Nitrogen uptake by Carex aquatilis during the winter–spring transition in a low Arctic wet meadow

Authors


Correspondence author. E-mail: kate.edwards@utoronto.ca

Summary

1. Processes of decomposition occur year-round in tundra ecosystems and respond quickly to seasonal changes. Characterizing the phenology of plant nutrient uptake in relation to these processes is essential to understanding the current and future productivity of Arctic ecosystems.

2. In wet sedge meadows located near Churchill, Manitoba, Canada, soil microbial biomass as well as inorganic and organic nutrient pools fluctuate seasonally, with late-winter peaks followed by declines of these variables during the early stages of soil thaw; however, it is unknown if the dominant plant in this community takes up nitrogen when levels of this nutrient are high but soil temperatures are 0 °C or below.

3. Stable isotope tracing was utilized by injecting 15NH4Cl into soil cores and incubating for 1 or 8 days during spring thaw to determine the short-term capacity for uptake and transport of inorganic nitrogen into Carex aquatilis (roots, shoots and rhizomes), moss and soil micro-organisms during this transitional time of year.

4. During three 8-day experimental trials in April and May 2007, C. aquatilis roots accumulated a substantial amount of the added nitrogen (33.5% increasing to 63.4%), when inorganic nitrogen was readily available in the soil, but declining. A smaller proportion of injected nitrogen was recovered from soil microbes (30% decreasing to 7%), and only trace amounts of injected 15N were measured in plant shoots, shoot bases, rhizomes and mosses (2% or less).

5.Synthesis. Shifting seasonal patterns in northern ecosystems resulting from climate change are likely to alter the progression of events that lead up to the summer growing season. A substantial pool of inorganic nitrogen resides temporarily in the soil at the end of winter, and we have shown here that plants are able to take up nitrogen at this time. Increases in the frequency and temperature highs of late-winter warming events are likely to trigger early episodes of soil thaw, potentially reducing the capacity of plants to take up this large ephemeral supply of nitrogen in early spring.

Introduction

The growth of Arctic vegetation is constrained by short summer seasons, cold temperatures and low levels of available nutrients (Shaver & Chapin 1980; Nadelhoffer et al. 1991). However, while plant growth is generally restricted to summertime, soil microbial activity continues throughout the cold seasons, and nutrient transformations prior to the plant growing season contribute to a large pool of inorganic nitrogen (N) that is available for absorption by plants (Lipson, Schmidt & Monson 1999; Grogan & Jonasson 2003; Schmidt & Lipson 2004; Kielland et al. 2006). The timing of the onset of plant nutrient uptake and seasonal patterns of uptake are not well described for northern ecosystems, which are particularly vulnerable to changing seasonal patterns as winters in high-latitude environments become warmer and temperatures become more variable (IPCC 2007).

Several studies conducted in Arctic and alpine tundra sites have found that soil microbes reach their annual peak biomass in late winter (Brooks, Williams & Schmidt 1998; Lipson, Schmidt & Monson 1999; Grogan & Jonasson 2003; Schmidt & Lipson 2004; Edwards et al. 2006; Larsen et al. 2007). High levels of nutrients (N, phosphorus [P]) have also been observed during the winter–spring transition, either concurrently with (Edwards et al. 2006) or shortly after the microbial biomass (MB) decline (Schmidt & Lipson 2004), although Brooks, Williams & Schmidt (1998) did not observe a nutrient flush at either time. Springtime is known to be a crucial time for N uptake by plants in Arctic and alpine systems (Bilbrough & Caldwell 1997; Jaeger et al. 1999; Bilbrough, Welker & Bowman 2000; Grogan & Jonasson 2003; Schmidt et al. 2007); however, it is less well known whether vascular plant roots can take up N during the late winter and very early spring. Bilbrough, Welker & Bowman (2000) demonstrated, using a 15N tracer applied to the snowpack, that some plants in an alpine moist meadow took up large amounts of snowpack-N during snowmelt (over 12% of annual uptake for Deschampsia caespitosa), but in an Arctic site less than 1% of annual uptake in Eriophorum vaginatum happened during snowmelt. Tye et al. (2005) applied 15N tracer to high-Arctic tundra sites immediately after snowmelt to simulate a snowpack-derived N pulse and found significant uptake by bryophytes, MB and vascular plants after an incubation period of 19 days or more, with a large proportion (c. 40%) of tracer lost presumably due to leaching. This incubation extended well into the growing season, so it is difficult to know from these data whether vascular plants were taking up N early in the spring.

Wet sedge meadow soils near Churchill, Manitoba, Canada, are characterized by large MB and nutrient pools in late winter, which then decline during soil thaw and remain in low abundance throughout spring and summer (Edwards et al. 2006). We injected 15N-ammonium into soil cores and then incubated the cores for 1 and 8 days, to determine whether Carex aquatilis, a non-mycorrhizal sedge and the dominant vascular plant in this system, would take up inorganic nutrients during the winter–spring transition, when soil temperatures are below 0 °C and inorganic N is relatively abundant for a short period of time. We observed the partitioning of labelled N throughout different tissues of the sedge, in mosses and in the soil MB. Given our observations in the field of fine white roots in the frozen soil and of over-wintering shoots embedded in ice and snow (Tieszen 1972; Bell & Bliss 1977; Jonasson & Chapin 1985), we hypothesized that C. aquatilis would be able to take up nutrients very early in spring, prior to the onset of rapid above-ground plant growth.

Materials and methods

Site description

We sampled cores from a wet sedge meadow (calcareous fen mire) located at 58°44′ N, 93°48′ W, about 1 km east of the Churchill Northern Studies Centre, Churchill, Manitoba, Canada. The region is at the boundary between sub-Arctic and Arctic biomes and near to both boreal forest and the Hudson Bay shoreline. The vascular plant community is dominated by C. aquatilis Wahlenb. The soil is highly organic and about 30 cm deep, underlain by a mineral layer of silt and fragmented limestone, and permafrost at a depth of less than 1 m. Soils are water saturated throughout most of the year, except in late summer. Soil pH varies between 6.2 and 7.2, and oxidation–reduction potential (Eh) is typically between −50 and +150 mV at thaw (10 cm depth; platinum electrode with silver chloride reference, calibrated using Zobell’s solution). Total carbon (C) and N values are approximately 37% and 2.5% of soil dry weight respectively.

The depth of standing water varies across the site. Experimental cores were sampled from an area of approximately 20 m2, along the margin of a wet depression, so that soil cores were water saturated, but not submerged after thaw. This specific location differs somewhat from the wetter microsites normally utilized for long-term monitoring (Edwards et al. 2006), by having soil fresh-weight to dry-weight ratios (FW:DW) that are approximately 25% higher, a shallower moss layer (< 2 cm rather than 2–10 cm) and a c. 50% reduction in the maximum height of C. aquatilis.

Core sampling and processing

Core processing was modified from Hart et al. (1994). Cores with intact C. aquatilis shoots (2 or 3 shoots between 3 and 10 cm tall) were sampled using a custom made CRREL (U.S. Army Cold Regions Research and Engineering Laboratory) permafrost drill on 26 April 2007, and these frozen samples were thawed gradually by keeping them in a refrigerator (4 °C) for approximately 40 h. Small amounts of ice were still present in some cores, indicating that internal core temperatures did not rise significantly above 0 °C. On 8 May and 21 May 2007, we sampled thawed soils using a polyethylene tube and bread knife. These cores were stored at 4 °C for 20 h before injection.

Cores from 26 April were sorted into triplicate groups based on similarity in C. aquatilis size, presence of moss, amount of litter and colour and texture of soil. Each triplicate group, of which there were five in total, contained cores for 24- and 8-day incubations, and a core designated for immediate processing (to test the procedure for quantitative recovery, Buresh, Austin & Craswell 1982; herein referred to as a ‘time-zero’ core). Cores from subsequent sampling dates were sorted into matching pairs, one designated as the time-zero core and the other for 8-day incubation. Prior to injection, the cores, which were 7.4 cm in diameter, were cut to 7.5 cm depth and were wrapped in plastic wrap (covering the sides and bottom). Cores were injected with 7 mL of 1 mm15NH4Cl (Isotech Laboratories Inc., Champaign, Illinois, USA; > 98 atm.%15N) in a hexagonal + centre pattern, using a 15-cm double-sideport spinal syringe (Popper and Sons, Hyde Park, New York, USA). The cores were inverted, and the needle was inserted into the bottom of the core and then ejected gradually while spinning the needle and withdrawing it. The total amount of N injected into each core was 105 μg, theoretically increasing NH4+-N levels by approximately 10%, which is well within the natural variability of NH4+-N for these soils. Time-zero cores were processed within 30 min of injection, while 24-h incubations were placed in the refrigerator (4 °C) and 8-day incubations were put in small, clear plastic Ziploc (S.C. Johnson & Son Inc., Racine, WI, USA) bags and planted back into the ground at the sampling site. Bags were kept open such that cores in the field were exposed to the air but somewhat protected from surface runoff and precipitation. Core temperatures were checked periodically during the final incubation period when air temperatures rose above 20 °C. The cores were not more than 1 °C warmer than the surrounding soil, even near the surface, so the Ziploc bags were effective at protecting the cores without causing significant soil warming through greenhouse-like conditions.

We processed cores by first removing and sorting live and dead shoot, shoot base, rhizome and root material, and collecting any living moss. Plant material was rinsed in calcium rich water to remove any 15N from root surfaces and was dried at 50 °C for at least 48 h before weighing for dry mass estimation. After the removal of plant tissues, each soil core was broken up and hand-mixed in a large Ziploc bag and portions were weighed for MB extractions and for FW:DW measurements (c. 10 g fresh weight, dried at 50 °C for a minimum of 48 h).

Microbial biomass extractions followed a chloroform fumigation and extraction method modified for wet soils (Witt et al. 2000; Henry & Jefferies 2003) using 25 g fresh soil in 50 mL 0.5 m K2SO4, shaken frequently for 1 h, then filtered through pre-leached Whatman 1 filter papers (Whatman International Ltd, Kent, UK). For the chloroform-fumigated fraction, 2 mL ethanol-free CHCl3 was applied directly to the soil and incubated in sealed jars for 24 h, followed by K2SO4 extraction. Extracts were kept frozen until further analysis. Microbial biomass C and N were determined as the difference between fumigated and unfumigated organic C and organic N in K2SO4 extracts. Organic C was determined using the potassium dichromate volumetric method (Nelson & Sommers 1996; Henry & Jefferies 2003) and the same extracts were used to determine total N using chemiluminescence (TOC-TN autoanalyzer; Shimadzu, Kyoto, Japan).

Isotope sample preparation

Dried plant tissues and soil were ground using a mortar and pestle, and weighed into tin capsules (2 ± 0.25 g for plant tissue, 0.7 ± 0.1 g for soil) for analysis by mass spectrometry (EA-IRMS; University of Waterloo Environmental Isotope Lab, Waterloo, ON, Canada). Actual mass values accurate to 0.001 g were used for delta-15N calculations of each sample. Total plant N per core was determined from total mass values of each sample and tissue N concentrations from mass spectrometry analysis.

K2SO4 extracts for microbial 15N were oxidized by alkaline persulphate digestion (Cabrera & Beare 1993) followed by a modified diffusion technique for recovering 15N (Stark & Hart 1996). Sample extracts were oxidized (autoclaved after mixing with alkaline persulphate solution) and then poured into a 250-mL glass jar along with 2 mL 10 m NaOH and 20 mL alkaline persulphate blank solution. Jars were left open with daily stirring for 3 days and replacement water was added to adjust for evaporative loss. A carrier solution of 0.7143 m KNO3 (5 μL) was then added to jars, along with 0.3 mL 10 m NaOH, 0.4 g Devarda’s alloy, and an acid trap that consisted of two pre-leached filter paper discs (7 mm diameter; Whatman no. 1) each containing 5 μL 2.5 m KHSO4, sealed in a strip of polytetrafluroethylene tape. Jars were sealed and diffused for 6 days, mixed daily; then traps were removed, rinsed in water, and filter discs were dried in a dessicator. Discs were weighed and folded into tin capsules, and these were analysed for isotopic composition by mass spectrometry (EA-IRMS; University of California, Davis Stable Isotope Facility, Davis, CA, USA).

Isotope calculations

Delta-15N values were converted to atm.% excess (Shearer & Kohl 1993), which is the level of 15N enrichment beyond the standard of 0.3663 (15N natural abundance of atm.-N). Calculations of 15N recovered in plant and soil pools were then calculated using the following equation (Powlson & Barraclough 1993), where F represents N recovered from the labelled addition (μg N g−1 tissue), T is the total weight of N in the sample tissue (μg N g−1 tissue), As is the atm.% excess 15N in the treated sample, AB is the atm.% excess 15N in the unlabelled control and AF is the atm.% excess 15N in the label:

image

The percentage of label-N that was recovered in each pool was calculated by multiplying F by the mass of tissue in the core and dividing by the mass of N injected into the core.

To calculate microbial immobilization of added N, atm.% excess values were corrected for carrier dilution (Powlson & Barraclough 1993) and then corrected using a calculated blank (Stark & Hart 1996) in which the difference between diffused and non-diffused isotope standards is used to account for incomplete nitrate recovery during diffusion. Finally, atm.% excess of background samples (no added 15N) was subtracted from corrected sample atm.% excess values, and the resulting sample atm.% excess was multiplied by the total microbial N pool size and the core volume, and then expressed as a percentage of the total injected label-N. Microbial N values were not corrected using a KN factor that would account for the proportion of microbes unsusceptible to chloroform fumigation, so the values presented for microbial 15N immobilization may be underestimates.

Seasonal measurements of soil microbial biomass and inorganic nitrogen

Soils were sampled periodically in 2007 (monthly to twice weekly, throughout the year) from the sedge meadow described above, and from a similar wet sedge area about 5 km west as part of a long-term monitoring project (Edwards, unpubl. data, but see Edwards et al. 2006 for description of methods and sites). In most cases, six replicates were sampled per sampling date, using a CRREL permafrost drill or axe when frozen, or a bread knife when the soil was thawed. Depth of soils sampled and soil processing were the same as described above for 15N experimental cores. Microsites differed slightly, as described previously, with long-term monitoring samples often taken from areas with considerable standing water. Soils collected in the frozen state were kept frozen and processed within 1 day of sampling. Extractions using frozen soils were done by chopping the sample into small pieces (1 cm3 or smaller) and allowing 15 extra minutes of extraction time to account for thawing during extraction. Microbial C was quantified as described above and results were not corrected using a KC factor. Thus, the MB results reported here may underestimate the true microbial component of C in the soils.

To quantify soil exchangeable ammonium (NH4+), 10 g of fresh soil was mixed with 50 mL of 1 m KCl and shaken frequently for 2 h, followed by filtration through pre-leached Whatman no. 1 filter papers. Extracts were stored frozen and NH4+ levels were then determined using an autoanalyser (Technicon AAII, Tarrytown, NY, USA). Microbial biomass and NH4+ levels are expressed on a per-volume basis, because the 15N experiment was conducted with cores of consistent soil volume, and because we are most interested here in plant root uptake.

Seasonal measurements of soil temperatures

Soil temperature readings were taken in a nearby site that is similarly wet and sedge dominated, but also contains some shrubs (G. P. Kershaw, pers. comm.; see Edwards et al. 2006 for details on temperature monitoring). Soil temperatures at this site correlate well with measurements recorded from October 2008 to June 2009, from sensors that are located within the 15N sampling area, and integrate between 5 and 10 cm soil depth (K. A. Edwards, data not shown). Thus, the soil daily minimum and maximum temperature data reported here are not specific to the site from which the samples came, but provide a reasonable proxy for the conditions at the experimental site.

Statistics

Root biomass, shoot biomass, MB and total plant N from experimental cores were fitted to simple linear and linear spline regression models. Normality was checked using the Shapiro–Wilks test, and variables were square-root transformed as necessary. The absence of curvilinearity and heteroscedasticity in residual plots was verified by visual inspection. Two spline regression models were evaluated which differed in the placement of single pre-determined knots (6 May and 9 May). Linear and spline models were compared based on adjusted R2 values, error terms and P-values, and for each variable the model of best fit is presented.

Both 1- and 8-day incubations were carried out for the first experimental trial, and a Student’s t-test was used to detect differences in 15N uptake between the two incubation times, using α = 0.05. One-tailed tests were utilized to detect increases in microbial and plant 15N accumulation and a decrease in soil 15N accumulation over time. Normality and homogeneity of variances were verified (Shapiro–Wilks test, and both Levene and Bartlett’s tests, respectively) and variables were transformed as necessary.

Accumulation of label-N in microbial and plant pools after 8 days of field incubation was evaluated using simple linear regression. Assumptions for regression analysis were checked as outlined above and data was transformed when necessary. All analyses were computed with jmp 7.0 (SAS, 2007).

Results

Timing of the experiment in relation to soil thaw, microbial biomass and inorganic nitrogen

Within the larger study site, soil inorganic N levels declined over the period of the isotope experiment from 27 April through May 2007, during the time that soil temperatures reached and rose above 0 °C (Fig. 1a). Microbial C levels also declined over this time (Fig. 1b). By the end of the experiment in June 2007, NH4+ and MB levels had reached low levels that are typical of this system during the summer months (Edwards et al. 2006). This decline corresponds with the onset of above-zero soil temperatures and steep increases in the amount of liquid water in the system, which was observed in the field beginning on 2 May 2007, resulting in the presence of standing water at the site. Snow melt and soil thaw continued at the site as air temperatures warmed, but cold events occurred on two occasions in May, most notably beginning 16 May when fresh snow covered the site, which had previously been snow free, and the surface ground refroze for a couple of days.

Figure 1.

 Levels of (a) ammonium and (b) microbial biomass carbon (mean ± SE; n = 5) in wet sedge meadows near Churchill, Manitoba, Canada, during April and May 2007. Vertical lines are daily maximum–minimum soil temperature intervals as measured at 10 cm depth. The three periods of time shown by horizontal brackets indicate the timing of three experimental trials, each lasting 8 days.

Inorganic nitrogen uptake into microbial and plant pools

Time-zero cores were processed to verify the recovery of added 15N in the soil prior to incubation (Buresh, Austin & Craswell 1982). Recovery of 15N from the soil was usually between 70% and 100%, but some samples were outside of this range, probably reflecting inadequate mixing of soils before partitioning for extractions. Microbial recovery in time-zero cores tended to be unexpectedly high, close to 10% on average. This could also be the result of inadequate mixing, or rapid uptake by microbes may have taken place (within 30 min of injection), exacerbated by the disturbance of soil mixing, which could stimulate microbial activity. Significant levels of 15N were not detected in plant and moss tissues from time-zero cores.

Recovery of injected-N was greatest in soil, root and microbial pools (Table 1). One-tailed t-tests did not detect differences between the 24-h and 8-day incubation periods (sampled on 21 April) of 15N pools in microbes (t8 = 0.40, = 0.351) or roots (t8 = 1.09, = 0.15); however, soil 15N recovery was lower after 8 days of incubation (mean = 25.5%) than after 24 h of incubation (mean = 76.2%; t8 = −1.90, = 0.047). This is likely an artefact of poor soil mixing and sampling for isotope measurement, as the two subsequent 8-day incubation trials resulted in soil 15N recovery that was more similar to the 24-h incubation (means of 78.0% and 53.6%).

Table 1.   Percentage recovery of added N into soil (excluding microbial), microbial, various Carex aquatilis and moss pools. Means and minimum–maximum intervals (n = 5) are shown for each incubation period. All cores were sampled from a wet sedge meadow near Churchill, Manitoba, Canada, in April and May 2007
Sampling date28 April25 April9 May22 May
Incubation time1 day8 days8 days8 days
Soil76.2 (13.5–137.9)25.5 (2.6–74.9)78.0 (61.9–105.6)53.6 (18.5–97.6)
Microbes25.7 (2.3–54.0)30.4 (15.5–52.0)15.6 (6.9–24.1)7.7 (4.1–13.2)
Roots21.0 (11.3–30.0)33.5 (13.3–65.6)40.7 (15.8–69.1)63.4 (15.8–37.3)
Rhizomes1.5 (1.0–2.7)2.1 (0.8–4.2)1.2 (0.0–4.3)0.8 (0.0–0.3)
Shoot bases1.7 (0.0–7.3)2.1 (0.0–3.8)0.7 (0.3–1.4)0.6 (0.3–0.1)
Dead shoots0.0 (0.0–0.0)0.0 (0.0–0.1)0.0 (0.0–0.0)0.0 (0.0–0.0)
Live shoots0.0 (0.0–0.0)0.0 (0.0–0.1)0.2 (0.0–0.2)0.9 (0.4–2.1)
Mosses0.0 (0.0–0.0)0.0 (0.0–0.1)0.0 (0.0–0.0)0.0 (0.0–0.0)

Root N-uptake appeared to increase over the entire experiment, with the mean label-N reaching 63% after the final 8-day incubation period (Fig. 2; R2 = 0.17, F1,13 = 3.78, = 0.074). Shoot 15N was a smaller component of the total label-N recovery, reaching a mean of only 0.9% during the final 8-day trial, although this increase represents an upward linear trend (R2 = 0.71, F1,13 = 34.70, < 0.001). Recovery of label-N from the microbial N pool declined over the experiment from a mean of 30.4% in the first trial to 7.7% after the final trial (Fig. 2; R2 = 0.60, F1,13 = 22.26, < 0.001). Small amounts of tracer N were measured in rhizomes, shoot bases and shoots of C. aquatilis, while mosses took up only trace amounts of the N label (Table 1).

Figure 2.

 Recovery of injected 15N-ammonium in Carex aquatilis roots and soil microbes in experimental soil cores collected on three occasions in April and May 2007 from wet sedge meadows near Churchill, Manitoba, Canada, and incubated in the field for 8 days. The proportion of label-N recovered from microbes decreased over the three trials (R2 = 0.60, < 0.001) while root label-N uptake showed a marginally insignificant increasing trend (R2 = 0.17, = 0.074). Untransformed means and standard errors are plotted (n = 6) with linear regression lines.

Plant biomass, plant N and microbial biomass in experimental cores

The biomass of C. aquatilis roots in the experimental cores increased throughout the experiment (Fig. 3; linear regression R2 = 0.53, F1,40 = 47.43, < 0.001). Shoots from the same plants best fit a linear spline regression with a knot designated on 9 May, after which linear growth was evident (Fig. 3; linear spline regression R2 = 0.15, F2,39 = 4.64, = 0.012). Mean root : shoot ratios therefore increased from less than 15 to greater than 20 over the time of the experiment.

Figure 3.

 Biomass of shoots and roots from experimental soil cores processed between 28 April and 30 May 2007 that were sampled on three occasions in April and May 2007 from a wet sedge meadow near Churchill, Manitoba, Canada, and then incubated for 0, 1 or 8 days before plant tissue harvesting. Root biomass increases linearly over the period (R2 = 0.53, < 0.0001) while shoot biomass is best fitted to a spline regression model containing a knot at 9 May (R2 = 0.15, = 0.016). Untransformed means and standard errors are plotted (n = 6) with linear regression lines.

Total plant N from experimental cores increased over the experiment from a low of 0.6 mg core−1 on 29 April to 2.2 mg core−1 on 30 May (Fig. 4; linear regression R2 = 0.55, F1,40 = 50.89, < 0.001) revealing that in the order of 1.6 mg N was taken up over a 33-day interval. Microbial biomass levels in these cores decreased from above 300 mg microbial C m−3 to below 200 mg m−3 during the experiment (Fig. 4; linear regression R= 0.35, F1,40 = 22.75, < 0.001). This decline was similar to that documented for the larger site, but the magnitude of the decline was dampened, and the absolute values were higher in experimental cores, relative to other areas of the site, at least on a per-volume basis.

Figure 4.

 Microbial carbon and total plant nitrogen from experimental soil cores processed between 28 April and 30 May 2007 that were sampled on three occasions in April and May 2007 from a wet sedge meadow near Churchill, Manitoba, Canada, and then incubated for 0, 1 or 8 days before plant tissue harvesting. Microbial C (open diamonds) declined over the period (R2 = 0.35, < 0.001) while plant N (black diamonds) increased (R2 = 0.55, < 0.001). Untransformed means and standard errors are plotted (n = 6) with linear regression lines.

Control cores (injected with water instead of 15NH4+) were not different than treatment cores in terms of plant biomass, plant N concentration or MB (data not shown), suggesting that the injection of NH4+ did not produce a fertilization effect and that our results are not due to the small increase in inorganic N provided by the isotope tracer.

Discussion

Carex aquatilis took up large proportions of added 15N-ammonium over short-term incubations during the seasonal transition between winter and spring, when snowmelt was well underway but the soil was still partially frozen at the site. Carex aquatilis is not known to be mycorrhizal (Muthukumar, Udaiyan & Shanmughavel 1994), so N-uptake reported here is not likely to be attributable to fungal associations.

The ability of plants to take up nutrients at or below freezing temperatures is well known. Billings et al. (1977) demonstrated that C. aquatilis and Eriophorum angustifolium roots in Barrow, Alaska, USA, could grow at temperatures below 1 °C, and roots of E. angustifolium grew along the interface between the active layer and frozen soil. Further, roots from these two species as well as Dupontia fischeri that were frozen for several days were able to recover to normal root elongation rates after being thawed for 12 h. More recently it has been observed that Corydalis conorhiza, an alpine herb, has snow roots that are specifically adapted to take up N in the snowpack (Onipchenko et al. 2009).

We have demonstrated here that C. aquatilis roots can acquire significant quantities of inorganic N within a few hours of soil thaw. As we thawed soils prior to treatment, we cannot be sure that N-uptake by plants occurred earlier than this; however, it is quite likely that plants can take up nutrients during soil thaw, when channels of liquid water form around plant roots, accelerated by early snowmelt around standing dead plant material (caused by lowered albedo). Additional heat produced by increased plant and microbial metabolic activity in the rhizosphere may also contribute to this localized warming, providing suitable microsite conditions for nutrient uptake by plant roots even when much of the bulk soil water is frozen.

Carex aquatilis accumulated more NH4+ during the winter–spring transition than did soil microbes, in contrast to the prevailing wisdom that over short-term intervals (days to weeks) soil microbes outcompete plants for N (Jackson, Schimel & Firestone 1989; Zak et al. 1990; Lipson, Schmidt & Monson 1999; McFarland, Ruess & Kielland 2002; Bardgett, Streeter & Bol 2003; Nordin, Schmidt & Shaver 2004; Andresen & Michelsen 2005; Grogan & Jonasson 2005; Harrison, Bol & Bardgett 2007; Nasholm, Kielland & Ganeteg 2009). However, some studies have found plant uptake of 15N-labelled N to be comparable to net microbial immobilization (Norton & Firestone 1996; Lipson & Monson 1998), and in an Arctic salt marsh just east of our experimental site, 29% of injected NH4+ was recovered in the dominant graminoid, Puccinellia phryganodes, while 34% was immobilized by soil microbes (using a KN correction, for incomplete microbial recovery, of 0.35; Henry & Jefferies 2003).

Previous authors have suggested that over the long term, plants may compensate for their short-term competitive disadvantage by maintaining the ability to acquire nutrients throughout the year (Andresen & Michelsen 2005). This view is supported by studies that show decreasing microbial sink strength during experiments lasting several months to 1 year (Hart et al. 1993), and by experiments demonstrating the importance of pulsed nutrient supplies as compared with steady supplies of nutrients for spring plant growth (Bilbrough & Caldwell 1997). Our study lends further support to the view that ephemeral nutrient supplies are an important component of the annual N budget of at least some plants.

Several features of the late winter and early spring period in wet sedge meadows make this a favourable time for plant nutrient uptake, namely, large pools of nutrients, moderately flowing water and weakened microbial populations. Nutrient pools are at or near annual peaks at this time of year (Lipson, Schadt & Schmidt 2002; Nemergut et al. 2005; Edwards et al. 2006; Buckeridge & Grogan 2008). The presence of moderately flowing water during spring snow melt facilitates nutrient mobility, increasing nutrient flux to root surfaces (Chapin et al. 1988). Moving water may also temporarily increase levels of dissolved oxygen in sites that are ice-bound in winter and water-logged throughout much of the summer. However, in some cases spring runoff may result in significant N losses, particularly when large volumes of water are flowing rapidly, where nitrate-N is abundant, and where N is deposited on soil surfaces, i.e. by atmospheric deposition (Joseph & Henry 2008, 2009). With soil microbes in decline during the transition from winter to spring, the competitive ability of soil microbes for nutrient acquisition may be lower than it would be during the rest of the growing season, when turnover of N by microbes is relatively high despite low MB (Buckeridge & Jefferies 2007). However, data from a Carex bigelowii– dominated heath site in Scotland suggested that microbes in spring could be strongly N-limited, and therefore be intense competitors for N, despite low MB at this time of year (Bardgett et al. 2002).

Our results reveal that about 1.6 mg of N was taken up in 33 days by the two or three small C. aquatilis ramets contained in each experimental core. These ramets would each accumulate about 15 mg of N by the end of the growing season (Edwards, unpubl. data), so this early season uptake represents on the order of 5% of the total N for the season. This initial gain early in the season would be important given the rapid growth that follows once temperatures warm. Daily minimum soil temperatures at the end of this experiment were still hovering just above zero (Fig. 1), so the time of greatest nutrient uptake likely followed soon after, with maximal growth and tissue N accumulation expected to occur during the first half of the growing season (Jaeger et al. 1999). The 5% estimate is lower than the mean 12% of annual N uptake reported for alpine tundra graminoid species during snowmelt (Bilbrough, Welker & Bowman 2000), but is much higher than estimates reported by the same authors for Alaskan tundra species (< 0.1%). However, Bilbrough, Welker & Bowman (2000) applied 15N to the snowpack, so it is unclear whether differing results between the two studies are due to methodological differences or due to biological differences in ecosystems, sites or plant species.

Although spring thaw is a favourable time for N uptake by C. aquatilis, we observed that the inorganic N that was taken up remained in the roots and was not translocated in significant amounts to other plant tissues such as the rhizome, shoot base or shoot. An increasing amount of N was translocated to shoots throughout the experimental period, but this was a small amount relative to what remained in roots, even after shoot development and growth were observed during the final experimental trial. This is in contrast to several studies conducted later in the growing season that found high rates of 15N-tracer allocation to graminoid shoots (Schimel & Chapin 1996; Grogan & Jonasson 2003; Sorensen et al. 2008) and 32P (Jonasson & Chapin 1991), suggesting that N acquired in early spring is used for root growth, and/or stored in roots for weeks or even months, and that initial shoot growth in the spring relies on N taken up the previous year and stored in rhizomes over the winter (Jaeger & Monson 1992).

Recovery of 15N from injected cores was high, often totalling well over 100%. This is not unexpected or uncommon (Finzi & Berthrong 2005; Clemmensen et al. 2008) given the large volume of soil injected and the difficulty of mixing soils thoroughly and quickly. Based on this, we believe that soil and microbial pools are more prone to error than are plant tissue pools, which are more completely homogenized prior to subsample collection for isotopic collection.

The winter–spring transition is recognized as an important time of year in the functioning of northern ecosystems and especially in understanding the consequences of broad-scale environmental disturbances including climate change. This time of year is marked by fluctuations in temperature, snow depth and the physical state of soil water (Jefferies et al. 2010), and the phenology of events below ground that impact the summer growing season are likely to be altered in the north as winters become warmer and snow fall increases (IPCC, 2007). The effects of these changes on plant productivity and nutrient cycling are unknown. It is possible that as springtime arrives earlier in the year, nutrient flushes that result from late-winter microbial die-off and spring meltwater discharge will occur before some plants are metabolically ready to utilize resources, and a large proportion of the annual N supply to Arctic plants could be lost through microbial immobilization, gaseous efflux and/or leaching. Alternatively, some plants may respond to these warming events by increasing uptake of N (Turner & Henry 2009), enhancing primary productivity and carbon sequestration potential. However, if plant nutrient uptake and growth is initiated during a winter warming event followed by subsequent freezing, development and reproduction during the growing season that follows can be dramatically compromised (Bokhorst et al. 2008, 2009; Joseph & Henry 2008). Further monitoring and experimental work, including during the fall and spring shoulder seasons, is needed to properly anticipate the effects that shifting climate patterns will have on soil nitrogen cycling and resultant plant productivity.

Acknowledgements

This work was funded by the Natural Sciences and Engineering Research Council of Canada through research grants to R.L.J. and a postgraduate scholarship and northern research internship to K.A.E., the Northern Scientific Training Program and the Churchill Northern Studies Centre (CNSC). This research is part of IPY WOLVES supported by IPY Canada and NSERC. We thank G.P. Kershaw for sharing soil temperature data with us, F. Forsythe, E. Horrigan, C. Jeon, A. Simonsen and A. Xu for assistance in the field and lab, L.A. Fishback and C. Basler (CNSC) for logistical and field help, D. Tam (University of Toronto) and L. Cameron (Queens University) for technical support and sample analysis, and labs at both University of Waterloo and University of California, Davis, for isotopic determination of samples. K.A.E. is grateful to H.A.L. Henry and two referees for helpful comments on earlier versions of the manuscript, and thanks R.L.J. for sharing a lifetime of wisdom.

Ancillary