Recent peat and carbon accumulation following the Little Ice Age in northwestern Québec, Canada

We selected two regions in northwestern Québec: one within the sporadic and the other within the discontinuous permafrost zone (Allard and K-Seguin 1987, Thibault and Payette 2009). Kuujjuarapik (K1P and K2) represents the subarctic, forested tundra ecoregion at the southernmost limit of the discontinuous permafrost zone (figure 1). Palsa mounds are characteristic at K1P peatland, whereas K2 is a small fen with peat thickness of 1–2 m. Radisson (LG2 and Rad) represents the northernmost boreal ecoregion (figure 1). These two peatlands, in region of sporadic permafrost, are ombrotrophic with peat thickness of 3−4 m (Thibault and Payette 2009).

Zoom In Zoom Out Reset image size

In the two study regions, mean annual temperatures are below 0 °C (table 1). Mean annual temperatures have increased in both regions (table 1). In Kuujjuarapik mean summer temperature has increased 1.0 °C since 1961, while mean winter temperature has increased 0.5 °C since 1971 (Environment Canada 2018). In Radisson, since 1971, the biggest seasonal increase of 0.5 °C is in mean autumn temperatures. In Kuujjuarapik, the increase in growing degree-days (GDD0) (between 1971–2000 and 1981–2010) is ca. 6% and in Radisson ca. 4% (table 1). In both regions, autumn rainfall especially has increased. 35%–40% of the precipitation falls as snow.

In the James Bay lowlands, peatlands cover ca. 20%–30% of the land area (Arseneault and Sirois 2004). In Radisson, peat started to accumulate following post glacial land uplift after 8000 calibrated (cal.) yr. before present (BP) (Dyke and Prest 1987, Beaulieu-Audy et al 2009). Radisson study sites are characterised by hummock-hollow microforms with e.g. Sphagnum lindbergii and Warnstorfia exannulata group abundantly present in low lawns and hollows and S. fuscum on hummocks while Picea mariana is sparsely present (table 1). Along the Hudson Bay lowlands, at the K1P, peat started to accumulate over marine sediments around 5950–5100 cal. BP and palsa formation is connected to the LIA cooling (Arlen-Pouliot and Bhiry 2005; Lamarre et al 2012). Ombrotrophic habitat conditions prevail in the central palsa mound part and hummock vegetation in K1P and K2 is largely similar to Radisson peatlands (table 1) but supplemented by e.g. Polytrichum strictum, and Sphagnum russowii. Sedges dominate the fen habitats together with Myrica gale, Menyanthes trifoliata, Eriophorum russeolum, S. papillosum, S. riparium, and brown mosses such as Scorpidium scorpioides. Larix laricina is occasionally present.

In early July 2017, twelve peat monoliths: three from each four peatlands, K1P, K2, LG2, and Rad, were collected (table 1, figure 1). The core-lengths varied between 32 and 39 cm (table 2). Peat sections (10 cm in diameter) were sampled by hand sawing the layer overlying seasonal frost or permafrost from intermediate lawn microforms (water tables from 13 to 23 cm) inhabited mainly by Sphagnum fuscum. Such microhabitats are considered the most sensitive to reflect climate-induced changes in hydrology and consequent plant community shifts (e.g. Väliranta et al 2007, Frolking et al 2014). The cores were collected from relatively central parts of the peatlands, some tens of meters apart from each other. Peat monoliths were wrapped in plastic film and aluminium foil, transported and frozen at the University of Helsinki, Finland. For the analyses, monoliths were cut into 1 cm slices. To avoid contamination the outmost ca. two centimetres of peat was removed. The subsamples were stored in a cold room at 6 °C in plastic bags.

Plant macrofossil analysis was performed at 2 cm resolution. Where prominent changes in plant composition occurred, the resolution was increased to contiguous centimetres. Volumetric samples of 5 cm³ were carefully rinsed under running water using a 140 μm sieve. From the residue, proportions of the main peat components (such as Sphagnum, non-Sphagnum mosses, Cyperaceae remains, wood, roots, and leaves) were estimated as percentages of a total sample volume under a stereomicroscope and a high-power light microscope. Seeds, leaves and charred remains were counted as exact numbers. Plant macrofossil analysis followed Mauquoy and van Geel (2007), modified by Väliranta et al (2007). If plant remains were unidentifiable to plant type level, proportion of unidentified organic matter (UOM) was estimated. Identification followed e.g. Laine et al (2009), Mauquoy and van Geel (2007), and Eurola et al (1992). A reference collection was also available. Software C2 (v. 1.7.7) (Juggins 2007) and Tilia 2.0.41 were used to produce the diagrams.

Radiocarbon (¹⁴C) accelerator mass spectrometry (AMS) and lead (²¹⁰Pb) dating methods were combined to establish accurate chronologies and build age-depth models (van Der Plicht 2004, Ali et al 2008). In total 16 samples were sent to Poznan Radiocarbon Laboratory (Poznan, Poland) and three samples to the Finnish Museum of Natural History (LUOMUS, Helsinki, Finland) for ¹⁴C dating (table S1 is available online at stacks.iop.org/ERL/14/075002/mmedia). Bulk peat samples, cleaned of roots and rootlets when possible (Holmquist et al 2016), were used for ¹⁴C dating. Samples from the base of each peat monolith, layers representing vegetation changes and/or intermediate layers (LG2.2 and K2.3) were selected for dating.

To create accurate chronologies for recent decades, ²¹⁰Pb dating was performed at the University of Exeter, UK, using alpha-spectrometry. ²¹⁰Pb was measured at 2 cm intervals for each full-length monolith. Freeze-dried subsamples of 0.14−0.51 g were analysed for ²¹⁰Pb activity after spiking them with a ²⁰⁹Po yield tracer, following the methods described by Kelly et al (2017) and Estop-Aragonés et al (2018). ²¹⁰Pb ages were obtained using the Constant Rate of Supply model (CRS) (Appleby and Oldfield 1978). Age-depth models for each core (figure 2) were created with BACON v2.3.3 package in R version 3.4.3 (R Development Core Team 2016). BACON divides the dated cores into sections and applies Bayesian statistics with prior information to reconstruct the accumulation, providing weighted mean ages (Blaauw 2010, Blaauw and Christen 2011) that are further used for calculations without chronological error ranges. ¹⁴C ages were internally calibrated using the INTCAL 13 calibration curve (Reimer 2013) and modern dates (pMC % modern carbon) were converted to radiocarbon ages applying NH Zone 1 post bomb curve (Hua et al 2013).

Zoom In Zoom Out Reset image size

For peat and carbon accumulation calculations, dry bulk density (g cm⁻³) was measured contiguously for every cm after freeze-drying volumetric subsamples of 5 cm³ and by dividing the dry mass (g) by the peat fresh volume (cm³). Carbon and nitrogen (C/N) content measurements were performed at 4 cm intervals, at the University of Helsinki, using a LECO TruSpec micro Elemental Determinator and these results were applied to calculate average carbon values.

To estimate the temporal variations in apparent carbon accumulation rates (ACAR, g C m⁻² yr⁻¹), the carbon mass of every 1 cm increment (g m⁻³) was multiplied by the corresponding vertical peat accumulation rate (m yr⁻¹) (Turunen et al 2002) based on the age-depth models. Average, non-cumulative, apparent recent rates of carbon accumulation (RERCA) (see Turunen 2003) were calculated for three periods; post AD 1950 until present, post AD 1900 until present, and post AD 1850 until present, following a procedure introduced in Lamarre et al (2012). These periods mainly represent incompletely decomposed (acrotelm) peat and are not straight forwardly comparable with results yielded from older highly decomposed, water saturated, and anoxic (catotelm) peat.

Age-depth models show that peat accumulation has not been constant over the accumulation history and ages at the base of the monoliths vary from ca. 2500 cal. BP (K1P2) to ca. cal. AD 1960 (K2.3 and LG2.2) (figure 2). However, in this study, we focus on the period of the last ca. 200 years. Mean peat accumulation rates range from 1.6 to 5.4 mm yr⁻¹ (table 2). On average, accumulation rates increase towards the present, starting from the 1950s (figures 3–6). Within both regions, peatland specific variation in accumulation rates exists (table 2). The K1P2 age-depth model suggests a slowdown in peat accumulation between ca. 1800 cal. BP and 95 cal. BP (19–15 cm) when the peat accumulation rate is only 0.02 mm yr⁻¹. Alternatively either of the ¹⁴C ages is an outlier. ²¹⁰Pb activity ceases already at 13 cm below the surface in K1P2 (figure 2). For cores K2.3 and LG2.2 (both 33 cm in length), the ²¹⁰Pb analyses did not reach the zero activity level (table S2). In order to establish age-depth models for these two cores, we ¹⁴C-dated additional peat samples (table S1 and figure 2). Additionally, the ¹⁴C basal-age of the record LG2.3 was younger than the corresponding age suggested by the ²¹⁰Pb chronology (table S2) and the BACON age-depth model therefore excluded the ¹⁴C age as an outlier (figure 2). A charcoal layer with charred plant remains in LG2.2 was dated to ca. cal. AD 1940 (21 cm) which corresponds to a previously reported fire in AD 1941 (SOPFEU 2004) and supports the reliability of the chronology (figures 2 and 5).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The average core-specific bulk density (BD) with standard deviation (±SD) varies between 0.05 ± 0.01 g cm⁻³ (LG2.3) and 0.11 ± 0.04 g cm⁻³ (K1P2) with an increasing trend towards older peat layers (figures 3–6). The mean BD of all the cores was 0.07 ± 0.03 g cm⁻³. Carbon content analysis yielded an average of 45.5% ± 1.7%. Carbon averages calculated over an individual peat core ranged from 43.9% ± 1.4% (K2.3) to 47.1% ± 2.7% (K1P2). C/N ratio varied between 14.3 and 62.8 with an average of 36.4 ± 11.9. The overall trend in C/N values increased towards the surface (figures 3–6).

Kuujjuarapik (K1P)

Kuujjuarapik (K2)

Radisson (LG2)

Radisson (Rad)

ACAR and RERCA variability

Because the ages at the base of the monoliths vary, the oldest dated to 2510 cal. BP (K1P2) and youngest to −10 cal. BP (cal. AD 1960, K2.3 and LG2.2), we focused on three defined RERCA periods: from AD 1850, from AD 1900, and from AD 1950 to the present-day (table 2). This approach enables us to identify and compare possible changes in recent carbon accumulation rates. In general, the post AD 1950 RERCA rates are the highest (table 2, figures 3–6), partly due to incomplete decomposition. At K2, RERCA values were consistently high (table 2). The lowest post AD 1950 RERCA rates were detected for the Radisson region peatlands: LG2.1: 65.3 g C m⁻² yr⁻¹ and Rad3: 50.7 g C m⁻² yr⁻¹. However, for these same peatlands, much higher rates were also derived: LG2.2: 136.9 g C m⁻² yr⁻¹ and Rad1: 103.3 g C m⁻² yr⁻¹.

ACAR variability is high (figure 7) and mainly varies with depth, with higher rates towards the peat surfaces (figures 3–6). RERCA ranges and exact values for the three focus periods are higher for Kuujjuarapik than for Radisson peatlands (e.g. AD 1950 RERCAs are 24% higher in Kuujjuarapik region than in Radisson) (figure 7). The site-combined averages for the three periods were consistently higher in Kuujjuarapik than in Radisson, but also the lowest apparent carbon accumulation rates were calculated for Kuujjuarapik (figure 7).

Zoom In Zoom Out Reset image size

Our BD values corresponded to previous studies from Northern Ontario where BD ranged from 0.0034 to 0.62 g cm⁻³ with an average of 0.093 ± 0.041 g cm⁻³ (Holmquist et al 2014). Our average carbon content was slightly lower than the 50% which is often used for peat carbon content calculations for boreal and subarctic to Arctic regions (Turunen et al 2002, Treat et al 2016), but close to an average reported for northern peatlands 46.8% ± 6.1% (Loisel et al 2014). Our C/N averages are comparable with other studies, but our lowest C/N values were lower than those reported by Treat et al (2016) (C/N 30–62).

Previous results from Kuujjuarapik have shown a notable rise of peat accumulation rates since the beginning of the 20th century, with average RERCAs: 133–147 g C m⁻² yr⁻¹ from cal. AD 1950 to present, 73–81 g C m⁻² yr⁻¹ from cal. AD 1900 to present, and 52–62 g C m⁻² yr⁻¹ from cal. AD 1850 to present (Lamarre et al 2012). The highest individual RERCA values of Lamarre et al (2012) exceeded 160 g C m⁻² yr⁻¹ and were interpreted to reflect the recent increase in annual temperatures.

High peat accumulation rates of 0.5 cm yr⁻¹, with an average of 0.037 cm yr⁻¹, and high RERCA values (form cal. AD 1850) between 52.8 and 114.5 g C m⁻² yr⁻¹ with a mean of 73.6 g C m⁻² yr⁻¹, have been reported for Eastmain, James Bay area south from Radisson (Loisel and Garneau 2010). Moreover, a mean carbon accumulation rate over the past ca. 300 years of 56.4 g C m⁻² yr⁻¹, has been obtained from the same region (van Bellen et al 2011b). Turunen et al (2004) reported a carbon accumulation rate of 73 g C m⁻² yr⁻¹ for eastern Canadian bogs, covering the last 150 years. High carbon accumulation rates of > 150 g C m⁻² yr⁻¹ over the past ca. 200 years in northeastern maritime Québec have been put forward by Magnan and Garneau (2014a). In contrast, Hudson Bay Lowland records from Northern Ontario indicated no rise in carbon accumulation rates for the recent decades, i.e. less than 30 g C m⁻² yr^{- 1} (Bunbury et al 2012). There, the lead activity covered only the topmost 6 cm, while in the current study, where the core lengths were between 31 cm and 39 cm, the zero ²¹⁰Pb activity level was sometimes not reached at all (K2.3 and LG2.2), which creates uncertainty for the chronologies. However, in K1P2, the lead activity covered only the topmost 13 cm, resembling the results of Bunbury et al (2012). These previous studies were conducted in peatlands resembling ours and samples were mainly collected from similar microhabitats. It has been suggested that peat and carbon accumulation rates are fast for Sphagnum-dominated habitats under high effective moisture conditions and slower when sedges dominate in warm and dry conditions (Nichols et al 2014). In the La Grande region, a rise in peat accumulation rates together with a recent change towards Sphagnum-dominated plant communities has been detected (Beaulieu-Audy et al 2009, Pratte et al 2017). In our study, however, we found no correlation between Sphagnum assemblages and carbon accumulation (Pearson correlation: r = 0.09, p = 0.21).

Intermediate peatland microforms, selected for this study, have higher vertical peat and carbon accumulation rates than, for instance, wet hollows or high and dry hummocks (Belyea and Clymo 2001). Sphagnum establishment promotes rapid peat accumulation and hence peatland vegetation community structure strongly influences carbon sequestration (Tuittila et al 2012, Loisel and Yu 2013). However, it is often a challenge to separate internal and site-specific and external, e.g. climate driven, forcing (Loisel and Garneau 2010, van Bellen et al 2011b, Tuittila et al 2007, 2012, Loisel and Yu 2013). Changes in hydrology and vegetation can also be caused by long-term hydroseral succession from fen to bog (ombrotrophication) that is driven by both internal and external forcing (Yu et al 2009, Väliranta et al 2017).

PAR0 and growing season length have been connected to Sphagnum productivity (NPP) (Loisel et al 2012, Gallego-Sala et al 2018). Accordingly, peat accumulation should be positively correlated to GDD0 (Clymo et al 1998, Charman et al 2013, Holmquist et al 2014). In addition, moisture is an important factor controlling carbon accumulation and peatland dynamics (Swindles et al 2015, Gałka et al 2017, Zhang et al 2018a). Both precipitation and GDD0 have increased since the 1960s and 1970s in our study sites (Environment Canada 2018). In the Hudson Bay region, sea ice cover has decreased up to 11.3% ± 2.6% per decade from AD 1971 to AD 2008, which is affecting moisture and heat transfer patterns of the area (Tivy et al 2011) and may also affect peat accumulation processes.

Our aim was to study recent peat and carbon accumulation patterns by collecting peat records from uniform lawn Sphagnum microforms. Because establishment of robust chronologies for young peat sections is challenging (see however e.g. Goslar et al 2005), in this study we combined two dating methods that support each other (e.g. Turetsky et al 2004). Our results indicated high accumulation rates during recent decades. Previous studies focussing on both the last millennium (Charman et al 2013, Garneau et al 2014, Loisel et al 2014) and recent decades (Zhang et al 2018b) showed that relatively high accumulation rates are likely caused by increased carbon inputs rather than by reduced decomposition. We acknowledge the difficulty of interpreting the recent peat accumulation rates due to the incomplete decomposition process and lower compaction of the surface peat and restraints on the chronologies. Despite these inaccuracies, by using specified focus periods, the data nevertheless allow spatio-temporal comparisons and the assessment of data against the previously published data carried out using similar protocols.

Observed and anticipated warmer and wetter climate conditions for the coming decades (Kirtman et al 2013, Environment Canada 2018) should be beneficial for peat accumulation (Payette et al 2004). Following this, a potential increase of carbon storage has been suggested (Holmquist et al 2014, Gallego-Sala et al 2018). However, contrasting views also exist where modelling exercises suggest that in eastern Canada, peatlands will turn into a carbon source due to increased decomposition under warmer climate, despite the potential increase in carbon sequestration (Chaudhary et al 2017). McGuire et al (2018) modelled the vegetation response to climate warming and CO₂ fertilization under different warming scenarios. They predicted that depending on the magnitude of warming, northern permafrost soils could act as a net sink of carbon, but if the warming is more prominent, substantial soil carbon losses could appear after 2100 (McGuire et al 2018). In some of the projections, vegetation was largely responsible for the carbon intake, but the other projections indicated that vegetation carbon intake could not compensate the losses (McGuire et al 2018). To address the question how high-latitude peatland vegetation and carbon dynamics will respond and have responded to recent warming, more data are needed including high-resolution chronologies and multiple cores, as presented in this study. The presented data highlight the importance of studying multiple peat sections from one study site and preferably the approach should be extended to regional-scales as stated by our third research question and the internal variability needs to be take into account when thinking about upscaling and modelling of results (Loisel and Garneau 2010, Lamarre et al 2012, Mathijssen et al 2017, Zhang et al 2018a).