Nitrogen transport in a tundra landscape: the effects of early and late growing season lateral N inputs on arctic soil and plant N pools and N2O fluxes

Understanding N budgets of tundra ecosystems is crucial for projecting future changes in plant community composition, greenhouse gas balances and soil N stocks. Winter warming can lead to higher tundra winter nitrogen (N) mineralization rates, while summer warming may increase both growing season N mineralization and plant N demand. The undulating tundra landscape is inter-connected through water and solute movement on top of and within near-surface soil, but the importance of lateral N fluxes for tundra N budgets is not well known. We studied the size of lateral N fluxes and the fate of lateral N input in the snowmelt period with a shallow thaw layer, and in the late growing season with a deeper thaw layer. We used 15N to trace inorganic lateral N movement in a Low-arctic mesic tundra heath slope in West Greenland and to quantify the fate of N in the receiving area. We found that half of the early-season lateral N input was retained by the receiving ecosystem, whereas half was transported downslope. Plants appear as poor utilizers of early-season N, indicating that higher winter N mineralization may influence plant growth and carbon (C) sequestration less than expected. Still, evergreen plants were better at utilizing early-season N, highlighting how changes in N availability may impact plant community composition. In contrast, later growing season lateral N input was deeper and offered an advantage to deeper-rooted deciduous plants. The measurements suggest that N input driven by future warming at the study site will have no significant impact on the overall N2O emissions. Our work underlines how tundra ecosystem N allocation, C budgets and plant community composition vary in their response to lateral N inputs, which may help us understand future responses in a warmer Arctic.

Abstract Understanding N budgets of tundra ecosystems is crucial for projecting future changes in plant community composition, greenhouse gas balances and soil N stocks. Winter warming can lead to higher tundra winter nitrogen (N) mineralization rates, while summer warming may increase both growing season N mineralization and plant N demand. The undulating tundra landscape is inter-connected through water and solute movement on top of and within near-surface soil, but the importance of lateral N fluxes for tundra N budgets is not well known. We studied the size of lateral N fluxes and the fate of lateral N input in the snowmelt period with a shallow thaw layer, and in the late growing season with a deeper thaw layer. We used 15 N to trace inorganic lateral N movement in a Low-arctic mesic tundra heath slope in West Greenland and to quantify the fate of N in the receiving area. We found that half of the early-season lateral N input was retained by the receiving ecosystem, whereas half was transported downslope. Plants appear as poor utilizers of earlyseason N, indicating that higher winter N mineralization may influence plant growth and carbon (C) sequestration less than expected. Still, evergreen plants were better at utilizing early-season N, highlighting

Introduction
Arctic ecosystems are often limited by the availability of nutrients (Marion and Miller 1982;Sistla et al. 2012;Wild et al. 2013) -especially nitrogen (N) and phosphorus (P). While most N turnover occurs during Arctic summer, winter N mineralization may provide an important N source upon thaw in spring, when a pulse of water-soluble nitrate (NO 3 -) in soil solution has been observed (Schimel et al. 2004;Buckeridge et al. 2013;Rasmussen et al. 2020), resulting in plant N uptake (Semenchuk et al. 2015).
A growing body of research has focused on Arctic soil N turnover at plot scales, i.e. in one dimension (e.g. Schimel et al. 2004;Sistla and Schimel 2013;Semenchuck et al. 2015;Rasmussen et al. 2020). However, during the spring pulse of N in solution, the shallow thaw layer can be infiltrated by meltwater from snow melt (Slater et al. 2007), and consequently, the spring N pulse can be transported and utilized biologically downslope (Giblin et al. 1991;Yano et al. 2010).
Such coupled transfer of N and water flow due to landscape hydrological characteristics and N cycling may lead to N-loss in one part of the landscape, but N input to other parts, and may be relevant especially at the beginning of the snow-free season, when plant demand for N is limited and the internal turnover is relatively slow (Bilbrough et al. 2000). The redistribution of N across the landscape can be important for understanding the spatial variability of N cycles in the Arctic landscapes. The availability of N impacts vegetation community composition and plant growth (Bobbink et al. 2010;Liu et al. 2018), N competition with microbes (Sistla et al. 2012) as well as ecosystem N losses of N 2 O and N 2 (Voigt et al. 2020). Because Arctic ecosystems have only little external N input (Rousk et al. 2017), a naturally occurring, unaccounted for, N input to downslope ''receiving'' ecosystems after thaw, would therefore be an important yet not quantified N source into lower-lying Arctic ecosystems that should be taken into account in the future. With future Arctic winter warming (IPCC 2019), and corresponding increase in winter N mineralization (Rustad et al. 2001;Weintraub and Schimel 2003;Salmon et al. 2018), summer belowground N cycling (Zhang et al. 2015), and changes in soil moisture content, microbial community structure and litter quality (Rinnan et al. 2008;Rousk et al. 2018;Rasmussen et al. 2020;Salazar et al. 2020), the importance of N redistribution in the landscape could increase. If the lateral N input is an important source for downslope ecosystems, this additional N could alleviate plant N limitation, thus increase tundra C uptake (Weintraub & Schimel 2005), however, it could also be substrate for increased N 2 O emissions (Voigt et al. 2017a, b), depending on N forms and redox conditions. In N-limited ecosystems, however, N might not move far before being re-incorporated into the ecosystem, making laterally transported N a negligible source for downslope vegetation or soil (Rastetter et al. 2004). With the highest root density and most microbial biomass concentrated in the upper layers of the soil (Ouhlele et al. 2016;Xu et al. 2013), the retention of lateral N could be highest in this horizon. With differences in rooting depth and N access between different plant species ), future increased lateral N transport in the later season could offer comparative advantages to some species. The importance of lateral N inputs following early spring thaw and throughout the growing season for plant growth, soil N stocks or N 2 O/N 2 fluxes is thus necessary to quantify in order to understand present and future downslope ecosystem N dynamics, but is yet unclear.
The aim of this study is to investigate the importance of lateral N flows from upslope to downslope, receiving ecosystems, and to quantify which parts of the receiving ecosystems that benefit most from the lateral N input. We do this by conducting a field experiment along a naturally vegetated arctic hillslope, where we use a 15 N tracer experiment to quantify N pathways and address the following questions: Does NO 3 --N released upon thaw move downslope through the ecosystem, or is it retained, making lateral transport a negligible N source-and how does this depend on the season (i.e. thaw depth)? If retained, will plants benefit from the lateral NO 3 --N input in the receiving area? Will the early season mineral N pulse and transport fuel a pulse of N 2 O emissions?
We hypothesize that (1) a major part of laterally transported NO 3 is utilized by the downslope ecosystem, and the amount of N transported further downslope depends on seasonal variations in lateral water flow related to snowmelt and plant demand/ uptake along flow paths, (2) more laterally transported NO 3 is retained in the early season, when the lateral input is in near-surface layers with greater root density and microbial biomass compared to the later season when lateral N input occurs further down in the soil profile, (3) the input of lateral N is small compared to internal N turnover in the system, but is an important source for plants in the early season, and (4) the lateral N input in the early season with shallow thaw layer fuels an early-season pulse of N 2 O production and gaseous N loss from the ecosystem.

Study site
Disko Island is located off the coast of Western Greenland. The southern part of the island consists of Early Tertiary basaltic plateau mountains cut through by ancient glaciers forming U-shaped valleys such as the study site valley Blaesedalen (69°18 0 40.9 00 N; 53°30 0 40.9 00 W) (Humlum et al. 1976).
The climate in Blaesedalen is Arctic Maritime with mean annual air temperatures of -3 ± 1.8°C  and an increase of 0.16°C y -1 in the same period (Zhang et al. 2019). The average precipitation (1991-2017) was 418 ± 131 mm y -1 , with about 40% falling as snow (Hansen et al. 2006;Zhang et al. 2019).
The area is affected by permafrost with active layer depths largely related to drainage conditions. On the well-drained tundra slopes, maximum active layer depths reach 2-3 m, whereas the water-saturated fen areas have maximum thaw depths of 40-80 cm (D'Imperio et al. 2017). On the eastern slope with aspect towards the west, a snow accumulation area supplies meltwater from a semi-permanent snow fan throughout most of the summer. The slope is made up of colluvial stones to boulders, intersected by alluvial and nival deposits and with a soil depth of 40-90 cm.
In 2014, a study site was established on the slope, covering an area of 35 9 50 m. The slope site represents a vegetation gradient from top to bottom, with gradually different environmental conditions. The growing season is longest at the footslope ([ 10 weeks) and shorter upslope as a snow drift melts back. Some years, the snow drift persist the entire summer, some years it is gone by mid-July. This affects the vegetation types along the slope; with upslope dominated by Salix herbacea and downslope featuring dwarf and low shrubs (e.g. Salix glauca, Betula nana, Cassiope tetragona and Empetrum nigrum). The shrubs are interlaid with mesic tundra mosses such as Tomentypnum nitens, Racomitrium lanuginosum and Sphagnum sp. The active layer development is delayed upslope with later snowmelt, causing generally lower temperatures and higher soil moisture upslope in the early thaw season. In the peak and later parts of the growing season, snow meltwater drains downslope through deeper active layers, restricted by underlying frozen ground. The resulting conditions are generally warmer and drier upslope compared to downslope. Water table depth at the footslope moves downward with the thawing of the active layer and fluctuates over the growing season from * 50-100 cm depending on evaporation and precipitation.

Experimental design
At the foot of the slope site, five plots were established in June 2018 for monitoring of the physical and chemical ambient conditions (Fig. 1). The plots were positioned on a line perpendicular to the slope on mesic tundra heath with 2-5 m apart and had similar vegetation community (Fig. 1A). The plots thus represent the same position on the slope based on the vegetation composition. The study site receives meltwater from the above-lying snow drift area and is henceforth referred to as the Reception area. Water from the study site drains further into a lower-lying fen area 100 m further away.
In July 2018, plots ( Fig. 1B) were equipped with sensors for hourly measurements of volumetric soil moisture and temperature at 2, 10 and 20 cm depth, in one plot also at 40 and 60 cm depth (Tinytag, Gemini Datalogger, UK; Decagon Devices, METER Group, USA).
Soil water suction cups (Prenart Equipment Aps, DK) were installed in all plots at 10-20 and 20-30 cm depth and water extracted twice a week during the growing seasons 2018 and 2019. Water was analyzed for Total Organic Carbon (TOC) (Shimadzu TOC analyzer; Kyoto, Japan), NH 4 ? -N, NO 3 --N and total dissolved nitrogen (TDN) (FIAstar 5000; Höganäs, Sweden). Soil moisture may affect the volume from which water is extracted, but the method is regarded a reasonable measure of soil water solute concentration in natural ecosystems (Singh et al. 2018). In immediate vicinity of the plots, a HOBO-20 pressure transducer (Onset Computer Corporation, MA, USA) measured water table depth during the thaw season.
Soil cores from 0-30 cm taken near the plots were split 0-10, 10-20 and 20-30 cm and analyzed for grain size distribution, total C and N content (Flash 2000, Thermo Scientific, Bremen, Germany), root biomass, root C and N, gravimetric water content (GWC%), and microbial C and N using chloroformfumigation direct-extraction (Brookes et al. 1985). Aboveground vegetation was sampled in a 10 9 20 cm area in each plot; leaves and stems were ground and analyzed for total C and N.
In all plots, the surface exchange of CO 2 and N 2 O was measured weekly from late June to late August using static chambers (Ambus et al. 1993). Chambers with fans to ensure air mixing and temperature loggers were placed on pre-installed frames in the plots for three hours, and every 45 min, an air sample was withdrawn and sent for analysis for content of trace gases (Agilent 7890A, CA, USA). On 30th June and 4th August 2018, six top soil samples (4 cm depth, 5.5 cm diam.) were taken adjacent to the plots, wrapped in plastic foil and bags, and frozen during transport. The N 2 O and N 2 emissions were determined at in-situ water content at the Austrian Research Centre for Forests in Vienna using the gas-flow-core method, where gas exchange is measured in the headspace atmosphere, where N 2 has been replaced prior to the accumulation phase with He/O 2 , such as described in Butterbach-Bahl et al. (2002).

N tracer experiment
In order to test the fate of NO 3 moving laterally, ten plots were established downstream of the Reception area plots (a view of one from above in Fig. 1C). On 3rd of July, when active layer had reached a depth of 30 cm, and 3rd of August with an active layer depth of 90 cm, a 15 NO 3 solution was applied on to the top of the frozen surface in a line perpendicular to the slope above the tracer experiment plots so that the solution would move downslope and into the plot with the lateral water input on the frozen surface (crosssectional view (A) and conceptual view from above (B) in Fig. 2). The solution consisted of KNO 3 -N (99% 15 N) dissolved in 1.1 L H 2 O (0.085 g N L -1 ), corresponding to 0.15 g N m -2 . A 110 mL solution was injected through a thin hollow rod at the top of the frozen surface for each 10 cm in a line of 125 cm perpendicular to the slope for each of the five replicates (Fig. 2B). Due to the low concentrations, the added N is not considered to have any fertilization effect, but is used only as a highly enriched tracer. To investigate the fate of the added 15 N, soil samples 0-30 cm were taken and vegetation was harvested in an area of 10 9 20 cm on day 1, 3, 7 and 25 after injection, and the following year on the day with thaw depth corresponding to the thaw depth at injection (Fig. 2). The horizontal distance between injection and upper limit of sampling areas was 5-10 cm.
The following procedure was adopted and the following samples were obtained inside the plot and analyzed for % C, % N and 15 N recovery using elemental analysis (Flash 2000, Thermo Scientific, Bremen, Germany) coupled to an isotope ratio mass spectrometer (Thermo Delta V Advantage IRMS, Thermo Scientific, Bremen, Germany). Soil samples were split in depths 0-10, 10-20 and 20-30 cm. As many roots as were feasible were manually removed from the samples, which were subsequently homogenized and analyzed for grain size distribution, gravimetric water content, bulk C and N and d 15 N of the bulk soil N. Soil bulk density was measured on a subset of samples. Furthermore, subsamples were obtained and analyzed for microbial C, N and 15 N recovery using an Eurovector elemental analyzer (Eurovector, Pavia, Italy) coupled to an Isoprime isotope ratio mass spectrometer (Elementar, Cheadle Hulme, UK), by comparing water extractable N and C with replicates subject to chloroform fumigation (Brookes et al. 1985) for 24 h prior to extraction. Root dry weight was quantified, and the roots were crushed and analyzed for total C, N and d 15 N. In an area of 10 9 20 cm (Fig. 2), all aboveground biomass was harvested and separated based on plant species before air drying. Leaf and stem was separated, and each fraction was crushed and analyzed for % C, % N and d 15 N. The ratio of the occurrence of stable isotopes 15 N/ 14 N compared to the reference 15 N/ 14 N ratio in Air-N 2 is expressed as the d 15 N value. The d 15 N value of all sample materials was calculated as: In order to achieve the recovery, the measured d 15 N relative to the amount of N g -1 in the background samples was converted to atom percentage of 15 N and subtracted from the measured amount in the experiment samples, and all pools were scaled to plot size using dry soil bulk density and the known size of the sampling areas. The scaled amount of 15 N was compared to the known added 15 N to obtain recovery in each pool, respectively, and the total recovery in percentage of added 15 N. Figure 3 shows the N pools and budget of the measured variables in the Reception area in 2018. The largest pool was total soil N (including humus), followed, orders of magnitude below, by microbial N. Smaller yet were the pools of root, stem and leaf N, and two orders of magnitude smaller again were the average pools of dissolved N in soil solution over the summer season 2018. N 2 O fluxes varied from -1.6 9 10 -4 to 2.4 9 10 -4 g N m -2 d -1 with a mean of 8.0 9 10 -7 g N m -2 d -1 . The measured ratio of N 2 / N 2 O was 16, and the estimated average N 2 loss was 4.5 9 10 -6 g N m -2 d -1 .

Measured C and N pools
Pool sizes varied slightly over the season. Pools of C and N at each specific sampling date, and divided into depth intervals, are specified in Tables S1-4 as well as C/N ratios of leaf and stem of the aboveground vegetation.

Soil physics and chemistry 2018 and 2019
The year 2019 had an early snowmelt (50% of the Reception area plots were snow free on 15th of May), which resulted in higher summer soil temperatures and lower soil moisture levels in 2019 than on the same day of the year in 2018. The summer 2018 had a late snowmelt (50% snow free 13th of June 2018). Soil moisture varied from almost 0% vol. when the soil was frozen in winter to 50% vol. after snowmelt, with growing season averages in the top 30 cm of 35 vol. %, whereas soil temperatures varied from -7.4°C in winter to 13°C in the upper 15 cm in mid-summer. Figure S1 shows the soil moisture levels and soil temperatures to 60 cm in the Reception area throughout this period. Soil saturated vertical hydraulic conductivity was 127 ± 50 cm d -1 . Horizontal hydraulic conductivity was smaller with an average of 18 ± 4 cm d -1 .
Soil water NO 3 --N and NH 4 ? -N concentrations varied over the summer seasons 2018 and 2019, with maximum concentrations immediately after soil thaw and minimum concentrations in the peak growing season (Fig. 4). The mean of five replicate NO 3 --N concentrations spanned from 3-19.2 lg L -1 (10-20 cm) and 2.2-12.2 lg L -1 (20-30 cm) in 2018, whereas NH 4 ? -N varied more and spanned from 3.7-17.5 lg L -1 (10-20 cm) and 4.0-8.2 lg L -1 (20-30 cm) over the 2018 season. In 2019, soil water collection began as soon as liquid water was present in 10-20 cm depth and therefore has more focus on the early growing season and less detail in the mid-growing season. Mean NO 3 --N concentrations in 2019 (N = 5) varied from 0.46 to 19.5 lg L -1 (10-20 cm) and below detection to . Pool units are g N m -2 to a depth of 100 cm. Root N is estimated based on root C with C/N ratio of 28 such as observed by e.g. Sullivan et al (2007). Flux units are g N m -2 d -1 . N-fixation is estimated from literature and scaled down from kg h -1 y -1 (Hobara et al. 2006;Rousk et al. 2017) 13.5 lg L -1 (20-30 cm). Mean NH 4 ? -N varied from 8.2 to 19 lg L -1 (10-20 cm) and below detection to 29 lg L -1 (20-30 cm).
TDN concentrations reached a maximum of 210 lg L -1 in 10-20 cm depth during late June of 2019. A similar, but smaller, peak was observed at further depth (Fig. 4). In 2018, the peak of 220 lg L -1 (10-20 cm depth) and 117 lg L -1 (20-30 cm depth) occurred in mid-July and then decreased to about half in mid-August. In both depths, 2019 had a higher with an average of 0.01 mg N d -1 m -2 as net balance (Fig. 5). The numbers reported here were used for Fig. 3, but are broken down into seasonally specific fluxes.
Fluxes were generally stable over the season, varying insignificantly around net 0 mg N d -1 m -2 .
The incubated topsoil (0-4 cm) samples had on average small N 2 O fluxes, in the order of magnitude comparable to in situ measurements (Table 1). N 2 emissions were one to two orders of magnitude higher, with the largest N 2 /N 2 O ratio in the early season.
Scaled to the whole growing season (here calculated as 15th of June-31st of August), the total gaseous N loss from the soil in one plot was 0.57 ± 0.23 g N m -2 (Table 1).

N recovery
In total appr. 50% of the added 15 N in July (30 cm) was recovered in soil and vegetation (Fig. 6A), whereas 50% likely leached through the profile and downslope. On day 1 after injection, microbes and bulk soil (excluding microbes) had recovered 9.4% and 28.2%, respectively, while less than 1% was recovered in roots, leaves and stem. On day 3, 7 and 25, bulk soil and microbial 15 N recovery were still high, but a minor, but significant, fraction of the added N was also recovered in vegetation: Leaf dominated with 2.4% recovery on day 25, where roots and stem 15 N recovery amounted to 0.6% and 0.4%, respectively. One year after addition, roots still held just below 1% of added N, whereas leaf and stem held  Reception area 04-08-2018 55.4 ± 6.8 0.106 ± 0.041 0.02 ± 0.01 5.7 ± 3.1 0.6% and 0.4%, respectively. 18.7% was still present in the soil.
At sites with addition in 90 cm depth in August (Fig. 6B), a maximum of 35% of added 15 N was recovered on day 7 after addition. Bulk soil and microbes were the dominating retainers of added 15 N, with 3.3% and 18% recovery on day 1, respectively. On Day 3 and 7, the bulk soil recovered 10 and 30%, respectively. Roots recovered less than 1% of added 15 N, but among the vegetation pools, roots contained most of the added N. A year after addition, only 6.4% was present in total, of which 5% was in the soil.
The small plant N recovery was split between species falling within the three functional plant type groups: Evergreen shrubs, deciduous shrubs and mosses ? lichen (pooled because of their shared role as potential association with N 2 -fixers). In the early thaw season, the evergreen shrubs gained most lateral 15 N input during the first days and generally benefitted the most from lateral N (Fig. 6C). However, from day 7, deciduous shrubs and moss-lichen had incorporated similar amounts of the lateral N into their tissue. After one year, the evergreens still contained more than half of the early season lateral N input.
In the later growing season with a 90 cm active layer (Fig. 6D), deciduous plants dominated the (small) lateral N uptake during the first day, and although the evergreen shrubs at day 7 retained more of the lateral moving N, in the long term (1 year), deciduous shrubs were the main benefactors of the late season lateral N input.
The full dataset of N recovery is shown in Table S5.

Discussion
To what extent is lateral N flow retained or transported away-and does the vegetation benefit from the N?

Ambient soil water chemistry
The low content of inorganic N in soil solution at the study site compared to e.g. temperate heathlands (Andresen et al. 2010) is in line with the general understanding of N-limited tundra heath ecosystems (Fig. 4). The levels are comparable to previously reported inorganic N values at other sites in the Blaesedalen (Rasmussen et al. 2020) and with measurements made in soils from High arctic Greenland during peak growing season (Lopez-Blanco et al. 2020). The variations in soil water N concentrations over the growing season agree with observations of early season inorganic N release in response to thawing soil, making N from winter mineralization biologically available (Buckeridge et al. 2010;Larsen et al. 2012;Semenchuk et al. 2015). The peak in inorganic N is subsequently followed by a peak in TDN, but no peak in N 2 O emissions, which suggests incorporation of inorganic N into microbial biomass and not direct denitrification. TOC was significantly higher in 2019 than in 2018, which is likely due to the warmer summer and earlier snowmelt of 2019, with consequently higher mineralization rates, and possibly more root exudates as result of deeper snow (Rasmussen et al. 2020).

N 2 O emissions and gaseous N loss
No peak in N 2 O emissions was observed at times with lateral N input in the early thaw and late growing season, i.e. following peaks in mineral soil solution N, and where the soil took up much lateral N. This is in line with Kolstad et al. (2021) from the same site, but contrary to e.g. Buckeridge et al. (2009) and our hypothesis (4) that lateral N input in the early season would fuel an early-season pulse of N 2 O. The results suggest that other limitations inhibited N 2 O production, or, in line with the results, that a strong microbial sink immobilized N.
However, the N 2 /N 2 O ratios offer an alternative explanation for the results obtained here. The ratio of 16.8 ± 7.4 in the early season compared to 5.7 ± 3.1 in the late season suggests a more complete denitrification or less nitrification-sourced N 2 O emissions in the early thaw period (Butterbach-Bahl et al. 2013). The fact that soil water content was higher in the early thaw season with snowmelt water moving through the shallow thaw layer, supports the explanation that more lateral N input and mineral N released in situ may have ended as N 2 in the early season, resulting in little effect on N 2 O fluxes of early-season lateral N input.
Shallow active layer: half of released N is retained, mostly within the soil A 50% 15 N recovery of NO 3 --N added laterally in the early season shows that some N, when moving on top of the frozen surface in a shallow active layer, is taken up immediately by the downslope ecosystem, supporting our hypothesis 1, which stated that a significant proportion of lateral N input would be utilized directly by the downslope ecosystem. Most of the N is retained as microbial biomass and soil organic compounds. This is aligned with several other N cycling studies based on stable isotopes within soil profiles (Sørensen et al. 2008;Larsen et al. 2012;Zhu et al. 2016). Lateral N input in the early season thus follows a pattern similar to N from deposition or mineralized within the plot-fast microbial uptake and incorporation into organic matter, followed by mineralization of a fraction and movement into plant biomass. In contrast to expectations in hypothesis 3, only a minor fraction of early-season added N was, however, accumulated in plant biomass after 25 days, which can be explained by a limited active root biomass (Oulehle et al. 2016), but it may also imply that microbes are better than plants at utilizing NO 3 at this time of the year ). Additionally, simple physics of root acquisition may decrease recovery when water moves through the root zone at a flow rate of up to 127 cm d -1 during snowmelt than if water was pooling for longer (Yano et al. 2010). Combined, microbial competition and hydrology of the Arctic tundra may explain the small plant recovery of lateral N-input. A year after lateral input, however, N is still retained in soil and vegetation, suggesting that lateral N input is a natural part of the ecosystem N budget.
Half of released N is transported further downslope With 50% of laterally transported N recovered in the early growing season, 50% was not recovered and therefore potentially drained further downslope or degassed. Low levels of measured N 2 O and N 2 fluxes (Fig. 3) suggest that most of the N not recovered drained out of the profile and thus offered potential for biological uptake further downslope. This is consistent with Giblin et al. (1991) and Yano et al. (2010), who concluded that mineralization rates within an Arctic permafrost-affected lowland were too low to explain soil water plant-available N concentrations, and that nutrients therefore must be transported into the lowlands from surrounding areas. It is in apparent contrast to the Rastetter et al. (2004) model calculations, which estimated that an N atom moved only on average 26 cm y -1 across an Arctic tussock tundra slope, however, the lateral N input may have been picked up further downslope by roots or microbes, and exactly how far the lateral N input moved cannot be quantified from this study. Nevertheless, our study suggests that early season lateral N flow is likely a naturally occurring input, which can explain phenomena such as the higher nutrient content found in a High arctic tundra surrounded by cation and nutrient-rich rock and sediment compared to a Low arctic tundra surrounded by more silicate rock and poor soils (Lopez-Blanco et al. 2020), or the lack of seasonal patterns in soil water inorganic N content found in a tundra fen with through-flow of drainage water (Rasmussen et al. 2020).
Deep active layer means less retained N and more lateral N leaching Later in the growing season, with an active layer of 90 cm, the amount of 15 NO 3 recovered within the soil profile was smaller than in the early season, such as expected in hypothesis 2. Based on a 35% total recovery, soil microbes and soil bulk N were again the largest N recipients, whereas plants recovered even less than following July addition in 30 cm depth. About 65% of lateral N input likely drained further downslope such as seen in the early growing season. The smaller recovery in the later growing season can be explained by physical separation of lateral N movement and the root zone, which was also seen by Treat et al. (2016). Few roots penetrate to 80-90 cm in the permafrost-affected tundra (Zhu et al. 2016), and assimilation of N from these depths is limited . As such, 35% recovery of deep N input emphasizes the adaptation of tundra heath to N limitation, with few deep roots able to move N into shallow layer biological pools.
Evergreens shrubs access early season N, deciduous shrubs access deeper N The evergreen plant type benefitted the most from early season N input in 30 cm depth. In the late season, deeper N input benefitted deciduous types more than evergreens, especially in the long term (Year 1). Many evergreen species are able to take up early thaw season N, whereas deciduous plant types will rather utilize N made available during later spring and summer (Larsen et al. 2012) and may even acquire N in autumn (Blume-Werry et al. 2016), which is in line with our results. Although plant N recovery was only 2% of lateral N input, this could give evergreen shrubs a competitive advantage in the early season. Rooting depth was suggested as a plant functional type-specific trait important for understanding depth-specific (tracer) N uptake (Oulehle et al. 2016;Zhu et al. 2016;Pedersen et al. 2020). Deciduous shrubs such as S. glauca were comparatively efficient at accessing N released the in mid-active layer (45 cm depth) , and that pattern is repeated in our results with most of plant-retained lateral late season N input was acquired by deciduous shrubs one year after addition.
However, the fact that less than 1% of lateral N input was recovered by plants in the later growing season suggests that the ability to utilize early season N is more important than late growing season uptake.

Implication of ecosystem responses to increased lateral N inputs in a warmer future
Future increases in winter air temperatures may increase soil temperatures, thus winter N mineralization and the corresponding N release in spring (Rustad et al. 2001;Weintraub and Schimel 2003). Warmer winter air temperature can also result in more precipitation and thicker snowpack, which, in addition to also increasing soil temperatures (e.g. Rasmussen et al. 2018), will increase the amount of lateral water moving along slopes .
The downslope ecosystem may be affected by increasing winter and summer temperatures directly, but also by increasing the lateral N input. The effects of increased lateral N may affect the ecosystem in various ways, which, based on our detailed observations of an arctic ecosystem and literature regarding warmer tundra, will be discussed in the following.
Plant functional type: competition may be altered by higher lateral N input In a future warmer climate, where lateral N input may be larger, functional plant type specific advantages in accessing the N are exacerbated. More lateral N input in the early season may thus, based on our results, disproportionately benefit evergreen species, whereas deeper-rooted species may have advantages of less competition at depth (Xu et al. 2013), of reaching lateral N moving at deeper depths and accessing newly available mineral N from the thawing permafrost Salmon et al. 2018). The difference in ability to access N at depth suggested here may be one of the mechanisms behind vegetation community changes observed in the last decades Increase in easily available N in the early thaw season can increase field observed N 2 O emissions in a tundra ecosystem (Buckeridge, 2009), as can summer warming (Salazar et al. 2020). However, warming may also lead to a decrease in soil moisture (Andresen et al. 2020;Kolstad et al. 2021), which can create more aerobic conditions and inhibit an N 2 O pulse related to denitrification following early season mineral N release (Altschuler et al. 2019), or, if slumping or thermokarst formation occurs due to warming, waterlog the soil and suppress N 2 O emissions (Voigt et al. 2017a). Following the patterns in our study, higher mineral N availability just after snowmelt did not lead to an N 2 O pulse, suggesting that substrate availability was not the control on N 2 O production.
Higher winter mineralization rates may thus in fact not result in increased N 2 O emissions, but predicting the consequences of warming on N 2 O fluxes at this site could rather depend on predictions of microbial immobilization rates and soil moisture changes. This is challenging in permafrost environments, where thermokarst phenomena and patterned ground features result in heterogeneous flow patterns (Grosse et al. 2013), and projections of future soil moisture conditions in the active layer thus also span widely (Andresen et al. 2020). A key question remains therefore: which parts of the tundra will further dry out as a result of warming, and which parts will become moister and facilitate increasing N 2 O production.
Loss of ecosystem N as N 2 may also decrease with near-surface warming and drying, such as seen from early to later growing season in this study. Although N loss was small under ambient conditions at this site, the effect could contribute to increasing ecosystem N content and alleviate biological N limitation.
C and N interactions in future scenarios N availability is projected to be a strong limitation on future increases in plant C sequestration as result of climate warming (Meyerholt et al. 2020). Arctic 'greening', which has been observed with warming (e.g. Jia et al. 2009;Vickers et al. 2016;Edwards and Treitz 2017) may increase CO 2 uptake and counter future increases in greenhouse gas emissions from Arctic areas (IPCC 2019; Natali et al. 2012). However, biomass growth in a warmer climate depends on future availability of N (Rinnan et al. 2009;Rasmussen et al. 2020), which is thus critical for predicting vegetation response to warmer air temperatures. Lopez-Blanco et al. (2020) suggested that higher nutrient availability, which they attributed to lateral input from adjacent slopes, meant that a High arctic tundra was presently a stronger C sink than a low arctic tundra, in which vegetation was of larger stature. Foliar C/N ratios were lower and the primary productivity was higher at the High arctic site with richer soil in spite of a considerably cooler climate. If mineralization rates increase such as indicated in a warmer climate (Schimel et al. 2004), the bioavailable N could relieve plant N limitation and increase overall tundra plant C uptake. However, this study suggests that vegetation may not benefit much from increased N availability in the surrounding landscape, and as such emphasizes that the link between N availability and C storage in vegetation is likely indirect and depends on the soil N sink in a warmer climate.

Conclusions
Our combined results indicate that lateral N transport does occur in the early and the late growing season despite N limitation of the ecosystem, and lateral N flow is thus a component of N budgets of lowlands. The soil benefits the most from the retained N and is the pool which can be impacted by lateral N input. With soil N being also the by far the largest pool, the relatively small amount of lateral N input may, however, not have a large influence on the pool size. Microbes were better competitors for early season lateral N compared to vegetation, and will likely benefit the most from enhanced lateral N input as a result of winter warming and higher N mineralization rates. Thus, while vegetation benefited only little from lateral N input directly, it may benefit gradually during the growing season, as microbial turnover frees some of the immobilized N.
While some microbes obtained early season lateral N input, landscape N fluxes did not fuel N 2 O emissions, indicating that the early season mesic tundra heath is not a notable present source of the powerful greenhouse gas. Whether N 2 O emissions will increase in the future, depends on warminginduced changes in surface layer soil moisture and future soil mineralization rates.
Deciduous shrubs and deep-rooted species could gain overall competitive advantages in a future with more active layer N mineralization, however, evergreen shrubs were the best at utilizing N released in the early season. Future studies of spatial N connectivity and of how the vegetation community composition will be affected thus need temporal resolutions small enough to capture the dynamic nature of the snowmelt and later growing season.
Our conclusions have implications for the Arctic C sink strength, as we suggest that lateral N input is more readily retained in the soil compared to plant tissue. Plant N limitation will increase with warmer temperatures, however, with warmer temperatures, N mineralization will also increase. Whether plant N demand is met is a key question to in order to constrain the future C balance in the Arctic tundra in the future, and whether the soil will continue to immobilize and adsorb also increased lateral N input in competition with plants is crucial to address. Studies with experimental or simulated increase in lateral N inputs and direct effects of surface warming are needed to assess the future implications of that issue for a wider arctic region.