SEARCH

SEARCH BY CITATION

Keywords:

  • Anser brachyrhynchus;
  • geese;
  • growth form;
  • grubbing;
  • moss;
  • plant functional types;
  • recovery;
  • soil moisture;
  • Svalbard;
  • tundra

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Acknowledgements
  7. References
  8. Supporting Information

1. Understanding the impact of disturbance on vegetation and the resilience of plant communities to disturbance is imperative to ecological theory and environmental management. In this study predictors of community resilience to a simulated natural disturbance are investigated. Responses to disturbance are examined at the community, plant functional type and species level.

2. Field experiments were set up in seven tundra plant communities, simulating disturbance based on the impact of grubbing by an increasing herbivore population of pink-footed geese (Anser brachyrhynchus). The short-term resilience of communities was assessed by comparing community dissimilarity between control plots and plots subject to three disturbance intensities based on the foraging impact of these geese. Potential for long-term recovery was evaluated across different disturbance patch sizes.

3. Resilience to disturbance varied between communities; those with higher moss cover and higher soil moisture, such as wetlands and mires, were most resilient to disturbance.

4. The wetter communities demonstrated greater long-term recovery potential following disturbance. In wetland communities, vegetative recovery of vascular plants and moss was greater in smaller disturbed patches and at the edges of patches.

5. The response of vegetation to disturbance varied with intensity of disturbance, plant community and plant species. The use of functional type classifications only partially explained the variation in species responses to disturbance across communities, thus their use in predicting community changes was limited.

6.Synthesis. The impact of disturbance is shown to be plant-community specific and related to the initial abiotic and biotic properties of the community. By showing that resilience is partly predictable, the identification of disturbance-susceptible communities is possible, which is of relevance for ecosystem management.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Acknowledgements
  7. References
  8. Supporting Information

The impact of disturbance on vegetation and the prediction of plant community responses to disturbance are important themes in ecology and environmental management across biomes from tropical forests to the arctic tundra (Lavorel et al. 1997; Forbes, Ebersole & Strandberg 2001). Responses to perturbation within a system can vary both between plant communities and between species (e.g. Pakeman 2004; Bråthen et al. 2007). Thus, plant functional types or growth forms are often used as a basis for simplifying or predicting vegetation responses, particularly within modelling frameworks (Chapin et al. 1996; Dorrepaal 2007). However, such classifications may not capture variability in community response to perturbations (Dormann & Woodin 2002; Dorrepaal 2007; Sasaki et al. 2008). Here, to broaden understanding of how plant communities and the species within them respond to disturbances, the impact of disturbance on a range of high-arctic tundra plant communities and their recovery potential are investigated at the community, functional type and species level.

Resilience, the ability of a system to return to its original state following a disturbance, is generally thought to be higher within species-rich communities, as the diversity in species’ responses to perturbation potentially allows ecosystem functioning to be maintained following disturbance (Elmqvist et al. 2003; Folke et al. 2004). However, variation in response to perturbation can also lead to community changes if, for example, formerly abundant species decrease and less abundant species increase, and these community changes may result in changes in ecosystem properties (Mulder 1999). In cases where the performance of all species within a community is strongly and disproportionately negatively affected by disturbance, reduced ecosystem functioning and shifts in ecosystem state are likely (Elmqvist et al. 2003; Folke et al. 2004). Herbivores, particularly larger vertebrates, may cause such shifts in ecosystem state in many systems through disturbing the vegetation and soil, and as a consequence of their selectivity. For example, the impact of vertebrate herbivores has led to the formation of alternate stable states through widespread vegetation loss and soil degradation in both the Sahel (Rietkerk & Van de Koppel 1997) and nearctic salt marshes (Jefferies, Jano & Abraham 2006). However, herbivore-driven state shifts are not necessarily so catastrophic, and may result in predictable and reversible vegetation state changes without dramatic reductions in ecosystem productivity (Van der Wal 2006).

Disturbance may be characterized by its intensity, frequency and spatial extent (Walker & Walker 1991). However, recovery of plant communities from disturbance will depend not only on the nature of the disturbance, but also on the sensitivity of the vegetation and ecosystem to that particular disturbance and to the subsequent exposure to the modified abiotic conditions. Plant responses to disturbance are likely to be complex and involve a range of processes including stress resistance, vegetative reproduction and the ability of propagules to establish in disturbed patches (Walker & Walker 1991; Forbes, Ebersole & Strandberg 2001). The impact of disturbance on a plant species will vary with the nature of the disturbance; some disturbances will favour some trait types, whilst the same traits may be disadvantageous when subject to other disturbances. For example, different types or scales of disturbance benefit either sexual or vegetative reproduction in forest tree species (Bond & Midgley 2001).

Due to plant–plant interactions, which are frequently net-positive in abiotically stressed environments (Callaway et al. 2002) such as the High Arctic, the resilience of a plant species to perturbation will depend on the response of neighbouring plants to the disturbance. Where vegetation is decreased or lost, these positive interactions are likely to decrease in both magnitude and frequency. Furthermore, vegetation loss can accentuate abiotic conditions in some communities, for example salinity in nearctic salt marshes increased with vegetation loss (Jefferies, Jano & Abraham 2006). Such accentuation of abiotic conditions can prevent vegetation recovery, particularly when the size of disturbed patches is large (McLaren & Jefferies 2004). Where vegetation loss occurs, recovery is likely to be extremely slow if the disturbance also causes loss of soil organic layers and vegetative propagules (Vavrek et al. 1999; Forbes, Ebersole & Strandberg 2001). On the other hand, disturbance can increase seedling establishment by creating gaps (i.e. safe sites) in otherwise closed vegetation (Hobbs & Huenneke 1992; Cooper et al. 2004; Sutton, Hermanutz & Jacobs 2006). The response of a community to a disturbance will thus partly depend on both the modification of abiotic conditions by disturbance and the facilitative effect of remaining vegetation. As these parameters are likely to vary between communities, community-specific responses can be expected.

Understanding the impact of disturbance on vegetation is particularly imperative in systems with extreme abiotic conditions, as it has been suggested that catastrophic state shifts are more likely in these systems (Didham, Watts & Norton 2005). Arctic tundra is one such system, it may be limited by low temperature and moisture availability, permafrost processes and growing season length. The arctic vascular plant flora is relatively rich in hemicryptophytes (sensuRaunkiaer 1934) and is thus particularly vulnerable to herbivory and disturbances which penetrate the soil or moss surface. Geese from the genera Anser and Chen forage for below-ground plant parts during spring (this foraging is known as grubbing) and populations of these species have risen over the past few decades in both the Nearctic and Palaearctic because of changes in agricultural practice, decreasing hunting pressure and conservation measures in their winter ranges and along migratory flyways (Jefferies, Rockwell & Abraham 2004; Fox et al. 2005). Grubbing, as a mechanism of foraging, causes disturbance and destruction to vegetation and, with increasing goose populations, has led to large-scale and persistent vegetation degradation in some Nearctic systems (Jefferies, Jano & Abraham 2006). In Svalbard, in the European High Arctic, grubbing also occurs on a large spatial scale (Speed et al. 2009); the landscape effect of this grubbing is unknown and will depend on the impact of this disturbance on vegetation, and the vegetation’s ability to recover.

The current study aims to assess the impact of disturbance as caused by grubbing geese on high-arctic vegetation, and the recovery potential of the vegetation. The study addresses the following questions: (i) Are all plant communities equally resilient to the same disturbance, and, if not, are differences predictable based upon abiotic conditions and biotic properties of the community? (ii) Does the potential for long-term recovery of plant communities depend on scale of disturbance? and (iii) Can functional grouping (based on plant functional types or growth forms) be used to predict disturbance response across plant communities? To address these questions, field experiments were undertaken in the High Arctic, in which the disturbance caused by grubbing geese was simulated over a range of intensities and patch sizes. The impact of disturbance was measured three growing seasons after treatment application and is thus the net result of the original effect of disturbance and initial recovery from it. This can be taken to represent short-term resilience; it was assumed that resistance (the ability of a community to remain unchanged by disturbance) did not factor in the vegetation response as the disturbance was applied indiscriminately of the initial vegetation state, thus the extent of disturbance was fixed. Furthermore, the potential for long-term recovery in relation to disturbance patch size and proximity of surrounding vegetation was addressed by quantifying seedling and vegetative establishment following high intensity disturbance (i.e. complete vegetation removal) across patch sizes and between the edges and centres of disturbed patches.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Acknowledgements
  7. References
  8. Supporting Information

The experimental study was based within Adventdalen, Svalbard, Norway (78°10′ N, 16°05′ E), and experiments were carried out during the summers 2006–2008. Adventdalen is a wide, formerly glaciated valley running from east to west on the island of Spitsbergen. The mean annual temperature in Adventdalen was −2.7 °C over the study period with a mean summer (June to August) temperature of 5.9 °C over the same years, whilst the mean annual precipitation was 186 mm (recorded at Svalbard airport, approximately 10 km from the field site; http://www.eklima.no). Five replicate blocks of each of seven plant communities were chosen within a 5-km2 area on the southern side of the valley. Six of these communities were on a gradient of soil moisture, the seventh was a salt marsh; moss cover, moss depth (including live and senescing parts of moss shoots) and soil moisture (conductivity measured with a Delta T theta probe) in these seven communities are summarized in Table 1. The selected communities were based on descriptions of Rønning (1996). The Dryas ridge community [RD] was characterized by the shrubs Dryas octopetala and Salix polaris and the herb Saxifraga oppositifolia, with low moss cover and a high degree of bare ground. Cassiope heath [CH] was also dry, and dominated by the ericaceous shrub Cassiope tetragona, with D. octopetala and a few scattered herbaceous species also present. Luzula heath [LH] had a higher moss cover than the previous two communities and was characterized by the wood rushes Luzula confusa and Luzula arctica, along with S. polaris. This community had higher soil moisture than the Dryas ridge and Cassiope heath during early summer, but lower soil moisture later in the summer. The wet moss tundra community [WM] had wetter soil and a lush moss-mat mainly comprising species from the genus Aulacomnium and Tomentypnum nitens, whilst Alopecurus borealis, S. polaris and Equisetum arvense characterized the vascular plant community. Dupontia wetland [DW] had a similar bryophyte community to the wet moss tundra but wetter soil and high abundance of the graminoid Dupontia fisheri and the horsetail E. arvense. The freshwater mire [FM] was the wettest community with Calliergon spp dominating the bryophyte community and Eriophorum scheuchzeri and Ranunculus hyperboreus being the only vascular species regularly encountered. In addition, a salt marsh community [SM] with very different abiotic conditions was used, where moss was almost completely absent and Carex subspathacea and Puccinellia phryganodes were the only common vascular plant species found. Soil salinity was assessed in salt marsh plots by taking soil cores to 50 mm depth, drying for 48 h at 80 °C and assessing conductivity in a shaken suspension of 5-g subsample of homogenized soil in 100 mL deionized water (Primo 5 conductivity meter; HANNA Instruments Inc., Woonsocket, RI, USA).

Table 1.   Summary of the characteristics of the seven studied tundra communities
Community (blocks)Vascular species richness (± SD)Moss cover % (± SD)Moss depth cm (± SD)Soil moisture averaged from June to August %vol (± SD)
  1. Data collected from the unmanipulated control plot (35 × 35 cm) of each block (n in parentheses) during August 2008. Moss depth includes green and brown moss, not dead moss. Soil moisture averaged from three readings (June, July and August 2008). Means and standard deviations (SD) are shown. Salinity for the salt marsh community is given in the Results section

Dryas ridge (n = 4)9.8 (3.7)35.93 (10.27)4.80 (1.98)28.73 (8.45)
Cassiope heath (n = 5)9.4 (2.2)71.28 (10.57)8.42 (3.70)34.72 (9.89)
Luzula heath (n = 5)10.2 (2.5)75.84 (13.18)7.66 (2.35)27.73 (4.89)
Wet moss tundra (n = 5)7.4 (2.0)89.25 (6.20)10.37 (1.50)67.60 (22.40)
Dupontia wetland (n = 5)6.2 (1.3)89.36 (5.98)9.16 (1.82)81.17 (16.44)
Freshwater mire (n = 5)3.8 (0.8)88.92 (10.20)11.10 (2.42)100 (0)
Salt marsh (n = 5)2.6 (2.1)10.26 (15.38)0.62 (0.57)39.57 (7.85)

Disturbance treatments modelled on grubbing by pink-footed geese (Anser brachyrhynchus), with variations in both the intensity and spatial scale of disturbance, were applied to plots in all five blocks in all communities during early June 2006. The design of the treatments was based upon detailed observation of naturally grubbed vegetation within Adventdalen. Three levels of disturbance intensity treatments were applied and undisturbed controls selected within 35 × 35 cm plots to investigate the impact of disturbance on vegetation. Low-intensity grubbing disturbance was implemented using a sharpened 20 mm diameter steel tube inserted to a depth of 50 mm and twisted to remove and export contents from the plot; this treatment was applied in a regular fashion to 31% of the plot area. In the mid-intensity treatment 50% of the vegetation was removed by knife with the remainder being fragmented across the plot using the steel tube. High-intensity disturbance was created by using a knife to remove 100% of the vegetation to a depth of 50 mm. To test the effect of different disturbance patch size on vegetation recovery potential, high-intensity disturbance was applied to plots at three spatial scales; 10 × 10 cm, 35 × 35 cm and 50 × 50 cm. Within blocks, plots were positioned within homogenous vegetation in terms of cover, community composition and slope and aspect, and treatments were randomly assigned to plots.

During July 2008, the third growing season following treatment application, vascular plant shoots were counted at the species level (terminal shoots counted for shrubs and herbs, basal shoots for graminoids), seedlings were counted as herbs or graminoids, and total moss percentage cover and depth were assessed by eye in each cell of a gridded quadrat (quadrat cells 7 × 7 cm (7.14 × 7.14 cm in the 50 cm plots); the 10 × 10 cm plots were recorded as a single 10 × 10 cm cell). To indicate the potential for long-term recovery, grid cell data were used for comparing the recovery of vascular plants, moss cover and the number of seedlings (taken to include growth from vegetative bulbils) between the edges (7-cm strip comprising the cells around the perimeter of the plot) and centre (the centre cells of the plot, excluding the outer 7 cm) of high-intensity disturbance patches. Vascular shoot counts and seedling numbers were also summed for whole plots, and mean plot values were calculated for moss cover and depth. Data were expressed on a per m2 basis.

The impact of disturbance on vegetation was expressed as community dissimilarity between treatments and control plots within each block. It was assumed that because of the direct and standardized nature of the disturbance treatment, observed responses constituted (short-term) resilience only, and not resistance. Thus, if a treated plot is similar to the control plot, it is deemed to be resilient to that disturbance in the short-term. Dissimilarity was calculated based on vascular species composition and abundance using the Ružička index (RI); RI = 2D/(1 + D), where D is dissimilarity between communities j and k given by inline image, within which xi is the number of shoots of species i in community j or k, summed for all species present in communities j and k. This was calculated within the vegan package in R.2.7 (Oksanen et al. 2008; R Development Core Team 2008). This index was selected as it is a proportional difference, thereby allowing for direct comparisons between communities. An RI value of 1 indicates maximum dissimilarity where no species are shared, whilst 0 indicates the same species are present at equivalent abundances.

To infer the importance of prior abiotic and biotic conditions on the impact of disturbance on plant communities, the dissimilarity (estimated with RI) was modelled against moss cover, soil moisture and vascular plant species richness of control plots using these as linear or, where appropriate, quadratic terms within general linear models. The potential explanatory variables were assessed for collinearity by calculating variance inflation factors. In all cases the variance inflation factor was below 2.0, indicating no substantial collinearity between these predictors (Graham 2003).

To assess the efficiency of functional groupings in explaining response to disturbance across communities, general linear models were used that test the ability of functional groups to predict species response across communities. Species response was expressed as the proportional difference in loge-transformed shoot abundance between treatment and control plots for all species present in each block.

  • image

General linear models were run for species response at each level of disturbance intensity, with community and one of five functional groupings as predictor variables. Based upon likelihood ratio tests and Akaike’s Information Criterion (AIC), models of species response as a function of functional group were compared to models in which species identity was used in place of functional grouping (see Appendix S1 in Supporting Information for functional groupings; each species is represented exactly once in each of the five functional grouping classifications). Three of the functional groupings were based on the functional type classification proposed by Chapin et al. (1996) at three different hierarchies, the fourth was the growth form grouping of Raunkiaer (1934) and the fifth a dichotomous division based on whether or not species were selectively grubbed by geese (Fox & Bergersen 2005; Fox, Francis & Bergersen 2006; Van der Wal et al. 2007 and personal observations).

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Acknowledgements
  7. References
  8. Supporting Information

Across plant communities, the abundance of vascular plants decreased with increasing intensity of disturbance (Fig. 1a). The number of shoots in high disturbance intensity plots was lower by on average 74% than that of shoots in control plots, although the decrease in shoot number was greatest in the drier communities (Fig. 1a). The number of shoots in the wettest community, the freshwater mire, was generally much lower than that in other communities and relatively constant between treatments, with the number of shoots after high-intensity simulated grubbing lower by only 30% than that in control plots (Fig. 1a). Vascular species richness (number of species per 35 × 35 cm plot) decreased under high-intensity disturbance relative to control and low-intensity disturbance in all but the two wettest communities (Dupontia wetland and freshwater mire) and the salt marsh. In the Dryas ridge, the vascular species richness decreased from a mean of 9.8 in the control treatment to 4.0 in the high-intensity disturbance treatment, whilst in the wetter communities and salt marsh species richness remained more or less unchanged across treatments (Fig. 1b). Moss cover consistently decreased with disturbance intensity across all communities (although not significantly in the salt marsh) with moss cover in high-intensity treatments being reduced by between 99.6% (Dryas ridge) and 44% (freshwater mire; Fig. 1c).

image

Figure 1.  (a) Vascular plant shoot density (sum of all species) (b) species richness and (c) % moss cover under different disturbance intensity treatments in different communities. The first six communities are on a gradient of increasing soil moisture, whilst the salt marsh is separated (as salinity differentiates it from the other communities). Measurements were taken three summers after treatment application. Means and standard errors of the mean between replicates of communities shown; n = 5 except for RD where n = 4. RD: Dryas ridge, CH: Cassiope heath, LH: Luzula heath, WM: Wet moss tundra, DW: Dupontia wetland, FM: Freshwater mire, SM: Salt marsh. Treatments sharing a letter above the bar do not significantly differ within community at P = 0.05 (Tukey’s HSD test).

Download figure to PowerPoint

Increasing intensities of disturbance also led to changes in vascular plant community composition, as shown by the increases in the Ružička index of dissimilarity in vascular plant communities between treated and control plots in all communities (Fig. 2). Generally, community dissimilarity increased with the intensity of disturbance (Fig. 2). The degree of dissimilarity caused by disturbance varied between communities, with the Dryas ridge, Cassiope and Luzula heaths being most altered at all intensities of disturbance and the salt marsh being heavily impacted by disturbance at mid and high intensities of disturbance. The wetter communities, i.e. wet moss tundra, Dupontia wetland and the freshwater mire, were the least impacted by disturbances at all intensities, although there was a large variation in dissimilarity between high-intensity and control plots within the freshwater mire (Fig. 2).

image

Figure 2.  Community dissimilarity (Ružička index) between treated plots and control plots within blocks, based on abundance of vascular species within communities, measured three summers after treatment application. High community dissimilarity is assumed to reflect a high impact of disturbance on that community and a low short-term resilience. Boxes represent interquartile range, bold lines the median and whiskers 95% confidence intervals, outliers shown by open circles. RD: Dryas ridge, CH: Cassiope heath, LH: Luzula heath, WM: Wet moss tundra, DW: Dupontia wetland, FM: Freshwater mire, SM: Salt marsh. Communities sharing a letter above the whiskers do not significantly differ at P = 0.05 (Tukey’s HSD).

Download figure to PowerPoint

The impact of disturbance on vegetation (measured by treatment–control dissimilarity in vascular plant communities) was significantly and negatively related to soil moisture in the control plots, with the highest impact where soils were drier (Table 2, Fig. 3). In addition, across all communities and at all intensities of disturbance a significant nonlinear relationship between the impact of disturbance and moss cover in the control plots was found (Table 2). At low-intensity disturbance, the highest impact of disturbance was found where moss cover was intermediate (around 60%); whilst at mid and high intensities of disturbance, the impact was highest when moss cover was below 70% (Fig. 3), and gradually decreased at higher moss cover. At all intensities of disturbance, a significant degree of variation in disturbance impact across plant communities was explained jointly by soil moisture and moss cover, but not by vascular species richness (Table 2). Within the salt marsh community, salinity (measured by conductivity) was 135.9 ± 45.3 μS cm−1 (mean ± SE) in the control plots. Salinity was significantly higher under mid (269.4 ± 70.0 μS cm−1) and high (260.9 ± 64.5 μS cm−1) intensity disturbances (P < 0.01 for both), whilst under low intensity disturbance, there was a tendency for an increase in salinity (174.7 ± 37.4 μS cm−1), but this was not statistically significant.

Table 2.   Predictors of the impact of disturbance (treatment–control community dissimilarity) on tundra plant communities at different intensities
Predictors of toleranceModelP
Disturbance intensitySoil moisture FSoil moisture PMoss cover FMoss cover PR2
  1. Model selection did not retain vascular species richness as a predictor in any model. Moss cover as a quadratic term significantly improved models relative to a linear term at all disturbance intensities. The linear term was only retained in the low-intensity disturbance model. Variance inflation factors were below 2.0 for all variables indicating no substantial collinearity.

Low8.37<0.01Linear: 11.97 Quadratic: 8.68<0.01 <0.010.44<0.001
Mid65.11<0.001Quadratic: 13.95<0.0010.70<0.001
High48.81<0.001Quadratic: 15.13<0.0010.65<0.001
image

Figure 3.  Relationship between treatment and control plot community dissimilarity (Ružička index) three summers after treatment application and (a) soil moisture and (b) moss cover at different disturbance intensities. A value of 1 shows high dissimilarity (no shared species) and a value of 0 indicates that communities are identical in terms of species composition and abundance. High community dissimilarity is assumed to reflect a high impact of disturbance on that community and a low short-term resilience. Polynomial regression lines are shown for moss cover, linear regressions for soil moisture.

Download figure to PowerPoint

In addition to the intensity of disturbance, the spatial scale of disturbance also affected recovery within communities. Whilst generally a greater number of vascular plant shoots per unit area (totalled across species) was found in smaller disturbed patches (Fig. 4a) than in larger patches, these differences were not significant. There was no significant interaction between disturbance patch size and habitat, indicating that the size of disturbed patches had a consistent effect across communities, although the level of reduction in shoot density in the largest disturbed patches relative to the smallest patches ranged from over 70% in the Dryas ridge to under 10% in the Cassiope heath. Furthermore, there were more vascular shoots at the edge relative to the centre of high-intensity disturbed plots across all plot sizes, with the number of shoots in the centre of plots being between 32% (wet moss tundra) and 75% (Luzula heath) lower than the number of shoots at the edges of plots (Fig. 5a). Moss cover following disturbance was also greater at the patch edges relative to the centre (Fig. 5b) and in smaller patches (Fig. 4b), with notably high moss cover at the edges of disturbed patches in the freshwater mire. Whilst vascular plant shoot density and moss cover were often greater in smaller disturbed plots, seedling density was independent of disturbance patch size, and seedlings were equally abundant at the centre of patches and at the edge in all communities (Figs 4c and 5c).

image

Figure 4.  (a) Density of vascular plant shoots (b) moss cover (%) and (c) density of seedlings in control plots and different high-intensity disturbed patch sizes. Means ± SE shown, n = 5 except for RD where n = 4. Data were gathered three summers following treatment application. RD: Dryas ridge, CH: Cassiope heath, LH: Luzula heath, WM: Wet moss tundra, DW: Dupontia wetland, FM: Freshwater mire, SM: Salt marsh. Treatments sharing a letter above the bar do not significantly differ within community at P = 0.05 (Tukey’s HSD test).

Download figure to PowerPoint

image

Figure 5.  (a) Density of vascular plant shoots (b) moss cover (%) and (c) density of seedlings, at the edge (within 7 cm of edge) and centre of high-intensity disturbed plots (35 × 35 cm plots). Means ± SE are shown, n = 5 except for Dryas ridge (RD) where n = 4. Data were gathered three summers after treatment application. RD: Dryas ridge, CH: Cassiope heath, LH: Luzula heath, WM: Wet moss tundra, DW: Dupontia wetland, FM: Freshwater mire, SM: Salt marsh. Locations sharing a letter above the bar do not significantly differ within community at P = 0.05 (Tukey’s HSD test).

Download figure to PowerPoint

There were species-specific responses to disturbance across communities. Poa and Stellaria species along with R. hyperboreus were the least negatively impacted by disturbance at all intensities across communities, whilst Bistorta vivipara and A. borealis showed no response or a positive response after low-intensity disturbance, but were negatively impacted at high intensities of disturbance (Table S1). Dryas octopetala and E. arvense were negatively affected by disturbance; S. polaris was generally negatively affected, except for within the freshwater mire community where it had a positive response to low-intensity disturbance (Table S1). Shoot densities of S. polaris were also much higher at the edges of disturbed patches than in patch centres in all five communities in which the species occurred (Table S2). Bistorta vivipara and the graminoids C. subspathacea, A. borealis and D. fisherii also had higher shoot densities at the edges of disturbed patches than in the centre in some communities (Table S2).

The response to disturbance, in terms of vascular shoot number, is thus seen to vary depending on community, intensity of disturbance and species identity. Community type and species identity best predicted the variance within each disturbance intensity (R2 ≥ 0.45; Table 3), with species identity having a significant effect on plant response at all intensities of disturbance (Table 3). Functional classifications explained much less of the variance in response: the second of the Chapin functional classifications (Chapin II in Table 3) was the only classification to explain over half of the variance explained by species identity (at mid- and high-intensity disturbance). However, each of the functional groupings did have a significant effect, or interactive effect, on plant response for at least one of the disturbance intensities. The interactions observed between functional classifications and community at all intensities indicate that the response of the functional groups to disturbance differed between communities. Despite explaining much less variance than the species model, many of the functional grouping models were selected by AIC (Table 3). The first of the Chapin functional groupings (Chapin I in Table 3) was selected by AIC at high disturbances, while the grubbed/ungrubbed species grouping was selected at mid intensity and the Raunkiaer classification at low intensity of grubbing.

Table 3.   Model parameters comparing the use of functional classifications in explaining plant response to disturbance at different intensities across plant communities
IntensityFunctional groupCommunity (F, P)Functional group (F, P)Interaction (F, P)R2AICComparison with species model (F, P)
  1. Individual models were run for each disturbance intensity and functional classification. Plant response was calculated as the standardized difference in shoot density between treatment and control plots: Plant response = (loge[Treatment shoots+1]−loge[Control shoots+1])/(loge[Control shoots+1] +1). Species were removed where the species was not present in any plots within block (Model: Plant response = Community × Functional Group). F-values and significance levels shown, *P < 0.05, **P < 0.01, ***P < 0.001, along with R2 and Akaike’s Information Criterion (AIC). Each of the functional group models are compared with a model using species identity in place of functional grouping (F test within anova). Groupings within each classification are given as footnotes; for groups that the species are assigned to see Appendix S1

  2. †Chapin 1: shrub, herb, graminoid, pteridophyte.

  3. ‡Chapin 2: aerenchymatous, deciduous shrub, evergreen shrub, non-aerenchymatous.

  4. §Chapin 3: deciduous shrub, evergreen shrub, Cyperaceae, Poaceae, Juncaceae, herbs, pteridophytes.

  5. ¶Raunkiaer: chamaephytes, hemicryptophytes, cryptophytes.

  6. ††Grubbed: species selectively grubbed, species not selectively grubbed.

LowSpecies1.441.81**1.46*0.45772
Chapin I†1.311.672.35**0.16745.11.35*
Chapin II‡1.301.001.85**0.18757.11.43*
Chapin III§1.321.192.29**0.18746.91.33
Raunkiaer1.310.043.19***0.14739.81.32
Grubbed††1.233.331.540.07751.51.53**
MidSpecies8.63***3.03***0.810.50620.3
Chapin I†7.80**1.571.97*0.235951.37
Chapin II‡7.77***1.791.480.26606.21.44*
Chapin III§7.18***2.110.400.18624.41.74**
Raunkiaer7.48***6.21**0.490.19603.31.52**
Grubbed††7.65**1.342.83*0.18589.71.41*
HighSpecies4.87***1.91**1.38*0.48601.3
Chapin I†4.46***1.162.49**0.20571.91.32
Chapin II‡4.52***1.891.96**0.25577.71.30
Chapin III§4.35***2.78*1.460.20583.91.44*
Raunkiaer¶4.12***2.740.710.12591.11.61**
Grubbed††4.14***1.671.160.11582.91.56**

Disturbance impact

Both abiotic and biotic properties of ecosystems have been associated with a community’s ability to withstand and recover from disturbance. Diverse communities are often the most resilient to disturbances (Elmqvist et al. 2003). In contrast, in the current study the least impacted community was the freshwater mire, which was the community with the lowest vascular plant species richness. However, the freshwater mire was also the wettest community and had the highest moss cover. At all intensities of disturbance there were significant relationships between resilience (the inverse of community dissimilarity) and both soil moisture and moss cover, with community resilience to this disturbance being higher when soil moisture was high and when moss cover of the original community was greater than 70%. This resilience may be due to rhizomes being protected from environmental exposure and further disturbance by the moss cover that remains or recovers after disturbance, and that rhizomes extend deep into the organic soil layers that are found in wetter communities. Another abiotic factor, salinity, might have been expected to interact with the impact of disturbance on the salt marsh community, as has been shown in nearctic salt marshes subject to grubbing by geese (McLaren & Jefferies 2004). In the Svalbard salt marsh, salinity did increase with increasing disturbance intensity and the salt marsh plant community was heavily impacted by mid- and high-intensity grubbing, although in contrast to the Nearctic system, the size of the disturbed patches did not affect resilience. However, salt marsh communities in Svalbard are spatially limited and rarely grubbed and so, considering the broader range and extent of tundra communities, it appears to be the interaction of grubbing with soil moisture and moss cover that is most important in controlling vascular plant response to such disturbances in Svalbard. Indeed, this may even be true for the salt marsh as it had the lowest moss cover recorded within control plots and was also one of the communities most impacted by simulated grubbing at mid-and high-intensities.

Disturbance impact was measured by community dissimilarity (Ružička index) three growing seasons after disturbance and hence reflects the net outcome of the original effect of disturbance and of initial recovery from it. This measure of impact could be regarded as the inverse of community resilience. As expected, the impact of disturbance on plant communities increased with its intensity. Across communities, disturbance had a negative impact on the abundance and diversity of vascular plants; the impact increased with increasing intensities of disturbance (Fig. 1), similar to findings in many other systems (e.g. Cole 1995; Walker, Reynolds & Tenhunen 1996; Rapport et al. 1997). Communities differ in their response to disturbance of this nature as shown by the variation in dissimilarity caused by the disturbance treatment between communities. The vascular plant communities in the drier habitats of Dryas ridge, and Cassiope and Luzula heaths were more affected than those in the wetter habitats of freshwater mire, Dupontia wetland and wet moss tundra, even at low disturbance intensities.

Olofsson (2006) found that within tundra communities, resilience to grazing did not increase with species richness, and proposed that because low-nutrient conditions were related to both increases in diversity and decreases in regrowth capacity, resilience of diverse communities was limited by nutrient availability rather than diversity. The current study provides findings complementary to this: diversity (species richness) was not a predictor of the impact of disturbance, whilst other abiotic conditions (habitat wetness) and biotic conditions (moss cover) were strong predictors at all disturbance intensities. Previous studies have attributed differences among arctic tundra communities in the impact of disturbance on vegetation to both differing resistance and resilience (e.g. Felix et al. 1992; Strandberg 1997). Here it has been shown that the short-term resilience of a tundra plant community to a disturbance modelled on goose herbivory is predictable, based on the soil moisture and moss cover in the community.

Long-term recovery potential

The high-disturbance treatment in this study involved complete removal of the above-ground vegetation; therefore the presence of vegetation three years following this treatment indicates recovery. We used the comparison between vascular plant shoot and seedling numbers and moss cover regenerating in high-disturbance plots of different sizes, and at plot edges vs. centres, to predict the potential for long-term recovery and to investigate whether adjacent intact vegetation facilitates recovery. In some communities, both mosses and vascular plant shoots showed greater potential for recovery in small than in larger, high-disturbance plots. This is similar to the result of McLaren & Jefferies (2004), who found a nonlinear effect of disturbance patch size on vegetation cover in a Nearctic salt marsh. Comparing between communities, across all plot sizes, recovery of vascular plant shoot density and moss cover from high disturbance was greatest in the wetter communities. Moss cover recovery was particularly high in the freshwater mire, presumably due to the wet conditions allowing moss shoots to grow and spread inwards.

Secondary succession can occur through both seedling establishment and clonal growth from existing vegetation. Recovery of vascular plant shoots was greater at the edges than the centres of high-disturbance plots, resulting from the clonal growth of plants in the surrounding vegetation in moist and wet communities that are dominated by plants with a relatively high potential of lateral growth, such as S. polaris and rhizomatous graminoids, such species may also benefit from resource translocation within extensive clonal networks (Jónsdóttir & Callaghan 1988; D’Hertefeldt & Jónsdóttir 1999). Accordingly, this recovery process was less pronounced in the communities dominated by evergreen shrubs with relatively slow lateral growth (the Dryas ridge and Cassiope Heath). The potential for vegetation recovery through seedling (and bulbil) establishment was slightly higher in the drier communities, where the number of seedlings in disturbed plots did not differ from control plots, than in wetter communities where seedling establishment was reduced by disturbance. Overall, seedling establishment was unrelated to disturbance patch size and to plot edge effects. Disturbance did not increase seedling establishment as expected; this suggests that the provisioning of gaps is less important than the facilitatory effect of existing vegetation on seedling establishment in these communities.

It has been proposed that recovery from disturbance in arctic tundra depends on the nature of the disturbance with regard to the fate of vegetative propagules, rather than on changes in abiotic conditions (Vavrek et al. 1999); however, in the case of the Nearctic salt marsh, increasing salinity caused by disturbance is a clear inhibitor of vegetation recovery (McLaren & Jefferies 2004). In the current study, plant communities responded differently to disturbance, depending in part on the abiotic factor of soil moisture. Nevertheless, the mechanism of recovery from high-intensity disturbance was relatively consistent across communities; vascular plants and mosses exhibit clonal growth, resulting in recovery that can be patch size and edge-dependent, whilst seedlings establish independently of patch size and location in relation to disturbance patch edges.

Functional groups as predictors of vegetation response

Functional groupings of plants are often used for generalizing ecological responses to perturbations, although their suitability for making predictions has been found to be inconsistent in tundra communities (e.g. Chapin et al. 1996; Dormann & Woodin 2002). Functional types were originally proposed to be based upon species’ responses to environmental factors (e.g. Grime 1979). In practice, however, functional groups are more often based upon growth form (Lavorel & Garnier 2002), although functional traits are becoming more frequently utilized (McGill et al. 2006). In this study, the response of vascular plant abundance to disturbance varied between communities, intensities of disturbance and species. The ability of several functional groupings of plants to predict the response to disturbance across several communities was compared, but as expected, use of any of the group classifications explained far less of the variation in disturbance response than the use of species identity. The functional groupings differed in terms of the variation explained in species responses to disturbance. As the presence of subterranean propagules may make a plant population resilient to surface disturbances, it might be expected that the Raunkiaer growth form, which groups species by the location of overwintering buds (Raunkiaer 1934), would be the most appropriate to predict the response of individual species response to disturbances. However, the Raunkiaer classification was one of the groupings which explained least of the variance in plant response at higher intensities of disturbance. The classifications based on Chapin et al. (1996) were generally better models, although the hierarchical levels differed in terms of variance explained and efficiency.

The functional group models were more efficient AIC than the species model and thus do provide some basis for the prediction of the response of vegetation to disturbance. However, in many cases the functional group models were significantly different from the model including a species identity term, and in most cases, explained less than half as much variance in plant response. Variation in the nature and magnitude of species response to perturbation is generally regarded as enhancing community resilience (e.g. Elmqvist et al. 2003), although species richness did not actually influence resilience in our study. Thus to ignore variation in response within functional groups may be to ignore or underestimate a community’s potential resilience, and resulting shifts in community composition or ecosystem states may not be predicted.

Implications for the tundra

Disturbance differentially impacts tundra plant communities, and this is predictable based on the wetness of the community and the amount of moss cover. It has been predicted that systems under strong abiotic regulation are more likely to undergo catastrophic shifts in system state (Didham, Watts & Norton 2005). Here it has been found that drier plant communities are both less resilient to disturbance modelled on foraging geese in the short term and also show less potential for recovery and so can be predicted to be less resilient in the long term. Although wetter communities are more likely to be grubbed by geese, drier communities are also utilized, especially when they occur close to the wet habitats that the geese favour (Speed et al. 2009).

The increase in above-ground biomass of B. vivipara and A. borealis following low-intensity disturbance based on goose grubbing was notable. These are species which are preferentially chosen by grubbing geese (Fox & Bergersen 2005). Although the experimental disturbance was applied in a regular pattern, rather than selectively as would be the case with geese, the same positive response of these species may occur under low-intensity natural grubbing. This suggests that at low intensities of grubbing, geese could improve the abundance of some of their forage species. However, other forage species such as Equisetum arvense were negatively impacted by low-intensity simulated grubbing, and thus this study provides no evidence that grubbing could increase the density of all goose forage species. However, deep moss can delay the spring soil thaw (Gornall et al. 2007), thus the fragmentation of moss by grubbing may still increase the availability of forage during spring.

Herbivory can lead to the formation of de-vegetated stable states (Rietkerk & Van de Koppel 1997), although within a range of communities in Svalbard it appears that recovery is possible in the long term, through both clonal growth and seedling establishment. However, as recovery from high-intensity disturbance in all but the wettest of the communities was limited within the three-year time scale of this study, it is estimated that full recovery will take many years more, especially in the drier communities. As recovery appeared to be influenced by the size of disturbance patches and the proximity to intact edges, increasing grubbing extent will also delay the process of recovery. If goose populations continue to increase, their grubbing activity may lead to a threshold effect on vegetation loss and thus widespread shifts in tundra community state.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Acknowledgements
  7. References
  8. Supporting Information

The authors are grateful to Adam Bratt, Jonathan McAllister and Graeme Abel for field assistance, and the UNIS logistic department for field support and safety training. The Sysselmannen of Svalbard gave permission for field experiments in Adventdalen. Funding was provided by the Natural Environment Research Council (NER/S/A/2005/13880) and constructive comments that were gratefully received from four anonymous reviewers improved this manuscript.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Acknowledgements
  7. References
  8. Supporting Information
  • Bond, W.J. & Midgley, J.J. (2001) Ecology of sprouting in woody plants: the persistence niche. Trends in Ecology & Evolution, 16, 4551.
  • Bråthen, K.A., Ims, R.A., Yoccoz, N.G., Fauchald, P., Tveraa, T. & Hausner, V.H. (2007) Induced shift in ecosystem productivity? Extensive scale effects of abundant large herbivores. Ecosystems, 10, 773789.
  • Callaway, R.M., Brooker, R.W., Choler, P., Kikvidze, Z., Lortie, C.J., Michalet, R., Paolini, L., Pugnaire, F.I., Newingham, B., Aschehoug, E.T., Armas, C., Kikodze, D. & Cook, B.J. (2002) Positive interactions among alpine plants increase with stress. Nature, 417, 844848.
  • Chapin, F.S., BretHarte, M.S., Hobbie, S.E. & Zhong, H.L. (1996) Plant functional types as predictors of transient responses of arctic vegetation to global change. Journal of Vegetation Science, 7, 347358.
  • Cole, D.N. (1995) Experimental trampling of vegetation. 1. Relationship between trampling intensity and vegetation response. Journal of Applied Ecology, 32, 203214.
  • Cooper, E.J., Alsos, I.G., Hagen, D., Smith, F.M., Coulson, S.J. & Hodkinson, I.D. (2004) Plant recruitment in the high Arctic: seed bank and seedling emergence on Svalbard. Journal of Vegetation Science, 15, 115124.
  • D’Hertefeldt, T. & Jónsdóttir, I.S. (1999) Extensive physiological integration in intact clonal systems of Carex arenaria. Journal of Ecology, 87, 258264.
  • Didham, R.K., Watts, C.H. & Norton, D.A. (2005) Are systems with strong underlying abiotic regimes more likely to exhibit alternative stable states? Oikos, 110, 409416.
  • Dormann, C.F. & Woodin, S.J. (2002) Climate change in the Arctic: using plant functional types in a meta-analysis of field experiments. Functional Ecology, 16, 417.
  • Dorrepaal, E. (2007) Are plant growth-form-based classifications useful in predicting northern ecosystem carbon cycling feedbacks to climate change? Journal of Ecology, 95, 11671180.
  • Elmqvist, T., Folke, C., Nystrom, M., Peterson, G., Bengtsson, J., Walker, B. & Norberg, J. (2003) Response diversity, ecosystem change, and resilience. Frontiers in Ecology and the Environment, 1, 488494.
  • Felix, N.A., Raynolds, M.K., Jorgenson, J.C. & Dubois, K.E. (1992) Resistance and resilience of tundra plant-communities to disturbance by winter seismic vehicles. Arctic and Alpine Research, 24, 6977.
  • Folke, C., Carpenter, S., Walker, B., Scheffer, M., Elmqvist, T., Gunderson, L. & Holling, C.S. (2004) Regime shifts, resilience, and biodiversity in ecosystem management. Annual Review of Ecology Evolution and Systematics, 35, 557581.
  • Forbes, B.C., Ebersole, J.J. & Strandberg, B. (2001) Anthropogenic disturbance and patch dynamics in circumpolar arctic ecosystems. Conservation Biology, 15, 954969.
  • Fox, A.D. & Bergersen, L. (2005) Lack of competition between barnacle geese Branta leucopsis and pink-footed geese Anser brachyrhynchus during the pre-breeding period in Svalbard. Journal of Avian Biology, 36, 173178.
  • Fox, A.D., Madsen, J., Boyd, H., Kuijken, E., Norriss, D.W., Tombre, I.M. & Stroud, D.A. (2005) Effects of agricultural change on abundance, fitness components and distribution of two arctic-nesting goose populations. Global Change Biology, 11, 881893.
  • Fox, T.A., Francis, I.S. & Bergersen, E. (2006) Diet and habitat use of Svalbard Pink-footed Geese Anser brachyrhynchus during arrival and pre-breeding periods in Adventdalen. Ardea, 94, 691699.
  • Gornall, J.L., Jonsdottir, I.S., Woodin, S.J. & Van der Wal, R. (2007) Arctic mosses govern below-ground environment and ecosystem processes. Oecologia, 153, 931941.
  • Graham, M.H. (2003) Confronting multicollinearity in ecological multiple regression. Ecology, 84, 28092815.
  • Grime, J.P. (1979) Plant Strategies and Vegetation Processes, John Wiley & Sons, Chichester, UK.
  • Hobbs, R.J. & Huenneke, L.F. (1992) Disturbance, diversity, and invasion – implications for conservation. Conservation Biology, 6, 324337.
  • Jefferies, R.L., Jano, A.P. & Abraham, K.F. (2006) A biotic agent promotes large-scale catastrophic change in the coastal marshes of Hudson Bay. Journal of Ecology, 94, 234242.
  • Jefferies, R.L., Rockwell, R.F. & Abraham, K.E. (2004) Agricultural food subsidies, migratory connectivity and large-scale disturbance in arctic coastal systems: a case study. Integrative and Comparative Biology, 44, 130139.
  • Jónsdóttir, I.S. & Callaghan, T.V. (1988) Interrelationships between different generations of interconnected tillers of Carex bigelowii. Oikos, 52, 120128.
  • Lavorel, S. & Garnier, E. (2002) Predicting changes in community composition and ecosystem functioning from plant traits: revisiting the Holy Grail. Functional Ecology, 16, 545556.
  • Lavorel, S., McIntyre, S., Landsberg, J. & Forbes, T.D.A. (1997) Plant functional classifications: from general groups to specific groups based on response to disturbance. Trends in Ecology & Evolution, 12, 474478.
  • McGill, B.J., Enquist, B.J., Weiher, E. & Westoby, M. (2006) Rebuilding community ecology from functional traits. Trends in Ecology & Evolution, 21, 178185.
  • McLaren, J.R. & Jefferies, R.L. (2004) Initiation and maintenance of vegetation mosaics in an Arctic salt marsh. Journal of Ecology, 92, 648660.
  • Mulder, C.P.H. (1999) Vertebrate herbivores and plants in the Arctic and subarctic: effects on individuals, populations, communities and ecosystems. Perspectives in Plant Ecology, Evolution and Systematics, 2, 2955.
  • Oksanen, J., Kindt, R., Legendre, P., O’Hara, B., Simpson, G.L., Solymos, P., Stevens, M.H.H. & Wagner, H. (2008) vegan: Community Ecology Package. R package version 1.15-0. http://CRAN.R-project.org/package=vegan.
  • Olofsson, J. (2006) Plant diversity and resilience to reindeer grazing. Arctic Antarctic and Alpine Research, 38, 131135.
  • Pakeman, R.J. (2004) Consistency of plant species and trait responses to grazing along a productivity gradient: a multi-site analysis. Journal of Ecology, 92, 893905.
  • R Development Core Team (2008) R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing, Vienna, Austria.
  • Rapport, D.J., Hilden, M., Roots, E.F. & Crawford, R.M.M. (1997) Transformation of northern ecosystems under stress. Disturbance and Recovery in Arctic Lands (ed. R.M.M.Crawford), pp. 7398. Kluwer Academic Publishers, Netherlands.
  • Raunkiaer, C. (1934) The Life Forms of Plants and Statistical Plant Geography, Oxford University Press, Oxford, UK.
  • Rietkerk, M. & Van de Koppel, J. (1997) Alternate stable states and threshold effects in semi-arid grazing systems. Oikos, 79, 6976.
  • Rønning, O.I. (1996) The Flora of Svalbard, Norwegian Polar Institute, Oslo, Norway.
  • Sasaki, T., Okayasu, T., Jamsran, U. & Takeuchi, K. (2008) Threshold changes in vegetation along a grazing gradient in Mongolian rangelands. Journal of Ecology, 96, 145154.
  • Speed, J.D.M., Woodin, S.J., Tommervik, H., Tamstorf, M.P. & Van Der Wal, R. (2009) Predicting habitat utilization and extent of ecosystem disturbance by an increasing herbivore population. Ecosystems, 12, 349359.
  • Strandberg, B. (1997) Vegetation recovery following anthropogenic disturbances in Greenland. Disturbance and Recovery in Arctic Lands (ed. R.M.M.Crawford), pp. 381390. Kluwer Academic Publishers, Netherlands.
  • Sutton, J.T., Hermanutz, L. & Jacobs, J.D. (2006) Are frost boils important for the recruitment of arctic-alpine plants? Arctic Antarctic and Alpine Research, 38, 273275.
  • Van der Wal, R. (2006) Do herbivores cause habitat degradation or vegetation state transition? Evidence from the tundra. Oikos, 114, 177186.
  • Van der Wal, R., Sjogersten, S., Woodin, S.J., Cooper, E.J., Jonsdottir, I.S., Kuijper, D., Fox, T.A.D. & Huiskes, A.D. (2007) Spring feeding by pink-footed geese reduces carbon stocks and sink strength in tundra ecosystems. Global Change Biology, 13, 539545.
  • Vavrek, M.C., Fetcher, N., McGraw, J.B., Shaver, G.R., Chapin, F.S. & Bovard, B. (1999) Recovery of productivity and species diversity in Tussock tundra following disturbance. Arctic Antarctic and Alpine Research, 31, 254258.
  • Walker, D.A., Reynolds, J.F. & Tenhunen, J.D. (1996) Disturbance and recovery of Arctic Alaskan vegetation. Landscape Function and Disturbance in Arctic Tundra (eds J.F.Reynolds & J.D.Tenhunen), pp. 3571. Springer-Verlag, Berlin Heidelberg.
  • Walker, D.A. & Walker, M.D. (1991) History and pattern of disturbance in Alaskan Arctic terrestrial ecosystems – a hierarchical approach to analyzing landscape change. Journal of Applied Ecology, 28, 244276.

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Acknowledgements
  7. References
  8. Supporting Information

Appendix S1. List of species and the functional groups they were assigned to.

Table S1. Mean and standard deviation of shoot numbers per species across disturbance intensities and communities.

Table S2. Mean and standard deviation of difference in shoot numbers between edge and centre of high intensity plot (35 × 35 cm) per species across communities.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

FilenameFormatSizeDescription
JEC_1685_sm_apps1.doc65KSupporting info item
JEC_1685_sm_tables1.xls40KSupporting info item
JEC_1685_sm_tables2.xls27KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.