Corresponding author: P. S. Levi, Department of Bioscience, Aarhus University, Ny Munkegade 116, DK-8000 Aarhus C, Denmark. (email@example.com)
 Biogeochemical transformations may represent an important pathway influencing the fate of nutrient subsidies in stream ecosystems. Pacific salmon (Oncorhynchus spp.) provide an ammonium (NH4+) subsidy to streams during their annual spawning runs, which may be transformed to nitrate (NO3−) via sediment nitrification. Increases in either forms of dissolved inorganic nitrogen may have ecosystem effects both at the reach and watershed scales, including the fertilization of algal biofilms and elevated export of nutrients to downstream ecosystems. In the nonnative range of salmon, where spawning runs are a relatively new phenomenon, few studies have explored the effect of introduced salmon on ecosystem processes. To assess the effect of nonnative salmon on dissolved inorganic nitrogen dynamics in Great Lakes tributaries, we quantified sediment nitrification in five streams before, during, and after the spawning run in 2009. Overall, sediment nitrification rates were higher in the channel thalweg (mean ± SE = 1.9 ± 0.1 mg N/gAFDM/d) compared to channel margins (mean ± SE = 0.9 ± 0.1 mg N/gAFDM/d). In the two streams with the largest salmon runs, nitrification was highest in the channel thalweg prior to salmon, but margin sediments had higher nitrification during the run. Among all streams, variation in nitrification rates was habitat specific, predicted by exchangeable NH4+ in sediments from the thalweg and predicted by salmon biomass for sediments in the channel margin. Nonnative salmon provide a pulsed source of inorganic nitrogen to Great Lakes tributaries, yet dissimilatory biogeochemical transformations such as nitrification may alter the form of the NH4+ subsidy and potentially influence downstream lakes via export of both NH4+ and NO3−.
 Nonnative species in freshwater ecosystems may alter biogeochemical transformations in their introduced ranges and, as a result, cause shifts in ecosystem function. Directly, nonnative species may influence biogeochemical transformations when they mobilize and/or excrete inorganic nitrogen (N), phosphorus (P), and carbon (C). For example, zebra mussels (Dreissena polymorpha) in lakes excrete ammonia (NH3) and carbon-rich pseudofeces, which fuel a cascade of nutrient transformations, including nitrification and subsequent denitrification [Bruesewitz et al., 2008]. Indirectly, nonnative species may alter community structure or their physical habitat resulting in food web shifts and subsequent changes in nutrient transformations. For example, benthic grazing by New Zealand mud snails (Potamopyrgus antipodarum) increased the biomass of N-fixing diatoms, resulting in a higher flux of N into streams containing high mud snail densities [Arango et al., 2009]. Through these biogeochemical pathways, nonnative species may elevate nutrient transformations above background levels [Ehrenfeld, 2010].
 Pacific salmon have been introduced via fisheries management throughout the world, including in the Laurentian Great Lakes, New Zealand, and Chile [Crawford, 2001; Quinn, 2005]. Nonnative Pacific salmon can increase stream water NH4+ concentrations via the excretion of NH3 during their migration and spawning [e.g., Gende et al., 2002; Collins et al., 2011]. Several studies conducted in the native range of salmon have also observed concurrent increases in stream water NO3− during salmon runs [Johnston et al., 2004; Levi et al., 2011], though salmon do not directly excrete NO3− [Wood, 1995]. The elevated NO3− concentrations observed during salmon runs suggest that salmon-derived NH4+ stimulates sediment nitrification. For example, nitrification rates tripled when spawning salmon were present in Southeast Alaska streams [Levi et al., 2012, 2013]. Therefore, the pulse of salmon-derived NH4+ potentially represents a novel resource to stream ecosystems in their introduced range. In addition, the transformation of salmon-derived NH4+ may result in increased NO3− concentrations in these streams, which could have implications for nitrogen dynamics in sensitive downstream water bodies (e.g., Great Lakes).
 Biogeochemical transformations may change the form of a nutrient subsidy and have cascading effects on the recipient ecosystem, but these transformations have been largely overlooked in most studies on subsidies [e.g., Polis et al., 2004]. Instead, research on nutrient subsidies, and on salmon-derived nutrients in particular, have generally focused on the food web effects of the subsidy on higher trophic levels [e.g., Janetski et al., 2009]. Nitrification of salmon-derived NH4+ has several potential ramifications given the unique qualities characterizing each dissolved nitrogen species. Nitrate is often less retentive than NH4+ in stream ecosystems, generally traveling further downstream prior to removal from the water column [Mulholland et al., 2000; Hall and Tank, 2003] and requiring additional energy for biotic assimilation compared to NH4+ [Dortch, 1990]. Furthermore, NO3− has a weak affinity for anion exchange sites in stream sediments, whereas NH4+ has a strong affinity for cation exchange sites [Orcutt and Nilsen, 2000]. For these reasons, NO3− export is often higher from stream ecosystems and considered “leaky” compared to NH4+ [Tank et al., 2008]. As a result, nitrification of salmon-derived NH4+ and subsequent downstream NO3− export may contribute more dissolved N to the Great Lakes, where anthropogenic inputs, such as fertilizer associated with agriculture and naturally sandy soils that promote drainage, already contribute to elevated stream nutrient export in these systems [Han and Allan, 2008]. Therefore, the nutrient subsidy provided by nonnative salmon may influence dissolved nitrogen dynamics at the habitat scale as well as across the watershed.
 Our research goal was to investigate the effect of nonnative Pacific salmon on biogeochemical transformations in streams within their introduced range. We predicted that nonnative salmon would increase stream water NH4+ concentrations and, in turn, sediment nitrification rates. In addition to the nutrient enrichment associated with salmon runs, salmon cause benthic disturbance during their migration and spawning [Peterson and Foote, 2000], which may affect ecosystem processes at the habitat scale. In order to assess the effect of introduced salmon at the habitat and watershed scales, we measured sediment nitrification rates in the channel thalweg and channel margin in five tributaries to the Great Lakes in northern Michigan and southern Ontario. The salmon spawning runs were large in two of our study streams, and their biomass at the reach scale was comparable to salmon-bearing streams in the Pacific Rim [Janetski et al., 2009]. In contrast, salmon biomass was low in the other three streams. Our study addressed the transformation of NH4+ at the habitat scale, which may influence N dynamics and NO3− export at the reach scale, as well as the management of nonnative Pacific salmon at the watershed scale.
2 Materials and Methods
2.1 Study Streams and Salmon Biomass
 We conducted our study in five streams located in Michigan, USA, and Ontario, Canada (Figure 1 and Table 1), designating a representative 400 m reach in each stream as our study reach. The streams are located in watersheds dominated by northern hardwood vegetation and managed for forestry and recreation (i.e., Hiawatha and Manistee National Forests in Michigan and Garden River First Nation Reserve in Ontario) with negligible agricultural or urban land use in the basins. In addition, there is a fish hatchery located upstream of our study reach in Thompson and upstream reservoirs in Crystal and Pendill's. The streams are tributaries to Lakes Superior, Michigan, and Huron, and receive annual spawning runs of introduced Chinook (O. tshawytscha) and coho (O. kisutch) salmon. The majority of salmon in the Great Lakes are stocked via hatcheries, though a small population reproduces naturally in these streams [Carl, 1982]. Due to the low reproductive success, the annual runs to these streams vary considerably in both space and time [Crawford, 2001]. To estimate salmon biomass during the spawning run (kg m−2), we counted all live salmon and salmon carcasses in the wetted channel of each 400 m study reach on three to five dates during the migration in each stream and noted the species (i.e., Chinook or coho). Using average weights obtained from salmon collected in the field [i.e., Chinook = 4.9 kg, coho = 2.2 kg], we determined the total salmon biomass in the stream and scaled these estimates for stream size by dividing by the area of the wetted channel. Salmon biomass upstream of our study reaches was comparable to the biomass observed within our reach (P. S. Levi, D. J. Janetski, personal observation), though we did not quantify this directly.
Table 1. Stream Characteristics of Our Five Study Sites in Northern Michigan and Southern Ontario. Streams Are Ordered in Terms of Increasing Salmon Biomass at the Peak of the Run
Numbers following stream name correspond with stream location on Figure 1.
Ambient nutrient concentrations refer to concentrations during the August sampling before the arrival of salmon. NH4+ = ammonium, NO3− = nitrate, and SRP = soluble reactive phosphorus.
42.4 ± 26.4
45.5 ± 29.0
23.9 ± 19.8
4.6 ± 8.7
17.1 ± 38.6
2.2 Sediment Nitrification Assay
 We collected sediment samples for nitrification assays from each of the five study streams before, during, and after the salmon run in 2009. During each time period at each stream, we collected sediments from both the thalweg and margin of the stream channel to examine spatial differences in sediment nitrification throughout the study reaches. In both benthic habitat types, we collected three to five sediment cores (30 cm2 by 2 cm deep; ~250 ml total) and pooled them to create a composite sediment sample. We replicated the sediment collection in each of the two habitats along five transects for a total of 10 sampling locations in each stream. We returned the sediments to the laboratory and immediately began the nitrification assays.
 We used the nitrapyrin-inhibition method to determine sediment nitrification rates [Hall, 1984; Kemp and Dodds, 2001]. For each sample replicate, we placed 25 ml of sediment each into a “blocked” and a “reference” flask along with 50 ml of unfiltered stream water. In the blocked flask, we inhibited nitrification by blocking the conversion of NH4+ to NO3− with 20 µl of a 10% solution of nitrapyrin dissolved in dimethyl sulfoxide (DMSO). In the reference flask, we added 20 µl DMSO alone, which allowed nitrification to occur. We incubated the flasks on a rotary shaker at 175 rpm for 24 to 48 h. After incubation, we subsampled 25 ml of the sediment slurry, added 25 ml of 2M KCl to flush NH4+ from cation exchange sites, and shook the samples for 1 h. We centrifuged each sample, filtered the supernatant into bottles, and froze the samples for later NH4+ analysis (described below). To determine sediment nitrification rates for each paired set of flasks, we calculated the difference in NH4+ between the blocked and reference flasks and scaled these differences with the duration of the assay and expressed the nitrification rates in two ways: nitrification per unit sample area of stream bottom (mg N m−2 d−1) and nitrification per unit sediment organic matter content (mg N gAFDM−1 d−1). The nitrification rates we measured using the nitrapyrin-inhibition method were likely higher than in situ rates because the slurries were left open to the atmosphere, thereby optimizing redox conditions (e.g., well oxygenated), and they were also incubated at room temperature [Strauss et al., 2004]. However, we did not amend the sediments with NH4+ [e.g., Arango and Tank, 2008]; therefore, the nitrification rates we report here represent a somewhat conservative estimate of the potential rates.
2.3 Stream Sediment Characterization
 We analyzed the exchangeable NH4+ and NO3− (i.e., sediment pore water) and organic matter content of each sediment sample used in the nitrification assays [Bruesewitz et al., 2008]. To determine exchangeable NH4+ and NO3−, we subsampled 30 g of wet sediment from each sample, added 30 ml of 2M KCl, and shook these samples for 1 h to flush all NH4+ and NO3− from ion exchange sites. Then we centrifuged these samples, filtered the supernatant, and froze the samples for subsequent NH4+ and NO3− analysis (described below). To determine sediment organic matter content, we subsampled 5 ml of sediment from each pooled sample and dried these samples at 60°C for 24 h. We weighed these samples for the dry mass, ashed the samples at 550°C for 4 h, and re-weighed the samples. To calculate percent organic matter (%OM), we subtracted one from the ratio of ash-free dry mass (AFDM) to dry mass and multiplied by 100. In the field, we estimated mean sediment size for each study stream by measuring the median diameter of five random particles collected along 30 lateral transects throughout the reach at each time point (n = 450).
2.4 Water Chemistry Analysis and Environmental Characteristics
 We collected three replicate water samples for dissolved inorganic nutrient analyses during each visit to the study streams before, during, and after the salmon run, which ranged from 1 to 6 sampling dates per stream per time period. We filtered stream water through a 0.7 µm glass fiber filter in the field into 60 ml Nalgene bottles and froze the water samples for subsequent laboratory analysis. We analyzed the samples for NH4+–N, NO3−–N, and SRP using the sodium-hypochlorite, cadmium-reduction, and ascorbic acid methods, respectively, on a Lachat QC-8500 Flow Injection Autoanalyzer (Lachat Instruments, Loveland, CO, USA) [APHA, 2005].
 During each site visit, we also quantified stream discharge by measuring water velocity at fixed transects across the channel with a Marsh McBirney Flo-Mate flow meter (Hach Company, Loveland, CO, USA) [Gore, 2006]. We created rating curves using discharge and water height measurements (R2 range = 0.57 to 0.95) to estimate continuous discharge using deployed capacitance meters (Dataflow Systems, Inc., Christchurch, New Zealand), which collected water height data every 30 min from June to November.
2.5 Statistical Analyses
 We used an analysis of co-variance (ANCOVA; α = 0.05) [Zar, 2009) to test for differences in stream water NH4+ and NO3− with time period as the factor (i.e., before, during, and after the salmon run) and salmon biomass as the co-variate. To test for differences in exchangeable NH4+ and NO3− and sediment nitrification rates, we used an ANCOVA with two main factors, time period and channel habitat (i.e., thalweg and margin), and salmon biomass as the co-variate. Given the variation in salmon biomass among our study streams (i.e., two streams with biomass >0.3 kg m−2 (“high”) and three with biomass <0.1 kg m−2 (“low”); Table 1), we conducted a two-way ANOVA on data from the high salmon biomass streams using time period and habitat as the factors (Bonferroni-corrected α = 0.02) [Zar, 2009]. We also used simple linear regression (SLR; α = 0.05) [Zar, 2009] to test whether stream water NH4+, exchangeable NH4+, or salmon biomass predicted sediment nitrification rates.
 To place our study in the context of other studies, we compared our results to previously published sediment nitrification rates from streams throughout the United States [Bernhardt et al., 2002; Webster et al., 2003; Starry et al., 2005; Arango and Tank, 2008, LINX II, unpublished data; Levi et al., 2012]. For the purposes of comparison, we defined “reference” streams as those with background NH4+ <30 µg N L−1 and in watersheds dominated by native vegetation. Several of the studies used in our literature review also included streams draining agricultural and urban watersheds. To analyze these data for differences in stream water NH4+ and sediment nitrification rates, we used a one-way ANOVA (α = 0.05) with land use as the main factor (i.e., our study streams, reference, agriculture, and urban). When necessary, we used log or square-root transformations to meet statistical assumptions. We performed all statistical analyses using SYSTAT 12 (2007, Systat Corp., Chicago, IL, USA).
3.1 Nitrification Rates Increase During the Salmon Migration
 Stream water NH4+ concentrations were influenced by salmon biomass (p < 0.001; Figure 2 and Table 2). Also, concentrations were significantly higher in the streams with large runs relative to those with small runs among all time periods (mean NH4+ ± SE = 80.7 ± 12.8 and 9.4 ± 0.9 µg N L−1, respectively). Variation among streams was considerable when all data were pooled; thus, NH4+ concentrations did not vary consistently through time (p = 0.4; Figure 2a). In general, nitrate concentrations were higher than NH4+ but did not differ among time periods or with salmon biomass (Figure 2b and Table 2).
Table 2. ANCOVA and ANOVA Statistical Test Results From Stream Water Nitrogen Concentrations, Sediment Exchangeable Nitrogen Concentrations, and Sediment Nitrification (NIT) Rates With All Data From Our Study and Data From High Salmon Biomass Streams Onlya
Time Period × Habitat
aBold p-values indicate statistical significance (p < 0.05). NH4+ = ammonium, NO3− = nitrate, and df = degrees of freedom.
Stream water NH4+
Stream water NO3−
Sediment NIT rates
High biomass NIT
 Exchangeable NH4+ and NO3− varied by habitat type (i.e., channel thalweg versus margin; Figure 3) as well as by time period and with salmon biomass. Overall, exchangeable NH4+ was an order of magnitude higher than surface water NH4+ (mean ± SE = 453.1 ± 61.5 versus 38.5 ± 9.4 µg N L−1). Among habitats, exchangeable NH4+ was approximately double in the channel margin sediments compared to the sediments in the thalweg (mean ± SE = 625.6 ± 46.1 versus 246.8 ± 19.6 µg N L−1; Figures 3a and 3b). In contrast, exchangeable NO3− concentrations were lower in the channel margin relative to the thalweg (mean ± SE = 22.6 ± 14.6 versus 71.6 ± 14.6 µg N L−1; Figures 3c and 3d and Table 2). Both exchangeable NH4+ and NO3− differed by time period (p < 0.001; Table 2), where NH4+ was generally lower during and after the salmon run compared to before, while NO3− was higher during the run. Furthermore, salmon biomass was a significant co-variate for both forms of exchangeable nitrogen (p = 0.02 and <0.001 for NH4+ and NO3−, respectively); thus, the variation in exchangeable N across time periods was influenced by not only the presence of salmon, but also by the size of the run.
 Sediment nitrification rates were twice as high in the channel thalweg (mean ± SE = 1.9 ± 0.1 mg N gAFDM−1 d−1) compared to the channel margin (0.9 ± 0.1 mg N gAFDM−1 d−1; p <0.001; Figure 4 and Table 2). Although nitrification rates did not vary consistently through time (p = 0.5; Table 2), salmon biomass was a significant co-variate (p < 0.001) mediating the nitrification response through time. Therefore, partitioning the two streams with large salmon runs, we found that nitrification rates did differ among time periods, but the change in nitrification was dependent on stream habitat (Table 2). For example, nitrification increased in the channel margin during and after the salmon run but generally decreased in the channel thalweg (Figure 4). In the streams with low salmon biomass, sediment nitrification rates did not change in response to salmon.
3.2 Exchangeable NH4+ and Salmon Biomass Predict Sediment Nitrification Rates
 Variation in sediment nitrification rates among time periods and streams was explained by variation in exchangeable NH4+ concentrations and salmon biomass (Figure 5). Exchangeable NH4+ positively predicted nitrification rates in sediments of the channel thalweg (SLR, R2 = 0.48, p <0.001) but was not a significant predictor of rates in the channel margin (p = 0.2; Figure 5a). Conversely, nitrification rates in sediments of the channel margin were positively related to total salmon biomass (SLR, R2 = 0.60, p = 0.001), though salmon biomass did not predict variation in the channel thalweg (p = 0.5; Figure 5b). In summary, the relationships between the explanatory variables and nitrification rates were habitat specific rather than consistent across the stream reach.
3.3 High NH4+ and Sediment Nitrification in Great Lakes Streams
 A comparison between our study and previous studies demonstrated that NH4+ concentrations and sediment nitrification rates in the Great Lakes tributaries were higher than concentrations and rates measured in other biomes (Figure 6). Surface water NH4+ concentrations in our study streams were comparable to those reported for urban streams and higher than agricultural and reference streams (one-way ANOVA, df = 3, 150; F-stat. = 8.5; p-value < 0.001), where we define “reference” streams as those with ambient NH4+ <30 µg N L−1 and in watersheds dominated by native vegetation. Similarly, nitrification rates were significantly higher in our study streams relative to previous studies from all land use types (one-way ANOVA, df = 3, 150; F-stat. = 22.9; p-value < 0.001).
4.1 Nitrification in Great Lakes Tributaries High Relative to Other Studies, Salmon Streams
 Nitrification in streams has been measured extensively throughout North America. In comparison, both NH4+ concentrations and sediment nitrification rates were high in our study streams relative to other “reference” streams despite the forested land cover in the watersheds of our study streams [Bernhardt et al., 2002; Webster et al., 2003; Starry et al., 2005; Arango and Tank, 2008, Levi et al., 2012] (Figure 6). The difference in NH4+ concentrations between our study and other reference streams may be due to the presence of salmon (i.e., Pine Creek) or elevated background concentrations (i.e., Thompson Creek). In addition, the nitrification rates quantified in the Great Lakes tributaries were more comparable to rates in streams draining watersheds with urban and agricultural land use [Arango and Tank, 2008; LINX II, unpublished data]. These high nitrification rates, even before salmon, may be because these tributaries have ideal conditions for nitrification (i.e., high exchangeable NH4+, fine sediments, and low gradient) [Strauss et al., 2002]. Overall, the comparison suggests that the transformation of NH4+ was high in our study streams, despite the forested land cover in the surrounding watersheds.
 Nonnative Pacific salmon did not consistently increase stream water NH4+ or NO3−, but we did find stream-specific changes in sediment nitrification. Previous research examining nitrification of salmon-derived NH4+ demonstrated that nitrification rates and NO3− concentrations increased markedly during the salmon run in streams of Southeast Alaska [Levi et al., 2012] (Figure 7). In part, the different responses to salmon in the Great Lakes and Southeast Alaska can be attributed to the high spatial and temporal variability in salmon run size and anthropogenic influences on stream water chemistry in Great Lakes streams (e.g., high background NH4+ concentrations from an upstream fish hatchery or reservoir) [Crawford, 2001]. Only two of our five study streams had large salmon runs with biomass that was comparable to streams with healthy runs in the native range of salmon [Janetski et al., 2009]. Also, while studies from the native range of salmon report increased NO3− during the run [Johnston et al., 2004; Levi et al., 2011], NO3− in the Great Lakes tributaries was high in all streams, including those with small salmon runs (i.e., Crystal), perhaps due to the underlying geology of or greater human impacts in these watersheds (e.g., upstream reservoirs) [Johnson et al., 1997]. Therefore, increased stream water N and nitrification rates associated with salmon spawning runs were less pronounced in the Great Lakes tributaries compared to Southeast Alaska but significant at the habitat scale in streams with high salmon biomass.
4.2 Reach-Scale Nitrification Rates Reflect Habitat-Specific Response to Salmon
 The habitat-specific patterns in nitrification among our study streams and through time were predicted by exchangeable NH4+ concentrations and salmon biomass. In the channel thalweg, exchangeable NH4+ controlled sediment nitrification rates, which has been shown to be a strong predictor of nitrification in other studies [Kemp and Dodds, 2001; Strauss et al., 2002]. While exchangeable NH4+ was often higher in the sediments of the channel margin, nitrification remained low. These low rates relative to high NH4+ availability may be due to NH4+ saturation resulting in a shift to secondary controls on nitrification, such as dissolved oxygen or pH [Strauss et al., 2004; Bernot and Dodds, 2005]. In low gradient streams dominated by fine sediments, such as those in the Great Lakes basin, the accumulation of organic matter may result in higher heterotrophic respiration [Hoellein et al., 2009], thereby reducing dissolved oxygen and pH and inhibiting nitrification.
 In the channel margin, sediment nitrification rates were higher in the presence of live salmon and carcasses. This positive relationship may reflect an increase in the availability of NH4+ via salmon excretion or a shift in secondary controls on nitrification via benthic disturbance from spawning salmon [e.g., Peterson and Foote, 2000; Gende et al., 2002; Tiegs et al., 2009]. Given the high background NH4+ in the margin sediments, the elevated nitrification during and after the run in this habitat is likely related to bioturbation, which may increase oxygen concentrations or decrease accumulated organic matter in the channel margin, both of which would improve the habitat for nitrifying bacteria.
 The response of sediment nitrification to nonnative Pacific salmon varied by stream habitat in our study of Great Lakes tributaries. Nitrification rates in the channel margin increased during the spawning run on a per unit area basis (mg N m−2 d−1) but decreased in the channel thalweg (Figure 7). In contrast, sediment nitrification rates consistently increased in both the channel thalweg and margin during the salmon run in Southeast Alaska streams [Levi et al., 2012] (Figure 7), likely due to less of a contrast between the habitats (e.g., similar exchangeable NH4+, sediment size) and generally low organic matter content in those cobble-dominated streams. Furthermore, in Southeast Alaska streams, salmon spawned across the wetted width of the channel, but in the Great Lakes tributaries, they spawned only in the channel thalweg (PS Levi, DJ Janetski, personal observation). As a result, salmon likely had a similar influence on both habitats in Southeast Alaska streams but may have affected thalweg and margins differently in the Great Lakes streams, especially given their ecological roles as both nutrient subsidies and ecosystem engineers [e.g., Peterson and Foote, 2000; Gende et al., 2002; Tiegs et al., 2009]. Our results emphasize that in Great Lake tributaries, the reach-scale response of nitrifying bacteria to salmon was mediated by variability in habitat conditions between sediments of the channel thalweg and margin.
4.3 Biogeochemical Transformations May Alter the Effect of Salmon at the Watershed Scale
 The nitrification of salmon-derived NH4+ at the reach scale may influence N dynamics at the watershed scale. Previous research has demonstrated that salmon runs increase stream water NH4+ and NO3− concentrations [Johnston et al., 2004; Levi et al., 2011]. These results suggest that nitrifying bacteria may modify the nutrient subsidy by increasing available NO3−, which has important implications for N export to downstream ecosystems. Ammonium is more retentive in stream ecosystems than NO3− due to both biological and physical mechanisms [Mulholland et al., 2000]. Stream autotrophs prefer NH4+, which is more energetically favorable for assimilation compared to NO3− [e.g., Dortch, 1990]. In addition, NH4+ has a stronger affinity to bind to ion exchange sites than NO3− [Orcutt and Nilsen, 2000], making NO3− more “leaky” relative to NH4+. Given the differences between these two forms of dissolved inorganic nitrogen in aquatic ecosystems, nonnative salmon may influence nitrogen dynamics in streams within their introduced range via multiple mechanisms. Streams may become enriched in NH4+ due to salmon excretion at the reach scale, though the level of enrichment will vary based on salmon run size and antecedent stream water conditions. The annual pulse of salmon-derived NH4+ to Great Lakes tributaries may alleviate nutrient limitation of stream autotrophs and increase primary production [Rüegg et al., 2011; Levi et al., 2013], which may alter food web structure and subsequent trophic dynamics. In addition, watersheds in the Great Lakes with salmon-bearing streams may have increased NO3− export due to salmon-stimulated nitrification, possibly increasing NO3− export to the Great Lakes from tributaries with large salmon runs. A potential outcome of elevated NO3− delivered to freshwater estuaries are algal blooms and, in turn, water column hypoxia [e.g., Howarth et al., 2011]. Thus, both forms of dissolved inorganic N associated with nonnative salmon have the potential to shift nutrient limitation status in freshwater ecosystems in the Great Lakes if exported from rather than retained within the tributaries.
 Nonnative Pacific salmon represent a managed, human-mediated N subsidy to the Great Lakes tributaries and have multiple effects on streams throughout their introduced range. The ecosystem effects vary and include nutrient enrichment [Collins et al., 2011], stimulating or decreasing whole-stream metabolism [Levi et al., 2013] and heavy metal contamination [Janetski et al., 2012], and enhancing biogeochemical transformations (this study). However, these numerous effects appear to be less pronounced when compared to those seen within their native range. For instance, NH4+ concentrations tripled in Pine Creek but did not change in the other four streams during the salmon run; whereas NH4+ often increases an order of magnitude in similar studies within the native range of salmon [Janetski et al., 2009]. Ecosystem processes within the native range of salmon have evolved with salmon runs for millennia [Quinn, 2005]. In contrast, streams in the Great Lakes watershed have received salmon runs for less than 50 years [Crawford, 2001], and these systems may still be adjusting to the nutrient enrichment and benthic disturbance associated with annual spawning activities. Therefore, the long-term effects of nonnative salmon on ecosystem processes in their introduced range remain to be seen, though our study suggests a habitat specific rather than generalized response for the relationship between the salmon subsidy and a key biogeochemical transformation in Great Lakes tributaries.
 Nonnative Pacific salmon may alter biogeochemical transformations at the reach scale and therefore nutrient availability and export at the watershed scale. The five Great Lakes streams we studied had high rates of ambient sediment nitrification relative to previous studies in North American streams (Figure 6). Furthermore, the presence of salmon resulted in a shift of the highest nitrification rates per unit stream area between the channel thalweg and margin (Figure 7). The transformation of salmon-derived NH4+ to NO3− has subsequent watershed-scale effects on inorganic nitrogen availability and export. As NO3− is often less retentive in stream ecosystems [Mulholland et al., 2000], the transformed N may travel further downstream, extending the influence of nonnative salmon to ecosystems beyond their immediate vicinity in the stream reach. While the indirect inputs of anthropogenic N in freshwaters have been extensively studied [e.g., Carpenter et al., 1998], nonnative species represent an additional indirect source of anthropogenic N that also alters these ecosystems [Ehrenfeld, 2010]. The continued management of nonnative Pacific salmon in the Great Lakes must consider the effects of these organisms on ecosystems beyond the lakes themselves. Salmon-derived nutrients transported during annual spawning runs may have unintended consequences on recipient stream ecosystems, including increased fluxes of inorganic N in multiple forms to and from the stream.
 We thank Ashley Moereke at Lake Superior State University for logistical support and use of laboratory facilities, David Janetski for salmon biomass estimates, and Dominic Chaloner for providing statistical expertise and the map of our study sites. We are grateful for invaluable field and technical support provided by Michael Brueseke, Ross Gay, Kate Harriger, Thomas Levi, and James Summers. Our research funding was provided to P.S.L. by an NSF-Doctoral Dissertation Improvement Grant (DEB-0910124) and an NSF-IGERT Graduate Fellowship (DGE-0504495).