SEARCH

SEARCH BY CITATION

Keywords:

  • Arctic;
  • climate change;
  • dwarf shrub;
  • Empetrum hermaphroditum;
  • extreme weather events;
  • NDVI;
  • shrub expansion;
  • snow cover;
  • winter warming

Summary

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

1.  The Arctic is experiencing considerable change in climate, particularly in winter, and a greater frequency of extreme climatic events is expected. However, the impacts of winter climate change and extreme events have received far less attention than the impacts of season-long summer warming. Here we report findings from observations following a natural event and from experimental studies to show that short (<10 days) extreme winter warming events can cause major damage to sub-Arctic plant communities at landscape scales.

2.  In the landscape observations, impacts were assessed following an extreme winter warming event that occurred in December 2007 in northern Scandinavia. During this event, temperatures rose up to 7 °C resulting in loss of snow cover and exposure of vegetation to firstly warm and then returning cold temperatures.

3.  In the following summer, extensive areas of damaged dwarf shrub vegetation could be observed. Ground observations showed damaged areas to have a 16 times greater frequency of dead shoots of the dominant shrub Empetrum hermaphroditum, resulting in 87% less summer growth compared to neighbouring undamaged areas. The landscape scale extent of this damage was confirmed by satellite-derived Normalized Differential Vegetation Index values that showed a considerable 26% reduction (comparing July 2007 with July 2008 values) over an area of 1424 km2. This reduction indicates a significant decline in either leaf area or photosynthetic capacity or efficiency at the landscape scale.

4.  Strikingly similar damage was also observed in a field manipulation experiment using heating lamps and soil warming cables to simulate such extreme events in sub-Arctic heathland over two winters. Here, an up to 21 times greater frequency of dead shoots and 47% less shoot growth of E. hermaphroditum was observed in plots exposed to simulated winter warming events compared to unmanipulated controls.

5.Synthesis. These combined landscape observations and experimental findings provide compelling evidence that winter warming events can cause considerable damage to sub-Arctic vegetation. With increasing winter temperatures predicted, any increase in such damage may have major consequences for productivity and diversity of these sub-Arctic ecosystems, in contrast to the greening of parts of the Arctic currently attributed to summer warming.


Introduction

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

The insulating properties of snow provide a vital winter service in Arctic ecosystems, effectively decoupling plant canopy and soil temperatures from colder ambient air temperatures and temperature fluctuations. This provides a relatively mild sub-nivean microclimate for plants, animals and soil beneath (Kausrud et al. 2008). However, there has been an increasing occurrence of warm air masses over some Arctic regions in winter (Visbeck et al. 2001), which can lead to either loss of insulating snow cover through melt, or formation of ice layers from refreezing of partially melted snow (Bamzai 2003). Ice layers have already been implicated in damage to plants and the population declines of voles, reindeer and musk ox unable to access their food plants trapped beneath the impenetrable ice (Forchhammer & Boertmann 1993; Robinson et al. 1998; Aanes et al. 2002). To date, however, the impacts of loss of snow cover and subsequent exposure to first warm and then returning winter-cold temperatures have previously remained unknown. The potential for damage to vegetation is considerable given that the loss of snow cover not only exposes plants to sub-zero ambient temperatures and large temperature fluctuations, but may also lead to damage by winter desiccation, repeated freeze–thaw cycles and abrasion by windblown ice particles before the system is covered with fresh snow (Sonesson & Callaghan 1991; Walker, Billings & de Molenaar 2001). Freezing damage can occur for several reasons including dehydration beyond cell tolerance, when limits to deep supercooling are exceeded, and when freezing-induced embolisms in xylem vessels persist (Pearce 2001).

If we are to fully understand the consequences of climate change on Arctic ecosystems, assessment of the impacts of such winter climatic change is essential, especially given that the Arctic is experiencing, and will continue to experience, greatest climatic change in winter rather than in summer (ACIA 2005). This need is amplified further given that to date the focus on Arctic climate change impacts has generally been on season-long summer warming (Van Wijk et al. 2003; Walker et al. 2006), and much less in known of the impacts of winter climatic change.

Assessment of climate change impacts can be made both through experimental simulations in field plots and through observations of temporal change in the landscape from ‘naturally’ changing climate. Manipulations provide a controlled experiment from which direct causal links between manipulation and response can be made, but may not perfectly simulate climate change. In contrast, observations of landscape change over time with changing climate provide data on real changes occurring, but direct causal links between the changes observed and any one environmental driver may be difficult to make with certainty. The combination of both approaches can provide the strongest evidence for climate change impacts; for example, it was recently seen that the observed shrub expansion in some regions of the Arctic (Sturm et al. 2005; Tape, Sturm & Racine 2006) was in close agreement with the increases in shrub cover seen in Arctic warming manipulation experiments (Van Wijk et al. 2004; Walker et al. 2006).

Here, we report findings from both landscape observations and a field manipulation experiment made possible from the timely coincidence of both, which in combination provide particularly compelling evidence for the impacts of extreme winter warming on Arctic vegetation. In the case of the field manipulation experiment, a unique study was established in sub-Arctic Sweden in the winter of 2006/7 on dwarf shrub heathland – a common vegetation type with circumpolar distribution. The system uses infrared heating lamps run with and without soil warming cables (soil warming was included to determine the importance of soil thaw). It was originally hypothesized that extreme winter warming would result in damage to the dwarf shrubs (dominant in the heathland), leading to reduced growth, and that this damage would be greatest in plots without soil warming, since greater desiccation damage would occur where soils remained frozen. However, the first winter warming simulation (winter of 2006/7) did not result in clear damage or reductions in growth, although considerable delays of bud burst in spring and reductions in flowering and berry production were observed (Bokhorst et al. 2008). As the experiment has remained ongoing with a further simulation run in the following winter (2007/8), we can now report critical new findings.

During winter of 2007/8, an acute warming event occurred in north-western Scandinavia, where temperatures rose rapidly up to 7 °C, resulting in snowmelt at the landscape scale, which exposed vegetation to at first warm temperatures and then returning winter cold. Here we describe the quantification of the amount of damage at the landscape scale apparent in the following summer as elucidated by the satellite-derived NDVI (Normalized Differential Vegetation Index). Further, we undertook measurements of shoot mortality and impacts on growth taken on the ground at a number of sites within the damaged area to allow direct comparison with identical measurements taken within the winter warming simulation study. This combination of landscape and experimental manipulation observations provides an exceptional opportunity to understand the consequences of winter warming events on sub-Arctic vegetation.

Materials and methods

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

Landscape scale winter warming event

During initial ground observations in spring of 2008 following the warming event in the winter of 2007/8, damaged Empetrum hermaphroditum was observed in the region running from Abisko in northern Sweden (68 °21′N, 18 °49′E) westwards to Narvik (68 °25′N, 17 °33′E) on the Norwegian coast. Empetrum hermaphroditum contributes most of the biomass in dwarf shrub heathland. As it is practically impossible to accurately quantify the amount damage in a landscape using ground surveys, landscape-scale damage for the study area was assessed by changes in NDVI between summer of 2007 and 2008 (i.e. the summers before and after the event). This was done using NDVI data covering periods of 16 days (centred on mid-July) from the MODerate-resolution Imaging Spectroradiometer (MODIS) sensors on NASA’s Terra satellites (the MOD3A1 and MOD13Q1 products (Huete et al. 2002)). The 16-day NDVI data product was found to be the most optimal product in this part of the world under conditions of frequent cloudy weather during summer (Karlsen et al. 2008). We also used medium to high spatial resolution NDVI data from the SPOT-5, Landsat-5 TM, Landsat-7 ETM+ and the IRS-PS6 LISS-3 satellites acquired in July and August 2007 and 2008. These data were re-sampled to a consistent spatial resolution (e.g. SPOT to the same spatial resolution as Landsat) and inter-calibrated in order to compare and control the measurements from the MODIS sensors. We checked past July NDVI values for the study area (available from MODIS back to 2000) to ascertain that any reduction in NDVI resulting from the observed damage was in fact unusual and not within the range of typical July NDVI values.

Temperature and snow depth data for this region were collated from three meteorological stations: the Abisko Scientific Research Station, Narvik (Norwegian Meteorological Institute) and Katterjåk (Swedish Meteorological and Hydrological Institute) approximately mid-way between Abisko and Narvik.

For ground measurements, we studied five sites between Abisko and Narvik in more detail (Fig. 3a). At each site, five damaged and non-damaged E. hermaphroditum-dominated plots (60 × 60 cm) were selected. The ratio of live to dead E. hermaphroditum shoots was recorded in three 10 × 10 cm squares within each plot (n = 50). Shoots were considered dead when all leaves on the whole length of the stem had died. Marked live and dead shoots were monitored in each quadrant for the remainder of the growing season to ascertain that no regrowth occurred. Measurements of current years’ shoot growth were taken at the end of the growth season on five marked shoots per subplot with digital callipers. In addition to confirming the visibly clear differences between damaged and undamaged plots, these comparisons were also made to allow comparison with the differences between damage quantified in the simulation experiment.

image

Figure 3.  (a) Vegetation map of the affected area, numbers indicate study sites where ground observations of damaged Empetrum hermaphroditum shoots were made (1–5), 6 represents a birch forest without obvious dwarf shrub damage and 7 is the experimental winter warming study site in Abisko. (b) Normalized Difference Vegetation Index (NDVI)-images (black and white images) across the 1424-km2 area in northern Scandinavia from mid-July 2007 and 2008. White indicates the highest NDVI values with darker shades indicating lower NDVI values. The colour map shows NDVI change between 2007 and 2008, showing decreased NDVI values in grey to blue and increased NDVI values in red (scale: 1 = 0–10%, 2 = 10–20%, 3 = 20–30% etc.). Comparing 2008 with 2007, more than 50% of the areas showed NDVI values decreasing by >10% with a further 29% showing a decrease <10%. (c) NDVI distribution in mid-July, before (2007) and after the winter warming event of 2008, across the above-mentioned area in northern Scandinavia.

Download figure to PowerPoint

Simulation of extreme winter warming

The extreme winter warming simulation was run during the 1st week of March in the winters of 2006/7 and 2007/8 on an E. hermaphroditum-dominated dwarf shrub heathland near the Abisko Scientific Research Station. The experiment consists of 18 plots (2.1 × 1.0 m) representing six control plots and six each of two warming treatments: canopy warming, and combined canopy and soil warming. For full details of the experimental design see Bokhorst et al. (2008). Snow melt was complete after 3 days, and warming of the vegetation continued for another 4 days before the lamps were switched off, allowing plots to return to ambient cold temperatures. Canopy temperature was recorded throughout the year by thermistors in all plots.

In the summer after each winter warming simulation, vegetation surveys were carried out in each plot using the point–intercept method. Outside experimental plots, point–intercept hits of E. hermaphroditum were correlated (Pearson correlation, n = 20, r= 0.99, P < 0.01) to above-ground biomass harvests (Jonasson 1988) allowing biomass estimations to be made within plots without destructively harvesting. The ratio of live to dead shoots was calculated in three 30 × 30 cm squares in each plot and current seasons’ shoot growth was measured on 10 previously marked shoots within each plot.

Statistical analyses

NDVI pixel (6642 grid cells of 0.25 km2) values from 2007 were compared to those from 2008 using a t-test. High-resolution NDVI measurements from IRS-PS6, Landsat TM/ETM+ and SPOT were treated in the same way. Empetrum hermaphroditum shoot growth and the ratio of live to dead shoots were analysed with one-way anova while E. hermaphroditum point cover surveys in the experimental simulation plots were analysed with repeated-measures anova to identify differences between 2007 and 2008. Correlation between point cover hits and above-ground biomass was made by a Pearson correlation. Data were tested for homogeneity with Levene’s test and transformed where appropriate. All statistical tests were performed with spss 14.0 for Windows (Chicago, IL, USA).

Results

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

Landscape-scale winter warming event

Data from the three meteorological stations show that temperatures rose rapidly to as high as 7 °C and remained elevated for at least a week before the return of normal winter temperatures (Fig. 1a). Critically, this also resulted in an average 75% reduction in snow depth for at least 7 days before fresh snow fell (Fig. 1b).

image

Figure 1.  Temperature (a) and snow depth (b) data from weather stations along the studied area between Abisko and Narvik during the winter of 2007/8 (presented in Fig. 3a). Note the increase in temperature during the 2nd week of December and resulting decreases in snow cover. Data obtained from the Abisko Scientific Research Station, Swedish and Norwegian meteorological institutes (http://www.smhi.se, http://www.senorge.no). Katterjåk (68º25′N, 18º10′E) is a mountain weather station (516 m a.s.l.) approximately half-way between Abisko and Narvik. (c) and (d) are temperature and snow depth from the experimental winter warming simulation field experiment in 2008. Simulation of winter warming commenced on the 1 March (warming period indicated by grey boxed area) (c) and resulted in almost complete snow melt after 3 days (d). For clarity, only the temperature data from the canopy warming experiment (indicated as ‘winter warmed plots’) is shown because the combined canopy and soil warming treatment had the same temperature development during and after the winter warming simulation. Grey diamonds indicate canopy temperature during the winter warming simulation of 2007 (as described by Bokhorst et al. 2008).

Download figure to PowerPoint

During the following summer of 2008, large areas of damaged E. hermaphroditum dwarf shrub heathland were clearly apparent (Fig. 2a,b), including heathlands found in contrasting habitats of open birch forest, dwarf shrub tundra and pine forest. NDVI data showed a considerable reduction of 26% in July 2008 compared to the same period in 2007 (NDVI: 0.42 after and 0.57 prior to the winter warming event) over a 1424-km2 area surrounding the met stations (Fig. 3b) (t-test: d.f. 13282, 32.5, P < 0.001), with a decline in the frequency of high NDVI values for the region being particularly apparent (Fig 3c). NDVI measurements made by SPOT-5, Landsat-5 TM, Landsat-7 ETM+ and IRSPS6 LISS-3 in July and August 2007 and 2008 were consistent with the MODIS-based measurements.

image

Figure 2. Empetrum hermaphroditum damage in an alpine heathland (a) and a coastal pine forest (b). The alpine heathland (a) and the coastal pine forest (b) experienced on average a 23% and a 16% reduction in NDVI values from 2007 to 2008 respectively. In (a), note the non-affected greener areas near the boulders, while further away from the boulders there is visible damage to E. hermaphroditum (see insets 2a). Dead E. hermaphroditum shoots in the pine forest were 50% taller (one-way anova: F1,8=50.3, P < 0.001) than alive shoots. Comparable damage occurred after 2 years of experimental simulated winter warming (c), compared with undamaged control plots (d).

Download figure to PowerPoint

The vegetation of the region consists of a mosaic of plant communities that includes birch or pine forests (Pinus sylvestris, Betula pubescens ssp. czerepanovii), fens, meadows, heath grasslands, heathlands including alpine heathlands, lakes and barrens. However, the majority of the vegetation in this region consists of heathland or shrub-containing vegetation interspersed with smaller patches of mountain birch forest (Fig. 3a). NDVI values of a mountain birch forest near the Abisko Scientific Research Station, which we observed to have no apparent damage to dwarf shrubs of the understorey (site 6 on Fig. 3a), showed a near-significant reduction (paired t-test, t = 3.118, d.f. = 3, P = 0.053) in NDVI between 2007 (NDVI = 0.86 ± 0.01) and 2008 (NDVI = 0.81 ± 0.02). This suggests the additional possibility of birch damage.

The ground measurements comparing damaged with undamaged plots within the five sampled sites showed that the ratio of live to dead shoots of E. hermaphroditum was reduced from 22.1 (±3.5) in undamaged plots to 1.4 (±0.4) in damaged plots (one-way anova: F1,8=25.6, P < 0.001) (Fig. 4). Empetrum hermaphroditum shoot growth during the growing season of 2008 was reduced by 87% (±3%) (one-way anovaF1,32 203, P < 0.001) in damaged plots compared to undamaged plots. All five of the ground-sampled areas showed reductions in NDVI (MODIS) from 2007 to 2008, with this reduction ranging from 16% to 91% (in a 0.25-km2 grid cell, which is the image resolution).

image

Figure 4.  Ratio of live : dead Emetrum hermaphroditum shoots in the study sites of the affected area comparing damaged patches with undamaged patches of vegetation, and in the experimental winter warming plots. An asterisk indicates a significant reduction in live-to-dead shoot ratio between damaged and undamaged areas and between winter-warmed and control plots in the field simulation experiment (grey-boxed area). The five study sites are named after the nearest village, lake or valley and are shown from west (Langstranda at the Norwegian coast) to east near the Abisko Scientific Research Station. Error bars are one standard error. For brevity only the canopy warming treatment data for the simulation experiment are shown (data for canopy and soil warmed plots stated in the text). [Correction added on 21 August 2009, after first online publication: key transposed].

Download figure to PowerPoint

The decrease in NDVI values was unprecedented over the period for which MODIS data is available for this region (2000–2007) (Fig. 5). While the past NDVI values showed no reduction similar to that seen in 2008, we did detect an extreme winter warming event in the meteorological data for the Katterjåk station located in the alpine heathland – the station at the centre of our study area – in the winter of 2000/1. However, there was no associated NDVI reduction for either our larger study region, or the area immediately around the Katterjåk station (Fig. 5). In contrast to the winter event of 2007/8, the 2000/1 event had reduced snow cover for only 3 days and temperatures fell to only −1 °C before snowfall re-occurred. This was therefore a much milder and shorter event than that of 2007/8, which we propose caused the landscape-scale damage.

image

Figure 5.  Normalized Difference Vegetation Index (NDVI) for the past 9 years across the study region, and NDVI and snow depth for the Katterjåk weather station area. Error bars are one standard error. Average winter (December–March) snow depth is given for every year, and lowest snow depth is indicated for years with winter warming events (2001 and 2008) by hatched boxes. Extreme winter warming during mid-December 2001 resulted in 3 days with minimum snow depth (12 cm) and temperature of −1.1 °C while in 2008 minimum snow (6 cm) lasted for 7 days with a minimum temperature of −6.8 °C.

Download figure to PowerPoint

The study region did not suffer from drought or extreme spring or summer heat, which might otherwise have resulted in similar damage.

Experimental winter warming simulation

The winter warming simulations conducted in two consecutive winters (March 2007 and 2008) caused complete snowmelt after 3 days and maintained canopy temperatures between 0 °C and 5 °C (Fig. 1c and Bokhorst et al. 2008). Following the winter warming simulation, leaves experienced much lower temperatures than those in the control plots due to a shallower snow depth (Fig. 1c,d). While only impacts on timing of bud burst and reproductive effort (flowering and berry production) were observed in the spring and summer after the first winter’s warming simulation (Bokhorst et al. 2008), following the second winter’s warming event, extensive damage to the dwarf shrub vegetation became apparent (Fig. 2c,d). Quantitatively, the ratio of live to dead E. hermaphroditum shoots declined from 34.1 (±20.7) in control plots (which remained covered by snow) to 2.5 (±0.6) and 1.6 (±0.4) in the canopy and canopy and soil warmed plots, respectively (one-way anova: F2,15 16.6, P < 0.001). The experiment therefore showed very similar changes to those observed in the landscape ground measurements (Fig. 4). Further, shoot growth was reduced by 14.2% (±5.3) and 47.4% (±1.5) for canopy and combined canopy and soil winter-warmed plots, respectively (one-way anova: F2,15 3.6, P < 0.05). Point cover surveys indicated this damage reduced total live above-ground biomass by 13.9% (±9.0) and 30.9% (±12.8) between 2007 and 2008 in the canopy and combined canopy and soil warmed plots, respectively (repeated-measures anova: F1,10 27.7, P < 0.001) compared to control plots.

Discussion

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

The similarity in responses from our experimental winter warming simulation and the natural winter warming event provides strong evidence that it was indeed the rising temperatures followed by snow melt and subsequent exposure to first warm and than much colder temperatures that were the cause of damage observed across a large area in north-western Scandinavia. The lack of large NDVI changes between 2000 and 2007 during which no major winter warming events combined with snow loss occurred also supports this. The short duration (3 days) and minimum temperature of only −1 °C during the minor winter warming event of 2000/1 should not have posed a major threat to plants and would explain the lack of NDVI change following that event. Furthermore, while the field simulation has taken two winters warming events to cause a similar level of damage as that resulting from the single natural event, each winter’s simulation was less severe than the natural event (lower maximum temperatures and shorter duration of exposure). It is therefore reasonable to expect damage resulting from two simulated events to be similar to that resulting from a single natural event. This may also indicate that even less-severe events – if they occur in consecutive years – may have as strong an impact as one larger event.

The landscape of the study area is classified mostly as heathland or shrub-containing vegetation (i.e. dwarf shrub heathland in particular or alpine meadows with scrub), and this vegetation also dominates the region showing reduced NDVI. Nonetheless, there are also other important vegetation types, particularly birch forest, which may be impacted by the winter warming event and which can contribute to the decline in high NDVI values. We did not observe damage to birch forest on the ground, but given the impossibility of surveying in detail such a large area, we cannot rule this out. Moreover, assigning damage to particular plants from NDVI data is complicated since it is not possible to tell which plant or plants within that vegetation is or are damaged: a decline in NDVI values of birch forest, for instance, does not necessarily have to stem from damage to birch but could come from damage to the shrub understorey. Similarly, while the decrease in NDVI occurred mostly among the higher NDVI values (c. 0.7–0.9) (Fig. 3c), such a decline should not be assumed to only come from (birch) forests since other vegetation types such as Empetrum heathlands also have similarly high NDVI values (Street et al. 2007). Nonetheless, since mountain birches in Scandinavia are generally sufficiently exposed to chilling required for winter dormancy release early on in winter (Myking & Heide 1995; Myking 1999), they might therefore be susceptible to a winter warming event, potentially leading to mid-winter bud burst and then frost damage to buds. This, in turn, could delay spring phenology (Bokhorst et al. 2008), but the MODIS data were acquired in mid-July, which is 5–6 weeks after Betula pubescens tends to burst bud in this region (Karlsson et al. 2003); so any delays in phenology would most likely have been compensated for. Overall, we cannot rule out that birch was influenced by the winter warming event and contributed to the decline in NDVI, even though no visual damage was observed to birches in the area while very obvious damage had occurred to E. hermaphroditum communities.

The observed increase in lower NDVI values and more negative ones are most likely caused by shrub damage and subsequent opening of bare ground.

While it must be noted that the field simulation experimental site was within the Abisko area that also experienced snow melt during the landscape event of 2007/8, the experimental plots appear not to have been affected by the natural winter warming event (as seen by the absence of damage to control plots). This may be the result of local topographical heterogeneity and vegetation impacts on snow build up. Sheltered areas generally tend to accumulate deeper snow, and vegetation can act as a snow trap as well (Grogan & Jonasson 2006). During winter warming events, plants under deeper snow are more likely to remain covered by the snow layer and experience less extreme temperature fluctuations than plants under shallower snow that lose most or all snow cover. Indeed, such variation in snow depth will contribute to the patchy heterogeneity of the damage seen in the landscape and variation in NDVI change. Since the experimental site was within a dense mountain birch forest, causing greater snow accumulation, experimental plots may not have lost substantial snow cover during the natural event. Further, since birch buds are situated above the snow layer in these regions, snow depth is not a regulating factor in damage to established trees.

The importance in heterogeneity of snow lay is also reflected in observations after the natural winter warming event, where plant damage in the field was most apparent in exposed areas, while the damage was less obvious in more sheltered areas, such as inside the mountain birch forest or beside boulders and below ridges (Fig. 2a). Snow depth will be a strong regulator for plant responses to extreme winter warming events and the impact of extreme winter warming events will depend greatly on the timing and intensity of the event in relation the local snow conditions.

Empetrum hermaphroditum is a successful, dominant plant of sub-Arctic heathlands and has a circumpolar distribution (Tybirk et al. 2000). Given that NDVI is an indicator of leaf area, vitality and photosynthetic capacity, and the tight coupling between NDVI and gross primary productivity in Arctic vegetation (Street et al. 2007), the damage and NDVI decline resulting from the landscape event is likely to represent a considerable reduction in productivity and carbon uptake by these ecosystems over a large area. In addition, changes in community composition of the affected vegetation may well occur in the following seasons because the competitive strength of Empetrum hermaphroditum is likely lessened by the considerable reduction (13–31%) of live biomass. Indeed, increases in biodiversity and herbivore-available biomass might increase since E. hermaphroditum is allelopathic, i.e. suppressing growth of some other species many of which are preferred food plants for herbivores (Tybirk et al. 2000).

It should be noted that the winter warming event in December 2007 is not a new phenomenon: the longest-running climate records for the region show that such events have occurred almost every decade during the 20th century (data records of the Abisko Station). The fact that such events have been previously overlooked is perhaps unsurprising given that: (i) such events are unpredictable; (ii) they are short-lived and (iii) researchers working on climate change impacts on Arctic ecosystems have generally not focused on winter processes in explaining summer responses of vegetation.

The reductions in shrub cover and declines in NDVI values resulting from acute winter warming in this sub-Arctic region are in sharp contrast to increased shrub abundance and increased NDVI values that are currently observed in many regions of the Arctic and have been attributed to spring and summer warming (Myneni et al. 1997; Jia, Epstein & Walker 2003; Sturm et al. 2005). As current winter warming events are occurring during a trajectory of climate warming, it seems likely that the frequency with which winter temperatures will rise above the critical (snow-melting) threshold of 0 °C will increase. The full potential impacts of increased frequency of extreme winter warming events on sub-Arctic ecosystems could be considerable both in terms of ecosystem carbon sequestration and floristic composition, but to date remain unknown. Furthermore, as temperatures continue to rise in colder regions of the Arctic, it may be that the damage observed in warmer sub-Arctic communities is indicative of the impact to come in a warmer Arctic. Given these winter events result in opposite effects to spring and summer warming, they provide a considerable challenge and uncertainty in predicting the future of Arctic and sub-Arctic ecosystems in a warmer world.

Acknowledgements

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

We would like to thank the staff of the Royal Swedish Academy of Sciences Abisko Scientific Research Station for their assistance during the set up of the experimental site. This research was supported by a Leverhulme Trust (UK) grant to G.K.P. and T.V.C. (grant F/00 118/AV), by a grant from the Norwegian Research Council (project no. 171542/V10) awarded to J.W.B., and by ATANS grants (EU Transnational Access Programme, FP6 Contract no. 506004) to S.B., J.W.B. and G.K.P. Infrastructure and equipment support was supplied by the Royal Swedish Academy of Sciences and by Frank Bowles and Jerry Melillo from the Marine Biological Laboratory in Woods Hole, MA, USA, who also contributed to the experimental design and instrumentation. We are grateful to the Swedish Meteorological and Hydrological Institute for providing long-term temperature and snow depth records of the Katterjåk area. This manuscript was improved through useful comments from two anonymous referees.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Aanes, R., Sæther, B.E., Smith, F.M., Cooper, E.J., Wookey, P.A. & Øritsland, N.A. (2002) The Arctic Oscillation predicts effects of climate change in two trophic levels in a high-arctic ecosystem. Ecology Letters, 5, 445453.
  • ACIA (2005) Arctic Climate Impact Assessment. Cambridge University Press, New York.
  • Bamzai, A.S. (2003) Relationship between snow cover variability and Arctic oscillation index on a hierarchy of time scales. International Journal of Climatology, 23, 131142.
  • Bokhorst, S., Bjerke, J.W., Bowles, F.P., Melillo, J.M., Callaghan, T.V. & Phoenix, G.K. (2008) Impacts of extreme winter warming in the sub-Arctic: growing season responses of dwarf-shrub heathland. Global Change Biology, 14, 26032612.
  • Forchhammer, M. & Boertmann, D. (1993) The muskoxen Ovibos moscatus in north and northeast Greenland - population trends and the influence of abiotic parameters on population dynamics. Ecography, 16, 299308.
  • Grogan, P. & Jonasson, S. (2006) Ecosystem CO2 production during winter in a Swedish subarctic region: the relative importance of climate and vegetation type. Global Change Biology, 12, 14791495.
  • Huete, A., Didan, K., Miura, T., Rodriguez, E.P., Gao, X. & L.G, F. (2002) Overview of the radiometric and biophysical performance of the MODIS vegetation indices. Remote Sensing of Environment, 83, 195213.
  • Jia, G.J., Epstein, H.E. & Walker, D.A. (2003) Greening of the Alaskan Arctic over the past two decades. Geophysical Research Letters, 30, 2067.
  • Jonasson, S. (1988) Evaluation of the point intercept method for the estimation of plant biomass. Oikos, 52, 101106.
  • Karlsen, S.R., Tolvanen, A., Kubin, E., Poikolainen, J., Høgda, K.A., Johansen, B., Danks, F.S., Aspholm, P., Wielgolaski, F.E. & Makarova, O. (2008) MODIS-NDVI-based mapping of the length of the growing season in northern Fennoscandia. International Journal of Applied Earth Observation and Geoinformation, 10, 253266.
  • Karlsson, P.S., Bylund, H., Neuvonen, S., Heino, S. & Tjus, M. (2003) Climatic response of budburst in the mountain birch at two areas in northern Fennoscandia and possible responses to global change. Ecography, 26, 617625.
  • Kausrud, K.L., Mysterud, A., Steen, H., Vik, J.O., Østbye, E., Cazelles, B., Framstad, E., Eikeset, A.M., Mysterud, I., Solhøy, T. & Stenseth, N.C. (2008) Linking climate change to lemming cycles. Nature, 456, 9397.
  • Myking, T. (1999) Winter dormancy release and budburst in Betula pendula Roth and B-pubescens Ehrh. ecotypes. Phyton-Annales Rei Botanicae, 39, 139145.
  • Myking, T. & Heide, O.M. (1995) Dormancy release and chilling requirement of buds of latitudinal ecotypes of Betula pendula and B. pubescens. Tree Physiology, 15, 697704.
  • Myneni, R.B., Keeling, C.D., Tucker, C.J., Asrar, G. & Nemani, R.R. (1997) Increased plant growth in the northern high latitudes from 1981 to 1991. Nature, 386, 698702.
  • Pearce, R.S. (2001) Plant freezing and damage. Annals of Botany, 87, 417424.
  • Robinson, C.H., Wookey, P.A., Lee, J.A., Callaghan, T.V. & Press, M.C. (1998) Plant community responses to simulated environmental change at a high arctic polar semi-desert. Ecology, 79, 856866.
  • Sonesson, M. & Callaghan, T.V. (1991) Strategies of survival in plants of the Fennoscandian Arctic. Arctic, 44, 95105.
  • Street, L.E., Shaver, G.R., Williams, M. & Van Wijk, M.T. (2007) What is the relationship between changes in canopy leaf area and changes in photosynthetic CO2 flux in arctic ecosystems? Journal of Ecology, 95, 139150.
  • Sturm, M., Schimel, J., Michaelson, G.J., Welker, J.A., Oberbauer, S.F., Liston, G.E., Fahnestock, J.T. & Romanovsky, V.E. (2005) Winter biological processes could help convert Arctic tundra to shrubland. BioScience, 55, 1726.
  • Tape, K., Sturm, M. & Racine, C. (2006) The evidence of shrub expansion in Northern Alaska and the Pan-Arctic. Global Change Biology, 12, 686702.
  • Tybirk, K., Nilsson, M.C., Michelson, A., Kristensen, H.L., Shevtsova, A., Strandberg, M.T., Johansson, M., Nielsen, K.E., Rils-Nielsen, T., Strandberg, B. & Johnsen, I. (2000) Nordic Empetrum dominated ecosystems: Function and susceptibility to environmental changes. Ambio, 29, 9097.
  • Van Wijk, M.T., Clemmensen, K.E., Shaver, G.R., Williams, M., Callaghan, T.V., Chapin, F.S. et al. (2004) Long-term ecosystem level experiments at Toolik Lake, Alaska, and at Abisko, Northern Sweden: generalizations and differences in ecosystem and plant type responses to global change. Global Change Biology, 10, 105123.
  • Van Wijk, M.T., Williams, M., Laundre, J.A. & Shaver, G.R. (2003) Interannual variability of plant phenology in tussock tundra: modelling interactions of plant productivity, plant phenology, snowmelt and soil thaw. Global Change Biology, 9, 743758.
  • Visbeck, M.H., Hurrell, J.W., Polvani, L. & Cullen, H.M. (2001) The North Atlantic Oscillation: Past, present, and future. Proceedings of the National Academy of Sciences of the United States of America, 98, 1287612877.
  • Walker, D.A., Billings, W.D. & De Molenaar, J.G. (2001) Snow-vegetation interactions in tundra environments. Snow Ecology: An Interdisciplinary Examination of Snow-covered Ecosystems (eds H.G.Jones, J.W.Pomeroy, D.A.Walker & R.W.Hoham), pp. 266324. Cambridge University Press, Cambridge.
  • Walker, M.D., Wahren, C.H., Hollister, R.D., Henry, G.H.R., Ahlquist, L.E., Alatalo, J.M. et al. (2006) Plant community responses to experimental warming across the tundra biome. Proceedings of the National Academy of Sciences of the United States of America, 103, 13421346.