Effect of unsteady flow on nitrate loss in an oligotrophic, glacial

meltwater stream

Joshua C. Koch,1 Diane M. McKnight,1 and Jenny L. Baeseman2

Received 13 April 2009; revised 21 September 2009; accepted 22 September 2009; published 8 January 2010.

[1] The McMurdo Dry Valleys of Antarctica are among the coldest, driest ecosystemson Earth. During the austral summer, glacial meltwater supports cyanobacterialmat communities in some streams, but they are not ubiquitous. We conducted a nitrate(NO3

�) enrichment tracer injection in Huey Creek to quantify NO3� loss in a Dry Valley

stream where algal mats would not obscure hyporheic microbial processes. Unsteadystreamflow led to diel variability in the tracer concentration and in surface/subsurfacewater and solute exchange. Subsequently, concentrations of NO3

�, nitrite (NO2�),

ammonium (NH4+), and dissolved organic carbon (DOC) varied significantly during the

injection, with a net loss of NO3�, NO2

�, and DOC, and production of nitrous oxide. Thesemass changes within a reach were often coincident with high streamflows. Reactivityalso coincided with the highest DOC concentrations, suggesting that DOC is the primarylimitation to heterotrophic microbial activity in the stream. Together, streamflow and DOCavailability create the hot spots and hot moments that dominate NO3

� reactivity andremoval in this polar desert ecosystem. The combination of spatially and temporallyvariable hyporheic dynamics and solute availability underscore the limitations of commonnutrient uptake metrics and transient storage models when unsteady flow conditions exist.

Citation: Koch, J. C., D. M. McKnight, and J. L. Baeseman (2010), Effect of unsteady flow on nitrate loss in an oligotrophic, glacial

meltwater stream, J. Geophys. Res., 115, G01001, doi:10.1029/2009JG001030.

1. Introduction

[2] Recent studies in stream ecosystem dynamics havedemonstrated the importance of spatial and temporal vari-ability of biogeochemical activity [Harms and Grimm,2008; McClain et al., 2003]. Hot spots and hot momentsare the locations and periods in which a number of variablescome together to allow elevated rates of biogeochemicalactivity. In semiarid riparian systems, hot spots and hotmoments are crucial to the cycling of carbon (C) and nitrogen(N). Such events are often tied to hydrology and the shorttimescales on which rainstorms, ephemeral inflows, andflooding events occur. Unsteady flow can influence exchangerates of water and solutes across the sediment/water interface[Harvey et al., 1996], which affects the distribution of hotspots and/or hot moments in lotic ecosystems [Holmes et al.,1998; Jones et al., 1995; Valett et al., 2005]. In this paperwe expand the hot spot/hot moment concepts to includesemiarid streams in polar deserts.[3] Hot spots and hot moments in streams are particularly

important in the hyporheic zone, an area proximal to thestream in which surface and subsurface waters mix, result-ing in a region of strong physical gradients in light, velocity,and temperature, and in chemical gradients of oxygen content

and subsequently redox potential. Hyporheic storage size andexchange rates can be spatially variable, due to variability inthe relative fluxes of groundwater return flows and fine-scalevertical hydraulic gradients in the stream bottom [Harvey andBencala, 1993]. The resultant flux of water and solutes acrossthe boundary between the stream and subsurface create adynamic setting in which chemical reactants and productsmay be delivered and removed from reaction sites. For thisreason, storage and exchange of hyporheic waters oftenstrongly influence rates of biogeochemical processes [Brunkeand Gonser, 1997; Dahm et al., 1998; Dent et al., 2007;Gooseff et al., 2004b; McKnight et al., 2004; Miller et al.,2006].[4] The McMurdo Dry Valleys of Antarctica are con-

ducive to measuring hyporheic parameters via tracer studiesdue to low streamflows, a lack of regional groundwater flow,absence of terrestrial vegetation, high hydraulic conductivity,and no human impacts. Tracer experiments have been con-ducted in this environment to study the potential for sub-surface storage [Gooseff et al., 2003; Runkel et al., 1998], toelucidate sources and sinks of solutes and nutrients [Gooseffet al., 2002; Joslin, 2005; McKnight et al., 2004], and toassess the retention and processing of NO3

� by cyanobacterialmats [Gooseff et al., 2004b; McKnight et al., 2004].[5] Nitrogen cycling is of particular interest in the

McMurdo Dry Valleys for many reasons. Due to atmo-spheric production of NO3

� by auroral activity and theabsence of vegetation, the McMurdo Dry Valleys arenaturally N-saturated, with high concentrations of NO3

� onvalley surfaces [Green et al., 1988] and in stream waters

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 115, G01001, doi:10.1029/2009JG001030, 2010ClickHere

for

FullArticle

1Institute of Arctic and Alpine Research, University of Colorado atBoulder, Boulder, Colorado, USA.

2International Arctic Research Center, University of Alaska Fairbanks,Fairbanks, Alaska, USA.

Copyright 2010 by the American Geophysical Union.0148-0227/10/2009JG001030$09.00

G01001 1 of 15

and the hyporheic zone [McKnight et al., 2004]. Under-standing ecosystem response to the high N loading in thissystem may provide insights into the many regions experi-encing similar anthropogenic N loading. McKnight et al.[2004] showed that NO3

� concentrations are higher in DryValley streams without mats, resulting in greater NO3

�

delivery to the lake ecosystems from streams such as HueyCreek. Reduced N species (i.e., NH4

+ and N2O) are detectedin mat-less streams [McKnight et al., 2004], suggesting thatsome amount of NO3

� loss may occur even in the absenceof cyanobacterial mats.[6] The goal of this study was to quantify the relative

importance of several N redox pathways in a Dry Valleystream devoid of microbial mats, where periphyton uptakewould not obscure hyporheic microbial processes. An injec-tion of greater than 24 h was performed in order to witnessthe potential effects of diel flow variability on hyporheicdynamics and microbial activity. This is the first nutrientenrichment tracer experiment in an Antarctic stream withoutvisible microbial mats and may provide new insights intomicrobial activity in one of the harshest environments on theplanet.

2. Site Description

[7] While 98% of Antarctica is covered by ice sheets, theMcMurdo Dry Valleys are representative of the ice-freedesert oases on the coasts of the continent. The valleys arerelatively free of snow and ice, receiving an average of lessthan ten centimeters of precipitation annually as snow[Doran et al., 2002]. Snow collects in depressions andentrenched stream channels during the winter months, andmelts at lower elevations during the austral summer. Ninety-five percent of the valley surfaces are composed of arid soils[Burkins et al., 2001] derived from tills composed of granites,sandstones, dolerites, and meta-sedimentary rocks. There isno terrestrial vegetation, except for mosses near some streamchannels and in wetted areas surrounding lakes and ponds. Ahalf meter below the surface the ground is permanentlyfrozen. Streams typically flow for up to 12 weeks in theaustral summer [McKnight et al., 1999], during which timethe sun is above the horizon 24 h a day. Water is derived fromglacial melt [Fountain et al., 1999] and follows establishedchannels into perennially ice-covered lakes on the valleyfloors. Stream solutes are weathered and leached from sedi-ments in the hyporheic zones [Gooseff et al., 2002; Green etal., 1988]. Streamflow fluctuates on a daily timescale withpulses of high flow generated when the sun shines directly onglacier surfaces [Conovitz et al., 1998]. The parafluvial zoneis defined as the wetted perimeter around streams and isvisually evident in the Dry Valleys, due to the lack ofvegetation and soils. Water is evaporated from the streamand parafluvial zone [Cozzetto et al., 2006; Gooseff et al.,2003]. Evaporation concentrates and precipitates dissolvedsolutes, leading to salt accumulation on the stream bank thatmark the furthest lateral extent of the parafluvial zone. Thesedynamics result in large streamflow and aqueous soluteconcentration fluctuations on a diel timescale.[8] Dry Valley streams are useful model systems to study

interactions between microbial communities and nutrients,specifically N. Sources of N to the stream include atmo-spheric deposition, and leaching of N from hyporheic and

parafluvial sediments. Many Dry Valley streams containmicrobial mats composed primarily of filamentous cyano-bacteria and diatoms. Streams with cyanobacterial mats havelower concentrations of NO3

� relative to mat-less streams,subsequently delivering smaller N yields to the lake eco-systems [McKnight et al., 2004]. Once in the stream, Nconcentrations are affected by storage in interstitial water inthe cyanobacterial mats and the subsurface, assimilation bycyanobacterial mats and microbes, and loss of N duringbiogeochemical reactions to gaseous species such as N2Oand N2. Stream-scale experimental studies indicate thatdecreased NO3

� concentrations in streams with cyanobacte-rial mats result from uptake and denitrification within thecyanobacterial mats [Gooseff et al., 2004b]. A NO3

� injectionin an mat-rich stream resulted in an approximate 20% loss ofNO3� attributable to dissimilatory reduction [McKnight et al.,

2004]. Whereas most of this activity occurred while waterwas stored within the cyanobacterial mats, 7–16% of NO3

�

uptake was due to microbial processes in the hyporheic zone.A NO2

� pulse accompanied this NO3� injection.Gooseff et al.

[2004b] inferred that denitrification was occurring, but notproceeding to completion due to either transport or reductaselimitation.[9] Huey Creek is a stream in the McMurdo Dry Valleys

with a steep slope. This stream is characterized by a shallowchannel with a narrow floodplain that is incised below thevalley fill level. A previous tracer injection performed onHuey Creek quantified hyporheic exchange rates rangingfrom 4.67 E�4 to 1.62 E�2 s�1, which are high relative tomany small streams, and storage zone areas ranging from0.8 to 3.07 m2 [Runkel et al., 1998]. There is a 182 m streamreach that is characterized by a shallow stream slope. Duringthe daily high flow period this reach is dominated byanabranches, which are defined as multiple channels sepa-rated by semipermanent alluvial islands [Nanson andKnighton, 1996]. These branches dry when the daily floodpulse recedes. Anabranch formation often occurs in streamswith flood-dominated flow regimes, and likely occurs at thislocation in Huey Creek due to the decrease in stream slopewithin this reach. The anabranches in Huey Creek aredecimeters wide, nonmigratory braids, separated by gravelbars that route water laterally away from the main channel.The length of the anabranches ranges from meters to tens ofmeters.Water and solute storage in the surface and subsurfaceof the anabranching reach may have important implicationsfor the timing and location of microbial activity in thissystem. While no mats were seen during our injection, sparsepopulations of microbial mats have been observed in distalanabranches of Huey Creek in previous studies [Alger et al.,1997]. These mats appeared predominantly in the far-lateralanabranches. Their absence in other stream reaches ofHuey Creek may be related to the steep stream slope, whichresults in high water velocities, turbulent flow, an unstablesandy substrate, and overall an environment too harsh tosupport perennial cyanobacterial mats [Alger et al., 1997;Esposito et al., 2006].

3. Methods

3.1. Sample Collection

[10] NaCl, LiCl, and NaNO3 were injected into HueyCreek for 31 h. Solute injection began at 7 P.M. on 9 January

G01001 KOCH ET AL.: UNSTEADY FLOW AND NITRATE LOSS

2 of 15

G01001

2006 (t = 0 h) on the tail of the daily flood pulse, and endedat 2 A.M. on 11 January 2006 (t = 31 h). A multiple dayinjection in Huey Creek was employed to capture dielvariability in flow, and the subsequent potential variabilityin solute storage and biological activity. The injection sitewas located in a steep, narrow reach, which facilitated mixingof the tracer in the stream. Above this point the stream flowed

beneath remnant winter snowdrifts for much of its length.Five sampling sites that reflected clear changes in thedominant stream morphology were selected to collect waterchemistry. These five sites (Figure 1) were labeled A throughE, and are 16, 123, 305, 451, and 665 m downstream of theinjection, respectively, as determined by a GPS survey.Samples were also collected directly above the injection, at

Figure 1. (a) The Lake Fryxell Basin with the tracer length indicated by the bracket. The inset showsthe relative location of the McMurdo Dry Valleys in Antarctica. (b) An aerial view of Huey Creekdisplaying the tracer reach. Sampling sites are indicated with a letter and distance from the injection site.

G01001 KOCH ET AL.: UNSTEADY FLOW AND NITRATE LOSS

3 of 15

G01001

a site labeled site I. Reaches 1, 2, 3, and 4 were defined asthe length of stream bounded by sites A and B, B and C,C and D, and D and E, respectively. Reach 2, betweensites B and C, displayed significant anabranching duringhigh flows. Stream stage in Huey Creek was measured at agaging station located just downstream of site E. Stagewas recorded every 15 min by a pressure transducer andconverted to discharge using a stage-discharge relationshipdeveloped at the gaging station [Von Guerard et al., 1995].[11] Anion, cation, and nutrient samples were collected at

each site every 5 min during the first and last hours of theinjection, and once every hour during the 31 h of steadyinjection. Samples were collected every 2 h at site I, justabove the injection pump, to monitor background streamchemistry. After the injection, samples were collected att = 33, 34, 35, 36, 38, 40, 44, 48, 52, and 56 h. DOC and N2Osamples were collected at sites I, B, D, and E at t = 0, 0.5, 1, 2,6, 12, 18, 24, 36, 48, and 52 h. All sites were sampled twiceduring the hour preceding the start of the tracer injection.Stream samples were collected using Geopump peristalticpumps, and filtered through a 147 mm diameter 0.45 mmfilter. The filters were rinsed with stream water immediatelybefore sample collection to minimize carryover from theprevious sample. Samples for nutrient analysis were frozenusing liquid N within 4 h of collection and remained frozenuntil analysis. All other samples were stored in the dark at4�C or less in the field and at the Crary Lab at McMurdoStation. Anions were analyzed on a Metrohm 761 CompactIon Chromatograph and NH4

+ on an OI Analytical FlowSolution IV Spectrophotometric Analyzer at the KiowaLaboratory at INSTAAR’s Mountain Research Station. Pre-cision for Cl�, NO2

�, NO3�, and NH4

+ was 1.98, 0.60, 0.81,and 0.55%RSD, respectively, with detection limits of 0.14,0.05, 0.02, and 0.13 mM, respectively. Li+ was analyzed atINSTAAR using a Perkin Elmer Flame Atomic AdsorptionSpectrometer, with precision estimated at 2.00% RSD and adetection limit of 2.9 mM. DOC was analyzed at the CraryLab in McMurdo station on a Shimadzu TOC-V CPN TotalOrganic Carbon Analyzer, with a precision estimated at1.34%RSD and a detection limit of 8.26 mM. N2O wasmeasured on an HNU Systems Gas Chromatograph 301fitted with a 63Ni, electron capture detector, with a precisionestimated at 3.61%RSD and a detection limit of 1.5 nM.

3.2. Data Analysis

[12] A mass balance approach was used to quantify waterflow and reactive species concentration in the anabranchingreach, between sites B and C. This reach was unique in itshydrology and subsequently its ecosystem potential, requir-ing a method to quantify the significant subsurface flow andsolute transport. When conservative solute variability indi-cated mixing of stream and subsurface waters, return flowswere calculated using:

QC ¼ QRF þ QB ð1Þ

and

QC * CC ¼ QRF * CRF þ QB * CB; ð2Þ

where Q represents flow in liters per second, and C isconcentration in micromoles per liter. Subscripts denote the

measurement location, where B and C represent samplingsites B and C, respectively, and RF is the subsurface returnflow. QC was estimated as flow measured at the stream gagebelow site E. Solute concentrations were measured at sitesB and C and estimated for CRF as the stream water con-centration at site B during the hours the anabranches wereactively recharging the subsurface. QRF and QB werecalculated by solving equations (1) and (2) simultaneouslyusing Cl� concentrations. Once all three flow terms wereknown, subsurface production/loss of N species and DOCwas calculated as the difference between the observed andcalculated values of return flow mass (CRF * QRF).[13] We assumed that streamflows were steady between

sites above and below the anabranches for all times.Therefore, flow at site B (QB) was assumed equal to flowat sites I and A. Similarly, flow at site C (QC) was assumedequal to flow at sites D and E. The average return flow forboth low-flow periods was divided by the return flowinterval to determine an hourly difference in flow betweensites B and C. This value was subtracted from QA and QB

during the low-flow periods, and added to QA and QB duringthe preceding high-flow periods.[14] Areal Uptake (U) is a metric defined by the Stream

Solute Workshop [1990] that relates nutrient availability anddemand by considering the rate of solute uptake or produc-tion given the physical stream properties velocity and depth.Proper use of this metric requires steady state conditions ofstreamflow, solute concentrations, and hyporheic dynamics,and was calculated during the quasi-stable low-flow periods(Q < 15 L/s) for reactive species. Reach 4 was chosen forU calculation because: (1) DOC was measured at theupstream and downstream sampling sites, (2) this reachis downstream of the anabranches and thus influenced byreturn flow chemistry, and (3) it is not a gaining reach atlow flow. Uptake was calculated according to:

U ¼ u * h=x * ln Coð Þ � ln Cx * Qx=Qoð Þ½ � * Cx; ð3Þ

where u is the stream water velocity, h is the depth of thewater, x is the distance between two sampling sites, C issolute concentration in moles per liter, Q is streamflow inliters per second, x is a downstream sampling site, and o isan upstream sampling site.[15] Mass change (DM) was calculated at all times and

sites for conservative and reactive constituents according to:

DM ¼ Qx * Cx � Qo * Co; ð4Þ

where DM is mass of solute produced or lost in each reachin mols/h. Total mass production/loss during the injectionwas calculated using equation (4), with results summedfrom t = 0 through 32 h. Error for each calculation wasdetermined by considering laboratory analyses uncertainty,and an estimated flow measurement error of 5%.

4. Results

[16] Discharge and background solute concentrationsduring the injection are displayed in Figure 2. Flow variedbetween 3.5 and 25.5 L/s, with maxima at t = 4 and 21 h andminima at t = 14 and 32 h. Anabranches were active inReach 2 during the two high-flow periods of the injection

G01001 KOCH ET AL.: UNSTEADY FLOW AND NITRATE LOSS

4 of 15

G01001

(Figure 3). Many of the anabranches were only meters longand remained hydrologically connected to the stream. How-ever, others routed water far from the main channel wherethey became stagnant and eventually dried. Presumably, thiswater infiltrated the subsurface. In addition to the diel flowpattern, there was also a trend of decreasing streamflowrelated to decreasing temperatures and overcast skies duringthe second day of the injection. Background concentrationsof Cl� and NO3

� fluctuated with both the diel and longer-termtrends in streamflow, with an overall increase in concentra-tion of about 50%.[17] Solute injection increased Cl� and NO3

� from back-ground concentrations of 358 and 11.5 mM to averageconcentrations of 515 and 52 mM, respectively (Figure 4).Lithium was initially undetectable, and increased to anaverage of 29 mM during the injection. Injected soluteconcentrations were variable during the injection, and dis-played travel times between sites A and E of approximately30 min. Chloride and Li+ concentrations only neared steadystate during low-streamflow periods, between t = 5 and 16 h,and between t = 24 and 31 h. Li+ concentrations varied withflow, and also decreased in magnitude from site A to site E.Nitrate concentrations decreased between tracer time t = 5and 20 h and increased from t = 20 h until the end of theinjection. Sodium concentrations varied similarly to Cl�

values, and therefore will not be discussed in this paper.

[18] Reach 2 displayed substantial anabranching duringthe tracer injection, and displayed trends in conservative andreactive solutes dissimilar from trends in the other reaches.There was a significant decrease in Cl� and Li+ concentra-tions in this reach during two low-flow periods, from t = 5 to16 h and t = 26 to 31 h (Figure 5). The dilution lasted for6 h during the first low-flow period. During the secondflood recession dilution is less clear due to (1) a largeconcentration spike at site C at t = 29 h and (2) cessation ofthe injection at t = 31 h. Using Cl� as the conservative tracer,average subsurface return flows for the first and second low-flow periods were calculated using equations (1) and (2) as1.90 and 3.15 L/s. Return flows accounted for 27 and 38%of the total streamflow during the two low-flow periods,respectively.[19] Noninjected reactive N species displayed a range of

responses during the injection. NO2� concentrations were

significantly higher at site A relative to the downstream sitesfrom t = 0.5 to 6 h, and concentrations were highly variableat site C. NH4

+ displayed large spikes in concentration thattypically lasted only 1–2 h, occurring at sites B, C, and D atdifferent times, and often occurred during high and recedingstreamflows. Mean N2O concentrations increased with dis-tance from the injection, and displayed the highest concen-trations at all measured sites during high and recedingstreamflows.[20] DOC concentrations were temporally and spatially

variable (Figure 6). The highest DOC concentrations were240 and 755 mM, and were measured at site I at t = 1 and 2 h,respectively. Mean concentrations decreased downstreamof this pulse at site I. At site D, DOC concentrations weresignificantly higher than other sites during low-flow periods,and not statistically different from other sites during high-flow periods. Even with the pulses of DOC at site D at lowflows, DOC concentrations at site E were never significantlydifferent from the minimum detection limit.[21] Areal uptake calculated in Reach 4 was variable

between species and between return flows (Figure 7).Nitrate uptake was variable, with greater uptake duringthe second return flow. Nitrite was taken up during the firstreturn flow and produced during the second return flow.Ammonium rates were variable, with the greatest productionduring the second return flow. Nitrous oxide was producedat similar rates during both periods. DOC was taken up, witha greater loss during the second return flow.[22] Total changes in injected solute mass during the

tracer are summarized in Figure 8. There was no significantdifference in the Cl� mass passing sites A and E during theinjection. In contrast, Li+ recovery at site E was only 47%(±3.8) of that measured at site A, with 74% of this lossoccurring in the anabranching reach. NO3

� recovery was86% (±4.0), which corresponds to a total loss of 10.6 mol.Relative to NO3

�, other N species mass changes were small(Table 1). NO2

� remained a small portion of total dissolvednitrogen (TDN) at all sites. NH4

+ mass was 3.8 times greaterat site B than at site A, and decreased significantly withdistance downstream between sites B and E.[23] Maximum, minimum, and mean mass changes were

calculated for each biogeochemically active species mea-sured during the injection using equation (4). These resultswere calculated hourly for NO3

�, NO2�, and NH4

+ and lessfrequently for N2O and DOC (Tables 2a and 2b). Figures 9a

Figure 2. Huey Creek discharge over 60 h, displayingthree daily flood pulses. The tracer injection began at timeequal to 0 h and ended at time equal to 31 h. Over thisperiod there is a trend of decreasing flow related to decliningweather conditions. Background concentrations of theprimary injected solutes (measured above the injection atsite I) reacted to high flows and increased significantly duringthe tracer, between t = 16 and 18 h.

G01001 KOCH ET AL.: UNSTEADY FLOW AND NITRATE LOSS

5 of 15

G01001

and 9b display these changes on a per-hour basis, and alsoinclude Cl� and Li+ mass changes. In general, Figures 9aand 9b show that Cl� and NH4

+ mass was constant, Li+,NO3�, NO2

�, and DOC were lost, and N2O was produced.Reactivity was focused around high-flow periods. Cl� masschanged only slightly between sites A and E, with the largestinstances of production/loss in Reach 2, the anabranchingreach. Li+ loss occurred mainly on the regression limb of thedaily flow pulse, and occurred predominantly in the Reach 2.NO3� was produced and lost in all reaches, with the greatest

loss during high-flow periods in Reach 1 and 2. NO2�was lost

predominantly in Reach 1. The greatest NO2� loss occurred

during the first high-flow period, with a second period of lossduring the second high flow. Discrete pulses of NO2

� wereproduced in Reach 2 during the second high-flow period, andlost in Reach 3. Ammonium pulses were produced in Reach 1during the second high flow, and lost in reach 2. N2Oproduction was greatest in Reaches 2 and 3, and mainlyoccurred during high-flow periods. DOC loss was greatestin the upstream reaches during the first flood pulse.[24] Mass change calculations (Figures 9a and 9b) show

that the greatest rates of NO3�, NO2

�, N2O, and DOC lossoccurred during the initial hours of the experiment. Thegreatest NO3

� loss occurs in Reach 1 at t = 1 h, while NO2�

loss is greatest at t = 3 to 4 h, with significant loss throught = 7 h. These losses coincide with the highest streamflowand highest concentration of DOC measured over the

sampling interval. This period of high activity correspondedto DOC decreasing from 240 to 57 mM and from 755 to26 mM at t = 1 and 2 h, respectively, between sites I andB. Extended periods of NO3

� and NO2� loss also occur at

early times between reaches C and E, coincident withelevated DOC. NO3

� and NO2� loss rates and DOC concen-

trations are smaller during the second high-flow period, butstill occur to some extent at both locations.[25] N species and DOC in the return flows were calcu-

lated from equations (1) and (2) (Table 3). Differencesbetween the calculated and observed reactive species indi-cate variations in N species and DOC mass change in thesubsurface of the anabranching reach between the two low-flow periods when return flow occurred. The first return flowdisplayed production of NO2

�, NH4+, N2O, and DOC. The

second return flow displayed uptake of NO3� and NO2

�,decreased production of NH4

+, and increased production ofN2O and DOC.

5. Discussion

5.1. Parsing Hydrology and Chemistry

[26] Dynamic variations in N species over the length ofthe 31 h Huey Creek NO3

� enrichment injection provideevidence of microbial activity. However, extreme variabilityof both reactive and conservative breakthrough curves(Figure 4) highlight the necessity of considering unsteady

Figure 3. The anabranching reach of Huey Creek, looking upstream. Site B is located to the right of thelarge snowfield. A radio is in the foreground for scale. The stream channel meanders from the centerbackground to the left foreground. Note anabranches leaving the frame on the right side, and infiltratingin the right foreground.

G01001 KOCH ET AL.: UNSTEADY FLOW AND NITRATE LOSS

6 of 15

G01001

Figure

4.

Breakthroughcurves

forthetracer

injection.InjectionofCl�,Li+,andNO3�began

attimeequal

to0hand

ended

attimeequalto31h,asindicated

bythedashed

line.Thelargeincrease

inCl�

aftertheendoftheinjectionresultsfrom

increasingbackgroundCl�

concentrationsas

stream

flow

decreased

(shownin

detailin

Figure

2).DOC

andN2O

were

sampledatsitesB,D,andE,andonly

atcertainhours.

G01001 KOCH ET AL.: UNSTEADY FLOW AND NITRATE LOSS

7 of 15

G01001

streamflow in order to understand heightened biogeochem-ical activity in this oligotrophic stream ecosystem. Conser-vative solutes are chemically inert; therefore changes in suchconstituents are indicative of hydrologic changes in thesystem (high versus low flows, lateral inflows, etc). Changesin reactive solutes result from both hydrology and chemicalreactivity. By comparing the conservative and reactive soluterecords we attempt to parse effects of microbial activity fromthe overall variability in the tracer data.[27] Several solutes appeared nonreactive in Huey Creek

during the injection. In-stream concentrations of injectedCl�, and background solutes PO4

� and SO42� (not shown)

varied similarly in time and space. Such covariance isindicative of changes in hydrology, which acts to dilute orconcentrate solutes in the stream. While Li+ acted conserva-tively during low flows, there was significant loss of Li+

during high-flow periods, especially in the anabranchingreach (Figure 9a). Similar loss of Li+ has been witnessed inprevious injections in a Dry Valley stream and attributed tosorption and/or cation exchange [Gooseff et al., 2004a]. Thelarge loss in Li+ during the Huey Creek injection was likely aresult of similar processes, and occurred predominantly inthe anabranching reach and during high flows due to theheightened interaction between water and sediments. Thisevidence highlights this reach and high-flow periods aspotentially biogeochemically active and therefore importantfoci of our current study.

5.2. Anabranch Hydrology

[28] Evidence from breakthrough curves and mass changecalculations suggest that subsurface storage and exchange inthe anabranching reach is significant and temporally variable.Water and solute exchange is the result of unsteady flowand subsequent anabranching and subsurface storage. Obser-vations suggests that during high flows, which occurredbetween t = 0 to 4 h and t = 17 to 23 h, water was exchangedfrom the channel to the anabranches and into the subsurface.The loss of water from the stream to the anabranches andsubsurface is most clearly witnessed in Cl�mass loss in reach

2 during the second high-flow period (Figure 9a). DecreasedCl� concentrations between sites B and C (Figure 5) duringlow-flow periods provide evidence that a mixture of cleanand tracer-labeled water returned to the stream once theflood pulse receded.[29] Cl� dilution was a result of ‘‘clean,’’ tracer-less

water moving from the subsurface into the stream andlasted for 6 h during the first low-flow period. Cessationof the clean water pulse was concurrent with a NO2

� pulse att = 16 h, which signaled (1) the arrival of the injected tracerat the end of the subsurface flow path, and (2) incompletereduction of the NO3

� injection in the subsurface. During thesecond flood recession there was another instance of dilutedconcentrations and increased NO2

� concentrations at site C.These changes signaled a second subsurface return flow.While our mass balance calculations show that this secondreturn flow is entering the stream, the dilution is less visuallyapparent in Figure 5, because all of the water entering thesubsurface during the second high flow was tracer-labeled.Further evidence of the second return flow is provided byhigher Cl� at site C relative to site B after cessation of theinjection. This increase occurred because the return flowwater measured at site C was a mixture of the tracer-labeledwater exiting the anabranches and the unlabeled streamwaterfrom site B.

5.3. Anabranch Biogeochemistry

[30] The difference in anabranch water chemistry betweenthe two return flows (Table 3) provides evidence of theeffect of NO3

�, our reactive tracer, on subsurface biogeo-chemistry. The first return flow likely contains a mixture ofbackground and tracer-labeled water, and is therefore moreclosely related to background conditions than is the second

Figure 5. Chloride dilution between sites B and C occurredduring the first low-flow period (t = 5–16 h), signaling aninflow of clean, tracerless water from the subsurface. Thistrend is less obvious during the second low-flow period(t = 24–31 h), because injected Cl� entered the subsurfacethrough anabranches during the previous high-flow period.

Figure 6. DOC concentrations measured at sites B, D, andE during high and low flows. Sample means were comparedusing the nonparametric Mann-Whitney Test. Error barsrepresent the range of values. The minimum detection limit(MDL) for DOC is 8.36 mM. Values below MDL were set toone half MDL. Letters represent populations that are notstatistically different at the 0.90 level. During high flows,sites B and D have significantly higher concentrations thansite E. The large ranges in sites B and D likely occur fromthe high DOC pulse measured upstream during the first hoursof the injection. During low flows, site D has significantlyhigher DOC concentrations than sites B and E.

G01001 KOCH ET AL.: UNSTEADY FLOW AND NITRATE LOSS

8 of 15

G01001

return flow. The switch from NO3� and NO2

� productionduring the first return flow and uptake during the secondreturn flow suggests greater uptake of oxidized N species atlater times. Less NH4

+ and more N2O and DOC are producedduring the second return flow. These shifts may indicate achange in microbial activity as populations adjust to theelevated N availability. Alternatively, these shifts may be afunction of the increased residence time of NO3

�-rich porewater that remains in the subsurface for many hours afterthe first return flow and before flushing by the followinghigh flow/infiltration event. The production of DOC in theanabranches is the cause of significantly higher concentra-tions measured at site D at low flows (Figure 6), and likelyalso influences the calculated uptake rates (Figure 7).

5.4. Nitrogen Reactivity in Channelized Reaches

[31] Reactivity in the channelized reaches of Huey Creekis highly variable in time and space. Production or loss inany given reach often results in mass change in subsequent

Figure 7. Areal uptake rates for reactive species between sites D and E. Positive values representproduction. Calculations require steady state conditions, and were only calculated during low-flow (Q <15 L/s) periods. Error bars represent the maximum uncertainty for each calculation. The solid linerepresents streamflow. Vertical dashed lines denote cessation of the injection. Horizontal dashed linesindicate an aerial uptake rate of zero.

Figure 8. Total mass measured at sites A and E, integratedfrom time equal to 0 h to time equal to 32 h, which is 1 hafter the injection pump was shut off. Mass is expressed as apercentage of mass measured at site A, just below theinjection. Error bars represent the range of possible values foreach measurement and consider both chemical analyses andflow measurement uncertainty.

G01001 KOCH ET AL.: UNSTEADY FLOW AND NITRATE LOSS

9 of 15

G01001

reaches, creating chain reactions as limiting nutrients arecycled downstream.[32] Ammonium production in Reach 1 and rapid loss in

Reach 2 during the second high-flow period (Figure 9a)provide evidence of microbial activity in Huey Creek. The

source of the pulse is ambiguous; NH4+ production may be

related to NO3� and NO2

� reduction, or it may have beengenerated by the exchange with Li+ at sorption sites.Regardless of the source, the NH4

+ pulse emanates fromReach 1 during high flows and is completely lost in Reach

Table 1. Nitrogen Species Mass at Sites A, B, C, and Ea

Site A Site B Site C Site E

Nitrate 74 (1.5) 70 (1.4) 66 (1.3) 64 (1.3)Nitrite 0.23 (0.0021) 0.08 (0.0020) 0.11 (0.0013) 0.06 (0.0051)Ammonium 0.16 (0.0032) 0.63 (0.0012) 0.27 (0.0053) 0.12 (0.0024)N2O NA 0.026 (0.00045) NA 0.034 (0.00058)TDN 74 (1.5) 70 (1.4) 67 (1.3) 64 (1.3)

aUnits are in moles. N2O was not sampled at sites A and C. Values in parentheses represent the maximum error range. TDN was calculated as the sum ofall measured species. NA, not available.

Figure 9a. Mass production (positive) and loss (negative), in moles, for species measured at each timeinterval. Samples were collected from time equal to 0–52 h. The injection began at t = 0, and ended att = 31 h. Mass change is calculated as the difference in mass between the upstream and downstream sites.The solid line represents streamflow. Horizontal dashed lines represent the maximum level of uncertainty;therefore values outside of this range represent significant production/loss. The vertical dashed linerepresents the end of the injection period. Changes in Cl�mass represent storage and exchange between thestream and anabranching/hyporheic zones. Major changes in N species that are not coincident with changesin Cl� indicate microbial activity. Generally, Li+, NO3

�, and NO2� loss are coincident with high-flow

periods.

G01001 KOCH ET AL.: UNSTEADY FLOW AND NITRATE LOSS

10 of 15

G01001

Figure

9b.

Massproduction(positive)

andloss

(negative),in

moles,forspeciesmeasuredateach

timeinterval.Sam

ples

werecollectedfrom

timeequalto

0–52h.Theinjectionbegan

att=0,andended

att=31h.Masschangeiscalculatedas

thedifference

inmassbetweentheupstream

anddownstream

sites.Thesolidlinerepresentsstream

flow.Horizontaldashed

lines

representthemaxim

um

level

ofuncertainty;therefore

values

outsideofthisrangerepresentsignificantproduction/

loss.Theverticaldashed

linerepresentstheendoftheinjectionperiod.DOCloss

isgreatestatearlytimes

intheupstream

reach.DOC

productionis

greatestin

betweensitesB

andD

(whichincludes

theanabranchingreach)duringfalling/low

stream

flow.Thearea

ofinsignificantDOCchange(i.e.,thedifference

betweenthetwohorizontal,dashed

lines)isdifficult

toseebecause

itissm

allrelativeto

theplotscale.

G01001 KOCH ET AL.: UNSTEADY FLOW AND NITRATE LOSS

11 of 15

G01001

2. This loss coincides with the highest NO2� production of

the entire injection, N2O production, and significant DOCloss. This reactivity provides evidence of N oxidation in thechannel of Huey Creek.[33] The direct correlation between N reduction and high

streamflows in channelized reaches suggests the cycle ofwetting and drying of stream banks may be an importantcontrol on hyporheic redox conditions. N cycling in largerriver systems has been related to water table elevations,which determine whether soils are aerobic, anoxic, or anaer-obic, and subsequently whether nitrification or denitrificationdominates [Hefting et al., 2004]. In the channelized reaches

of Huey Creek, changes in N species mass are not associ-ated with Cl� mass changes (Figures 9a and 9b), suggestingthat biogeochemical processes are not related to largechanges in solute storage. Instead, N species reactivity andoxidation state may be primarily controlled by exchangebetween oxidized surface water and sub/anoxic sedimentsin the hyporheic zone. High streamflows have previouslybeen linked to decreased hyporheic exchange [Harvey etal., 1996]. In Huey Creek, high streamflow may similarlyretard hyporheic flushing, thereby allowing the subsurfaceconditions to become anoxic, and increasing contact timebetween NO3

� and reaction sites. Anoxic conditions havebeen inferred from denitrifying activity in other Dry Valleystreams [Gooseff et al., 2004b]. As streamflow decreases,exchange rates may increase, bringing reduced N (i.e., NH4

+

in reach 1 at t = 17 and 20) back in contact with oxygenatedsurface waters. This scenario would explain the NH4

+ peaksthat emanate from reach 1 at low flows. This model isuseful in understanding N dynamics in Huey Creek, andtestifies to the temporal and spatial variability in N storageand cycling that can be caused by unsteady flow conditions(Figure 10).[34] Reactive species mass change also occurs in Reach 4,

as inferred from uptake calculations (Figure 7). This is one ofonly a few instances of reactivity during low-flow periods,and is likely dependent on the DOC-rich return flows fromthe anabranching reach. Variability in uptake lengths maystem from several causes. During the first return flow, theswitch from a tracerless, to a labeled return flow could affectmicrobial processes and subsequently the reactive speciesinputs to Reach 4. Furthermore, even at low flows, stream-flow is not completely steady, which likely affects thecalculated uptake. Changes in uptake rates may be relatedto variability in the N species and DOC mass in the returnflows (Table 3), or may be associated with some lag inmicrobial reaction time, in which populations adjust theiractivity to most efficiently utilize the elevated N species andDOC concentrations.

5.5. Dissolved Organic Carbon

[35] The coincidence of high DOC concentrations andreactive species loss suggests that Huey Creek is significantlyC limited and that C availability leads to heightened micro-bial activity. DOC represents an important energy source formicrobes, and is in short supply in Huey Creek relative to DryValley streams with significant cyanobacterial mat material.High DOC mass from an unknown source was measured atsite I in the early, high-flow hours of the injection. DOC

Table 2a. Hourly Average, Maximum, and Minimum Production

and Loss Rates of Mass Change for Reactive Species Measured

Each Hour During the 32 h in Which Tracer Mass Was Stored in

the Experimental Reachesa

Nitrate (mol/h) Nitrite (mol/h) Ammonium (mol/h)

Reach 1 (A–B)Average �0.12 (0.22) �0.0038 (0.00067) 0.012 (0.0018)Maximum 1.1 (0.093) 0.0028 (0.00032) 0.25 (0.00062)Minimum �3.60 (0.98) �0.034 (0.0041) �0.10 (0.0019)

Reach 2 (B–C)Average �0.082 (0.21) 0.00058 (0.00023) �0.0094 (0.0018)Maximum 1.3 (0.56) 0.014 (0.00043) 0.067 (0.00023)Minimum �2.5 (0.89) �0.0050 (0.000075) �0.24 (0.028)

Reaches 3 and 4 (C–E)Average �0.073 (0.19) �0.0013 (0.00016) �0.0039 (0.00034)Maximum 1.4 (0.21) 0.0022 (0.00024) 0.018 (0.0026)Minimum �1.0 (0.60) �0.013 (0.00018) �0.068 (0.0077)

All Reaches (A–E)Average �0.27 (0.22) �0.0045 (0.00067) �0.0011 (0.0018)Maximum 2.44 (0.21) 0.00063 (0.00032) 0.0094 (0.0026)Minimum �3.32 (0.98) �0.035 (0.0041) �0.096 (0.0077)

aProduction is positive values, and loss is negative values. Values inparentheses represent the maximum error range.

Table 2b. Hourly Average, Maximum, and Minimum Production

and Loss Rates of Mass Change for N2O and DOC During the 32 h

in Which Tracer Mass Was Stored in the Experimental Reachesa

N2O (mmol/h) DOC (Kmol/h)

Reach Upstream and 1 (I–B)Average �0.0041 (0.15) �1.3 (0.10)Maximum 0.21 (0.12) 0.42 (0.032)Minimum �0.44 (0.26) �7.6 (0.80)

Reaches 2 and 3 (B–D)Average 0.27 (0.18) �0.0040 (0.040)Maximum 0.61 (0.26) 0.47 (0.076)Minimum 0.056 (0.28) �0.50 (0.019)

Reach 4 (D–E)Average 0.16 (0.20) �0.16 (0.022)Maximum 0.36 (0.25) 0.47 (0.047)Minimum �0.0043 (0.077) �0.77 (0.031)

All Reaches (I–E)Average 0.50 (0.32) �1.5 (0.16)Maximum 1.0 (0.084) 0.0060 (0.011)Minimum 0.052 (0.0071) �7.4 (0.80)

aProduction is positive values, and loss is negative values. Values inparentheses represent the maximum error range. Samples for these twospecies were collected at hours 0, 0.5, 1, 2, 6, 12, 18, 24, and 30 h.

Table 3. Reactive Species Mass Production and Loss in Return

Flows From the Anabranchesa

Return Flow 1 Return Flow 2

Time (h) 5–16 24–31Nitrate (mmol) �31 (7.9)Nitrite (mmol) 0.031 (0.0063) �0.19 (0.0074)Ammonium (mmol) 0.28 (0.030) 0.11 (0.014)Nitrous oxide (mmol) 0.045 (0.021) 0.049 (0.023)DOC (mol) 195 (11) 447 (12)

aProduction is positive values, and loss is negative values. Valuesrepresent the difference between the expected and calculated mass.Maximum errors from chemical analyses are shown in parentheses. Valuesthat were not significantly different from 0 at a 95% confidence interval arenot reported.

G01001 KOCH ET AL.: UNSTEADY FLOW AND NITRATE LOSS

12 of 15

G01001

concentrations decreased significantly in Reach 1, accompa-nied by the highest rates of NO3

�, NO2�, and N2O loss for the

entire experiment. This pulse continues downstream, leadingto NO3

� and NO2� loss between sites C and E as well.

[36] The anabranching reach is another significant sourceof DOC as evidenced by DOC pulses measured at site Dduring subsurface return flows (Figures 4 and 6), and DOCproduction calculated from the anabranch reach mass bal-ance (Table 3). Mass balance calculations also suggest thatreturn flows from the anabranches experienced significantreduction of NO3

� and NO2� (Table 3). DOC pulses that exit

the anabranches are first measured at site D and decreasebelow the minimum detection limit by site E (Figure 6). Thisdecrease testifies to the rapid uptake of DOC in Huey Creek.[37] While the importance of DOC to ecosystem function

in Huey Creek is clear, its source is less so. Microbial matgrowth in Dry Valley streams is often limited by high watervelocity, unsteady streamflow, and unstable substrates[Alger et al., 1997]. In previous years cyanobacterial matshave been found growing in the anabranching reach ofHuey Creek [Alger et al., 1997], where low water velocitiesand stagnant pools dominate. In lieu of visible mats in theanabranching reach during our injection, a more likely DOCsource may be leaching of dead or dormant mat materialthat has been buried in the anabranch sediments fromprevious periods of high cyanobacterial mat growth. Infil-trating water may transport senescent material from thissource through the anabranches, allowing NO3

� reductionto occur in the subsurface, and creating a DOC pulse as waterreturns to the stream channel hours later.[38] The source of DOC measured at site I is similarly

unclear. The stream slope steepens above the tracer reach,

precluding the possibility of anabranching upstream. Fur-thermore, the DOC pulse at site I occurred only during thehigh flows, while DOC pulses from the anabranching reachwere measured at low flows. It is possible that the source ofthe DOC is the ice field from which Huey Creek originates.During periods of significant melt, supraglacial streams mayflush material from cryoconite holes and other DOC-richglacial features [Fortner et al., 2005; Fountain et al., 2007;Porazinska et al., 2004], providing a critical C source toHuey Creek. Because the upstream DOC pulse only occursduring the first high-flow period, we do not believe that it isrelated to the low-flow DOC source emanating from theanabranches at both low-flow periods.

5.6. Implications of Unsteady Flow

[39] The unsteady flow regime that produces temporallyand spatially heightened biogeochemical activity in HueyCreek also precludes the use of transient storage modeling(TSMs) and limits the use of common nutrient uptakemetrics. Based on a previous conservative tracer experimentin Huey Creek, unsteady flow has been incorporated intoTSMs by calculating the surface water kinematic wave[Runkel et al., 1998]. However, our results suggest thenecessity of also considering the subsurface effect of theflood wave, which leads to spatial and temporal variabilityin subsurface storage and exchange rates. Coupling surfacewater flow, groundwater flow, solute transport, and bio-geochemical reactivity is beyond the scope of this work,but is a worthy goal, because it would provide an integratedview of the hydrologic influences on stream ecosystems.Similarly, given Huey Creek’s unsteady flow, variable hypo-rheic dynamics, and subsequent hot spots and hot moments

Figure 10. Conceptual model of unsteady flow and biogeochemistry in Huey Creek. Unsteady flowaffects the interaction of water, solutes, and sediments, and subsequently N reactivity in Huey Creek. Inthe anabranching reach, high flows lead to significant storage in the subsurface, with exchange back tothe stream at low flows. This return flow contains elevated DOC, which spurs activity in the downstreamreaches. In the channels, N reduction is focused at high flows, when interaction between water andsediments is highest.

G01001 KOCH ET AL.: UNSTEADY FLOW AND NITRATE LOSS

13 of 15

G01001

in microbial activity, our system does not lend itself tomeaningful application of areal uptake (U) and/or othersimilar metrics defined by the Stream Solute Workshop[1990]. While such metrics are useful for comparing dis-parate systems and extrapolating laboratory experiments,their calculation is highly dependent on steady state stream-flow, hyporheic exchange, and solute transport character-istics [Runkel, 2007]. Conditions during this injection wereonly stable at certain sites and times, precluding a thoroughanalysis using uptake metrics.

6. Conclusions

[40] In Huey Creek, unsteady flow regulates biogeochem-istry by controlling the interactions between water, sedi-ments, and solutes and subsequently the location, duration,and magnitude of microbial activity. Li+ loss results fromsorption, and therefore Li+ mass change may be used as aproxy for water/sediment interaction. This process is greatestat high flows and in the anabranching reach. Nitrate andNO2� loss are concurrent with high flows and Li+ loss,

suggesting that microbial activity is dependent on N transportreaching the parafluvial zone. Unsteady flow also plays animportant role by significantly increasing water and soluteresidence times within the anabranching reach. Shifts inreactive solute export from this reach implicate the anab-ranches as a hot spot of microbial activity. Furthermore, thisreach is a DOC and reduced N source, and therefore waterrecharging the channel from the anabranches also deliversnutrients to reaction sites further downstream. Flushing fromupstream/glacial sources may explain the large spikes inDOC at the beginning of our injection, which lead to thehighest NO3

� loss rate of the entire injection. Our resultssuggest that microbial activity in Huey Creek is dependenton DOC availability, and the solute transport facilitated bydaily high-flow periods. Moreover, without cyclical stream-flow, the exchange of subsurface water in the anabranchingreach would not occur, which might further limit microbialactivity by limiting nutrient delivery and reactant removal.Our tracer results indicate that even though day/night lightvariability is minor in polar desert streams, diel variations inmicrobial activity may still occur.[41] Including the effects of unsteady flow on hyporheic

storage and N cycling is an important step toward improvinghydrologic and biogeochemical modeling of stream eco-systems. We have shown how a combination of solute andwater mass balance calculations can be used to quantitativelyevaluate the complicated biogeochemical signature of asystem dominated by highly variable microbial activity. Thisdata set underscores the challenge inherent in analyzingstream data from harsh environments and unsteady flowregimes, characteristics common to many polar and aridenvironments.

[42] Acknowledgments. We are indebted to two anonymousreviewers for their comments, which substantially improved this manu-script. We would like to thank Raytheon Polar Services and PHI helicoptersfor invaluable field support, Chris Siebold and Amber Roche for laboratoryanalyses, Bess Ward for use of the gas chromatograph at Princeton Uni-versity, Ken Bencala and Michael Gooseff for advice on previous drafts, andeveryone who contributed time and talent to preparation and execution of the

tracer injection. This work was supported by NSF OPP-9810219 and theNSF Microbial Biology Postdoctoral Fellowship Program.

ReferencesAlger, A. S., D. M. McKnight, S. A. Spaulding, C. M. Tate, G. H. Shupe,K. A. Welch, R. Edwards, E. D. Andrews, and H. R. House (1997),Ecological processes in a cold desert ecosystem: The abundance andspecies of algal mats in glacial meltwater streams in Taylor Valley,Antarctica, Occas. Pap. 51, Inst. of Arct. and Alp. Res., Boulder, Colo.

Brunke, M., and T. Gonser (1997), The ecological significance of exchangeprocesses between rivers and groundwater, Freshwater Biol., 37(1), 1–33,doi:10.1046/j.1365-2427.1997.00143.x.

Burkins, M. B., R. A. Virginia, and D. H. Wall (2001), Organic carboncycling in Taylor Valley, Antarctica: Quantifying soil reservoirs and soilrespiration, Global Change Biol., 7(1), 113–125, doi:10.1046/j.1365-2486.2001.00393.x.

Conovitz, P. A., D. M. Mcknight, L. H. MacDonald, A. G. Fountain, andH. R. House (1998), Hydrologic processes influencing streamflow var-iation in Fryxell Basin, Antarctica, in Ecosystem Dynamics in a PolarDesert: TheMcMurdo Dry Valleys, Antarct. Res. Ser., vol. 72, pp. 93–108,AGU, Washington, D. C.

Cozzetto, K., D. McKnight, T. Nylen, and A. Fountain (2006), Experimentalinvestigations into processes controlling stream and hyporheic tempera-tures, Fryxell Basin, Antarctica, Adv. Water Resour., 29(2), 130–153,doi:10.1016/j.advwatres.2005.04.012.

Dahm, C. N., N. B. Grimm, P. Marmonier, H. M. Valett, and P. Vervier(1998), Nutrient dynamics at the interface between surface waters andgroundwaters, Freshwater Biol., 40(3), 427–451, doi:10.1046/j.1365-2427.1998.00367.x.

Dent, C. L., N. B. Grimm, E. Marti, J. W. Edmonds, J. C. Henry, and J. R.Welter (2007), Variability in surface-subsurface hydrologic interactionsand implications for nutrient retention in an arid-land stream, J. Geophys.Res., 112, G04004, doi:10.1029/2007JG000467.

Doran, P. T., C. P. McKay, G. D. Clow, G. L. Dana, A. G. Fountain,T. Nylen, and W. B. Lyons (2002), Valley floor climate observationsfrom the McMurdo dry valleys, Antarctica, 1986–2000, J. Geophys.Res., 107(D24), 4772, doi:10.1029/2001JD002045.

Esposito, R. M. M., S. L. Horn, D. M. McKnight, M. J. Cox, M. C. Grant,S. A. Spaulding, P. T. Doran, and K. D. Cozzetto (2006), Antarcticclimate cooling and response of diatoms in glacial meltwater streams,Geophys. Res. Lett., 33, L07406, doi:10.1029/2006GL025903.

Fortner, S. K., M. Tranter, A. Fountain, W. B. Lyons, and K. A. Welch(2005), The geochemistry of supraglacial streams of Canada Glacier,Taylor Valley (Antarctica), and their evolution into proglacial waters,Aquat. Geochem., 11(4), 391–412, doi:10.1007/s10498-004-7373-2.

Fountain, A. G., et al. (1999), Physical controls on the Taylor Valley eco-system, Antarctica, BioScience, 49(12), 961–971, doi:10.2307/1313730.

Fountain, A. G., M. Tranter, T. H. Nylen, K. J. Lewis, and D. R. Mueller(2007), Evolution of cryoconite holes and their contribution to meltwaterrunoff from glaciers in the McMurdo Dry Valleys, Antarctica (vol 50, pg35, 2004), J. Glaciol., 53(183), 723, doi:10.3189/002214307784409315.

Gooseff, M. N., D. M. McKnight, W. B. Lyons, and A. E. Blum (2002),Weathering reactions and hyporheic exchange controls on stream waterchemistry in a glacial meltwater stream in the McMurdo Dry Valleys,Water Resour. Res., 38(12), 1279, doi:10.1029/2001WR000834.

Gooseff, M. N., D. M. McKnight, R. L. Runke, and B. H. Vaughn (2003),Determining long time-scale hyporheic zone flow paths in Antarcticstreams, Hydrol. Process., 17(9), 1691–1710, doi:10.1002/hyp.1210.

Gooseff, M. N., D. M. McKnight, and R. L. Runkel (2004a), Reach-scalecation exchange controls on major ion chemistry of an Antarctic glacialmeltwater stream, Aquat. Geochem., 10(3–4), 221–238, doi:10.1007/s10498-004-2260-4.

Gooseff, M. N., D. M. McKnight, R. L. Runkel, and J. H. Duff (2004b),Denitrification and hydrologic transient storage in a glacial meltwaterstream, McMurdo Dry Valleys, Antarctica, Limnol. Oceanogr., 49(5),1884–1895.

Green, W. J., M. P. Angle, and K. E. Chave (1988), The geochemistry ofAntarctic streams and their role in the evolution of 4 lakes of the McmurdoDryValleys,Geochim. Cosmochim. Acta, 52(5), 1265–1274, doi:10.1016/0016-7037(88)90280-3.

Harms, T. K., and N. B. Grimm (2008), Hot spots and hot moments ofcarbon and nitrogen dynamics in a semiarid riparian zone, J. Geophys.Res., 113, G01020, doi:10.1029/2007JG000588.

Harvey, J. W., and K. E. Bencala (1993), The effect of streambed topogra-phy on surface-subsurface water exchange in mountain catchments,Water Resour. Res., 29(1), 89–98, doi:10.1029/92WR01960.

Harvey, J. W., B. J. Wagner, and K. E. Bencala (1996), Evaluating thereliability of the stream tracer approach to characterize stream-subsurface

G01001 KOCH ET AL.: UNSTEADY FLOW AND NITRATE LOSS

14 of 15

G01001

water exchange, Water Resour. Res., 32(8), 2441–2451, doi:10.1029/96WR01268.

Hefting, M., J. C. Clement, D. Dowrick, A. C. Cosandey, S. Bernal,C. Cimpian, A. Tatur, T. P. Burt, and G. Pinay (2004), Water tableelevation controls on soil nitrogen cycling in riparian wetlands along aEuropean climatic gradient, Biogeochemistry, 67(1), 113–134.

Holmes, R. M., S. G. Fisher, N. B. Grimm, and B. J. Harper (1998), Theimpact of flash floods on microbial distribution and biogeochemistry inthe parafluvial zone of a desert stream, Freshwater Biol., 40(4), 641–654,doi:10.1046/j.1365-2427.1998.00362.x.

Jones, J. B., S. G. Fisher, and N. B. Grimm (1995), Vertical hydrologicexchange and ecosystem metabolism in a Sonoran Desert stream, Ecology,76(3), 942–952, doi:10.2307/1939358.

Joslin, J. (2005), Determining the role of chemical weathering reactions andhyporheic exchange on silicate concentrations in Dry Valley streams,Antarctica, Univ. of Colo., Boulder.

McClain, M. E., et al. (2003), Biogeochemical hot spots and hot momentsat the interface of terrestrial and aquatic ecosystems, Ecosystems, 6(4),301–312, doi:10.1007/s10021-003-0161-9.

McKnight, D. M., D. K. Niyogi, A. S. Alger, A. Bomblies, P. A. Conovitz,and C. M. Tate (1999), Dry valley streams in Antarctica: Ecosystemswaiting for water, BioScience, 49(12), 985–995, doi:10.2307/1313732.

McKnight, D. M., R. L. Runkel, C. M. Tate, J. H. Duff, and D. L. Moorhead(2004), Inorganic N and P dynamics of Antarctic glacial meltwater streamsas controlled by hyporheic exchange and benthic autotrophic communities,J. N. Am. Benthol. Soc., 23(2), 171 – 188, doi:10.1899/0887-3593(2004)023<0171:INAPDO>2.0.CO;2.

Miller, M. P., D. M. McKnight, R. M. Cory, M. W. Williams, and R. L.Runkel (2006), Hyporheic exchange and fulvic acid redox reactions in analpine stream/wetland ecosystem, Colorado front range, Environ. Sci.Technol., 40(19), 5943–5949, doi:10.1021/es060635j.

Nanson, G. C., andA.D. Knighton (1996), Anabranching rivers: Their cause,character and classification, Earth Surf. Process. Landf., 21(3), 217–239,

doi:10.1002/(SICI)1096-9837(199603)21:3<217::AID-ESP611>3.0.CO;2-U.

Porazinska, D. L., A. G. Fountain, T. H. Nylen, M. Tranter, R. A. Virginia,and D. H. Wall (2004), The biodiversity and biogeochemistry of cryo-conite holes from McMurdo Dry Valley glaciers, Antarctica, Arct. Ant-arct. Alp. Res., 36(1), 84 – 91, doi:10.1657/1523-0430(2004)036[0084:TBABOC]2.0.CO;2.

Runkel, R. L. (2007), Toward a transport-based analysis of nutrient spiral-ing and uptake in streams, Limnol. Oceanogr. Methods, 5, 50–62.

Runkel, R. L., D. M. McKnight, and E. D. Andrews (1998), Analysis oftransient storage subject to unsteady flow: Diel flow variation in anAntarctic stream, J. N. Am. Benthol. Soc., 17(2), 143 – 154,doi:10.2307/1467958.

Stream Solute Workshop (1990), Concepts and methods for assessing solutedynamics in stream ecosystems, J. N. Am. Benthol. Soc., 9(2), 95–119,doi:10.2307/1467445.

Valett, H. M., M. A. Baker, J. A. Morrice, C. S. Crawford, M. C. Molles,C. N. Dahm, D. L. Moyer, J. R. Thibault, and L. M. Ellis (2005),Biogeochemical and metabolic responses to the flood pulse in a semiaridfloodplain, Ecology, 86(1), 220–234, doi:10.1890/03-4091.

Von Guerard, P., D. M. Mcknight, R. A. Harnish, J. W. Gartner, and E. D.Andrews (1995), Streamflow, water-temperature, and specific-conductancedata for selected streams draining into Lake Fryxell, Lower Taylor Valley,Victoria Land, Antarctica, 1990–92, U.S. Geol. Surv. Open File Rep.,94–434.

�����������������������J. L. Baeseman, International Arctic Research Center, University of

Alaska Fairbanks, Fairbanks, AK 99775, USA.J. C. Koch and D. M. McKnight, Institute of Arctic and Alpine Research,

University of Colorado at Boulder, Boulder, CO 80303, USA. (kochjc@colorado.edu)

G01001 KOCH ET AL.: UNSTEADY FLOW AND NITRATE LOSS

15 of 15

G01001