Despite a fivefold rise in nitrate concentration over the last century, many fundamental aspects of Lake Superior's N and C cycles are still very poorly understood. We present here the first measurements of inorganic N uptake and in situ C uptake rates in Lake Superior, one of the largest lakes in the world. A profile of C uptake suggests that more than 95% of production occurs in the top 30 m with highest productivity to biomass ratio in the epilimnion. High C uptake:N uptake and particulate C: N ratio compared to the Redfield ratio (6.6) in the epilimnion suggests higher turnover rate of C compared to N in epilimnetic particles. Experiments performed over a range of typical environmental conditions suggest a strong temperature dependence of N uptake with maximum rates observed during the warmest stratified period. Lakewide N uptake estimates derived from a temperature-based model suggest that on an annual basis, uptake is considerably higher than total N inputs from outside the lake. This difference indicates that the lake is recycling N rapidly, leading to a shorter turnover time in the water column than previously assumed. The long-term buildup of nitrate in the lake has been hypothesized to arise from limited assimilation of nitrate entering the lake. In contrast, our results suggest that nitrate accumulating in the lake is a result of internal N cycling, a finding consistent with recent studies based on a nitrogen budget and NO3− stable isotope analyses.
 Lake Superior is Earth's largest lake by area and holds almost 10% of Earth's liquid surficial freshwater [Hecky, 2000]. As the headwater lake of the Laurentian Great Lakes system, it has regional and global importance, yet many basic ecosystem properties remain poorly studied. Lake Superior's nitrogen (N) and carbon (C) cycles have received scant attention, with very few measurements yet reported of in situ N or C transformation rates. Sources and sinks of these two important elements are therefore poorly constrained and current models and budgets are unable to explain the reasons for the observed imbalances in N [Sterner et al., 2007] and C [Cotner et al., 2006; Urban et al., 2005]. A striking feature of the biogeochemistry of Lake Superior is the steady, century-long trend in nitrate concentration, which has increased by almost fivefold (∼5 to 25 μM) since the early 1900s to the present [Bennett, 1986; Sterner et al., 2007]. The nitrate increase and continued low phosphate have led to a severe stoichiometric imbalance resulting in a rising NO3−: PO43− ratio that in 2006 was around 10,000 (moles: moles) compared to the Redfield ratio of 16 [Sterner et al., 2007]. There is also a reported large imbalance in the C cycle of the lake, with measured respiration rates exceeding estimated inputs of organic C by between two to ten fold [Cotner et al., 2006; Urban et al., 2005].
 Nitrogen is one of the most bioactive elements and it is often limiting in terrestrial and aquatic systems [Elser et al., 2007]. Globally, inputs of reactive N to the biosphere have increased ∼ twofold compared to preindustrial periods [Galloway, 1998] due to accelerated anthropogenic activities. In Lake Superior, an increase in nitrate concentration is not only conspicuous but also unusual in its steady rise. Based on a linear model, nitrate concentration in Lake Superior has been increasing at a rate of 0.16 μM per year for decades [Sterner et al., 2007]. There is no evidence for a similar long-term rise in any other nutrient element or inorganic chemical species in the lake [Weiler, 1978]. Increasing nitrate concentrations in many terrestrial and aquatic systems are explained by increased anthropogenic activities [Galloway, 1998]. However, the build up of nitrate in Lake Superior is a puzzle due to the lack of sufficiently large anthropogenic sources of N in its watershed and airshed to account for the lakewide increases. Although some studies have attributed the nitrate increase in the lake to deposition of increased NOx emissions [Bennett, 1986; Ostrom et al., 1998], recent studies of the lake's N budget [Sterner et al., 2007] and natural abundance isotopic measurements of N and oxygen in nitrate in both the lake and precipitation [Finlay et al., 2007] reject this explanation. Instead, these studies suggest that the nitrate build up is largely due to in-lake nitrification of reduced forms of N [Sterner et al., 2007; Finlay et al., 2007]. In situ N dynamics in Lake Superior are still poorly understood, however N transformations in the lake need appropriate quantification to provide additional tests of hypotheses regarding the source of the nitrate increase in the lake. To date, there is no direct estimate of nitrification rates for Lake Superior. There is indirect evidence of nitrification at the sediment-water boundary with an estimated annual efflux of 44–20 × 108 moles of nitrate per year [Heinen and McManus, 2004].
 One of the most important terms in the N cycle of lakes is uptake of inorganic N (i.e., nitrate and ammonium) by phytoplankton and bacteria. Knowledge of uptake rates and their environmental controls can provide insights into the controls of N dynamics of lakes. Uptake of N by phytoplankton is regulated primarily by light and substrate availability [Gardner et al., 2004]. Under similar conditions, the energy requirements for ammonium assimilation are much less than for nitrate assimilation, because ammonium is already in reduced form. Nitrate uptake is also known to be inhibited by ammonium in marine [Eppley et al., 1969] as well as fresh water [Prochazkova et al., 1970] systems.
 Nitrogen uptake in the Laurentian Great Lakes has seldom been measured. A nitrogen flux study performed in southern Lake Michigan reported coupled ammonium uptake and regeneration with rates of uptake of ∼0.12 mmol N m−3 d−1 (based on average light uptake) and regeneration rates of ∼0.07 mmol N m−3 d−1 [Gardner et al., 2004]. No temporal pattern in N cycling rates was observed in Lake Michigan despite variation in mean temperature from 1.5 to 11°C from March to June. Measurement of inorganic N uptake rates in Lake Ontario (2 stations) and Lake Erie (3 stations) reported variation in nitrate uptake from <0.0008 mmol N m−3 d−1 (Lake Erie) to 0.360 mmol N m−3 d−1 (Lake Ontario) [Murphy, 1980]. Ammonium uptake in Lake Erie ranged from 0.026 mmol N m−3 d−1 to 2.14 mmol N m−3 d−1. However, apprehension was expressed regarding the measured ammonium uptake rates in Lake Erie and Lake Ontario due to uncertainty in the ambient ammonium concentrations [Murphy, 1980]. To the best of our knowledge, there is only one study related to biological N fixation in Lake Superior [Mague and Burris, 1973]. In this study, N2-fixation was estimated using the acetylene reduction technique. Very low reduction activity of acetylene was found and the authors concluded that N2 -fixation in offshore Lake Superior was very low, a finding consistent with the phosphorus (P) poor but N rich nature of the lake [Mague and Burris, 1973].
 Reconstruction of a comprehensive C budget using published data indicates a large imbalance in the organic C cycle of the lake. According to current estimates, organic C disappears at much faster rate (14– 40 Tg a−1 [Cotner et al., 2006]; 13–81 Tg a−1 [Urban et al., 2005]) than rates of inputs (5.3 Tg a−1 [Cotner et al., 2004]; 3–8 Tg a−1 [Urban et al., 2005]). Though the organic C cycle in the lake undoubtedly is affected by subsidy of terrestrial dissolved organic carbon (DOC) runoff, this gross imbalance may reflect an underestimation of production due to technical aspects of the 14C method (particularly for some earlier studies) or insufficient study of the DCM and during cold dark conditions and in deeper layers of the lake.
 In the present study we quantify for the first time planktonic N uptake dynamics and in situ C uptake rates in Lake Superior. Our specific goals were to (1) estimate the rates of uptake of nitrate and ammonium in the lake, (2) examine the effect of light, temperature, and nutrient (P and Fe) addition on N uptake, (3) generate a lakewide estimate of annual N uptake and relate it to the lake's N cycle, (4) compare the N and C uptake rates and their stoichiometric ratios, and (5) develop a conceptual model to explain observed increases in nitrate levels in the lake over the last century.
2. Materials and Methods
2.1. Site Description
 The largest of the Laurentian Great Lakes by area and volume, Lake Superior is an internationally important natural and economic resource. The present study was carried out during 2005 and 2006 at locations distributed to provide representative coverage of major regions of the lake (Figure 1).
 Lake Superior has the lowest annual average lake temperature and summer surface temperature of the Laurentian Great Lakes [Bennett, 1978]. Typical temporal variation in the surface temperature of Lake Superior observed at CD-1 shows that the lake is stratified for almost 24% of the year and is well mixed otherwise (Figure 2). Over the period 1980 – 2005, lake temperature increased and winter ice cover decreased [Austin and Colman, 2007]. During the present study, the lake was well mixed during May with water column temperature around 3.4°C. The lake started to stratify on the southern margin during June (sites ON-2 and STE-I during 2005 and CD-1 during 2006) while remaining well mixed at others (PI-1). Surface temperature during June varied from 3.4°C at PI-1 during 2005 to 13.6°C at CD-1 during 2006. Stratification was lakewide during July and August with surface temperature varying between 11.6°C at RM and 19°C at SI-1. In general, the thermocline was around 20–40 m during July and August, except EM where it was 10 m.
 Ambient nitrate, ammonium and total dissolved nitrogen (TDN) were measured during the study period. Nitrate was measured using an autoanalyzer (AlpKem) based on standard colorimetric technique modified after Strickland and Parsons . A recently developed fluorometric technique [Holmes et al., 1999] was used for ammonium measurement [Kumar et al., 2007]. A Shimadzu TOC-Vcsh using standard sensitivity catalyst was used for TDN.
 Lake Superior is a nitrate-rich environment with a current (i.e., for 2005–2006) average concentration of around ∼25 μM [Sterner et al., 2007]. TDN in the lake is dominated by nitrate (90%). Analysis of lakewide nitrate data in the top 30 m of water column revealed a drawdown of 1.30 ± 0.88 μM (95% confidence intervals) from June (average ∼23.9 μM; n = 35) to August (average ∼22.6 μM; n = 182) during 2005 (data not shown). There was no significant spatial variation in nitrate concentration of the open lake. Ammonium in Lake Superior averages around 0.21 μM [Kumar et al., 2007]. Vertical profiles of nitrate and ammonium concentration, based on new data, observed during 2006 are shown in Figure 3. It is noteworthy that nitrate concentration in the lake is almost 125 times greater than ammonium.
 Chlorophyll a was estimated using standard fluorometric measurement after 90% acetone extraction [Welschmeyer, 1994]. Dissolved inorganic carbon (DIC) was measured via acidification followed by detection of CO2 by infrared gas detection using a Shimadzu analyzer. Samples for particulate C, N, and P were collected by filtering 1 – 1.5 L lake water though 25 mm GF/F filters. Filters for particulate P were rinsed with 1% HCl prior to sample filtration. Particulate C and N were measured using a CHN analyzer. Particulate P samples were digested with potassium persulfate and analyzed on a Shimadzu UV160U Spec or on an OI Analytical AlpKem, both using the ascorbic acid method.
2.2. Experiments and Analysis
 During 2005 (June, July, and August), N uptake experiments were performed using treatments designed to simulate conditions at several depths within the upper ∼23 m of the water column to assess values and controls on surface-layer N uptake. During 2006 (May, June, and August), both N (all stations) and C (EM and SI-1) uptake experiments were performed at multiple depths in situ to better understand the vertical structure of N uptake in the lake and to begin to understand linkages of C and N dynamics.
2.2.1. Nitrogen Uptake Experiments
 Nitrogen uptake experiments were performed at 10 sites during 2005 and 3 sites during 2006 (Table 1). Due to differences in goals between the years, protocols differed in terms of incubation technique and duration. During 2005, following the Joint Global Ocean Flux Study (JGOFS, June 1994- chapter 17) protocol, on-board incubations were performed using several light levels and a 4–6 h incubation period. These 2005 experiments were performed with water from three depths corresponding to 100, 50, and 10% light levels (approximately 0, 7, and ∼23 m depths) at each station. No dark incubations were done. In 2006, in situ incubations of duration 5–17 h were performed. These 2006 experiments were performed at six to eight depths (5, 10, 20, 30, 40, 60, 80, and 100 m) except at SI-1 where only four depths were studied (5, 10, 20, and 40 m) due to shallow station depth (∼63 m).
Table 1. Column-Integrated Nitrate, Ammonium, and Total N Uptake Rates Along With f-Ratio at Various Locations in Lake Superiora
Uptake Rate (mmol N m−2 d−1)
Uptake rates in surface waters (mmol N m−3 d−1) at different locations are within parentheses. The protocols for N uptake measurements differed between 2005 and 2006 (see text).
7 Jun 05
7 Jun 05
8 Jun 05
20 Jul 05
21 Jul 05
22 Jul 05
22 Jul 05
28 Aug 05
29 Aug 05
29 Aug 05
30 Aug 05
31 Aug 05
31 Aug 05
23 May 06
20 Jun 06
9 Aug 06
9 Aug 06
 Water samples were collected from Niskin water samplers into 2-L acid-washed polycarbonate Nalgene bottles in duplicate for ammonium and nitrate uptake at each depth. Incubation bottles were covered immediately with dark cloth to minimize light shock to the plankton. Samples were spiked with 98 atom%+ enriched K15NO3 or 15NH4Cl to trace nitrate and ammonium uptake respectively. A constant addition of 2 μM for nitrate and 0.02 μM for ammonium was made to each sample. Since nitrate concentration in Lake Superior is around 25 μM with relatively little variation in space and time, the tracer addition for nitrate was always ≤10% of ambient concentration. Due to lack of previous knowledge about the low-level concentration of ammonium in Lake Superior, tracer addition for ammonium at some depths exceeded the target of 10% of ambient concentration, with a maximum of 33% at EL-1 during July 2005.
 We also performed experiments to estimate the effect of ammonium concentration on uptake rates at three stations during June 2005 (ON-2, PI-1, and PW-1) by adding an additional 0.1 μM of labeled ammonium to some treatments. Size fractionation experiments to determine the contribution of smaller plankton to N uptake were performed at EL-3 and EM during August 2005 where some samples were passed through 10 μm Nitex mesh before being filtered on GF/F. We also performed experiments to estimate short-term responses to added P (1 μM) and Fe (55 nM) on nitrate and ammonium uptake rates at all stations at the 50% light level during 2005.
 During 2005, following tracer addition, samples were incubated in an on-deck water bath after putting the appropriate neutral density filters to simulate the ambient light conditions inside the bottles. The temperature of the water bath was within ±1.5°C of the lake surface. During 2006, sample bottles were strapped to a wire basket attached to a cable suspended from a float. A second smaller float carried a radar deflector and strobe light to allow for recovery. Post incubation samples were filtered on precombusted filters as described for particulate nutrients. Samples were stored frozen and later dried for mass spectrometric analysis where 15N atom% and particulate N content were measured using continuous flow stable isotope mass spectrometer (Fisons Optima) connected to elemental analyzer (Carlo Erba Model NA 1500). The coefficient of variation in 15N atom% measurement was around 1% and 3% for nitrate and ammonium uptake samples respectively. The average difference in PON measurement of duplicate samples was less than 8%. The uptake rate was calculated using the equation [Dugdale and Wilkerson, 1986; Kanda et al., 2003],
where P is the amount of particulate N in post incubation sample, Δ Ip is the increase in 15N atom% in particulate N during incubation, Sa and St are ambient and added nitrate (or ammonium) concentration, respectively, Ir and I0 are 15N atom% of added tracer and natural 15N atom%, and T is the incubation time. No time course experiments were performed to correct ammonium uptake rates for isotope dilution [Laws, 1984]. An underestimation of ammonium uptake rates by 6 – 35% has been observed in large temperate oligotrophic lakes due to isotope dilution [Dodds et al., 1991]. Therefore, ammonium uptake rates reported here may be underestimated.
 Column integrated uptake rates were calculated using the trapezoidal integration method from the surface to the 10% light level (∼23 m) during 2005 and from the surface to 100 m during 2006 (except SI-1 where it was 40 m depth). We considered the sum of nitrate and ammonium uptake rate as the total N uptake and the ratio of column integrated nitrate uptake to column integrated total N uptake as the f-ratio. Classically, the f-ratio is defined as the fraction of total N uptake due to sources other than regeneration within the photic zone and has been taken to be an estimate of export production [Eppley and Peterson, 1979]. Due to low vertical and horizontal gradient of nitrate and poor understanding of seasonal upwelling, if any, in the water column of Lake Superior, we used the f-ratio simply to measure the percentage contribution of nitrate uptake to total N uptake, and not to interprete export production.
2.2.2. Carbon Uptake Experiments
 Carbon uptake was measured at EM and SI-1 during August 2006 using the 14C method (JGOFS protocol 1994 - Chapter 19). Before sunrise, water was collected using Si-gasket-equipped Niskin bottles from similar depths as N uptake experiments. For each depth, 250 ml of sample water was distributed into five acid-washed polycarbonate bottles. Spikes of 14C-HCO3 were delivered to each bottle to equal 7 μCi l−1. One bottle was filtered immediately and used as a T0 correction. One bottle was wrapped with Al-foil and strapped to the inside of a wire basket along with three light bottles and these were returned to the lake to incubate as mentioned earlier. Incubation time was 17 h. After sunset, the bottle array was retrieved and the entire contents of all bottles were filtered onto GF/F, after first removing a 0.25 × 10−3 l aliquot to measure the total 14C activity. Production was calculated using measured values for DIC and correcting for T0 and dark uptake. For depths ≤ 30 m, the T0 and dark uptake corrections each were <5% of the final values. For the deep measurements, T0 corrections were ≈30% of the final value, and uptake rates measured in the dark bottles were approximately equal to rates measured in the light bottles.
 Production was calculated using the following formula:
where SDPM is DPMs in filtered sample, V is volume of filtered sample (l), TDPM is total 14C DPMs (in 0.25 ml), W is DIC concentration (mg l−1), and 1.05 is the correction for lower uptake of 14C compared to 12C, and T is time in days. Dark uptake rates measured during part of a 24 h period were assumed to apply for a full 24 h; thus dark rates were divided by T/24 where T was the incubation time as above. Measured light uptake rates represent one full daylight period and thus were not adjusted.
3.1. The 2005 Nitrogen Uptake
 During 2005, nitrate and ammonium uptake rates were determined at 11 station occupations across the lake on three cruises (June, July, and August; Table 1). This broad coverage was intended to allow exploration of spatial and temporal variability and to test for important physical and chemical controls on N uptake. Nitrate uptake rates in surface waters ranged from 0.001 mmol N m−3 d−1 at PI-1 during June to 0.030 mmol N m−3 d−1 at CD-1 during July. Ammonium uptake in surface waters ranged from 0.004 at STE-I during June to 0.062 mmol N m−3 d−1 at CD-1 during July.
 Column integration (surface to ∼ 23 m) of N uptake rates was performed to compare N uptake in the upper water column. Column-integrated total N uptake rates over all stations and over three months averaged ∼ 1.18 mmol N m−2 d−1 (s.d. = 0.65; station-wise data shown in Table 1). Average column-integrated nitrate and ammonium uptake rates during this period were 0.30 (s.d. = 0.17) and 0.89 (s.d. = 0.49) mmol N m−2 d−1 respectively. The variability around these means reflects geographical and seasonal variability. The nitrate (p = 0.0073), ammonium (p = 0.0068), and total uptake (p = 0.0068) rates were significantly different between months (ANOVA), though this difference should be interpreted cautiously because different sites were studied during each months. The average total N uptake rate was lowest during June (0.42 mmol N m−2 d−1), highest during July (1.75 mmol N m−2 d−1) and decreased again during August (1.19 mmol N m−2 d−1).
 Ammonium uptake rates at nearshore stations (ON-2 and PI-1) were higher than at the open lake station (STE-I) during June (Table 1). Nitrate and total uptake rates during June did not show much difference between nearshore and open lake stations. However, the opposite was observed during August where nitrate, ammonium, and total N uptake rates at the nearshore PW-1 station were the lowest.
3.2. The 2006 Nitrogen and Carbon Uptake
3.2.1. Nitrogen Uptake
 In 2006 a different protocol was used to better resolve the depth dependence of N uptake as well as to measure the coupling between N and C uptake in the water column. The lake was unusually warm and stratified earlier in 2006 than in previous years (Figure 2). Hence, values from 2006 may not necessarily be easily comparable to 2005. In situ N uptake was measured during four station occupations (Table 1) and in situ C uptake was measured at two of those, one nearshore and one offshore (Table 2). At offshore site EM in August, the water column was stratified with a thermocline at 10 m and a DCM at ∼40 m, at approximately the base of the metalimnion (Table 2). Nitrate was ∼2–3 μM lower in the epilimnion than at lower depths whereas ammonium and DIC showed little variation with depth. Particulate C:N did not vary systematically with depth whereas C:P was clearly elevated at depths above the DCM. The chl a:C ratio was maximum in the DCM. Nearshore site SI-1 was also stratified, and showed similar patterns in POC, PON, chl a and elemental ratios as site EM (Table 2).
Table 2. Dissolved and Particulate Nutrient Concentrations Observed at EM and SI-1 During 2006 in Situ Experiments
Dissolved DIC (mg/L)
Chl a (μg/L)
 Column-integrated (surface to 100 m except SI-1) N uptake rates (nitrate, ammonium as well as total N) during 2006 were highest during June (CD-1; Table 1). Uptake rates in June were almost twice as high as the previous month (Table 1). The minimum uptake rate was observed at SI-1 in August. Since the column integration depths during the two study years were different, integration to ∼20 m was calculated for 2006 to compare to 2005. Comparison of integrated uptake rates for this similar depth range revealed nitrate uptake rates to be very similar (∼0.30 mmol N m−2 d−1) for both years. Ammonium uptake rates however were higher in 2006 than in 2005; average ammonium uptake increased from 0.89 mmol N m−2 d−1 during 2005 to 1.55 mmol N m−2 d−1 during 2006.
 Vertical profiles of nitrate and ammonium uptake indicated significant uptake in the top 40 m for nitrate and the top 60 m for ammonium (Figure 4). Although the rates for both forms at lower depths were very low, N uptake did not cease entirely even at 100 m. We are unsure whether the seemingly anomalous value of high nitrate uptake at 100 m is a reliable observation. However, higher nitrate uptake at depth > 75 m has been commonly observed in the subtropical ocean [Painter et al., 2007]. During each year, nitrate and ammonium uptake were positively correlated with a stronger relationship during 2005 (R2 ∼ 0.71) than 2006 (R2 ∼ 0.55; Figure 5).
3.2.2. Carbon Uptake
 For C uptake, we focused primarily on the primary productivity profile of the open lake site EM. We observed highest volumetric C uptake in the epilimnion, decreasing values through the thermocline, and near-zero uptake at 60 m and below (Figure 6 and Table 3). Peak volumetric productivity occurred at 10 m; well above the biomass peak at 40 m. Uptake to chlorophyll a ratios were high in the epilimnion and decreased with depth (Table 3). More than 95% of the areal primary production occurred at depths ≤30 m. Volumetric productivity integrated numerically from 0 to 100 m (5 m value extrapolated to the surface) was 30.8 mmol C m−2 d−1 (i.e., 369 mg C m−2 d−1). Similarly, at site SI-1, volumetric productivity was contained within the upper 20 m of the water column (Table 3). Productivity at 40 m at SI-1 was near-zero (not shown in Table 3).
Table 3. Nitrate, Ammonium, Total N, and Inorganic Carbon Uptake Rates at EM and SI-1
Uptake Rate (μmol N m−3 d−1)
C Uptake (μmol C m−3 d−1)
C Uptake: N Uptake
C Uptake: Chl a (μgC (μg Chl)−1 d−1)
 We compared C and N uptake versus depth for both open (EM) and nearshore locations (SI-1; Table 3). At EM, all uptake values generally decreased with depth, except for one relatively high nitrate uptake value at 100 m. Nitrate and C uptake decreased faster with depth than ammonium uptake so that nitrate:ammonium and C:N uptake ratios fell with depth. C:N uptake ratios in the surface layer were higher than the Redfield ratio (6.6) and higher than observed particulate C:N ratios for both EM and SI-1 (Table 3). C:N uptake ratios were close to the Redfield ratio around the biomass peak.
3.3. Nitrate: Total Uptake
 In spite of considerable spatial and temporal variability in absolute uptake rates, the ratios of nitrate: total uptake (f-ratio) observed during different months of 2005 were very similar to one another, averaging 0.25 (s.d. = 0.07; Table 1). Values during 2006 were similar but somewhat lower, averaging 0.17 (s.d. = 0.03; Table 1). The f-ratios were least consistent during June 2005 (ranging from 0.06 at PI-1 to 0.37 at STE-I). During 2006, the f-ratio was lowest at CD-1 during June (0.12) and greatest at the nearshore location SI-1 (0.20) during August. Vertical profiles of the f-ratio during 2006 suggest higher values in deeper waters at EM (0.57 at 100 m) and CD-1 (0.55 at 80 m during May; Figure 7).
 To assess the effect of ammonium concentration on uptake, two levels (0.02 μM and 0.1 μM) of ammonium additions were made. A small increase in uptake rate was observed but the difference was not significant (ANOVA; p = 0.14, n = 12; figure not shown). The low sensitivity of uptake to ammonium concentration suggests that it is constrained by some other factor, perhaps P and/or Fe limitation.
3.4. Size Structure of Uptake
 Two size-fractionated N uptake measurements indicated that 76% (EL-3) and 85% (EM) of total N uptake in the top 23 m of the water column was due to plankton of size range less than 10 μm (Table 1). This small size fraction accounted for 94% of ammonium uptake and a lower percentage (56%) of nitrate uptake at EM. This contribution was 72 and 78% for nitrate and ammonium respectively at EL-3.
3.5. Effect of Light and Temperature on N Uptake
 Multiple factors including light, nutrients and temperature, as well as algal biomass, could be important in regulating N uptake dynamics of Lake Superior. We used results from 2005, where we studied a wider range of conditions compared to 2006, to assess the effect of light, temperature, and biomass on uptake rates and to construct the lakewide N uptake estimate.
3.5.1. Effect of Light
 Nitrate uptake rates did not show significant response to light intensity at most of the stations during 2005 (Figure 8). In contrast, ammonium uptake rates showed a light inhibition, usually decreasing considerably with increasing light levels (Figure 8). One exception occurred during June at nearshore station ON-2 which showed an increase in nitrate and ammonium uptake from 10% to 50% light level (Figure 8a). When data from all three months were analyzed together, the results were in agreement, although not significant, with higher response of light on ammonium uptake (p = 0.06) than nitrate uptake (p = 0.93).
3.5.2. Effect of Temperature
 A strong positive relationship (R2 = 0.55, n = 30) between volumetric (depth-wise) N uptake rates and incubation temperature was observed (Figure 9). This relationship was slightly better for nitrate (R2 = 0.62, n = 30) than ammonium uptake (R2 = 0.49, n = 30). A similar trend was also observed for column-integrated (∼23 m) uptake rates and incubation temperature (R2 = 0.52 for total N; figure not shown). A positive relationship between volumetric uptake rates and temperature was also observed during 2006 (data not shown) but was relatively weak (R2 = 0.33) due to limited data and scatter at higher temperature.
 Multiple regressions relating volumetric uptake rates to light, incubation temperature, and chlorophyll a indicated a highly significant relationship between temperature and N uptake rates (Table 4). Light was significant only for ammonium uptake whereas chl a was not significant at all. Backward stepwise variable section revealed temperature to be the most important factor affecting uptake rates in Lake Superior. In addition to the multiple regression model with each of the three predictor variables, models were run with all two-factor interaction terms (temp x light, light x chlorophyll, and chlorophyll x temp). None of the interaction terms were significant. The three predictor variables were only modestly correlated with one another (R2 < 0.35) and tolerance tests indicated acceptable redundancy in the regression.
Table 4. Multiple Regression Parameters of Uptake Rate, Percent Light, Temperature, and Chl a During 2005a
Uptake Rate (mmol N m−3 d−1)
L, T, and C denote percent light, temperature (°C), and Chl a, respectively. The β parameters represent the coefficients for the slopes with standardized variables (mean = 0, s.d. = 1), which allows for comparison of the relative contribution of each independent variable in the prediction of the dependent variable.
 Lake Superior is a nitrate-replete environment with poorly understood N and C biogeochemistry, which has led to speculations regarding the reasons for rising nitrate levels in the lake. Recent studies have indicated that internal N cycling in the lake is more important in the nitrate rise than previously assumed [Sterner et al., 2007; Finlay et al., 2007]. However, rates of many internal N-cycling processes in Lake Superior are not properly quantified or not known at all. The uptake of N is one such important rate. During the present study, we found that nitrate and ammonium uptake rates in Lake Superior are considerably lower than those observed in warm, eutrophic lakes [e.g., Gu et al., 1997], but close to those observed in other Great Lakes [Murphy, 1980; Gardner et al., 2004]. Ammonium uptake in Lake Superior was almost three times higher (Table 1) than nitrate uptake, but the positive relationship between the two (Figure 5) suggested that similar factors influenced uptake of both forms of N. Both nitrate and ammonium uptake rates showed increase with rising temperature, indicating N uptake to be greater in the surface layer during summer than during onset of spring.
4.1. Interannual Variability
 Nitrogen uptake rates were estimated during 2005 and 2006. Although nitrate uptake rates were similar for both years, we observed higher ammonium uptake rates (20 m water column) in 2006 compared to 2005. This interannual variability could be attributed to several factors. First, a limited number of stations were sampled during 2006 and spatial variability could have biased our estimates. Second, interannual variability in the lake's biological and physical conditions. The lake, in general, was warmer during 2006 compared to 2005 (Figure 2). However, algal biomass and thermal conditions at EM, a station sampled during same month (August) of both years, were similar. Surface temperatures at EM were 18.1°C and 17°C during August of 2006 and 2005, respectively, whereas ammonium uptake during 2006 (1.32 mmol N m−2 d−1) was considerably higher compared to 2005 (0.97 mmol N m−2 d−1) for 20 m water column. Third, and perhaps most likely, factor could be difference in methods used during each year. Overall, the two years with different methods provide a range of plausible uptake rates in the lake and allow for further testing of the importance of in-lake N cycling in terms of the nitrate rise in Lake Superior.
4.2. Preference for Ammonium in Lake Superior
 The f-ratios observed in Lake Superior indicate that approximately 75% of the N sustaining biological production is supported by the small pool of ammonium, with the remainder sustained by the much larger pool of nitrate. A preference for ammonium relative to nitrate is consistent with observations from other large oligotrophic lakes [Dodds et al., 1991], though the striking consistency of the f-ratios we observed in Lake Superior were unexpected and at present cannot be satisfactorily explained.
 Low levels of P and Fe along with cold and dark conditions may be an important factor in lower nitrate utilization by the plankton. Our work suggests that the limited availability of P and Fe keeps the biomass pool low in the lake and the small ammonium pool is replenished quickly enough to sustain this biomass, leaving the large nitrate pool less utilized. Measurement of nitrate utilization in Lake Superior by monitoring bioreporter luminescence revealed that nitrate utilization was often enhanced following addition of P [Ivanikova et al., 2007]. Enhancement in the P effect due to Fe addition at many stations led to the conclusion that P-limited algae are deficient in their ability to assimilate nitrate in Lake Superior. Phosphorus limitation is known to control the utilization of excess nitrate in estuarine systems as well [Kedong et al., 2004].
 During the present study, however, additions of P and Fe did not increase the rate of N uptake within the time of incubation for experiments (4–6 h; data not shown). This short interval of time probably was not long enough to alleviate possible P or Fe limitation on uptake of nitrate and ammonium.
4.3. Role of Smaller Plankton
 Smaller size phytoplankton and bacteria (<10 μm) were of major importance in nitrate and especially ammonium uptake during the present study, consistent with certain other marine and freshwater systems [Wheeler and Kirchman, 1986; Suttle and Harrison, 1988; Dodds et al., 1991] as well as Lake Superior. Small phytoplankton (<5 μm) can contribute almost 50% and the <1 μm size class accounts for almost 20% of total primary production in Lake Superior [Munawar et al., 1978; Fahnenstiel et al., 1985]. In large oligotrophic lakes, smaller size phytoplankton (<3 μm) have also been observed to play a major role in ammonium regeneration along with uptake [Dodds et al., 1991]. However, their role in ammonium regeneration is unknown in Lake Superior. Higher ammonium uptake compared to nitrate uptake in the smaller size fraction, as observed during the present study, has also been observed in regions such as the North Sea [Rees et al., 2002].
 Heterotrophic bacteria also assimilate ammonium [Laws et al., 1985] and their contribution can be as high as 70% of total ammonium uptake as observed in Lake Michigan [Gardner et al., 2004]. We did not separately estimate the effect of heterotrophic bacteria on ammonium uptake in Lake Superior. The importance of bacteria to biogeochemical processes in oligotrophic waters of Minnesota, including Lake Superior, compared to eutrophic lakes is known [Biddanda et al., 2001]. Bacteria may have a significant role in N uptake dynamics of the lake and this issue warrants further attention in future studies. Also, the extent of dissolved organic nitrogen as an alternative source of N was not considered in this study and could be a component of N uptake in the lake. In general, the significant contribution of small plankton to total N uptake found here is consistent with both carbon based productivity measurements in Lake Superior and results from other oligotrophic systems.
4.4. Effect of Light and Temperature
 We observed a decrease in ammonium uptake with increasing light (Figure 8). This probably demonstrates the potential for higher ammonium uptake in the subsurface layer compared to surface, also evident by subsurface ammonium uptake maximum at majority of stations during 2006 (Figure 4). Earlier studies from Lake Superior, which concentrated mostly on the effect of light on surface communities, also observed a decrease in photosynthesis with increasing light [Nalewajko et al., 1981; Nalewajko and Voltolina, 1986]. No significant effect on nitrate uptake rate with increasing light level was observed during the present study. The relatively higher influence of light on ammonium uptake compared to nitrate uptake during the present study is contrary to observations in another large oligotrophic lake [Dodds and Priscu, 1989] where nitrate uptake was found to be relatively more influenced by photosynthetic photon flux density. Suppression of both nitrate and ammonium uptake by intense light has been observed in freshwater [Murphy, 1980; Dodds and Priscu, 1989] as well as in marine environment [McCarthy et al., 1999]. However, we cannot make any such conclusion for Lake Superior based on the present study. Greater numbers of light level manipulations using phytoplankton communities from surface as well as deeper layers may be needed to comprehensively study the effect of light on N uptake rates in Lake Superior.
 Our experimental results point to an overriding importance of temperature in controlling N uptake rates. This conclusion is consistent with other studies of plankton metabolism in Great Lakes, including Lake Superior, where temperature has been found to be the single most important factor [Schelske et al., 1974; Nalewajko and Voltolina, 1986]. Maximal algal growth rates in Lake Superior have also been observed at warmer temperatures [Sterner et al., 2004]. Similarly, seasonal analysis of N uptake rates from another large oligotrophic lake revealed lowest nitrate and ammonium uptake rate during winter and low temperature was concluded to be one of the responsible factors [Dodds and Priscu, 1989]. Due to the predictive power of temperature on N uptake, lakewide estimates based solely on temperature are a reasonable first approach to determine annual N uptake in this lake.
4.5. Lakewide Uptake Estimates and Significance for the N Cycle of Lake Superior
 With these first measurements of nitrate and ammonium uptake rates in Lake Superior, it is now possible to explore how N uptake relates to water column N cycling and the nitrification of Lake Superior. We estimated lakewide annual uptake rate using a temperature-dependent model. A linear model of the relationship between uptake and temperature (Figure 9) was found to be inadequate because when temperature is around 0°C, total uptake is negative. We therefore fit a Boltzman function (Loge (uptake rate) = β1-β2(1/T), where T = temperature in Kelvin) to the data. The relationships obtained for nitrate uptake was [Loge (nitrate uptake) = – 15765 × T−1 + 50.6; R2 = 0.75, n = 30] and for total uptake was [Loge (total uptake) = – 11222 × T−1 + 36.1; R2 = 0.63, n = 30]. Next, using water column temperature data from station CD-1 for different months (from cruises conducted between 1996 and 2005), we used this model to estimate the annual N uptake assuming 30, 50 or 100 m of water column as the effective uptake zone (Table 5).
Table 5. Comparison of Annual Uptake Rates of Nitrate and Total Nitrogen Integrated Over 30, 50, and 100 m Water Column Based on a Temperature-Dependent Estimatea
Uptake (mmol N m−2 a−1)
Ammonium uptake is the difference between total and nitrate uptake.
 The annual total N input to the lake, including rivers and precipitation, is around 6,614 × 106 mol a−1 [Hecky, 2000] with nitrate accounting for ∼38% (2,513 × 106 mol a−1) and ammonium ∼ 25% (1,653 × 106 mol a−1) [Sterner et al., 2007]. Even considering only the shallowest (30 m) integration zone, annual lakewide total N (16,255 × 106 mol a−1) and nitrate (3,116 × 106 mol a−1) uptake were high compared to these inputs to the lake. Lakewide uptake estimates were calculated by multiplying annual column-integrated uptake rates with the surface area of the lake. Another potential source of N input to the lake, N2-fixation, should be examined in the future. However, based on one previous study [Mague and Burris, 1973] and very low P levels and high concentrations of nitrate in the lake, N2-fixation is not likely to be a large component of the N cycle of Lake Superior.
 Lakewide average nitrate concentrations in the upper 30 m recorded by this study indicate a nitrate drawdown of 1.30 ± 0.88 μM or (3201 ± 2172) × 106 moles (95% confidence intervals) from June to August 2005, approximately 90 d. Our temperature based nitrate uptake estimate (1178 × 106 moles) for the same period and same depth is lower but within the 95% confidence interval. To put a better constraint on nitrate uptake rate using direct concentration change, very high-resolution sampling would be required. Overall, this independent estimate of net column uptake (30 m) over the 90-d period provides further, independent support that nitrate uptake is higher than nitrate inputs to the lake.
 Rates of total N and nitrate uptake that exceed annual inputs provide additional evidence that the lake is turning over water column nitrate and N rapidly relative to the century-long increase of nitrate in the lake waters. The residence time of ammonium and nitrate in the lake based on average concentrations (0.21 and 25 μM respectively) and annual average uptake rate in the top 30 m of the water column based on the temperature-dependent uptake model are 14 d and 18 years, respectively. Residence time is shorter during summer (June–August; 10 d and 11 years, respectively) than during winter (January–March; 24 d and 50 years, respectively). As in many other aquatic systems, ammonium is recycled rapidly. It also appears from these calculations that external nitrate transported to the lake is actively assimilated and recycled relative to the century long build of nitrate in the lake.
 A parsimonious hypothesis for the nitrate rise in Lake Superior is that atmospherically delivered nitrate is incompletely assimilated due to the oligotrophic nature of the lake and builds up in concentration in the lake over many years. Three independent lines of evidence now indicate that this hypothesis is incorrect. First, lake-wide budgeting of loadings and outflows [Sterner et al., 2007] shows insufficient delivery of nitrate to support the nitrate rise plus the outflow to the lower Great Lakes. Second, 18O-NO3− in lake waters are strongly differentiated from rainfall and watershed inputs [Finlay et al., 2007], but are consistent with internal generation of NO3− in the lake. Third, the present study compares for the first time, direct in-lake estimates of N cycling to loading rates. High rates of uptake relative to loading suggest that the nitrate rise is not due to low assimilation of nitrate inputs but rather can only be explained by extensive within-lake biogeochemical oxidation of reduced forms of N. Nitrogen, and in particular nitrate, thus has a shorter turnover time in the water column compared to the traditional hypothesis to explain nitrate accumulation in the lake.
4.6. Rationale for Increasing Nitrate Concentration in Lake Superior
 It is well known that Lake Superior is an oligotrophic lake where primary production is limited by nutrients like P and possibly Fe [Sterner et al., 2004]. One of the most likely consequences of these oligotrophic conditions is low denitrification due to limited labile organic carbon (OC) availability in the sediments. Published reports of OC flux from surface water to the deeper layer in Lake Superior vary from 27.6 mg C m−2 d−1 [Heinen and McManus, 2004] to around 60–90 mg C m−2 d−1 [Baker et al., 1991]. As a result of degradation during settling, the amount of labile organic matter reaching the bottom of the lake is smaller still. Studies from sediment traps and cores suggest that approximately 75% of OC degrades during transportation from the surface to the bottom on a timescale of one year [Baker et al., 1991]. Subsequently, only 5% of settling OC accumulates in bottom sediments [Baker et al., 1991]. This finding is confirmed by another study, which reported OC sedimentation rate in Lake Superior to be very low (0.004 mg C m−2 d−1) and similar to oceans [Johnson et al., 1982]. This very low availability of organic matter does not cause a large oxygen demand at the sediment-water interface. Therefore, the upper layer of Lake Superior sediment (8 to 24 mm) remains oxic [Carlton et al., 1989]. Nitrate concentration in the surficial sediment layer has been reported to be higher than overlying water indicating nitrification [Carlton et al., 1989] and consequent nitrate efflux (average ∼0.15 mmol m−2 d−1) from sediment to the water column [Heinen and McManus, 2004]. In some cores, diffusion of nitrate in anaerobic sediment has also been reported [Carlton et al., 1989]. Maximum observed denitrification rate in Lake Superior sediments is 0.07 mmol m−2 d−1, only half of average nitrate efflux to the water column [Carlton et al., 1989; Heinen and McManus, 2004]. Overall, oxygen supports at least 94% of carbon metabolism in Lake Superior sediment and only 6% is supported by denitrification [Carlton et al., 1989].
 This brief synthesis of existing knowledge of organic matter, oxygen, and nitrate dynamics in the lake (summarized in Figure 10) supports the hypothesis that N removal is strongly limited by the availability of organic matter in Lake Superior. In general, it appears that almost all nitrate and ammonium inputs in the lake are taken up by phytoplankton during primary productivity on an annual scale. A fraction of this primary productivity gets transported to the deeper layers and the rest gets remineralized and eventually nitrified in the surface layer of the water column. Nitrogen associated with settling organic matter may also be remineralized while sinking or at the oxic sediment surface. Thus, most N transported to the sediments also gets nitrified and released to the water column. Therefore, nitrate in Lake Superior does not currently appear to have an efficient sink, except some seasonal uptake by phytoplankton and a likely small removal via denitrification in sediments, and this probably contributes to its build up in the lake on a longer timescale (Figure 10).
4.7. Carbon: Nitrogen Linkages
 Nitrogen uptake dynamics are strongly tied to biological productivity but the coupling of C and N in Lake Superior has not previously been examined. As a first step toward better resolving C uptake in the lake and linking it with this new information on N uptake, we simultaneously examined C and N uptake.
 The reported range of primary productivity for Lake Superior is around 20–220 mg C m−2 d−1 [Vollenweider et al., 1974]. Volumetric productivity integrated numerically from 0 to 100 m (5 m value extrapolated to the surface) at EM was 30.8 mmol C m−2 d−1 (i.e., 369 mg C m−2 d−1), higher than most previous estimates, which have used other methods and emphasized nearshore samplings [Olson and Odlaug, 1966; Putnam and Olson, 1966; Hecky, 2000; Urban et al., 2005]. Primary productivity around 100–250 mgC m−2 d−1 has been reported at a location very close to EM during June 1990 [Fee et al., 1992]. Although limited to only one offshore profile, our results suggest that previous measurements of primary production in Lake Superior, none of which were incubated in situ, may have underestimated this important parameter.
 The higher C:N uptake ratios in the surface layer than the Redfield ratio (6.6) and observed particulate C:N ratios (Table 3) possibly suggest faster turnover rate of C than N in epilimnetic particles. The higher C: N uptake than Redfield ratio is not uncommon in aquatic systems [Watts and Owens, 1999]. However, the concurrent observation of non-correlation of N uptake with chlorophyll and higher C:N uptake than Redfield ratio may be due to constrained range of chlorophyll in the lake or small sample size. It is possible that due to constrained range of chlorophyll, this variable does not provide a good measure of overall activity of the algal biomass. There are lots of mixotrophs in this lake and they are poorly studied with respect to stoichiometry in general. The difference in C:N uptake ratio with contemporaneous particulate C:N also suggests caution while estimating nutrient demand using productivity and seston measurements.
 The present study provides the first measurements of N uptake dynamics and in situ C uptake rates in Lake Superior. Study of in situ carbon uptake indicates higher water column integrated primary production than reported in previous studies and also highlights the top 30 m (generally above the DCM) as the active zone for production and a higher turnover of carbon than nitrogen in epilimnion. However, to characterize the effect of season and spatial variability in C uptake rates, there is a need for more in situ carbon based productivity study in Lake Superior. Experimental studies of nitrogen cycling during the present study highlight the significant role played by smaller size classes of plankton in N uptake and inability of plankton to utilize the available nitrate due to possible limitation of P and Fe. Multiple lines of evidence from this study and others indicate rapid cycling of nitrate relative to the century-long rise of nitrate concentration in the lake. Estimates of N uptake based on temperature-dependent model support the hypothesis of in-lake generation of nitrate, in agreement with N budget and natural abundance stable isotope based studies, as potential source for increasing nitrate concentration in Lake Superior. Similarly, estimates of nitrate uptake based on observed seasonal drawdown of nitrate concentrations indicate even larger total assimilation of nitrate in the water column. Overall, based on earlier studies and the results of this study, it appears that there is no efficient sink of nitrate in the lake, particularly due to lack of significant denitrification as a result of limited supply of labile C to the lake bottom, leading to nitrate build up. We suggest additional research to solidify lakewide uptake estimates, particularly during winter, and measurements of other transformation rates, including nitrification.
 We thank captain and crew of R/V Blue Heron for their excellent support. We also thank Sandy Brovold and Mike Dolan for help during sample analysis. Constructive suggestions about sampling were provided by R. Sherrell, G. Bullerjahn, and R. M. L. McKay. This research was funded by an NSF-Chemical Oceanography grant 0352291.