Introduction

Northern peatlands account for approximately 25% of the global soil carbon (C) pool, with most of the peatlands occurring in Russia and Canada (Yu, 2012). Peatlands are categorized as ombrotrophic bogs (receiving water primarily from precipitation) and minerotrophic fens (receiving water from precipitation and groundwater inflow). Bogs and fens cycle C differently (Bauer et al., 2009), and their C sink strength may change as peat warms and becomes drier or wetter (Webster et al., 2013). However, vulnerability and adaptation of peatland C budgets in a changing climate are largely unknown (Frolking et al., 2011).

The Far North of Ontario (located between approximately 50 and 57°N, 79 and 94°W) is a highly variable landscape that consists of about 45 million hectares distributed relatively evenly between the Canadian Shield and Hudson Bay Lowlands (HBL). Here, peatlands are the dominant land class, accounting for nearly 50% of the area (Fig. 1). These peatlands store an estimated 36 Gt (1 Gt = 1 Pg = 10¹⁵ g) of C as peat (Far North Science Advisory Panel, 2010), which is approximately 25% of the C stored in all of Canada's peatlands. The HBL, which contains nearly 75% of the C stored in the Far North, is the largest peatland complex and has the southernmost extent of non-alpine permafrost (subsurface earth materials that remain below 0 °C for two or more consecutive years [Brown, 1967]; see Table 1 for permafrost features and their definitions) in North America (Martini, 2006).

Peatlands in the HBL have been important in Holocene climate cooling because since deglaciation they have sequestered large amounts of atmospheric carbon dioxide (CO₂) (van Bellen et al., 2011; Bunbury et al., 2012). However, when disturbed, these peatlands may release much of this C back to the atmosphere as CO₂ or methane (CH₄), which may speed global warming (Webster and McLaughlin, 2011). For example, although warmer and drier peat may increase tree and shrub biomass (Weltzin et al., 2003), peat decomposition and CO₂ loss may dominate over its gain through net primary production (NPP). In contrast, warmer and wetter peat enhances sedge and moss biomass, characteristic of fens that support rapid C accumulation in, and CH₄ emission from, peat (Webster et al., 2013). Therefore, although CO₂ is sequestered and accumulated as peat, CH₄ losses may offset C gain and enhance atmospheric global warming potential (GWP; sum CO₂ and CH₄ fluxes expressed as CO₂ equivalents].

Ontario House Bill 191, the Far North Act ( http://www.ontla.on.ca/bills/bills-files/39_Parliament/Session2/b191ra.pdf), which achieved Royal Assent in October 2010, describes four land-use planning objectives. One of those objectives is “the maintenance of biological diversity, ecological processes and ecological functions, including the storage and sequestration of carbon in the Far North.” Furthermore, Far North land-use planning teams may request the development of several policy statements, one of which concerns “ecological systems processes and functions, including considerations for cumulative effects and for climate change adaptation and mitigation.”

Achieving the act's planning objective and developing an appropriate policy statement for land-use planning requires an understanding of how climate change affects peatland C storage and sequestration at the landscape scale. Spatial configurations of permafrost and peatland types, succession and plant communities, along with subsequent evapotranspiration (ET) are important controls on landscape C storage and sequestration. For example, plant communities alter peatland C cycling through plant photosynthesis and the quantity and quality (i.e., decomposability) of litter added to peat (Laiho, 2006). Permafrost amount and thaw rates can also alter landscape-scale C budgets by creating drier or wetter conditions (Malmer et al., 2005). The latter inhibits organic matter decomposition relative to plant NPP, but enhances CH₄ emissions, whereas the opposite occur when peat dries (Johansson et al., 2006b). Furthermore, ET is positively correlated with CO₂, sequestration under wet conditions (Sonnentag et al., 2010). However, dry peat created by prolonged elevated ET relative to water input is also more susceptible to fires that burn deeper into peat (Turetsky et al., 2004) than occur in current HBL landscapes (Balshi et al., 2007), further contributing net peatland C losses to the atmosphere.

FIGURE 1.

Ontario land cover map. Solid white line approximates the Far North of Ontario boundary (adapted from OMNR, 2000, with permission).

TABLE 1

Features common in the continuous (CPZ) and discontinuous (DPZ) permafrost zones in the Far North of Ontario.

In this paper, we synthesize the processes controlling peatland C accumulation and greenhouse gas (GHG) exchange in the HBL. The approach was to compile, analyze, and discuss the data on peatland C fluxes, succession, permafrost, and ET from the literature. We conclude by presenting a climate change vulnerability and adaptation assessment framework for HBL peatlands, discussing challenges to its development and its application in land-use planning.

Peatland Carbon Indicators

Successful communication among researchers, policymakers, and land managers requires common terminology. For this paper, peat C content is defined as mass of C contained in a unit area (e.g., kg C m^-2) of dry peat. Carbon content is calculated as the product of C concentration, peat bulk density, and sample increment thickness. Peat C storage is defined as the product of C content, area of peatland in a landscape, and peat depth—the latter not well documented in subarctic peatlands. Carbon dioxide sequestration is defined as short-term (annual to decadal) balance of atmospheric CO₂ uptake by plants and its release through plant respiration and organic matter decomposition. Carbon dioxide sequestration is commonly measured using gas flux chambers and eddy covariance flux towers. Carbon accumulation is defined as net C gain in peat at century to millennial scales. Peat C accumulation is measured by (1) dating of surface peat with specified tracers, such as ²¹⁰Pb and ¹³⁷Cs, that provide decadal to century (e.g., 50 to 200 years) time scales of peat accumulation (i.e., recent rate of C accumulation [RERCA]; Oldfield et al., 1995) and (2) radiocarbon (¹⁴C) dating of basal peat and peat profile age/depth relationships to estimate century to millennial C accumulation (i.e., long-term rates of C accumulation [LORCA]; Kuhry and Vitt, 1996). These rates are considered apparent because without intensively dating individual peat cores and identifying wetness using paleoecological indicators, timelines for and causes of LORCA and RERCA changes during peatland development cannot be inferred (Yu, 2012). However, differentiating sequestration from accumulation is critical in climate change vulnerability assessments because CO₂ sequestration in plant biomass and surface litter does not equate to long-term C accumulation as peat.

TABLE 2

Number of peatland studies reporting data on carbon accumulation and sequestration and methane release that were used in this synthesis by peatland type/feature.

FIGURE 2.

Box plots of long-term rates of carbon accumulation (LORCA) and recent rates of carbon accumulation (RERCA), where the box height indicates the interquartile range, the solid white line within the box represents the median value, and the dashed lines from the top and bottom extend to the highest and lowest values, respectively. Different letters for LORCA and RERCA, respectively, indicate statistical differences at p < 0.05.

Data sets for peatland C indicators analyzed in this study were compiled from 79 published papers; the number of studies and data points available for LORCA (g C m^-2 yr^-1), RERCA (g C m^-2 yr^-1), CO₂ sequestration (g C m^-2 d^-1), and CH₄ emission (mg C m^-2 d^-1) are shown in Table 2. A total of 680 values were retrieved from the publications, with approximately 70% corresponding to GHG exchange and the remainder to C accumulation rates. Mean values were calculated for each study reporting multiple rates by peatland type; this resulted in 129 data points to conduct the statistical analyses described below. Because of small sample sizes and infrequent differences among bog and fen hummocks, permafrost palsas, peat plateaus, and polygonal peatlands, they were classified as “dry features.” Similarly, collapse scars, lawns, carpets, hollows, and pools in bogs and fens were classified as “wet features.” Although only four ponds were represented in the final data set for CO₂ sequestration, they were considered a separate category because all were net CO₂ and CH₄ sources to the atmosphere. A meta-analysis to assess the effects of experimental manipulations on C fluxes was beyond the scope of this paper, as the primary interest was an initial evaluation of potential C indicator variability. Meta-analysis is, however, recommended as the next step in peatland C indicator selection.

Differences in LORCA and RERCA between dry and wet features were assessed with the Mann-Whitney test, while a Kruskal-Wallis (Conover, 1980) test was used to assess differences in CO₂ sequestration and CH₄ emission from wet, dry, and pond features. Coefficient of variation (CV) and sample sizes were used to calculate the power (probability of failing to reject a false null hypothesis; Type II error) of the statistical tests. In addition, sample sizes required to detect a 20% difference in mean values of each indicator at the 0.05 level of significance (α) and a power of 0.70 were calculated following the methods of McLaughlin and Phillips (2006). Quartile analysis was used to define the bounds of natural variation for the indicators using the inter-quartile range between the first through third quartiles (25th through 75th percentiles), to which data compiled for HBL peatlands were compared.

Long-term C accumulation rate is summarized in Table 3; it was the most promising indicator (power = 0.78) analyzed in this study. No difference (p = 0.281) occurred between dry and wet features; therefore, data were combined into one peatland class. Natural variation of LORCA ranged from 18.0 to 24.5 g C m^-2 yr^-1 (Fig. 2), and its mean value in HBL peatlands (16.4 ± 4.9 g C m^-2 yr^-1 ; n = 11) was below the 25th percentile (18.0 g C m^-2 yr^-1) of the overall data set. Although the power of the Mann-Whitney test was 0.40 for RERCA, it was higher (p = 0.009) in wet than dry features (Fig. 2). Furthermore, mean RERCA in HBL peatlands (77 ± 42 g C m^-2 yr^-1; n = 9) was similar to the 50th percentile (78 g C m^-2 yr^-1) of the overall data set. The above results suggest RERCA may be a viable decade-to-century indicator of C accumulation. However, it would be beneficial to conduct additional RERCA measurements in boreal and subarctic peatlands, including the HBL, to increase statistical power.

TABLE 3

Recent and long-term carbon accumulation rates reported for northern peatlands.

Greenhouse gas fluxes are summarized in Table 4; CO₂ sequestration (power = 0.17) and CH₄ emission (power = 0.24) were characterized by high variation (Fig. 3, parts a and b) and low statistical power. Furthermore, CO₂ sequestration did not differ (p = 0.766) between dry and wet features, and they were weak CO₂ sinks (Fig. 3, part a). Ponds may be a CO₂ source; however, statistical inferences are limited by the sample size (n = 4) and caution is recommended when interpreting the pond CO₂ exchange data. For example, ponds covering mineral sediments are large CO₂ sources to the atmosphere, whereas those over organic sediments alternate between being CO₂ sinks and sources and tend to be smaller sources when assessed on an annual basis (Macrae et al., 2004). Differences in mineral and organic sediments should, therefore, be considered in scaling and modeling CO₂ sequestration in peatland ponds.

Mean CO₂ sequestration in dry features of the HBL (-0.62 ± 1.04 g CO₂-C m^-2 d^-1) was similar to the 50th percentile (-0.64 g CO₂-C m^-2 d^-1) of the overall data set (Fig. 3, part a). However, 7 of the 11 data points from the HBL analyzed here came from the 1990 Northern Wetland Studies in the southern HBL (Glooschenko et al., 1994) and may not be applicable to the entire HBL or its current landscape structure. Nearly all the peatlands (97%) analyzed in this study were net emitters of CH₄ to the atmosphere, with wet features and ponds releasing approximately 25 times more CH₄than dry features (p < 0.0001; Fig. 3, part b). Variation in CH₄ emission was least in ponds (CV = 0.55), followed by wet (CV = 0.77) and dry (CV = 0.91) features. In the HBL, mean CH₄ emission from wet (41 ± 41 mg CH₄-C m^-2 d^-1, n = 10) and dry (5.4 ± 4.4 mg CH₄-C m² d^-1, n = 8) features, as well as ponds (87 ± 41 mg CH₄-C m^-2 d^-1 n = 8), fell within the bounds of natural variation established in this study (Fig. 3, part b).

FIGURE 3.

Box plots of greenhouse gas fluxes from boreal and subarctic peatlands to the atmosphere calculated from the literature: (a) CO₂, where the box height indicates the interquartile range, the solid white line within the box represents the median value, and the dashed lines from the top and bottom extend to the highest and lowest values, respectively, and (b) CH₄, where the box height indicates the interquartile range, the solid white line within the box represents the median value, and the dashed lines from the top and bottom extend to the highest and lowest values, respectively. Different letters for CO₂ and CH₄, respectively, indicate statistical differences at p < 0.05.

Ecological Processes and Carbon Associations

PEATLAND SUCCESSION

Peatland development in the HBL was primarily controlled by recession of the Tyrell Sea and subsequent isostatic rebounding of the land (Glaser et al., 2004). These peatlands emerged more than 7000 years ago, although ¹⁴C basal peat dates produced variable ages (Table 5). The results reported in Table 5 were summarized from 54 peat depth and 42 ¹⁴C basal peat dates, respectively, collected in the Hudson Bay region. These data produced mean peat depth and age of 236 ± 92 cm and 4826 ± 1356 ¹⁴C years, respectively, and neither depth (p = 0.256) nor age (p = 0.523) differed among peatland types; statistical power was near 0.70 for both depth and age. Natural variation for the combined data was 168 to 266 cm for depth and 4003 to 5810 for ¹⁴C age. Variation in both depth (CV = 0.39) and age (CV = 0.28) was comparable to variation of LORCA.

In the HBL, peat and C accumulation rates were generally high, between 6500 and 4200 cal yr BP (calibrated years before present), then decreased as rich and intermediate fens transformed to poor fens, with Larix laricina (larch) and ericaceous plants present (Arlen-Pouliot and Bhiry, 2005; Bunbury et al., 2012). Since 2500 cal yr BP, poor fens generally transformed to bogs, and in the northern HBL bogs may have contained permafrost. In general, bogs may have accumulated the least amount of peat during development (Arlen-Pouliot and Bhiry, 2005). Further south, low peat and C accumulation rates were also reported during bog and poor fen successional stages (Bunbury et al., 2012). Here, however, the fastest accumulation rates occurred between 1000 and 600 cal yr BP and have since declined. At the southernmost portion of the HBL (Eastmain region of Québec), high C accumulation rates occurred in bogs during dry conditions because of high bulk density and C concentration in buried wood (Loisel and Garneau, 2010). Also, rapid accumulation rates during wet conditions were dependent on the presence of hummock-mosses because of their slow decomposition rates (van Bellen et al., 2011).

Peat temperature and wetness are important controls on peatland succession, and plant communities can evolve rapidly in response to modified peatland and permafrost conditions. For example, sampling in a permafrost peatland complex in northern Sweden over 30 to 50 years showed more trees, ericaceous shrubs, and hummock-mosses in drier areas after permafrost thaw (Malmer et al., 2005). However, sedges expanded at the landscape scale due to spreading of wet areas at the cost of hummocks as permafrost thawed, contributing to increased CO₂ sequestration and ch₄ emissions (Johansson et al., 2006b). Resampling of another Swedish peatland complex indicated that tree cover increased from less than 5% of plots initially sampled to approximately 30% of plots sampled 42 years later (Gunnarsson et al., 2002). Percentage of plots containing mosses also increased, whereas those containing sedges decreased. In a large-scale water table and temperature manipulation experiment in northern Minnesota (Weltzin et al., 2003), ericaceous shrub cover increased 50% four years after peat in a bog mesocosm was warmed from 1.6 to 4.1 °C; again, sedge cover decreased. Furthermore, above- (30%) and belowground (100%) biomass increased when water tables were low (Weltzin et al., 2003). Higher water table levels in a fen mesocosm resulted in increased sedge cover, whereas lower water table levels produced shrub results similar to those in the bog.

Remote sensing studies also indicated increased tree and shrub vegetation under warmer and drier conditions, or sedges, hollow-mosses, and tall shrubs under wetter conditions within the past 50 years (Cornelissen et al., 2001; Tape et al., 2006). Therefore, establishing plant function types (PFTs) should facilitate scaling and modeling C budget responses to landscape-level drying or wetting. Combining remote sensing of peatland PFTs and peat accumulation rates and GHG exchange measured at research sites and monitoring networks allows the use of calibrated and validated models to estimate C balances (St-Hilaire et al., 2010). The PFTs used depend on scale. For large landscapes, vascular (e.g., treed bogs and fens) and non-vascular (e.g., open bogs and fens) plant groups have been used (Dorrepaal, 2007). Results presented in this paper suggest dry, wet, and pond classifications may also be useful for scaling and modeling current and future C accumulation and GHG changes of HBL landscapes. Peatlands may also be more narrowly defined using vegetation classes, such as (1) mosses, sedges, shrubs, and trees or (2) genera- and species-level PFTs for smaller areas (Dorrepaal, 2007). For example, sedges, tall shrubs, and Sphagnum mosses have been used successfully to scale CH₄ emissions at site (Johansson et al., 2006b) and regional (Dorrepaal, 2007) scales.

TABLE 4

Carbon dioxide (CO₂) sequestration and methane (CH₄) emissions for northern peatlands.

TABLE 5

Basal peat ages in peatlands in permafrost zones of Canada, Peat depth Age

PEATLAND EVAPOTRANSPIRATION

Evapotranspiration is reported as actual (ET_a) and potential (ET_p), and northern peatlands frequently evaporate water at rates lower than their potential (Rouse, 2000); the rate of ET is strongly influenced by the surface energy balance (i.e., latent heat flux; Q_e). Evapotranspiration affects peatland C exchange at landscape scales primarily through its contribution to water balances. For example, when water was abundant, ET and gross primary production in bogs were positively correlated (Sonnentag et al., 2010), but under water-stressed conditions, photosynthetic depressions were linked to elevated ET in a variety of peatland plants (Euskirchen et al., 2012). Furthermore, elevated ET relative to water inputs may contribute to long-term peatland drying, leading to higher frequency and severity of fires, potentially contributing further to weakening peatland C sink strengths (Turetsky et al., 2004). Because of its relationship with C exchange, accurate estimates of ET are required to construct water balances and assess their potential responses to climate change. However, the amount of water withdrawn through ET varies between 20 and 80% of the water balance, depending on terrain unit (e.g., site type), permafrost conditions, and temperature and precipitation regimes (Rouse, 1998).

Since large areas of the northern HBL occur as palsa-bogfen-pool/pond complexes and small lakes (Martini, 2006), ET estimates for dominant features are needed before a scaled value can be derived for a specific peatland complex (e.g., Petrone et al., 2000). Although ET rates vary (Table 7), tundra ponds, small lakes, and wet sites including fens, marshes, and tundra wetlands displayed the highest ET rates. Dry features, including spruce forests, non-forested bogs, lichen/heath, heath/shrub, tussock tundra, and peat plateaus/polygonal tundra had the lowest ET rates. Analysis of data compiled from 28 studies (Tables 7 and 8) indicated ET was higher (p = 0.002) in wet and pond than in dry features (power = 0.61). Evapotranspiration rates in wet features in the HBL (2.8 to 3.5 mm d^-1) were within the range of natural variability (2.5 to 3.5 mm d^-1) calculated from the overall data set. Wet features (0.71 ± 0.15) also partitioned more energy (p = 0.003) to ET than dry features (0.49 ± 0.08), as defined by the Q_e-to-net radiation (Q*) ratio (Table 8). The statistical power was 0.58, but the smallp value suggests Q_e/Q* should be further pursued as an indicator of wet and dry feature energy partitioning and water budgets in developing methods to assess climate change vulnerability and adaptation.

Looking at estimates of how expected changes in climate might influence annual ET patterns in a sedge fen at Churchill, Manitoba, Rouse (1998) reported that a two times CO₂ climate warming scenario with a MAAT increase of 4 °C and no precipitation change increased summer water deficit, triggered mainly by greater ET during May. The water deficit could be counterbalanced by a 23% increase in summer rainfall (Rouse, 1998). Similar results were recorded in the James Bay Lowland (Lafleur et al., 1993). Roulet et al. (1992) assumed a 3.5 °C increase in temperature and 1 mm day^-1 increase in precipitation for several northern fens and reported an increase in the soil moisture deficit in excess of 100%. Other studies across Canadian peatlands have since shown that thresholds of about a 25% increase in water supplied by precipitation or permafrost thawing would be required to maintain soil moisture levels under warminginduced enhancement of ET (Quinton et al., 2011).

FIGURE 4.

(a) Permafrost distribution across Canada illustrating permafrost zones, which are defined by the proportion of area underlain by permafrost. (b) Mean annual near-surface temperature distribution across Canada. Image (a) adapted from Department of Natural Resources Canada, < http://geogratis.gc.ca/api/en/nrcan-rncan/ess-sst/092b663d-198b-5c8d-9665-fa3f5970al4f.html>. Image (b) from < http://www.ccin.ca/home/ccw/permafrost/future>.

TABLE 6

Predicted change in permafrost area and active layer depth during the 21st century in Canada.

Effects of climate change on ET are more difficult to predict for dry than for wet features. This has been attributed to the phenology of vascular plants, predominantly leaf area index (LAI), and its control on energy partitioning in sparsely vegetated bogs (LAI < 1.0) (Admiral et al., 2006). Leaf area index is minimal in spring, when peat is generally saturated (Lafleur and Humphreys, 2008). This corresponds to the time of highest solar energy reaching bog surfaces and high Q_e/Q* values due to Sphagnum mosses, which account for 50 to 70% of the total annual ET from bogs (Kellner, 2001). Increasing LAI decreases the amount of radiation reaching the surface (Admiral et al., 2006). At this time, ET switches from moss- to vascular plant—dominated, and subsequently dries the peat (Kellner, 2001).

Understanding moss and vascular plant interactions with projected earlier snowmelt is critical to comprehend future ET trends of dry features because of the earlier onset of ET by mosses. However, few studies have documented the competition between mosses and vascular plants following snowmelt. Therefore, it is uncertain if earlier snowmelt will only shift moss ET to earlier in the year or if soil warming during spring will enhance nutrient mineralization and subsequent vascular plant biomass and LAI, prolonging soil moisture deficits. Most model results project that during the 21st century vascular plant biomass and LAI will increase. Such increases have been positively correlated with aboveground biomass and soil C accretion (Frolking et al., 2011). However, LAI increases are also positively correlated with fire frequency and area burned because of dry surface peat (Turetsky et al., 2004). As such, we view (1) moss and vascular plant competition following earlier snowmelt and (2) extended growing season, energy partitioning, and peatland drying on C accumulation rates and GHG exchange rates as key research areas, where data are vital to understand and improve future bog C budget projections.

Permafrost is a strong heat sink and generally has a larger ground heat flux (Q_g) than non-permafrost peatlands, and changes in temperature and moisture may alter permafrost thaw rates (Quinton et al., 2011). However, few data exist to confidently describe and model the energy balance in permafrost features. Boike et al. (2008) reported that during a wet year in polygonal tundra of the continuous permafrost zone (CPZ), energy partitioning remained typical of wet features (Table 8). However, when soil was dry, a relatively even partitioning of energy occurred (Table 8). In addition, during dry years, energy partitioning was not affected by differences in air temperature. However, during wet years Q_e increased with increasing temperatures (Boike et al., 2008). Similar patterns were shown for wet and dry tussock tundra underlain by permafrost (Harazono et al., 1998) and mesic and heath/shrub tundra (Lafleur and Humphreys, 2008).

Climate Change Vulnerability and Adaptation Assessment Framework to Assess Peatland Carbon Indicators

The large contribution of the HBL to provincial and national C budgets implies that understanding peatland C accumulation GWP gains and losses is vital to sustaining C storage and sequestration in the ever-changing HBL landscape. Although scant, evidence is convincing that PFTs, permafrost, and ET are changing in the HBL. However, rates and direction of peatland changes are variable and depend on whether peat becomes drier or wetter as it warms (Fig. 5). Some changes, such as thermokarst formation (Arlen-Pouliot and Bhiry, 2005) and transformations within fen types as they convert to bogs (McLaughlin and Webster, 2010), occur rapidly (i.e., decades); those two processes have opposite effects on peatland C balances (Fig. 5). However, successional trajectories in the HBL are poorly documented (although recent summaries of data collected from the 1970s and early 1980s are available [Riley, 2011]). Also, the remoteness of the region hinders large-scale sampling endeavors due to high costs.

TABLE 7

Evapotranspiration (ET) rates for common site types in high boreal and subarctic drainage basins north of 50° latitude.

TABLE 8

Energy balance partitioning, Bowen ratio [latent heat flux (Q_e)/net radiation (Q^*); sensible (Q_h)/Q^*; ground heat flux (Q_g)/Q^*; lake heat storage Q_w/Q^*], Bowen ratio energy balance (β), and Priestly-Taylor alpha (α) for different site types in peatlands and permafrost features north of 50° latitude.

Remote sensing using various platforms and sensors and modeling techniques have been employed to assess PFTs, permafrost thawing, ET, and C indicators at the landscape scale (Dorrepaal, 2007). Furthermore, the ecosystem processes presented here have consistently been correlated with various components of C budgets and successfully scaled across landscapes (e.g., Johansson et al., 2006b; Beilman et al., 2009). However, few researchers have attempted to develop correlations between climate change and peatland C indicators for the HBL. Significant efforts are required to sample peat and calibrate satellite and aerial photographs for PFTs, permafrost coverage, and ET to calibrate permafrost and peatland C models.

In terms of C indicators assessed in this study, low variation and high statistical power in LORCA comparisons point to it being a promising C indicator. Coupled with relatively small variability in peat depth and ¹⁴C basal peat age, long-term peatland processes may be effective metrics in climate change assessments. However, LORCA in the HBL peatlands was on the low end of the natural variation calculated across boreal and subarctic peatlands. Furthermore, LORCA values were from a limited number of locations and peatland types in the HBL; thus, additional measurements in the HBL are warranted to further evaluate LORCA variability. Although statistical power was relatively low for RERCA and CH₄ emissions, rates were much higher in wet (and CH₄ for ponds) than dry features, suggesting their potential usefulness in evaluating landscape-scale peatland C balances in response to decade-tocentury permafrost and PFT changes. However, much of the data from the HBL analyzed in this synthesis pre-dates the mid-1990s and may not be applicable across the region or to its current landscape structure. In addition, high variation and low statistical power in CH₄ emission (and CO₂ sequestration) hinders construction of viable GHG balances in HBL peatlands. Assigning to landscape GHG changes may be possible but requires further analyses.

Current and future peatland C accumulation and GHG emissions in northern peatlands may also be underestimated due to (1) poorly defined climate change rates and subsequent peatland adaptation and (2) whether new climate states will be within the range of Holocene variation (Frolking et al., 2011). As indicated by Frolking et al. (2011), the only approach available to address climate change and C budgets is rudimentary process-based modeling, the results of which are inconsistent with one another and field data. For example, warming may increase peatland C losses by 40 to 85% in coming centuries (Ise et al., 2008), or peatland C stocks may stabilize in response to drying (Frolking et al., 2010). However, none of the available permafrost (e.g., NEST) and C (e.g., McGill Wetland Model [St-Hilaire et al., 2010], Wetland-DNDC [Zhang et al., 2002]) models allow plant communities or spatial patterns in permafrost and peatland features to vary during simulations, significantly limiting their applicability to functional landscapes.

Information about the quality (decomposability) of peat exposed to aerated or flooded conditions with warming temperatures is also lacking. Deeply buried peat is primarily “recalcitrant” material due to prior decomposition in aerobic surface peat layers (Laiho, 2006). However, the effects of warmer and drier (or wetter) conditions on decomposition rates of “recalcitrant” peat are conflicting, with some studies reporting faster decomposition (Dorrepaal, 2007) and others showing no rate changes (Davidson and Janssens, 2006) under warmer temperatures. Furthermore, peat enclosed in permafrost tends to be less decomposable than that of enclosed mineral soil organic matter (Turetsky et al., 2007). However, data on short- and long-term decomposition of C released through ice melting in permafrost regions are scarce. Finally, dissolved organic C (DOC) may be an important regulator of microbial decomposition in peatlands (Laiho, 2006). The quality of DOC to support decomposition is known to differ among peatland types and between dry and wet conditions (Webster and McLaughlin, 2010). How those changes will be manifested in future organic matter decomposition in peatlands, however, is virtually unknown.

Other uncertainties and challenges also impede the use of climate and C indicators for assessing climate change vulnerability and adaptation. For climate change, lack of weather and permafrost monitoring networks in the HBL significantly limits understanding of past and current temperature, precipitation, and permafrost regimes, contributing to highly variable projections in future conditions. For example, mean monthly precipitation projections differed by more than 200% (Gagnon and Gough, 2005; McKenney et al., 2010). Precipitation amounts are important to landscape permafrost coverage within the same temperature zone (Rouse, 2000; Johansson et al., 2006a). Interactions between air temperature and precipitation with peat temperature and moisture dynamics are also some of the most poorly calibrated coefficients used to project permafrost and peatland C dynamics (Zhang et al., 2008; St-Hilaire et al., 2010).

Few valid estimates of permafrost amount and its thaw rate exist for the HBL. Payette et al. (2004) reported that between 1957 and 2003 percent annual permafrost loss increased from 2.5 to 5.3%, which is much higher than the natural range of variation (e.g., 0.33% yr^-1 to 0.73% yr^-1 assuming a 100-year time frame over the 21st century) calculated from coarse-scale climate and process-based models. Current models, such as NEST, may constrain variability by accounting for plant structure, snow depth, and soil organic layer thickness effects on energy and water transfer in soil-vegetation-atmosphere systems, and subsequent ground thermal dynamics. However, all models fail to address changes in landscape pattern, which is a critical modeling gap. For example, assuming a 50% increase in fen and 20% increase in pond areas in the northern HBL (Payette et al., 2004) during the 21st century, landscape-level CO₂ and CH₄ emissions to the atmosphere may increase significantly and would not be detectable in coarsescale projections. Such changes have been shown to significantly weaken C and GWP sinks (Malmer et al., 2005; Johansson et al., 2006b).

FIGURE 5.

Conceptual model of permafrost and non-permafrost peatland succession (arrows) in response to changing peat conditions (gray boxes) in the Hudson Bay Lowlands of Ontario. C = carbon, CO₂ = carbon dioxide, CH₄ = methane, GWP = global warming potential.

In conclusion, this synthesis indicates that changes in C accumulation rates and CH₄ emissions in response to altered ecosystem processes, such as succession, permafrost loss, and ET may be useful to assess peatland climate change vulnerability and adaptation for large HBL landscapes. Furthermore, C accumulation and CH₄ emission may be evaluated through remote sensing identification of changes in wet, dry, and pond landscape features. However, an improved understanding of the interactions between GHG fluxes and permafrost dynamics (loss and ALT) is needed before such assessments are possible for small landscapes. Therefore, enhanced research and monitoring studies are needed to estimate current and project future peatland C storage and sequestration with reasonable certainty to support land-use planning in the HBL.