Introduction

Biologically available nitrogen (N) is closely associated with the quantity and quality of soil organic matter (SOM), particularly in high-latitude ecosystems where decomposition is slow and large amounts of N are stored in SOM (Robinson and Wookey, 1997; Jonasson et al., 1999). Nitrogen is generally considered the most limiting nutrient in high-latitude arctic/subarctic ecosystems, and N availability (of both inorganic and organic forms) is therefore likely to modulate ecosystem response to climate warming (Sveinbjörnsson et al., 1996; Press et al., 1998; Arft et al., 1999; Jonasson et al., 1999; Rustad et al., 2001; Hobbie et al., 2002). Since decomposition of SOM and the related release of N are sensitive to increased temperature (Nadelhoffer et al., 1991; Lloyd and Taylor, 1994; Peterjohn et al., 1994; Rustad et al., 2001; Fang and Moncrieff, 2001; Sjögersten and Wookey, 2002), the warming predicted at high latitudes (Hagen et al., 2001) will likely increase the bioavailable N. Increased N availability is likely to increase tree growth at the tree line in subarctic Fennoscandia (Sveinbjörnsson et al., 1996) and current predictions are indeed for an expansion of trees into tundra areas (Emanuel et al., 1985; Kittel et al., 2000; White et al., 2000).

In addition to climatic controls, the storage and release of nutrients are strongly linked both to the quantity and quality of litter inputs, and to the nature of the decomposer community (Swift et al., 1979). SOM quality is therefore expected to vary among areas with contrasting litter inputs and climate (Swift et al., 1979; Dyer et al., 1990; Raich and Schlesinger, 1992; Coûteaux et al., 1995). Indeed, many studies carried out in subarctic and tundra environments have noted differences in respiration and net mineralization rates over short distances between contrasting plant communities (e.g., Giblin et al., 1991; Nadelhoffer et al., 1991; Hobbie, 1996; Weih, 1998; Rustad et al., 2001; Sjögersten and Wookey, 2002). Contrasting decomposition, for example, and carbon and nutrient storage, have been observed across the mountain birch forest-tundra ecotone in the Fennoscandian mountains (Sjögersten et al., 2003; Sjögersten and Wookey, 2004). Greater storage of labile carbon in soils was found in mesic tundra heath areas compared to nearby mountain birch forest, which is attributed to lower decomposition rates in tundra areas (Sjögersten et al., 2003). There are also indications of higher SOM quality in forest soils (Sjögersten and Wookey, 2002; Sjögersten et al., 2003), including lower C:N ratios, that might be related to the higher quality and quantity of litter inputs that these soils receive compared to nearby tundra heaths dominated by evergreen ericaceous dwarf shrubs (Cornelissen, 1996; Quested et al., 2002). Indeed, Sveinbjörnsson et al., (1996) found 10 times lower ammonium concentrations in soils above the treeline. Thus, we hypothesize that the amount of N mineralized will be greater in mountain birch forest soils compared with tundra heaths, in response either to more favorable climatic conditions within the forest canopy, or to higher litter quality. The decomposer community would likely be responsive to alteration in the above-ground vegetation, together with the associated change in microclimate and substrate quality (Robinson, 2001). Although we focus here on the release of inorganic N from SOM, we acknowledge the demonstrated role of direct uptake of organic N by several boreal and tundra plant species (Kielland, 1994; Näsholm et al., 1998).

There is growing evidence of significant microbial activity beneath the snow cover in high-latitude and alpine areas during the winter period. This activity is fuelled by plant litter inputs at the end of the growing season, as well as by rapid recycling of microbial residuals (Schmidt and Lipson, 2004; Schimel et al., 2004). There is, however, still a need for information on how variation in latitude and SOM and litter quality between contrasting vegetation communities influence the winter activity. In subarctic and arctic environments, the long duration of the winter period, compared with a much shorter growing season for plants, makes research on winter processes imperative.

The aim of the current study was to answer the following specific questions: Does N-mineralization differ between forest and tundra soils, and is the difference consistent over similar vegetation communities along the Fennoscandian mountain range? If so, can the difference be linked to climate or SOM quality? Does significant mineralization occur during the winter period beneath the snowpack?

Materials and Methods

Results

Surface air and soil temperatures at 50 mm depth were generally higher farther south, and forest sites were warmer than tundra, with the largest difference during the winter period due to the deeper snow cover at forest sites (Table 1 and 2). The period with subzero soil temperatures (50 mm depth) differs between areas, and between forest and tundra sites. During the winter of 1999/2000, the period with subzero temperatures lasted 23, 27, and 30 weeks at Dovrefjell, Abisko, and Joatka, respectively. The duration of below-zero temperatures was 3, 1, and 2 weeks longer at tundra sites than at forest sites at Dovrefjell, Abisko, and Joatka, respectively. Forest soils were drier than tundra soils (F = 14.8, p < 0.001) with 51.4% and 63.1% gravimetrical water content, respectively, except at Dovrefjell, where no difference was found (Table 3). Soils at Joatka were drier than at the other sites (F = 11.8, p < 0.001) (Table 3). Despite these differences in moisture content, regression analysis between soil moisture content and nitrification and ammonification rates, respectively, explained none of the variance in mineralization rates.

We measured the highest rates of NH₄⁺ release at Dovrefjell and Abisko. Ammonification rates were also higher at forest sites (Table 4, and Figs. 3 and 4). We recorded immobilization of NH₄⁺ at Joatka during the winter period. The significant interaction between site and season for NH₄⁺ reflect the reduced mineralization rates during the winter period at Joatka and Dovre fjell, while mineralization actually tended to increase during the winter period at Abisko and Vassijaure.

Tundra soils had generally higher rates of NO₃⁻ release (nitrification), and the rates were higher during the winter period. However, the pattern of higher mineralization during winter was not repeated at Dovrefjell; here the mineralization rates were highest in the summer. It is also worth noting that we recorded immobilization of nitrate in the summer period at Joatka.

We found a significant negative relationship between N-mineralization and the degree of continentality at each site (Fig. 5). No significant regressions were found between N-mineralization rates and summer and winter soil temperature or pH.

Further, we found that significant variation in N-mineralization could be explained by the C and N content in the soil: Positive regressions were detected between summer ammonification rates and total N content in the soils (R² = 0.71; p = 0.03), and between winter nitrification rates and total C content (R² = 0.76; p = 0.02). In addition, nitrification rates also tended to be positively associated with the concentration of labile carbon (N-alkyls + O-alkyls + acetals) in the soil (R² = 0.57; p = 0.08).

The transplantation of soils between sites generally increased immobilization of both NH₄⁺ and NO₃⁻ irrespective of soil origin, and no effect of incubation (recipient) site was detected (Fig. 6). This response was significant for nitrate in all transplanted soils, while for ammonium this pattern was only significant for soils originating from tundra at Vassijaure and forest in Abisko. Furthermore, significantly higher NO₃⁻ immobilization occurred if the recipient site was warmer than the original site (i.e., from tundra to forest), both at Vassijaure (p = 0.007) and Abisko (p = 0.002) (Fig. 6).

Discussion

Ammonification (release of NH₄⁺) was consistently higher in forest soil than tundra across the ca. 8° latitudinal gradient. Similar variation in ammonification rates between forest (ca. 400 mg m⁻²) and tundra (ca. 100 mg m⁻²) soils was noted in the Abisko region by Robinson et al. (1995) and Schmidt et al. (1999), respectively. The values from the current study are, however, lower than those previously reported: ca. 135 and 12 mg m⁻² (in the upper 50 mm of soil), at Lower Forest and Upper Tundra sites, respectively. By way of explanation, it is possible that the soils investigated by Robinson et al. (1995) which were frozen before extraction, may have increased the release of extractable constituents. The difference between tundra sites is probably linked to the higher total N content in the soils studied by Schmidt et al. (22.8 mg g⁻¹, compared to 13.3 mg g⁻¹ at the Abisko Upper Tundra site in the current study), and the higher pH (7.1, compared to 3.9 at our tundra site). The higher ammonification in forest soils compared with tundra heaths is probably associated with the higher N content in forest soils. Increased N availability in forest soils could be explained by a combination of factors: (1) higher litter quality in the birch forest (Cornelissen, 1996; Sveinbjornsson et al., 1996; Weih and Karlsson, 2001), and (2) the deeper snow accumulation at forest sites, which increases both winter soil temperatures and the flux of airborne N-containing compounds (Weih, 1998) compared with nearby tundra heaths. The greater difference in ammonification rates between forest and tundra sites at Dovrefjell, compared to Joatka, could indeed be related to the higher snowfall at the southerly site, and possibly also linked to increased N deposition in precipitation at Dovrefjell. Tissue N concentrations in mountain birch trees have been shown to increase as a physiological response to decreased temperature at higher altitudes (Weih and Karlsson, 2001). This provides an interesting link with the higher N-mineralization rates found at forest sites in this study, probably partly explaining the significant difference in ammonification rates between forest and tundra.

Although temperature has been shown to determine the position of the treeline on a macroscale (Körner and Paulsen, 2004), small-scale variation is likely to relate to topography, snow cover, hydrology, and soil nutrient status (Walker et al., 1993; Kjällgren and Kullman, 1998). This study suggests that a positive feedback could be possible whereby the establishment of birch trees on former tundra heath has the capacity to ameliorate the local physicochemical environment (as discussed above) by increasing N availability. Mountain birch is most likely, nonetheless, to become established at sites that are topographically predisposed to accumulate snow, and hence both additional N fluxes and also eolian inputs of inorganic and organic material.

The apparent relationship between summer N-mineralization rates and continentality (Fig. 5) suggests that climate exerts some form of control on the overall rates of summer N mineralization. This relationship is not readily explained, however, on the basis of our directly measured climatic variables: We saw no clear differences, for example, in mean July temperatures between the main study sites over the measurement periods. It is also important to note that a regression relationship based on 4 sites (two of which with the same Gorzinski continentality index—Abisko and Vassijaure—but with markedly contrasting precipitation; Table 2, and Fig. 5) should be interpreted with a degree of caution. If, nonetheless, a climatic control on N mineralization is accepted, it appears likely that it is a long-term effect of climate, or one exerted indirectly. This might be linked, for example, with a microbial community that is increasingly conservative of assimilated N further at more continental sites, but mineralization rates could also be affected by indirect factors such as quality and quantity of litter input. The total N content in the soils is, however, inversely related to continentality, which makes the two potential drivers difficult to separate. The relatively high mineralization rates at the tundra site at Dovrefjell could be linked to the higher pH in the organic horizon (6.2 at Dovrefjell, compared with 3.9–5.2 at the other sites), but we found no significant regression relationship between pH and either ammonium or nitrate fluxes. It is also worth noting, however, the large difference in the deposition of airborne N-containing pollutants between these areas. N-deposition rates reported from the Dovrefjell region (Osen; 61°15′N, 11°47′E) and Finnmarksvidda (Karasjokk; 69°28′N, 25°13′E) show three times higher N deposition at the more southerly site (325.3 compared to 104.3 mg N m⁻², average from 1998, 2000, and 2001) (Hjellbrekke, 2002). Thus, our continentality/latitude gradient has, superimposed upon it, a further gradient in terms of N inputs to the ecosystems.

Nitrification rates were highest in tundra soils and during the winter period. This is in agreement (but see comments in paragraph below) with a gradient study by Ineson et al. (1998) in the Pennines in the U.K., where the highest NO₃⁻-N leaching was found at the higher elevations sites, and suggests increased nitrification at colder sites. The significant regression relationship between winter nitrification rates and both total C and labile C content in the soils suggest that the nitrifiers depend upon the amount of available C for growth in these soils. It is, however, surprising that bacteria in these C-rich soils should be generally C limited, and we therefore propose that it is the presence of labile C that is of importance to the nitrifying bacteria (if we assume that these are of greater significance than autotrophic nitrifiers; see Hart et al., 1994). The regression is, however, driven strongly by the immobilization rates measured at the forest site in Abisko, and should therefore be interpreted cautiously. Schimel et al. (2004) and Schmidt and Lipson (2004) suggest that nitrification in tundra soil during the winter period is due to the heterotrophic microbial community becoming C-limited during the winter, leading to increased NH₄⁺ availability and hence nitrification. This mechanism could explain the higher nitrification rates at the tundra site in the current study, where temperatures are lower during the winter period: In addition, litter quality could also be important here.

The transplant experiment (Fig. 2) showed that sudden alterations in the physical environment have the potential to shift the system from net N mineralization to immobilization, but it did not provide a clear answer as to the importance of temperature vs. SOM for N mineralization. The greater increase in NO₃⁻ immobilization in response to a move to a warmer environment suggests that the increased immobilization is related to temperature, but the transplanted soils did not adjust to the mineralization rates of native soils at the recipient site. Similar results were found in the transplant study by Ineson et al. (1998), in which transplanting soil down an altitudinal gradient (to a warmer environment) reduced nitrate leaching from lysimeter cores. The study by Ineson et al. (1998), however, involved the use of intact lysimeter cores, so that leachate chemistry was the net product of mineralization and assimilation, both by the decomposer community and the vegetation. Direct comparison with the present study is not, therefore, straightforward.

Since the methodology used in the present study (soil coring, and incubation in polyethylene bags) will undoubtedly kill fungal hyphae and plant roots, the technique produces several artifacts (Adams et al., 1989; Robinson et al., 1995; Hartley et al., 1999). In a nearby tundra heath area, ca. 5% of the total N in SOM was contained within the microbial biomass (Schmidt et al., 1999). Specifically, decaying hyphae and fine roots will provide labile resource to the surviving microbial community during the incubation period, nutrient and water uptake by plants is prevented, and root exudates and fresh litter inputs are excluded. Our cores are therefore not likely to reflect the natural rates of N mineralization and should only be used for relative comparisons; additionally, the results will inevitably reflect initial responses to disturbance.

N mineralization/immobilization proved rather tolerant of low temperatures. Although net NH₄⁺ release was lower during the winter period at some sites, NO₃⁻ release rates were actually higher during the winter period (Figs. 2 and 3). This activity could have occurred mainly during the late autumn, when soil temperatures were generally above zero (soils froze first in mid–late October at tundra sites in Joatka and Abisko, and last in early–mid December at forest sites in Dovrefjell). However, there is growing evidence that soil microbial activity can continue during the winter period despite below-zero temperatures (Zimov et al., 1996; Fahnestock et al., 1998; Schmidt and Lipson, 2004; Schimel et al., 2004), and the January temperature range at the research sites in this study suggests that this could be the case at both tundra and forest sites across the mountain range. The higher winter ammonification rates at forest sites at Dovrefjell and Abisko are likely related to the longer active periods, as well as to the factors discussed above.

Finally, we want to stress that the links between above-ground vegetation and N mineralization presented here have important implications in a changing environment. We suggest that an expansion of the mountain birch forest in response to a warmer climate would alter the N cycling in these soils more drastically than the actual temperature increase itself, both due to the input of higher quality litter to the soils and to the changes in microclimate that are associated with the mountain birch forest, especially snow accumulation and higher winter soil temperatures.