Stream tracer methodology

The tracer injection experiment was conducted on a supraglacial stream on Mendenhall Glacier near Juneau, Alaska (Fig. 1). A 100 m study reach was selected, which consisted of an initial 15 m narrow reach for mixing, followed by a shallow and wide 45 m reach (together reach 1), and a 40 m narrow, steep reach (reach 2). Samples were collected above the injection site, and at the beginning and end of each reach (e.g. 15, 60 and 100 m downstream from injection). The 45 m section of reach 1 had observable regions of slow-moving water (e.g. pools with abundant ‘surface storage’), with some visible sediment and organic material (wind-blown leaves). Reach 2 was devoid of any bottom substrate other than glacial ice.

Fig. 1. Location of Mendenhall Glacier in southeast Alaska. The study stream, which was located 2 km from the glacier terminus, is shown in the photograph on the right.

The stream tracer experiment consisted of co-injecting conservative and non-conservative solutes. NaCl was injected into the supraglacial stream at a rate of 59 mLmin⁻¹ for 3 hours. A second solution of labile carbon (dextrose) was added in conjunction with a third solution containing a mixture of HNa₂PO₄ and NH₄Cl. The objective was to increase background concentrations of DOC from <15 to 125 μm, PO₄ ^3– from <1 to 2 μm, and NH₄ ⁺ from <1 to 5 μm in order to determine the potential uptake within this environment. Since only one nutrient addition experiment was possible due to logistics, we acknowledge that our results will be an underestimate of the actual uptake rate within our study stream (Reference MulhollandMulholland and others, 2002). This is because nutrient uptake rate is only directly proportional to small perturbations of the in-stream nutrient concentrations under strongly nutrient-limiting conditions, and becomes nutrient-saturated at higher concentrations (thus following a nutrient saturation model (e.g. Michaelis–Menten)).

Sampling for the stream tracer experiment was performed in three phases. For the first 60 min after the beginning of the injection, samples for Cl⁻ were collected every 30 s to 5 min to capture the rise of and shoulder of the Cl⁻ breakthrough curve at each site. From 60 to 180 min during the plateau of the stream injection, samples were collected every 15 min for nutrients and Cl⁻. At 180 min, the injection was turned off, and samples were collected frequently (from 30s to 5 min) to capture the decline and tail of the breakthrough curve at each site. The conservative solute breakthrough curve facilitates quantification of advection, dispersion and transient storage within the surface pools. Lateral inflow is quantified using observations during the plateau portion of the breakthrough curve. Synoptic samples were collected during the plateau every 10 m along the experimental reach. Samples for anions and nutrients were immediately filtered through 0.45 μm into 20 mL scintillation HDPE vials. Anion samples were stored in the dark, and were analyzed within 30 days using ion chromatography (Dionex 8000). Nutrient samples were stored at 0°C and analyzed within 45 days, and were measured using standard colorimetric techniques on a SEAL flow injection analyzer. The standard deviations of check standards analyzed throughout the analyses were 0.1 μm NH₄ ⁺ , 0.1μm NO₃ ⁻, 0.01 μm PO₄ ^2– and 18μm DOC. DOC samples were filtered through 0.45 μm filters into pre-combusted amber glass vials, and DOC was measured on a Shimadzu TOC-V carbon analyzer within 48 hours. To examine non-conservative nutrient behavior, the ratio of nutrient/chloride along the experimental reach normalized to a value of 1 at x = 0m was calculated using

where x is the distance downstream.

Stream discharge

During the experiment, stream discharge increased by 66% (from 10.5 to 17.4 L s⁻¹) down the 100 m study reach (Fig. 2). Measuring the change in stream discharge over the study reach was important for determining if the added reactive nutrients were simply diluted by inputs of melting ice or if they were retained during downstream transport. Most of the streamflow increase occurred in reach 1, across which streamflow increased by ∼45%. Along reach 2, stream discharge increased at a lower rate, presumably due to the steeper gradient and smaller contributing area.

Fig. 2. Discharge within the supraglacial stream increased throughout the experimental reach, most notably within the zone of the highest-contributing area.

The chloride measurements over the 3 hour injection period also suggested variable streamflow in response to solar inputs, as seen by the increasing Cl– concentrations towards the end of the 180min (Fig. 3). During the last 60 min of the injection period (from 120 to 180 min), the glacier was no longer receiving direct sunlight, which began to decrease lateral inputs of water into the channel, thereby increasing Cl– concentrations.

Fig. 3. Chloride observations were used to model the physical transport processes within the stream reach using a one-dimensional advection–dispersion model. The modeled fit closely matched the observations over the breakthrough curves within reaches 1 and 2.

Conservative solute modeling

The chloride observations at the end of the two study reaches provide the ability to calculate travel time and estimate the physical parameters of D, q _L, A, A_s and α for the OTIS model. Using the inverse modeling approach, we estimated a unique set of parameters for both stream reaches (Table 1). Although streamflow began to decrease at the end of the injection period, we modeled the stream using a constant discharge assumption since the decrease was only ∼8% over the injection period. We found that the first reach’s mean travel time was ∼13min, and the second reach’s travel time was only 2–3 min. The calculated velocity using the model parameters in the first reach was only 0.04ms⁻¹, in contrast to 0.34 m s⁻¹ in reach 2. This result is consistent with the channel geomorphology, and suggests that there is more opportunity for nutrient transformations in the first reach.

Our modeling results also confirm differences in transient storage between the first and second reaches. We found that inclusion of transient storage within reach 1 provided a statistically more robust model to explain the observed Cl⁻ tailing in the breakthrough curve. This result is consistent with the field observations of a pre-tracer dye used to plan for the injection. Water within the first reach was transported slowly, mixing with zones of much slower surface water that were acting as surface dead zones (Reference Bencala and WaltersBencala and Walters, 1983). Although transient storage within forested streams consists of both surface and subsurface storage, supraglacial streams are not expected to have subsurface storage owing to the frozen stream bed. In our study, reach 2 had neither surface nor subsurface transient storage due to the lack of surface water pools and the frozen stream bed.

Nutrient fate

There are relatively few reports documenting biogeochemical processes within supraglacial streams (Reference HodsonHodson and others, 2008). Supraglacial streams can be expected to have very low levels of C, N and P, given that the meltwater does not come into contact with substantial nutrient pools and thus largely reflects inputs via precipitation. These streams are in sharp contrast to streams within forested watersheds, where water entering the stream has likely been in contact with upland and riparian soils that can act as a source of organic and inorganic nutrients. Background nutrient measurements within our study stream were all near the instrument detection level (Table 2), with the exception of NO₃ ⁻ (0.7 μm), which was similar in concentration to the NO₃ ⁻ concentration measured in a supraglacial stream on Canada Glacier, Antarctica (Reference Fortner, Tranter, Fountain, Lyons and WelchFortner and others, 2005). In addition, the background supraglacial stream water at our site was so dilute that even concentrations of Cl⁻ were below instrument detection limits.

The nutrient injection was designed to determine if there was any short-term nutrient removal within the supraglacial stream environment. The NH₄ ⁺ and PO₄ ^3– injection resulted in steady-state concentrations of 18 μm NH₄ ⁺ and 4.4 μm PO₄ ^3– at the beginning of reach 1 (Table 2). DOC concentrations entering reach 1 were at a maximum value at 60 min (190 μm), and decreased in response to the change in the carbon injection input over the course of the experiment.

We found that the added NH₄ ⁺ and PO₄ ^3– were generally conservative over the course of the short-term experiment, as seen by the NH₄ ⁺/Cl⁻ and PO₄ ^3–/Cl⁻ ratios (Fig. 4). Our analysis shows that the nutrient addition did not stimulate growth, implying that there is limited net uptake of P and NH₄ ⁺ and likely little primary production. Because supraglacial streams frequently disappear within tens of meters into moulins or crevasses, the potential for measurable nutrient uptake is already limited by short travel times. In our study, the first experimental reach was specifically chosen because of the observed surficial transient storage zones, lower stream gradient and some observable sediment accumulation zones. The second reach was more typical of other observed supraglacial streams, with a narrow profile, steeper gradient and no bed sediments. Even within the first reach where calculated velocities were on the order of 0.04 ms⁻¹, there was no detectable removal of either NH₄ ⁺ or P. Thus it appears that the short travel times, combined with stream temperatures at or near 0°C and limited surface area for colonization, inhibit in-stream nutrient uptake.

Fig. 4. Normalized NH₄ ⁺ , PO₄ ^3– and NO₃ ⁻ to chloride ratios illustrate whether a solute is conservative, increasing or decreasing along the stream reach.

Although we are not aware of any previous efforts to quantify nutrient uptake within supraglacial streams, other extreme environments have been shown to support algal communities. For example, freeze-dried algal mats within Antarctic streams respond within days of receiving melting glacial water from upstream after being dry for >40 weeks (Reference McKnight, Niyogi, Alger, Bomblies, Conovitz and TateMcKnight and others, 1999). One large difference with these Antarctic streams is their relative stability over time in contrast to supraglacial streams, which frequently appear in new locations on the glacier surface each season. As a result, supraglacial streams lack the potential for accumulating a supportive substrate for long-term colonization by algal communities.

Other studies have found evidence for primary production within cryoconite holes on glacier surfaces (Reference Anesio, Hodson, Fritz, Psenner and SattlerAnesio and others, 2009). However, the higher hydrologic retention times as well as the presence of fine sediments (Reference Kaštovská, Elster, Stibal and ŠantréčkováKaštovská and others, 2005) within cryoconite holes make them more suitable for algal growth. Although algal communities within cryoconite holes were observed at our study site, the study stream had little fine sediment to support algal communities and no net N and P uptake. Our findings and observations suggest that biological activity in cryoconite holes likely has little influence on nutrient concentrations in supraglacial streams, except in circumstances where cryoconite holes are directly hydrologically connected to the stream.

Interestingly, nitrate increased through the experimental reach (as seen in the upward trend in NO₃ ⁻/Cl⁻ ratios; Fig. 4), even though nitrate was not injected into the stream. Concentrations more than doubled by the end of the reach to 2.4 mm NO₃ ⁻, even though streamflow nearly doubled over the same distance. Prior to the NH₄ ⁺ addition, background NO₃ ⁻ concentrations were 0.6 and 1.1 mm at the beginning and end of the first reach respectively (NH₄ ⁺ was non-detectable). Assuming conservative transport, the incoming lateral inflow into reach 1 would have needed concentrations of >2.7 μm NO₃ ⁻ to account for the observed increase in mass flux. An alternate assumption is that the incoming lateral inflow had an average concentration of 0.85 p.m, the average in-stream concentration prior to the experiment, and that nitrification produced NO₃ ⁻ within the stream. This assumption results in an estimated aerial stream-bed nitrification rate of 1.6 (μmoles m⁻² min⁻¹, which is at the lower end of the reported range for forested streams (0.1–6 (μmoles m⁻² min^{- 1} ) (e.g. Reference PetersonPeterson and others, 2001).

Previous studies have also documented ‘excess’ NO₃ ⁻ as evidence of nitrification in subglacial and ice-marginal environments in a variety of glacial ecosystems (Reference Tockner, Malard, Uehlinger and WardTockner and others, 2002; Reference Hodson, Mumford, Kohler and WynnHodson and others, 2005; Reference HodsonHodson 2006; Reference Williams, Knauf, Cory, Caine and LiuWilliams and others, 2007). Because there was no net NH₄ ⁺ removal along the reach, the organic matter mineralization rate would need to have compensated for any loss of NH₄ ⁺ through nitrification. These findings suggest that supraglacial streams have some resident nitrifying and heterotrophic bacteria, but further examination is needed to elucidate coupled N reactions.

A further indication of heterotrophic activity is that the added bioavailable dissolved organic matter (DOM) was not conservatively transported (Fig. 5). Although DOC/Cl ratios could not be used due to the absence of a plateau, the fate of DOM was analyzed within OTIS. Within reach 1, the conservative simulation initially matched the observations prior to the observed spike between 75 and 90 min, then over-predicted DOC concentrations through 165 min. We applied a first-order decay term in both the main water column and transient storage zone within reach 1 to account for the DOC overestimation from 90 to 165 min, and found the decay variable significant only in the main water column. The estimated decay, A, from the inverse modeling was 2 × 10–₄s⁻¹, translating to a mass transfer coefficient from the water to the sediment-water interface of 10 μm s⁻¹. At this rate, it would take a travel time of nearly 1 hour for the concentration to decrease 50%. Although there was no detectable decay in reach 2, the travel time was <3 min, which at the estimated decay rate would only result in a decrease of ∼4μm, which is within the uncertainty of the DOC measurements. Other studies measuring the removal of labile organic carbon (e.g. sucrose, acetate, glucose) have found mass transfer coefficients ranging from 9 to 146 μm s⁻¹ in forested streams (Reference Munn and MeyerMunn and Meyer, 1990; Reference Hall and MeyerHall and Meyer, 1998; Reference Wiegner, Kaplan, Newbold and OstromWiegner and others, 2005). Our measured mass transfer coefficient is on the low side of this range, suggesting lower heterotrophic metabolism in comparison to forested streams.

Fig. 5. DOC was modeled in reach 1 under two scenarios: conservative transport (solid curve) and transport including first-order decay (dotted curve).

The observed NO₃ ⁻ production and DOM removal, particularly within reach 1 , suggest that resident populations of chemotrophs and heterotrophs respond rapidly to short-term nutrient perturbations. These populations likely reside in the limited stream substrate, which was only visually identifiable in reach 1 . The measured DOM removal in the supraglacial stream we studied was likely due to mineralization of the added carbon by heterotrophic organisms. Several recent studies have documented the export of labile, protein-rich DOM from glaciers in a variety of environments (Reference Barker, Sharp, Fitzsimons and TurnerBarker and others, 2006; Reference HoodHood and others, 2009; Reference Bhatia, Das, Longnecker, Charette and KujawinskiBhatia and others, 2010). Other research has documented the presence of heterotrophic species at the glacier–sediment interface (e.g. Reference Skidmore, Foght and SharpSkidmore and others, 2000). Taken together, these findings suggest that there may be both production and consumption of organic matter along hydrologic flow paths within glaciers.