The Fractionation of Nitrogen and Oxygen Isotopes in Macroalgae during the Assimilation of Nitrate

In order to determine and understand the stable isotope fractionation of 18 O and 15 N manifested during assimilation of NO − 3 in marine macro-benthic algae, two species (Ulva sp. and Agardhiella sp.) have been grown in a wide range of NO − 3 concentrations (2–500 µM). Two types of experiments were performed. The first was one in which the concentration of the NO − 3 was allowed to drift downward as it was assimilated by the algae, between 24 hour replacements of media. These experiments proceeded for periods of between 7 and 10 days. A second set of experiments maintained the NO − 3 concentration at a low steady-state value by means of a syringe pump. The effective fractionation during the assimilation of the NO − 3 was determined by measuring the δ 15 N of both the (i) new algal growth and (ii) residual NO − 3 in the free-drift experiments after 0, 12, 24 and 48 h. Modelling these data show that the fractionation during assimilation is dependent upon the concentration of NO − 3 and is effectively 0 at concentrations of less than ∼ 2 µM. The change in the fractionation with respect to concentration is the greatest at lower concentrations (2–10 µM). The fraction-ation stablizes between 4 and 6 ‰ at concentrations of between 50 and 500 µM. Although the δ 18 O and δ 15 N values of NO − 3 in the residual solution were correlated, the slope of relationship also varied with respect to NO − 3 concentration, with slopes of greater than unity at low concentration. These results suggest shifts in the dominant fractionation mechanism of 15 N and 18 O between concentrations of 1 and 10 µM NO − 3. At higher NO − 3 concentrations (> 10–50 µM), frac-tionation during assimilation will lead to δ 15 N values in algal biomass lower than the ambient NO − 3 and 15 N enrichments in the residual NO − 3 .


Introduction
Nitrogen availability is an important factor in controlling algal growth in marine environments, representing a limiting nutrient throughout much of the global ocean (Dugdale and Wilkerson, 1986). In many studies, information on nitrogen sources and its cycling has been obtained by examining the ratio of the stable isotopes of nitrogen ( 14 N and 15 N) as well as oxygen ( 18 O and 16 O) in the case of NO − 3 . Isotope ratios are expressed using the conventional "delta" notation (δ 15 N or δ 18 O) in parts per thousand (‰) deviation from the atmo- The term epsilon (ε) is also commonly used and is related to α by Eq. (2).
The term ε can refer to fractionation of either 15 N ( 15 ε) or 18 O ( 18 ε) relative to the more abundant isotope of the element. In some of these processes, such as the fixation of atmospheric nitrogen, no significant isotopic fractionation takes place ( 15 ε ∼ 0.0 ‰) (Hoering and Ford, 1960), and consequently the δ 15 N of N 2 fixing organisms is similar to that of atmospheric N 2 (0 ‰ by convention). In other processes, such as the denitrification of NO − 3 , 15 ε values are higher than 20 ‰ (Barford et al., 1999;Delwiche and Steyn, 1970;Granger et al., 2006;Miyake and Wada, 1971), leading to large increases in the δ 15 N of the residual reservoir of NO − 3 . While the δ 15 N of microalgae has been studied in order to understand its use as a paleoceanographic proxy (Altabet, 1989;Altabet et al., 1991;Haug et al., 1998;Sigman et al., 2003), variations in the δ 15 N of macroalgae have also been widely used as possible indicators of anthropogenic influences (Carballeira et al., 2013;Costanzo et al., 2001;Heaton, 1986). Generally speaking, nitrogen derived from sewage is isotopically enriched in 15 N, and it has been argued that even modest enrichments of 15 N in macroalgae might reflect enhanced input from such sources (Lapointe et al., 2004). Other studies have shown that such enrichments could occur through normal processes including fractionation during assimilation (Lamb et al., 2012;Stokes et al., 2011) and that there are not always simple relationships between the input of anthropogenic wastes and δ 15 N values (Viana and Bode, 2013).
Studies of isotope fractionation during the assimilation of dissolved inorganic nitrogen by marine microalgae have reported a wide range of values. In one study, reported 15 ε values ranged from 0.7 to 23 ‰ for the assimilation of NO − 3 by Phaeodactylum tricornutum (Wada and Hattori, 1978), a marine diatom. Another study reported 15 ε values between 2.2 and 6.2 ‰ for 12 different marine phytoplankton cultures kept at a NO − 3 concentration of 100 µM (Needoba et al., 2003). Other research also reports wide ranges in 15 ε values for both NO − 3 and NH + 4 for a variety of different microalgae Lajtha and Michener, 1994;Montoya et al., 1990;Wada and Hattori, 1978). At least part of these large ranges in 15 ε values probably resulted from variations in experimental conditions and are perhaps artefacts resulting from differences in aeration, light and nutrient drawdown. In addition, changing nutrient concentration might be an important controlling parameter, and several studies have shown that microalgae show varying fractionation as a function of concentration (Hoch et al., 1992;Pennock et al., 1996;Waser et al., 1998) that is likely due to changes in physiology and perhaps uptake mechanism.
In contrast to microalgae, there have been relatively few studies of 15 N fractionation in macroalgae. Some of these studies have relied on spiking the natural environment with high nitrate and ammonium concentrations (Teichberg et al., 2007), while others have used transplant experiments (Deutsch and Voss, 2006). Neither of these investigations reported 15 ε values for fractionation during the assimilation of NO − 3 . The study of Cohen and Fong (2005) grew the green algae Enteromorpha intestinalis under varying concentrations of NO − 3 and NH + 4 , and, although they did not report values for 15 N fractionation, they concluded that the δ 15 N of the algae was not dependent upon concentrations of dissolved inorganic nitrogen. These experiments used a combination of increases in NO − 3 and NH + 4 , with the lower NO − 3 concentration experiments containing high amounts of NH + 4 and vice versa. Under such experimental conditions it would have been difficult to isolate any potential concentration dependence upon fractionation manifested during assimilation. Given the possibility of a concentration dependence of 15 N fractionation for NO − 3 in microalgae, we revisit here whether such a dependency is found in macroalgae. We have used two different approaches over a range of different concentrations. In the first series of experiments, two species of macroalgae, Ulva sp. and Agardhiella sp, were grown over a range in nominal NO − 3 concentrations of 10, 50, 100 and 500 µM. As the algae within each culture consumed the NO − 3 in the solution, the solutions were replaced every 24 h. These were the so-called free-drift experiments. In the second set of experiments, NO − 3 levels were maintained at a low level (< 2 µM) by continual addition from a syringe pump. Hence these experiments cover the range of NO − 3 concentrations used in most previous experiments (> 100 µM) as well as those seen under natural conditions.

Methods
Samples of the green algae Ulva sp. and the rhodophyte algae Agardhiella sp. were collected from cultures held at the Aplysia Mariculture Laboratory's algal aquaculture facility (University of Miami). These species were maintained in a system of seven, 9000 L fiberglass tanks supplied with filtered seawater at a rate of ∼ 22 L min −1 . Radiant energy and temperature are monitored constantly, and algal growth rates are optimized by adjusting nutrient levels weekly. These stocks are kept continually as a food source for other organisms in the facility. In preparation for these experiments the algal thalli were rinsed with filtered seawater and gently scrubbed to remove surface epiphytes. Prior to experimentation, the macroalgae were maintained within 2 L flasks at 26 • C and approximately 100 µmol photons m −2 s −1 for a 14-day acclimation period. During the acclimation period, filtered and autoclaved seawater was changed every 2 days, enriched to 500 µM N (250 µM NaNO 3 and 250 µM NH 4 Cl) and 44 µM KH 2 PO 4 , with f/2 medium supplements of B   (Guillard, 1975). The cultures were continually aerated throughout the incubations.

Free-drift experiments
In these experiments the effect of varied nutrient availability on the nitrogen isotopic composition of new algal growth with respect to varied NO − 3 concentration was investigated. Nominal concentrations of 10, 50, 100 and 500 µM N (NaNO 3 ) were supplied in a medium of autoclaved, filtered (0.2 µm cartridge filter) seawater enriched with the same KH 2 PO 4 , B vitamin and trace metal supplements outlined for the acclimation medium (note that the actual targeted and measured concentrations were slightly different and the values used are reported in Tables 1 and 2). Subsamples of Ulva and Agardhiella (0.25-0.5 g wet weight; 2.5-3.0 cm) were taken from acclimation flasks, any visible epiphytes were again removed and the algae samples were placed in 2 L flasks filled with incubation medium. The media were replaced every 24 h, at which time each algal sample was rinsed to prevent epiphyte accumulation. The experiments proceeded for a period of 7-9 days. Water samples were collected after each 24 h period and analysed for the concentrations of NO − 3 and NH + 4 . At the conclusion of the incubations, final accumulated biomass was weighed and, as the new algal growth produced was clearly visible, material which had grown only under the experimental conditions was trimmed off (Fig. 1). This material was dried (40 • C, 48 h) and then ground with mortar and pestle for subsequent N isotopic analyses and C / N determination. In order to examine the effect of assimilation on the δ 15 N of residual NO − 3 , special experiments were performed in which the same water was kept in the algal cultures for periods of up to 48 h. After All experiments in which NO − 3 was added showed approximately similar growth rates, but reduced uptake of N at lower N concentrations. At the end of the experiment the ends of the algae were trimmed and analysed for their δ 15 N, δ 13 C and C / N ratio. The new growth could be distinguished by comparison with the size of the original fragment (see Fig. 1a) and the change in colour.
12, 24 and 48 h, water samples were taken and the δ 18 O and δ 15 N of the NO − 3 measured.

Constant NO − 3 concentration experiments
At low concentrations of NO − 3 (< 10 µM), the algae rapidly assimilated NO − 3 and concentrations decreased to values of less than 3 µM within a few hours. In order to maintain a consistent low concentration and provide sufficient NO − 3 for the algal growth, NO − 3 was continuously added by means of a syringe pump. The rate of addition was initially determined by using the uptake rates calculated from the freedrift experiments and then adjusted slightly after the analysis of the NO − 3 concentration in the experiment. In these experiments concentrations started at ∼ 10 µM and stabilized at 3 µM throughout the growth period.

Algal biomass
The organic carbon and nitrogen content as well as the stable nitrogen (δ 15 N) and carbon (δ 13 C) isotopic composition of the algae was determined using a CN analyser (ANCA, Europa Scientific) interfaced with a continuous-flow isotoperatio mass spectrometer (CF-IRMS) (20-20, Europa Scientific). Prior to analysis the algae samples were dried and 3-6 mg were placed in tin capsules. Data obtained from the mass spectrometer provide the C / N ratio of the samples in addition to the isotopic content of the organic matter. Samples of the nutrient salts added were analysed in a similar manner to determine the initial δ 15 N of the medium. The δ 15 N and δ 18 O of the initial NO − 3 were also analysed as dissolved inorganic nitrogen (see below). Internal laboratory standards, calibrated to VPDB and atmospheric N 2 , were analysed every 10 samples, and data were corrected relative to the mean of the two nearest standards. External precision is approximately ±0.2 ‰ for δ 15 N and ±0.1 ‰ for δ 13 C. The C / N ratio was calculated by comparing the integrated area of the major beams (mass 28 for N and mass 44 for C) to standards with known C / N ratios. The external precision for this method is < 0.1 %.

Dissolved inorganic nitrogen
The δ 15 N and δ 18 O composition of the samples were determined using a GV IsoPrime with an external automated purge-and-trap system at the University of Massachusetts, Dartmouth, SMAST campus. The NO − 3 was converted to N 2 O using Cd reduction to NO − 2 followed by azide treatment (McIlvin and Altabet, 2005). Data are reported relative to atmospheric N 2 and VSMOW for nitrogen and oxygen, respectively. Each run of NO − 3 samples consisted of one operational blank (low-nutrient seawater treated with azide), three NO − 2 standards, a cadmium blank (low-nutrient seawater treated with cadmium) and three NO − 3 standards (USGS 34, 35 and an internal Altabet lab standard), followed by the prepared samples. Three randomly selected samples were also prepared in triplicate to check for method and machine reproducibility. The run ended with three more NO − 2 standards, three NO − 3 standards, a Cd blank and an operational blank. Analytical precision measured from multiple determinations on standards was approximately ±0.2 ‰ for δ 15 N and ±0.7 ‰ for δ 18 O (NO − 3 only). Isotopic data produced from each run were scrutinized for standard precision throughout individual runs. Samples were corrected for the small amount (∼ 15 %) of oxygen exchange that occurs between the sample and water during the conversion to nitrous oxide, fractionation due to oxygen removal and the 1 : 1 addition of azide N to NO − 2 -N in the formation of N 2 O (see McIlvin and Altabet, 2005, for an in-depth discussion of δ 15 N and δ 18 O corrections).

.2 Nutrient concentrations
Concentrations of NO − 3 , NO − 2 and NH + 4 in the growth solutions were analysed prior to, during and after each experiment. Nitrate and nitrite concentrations were determined by diazotization before and after reduction with cadmium (Grasshoff, 1976). Ammonium concentrations were determined with the indophenol-blue method. Note that the measured concentrations of the NO − 3 were slightly different than initial target concentrations.

Free-drift experiments
Results from the free-drift nutrient experiments from Ulva and Agardhiella are presented in Table 1. In each of the treatments the δ 15 N of the new algal growth during each experiment and the residual NO − 3 concentrations left in each treatment after the 24 h incubations were determined. Although concentrations of NO − 2 and NH + 4 were measured, none was detected. The δ 15 N of the newly grown Agardhiella material decreased from 1.8 ‰ (14 µM) to 1.6 ‰ in the 50 µM treatment, to 0.7 ‰ in the 103 µM treatment and finally to −3.0 ‰ in the 485 µM experiment. Similar results were found in the experiments using Ulva although the δ 15 N values were all higher (Table 1). For example, in the lowest two NO − 3 treatments, the δ 15 N of the Ulva was actually more positive than that of the NO − 3 in the growth medium (Table 1). The C / N ratios and the δ 13 C values of the algae are included in the supplementary information.

Syringe experiments
The results from all the syringe experiments are listed in Table 2. The δ 15 N value of Ulva and Agardhiella exhibited small decreases.

Free-drift experiments
The data from the free-drift experiments are presented in Table 1

Discussion
In order to calculate the fractionation during assimilation, the change in the δ 15 N and δ 18 O of the NO − 3 and the algal tissue was modelled using a Rayleigh distillation model. In the case of N, the 15 N / 14 N of the new algal growth (RA) at time (t) is given by Eq. (3), while the 15 N / 14 N of the residual NO − 3 (R) at t is given by Eq. (4).
In these equations, f represents the fraction of the initial NO − 3 remaining; Ri the 15 N / 14 N ratio of the initial NO − 3 ; Rt and RAt the 15 N / 14 N ratio of the NO − 3 and new algal growth, respectively, after a specific time during which f has been determined; and α the fractionation factor. The fractionation ( 15 ε) can also be calculated using the approach of Mariotti et al. (1981) which utilizes a plot of the isotopic composition of the NO − 3 with respect to ln f or ln (NO − 3 (t) / NO − 3 (i)) as in Eq. (5).
In Eq. (5), ε ( 15 ε) is the slope of the relationship between δ 15 Nt and ln f. The term δt is equal to the δ 15 N of the NO − at time (t) when the concentration is equal to NO − 3 (t), and δi is equal to δ 15 N of the NO − 3 at the initial time when the concentration is equal to NO − 3 (i). In the free-drift experiments where the δ 15 N of the solution was sampled multiple times, the δ 15 N values were measured at various concentrations as the NO − 3 was assimilated by the algae and hence values of ln f calculated. A similar approach was used to calculate 18 ε using the δ 18 O data.
An alternative method for calculating the fractionation factor used the measurement of the δ 15 N of new algal tissue as a function of the expression in Eq. (6).
As f tends to 0 (all the NO − 3 was consumed), the δ 15 N of the algae (δA) tended to approach the δ 15 N of the initial NO − 3 . Hence, utilizing Eq. (6), the slope of the relationship was equivalent to (α − 1) × 1000 or ε.
In each of the experiments the slope of the line was determined by plotting the initial δ 15 N of the NO − 3 at a f value of 0 and the measured δ 15 N of the algae at the appropriate f value corresponding to the decrease in the concentration of NO − 3 at the end of 24 h. While it is possible to arrive at an estimate of fractionation using either the solid sample or the NO − 3 data, the method of measuring the δ 15 N of the residual NO − 3 may provide a more accurate method for a number of reasons. First, in the case of the measurement of the δ 15 N of the tissue in the free-drift experiments, the f factor is calculated by averaging the amount of NO − 3 utilized during a 24 h period. This assumes that the algae grows equally throughout the 24 h period, rather than perhaps faster when the NO − 3 concentration is high and lower as the concentration is reduced. In addition, as the concentration of NO − 3 is reduced to low concentrations, the fractiona-tion of 15 N will also change (see later discussion). It might be possible to model these changes, but the interpretation would be dependent upon a number of assumptions which could not be validated with the present data set. In this regard the δ 15 N of the tissue grown in the syringe experiments might be more reliable in providing an estimate of fractionation as a constant amount of NO − 3 is supplied throughout the growth period, and therefore Rayleigh type modelling is unnecessary. In addition, both the syringe and the free-drift tissue measurements might suffer from the inability to precisely separate new and old algal tissue growth and the possibility of translocation of N-bearing compounds in the algal tissue. In contrast, the δ 15 N of the residual NO − 3 provides a direct measurement of the fractionation during assimilation. While the results obtained between the two methods are similar, in cases where there are differences we feel that the data obtained from the δ 15 N of the NO − 3 provide the best estimate of fractionation.

Modelling
As the NO − 3 removed from the medium was balanced by algal assimilation, isotopic fractionation produced corresponding changes in both the δ 15 N of the residual NO − 3 and the δ 15 N of new algal growth. These data are reported in Table 1, and the fractionation factors estimated for 15 N (and 18 O when applicable) using Eqs. (5) and (6) are reported in Table 3.

Ulva
The 15 ε values calculated from the δ 15 N of the algal growth and the NO − 3 show a decrease towards 0 with decreasing concentration of NO − 3 (Table 3, Fig. 3). At the higher concentrations, the estimate of 15 ε obtained from the algal growth (∼ 3 ‰) and that obtained from residual NO − 3 δ 15 N are statistically the same, while at the lower initial NO − 3 concentrations values of 15 ε obtained from the algal δ 15 N are significantly lower (Fig. 4). If the observation that fractionation varies as a function of the concentration of NO − 3 is correct, then Eq. (3) can only yield a mean estimate of 15 ε as during the experiment the concentration of NO − 3 changes considerably as it is assimilated. In fact the data from the NO − 3 free-drift experiment (Table 1) is best fitted by a quadratic equation confirming a change in fractionation with changing concentration (Fig. 5). Using a Chi-squared test, the improvement in the fit between the linear and non-linear model can be shown to be statistically significant at the 99 % level in both the 60 and 103 µM experiments. The first differential of the quadratic equation therefore provides an estimate of ε at any value of f . Using the data from the experiments which were initiated at concentrations of 14, 60 and 103 µM NO −  Table 1).
( Fig. 6). Data from the 500 µM experiment were not used in this estimate as a result of the small change in concentration of NO − 3 (and as a consequence a small change in f which occurred during the experiment at high concentration). Although the estimates of 15 ε obtained from the nonlinear equation predict a value of less than 0 at concentrations lower than ∼ 1 µM, none of the experiments attained these low concentrations and therefore this observation will need to be confirmed. In addition the one syringe experiment performed with Ulva at a constant concentration of ∼ 3 µM yielded a 15 ε value of 1 ‰, higher than the values estimated from the NO − 3 drawdown experiments. Hence such data were inconsistent with a 15 ε value below 0. In the free-drift experiments however, the δ 15 N of the measured algae was greater than that in the initial NO − 3 (5.1 and 4.0 ‰ in the 14 µM and 60 µM treatments, respectively, compared to the initial NO − 3 and algal values of 3.3 and 3.1 ‰, respectively) (Table 1), giving estimates of 15 ε less than 0 (ε = −3.2 ‰). While the data of the solids appear inconsistent with the data measured on the δ 15 N of the NO − 3 , based on previous discussion our feeling is that assimilation factors calculated from the analysis of the δ 15 N of the NO − 3 are correct and that the δ 15 N of the solid material might therefore be some kind of artefact as discussed earlier. Regardless of whether 15 ε is less than 0, both approaches show a decrease in 15 ε with decreasing concentrations, and a rate of change of appears to be greatest at the lowest concentration, i.e. between 1 and 10 µM (Fig. 6). With higher concentrations (between 10 and 50 to 500 µM) the fractionation appears to reach a constant positive value (ε = 3 to 4 ‰).

Agardhiella
Based on both the algal and NO − 3 δ 15 N data, this species also exhibited a strong dependence between fractionation and NO − 3 concentration. Values of 15 ε were close to 0, or slightly negative, at low concentrations (< 10 µM) and increased between 100 and 500 µM, reaching a value of ∼ 8 ‰ at 500 µM (Fig. 7). As a result of the fact that at most only three samples were taken for measurement of the δ 15 N (and δ 18 O) of the NO − 3 during the free-drift experiments, it was not considered valuable to fit anything more than a straight line to the data, and therefore a more refined equation relating the change in ε to the concentration of NO − 3 was not calculated. As in the case of Ulva, there was a suggestion that 15 ε values might fall below 0 at low concentrations, although the δ 15 N of the solid material did not increase at low NO − 3 concentrations as seen in Ulva sp.

Concentration dependence of the fractionation factor
In microalgae and bacteria the uptake and fractionation of NO − 3 has been proposed to be a three-step process (Granger et al., 2004;Hoch et al., 1992;Karsh et al., 2012Karsh et al., , 2014Mariotti et al., 1982;Shearer et al., 1991): first a transport step across the cellular membrane (ε in ), then a nitrate reductase step (ε NR ) and finally a flux out of the cell (ε out ). The overall fractionation manifested by the organism, expressed as ε org , is related to the influx, efflux and nitrate reductase Figure 5. The changing fractionation as a consequence of decreasing NO − 3 concentration, as shown in Fig. 3, necessitates the use of a non-linear curve fitting to the data. The use of a quadratic equation shows an improved fit to the data and allows the slope of the relationship to be calculated at specific concentrations using the first differential of the equation. Data from the 485 µM experiment have been omitted as a result of the small change in the f value. fractionation by Eq. (8), in which γ is the proportion of efflux relative to influx (Karsh et al., 2014).
The estimated fractionation associated with these processes in a marine diatom (Thalassiosira weissflogii) are 15 ε in = 2 ‰, 15 ε out = 1.2 ‰ and 15 ε NR = 26.6 ‰ (Karsh et al., 2012(Karsh et al., , 2014. As the majority of the fractionation is associated with the NR step, the degree to which this is expressed in the external medium and also in the organism is controlled by the amount of efflux relative to influx (γ ). Accepting the possibility that there may be differences between microalgae and the organisms used in this study, we have nevertheless used this model as a basis with which to explain the observations of a concentration dependence on 15 ε org made in this paper. In this regard it is helpful to examine the work of Needoba et al (2004), who measured the δ 15 N of the internal and external NO − 3 pools. They determined that the maximum difference in δ 15 N occurred in situations in which ε org was at a minimum, thus indicating that the efflux from the cell was small. Conversely when fractionation was high, the difference between the δ 15 N of the external and internal pools was at a minimum and efflux maximal. As in both cases, the greatest potential for isotope fractionation is at the NR step (Karsh et al., 2012;Ledgard et al., 1985), the principal explanation for dependence on external concentration must relate to the ratio of NO − 3 uptake to efflux from the cell. At lower external concentrations, NO − 3 is limiting and the δ 15 N of the internal pool is highly elevated. However, most of the NO − 3 is consumed and efflux is minimal, and, although the same amount of fractionation at the NR step takes place, this isotopic signal is not communicated to the external environment. At high concentrations the reverse is true: NO − 3 is not limiting, and the fractionation experienced at the NR step is translated to the external environment. Based on our findings we propose that macroalgae may behave similarly in many respects to microalgae. However, the only study we are aware of dealing with macroalgae concluded that the concentration of NO − 3 did not influence the fractionation of 15 N (Cohen and Fong, 2005) and would therefore appear to be in conflict with the results of this study. However, in the Cohen and Fong (2005) research the only experiments in which the concentration of NH + 4 was not altered, in addition to NO − 3 , were carried out at relatively high concentrations of NO − 3 (> 50 uM). This is above the level at which the fractionation appeared to be constant in our study.

3
The measurement of the δ 18 O of nitrate is a relatively new technique which has helped explain both the source of NO − 3 and the mechanism of fractionation of N and O isotopes during assimilation (Granger et al., 2004(Granger et al., , 2010Leichter et al., 2007;Wankel et al., 2006Wankel et al., , 2009. The data presented here suggest that in a manner similar to 15 N the fractionation of 18 O is dependent upon the concentration of NO − 3 in the external environment (Tables 1 and 2; Fig. 8). While generally the fractionation of δ 18 O and δ 15 N are related in a 1 : 1 ratio (Granger et al., 2004), in this study the slope of the data seems to have a value of greater than the ideal 1 : 1 relationship (Fig. 2). It was argued by Granger et al. (2004) that this 1 : 1 relationship was consistent with fractionation of N and O during NR, whereas fractionation during uptake would give a 2 : 1 relationship. In more recent work it was shown that there are different degrees of fractionation for N compared to O during uptake and efflux, which would cause the relationship between 18 ε and 15 ε to rise significantly above unity, when fractionation is low (Karsh et al., 2014). Such data are in agreement with our study in that the 18 ε : 15 ε ratio is closest to unity in the highest-concentration (∼ 500 µM) experiments and increases with lower initial concentrations of NO − 3 , reaching a value of ∼ 2 at 10 µM. This 2 : 1 relationship corresponds to the lowest amount of fractionation observed (ε ∼ 0 ‰). Using the rationale suggested by Granger et al. (2004), this pattern is consistent with a change in fractionation from a process predominantly controlled by NR, to one in which fractionation is controlled by the relative difference between the fractionation of O and N during up- take (1.4) and efflux (2.3) (Karsh et al., 2014). If the results of these experiments are correct, then the relationship between δ 18 O and δ 15 N also should not be linear, but rather a quadratic, similar to that observed between the δ 15 N and concentration discussed earlier. However, as a result of the larger error on the δ 18 O compared to δ 15 N (0.7 vs. 0.2 ‰), this pattern was not evident in the data collected in these experiments.

Biogeochemical implications
The observation of a concentration dependence upon 15 N fractionation during denitrification has been previously made for microbes . Both the results of that study and the data presented here suggest that there is a relationship between fractionation and concentration during assimilation, which has implications for the application of nitrogen isotopes for detection of N sources. It is clear that under typical N-limiting conditions both micro-and macroalgae have the same isotopic composition as the ambient nitrate. However, when NO − 3 concentrations are elevated, algae fractionate the external NO − 3 pool, forming biomass which is relatively isotopically more negative than the ambient NO − 3 . The residual NO − 3 effluxed from the cell consequently becomes isotopically more positive regardless of the δ 15 N of the original NO − 3 . Consider a hypothetical coastal estuary in which there is a significant input of NO − 3 from artificial fertilizers (with a δ 15 N ∼ 0 ‰) applied to adjacent agricultural areas. As a result of the high NO − 3 concentrations, the fractionation during assimilation by algae would be