Continuous permafrost zones with well-developed polygonal ice-wedge networks are particularly vulnerable to climate change. Thermo-mechanical erosion can initiate the development of gullies that lead to substantial drainage of adjacent wet habitats. How vegetation responds to this particular disturbance is currently unknown but has the potential to significantly disrupt function and structure of Arctic ecosystems. Focusing on three major gullies of Bylot Island, Nunavut, we estimated the impacts of thermo-erosion processes on plant community changes. We explored over 2 years the influence of environmental factors on plant species richness, abundance and biomass in 62 low-centered wet polygons, 87 low-centered disturbed polygons and 48 mesic environment sites. Gullying decreased soil moisture by 40 % and thaw-front depth by 10 cm in the center of breached polygons within less than 5 years after the inception of ice wedge degradation, entailing a gradual yet marked vegetation shift from wet to mesic plant communities within 5 to 10 years. This transition was accompanied by a five times decrease in graminoid above-ground biomass. Soil moisture and thaw-front depth changed almost immediately following gullying initiation as they were of similar magnitude between older (> 5 years) and recently (< 5 years) disturbed polygons. In contrast, there was a lag-time in vegetation response to the altered physical environment with plant species richness and biomass differing between the two types of disturbed polygons. To date (10 years after disturbance), the stable state of the mesic environment cover has not been fully reached yet. Our results illustrate that wetlands are highly vulnerable to thermo-erosion processes, which drive landscape transformation on a relative short period of time for High Arctic perennial plant communities (5 to 10 years). Such succession towards mesic plant communities can have substantial consequences on the food availability for herbivores and carbon emissions of Arctic ecosystems.
Warming in the Arctic is occurring twice as fast as the global average
(USGCRP, 2009; New et al., 2011; NOAA, 2014). Perennially frozen ground
(permafrost) has consequently warmed by 2
Permafrost is tightly associated with biophysical components such as air temperatures, soil conditions, surface water, groundwater, snow cover and vegetation (Jorgenson et al., 2010; Sjöberg, 2015). Permafrost impedes water to drain to deeper soil layers and maintains a perched water table and saturated soils which favors the existence of wetlands (Woo, 2012; Natali et al., 2015). Permafrost degradation that would increase subsurface drainage and reduce the extent of lakes and wetlands at high latitudes (Avis et al., 2011; Jorgenson et al., 2013; Beck et al., 2015) would thus have major consequences on ecosystem structure and function (Collins et al., 2013; Jorgenson et al., 2013). It would also strongly influence variations of active layer depths (Wright et al., 2009; Shiklomanov et al., 2010; Gangodagamage et al., 2014), as illustrated by long-term monitoring sites throughout the circumpolar North (Tarnocai et al., 2004; Nelson et al., 2008; Smith et al., 2009; Shiklomanov et al., 2010).
Several forms of ground and massive ice can be found within permafrost
(Rowland et al., 2010), especially ice wedges in regions where winter
temperatures enable thermal contraction cracking (Fortier and Allard, 2005; Kokelj et al., 2014; M. T. Jorgenson et al., 2015; Sarrazin
and Allard, 2015). Continuous permafrost zones with well-developed polygonal
ice-wedge networks are particularly vulnerable to climate change because ice
wedges are usually found near the top of permafrost (Smith et al., 2005;
Jorgenson et al., 2006; Woo et al., 2008; Vonk et al., 2013). In these
regions, thawing permafrost can result in ground ice erosion and displacement
of sediments, carbon and nutrients by drainage (Rowland et al., 2010; Godin
et al., 2014; Harms et al., 2014). This thermo-erosion process has especially
been observed across North-America (Grosse et al., 2011), in Siberia
(Günther et al., 2013) and in the Antarctic Dry Valleys (Levy et al.,
2008). On Bylot Island, Nunavut, thermo-mechanical erosion by water has
initiated permafrost tunneling and the development of gully networks in
aeolian, organic and colluvial depositional environments of nearly
158 000 m
Many observational and experimental studies have highlighted shifts in tundra plant community structure and plant species productivity in response to warming temperatures (Jonsdottir et al., 2005; Hudson and Henry, 2010; Epstein et al., 2013; Naito and Cairns, 2015). In contrast, little is known about how thermo-erosion gullying affects plant community structure and plant species abundance. Yet, this information is urgently needed as vegetation plays an important role in structuring Arctic ecosystems and regulating permafrost response to climate change (Jorgenson et al., 2010; Gauthier et al., 2011; Legagneux et al., 2012). Wetlands serve as preferred grounds for Arctic herbivores such as snow geese (Gauthier et al., 1996; Massé et al., 2001; Doiron et al., 2014). They are also expected to produce more methane compared to shrub-dominated areas (Olefeldt et al., 2013; Nauta et al., 2015; Treat et al., 2015).
The present study aimed at examining plant community patterns following thermo-erosion gullying processes. Bylot Island, where geomorphological and ecological processes in response to climate change have been monitored for over 2 decades (Allard, 1996; Fortier and Allard, 2004; Gauthier et al., 2013; Godin et al., 2014), offered a unique opportunity to specifically assess the response of wetlands to gullying. The following questions were addressed: (1) to what extent thermo-erosion gullying modifies environmental conditions of low-centered wetland polygons? (2) How do plant communities cope with these geomorphological changes, i.e. do we observe shifts in plant diversity, abundance and productivity?
This study took place in the Qarlikturvik valley of Bylot Island, Nunavut,
Canada (73
Location of the study area.
Our work was specifically conducted around three gullies that were selected
among the 36 identified in the valley (Godin and Fortier, 2012b). These gully
networks have originated from snowmelt water infiltration into cavities of
the frozen active layer and the subsequent formation of underground tunnels
that have ended up collapsing (Fortier et al., 2007; Godin and Fortier,
2010). The gullies R08p and R06, respectively 835 and 717 m long, are
characterized by ongoing thermo-erosion processes (Fortier et al., 2007; Godin and
Fortier, 2012b) whilst the gully RN08, 180 m long, has not been actively eroding in
recent years. A total of 197 sampling sites were randomly selected around the
three gullies (Table 1; Fig. 1b) and classified into one of four categories
(referred hereafter as habitats) that represented the two baseline vegetation
types (wet and mesic) as well as increasing levels of disturbance related to
thermo-erosion processes. The disturbed habitats were sorted via a visual assessment of
the low-centered polygon rim integrity coupled with a recent close monitoring
of drainage system development along the gullies (Fortier et al., 2007; Godin
and Fortier, 2012a, b). The habitats were defined as follows: (i) intact
low-centered wet polygons (
Distribution of the sampling sites per habitat type and per gully.
Daily precipitation was recorded with a manual rain gauge throughout summer
2010 at the base camp, located 700 m west of the gully R08p (Gauthier et
al., 2010). Soil (top 10 cm) moisture was recorded at the center of each
sampling site using ECH
Species richness and abundance were determined at each site in July 2009 or
2010 using three randomly placed 70 cm
The four habitat types studied in the Qarlikturvik valley,
Bylot Island, Nunavut. The close view at the bottom right of each picture
represents the 70 cm
Five sampling sites per habitat were also randomly selected along the gullies
R08p and R06 to measure above-ground biomass of graminoid species. At each
site, an exclosure of 1 m
Differences in soil moisture, thaw-front depth and graminoid above-ground
biomass among habitats were tested with a generalized linear mixed model
(procedure MIXED, REML method in SAS, version 9.4, SAS Institute, Cary, NC,
USA). Soil moisture, thaw-front depth as well as date or year of measurements
and the interaction terms were treated as fixed factors and gully as a random
factor. Type III sums of squares were used for the calculation of fixed
effect
In 2009, above-average spring temperatures led to a rapid snowmelt (16 June)
while summer was one of the driest on record (Gauthier et al., 2009). In
2010, despite a relatively warm spring (0.26
Soil moisture monitored early and late July 2010 in the four habitat
types studied in the Qarlikturvik valley, Bylot Island, Nunavut. The 10th
percentile, lower quartile, median (solid line), mean (dashed line), upper
quartile and 90th percentile are shown. See Table A1 for sample sizes and
post-hoc contrasts.
Thaw-front depth monitored in July 2009 and 2010 in the four
habitat types studied in the Qarlikturvik valley, Bylot Island, Nunavut. The
10th percentile, lower quartile, median (solid line), mean (dashed line), upper
quartile and 90th percentile are shown. See Table A1 for sample sizes
and post-hoc contrasts.
A total of 18 vascular plant families encompassing 59 species were sampled
throughout the study (Table A2). The greatest species richness was found in
polygons that were disturbed for at least 5 years and where both hydrophilic
and mesic species were present (Table 2). The transition from wet polygons to
mesic environments was accompanied by significant changes in vascular plant
community composition, especially with the decline in Cyperaceae and Poaceae
cover and the emergence of Salicaceae species (Table 2).
Species richness, family total cover and species mean cover of
vascular taxa as well as mean cover of non-vascular taxa in each of the four
habitat types studied in the Qarlikturvik valley, Bylot Island, Nunavut. Mean
species richness is given for sampled areas of 49 dm
Moreover, we observed vegetation changes through the decline of graminoid
above-ground biomass which varied significantly among habitats (d
Above-ground biomass of graminoids growing in the four habitat types
studied in the Qarlikturvik valley, Bylot Island, Nunavut.
The first two axes of the canonical correspondence analysis retained 14 %
of the vegetation data variance and 80 % of the vegetation–environment
relationship variance (Table 3). Five of the eight environmental variables
tested were significant within the canonical model (
Canonical correspondence analysis of the vegetation and environmental data gathered in four habitat types in the Qarlikturvik valley, Bylot Island, Nunavut. CCA-1: first canonical axis; CCA-2: second canonical axis.
Sustainability of wetlands at high latitudes essentially relies on perennial frozen ground that prevents drainage and allows wet soil conditions (Woo and Young, 2006; Ellis et al., 2008). However, snowmelt water run-off through ice-wedge polygon landscapes can initiate thermal erosion of the permafrost and the development of gullies (Fortier et al., 2007; Godin and Fortier, 2014). We showed here that permafrost gullying significantly altered wetlands by changing the original polygon microtopography, and decreasing soil moisture and thaw-front depth of disturbed polygons along the gullies. Vegetation was sensitive to this process, and mesic environment plant species gradually replaced hydrophilic species within 5 to 10 years, although the full transition has yet to be reached. This vegetation turn-over can have substantial consequences on wildlife biology, permafrost stabilization and ecosystem-level greenhouse gas emissions (Blok et al., 2010; Doiron et al., 2014; M. T. Jorgenson et al., 2015; McEwing et al., 2015).
Thermo-erosion gullying led to a significant decrease in soil moisture and thaw-front depth of breached polygons. Both older and recently disturbed polygons had similar soil moisture and thaw-front depth while differing in time since disturbance, which shows that the change in polygon environmental conditions after permafrost disturbance was rapid. The decrease in soil moisture following polygon rim erosion is consistent with what has been previously observed in gullied areas (Seppälä, 1997; Godin and Fortier, 2012a, 2014, 2015; Harms et al., 2014) and concurs with a modeling analysis showing that the transformation of low-centered to high-centered polygon landscape following ice wedge degradation is accompanied by a significant alteration in the water balance partitioning (Liljedahl et al., 2012). In our study, all types of polygons were recharged by snowfall and summer rainfall, yet disturbed habitats had lower soil moisture than wet polygons and a thorough examination of moisture evolution throughout an entire summer showed that soil moisture of breached polygons was significantly more variable than that of wet polygons at both intra- and inter-polygonal scales (Godin et al., 2015). Given that soil moisture is an important driver of plant community composition (Dagg and Lafleur, 2011), it is no surprise that we observed a shift in vegetation following changes in moisture regime.
Decreasing soil moisture in the center of disturbed polygons came with decreasing thaw-front depth, which was expected given that active layer thickness is closely related to soil moisture (Nelson et al., 1999; Hinzman et al., 2005; Minke et al., 2009; Wright et al., 2009; Gangodagamage et al., 2014). This result, however, contrasts with the active layer thickening generally observed in response to climate warming (Tarnocai et al., 2004; Woo et al., 2007; Akerman and Johansson 2008; Smith et al., 2009; Nauta et al., 2015), and this is likely due in part to ground surface subsidence and drainage which follows ice-rich permafrost thawing (Shiklomanov et al., 2013) and in part to snow accumulation patterns (Godin et al., 2015). Within 5 years of drainage, thaw-front depth in disturbed polygons decreased by 37 % compared to that in wet polygons. This is mainly explained by heat capacity of water and the higher thermal conduction rates in wetter polygons that provide substantial melt energy to the frost table (Nelson et al., 1997; Hinzman et al., 2005; Wright et al., 2009; Romanovsky et al., 2010). This effect is also sharpened by the low thermal conductivity of drier moss carpets in disturbed habitats (Wright et al., 2009) and reduced local snow conditions within the polygons adjacent to the gullies (Godin et al., 2015).
Canonical correspondence analysis ordination diagram of the 197
sites sampled in the Qarlikturvik valley, Bylot Island, Nunavut. Wet polygons
(
Canonical correspondence analysis of the vegetation sampled along
three gullies in the Qarlikturvik valley, Bylot Island, Nunavut, and biplot
scores for environmental variables. CCA-1: first canonical axis; CCA-2:
second canonical axis. Statistically significant values (
Overall, the floristic composition of our sampling sites is in line with
previous field surveys conducted in the same area (Gauthier et al., 1996;
Duclos, 2002; Doiron, 2014). The presence of
The development of gullies in the Qarlikturvik valley and the subsequent
drainage of adjacent low-centered polygons have led within 5 to 10 years to
a gradual change in plant communities with vegetation of disturbed polygons
leaning toward a new equilibrium, that of mesic environments. Mesic
environment species such as
In the canonical ordination analysis, the soil moisture gradient discriminated wet polygons from the other habitats as well as recently disturbed from older disturbed habitats. The 37 % decrease in soil moisture between wet and disturbed polygons represents a drastic change of conditions for plant communities and is of similar magnitude than what has been documented in Alaskan drying wetlands as a result of increasing temperatures (Klein et al., 2005). The strong influence of soil moisture in separating plant community types at high latitudes has indeed been well documented (Hinzman et al., 2005; Daniëls and de Molenaar, 2011; Daniëls et al., 2011; Sandvik and Odland, 2014). Four other variables significantly influenced the distinction among habitats: (i) thaw-front depth discriminated habitats in the same direction than soil moisture with a 30 % decrease in disturbed polygons and mesic environments compared to wet polygons, which was expected since these two factors are closely related (see Sect. 4.1); (ii) litter cover separated mesic environments from the others, which may be explained by increased organic matter related to greater shrub abundance in mesic environments (Zamin et al., 2014); (iii) vascular plant standing dead separated wet and recently disturbed polygons from the other habitats, which can be explained by the senescence of Cyperaceae tillers that are highly abundant at these locations (Fig. 5); (iv) goose feces were mainly associated with older disturbed and mesic environments. While this may suggest a higher use of these habitats by geese, the slower degradation of feces in dryer habitats cannot be ruled out; this has yet to be tested.
The shift in vegetation composition in disturbed polygons was accompanied by
significant changes in biomass. Above-ground biomass of graminoids was the
greatest in wet polygons, which is concordant with the fact that wetlands are
the most productive habitats of forage plants in the Arctic (Sheard and
Geale, 1983; Duclos, 2002; Doiron, 2014). It gradually decreased in disturbed
polygons as conditions became closer to those of mesic environments. Compared
to the immediate change in environmental conditions, we nonetheless observed
a lag-time in vegetation response to thermo-erosion related disturbances as
graminoid biomass differed significantly between recently and older disturbed
polygons. In our study, graminoid above-ground biomass of wet polygons was
35 % lower than what Cadieux et al. (2008) found via a long-term plant
monitoring on Bylot Island (45.2 g m
It is likely that the replacement of hydrophilic plants by mesic vegetation
will severely impact wildlife biology. The Qarlikturvik valley of Bylot
Island represents an important summer habitat for greater snow geese
(Legagneux et al., 2012). It is well documented that this species mostly
relies on wetlands for food resources (Gauthier et al., 1995, 2011),
especially because graminoids are easily digested due to their low fiber
concentration and rich nutritive elements (Sedinger and Raveling, 1989;
Manseau and Gauthier, 1993; Audet et al., 2007). For instance, geese removed
respectively 40 and 31 % of the total annual production of
Effects of gullying-induced vegetation changes may finally be visible on
variations of greenhouse gas emissions. There is evidence for a strong
vegetation control on methane emission from wetlands (Olefeldt et al., 2013;
Lara et al., 2015; McEwing et al., 2015; Tveit et al., 2015). In wet
polygonal tundra of Northern Siberia, Kutzbach et al. (2004) found for
instance that dense
This study illustrates that changes in the hydrological and thermal regimes following thermo-erosion gullying processes boost landscape transformation from wet to mesic habitats, providing evidence that permafrost disturbance is a critical component of ecosystem modification at high latitudes. Ecological studies should consequently start using an approach that integrates disturbed permafrost monitoring if one wants to more efficiently document climate change effects on arctic terrestrial ecosystems. In addition, our latest field observations showed that hydrology and thaw regimes of breached polygons have yet to reach equilibrium with new conditions. Similarly, vegetation remains in transition given that, 10 years after disturbance, the cover of dominant shrubs and mesic bryophytes in disturbed polygons is still lower than in adjacent mesic environments. It is currently not possible to predict how long these species would take to out compete declining species and cryptogamic crust and reach a new mesic environment equilibrium. This current state underscores the importance of long-term monitoring of permafrost and its associated vegetation. In addition, more work should be devoted to the feedback effects of plant communities and vegetation succession on thermal and mechanical stabilization dynamics of disturbed permafrost terrains. This is especially needed since plant community differences between disturbed and intact sites can last several centuries (Cray and Pollard, 2015).
Sample sizes and means (
List of the vascular plant species
inventoried in the Qarlikturvik valley, Bylot Island, Nunavut during the 2009
and 2010 field seasons. Species names were retrieved (2011) from the
Integrated Taxonomic Information System (ITIS) (
The authors are grateful to the Inuit community of Pond Inlet and to Parks Canada-Sirmilik National Park, Centre d'études nordiques (CEN) and Gilles Gauthier (Université Laval) for the access to the field camp during summers 2009, 2010 and 2012. We also thank Alexandre Guertin-Pasquier, Étienne Godin, Jonathan Lasnier, Stéphanie Coulombe and Coralie Henry-Brouillette for their fieldwork support as well as Alexandre Moreau, Stephan Ouellet, Noémie Boulanger-Lapointe for their help with statistical analyses. This project was funded by the International Polar Year program of the Government of Canada, Fonds Québécois de la Recherche sur la Nature et les Technologies (FQRNT), Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Frontiers grant “Arctic Development and Adaptation to Permafrost in Transition” (ADAPT), Network of Centers of Excellence of Canada ArcticNet, Northern Scientific Training Program (NSTP), NSERC CREATE Training Program in Northern Environmental Sciences (EnviroNorth), Groupe de Recherche en Biologie Végétale (GRBV) of Université du Québec à Trois-Rivières and Cold Regions Geomorphology and Geotechnical Laboratory (Geocryolab) of Université de Montréal. Essential logistic support was provided by Polar Continental Shelf Program (Natural Resources Canada). Edited by: W. F. Vincent