STUDY SITES

Communities dominated by C. tetragona were investigated in northern Sweden near Abisko (68°18′N, 18°51′E), in Svalbard near Longyearbyen (78°25′N, 15°40′E), and in northeast Greenland near Zackenberg (74°47′N, 20°55′W) (Fig. 2). At Abisko the climate is subarctic montane, at Longyearbyen north Atlantic–Arctic, at Zackenberg continental high Arctic. Mean annual temperature and precipitation are −1.0, −5.7, and −9.4 °C, and 304, 180, and 228 mm, at Abisko, Longyearbyen, and Zackenberg, respectively ( www.linnea.com/˜ans,  www.met.no; Sigsgaard et al., 2006). The growing season is about three months at Abisko and two months at Zackenberg and Longyearbyen. The sampling took place in mid season: at Abisko in late July–mid August 2005 and 2007, at Longyearbyen in mid–late July 2005, and at Zackenberg in mid July–early August 2006. The glacial history of the sites does not differ substantially with deglaciation and emergence from sea level occurring around 8000–10,000 years Before Present (Christiansen and Humlum, 1993; Berglund et al., 1996, Lønne and Nemec, 2004).

Figure 2

Arctic locations investigated in this study: (1) Abisko, (2) Longyearbyen, and (3) Zackenberg (modified from ACIA, 2004).

Two plant communities were investigated at Abisko: (i) a tree-line heath at 450 m a.s.l., and (ii) a high-elevation fellfield at 1150 m a.s.l. The tree-line heath is characterized by species-rich dwarf shrub vegetation and a sub-canopy cover of mosses and lichens. Sampling took place in areas with uniform, mostly C. tetragona cover, avoiding particularly productive or unproductive spots. At the fellfield, the vegetation is sparse, with few vascular species scattered among a discontinuous cover of mosses and lichens; we sampled within patches dominated by C. tetragona. The population of C. tetragona at the tree-line heath is truly subarctic and near the lower distributional limit of the species. On the other hand, because of the marked elevation, the C. tetragona population at the fellfield has physiological and environmental characteristics similar to that of C. tetragona at high-arctic locations (Callaghan et al., 1989; Havström et al., 1993). Further information about Abisko sites can be found in Havström et al. (1993), Michelsen et al. (1996), Jonasson et al. (1999b), Graglia et al. (2001), and Campioli et al. (2009).

At Zackenberg, we sampled a dry sparse heath on the valley bottom, a heath facing south, an abrasion plateau1, and a snow bed, all at about 30–40 m a.s.l. (Bay, 1998). At Longyearbyen, we investigated a dry heath and a species-rich dwarf shrub heath. At Zackenberg and Longyearbyen, samples comprised a range of vegetation cover, varying from sparse to lush spots of C. tetragona. All data from the Zackenberg and Longyearbyen sites are available at  http://ecosystems.mbl.edu/arc/terrest/tracegas/index.shtml.

LAI AND BIOMASS

We estimated LAI and biomass by harvesting replicate plots representative of each plant community. Harvesting took place after measurements of reflectance were completed as described below. At Abisko, we selected 0.04 m² plots (n  =  15–16). Deciduous shrubs and herbs (forbs plus graminoids) are patchy at the sites and we sampled their leaves for the entire 0.04 m² plot. On the other hand, the cover of evergreen species (mostly C. tetragona) is uniform and we sampled it only in a representative 0.01 m² subplot. At Zackenberg and Longyearbyen, covers are more heterogeneous and we selected plots of 0.09 m² (n  =  2–8). However, due to time limitation, for most of the replicates (ca. 70%) only representative 0.03–0.05 m² subplots were harvested. Samples were brought to the laboratory, kept refrigerated in sealed bags, and processed within two days after collection. For each plant community, LAI was calculated as the sum of (i) LAI of deciduous shrubs, herbs, and evergreen shrubs other than C. tetragona, determined from measurements of one-sided leaf area as in Street et al. (2007); and (ii) LAI of C. tetragona, which was determined using the new method (see below). All green biomass used in the LAI assessment was dried at 70 °C for 48 h and subsequently weighed.

VEGETATION REFLECTANCE AND NDVI

We determined NDVI from measurements of vegetation reflectance in the near infrared (R_NIR; reflectance at 0.725–1.0 µm) and red light (R_RED; reflectance at 0.56–0.68 µm) following the equation:

1

R_NIR and R_RED were measured for a subset of the plots harvested for the LAI assessment (8 plots at Abisko, 17 at Zackenberg, and 4 at Longyearbyen) as described by Street et al. (2007) using a UniSpec Spectral Analyser (PP Systems International, Amesbury, Massachusetts, U.S.A.) at Zackenberg, or a Skye Instruments SpectroSense2 (Skye Instruments, Powys, U.K.) at Abisko and Longyearbyen. The UniSpec analyzer records complete reflectance spectra from 0.3 µm to 1.0 µm; we selected wavebands of R_NIR (0.725–1.0 µm) and R_RED (0.56–0.68 µm). The Skye sensor measures total reflected radiation R_NIR and R_RED. The two instruments agreed well in cross-calibration (Street et al., 2007).

LEAF TRAITS

Leaf traits of C. tetragona and other common shrub species (Betula nana, Vaccinium uliginosum, Vaccinium vitis-idaea, Salix polaris × herbacea) were analyzed in detail for material from the plots used for the LAI assessment at the Abisko tree-line heath and the fellfield sites. Specific leaf area (SLA; m² leaf kg⁻¹ leaf) was determined from measurement of leaf area and biomass. SLA of C. tetragona was calculated using area of leaves attached to stems and biomass of leaves detached from stems by peeling. Leaf N concentration (mg N g⁻¹ biomass) was determined with a TruSpec CN Determinator (LECO Corporation, St. Joseph, Missouri, U.S.A.) on ground leaf samples. Because peeling of C. tetragona is very time-consuming and N analyses costly, we determined SLA and leaf N concentration only for subsets of the LAI plots (n  =  8–9). However, for 3 plots of such subsets, we performed extra analyses on C. tetragona to test for N concentration differences between (i) current-year vs. non current-year leaves, and (ii) total green leaves vs. total green biomass. Because C. tetragona produces smaller leaves in spring than in summer (Callaghan et al., 1989), we considered as current-year growth the biomass from the shoot apex till the first double pair of small leaves. For these samples, the biomass ratio of leaf or stem vs. total green biomass was additionally determined. The water content of the green biomass of C. tetragona was determined gravimetrically from measurements of fresh- and dry weight of extra plant samples collected randomly at the tree-line heath and at the fellfield in mid August 2007 (n  =  6–8).

TFN

TFN was estimated from leaf N concentration and leaf standing biomass specific for each plot and species. As mentioned above, because of time and financial constraints on N analyses, TFN was estimated only for a subset of the plots sampled for LAI (n  =  8–9 at Abisko and n  =  2 at Longyearbyen). For plots at Abisko, we used leaf N concentrations as determined above, whereas leaf N concentrations at Longyearbyen were determined with a 2400 Series II CHNS/O analyzer (PerkinElmer, Waltham, U.S.A.). For species not analyzed for N concentration (7% of the total leaf biomass), leaf N concentration was obtained from samples of the same species close to the investigated plot (2% of the total leaf biomass) or assumed equal to leaf N concentration of the dominant species (evergreen for evergreen species; deciduous shrubs for deciduous shrubs and herbs; 5% of the total leaf biomass). Bias due to the last assumption was considered negligible because of the small amount of leaf biomass involved and because leaf N concentration of species of the same functional type (or deciduous and herbs) at the study sites, or at similar plant communities, are comparable, i.e. never exceed a twofold difference (Wielgolaski et al., 1975; Jonasson et al., 1999b; van Wijk et al., 2005; Campioli, 2008).

STATISTICAL ANALYSES

Statistical analyses were conducted using S-PLUS 6.1 (Insightful Corporation, Seattle, U.S.A.). The difference between values of LAI or leaf N concentration was tested with the Student t-test (for comparison of two samples) or with one-way ANOVA (for comparison of more than two samples) (p  =  0.05).

LAI, TFN AND CANOPY BIOMASS

LAI was 0.8–1.1 m² m⁻² at Abisko, 0.4–0.5 m² m⁻² at Zackenberg, and 0.7–0.9 m² m⁻² at Longyearbyen (Table 1). LAI of the subarctic heath was substantially larger (1.3–2.8 fold) than LAI of the fellfield and high arctic heaths. This is in agreement with biometric measurements which revealed lower growth and leaf area for C. tetragona populations in harsher environments (Havström et al., 1993). Except for Abisko, LAI of communities sampled in the same area did not differ significantly (Table 1). The contribution of C. tetragona to total LAI was substantial, varying from 54% to 90% of the community total across all sites. LAI at Zackenberg was larger than indirect estimates reported for C. tetragona vegetation in the area (0.2–0.3 m² m⁻²; Soegaard et al., 2000; Groendahl et al., 2007). Apart from these, we are not aware of any other information in the literature about LAI of C. tetragona tundras. The LAI of the subarctic tree-line heath is comparable to the LAI of other alpine and subarctic mesic-dry sites such as Empetrum hermaphroditum heath (1.3 m² m⁻²; Street et al., 2007) and dry meadow and birch forest understory (about 1.0 and 1.2 m² m⁻²; Berg et al., 1975). On the other hand, the LAI of the fellfield and high arctic C. tetragona tundras is similar to the LAI of low elevation heaths in northern Alaska (0.6–0.8 m² m⁻²; Williams and Rastetter, 1999; Williams et al., 2000, 2006) and of mesic-dry tundras dominated by dwarf-shrub vegetation (e.g. Dryas spp. and Salix spp.) near the study sites in Svalbard and Greenland (0.4–0.5 m² m⁻²; Soegaard et al., 2000;  http://ecosystems.mbl.edu/arc/terrest/tracegas/index.shtml).

Table 1

Mean values (with SE when appropriate) of leaf area index (LAI), canopy biomass, total foliar nitrogen (TFN), and nitrogen per leaf area (NLA) obtained by plot harvesting of different plant communities dominated by Cassiope tetragona in the subarctic (Abisko) and High Arctic (Zackenberg and Longyearbyen). Plot size varied from 0.01 to 0.04 m² at Abisko and 0.03 to 0.09 m² at Zackenberg and Longyearbyen.

TFN of C. tetragona communities ranged from 1.4 to 1.7 g N m⁻² ground (Table 1), similar to the TFN of the vascular component of many shrub and heath tundras in northern Sweden and Alaska (Williams and Rastetter, 1999; van Wijk et al., 2005). Canopy biomass was highest for the tree-line heath (150 g m⁻²), intermediate for the fellfield and high arctic heath at Longyearbyen (100–130 g m⁻²), and lowest for the high arctic sites at Zackenberg (50–80 g m⁻²) (Table 1). These estimations match similar records for mesic-dry sites in Alaska (Williams and Rastetter, 1999), high arctic Canada (Svoboda, 1987; Bliss et al., 1987), and Svalbard and Greenland ( http://ecosystems.mbl.edu/arc/terrest/tracegas/index.shtml).

METHODOLOGY OF LAI DETERMINATION

Methods of LAI determination based on leaf collection and direct measurement of leaf area are the most accurate and are important for calibration and corroboration of indirect methods (Jonckheere et al., 2004). The most important assumption of our method to estimate LAI of C. tetragona, i.e. that green leaf tissue of C. tetragona does not overlap or present gaps along the stems, was corroborated by visual inspection (Campioli, personal observation). In particular, the arrowhead shape of the leaves (Fig. 1A(1)) assures a negligible overlapping between the bottom of an upper leaf and the top of a lower leaf in the leaf scale-like composition of the shoots (Fig. 1A(2)). The absence of significant overlap and gap agrees with basic physiological principles: (i) continuous self-shading of energy-costly green tissue is inefficient, and (ii) a leaf structure with large gaps and increased water losses contrasts to the xeromorphic nature of C. tetragona. An accurate comparison between LAI estimates obtained from the surface of the C. tetragona leaf-stem combinations and from the total area of C. tetragona leaves peeled off the stems cannot be done because removing individual leaves causes leaf damage and leaf desiccation. Furthermore, the assumption that auxiliary green branches and small stems with a rather flat shape can be considered square-based parallelepipeds, as the main stems, does not affect the LAI estimation. Because of their low amount (5% of the total green biomass), the LAI overestimation caused by neglecting such evidence is minor (3–8%) and not significant. These considerations and, as shown above, the very close agreement between the LAI of C. tetragona vegetation and the LAI of other mesic-dry arctic sites with similar climate and soil fertility make us confident about the soundness of the method applied.

LAI-NDVI RELATIONSHIP

LAI and NDVI for C. tetragona tundras showed a high correlation (p < 0.0001), with a low root mean squared error of LAI prediction (0.14 m² m⁻²) (Fig. 3a), as observed for other tundra plant communities (van Wijk and Williams, 2005; Williams et al., 2006; Street et al., 2007). Such a relationship can therefore be used to determine indirectly LAI of plant communities dominated by C. tetragona from measures of vegetation reflectance.

Figure 3

Relationship of leaf area index (LAI) vs. normalized difference vegetation index (NDVI) for (A) arctic tundras dominated by Cassiope tetragona and generic arctic tundra, and (B) different types of arctic tundras (the LAI-NDVI relationship for generic arctic tundra and for tundras dominated by species other than C. tetragona are from Shaver et al., 2007, and Street et al., 2007, respectively). The equation of exponential regression does not include one outlier at LAI  =  1.67 m² m⁻².

The LAI-NDVI relationship may vary significantly between tundra vegetation types according to canopy structure (e.g. vertical vs. horizontal canopy development, exposed ground fraction, presence of nonvascular vegetation and litter) and leaf characteristics (e.g. structure, orientation, and color) (Steltzer and Welker, 2006; Street et al., 2007). Cassiope tetragona communities generally have low NDVI (especially for LAI < 0.8 m² m⁻²; Fig. 3b), likely due to the erect nature of its green shoots which do not cover the bare ground and the light-grayish and very abundant attached dead leaves. The LAI-NDVI relationship for C. tetragona was, however, similar to those of other tundra vegetation types characterized by low NDVI such as sedge, affected also by abundant litter, and, at low LAI, Salix vegetation, characterized by gray-green leaves (Fig. 3b; Street et al., 2007). On the other hand, the LAI-NDVI relationship for C. tetragona vegetation differed markedly from generic LAI-NDVI relationships for arctic tundra (van Wijk et al., 2005; Shaver et al., 2007; Fig. 3a), which are therefore inadequate to estimate C. tetragona LAI. Such discrepancy was recorded because tussock, B. nana, and heath communities showed, for similar values of NDVI, much lower LAI (Fig. 3b).

Soegaard et al. (2000) estimated LAI for a high arctic C. tetragona heath by using a linear relationship between LAI and the ratio between reflectance in the near infrared and red light. However, tests revealed that the correlation between LAI and the R_NIR/R_RED ratio at our sites was low (R²  =  0.18) and therefore inadequate to be used as an indirect method to estimate LAI. Regression of measured LAI vs. predicted LAI using the model of Soegaard et al. (2000) revealed that it performs rather poorly at our sites (slope of 0.97 but intercept of 0.34; n  =  29).

LEAF TRAITS

SLA of C. tetragona was 12 m² kg⁻¹ (Table 2), which is similar to or less than the SLA of deciduous species common at the investigated sites (11–17 m² kg⁻¹) but larger than the SLA of arctic evergreen shrubs (4–7 m² kg⁻¹; van Wijk et al., 2005). These results agree with observations on leaf thickness, found for C. tetragona to be intermediate between leaf thickness of evergreen and deciduous shrubs (Kudo et al., 1999). SLA of C. tetragona did not differ significantly between the subarctic tree-line heath and the fellfield (Table 2). Populations of C. tetragona growing in different climatic conditions might vary in leaf size because of trade-offs between protection against desiccation and light interception, but larger leaf size regularly corresponds to larger leaf biomass and thus constant SLA in C. tetragona (Havström et al., 1993).

Table 2

Leaf traits of Cassiope tetragona at a subarctic tree-line heath and fellfield: nitrogen per leaf mass (NLM; mean and SE; n  =  3–9), leaf area per mass (LAM; mean and SE; n  =  8–9) and nitrogen per leaf area (NLA; derived from NLM and LAM) (samples taken in late July–mid August). LAM coincides with the specific leaf area (SLA) when the mass of total green leaves rather than total green mass is considered.

Nitrogen concentration of current-year leaves of C. tetragona was significantly larger than N concentration of older C. tetragona leaves (Table 2), as observed for other evergreen species (Jonasson, 1989; Campioli, 2008). Similarly, the leaf N concentration of C. tetragona was comparable to other arctic evergreen shrubs (van Wijk et al., 2005; Campioli, 2008) but lower than in deciduous species at the same sites (e.g. Betula nana at the tree-line heath: 14.8 mg N g⁻¹; Salix polaris × herbacea at the fellfield: 17.4 mg N g⁻¹) or at other arctic locations (van Wijk et al., 2005). The leaf N concentration of C. tetragona at the fellfield was larger than at the tree-line heath (Table 2), likely because of the shorter growing season at the fellfield. Nitrogen concentration of the total green biomass of C. tetragona was slightly lower than the N concentration of the total green leaves (3–10%), because of the lower N content of the stem (e.g. 7.9 mg N g⁻¹ stem at the tree-line heath) which accounted for about 14–15% of the total green biomass. However, the difference was not significant and therefore N concentration of the total green tissue can be used as a surrogate of foliar N for C. tetragona.

The water content of the green biomass of C. tetragona at the tree-line heath and at the fellfield did not differ significantly (67–73 g H₂O 100 g⁻¹ green biomass).

LEAF AREA-BIOMASS RELATIONSHIP

The constancy of SLA and water content for C. tetragona and the strong correlation between leaf area and green biomass (p < 0.0001; Fig. 4), allow the latter relationship to be used to estimate leaf area of C. tetragona from measurement of biomass. The leaf area–green biomass relationship for C. tetragona can be used as a straightforward indirect method to estimate LAI in case of time limitation (e.g. for communities with lush C. tetragona vegetation, the direct measurement of leaf area is time consuming) or when a planimeter for leaf area detection is not readily available at the field laboratory for a quick processing of the samples.

Figure 4

Relationship of leaf area vs. green biomass of Cassiope tetragona for arctic tundras dominated by C. tetragona.

TFN-LAI relationship

The relationship between TFN and LAI showed high correlation (p < 0.0001; Fig. 5), indicating that TFN can be derived from measurements of LAI for C. tetragona tundras. The slope of the TFN-LAI relationship, i.e. the nitrogen per leaf area (NLA), was 1.78 g N m⁻² (Fig. 5), with some variation across communities (1.7–2.2 g N m⁻²; Table 1). At species level, the NLA of C. tetragona (1.89–1.94 g N m⁻²; Table 2) was larger than the NLA of the deciduous species (1.30–1.78 g N m⁻²).

Figure 5

Relationship of total foliar nitrogen (TFN) vs. leaf area index (LAI) for different arctic communities dominated by Cassiope tetragona. The TFN-LAI relationship for generic arctic tundra (van Wijk et al., 2005) is also shown.

A tight correlation between TFN and LAI is common for tundra ecosystems (Williams and Rastetter, 1999; van Wijk et al., 2005). Furthermore, van Wijk et al. (2005) found a generic TFN-LAI relationship for both Swedish and Alaskan tundra up to an LAI of 1 m² m⁻². A comparison of our data with the ones published by van Wijk et al. (2005) revealed no significant difference in the slope of our TFN-LAI relationship for C. tetragona tundra and their generic value for arctic tundra (1.9 g N m⁻²). The inclusion of C. tetragona tundra would have a minimal impact (1% variation) on the slope and the correlation parameters of the generic relationship. Thus, the latter can be used to derive TFN also in arctic landscapes with plant communities dominated by C. tetragona.