This study focused on the 70 km channelized reach of the Sacramento River extending from the city of Sacramento downstream to Isleton (Fig. 1), where the river enters the more hydrodynamically complicated network of open water, channels, and sloughs called the Cache Slough Complex. The only significant inflow within the study reach is just below the Freeport Bridge where treated effluent from the Sacramento Regional WWTP enters the river. River flows are monitored at two USGS stations located at Freeport Bridge and Walnut Grove (http://waterdata.usgs.gov/usa/nwis).

Field sampling was conducted as part of a larger experiment designed to examine changes in phytoplankton abundance and community composition in the presence and absence of wastewater in the Sacramento River. For details of the field methods employed see Kraus et al. (2017a). Briefly, field sampling was conducted using a Lagrangian sampling approach during 24 to 29 October 2013, and 30 May to 4 June 2014 (hereafter referred to as the “October” and “June” experiments). During both October and June, sampling was coordinated with ∼ 20 h WWTP effluent discharge holds, creating a ∼ 15 km stretch of effluent-free river to allow comparison of two parcels of river water; one containing effluent high in NH4+ (+EFF) and one without effluent (-EFF). During both experiments, +EFF and -EFF parcels were tracked using small drifters and a high-speed mapping boat equipped with a custom-designed flow-through instrument package that continuously displayed surface–water measurements of specific conductance (a conservative tracer), to assure that samples were collected from within the designed parcel of water. On both dates +EFF and -EFF parcels were tracked and sampled over ∼ 80 h (3.5 days) as they traveled ∼ 70 km downstream (Fig. 1). Water samples were collected from both parcels each day at approximately 2 to 3 h intervals between 08:00 and 17:00 PST. Discrete water samples were collected from 1 m depth using a 3k Shurflo pump with clear 1/2 in. tubing using USGS protocols (USGS, 2006). Samples were pumped into 8 L Teflon Jerri cans and then transferred into a 20 L churn (USGS, 2006) for subsampling. Subsamples were collected for nutrients, chlorophyll a (Chl a), plankton identification and enumeration, flow cytometry, and stable isotope analysis.

Nutrient and Chl a concentration analyses were performed at the San Francisco State University Romburg Tiburon Center as described in Parker et al. (2012a). Water samples for nutrient analysis were immediately filtered through Whatman GF/F filters using a 50 mL syringe into either 20 mL HDPE scintillation vials or 50 mL centrifuge tubes, placed on dry ice, and then stored at -20 ∘C until analysis. Concentrations of NO2- and NO3- plus NO2- were analyzed independently on a Bran and Luebbe AutoAnalyzer II. Samples for NH4+ determination were collected separately into 50 mL centrifuge tubes after similar filtration. These samples were also immediately frozen for later analysis by colorimetry using a Hewlett Packard diode array spectrophotometer with a 10 cm path cell length.

Chl a samples were concentrated onto 25 mm, 0.70 µm, Whatman GF/F filters using a low vacuum (< 250 mm Hg). Filters were stored dry at 4 ∘C for up to 1 week. Prior to analysis, Chl a was extracted from the filters in 90 % acetone for 24 h at 4 ∘C. Analysis was performed fluorometrically with a Turner Designs model 10AU using 10 % hydrochloric acid to correct for and measure phaeophytin. The fluorometer was calibrated with commercially available Chl a (Turner Designs chlorophyll a standard).

Stable isotope analysis of POM, NO3-, and NH4+ were performed at the US Geological Survey's Menlo Park Stable Isotope Laboratory. Nitrogen has two stable isotopes, 14N and 15N, and the relative abundance of 14N and 15N is expressed as δ15N (‰) = [(Rsample/Rstandard)-1] × 1000, where Rsample is the ratio of 15N to 14N in a sample, and Rstandard is the ratio of the isotopes in AIR, the recognized reference material for δ15N values. Replicate samples for isotopic analysis of POM were filtered through pre-combusted GF/F filters, and the filters were frozen at -4 ∘C until analysis. Filters were freeze-dried, ground, and vapor-acidified to remove any carbonate prior to analysis for δ15N and C : N atomic ratio using a Carlo Erba NA 1500 elemental analyzer connected to a Micromass Optima mass spectrometer. Analytical precision for duplicate analyses of the same POM sample was < 0.5 ‰ for δ15N. Samples for NO3- isotopes (δ15N-NO3-) were filtered through 0.45 µm nucleopore filters, and the filtrate was kept frozen until analysis using a minor modification of the Sigman et al. (2001) and Casciotti et al. (2002) microbial denitrifier method using a modified Gilson autosampler connected to an IsoPrime mass spectrometer. Analytical precision for sample replicates was 0.3 ‰ for δ15N-NO3-. Samples for δ15N analysis of NH4+ (δ15N-NH4+) were prepared using a slightly modified version of the method of Holmes et al. (1998) and analyzed on a Carlo Erba NA 1500 elemental analyzer connected to a Micromass Optima mass spectrometer (Kendall et al., 2001). Analysis of δ15N-NH4+ was only possible on samples with NH4+ concentrations greater than 15 µM. Precision for this method based on replicate analyses of samples in this study was < 0.4 ‰.

At a subset of stations, samples were collected (n=27) for flow cytometric separation of phytoplankton from bulk POM for N-isotopic analysis using the method described in detail in Fawcett et al. (2011). For each sample, 1 L of river water was pre-concentrated onto four 47 mm 0.2 µm polycarbonate filters using a gentle vacuum (less than < 250 mm Hg). Filters were transferred to 15 mL Falcon tubes containing 10 mL of river water and 500 µL of 10 % paraformaldehyde (PFA) for a final concentration of 0.5 % PFA. During preliminary investigations of Sacramento River water samples, we observed substantial loss of intact phytoplankton cells when freezing and thawing the pre-concentrated samples, and a degradation of Chl a fluorescence after > 9 days of storage. Because both of these processes reduce the number of detectable phytoplankton cells in a sample over time and potentially impact δ15N values of the sorted populations, all of the samples included in this study were sorted from unfrozen samples (stored in the dark at 4 ∘C) within 1 week of collection.

Phytoplankton were sorted with an Influx Cell Sorter in logarithmic mode (BD Biosciences, San Jose, CA, USA). Prior to sorting, sample concentrates were pre-filtered using 50 µm mesh size filters to prevent clogging of the 70 µm diameter flow cytometer nozzle with large particles. Phytoplankton were detected and sorted into a gate using forward scatter (proxy for cell size) and chlorophyll fluorescence at 692 nm. Chlorophyll autofluorescence was excited using a 200 mW, 488 nm sapphire laser (Coherent, Santa Clara, CA, USA). To ensure a sufficient mass of nitrogen, approximately 10 million cells were sorted directly into 5 mL polystyrene Falcon tubes orientated at a low angle to the sort stream. Regular analysis of sort purity (calculated by analyzing a sorted sample to determine the proportion of events that fall within the target gate as a percentage of total event rate) was approximately 95 %. Because the river water contained abundant detritus and sediment, long sort times were required to achieve high sample purity. Unfortunately, the need to sort unfrozen samples within 1 week of collection precluded sorting sufficient cells for N isotopic analysis of individual populations such as diatoms or cyanobacteria. Instead, all phytoplankton cells were sorted into a single sample from each location.

Sorted cells were transferred to 20 mL glass vials and dried down under vacuum using a centrifugal evaporator. Dried phytoplankton samples were redissolved in 20 µL ultra-high-purity deionized water and transferred into tin capsules. Capsules were dried overnight at 60 ∘C and then crushed into small cubes. δ15N analysis of sorted phytoplankton was conducted using elements of a coupled Carlo Erba CHNS-O EA1108 elemental analyzer and Thermo Finnigan GasBench II system with automated cryo-trapping system that is connected to an isotope ratio mass spectrometer (Thermo Fisher Scientific) at the UC Santa Cruz Stable Isotope Laboratory facility. The elemental analyzer and GasBench were configured to run small samples using the methods described in Polissar et al. (2009). In this configuration, samples as small as 35 nmol N could be analyzed with a precision of 0.5 ‰. Phytoplankton samples analyzed for this study (δ15N-PHY) ranged in size between 50 and 100 nmol N. Analysis of duplicate samples (sorted and analyzed independently) indicated a precision of 0.8 ‰ for the entire method.

A two-end-member stable isotope mixing model approach was used to quantify changes in the N source used by phytoplankton encountering elevated NH4+ concentrations downstream of the WWTP. Assuming that NO3- and NH4+ were the only available N sources, the percentage of N uptake from NH4+ (%NH4) was calculated according to Eq. (1), where δ15NNO3, δ15NNH4, and δ15NPHY are the N isotopic ratios for NO3-, NH4+, and phytoplankton, and εNO3 and εNH4 are the enrichment factors for NO3- and NH4+, respectively (York et al., 2007). %NH4=[(δ15NNO3-εNO3)-δ15NPHY][(δ15NNO3-εNO3)-(δ15NNH4-εNH4)]×100 The enrichment factor (ε) is an expression of the magnitude of fractionation between the substrate (e.g., NH4+ or NO3-) and product (e.g., phytoplankton). Phytoplankton are able to process 14N faster than 15N (due to a difference in energy required to break bonds) and this results in lower δ15N values in phytoplankton cells compared to their nutrient source as long as the nutrient source is not completely exhausted. In an open system when N substrates are not used to exhaustion, ε can be approximated as the instantaneous difference between δ15Nsubstrate and δ15Nproduct.

Nitrate concentrations increased downstream during the October and June campaigns in both +EFF and -EFF parcels; however, the downstream gains in NO3- were more modest in -EFF parcels (Fig. 2). During both the October and June sampling campaigns, NH4+ concentrations were low (< 1.0 µM) upstream of the WWTP. Immediately downstream of the WWTP NH4+ concentrations increased in the +EFF parcel to ∼ 100 µM in October and ∼ 60 µM in June. The maximum NH4+ concentration in October was higher than in June due to a greater percentage of effluent in the river (4.0 % in October compared to 2.7 % in June; Kraus et al., 2017a). During the 3 days of downriver travel following effluent addition, riverine NH4+ concentrations decreased modestly in the +EFF parcels. In contrast, NH4+ concentration remained less than 20 µM downstream of the WWTP in the -EFF parcels, and a slight increase in concentration was observed during downstream transit. In June, phytoplankton growth rates may have been N limited at the most upstream locations, as the concentration of DIN was close to the half-saturation constant of 7 µM frequently used to model phytoplankton growth (Travis et al., 2015).

Table 1 presents δ15N values of NO3- and NH4+ for all samples collected during the October and June campaigns with concentrations sufficient for analysis. During the October field campaign, δ15N-NO3- started at 8.5 ‰ upstream of the WWTP and decreased downstream in both the +EFF and -EFF parcels. The magnitude of the decrease was greatest in the parcel containing wastewater effluent; over 83 h of travel downstream of the WWTP δ15N-NO3- decreased by 7.5 ‰ in the +EFF parcel, whereas δ15N-NO3- only decreased by 1.6 ‰ in the -EFF parcel. In June, upstream values for δ15N-NO3- were lower than in October (3–5 ‰), and remained relatively stable as both the +EFF and -EFF parcels traveled downstream (Fig. 5).

Due to low concentrations of NH4+ upstream of the WWTP, it was only possible to measure δ15N-NH4+ in the +EFF parcels downstream of the WWTP. In the +EFF parcels δ15N-NH4+ increased from 7.9 to 9.7 ‰ in October and from 8.0 ‰ to 10.7 ‰ in June with downstream travel. We also observed that δ15N-NH4+ increased while NO3- concentration increased and δ15N-NO3- values decreased during transit in parcels containing effluent, which suggests nitrification was occurring. This observation is consistent with high rates of nitrification reported in the Sacramento River (Hager and Schemel, 1992; Parker et al., 2012a; O'Donnell 2014; Damashek et al., 2016; Kraus et al., 2017b).

δ15N-POM values decreased over the study reach in all parcels, though downstream trends were not monotonic: in particular, we observed periods during the night when δ15N-POM increased > 2 ‰ (Fig. 3). The addition of effluent caused a larger decrease in δ15N-POM values during transit compared to the -EFF parcels in October and June. C : N atomic ratios in all POM samples were near the Redfield ratio, ranging from 6.8 to 8.9, suggesting the POM was primarily phytoplankton (Tables 1 and 2). C : Chl a (weight : weight) ratios of POM ranged from 10 to 170 for all samples, with a median value of 35, which is also consistent with phytoplankton-dominated POM during October and June (Table 1).

Similar to δ15N-POM, δ15N-PHY values decreased with downstream travel in all parcels, but the magnitude of decrease was much greater in the +EFF parcels despite the fact that effluent contained NH4+ with a higher δ15N value than NO3- present upstream of the WWTP (Fig. 3). For many samples, the difference between δ15N-POM and δ15N-PHY was less than 1 ‰ (Fig. 4). However, δ15N-POM values diverged from δ15N-PHY after more than 20 h of travel past the WWTP in the +EFF parcels. The largest difference between δ15N-POM and δ15N-PHY was ∼ 7 ‰ in the October +EFF parcel at the most downstream site (Figs. 3, 4).

C : N and C : Chl a ratios suggest that POM in the Sacramento River collected in October and June was primarily phytoplankton. Previous studies in the SFE have also reported POM C : N ratios near the Redfield ratio (Canuel et al., 1995; Cloern et al., 2002), and in a survey of POM across the SFE, Wienke and Cloern (1987) reported a median C : Chl a ratio of 50 over a range of phytoplankton community composition and productivity. The median C : Chl a ratio of 35 observed in this study is also in agreement with previous studies which have used a C : Chl a ratio of 35 as a conservative estimate of the phytoplankton C : Chl a ratio in the SFE (Cloern et al., 1995; Canuel et al., 1995; Sobczak et al., 2005). Consistent with the interpretation that POM contained primarily phytoplankton, we found general agreement between δ15N-POM and δ15N-PHY, particularly in samples that did not contain effluent. However, we also observed a divergence between in δ15N-PHY and δ15N-POM values within 24 h following the addition of effluent containing high concentrations of NH4+. The slower response to a change in N sources observed in bulk POM compared to intact phytoplankton isolated by flow cytometry suggests that the bulk POM pool contained a significant fraction of dead or inactive phytoplankton not actively taking up nitrogen.

During the October and June experiments, phytoplankton samples were collected for quantitative enumeration and qualitative evaluation (for details see Kraus et al., 2017a). Patterns in the phytoplankton assemblages were examined and potential differences between the +EFF and -EFF parcels were tested for significance using analysis of similarity (ANOSIM). The results of this analysis showed no statistical difference between the assemblages present in +EFF and –EFF parcels in either October or June (Kraus et al., 2017a). During both experiments it was observed that diatoms accounted for ∼ 90 % of the algal biovolume, and that upstream of the WWTP colonies appeared more vibrant compared to downstream samples which contained abundant decrepit cells and partially empty frustules (Kraus et al., 2017a). A downstream decline in cell health, which mirrored decreasing Chl a concentration, was observed in both +EFF and -EFF parcels. The observation of declining cell health in phytoplankton may help explain how bulk POM could contain primarily phytoplankton cells and yet also display a different downstream trend in δ15N values when compared to sorted phytoplankton. Because phytoplankton cells were sorted from bulk POM based on a ratio of cell size and Chl a fluorescence associated with healthy and intact cells, an increasing abundance of decrepit cells would not influence δ15N-PHY, but it could dilute the contribution of live phytoplankton to the δ15N value of bulk POM.

It is also possible that the presence of effluent caused an increase in heterotrophy by bacteria and zooplankton in the +EFF parcels, resulting in δ15N-POM values greater than δ15N-PHY downstream of the WWTP. Heterotrophic bacterial abundance was not measured during this study, but dissolved oxygen concentration was monitored in the both parcels, and in both October and June experiments dissolved oxygen concentrations were lower in the parcels containing effluent (Kraus et al., 2017a). During the June experiment, zooplankton biomass was measured in the -EFF and +EFF parcels, and a decrease in zooplankton biomass was observed in the +EFF parcel downstream of the WWTP (Kraus et al., 2017a). Thus, while we cannot rule out the possibility that zooplankton growth elevated δ15N-POM downstream of the WWTP in October, it is unlikely that zooplankton caused the divergence between δ15N-POM and δ15N-PHY in the +EFF parcel in June. Temporal variability in δ15N-POM also provides some indirect evidence of bacterial reworking of bulk POM during downstream transport. On multiple occasions in this study we observed that δ15N-POM increased overnight in both +EFF and -EFF parcels. Isotopic fractionation during remineralization by bacteria has been shown to produce NH4+ ∼ 3 ‰ lower than its organic matter source (Hoch et al., 1994), thus increasing the δ15N value of the remaining organic matter. If N remineralization exceeded uptake under low light conditions, it could explain observed increases in δ15N-POM at night. Remineralization of labile POM during downstream transport would also dilute the isotopic signal of NH4+ uptake by phytoplankton, resulting in δ15N-POM values greater than δ15N-PHY downstream of the WWTP in the +EFF parcels. This interpretation would also be consistent with the results of a previous investigation into spatial and temporal variability of δ15N of POM in the freshwater portion of the SFE by Cloern et al. (2002). In that study, the authors reported little overlap between δ15N-POM values and the δ15N values of potential organic matter sources to POM, suggesting that bacterial processing had overprinted the isotopic composition of a significant fraction of the organic matter sources present in bulk POM.

To trace the movement of WWTP-derived NH4+ into phytoplankton downstream of the WWTP, we employed a two-end-member mixing model approach. The data used for mixing model calculations of phytoplankton N source using Eq. (1) are shown in Fig. 5. Use of this approach requires knowledge of enrichment factors for NO3- (εNO3) and NH4+ (εNH4). Enrichment factors for phytoplankton N use have been measured in numerous laboratory and field investigations (Cifuentes et al., 1989; Waser et al., 1998; Altabet et al., 1999; Needoba et al., 2003; Karsh et al., 2014). While a range of values have been reported for fractionation during assimilation, in general, εNO3 tends to be lower (2–7 ‰) than εNH4 (0–25 ‰). However, enrichment factors are known to be impacted by light, growth rate, and N concentration, and both N-limited conditions and high growth rates have been shown to result in lower enrichment factors (Finlay and Kendall, 2007).

There have been relatively few field investigations of εNO3 in fresh water settings. In single species culture studies, values of εNO3 from < 1 ‰ to as high as 20 ‰ have been reported (Granger et al., 2004). In other coastal and estuarine field investigations εNO3 values between 2 and 7 ‰ have been reported (York et al., 2007, and references therein) and in a recent study in the Danube Delta a value of 2.7 ‰ was reported based on a Rayleigh distillation model (Möbius and Dähnke, 2015). In this study, we estimated εNO3 from the difference between δ15N-NO3- and δ15N-PHY in river water upstream of the WWTP where NH4+ concentrations are low and NO3- is the dominant N source. The average offset between δ15N-NO3- and δ15N-PHY in this portion of the river was 3 ‰ (n=6), which fits within the range of previously reported values from culture and field investigations of εNO3 where growth was not nutrient limited. Field investigations of εNH4 are even less common than NO3- investigations. Values ranging from 0 to 25 ‰ have been reported from both culture and field studies (Waser et al., 1999). In this study, the minimum value of εNH4 that resulted in solutions to Eq. (1) between 0 and 100 was 17 ‰ (using a εNO3 value of 3 ‰). An enrichment factor of 17 ‰ is relatively high compared to other studies; however, previous investigations of εNH4 focused on nutrient-limited conditions. Elevated NH4+ concentrations in this study may help explain the large apparent enrichment factor.

The percentage of the phytoplankton N pool that was derived from NH4+ (%NH4) versus NO3- calculated using values of εNH4 ranging from 17 ‰ (the minimum value required for valid solutions to the model) to 25 ‰ (highest reported value) illustrates that model solutions become increasingly sensitive to εNH4 as δ15N-PHY decreases downstream. Nevertheless, despite the uncertainty in enrichment factors, a gradual increase in the percentage of N derived from NH4+ is apparent (Fig. 6). In the October +EFF parcel the estimated portion of phytoplankton N derived from NH4+ increased to 50 % over 60 h of travel time and may have reached as high as 88 % after 80 h. Similar patterns were observed between the October and June samplings. However, mixing model calculations were only completed at three locations in June because δ15N-PHY was greater than δ15N-NO3- and δ15N-NH4+ at the top of the study reach (discussed below).

Mixing model calculations suggest that downstream of the WWTP, a significant portion of the phytoplankton N pool was derived from NO3- despite the presence of high concentrations of NH4+. However, N uptake experiments conducted using 15N-tracer incubation techniques as part of this effluent hold study indicate a near immediate switch from NO3- uptake to NH4+ uptake when NH4+ concentrations were elevated in this portion of the river (Travis, 2015; Kraus et al., 2017a). The apparently gradual increase over time in the proportion of δ15N-PHY derived from NH4+ suggests either that simultaneous uptake of NO3- and NH4+ occurs in the river under conditions not captured by the 15N-tracer incubations, or that the N turnover time of phytoplankton is much longer than the ∼ 80 h travel time covered in this study.

We can estimate potential N turnover time as the mean concentration of POM-N (µM) divided by the mean rate of N uptake (µM d-1) measured downstream of the WWTP during 24 h 15N-tracer experiments. This estimate assumes that all POM-N comes from phytoplankton and it represents a minimum turnover time because it is based on potential N uptake rates measured under high light conditions, in bottles where phytoplankton are isolated from the impacts of turbulence and mixing which may transport cells into lower light environments. In the October +EFF parcel, the concentration of particulate N (mean ± standard deviation) downstream of the WWTP was 4.0 ± 1.0 µM, while the potential NH4+ uptake rate was 1.4 ± 0.6 µM d-1 (mean ± standard deviation; Travis, 2015). This implies that it would take ∼ 66 h to completely turnover phytoplankton N with newly assimilated NH4+ if phytoplankton switched to 100 % NH4+ uptake. After 60 h of travel time in the presence of elevated NH4+ concentrations, mixing model calculations indicate that only ∼ 50 % of the phytoplankton N was derived from NH4+, suggesting NH4+ uptake rates in the river are much lower than the potential growth rates measured in 15N uptake experiments. Given that phytoplankton in the river likely experienced light limitation, it is reasonable to infer that lower in situ growth rates resulted in an N turnover time greater than 80 h. Because δ15N-PHY reflects a time-integrated mixture of N uptake, an N turnover time > 80 h would mute changes in δ15N-PHY following an abrupt switch to NH4+ uptake.

While there is a general consensus that phytoplankton preferentially take up NH4+ when NH4+ concentrations are elevated (for reviews see Dortch, 1990; Glibert et al., 2016), simultaneous uptake of NO3- and NH4+ has been documented in several field studies. For example, Berg et al. (2001) report nearly equal percentages of NO3- and NH4+ uptake with a small percentage of N uptake as urea for the spring bloom diatom Thalassiosira baltica. Likewise, Twomey et al. (2005) report near parity of uptake of N as NO3- and NH4+ for the phytoplankton community in the Neuse River estuary. Both field- and laboratory-based studies make it clear that different cells respond differently to the presence of multiple N sources (Dortch, 1990). Diatoms, for example, can reach maximum growth rates when using both NO3- and NH4+, whereas cyanobacteria appear to be NH4+ specialists (Senn and Novick, 2014, and references therein). During this study, diatoms were the most abundant type of algae, which would be consistent with simultaneous use of NO3- and NH4+. However, 15N-tracer uptake measurements made during 24 h bottle incubations in this section of the Sacramento River have consistently found that 15N-NO3 uptake rates are near zero when NH4+ concentrations are elevated (Parker et al., 2012a; Kraus et al., 2017a) meaning that if simultaneous uptake of NO3- and NH4+ occurred in the river it would appear to require conditions (such as light limitation) not captured in these incubations.

Another possible explanation for the observed gradual increase in the %NH4 making up the phytoplankton N pool in the +EFF parcels is that δ15N-PHY represents a mixture of phytoplankton actively taking up NH4+ (as observed in bottle incubations) and phytoplankton subsisting on an internal supply of NO3- acquired upstream of the WWTP. Previous investigations have shown that diatoms are capable of accumulating an internal DIN pool under both N-sufficient and N-deficient conditions and that DIN accumulation is impacted by prior conditioning of the cells (Dortch, 1982; Collos, 1982; Lomas and Glibert, 2000). During the June transect δ15N-PHY was greater than δ15N-NO3 at the top of the study reach and remained greater than both δ15N-NO3- and δ15N-NH4+ for > 40 h of downstream transport. Because δ15N-PHY should be equal to or lower than the δ15N value of its source N, it appears that during the June experiment phytoplankton N was acquired above the study reach. A similar observation was made in the Childs River of Massachusetts, where phytoplankton maintained a stable δ15N value over several days of downstream transport while both δ15N-NO3- and δ15N-NH4+ decreased, leading York et al. (2007) to infer that phytoplankton growth was sustained by internal N stores. If a portion of the phytoplankton community acquired N upstream of the study reach and then was advected downstream without taking up additional N, this could account for the 20 % apparent contribution of NO3- to δ15N-PHY after 80 h of transport.