Can boreal peatlands with pools be net sinks for CO2?

Peatland open-water pools, a common feature on temperate to subarctic peatlands, are sources of carbon (C) to the atmosphere but their contribution to the net ecosystem carbon dioxide exchange (NEE-CO2) is poorly known; there is a question as to whether peatlands with pools are smaller sinks of atmospheric C, or even C-neutral, compared to other peatlands. We present growing season NEE-CO2 measurements using the eddy covariance technique in a peatland with pools. We found the maximum photosynthetic uptake and ecosystem respiration rates at 10 °C to be in the lower range of the published data. The lower total vegetation biomass, due to the presence of pools, reduced CO2 uptake during day and the autotrophic component of ecosystem respiration. The low CO2 uptake combined with reduced CO2 loss resulted in the site being a net sink for CO2 of a similar magnitude as other northern peatlands despite the inclusion of pools.


Introduction
Peatland open water pools are autogenic features that form through interaction between the biotic components of the ecosystem. The water bodies are, as opposed to the vegetated portions of peatland sites, net sources of carbon (C) to the atmosphere (23-419 gC m −2 yr −1 ) (Hamilton et al 1994, Waddington and Roulet 2000, Repo et al 2007, McEnroe et al 2009, Pelletier et al 2014. This release of C is due to peat decomposition at their bottom, limited emergent vegetation to uptake CO 2 , and microbial and photo-degradation of dissolved organic carbon (DOC). The published rates of C release from water bodies on peatlands are of the same magnitude, but with an opposite sign, as the published net ecosystem carbon balance (NECB) for peatlands without pools (e.g. from a source of 14 to a sink −101 gC m −2 yr −1 ) (Roulet et al 2007, Nilsson et al 2008, Billett et al 2010, Koehler et al 2011, Olefeldt et al 2012. Peatlands with pools are found from temperate to subarctic regions in both the northern and southern hemispheres (Glaser 1999) and are of varying age (e.g., Foster and Wright 1990, Beilman et al 2009, van Bellen et al 2011, Magnan and Garneau 2014; the long-term C accumulation in the vegetated areas of these peatlands has to exceed the C loss from the pools. However, assuming peatlands with pools have a similar uptake as those without pools could result in a significant overestimation of the C uptake attributed to peatlands. Pools form from differential biomass accumulation and decomposition and their development is influenced by climate, topography, and geographical setting (e.g., Foster and Wright 1990, Belyea and Lancaster 2002, Belyea 2007, Eppinga et al 2009, Morris et al 2013. Pool depth appears to vary from <0.5 to >2 m and width from 1 m to >100 m (e.g., Foster andWright 1990, Karofeld andTõnisson 2012). Despite their wide geographic coverage, there are only a few estimates of the surface area of peatlands covered by pools. In the Hudson Bay Lowlands, pool coverage is >40% in some areas (Roulet et al 1994), >50% in fens in northeastern Quebec, Canada (White 2011), and between 5 and 40% in some of the major peatland types in Russia (Botch et al 1995). Recently there has been an effort to include peatlands (e.g., Wania et al 2009, Kleinen et al 2012, Spahni et al 2012, Wu et al 2012 in models that simulate climate-C connections, but the resolution of these models is far too coarse to include pools. Therefore it is relevant and timely to determine if the C exchange from peatlands with pools is different than that of peatlands without pools to determine if the simple generalized model parameterization might be used for peatlands with pools. Measurements of net ecosystem carbon dioxide exchange (NEE-CO 2 ) using the eddy covariance (EC) method have been made in several peatlands in temperate, boreal and surbarctic regions, covering multiple years of continuous measurements (e.g., Aurela et al 2004, Roulet et al 2007, Sagerfors et al 2008. However, these peatlands have relatively homogeneous surface vegetation (e.g., Lafleur et al 2003, Aurela et al 2009) and no pools, with the possible exception of the measurements from Kaamanen in northern Finland where there are ephemeral pools (Aurela et al 2001(Aurela et al , 2002(Aurela et al , 2004. To our knowledge, no NEE-CO 2 measurements have been reported for peatland with deeper and permanent open water pools. The magnitude of the published annual release of C from open water pools raises the question as to whether the generalized uptake figures for peatlands without pools apply to peatlands with permanent openwater pools. Considering the efforts to integrate peatlands into global climate models, it is important that the C exchange from different peatland types be documented in order to provide guidance on how to parameterize these models (Frolking et al 2009).
Based on the reported net loss of CO 2 from pools (e.g., Waddington and Roulet 2000, Pelletier et al 2014) and the NEE-CO 2 uptake for vegetated peat surfaces (e.g., Lafleur et al 2003, Sagerfors et al 2008), we hypothesize that peatlands with pools are either NEE-CO 2 neutral or a smaller sink for CO 2 during the growing than peatlands without pools. Here we present the results of one growing season (May-October) of NEE-CO 2 measurements in a boreal ombrotrophic peatland with pools and compare these results with those reported in the literature for peatlands without pools.

Study site and methods
We measured the NEE-CO 2 using the EC technique (Baldocchi 2003)  In 2012, the pools were ice-covered from mid-November to the end of April, and the vegetated area was frozen to a depth of ∼0.1 m for four months of the year.
The EC system consisted of a fast response threedimensional sonic anemometer (CSAT-3, Campbell Scientific, Edmonton, Canada), a fine-wire thermocouple (FW05, Campbell Scientific, Edmonton, Canada), and an enclosed CO 2 /H 2 O analyzer (LI-7200, LI-COR, Lincoln, NE). The instruments were mounted on a tripod 2.5 m above the surface of the peatland. The variables used to calculate the flux were recorded and stored on a 4 GB industrial grade USB flash drive using an analyzer control unit (LI-7550, LI-COR, Lincoln, NE) at 10 Hz. Air density fluctuations due to temperature were accounted for using a posteriori correction from a revision of the WPL formulation (Ibrom et al 2007). The 30 min CO 2 fluxes were computed from the 10 Hz data using the Eddy-Pro processing software (V5.1.0, LI-COR, Lincoln, NE). The CO 2 fluxes were derived from the covariance between vertical wind speed and CO 2 mixing ratio (Burba et al 2012). A two-dimensional coordinate rotation was applied. The EC CO 2 data were cleaned for quality flags output by the EddyPro processing software (Mauder and Foken 2004). The CO 2 data showing uptake at night were removed using a photosynthetically active radiation (PAR) threshold of <20 μmol m −2 s −1 (Lafleur et al 2003). Following this step, the CO 2 data were separated into day and night, and data were discarded if deviating more than ±3 standard deviations of the monthly means (Baldocchi et al 1997). The nighttime NEE-CO 2 were plotted (not shown) against friction velocity (u*), and a threshold of 0.1 m s −1 was used to identify insufficient turbulent mixing to assess reliable fluxes (e.g., Lafleur et al 2001); data not meeting the criteria were discarded. The cleaning procedure resulted in 43% of the fluxes being rejected. Due to the complexity of the landscape surrounding the EC tower, no gap filling procedure was applied to the data set for the analysis we present below. The monthly daily average NEE-CO 2 was therefore evaluated by averaging the mean monthly diurnal pattern of NEE-CO 2 presented in figure 3. The monthly NEE max was evaluated by averaging the individual NEE-CO 2 measurements for PAR > 1000 μmol m −2 s −1 . The CO 2 fluxes are presented following the micrometeorological convention where an uptake by the ecosystem is represented by a negative flux, while a loss of CO 2 to the atmosphere is represented by a positive flux.

Results
The 2012 monthly mean air temperatures between May and October were above the 30-year normal  [Environment Canada, data available at http://climate.weatheroffice.gc.ca]. The average monthly temperatures were higher by 1.0-2.4°C with largest differences observed in August. These differences represent 0.9-2.2 times the standard deviation from the normal monthly average temperature. July precipitation was approximately half the normal value while October precipitation was double. Despite the warmer and drier conditions in July, the vegetation at the site showed no sign of desiccation.
The NEE-CO 2 measurements made between May and October 2012 covered the peatland surface between wind directions 180°-240°(36%), 270°-360°( 25%), and 30°-60°(12%) (figure 2). The same wind directions dominated for nighttime ecosystem respiration (ER = NEE-CO 2 for PAR < 20 μmol m −2 s −1 ). The dominant wind directions were also relatively constant between months with the exception of June where the contribution from 30°to 60°was more important (22%). The monthly average diurnal trends in NEE-CO 2 showed CO 2 uptake during the day and CO 2 release at night (figure 3). The ER and NEE max (NEE-CO 2 when PAR > 1000 μmol m −2 s −1 ) varied statistically (p < 0.05) between months over the measurement period ( figure 4). The monthly average ER rate increased from early (May) to mid-growing season (July-August), before decreasing until October (figure 4). The monthly average NEE max increased from early to late growing season, reaching a maximum uptake of −4.1 μmol m −2 s −1 in September ( figure 4). Overall, the monthly mean daily NEE-CO 2 flux showed uptake for all months with a range of −1.02 (SE ± 0.04) to −2.76 (±0.06) g CO 2 m −2 d −1 and was higher in the first half of the growing season (May-July) (figure 4). The mean daily uptake for the entire study period was −1.84 g CO 2 m −2 d −1 . Data were binned by direction to differentiate the signals from sectors with different pool coverage. However, because of the proximity of the Saint Lawrence River and the Gulf of Saint Lawrence, easterly winds generally bring clouds and rainy conditions and lower CO 2 exchange rates are typically measured during such conditions. Therefore, different processes (lower daytime PAR; presence of pools) yield numerically similar fluxes and the analysis of variability in fluxes by wind sector is compromised. Similarly, sorting ER by wind direction resulted in some bins having a very small number of data reducing the ability for statistical analysis.

Discussion
This study is the first to report EC NEE-CO 2 measurements made over a boreal peatland with permanent    Despite the lower maximum photosynthetic uptake and respiration rates, the measured mean daily NEE-CO 2 for June-September (−1.83 g CO 2 m −2 d −1 ) is within the range of published mean daily NEE-CO 2 measured in pool-free peatlands (−1.51 to −5.20 g CO 2 m −2 d −1 ; table 1). This suggests that the lower photosynthetic rates measured at our site were offset by lower loss through ER, making this peatland with pools a net sink for CO 2 for the 2012 growing season, in the same range as that of peatlands without pools. It is unknown how the higher than normal temperatures observed during the measurement period affected the ecosystem level NEE-CO 2 . Pelletier et al (2014) observed a strong positive correlation between pool water temperature and their C fluxes at the same site suggesting that pool C release may have been greater than 'normal' during the 2012 growing season.
Our results refute our hypothesis: the study peatland including its pools is not C-neutral nor a smaller sink for CO 2 during the growing season than what has been observed in other peatlands. This more generally suggests that the presence of pools on a northern peatland does not necessarily reduce the C sink potential. Olefeldt et al (2012) showed that low productivity combined with lower ER led to the NECB of a permafrost peatland having a similar net overall sink to boreal peatlands. For the permafrost peatland, the combined effect of limited vegetation biomass, low ER linked to the presence of permafrost, and extended winter periods still resulted in an average NECB of 56 gC m −2 yr −1 . In our studied peatland, the pools play a similar role in reducing the vegetation biomass therefore reducing both photosynthesis and autotrophic respiration. While the low ER in a permafrostaffected landscape is probably more due to lower soil respiration because of the low temperatures, the effect of the pools on ER is likely experienced through a decrease in the ecosystem autotrophic respiration. Simultaneously, chamber measurements of CO 2 exchange performed over the different microforms found on the studied site showed high CO 2 uptake on Sphagnum hummocks with P mariana (Pelletier et al in review). These high CO 2 uptake rates combined with surface coverage of this microform (figure 1) could represent an explanation as to why the vegetated surface offset the CO 2 loss from the pools (Pelletier et al in review). Winter CO 2 loss from peatlands represents an important part of the annual budget (Aurela et al 2002). Although we did not do winter measurements, the cold season CO 2 loss is likely to be low since the R 10 value is low (figure 5) and cold temperatures persist for more than five months of the year. Even without winter measurements we are confident that the studied peatland is a net sink for CO 2 . Using the NEE-CO 2 data from the Mer Bleue temperate bog (Humphreys et al 2014) in place of the periods January-April and November-December, and assuming that May and October are CO 2 neutral (in reality likely a weak sink), we found our site to be a net annual sink of 48.8 g CO 2 m −2 yr −1 . This estimation is conservative considering that the winter ER is likely greater at Mer Bleue because of the warmer peat temperatures and the absence of pools.

Conclusion
The results from the present study suggest that peatlands with pools can be net sinks for CO 2 at the ecosystem level during the growing season and potentially on an annual basis though we did not test this directly. Although the pools at our site represented net sources of CO 2 to the atmosphere, the reduced ecosystem CO 2 uptake capacity is compensated by the limited CO 2 loss through respiration. This study is the first to present spatially integrated NEE-CO 2 for a peatland with pools; we present data from a single growing season and for one specific site, which is an example of peatland with pools. The representativeness of our site and results will only be determined if our work stimulates others to do the same sort of measurements in other peatlands with pools in similar and different geographical settings. We also recognize the importance of long-term C exchange studies as those have, in some cases, shown significant inter-annual variability in NEE-CO 2 (e.