Volume 9, Issue 5 e2020EF001858
Research Article
Open Access

Mapping the Vulnerability of Arctic Wetlands to Global Warming

Elisie Kåresdotter

Corresponding Author

Elisie Kåresdotter

Department of Physical Geography and Bolin Centre for Climate Research, Stockholm University, Stockholm, Sweden

Correspondence to:

E. Kåresdotter,

[email protected]

Search for more papers by this author
Georgia Destouni

Georgia Destouni

Department of Physical Geography and Bolin Centre for Climate Research, Stockholm University, Stockholm, Sweden

Search for more papers by this author
Navid Ghajarnia

Navid Ghajarnia

Department of Physical Geography and Bolin Centre for Climate Research, Stockholm University, Stockholm, Sweden

Search for more papers by this author
Gustaf Hugelius

Gustaf Hugelius

Department of Physical Geography and Bolin Centre for Climate Research, Stockholm University, Stockholm, Sweden

Search for more papers by this author
Zahra Kalantari

Zahra Kalantari

Department of Physical Geography and Bolin Centre for Climate Research, Stockholm University, Stockholm, Sweden

Department of Sustainable Development, Environmental Science and Engineering (SEED), KTH Royal Institute of Technology, Stockholm, Sweden

Search for more papers by this author
First published: 24 April 2021
Citations: 15

Abstract

Wetlands provide multiple ecosystem services of local and global importance, but currently there exists no comprehensive, high-quality wetland map for the Arctic region. Improved information about Arctic wetland extents and their vulnerability to climate change is essential for adaptation and mitigation efforts, including for indigenous people dependent on the ecosystem services that wetlands provide, as inadequate planning could result in dire consequences for societies and ecosystems alike. Synthesizing high-resolution wetland databases and datasets on soil wetness and soil types from multiple sources, we created the first high-resolution map with full coverage of Arctic wetlands. We assess the vulnerability of Arctic wetlands for the years 2050, 2075, and 2100, using datasets on permafrost extent, soil types, and projected mean annual air temperature from the HadGEM2-ES climate model for three change scenarios (RCP2.6, RCP4.5, and RCP8.5). Our mapping shows that wetlands cover approximately 3.5 million km2 or roughly 25% of Arctic landmass and 99% of these wetlands are in permafrost areas, indicating considerable vulnerability to future climate change. Unless global warming is limited to scenario RCP2.6, robust results show that large areas of Arctic wetlands are vulnerable to ecosystem regime shifts. If scenario RCP8.5 becomes a reality, at least 50% of the Arctic wetland area would be highly vulnerable to regime shifts with considerable adverse impacts on human health, infrastructure, economics, ecosystems, and biodiversity. The developed wetland and vulnerability maps can aid planning and prioritization of the most vulnerable areas for protection and mitigation of change.

Key Points

  • Multi-data synthesis shows that wetlands cover 25% of the Arctic landmass

  • Around 50% of Arctic wetlands are vulnerable to permafrost thaw by the year 2100 in climate scenario RCP8.5

  • Much of the permafrost underlying Arctic wetland areas can remain stable up until the year 2100 under climate scenario RCP2.6

Plain Language Summary

Wetlands play an important role in the Arctic; they cool the global climate, hold freshwater for animals and plants, regulate water, carbon, and nutrient cycling, and are biologically diverse and of great importance for indigenous human activities. Large areas lie on frozen ground (permafrost), but this could thaw out under expected future global and regional warming. Permafrost thaw can lead to wetland change leading to ecosystem as well as societal problems since the permafrost also acts as the foundation for roads and buildings in the Arctic. Through combined multi-variable mapping, we estimate that wetlands cover 25% of land in the Arctic region, almost entirely on permafrost areas. Air temperatures above −2 degrees Celsius may lead to permafrost thaw and associated wetlands drainage, and this is enhanced by higher temperatures and longer durations of temperature elevation. If global temperatures increase by more than 2 degrees Celsius from preindustrial levels to year 2100, 30%–50% of Arctic wetlands are vulnerable to change. Limiting global warming is critical for preserving Arctic wetlands and reducing societal and ecosystem impacts of their changes.

1 Introduction

The Arctic region is experiencing more rapid temperature changes than other regions worldwide, with continued warming at approximately 2-to 3-fold the global average (IPCC, 2018). Increased temperatures threaten livelihoods, increase morbidity, and increase mortality by increasing the frequency of extreme events such as storms and floods (Nilsson et al., 2013; Parkinson & Evengård, 2009; Seifollahi-Aghmiuni, Nockrach et al., 2019). Increased temperatures also cause changes in the properties and extent of wetlands through permafrost thaw and altered runoff and evapotranspiration patterns, leading to loss of ecosystem services (Haynes et al., 2018; Malinauskaite et al., 2019; Markkula et al., 2019). Two climate-driven and hydrological mediated regime shifts relevant for Arctic wetland ecosystems on permafrost (Karlsson et al., 2011) are: shift from terrestrial ecosystems to thermokarst lakes and wetlands and shift from thermokarst lakes and wetlands to terrestrial ecosystems. Such regime shifts have extensive impacts on local hydrology and ecology, with feedbacks on regional and global climate.

Wetlands are ecosystems with high productivity and biodiversity acting as buffers between terrestrial and aquatic ecosystems (Basatnia et al., 2018; Kharazmi, 2018) and providing unique values both ecologically and economically (Basatnia et al., 2018; Thorslund et al., 2017). However, wetlands and the service and values they provide are also vulnerable to changes, for example, in land use, anthropogenic pollution, and climate change, with potential negative wetland effects and alterations (Ghajarnia et al., 2020; Kharazmi, 2018; Thorslund et al., 2017). The Arctic region contains a substantial proportion of the world’s wetlands and freshwater resources (CAFF, 2019). Thawing of Arctic permafrost could change the abundance and distribution of animals, plants, and macro- and micro-organisms, causing disturbances to ecosystems and ultimately negative impacts on food and water security, infrastructure, and livelihoods (CAFF, 2019; Hjort et al., 2018; Pörtner et al., 2019). It could further release contaminants or pathogens currently trapped in the frozen ground (Revich et al., 2012; Schuster et al., 2018; Seifollahi-Aghmiuni, Kalantari, et al., 2019; Selroos et al., 2019) and have other adverse health impacts (Ma et al., 2019) for the millions of people living in the Arctic region. Moreover, wetland regime shifts could affect the local potential for climate change adaptation, as wetlands act as buffers against floods, storms, increased sea level rise, and wildfires, all of which are expected to increase with future warming (Arctic Monitoring and Assessment Programme, 2017; Box et al., 2019; Markkula et al., 2019; Seifollahi-Aghmiuni, Kalantari, et al., 2019). Previous efforts to assess regime shift vulnerability and impacts of permafrost thaw in Arctic wetlands include (Hugelius et al., 2020; Karlsson et al., 2011), showing regions in the southern border of the Arctic region to be particularly vulnerable.

Among Arctic changes, permafrost thaw can cause wetting or drying of wetland landscapes depending on terrain and ground characteristics (Shur & Jorgenson, 2007). Permafrost thaw in ice-rich ground cause thermokarst (ground collapse), which often leads to wetland formation, but thaw can also contribute to drying of the ground surface through the formation of through-taliks (when the barrier of frozen ground thaws and allows drainage into the subsoil) (Jorgenson et al., 2010). In general, the most drastic regime shifts in wetlands are expected to occur with complete permafrost thaw (Avis et al., 2011). The potential for such regime shifts is the highest for mineral soils, as peat soils tend to have more initial permafrost thickness, thaw less rapidly and persist longer at higher temperatures, mainly due to the insulating capacity of peat with its higher heat capacity and greater water retention than mineral soil (Selroos et al., 2019; Shur & Jorgenson, 2007; Yi et al., 2007). Their greater water retention also tends to imply less runoff from peatlands (Yi et al., 2007) compared to mineral soils. Changes in permafrost are related to ground temperatures (mainly determined by air temperature), snow depth, and local soil properties (Selroos et al., 2019; Wisser et al., 2011). Permafrost is vulnerable to shifting temperatures to above 0°C (Callaghan et al., 2010; Chadburn et al., 2017; Zhang, 2005), especially in the active layer near the surface (Isaksen et al., 2007). Areas with discontinuous, isolated, or sporadic permafrost are especially sensitive to increased temperatures, as initial ground temperatures already relatively high, and permafrost layers already thin and can thaw entirely with even small temperature increases in such areas (Quinton et al., 2009; Riordan et al., 2006). It has been estimated that permafrost persists up to around +2°C (Jorgenson et al., 2010) or somewhere between +1 and +1.5°C in organic soil (Smith & Riseborough, 2002), while permafrost in mineral soil is largely absent at ground temperatures above −2°C (Smith & Riseborough, 2002). Over time, however, if temperatures remain above the thaw threshold for that soil type, all permafrost will thaw, with the timing of total thaw depending on both initial permafrost extent (e.g., thickness) and soil characteristics (Selroos et al., 2019). Even at lower temperatures, changes to the permafrost layer occur at increasingly higher rates from ∼around −7°C toward higher temperatures (Chadburn et al., 2017).

Despite the importance of Arctic wetlands, the numerous datasets currently available (e.g., those reported by Bontemps et al., 2011; Chen et al., 2015; Kobayashi et al., 2017; Lehner & Döll, 2004; Schroeder et al., 2015) are considerably lacking in coverage of and accuracy for the Arctic region. Some are also potentially outdated, as the region is rapidly changing. A consistent high-resolution data set with full coverage is needed, both to identify current locations and extents of Arctic wetlands and to assess their vulnerability to forthcoming regional changes to limit negative impacts. Such information can improve the ability to fulfill sustainability goals, such as the Sustainable Development Goals (SDGs), related to wetlands (Jaramillo et al., 2019), and be vital for adaptation to and mitigation of wetland shifts driven by local and global change impacts. High-resolution maps can improve planning for, mitigation of, and adaptation to various change pressures, with relevant prioritization and focus on the most vulnerable areas, as identified by such mapping, to prevent dire consequences to humans and ecosystems with the greatest benefits at the least cost. This study focuses on identifying the area locations and extents and regime shift vulnerabilities of Arctic wetlands in response to regional temperature changes. As a first step, we have combined various available data to map existing wetlands over the entire Arctic region. We have further evaluated how permafrost thaw due to projected future temperature changes may affect the regime shift vulnerability of wetlands. We have assessed wetland vulnerability based on the observed temperature in 2018 and projected future temperatures in 2050, 2075, and 2100, under three different Representative Concentration Pathways (RCP) scenarios, in combination with current permafrost coverage and soil types over the Arctic region.

2 Materials and Methods

2.1 Study Area

The definition used for the Arctic region in the present analysis is that provided by the Conservation of Arctic Flora and Fauna (CAFF), which includes the countries Denmark (Greenland and Faroe Islands), USA (Alaska), Canada, Russia, Iceland, Norway, Sweden, and Finland (in part or in full). Definitions of what constitutes a wetland vary across the globe. The definition used here is that all organic soils (peatlands) are considered as wetlands and, additionally, that wetlands are water areas covered with vegetation and areas where water is close to, in, or above the soil surface for a substantial part of the year (Gunnarsson & Löfroth, 2009). At least 50% of the vegetation should be hydrophilic for an area to be called a wetland. Oceans and permanent lakes are not considered as wetlands.

Overall, the Arctic region has low population density, with around 90% of settlements having less than 5,000 inhabitants, and a majority of these settlements being in permafrost areas (Heleniak et al., 2019; Jungsberg et al., 2019). For indigenous peoples living in the region, traditional livelihoods depend strongly on services provided by Arctic wetland ecosystems over large areas with traditional land use, such as reindeer or caribou herding, hunting, and fishing (CAFF, 2019). Locally, the indigenous proportion of the population varies between zero and almost 90%, depending on country and area, and the indigenous peoples represent around 9% of the total population in the Arctic region (Jungsberg et al., 2019)

2.2 Mapping of Wetlands

Wetlands, especially peatlands, are known to be difficult to map with high accuracy (Hugelius et al., 2020). As there are considerable differences between wetlands, determining a standard method to define all wetlands is challenging, but several indicators are available to locate areas likely to be wetlands. Type of vegetation, hydrology, soil type, proximity to water, and landform are examples of such indicators (Tiner, 2009). Numerous databases, national, regional, and global, to some extent map wetland areas, but most available databases lack sufficient resolution to make it possible to identify wetland areas and their vulnerability to future change with accuracy. Moreover, none of the high-resolution datasets for soil moisture (1 km) and soil type (250 m) covers the entire Arctic region, implying a need to consider and synthesize several datasets.

To create a comprehensive wetland data set for the Arctic region in the present study, we combined three different types of datasets in a Geographic Information System (GIS): wetland databases (Finnish Environment Institute [SYKE], 2018; Länsstyrelsen i Norrbottens län, 2017; Miljødirektoratet, 2019; Náttúrufræðistofnun Íslands–Icelandic Institute of Natural History, 2014; Naturvårdsverket et al., 2019; Schneider et al., 2009; U.S. Fish and Wildlife Service, 2019); a soil wetness database identifying soils that are wet enough to be wetlands (Widhalm et al., 2015a2015b); and a database on soil types based on the World Reference Base (WRB) and USDA classification systems using soils known to be wetlands soils (Kempen & Hengl, 2017). The wetland databases (Figure 1a) include spatial data on wetland extent. All of these databases, apart from data from the Lena Delta, are provided by national authorities. From the soil-type database provided by ISRIC World Soil Information, wet gleysols and all histosols were extracted, as these are wetland soils (Figure 1c). Together with soil wetness data provided from a research project by Widhalm et al. (2015a2015b) (Figure 1b), these three types of dataset all fit the definition of wetlands used in this study. All the different wetland extent maps were reclassified to binary maps of wetland coverage and overlain in their original spatial resolution. Overlap in mapped wetland areas between one or more datasets is dissolved to create a single map layer of wetland areas with no overlapping features. Oceans and permanent lakes were removed from these datasets to ensure consistency within the considered definition of wetland areas. The used ocean and lakes datasets for doing this are a combination of a global dataset (Buchhorn et al., 2019), national datasets, and the HydroLAKES database (Messager et al., 2016), which is used where high-latitude water-cover data was missing. Parts of the inner Greenland ice sheet (not part of coastal Greenland) have also been mapped as wetlands in Widhalm et al. (2015a2015b) but are excluded from this study. In the soil types database (Kempen & Hengl, 2017), all Gleysols overlaying a wet pixel in Widhalm et al. (2015a2015b) and all Histosols were assumed to be wetlands.

Details are in the caption following the image

Datasets are used for the creation of the wetland map. (a) Wetland databases. (b) Soil wetness data (Widhalm et al., 2015a2015b). (c) Coverage of soil types that are wetlands (Kempen & Hengl, 2017). Projection: Azimuthal Equidistant.

All types and sources of data used to create the map of current Arctic wetlands and assess their regime shift vulnerability under projected Arctic warming and associated permafrost thaw are is listed in Table 1. Spatial data on population density relevant for the year 2018 (WorldPop, www.worldpop.org) and protected areas in June 2020 (UNEP-WCMC and IUCN, 2020) were used to assess how the most vulnerable areas coincide with where individuals live and areas that currently are protected.

Table 1. Data Used in GIS to Map Arctic Wetlands and Their Vulnerabilities in This Study
Type of data Source
Oceans and permanent lakes (Alaska Department of Natural Resources - Information Resource Management, 2007; Buchhorn et al., 2019; Finnish Environment Institute [SYKE], 2018; Lantmäteriet, 2017; Norwegian Polar Institute, 2014; Messager et al., 2016; Náttúrufræðistofnun Íslands – Icelandic Institute of Natural History, 2014; Norwegian Water Resources and Energy Directorate, 2019)
Permafrost areas (Obu et al., 2018)
Population density 2018 (WorldPop, www.worldpop.org)
Protected areas (UNEP-WCMC and IUCN, 2020)
Soil types: Histosols, wet Gleysols (Kempen & Hengl, 2017)
Soil wetness (Class 1, wet) (Widhalm et al., 2015a2015b)
Temperature projection (Fan & van den Dool, 2008; Martin et al., 2011)
Wetland databases from Alaska, Finland, Iceland, Lena river delta (Russia), Norway & Sweden (Finnish Environment Institute (SYKE), 2018; Länsstyrelsen i Norrbottens län, 2017; Miljødirektoratet, 2019; Náttúrufræðistofnun Íslands – Icelandic Institute of Natural History, 2014; Naturvårdsverket et al., 2019; Schneider et al., 2009; U.S. Fish and Wildlife Service, 2019)

2.3 Vulnerability of Wetlands to Permafrost Thaw

Numerous factors can cause regime shifts in Arctic wetlands. In this study, we focused on three factors affecting climate-driven shifts by affecting the propensity for permafrost thaw (Figure 2): (i) current permafrost coverage and type, (ii) soil type (peat or mineral soil), and (iii) projected mean annual average temperature (MAAT). These were selected because: (i) discontinuous, isolated and sporadic permafrost is more vulnerable to temperature changes and undergoes more abrupt thaw (Quinton et al., 2009; Riordan et al., 2006) with higher associated drainage of surface water than continuous permafrost (IPCC, 2018); (ii) there is a strong connection between temperature and permafrost thaw (Callaghan et al., 2010; Chadburn et al., 2017; Isaksen et al., 200720072007; Jorgenson et al., 2010; Selroos et al., 2019; Smith & Riseborough, 2002; Wisser et al., 2011; Zhang, 2005); and (iii) soil type affects the rate of thaw, as mineral soils tend to thaw more rapidly and at lower temperatures than peat soils (Jorgenson et al., 2010; Selroos et al., 2019; Shur & Jorgenson, 2007; Yi et al., 2007). Based on these factors, we developed a set of ranking indices for wetland regime shift vulnerability, ranging from 1 to 5, with 5 being the highest vulnerability (Figure 3) under projected MAAT for the years 2050, 2075, and 2100 in climate change scenarios RCP2.6, 4.5, and 8.5. For example, as permafrost in peat soils is more resistant to thaw, peat soil settings get an overall lower score in the ranking indices. If a wetland is located in an area with peat soil and continuous permafrost, the ranking index will be even lower as the continuous permafrost under that wetland is even more resistant to thaw under warming. In the highest temperature class of 2°C or higher, where permafrost thaw is the most likely to happen, this condition combination would result in a risk ranking index of 4, instead of 5 (the highest score) that would be the case for a wetland in discontinuous permafrost or mineral soil. The ranking is relative to baseline conditions of current permafrost coverage, soil type, and observed MAAT for the year 2018 (Figure 2). Chosen temperature spans reflect the different temperatures where permafrost thaw has been found to occur in previous studies. The +2°C reflects the limit for organic soils found in Jorgenson et al., (2010) and the −2°C limit applies to mineral soil, meaning that permafrost is largely absent at ground temperatures above these limits (Smith & Riseborough, 2002). −7°C is the ground temperature limit above which permafrost thaw starts occurring at increasingly higher rate (Chadburn et al., 2017).

Details are in the caption following the image

Baseline datasets for evaluation of vulnerability ranking indices. (a) Current permafrost coverage (Obu et al., 2018). (b) Soil type divided into mineral and peat soils (Kempen & Hengl, 2017). (c) Mean annual average temperature (MAAT) for the year 2018. Projection: Azimuthal Equidistant.

Details are in the caption following the image

Ranking indices for wetland regime shift vulnerability due to permafrost thaw for different combinations of soil and permafrost types and mean annual average temperature. Higher rank index value implies higher vulnerability.

2.4 Current and Projected Temperature

To evaluate the current status and possible future changes in annual average near-surface air temperature, we used observational data for 2018 and climate projections for the years 2050, 2075, and 2100. For the base year 2018, we employed temperature data from NOAA/NCEP GHCN-CAMS Version 3.01 (Fan & van den Dool, 2008). The climate projections are obtained from the HadGEM2-ES climate model (Martin et al., 2011), which is one of the climate models included in the Coupled Model Intercomparison Project 5 (CMIP5). HadGEM2-ES provides results for relevant land and related atmospheric variables (including temperature data) based on different climate scenario projections. In this study, we first downloaded the monthly temperature projections of the HadGEM2-ES climate model in three different projection scenarios of future climate change (RCP2.6, RCP4.5, RCP8.5) from the CMIP5 data center. Then relevant variable time series were extracted over the Arctic region, and finally, the mean annual temperature for the target years was calculated for every grid cell within the study area. The HadGEM2-ES model is used because it has been found to perform best among various tested GCMs for temperature data in comparison with observation data for the Nordic-Arctic region (Bring et al., 2019). With monthly temperature data from these products, we calculated the mean annual temperature for the target years for every grid within the Arctic region.

3 Results

3.1 Location and Extent of Arctic Wetlands

By combining previous wetland maps (Figure 1a), soil moisture data (Figure 1b), and soil type data (Figure 1c), and removing oceans and permanent lakes, we have created a wetland map covering the entire CAFF area (Figure 4). This map reveals that a significant proportion of the Arctic region is likely wetland-covered; current wetland areas are mapped to extend over approximately 3.5 million km2, representing roughly 25% of the total Arctic land area. Of these wetland areas, about 64% are on peat soils, and 36% are on mineral soils. Around 99% of the wetlands are in areas with some permafrost coverage, with 73% being on continuous permafrost, defined as at least 90% permafrost coverage (Obu et al., 2018).

Details are in the caption following the image

Combined wetland area map for the entire Conservation of Arctic Flora and Fauna (CAFF) region. Projection: Azimuthal Equidistant.

3.2 Vulnerability of Wetlands

Permafrost covers a large extent of the Arctic region (Figure 2). By layering permafrost, MAAT, soil types, and the wetland map together for the different climate change scenarios, we can identify wetland areas particularly vulnerable to future warming. The vulnerability index of wetland areas increases both with time for each climate scenario from 2018 up until 2100 and between scenarios from RCP2.6 to RCP8.5 (Figures 5 and 6). The most considerable changes are in the southern parts of the Arctic region, where temperatures are expected to rise above the thresholds for more extensive permafrost thaw during relatively extended time periods. In the base year 2018, almost 8% of wetlands already have a high vulnerability index of 4 or 5. There is little change between years in the RCP2.6 scenario and compared with the 2018 case, indicating that wetlands would remain largely stable if climate change can be limited to this scenario. The scenario RCP4.5 is at first stable, with little difference between 2018 and 2050 conditions, after which the areas of vulnerable wetlands with index 4 and 5 increase by almost 25% until year 2100. In scenario RCP8.5, the change in these wetland areas is more rapid and more considerable, with a 40% increase until 2100 compared to the 2018 conditions.

Details are in the caption following the image

Relative areas with different vulnerability ranking indices. These areas are expressed as a percentage of total Arctic wetland area on permafrost, and given for different years and Representative Concentration Pathway (RCP) scenarios. Higher rank index value implies higher vulnerability.

Details are in the caption following the image

Vulnerability ranking indices shown for all mapped wetlands in permafrost areas in different years and for different Representative Concentration Pathway (RCP) scenarios in the Conservation of Arctic Flora and Fauna (CAFF) area. Higher rank index value implies higher vulnerability. Projection: Azimuthal Equidistant.

Even though most vulnerable areas are found in the southern parts of the Arctic region, some wetland areas at the southern boundary exhibit no vulnerability increase in any climate scenario, and some areas at higher latitudes exhibit higher vulnerability than some areas at lower latitudes (Figure 6). For example, about half of identified wetlands on Svalbard have a vulnerability index of 4–5 already in the 2050 RCP4.5 scenario.

3.2.1 Affected Populations and Protected Areas

Among the areas with the highest population in 2018 identified as most vulnerable in the 2100 RCP8.5 scenario, the ones with the highest population count in vulnerable areas are in the coastal areas in southwest Alaska and the northwestern part of the Ural region in Russia, with some highly vulnerable and populated areas also in northern parts of Sweden, Finland, the Murmansk Oblast, Russia, the coastal areas in southern Greenland, and the Northwest Territories, close to Lac La Martre and the Great Slave Lake and along the eastern coastline along Hudson Bay, including an area south of Fort Albany in Canada. A large proportion of the population in these areas is indigenous, with Fort Albany being a reserve for an indigenous community. Apart from the area south of Lac La Martre and the Ural region in Russia, these areas are, to some extent, highly vulnerable already under 2018 conditions.

Of the wetland areas with the highest vulnerability index in the year 2100, 17%–30% are in a protected area based on the International Union for the Conservation of Nature (IUCN) definition of such areas (Conservation of Arctic Flora and Fauna, CAFF, 2017). The highest percentage is in scenario RCP2.6, with a significantly lower total area in the high vulnerability range than in scenarios RCP4.5 and RCP8.5. Overall, results are robust and only slightly affected by considering alternative vulnerability ranking indices, with more emphasis on permafrost type or temperature changes (Figure S1).

4 Discussion

4.1 Mapping the Extent of Arctic Wetlands

The wetland map developed in this study identifies some wetland areas that were not classified as such in previous mapping attempts (Bontemps et al., 2011; Chen et al., 2015; Kobayashi et al., 2017; Lehner & Döll, 2004; Schroeder et al., 2015). It could be that wetlands do not exist in some of these areas, as we have not been able to check the ground truth of all mapping results. However, the existence of wetlands is verified in 43% of the wetland area, either with multiple sources (8.9% of the total wetland area), with soil wetness data previously recognized as wetland (38% of wetland area) (Widhalm et al., 2015a2015b), or with national databases (4.4% of wetland area) likely to have a high accuracy; e.g., the Swedish database has been quality checked with infrared images, satellite imaging, and orthophotos (Naturvårdsverket et al., 2019).

The remaining main mapping uncertainty relates to how accurate organic soils (peatlands) are mapped in the SoilGrids dataset (Kempen & Hengl, 2017). Because the surface expressions of peat depth are often gradual or lacking, peatland soils are difficult to map using remote sensing. Hugelius et al. (2020) evaluated the accuracy of different soil maps for peatland mapping and found that the SoilGrids dataset performed well overall but tended to overestimate peatland coverage in some regions. The wetland area mapped as peatland using only the SoilGrids dataset (Kempen & Hengl, 2017) accounts for almost 57% of our total mapped wetland area. Compared with other peatland datasets, 26% of this wetland area is verified using the PEATMAP dataset (Xu et al., 2017) that combined four different datasets, none of which was used in the making of the wetland map for this paper. The approaches discussed above all focus on mapping of organic soils based on statistical analyses of environmental variables or traditional soil survey data. Direct mapping of wetlands using various remote sensing approaches is sometimes challenging, as varying characteristics and dynamics of wetlands make them difficult to distinguish from other land covers (Bartsch et al., 2016; Bourgeau-Chavez et al., 2015; Gallant, 2015; Ozesmi & Bauer, 2002), but many different types of data can be used to good effect (Gallant, 2015). This includes approaches associated with land cover mapping using multispectral remote sensing (e.g., Amani et al., 2019) or soil wetness (Widhalm et al., 2015a2015b) and surface inundation mapping (Aires et al., 2018) using active remote sensing platforms. It has been argued that future generations of Arctic wetland maps should be of high-resolution and have the capacity to delineate discrete wetland complexes (Hugelius et al., 2020), which would require high-resolution applications based on remote sensing. The use of satellite imagery for wetland mapping may be the most promising method for updating wetland maps with changes over time.

Comparison between databases of permanent lakes, identified wetland areas and a recent dataset of global surface water (Pekel et al., 2017) shows high consistency between the latter and the databases of permanent lakes considered in our study, but little overlap with the identified wetland areas (overall 2% overlap). This implies that such water cover data represent primarily permanent lakes and streams but are not useful for identifying wetland areas and their shifts in the Arctic. Further studies distinguishing between peatlands of different types are called for to provide essential information about the wetness status and drainage vulnerability of these areas. The used vulnerability ranking index values are, to some degree, subjective but with an objective ranking-order basis that can still provide valuable insights into the spatial distribution of wetland regime shifts vulnerability to future warming. Our comparison of different ranking approaches and indices shows result robustness in identifying areas with the highest wetland vulnerabilities (Figure S1).

4.2 Vulnerability and Impacts of Changing Wetlands

A substantial proportion of Arctic wetlands lie on continuous permafrost (73%), indicating smaller regime shift vulnerability to warming than in discontinuous, isolated, and sporadic permafrost. Nevertheless, large permafrost areas will still thaw due to increased temperatures in scenarios RCP4.5 and RCP8.5. If climate warming can be limited to the RCP2.6 scenario, wetlands may be largely stable in relation to permafrost thaw until 2100. However, more pronounced impacts should be expected to Arctic wetlands in a future more in line with scenario RCP4.5 or RCP8.5. Findings are consistent with those of (Hugelius et al., 2020), showing higher vulnerability of permafrost peatland in southern parts of the Arctic, with increased vulnerability for increased MAAT.

Some hydrological shifts with increasing temperatures will likely also form completely new wetlands, implying ecosystem regime shifts from terrestrial to aquatic systems (Karlsson et al., 2011). A study by Hugelius et al. (2020) shows shifts in permafrost peatlands from thaw impacting greenhouse gas fluxes and loss of peat carbon, having considerable implications for aquatic biogeochemistry and ecosystems. Peatland transition due to permafrost thaw can shift the peatlands from sinks to sources of greenhouse gas emissions, with young thermokarst lakes (50–150 years) being sources of both CO2 and CH4 (Hugelius et al., 2020). This study has focused on regime shift vulnerabilities of Arctic wetlands in response to regional temperature changes. Wetland regime shifts have the potential to increase greenhouse gas emissions and involve feedback loops that can enhance warming and associated vulnerabilities. Increased temperature also causes additional hydrological changes, for example, in runoff and evapotranspiration (Bosson et al., 2012), which can further affect wetlands along with additional factors also influencing the thaw of underlying permafrost, such as drilling for oil and gas, warm pipelines under and above ground, and changes in vegetation, snow cover, and wildfires (Jorgenson et al., 2010). These types of feedbacks were not considered in the study, implying that the vulnerabilities and consequences of wetland regime shifts could be even more extensive than assessed here. In this study, the uncertainty in the climate change scenarios and the vulnerability analysis were not assessed beyond testing the robustness of results by a shifted yet still reasonable ranking system (found in supporting information).

A changing climate with thawing permafrost, in turn, threatens infrastructure, food and water resources, and human health, especially for the indigenous peoples in the Arctic region, who live in permafrost areas and depend on local resources. It has been estimated that 48%–87% of essential human infrastructure in the Arctic currently exists in areas with a high potential for thawing of near-surface permafrost by 2050 (Hjort et al., 2018). Regional climate change, with thawing of previously frozen land and changing vegetation, affect opportunities for hunting and animal herding by limiting food availability for herding animals, increasing the spread of pathogens, and creating physical barriers to animal movement (Arctic Monitoring and Assessment Programme [AMAP], 2015; IPCC, 2018). Overall, increased Arctic temperatures can change local ecosystems and disease spreading, while global sea-level rise and changes in atmospheric and ocean circulation also negatively affect the ability to live a healthy life and achieve SDGs in the region (IPCC, 2018; Overland et al., 2019; Schuur et al., 2015).

Furthermore, Arctic wetlands are also threatened by direct human impacts, in addition to climate change. Large areas of wetlands have disappeared in response to expanded agriculture and forestry activities, and peat has been harvested for use as fuel (Gunnarsson & Löfroth, 2009), further increasing negative wetland impacts. Sea-ice decline would also make previously inaccessible land areas accessible to ship-based transport, making them a target for mineral, gas, and oil exploitation (Heleniak et al., 2019) and also adding to the negative wetland impacts. In general, such exploitations can increase conflicts with indigenous peoples, as economic development in the Arctic has been linked to nature management conflicts, such as conflicting interests between conservation efforts and shipping or tourism, or between forestry, mining, or oil extraction and animal herding or other indigenous rights (Evseev et al., 2018; Kyllönen et al., 2006; Österlin & Raitio, 2020). Land-use changes for natural resource exploitation can provide job opportunities (Heleniak et al., 2019), but may also increase pollution and disturbances to local ecosystems and biodiversity, with potential long-term consequences, for example of oil spills, increased transportation, and expansion of roads to reach new oilfields and mining areas (CAFF International Secretariat, 2012; Yang et al., 2016).

The importance of the Arctic region is recognized, with more than 20% of its terrestrial area currently classified as protected and with the Arctic Council now trying to increase the proportion of protected areas, especially those important for biodiversity and ecosystem services (CAFF & PAME, 2017). Existing Arctic wetlands should be protected and preserved, considering the possible severe ecologic and socio-economic consequences associated with wetland regime shifts, and also to contribute to control and mitigation of greenhouse gas emissions and global warming as existing wetlands mainly act as carbon sinks while new and restored wetlands are more likely to be sources of greenhouse gases (Hugelius et al., 2020; Kuhn et al., 2018; Turetsky et al., 2020). In general, further research is called for to investigate the various factors affecting and being affected by wetlands and their vulnerabilities in the Arctic. The high-resolution wetland and vulnerability maps developed in this study provide important wetland overview and insights for the Arctic region, useful for identification and prioritization of the most vulnerable areas to changing temperatures. This type of mapping and its underlying data synthesis are needed both for further Arctic studies and for improving efforts to plan for, adapt to, and mitigate negative change impacts and to achieve long-term sustainability for the Arctic wetlands.

5 Conclusions

The new comprehensive map of Arctic wetlands created in this study provides an essential basis for assessing wetland areas and their vulnerabilities to future warming and permafrost thaw. The map can support regional planning of mitigation and adaptation measures to limit negative change impacts on the Arctic wetlands and associated societal and ecosystem effects. Wetlands cover large parts of the Arctic region and provide vital ecosystem services locally, regionally, and globally (e.g., via weather systems), with loss of wetlands potentially resulting in devastating local, regional, and up to global impacts. The Arctic wetlands may remain relatively stable if climate scenario RCP2.6 can be attained. However, if scenario RCP8.5 becomes a reality, severe wetland impacts can be expected, as at least 50% of total Arctic wetland areas are vulnerable to warming in that scenario. Associated effects on health, socio-economics, and ecosystems imply urgent needs to limit climate change and preserve permafrost and related wetlands in the Arctic region.

Acknowledgments

This study arose from three research projects: (1) project Resilience and management of Arctic wetlands under the Arctic Council Conservation of Arctic Flora and Fauna (CAFF) Working Group and funded by the Swedish Ministry of the Environment and Energy in collaboration with CAFF; (2) Nordforsk Centre of Excellence CLINF (grant number 76413); (3) A project funded by Formas, 2017-00,608.

    Data Availability Statement

    Databases on national and local wetland datasets and for the removal of permanent lakes was provided from numerous sources, among which one is from a research project by Schneider et al. (2009) covering wetland data from the Lena Delta provided through PANGAEA, and the rest are provided by national authorities from Finland, Iceland, Norway, Sweden and US (Alaska) (in alphabetical order of provider country): Finland’s environmental administration via the website https://www.syke.fi/en-US/Open_information/Spatial_datasets/Downloadable_spatial_dataset (wetland and lake data); Icelandic Institute of Natural History via https://en.ni.is/resources/publications/maps/vegetation-maps (wetland and lake data); Norwegian Environment Agency and Norwegian Water Resources and Energy Directorate via https://kartkatalog.miljodirektoratet.no/Dataset/Details/2031 (wetland and lake data); Norwegian Polar Institute via https://data.npolar.no/dataset/645336c7-adfe-4d5a-978d-9426fe788ee3 (lake data for Svalbard); County Administrative Board of Norrbotten via http://mdp.vic-metria.nu/miljodataportalen/GetMetaDataById?UUID=a41d282b-4c2a-4838-bfaa-cf718c4a7df3 (wetland data); Swedish Environmental Protection agency via http://www.naturvardsverket.se/Sa-mar-miljon/Kartor/Nationella-Marktackedata-NMD/Ladda-ned/(wetland data); Lantmäteriet via https://zeus.slu.se/get/?drop=get (lake data); Alaska Department of Natural Resources - Information Resource Management via https://catalog.data.gov/dataset/alaska-hydrography-1-63360 (lake data); and US Fish and Wildlife Service via https://www.fws.gov/wetlands/Data/State-Downloads.html (wetland data). Soil moisture data were provided from a research project by Widhalm et al., 2015a2015b, provided through PANGAEA, which is hosted by the Alfred Wegener Institute, Helmholtz Center for Polar and Marine Research (AWI), and the Center for Marine Environmental Sciences, University of Bremen (MARUM) via https://doi.pangaea.de/10.1594/PANGAEA.840548. SoilGrids250 m data of different soil types was provided by ISRIC World Soil Information, Wageningen, Netherlands, via https://data.isric.org/geonetwork/srv/api/records/5c301e97-9662-4f77-aa2d-48facd3c9e14. Permafrost zonation data were provided through PANGAEA via https://doi.pangaea.de/10.1594/PANGAEA.840548. For the removal of oceans and permanent lakes, data was also here provided from numerous sources, with the majority of the data provided by Copernicus Global Land Service through http://doi.org/10.5281/zenodo.3243509. Water datasets from national authorities already mentioned were included to complement the primary dataset, together with HydroSHEDS provided from the World Wildlife Fund (WWF) via https://www.hydrosheds.org/. GHCN-CAMS Gridded V2 data of near-surface air temperature was provided by NOAA/OAR/ESRL PSL, Boulder, Colorado, USA, via https://psl.noaa.gov/. For evaluation of affected populations and protected areas, population data were provided through Worldpop via https://dx.doi.org/10.5258/SOTON/WP00647 and protected areas were provided through Protected Planet: The World Database on Protected Areas (WDPA) via https://www.protectedplanet.net/en/thematic-areas/wdpa?tab=WDPA.