STUDY SITE

The study area was located in a low arctic tundra region at the Tundra Ecosystem Research Station, Daring Lake, Northwest Territories (64°52′N, 111°35′W, 414–470 m a.s.l.) (Fig. 1), approximately 90 km northeast of the northern limit of continuous trees within the physiographic zone of the Bear–Slave Upland of the Canadian Shield (Obst, 2008). Landscape features include eskers, boulder fields, exposed bedrock, upland and lowland tundra, wetlands, and various sizes of lakes, ponds, and streams. Continuous permafrost is present at the site with a soil active layer ranging from 0.3 to 2 m (Obst, 2008).

FIGURE 1

Location of the study site at Daring Lake, Northwest Territories, Canada (64°52′N, 111°35′W) (adapted from  http://www.enr.gov.nt.ca/_live/pages/wpPages/Tundra_Ecosystem_Research_Station.aspx).

Mean monthly air temperature is −30 °C in January and +13 °C in July (INAC, 2007; Obst, 2008). Mean monthly precipitation from May to October is 25 mm as rain. Snow accumulation is highly variable across the landscape, but usually ranges from 15 to 60 cm in low-lying heath vegetation by mid to late May (1996–2008; Bob Reid, INAC, unpublished data). Snowmelt usually starts after mid-May ending in early June, with some snow beds persisting on slopes until late June or early July. The plant growing season generally begins in late May or early June and ends by late August (Nobrega and Grogan, 2008; Lafleur and Humphreys, 2008).

The landscape study area encompasses the East Daring Lake Basin (26.7 km²). Ecosystem mapping and distribution of landscape units for the landscape study area follows Obst (2008). A 1-m resolution IKONOS image provided detailed information on 15 classes (plus unclassified areas) of land covers, vegetation communities, and ecosystem types present in the study area (Obst, 2008). We focused our study on the dominant ecosystem types that together occupy a total of 68% of the study area: Heath-Lichen/Heath-Mat Tundra (42%), Birch Hummock (13%), Wet Sedge Meadow (8%), and Xerophytic Herb Tundra (5%). The distribution of these four ecosystem types is largely driven by esker topography (Table 1).

TABLE 1

The principal topographic position, substrate, drainage, and characteristic plant species for each ecosystem type included in the landscape study at Daring Lake, Northwest Territories, Canada (follows Obst, 2008).

N₂-FIXING ASSOCIATIONS

Four predominant cyanobacterial associations were identified within the selected landscape study area at Daring Lake: Biological Soil Crusts (BSCs) in hollows, BSCs on mineral soil mounds, Sphagnum spp., and Stereocaulon paschale sensu lato. Each cyanobacterial association was found in all of the ecosystem types included in the landscape study; however, the abundance of each association varied between ecosystem types. Vascular plant species with N₂-fixing associations, such as Oxytropis nigrescens (Pall.) Fisch. ex DC., and Alnus crispa (Aiton) Pursh., do occur in the area but are rare (Obst, 2008).

Two major BSC communities were found in association with hummock-hollow microtopography. Hollow BSCs were found mainly in the depressions between hummocks and were composed of liverworts growing in dense mats generally underlain by varying depths of organic matter. The main components of Hollow BSCs were Anastrophyllum minutum Schreb. and Cephalloziella spp. including C. rubella (Nees) Warnst. and C. hampeana complex (O. Lee, unpublished data). Cyanobacteria on Hollow BSCs were mostly the filamentous and heterocystous cyanobacterium Stigonema cf. turfaceum (Berk.) Cooke (B. Büdel, unpublished data). However, on some samples filamentous and heterocystous Tolypothrix sp. and the filamentous, non-heterocystous Schizothrix cf. cuspidata W. et G.S. West, were found growing in between the leafy liverworts and on the Stigonema filaments. Stigonema minutum (C. Agardh) Hass. and Calothrix sp. were also found on Hollow BSC samples. Hummock BSCs were cohesive, well-developed crusts (1–2 cm thick) found on cryoturbated mineral soil mounds. Small, less well-developed patches of Hummock BSCs also occurred in sandy well-drained areas on ridge tops. Hummock BSCs were complex communities made up of lichens, mosses, and liverworts. Lichen species included Placynthiella uliginosa Schrader., Bryocaulon divergens Ach., Bryoria tenuis (E. Dahl) Brodo & D. Hawksw., Cladonia spp., Japewia tornoensis Nyl., Ochrolechia frigida Sw., and Solorina crocea L. (C. Bjork, unpublished data). Moss species (Funaria sp., Pohlia sp., Ditrichum sp., and Polytricum piliferum Hedwig.) and liverwort species (Cephalozia sp., Cephaloziella sp., Anastrophyllum sp., Anthelia sp., Lophozia sp., and Lophozia incise Schrad.) were also key components of these diverse communities. Stiognema turfaceum, S. minutum, and S. hormoides (Kutz.) Born. & Flah. dominated Hummock BSCs; however, Gloeocapsa decorticans A. Braun., G. novacekii (Komárek & Anagnotid), S. cuspidata, Anabaena sp., and Chroococcidiopsis sp. were also present.

Sphagnum spp. were the dominant ground cover in Wet Sedge Meadows and were found scattered in damp depressions throughout the landscape. The majority of Sphagnum spp. samples were composed of Sphagnum aongstroemii C. Hartm. and S. subsecundum complex, with occasional fine strands of S. balticum (Russ.) C. Jens. (O. Lee, unpublished data). In addition, other moss (Drepanocladus aduncus (Hedw.) Warnst.) and liverwort (Gymnocolea inflate Huds.) species were found intermingled within Sphagnum spp. samples. The cyanobacteria G. decorticans was found in association with Sphagnum spp. samples (B. Büdel, unpublished data).

Stereocaulon paschale was predominantly found in small continuous mats on high/mid-slope positions, often in areas where late-lying snow patches occurred. Patchy distribution of S. paschale also occurred on well-drained ridge tops and in hummock-hollow complexes.

N₂-FIXATION RATES

Measurements of N₂-fixation were made using acetylene reduction assays (Stewart et al., 1967). Acetylene gas (C₂H₂) was generated on-site from CaC₂ and water, with incubations injected with 10% (v/v) acetylene. Ethylene concentrations were measured in the field with a portable gas chromatograph (SRI 8610A, Wennick Scientific Corporation, Ottawa, Ontario, Canada) fitted with a Porapak column (Alltech Canada, Guelph, Ontario, Canada) and a flame ionization detector. A Stand-Alone Hydrogen Generator (SRI H₂-50, Alltech Canada, Guelph, Ontario, Canada) provided hydrogen as the carrier gas, which was held at a constant pressure of 26 psi. Column temperature was held at 65 °C. The gas chromatograph was calibrated for each incubation with ethylene (BOC Canada Ltd., Mississauga, Ontario, Canada; C₂H₄, 98%+) that was kept at the same temperature as incubation gas samples.

Cores of Hollow BSCs (n  =  12), Hummock BSCs (n  =  12), and S. paschale (n  =  12) were randomly sampled for each incubation from an area of ∼5 km² that was representative of the larger landscape study area. Samples were taken from multiple positions on both north- and south-facing slopes of the main east-west-oriented esker. BSC samples were trimmed to an area of 19 cm² and 0.75 cm depth such that each sample had a thin underlying soil substrate. Stereocaulon paschale was trimmed to an area of 19 cm² and 2 cm depth, but no underlying soil substrate was included. Samples were enclosed in 250 mL glass canning jars with modified lids containing a rubber septum. The mean headspace of ARA incubations was 235.75 mL (250 mL jar volume minus 14.25 mL sample) for Hollow and Hummock BSCs and 212 mL for S. paschale (250 mL jar volume minus 38 mL sample). Sphagnum spp. cores (n  =  12) were sampled from an area of ∼0.5 km² within the Wet Sedge Meadow ecosystem type only. Samples were trimmed to an area of 56 cm² and 6 cm depth and included both live (green) and underlying decaying stems. Sphagnum spp. samples were incubated in 1 L canning jars with modified lids containing rubber septa and had a mean headspace of 664 mL (1000 mL jar volume minus 336 mL sample). All Sphagnum spp. jars were incubated in situ in the Wet Sedge Meadow with the glass bottom facing up and the Sphagnum spp. sample level with the surrounding vegetation. For each set of incubations one sample for each N₂-fixing association was used as a control, which served as both a temperature control and a blank not injected with acetylene. Control samples did not show any natural evolution of ethylene. Contamination of generated acetylene with ethylene was monitored and corrections were made for each set of incubations, as required.

Daytime ARA incubations occurred between 10:00 and 16:00 hr (6 hours) and nighttime ARA incubations between 21:00 and 7:00 hr (10 hours). A pilot study conducted in 2007 indicated net respiration in both Hollow BSCs (108 µL L⁻¹ CO₂/hr) and Hummock BSCs (162 µL L⁻¹ CO₂/hr) under average light conditions, suggesting that CO₂ limitation was unlikely to limit N₂-fixation despite longer incubation periods. However, we injected the S. paschale incubations with 1% (v/v) CO₂ after 3 hours for daytime incubations and after 1 hour for night-time incubations because the lichen samples lacked an underlying soil substrate to provide a CO₂ source.

Destructive sampling was used for each incubation, with new samples of each cyanobacterial association (n  =  12) collected per incubation. Nine consecutive sets of in situ incubations under ambient field conditions were conducted over a 6 day period (5 nighttime and 4 daytime) in each growing season month for each N₂-fixing association, with the exception of Sphagnum spp. in 2007. Incubations for Hollow BSCs and Hummock BSCs were conducted on 19–24 June, 6–11 July, and 9–13 August in 2007; and on 12–17 June, 1–6 July, and 5–10 August in 2008. Sphagnum spp. were incubated for a 24 hr period over 5 consecutive days on 25–29 June, 9–13 July, and 17–22 August in 2007 due to logistical constraints. In 2008, Sphagnum spp. were incubated in the same manner as other N₂-fixing associations on 18–23 June, 7–12 July, and 17–22 August. Sphagnum spp. N₂-fixation rates were not significantly different between 2007 and 2008; therefore, the difference in incubation length likely had little influence on the overall rate estimation. Stereocaulon paschale was incubated only in 2008 on 4–9 June, 7–12 July, and 11–16 August. Over the 2007–2008 growing seasons a total of 571 (Hollow BSCs), 572 (Hummock BSCs), 794 (Sphagnum spp.), and 294 (S. paschale) individual samples were incubated in situ under ambient field conditions.

With the exception of Sphagnum spp., all in situ samples were incubated outdoors near the research station laboratory under ambient field conditions. Incubation chambers were placed in water baths and bath temperature was altered to ensure that incubation temperatures reflected ambient conditions. Photosynthetically Active Radiation (PAR), air temperature, incubation temperature, and ambient temperature of Hollow BSCs, Hummock BSCs, and S. paschale were monitored every 30 minutes during daytime ARA incubations. On average, the surface temperature of incubation samples were within 1.5 °C of the surface temperature of the respective N₂-fixing associations under ambient conditions. Heating of incubation chambers via solar radiation was not a concern for nighttime incubations where microclimate was monitored for the first and last hour only. Moisture of Hollow BSCs, Hummock BSCs, and S. paschale were determined both pre- and post-incubation to ensure that drying of specimens did not occur during the incubation period. Average loss of moisture during incubations was less than <1.8% for all N₂-fixing associations. Following incubation, all samples were weighed, air dried, and then re-weighed to determine moisture content. Moisture content of samples over the growing season was later used for modeling N₂-fixation potential.

In addition to in situ incubations under ambient field conditions, N₂-fixing associations were also incubated in situ under optimal environmental conditions (200 µmol PAR m⁻² s⁻¹, 20 °C) at the end of each set of incubations in June, July, and August 2007/2008. For each of our N₂-fixing associations we likely had several different cyanobacterial species present with varying optimal operating environments; however, an optimal temperature of ∼20 °C has been demonstrated for several species/environments (Basilier and Granhall, 1978; Chapin et al., 1991; Liengen, 1999a), and light saturation was been demonstrated at ∼100 µmol PAR m⁻² s⁻¹ (Coxson and Kershaw, 1983a; Chapin et al., 1991). Samples (n  =  12) were treated in the same way as field incubations, with the exception of a 24 hr wetting pretreatment at optimal hydration levels.

N₂-fixation rates for both in situ incubations under ambient and optimal conditions were calculated as micromoles of ethylene reduced per hour per m² based upon the length of incubation and area of each sample (19 cm² BSCs and S. paschale; 56 cm² Sphagnum spp.). Conversion ratios determined for each N₂-fixing association (see below) were used to convert ethylene reduced to N₂ reduced. ARA values were corrected for differences in incubation jar volume, mean sample volume, and area. Different N₂-fixing associations were allowed to vary in sample depth (i.e. 0.75 cm BSCs, 2 cm S. paschale, 6 cm Sphagnum spp.) to help ensure sampling units were kept intact and that N₂-fixing surfaces were representative of the different associations under natural conditions.

¹⁵N-INCUBATIONS

Samples of each cyanobacterial association were collected from the Daring Lake landscape in August 2008 to determine conversion ratios for each of the N₂-fixing associations following the methods of Liengen (1999b). Samples were kept cool (∼4 °C) and shipped to the lab at the University of Northern British Columbia. Prior to incubation samples were kept at optimal hydration in a growth chamber for 72 hours under a 17/7 hr light (200 µmol PAR m⁻² s⁻¹)/dark cycle with temperatures at 15 °C during light hours and 5 °C during dark hours. Cores for each N₂-fixing association (n  =  8) were similar in area (19 cm²) and depth (0.75–2 cm) to those used for field incubations.

In order to achieve detection of ¹⁵N enrichment it was determined that 48 hr laboratory incubations (200 µmol PAR m⁻² s⁻¹, 20 °C) were required. Air (10% v/v) was replaced with 10% (v/v) ¹⁵N gas (Cambridge Isotope Laboratories Inc., Andover, Massachusetts, U.S.A.; ¹⁵N₂, 98%+). To reduce the potential for CO₂ limitation due to the long incubation period, each chamber was injected with 5% (v/v) CO₂. After the 48 hr incubation samples were immediately dried at 105 °C. Dry samples were ground in a ball mill and sent for ¹⁵N and total N analysis (Stable Isotope Facilities, University of Saskatchewan, Saskatoon, Saskatchewan, Canada). Control samples (n  =  8 for each N₂-fixing association) treated in the same manner but incubated with C₂H₂ were used to determine the natural abundance of ¹⁵N and the acetylene reduction rate. The amount of N fixed was calculated using (Liengen, 1999b, p. 224):

TABLE 2

Mean monthly N₂-fixation rates (µmol N m⁻² hr⁻¹) in incubations under field and optimized environmental conditions for each of the principal N₂-fixing cyanobacterial associations in the low arctic tundra landscape near Daring Lake, Northwest Territories, Canada. Acetylene reduction conversion ratios based on optimal conditions are included for each N₂-fixing association. Parentheses indicate standard errors. BSC  =  biological soil crust.

ESTIMATION OF LANDSCAPE LEVEL N INPUTS

Modeling N₂-Fixation Potential

Models of N₂-fixation potential were determined for BSCs using N₂-fixation rates and microclimatic data recorded in the in situ ambient incubations. The Sphagnum spp. N₂-fixation model was based on N₂-fixation rates under controlled laboratory conditions, and the S. paschale model was based on N₂-fixation rates recorded during in situ ambient incubations and macroclimatic data recorded at a local Daring Lake weather station (∼500 m from the research station) (2007–2008; Bob Reid, INAC, unpublished data). Spearman correlations were determined between mean N₂-fixation rate, mean light (PAR), mean incubation temperature, and mean % moisture for each incubation across all in situ ambient incubations (2007 and 2008) for Hollow (n  =  54) and Hummock BSCs (n  =  54). Temperature had the highest correlation with N₂-fixation for both Hollow (r  =  0.78) and Hummock BSCs (r  =  0.64). Light and temperature had a high covariance for Hollow BSCs (r  =  0.84) and Hummock BSCs (r  =  0.81); therefore, only temperature and moisture were used in the models. Separate models were determined for high and low moisture conditions. High and low moisture classes were based on percent moisture values above (‘high’) and below (‘low’) the median % moisture detected for Hollow or Hummock BSCs incubated in the field over the growing season in 2007 and 2008.

Sphagnum spp. were sampled from the field site in early August 2009, kept cool (∼4 °C), and shipped to the lab at the University of Northern British Columbia. The samples were incubated under a range of laboratory conditions reflecting field conditions in 2007–2008 to determine the response of N₂-fixation to temperature and light. Moisture was not included because >70% of Sphagnum spp. within the landscape occurred in Wet Sedge Meadows where moisture remains relatively high throughout the growing season and is likely not limiting. Sphagnum spp. N₂-fixation rates were significantly correlated with temperature (r  =  0.62). N₂-fixation rates under light conditions ranging from 0 to 1000 µmol PAR m⁻² s⁻¹ were not significantly different (ANOVA, Tukey post hoc, p  =  0.09); therefore, only temperature was included in the model.

Mean N₂-fixation by S. paschale under in situ ambient incubations (n  =  27) was highly correlated with mean % moisture (r  =  0.92). Days since precipitation was used as the moisture variable for S. paschale since direct measurements of field moisture content were not possible. A considerable lag period often occurs following saturation of Stereocaulon mats from rainfall events before steady nitrogenase activity is recovered (Crittenden and Kershaw, 1978). Therefore, a 24 hour lag was incorporated into the days since precipitation variable. The 24 hours following a precipitation event (≥1 mm) was coded as 0 and every subsequent day without a precipitation event coded with an increasing value of 1. Days since precipitation was highly correlated with mean % moisture (r  =  −0.76) and with N₂-fixation (r  =  −0.84).

The above models were used to estimate hourly N₂-fixation rates for each association over a growing season based upon microclimate and macroclimatic monitoring in the study landscape. Hourly N₂-fixation rates were summed to provide daily (Fig. 2) and seasonal totals (Table 4). We defined the start of the growing season as the first set of three or more consecutive days with no snow cover and mean air temperature above 0 °C, and the end of the growing season as the first occurrence of three or more consecutive days with mean air and soil surface temperature <0 °C. In 2007 and 2008 these conditions occurred between Julian days 152 and 257 and 144 and 259, respectively. Estimates of N₂-fixation for all of the above growing season days were not possible for every N₂-fixing association due to unavailable microclimatic data. Therefore, the total mean N input for each association was based on the average of 2007 and 2008 estimates over a 103 day growing season from 154 to 257 in both years. Air temperature, snow depth, and precipitation were determined from macroclimatic data from the local Daring Lake weather station (2007–2008; Bob Reid, INAC, unpublished data).

FIGURE 2

Seasonal trends in N₂-fixation rate (μmol N m⁻² day⁻¹) estimated from potential N₂-fixation rate models and field environmental data records for each N₂-fixing association for 2007 and 2008 at Daring Lake, NWT. See Table 3 for further model details. ARA to N₂-fixation conversion ratios (Table 2) were applied for each N₂-fixing association.

N₂-FIXATION RATES OF THE PRINCIPAL CYANOBACTERIAL ASSOCIATIONS

Mean monthly N₂-fixation rates under field conditions differed significantly among N₂-fixing associations (F_(3,2232)  =  156.51, P < 0.01) and were significantly different between months (F_(2,2232)  =  3.40, P  =  0.03) (Table 2). The interaction of month and N₂-fixing association was also significant (F_(6,2232)  =  27.67, P < 0.01) with patterns of N₂-fixation over the growing season (June–August) varying among the different associations. The highest rates of N₂-fixation for all of the associations with the exception of S. paschale occurred in July; however, Hollow BSCs had lower rates in June compared with July and August, and Sphagnum spp. had lower rates in August compared with June and July.

N₂-fixation rates under optimal conditions (200 µmol PAR m⁻² s⁻¹, 20 °C) differed among cyanobacterial associations (F_(2,154)  =  181.15, P < 0.01) and between months (F_(2,154)  =  8.17, P < 0.01), and there was a significant interaction between these two factors (F_(4,154)  =  2.70, P  =  0.03). The highest N₂-fixation rates under optimal conditions for all associations were in July (Table 2). The lowest rates under optimal conditions occurred in June for both Hollow BSCs and S. paschale, while the lowest rates occurred in August for Hummock BSCs.

Comparison of N₂-fixation rates under field and optimal conditions clearly indicated that adverse in situ environmental factors severely curtailed N₂-fixation, and that the extent of this constraint varied substantially among cyanobacterial associations. BSC associations had N₂-fixation rates under optimal conditions that were 2–3 times higher than those observed under field conditions, while rates in S. paschale were 4–7 times higher under optimal conditions (Table 2).

LANDSCAPE-SCALE PATTERNS OF N INPUT

Microclimatic Models of Potential N₂-Fixation for Each Cyanobacterial Association

Our simple models of N₂-fixation rates in relation to either temperature and/or moisture explained at least 50% of the variation in the field incubation data (Table 3). N₂-fixation rates were correlated with temperature in the high moisture and low moisture category samples (R²  =  0.77, P < 0.001; R²  =  0.78, P < 0.001, respectively) of Hollow BSC associations (Table 3). Similarly, N₂-fixation rates were correlated with temperature in the high moisture and low moisture category samples (R²  =  0.56, P < 0.001; R ²  =  0.50, P < 0.001, respectively) of Hummock BSC associations (Table 3). N₂-fixation rates for the Sphagnum spp. cyanobacterial associations were also correlated with temperature (R²  =  0.72, P < 0.001), while S. paschale rates were significantly correlated with days since precipitation (R²  =  0.69, P < 0.001) (Table 3).

TABLE 3

Potential N₂-fixation rate logistic regression models based on acetylene reduction (AR) rates in field incubations for each of the principal N₂-fixing associations. Hollow and Hummock BSC data were each separated into two moisture classes as indicated. Environmental variables included in models are surface temperature of Hollow (T_ho) and Hummock (T_hu) BSCs, Sphagnum spp. temperature at 2 cm depth (T_s), and Days since precipitation (D_sp). The dependent variable for all models is log acetylene reduction (μmol C₂H₄ m⁻² hr⁻¹).

Seasonal trends of N₂-fixation estimated by using full growing season field microclimatic records in the models indicated that each N₂-fixing association had similar patterns of activity in 2007 and 2008 (Fig. 2). N₂-fixation inputs in Hollow BSC, Hummock BSC, and Sphagnum spp. associations fluctuated dynamically during the first half of the season but tended to generally increase toward peak values in mid to late July, and then to decline fairly steadily afterwards. No clear seasonal trend could be observed for S. paschale because estimates were based solely on days since precipitation. The model estimates of mean total N input across the growing season (3 June to 13 September) for each cyanobacterial association ranged from 3.4 kg N ha⁻¹ yr⁻¹ (Hollow BSCs) to 24.9 kg N ha⁻¹ yr⁻¹ (S. paschale) for a 103 day growing season (Table 4).

TABLE 4

Mean total N fixed over the growing season (3 June to 13 September) based on estimates of N₂-fixation by Hollow BSC, Hummock BSC, Sphagnum spp., and Stereocaulon paschale at Daring Lake, Northwest Territories, in 2007 and 2008. Microclimatic models were used to predict hourly acetylene reduction rates per m² (See Table 3). ARA to N₂-fication conversion ratios (Table 2) were applied for each N₂-fixing association. Rates were summed to give total mg N m⁻² yr⁻¹ based on the 2007 and 2008 growing seasons indicated by Julian days. The mean of 2007 and 2008 estimates based on a 103 day growing season (154–257) in both years was used to determine mean total N.

N Input by N₂-Fixing Associations and by Ecosystem Types

No single N₂-fixing association dominated N inputs across all of the ecosystem types. Stereocaulon paschale was the largest source of N input in both Xerophytic Herb Tundra and Heath-Lichen/Heath-Mat Tundra followed by Hummock BSC (Table 5). Sphagnum spp. was the largest source of N input in both Birch Hummock and Wet Sedge Meadow ecosystems followed by Hollow BSC. Despite having the highest mean N₂-fixation rate, S. paschale did not have the highest overall landscape N input (549.84 kg; Fig. 3). Sphagnum spp. had the highest N input (1030.72 kg) due to its relatively high mean N₂-fixation rate and greater area within the landscape (50.28 ha) compared with S. paschale (22.11 ha) (Fig. 3).

FIGURE 3

(a) Mean N₂-fixation rate measured over the growing seasons of 2007 and 2008 (see Table 2), (b) total area of N₂-fixing associations in landscape study area, and (c) total N input for Hollow biological soil crusts (BSCs), Hummock BSCs, Sphagnum spp., and Stereocaulon paschale determined by potential N₂-fixation models in the landscape study area at Daring Lake, NWT. Total area of N₂-fixing associations and total N input in landscape study area were calculated from values in Table 5. Error bars in (a) indicate standard error.

TABLE 5

The contributions of individual N₂-fixing associations to total biological N input across the selected landscape study area and to N inputs per unit area for each of the major ecosystem types at Daring Lake, Northwest Territories, over the growing season (Julian days 154 to 257).

We used the model estimates of growing season N₂-fixation by each cyanobacterial association along with the mapping data of the distribution of ecosystem types in our landscape study area to estimate overall N inputs in each ecosystem type. Total N input per unit area was ∼10 times higher in the Wet Sedge Meadow than in any other ecosystem type (Table 5). Our spatially explicit analyses indicate that this effect can be explained by particularly high inputs by Sphagnum spp. cyanobacterial associations due to relatively large fixation rates (Table 4) in combination with relatively high proportional cover of this association in the Wet Sedge Meadow (Table 5). Heath-Lichen/Heath-Mat Tundra had relatively low N inputs per unit area but had the largest total N input in our selected landscape study area because of its abundant coverage in the landscape. Birch Hummock tundra had similar total N₂-fixation rates per unit area to Heath-Lichen/Heath-Mat Tundra, but its coverage was low in the study area, resulting in low total N inputs. Finally, N₂-fixation rates per unit area within Xerophytic Herb Tundra were lowest, and its coverage was also low, resulting in relatively small N inputs into the selected landscape study area. Total N input for 68% the Daring Lake landscape study area over the 103 day growing season was 0.68 kg ha⁻¹ yr⁻¹.

MICROCLIMATIC CONTROLS ON SEASONAL AND SPATIAL VARIATION IN N₂-FIXATION

Several studies have detected distinct seasonal patterns in N₂-fixation rates in the Arctic (Alexander and Schell, 1973; Henry and Svoboda, 1986; Chapin et al., 1991; Zielke et al., 2005). Our data (Table 2 and Fig. 2) are consistent with the general pattern of detectable N₂-fixation rates soon after snowmelt (March–June) followed by the highest rates coinciding with the peak growing season and declining rates in late July to August depending on latitude. Our field measurements of N₂-fixation demonstrate that seasonal patterns vary among the vegetation communities, which may account for the lack of seasonal variation detected in studies that average across both topography and vegetation type (Hobara et al., 2006).

Moisture appears to be one of the most important environmental factors controlling N₂-fixation across various arctic environments (Chapin and Bledsoe, 1992; Nash and Olafsen, 1995; Zielke et al., 2002, 2005). Moisture can affect seasonal patterns of N₂-fixation within individual N₂-fixing communities, and spatial heterogeneity in N₂-fixation is often a reflection of differences in cyanobacterial biomass due to the long-term characteristics of the community moisture regime (Chapin et al., 1991). Stereocaulon paschale, which was primarily located on xeric esker tops and well-drained upper esker slopes, had the highest mean rate of N₂-fixation over the growing season, but demonstrated a strong sensitivity to desiccation. Lichens are often established on drier exposed habitats where nitrogenase activity may be reduced to a few comparatively short episodes when moisture conditions are suitable (Crittenden and Kershaw, 1979). Rainfall in July 2008 was 6.7 mm compared to 24.6 mm and 28.1 mm in June and August, respectively. Accordingly, the average percent moisture of S. paschale incubated in July was 26% as compared to June (56%) and August (45%), perhaps explaining the relatively low July N₂-fixation rates. The relatively high and consistent N₂-fixation rates associated with Sphagnum spp. over the growing season are likely due to the consistently high moisture conditions of the low-lying Wet Sedge Meadow where Sphagnum spp. are the dominant vegetation.

Temperature has also been significantly correlated with N₂-fixation rates in the Arctic (Smith, 1984; Lennihan et al., 1994; Liengen and Olsen, 1997b; Zielke et al., 2002). Our strong correlations between N₂-fixation and temperature for Hollow BSC, Hummock BSC, and Sphagnum spp. indicate that seasonal temperature fluctuations are important in determining seasonal rates of N₂-fixation. Like Zielke et al. (2005), we also found that temperature was a good predictor of N₂-fixation provided different models were used depending on moisture condition. Hollow BSC had lower field and optimal rates of N₂-fixation in June compared with other cyanobacterial associations (Table 2). Mean Hollow BSC temperature in June was 6.8 °C compared to 9.1 °C for Hummock BSC. Therefore, the lower rates of N₂-fixation for Hollow BSC are likely due to the persistence of snow in these depressions resulting in relatively low temperatures, as well as restricted light inputs that together may impede the recovery and development of cyanobacterial communities in the early growing season (Fig. 4).

FIGURE 4

Hummock-hollow complexes at Daring Lake, NWT, 17 June 2008. Differences in early season fixation rates between Hollow and Hummock BSC may result from snow covering hollows of hummock-hollow complexes whereas Hummock BSC is exposed.

Some studies have found N₂-fixation to be light dependent (Granhall and Lid-Torsvik, 1975; Alexander et al., 1978) while others have found little light dependence as photosynthetic rates tend to saturate at relatively low light levels (<500 µmol PAR m⁻² s⁻¹) (Coxson and Kershaw, 1983b; Smith, 1984; Chapin and Bledsoe, 1992; Nash and Olafsen, 1995; Zielke et al., 2002). Varying light conditions (0 to 1000 µmol PAR m⁻² s⁻¹) did not affect Sphagnum spp. N₂-fixation rates in our study, supporting the concept that stored energy for N₂-fixation, combined with continuous or near continuous daylight and a limited plant canopy, reduce the potential for light to act as a controlling factor on N₂-fixation in the Arctic (Chapin and Bledsoe, 1992). Nevertheless, remote sensing, repeat photography, and experimental nutrient addition studies all suggest that current warming trends in the low Arctic may be promoting shrub growth and expansion within various topographic positions (Goetz et al., 2005; Sturm et al., 2001; Chapin et al., 1995). Declining macrolichen abundance in warming sub- and mid-arctic environments may be a function of increased growth and abundance of deciduous shrubs, which may inhibit lichen performance and persistence through shading (Cornelissen et al., 2001). N₂-fixation rates and persistence of other N₂-fixing associations in these environments may also be similarly influenced by reduced light availability.

The higher N₂-fixation rates detected under optimal conditions for all N₂-fixing associations indicate that microclimatic conditions in the field are limiting N₂-fixation for all of the principal N₂-fixing associations. We conclude that climate change scenarios that result in warmer surface temperatures without increased surface desiccation are likely to lead to higher rates of N₂-fixation (Chapin and Bledsoe, 1992).

THE SIGNIFICANCE OF BIOLOGICAL N₂-FIXATION TO N CYCLING IN A LOW ARCTIC LANDSCAPE

Since N availability is commonly a major limitation on tundra plant growth, our results provide important insights to understanding the functioning of low arctic terrestrial ecosystems. Our estimate of mean seasonal N₂-fixation in Birch Hummock (Table 5; 1.4 kg⁻¹ ha⁻¹ yr⁻¹ over 103 days) is ∼1/300th the late summer rate of gross N mineralization by soil microbes in the same ecosystem type (Buckeridge et al., 2010). Therefore, internal recycling of N from soil organic matter is undoubtedly the critical N supply process within the Birch Hummock ecosystem type at least. Nevertheless our data suggest a significant influence of N₂-fixation on N cycling and carbon uptake at a larger scale. N₂-fixation during the growing season was highest in the Wet Sedge Meadow, which is also the ecosystem with the largest annual plant primary production in this landscape (Nobrega and Grogan, 2008). Nutrient inputs associated with run-off and leachates from higher elevation ecosystems toward the valley floor where wet sedge ecosystems predominate may facilitate the high rates of primary production there. Our data here suggest that in addition to that process, in situ N₂-fixation inputs may be an important pathway supplying N to support the high primary productivity of this ecosystem type.

Total biological N₂-fixation input across the study landscape area was estimated at 0.68 kg N ha⁻¹ yr⁻¹. Previous estimates of arctic N₂-fixation inputs range from 0.06 to 3 kg N ha⁻¹ yr⁻¹, with the majority of estimates ranging from 0.10 to 1.20 kg N ha⁻¹ yr⁻¹ (Alexander and Schell, 1973; Barsdate and Alexander, 1975; Chapin and Bledsoe, 1992; Hobara et al., 2006). Summertime mean atmospheric N inputs from wet deposition at the nearest monitoring station (63.52°N, 116.00°W) to Daring Lake (∼240 km away) were 0.39 kg N ha⁻¹ yr⁻¹ (1991–2006; CAPMon, Environment Canada, unpublished data). Wintertime atmospheric N deposition inputs as total inorganic N accumulation in ambient snow packs (0.3 m) at Daring Lake in 2007 were 0.05 kg N ha⁻¹ (Buckeridge and Grogan, 2010). Together, these numbers suggest that total biological N₂-fixation input for the landscape study area at Daring Lake is approximately twice the amount of N deposited via atmospheric deposition. While some studies have found N₂-fixation contributed 80% or higher to total landscape N inputs (Hobara et al., 2006; Solheim et al., 2006), other studies, including ours, have found the contribution of N₂-fixation to ecosystem N inputs is approximately 50–70% (Chapin and Bledsoe, 1992; Henry and Svoboda, 1986).

We found N₂-fixation across a low arctic tundra landscape was concentrated in the Wet Sedge Meadow ecosystem type where N₂-fixation per unit area was ∼10 times higher than in any of the other ecosystem types (Table 5). Of the four principal N₂-fixing associations, Sphagnum spp., which had the highest percent cover in Wet Sedge Meadows, made the largest contribution (55.2%) to total N input. Several other studies have also found the highest rates of N₂-fixation in arctic landscapes are associated with cyanobacteria moss associations (Alexander and Schell, 1973; Henry and Svoboda, 1986; Solheim et al., 1996).

Five methodological constraints may have affected our landscape estimates of biological N₂-fixation inputs. Firstly, conversion ratios can vary depending on the operating environment of a given N₂-fixing association (Millbank, 1981; Gunther, 1989), and seasonal variation in conversion ratios has been detected for free-living cyanobacteria from high arctic habitats (Liengen, 1999b). Secondly, we used visual estimates of abundance for the N₂-fixing associations without accounting for variations in cyanobacterial biomass that can impact rates of N₂-fixation. Thirdly, the ecosystem types in our analysis account for only 68% of the Daring Lake study area. Some excluded ecosystem types (Exposed Sand and Gravel, and Rocky Outcrops) probably contribute little or nothing to landscape N input. However, other ecosystem types such as Dry Sedge Meadows (8.2%) may contain considerable Sphagnum spp. cyanobacterial associations and therefore may make significant contributions to landscape N input, albeit for a limited duration due to less favorable microclimatic conditions. Fourthly, our estimates of modeled N inputs would have been improved by more accurate quantification of spatial variability in soil surface microclimate by using a much larger number of climate sensors. This is a limitation that is common to many studies of arctic and subarctic ecosystems (Rouse, 1976; Young et al., 1997). Fortunately, the type of conditions that favor N₂-fixation by most cyanobacterial associations (i.e. during or immediately following growing season precipitation events) will tend to minimize between site variability in soil surface microclimate, reducing the impact of this factor on our estimates. Fifthly, we used a 103 day growing season as the basis for yearly N input. However, N₂-fixation likely occurs outside of this period whenever microclimatic conditions are favorable (Davey, 1983; Liengen, 1999a; Zielke et al., 2002; Hobara et al., 2006). In summary we conclude that, provided our conversion ratios, percent cover of cyanobacterial associations, and models of potential N₂-fixation are sufficiently accurate, our estimates of ecosystem N₂-fixation inputs and of total landscape-level N input over the growing season are minimum values.