Isotopic constraints on oxygen sink partitioning in the Lower St. Lawrence Estuary

Aerobic respiration and hypoxia in the Lower St. Lawrence Estuary: Stable isotope

ratios of dissolved oxygen constrain oxygen sink partitioning

Moritz F. Lehmann,a,b,* Bruce Barnett,c Y. Gelinas,b,d Denis Gilbert,e Roxane J. Maranger,f

Alfonso Mucci,b,g Bjorn Sundby,g,h and Benoit Thibodeaub

a Institute for Environmental Geosciences, University of Basel, Basel, SwitzerlandbGeochemistry and Geodynamics Research Center (GEOTOP-UQAM-McGill), Montreal, Quebec, Canadac Department of Geosciences, Princeton University, Guyot Hall, Princeton, New JerseydDepartment of Chemistry and Biochemistry, Concordia University, Montreal, Quebec, Canadae Fisheries and Oceans Canada, Maurice–Lamontagne Institute, Mont-Joli, Quebec, Canadaf Department of Biological Sciences, Universite de Montreal, Montreal, Quebec, Canadag Department of Earth and Planetary Sciences, McGill University, Montreal, Quebec, CanadahInstitut des Sciences de la Mer de Rimouski (ISMER), University of Quebec, Rimouski, Quebec, Canada

Abstract

We measured the concentration and the stable isotope ratios of dissolved oxygen in the water column in theEstuary and Gulf of St. Lawrence to determine the relative importance of pelagic and benthic dissolved oxygenrespiration to the development of hypoxic deep waters. The progressive landward decrease of dissolved oxygen inthe bottom waters along the axis of the Laurentian Channel (LC) is accompanied by an increase in the 18O : 16Oratio, as would be expected from O-isotope fractionation associated with bacterial oxygen respiration. Theapparent O-isotope effect, eO-app, of 10.8% reveals that community O-isotope fractionation is significantly smallerthan if bacterial respiration occurred solely in the water column. Our observation can best be explained by acontribution of benthic O2 consumption occurring with a strongly reduced O-isotope effect at the scale ofsediment–water exchange (eO-sed , 7%). The value for eO-sed was estimated from benthic O2 exchange simulationsusing a one-dimensional diffusion-reaction O-isotope model. Adopting this eO-sed value, and given the observedcommunity O-isotope fractionation, we calculate that approximately two thirds of the ecosystem respirationoccurs within the sediment, in reasonable agreement with direct respiration measurements. Based on thedifference between dissolved oxygen concentrations in the deep waters of the Lower St. Lawrence Estuary and inthe water that enters the LC at Cabot Strait, we estimate an average respiration rate of 5500 mmol O2 m22 yr21

for the 100-m–thick layer of bottom water along the LC, 3540 mmol O2 m22 yr21 of which is attributed tobacterial benthic respiration.

The Laurentian Channel (LC) is a 1200-km–long andmore than 300-m–deep submarine valley that originates onthe Atlantic continental shelf off Nova Scotia and endsnear the mouth of the Saguenay Fjord (Fig. 1). The deepand slow landward flow in the deep waters brings oxygen-rich water from the Atlantic Ocean into the Gulf of St.Lawrence (Gilbert et al. 2005). The deep water is separatedfrom the oxygenated surface and the cold intermediatesubsurface layer (Gilbert and Pettigrew 1997) by a strongdensity gradient that inhibits vertical mixing of oxygen-richsurface waters with oxygen-poor bottom water (Fig. 2).Even during winter, water-column convection does notreach beyond 150 m in depth (Galbraith 2006). Thus,isolated from the atmosphere, the bottom water losesoxygen gradually through organic matter respiration as itflows landward along the LC.

The bottom water in the Lower St. Lawrence Estuary(LSLE) at the western end of the LC is hypoxic, withdissolved oxygen (O2) concentrations as low as 55 mmol L21

(Gilbert et al. 2005). The O2 concentration in the LSLEbottom water has decreased by 50% since 1930 (Gilbertet al. 2005), corresponding to an average depletion of

approximately 1 mmol L21 yr21. Gilbert et al. (2005)attributed one half to two thirds of the oxygen depletionto changes in the properties of the deep-water mass thatenters the Gulf of St. Lawrence through Cabot Strait. Theremaining O2 depletion was attributed to an increase in theorganic particle flux to the seafloor, possibly in response toeutrophication (Gilbert et al. 2005; Benoit et al. 2006;Thibodeau et al. 2006), with oxygen consumption bybacteria metabolizing organic matter in the water columnand in the sediment.

The relative importance of pelagic (water-column) andbenthic (sediment) respiration is unresolved but is critical toour understanding and ability to model the developmentand effects of hypoxic waters in the LSLE and the Gulf ofSt. Lawrence. In a previous study, Benoit et al. (2006)assumed that the sediments were the dominant oxygen sink,but this assumption has not been tested. Availablerespiration data are based on direct measurements ofoxygen uptake by sediments, but these measurements aredifficult to extrapolate at the scale of the whole estuarybecause of spatial and temporal variability. The purpose ofthis study is to apply an isotope method that integrates overlarge spatial and long temporal scales in order to examinewhere below the permanent pycnocline in the LSLE oxygenis consumed.* Corresponding author: [email protected]

Limnol. Oceanogr., 54(6), 2009, 2157–2169

E 2009, by the American Society of Limnology and Oceanography, Inc.

2157

Oxic respiration by various microorganisms (protozoa,bacteria, and phytoplankton) exhibits a significant organ-ism-level O-isotope fractionation as a result of the prefer-ential consumption of the lighter isotopologue 16O2. Hence,respiration leads to a detectable increase in the 18O : 16Oratio of O2 in oxygen-deficient marine and freshwaterenvironments (Bender 1990; Kiddon et al. 1993; Parker etal. 2005). If the isotope fractionation during respiration(eO) is known, the spatial distribution of the O2

18O : 16O, orits temporal variation, can yield quantitative informationon community respiration rates (Luz et al. 2002; Tobias etal. 2007). Previous work (Guy et al. 1993; Kiddon et al.1993; Bender et al. 1994) indicates that the O-isotope effectfor dark respiration by bacteria, through the enzymecytochrome oxidase, is relatively robust, with estimatesclustering around 18%. Significantly lower and higherO-isotope effects have been reported for other biological

processes that consume oxygen (Raven and Beardall 2005),but they do not occur, or can be considered insignificant, inthe aphotic zones of the ocean and other aquaticenvironments (Raven and Beardall 2005).

Oxygen respiration in sediments is often limited by thediffusive supply of oxygen, and, therefore, the community(or apparent) O-isotope effect can be significantly reduced(Bender 1990; Brandes and Devol 1997). Diffusion-limitedoxygen supply has been invoked to explain the reduction ofeO during root respiration (Angert and Luz 2001), as well asduring respiration in agricultural streams (Tobias et al.2007) and in coastal marine sediments (Brandes and Devol1997). Where the diffusion of O2 to the reaction site is veryslow with respect to the reaction rate, O2 is completelyconsumed and there is no apparent O-isotope effect.Brandes and Devol (1997) showed that the apparentsediment respiration O-isotope effect for the highly reactivesediments of Puget Sound was only 3% as a consequence ofdiffusion limitation around reactive microsites. Likewise,nitrate is often completely consumed within sedimentdenitrification zones, and sedimentary nitrate reductiondoes not cause much net isotope enrichment of nitrate inthe overlying water column of benthic environments(Brandes and Devol 1997; Lehmann et al. 2004, 2007), incontrast to denitrification occurring in the open watercolumn (Voss et al. 2001). The differential isotopicbehavior of respiration in the water column and insediments has been used to constrain the relative contribu-tion of the sediments to the total loss of nitrate bydenitrification, both in the global ocean (Brandes andDevol 2002) and in specific marine environments (Sigmanet al. 2003; Lehmann et al. 2005). By analogy, if the twoend-member isotope effects of oxic respiration are known,O-isotope systematic should provide a means to distinguishbetween bacterial O2 consumption in the water column andin sediments of marine or estuarine basins (Quinones-Rivera et al. 2007). The bottom water in the LC is wellsuited for the O-isotope approach, because it is isolated

Fig. 2. Potential density (sh) transect between Sta. 23 in the eastern part of the LSLE and Sta. 16 in the Gulf of St. Lawrence(August 2006). The pronounced density gradient in the upper water column results from a strong salinity gradient (data not shown). Thex-axis indicates distance from the mouth of the Laurentian Channel.

Fig. 1. Map showing hydrocast sampling locations duringcruises in August 2006 and May 2007. The bathymetric contoursoutline the Laurentian Channel. SF, Saguenay Fjord.

2158 Lehmann et al.

from the atmosphere and from the oxygen-rich surfacewater. It can therefore be treated as a quasi-closed systemwith respect to dissolved oxygen (DO).

Here we report on the concentration and isotopecomposition of O2 in the water column of the LSLE andGulf of St. Lawrence. These data show that the 18O : 16Oratio increases significantly with decreasing O2 concentra-tion. We apply a one-dimensional (1D) diffusion-reactionmodel to constrain the apparent O-isotope effect of benthicO2 respiration in the LSLE. The observed net O-isotopeeffect for community respiration in the deep waters of theLSLE and Gulf of St. Lawrence is used to estimate therelative contributions of benthic and water-column respi-ration, and these are compared to results of directmeasurements of bacterial production. We show that alarge fraction of the total aerobic bacterial respiration inthe deep estuary takes place within the sediments, empha-sizing their role in generating hypoxic conditions in thedeep water column of the LSLE.

Methods

Water samples were collected using a 12–Niskin bottlerosette sampling system (SeaBird SBE 911) at multiplestations along the LC in the LSLE and Gulf of St.Lawrence during two cruises in August 2006 and one cruisein May 2007 aboard the RV Coriolis II (Fig. 1). DO wasdetermined by Winkler titration (Grasshoff et al. 1983),with a relative standard deviation of better than 1%. ForO2 isotopic analysis, approximately 150 mL of water wascollected in evacuated 300-mL bottles containing 100 mL ofsaturated HgCl2 solution to stop biological activity(Emerson et al. 1995; Luz et al. 2002). The dissolved gaseswere allowed to equilibrate with the headspace over severaldays at room temperature. The water phase was carefullyremoved using a rotary vane pump, and the neck of thesample bottle was then dried prior to extraction. Theheadspace gas extraction line was evacuated, and afterremoval of all water and CO2 in cold traps, headspace gaseswere extracted into pre-evacuated 0.63-cm stainless-steeltubes immersed in liquid helium. The tubes were sealedwith a Nupro valve, removed from the liquid helium, andallowed to warm for 1 h prior to mass spectrometricanalysis. The gas mixture was expanded into a dual-inletFinnigan-Mat Delta plus XP isotope ratio mass spectrom-eter, and m/z ratios of 32, 33, 34, and 40 (argon [Ar]) weremeasured simultaneously relative to a standard of com-pressed atmosphere collected from Niwot Ridge Colorado.Oxygen isotope ratios are reported using the common deltanotation in %, thus:

d18O~18O�

16Osample

18O=16Ostandard

{1

� �|1000

� �ð1Þ

All d18O values are reported relative to atmospheric O2.O2 : Ar ratios were determined in the same samples, andd18O corrections were applied to account for the depen-dence of ionization efficiencies for the O2 isotopologues onthe O2 : Ar ratio (Bender et al. 1994). The correction factorwas determined for each measurement day by multiple

measurements of the compressed air standard with differ-ent amounts of Ar added. Additional corrections wereapplied to account for the sensitivity of measured d18Ovalues resulting from the pressure differentials of the twobellows (Bender et al. 1994). Single-run duration wasapproximately 30 min, during which the relative ion beamintensities for the sample and reference gases were bothdetermined 24 times. The reported d-values represent theaverage isotope ratios for each sample run. The analyticalprecision for d18O measurements and the reproducibilityfor replicate environmental samples were generally betterthan 0.01% and 0.2%, respectively.

Bacterial biomass production (BP) was estimated fromthe incorporation of tritiated leucine into protein (Kirch-man et al. 1985; Smith and Azam 1992). Briefly, triplicate1.5-mL water sample aliquots were dispensed into micro-centrifuge vials preloaded with 50 mL 3H-leucine (88 Cimmol21) to a final leucine concentration of 40 nmol L21.Samples were incubated at in situ temperatures, andincubations were stopped after 60 min by the addition of200 mL 50% trichloracetic acid (TCA). Samples werevortexed and centrifuged at 14,000 rounds min21 (rpm)for 10 min. Following centrifugation, the supernatant wasaspirated, and 1.5 mL of 5% cold TCA was added. Sampleswere vortexed, re-centrifuged, and aspirated as before.After addition of a scintillation cocktail (BD ScintiVerse,Fisher Scientific), radioactivity was measured using aPerkin Elmer scintillation counter. Rates of leucineincorporation were converted to estimates of bacterialcarbon (C) production using conversion factors reported inSimon and Azam (1989). An isotope dilution factor of 1.7was empirically determined by diluting 40 nmol L21

radioactive leucine with five different concentrations ofunlabeled leucine ranging from 0 to 150 nmol L21.

Results

DO concentrations—The concentration of DO decreaseswith distance from Cabot Strait along the sh 5 27.3 60.1 kg m23 isopycnal (corresponding to water depthsbetween 260 and 300 m; Fig. 2), from 175 mmol L21 atSta. 16 to below 60 mmol L21 O2 at Sta. 23 (2006 data,Figs. 3, 4). Dissolved O2 is continuously consumed as thedeep-water mass entering the Gulf of St. Lawrence slowlytravels landward. Figure 5a depicts O2 water-columnprofiles from the LSLE water column in 2007. Surfaceand cold intermediate layer waters contain O2 concentra-tions close to, or slightly above, saturation levels, as onewould expect for water in contact with the atmosphere. O2

concentrations decrease with depth at all stations, frombetween 340 and 380 mmol L21 in the photic zone tobetween 70 and 110 mmol L21 at ,250 m in depth. Asomewhat different O2 distribution at Sta. 20 (Fig. 5a),with significantly lower subsurface–depth O2 concentra-tions, is explained by local cyclonic circulation-inducedisopycnal doming, locally rising waters with higher salini-ties, and lower O2 concentrations in the center of theAnticosti gyre (salinity data not shown), a salient andpersistent local feature close to Sta. 20 (Saucier et al. 2003).Below ,200 m, the O2 concentration is essentially invari-

Respiration in the St. Lawrence Estuary 2159

ant with depth, but variations of up to 50 mmol L21 existbetween the most seaward and landward stations (Fig. 5a).The deep-water O2 concentrations at the proximal estua-rine stations are significantly lower than at the moreseaward stations, an observation that is consistent withcumulative bacterial respiration as the deep-water massestravel toward the head of the LSLE.

18O : 16O ratios in DO—In the deep waters, which areisolated from the upper water column by a strong andpermanent pycnocline, the d18O of the DO, d18OO2

, isstrongly anti-correlated with DO concentrations (R2 50.995) (Figs. 4, 5). Water column d18OO2

at Sta. 23increases from 1.2% at 50 m to .14.3% at 200 m andbelow, reflecting the .270 mmol L21 difference in [O2] withdepth (Fig. 5). Similarly, in 2006, we observed an increasein d18OO2

from 7.7% to 16.4% as the O2 decreased by.110 mmol L21 along the isopycnal transect (sh 5 27.3 60.1 kg m23) from Sta. 16 to Sta. 23 (Fig. 4). Theenrichment in 18O with decreasing O2 concentrations isconsistent with 18O-isotope discrimination during oxicbacterial respiration, the extent of which appears to besimilar along both horizontal and vertical O2 gradients.

A more quantitative assessment of the degree of18O-isotope enrichment (i.e., the apparent O-isotope effect,eO-app, is provided by a plot of d18OO2

vs. the naturallogarithm of the fraction of O2 that remains after partial O2

consumption by bacterial respiration (f) (‘Rayleigh plot’;Fig. 6). The fraction f is estimated from the ratio of theobserved O2 concentration over the equilibrium O2 concen-tration at the temperature and salinity of the sample, bothnormalized to Ar in order to account for mixing. Withongoing respiration, the O2 pool changes along a straightline in the Rayleigh diagram of d18OO2

vs. ln[O2], with theslope of the line approximating the O-isotope effectaccording to

d18OO2~d18OO2 init{eO-app ln f ð2Þ

(Mariotti et al. 1981). The Rayleigh model assumes thatthe respective water masses have the same initial O2 iso-tope composition, d18OO2

_init, and behave subsequently asclosed systems in which O2 is consumed. The data from theLC bottom-layer waters fit very well on a single Rayleighconsumption line, corresponding to an apparent orRayleigh O-isotope effect, eO-app, of 10.8% 6 0.1%(Fig. 6). This is only about 60% of the O-isotope effectreported for open water-column respiration (Bender 1990).Interestingly, analyses of d18OO2

data from the along-channel deep-water transect in 2006 yielded, within theerror of the method, the same eO-app as the d18OO2

datafrom water-column profiles collected in 2007 (Fig. 6a–c). Itis important to note that the O-isotope effect derived fromthe Rayleigh model represents a bulk community isotopeeffect, as it integrates the effect of respiration in the watercolumn and the sediments. We elucidate the potential

Fig. 3. Water column O2 concentration transect between Sta. 23 in the LSLE and Sta. 16 in the Gulf of St. Lawrence (August 2006).Hypoxic conditions prevail in the proximal portion of the estuary ([O2] , 62 mmol L21). The x-axis indicates distance from the mouth ofthe Laurentian Channel.

Fig. 4. O2 concentration (solid circles) and d18OO2(open

circles) transects in the LC along isopycnal surface sh 5 27.3 60.1 kg m23 in August 2006 (symbols represent mean of repli-cate measurements).

2160 Lehmann et al.

O-isotope dynamics during sedimentary bacterial respira-tion in detail below.

Bacterial production and respiration in the water column—Leucine incorporation–derived BP rates show a typicaldecline with depth in the LSLE water column (Sta. 23),with high BP rates in the upper 30 m of the water columnranging between 76 and 346 nmol C L21 d21, and they alsoshow BP values that are up to two orders of magnitudelower further below in the water column (10 6 2 nmol CL21 d21 at depths of ,150 m) (Fig. 7). Bacterial respira-tion (BR) was not measured directly but was calculatedfrom BP and an assumed bacterial growth efficiency (BGE)of 25% (Del Giorgio and Cole [1998] and referencestherein; BGE is the ratio of bacterial production to organicsubstrate metabolized) according to BR 5 (BP : BGE) 2BP. Assuming a respiratory C-to-O quotient of 0.89 foralgal organic matter (OM) decay (Hedges et al. 2002), weobtained BP-based O2 respiration rates of 33 6 6 nmol O2

L21 d21 for the deep LSLE water column (.150 m) at Sta.23.

Discussion

Numerical simulations: Estimating the apparent O-isotopeeffect of benthic O2 respiration on the water column—Wedistinguish between (1) the intrinsic (organism-level)O-isotope effect of bacterial O2 respiration (eO, theO-isotope effect at full expression, e.g., during water-column respiration), (2) the effective O-isotope effect ofbenthic respiration on the water column (eO-sed, lower thaneO as a result of diffusion limitation), and (3) thecommunity respiration O-isotope effect (eO-app, the mixedO-isotope fractionation signal in the water column

resulting from both the water-column and the benthicrespiration). In contrast to eO, eO-sed is, thus far, not wellconstrained. Based on O-isotope data from in situincubation experiments, Brandes and Devol (1997) esti-mated eO-sed in Puget Sound to be ,3%, indicating that themagnitude of underexpression of the organism-levelisotope effect of O-respiration at the scale of thesediment–water interface (SWI) is as low as for benthicdenitrification. Lehmann et al. (2007) demonstrated thatthe diffusive distance of nitrate to the denitrification zone isthe main constraint on the suppression of the organism-level isotope effect of denitrification in sediments, withgreater distances yielding lower nitrogen (N)–isotopeeffects. In contrast to denitrification, the diffusive distanceof oxygen to the active sites of respiration in the sedimentsis presumably zero, as aerobic respiration continues directlyat the SWI. Thus, the eO-sed for benthic O respiration, whilecertainly underexpressed with respect to the intrinsicO-isotope effect, is expected to be significantly higher than0–3% (Bender 1990).

In order to assess the effect of O2 consumption bybenthic bacteria on the O2 pool in the overlying bottomwater, we calculated the net fluxes of 18O and 16O acrossthe SWI from simulated pore-water O2 concentration andd18OO2

profiles, analogous to the approach used byLehmann et al. (2007) for nitrate N-isotopes. Profiles werecomputed using a 1D diffusion-reaction model, withequations for bacterial respiration implemented in version2.1f of AQUASIM (Reichert 1994, 1998; Lehmann et al.2007). The most important parameters that control thebalance of oxygen supply and consumption in sedimentsare bottom-water oxygen content and respiration rate(which in turn is a function of the sediment reactivity).These parameters modulate the O2 sediment penetration

Fig. 5. Water-column profiles for (a) O2 concentrations and (b) d18OO2for selected stations

in the LSLE in May 2007. Open symbols indicate single measurements north and south of thecentral channel at Sta. 20 and Sta. 23, respectively.

Respiration in the St. Lawrence Estuary 2161

depth. Complete consumption in an infinitely thin O2-respiration zone within the sediments would result in aneO-sed value of near zero. This idealized case cannot describethe situation in coastal environments, where benthicrespiration generally occurs over millimeter to centimeterscales within the sediment column and where, as aconsequence, finite d18OO2

gradients do exist.To better understand the effect of variations in sediment

reactivity and bottom-water oxygenation on eO-sed, wegenerated a suite of environmental scenarios that lead todifferent distributions of O2 concentrations and d18OO2

within the sediment pore water (Fig. 8). Our model resultsindicate that within reasonable limits of the governingparameters, net fluxes of 16O2 and 18O2 into the sedimentsalways combine to yield a robust value of approximately7% for eO-sed. This is slightly less than half the organism-

level O-isotope effect of 18% used in the model simulationsand is independent of reasonable (for coastal environ-ments) estimates of bottom-water O2 concentrations andthe respiration rate coefficient k (Table 1). Our model-derived estimates of eO-sed are not necessarily inconsistentwith the lower estimates of eO-sed of 3%, derived fromactual measurements by Brandes and Devol (1997) (seeabove). Whereas overall O2 consumption rates in PugetSound (4400 mmol m22 d21 on average) are well within therange considered in our simulations, the diffusion-reactiongeometry may be different. Brandes and Devol (1995, 1997)argued that both oxic respiration and denitrification inPuget Sound occur primarily within reactive microsites,rendering average diffusion distances greater than in otherbenthic environments (at least for O2 molecules) andpushing eO-sed closer to 0%. Yet, based on our modelresults (and ignoring the possible effect of microsites uponthe apparent O-isotope fractionation factor), an eO-sed valueof between 6.5% and 7% seems to apply well to benthicoxygen respiration in the SL system (Table 1), whereoxygen fluxes range between 2000 and 7000 mmol m22 d21

and where the average O2 penetration depth is ,7 mm(Katsev et al. 2007). We will therefore use this value in thefollowing discussion regarding the O-isotope budgets forrespiration in the LC. Nevertheless, it should be noted thatour model-based estimate of benthic O-isotope fraction-ation only applies to O2 consumption during oxic bacterialrespiration and does not consider O2 consumption throughoxidation of reduced compounds in the sediments otherthan carbon. Neither does it consider the possible effects ofbio-irrigation on the communication of the pore-water O2

Fig. 6. Plots of d18OO2vs. the natural logarithm of f, where f

is the residual substrate fraction after partial O2 consumption: (a)2006 data along the deep LC channel at sh 5 27.3; (b) 2007 datafrom water-column profiles at depths of .50 m in water depth(see Fig. 5); (c) all 2006 and 2007 data combined (2006 data alsoinclude water depths not shown in [a]). The slope of the regressionlines indicates the apparent O-isotope effect eO-app (see textfor discussion).

Fig. 7. Bacterial production as a function of water depth inthe LSLE water column (Sta. 23).

2162 Lehmann et al.

18O enrichment into the water column. Bio-irrigation canrender the benthic system more open with respect to waterand O2 exchange. On the other hand, it has been shown fornitrate isotopes that diffusion limitation can still dominatethe benthic nitrate isotope fractionation behavior inenvironments in which bio-irrigation is an importantmechanism facilitating sediment–water solute exchange bynondiffusive processes (Lehmann et al. 2004).

Estimating the relative importance of benthic vs. water-column respiration from community O-isotope fraction-ation—The main finding of our analysis of the oxygenisotope data is the anomalously low degree of 18Oenrichment associated with O2 loss in the LSLE. We

attribute this lowered community O-isotope fractionationto the contribution of benthic respiration that occurs with alowered net O-isotope effect. Bottom water enters the LCthrough Cabot Strait, and as it travels landward, O2 iscontinuously removed by bacterial respiration both in thewater column and in the sediments. If we assume thatreplenishment of O2 through mixing with overlying watersis negligible (Benoit et al. 2006), the deep LSLE can beconsidered a quasi-closed system that is adequately welldescribed by Rayleigh-fractionation dynamics (Fig. 6).From the observed community O-isotope effect, eO-app,and the end-member water-column and benthic respirationO-isotope effects, we can estimate the relative contributionof the individual processes to the total ecosystem respira-

Fig. 8. 1D diffusion-reaction model-derived pore-water O2 concentration and d18OO2for

different benthic O2 respiration rates (upper panels) and variable bottom-water O2 concentrations(lower panels). The simulated sediment pore-water profiles, representing the bulk of naturallyoccurring sedimentary O2 flux regimes (280–12,000 mmol m22 d21), were used for computationof the benthic bacterial respiration O-isotope effects (eO-sed) listed in Table 1.

Respiration in the St. Lawrence Estuary 2163

tion according to the following:

eO-app~eO-sed|RSedzeO|RWC

RSed|RWCð3Þ

where RSed and RWC are the respective rates of benthic andwater-column respiration and eO-sed and eO are theassociated end-member O-isotope effects. Assuming valuesof 6.8% and 18% for eO-sed and eO, respectively, and giventhe observed eO-app of 10.8% (Fig. 6), the ratio of water-column to benthic respiration (RWC : RSed) is approximately1 : 2. In other words, the O-isotope approach indicates that64% of the ecosystem respiration in the deeper LSLE is dueto O2 consumption at and below the SWI, with theremainder of the O2 being respired within the water column.

The estimated RWC : RSed ratio hinges on the valuesassigned to eO-sed and eO, the two end-member O-isotopeeffects. The model-derived eO-sed value does not considersediment oxygen demand (SOD) other than by aerobicbacterial O2 respiration of organic matter. Reduced dis-solved compounds (e.g., ammonium, sulfides, Fe2+ andMn2+), produced diagenetically in the sub- and anoxicsediment zones, may diffuse to, and be oxidized in, the oxiclayer. In organic-rich coastal environments, their oxidationmay contribute significantly to the total SOD (Soetaert etal. 1996), but both the intrinsic O-isotope effects associatedwith these reactions as well as their possible expression inthe water column are unknown. In the LC, however, anoxicprocesses do not seem to significantly contribute to theSOD (Silverberg et al. 1987; Colombo et al. 1996; Anschutzet al. 2000). For example, Anschutz et al. (2000) concludedthat most of the Fe2+ and Mn2+ diffusing upward from thesub-oxic sediments of the LSLE is oxidized by nitraterather than by O2. Nevertheless, regardless of what ourmodel simulations indicate, even bacterial OM respirationwithin sediments may not always produce the sameO-isotope signature in the overlying waters. Physical andbiogeochemical processes neglected in our model (e.g., therespiration in high-reactivity microsites) may combine toyield sedimentary O-isotope effects lower than 7%. In our

end-member approach, the use of a eO-sed value of 3%instead of 7% (as suggested by Brandes and Devol [1997])would reduce the calculated contribution of sedimentrespiration by ,17% of the total community respiration,rendering water-column respiration slightly more impor-tant than benthic respiration. More experimental con-straints on the benthic respiration O-isotope effect areneeded to verify our model-based estimates of eO-sed.

eO can also be expected to display some natural variability(Bender 1990; Luz et al. 2002; Quinones-Rivera et al. 2007).eO can vary, because bacterial respiration is not the onlycomponent of the total community oxygen demand in thewater column. Community respiration includes heterotro-phic respiration from bacteria, protozoa, micro- andmesozooplankton, and higher animals (e.g., fish andinvertebrates) as well as phytoplankton (autotrophic)respiration, which we consider to be negligible in the deepGulf and LSLE. Heterotrophic respiration by protozoa andzooplankton discriminates against 18O on a similar orslightly higher level, compared to our assumed value of eO

(18%) for bacterial water-column respiration (Kiddon et al.1993). However, its effect on the d18O of the O2 pool below150 m of an estuarine environment should be minimal (Joiriset al. 1982; Harvey et al. 2009). Finally, eO may not representthe biological O-isotope discrimination at the organismlevel. Respiration in small sinking organic particles couldresult in the development of microenvironments, in which O2

limitation may cause the underexpression of the biologicalO-isotope fractionation (i.e., eO values significantly smallerthan 18%). For larger aggregates, however, it can beassumed that their sinking velocities (i.e., the fluid flowthrough aggregates) are large enough (Ploug et al. 2002) toprevent oxidant limitation. Without knowing the sizedistribution of sinking particles, the degree of underexpres-sion of the biological O-isotope effect due to particle-associated respiration is difficult to constrain.

Whereas eO has not been characterized specifically forthe LSLE, it is fair to say that given previous reports of theaverage fractionation by heterotrophic respiration (Kiddon

Table 1. Oxygen flux, oxygen penetration depth, and sedimentary respiration O-isotopeeffect eO-sed as function of bottom-water oxygenation and respiration rate. The organism-levelO-isotope effect for bacterial respiration (eO) is set at 18%.

Bottom-water[O2] (mmol L21)

Respiration ratecoefficient (d21)

O2 flux(mmol O2 d21 m22)

O2 penetrationdepth (cm) eO-sed (%)

300 0.5 25249 1.75 6.6200 0.5 24205 1.45 6.7100 0.5 22827 1.25 6.950 0.5 21834 1.05 7.020 0.5 2950 0.85 7.010 0.5 2537 0.65 6.85 0.5 2288 0.55 6.7

100 0.01 2391 8.25 7.0100 0.1 21251 2.65 7.1100 0.25 21990 1.75 7.0100 1 24012 0.85 7.0100 5 210,029 0.45 7.0100 10 211,687 0.35 5.0

Average 6.860.5

2164 Lehmann et al.

et al. 1993), and assuming that oxidant limitation in sinkingaggregates does not play a very important role, an eO valuelower than 15% seems very unlikely for deep waters in theGulf and the LSLE. Values of 15% and 21%, for example,would yield slightly different relative contributions forbenthic respiration in the LSLE (51% and 72%, respec-tively, instead of 64%). Regardless, the O-isotope dataindicate that benthic respiration is the dominant O2

consumption process in the LSLE. Even when calculatedconservatively, it contributes to more than half, andprobably two thirds, of the total community respiration.

Openness of the system—Previous studies have demon-strated that open-system aspects and diffusion have a first-order influence on the distribution of isotope tracers in theocean or other aquatic environments (Bender 1990; Quay etal. 1993; Lehmann et al. 2003). The suitability of theRayleigh approach to calculating eO-app (i.e., the basis forthe estimation of RWC : RSed) depends on rates of O2

replenishment by O2 production through photosynthesis orvertical mixing. Openness of the system and diffusion tendto lower the apparent community O-isotope effect, eO-app,and could contribute to the underexpression of thebiological O-isotope effect, eO, in the water column (Bender

1990; Levine et al. 2009). The contribution of photosyn-thetic oxygen production can be excluded for the deepwaters far below the photic zone, and eO-app calculationsfrom water samples along the deep LC should, in thiscontext, not be biased (Fig. 6b). As to O2 recharge bymixing, we can probably assume that vertical transport ofoxygen across the pycnocline is negligible relative to thealong-channel advection and community O2 consumptionrates (Benoit et al. 2006; see below). If open-system ratherthan closed-system conditions applied, we would expect anonlinear relationship between d18O and ln f, particularlyin the concentration range that corresponds to low valuesof f. Both the Rayleigh and open-system models represent asimplification of nature, and slow vertical diffusion of O2

must occur to some extent, so that strict closed-systemconditions do not fully apply for the deep LSLE. Yetthe fact that closed-system Rayleigh dynamics almostperfectly predict the data indicates that mixing has littleeffect on our estimate of the community O-isotopefractionation eO-app.

In spite of the above conclusions and following theapproach of Quinones-Rivera et al. (2007), we evaluatedthe possible effect of gas exchange across the pycnocline onthe O-isotope dynamics in the LC bottom water by

Fig. 9. Measured (circles) and simulated (lines) relationship between oxygen saturation andthe d18O of dissolved O2. Full lines represent respiration-driven O-isotope dynamics in a closedsystem according to e values of 6.8% (100% benthic respiration), 18% (100% water-columnrespiration), and 10.8%, the O-isotope effect that describes best respiratory communityfractionation in the LSLE and that most probably results from the combined effects of benthicrespiration (64% 6 13%) and water-column respiration (36% 6 13%). Dashed and dotted linesrepresent open-system scenarios, in which oxygen respiration is partly (20%) compensated by theaddition of dissolved O2, with various d18O values for the input DO. It can be seen that gasexchange with water masses in the upper water column has not had a strong effect on therelationship between O2 saturation and the d18O for a given O-isotope effect. See textfor discussion.

Respiration in the St. Lawrence Estuary 2165

allowing 20% of the respired oxygen (for each increment ofcumulative respiration) to be added back into bottomwaters through mixing with surface or intermediate waters(d18O 5 21% and 8%, respectively). Figure 9 indicatesthat this moderate addition of O2 from either part of theupper water column does not strongly affect the O-isotopeenrichment in the deeper LSLE for a given O2 depletion(expressed as % O2 saturation). Hence, replenishment evenof strongly 18O-depleted oxygen by gas exchange cannotexplain the low apparent community isotope fractionationobserved in the LSLE. This confirms the validity of usingthe closed-system community effect eO-app to constrain therelative importance of water-column vs. benthic respira-tion.

Comparison with direct constraints on O2 respiration rates—Following the above arguments, benthic O2 consumptionemerges as the principal cause for the local O2 depletion in thedeep LSLE. We now compare our findings with measuredbacterial production rates at Sta. 23, published data onrespiratory electron transport system (ETS) activity in theGulf, and measured O2 fluxes at the SWI. Oxygen fluxes atthe SWI, calculated from onboard short-term (,24–36-h)sediment incubations or microelectrode measurements, in theLC vary between 3 and 11 mmol m22 d21 (Silverberg et al.1987, 2000; Katsev et al. 2007), with a conservative meanvalue of 4.7 mmol m22 d21 or 1715 mmol m22 yr21 for Sta.24 to Sta. 18. Measured BP rates for deep waters (.150 min water depth) ranged from 5 to 10 nmol C L21 d21 (Fig. 7).Using the previously calculated BP-based average O2

respiration rate of 33 nmol O2 L21 d21, the computed totalwater-column respiration rate per square meter, integratedover the 100 m of the water column above the sediment–waterinterface, is approximately 1200 mmol m22 yr21, or 41% ofthe total community oxygen depletion rate, in goodagreement with our estimate based on the O-isotope approach(36% 6 13%). We note, however, that the water-columnrespiration rate is based on a single BP profile and that theestimated integrated O2 respiration rate is sensitive to thechoice of the BGE coefficient as well as the dimension (i.e.,thickness) of the respiration zone. These uncertainties andothers (e.g., geometry, bathymetry, extrapolation of discreteflux measurements to the whole LSLE) propagate into arelatively large error with respect to the average benthic O2

consumption rate and its contribution to oxygen depletion inthe water column. Additional independent support of ourresults, however, comes from a study by Savenkoff et al.(1996). In this study, a comparison of ETS-based respirationrates in the water column and core incubation-based rates ofbenthic respiration between the Anticosti gyre and CabotStrait revealed that benthic respiration was the maincontributor to O2 consumption in the Gulf of St. Lawrence(between 59% and 83% of the total community respiration inthe lower 150 or 65 m of the water column, respectively).Thus, in spite of the uncertainties listed above, directconstraints on community respiration seem to validate ourO-isotope–derived estimate of the RWC : RSed ratio(0.36 : 0.64).

The O-isotope approach can constrain the relative ratesof benthic vs. water-column respiration but not their

absolute rates. The isotope results can be converted toabsolute oxygen respiration rates if one of the two principalO2 consumption pathways in the LSLE is quantified, but,as noted above, rate measurements based on discretesamplings carry large uncertainties and are subject tospatial and temporal variability. As is the case withO-isotope ratios, water-column oxygen concentrationmeasurements integrate processes over large spatial andtemporal scales and, thus, provide a more robust estimateof the mean estuary-wide respiration, at least with the rightcomplementary constraints (e.g., on the water residence timeand basin geometry). The deep-water dissolved O2 concen-trations close to Sta. 23 are approximately 110 mmol L21

lower than the equivalent (same isopycnal) values for deepwater at Sta. 16 (,170 mmol L21). Assuming that O2

depletion is solely due to bacterial respiration of OM withinthe deep estuary, that the travel time of water massesbetween Sta. 16 and Rimouski (Sta. 23) is approximately2 yr (1 cm s21; D. Gilbert unpubl.), and ignoring O2

replenishment by vertical diffusion, we calculate an averagecommunity bacterial respiration rate of 5500 mmol O2

m22 yr21 for the bottom 100 m of the water column.According to our estimated range for the partitioningbetween benthic and water-column respiration, approxi-mately 3520 mmol O2 m22 yr21 (between 2850 m22 yr21

for 51% benthic respiration and 3960 m22 yr21 for 72%benthic respiration) of this total O2 consumption rate is dueto bacterial respiration within the benthic environment,whereas the remainder (between 1540 and 2650 mmol O2

m22 yr21, respectively) occurs in the water column. Inaddition to the single BP profile obtained in this study, we donot have reliable direct constraints on the ecosystem water-column respiration rate. The integrated rate of benthic O2

consumption falls, however, within the range of sedimentaryO2 fluxes in the LSLE (985–4197 mmol O2 m22 yr21;Silverberg et al. 1987, 2000; Katsev et al. 2007) confirmingthat hypoxia in the deep waters of the LSLE is, in large part,the result of benthic respiration.

The degree of oxygen isotope enrichment (eO-app)associated with O2 depletion in the LSLE is low comparedto the canonical O-isotope effect (eO) of 18% for bacterialwater-column respiration. The low value for eO-app can bestbe explained by the contribution of benthic O2 consump-tion, occurring with a strongly reduced isotope effect at thescale of sediment–water exchange. We have shown with a1-D diffusion-reaction model that the apparent O-isotopeeffect for benthic oxygen respiration is on the order of 7%and largely independent of environmental conditions (O2

consumption rates) typically encountered in coastalmarine sediments. Based on the end-member O-isotopeeffects for benthic (6.8%) and water-column (18%)respiration and the observed community O-isotope effectof 10.8% we calculated that between one half and threequarters of the ecosystem respiration in the deep LSLEoccurs in the sediments, with the remainder taking placein the water column, a finding that is in good agreementwith results from direct measurements.

The RWC : RSed ratio and the calculated communityrespiration rate derived in this study apply only to the

2166 Lehmann et al.

bottom 100 m of the water column—the low-oxygen zone.The whole-ecosystem respiration rate and RWC : RSed ratioare significantly different, because pelagic O2 respirationrates (heterotrophic and autotrophic) are much higher inthe upper water column. Moreover, our RWC : RSed ratiorepresents an average value for the LC, as we expect it to bequite variable both in time and space. In this context, ourdata set also begs the question of why we see no along-channel and/or depth-dependent variation in the commu-nity O-isotope effect, although the RWC : RSed ratio can beexpected to vary as a function of basin geometry (i.e.,changes in the sediment area to water volume ratio). We arenot able to adequately explain how multiple O2-consumingprocesses produce [O2] and d18OO2

data that fall on a single,perfect Rayleigh line. The simplest explanation would bethat the end-member O-isotope effects for both benthic andwater-column respiration are in fact similar to each other,that is, close to 11%. However, following the discussionabove, and based on existing field and model data, we tendto discard this possibility.

In spite of some uncertainties, the integrative O-isotopeapproach used to assess community respiration is apowerful tool, the sensitivity of which increases with theisotopic difference between the end-member O-isotopeeffects. Discrete measurements of bacterial production inthe water column and sedimentary O2 fluxes are inherentlymore direct, but they are hard to extrapolate to the scale ofthe estuary because of spatial and/or temporal variability.In contrast, the oxygen isotope systematic in the deep watercolumn of the LSLE and the Gulf may be more ambiguouswith respect to the actual biogeochemical processes atwork, but the isotopes integrate processes over large scalesof time and space and thus provide valuable complemen-tary constraints. The uncertainty in the calculation ofRWC : RSed using eO-app arises directly from uncertainties inthe end-member O-isotope effects for individual processesthat consume oxygen. In this study, neither eO nor eO-sed

was measured directly, and future work should aim atconstraining the O-isotope effects that are intrinsic to theSL system. In the water column, abiotic mechanisms mayreduce the O-isotope effect for respiration (e.g., diffusivemixing in the water column or diffusion limitation insinking particles). In the sediments, eO-sed may be lowerthan the modeled value (7%) if bacterial O2 respirationoccurs primarily in reactive microsites (Brandes and Devol1997) or higher than the modeled value if sediments areheavily bio-irrigated through nondiffusive exchange ofsolutes between the sediment pore waters and the overlyingwater. Furthermore, bacterial mediation may not be theonly cause of benthic O-isotope fractionation. Whereasprevious work in the LC indicates that the aerobicoxidation of organic material outweighs other diageneticreactions that contribute to the total sediment oxygendemand (oxidation of sulfides, Fe2+ and Mn2+), we cannotrule out that such reactions occur. The specific O-isotopeeffects of these oxidation processes and their propagationinto the water column are unknown, and thus the influenceof these alternative benthic O2 sinks on the communityO-isotope effect eO-app and, in turn, our estimate of theRWC : RSed ratio in the LSLE cannot readily be ascertained.

In the absence of a more reliable representation of thewhole-community (diagenetic) sedimentary O-isotope ef-fect, we conclude that aerobic bacterial respiration in thebenthic environment is more important than water-columnrespiration in maintaining hypoxic conditions in thebottom waters of the LSLE.

AcknowledgmentsWe thank the captain and crew of the Coriolis II for their

technical assistance during the sampling campaigns. We also thankT. Kana and an anonymous reviewer for their helpful sugges-tions. Figure 1 was prepared using Online Map Creation (OMC) byM. Weinelt. Figures 2 and 3 were prepared using Ocean Data View.This study was funded by the Natural Sciences and EngineeringResearch Council of Canada (NSERC) through Strategic Re-search, Ship-Time and Discovery grants to M.F.L, Y.G., A.M., andB.S.

References

ANGERT, A., AND B. LUZ. 2001. Fractionation of oxygen isotopesby root respiration: Implications for the isotopic compositionof atmospheric O2. Geochim. Cosmochim. Acta 65: 1695–1701.

ANSCHUTZ, P., B. SUNDBY, L. LEFRANCOIS, G. W. LUTHER, AND A.MUCCI. 2000. Interactions between metal oxides and species ofnitrogen and iodine in bioturbated marine sediments.Geochim. Cosmochim. Acta 64: 2751–2763.

BENDER, M. L. 1990. The d18O of dissolved O2 in seawater—aunique tracer of circulation and respiration in the deep sea. J.Geophys. Res. Oceans 95: 22243–22252.

———, P. P. TANS, J. T. ELLIS, J. ORCHARDO, AND K. HABFAST.1994. A high-precision isotope ratio mass-spectrometrymethod for measuring the O2/N2 ratio of air. Geochim.Cosmochim. Acta 58: 4751–4758.

BENOIT, P., Y. GRATTON, AND A. MUCCI. 2006. Modeling ofdissolved oxygen levels in the bottom waters of the Lower St.Lawrence Estuary: Coupling of benthic and pelagic processes.Mar. Chem. 102: 13–32.

BRANDES, J. A., AND A. H. DEVOL. 1995. Simultaneous nitrate andoxygen respiration in coastal sediments—evidence for discretediagenesis. J. Mar. Res. 53: 771–797.

———, AND ———. 1997. Isotopic fractionation of oxygen andnitrogen in coastal marine sediments. Geochim. Cosmochim.Acta 61: 1793–1801.

———, AND ———. 2002. A global marine-fixed nitrogenisotopic budget: Implications for Holocene nitrogen cycl-ing. Glob. Biogeochem. Cycles 16: 1120, doi: 10.1029/2001GB001856.

COLOMBO, J. C., N. SILVERBERG, AND J. N. GEARING. 1996.Biogeochemistry of organic matter in the Laurentian Trough.2. Bulk composition of the sediments and relative reactivity ofmajor components during early diagenesis. Mar. Chem. 51:295–314.

DEL GIORGIO, P. A., AND J. J. COLE. 1998. Bacterial growthefficiency in natural aquatic systems. Ann. Rev. Ecol. Syst. 29:503–541.

EMERSON, S., P. D. QUAY, C. STUMP, D. WILBUR, AND R.SCHUDLICH. 1995. Chemical tracers of productivity andrespiration in the subtropical Pacific Ocean. J. Geophys.Res. Oceans 100: 15873–15887.

GALBRAITH, P. S. 2006. Winter water masses in the Gulf ofSt. Lawrence. J. Geophys. Res. 111: C06022, doi: 10.1029/2005JC003159.

Respiration in the St. Lawrence Estuary 2167

GILBERT, D., AND B. PETTIGREW. 1997. Interannual variabi-lity (1948–1994) of the CIL core temperature in theGulf of St. Lawrence. Can. J. Fish. Aquat. Sci. 54: 57–67.

———, B. SUNDBY, C. GOBEIL, A. MUCCI, AND G. H. TREMBLAY.2005. A seventy-two-year record of diminishing deep-wateroxygen in the St. Lawrence Estuary: The northwest Atlanticconnection. Limnol. Oceanogr. 50: 1654–1666.

GRASSHOFF, K., M. EHRHARDT, AND K. KREMLING. 1983. Methodsof seawater analysis. Verlag Chemie.

GUY, R. D., M. L. FOGEL, AND J. A. BERRY. 1993. Photosyntheticfractionation of the stable isotopes of oxygen and carbon.Plant Physiol. 101: 37–47.

HARVEY, M., P. GALBRAITH, AND A. DESCROIS. 2009. Verticaldistribution and diel migration of macrozooplankton in theSt. Lawrence marine system (Canada) in relation with thecold intermediate layer thermal properties. Prog. Oceanogr.80: 1–21.

HEDGES, J. I., J. A. BALDOCK, Y. GELINAS, C. LEE, M. L.PETERSON, AND S. G. WAKEHAM. 2002. The biochemical andelemental compositions of marine plankton: A NMRperspective. Mar. Chem. 78: 47–63, doi: 10.1016/S0304-4203(02)00009-9.

JOIRIS, C., AND oTHERS. 1982. A budget of carbon cycling in theBelgian coastal zone—relative roles of zooplankton, bacter-ioplankton and benthos in the utilization of primaryproduction. Neth. J. Sea Res. 16: 260–275.

KATSEV, S., G. CHAILLOU, B. SUNDBY, AND A. MUCCI. 2007. Effectsof progressive oxygen depletion on sediment diagenesis andfluxes: A model for the Lower St. Lawrence River Estuary.Limnol. Oceanogr. 52: 2555–2568.

KIDDON, J., M. L. BENDER, J. ORCHARDO, D. A. CARON, J. C.GOLDMAN, AND M. DENNETT. 1993. Isotopic fractionation ofoxygen by respiring marine organisms. Glob. Biogeochem.Cy. 7: 679–694.

KIRCHMAN, D., E. KNEES, AND R. HODSON. 1985. Leucineincorporation and its potential as a measure of protein-synthesis by bacteria in natural aquatic systems. Appl.Environ. Microbiol. 49: 599–607.

LEHMANN, M. F., P. REICHERT, S. M. BERNASCONI, A. BARBIERI,AND J. A. MCKENZIE. 2003. Modelling nitrogen and oxygenisotope fractionation during denitrification in a lacustrineredox-transition zone. Geochim. Cosmochim. Acta 67:2529–2542.

———, D. M. SIGMAN, AND W. M. BERELSON. 2004. Coupling theN15/N14 and O18/O16 of nitrate as a constraint on benthicnitrogen cycling. Mar. Chem. 88: 1–20.

———, ———, D. C. MCCORKLE, J. GRANGER, S. HOFFMANN, G.CANE, AND B. G. BRUNELLE. 2007. The distribution of nitrate15N/14N in marine sediments and the impact of benthicnitrogen loss on the isotopic composition of oceanic nitrate.Geochim. Cosmochim. Acta 71: 5384–5404.

———, AND oTHERS. 2005. Origin of the deep Bering Sea nitratedeficit: Constraints from the nitrogen and oxygen isotopiccomposition of water column nitrate and benthic nitratefluxes. Glob. Biogeochem. Cy. 19: GB4005, doi: 10.1029/2005GB002508.

LEVINE, N. M., M. L. BENDER, AND S. C. DONEY. 2009. The d18Oof dissolved O2 as a tracer of mixing and respiration in themesopelagic ocean. Glob. Biogeochem. Cy. 23: GB1006, doi:10.1029/2007GB003162.

LUZ, B., E. BARKAN, Y. SAGI, AND Y. Z. YACOBI. 2002.Evaluation of community respiratory mechanisms withoxygen isotopes: A case study in Lake Kinneret. Limnol.Oceanogr. 47: 33–42.

MARIOTTI, A., J. C. GERMON, P. HUBERT, P. KAISER, R. LETOLLE,R. TARDIEUX, AND P. TARDIEUX. 1981. Experimental determi-nation of nitrogen kinetic isotope fractionation: Someprinciples; Illustration for the denitrification and nitrificationprocesses. Plant Soil 62: 413–430.

PARKER, S. R., S. R. POULSON, C. H. GAMMONS, AND M. D.DEGRANDPRE. 2005. Biogeochemical controls on diel cyclingof stable isotopes of dissolved O2 and dissolved inorganiccarbon in the Big Hole River, Montana. Environ. Sci.Technol. 39: 7134–7140.

PLOUG, H., S. HIETANEN, AND J. KUPARINEN. 2002. Diffusion andadvection within and around sinking, porous diatom aggre-gates. Limnol. Oceanogr. 47: 1129–1136.

QUAY, P. D., S. EMERSON, D. O. WILBUR, C. STUMP, AND M. KNOX.1993. The d18O of dissolved O2 in the surface waters of thesub-arctic Pacific—a tracer of biological productivity. J.Geophys. Res. Oceans 98: 8447–8458.

QUINONES-RIVERA, Z. J., B. WISSEL, D. JUSTIC, AND B. FRY. 2007.Partitioning oxygen sources and sinks in a stratified,eutrophic coastal ecosystem using stable oxygen isotopes.Mar. Ecol. Prog. Ser. 342: 69–83.

RAVEN, J., AND J. BEARDALL. 2005. Respiration in aquaticphotolithotrophs, p. 36–46. In P. A. delGiorgio and P. J. deB. Williams [eds.], Respiration in aquatic ecosystems. OxfordUniv. Press.

REICHERT, P. 1994. AQUASIM—a tool for simulation and dataanalysis of aquatic systems. Water Sci. Technol. 30: 21–30.

———. 1998. AQUASIM 2.0—user manual, technical report.Swiss Federal Institute for Environmental Science andTechnology (EAWAG).

SAUCIER, F. J., F. ROY, D. GILBERT, P. PELLERIN, AND H. RITCHIE.2003. Modeling the formation and circulation processes ofwater masses and sea ice in the Gulf of St. Lawrence, Can-ada. J. Geophys. Res. Oceans 108: 3269, doi: 10.1029/2000JC000686.

SAVENKOFF, C., AND oTHERS. 1996. Distributions of oxygen,carbon, and respiratory activity in the deep layer of the Gulfof St. Lawrence and their implications for the carbon cycle.Can. J. Fish. Aquat. Sci. 53: 2451–2465.

SIGMAN, D. M., R. ROBINSON, A. N. KNAPP, A. VAN GEEN, D. C.MCCORKLE, J. A. BRANDES, AND R. C. THUNELL. 2003.Distinguishing between water column and sedimentarydenitrification in the Santa Barbara Basin using the nitrogenisotopes of nitrate. Geochem. Geophys. Geosys. 4: 1040, doi:10.1029/2002GC000384.

SILVERBERG, N., J. BAKKER, H. M. EDENBORN, AND B. SUNDBY.1987. Oxygen profiles and organic-carbon fluxes in Laur-entian Trough sediments. Neth. J. Sea Res. 21: 95–105.

———, AND oTHERS. 2000. Remineralization of organic carbon ineastern Canadian continental margin sediments. Deep-SeaRes. II 47: 699–731.

SIMON, M., AND F. AZAM. 1989. Protein content and proteinsynthesis rates of planktonic marine bacteria. Mar. Ecol.Prog. Ser. 51: 201–213.

SMITH, D. C., AND F. AZAM. 1992. A simple, economical methodfor measuring bacterial protein synthesis rates in seawaterusing 3H-leucine. Mar. Microb. Food Webs 6: 107–114.

SOETAERT, K., P. M. J. HERMAN, AND J. J. MIDDELBURG. 1996.A model of early diagenetic processes from the shelfto abyssal depths. Geochim. Cosmochim. Acta 60: 1019–1040.

THIBODEAU, B., A. DE VERNAL, AND A. MUCCI. 2006. Recenteutrophication and consequent hypoxia in the bottom watersof the Lower St. Lawrence Estuary: Micropaleontological andgeochemical evidence. Mar. Geol. 231: 37–50.

2168 Lehmann et al.

TOBIAS, C. R., J. K. BOHLKE, AND J. W. HARVEY. 2007. The oxygen-18 isotope approach for measuring aquatic metabolism in high-productivity waters. Limnol. Oceanogr. 52: 1439–1453.

VOSS, M., J. W. DIPPNER, AND J. P. MONTOYA. 2001. Nitrogenisotope patterns in the oxygen-deficient waters of the EasternTropical North Pacific Ocean. Deep-Sea Res. I 48: 1905–1921.

Associate editor: John Albert Raven

Received: 27 March 2009Accepted: 02 July 2009Amended: 21 July 2009

Respiration in the St. Lawrence Estuary 2169