf Research 28 (2008) 2031–2047

Fig. 7. Example cores from the canyon head and thalweg mud layer (A and B), revealing the extent of the mud layer (C). Core CC2-150 shows an example from the canyon head, where relatively clean sand unconformably overlies consolidated muds (A). A mud layer, 4–22 cm thick, drapes the thalweg at 400–780 m water depth (the end of the study area) (B and C). This mud layer has high water content and consistent levels of excess 210Pb. It overlies a sand layer and consolidated mud layer (B). supported levels, indicating that they are also less than 100 (CTM571a) collected in April 2005, a re-occupation of site years old. One thalweg core, CTM492, reveals consolidated CTM571 from October 2004, did not resemble the thalweg mud unconformably underlying the sands seen elsewhere at mud layer. However, the ability to exactly reoccupy a site is the base of cores (Fig. 7B). difficult in canyons so it is unknown if the absence of the mud layer was due to erosion of the previously observed mud layer, or simply due to sampling variability. 4.3.2. Northern and southern flanks Accumulation rates on the flanks are variable, ranging 4.4. Construction of the sediment budget from 0 to 44.1 mm/yr (Fig. 6). An across-canyon profile from transect A–A0 shows a clear asymmetry between the In order to determine the relative importance of the canyon flanks: significantly more accumulation occurs on western GOL in sequestering sediments, a semi-quantita- the northern flank than the southern (Fig. 8). Most cores tive budget was created for the shelf and Cap de Creus collected on the northern flank have accumulation rates upper canyon. The annual amount of sediment deposited greater than 1.5 mm/yr (Fig. 3); in contrast, those on the was calculated based on the following equation: southern side are either erosional (as evidenced by the X presence of only consolidated muds or coarse grains) or Mt ¼ ðAiÞðMi riÞ, have very low accumulation rates (Fig. 6). The cores taken where Mt is the total sediment mass, Ai is the surface area for within the thalweg have high accumulation rates associated a specific region, ri is the average bulk density (based on with the thalweg mud layer discussed previously (e.g. core modern sediments) forthatregion,andMi is the average CTM571). sediment thickness for that region. This is defined as: Reoccupation of sites within the canyon showed little evidence of seasonal variation from March 2004 to Mi ¼ðRiÞð1yearÞ, 210 February 2005. Within the thalweg mud layer, thicknesses where Ri is the mean Pb-derived accumulation rate for that varied slightly but spatial resolution was too low to know if region. For calculations involving the thalweg mud layer, the this was simply an artifact of sampling variability. One core thickness used was the average of the mud-layer thicknesses ARTICLE IN PRESS A.L. DeGeest et al. / Continental Shelf Research 28 (2008) 2031–2047 2039

Fig. 8. Transect A–A0 across the canyon at 571 m thalweg water depth (see Fig. 2 for transect location). The canyon profile shows X-radiographs of cores taken from both flanks. Cores from the southern flank show coarse-grained material, consolidated muds, and low accumulation rates. Cores from the northern flank are finer grained and actively accumulating. Thalweg cores (e.g. CTM 571) show the expected mud layer overlying coarse-grained material.

Table 1 Shelf budget calculations

Area (m2) Mean accum # of cores Standard Mean bulk Total sediment (kg/yr) ratea (mm/yr) used deviation density (g/cm3)

Southwestern shelfb 3.27 107 1.3 9 0.5 1.0670.15 (4.671.8) 107 (4.671.8) 104 ((metric tons sed)/yr) Northwestern shelfb 3.85 108 1.6 4 0.5 0.93070.16 (5.772.0) 108 (5.772.0) 105 ((metric tons sed)/yr)

aMaximum apparent accumulation rates because of uncertainty in the influence of biological mixing on 210Pb profiles. bSee Fig. 5 for shelf regions. in all cores collected in that region. In regions without water depth) and the Zone of Bypassing (Fig. 5). Within complex bathymetry (such as the continental shelf), the the canyon, the four main regions used in calculations are accumulation rates and bulk densities will likely be similar similar to those defined previously: the canyon head, the over broad areas. However, within the canyon, where there mud layer (in the thalweg), the northern flank, and the are dramatic bathymetric changesoversmallerscales,there southern flank (Fig. 2). may be significant variability at the canyon scale. Mt is defined for each physiographic region (see inset of Fig. 2), 5. Discussion using mean accumulation rates and bulk densities for each region individually. A rough estimate of the total amount of 5.1. Shelf sedimentation processes sediment accumulating within Cap de Creus upper canyon and on the western GOL continental shelf is then calculated Grain-size patterns and accumulation rates on the shelf by summing the mass of all regions (Tables 1 and 2)and indicate that fine grains are primarily accumulating at mid- compared to Rhone River discharge (Tables 3 and 4). shelf depths (30–85 m), which is in agreement with previous Regions are defined by the depocenters previously studies (Martin et al., 1981; Got and Aloisi, 1990, Certain discussed in this study. All calculations are based on fine- et al., 2005)(Figs. 4 and 5). Got and Aloisi (1990) describe grained sediment only (this assumes that most coarse- how across-shelf transport of a benthic nepheloid layer grained material would be re-worked sediment that is not (BNL) can create this feature. Sediments are transported applicable to this budget). On the shelf, boundaries of beyond wave base, leaving the inner shelf bare of fine regions are set as the extent of the MSMD (30–85 m grains. Those sediments reaching the outer shelf remain in ARTICLE IN PRESS 2040 A.L. DeGeest et al. / Continental Shelf Research 28 (2008) 2031–2047

Table 2 Canyon budget calculations

Area (m2) Mean accum. # of cores Standard Mean bulk Total sediment (kg/yr) ratea (mm/yr) used deviation density (g/cm3)

North Flankb 3.33 107 2.0 5 1.4 0.89270.18 (6.074.4) 107 ((kg sed)/yr) (6.074.4) 104 ((metric tons sed)/yr)

Area (m2) Mean thickness # of cores Standard Mean bulk Total sediment (kg/22 yr) (m) used deviation density (g/cm3)

Mud Layerb 9.57 106 0.10 10 0.007 0.74270.14 (6.771.4) 108 ((kg sed)/(22 yr)) (6.771.4) 105 ((metric tons sed)/22 yr)) (3.070.6) 104 (metric tons sed)/yr))

aMaximum apparent accumulation rates because of uncertainty in the influence of biological mixing on 210Pb profiles. bSee Fig. 2 for canyon regions: north flank—Region 3, mud layer—Region 2.

Table 3 Comparison of shelf budget data with Rhone discharge

Rhone Sed discharge Southwestern shelfb,c Northwestern shelfb,c Total western shelfb,c (tons/yr) (% of value) (% of value) (% of value)

Low estimatea 7.4 106 0.6 7.7 8.3 High estimatea 10.1 106 0.5 5.6 6.1 Total sediment (assuming Rhone is 94%) Low estimatea 7.9 106 0.6 7.2 7.8 High estimatea 10.7 106 0.4 5.3 5.7

aEstimates from Pont et al. (2002) and Bourrin and Durrieu de Madron (2006). bSee Fig. 5 for shelf regions. cUpper-bound budget values based on maximum apparent accumulation rates.

Table 4 Comparison of canyon budget data with Rhone discharge

Rhone Sed discharge North flankb,c Mud layerb,c Upper canyonb,c (tons/yr) (% of value) (% of value) (% of value)

Low estimatea 7.4 106 0.8 0.4 1.2 High estimatea 10.1 106 0.6 0.3 0.9 Total sediment (assuming Rhone is 94%) Low estimatea 7.9 106 0.7 0.4 1.1 High estimatea 10.7 106 0.5 0.3 0.8

aEstimates from Pont et al. (2002) and Bourrin and Durrieu de Madron (2006). bSee Fig. 2 for canyon regions: north flank—Region 3, mud layer—Region 2. cUpper-bound budget values based on maximum apparent accumulation rates. suspension and move southwestward in the regional (Fig. 4). This pattern of mid-shelf deposition, due to the currents, leaving the outer shelf bare of modern sediment inability of surface waves to impact the seafloor under deposits (Got and Aloisi, 1990). The across-shelf sampling increasing water depths and the lack of mechanisms to in this study is somewhat limited, but the results support transport sediment across the shelf at greater depths, this model: coarser sediments are deposited at shallower is observed on shelves worldwide, including the US west depths (e.g. XX50) but finer sediments are held in coast, the Amazon shelf, and other locations within the suspension until deeper depths (e.g. XX80). There, mud Mediterranean (Nittrouer and Sternberg, 1981; Got and accumulation rates are elevated and only relict sands Aloisi, 1990; Nittrouer and Wright, 1994; Palaques et al., with low mud accumulation rates exist on the outer shelf 2002). ARTICLE IN PRESS A.L. DeGeest et al. / Continental Shelf Research 28 (2008) 2031–2047 2041

Along-shelf sedimentation patterns within the western Harris, 2000). In deeper water, the strength of the eddy on GOL are more complex (Figs. 3–5). The primary forcing is the bottom is reduced, allowing deposition. The conceptual a regional southward flow that moves along the entire model presented here is that the coastal current is deflected shelf. This flow, which can be enhanced by meanders of the by Cap Bear, part of the flow moves off the shelf around LPC current onto the shelf, transports particles southward Cap de Creus, while another part circulates as an eddy on for a long distance before they accumulate on the bed. the southwestern shelf. It is hypothesized that such eddies However, small-scale flow variations can regulate both the enhance accumulation near their center, and may be location and characteristics of deposits (i.e. faster flows responsible for enhanced rates on the southwestern shelf. tend to leave coarser grains) (Davies et al., 1995; Pawlak Seismic profiles also show a ‘‘mound’’ at this location, and MacCready, 2002). which points to similar processes operating over the The northwestern shelf has a smooth coastline and Holocene highstand (Ercilla et al., 1995). experiences a general southward flow throughout the year Near Cap de Creus Canyon, accumulation rates and with little flow disturbance. The similarity between this grain-size distributions on the northern and southern rim region and other parts of the Gulf is reflected by similar are drastically different (Figs. 5 and 6). Northern rim cores sedimentation processes and depositional patterns, i.e. the were composed of a muddy, fine sand (35% mud) and had MSMD (Got and Aloisi, 1990; Courp and Monaco, 1990; appreciable steady-state accumulation rates (1.6 mm/yr at Certain et al., 2005). CFM138; Fig. 3), indicating that sediment reaches this area Near Cap Bear, the seabed is sandier than the northern regularly over decadal timescales. In contrast, the southern shelf (36% sand at R55 relative to less than 20% sand at rim is composed of very coarse sediments (gravel up to K50) (Fig. 4). Combined with low accumulation rates 4 cm at J01) and consolidated material, which is indicative (o1.0 mm/yr) this suggests inhibited deposition in a zone of currents scouring the shelf and preventing fine sediment of likely sediment bypassing. ADCP data recorded near deposition. This comparison suggests that the north rim Cap Bear in November 2003 and Cap Creus in February represents a potential sediment pathway to the upper 2005 (Fig. 4 inset) capture significant near-bottom currents canyon; in contrast, the southern rim is an area of sediment moving towards the southeast, suggesting current accel- bypassing that may provide a pathway to the deeper parts eration around the headlands. This acceleration is of the canyon. hypothesized to be caused by the Venturi effect, which dictates that when a confined flow encounters a restriction, 5.2. Canyon asymmetry it must speed up to compensate for reduced pressure created by the formation of a pressure gradient behind the Within Cap de Creus Canyon, each flank is classified as restriction (i.e. acceleration due to conservation of mass depositional or non-depositional with respect to fine through a constriction). In the western GOL, the flow sediment. The northern flank has relatively high accumula- restrictions are Cap Bear and Cap de Creus. The offshore tion rates (up to 4.1 mm/yr) and fine-grain sizes (primarily flow boundary (which is required for the Venturi effect to silty clays), indicating an area of fine-sediment deposition be valid) is likely the intrusion of the LPC current on the over a 100-year timescale (Figs. 6 and 8). The southern outer shelf. Analogous acceleration around a headland has flank, however, has erosional features (i.e. CFM560), low been documented by Geyer and Signell (1990) and Geyer accumulation rates (0.5 mm/yr at CFM500), and the (1993). The variability of these strong flows around Cap presence of gravel (J01), suggesting that it is an area of Bear and Cap de Creus is poorly known. However, the bypassing. As such, it is classified as non-depositional for geographic correlation between the seabed data and fine grains on a 100-year timescale. current information suggests that this acceleration around Enhanced deposition of mud on the northern flank the headland likely controls the deposition of fine suggests that fine sediments are entering Cap de Creus sediments. Canyon from the northern rim (Fig. 6). The significant The increase in accumulation rates (2.5 mm/yr at U50) asymmetry of accumulation rates within the canyon (high and fine-grain sediment (o25% sand at R55) on the rates on the northern flank, lower rates or erosion on the southwestern shelf suggest that flow separation (associated southern), as well as modern sediment accumulation on the with current deflection around Cap Bear) may allow the rim adjacent to the northern flank (1.6 mm/yr at site development of an eddy that causes increased deposition CFM138), provides evidence of this preferential deposi- south of the cape. Such eddies were observed during the tion, as well as a source for the sediment. It is hypothesized February 2005 ADCP survey (Fig. 4 inset). Geyer (1993) that the south-flowing regional currents transport the shelf documented the creation of eddies on the down-flow side of BNL over the canyon rim, where it splits into a canyon coastal headlands on a smaller scale, while Davies et al. BNL and an intermediate nepheloid layer (INL), supplying (1995) modeled eddy formation at the lee side of a cape sediment to the northern (and southern) flank. (assuming flows were sufficiently fast) at larger scales. This The strongest evidence for this conceptual model is type of separation has been shown to erode fine sediments mooring data collected within the canyon. Current meters and move sands in shallow water (o20 m), suggesting that at 200 and 500 m (both 5 mab) in the canyon show a they can have significant impacts on the seabed (Signell and distinct southward flow (maximum speed of 50 cm/s) ARTICLE IN PRESS 2042 A.L. DeGeest et al. / Continental Shelf Research 28 (2008) 2031–2047

Fig. 9. Current meter readings from 2004 to 2005 within Cap de Creus Canyon (unfiltered, recorded every 20 min; Puig et al., this issue). Currents flow to the SE in the upper canyon, where the thalweg is narrow. However, deeper within the canyon, currents travel directly east/northeast at speeds up to 80 cm/s at 5 mab, suggesting that these currents may be capable of eroding sediments. When the currents turn east/northeast, water temperature decreases (shown by grayscale change in polar plots). These currents are caused by dense-water cascading off the southern rim of Cap de Creus Canyon, likely causing the significant asymmetry in accumulation within the canyon. The current meter at 500 m water depth also shows a distinct flow to the south, confirming the likelihood of hemipelagic sediment transport from the northern rim that can supply sediment to the canyon.

within the canyon (Fig. 9). These southward flows have 1986). Similar to Quinault canyon, it is hypothesized that appreciable suspended-sediment concentrations (generally as the regional currents move over the Cap de Creus 5 mg/l), indicating that they are probable sources of Canyon, the increase in depth and deflection of the isobars sediment to the canyon (Puig et al., this issue). Durrieu de cause them to slow, allowing enhanced deposition relative Madron and Panouse (1996) and Palanques et al. (2006) to the shelf (Carson et al., 1986; Hickey et al., 1986). While have reported the presence of a well defined BNL the specific dynamics of sediment movement vary between (concentrations, 1 mg/l) extending to the shelf break. margins, nepheloid-layer advection facilitating deposition Nepheloid-layer advection of sediment into Lacaze- in canyons is a well-documented process. Duthiers Canyon (located directly to the northeast of If advective transport of nepheloid layers were the only Cap de Creus Canyon) and Grand-Rhone canyon also has process occurring within Cap de Creus Canyon, one would been observed, indicating this type of transport is active in expect fine-grained deposition across the entire canyon, the GOL (Durrieu de Madron et al., 1990; Durrieu de including the southern flank. However, data from the Madron et al., 1999; Frignani et al., 2002). southern flank indicate that sediments were generally Advection of nepheloid layers off the adjoining shelf has consolidated muds or coarse-grained material devoid of been seen on other margins, such as Quinault canyon on modern sediment. These types of deposits are generally the Washington coast (Hickey et al., 1986; Baker and representative of a high-energy environment, not hemi- Hickey, 1986; Carson et al., 1986; Snyder and Carson, pelagic sedimentation from nepheloid layers. In addition, ARTICLE IN PRESS A.L. DeGeest et al. / Continental Shelf Research 28 (2008) 2031–2047 2043 data collected within the canyon at 750 m water depth (and to 5.3. Canyon Thalweg processes a lesser degree at the 500 m site) show currents that flow due east (along the thalweg) with frequent and significant cold- There are two regions within the canyon thalweg that water currents (up to 80 cm/s) directed towards the northeast can be distinguished by surface sediment characteristics. (Fig. 9). These are opposite to the southward-flowing across- The mid-depth thalweg (400–780 m) during March canyon currents observed at shallower depths in the canyon. 2004–January 2005 was characterized by the presence of These data point to fundamentally different environmental a mud layer, which has uniform excess 210Pb activity, conditions on the southern flank, which (1) prohibit sediment overlying sand with detectable excess 210Pb activity. This deposition in this part of the canyon, and/or (2) removes indicates rapid sediment accumulation in this region recent sediment after it is deposited. (4–22 cm in less than one half life, 22 years) (Fig. 7B). In Mooring data from within the canyon and modeling contrast, the canyon head (o400 m water depth) contains work show that such strong currents occur on the southern surface sediments composed of recent sands with negligible rim and flank of the canyon during the cascading of cold, fine sediment (generally 490% sand). Therefore, the dense water off the GOL continental shelf (Fig. 9) canyon head is classified as non-depositional for modern (Palanques et al., 2006; Ulses et al., 2008). These mooring muds. It is, however, likely an area of coarse-grained data clearly indicate intense cascading events in Cap de transport and deposition. Creus Canyon: from January to April, a distinct correla- The mud layer that overlies the thalweg sands at tion (particularly at the 750-m mooring site) between fast 400–780 m water depth is indicative of rapid, non-steady currents (up to 80 cm/s) and decreased temperature state deposition within the canyon (Fig. 7). One of the most (10–12 1C, relative to the 13.2 1C in the ambient slope distinguishing characteristics of this layer is the minimal 210 water; Fig. 9) was recorded, which is consistent with amount of Pbxs decay with depth in the core. If this characteristics of dense-water formation (Durrieu de deposit had accumulated slowly over time (even over one Madron et al., 2005, Puig et al., this issue). Previous half life), the activity at the base of the layer should be only hydrographic studies have identified the western Gulf as a half of the surface activity (Fig. 7B). Rather, there is a primary location of annual dense-water formation and distinct break between relatively constant excess activity cascading off the shelf (Millot, 1990; Durrieu de Madron levels and the underlying supported levels. This suggests et al., 2005; Canals et al., 2006). Recent work has suggested that the layer was deposited very rapidly (i.e. non-steady that after formation, much of this dense water is pushed state) (Mullenbach, 2002). A conservative estimate for southward along the shelf by Tramontane winds and the accumulation rates can be calculated assuming the entire general current regime within the Gulf (Ulses et al., 2008; deposit was laid down within one half life (over 22.3 years). Bourrin et al., this issue). Once it encounters Cap de Creus This estimate gives accumulation rates up to 1.0 cm/yr for headland, the flow is impeded and the dense water begins the mud layer. However, the low bulk density throughout to pool on the outer southwestern shelf until it spills into the layer also suggests that the layer was deposited in a Cap de Creus Canyon from the southern rim. These near- time frame much shorter than the half life. bottom flows inhibit sediment deposition and erode The lack of sedimentary structure within this mud layer sediments that may have been deposited on the southern opens the possibility that these mud deposits have flank. This conceptual model is consistent with the accumulated steadily over a 100-year timescale, but are erosional seabed observed on the southern rim (Figs. 4 so biologically mixed that they appear to be episodic layers. 210 and 5). If the initial Pbxs activity of sediment reaching the Moorings also recorded a strong correlation between seabed is consistent across the canyon, then an estimate of dense-water flows and elevated suspended-sediment con- the excess activity of a biologically homogenized sediment centrations (up to 40 mg/l at 5 mab), suggesting that these layer can be calculated using steady-state profiles from the flows are able to move large amounts of sediment into and northern flanks. The assumption is that the thalweg through the upper canyon (Puig et al., this issue). Canals et deposits could have looked similar to the northern-flank al. (2006) found that cascading events can be strong cores prior to homogenization (e.g. steady-state deposi- 2 210 enough to resuspend sand (shear stresses 0.7 N/m ) and tion). This calculation shows that Pbxs activities would are thought to be responsible for creating over-consoli- be lower (10–18 dpm/g of clay) throughout the homo- dated furrows on deep portions of the southern flank. genized layer (if biological mixing were to produce the Moreover, because the measured currents and concentra- uniform activity with depth) than is actually found in the tions are still very large at 750-m water depth, sediments thalweg mud layer (25 dpm/g of clay). This suggests that must be carried past the upper canyon and deposited the mud layer is likely not the result of biological mixing of deeper in the canyon, seaward of the present study area steady-state deposits. (Fig. 9). These data combined with the seabed observations The high excess 210Pb activities recorded in both the indicate that large amounts of sediment are moving off the sands and mud layer, combined with the abrupt shift to shelf and through the upper canyon, making Cap de Creus supported 210Pb activities at depth (CTM492, Fig. 7B) Canyon an important conduit for sediment export from the indicate that the mud layer is periodically flushed (i.e. there GOL shelf. is a loss of stratigraphic time in the sediment column). ARTICLE IN PRESS 2044 A.L. DeGeest et al. / Continental Shelf Research 28 (2008) 2031–2047

Probable mechanisms are (1) dense-water cascading through the thalweg that erodes the mud layer or (2) gravity-driven sediment flows down the main thalweg. While neither can be completely discounted, results from previous studies and this work strongly point to dense- water cascading as the most likely mechanism for flushing the mud layer. Monitoring of Lacaze-Duthiers Canyon (located NE of Cap de Creus Canyon) since 1993 has shown that dense-water cascading occurs annually, with some variation in the magnitude of events (Canals et al., 2006). Near-bottom flows driven by cold-water density anomalies appear to be an annual event in the canyons of the western GOL; in contrast, there was no evidence of flows driven by sediment-induced density anomalies (i.e. turbidity currents) during this record. Cascading events also can induce strong shear stresses (0.7 N/m2) that are capable of eroding sand and can be sustained for days (Canals et al., 2006); this suggests they are competent enough to flush the unconsolidated mud layer. Because the dense-water cascades in Cap de Creus Canyon have been shown to carry significant suspended-sediment loads (Puig et al., this issue), it is reasonable to assume that sediment can be deposited following their passage. In addition, sediment-trap data from Cap de Creus Canyon at 500-m water depth (5 mab; Fig. 9) indicate that sand can be transported in suspension during cascading events (Puig et al., this issue). This suggests that the cascades can remove unconsolidated material and transport/emplace the coarse-grained material followed by a new ephemeral mud layer. All of these data point to dense-water cascading as an important mechanism for removal and formation of the Fig. 10. General processes affecting Cap de Creus Canyon as defined by core data. The dominant processes are along-shelf movement of sediment ephemeral mud layer. Regardless of the supply and to create mid-shelf mud deposits, acceleration of currents around Cap flushing mechanisms, the rapid deposition and removal is Bear to create a zone of bypassing, advection of nepheloid layers from the good evidence that significant amounts of material are northern flank of the canyon, dense-water cascading onto the southern moving through Cap de Creus Canyon over short (annual flank and scouring sediments, and possible entrance of coarse sediments to decadal) timescales. down the canyon head. Based on the prominence of sand in the canyon head and its general confinement within the thalweg, the canyon canyon head (o400 m water depth) and most of the head is the most probable site of entry for these coarse southern flank were classified as non-depositional areas grains. It is hypothesized that dense-water cascading also is and were given values of 0 tons of sediment (considering responsible for depositing sand in the canyon head. In fine-sediment accumulation only). The northern flank was addition to observations of coarse material being trans- classified as a depositional area and was calculated to ported in the dense water, Gaudin et al. (2006) found accrue (6.074.4) 104 tons of sediment annually (Table 2). similar sandy deposits in the head of the Boucart (Aude) The thalweg (from 400 to 780 m water depth), considering Canyon (next canyon east of Lacaze-Duthiers Canyon), only the mud layer, was found to have (67714) 104 tons which were attributed to reworked sand deposited by of sediment (Table 1). Because it is only definitively known dense-water flows. In summary, the dense-water cascading that this layer accumulated in less than 22.3 years, for the appears to flush material down canyon and leave a rapidly purpose of this annual budget, the total mass will be emplaced deposit in the canyon thalweg. divided by the half life to determine an average annual input value. Using this method, a minimum value of 5.4. Sediment budgets (3.070.60) 104 tons of sediment is deposited annually in this layer. In total, an average of (9.074.4) 104 tons of Budget calculations for the shelf reveal that sediment are annually deposited in Cap de Creus upper (57720) 104 metric tons of sediment are accumulating canyon. each year on the northwestern shelf and (4.67 Comparing these values with the total sediment input to 1.8) 104 tons accumulate yearly on the southwestern shelf the GOL provides a better understanding for the impor- (Fig. 10, Table 1). Within the Cap de Creus Canyon, the tance of the western region. Because good discharge data ARTICLE IN PRESS A.L. DeGeest et al. / Continental Shelf Research 28 (2008) 2031–2047 2045 were not available for the western rivers, only the Rhone issue) found that sediment fluxes were an order of River sediment influx (7.4–10.1 106 tons/yr) was used to magnitude higher in Cap de Creus than any other canyon. estimate the total influx to the Gulf, assuming the Rhone Cap de Creus Canyon is ideally situated to move sediment makes up 94% of the total (Pont et al., 2002; Bourrin and off the continental shelf, despite its presence on a passive Durrieu de Madron, 2006). It is also important to reiterate margin during a sea-level highstand. that the sediment mass estimates for the canyon are based This margin has implications for the study of sediment on maximum apparent accumulation rates; therefore export from passive margins during sea-level highstands. budget values must be assumed to be upper-bound The assumption that sediments are primarily trapped on estimates. Based on the range of sediment input values broad shelves can be complicated by oceanographic to the GOL, this study indicates that 6.1–8.3% of the conditions facilitating the movement of sediment over Rhone input (5.7–7.8% of the total sediment influx) is specific pathways. As such, an understanding of these sequestered on the western portion of the GOL shelf each conditions also affects our interpretation of deposits year (Table 3). This verifies that there is significant suspected of occurring during sea-level lowstands, which sediment transport to and accumulation in the western may have been influenced by a combination of factors portion of the GOL continental shelf, especially when allowing off-shelf sediment export during high stands in considering the large distance from the primary sediment sea level. source. This study also indicates that 0.9–1.2% of the Rhone 6. Conclusions input (0.8–1.1% of the total sediment input) is sequestered in upper Cap de Creus Canyon each year. This is an (1) Accumulation rates and grain-size patterns in the average over a 100-year timescale and does not account for western GOL reveal distinct depocenters on the shelf, shorter-term variability (Table 4). While these numbers similar to those defined in previous research. Surficial seem small, it is important to note that they only account grain sizes show a sharp transition from mud at mid- for what is currently deposited within the thalweg and water depths (30–85 m) to sand near the slope and northern flank, and do not include sediment that passes canyon head. Primary deposition on the shelf occurs at rapidly through the canyon (e.g. from periodic flushing by mid-shelf depths on the MSMD; however, these data dense-water cascading). Based on the high SSCs recorded show a discontinuity in the MSMD, with two primary by mooring instruments (Puig et al., this issue) and depocenters separated by a zone of bypassing. De- modeling results (Ulses et al., this issue, Ferre´et al., this creased accumulation in the bypassing zone is appar- issue), it is suggested that much more sediment (few ently due to acceleration around a headland, pointing 106 tons based on model results) actually passes through to the importance of the interaction between regional the canyon annually, but is not included in the sediment flow and coastal morphology in longer-term sediment budget for the upper canyon. Studies by Drexler et al. accumulation. (2008) at locations deeper in the canyon help to constrain (2) Within Cap de Creus Canyon, distinct grain size and the total amount of sediment moving through Cap de accumulation patterns clearly show localized zones in Creus Canyon. the canyon that experience different transport processes and strata formation. Accumulation rates are asym- 5.5. Implications for sediment export during sea-level metric across the canyon, with higher accumulation highstands rates on the northern flank compared to little or no accumulation on the southern flank. Grain-size dis- Cap de Creus Canyon appears to be an important tributions have even more dramatic asymmetry, with conduit for the transport of sediment to the deeper canyon finer grain sizes on northern flanks and adjacent shelf and slope based on seabed data from this study and water- compared to gravel and sand to the south. A sharp column data from Puig et al. (this issue). Little sediment is transition in surficial grain size also occurs in the main accumulating on 100-year timescales within the upper thalweg, with sandy deposits unconformably overlying canyon, but significant amounts of sediment appear to be consolidated mud in the upper canyon (o300 m), and passing through the canyon due to oceanographic (dense- unconsolidated muddy layers unconformably overlying water cascading, current interactions with coastal mor- sand in the mid canyon (400–700 m). 210Pb data from phology/bathymetry) and geologic (narrow shelf, canyon the mud layer indicate non-steady state accumulation incision) conditions. Seabed and mooring data indicate over decadal timescales, which suggests rapid deposi- scouring/sediment bypassing on the southern flank and tion and periodic flushing. Core data show that part of periodic flushing of an ephemeral mud layer in the thalweg, the canyon is acting as a long-term depocenter (north- likely the result of sediment-laden cascading events (Figs. 6 ern flank), whereas other parts appear to be more and 9)(Puig et al., this issue). These data support previous ephemeral (thalweg muds). studies that clearly showed high sediment fluxes through (3) These seabed data, along with time-series data within Cap de Creus Canyon, relative to the eastern canyons. the canyon, suggest the following off-shelf transport Specifically, Palanques et al. (2006) and Ulses et al. (this mechanisms: (a) sediment is preferentially supplied to ARTICLE IN PRESS 2046 A.L. DeGeest et al. / Continental Shelf Research 28 (2008) 2031–2047

the canyon from the shelf/slope adjacent to northern Canals, M., Puig, P., Durrieu de Madron, X., Heussner, S., Palanques, A., flanks via nepheloid transport by the regional current, Fabre` s, J., 2006. Flushing submarine canyons. Nature 444, 354–357. and (b) strong currents are scouring fines from the Carson, B., Baker, E.T., Hickey, B.M., Nittrouer, C.A., DeMaster, D.J., southern canyon flanks due to swift dense-water Thorbjarnarson, K.W., Snyder, G.W., 1986. Modern sediment dispersal and accumulation in Quinault submarine canyon—a cascading from the southern canyon head, and this summary. Marine Geology 71, 1–13. process carries sediment to deeper parts of the canyon. Certain, R., Tessier, B., Barusseau, J.P., Courp, T., Pauc, H., 2005. (4) Despite the significant distance from the primary fluvial Sedimentary balance and sand stock availability along a littoral source and the passive margin, there is sediment system. The case of the western Gulf of Lions littoral prism (France) escaping from the shelf via Cap de Creus Canyon. investigated by very high resolution seismic. Marine and Petroleum Geology 22, 889–900. However, the volumes are difficult to quantify due to Courp, T., Monaco, A., 1990. Sediment dispersal and accumulation on the the suspected volume of sediment bypassing through of the Gulf of Lions: sedimentary budget. the canyon and lack of information on sediment input Continental Shelf Research 10, 1063–1087. entering the GOL from the smaller rivers. It is Davies, P.A., Dakin, J.M., Flaconer, R.A., 1995. Eddy formation behind estimated that 1% of the Rhone River sediment a coastal headland. Journal of Coastal Research 11, 154–167. budget is accumulating in the upper canyon over 100- Drexler, T.M., Nittrouer, C.A., Ogston, A.S., Mullenbach, B.L., DeGeest, A.L., 2008. Off-shelf export from the Gulf of Lions continental shelf: year timescales. roles of Lacaze-Duthiers and Cap de Creus Canyons in the Gulf of Lions sediment dispersal system. In: Proceedings of AGU Ocean Overall, these results suggest that regional geology and Sciences Meeting. oceanographic setting must both be considered when Durrieu de Madron, X., 1994. Hydrography and nepheloid structures in the Grand-Rhone canyon. Continental Shelf Research 14, 457–477. determining off-shelf sedimentation patterns. Durrieu de Madron, X., Panouse, M., 1996. Transport de matie` re en suspension sur le plateau continental du Golfe du Lion-Situation estivale et hivernale. Comptes Rendus de l’Acade´mie des Sciences, Acknowledgments Se´rie IIa, Paris 322, 1061–1070. Durrieu de Madron, X., Nyffeler, F., Godet, C.H., 1990. Hydrographic We would like to thank the Office of Naval Research for structure and nepheloid spatial distribution in the Gulf of Lions funding this work (award numbers N00014-04-1-0082, continental margin. Continental Shelf Research 10, 915–929. N00014-99-1-0028 and N00014-04-1-0379). We also thank Durrieu de Madron, X., Radakovitch, O., Heussner, S., Loye-Pilot, M.D., the EuroSTRATAFORM participants for their help in the Monaco, A., 1999. Role of the climatological and current variability on shelf-slope exchanges of particulate matter. Evidence from the field and for their insightful discussions throughout the Rhoˆne continental margin (NW Mediterranean). Deep-Sea Research I course of the program that have greatly contributed to this 46, 1513–1538. research. We also thank the anonymous reviewers and Durrieu de Madron, X., Zervakis, V., Theocharis, A., Georgopoulos, D., editor for suggestions that significantly improved the final 2005. Comments on ‘‘Cascades of dense water around the world version of this paper. ocean’’. Progress in Oceanography 64, 83–90. Ercilla, G., Diaz, J.I., Alonso, B., Farran, M., 1995. Late Pleistocene- Holocene sedimentary evolution of the northern continental References shelf (northwestern ). Continental Shelf Research 15, 1435–1451. ´ ´ Arnau, P., Liquete, C., Canals, M., 2004. River mouth plume events and Ferre, B., Guizien, K., Durrieu de Madron, X., Palanques, A., Guillen, J., ´ their dispersal in the Northwestern Mediterranean Sea. Oceanography Gremare, A., 2005. Fine sediment dynamics study during a winter 17, 22–31. storm in the Gulf of Lion shelf (NW Mediterranean). Continental Shelf Research 25, 2410–2427. Baker, E.T., Hickey, B.M., 1986. Contemporary sedimentation processes in and around an active West Coast submarine canyon. Marine Ferre´, B., Durrieu de Madron, X., Estournel, C., Ulses, C., Le Corre, G. Geology 71, 15–34. Impact of natural (waves and currents) and anthropogenic (trawl) Berne, S., Aloisi, J.C., Baztan, J., Dennielou, B., Droz, L., Dos Reis, T., resuspension on the export of particulate matter to the open ocean. Lofi, J., Mear, Y., Rabineau, M., Satra, C., 2002. Notice de la carte Application to the Gulf of Lion (NW Mediterranean). Continental morpho-bathymetrique du Gulf du Lion. IFREMER et Region Shelf Research, this issue, doi:10.1016/j.csr.2008.02.002. Langeudoc Roussillon, Brest. Frignani, M., Courp, T., Cochran, J.K., Hirshberg, D., Vitoria Codina, Berne, S., Rabineau, M., Flores, J.A., Sierro, F.J., 2004. The impact of L., 2002. Scavenging rates and particle characteristics in and near the Quaternary global changes on strata formation: exploration of the Lacaze-Duthiers submarine canyon, northwest Mediterranean. Con- shelf edge in the northwest Mediterranean Sea. Oceanography 17, tinental Shelf Research 22, 2175–2190. 92–103. Gardner, W.D., 1989. Baltimore canyon as a modern conduit of sediment Bourrin, F., Durrieu de Madron, X., 2006. Contribution to the study of to the deep sea. Deep-Sea Research 36, 323–358. coastal rivers and associated prodeltas to sediment supply in Gulf of Gaudin, M., Berne´, S., Jouanneau, J-M., Palanques, A., Puig, P., Mulder, Lions (NW Mediterranean Sea). Vie et Milieu—Life and Environment T., Cirac, P., Rabineau, M., Imbert, P., 2006. Massive sand beds 56 (4), 307–314. attributed to deposition by dense water cascades in the Bourcart Bourrin, F., Durrieu de Madron, X., Estournel, C., Heussner, S. Impact of canyon head, Gulf of Lions (northwestern Mediterranean Sea). winter dense water formation on shelf sediment erosion (evidence from Marine Geology 234, 111–128. the Gulf of Lions, NW Mediterranean). Continental Shelf Research, Geyer, W.R., 1993. Three-dimensional tidal flow around headlands. this issue. Journal of Geophysical Research 98, 955–966. Canals, M., Casamor, J.L., Lastras, G., Monaco, A., Acosta, J., Berne, S., Geyer, W.R., Signell, R., 1990. Measurements of tidal flow around a Loubieu, B., Weaver, P.P.E., Grehan, A., Dennielou, B., 2004. The headland with a shipboard acoustic doppler current profiler. Journal role of canyons in strata formation. Oceanography 17, 80–91. of Geophysical Research 95, 3189–3197. ARTICLE IN PRESS A.L. DeGeest et al. / Continental Shelf Research 28 (2008) 2031–2047 2047

Got, H., Aloisi, J.C., 1990. The Holocene sedimentation on the Gulf of application to the Washington continental shelf. Marine Geology 31, Lions margin: a quantitative approach. Continental Shelf Research 10, 297–316. 841–855. Palanques, A., Durrieu de Madron, X., Puig, P., Fabres, J., Guille´n, J., Guille´n, J., Bourrin, F., Palanques, A., Durrieu de Madron, X., Puig, P., Calafat, A., Canals, M., Heussner, S., Bonnin, J., 2006. Suspended Buscail, R., 2006. Sediment dynamics during ‘‘wet’’ and ‘‘dry’’ storm sediment fluxes and transport processes in the Gulf of Lions submarine events on the Teˆt inner shelf (SW Gulf of Lions). Marine Geology 234, canyons-The role of storms and dense water cascading. Marine 129–142. Geology 234, 43–61. Heussner, S., Durrieu de Madron, X., Calafat, A., Canals, M., Carbonne, Palaques, A., Puig, P., Guillen, J., Jimenez, J.A., Garcia, V., Sanchez- J., Delsaut, N., Saragoni, G., 2006. Spatial and temporal variability of Archilla, A., Madsen, O.S., 2002. Near-bottom suspended sediment downward particle fluxes on a continental slope: lessons from an 8-yr fluxes on the microtidal low-energy Ebro continental shelf (NW experiment in the Gulf of Lions (NW Mediterranean). Marine Mediterranean). Continental Shelf Research 22, 285–303. Geology 234, 63–92. Paull, C.K., Ussler III, W., Greene, H.G., Keaton, R., Mitts, P., Barry, J., Hickey, B., Baker, E., Kachel, N., 1986. Suspended particle movement in 2003. Caught in the act: the 20 December 2001 gravity flow event in and around Quinault submarine canyon. Marine Geology 71, 35–83. Monterey Canyon. Geo-Marine Letters 22, 227–232. Jago, C.F., Barusseau, J.P., 1981. Sediment entrainment on a wave-graded Pawlak, G., MacCready, P., 2002. Oscillatory flow across an irregular shelf, Roussillon, France. Marine Geology 42, 279–299. boundary. Journal of Geophysical Research 107, 1–17. Kineke, G.C., Woolfe, K.J., Kuehl, S.A., Milliman, J.D., Dellapenna, Pont, D., Simonet, J.P., Walter, A.V., 2002. Medium-term changes in T.M., Purdon, R.G., 2000. Sediment export from the Sepik River, suspended sediment delivery to the ocean: consequences of catchment Papua New Guinea: evidence for a divergent sediment plume. heterogeneity and river management (Grand Rhoˆne river, France). Continental Shelf Research 20, 2239–2266. Estuarine, Coastal, and Shelf Science 54, 1–18. Puig, P., Ogston, A.S., Mullenbach, B.L., Nittrouer, C.A., Sternberg, Martin, R.T., Gadel, F.Y., Barusseau, J.P., 1981. Holocene evolution of the Canet-St. Nazaire lagoon (Golfe du Lion, France) as determined R.W., 2003. Shelf-to-canyon sediment-transport processes on the Eel from a study of sediment properties. Sedimentology 28, 823–836. continental margin (northern California). Marine Geology 193, 129–149. May, J.A., Warme, J.E., Slater, R.A., 1983. Role of submarine canyons on Puig, P., Palanques, A., Orange, D.L., Lastras, G., Canals, M. Dense shelf shelfbreak erosion and sedimentation: modern and ancient examples. water cascades and sedimentary furrow formation in the Cap de Creus In: Stanley, D.J., Moore, G.T. (Eds.), The Shelfbreak: Critical Canyon, northwestern Mediterranean Sea, this issue. Interface on Continental Margins. SEPM, Tulsa, OK, pp. 315–332 Shepard, F.P., Dill, R.F., 1966. Submarine Canyons and Other Sea Special Publication No. 33. Valleys. Rand McNally, Chicago, IL, 381pp. Milliman, J.D., Meade, R.H., 1983. World-wide delivery of river sediment Signell, R.P., Harris, C.K., 2000. Modeling sand bank formation around to the . The Journal of Geology 91, 1–21. tidal headlands. In: Spaulding, M.L., Blumberg, A.F. (Eds.), Estuarine Millot, C., 1990. The Gulf of Lions hydrodynamics. Continental Shelf and Coastal Modeling, 6th Int. Conf., ASCE, New Orleans, November Research 10, 885–894. 210 3–5, 1999. Miralles, J., Radakovitch, O., Aloisi, J.-C., 2005. Pb sedimentation Snyder, G.W., Carson, B., 1986. Bottom and suspended particle sizes: rates from the Northwestern Mediterranean margin. Marine Geology implications for modern sediment transport in Quinault Submarine 216, 155–167. Canyon. Marine Geology 71, 85–105. Mullenbach, B.L., 2002. Characterization of modern off-shelf sediment Ulses, C., Estournel, C., Durrieu de Madron, X., Palanques, A. Suspended export on the Eel margin, northern California. Ph.D. Dissertation, sediment transport in the Gulf of Lion (NW Mediterranean): Impact University of Washington, WA, 154pp. of extreme storms and floods. Continental Shelf Research, this issue, Mullenbach, B.L., Nittrouer, C.A., Puig, P., Orange, D.L., 2004. Sediment doi:10.1016/j.csr.2008.01.015. deposition in a modern submarine canyon: Eel Canyon, northern Ulses, C., Estournel, C., Bonnin, J., Durrieu de Madron, X., Marsaleix, California. Marine Geology 211, 101–119. P., 2008. Impact of storms and dense water cascading on shelf-slope Nittrouer, C.A., Sternberg, R.W., 1981. The formation of sedimentary exchanges in the Gulf of Lions (NW Mediterranean). Journal of strata in an allocthonous shelf environment—the Washington Con- Geophysical Research 113, C02010, doi:10.1029/2006JC003795. tinental Shelf. Marine Geology 42, 201–232. Walsh, J.P., Nittrouer, C.A., 2003. Contrasting styles of off-shelf Nittrouer, C.A., Wright, L.D., 1994. Transport of particles across accumulation in New Guinea. Marine Geology 196, 105–125. continental shelves. Reviews of Geophysics 32, 85–113. Zuo, Z., Eisma, D., Gieles, R., Beks, J., 1997. Accumulation rates and Nittrouer, C.A., Sternberg, R.W., Carpenter, R., Bennett, J.T., 1979. sediment deposition in the northwestern Mediterranean. Deep-Sea The use of 210Pb geochronology as a sedimentological tool: Research II 44, 597–609.