The experiment took place in an experimental glasshouse consisting of multiple heated cabins, each with a size of 64 m² and a roof top height of 4 m. Two of these cabins were used for this study, cabin no. 7 for pre-cultivating tomato plants (Solanum lycopersicum cv. ‘Cheramy F1’) and cabin no. 5 for conducting the experiment. Temperature in the cabins was set to 20/18°C (day/night), and roof top ventilation was opened at temperatures above 23/20°C (day/night). Shading was done automatically at photosynthetically active radiation (PAR) values above 900 μmol m⁻² s⁻¹ and artificial lighting was applied between 5:00 and 12:00 CET, if PAR values were below 180 μmol m⁻² s⁻¹. Air temperature and humidity in the cabins as well as roof top PAR were continuously monitored by a climate computer (Supplementary Figure S1). Tomato plants were sown on 26th July 2021 and after germination in moistened sand, 64 seedlings were transplanted into pre-weighed rock wool cubes (10 × 10 × 6.5 cm; Grodan B.V., Roermond, Netherlands) for further cultivation. On 2nd September each two planted rock wool cubes were put on one rock wool slab (100 × 20 × 7.5 cm; Grodan Vital, Grodan B.V., Roermond, Netherlands) at a distance of 50 cm. One-half of the planted rock wool slabs were installed in eight hydroponic units with elevated gutters in cabin no. 5, which included separate fertigation systems and were later used for the ¹⁵N labeling. The other half was further cultivated in cabin no. 7 in four gutters on the ground, which shared one fertigation system. In both cases, the collected drain solution (i.e., leachate) was re-used and mixed with fresh nutrient solution in storage tanks as needed (closed hydroponic system with re-circulating nutrient solution). The nutrient solution from the storage tanks was supplied to plants via pumps, PE tubes, and drippers inserted into the rock wool cubes. The tomato plants were supplied with a custom-made nutrient solution modified after the recipe of de Kreij et al. (2003), which had a high NO₃⁻ to NH₄⁺ ratio (~20:1) that was found optimal for tomato cultivation. Macro and micro nutrients were dissolved in de-ionized water targeting a pH of 5.6 and an electrical conductivity (EC) of 2 mS cm⁻¹. The pH and EC values in the storage tanks were regularly monitored (Supplementary Figure S2). Tomato seedlings were supplied with an N concentration of 361 mg L⁻¹ at the beginning (starter solution; 338 mg L⁻¹ NO₃⁻-N and 23 mg L⁻¹ NH₄⁺-N). After the development of the 5th truss and the first green fruits on, from 4th October, the N concentration in the nutrient solution was reduced to 165 mg L⁻¹ (refill solution; 151 mg L⁻¹ NO₃⁻-N and 14 mg L⁻¹ NH₄⁺-N). The composition of the different nutrient solutions used in this study can be found in Supplementary Table S1. Each hydroponic unit in cabin no. 5 consisted of a 4 m gutter in which three rock wool slabs, two with plants and one unplanted, were placed and a nutrient solution storage tank filled up to approximately 40 L (Supplementary Figure S3). Two sampling periods were selected according to expected differences in plant N uptake and associated assimilate distribution in the root-shoot system, representing high growth and N uptake rates during early development and a more balanced assimilate distribution during fruit ripening. The first sampling and ¹⁵N labeling campaign were performed on 22nd and 23rd September, when the tomato plants developed the 3rd truss and first flowers. Subsequently, the 16 planted rock wool slabs (32 plants) in cabin no. 5 were completely removed (destructive sampling, described below) and replaced by the other 16 planted rock wool slabs pre-cultivated in cabin no. 7 on 24th September. The eight unplanted rock wool slabs were also exchanged with fresh rock wool slabs. To avoid carryover of ¹⁵N label, the hydroponic gutters were covered with plastic film below the rock wool slabs until 23rd September to reduce contact with the ¹⁵N-enriched nutrient solution. Both, the gutters and pumps for nutrient solution, were thoroughly cleaned with a detergent/disinfectant (MENNO Florades^®, MENNO CHEMIE-VERTRIEB GMBH, Langer Kamp, Germany) before installing the unlabeled plants and rock wool slabs. Furthermore, the storage tanks and the tubing as well as the drippers for nutrient solution were completely replaced with new material. To ensure the supply of further growing plants with water and nutrients, larger storage tanks were used (Supplementary Figure S4) and filled up to approximately 200 l. The experiment ended with the second sampling and ¹⁵N labeling campaign on 3rd and 4th November, when the tomato plants developed the 8th truss and the first fruits were ripe.

For measuring the gas fluxes, the closed chamber method as described by Karlowsky et al. (2021) was used. Acrylic glass chambers with two small openings for plant stems were fitted around the rock wool slabs (planted and unplanted) and sealed with foam rubber to obtain a closed headspace with a volume of approximately 16 l (Supplementary Figure S5). Over a period of 1 hour after closing, four gas samples (each 30 ml) were taken in 20 min intervals with a 30 ml syringe through a sampling port on top of the chamber. The gas samples were transferred to 20 ml glass vials with silicone/PTFE septa (type N17, MACHEREY-NAGEL GmbH & Co. KG, Düren, Germany) for transport and were analyzed on the same day by a gas chromatograph (GC 2010 Plus, Shimadzu Corporation, Kyoto, Japan) equipped with an electron capture detector (ECD) for N₂O. The measured concentrations in μmol mol⁻¹ were converted to μmol m⁻³ by applying the ideal gas law, including a correction for the temperature at the time of sampling. Afterward, gas fluxes were calculated using the R package “gasfluxes” [version 0.4–4; (Fuss et al., 2020)] by robust linear regression (except one case with only 3 time points, for which standard linear regression had to be used). Input variables used were gas concentration (μmol m⁻³), chamber volume (m³), time after closing the chamber (h), and area covered (m²). The latter was set to 1 m² assuming a typical density of greenhouse-cultivated tomato plants of 2 plants m⁻². The resulting gas fluxes in μmol m⁻² h⁻¹ were further converted to g ha⁻¹ d⁻¹ based on molar masses.

Natural abundance samples were taken on 22nd September and 3rd November shortly before the ¹⁵N labeling from each hydroponic unit in cabin no. 5 (from here on called “experimental unit”). These included plant samples, nutrient solution samples, and gas samples from planted rock wool slabs. For the latter, 140 ml of air was collected from the headspace of rock wool substrate with a syringe at the end of gas flux measurements after 1 h of N₂O enrichment in the closed chambers. The gas samples were transferred into 120 ml crimp-cap glass vials closed with gray butyl septa (type ND20, IVA Analysentechnik GmbH & Co. KG, Meerbusch, Germany) for later stable isotope analysis. To determine natural abundance δ¹⁵N values of plants, the tips (first three leaflets) of 2–3 fully developed leaves from one plant in each experimental unit were sampled and dried at 80°C for at least 48 h. Approximately 15 ml of nutrient solution (mixture with leachates) was sampled from the storage tank of each experimental unit and then stored at −20°C for later δ¹⁵N analyses. In addition, three samples of de-ionized water were taken to determine the natural abundance δ¹⁸O values of the nutrient solution water.

On both dates, the ¹⁵N labeling took place directly after the natural abundance sampling at approximately 12:00 pm CET. The remaining nutrient solution in the experimental units was removed as far as possible and 15 l of ¹⁵N-labeled nutrient solution was added in the storage tanks of each unit. In a randomized way, four units received a nutrient solution with ¹⁵N-enriched NH₄⁺ (¹⁵NH₄⁺) and four units received a nutrient solution with ¹⁵N-enriched NO₃⁻ (¹⁵NO₃⁻). This was done by adding ammonium nitrate (NH₄NO₃; SIGMA-ALDRICH, Saint Louis, MO, United States) with 10.5/11 atom-% ¹⁵N (¹⁵NH₄⁺/¹⁵NO₃⁻) as only N source. The composition of the nutrient solution used for the ¹⁵N labeling can also be found in Supplementary Table S1. In total, 115 mg of ¹⁵N was applied to each ¹⁵NH₄⁺ unit and 120 mg of ¹⁵N to each ¹⁵NO₃⁻ unit (3.1 g NH₄NO₃ per unit), yielding an N concentration of 146 mg L⁻¹ (comparable to the standard refill solution). To distribute the ¹⁵N label in the hydroponic system, drip fertigation was run continuously for 30 min after adding the ¹⁵N labeled nutrient solution to the experimental units. After 4 h, a first sampling to determine the ¹⁵N enrichment in plant, nutrient solution and gas samples took place. The sampling was done analogously to the natural abundance sampling, including the determination of gas flux rates and the collection of gas samples for isotopic analyses as well as leaf and nutrient solution samples. Following the same scheme, the last sampling took place 24 h after the labeling. This time, also samples from the tomato stems, roots and fruits were taken. From the middle of the tomato plant ca., 10 cm of the stem was cut. Around 0.5 g of fresh roots was sampled from the interface of rock wool cubes and rock wool slabs, where a dense root net allowed to obtain root material without rock wool fibers. Root samples were washed in de-ionized water and dried with lint-free cellulose wipes to remove the ¹⁵N label from adhering nutrient solution. During the second sampling campaign, each three green fruits from different positions (top, mid, and bottom) of one plant per experimental unit were sampled. All plant samples were dried for a minimum of 48 h at 80°C before later processing for analysis. Different plants were used for obtaining plant material before labeling, 4 h after labeling, and 24 h after labeling in order to minimize sampling effects on ¹⁵N uptake. Gas samplings for stable isotope analysis always took place on the rock wool slab in the middle of each experimental unit, from which plant samples were taken only after the last gas sampling (24 h after labeling). On the unplanted rock wool slabs, additional gas flux measurements took place shortly before the 24 h sampling to determine the N₂O emission potential from re-circulated nutrient solution with leachate and therein contained organic carbon.

The concentrations of NO₃⁻ and NH₄⁺ [mg N L⁻¹] were determined using flow injection analysis with photometric detection (FIAmodula; MLE GmbH, Dresden, Germany). Measurements of δ¹⁸O values in water samples were done by TC/EA coupled to a Delta V plus IRMS (Thermo Finnigan, Bremen, Germany) via a ConFlo IV interface. The δ¹⁵N values of NH₄⁺ and NO₃⁻ were determined according to Dyckmans et al. (2021) using a sample preparation unit for inorganic nitrogen (SPIN) coupled to a membrane inlet isotope ratio mass spectrometer (MIRMS; Delta plus; Thermo Finnigan) via a ConFlo III interface. Additional nutrient solution samples taken one day after the labeling were analyzed for their dissolved organic carbon content (DOC) using a liquiTOC analyzer (Elementar Analysensysteme GmbH, Langenselbold, Germany). Dried plant samples were transferred into 20 ml HDPE vials (Zinsser Analytic GmbH, Eschborn, Germany) and ground to a fine powder using a steel ball mill (MM400; RETSCH GmbH, Haan, Germany). Plant samples were analyzed for total N content (N_t) and their δ¹⁵N values using an Elemental Analyzer (EA) Flash 2000 (Thermo Fisher Scientific, Bremen, Germany), coupled with a Delta V isotope ratio mass spectrometer via a ConFlo IV interface (Thermo Fisher Scientific, Bremen, Germany). Data were normalized to the international scale for atmospheric nitrogen, by analysis of the international standards USGS40 and USGS41 (L-glutamic acid). Gas samples were analyzed for N₂O isotopocules (δ¹⁵N_N2O, δ¹⁸O_N2O) using a Delta V Isotope ratio mass spectrometer (Thermo Scientific, Bremen, Germany), coupled to an automatic preparation system with Precon plus Trace GC Isolink (Thermo Scientific, Bremen, Germany). In this setup, N₂O was pre-concentrated, separated, and purified, and afterward m/z 44, 45, and 46 of the intact N₂O⁺ ions as well as m/z 30 and 31 of NO⁺ fragment ions were determined (Lewicka-Szczebak et al., 2014). All measured delta values (δ) were expressed in permil (‰) deviation from the ¹⁵N/¹⁴N and ¹⁸O/¹⁶O ratios of the international reference standards (i.e., atmospheric N₂ and Vienna Standard Mean Ocean Water (VSMOW), respectively).

Data from the analysis of natural abundance gas samples were evaluated for δ¹⁵Nα (δ¹⁵N of the central N position of the N₂O molecule), δ¹⁵Nβ (δ¹⁵N of the peripheral N position of the N₂O), and δ¹⁸O according to Toyoda and Yoshida (1999) and Röckmann et al. (2003). The ¹⁵N site preference (δ¹⁵N^SP) was defined as the difference of δ¹⁵Nα and δ¹⁵Nβ. The δ¹⁸O values of N₂O depend on δ¹⁸O values of precursors, i.e., for denitrification to >80% on H₂O-O of the nutrient solution (Lewicka-Szczebak et al., 2016). Therefore, δ¹⁸O values of the emitted N₂O (δ¹⁸O_N2O) were corrected for the δ¹⁸O values measured in the de-ionized water (δ¹⁸O_H2O) and expressed as δ¹⁸O_N2O/H2O values:

In the case of nitrification, the δ¹⁸O_N2O values depend on atmospheric oxygen (O₂) as a precursor (Kool et al., 2007). In contrast to bulk δ¹⁵N_N2O, δ¹⁵N^SP is known to be independent from source processes. During chamber air sampling, the collected N₂O was a mixture of atmospheric and substrate-emitted N₂O. Thus, δ values of substrate-emitted N₂O were corrected using a basic isotope mixing model according to Well et al. (2006). To calculate the contribution of N₂O production pathways and N₂O reduction to N₂, the isotopocule mapping approach based on δ¹⁵N^SP_N2O and δ¹⁸O_N2O values was applied (Lewicka-Szczebak et al., 2017; Buchen et al., 2018). For the mapping approach, literature values for δ¹⁸O and δ¹⁵N^SP_N2O of bD, fD, nD, and Ni were used as proposed by Yu et al. (2020) and Lewicka-Szczebak et al. (2020). To account for differences in oxygen precursors between denitrification and Ni, the literature values for δ¹⁸O_N2O of bD, fD, and nD were adjusted by the addition of δ¹⁸O_H2O (Lewicka-Szczebak et al., 2020). Based on the sample position in the map, the contribution of bD and/or nD, Ni, and fD was calculated based on mixing equations, while the contribution of N₂O reduction to N₂ was calculated from the Rayleigh equation. All calculations were done as described in detail by Buchen et al. (2018) and Zaman et al. (2021) (Chapter 7: “Isotopic Techniques to Measure N₂O, N₂ and Their Sources). Two possible cases of N₂O mixing and reduction were assumed: (i) N₂O, which is produced by bD is first partially reduced to N₂, followed by mixing of the residual N₂O with N₂O from other pathways or (ii) N₂O produced by various pathways is first mixed and then reduced to N₂. A detailed description is given in the supplement of Wu et al. (2019). Five samples from sampling 1 and four samples from sampling 2 with a low fraction of substrate-derived N₂O were excluded from the data analyses because the uncertainty in substrate-derived δ values increases exponentially as sample and atmospheric N₂O concentrations converge. Similar to Buchen et al. (2018), a threshold was used for the minimum difference between sample and atmospheric N₂O concentrations, which was determined based on measured N₂O concentrations in ambient air during the sampling. For sampling 1, the threshold was 337 ppb and for sampling 2, it was 359 ppb (65 ppb above the ambient air N₂O concentration). This was supported by a Gaussian error propagation, with the threshold limiting the propagated errors of δ¹⁵N^SP_N2O and δ¹⁸O_N2O to <6‰ and < 5‰, respectively.

Data from the analysis of ¹⁵N-enriched gas samples were only evaluated for bulk δ¹⁵N_N2O. For further calculations, δ¹⁵N values were converted to atom-%_15N to express the ¹⁵N enrichment:

with R_STD being the isotopic ratio (¹⁵N/¹⁴N = 0.0036765) of atmospheric nitrogen. Calculations of the contributions of N₂O originating from the labeled and non-labeled pools were based on the non-equilibrium distribution of N₂O isotopocules, as described by Spott et al. (2006) and Bergsma et al. (2001). For labeling with ¹⁵NO₃⁻, this approach directly determines the ¹⁵N enrichment of the labeled N pool producing N₂O (ap_N2O) and the fraction of N₂O derived from that pool. Considering, the fraction of atmospheric N₂O in the samples, the fraction of NO₃⁻-derived N₂O in the emitted N₂O (f_PN2O) can be calculated. A detailed procedure is given in Deppe et al. (2017). However, due to the experimental setup, labeled N₂O could originate from two pools (NO₃⁻, NH₄⁺, or a mixture of both pools). Thus, for labeling with ¹⁵NH₄⁺, f_PN2O was estimated based on the ¹⁵N atom fraction of emitted N₂O (¹⁵a_N2O) using a mixing equation:

with ¹⁵aNO₃⁻ being the ¹⁵N enrichment of the NO₃⁻ pool and ¹⁵aNH₄⁺ being the ¹⁵N enrichment of the NH₄⁺ pool (cf. Eq. 2). The N₂O flux from the NO₃⁻ pool (NO₃⁻-derived N₂O) was calculated from f_PN2O by ordinary linear regression using the measured N₂O concentrations at t0 and after 1 h of chamber closure to determine the total N₂O flux (total N₂O), assuming that the increase in the N₂O emitted from the ¹⁵N-labeled pool was also linear as shown for the emission of total N₂O (Buchen et al., 2016). The N₂O flux from the NH₄⁺ pool (NH₄⁺-derived N₂O) was calculated analogously based on the fraction of NH₄⁺-derived N₂O in the emitted N₂O (f_NH4), which was deduced from f_PN2O (f_NH4 = 1 – f_PN2O). Thus, the NH₄⁺-derived N₂O was calculated as the difference between total N₂O and NO₃⁻-derived N₂O.

To determine the amount of ¹⁵N tracer, which was recovered in the different pools 4 and 24 h after the labeling (excess ¹⁵N), atom-%_15N values were used to calculate atom-% ¹⁵N excess (APE):

with atom-%_15N,labeled being the atom-%_15N values of labeled samples and atom-%_{15N,natural abundance} being the atom-%_15N values of natural abundance samples. Afterward, excess ¹⁵N [mg ¹⁵N unit⁻¹] for each pool was calculated:

with N_pool being the N amount in each pool [mg N unit⁻¹] at the time of sampling (4/24 h after labeling). The N_pool values for plant biomass were calculated by multiplying the measured dry weight [g] of shoots (leaves + stems), roots and fruits per unit with their N_t content [g N g_{dry weight}⁻¹]. The N_pool values for NO₃⁻-N and NH₄⁺-N from the nutrient solution were calculated by multiplying the measured N concentrations [mg N L⁻¹] with the total volume of nutrient solution per unit [L]. The latter was a mixture of nutrient solution added for the labeling and remaining (unlabeled) nutrient solution in the rock wool substrate. The total volume of the nutrient solution was estimated based on the dilution of NH₄⁺-N concentrations from the labeled nutrient solution (73 mg N L⁻¹ in 15 l) at the 4 h sampling point, assuming that NH₄⁺-N concentrations in the unlabeled nutrient solutions were negligible (measured concentrations in natural abundance samples <2.5 mg N L⁻¹ at first sampling campaign and <7 mg N L⁻¹ at second sampling campaign) and that the N_t content as well as composition in the mixed nutrient solution did not substantially change during the 4 h. For the calculation of excess ¹⁵N, two neighboring units were excluded from the second sampling campaign, because of a spillover of labeled nutrient solution between these units. The N_pool values for N₂O were calculated from the measured gas flux rates [mg N h⁻¹] of planted and unplanted rock wool slabs. For the planted rock wool slabs, cumulative N₂O emissions [mg N] were calculated by linear integration between the natural abundance (0 h), 4 h, and 24 h samplings, and summation of hourly gas fluxes. For unplanted rock wool slabs, constant N₂O emission rates were assumed and used to calculate cumulative N₂O emissions, as they were not affected by plant activity. For calculating the N_pool value per unit, cumulative N₂O emissions from planted rock wool slabs were multiplied by 2 (two planted slabs per unit) and the cumulative N₂O emissions from unplanted slabs (one per unit) were added. Finally, the excess ¹⁵N values from the different pools were summed up to obtain the total amount of ¹⁵N recovered from the labeling (¹⁵N_total) and the ¹⁵N recovery rate [%] was calculated:

with ¹⁵N_label being the amount of ¹⁵N tracer [mg ¹⁵N unit⁻¹] added during the labeling.

All statistical analyses were done using the R software (version 4.2.0). Linear mixed-effects models were done using the R package ‘lme4’ (version 1.1–29), including the effects of individual hydroponic units as random intercept. Post-hoc tests on linear mixed-effects models were done using the R package “emmeans” (version 1.7.4–1), applying the Holm-Bonferroni correction method for multiple comparisons. If necessary, data were log- or square root-transformed prior to analysis to fulfill the requirements of normality and variance homogeneity.

The N₂O flux measurements from this study are summarized in Table 1. In general, all fluxes were in the same range, except for the measurement 24 h after labeling during the first sampling, which was significantly (p < 0.05) higher than the other measurements. There was no significant difference between planted and unplanted rock wool slabs from the same sampling campaign. The trend to higher N₂O emissions from unplanted substrate during sampling 2 was reflected by higher DOC contents in the nutrient solution compared to sampling 1 (Table 1).

Results from isotopic analyses of N₂O are shown in Figure 1 as a δ¹⁵N^SP_N2O/δ¹⁸O_N2O map. The δ values from both samplings clearly scatter around the reduction line of N₂O derived from bD, indicating that either bD or nD or a mixture of both was the main source of N₂O. Moreover, the increased δ¹⁵N^SP_N2O and δ¹⁸O_N2O values compared to the literature value for bD indicate that a high share of N₂O was reduced before emitted to the atmosphere. Altogether, the differences in isotopic results between the first and the second sampling campaign were negligible (Table 2). Depending on which scenario (mixing of bD and fD or bD and Ni) and case (first reduction than mixing or first mixing than reduction) was assumed, the fraction of bD varied between 0.85 and 0.90, while the N₂O/(N₂O + N₂) ratio of bD (r_N2O) varied between 0.08 and 0.14. In consequence, the calculated N₂ fluxes were between six to ten times higher than the measured N₂O fluxes.

Although the same amounts of NO₃⁻-N and NH₄⁺-N were added in the form of NH₄NO₃ during each ¹⁵N labeling, NO₃- concentrations were clearly higher than NH₄+ concentrations in the nutrient solution after labeling (Table 3). This indicated that a significant amount of unlabeled nutrient solution with a high NO₃⁻ to NH₄⁺ ratio was still present in the rock wool substrate during ¹⁵N labeling. Regardless of the higher dilution of ¹⁵NO₃⁻ label (Table 3; Supplementary Figure S6), the ¹⁵N tracer could be detected in the emitted N₂O independent of the applied form (¹⁵NH₄⁺ or ¹⁵NO₃⁻). The ¹⁵a_N2O values mirrored the ¹⁵N enrichments of the labeled NO₃⁻ and NH₄⁺ pools, with higher values in of ¹⁵NH₄⁺-labeled units compared to ¹⁵NO₃⁻-labeled units (Supplementary Figure S6). The label dilution was considered for calculating NO₃⁻-derived N₂O and NH₄⁺-derived N₂O. The NO₃⁻-derived N₂O (Figures 2A,B) reflected the N₂O emission rates measured by GC (Table 1), with highest values found 24 h after the first labeling. There was no clear difference in NO₃⁻-derived N₂O between the ¹⁵NH₄⁺ and ¹⁵NO₃⁻ labels. In general, the NH₄⁺-derived N₂O values (Figures 2C,D) were lower than the NO₃⁻-derived N₂O values, but also followed the dynamics of N₂O emission rates measured by GC. Notably, NH₄⁺-derived N₂O was higher for ¹⁵NO₃⁻-labeled units compared to ¹⁵NH₄⁺-labeled units during sampling 2. Consequently, the calculated average f_PN2O values varied from 0.4 to 0.9 between the applied label forms, sampling times, and sampling campaigns (Figures 2C,D). During both sampling campaigns, an increase of f_PN2O from 4 h to 24 h after labeling was present for the ¹⁵NO₃⁻-labeled units, while there was no effect of sampling time for the ¹⁵NH₄⁺-labeled units. The latter showed higher f_PN2O values during the second sampling campaign, which was also significantly higher than for the ¹⁵NO₃⁻-labeled units at 4 h after labeling.

The natural abundance δ¹⁵N values from both samplings were equal (leaves) or slightly lower (NH₄⁺, NO₃⁻ and N₂O) at the second sampling, indicating that no carryover of ¹⁵N label occurred from the first sampling. The amount of ¹⁵N tracer from the ¹⁵N-enriched NH₄NO₃ added during the labelings that was recovered in different pools (dissolved NH₄⁺ and NO₃⁻, N₂O, plant biomass) was calculated as excess ¹⁵N (¹⁵N_exc). At both samplings, most of the ¹⁵N label remained in its original form after 24 h, i.e., as dissolved NH₄⁺ and NO₃⁻ (Table 4). There was a notable increase of ¹⁵N_exc of dissolved NO₃⁻ in the ¹⁵NH₄⁺-labeled units, indicating the conversion of NH₄⁺ to NO₃⁻ by Ni (up to 2% of added label during sampling 1). On the other side, the ¹⁵N_exc of dissolved NH₄⁺ in the ¹⁵NO₃⁻-labeled units was comparably low (at maximum 0.3% of added label during sampling 1). The ¹⁵N_exc of N₂O strongly differed between the two samplings, with up to 20 times higher values at sampling 1, reflecting the APE values of N₂O (Supplementary Figure S7). Despite the higher dilution of ¹⁵N tracer in the NO₃⁻ pool (Table 3) and the resulting lower ¹⁵N enrichments in the ¹⁵NO₃⁻-labeled units compared to ¹⁵NH₄⁺-labeled units (Supplementary Figure S6), there were no significant differences between the label types regarding the amount of ¹⁵N tracer found in N₂O, as shown by the ¹⁵N_exc values (Table 4). In all cases, the ¹⁵N_exc of total plant biomass was higher than the ¹⁵N_exc of N₂O. The highest plant ¹⁵N uptake was observed during the second sampling in ¹⁵NH₄⁺-labeled units. Irrespective of the generally higher ¹⁵N-enrichment of roots (Supplementary Table S2), most ¹⁵N tracer was found in shoots (i.e., the sum of stem leaf biomass; Table 4), as a consequence of the biomass difference (root to shoot ratio of 0.23). Only marginal amounts of ¹⁵N tracer were found in tomato fruits during sampling 2. Overall, the majority of ¹⁵N added during labelings was recovered in the studied pools, with the calculated ¹⁵N recovery rates varying around 100%.