In situ measurements of advective solute transport in permeable shelf sands

ARTICLE IN PRESS

Continental Shelf Research 24 (2004) 183–201

In situ measurements of advective solute transport in permeable shelf sands Clare E. Reimersa,*, Hilmar A. Stecher IIIa, Gary L. Taghonb, Charlotte M. Fullerb, Markus Huettelc,1, Antje Ruschc,2, Natacha Ryckelyncka, Christian Wildc a

College of Oceanic and Atmospheric Sciences, Oregon State University, 104 Ocean Admin. Bldg., Corvallis, OR 97331, USA b Institute of Marine and Coastal Sciences, Rutgers University, New Brunswick, NJ 08901, USA c Max Planck Institute for Marine Microbiology, Celsiusstr. 1, Bremen D-28359, Germany Received 14 March 2003; received in revised form 13 June 2003; accepted 16 October 2003

Abstract Solute transport rates within the uppermost 2 cm of a rippled continental shelf sand deposit, with a mean grain size of 400–500 mm and permeabilities of 2.0–2.4
1011 m2, have been measured in situ by detecting the breakthrough of a pulse of iodide after its injection into the bottom water. These tracer experiments were conducted on the USA Middle Atlantic Bight shelf at a water depth of B13 m using a small tethered tripod that carried a close-up video camera, acoustic current meter, motorized 1.5 liter ‘‘syringe’’, and a microprofiling system for positioning and operating a solidstate voltammetric microelectrode. When triggered on shipboard, the syringe delivered a 0.21 M solution of potassium iodide and red dye through five nozzles positioned around and above the buried tip of the voltammetric sensor for 0.65– 5 min. Bottom turbulence rapidly mixed and dispersed the tracer, which then was carried into the bed by interfacial water flows associated with ripple topography. The advective downward transport to the sensor tip was timed by a sequence of repetitive voltammetric scans. The distance-averaged vertical velocity, expressed as the depth of the sensor tip in the sand divided by the time to iodide breakthrough, was found to vary from 6 to 53 cm h1 and generally to decrease with sediment depth. Because of episodic pumping and dispersion associated with the greatest 5% of wave heights and current speeds recorded, some concentration vs. time responses showed evidence of uneven solute migration. For reasons of mass balance, the advective flow field in the surface layers of permeable beds includes regions of water intrusion, horizontal pore-water flow and upwelling which also may explain some of the observed uneven migration. Pore-water advection was also evident in oxygen profiles measured before and after tracer injection with the voltammetric sensor. These profiles showed irregular distributions and oxygen penetration depths of 4–4.5 cm. Sand cores from the study site subjected to continuous pore fluid pumping showed that oxygen consumption was positively correlated with flow rate. The effect was calculated to be equivalent to increasing the benthic oxygen flux by 0.029 mmol m2 d1 for every 1 liter m2 d1 flushed through a 4 cm thick oxic zone. Thus, it is concluded that in situ oxygen consumption rates must be highly variable and dependent on the prevalent wave and current conditions. r 2003 Elsevier Ltd. All rights reserved. Keywords: Advection; Voltammetric electrode; Permeable sediments; Sand ripples; Oxygen consumption; Inner shelf

*Corresponding author. Fax: +1-541-737-2064. E-mail address: [email protected] (C.E. Reimers). 1 Present address: Department of Oceanography, Florida State University, Tallahassee, FL 32306-4320, USA. 2 Present address: Department of Earth and Planetary Science, Washington University, St. Louis, MO 63130, USA. 0278-4343/$ - see front matter r 2003 Elsevier Ltd. All rights reserved. doi:10.1016/j.csr.2003.10.005

ARTICLE IN PRESS 184

C.E. Reimers et al. / Continental Shelf Research 24 (2004) 183–201

1. Introduction Surface gravity waves and ensuing boundary layer currents can influence structure, composition and biogeochemical reactions in coastal and shelf sediments (Komar et al., 1972; Vanderborght et al., 1977; Malan and McLachlan, 1991; Styles and Glenn, 2002). When sediments are permeable and uneven, as is common for the sand beds covering large areas of the continental shelf, these flows trap and resuspend fresh particulate organic matter, winnow other fines, induce interstitial fluid advection, and create oscillatory ripple marks (Webster and Taylor, 1992; Lohse et al., 1996; Huettel et al., 1996; Nelson et al., 1999). Unfortunately, due to the difficulty of observation under wave action, there are few field studies that report quantitative measures of these phenomena or their consequences. This has led to a poor understanding of the role permeable sediments play in biogeochemical cycles and shelf carbon budgets (Rowe et al., 1988; Shum and Sundby, 1996; Boudreau et al., 2001). One of the most pioneering studies of sediments under wave action used the tracer fluorescein to follow interstitial water flow (Webb and Theodor, 1968). Divers injected 2 ml samples of seawater mixed with dye into sand ripples at depths of 2.5– 10 cm and then timed the rapid upward reemergence of the dye. It was emphasized later that the observed flow pattern could not have been caused by wave effects alone, but must have resulted from the interaction of the bottom swell and sediment ripples (Rutgers van der Loeff, 1981). Seawater anions also have made useful tracers for evaluating rates of solute transport through sediments. Most notably, Br has been used to estimate sediment irrigation rates resulting from the burrowing and feeding activities of sediment infauna (Martin and Banta, 1992). On a mathematical basis, mixing by waves and currents or mixing by benthic animals can be treated the same. For example, in one-dimensional models of biological irrigation rates, it is common to specify a non-local exchange function, aðxÞ; which has units of inverse time (Emerson et al., 1984; Martin and Sayles, 1987; Marinelli et al., 1998). Boudreau (1997, p. 143) suggests that the

integral of aðxÞ over the depth of exchange can be expressed as an average exchange velocity (i.e., pore fluid velocity). Depth distributions of average exchange velocities resulting from pressure variations associated with horizontal currents interacting with sand mounds in a flume were recently measured by Huettel et al. (1996). Mass balance requires that advection representing a net solute influx at a given location is balanced by a change in concentration with time, chemical reaction, or an advective or dispersive efflux. Dispersion is the type of fluid mixing within sediments that is usually modeled as an effective diffusion process associated with hydraulic flows (Boudreau, 1997). Empirically (after Lee, 1999), the hydrodynamic dispersion coefficient Dx in the x-direction may be related to the pore fluid velocity, nx ; by D x ¼ bx

nx þ Ds ; f

ð1Þ

where Ds is the sediment molecular diffusion coefficient in terms of area of sediment per unit time (Berner, 1980), f is the porosity, and bx is the dispersivity, a scale-dependent parameter with units of length.3 Under a rippled bed subject to irregular waves, flow paths and fluid velocities become time dependent, but wave tank experiments have confirmed that fluid intrusion occurs predominantly beneath ripple troughs and flanks, while pore-water extrusion is focused at ripple crests (Precht and Huettel, 2003). The same flow pattern was predicted by Shum (1992, 1993) using an analytical model for two-dimensional flow fields within ripples under oscillatory wave motion. Again, as was recognized by Rutgers van der Loeff (1981), the main driving force for such filtration systems is the pressure field generated by interaction between wave-driven bottom currents and sand ripples. 3

Some representations of dispersion describe Dx as an effective exchange coefficient equal to DD þ Ds ; where DD is described as the dispersion coefficient. For example, Svensson and Rahm (1991) separate molecular diffusion from dispersion and assume DD ¼ 0:05 dv; where d is the particle diameter of the sediment considered and v% is the magnitude of the velocity vector.

ARTICLE IN PRESS C.E. Reimers et al. / Continental Shelf Research 24 (2004) 183–201

The emphasis of this work was primarily to measure distance-averaged intrusion velocities in rippled sand deposits on the inner continental shelf of the USA Middle Atlantic Bight (MAB) under in situ conditions to confirm both modeling and experimental studies. A method was developed in which pulses of dissolved iodide were released to the sediment–water interface and sensed at discrete depths in the sediment with voltammetric electrodes. A video camera system, acoustic Doppler current meter, and sediment cores yielded sitespecific information on actively forming ripples, wave and current parameters, sediment properties, and oxidation rates of sediment organic matter as a function of flow. Aspects of the inner continental shelf of the MAB that divide it from outer shelf regions are its water depth (generallyo30 m), higher tidal and wave energy, proximity to fluvial sources, and turbidity. The inner shelf may also exhibit higher annual rates of primary productivity and more variable CO2 fluxes to the atmosphere compared to the outer shelf (Boehme et al., 1998; DeGrandpre et al., 2002). The latter is believed to be in part because very high rates of heterotrophic CO2 metabolism in both the water column and sediment can exceed even the rich production of organic matter near shore. There are many problems associated with applying standard methods for constraining rates of benthic metabolism to shelf sands, however, so we report rates determined by a method tailored to permeable substrates. In this way we have begun to set limits for the influences of advective pore-water exchange on rates of organic matter turnover in the shallow regions of the continental shelf.

2. Methods 2.1. Experimental design ‘‘Breakthrough’’ times of pulses of dissolved iodide at discrete depths in the sediment were measured. Although we have found ‘‘breakthrough time’’ defined in the literature as either the time of first arrival of a tracer (Fritsche et al., 2001) or as when a ‘‘chosen isopleth (line of given concentration) reaches a given location’’ (Lee,

185

1999), we define it as when the concentration maximum of a pulsed input reaches a given location. By this definition a distance-averaged linear velocity can be described as the migrating rate of the leading edge of the peak of a pulsed input, which is expected to decay and spread with distance because of dispersion (Boudreau, 1997). A tracer was prepared by adding 1 l of deionized water (colored with 1 ml of concentrated red food coloring for visualization) to 35.3 g KI to make a 0.21 M iodide solution equal in density to seawater. This solution was loaded in a custom-made, motor-driven syringe (Eastern Oceanics) capable of delivering up to 1.5 l through a manifold made of 0.32 cm outer-diameter (OD), Perfluoroalkoxy (PFA) tubing and stainless steel fittings. The injection rate of tracer was adjustable but set at 270 ml min1 (Bone quarter the maximum rate). Tracer was delivered from the PFA tubing to five 0.32 cm OD stainless steel tubes and then through diffusers (made by drilling several 0.04 cm OD holes in the caps shipped with disposable syringes which were affixed to the tips of the stainless steel tubes using silicone adhesive). The tubes and diffusers were spaced radially 4 cm from a centered voltammetric sensor and 60 apart (Fig. 1). The plane of the diffuser tips was placed from 1 to 4 cm above the tip of the voltammetric electrode depending on the anticipated bottom relief and the final measuring depth of the electrode (0.4– 2 cm). The iodide delivery array and voltammetric sensor (Fig. 1) were positioned relative to the seafloor using a motorized, vertical ball slide assembly (Eastern Oceanics, custom design) with a total coarse-scale travel of 20 cm, and a microprofiler unit with a total vertical travel of 8 cm at 0.25 mm resolution while viewed in live color video (Deep-Sea Power & Light, MultiSeaCam 2050). The microprofiler was mounted on the slide assembly, and the slide assembly was fixed to a benthic tripod (1.6 m pod-to-pod, 1.4 m high). The tripod also carried an acoustic current meter (Nortek, Aquadopp 3D-Vector averaging current meter with pressure and temperature sensors), the custom-made syringe, the video camera (positioned to view the sediment–water interface at ca. 35 ), and three underwater lights

ARTICLE IN PRESS 186

C.E. Reimers et al. / Continental Shelf Research 24 (2004) 183–201

During each seafloor experiment, the tripod was tethered to the Research Vessel (R./V.) Cape Henlopen while the vessel was on three-point anchor over the study site (39 27.01 N; 74 14.27 W; 12–13 m water depth). Eight experiments were conducted between July 24 and July 26, 2001 after a day of preliminary trials. 2.2. Study site

(A)

(B) Fig. 1. (A) Artist’s portrayal of the array for emitting tracer above the sediment–water interface around a centered voltammetric electrode with buried tip. The tracer streams flow, pool, and shift direction under oscillatory wave action. (B) Video frame grab of tracer release during D16-2.0.

The study site was located in the inner shelf region of the USA MAB near the cabled observatory known as LEO-15 (Glenn et al., 2000). Boundary layer studies have documented the seasonal variation of benthic flow conditions in this region as well as the significance of energetic wave events for initiating and maintaining sediment transport (Styles, 1998). Near bottom wave orbital velocities calculated from long-term records of wave pressure spectra range from 5 cm s1 to over 70 cm s1 but rarely exceed 30 cm s1 in summer months (Styles, 1998). During summer conditions an equilibrium ripple field is the most characteristic bedform with reported ripple heights of 3–15 cm and wavelengths of 20–100 cm (Traykovski et al., 1999; Styles and Glenn, 2002). The rippled sediments are well-sorted, dominantly quartz, medium to coarse sands with a median grain size ranging from 400 to 500 mm and an average bulk organic carbon content of 0.02% dry weight. 2.3. Sediment sampling and analyses

(Deep-Sea Power & Light, Micro-SeaLite). In addition, the tripod carried the microprofilercontroller electronics bottle (Reimers and Glud, 2000), an interface bottle for the video, syringe and slide (connected by cable to a surface unit with controlling switches and video input/output; Eastern Oceanics), and a transmitter/cable interface (Analytical Instrument Systems, Model DLKLCT-1) that connected the voltammetric working, counter and reference electrodes through a 30 m waterproof cable to a signal receiver (AIS, ModelLCR-1) and electrochemical analyzer (AIS, DLK100A) (Luther et al., 1999).

Cylindrical cores of surface sediment were collected by SCUBA divers from both ripple crests and troughs on July 24, 2001 at approximately 11 AM. The divers worked along the bottom approximately 50 m east of the site targeted for the tripod measurements. Different sized cores were collected for measurements of sediment permeability and porosity, percent fines by weight, oxygen consumption rates under varying flow rates, and percent organic C and total N; pigments and bacterial activity (the latter are presented in Rusch et al., 2003). Cores in 2.65 cm (ID) tubes were connected to a falling-head permeameter to assess sediment

ARTICLE IN PRESS C.E. Reimers et al. / Continental Shelf Research 24 (2004) 183–201

permeability (Gray, 1958). Afterwards, the length of the sediment column was measured to the nearest 0.1 cm, and the sediment solids were rinsed with deionized water and dried to constant weight. Porosity was calculated as: 1[(sample dry weight)/ (sample totalwet volume x particle density)]. A particle density of 2.65 g cm3 (quartz) was used. Percent fines by weight were determined from 1 cm depth intervals of 2.65 cm diameter cores based on the assumption that particle settlingvelocity followed Stoke’s Law. After sectioning a core, a small aliquot of wet sediment from each depth interval was placed in a test tube, then covered to a depth of 1.5 cm with filtered (Whatman GF/F) seawater. The sample was mixed vigorously, allowed to stand undisturbed for 5 s, then the overlying water and suspended particles were removed with a pipet. Under these conditions, all particles of quartz-density p60 mm remain suspended. The washing steps were repeated until the water was visibly clear, typically a

187

total of five times. All washings were combined and filtered through a pre-weighed Whatman GF/ C filter. The filter was rinsed briefly with deionized water then dried to constant weight. Sand-sized particles (> 60 mm) remaining in the test tube were also dried to constant weight.

2.4. Oxygen consumption rates Oxygen consumption rates per area of seafloor were determined as a function of flow rate through eight separate cores (3.6 cm diameter, 6.1–10.5 cm long) at different times after retrieval and laboratory setup (Table 1). Immediately after recovery the cores (henceforth called columns) were clamped and mounted vertically on a stand in a water bath held at 21.470.2 C (close to the temperature of the bottom water). The water volume overlying each core was reduced to less than 5 ml by displacing excess water with pierced

Table 1 Column experiment conditions Column number

Sediment column length (cm)

Mean flow rate (ml cm2 h1)a

Time since retrieval and setup under flow (h)b

1 1 1 2 2 2 3 3 3 4c 5 5 5 6 6 6 7 7 7 8c

6.1 6.1 6.1 9.0 9.0 9.0 7.3 7.3 7.3 10.0 10.5 10.5 10.5 8.0 8.0 8.0 8.5 8.5 8.5 9.5

5.33 2.67 2.73 5.56 2.78 2.67 5.56 2.78 2.75 2.37 4.00 1.33 1.20 4.00 1.33 1.30 4.00 1.22 1.15 1.23

3.25–4.75 5.0–6.5 20.5–26.0 3.25–4.75 5.0–6.5 20.5–26.0 3.25–4.75 5.0–6.5 20.5–26.0 21.1–26.0 3.25–4.75 5.0–6.5 21.1–26.5 3.25–4.75 5.0–6.5 21.1–26.5 3.25–4.75 5.0–6.5 21.1–26.5 21.8–26.5

To convert ml cm2 h1 to l m2 d1 multiply by the factor 240. These are the periods during which in- and out-flowing water were monitored for determining oxygen consumption. c Organic substrates were added to the in-flowing waters of columns 4 and 8 during the initial flow periods. The uptake of these substrates was monitored and will be reported in a later publication. a

b

ARTICLE IN PRESS 188

C.E. Reimers et al. / Continental Shelf Research 24 (2004) 183–201

inert StyrodurTM stoppers. Light was excluded by wrapping all columns with aluminum foil. The pierced rubber stoppers at the inflow of each column were connected with 50 cm long pieces of 1.02 mm ID TygonTM tubing to an aerated reservoir of filtered (0.2 mm) local seawater. The outflow of each column was connected with TygonTM tubing of the same dimensions via a peristaltic pump to 50 ml glass syringes that collected the water for later analyses. With this arrangement, seawater that was initially airsaturated was pumped continuously through the sediment columns at pre-set rates ranging from 1.20 to 5.56 ml cm2 h1 (288–1334 l m2 d1). Oxygen concentrations in the in- and outflowing water were monitored using separate flow-through cells fitted with integrated fiber-optic oxygen microsensors (PreSenss, Sensor type B2, tip diameter o50 mm) that were mounted via Luer Lock connectors into the tubing at the upper and lower end of each column. The microsensors were read after taking into account the time necessary for water to percolate the entire column length by a fiber-optic oxygen-meter (Microx-TX, PreSenss) that was connected to a standard PC for signal processing and data storage. Outflow water never became anoxic, and the sensors were calibrated before and after each experiment using a two-point calibration in oxygen-free (addition of sodium dithionite) and air-saturated seawater (airbubbled) from the inflow reservoir. Additionally, oxygen concentrations in the reservoir were measured using Winkler titration (Grasshoff, 1999). 2.5. Water column measurements Seawater samples from the study site were collected once or twice daily within 1 m of the seafloor and within 2 m of the sea-surface together with CTD measurements using a Seabird 911 Plus CTD-rosette system outfitted with 10 liter General Oceanic 1015 bottles. Although these samples were analyzed for a suite of parameters, only the bottom water dissolved oxygen data will be reported in this paper because these results are used for calibrating the pore-water oxygen measurements determined by voltammetry. These

oxygen analyses were performed on duplicate samples from each 10 l bottle using an automated Winkler titration system and procedures developed by Friederich et al. (1991). Six vertical profiles of upwelling radiance and downwelling irradiance were also measured to within 2 m of the bottom between 08:50 and 14:30 during the 4 days on station with a Biospherical Profiling Reflectance Refractometer (PRR-600). 2.6. Voltammetric methods Amalgamated Au electrodes were used to detect dissolved oxygen and the breakthrough of iodide in sands near the sediment–water boundary. These sensors were made according to methods developed by Brendel and Luther (1995) and Luther et al. (1999) by inserting 100-mm diameter Au wire into glass pipets pulled from 5-mm OD glass tubing to have 4–6 cm long tapers, with tip diameters of 0.3–0.5 mm. A non-conductive epoxy (West System 105) was used to fill the space between the Au and glass over the entire length (15–25 cm) of the pipet. The sensor tip was sanded against a rotating 1-cm2 piece of 400-grit SiC sandpaper to expose the Au wire as a core rimmed with solid epoxy and glass. Each tip was ground further with 1000- and 4000-grit SiC paper, then finely polished with a series of diamond-grit pastes and plated with Hg to create hemispheric sensing elements, 100 mm in diameter. In situ voltammetric analyses were carried out using the Analytical Instrument Systems, Inc. (AIS) DLK-100A electrochemical analyzer with long-cable transmitter/receiver interfaces, and single amalgamated Au electrodes in cells with a Ag/ AgCl reference and a Pt counter electrode. These analyses were controlled by a microcomputer aboard ship using software provided by the manufacturer (AIS). The analyzer and computer were run from separate DC power sources and grounded to the ship’s hull. When using linear sweep voltammetry, O2 gives two waves corresponding to the reduction of O2 to H2O2 and of H2O2 to H2O (Buffle and Tercier-Waeber, 2000). Oxygen concentrations were estimated from the peak (or half-wave) currents of the first reduction (ca. 0.3 V vs. Ag/AgCl) minus any non-zero

ARTICLE IN PRESS C.E. Reimers et al. / Continental Shelf Research 24 (2004) 183–201 80 Current (nA)

70 60 50 40 30 20 ip-oxygen b

10 0 -0.2

-0.4

(A)

-0.6 -0.8 -1 -1.2 Potential vs. Ag/AgCl (V)

-1.4

-1.6

Resultant Current (nA)

16 14 12 10

ih

8 6 4 2

ip-iodide

0 -0.1

(B)

-0.2

-0.3

-0.4

-0.5

-0.6

-0.7

-0.8

Potential vs. Ag/AgCl (V)

Fig. 2. (A) Representative linear sweep voltammogram measured in situ with the voltammetric sensor tip positioned at the sand–water interface. The potential scan was from 0.1 to 1.8 V vs. Ag/AgCl at 1000 mV s1 after 10-s conditioning at 0.1 V. The current peak to the left results from the reduction of dissolved O2 to peroxide at the electrode surface. Its height, ip-oxygen ; above baseline, b; is proportional to the concentration of dissolved oxygen under constant environmental conditions. (B) Square wave voltammogram showing the detection of dissolved iodide (top curve) and the resultant peak height, ip-iodide; after baseline subtraction. Values of ip-iodide increase linearly with iodide concentration. The measure ih is used at high concentration when peaks are too broad to observe the leading baseline.

baseline of scans run at 1000 mV s1 between 0.1 and 1.8 V vs. Ag/AgCl after a 10-s conditioning step at 0.1 V (Fig. 2A). It may be assumed that peak currents (ip-oxygen ) determined in this way are proportional to oxygen concentration, yielding ½O2 x ¼

ipx ½O2 BW ; ipx¼0

ð2Þ

where x represents a depth below the sediment– water interface (x ¼ 0) and [O2]BW is the dissolved oxygen concentration in the bottom water.

189

Above the sediment–water interface, the sensitivity of voltammetric electrodes of this size was observed to be influenced by the hydrodynamic flow such that scans were often irregular and poorly reproducible. The sediment–water interface and ðip-oxygen Þx¼0 were defined by approaching the sediment surface incrementally, observing the sensor by close-up video, and monitoring the form of linear sweep scans. However, this interface was not a truly fixed boundary. Instead it was observed to gain and lose up to 2 mm during an experiment as sand particles and shell fragments were moved to and fro by the forces of the water motion. Although iodide may also be detected by linear sweep voltammetry, we used a more sensitive square wave technique (0.4 mV scan increment, 200 mV s1, 15 mV pulse height) to detect the arrival of iodide tracer at discrete sand depths (Luther et al., 1998). Scans were programmed to run sequentially between 0.05 and 0.85 V (or in some cases 0.45 V) with 5 s of deposition time applied at 0.05 V at the beginning of each scan. Because iodide adsorbs as a surface film on Hg at X0.1 V, well-defined peaks in resultant current are produced due to the oxidative stripping of the adsorptive Hg complex (Luther et al., 1998). The heights of these peaks (ip-iodide ) were tabulated after subtraction of the baseline current as illustrated in Fig. 2B, or in the case of later control experiments simply as the maximum height of the primary peak above the cathodic baseline (ih ). For typical electrodes that are calibrated over an iodide concentration range of 2–25 mM in anoxic seawater at 20 C, ðip Þiodide (Fig. 2B) (or else the resolved peak area) is the preferred measure because it increases as a linear function of concentration with a calibration slope of approximately 1 nA mM1 (or 1 nC mM1), and such calibrations yield minimum detection limits of B1 mM, and zero intercepts. However, for iodide concentrations increasing beyond 25 mM, peaks become too broad to define the leading portion of the baseline. The measure (ih ) (Fig. 2B) continues to increase beyond concentrations of 300 mM but this trend is non-linear with a non-zero intercept created by the skewing effects of the chloride signal added to the baseline. Iodide calibrations run in the presence of dissolved O2

ARTICLE IN PRESS 190

C.E. Reimers et al. / Continental Shelf Research 24 (2004) 183–201

by either method of peak definition exhibit positive (non-zero) intercepts due to the overlap in the potential ranges that are electroactive for I and O2.

by stagnant seawater. Tracer was added to the overlying water with a pipet with very little imposed mixing. We later estimated a dispersion coefficient for iodide from the Einstein–Smoluchowski relation

2.7. Control experiments Dx ¼ To check whether tracer delivery or channeling along the electrode could contribute to the advection of iodide through sands, two types of control experiment were carried out. In the first, the tripod and all its equipment were positioned over a flat sand bed that was created from grab samples collected from the study site in November 2001. The sand bed was 6 cm deep, contained living benthic fauna, and was overlain by 9 cm of seawater (1072 C) in a 34 cm wide
67 cm long trough placed on the ship’s deck. The ship’s motion and wind caused some agitation of the water. Using the same procedures as during in situ experiments, the voltammetric sensor was lowered until the sediment–water interface was detected and then lowered until the tip was at 0.8 cm depth. As the diffuser tips of the iodide delivery array were set 2.5 cm above the electrode tip, their injection height was 1.7 cm above the sand. Without the benefit of bottom water exchange, the tracer pulse injected for 20 s during this control experiment, did not dissipate. Scans for detecting iodide at 0.8 cm depth were run in a continuous sequence for 4.75 h (or six times the length of the longest in situ experiment). Because no iodide was detected over this time, the voltammetric sensor was then raised to the overlying water and used to measure three consecutive depth profiles of dissolved iodide separated by horizontal distances of several centimeters and in steps of 0.5–2 mm. The third profile was completed 6.25 h after the tracer injection. The second type of control experiment was completed in our shore-based laboratory at 19.770.5 C. A voltammetric sensor was introduced with a micromanipulator into a bed of sand from the LEO-15 study site, contained within a 1 liter plastic jar. The arrival and then rate of change in iodide peak current were measured at one depth, 2.0 cm, when the sand bed was overlain

L2 2tD

ð3Þ

that relates the displacement of a concentration pulse to elapsed time (Boudreau, 1997). In this application, L was equated to the sediment thickness (or depth of the electrode tip), and tD to the time-lag before breakthrough that is approximated graphically as the zero concentration intercept of the linear portion of a concentration–time curve (Meier et al., 1988, 1991). The expectation was that if no pore-water advection was introduced by our methods, derived values of Dx should approximate the molecular diffusion coefficient for iodide in sediments (i.e., Ds of Eq. (1)) at the experimental temperatures. Gross deviation from diffusion behavior in either type of control experiment would be evidence of experimental artifacts.

3. Results 3.1. Seafloor properties and diffusion coefficients During the 4 days of this study the percent surface photosynthetically available radiation (PAR) reaching 10 m depth (or 2–3 m above bottom) averaged 0.370.1% (n ¼ 6). In this dim light, sand ripples were observed by video and divers to cover the seafloor and to be forming actively. Porosity, permeability and fine fraction data for both ripple crests and troughs are presented in Table 2. The surface intervals of sand from troughs were enriched in fine particles compared to deeper intervals and crests. However, the porosity and permeability determinations of whole cores (lengths: 6.8–8.8 cm) from either crests or troughs fell within a narrow range. Whole sediment diffusion coefficients for solutes may be estimated by dividing molecular diffusion coefficients by a tortuosity factor equal to the porosity times the sediment formation factor, F

ARTICLE IN PRESS C.E. Reimers et al. / Continental Shelf Research 24 (2004) 183–201

191

Table 2 Sediment properties Sample location in ripple field

Depth interval (cm)

N

Porosity

Permeability (m2)

Crest Trough Trough

0–6.8 0–8.0 0–8.8

1 1 1

0.38 0.37 0.37

2.4
1011 2.0
1011 2.3
1011

Crest Crest Crest Crest Crest Trough Trough Trough Trough Trough

0–1 1–2 2–3 3–4 4–5 0–1 1–2 2–3 3–4 4–5

3 3 3 3 3 3 3 3 3 3

0.06370.043 0.03770.003 0.04170.008 0.04470.006 0.05270.007 0.31070.073 0.11270.050 0.06270.018 0.05670.008 0.06770.014

(Berner, 1980): D : Ds ¼ fF

Weight % fines (o60 mm)

ð4Þ

We have made in situ measurements of electrical resistivity Rx in these sediments and calculated formation factors as F ¼ Rx =Rbottom water (McDuff and Ellis, 1979; Reimers et al., 2001). Throughout the uppermost centimeters of the sediment column F ¼ 323:5: Estimates for the molecular diffusion coefficient of I in seawater between 10 C and 20 C are from 1.4 to 1.8
105 cm2 s1 (Li and Gregory, 1974; Boudreau, 1997). Accordingly, during the experiments in this study Ds for iodide is estimated to have equaled approximately 1.1– 1.6
105 cm2 s1. 3.2. Control experiments No iodide was detected at 0.8 cm depth after 4.75 h of voltammetric monitoring during the first control experiment run with the complete tripod and iodide delivery system placed over an enclosed sand bed. Iodide concentration profiles measured with the same sensor over the next hour and a half (Fig. 3A), however, indicated tracer concentrations at 0.8 cm that were near the sensor’s detection limit and a concentration gradient

throughout the first centimeter of sand that is consistent with diffusive transport. The minimum time-lag for tracer arrival assuming only diffusive transport and Ds ¼ 1:1
105 cm2 s1 should have been 8 h according to Eq. (3). A larger Ds caused by a smaller than estimated tortuosity effect or a small degree of dispersion at the sediment–water interface could explain the slightly shorter time delay. Advective transport caused by tracer injection was not indicated. Similarly, the laboratory control experiment showed a pattern of tracer transport that may be predicted assuming vertical molecular diffusion. Sensor scans at 2.0 cm were monitored long enough to observe steadily increasing iodide concentrations (Fig. 3B). The tD value of this experiment (22.1 h) predicts a sediment dispersion coefficient (Dx ) of 2.5
105 cm2 s1. For our purposes this value is close enough to the predicted molecular diffusion coefficient to assert that introduction of a microelectrode into the sand does not provide a path that promotes tracer penetration. 3.3. In situ experiments The conditions that distinguish the eight in situ experiments from the controls are the presence of

ARTICLE IN PRESS C.E. Reimers et al. / Continental Shelf Research 24 (2004) 183–201

192 -0.6

Depth in Sand (cm)

-0.4 -0.2 0 0.2 0.4 0.6 P1 P2 P3

0.8 1 0

5

(A)

10 Iodide (µM)

15

20

120

Iodide Peak Height (nA)

110 100 90 80 70 60 50 40

tD

30 20 10

L= 2.0 cm

0 0

(B)

10 20 30 Time after Tracer Injection (hr)

40

Fig. 3. (A) Iodide profiles from the shipboard control experiment measured 4.75–6.25 h after tracer release from the diffuser array. The deepest point of each profile represents the last point at which an iodide peak could be clearly separated from the signal produced by dissolved O2. Horizontal error bars represent71 standard deviation of replicate measurements (usually 3) at each electrode position. (B) The results of a second laboratory control experiment presented as a breakthrough curve for tracer diffusion into sand from the study site, L ¼ 2:0 cm.

bedforms and oscillating flow. Table 3 characterizes these experiments according to the sediment depth (cm) monitored for tracer arrival, location relative to bedforms, duration of monitoring, and duration of tracer release. Each experiment is also designated by a tripod deployment code (e.g., D11). During tripod deployment D13, two sepa-

rate tracer experiments were conducted. The first was at 0.9 cm; the second was at 0.6 cm and initiated 50 min after the first. Fig. 4 illustrates the experimental operations overlain on the record of current speed and wave-generated pressure variations for experiment D13-0.9. This data format reveals that peak current speeds and wave heights often coincided. Average current and wave conditions for all eight in situ experiments were not significantly different, but the temperature of the bottom water decreased over the 3 days of operations (Table 4). This shift was caused by localized upwelling and the transport of offshore waters shoreward. The average wave period was 7 s (range 6–7 between experiments). By comparing ripples observed in video records to objects of known scale (such as the diffuser array), we estimated most ripples to have heights of B3 cm (Table 3). Traykovski et al. (1999) describe the predominant ripple pattern at LEO-15 as two-dimensional wave ripples with a mean ripple steepness (height/ wavelength=0.15). This would suggest ripple wavelengths B20 cm during this study. Using the relationships of Airy wave theory (Komar et al., 1972) to make some approximate calculations of the bottom water orbital motion under waves during our observations, an average orbital diameter do ¼ 17 cm is computed. This measure also is expected to approximately equal the ripple length (Komar et al., 1972). Using the Grant and Madsen (1982) model of bedform-generated roughness in oscillatory flow, average surface wave conditions during all deployments were more than sufficient to erode sand. The percentage of all waves recorded that generated wave shear velocities in excess of the threshold shear velocity for sand movement ranged from 50% for D11 to 87% for D12. Tracer penetration was detected in every in situ experiment except D21-2.0, the only one sampling at a ripple crest. Vertical oxygen profiles and iodide breakthrough records are presented in Fig. 5. Oxygen concentrations were measured at increasing depths into the sediment until reaching the depth chosen for tracer detection. Then after tracer detection and dissipation, the oxygen profile was resumed in some cases (Fig. 5). Experiment

ARTICLE IN PRESS C.E. Reimers et al. / Continental Shelf Research 24 (2004) 183–201

193

Table 3 Conditions of in situ experiments Experiment designation

Depth of sensor tip in sand (cm)

Location/microtopography/flow and experimental observations Duration of Tracer pulse scanning (mm:ss) (mm:ss)

D11-0.4

0.4

D12-0.6

0.6

D13-0.6

0.6

D13-0.9 D15-1.0

0.9 1.0

D14-1.5

1.5

D16-2.0

2.0

D21-2.0

2.0

Area with irregular relief, 1–2 cm in height; sand transport during surges Trough between ripples; ripple height B3 cm; oscillating sand transport especially at ripple crests Trough between ripples; ripple height B3 cm; oscillating sand transport especially at ripple crests Same location as D13-0.6 Trough between ripples, shell debris concentrated in trough; experiment was aborted after tripod lifted due to tension on tether Trough between ripples; oscillating sand transport especially at ripple crests Base of ripple flank, B8 cm away from crest; ripple height B4 cm and building Ripple crest; ripple height B4 cm; oscillating sand transport at ripple crest, produced episodic scouring and infilling of grains around sensor; no tracer arrival was detected

01:40

11:55

01:29

06:26

00:39

12:48 10:58

03:00 02:00

19:08

05:00

47:32

03:13

24:24

02:49

iodide detection at 0.9 cm 12

11.5

11

Pressure 1 mab (dbar)

tracer release

06:15

Current speed (m s-1)

0.5 0.4 0.3 0.2 0.1 0 7:46:30

7:51:30

7:56:30

Time

Fig. 4. A wave and current record from 1 m above the seafloor during the time interval voltammetric scans were run in a repetitive sequence during experiment D13-0.9. The time intervals of tracer release and detection at 0.9 cm are indicated, as are the greatest 5% of all current velocities (D) and wave heights (}) measured during this deployment.

D15-1.0 had to be aborted soon after the tracer was detected because the tripod lifted momentarily from the bottom due to tension on the tether.

Experiment D21-2.0 was also stopped (after 24 min 24 s) because of worsening sea conditions and a high degree of sediment movement and

ARTICLE IN PRESS 194

C.E. Reimers et al. / Continental Shelf Research 24 (2004) 183–201

Table 4 Physical conditions during tracer experiments Experiment

Horizontal flow speed at 1 mab (m s1)

Water column height at 1 mab (m)

Surface wave height (m)

Water temp. ( C)

D12-0.6 D11-0.4 D13-0.6 D13-0.9 D15-1.0 D14-1.5 D16-2.0 D21-2.0

0.1370.08 0.1070.05 0.1670.09 0.1570.09 0.1270.07 0.1370.08 0.1270.07 0.0970.05

11.570.2 11.870.1 11.970.2 11.770.1 12.070.1 12.670.1 11.670.1 12.970.1

0.3470.17 0.2170.11 0.3370.16 0.2970.16 0.2770.13 0.3070.18 0.2570.13 0.2370.11

18.070.4 20.470.4 16.670.0 17.770.3 18.070.8 17.470.4 14.770.2 15.671.2

Note: Except for wave height, each value represents the average of current meter readings every 1 s over the time interval when scanning for iodide (Table 3). Individual wave heights were derived from the difference of alternating maximum (crest) and minimum (trough) pressures and averaged over the scanning time.

associated scouring at the ripple crest (Table 3). We did not see (through the close-up video) indications of tripod movement or scouring around the sensors during all other experiments. However as indicated earlier, sand grains and shell fragments at the very surface of the sediment were often in oscillatory motion. Tracer advection rates were calculated as vertical velocities by dividing the sediment depth of tracer detection by the time delay. The latter was calculated as the difference between the tracer breakthrough time and the mid-point of the tracer release pulse (Fig. 5). Although the conditions during each experiment were in detail unique, the uniformity of water depth, wave characteristics, and sediment type allows intercomparison by sediment depth (Fig. 6). Error bars in Fig. 6 represent the ranges in computed velocities that result if delay times are calculated as breakthrough time minus either the start or end point of the tracer pulse. Since in D11-0.4, tracer arrival was detected before the end of the injection pulse, no upper limit is shown for the velocity at 0.4 cm (Fig. 6). The range for each experiment tightens with increasing sediment depth and as the ratio of breakthrough time to the duration of the release pulse increases. The derived velocities generally decrease with increasing depth in the sand, and an exponential curve is fit through the data to emphasize this trend.

3.4. Column experiments The flow rate of seawater through the column experiments was set to vary from 1.20 to 5.56 ml cm2 h1 or 288–1334 l m2 d1 (Table 1). Given an average sediment porosity of 0.37 (Table 2), these pumping rates would propagate a fluid front down column at 3.2–15 cm h1, in keeping with the in situ velocities between 0.4 and 2.0 cm depth (Fig. 6) and probably a few centimeters deeper. The oxygen consumption rates for each sediment column at known pumping rates were calculated from ([O2]inflow[O2]outflow) times the volume of water (pore space) in each column times the flow rate. These values were then normalized to a conservative flushing depth of 4 cm (assumed based on the observed depth of pore-water oxygen penetration, Fig. 5) to yield oxygen fluxes per area of seafloor (Fig. 7). Since the column oxygen measurements were recorded first with six freshly retrieved cores at two flow rates, then repeated the next day at the slower flow rate (plus two columns; Table 1), the results in Fig. 7 are separated into two groups of measurements. After 20.5 h or greater of incubation, the oxygen consumption rates were all higher than initial values at similar flow rates and more reproducible between columns. The relatively large scatter in the initial data may be due to sediment heterogeneity, or different densities of micro-, meio- and macrofauna in the

ARTICLE IN PRESS C.E. Reimers et al. / Continental Shelf Research 24 (2004) 183–201

columns.4 Assuming a flushing depth of 4 cm, these first sediment column experiments indicate that the oxygen flux to the sediment decreases on the order of 0.029 mmol m2 d1 when the rate of flushing decreases by 1 l m2 d1. Deeper flushing (e.g., if rates were normalized to 6 cm instead of 4 cm) would increase the estimated rates and the slope of the flow rate vs. flux relationship proportionally. We have measured sediment oxygen penetration depths with both voltammetric and amperometric microelectrodes at this site ranging from 0.4 to >6 cm at different times between April 2000 and July 2001. These observations indicate the flushing depth deepens as wave conditions and associated sand ripples build.

4. Discussion 4.1. Patterns of advective transport Streamlines and rates of wave-induced advective transport within a rippled sand bed have been predicted with two-dimensional steady state transport models, compared to diffusive transport, and used to argue for high organic matter processing rates in continental shelf sediments (Shum, 1992, 1993; Shum and Sundby, 1996). These patterns have also been followed in flumes and in field studies of rippled intertidal sandflats with dye tracers and natural solutes (Huettel et al., 1996, 1998; Ziebis et al., 1996; Huettel and Webster, 2001; Precht and Huettel, 2003). Although dependent on the prevalent wave conditions, the ripple profile, and sediment properties, net advective transport is expected to be downward into the sands of ripple troughs and flanks, and to lead out of the bed near crests. While following these courses the trajectories of discrete pore-water parcels are predicted to vary much more widely. The measurements of iodide transport from this study are in remarkable agreement with the physical models and provide the first estimates of 4 Macrofaunal counts were: column 1, 1 small bivalve; column 2, 1 small shrimp; column 3, 1 polychaete; column 4, none; column 5, 1 amphiurid; column 6, 2 small polychaetes; column 7, 2 small amphiurids; and column 8, none.

195

the magnitude of advective flows within shelf sediments under fairly low-energy sea conditions. The one experiment at a ripple crest (D21-2.0), although only maintained for 24.4 min, showed no tracer arrival, consistent with the crest being a zone of fluid extrusion. Interfacial advective velocities at ripple troughs and flanks were observed to be into the sand and to vary from approximately 53 to 6 cm h1 (Fig. 6, mid-point velocities, 0.4–2.0 cm, respectively). These velocities are approximately an order of magnitude greater than rates of penetration observed in the flume studies of Huettel et al. (1996). However, these flume studies were conducted under steady horizontal flow velocities selected to equal 10 cm s1 at 10 cm above the bed (compared to the non-steady oscillating in situ flow), and the flume sands were finer (median grain size 250– 300 mm compared to 400–500 mm). In this study, distance-averaged intrusion velocities generally decreased with sediment depth, and the depth of advective penetration can be inferred to have been approximately 4–4.5 cm (where pore fluids became anoxic; Fig. 5). Interestingly, the flume studies of Huettel et al. (1996) showed similar penetration depths, and their vertical intrusion velocities (derived by time lapse photography of tracer fronts) also decreased rapidly with sediment depth. The records of near-bottom current velocities and pressure fluctuations (Table 4; Fig. 4) from this study suggest that forces from waves and currents commonly peak every few minutes when the largest 5% of waves and current velocities coincide. In turn, the iodide tracer breakthrough records show evidence of non-steady interstitial advection. Breakthrough records that exhibited abrupt peaks that were shorter in duration than the corresponding tracer release pulse (e.g., D110.4, D14-1.5), or that were bimodal (D12-0.6), suggest bursts of intrusive flow followed by dispersive dissipation. Relatively long periods of tracer detection, such as was observed during D162.0, suggest an absence of vigorous flushing after an initial tracer intrusion event. Further evidence of in-bed temporal changes being driven by the changeable flow is found in the voltammetric electrode profiles of dissolved

ARTICLE IN PRESS C.E. Reimers et al. / Continental Shelf Research 24 (2004) 183–201 0

500

1

400 Peak Height (nA)

Depth in Sand (cm)

196

2 3

300 200 100

4

0

5 0

0

40 80 120 160 200 Oxygen (µM)

0

2

4 Elapsed Scan Time (min)

6

8

140 120

1 Peak Height (nA)

Depth in Sand (cm)

D11 0.4 cm

2 3 4

D12 0.6 cm

100 80 60 40 20 0

5 0

40

80

0

120 160

2

Oxygen (µM)

4 6 8 Elapsed Scan Time (min)

10

12

Peak Height (nA)

40

1

20

10

2 0

3

0

4

2 4 6 Elapsed Scan Time (min)

8

20

5 0

40 80 120 160 Oxygen (µM)

D13 0.9 cm

16 Peak Height (nA)

Depth in Sand (cm)

0

D13 0.6 cm

30

12 8 4 0 0

2

4 6 8 Elapsed Scan Time (min)

10

12

ARTICLE IN PRESS C.E. Reimers et al. / Continental Shelf Research 24 (2004) 183–201 30

2 3

20 -tripod lifted 10

4 0

5 0

0

40 80 120 160 Oxygen (µM)

2

4 6 8 10 Elapsed Scan Time (min)

12

14

70

0

60

1 Peak Height (nA)

Depth in Sand (cm)

D15 1.0 cm

1 Peak Height (nA)

Depth in Sand (cm)

0

2 3 4

D14 1.5 cm

50 40 30 20 10 0

5 0

0

40 80 120 160 Oxygen (µM)

0

20

1

16 Peak Height (nA)

Depth in Sand (cm)

197

2 3

4

8 12 Elapsed Scan Time (min)

16

20

D16 2.0 cm

12 8 4

4

0

5 0

40 80 120 160 Oxygen (µM)

0

10

20 30 Elapsed Scan Time (min)

40

50

Fig. 5 (continued).

Fig. 5. Oxygen profiles (left) and tracer breakthrough records (right) measured in situ with voltammetric electrodes. The results for each experiment are presented in order of increasing sediment depth for tracer detection. The error bars for each oxygen measurement represent the standard deviation about the mean of 3 or 4 scans with (~) representing measurements before the tracer release and (U) representing measurements after. Profiles were extended deep enough to penetrate anoxic sediments during only two deployments. Within the breakthrough plots the bold horizontal bar corresponds to the time interval of the tracer release pulse. The arrow points to the designated ‘‘breakthrough’’ time. The horizontal scales vary according to the duration of scanning (Table 3).

ARTICLE IN PRESS C.E. Reimers et al. / Continental Shelf Research 24 (2004) 183–201

198

0

60

15

60 1

40 Velocity = 47.7e-1.12x R2 = 0.74

20

0 0

0.5

1 1.5 2 Depth in Sand (cm)

Oxygen Consumption Rate (mmol m-2 d-1)

Velocity (cm hr-1)

Vertical Velocity (cm hr -1) 5 10

50 3

40

1 2

30 8 75 6

20

oxygen. Oxygen concentrations measured at discrete depths on time scales of several minutes (3–4 scans, then averaged to give points within profiles), or as sets of scans at the same location before and after tracer experiments, often exhibited high variance (Fig. 5). These examples have higher variance than the analytical precision reflected in bottom water values (Fig. 5) and can be explained if each single scan ‘‘sampled’’ a unique parcel of pore fluid while it and neighboring parcels were in motion under a rippled bed. The irregular shapes of the vertical oxygen profiles are also signatures of complex, pressure-driven transport (Ziebis et al., 1996; Lohse et al., 1996). Since streamlines must bend and run horizontally to connect zones of intrusion and extrusion (Huettel et al., 1996), the subsurface gradient reversal, seen for example in the oxygen profile from D13, indicates complicated pore-water exchange. 4.2. Oxygen fluxes to sands under waves The finite length of each tracer release and the probability that only some parcels of the tracer pulse entered the bed in the vicinity of the voltammetric sensor were reasons to calculate a

3 1

7 6

7 2

3 2

10

2.5

Fig. 6. Distance-averaged advection rates vs. depth in the sand in zones of bottom water intrusion. Symbols indicate rates calculated from delay times measured from the mid-point of the tracer release pulse to the breakthrough time. Ranges represent rates based on times estimated from either the beginning or end of the pulse.

4

6 5

5

0 0

200

400

600 800 1000 1200 1400 Flow Rate (L m-2d-1)

Fig. 7. Relationships between oxygen flux and flushing rate for surface sediments from LEO-15 in July 2001. Each oxygen flux determination represents oxygen consumption values from an individual numbered column normalized to a flushing depth of 4 cm (i.e., the measured column flux was multiplied by 4 divided by the sediment column length in centimeters). Error bars indicate standard deviations about the average of either four invs. out-flow measurements (closed symbols) or six in- vs. outflow measurements (open symbols) started after 3.25 or 20.5 h of shipboard incubation under flow, respectively (Table 1). Initial oxygen consumption measurements were run at two flow rates per column, dropping from the higher to lower rate. A linear regression of the initial measurements is plotted with a solid line (y ¼ 0:029x  2:27; R2 ¼ 0:56), while the regression of the data after 20.5 h is represented by a broken line (y ¼ 0:029x þ 13:9; R2 ¼ 0:75); where x=flow rate (l m2 d1) and y=oxygen consumption rate (mmol m2 d1).

range of pore-water velocities for each in situ experiment in this study (Fig. 6). However, it can be concluded without doubt that bottom water intrusion does occur between ripples and that the flow rates of 288–1334 l m2 d1 (3.2–15 cm h1, as applied to the sediment columns on shipboard) are realistic. As oxygenated bottom water is forced into permeable sediments by the oscillatory pressure field, suspended organic particles will be carried along, trapped and degraded (the biocatalytic filter effect, Rusch et al., 2001). Bacterial and algal cell abundances and particulate organic carbon and nitrogen contents in additional cores taken during this study showed near-surface enrichments but also high between-column

ARTICLE IN PRESS C.E. Reimers et al. / Continental Shelf Research 24 (2004) 183–201

variability (Rusch et al., 2003). This spatial variability, as well as differences in meio- and macrofauna, and the complex coupling between organic substrate lability, oxygen supply, and benthic respiration are presumed to explain the scatter in the oxygen consumption rates observed at individual flow rates in the column experiments beginning just 3.25 h after core retrieval (Fig. 7 open symbols). Although the oxygen consumption rates for the column experiments were not uniform initially, each decreased by similar amounts when flow rates were lowered. We interpret this result to indicate that rates of aerobic organic matter oxidation in permeable sediments are flow-dependent, probably because higher rates of flow cause greater dispersion and greater penetration of dissolved oxygen into microenvironments within a heterogeneous sand column. It is also suggested that oxygen consumption must vary considerably in space and time depending on the conditions controlling the depth (and thus volume) of sediment affected by flow. Chemical oxidation rates (e.g., by sulfide oxidation) were not a factor in our column experiments because the sands were kept oxic under flow. When the sand columns had incubated for > 20:5 h under steady flow, oxygen consumption rates rose to more consistent values. This may be due to an increase in the metabolic rate of sediment-attached bacteria (perhaps stimulated by decaying biota trapped in cores), or to the development of mutualism between microorganisms. Whatever the explanation, the importance of biological variables in regulating oxygen consumption in the presence of pore-water flow needs further investigation. The results presented in Figs. 6 and 7 when considered together suggest benthic respiration rates at LEO-15 range between 10 and 40 mmol O2 m2d1 under modest summer waves. Such rates bracket an earlier estimated mean (12.8 mmol O2 m2d1) for sandy regions of the continental shelf south of New England based on whole core incubations without flow (Rowe et al., 1988). However, neither the earlier core experiments nor our column experiments included light effects on net oxygen fluxes. Benthic microalgal primary production rates of oxygen have been observed to

199

nearly cancel dark oxygen consumption on the southeastern USA continental shelf in summer (Jahnke et al., 2000). The light levels at the seafloor of the southeastern USA shelf are nearly always significantly higher in intensity and percent surface irradiance (even at water depths of 30 m; Nelson et al., 1999) than what we have measured at LEO15, but longer-term and more wide-spread light records are needed to assess the role of light for benthic metabolism across the MAB. We can add anecdotally that we have measured steadily decreasing bottom water O2 levels (and rising total-CO2 levels) during several multi-day periods of summer water-column stratification at LEO-15, and local hypoxia occurs in years when these conditions persist. Thus, most available evidence suggests a dominance of benthic respiration over benthic primary production in the New Jersey coastal region in contrast to the southeastern USA shelf. We conclude that benthic oxygen consumption is significantly enhanced under waves by porewater flow-through ripples, but without more measurements under varying hydrodynamic conditions we cannot predict the full impact of such sand filtering processes on the MAB shelf ecosystem. More measurements of the kind pioneered in this study should also be undertaken with light and dark benthic chamber measurements in other regions with different types of permeable sediments. By such efforts, oceanographers may come to describe the effects of wave-induced advective transport on fluxes of solutes and the processing of organic detritus in the coastal ocean.

Acknowledgements This study would not have been successful without the assistance of several generous, patient and hard-working people. Dr. George Luther III (U. Delaware) helped us to master the voltammetric techniques and loaned us critical electrode polishing equipment from his laboratory for this work. Dr. Don Nuzzio (AIS) also helped us to trouble shoot the underwater voltammetric equipment and provided backup circuit boards. David Lovalvo (Eastern Oceanics) developed parts of the tripod specifically for these experiments and

ARTICLE IN PRESS 200

C.E. Reimers et al. / Continental Shelf Research 24 (2004) 183–201

helped design the diffuser array. A team of divers from Rutgers University collected the cores. The technicians, crew and captain of the R./V. Cape Henlopen kept the ship on three-point anchor and assisted with the deployments. Dr. Robert Key assisted with the deployments and water sampling. Drs. Stefan Forster and Bernard Boudreau provided helpful reviews of the initial manuscript. This work was supported by a grant from the US National Science Foundation to CR, GT and CF. Support for the participation of MH, AR and CW was provided by the Max Planck Institute for Marine Microbiology, Germany.

References Berner, R.A., 1980. Early Diagenesis: a Theoretical Approach. Princeton University Press, Princeton, NJ, 241pp. Boehme, S.E., Sabine, C.L., Reimers, C.E., 1998. CO2 fluxes from a coastal transect: a time-series approach. Marine Chemistry 63, 49–67. Boudreau, B.P., 1997. Diagenetic Models and Their Implementation Modeling Transport and Reactions in Aquatic Sediments. Springer, Berlin, 414pp. Boudreau, B.P., Huettel, M., Forster, S., Jahnke, R.A., McLachlan, A., Middelburg, J.J., Nielsen, P., Sansone, F., Taghon, G., Van Raaphorst, W., Webster, I., Weslawski, J.M., Wiberg, P., Sundby, B., 2001. Permeable marine sediments: overturning an old paradigm. Eos 82, 133–136. Brendel, P.J, Luther III, G.W., 1995. Development of a gold amalgam voltammetric microelectrode for the determination of dissolved Fe, Mn, O2, and S(-II) in porewaters of marine and freshwater sediments. Environmental Science and Technology 29, 751–761. Buffle, J., Tercier-Waeber, M.-L., 2000. In situ voltammetry: concepts and practice for trace analysis and speciation. In: Buffle, J., Horvai, G. (Eds.), In Situ Monitoring of Aquatic Systems. Wiley, Chichester, pp. 279–406. DeGrandpre, M.D., Olbu, G.L., Beatty, C.M., Hammar, T.R., 2002. Air–sea CO2 fluxes on the US Middle Atlantic Bight. Deep-Sea Research II 49, 4355–4367. Emerson, S., Jahnke, R., Heggie, D., 1984. Sediment–water exchange in shallow water estuarine sediments. Journal of Marine Research 42, 709–730. Friederich, G.E., Codispoti, L.A., Sakamoto, C.M., 1991. An easy-to-construct automated Winkler titration system. MBARI Technical Report No. 91-6. Fritsche, U., Koschinsky, A., Winkler, A., 2001. The different diffusive transport behaviors of some metals in layers of Peru Basin surface sediment. Deep-Sea Research II 48, 3653–3681.

Glenn, S.M., Dickey, T.D., Parker, B., Boicourt, E., 2000. Long-term real-time coastal ocean observation networks. Oceanography 13, 24–34. Grant, W.D., Madsen, O.S., 1982. Movable bed roughness in unsteady oscillatory flow. Journal of Geophysical Research 87, 469–481. Grasshoff, K., Kremling, K., Ehrhardt, M., 1983. Methods of Seawater Analysis, 2nd Edition. Verlag Chemie, Weinheim, 419pp. Gray, H., 1958. Suggested method of test for permeability of porous granular materials by the falling-head permeameter. In: Procedures for Testing Soils. American Society for Testing Materials, Philadelphia, pp. 253–255. Huettel, M., Webster, I.T., 2001. Porewater flow in permeable sediments. In: Boudreau, B.P., J^rgensen, B.B. (Eds.), The Benthic Boundary Layer: Transport Processes and Biogeochemistry. Oxford University Press, Oxford, pp. 144–179. Huettel, M., Ziebis, W., Forster, S., 1996. Flow-induced uptake of particulate matter in permeable sediments. Limnology and Oceanography 41, 309–322. Huettel, M., Ziebis, W., Forster, S., Luther III, G.W., 1998. Advective transport affecting metal and nutrient distributions and interfacial fluxes in permeable sediments. Geochimica et Cosmochimica Acta 62, 613–631. Jahnke, R.A., Nelson, J.R., Marinelli, R.L., Eckman, J.E., 2000. Benthic flux of biogenic elements on the Southeastern US continental shelf: influence of pore water advective transport and benthic microalgae. Continental Shelf Research 20, 109–127. Komar, P.D., Neudeck, R.H., Kulm, L.D., 1972. Observations and significance of deep-water oscillatory ripple marks on the Oregon continental shelf. In: Swift, D.J.P., Duane, D.B., Pilkey, O.H. (Eds.), Shelf Sediment Transport: Process and Pattern, Dowden. Hutchinson and Ross, Stroudsbury, PA, pp. 601–619. Lee, T.-C., 1999. Applied Mathematics in Hydrogeology. CRC Press, Boca Raton, FL, 382pp. Li, Y.-H., Gregory, S., 1974. Diffusion of ions in seawater and in deep-sea sediments. Geochimica et Cosmochimica Acta 38, 703–714. Lohse, L., Epping, E.H.G., Helder, W., van Raaphorst, W., 1996. Oxygen pore water profiles in continental shelf sediments of the North Sea: turbulent versus molecular diffusion. Marine Ecology Progress Series 145, 63–75. Luther III, G.W., Brendel, P.J., Lewis, B.L., Sundby, B., Lefrancois, L., Silverberg, N., Nuzzio, D.B., 1998. Simultaneous measurement of O2, Mn, Fe, I, and S(-II) in marine pore waters with a solid-state voltammetric microelectrode. Limnology and Oceanography 43, 325–333. Luther III, G.W., Reimers C, E., Nuzzio D, B., Lovalvo, D., 1999. In situ deployments of voltammetric, potentiometric, and amperometric microelectrodes from a ROV to determine dissolved O2, Mn, Fe, S(-2), and pH in porewaters. Environmental Science and Technology 33, 4352–4356. Malan, D.E., McLachlan, A., 1991. In situ benthic oxygen fluxes in a nearshore coastal marine system: a new approach

ARTICLE IN PRESS C.E. Reimers et al. / Continental Shelf Research 24 (2004) 183–201 to quantify the effect of wave action. Marine Ecology Progress Series 73, 69–81. Marinelli, R.L., Jahnke, R.A., Craven, D.B., Nelson, J.R., Eckman, J.E., 1998. Sediment nutrient dynamics on the South Atlantic Bight continental shelf. Limnology and Oceanography 43, 1305–1320. Martin, W.R., Banta, G.T., 1992. The measurement of sediment irrigation rates: a comparison of the Br tracer and 222Rn/226Ra disequilibrium techniques. Journal of Marine Research 50, 125–154. Martin, W.R., Sayles, F.L., 1987. Seasonal cycles of particulate and solute transport processes in nearshore sediments: 222 Rn/226Ra and 234Th/238U disequilibrium at a site in Buzzards Bay, MA. Geochimica et Cosmochimica Acta 51, 927–943. McDuff, R.E., Ellis, R.A., 1979. Determining diffusion coefficients in marine sediments: a laboratory study of the validity of resistivity techniques. American Journal of Science 279, 666–675. Meier, H., Zimmerhackl, E., Hecker, W., Zeitler, G., Menge, P., 1988. Measurement of diffusion of radionuclides in sediment rocks. Radiochimica Acta 44/45, 239–244. Meier, H., Zimmerhackl, E., Zeitler, G., Menge, P., 1991. Diffusion measurements of radionuclides in site-specific sediment/groundwater-systems. Material Research Society Symposium Proceedings 212, 617–622. Nelson, J.R., Eckman, J.E., Robertson, C.Y., Marinelli, R.L., Jahnke, R.A., 1999. Variability in biomasses of benthic and planktonic microalgae on the continental shelf on the South Atlantic Bight. Continental Shelf Research 19, 477–505. Precht, E., Huettel, M., 2003. Advective pore-water exchange driven by surface gravity waves and its ecological implications. Limnology and Oceanography 48, 1674–1684. Reimers, C.E., Glud, R.N., 2000. In situ chemical sensor measurements at the sediment–water interface. In: Varney, M. (Ed.), Chemical Sensors in Oceanography. Gordon and Breach Publishing, Amsterdam, pp. 249–282. Reimers, C.E., Jahnke, R.A., Thomsen, L., 2001. In situ sampling in the benthic boundary layer. In: Boudreau, B.P., J^rgensen, B.B. (Eds.), The Benthic Boundary Layer: Transport Processes and Biogeochemistry. Oxford University Press, Oxford, pp. 245–268. Rowe, G.T., Smith, S., Falkowski, P., Whitledge, T., Theroux, R., Phoel, W., Ducklow, H., 1988. Benthic carbon budgets for the continental shelf south of New England. Continental Shelf Research 8, 511–527.

201

Rusch, A., Forster, S., Huettel, M., 2001. Bacteria, diatoms and detritus in an intertidal sandflat subject to advective transport across the water–sediment interface. Biogeochemistry 55, 1–27. Rusch, A., Huettel, M., Reimers, C.E., Taghon, G.L., Fuller, C.M., 2003. Activity and distribution of bacterial populations in Middle Atlantic Bight shelf sands. FEMS Microbiology Ecology 44, 89–100. Rutgers van der Loeff, M.M., 1981. Wave effects on sediment water exchange in a submerged sand bed. Netherlands Journal of Sea Research 15, 100–112. Shum, K.T., 1992. Wave-induced advective transport below a rippled water–sediment interface. Journal Geophysical Research 97 (C1), 789–808. Shum, K.T., 1993. The effects of wave-induced pore water circulation on the transport of reactive solutes below a rippled sediment bed. Journal Geophysical Research 98 (C6), 10289–10302. Shum, K.T., Sundby, B., 1996. Organic matter processing in continental shelf sediments—the subtidal pump revisited. Marine Chemistry 53, 81–87. Styles, R.B., 1998. A continental shelf bottom boundary layer model: development, calibration and applications to sediment transport in the Middle Atlantic Bight. Ph.D. Thesis, Rutgers University, New Brunswick, NJ, 261pp. Styles, R., Glenn, S.M., 2002. Modeling bottom roughness in the presence of wave-generated ripples. Journal Geophysical Research 107(c8), 24-1–24-15. Svensson, U., Rahm, L., 1991. Towards a mathematical model of oxygen transfer to and within bottom sediments. Journal Geophysical Research 96, 2777–2783. Traykovski, P., Hay, A.E., Irish, J.D., Lynch, J.F., 1999. Geometry, migration, and evolution of wave orbital ripples at LEO-15. Journal Geophysical Research 104 (C1), 1505–1524. Vanderborght, J.P., Wollast, R., Billen, G., 1977. Kinetic models of diagenesis in disturbed sediments. Part 1. Mass transfer properties and silica diagenesis. Limnology and Oceanography 22, 787–793. Webb, J.E., Theodor, J.L., 1968. Irrigation of submerged marine sands through wave action. Nature 220, 682–685. Webster, I.T., Taylor, J.H., 1992. Rotational dispersion in porous media due to fluctuating flow. Water Resources Research 28, 109–119. Ziebis, W., Huettel, M., Forster, S., 1996. Impact of biogenic sediment topography on oxygen fluxes in permeable seabeds. Marine Ecology Progress Series 140, 227–237.