Nitrogen retention and transport differ by hillslope aspect at the rain-snow transition of the Colorado Front Range



Over a decade of research in the alpine zone of the Colorado Front Range has shown that atmospheric nitrogen (N) deposition originating from source areas in low elevation, developed areas, has changed ecosystem stoichiometry, nutrient transformations, and aquatic community structure. Less research has occurred in the montane zone, which sits at the current rain-snow transition and is vulnerable to climate change, land cover disturbances, and increased N loading. We conducted lithium bromide and 15N-nitrate (15NO3) tracer studies during spring snowmelt to determine the immediate fate of N in a forested catchment. Measurements of N species and applied tracers in ecosystem pools and soil solution on north and south facing slopes provided a means of determining export pathways and uptake of deposited N. Our results indicate that NO3 residence time is longer within north than south facing slope soils, due to longer contact with the soil matrix, greater microbial biomass N, and a larger soil organic matter pool. On the north facing slope, >50% of the 1 kg ha−1 of 15NO3 applied was retained in soil and vegetation pools. On the south facing slope, rapid transport during sporadic snowmelt events reduced total recovery of the 15N label in ecosystem pools to 16–34%. Our results suggest that snowmelt events quickly transport N through south facing slope soils, potentially contributing more N to aquatic systems than north facing slopes. Thus, it is important to consider how the fate of N differs by hillslope aspect when predicting catchment-scale N export and determining ecosystem N status across the Colorado Front Range.

1 Introduction

Reactive nitrogen (N) deposition to the biosphere has increased dramatically since 1970, due to a larger global population increasing the amount of reactive N created in urban, industrial, and agricultural areas [Galloway et al., 2008]. Large hydrologic losses have often been reported as an important mechanism causing terrestrial ecosystems to remain limited with respect to N [Hedin et al., 1995; Vitousek and Howarth, 1991], yet direct measurements of near-surface hydrologic response and N biogeochemical processes are not often made simultaneously. Watershed N biogeochemical studies have had a strong focus on understanding input and output budgets [e.g., Campbell et al., 2004; Likens et al., 1970] and how soil N biogeochemical processes mediate them. At smaller scales (i.e., within catchments), past process-based studies have documented a link between seasonal hydrologic events and ecosystem N dynamics, such as during spring snowmelt in high-elevation systems [Brooks and Williams, 1999; Brooks et al., 1998; Hood et al., 2003; Baptist et al., 2010; Fan et al., 2012] and winter rains in xeric environments [Davidson, 1992; Hungate et al., 1997]. However, these studies have not evaluated the links between event-based hydrologic response and ecosystem N uptake and transport.

Similar to many places around the world, elevated deposition of reactive N to the alpine zone of the Colorado Front Range has been a concern in the last three decades [e.g., Benedict et al., 2013]. The 4–5 kg N ha−1 yr−1 that reach the alpine zone is primarily ammonium nitrate (NH4NO3), with >75% coming from anthropogenic sources in surrounding low-elevation areas [Baron et al., 2000]. Baron et al. [2000] report that these sources—primarily urban, industrial, and agricultural—are more prevalent on the eastern than the western slope of the Rocky Mountains, leading to 3–5 kg ha−1 N yr−1 and 1–2 kg N ha−1 yr−1 deposited on the eastern and western slopes, respectively. The amount of N deposited is strongly tied to the amount of precipitation that falls across the elevation gradient, with the highest N deposition and precipitation in the alpine zone [see Williams et al., 2011]. The fate of reactive N within the alpine is of interest because the poorly developed soils and sparse vegetation make these ecosystems particularly vulnerable to the effects of small changes in N, climate, and other drivers; in addition, these watersheds serve as a drinking water source to communities in the Colorado Front Range and Denver Metropolitan area, and, thus, it is important to understand how incoming N deposition alters the health of terrestrial and aquatic ecosystems [Williams et al., 1996a; Williams and Tonnessen, 2000].

The impact of chronic elevated N deposition varies broadly by ecosystem, with some watersheds appearing far more sensitive to atmospheric deposition than others [Baron et al., 2011]. For example, studies in the alpine zone of the Colorado Front Range have shown that biota in this region (e.g., lichen and phytoplankton) are particularly sensitive to relatively small changes in N. Bowman et al. [1993] observed increased photosynthetic rates in plots in alpine dry (dominated by Kobresia suroides) and wet (dominated by Carex scopulorum) meadows that were amended with N and N plus phosphorus (P) (average inputs were 25 g m−2 yr−1 of each nutrient). They found additional ecological changes within these treatments, including shifts in vegetation community composition and grasshoppers feeding in areas with higher foliar N. Baron et al. [2000] showed that the eastern, compared with the western slope of the Colorado Front Range, has lower soil C:N ratios, higher potential N mineralization rates, lower foliage C:N, and higher lake NO3 concentrations with accompanied changes in diatom community structure from complex assemblages before 1900 to domination by Asterionella formosa around 1970. Physical features of this landscape also limit retention of deposited N; past studies in the alpine zone have reported that steep slopes, shallow soils, extensive talus cover, sparse vegetation, short growing seasons, snowmelt-dominated hydrology, and low rates of N cycling and primary productivity result in the majority of N deposition leaving the ecosystem via hydrological flow paths (e.g., Baron et al. [1994] and Williams et al. [1996a, 1996b] in Fenn et al. [2003]).

Compared to the large number of N studies in the alpine zone, there have been relatively few in forested montane catchments near the rain-snow transition in the Colorado Front Range. These forests have also been described as strongly N limited [Burns, 2004; Fenn et al., 2003] leading primarily to retention of deposited N in plant, soil, and microbial pools; current N loads at the rain-snow transition are approximately 2 kg N ha−1 yr−1 [Williams et al., 2011]. No studies to date have considered how spring snowmelt, the major hydrologic event in these systems, affects the amount of N retained in ecosystem pools. Recent advances in the conceptual model of N saturation in forested ecosystems suggest that researchers need to consider N losses simultaneously with the size and capacity of the sinks (or pools) of N within the ecosystem [Lovett and Goodale, 2011]. In systems where physical factors (e.g., soil properties and hydrologic forcing) are a predominant control over N cycling, deposited N can move both into biological pools (e.g., microbial communities and plants) and out of the terrestrial system via loss pathways. This observation serves to explain how terrestrial systems can remain perpetually N limited, never reaching N saturation, despite receiving relatively high N loads.

In the Colorado Front Range, hydrological flow paths differ by hillslope aspect. The Rocky Mountains trend north-south, leading to east-west trending valleys, with contrasting north and south facing hillslopes. In the elevation range of the montane forests, north facing slopes tend to develop a seasonal snowpack, while south facing slopes experience intermittent accumulation and melt throughout the winter and spring. Hinckley et al. [2012] found that two primary hydrologic regimes exist in these catchments: south facing slopes are characterized by preferential flow during melt events, followed by flow in unconnected soil pores between events, and north facing slopes experience sustained flow through connected soil pores during spring snowmelt (Figure 1). We hypothesized that these differences in hydrologic response would lead to heterogeneous N transport where and when preferential flow occurs in south facing slope soils, whereas north facing slope soils would have higher retention of N in ecosystem pools due to longer residence times providing more interaction with microbial communities and plant roots. Such differences would likely have implications for N cycling in the summer growing season, the amount of N delivered to the channel, and the overall N retention capacity of ecosystems at this elevation. To our knowledge, few studies have simultaneously explored development of hydrological flow paths and the fate of N deposition in montane forests during snowmelt.

Figure 1.

Conceptual diagram of north and south facing hillslope hydrological flow paths based on results from Hinckley et al. [2012]. (1) Inputs, (2) preferential flow, (3) shallow subsurface stormflow, (4) immobile matrix flow, and (5) groundwater flow (not measured directly).

We addressed the interactions of aspect, snowmelt, and N cycling using a 15N tracer approach in a montane catchment. Labeled N has been used as a tracer of N pools and fluxes in several studies of forested ecosystems [see Templer et al., 2012; Nadelhoffer et al., 1999b], both in the field [e.g., Matson et al., 1987; Tietema et al., 1998; Nadelhoffer et al., 1999a; Billbrough et al., 2000] and laboratory [e.g., Barraclough and Puri, 1995; Dail et al., 2001]. More recently, researchers have used 15N tracer studies to follow hydrologic losses of N from forested systems [e.g., Durka et al., 1994; Fotelli et al., 2004; Lohse and Dietrich, 2005; Perakis et al., 2005; Campbell et al., 2007; Kjønaas and Wright, 2007]. Coupled with a hydrologic tracer, such as bromide or deuterium, water can be followed independently of the labeled N. Such studies have provided insight into N movement via a particular flow path in their respective systems, such as infiltration and lateral subsurface storm flow [Lohse and Dietrich, 2005; Perakis et al., 2005] and runoff [Kjønaas and Wright, 2007]. In this study, we used 99.3 atomic percent (at. %) potassium nitrate (K15NO3) and lithium bromide (LiBr) as a tracer of N and water movement during snowmelt.

2 Site Description

The study area was described in Hinckley et al. [2012]. In brief, Gordon Gulch (40.01°N, 105.47°W) is a 2.7 km2 subcatchment within the Boulder Creek Critical Zone Observatory (BcCZO) in Colorado, USA. At 2440–2730 m elevation, Gordon Gulch is currently at the rain-snow transition in the Colorado Front Range of the Rocky Mountains; lower elevations experience more precipitation as rain, and at higher elevation precipitation is dominated by snow. Mean annual precipitation is 506 mm, based on the 20 year average (National Atmospheric Deposition Program (NADP) site CO94 at 39.99°N, 105.48°W), and the mean annual air temperature is 6.9°C [Western Regional Climate Center, 2011]; during the year of our study, approximately 70% of annual precipitation fell as snow. Vegetation is characteristic of the upper montane zone described by Marr in Birkeland et al. [2003]. Moderately dense lodgepole pine (Pinus contorta) forests dominate north facing slopes, whereas low-density ponderosa pine (Pinus ponderosa) stands dot the south facing slopes, with intervening grasses [Peet, 1981].

Soils are primarily Bullwark-Catamount families-Rubble land complex [National Resources Conservation Service (NRCS), 2009] on 5 to 40% slopes. Along the study transect discussed below, we found that depth to saprolite on hillslopes was typically 30–45 cm, with greater depths on the north than on the south facing slopes. The depth to saprolite is relatively uniform along each hillslope profile, with the exception of the toeslope (i.e., stream terrace in the riparian zone), a depositional area, where depth to saprolite is greater than 1.5 m. Basic soil characteristics on the north and south facing slopes are summarized in Table 1. In general, surface (i.e., 0–10 cm) soils on the north facing slope have higher C:N ratios and lower bulk density values, likely arising from higher organic matter contents than south facing slope soils. Soil pH is also slightly lower in north facing slope soils (5.35 ± 0.13) than south facing slope soils (5.81 ± 0.59) (Table 1).

Table 1. Mean (±1 SE) Soil Characteristics on the North and South Facing Slopes
 Soil C:NaSoil TextureaBulk Density (g cm−3)bSoil pHb
  1. a

    Soil C:N data are from Eilers et al. [2012], n = 4 soil pits (north facing slope) and n = 3 soil pits (south facing slope). Soil texture was measured on the same samples but reported in NRCS [2009].

  2. b

    Bulk density and pH data are from this study, n = 3; pH was measured in water.

North facing slope    
0–10 cm35.5 (8.5)Silt loam0.98 (0.05)5.35 (0.13)
10–25 cm26.0 (3.5)Gravelly sandy loam1.12 (0.12)5.47 (0.18)
South facing slope    
0–10 cm26.9 (2.7)Fine sandy loam1.145.81 (0.59)
10–25 cm21.6 (2.0)Fine sandy loam1.315.75 (0.49)

3 Methods

3.1 Experimental Design and Hydrological Observations

We established experimental plots across a north-south cross section of the study catchment. Plot locations included upper (U) and lower (L) positions on the north (NF) and south facing (SF) aspects, as well as the toeslope (TS) in the riparian zone adjacent to the creek (Figure 2). Each plot (one unlabeled/no tracers and one labeled/tracers added per hillslope position) was 4 m × 4 m, and their locations were chosen for representative vegetation cover and subsurface structure, as determined by vegetation sampling and soil pit excavations in August 2009 (S. Anderson, unpublished data, 2009). Complete description of the study design is in Hinckley et al. [2012]. In brief, we installed nested instrumentation within each plot, including tension (Prenart Equipment, ApS) and zero-tension lysimeters (dimensions: 10.2 (W) × 30.5 (L) × 5.0 (D) cm; design described in Hinckley et al. [2008]) to sample water held within the soil matrix and gravity-driven leachate moving through macropores, respectively. Lysimeters were installed in September and October 2009, before winter snowfall, and remained in situ for 6 months prior to sampling.

Figure 2.

Study area, location of experimental transect, and instrumentation of the plots. (a) LiDAR shaded relief image for Gordon Gulch, a study catchment within the BcCZO showing location of the experimental transect and (b) nested instrumentation within each study plot. Inset shows the location of Gordon Gulch within the state of Colorado, USA.

At the same time, soil temperature and soil water content sensors (Campbell Scientific model numbers CS-107 and 616, respectively) were installed. Lysimeters, soil temperature sensors, and water content probes were installed at two depths within the hillslopes dictated by the shallow soils at the U and L positions of each hillslope: (1) below the majority of the rooting zone of the grasses and shrubs (10 cm) and (2) near the soil-saprolite interface (30 cm). At the TS position where soils are deeper, lysimeters and both sets of sensors were installed at four depths within the soil profile (5, 10, 30, and 60 cm). Our plot sizes were limited (i.e., 16 m2), both due to the cost of 15N label and analyses and the need to minimize impact to the site; thus, only one instrument of each type was installed at each depth per plot. Instead of evaluating variability in N movement and recovery spatially, we focused our study on the frequent (i.e., daily) temporal sampling of soil water.

3.2 15N Tracer Studies

On 10 April 2010, rapid melting of the seasonal snowpack began on the north facing slope, and we applied the tracer solution to all plots. We wanted to investigate the fate of N through near-surface soils during snowmelt, and so we chose to apply the tracer solution to the soil surface at the same time on both slopes. This approach also increased the likelihood that we would recapture the tracer within our plots. At the time of application, there was no snow on the south facing slope, and snow depth was 0.24 ± 0.12 m on the north facing slope. On the north facing slope, snow was removed from each plot and piled onto a tarp. Tracer solution containing K15NO3 at 99.3 at. % (2.5 L of solution) was applied to the soil surface using a handheld pesticide sprayer at a rate of 1 kg 15NO3-N ha−1 and 0.44 mm h−1 to the entire area (16 m2) of each plot. This quantity of N represented the mean annual nitrate load at this elevation, which is approximately 50% of the total N received via atmospheric deposition to the region. According to Templer et al. [2012], 15N studies that apply < 2.5 kg N ha-1 yr-1 are considered ambient; our objective was to add a sufficient quantity of 15N to obtain measureable results while minimizing the potential for a fertilization effect. Simultaneously, 0.5 M LiBr (6.25 g Br m−2) was applied to track hydrological flow paths through the near surface. Hinckley et al. [2012] report the hydrological and LiBr data in detail, and we reference them herein to describe the snowmelt dynamics and transport during the study period.

After applying the tracer, the snow was replaced on the north facing slope plots, its structure destroyed, thereby reducing the snow depth by ~50% but maintaining the snow water equivalent (SWE). Because the snowpack was isothermal when the experiment began, we expect that structural effects on melt water movement were minimal. Several studies have shown that a transition from preferential to matrix flow is common as a snowpack ripens and warms [Colbeck, 1979; Lee et al., 2010; Williams et al., 2010], so we were not concerned about destroying structural features.

Tension lysimeters were immediately set to 40.6 cm Hg of vacuum after tracer application to begin sampling N species (NH4+, NO3, and total dissolved nitrogen (TDN)) and tracers in soil water. In the tracer plots, we sampled tension lysimeters daily for 40 days, then on an event basis until 27 May (total sampling period was 47 days), and collected leachate from zero-tension lysimeters when available. In the unlabeled plots, we sampled solution from lysimeters 6 times during the measurement period, in order to characterize background 15N (i.e., natural abundance) values at the site. Zero-tension lysimeters were emptied on 24 h intervals, so these data integrate over the same time period as soil water from the tension lysimeters.

3.3 Sampling Soil, Vegetation, and Microbial Biomass

We determined the amount of 15N tracer in ecosystem pools following spring snowmelt. On 26 May 2010, we sampled surface soils (0–10 and 10–25 cm) at nine locations within the experimental plots and three locations within the unlabeled plots. We then composited the cores into three samples from the experimental plots and one sample from the unlabeled plots. These samples were immediately taken to the laboratory for measurements of field moisture content, soil pH, determination of inorganic N (NH4+ and NO3) pools, and quantification of microbial biomass N (described below). Soils were sieved (2 mm mesh size) prior to analysis of N, eliminating all but very fine roots.

On 12 July 2010, we sampled new aboveground biomass in the labeled and unlabeled plots. On the south facing slopes, we quantitatively harvested grasses and new shrub leaves within 1 m × 1 m quadrats (one per hillslope position) located in the center of each 4 m × 4 m plot. On the north facing slope, where an understory was not present, we measured 15N uptake into new P. contorta needle growth by sampling 10–15 buds within labeled and unlabeled plots (a pooled sample in each plot). Because we could not destructively harvest biomass in the north facing slope plots, we calculated annual foliage increment using the value of dry litter mass represented by pine needles reported in a study of P. contorta by Turner et al. [2004], assuming steady state and, thus, new foliar biomass increment being equal to litterfall. We combined these data with information about our site that stand density is 0.25 trees m−2 and age is estimated at 124 years (H. Adams, unpublished data, 2013) to determine the quantity of new pine needles per plot. We believe that these estimates allowed us to place a reasonable bound on foliar biomass increment in order to calculate 15N recoveries.

3.4 Laboratory Analysis

All soil solution samples were filtered using 0.45 µm Supor membrane syringe filters (Acrodisc) in the field and stored in 60 mL high-density polyethylene bottles. Samples were analyzed in the Niwot Ridge Long-Term Ecological Research (LTER) Kiowa laboratory (Boulder, CO) within 1 week of collection. For all soil samples, inorganic N pools were determined by extracting 60 g field-moist soil in 150 mL 2 M potassium chloride that was shaken periodically for 18 h and then filtered through cellulose Whatman 1 filters. Microbial biomass N was extracted by preparing two sets of soil subsamples (60 g field-moist soil) in 150 mL 0.5 M potassium sulfate, and treating one set with 1 mL ethanol-free chloroform to serve as a fumigated sample, following Fierer and Schimel [2003]. The extraction period and filtration were the same as for inorganic N determination (discussed above). Samples were oxidized with persulfate digestion, according to Cabrera and Beare [1993], and microbial biomass N was calculated by differencing fumigated and unfumigated samples without correction for extraction efficiency [see Silver et al., 2005].

Diffusion of soil extracts for 15N isotopic analysis, including blank correction, was performed following Stark and Hart [1996]. We optimized our diffusion technique for extracts in the 20–60 mL range with 20–400 µg inorganic N at <30 at. %. Due to our use of highly enriched 15NO3 label and the presence of low NO3 concentrations in soils at our study site, we spiked the soil extracts with 60 µg NO3-N (natural abundance) in order to dilute the isotopic signal and have sufficient N for determination of 15N at. %. We subsequently subtracted the effect of the spike to get actual 15N abundance.

In both water and soil samples, NH4+ was measured on a BioTek Synergy 2 with a detection limit of 0.009 mg N L−1. Nitrate was measured on an OI Analytical FS-IV with a detection limit of 0.5603 µg N L−1 and TDN on a Shimadzu total organic carbon-V CSN with a detection limit of 0.014 mg N L−1. Nitrogen stable isotope composition in solution was measured at the University of California, Davis, Stable Isotope Laboratory using the denitrifier method [Sigman et al., 2001; Casciotti et al., 2002] on a ThermoFinnigan GasBench and PreCon trace gas concentration system. 15N values were calculated using working standards composed of N2 and N2O (e.g., 3% N2 + 1 ppm N2O with the balance He or 1 ppm N2O with balance N2). The N2 and N2O (following conversion to N2 + O2) were calibrated against the 15N value of air. 15N in bulk soils (i.e., 15N of total N) and vegetation was measured at the Stable Isotope Laboratory at ETH, Zurich, using a Thermo-Fisher 1112 flash elemental analyzer connected to a Thermo-Fisher DeltaV Continuous-flow Mass Spectrometer; data were corrected to two in-house standards calibrated to the international standards International Atomic Energy Agency (IAEA)-N1, IAEA-N2, and IAEA-N3. 15NO3 and 15NH4+ in soils and microbial biomass (15NO3 only), diffused onto filter disks, were measured at the University of California, Davis, Stable Isotope Laboratory using a PDZ Europa Automated Nitrogen Carbon Analyser-Gas Solids and Liquids elemental analyzer with PDZ Europa 20-20 Isotope Ratio Mass Spectrometer (long-term standard deviation of 0.3‰ for 15N). Values were calibrated using two in-house standards and to the international standard for air. Recovery of N masses in spiked and untreated samples were 85 ± 2.3% and 88 ± 6.3%, respectively.

3.5 Calculation of 15N Recovery in Ecosystem Pools

We calculated the recovery of 15N label in bulk soil N, inorganic N pools, and microbial biomass in 0–10 cm and 10–25 cm soil cores, foliar tissues, and leachate. After Nadelhoffer et al. [1999a], we used the following equation to determine the recovery of 15N relative to natural abundance values for each ecosystem component:

display math(1)

where 15Nrec is the mass of 15N tracer recovered in the labeled N pool (kg N ha−1), mpool is the N mass of the labeled N pool (kg N ha−1), at. % 15Npool is the at. % 15N of the labeled N pool, at. % 15Nref is the at. % 15N of the reference (prelabeled or nonlabeled) N pool, and at. % 15Ntracer is the at. % 15N of the tracer applied (i.e., 99.3 at. %). For reference 15N abundance, we used samples from unlabeled plots at the same landscape positions as the labeled plots (see Table 3). We express recoveries as proportions of the total 15N tracer applied to the plot. Note that 15N recoveries are not double counted in soil pools. All references to total ecosystem retention of N in each plot refer to uptake by the bulk soil (i.e., total N pool), not also the microbial biomass and inorganic N pools. The latter two pools are part of the soil N pool, and we calculate the amount of 15N tracer measured in each to provide further resolution on the fate of N within the soil N pool. For calculations of 15N recovery in leachate, we assumed that the volumes of leachate and concentrations of NO3 collected in the zero-tension lysimeters (an area of 300.5 cm2) were representative of the plot (16 m2). We scaled volumes and NO3 masses collected in each zero-tension lysimeter to calculate mass movement of NO3 at the plot scale and, using the equation above, the percent recovery of 15N tracer.

3.6 Statistical Analysis and Data Reporting

We report means and standard error of 15N recovery in bulk soils, inorganic N pools, and microbial pools by soil core increment (0–10 cm and 10–25 cm), using the plot as our experimental unit. In all cases, we report data in terms of N for each compound. We used a mixed effects model (residual maximum likelihood estimation, JMP Statistical Software, version 10) to determine whether or not there was a significant effect of hillslope position (i.e., plot) or soil depth (i.e., 0–10 cm and 10–25 cm) on recovery of 15N in soil pools. In our model, we treated plot and soil depth as fixed effects and sample number (i.e., sample 1, sample 2, and sample 3 per labeled plot) as the random effect.

4 Results

Within all five positions on our catchment transect, we measured the movement of 15NO3 tracer and N species in soil water and leachate, as well as the recovery of tracer in ecosystem pools (soil, microbial biomass, and foliar tissues) and leachate. Total recovery of 15N in measured pools and leachate at each hillslope position was as follows: 76.6% and 34.5% at SF-U and SF-L, respectively, and 52.1%, 55.7%, and 53.8% at TS, NF-L, and NF-U. We present the patterns of tracer movement observed at the onset of seasonal snowmelt on the north facing slope, followed by details of the tracer recovery results.

Immediately following tracer application on 10 April, we began measurement of N species (NH4+, NO3, and TDN) and 15NO3 in soil water and leachate at each hillslope position. During the 47 day period of measurement, we found that movement of all N species was closely coupled to start of snowmelt and changes in volumetric water content in the subsurface (Figures 3 and 4). At SF-U, which was snow free when the tracer was applied, concentrations of N species in soil water, as measured via the tension lysimeters, increased above background levels during a large (57 mm SWE) storm that occurred 21 through 26 April. At this time, both inorganic N (presumably from the melting snow) and the 15N tracer were mobilized at 10 and 30 cm (Figures 4 and 5). SF-U and SF-L differed in the degree to which N was mobilized. Concentrations of NO3 were an order of magnitude larger at SF-U than SF-L; in general, SF-U had persistently high background NO3, and 15NO3 was measurable throughout the study period at 30 cm. After 3 May, 15NO3 at SF-U dropped to 0.933 at. %, approaching natural abundance levels (0.3678 ± 0.0008, 15 samples from unlabeled plots), indicating that the majority of the tracer had passed out of the pore waters or had been diluted by native N. Concentrations of NO3 and volumes of solution sampled in the lysimeters were too low at SF-L (<0.404 mg NO3-N L−1) to get measurements of 15NO3 during the period of measurement.

Figure 3.

Volumetric water content during the period of measurement at each hillslope position. The black arrow marks the start of the 15NO3 and LiBr tracer experiment. The shaded area marks the duration of the seasonal snowpack at each hillslope position. Data are from Hinckley et al. [2012].

Figure 4.

Compilation of N species, (a) TDN, (b) NH4+-N, and (c) NO3-N, in soil water (tension lysimeters, closed symbols) and leachate (zero-tension lysimeters, open symbols) at each hillslope position. The shaded area marks the duration of the seasonal snowpack, and gray bars mark precipitation events >10 mm snow water equivalent.

Figure 5.

(a) 15NO3 and (b) bromide in soil water (tension lysimeters, closed symbols) and leachate (zero-tension lysimeters, open symbols) at each hillslope position. Bromide data are from Hinckley et al. [2012]. The shaded area marks the duration of the seasonal snowpack, and gray bars mark precipitation events >10 mm snow water equivalent.

In contrast, on the north facing slope, detection of all N species and movement of 15N tracer in soil water occurred immediately following application at the TS, NF-L, and NF-U positions. We continued to measure elevated concentrations of these constituents until the seasonal snowpack melted completely on 19 April at NF-L and NF-U (as determined by daily field observations) and 5 May at TS (Figures 4 and 5a). At TS, we measured 86.3 at. % 15 NO3 at 10 cm in tension lysimeters on 13 April and detected the tracer at 60 cm by 17 April; a maximum of 32.2 at. % 15 NO3 was measured at this deepest lysimeter on 18 April, indicating that significant mixing of solution waters containing native N likely occurred. At north facing slope positions, we measured near-natural abundance levels (<0.6 at. %) of 15 NO3 by 25 April. There was a second peak in NO3 during the 21 through 26 April storm, measured in soil water at 30 cm (see Figure 4c) at both NF-L and NF-U; however, the limited isotopic measurements that we obtained indicate that the tracer front had moved though on the N facing slope or was diluted by native N from the melting snowpack (Figure 5a).

We sampled leachate from zero-tension lysimeters immediately following large snowmelt events on the south facing slope and during melting of the seasonal snowpack on the north facing slope (i.e., when it was available on both slopes). These samples reflected the mobile water component of flow (see explanation in Hinckley et al. [2012]). At all hillslope positions except SF-U, NO3 and 15NO3 in leachate were the same as those measured in soil water; at SF-U, NO3 concentrations at 10 and 30 cm were higher in leachate than in soil water measured at the same time (see Figure 4c). Recovery of 15N tracer in leachate was highest at SF-U (60.5%); at SF-L it was 0.01%, and at north facing slope positions, it was 1.7–4.8% (see Table 4 and Figure 6). We conducted limited analysis of leachate samples for TDN and NH4+, due to low sample volumes. At all hillslope positions, concentrations of these constituents were higher in leachate than soil water throughout the study period (Figures 4a and 4b).

Figure 6.

Summary of 15N tracer recovery in ecosystem pools and leachate at each hillslope position. Values are expressed as a percent of the total 15N applied (0.1 g 15 NO3-N m−2). MB is microbial biomass, and inorganic N (NH4+ plus NO3) values are in brackets. Arrows represent leachate from plots.

The second part of the study involved quantifying 15N in new aboveground biomass, bulk soil (total N), the soil inorganic N pool (NH4+ and NO3), and soil microbial biomass following the snowmelt period on north and south facing slopes. We recovered approximately 50% and 16–34% of the 15N tracer in soil and vegetation pools on north and south facing slopes, respectively. Tables 2 and 3 summarize the background N masses and 15N natural abundance values of these ecosystem components measured in nearby unlabeled plots, and Figure 6 summarizes recovery of 15N in different ecosystem pools. Biomass harvests at SF-U, SF-L, and TS (dominated by short-stature vegetation; no trees) yielded a range of standing stock values from 19 to 123 g dry matter m−2 in both labeled and nonlabeled plots. For north facing slope plots (NF-U only, as NF-L did not have any live trees), we estimated new foliar biomass to be 12.5 g dry matter m−2 yr−1, following Turner et al. [2004] and using stand density and age estimates from our site (reported above). We found that vegetation within south facing slope plots took up 8–10% of applied 15N while new P. contorta needle growth on the north facing slope accounted for <1% of the label.

Table 2. Background N Contents in Ecosystem Pools at Each Hillslope Position (One Composite Sample Per Plot)
Ecosystem PoolSF-USF-LTSNF-LNF-U
N (%)N (g m−2)N (%)N (g m−2)N (%)N (g m−2)N (%)N (g m−2)N (%)N (g m−2)
Bulk soil          
(0–10 cm)0.11124.00.41471.70.16174.30.34317.10.50464.0
(10–25 cm)0.06115.30.10195.30.07123.80.0346.00.0573.1
Soil NH4+          
(0–10 cm)0.00040.4830.00050.6110.00040.3820.00050.4510.00111.026
(10–25 cm)0.00030.4950.00020.4610.00010.2190.00010.1740.00010.223
Soil NO3          
(0–10 cm)0.00020.2620.000020.0210.000010.0120.00110.0340.000050.046
(10–25 cm)0.00010.198<0.000010.0060.000010.0260.00010.0060.000020.033
Microbial biomass          
(0–10 cm)0.00050.5310.00131.5070.00111.2040.00181.6250.00222.059
(10–25 cm)0.00020.4340.00051.0690.00020.3050.00020.2480.00060.988
New foliage tissues1.960.870.991.220.880.340.73nd0.760.10
Table 3. Natural Abundance 15N (at. %) Used to Calculate 15N Tracer Recovery in Ecosystem Pools and Leachate
Ecosystem PoolSF-USF-LTSNF-LNF-U
  1. a

    One composite sample per plot.

  2. b

    Average abundance across available samples per plot.

Bulk soila     
(0–10 cm)0.36720.36680.36730.36620.3659
(10–25 cm)0.36780.36760.36800.36780.3676
Soil NH4+a     
(0–10 cm)0.36590.36550.36470.36410.3666
(10–25 cm)0.36660.36700.37680.37190.3827
Soil NO3a     
(0–10 cm)0.36430.33570.33300.35160.3632
(10–25 cm)0.35970.36710.32250.27550.3579
Microbial biomassa     
(0–10 cm)0.40870.36120.39460.35960.3648
(10–25 cm)0.36520.37550.36240.37240.3776
New foliage tissuesa0.36600.36590.36580.36490.3649
(5 cm)ndnd0.3690ndnd
(10 cm)0.36680.3663nd0.3656nd
(30 cm)ndndnd0.36560.3740

Soil organic pools took up a large proportion of 15N applied by the time they were sampled 46 days after the tracer application. Our bulk soil recoveries provide insight into uptake of 15N tracer within the total soil N pool, while 15N tracer recoveries in inorganic N and soil microbial biomass provide more resolution on the fate of 15N within the total pool. We found that bulk soils (0–25 cm) contained >50% of 15N applied to N facing slope positions; most applied 15NO3 retained was likely in the organic pool, as measurement of 15N recovery in the inorganic N soil pool (NH4+ + NO3) revealed relatively low levels (2–7% of applied tracer at north facing positions, Table 4 and Figure 6). Approximately 8–13% of the total tracer 15N was recovered in microbial biomass in the top 25 cm of soil at TS, NF-L, and NF-U. Overall, 15N tracer recoveries in soil were lower at the south facing than the north facing slope positions. On the south facing slope, we measured a maximum recovery of approximately 24% in bulk soil (0–25 cm, SF-L). Within the microbial pool (0–25 cm soil), we measured recoveries of 0.8% at SF-U and 6% at SF-L.

Table 4. Recovered 15N Tracer in Ecosystem Pools and Leachate at Each Plot Location
Ecosystem PoolSF-USF-LTSNF-LNF-U
  1. a

    Three composite samples per plot. Units are mg N m−2 (±1 SE).

  2. b

    Total recovery across depths in each plot. Units are mg N m−2.

Bulk soil total 15Na     
(0–10 cm)4.52 (1.36)12.28 (5.23)30.70 (8.07)33.55 (16.17)23.79 (20.72)
(10–25 cm)3.21 (1.36)12.21 (7.34)18.23 (2.82)17.34 (10.86)26.32 (3.44)
Soil 15NH4+a     
(0–10 cm)0.30 (0.09)1.33 (0.77)2.86 (1.06)2.09 (1.00)1.53 (0.77)
(10–25 cm)0.71 (0.48)2.15 (0.77)2.50 (0.42)1.00 (0.55)4.65 (3.38)
Soil 15NO3a     
(0–10 cm)0.82 (0.45)0.67 (0.16)0.36 (0.13)0.19 (0.18)0.21 (0.09)
(10–25 cm)0.40 (0.16)1.08 (0.24)0.41 (0.07)0.27 (0.13)0.96 (0.64)
Microbial biomassa     
(0–10 cm)0.24 (0.52)5.65 (4.58)6.70 (0.62)6.33 (3.47)6.59 (3.12)
(10–25 cm)0.53 (0.62)0.05 (0.31)2.67 (3.22)1.80 (1.63)6.64 (6.54)
New foliage tissuesa8.410.01.5nd0.08

These differences between 15N retention in soils on north and south facing aspects suggest overall higher retention of N on the north than the south facing slope. The mixed effects model indicates that the most statistically striking results were for the effect of hillslope position on 15N recovery in bulk soils (p = 0.11) and in the soil NO3 pool (p = 0.09). However, we did not find a significant effect (i.e., p < 0.05) of hillslope position or soil depth on 15N recoveries in soil pools (Table 5). We believe that our limited sample size and the heterogeneity of macropore features and, therefore, transport of 15N tracer, led to high uncertainty in our estimates of mean 15N recoveries in soil N pools.

Table 5. Results of a Mixed Effects Model to Test Whether Hillslope Position (Plot) and Soil Depth Have Significant Effects on 15N Tracer Recoveries in Soil N Pools
Response VariableF Ratiop Value
Bulk soil 15N  
Soil depth0.9970.34
Soil 15NH4+  
Soil depth0.5020.49
Soil 15NO3  
Soil depth0.7930.39
Microbial biomass 15N  
Soil depth1.5810.23

5 Discussion

Our study evaluated contrasting patterns of N transport and ecosystem retention on opposing hillslope aspects during spring snowmelt. We found that mobility of 15NO3 in soil water and leachate was largely controlled by soil moisture content in near-surface soils. On the north facing slope, we measured 15NO3 (as well as other N species) in soil water as soon as melting of the seasonal snowpack began (Figures 4 and 5). At TS, 15NO3 abundance in soil water decreased from a peak of 86.4 at. % on 12 April (2 days after tracer application) to 0.51 at. % on 25 May (45 days after tracer application) at 10 cm (Figure 5a). A decrease in the 15N abundance likely indicates mixing of native 15NO3 (natural abundance) and/or movement of the tracer front below the depth of the instruments.

On the south facing slope, we did not capture 15NO3 tracer in solution until a significant storm began on 21 April, 11 days after tracer application. At this time, N in snow was flushed through the soil matrix rapidly; the sharp peaks containing 63.8 at. % 15NO3 at SF-U (Figure 5a) demonstrate the rapid movement of the tracer front, which was corroborated by calculations of 15NO3 recovery in leachate, at 60.5%. However, it is important to note that the maximum 15NO3 abundance measured in leachate and soil water was much lower than that of the original tracer (99.3 at. %), most likely due to mixing with native NO3. We measured high background soil water NO3 throughout the study period at SF-U, relative to other hillslope positions. At 30 cm SF-U had approximately 2 mg NO3-N L−1, compared to 0.1 mg NO3-N L−1 at SF-L and 0.02–0.05 mg NO3-N L−1 at north facing slope positions. This result suggests that changes in the abundance of 15NO3 can be explained by mixing in the subsurface (Figures 4 and 5).

We did not measure rapid movement of tracer and N to depth at SF-L (Figures 4c and 5a), and 15NO3 recovery was only 0.01% in leachate, which is consistent with patterns in solution Br concentrations (applied simultaneously with 15NO3; see Figure 5b). We suspect that differences between the two south facing slope positions are due to the heterogeneity of macropore features at the submeter scale; preferential flow in near-surface fractures was captured by our instruments at SF-U, but not at SF-L.

Despite the rapid movement of 15NO3 in subsurface flow paths at all hillslope positions, it was decoupled from the movement of water to varying degrees on each slope, as indicated by retention in biological pools and soil organic matter (see Figures 5 and 6). We measured the majority of 15N recovery in the bulk soil N pool at each hillslope position except SF-U. N fertilization studies conducted in forested systems have reported that 10–50% of N applied is lost via denitrification, leaching, or volatization of ammonia (NH3), and approximately 70% remains in the soil [see Templer et al., 2012, and citations therein; Preston and Mead, 1994]. Given that we measured rapid vertical transport at SF-U, it is likely that 15NO3 was mobilized quickly after the storm on 21 April and perhaps had less contact time with the soil matrix than at all other plots where we measured 24–50% of the tracer in the top 0–25 cm of soil. Our results are consistent with those reported in Curtis et al. [2011] and Schleppi et al. [1998, 2004], as well as the conceptual model put forward by Lovett and Goodale [2011]. This previous research demonstrates that transport of N can occur simultaneously with ecosystem uptake of N in N-limited ecosystems; terrestrial N export does not only occur once forests are N-saturated. At our site, periods like snowmelt, which are characterized by rapid movement of water through the hillslopes [see Hinckley et al., 2012], likely favor transport over biological retention of N. Thus, the ecosystem remains N limited despite the trend in elevated N deposition to the region.

The degree to which 15N was retained in soils and microbial biomass on the north versus the south facing slope is particularly interesting. We measured higher recoveries of total 15N in soil (approximately 50% at north facing slope and 8–24% at south facing slope positions (0–25 cm)). Within the total N pool of the soil, recovery of the tracer in inorganic pools was merely 3–7% at north facing slope and 1–5% at south facing slope positions. These results suggest that stabilization of deposited N in the organic pool occurs on both the north and south facing slopes but that retention is greater overall in north facing slope soils. North facing slope soils also remained consistently wetter (Figure 3) and were not subject to the wetting and drying cycles of the south facing slope throughout the study period. Higher, consistent soil moisture throughout the spring on the north facing slope favors maintenance of organic-rich soils and, likely, a more stable microbial population with higher microbial activity than on the south facing slope [e.g., Fisk et al., 1998].

We recovered approximately 1–5% of applied 15N as 15NH4+ in soils (Table 4), suggesting that the label had made its way into (via uptake) and out of (via mineralization) the microbial pool before we assessed tracer recovery. The timing of our soil sampling (after the spring snowmelt period) likely determined the nature of these results. Based on other research in the Colorado Front Range [e.g., Brooks et al., 1998; Brooks and Williams, 1999], biota are adapted to respond quickly to available N during snowmelt and likely accessed it before we collected soil samples on 17 May. Several studies in other systems have described an early spring peak in microbial immobilization [e.g., Emmet and Quarmby, 1991; Groffman et al., 1993; Seely and Lajtha, 1997]. Emmet and Quarmby [1991] observed a decline in microbial retention following this peak, which they suggested was due to mineralization of added 15N following the death and decay of microbial cells. In their study, mineralized 15N tracer was available for plant uptake, leaching, or reimmobilization later in the season. Other studies in a variety of systems that measured the dual isotopes of NO3 in stream and soil water have found that the majority of NO3 has a microbial, as opposed to atmospheric signal even during periods of rapid runoff associated with snowmelt [e.g., Burns and Kendall, 2002; Campbell et al., 2002]. This finding suggests that N is quickly turned over and/or incorporated within soils. In our study, the highest recovery of 15NH4+ was in north facing slope soils, while the highest recovery of 15NO3 was in south facing slope soils. In the case of the latter, 15NO3 may have been tracer that was taken up neither by plants nor microbes (Table 4). These patterns are consistent with our results of total 15N recovery in soils and leachate, which suggest that more N deposition is cycled by the microbial community in north than south facing slope soils (Table 4 and Figure 6).

We estimated low (<1% of 15N applied) recoveries in P. contorta needle growth on the N facing slope. This result may represent an overestimation of 15N uptake into new needle growth, due to the assumption of steady state (i.e., litterfall equals new needle growth); as foliage matures, fewer buds may be produced. However, previous studies reported similarly low 15N uptake by trees. Preston and Mead [1994] recovered 5.8% and 6.8% of 15N in lodgepole pine trees (whole tree analyzed) during two separate years of measurement following a 15N tracer application. In an analysis of nine forested sites where 15N studies were conducted, Nadelhoffer et al. [1999b] report that 20% of 15N tracer recovery is in trees and that they are not a primary sink for N. Assuming that our biomass estimates for P. contorta on the north facing slope are correct, 15N uptake in foliar biomass was higher on the south facing slope, with 8.4% and 10% at SF-U and SF-L, respectively (Figure 6). Annual grasses comprise the majority of the vegetation community on the south facing slope (we did not have P. ponderosa in our study plots), and they are likely adapted to respond quickly to N availability in snowmelt before the comparatively drier summer months that follow. Billbrough et al. [2000] reported that alpine species showed a similar ability to respond rapidly to N availability. In their study, vegetation took up 50% of applied 15N upon initiation of snowmelt. In contrast, 15N uptake by P. contorta on the north facing slope may be delayed (i.e., measureable after several years).

In general, retention of the 15N tracer in measured ecosystem pools increased along the study transect from the south to the north facing hillslope positions (Figure 6). The balance of 15N tracer (i.e., the fraction not included in these pools) was accounted for, in part, by leachate, with the highest recovery at SF-U (60.5%), where we observed the most dramatic movement of 15NO3. However, heterogeneity of macropores and preferential flow within and between our plots may have caused us to overestimate recovery of 15NO3 in leachate at the plot scale at SF-U and underestimate it at our other plot locations. Still, these estimates provide a useful illustration of the potential pathways of reactive N deposition in this system. With the exception of total 15N recovery estimated at SF-L (34.5%), recoveries at the other hillslope positions (52.1–76.6%) fall within the range of those reported previously for forested systems (e.g., an average of 74.9% [see Templer et al. 2012]).

An important caveat to our study is that we applied 15N as NO3, but N deposition to the Colorado Front Range falls as NH4NO3 (see NADP, Ammonium tends to remain in soils longer than NO3, due to its greater affinity for exchange sites. We measured NH4+ transport during melting of the seasonal snowpack on the north facing slope and melt events that followed snowstorms on the south facing slope (see Figure 4b). Therefore, despite the potential for NH4+ to be retained in surface soils, our data suggest that when water availability is high, NH4+ may be mobile in this system, similar to NO3.

When comparing recovery of 15N on the two slopes, it is important to note that our study may not have accounted for all processes or reservoirs that control the fate of atmospheric N deposition. We did not measure uptake of N in fine roots or woody tissues, which have been reported as a significant sink for atmospherically derived N in other 15N tracer studies [Nadelhoffer et al., 1999a, 2004]. In their meta analysis, Templer et al. [2012] report that 15N uptake in fine- and coarse-root biomass accounts for 6.3% (n = 14) and 3.5% (n = 4) of tracer recovery, respectively. We also did not measure denitrification, which may occur during discrete events at the ground surface when soil moisture is high [e.g., Parkin, 1987; Robertson and Tiedje, 1987; Lloyd et al., 2006; Seitzinger et al., 2006], or gaseous losses during nitrification. We acknowledge also that differences in N use among P. contorta, P. ponderosa, annual grasses, and shrubs likely influenced our data; vegetation cover undoubtedly affects N retention in biological pools, as well as the forms and amounts transported out of surface soils. These relationships have been highlighted in previous studies [e.g., Fisk et al., 1998].

Despite not constraining all potential fates of N deposition, our results point to a compelling trend across this north-south cross section (Table 4 and Figures 5 and 6). The differences in the fate of N deposition that we observed on opposing aspects provide valuable insight into the immediate fates of atmospheric N deposition during the most significant hydrologic and N transport event of this system. Changes in the N status of midelevation systems are likely to be driven strongly by changes in climate. For example, an increase in the ratio of precipitation falling as rain versus snow may cause more pulsed inputs of water on the north facing slope (i.e., little or no snowpack development), similar to the current conditions on the south facing slope. The degree to which N is transported versus retained in biological pools may be largely dependent on the size and duration of rain events, which will, in turn, affect N loading to downgradient aquatic systems. Our study shows that as midelevation landscapes respond to changes in climate and N deposition, it will be important to evaluate responses and feedbacks on both north and south facing slopes to understand catchment-scale N export and ecosystem N status.

6 Conclusions

The overall patterns that we observed with respect to transport and retention of applied 15NO3 suggest stark contrasts in the fate of atmospheric N deposition on north and south facing slopes of midelevation catchments and have implications for catchment-scale patterns of N losses. Our solution N data support differences in the patterns of snowmelt delivery and subsurface transport reported in Hinckley et al. [2012] and summarized in Figures 1 and 3: sustained transport through wet soils with well-connected flow paths through the soil matrix and higher soil moisture levels in north facing slope soils, versus wetting and drying cycles on the south facing slope following spring snow storms and lower soil moisture present at the onset of the summer months. These differences in the physical and hydrological conditions of the two slopes resulted in higher microbial N (and uptake of 15N into the microbial pool), as well as greater retention of 15N in the organic pool of north than south facing slope soils. On the south facing slope, uptake by annual grasses was greater than by P. contorta on the north facing slope, suggesting that N retention by vegetation may be a more important rapid-response mechanism slowing N movement on the south than the north facing slope. Our data indicate that south facing slopes quickly transport atmospherically derived N to depth during pulses of melt water movement, perhaps stimulating deeper soil microbial communities. In contrast, north facing slope soils retain N in surficial soil organic pools as sustained snowmelt moves through the soil matrix.

More broadly, the resulting behavior on both slopes suggests that these forests are subject to simultaneous transport and ecosystem uptake of N. Depending upon the degree of connectivity between the hillslope and the channel, our data indicate that the south facing slope may contribute more of the annual stream N than the north facing slope; thus, south facing slope soils may behave more like alpine systems with lower N retention and higher transport of inorganic N than north facing slope soils. Under future climate change scenarios, the south facing slope may also be a window into how north facing slopes will behave if winter precipitation becomes characterized by periodic rainfall events rather than by accumulating snow.


E.S.H. and R.T.B. were funded by NSF EAR Postdoctoral Fellowships (NSF EAR 0847987 and 0814457) during the study. E.S.H. also received funds from the CZO International Scholar Program (NSF EAR 0725019). Faculty, staff, and students associated with the Boulder Creek CZO (NSF EAR 0724960 and 1331828) provided additional support. We thank Chris Seibold and staff at the Niwot Ridge LTER's Kiowa laboratory for rapid, careful analysis of all samples. We also thank Alexandra Czastkiewicz, Daniel Eldridge, Hana Fancher, Zan Frederick, Abigail Langston, Christina Pruett, Nathan Rock, and the Arvada Colorado Boy Scout Troop 667 for assistance with sampling. Data included in this paper are available for download via the Boulder Creek Critical Zone Observatory website (