Derivation of greenhouse gas emission factors for peatlands managed for extraction in the Republic of Ireland and the UK

Introduction Conclusions References


Introduction
Greenhouse gas (GHG) emissions to the atmosphere have increased significantly since pre-industrial times as a direct result of human activities, such as fossil fuel burn- 5 ing, cement production and land use changes (IPCC, 2013). The Intergovernmental Panel on Climate Change (IPCC) have estimated in their Fifth Assessment Report (AR5) that around one third of all anthropogenic emissions of carbon dioxide (CO 2 ) for the period 1750-2011, were caused by land use changes (IPCC, 2013). From 2000From -2009, the Agriculture, Forestry and Other Land-Use (AFOLU) sector accounted for 10 24 % of all global GHG emissions (around 10 Gt CO 2 − eq yr −1 ), with emissions from peatland drainage and burning alone estimated at around 0.9 Gt CO 2 − eq yr −1 , making this the third largest source of emissions in the entire AFOLU sector (IPCC, 2013). Natural (i.e. undrained) peatlands function as long term carbon (C) stores as the sequestration of CO 2 over time is greater than the amount of C that is emitted from the 15 peatland as methane (CH 4 ) and leached in waterborne exports (Roulet et al., 2007;Nilsson et al., 2008;Koehler et al., 2011;Gažovič et al., 2013). Key to this role is the position of the water table, which largely dictates the rate of decomposition within the peatland. When the water table is positioned close to the peat surface, the breakdown and degradation of organic matter typically proceeds very slowly in the absence of 20 oxygen. As a consequence there is an accumulation of peat (and C within) (Dise, 2009).
In the Republic of Ireland (ROI) and the United Kingdom (UK), peat has been extracted for energy use for many centuries (Chapman et al., 2003;Renou et al., 2006). Traditionally, this involved the manual removal of the peat i.e. hand cutting but this has been largely superseded by highly mechanised methods to extract the peat for Introduction  (Howley et al., 2012) and for use in horticulture. A further 0.4 million t year −1 is likely burned for domestic heating (Duffy et al., 2014) and may impact as much as 600 000 ha of peatlands (Wilson et al., 2013b). Although peat extraction areas in the UK have generally declined over the last few decades, approximately 0.8 million t of peat is still extracted each year in England and Scotland (Webb et al.,5 2014), although it is UK Government policy to phase out peat extraction in England by 2030 (Department of Environment Food and Rural Affairs, 2011). Peat extraction areas in Wales are small (482 ha) and have remained unchanged in the 1991-2010 period (Webb et al., 2014). In Northern Ireland, the area of peatland utilised for fuel (mechanical and hand cutting) has declined considerably in the 1990-2008 period, although a slight increase in the areas used for horticulture have been recorded (Tomlinson, 2010). In industrial peatlands, the extraction of peat is facilitated by the installation of drainage ditches at regular (typically 15-30 m) intervals across the peatland. For peat used for horticultural purposes, the more fibrous upper layers (e.g. Sphagnum peat) 15 are extracted and utilised. If the peat is to be used for energy production the more highly decomposed peat is milled, dried in the production fields and removed for immediate use or stockpiled for later requirements. In the latter, peat extraction ceases when either the sub-peat mineral soil is reached, large quantities of fossilised timber are encountered or drainage is no longer practical (Farrell and Doyle, 2003). For peatlands used for the provision of domestic heating, the peat is either removed by a digger from the margins of peatlands, placed in a tractor mounted hopper and extruded onto the surface of the peatland, or the peat is extruded onto the surface of the peatland from openings made in the peat by a chain cutter. Over a period of weeks the peat is dried in situ and removed from the site. The effect of peat extraction on the hydrological 25 functioning is marked by a large fall in the water level either throughout the peatland (industrial) or at the margins of the peatland (domestic). In the latter, significant water level drawdown is also experienced further inward towards the centre of the peatland (Schouten, 2002 The impact of drainage on C cycling in peatlands has been widely documented. In general, a lowering of the water table leads to increased CO 2 emissions Salm et al., 2012;Haddaway et al., 2014) as the aerobic layer is deepened and mineralisation rates are accentuated. Concurrently, CH 4 emissions (with the exception of ditches) may decrease or cease Turetsky et al., 2014), waterborne 5 C exports may increase (Strack et al., 2008;Evans et al., 2015) and there may be a heightened risk of C loss through fire (Turetsky et al., 2015). In the case of peat extraction, C cycling may be further altered by the removal of vegetation (Waddington and Price, 2000), and losses of windblown particulate organic carbon (POC) may be exacerbated from the bare peat surfaces (Lindsay, 2010). 10 Under the United Nations Framework Convention on Climate Change (UNFCCC) and the Kyoto Protocol, "Annex 1" countries (i.e. countries that have committed to targets that limit or reduce emissions) are obligated to prepare annual National Inventory Reports (NIR) and up-to-date annual inventories, detailing GHG emissions and removals from six different sectors. Emissions from peat extraction fields are reported 15 under Land use, Land Use Change and Forestry (LULUCF, Wetlands: category 5.D). The recent IPCC Wetlands Supplement (IPCC, 2014) to the 2006 Good Practice Guidance (GPG) (IPCC, 2006) derived new Tier 1 emission factors (EFs) for drained organic soils and differentiated between on-site emissions (e.g. CO 2 -C on−site ), emissions from fire (L fire ) and off-site losses (i.e. leaching of waterborne C). In the case of peatlands 20 managed for extraction in the temperate climate zone, the CO 2 -C on−site values have increased from the 0.2 (nutrient poor/bogs) and 1.1 (nutrient rich/fens) t CO 2 -C ha −1 yr −1 in the 2006 guidelines (IPCC, 2006) to a single higher EF of 2.8 t CO 2 -C ha −1 yr −1 (covering the entire boreal and temperate regions) in the Wetlands Supplement. On-site burning directly consumes aboveground C stocks (prescribed and wildfire burning) and Introduction from wildfires are addressed and an EF of 362 g CO 2 -C kg −1 dry fuel burned is provided with a proviso that it was derived from a very small dataset. Given the relatively large areas under peat extraction in both the ROI and the UK, a move from Tier 1 to higher reporting levels is desirable, particularly as (a) a wide range in uncertainty is associated with the IPCC Tier 1 values (1.1-5 4.2 CO 2 -C ha −1 yr −1 ), which reflects the disparity in emissions from drained peatlands from different climate zones and nutrient composition, (b) the most recently published annual CO 2 flux estimates (not included in the derivation of IPCC Tier 1 values) also display a very wide amplitude (cf. Järveoja et al., 2012;Mander et al., 2012;Salm et al., 2012;Strack et al., 2014), (c) no data from ROI or UK peatlands were included 10 in the IPCC derivation, which might mean that the Tier 1 value may not be appropriate for these countries, and (d) no distinction is made between industrial or domestic extraction sites, despite large differences in their drainage, vegetation cover and management characteristics. Countries are encouraged to disaggregate emissions at the higher reporting levels on the basis of local climate, land management practices or hy-15 drology for example, with the aim of producing more precise values with reduced associated uncertainty (IPCC, 2014). In addition, previous studies of peatland fire EFs have focused on the boreal peatlands of Alaska (Yokelson et al., 1997) and Canada (Stockwell et al., 2014); and the temperate peatlands of Minnesota (Yokelson et al., 1997) and North Carolina (Stockwell et al., 2014). These studies found that the smouldering 20 combustion of peats associated with low combustion efficiency leads to relatively lower CO 2 emissions (compared with other ecosystems), and much higher carbon monoxide (CO), CH 4 , and other non-CH 4 hydrocarbon emissions. Therefore, it is important to quantify emissions of these gases as they include strong GHGs (e.g. CH 4 ) and reactive gases responsible for tropospheric ozone formation and poor air quality (e.g. CO, 25 ammonia (NH 3 ), hydrogen cyanide, HCN). The objectives of the study are (1) to provide estimates of the annual CO 2 -C exchange (i.e. CO 2 -C on−site ) for 9 peat extraction sites in the ROI and the UK, (2) to derive regional specific CO 2 -C EFs for drained peat extraction areas that would per-

Study sites
The study sites were located at 9 peat extraction areas in the ROI and the UK with a history of either industrial peat (IP) or domestic peat (DP) extraction ( than −20 cm (Couwenberg and Fritze, 2012;Strack et al., 2014). Physico-chemical characteristics of all the sites are detailed in Table 1. At Clara (DP1), Glenlahan (DP2) and Moyarwood (DP3) the peat has been extracted from the margins of the sites for use in domestic heating. In the case of Clara, peat extraction was an ongoing activity at the time of our study despite the designation of the 20 site as a Special Area of Conservation (SAC). DP1 and DP3 are raised bogs and DP2 is a mountain blanket bog. The vegetation component at all the sites is species poor and is composed mainly of ling heather (Calluna vulgaris), cross leaved heather (Erica tetralix) and lichens (Cladonia spp.) A continuous water table level was not observed at DP2, as the relatively shallow peat deposit (∼ 40 cm) over bedrock at that site was 25 prone to drying out at various times throughout the study.

Environmental monitoring
At each site, 3-9 aluminium square collars (60 cm × 60 cm) were inserted to a depth of 30 cm into the peat. At IP6, smaller circular plastic collars were used (15 cm diameter) to facilitate the use of the PP Systems CPY-4 chamber. Weather stations were established at each site (exception IP5) and recorded photosynthetic photon flux den-10 sity (PPFD; µmol m −2 s −1 ) and soil temperature at 5, 10 and 20 cm depths at 10 min intervals ( • C). At DP3, volumetric moisture content (VMC %) was also recorded. Soil loggers (µ logger, Zeta-tec, UK, Hobo External Data Loggers, Onset Computer Corporation, MA, USA or Comark N2012 Diligence Loggers, Norwich, UK) were installed at each site and recorded hourly soil temperatures ( • C) at 5, 10 and 20 cm depths. 15 At sites IP5 and IP6, soil temperature was only measured manually during CO 2 flux measurements. In order to estimate soil temperature at times where data was lacking a regression based approach, between manually recorded T 6 cm and air temperature recorded at 15 min intervals by a logger on the site was used to gap fill the data (r 2 = 88.7 %). Water of peat extraction. However, at the DP sites a vegetation component is present and in order to incorporate the seasonal dynamics of the plants into CO 2 -C exchange models, the leaf area index (LAI) was estimated for each of the collars. This involved accounting for the green photosynthetic area of all vascular plants (leaves and stems) within the collar at monthly intervals. In short, the number of leaves and stems were counted from five subplots (8 cm × 8 cm) within each collar. The size (length, width) of the leaves was measured from sample plants outside the collars. The LAI was then calculated by multiplying the estimated number of leaves by an area estimate of the leaf. Moss and lichen % cover was estimated at the same time. Species-specific model curves were applied to describe the phenological dynamics of the vegetation of each collar, 10 and the models (vascular plants and moss) were summed to produce a plot-specific LAI. For a detailed description of the method see Wilson et al. (2007). At site DP1 only, the vegetation was removed by regular clipping from one third of the collars, in order to provide an estimate of the heterotrophic contribution (R H ) to ecosystem respiration (R eco ).

Field measurements
At sites IP1-5 and DP1-3, R eco was measured with a static polycarbonate chamber (60 cm × 60 cm × 33 cm) equipped with internal fans to ensure mixing of the air and a cooling system to maintain the temperature within the chamber close to the ambient 20 air temperature (for a more detailed description see Alm et al., 2007b intervals over a period of 60-180 using a portable CO 2 analyser (EGM-4; PP Systems, UK). Concurrently, air temperature ( • C) within the chamber and soil temperatures at 5, 10 and 20 cm depths were recorded at each collar (soil temperature probe; ELE International, UK). The WT position relative to the soil surface was manually measured with a water level probe (Eijkelkamp Agrisearch Equipment, the Netherlands). At the 5 DP sites, net ecosystem exchange (NEE) was measured under a range of ambient light levels (PPFD; µmol m −2 s −1 ) prior to R eco measurements. PPFD was recorded from a sensor (PAR-1. PP Systems) located at the top of the chamber. The portable CO 2 analysers were regularly calibrated with a CO 2 standard gas. 10 Flux rates (mg CO 2 -C m −2 h −1 ) were calculated as the linear slope of the CO 2 concentration in the chamber headspace over time, with respect to the chamber volume, collar area and air temperature. A flux was accepted if the coefficient of determination (r 2 ) was at least 0.90. An exception was made in cases where the flux was close to zero (mainly in winter time where soil processes are typically slower) and the r 2 is al- 15 ways low (Alm et al., 2007b). In these cases the flux data were examined graphically and fluxes with obvious non-linearity (due to chamber leakage, fan malfunction etc.) were discarded. The remainder were accepted provided that some of the environmental variables measured at the same time (e.g. soil temperature) were sufficiently low to account for the low flux values (Wilson et al., 2013a). In this study, we follow the 20 sign convention whereby positive values indicate a CO 2 -C flux from the peatland to the atmosphere (source) and negative values indicate a flux from the atmosphere to the peatland (sink). Gross primary production (GPP) was calculated as NEE minus R eco (Alm et al., 2007b).

Modelling
Statistical and physiological response models (Alm et al., 2007b) were constructed and parameterised for each study site. Model evaluation was based on the following criteria, (a) statistically significant model parameters (p < 0.05), (b) lowest possible standard error of the model parameters and (c) highest possible coefficient of determination (adjusted r 2 ) (see Laine et al., 2009). The R eco models, based upon the Arrhenius equation (Lloyd and Taylor, 1994), are non-linear models related to soil temperature. GPP model coefficients were estimated using the Levenberg-Marquardt multiple non-linear regression technique (IBM SPSS Statistics for Windows, Version 21.0. Armonk, NY, USA). During model construction, the relationship between R eco or GPP and a range 10 of independent environmental variables (recorded in conjunction with flux measurements) was tested. Only variables that increased the explanatory power of the model (i.e. improved r 2 values) were included. The models were accepted if the residuals were evenly scattered around zero. 15 The response functions estimated for R eco and GPP were used for the reconstruction of the annual CO 2 -C balance. R eco fluxes were reconstructed for each collar in combination with an hourly time series of (1) T 5 cm , (2) VMC (at DP3) recorded by the data loggers or (3) WT depths linearly interpolated from weekly measurements. The annual CO 2 -C balance (g C m −2 yr −1 ) was calculated for each sample plot by integrating the 20 hourly R eco values over each 12 month period. (Note that the integration periods vary between study sites; Table 1). At the DP sites, GPP was reconstructed in combination with (1) PPFD values recorded by the weather station, (2) plot specific modelled LAI and (3) an hourly time series of T 5 cm (DP1only). At the DP sites, annual NEE was calculated as annual GPP + annual R eco .

Statistical analysis
The CO 2 -C flux data (R eco for the IP sites, and R eco and GPP for the DP sites) had a non-normal distribution, so the non-parametric Kruskal-Wallis (p = 0.05) and Mann-Whitney tests were used to test for differences between sites. Uncertainty in reconstructed annual R eco and NEE was calculated by summing up the maximum and min-5 imum standard errors associated with each of the model parameters (e.g. Drösler, 2005;Elsgaard et al., 2012;Renou-Wilson et al., 2014). Uncertainty in the annual R eco or NEE estimate was calculated following the law of error propagation as the square root of the sum of the squared standard errors of GPP and R eco (IPCC, 2006). 10 Around 5 kg (dry mass) of loose Irish Sphagnum moss peat was used for measuring fire EFs. Subsamples of the peat were taken and placed into a 22 cm × 12 cm × 10 cm open-topped insulated chamber. The chamber was constructed from lightweight Celcon insulation blocks and was used to replicate natural surface combustion conditions, leaving only one surface of the peat exposed to open air thereby reducing heat loss 15 and oxygen exchange from the other surfaces, in accordance with the suggested peat combustion methodology of Rein et al. (2009). Each sample was dried in an oven overnight at 60 • C. In order to produce comparable replicates, the samples for the burning experiment had to be dried to an absolute dry base to increase ignition probability (Frandsen, 1997), and encourage pyrolysis (Rein et al., 2009). Following drying, the 20 chamber and sample were placed in a fume cupboard under controlled air flow conditions and the peat was ignited using a coiled nichrome wire heated to ∼ 600 • C and placed in contact with the surface of the peat. This also best represents natural ignition conditions (e.g. from a surface shrub fire), also in accordance with the methodology of Rein et al. (2009). Once ignited, each 1 kg sample proceeded to burn for ∼ 90 min. 25 The resulting smoke was continuously sampled using a pump and a 90 cm sample line with a funnel held ∼ 12 cm above the smouldering peat. The smoke was sampled into  15 Annual rainfall varied between sites and between years (Fig. 1). The wettest site was DP3 (1390 mm), and the driest was IP6 (746 mm) in the first year of measurements at that site. All multi-year sites displayed inter-annual variation in rainfall with the largest differences observed in IP4 (210 mm difference in annual rainfall between years). Annual rainfall at IP2, IP5, DP1, DP2 and DP3 was above the long-term average in all 20 years. IP1 and IP4 were wetter than the long-term average in one of the years and drier in the other. IP3 and IP6 were drier than the long-term average. The mean annual water table was below −20 cm at all sites in all years (Fig. 1). The deepest mean annual values were at IP1 (−60 cm) and the shallowest at IP3, 4 and 5 (−25 cm). Mean water table position tracked annual rainfall (i.e. higher rainfall resulted in higher water The highest mean annual soil temperature (T 5 cm ) value (12.7 • C) was recorded at IP4 and the lowest at IP5 (6.7 • C) and inter-annual variation was evident in the multiyear sites (Fig. 1). The lowest hourly T 5 cm value (−12.9

On-site carbon dioxide fluxes
At the IP sites, R eco fluxes ranged from 0 to 133 mg CO 2 -C m −2 h −1 and differed sig- where R eco is ecosystem respiration, T REF is reference temperature set at 283.15 K, parameter T is the temperature minimum at which respiration reaches zero, WT is water table depth, VMC is volumetric moisture content, a and b are fitted model parameters.
A strong relationship was observed between GPP and PPFD at the DP sites. It was 10 the sole explaining variable at DP2 where it accounted for 70 % of the variation (Eq. 4). The addition of LAI (Eq. 5) increased the explanatory power of the GPP model at DP3 (59 %) and the addition of LAI and T 5 cm resulted in 62 % of the variation explained at DP1.
where GPP is gross primary productivity, P max is maximum photosynthesis, PPFD is photosynthetic photon flux density, k PPFD is the PPFD value at which GPP reaches half its maximum, LAI is leaf area index, T 5 cm is soil temperature at depth of 5 cm.

Annual CO 2 -C balance
The annual CO 2 -C balance varied both spatially (between sites) and temporally (multi-year sites) (Figs. 4 and 5). In the IP sites, emissions ranged from 93 (IP5) to 304 g CO 2 -C m −2 yr −1 (IP4). Annual emissions varied considerably within the multi-year sites, where coefficient of variation values ranged from 4 (IP1) to 20 % (IP2). As would 5 be expected given the close relationship observed between soil temperature and CO 2 -C fluxes, a noticeable increase in modelled CO 2 -C emissions was observed during the summer months at all sites (Fig. 4), although the rate of the increase varied somewhat in strength between years in the multi-year sites as a function of measured T 5 cm and WT (where applicable). In the DP sites (Fig. 5), annual GPP and R eco were highest piration (R H ) at DP1 were 344 g CO 2 -C m −2 yr −1 , which equates to 49 % of R eco at that site. Applying this proportional value to the other DP sites, we estimate that R H emissions to be 337 and 213 g CO 2 -C m −2 yr −1 at DP2 and DP3 respectively.

Drivers of annual CO 2 -C on site
No relationships were observed between annual CO 2 -C balances (NEE) and nutrient 20 concentrations, water table levels (average, maximum or minimum) or the von Post scale at either the IP or DP (p > 0.05) sites. A strong relationship (r 2 = 0.63) between average soil temperature at 5 cm depth and R eco was very evident across the IP sites (Fig. 6); the highest annual emissions and highest average soil temperatures were associated with IP4 and the lowest at IP5. 25 The variation in NEE between the DP sites appeared to be related to differences in LAI  (Fig. 6), however the number of sites was very small (n = 3) and some caution must be used in this regard.

Emission factors
Using a single mean value for each multi-year site and for its associated uncertainty (IPCC, 2014), an EF was calculated for each land use category. The derived EFs for 5 the IP and DP sites were 1.70 (±0.47) and 1.64 (±0.44) t CO 2 -C ha −1 yr −1 respectively ( Table 2). The 95 % confidence intervals associated with the derived EFs were ±28 and ±26 % for the IP and the DP sites respectively. There was no significant difference in the EF values between the IP and DP sites (p = 0.90). 10 Mean modified combustion efficiency (MCE) and EFs with their SDs for eight trace gas species were calculated from measurements of five Irish sphagnum moss peat samples (Table 3). The peat burned with a mean MCE of 0.837 (±0.019) typical of smouldering combustion. Emissions of CO 2 amounted to 1346 (±31) g CO 2 kg −1 of dry fuel burned or 342 (±8) g CO 2 -C. Other carbonaceous emis-15 sions amounted to 218 g CO kg −1 ; 8.35 g CH 4 kg −1 ; 1.74 g C 2 H 4 kg −1 ; 1.53 g C 2 H 6 kg −1 ; and 0.60 g CH 3 OH kg −1 of dry fuel burned. Emissions of the nitrogenous compounds amounted to 2.21 g HCN kg −1 ; and 0.73 g NH 3 kg −1 .

Discussion
There is a very wide range in reported CO 2 emissions from both active and abandoned 20 peat extraction areas in the literature (Fig. 7). Much of this variation can be attributed to differences in climate, drainage level, peat type, peat extraction methods and the end use of the peat and, as such, provides a useful framework to examine the variations in this study.

Effects of climate
While the study sites in this paper are all located within the temperate zone, considerable variation in CO 2 -C emissions was evident, driven largely by differences in soil 5 temperatures between the sites (Fig. 6). The coldest site in terms of mean soil temperatures and lowest in terms of annual emissions was Muirhead Moss (IP5) in North-Eastern Scotland. Although rainfall and site water table conditions were similar to other sites, for a high proportion (14 %) of the year, soil temperatures at this site remained below 0 • C and are likely to have resulted in a slowdown of extracellular enzymatic dif-10 fusion (Davidson and Janssens, 2006), reduced microbial activity (Fenner et al., 2005) and consequently lower rates of CO 2 production (Basiliko et al., 2007). Indeed, it is likely that our value of 0.93 t CO 2 -C ha −1 yr −1 at this site may be an overestimation given that it was calculated from monthly mean values that were measured during day time hours (highest daytime temperature). As much of the peatlands in Scotland fall 15 within the same temperature regime (Chapman and Thurlow, 1998), CO 2 -C emissions data from a wider range of peat extraction sites in this region might significantly refine our EF derivation. At the other end of the spectrum, the highest emissions and soil temperatures were observed at Turraun (IP4) in the Irish Midlands. Data from this site had been previously 20 reported by Wilson et al. (2007). In this study, we only utilised CO 2 -C flux data from plots where the mean annual water table position was deeper than −20 cm. This resulted in a higher mean value (taken over two years) in this current study. Three of the ROI IP sites are located in the Midlands where more "extremes" in climate are generally experienced (lower winter temperatures, higher summer temperatures) than along 25 the Western coast (IP3). However, during this study, winter temperatures at all the ROI sites seldom decreased below 0 • C (Fig. 3)  received the lowest annual rainfall of all sites in year 1 of the study at that site (Fig. 1). However, mean annual soil temperatures were in the mid-range of the 9 study sites, hourly T 5 cm values were normally distributed (Fig. 3) and CO 2 -C onsite emissions were close to the derived EF value of 1.70 t CO 2 -C ha −1 yr −1 (Table 2), which would confirm that soil temperature rather than water table level is the main driver of emissions in 5 peatlands managed for extraction in this region. The DP sites are all located in the ROI and within a 35 km radius, but considerable variation in annual rainfall is apparent during this study (Fig. 1), with DP3 (the furthest west) receiving the highest rainfall of all sites in the study (on average 34 % more rainfall than the other DP sites). The eastwest rainfall gradient in the ROI is well documented and coincides with a change in 10 peatland types (i.e. raised bogs to Atlantic blanket bogs). This climatic variation is reflected in the annual R eco values, which were similar between DP1 and DP2 but much lower in DP3 (Fig. 5). There is an established relationship between rainfall amount and the moisture content of peat (Price and Schlotzhauer, 1999; Strack and Price, 2009). For the sites located in high rainfall areas, such as DP3, there is likely to be a suppression of aerobic activity within the peat matrix, and as a consequence R eco values may be lower than would be expected for a drained peat soil. Indeed, at some of these sites, occult precipitation (e.g. dew and fog droplets) may also contribute significantly to higher levels of soil moisture (Lindsay et al., 2014). Annual GPP showed a similar trend to annual R eco in these vegetated DP sites. GPP is strongly controlled by the amount 20 of light received by the plants (i.e. PPFD levels and LAI) and the efficiency with which the plants use it. PPFD values (data not shown) and the vegetation communities were broadly similar during the sampling periods, which would seem to indicate that LAI is the driver of both productivity and NEE at these sites (Fig. 6)

Effects of drainage level
While a close relationship between WT position and CO 2 -C emissions has been established in some peatland studies Blodau and Moore, 2003;Blodau et al., 2004), soil temperature proved to be the strongest determinant of CO 2 -C on−site emissions at our sites and this relationship has also been observed by other studies in peat extraction areas (e.g. Shurpali et al., 2008;Mander et al., 2012;Salm et al., 2012). At some of our sites, the addition of water table or VMC to the R eco model did significantly improve model performance. For those sites where water table did not appear to influence R eco dynamics it may be that fluctuations in WT level were missed with the interpolation approach and CO 2 -C flux measurement regimes that we employed here, although these methodologies have been widely used elsewhere (Riutta et al., 2007;Soini et al., 2010;Renou-Wilson et al., 2014). Instead, it is probable that our results reflect the complexity of the relationship between R eco and WT in very dry soils as outlined by Lafleur et al. (2005), where factors such as a stable, low surface soil moisture content, and decreased porosity (i.e. limited oxygen availability) at the depths that the 15 WT is mainly located, ensure that when CO 2 -C fluxes are measured, the WT is deeper than the zone where it has a discernible impact on R eco (Juszczak et al., 2013). As such, the soil temperature regime in these sites may act as a "proxy" for drainage level (i.e. higher soil temperatures are likely to occur in conjunction with deeper water table levels and vice versa) (Mäkiranta et al., 2009).

Peat characteristics
Industrial peat extraction involves the removal of surface vegetation and results in the exposure of decomposed peat at the surface. The level of decomposition in the peat is related to depth and as extraction proceeds, the more highly decomposed peat is exposed. The peat in industrial extraction sites tends to have a lower aerobic CO 2 pro-Introduction duced labile organic matter supply in the absence of plant communities (i.e. priming). In our study, the C content (with the exception of DP2) was similar across all sites (Table 1). Although, Glatzel et al. (2004) noted that CO 2 production was negatively correlated with the von Post scale of decomposition, no correlation with annual CO 2 -C emissions was evident in our study (p > 0.05). Similarly, despite obvious difference 5 in nitrogen content and pH values between IP sites, no relationships with CO 2 fluxes were discerned. However, the residual peat at IP4 is strongly influenced by the close proximity of limestone parent material, as evidenced by high pH values and the lowest C : N ratio (Table 1), and is highly minerotrophic. Given the high CO 2 -C emissions associated with this site, consideration should be given to disaggregation by nutrient type 10 should more data become available in the future.
Organic matter quality has been closely linked to the soil respiration rate, with lower emission rates associated with the poorer quality organic matter found at depth in drained peatlands (Leifeld et al., 2012). The lowest emissions at our sites occurred where the residual peat was either of Cyperaceous (IP3) or Sphagnum/Cyperaceous 15 (IP5) origin. However, while the slow decomposition rate of Sphagnum litter in comparison to other plant litter has been well documented (Verhoeven and Toth, 1995;Bragazza et al., 2007), there is insufficient data from our study sites to determine whether the limited relationship observed here between peat type and CO 2 -C emissions in our study sites is coincidental rather than causal.

Effects of peat extraction methods and peat end use
For peat utilised for horticulture, the more fibrous peat layers nearer the surface are extracted. This may result in the oxidation of more labile organic matter and may account for the very high emissions associated with Canadian peatlands for example (Fig. 7) in comparison to countries where the deeper peat layers are extracted (Mander et al.,25 2012). However, the IP sites in this study are highly decomposed peat and have been abandoned for 30 years or more in some cases (e.g. IP4) and have remained unvegetated. It is possible that CO 2 -C emissions from active extraction areas may be higher 7511 Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version

Interactive Discussion
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | than those derived in this study given that over the summer period the surface of the peat is regularly scarified and aerated. However, Salm et al. (2012) reported higher emissions from abandoned areas in comparison to active areas, although colonisation by vegetation in the former may have accentuated respiration losses. High annual CO 2 -C emissions following abandonment and recolonization have also been reported 5 by Strack and Zuback (2013) and are in close agreement with the R eco values reported here for the DP sites (Fig. 5).
We have estimated the contribution of heterotrophic respiration (R H ) to R eco at 49 %. Although, this is based on measurements at a single site (DP1), it is within the range reported by other studies (Frolking et al., 2002;Moore et al., 2002;Shurpali et al., 2008). 10 The R H values at DP1 (Fig. 5) and DP2 are higher than the R eco values at the IP sites, which would indicate that decomposition of the belowground biomass (following clipping) and subsequent "priming" effects may contribute significantly to CO 2 -C dynamics at vegetated extraction sites. Furthermore, the methods employed to extract the peat at some of the DP sites (the peat is extruded onto the surface of the peatland from narrow 15 openings made in the peat by a chain cutter) has led to the formation of deep fissures (ca. 4 cm wide and > 2 m deep) within the peat that may enhance oxidation throughout the peat profile. Nonetheless, fissures (ca. 10 cm wide and > 1 m deep) formed in the peat during climatically dry years and that were partially filled in during wetter/windier years were also observed at IP5 where the lowest annual emissions were observed.

Fire emission factors
The mean MCE reported here (0.837) is typical of smouldering combustion (Akagi et al., 2011) and comparable with the reported range of MCE in other studies of high latitude peats (Stockwell et al., 2014;Yokelson et al., 1997). Emission factors for CO 2 and CO are also typical of smouldering combustion and similar to those from other 25 peat studies, particularly Yokelson et al. (1997). As found in other studies of peat fire emissions, our measurements confirm that the CH 4 EF for Irish peat is particularly high (8.35 g kg −1 dry fuel burned) when compared with other forms of biomass burn-7512 Introduction ing. Given the high Global Warming Potential, where each gram of emitted CH 4 is equivalent to 34 g of CO 2 (100 year time horizon, IPCC, 2013), the CH 4 emissions from Irish peat fires may account for over 12 % of the CO 2 -equivalent emissions. This result emphasises the importance of understanding the full suite of trace gas emissions from biomass burning, rather than focussing solely on CO 2 and CH 4 emissions. Gen-5 erally the other EFs reported here lie within the range of variability observed by other peat burning studies, with the exception of NH 3 , which is particularly low, possibly as a result of the nitrogen-poor soils that are typical of Irish and UK blanket bogs. Here, we also report the first C 2 H 6 EF for peat (1.53 ± 0.17 g kg −1 dry fuel burned), similar in magnitude to C 2 H 6 emissions from boreal forests (1.77 g kg −1 dry fuel burned) ac-10 cording to Akagi et al. (2011). However, the use of prescribed fire in the UK to burn off old heather growth to encourage new growth (e.g. the muirburn practice) may not impact the underlying peat to any great extent, given that the practice is restricted to the October-April period when soil moisture conditions are highest. Emissions result from the burning of the woody aboveground biomass, and the underlying peat is gen-

Implications for National Inventory reporting
The ROI currently employs the 2006 GPG default value of 0.2 t CO 2 -C ha −1 (nutrient poor) in reporting of all peat extraction areas, and estimated emissions for 2012 (the most recent assessment year) were 9312 t CO 2 -C yr −1 (Table 4). In contrast, the approach in the UK has been to differentiate between peat extracted for fuel and horticul- 25 ture and then applying the default EFs for nutrient rich (1.1 t CO 2 -C ha −1 ) and nutrient poor peat (0.2 t CO 2 -C ha −1 ) respectively. For 2012, CO 2 -C emissions from UK extraction peatlands were estimated at 2118 t CO 2 -C yr −1 (Table 4).
Reported annual emissions are likely to increase considerably if the Tier 1 values in the IPCC Wetlands Supplement are adopted by inventory compilers. We estimate that emissions from peatlands managed for extraction will be approximately 16 and 10 5 times higher for the ROI and UK respectively ( Table 4). The EFs derived in this study for CO 2 -C onsite for both industrial and domestic peatlands (Table 2) are considerably lower than the Tier 1 value of 2.8 t CO 2 -C ha −1 yr −1 provided in the IPCC Wetlands Supplement (2014). Although the EFs derived in this study fall within the lower confidence margin of the Tier 1 range, our new EFs have a marked reduction in associated 10 uncertainty. As the Tier 1 is a generic value based on published literature rather than a targeted measurement programme, it is naturally subject to a certain level of bias, which result when the underlying studies are not representative of management practices, climatic zones, or soil types in a particular region (Ogle et al., 2004), and may lead to either an over-or underestimation of CO 2 -C emissions. As such, a progres-15 sion to higher reporting Tiers, where country specific or indeed regional data can be disaggregated "to develop more precise, locally appropriate emission factors" (IPCC, 2014) is highly appropriate. Given that no significant difference exists between the EFs derived for the IP and DP sites in this study, we propose a single EF for CO 2 -C on−site of 1.68 t CO 2 -C ha −1 yr −1 to be applied to peatlands managed for extraction in the ROI 20 and UK regardless of peat type. This EF value could be further disaggregated by regional climate, domestic peat extraction intensity (based on extraction rates) or by end use of the peat (horticulture or energy) if more data becomes available. For the latter, it would be highly useful to determine quantitatively whether CO 2 -C on−site emissions vary between the less decomposed residual peat utilised for horticulture and the more 25 decomposed residual peat used for energy production. As the EFs derived in this study have come from sites located within the same "climatic" region, we feel that they are more appropriate for the ROI  used in annual NIRs, we estimate that annual emissions would be 9.5 and 6 higher for the ROI and UK respectively, in comparison to the emissions calculated with the 2006 GPG Tier 1 value, and 40 % lower than emissions calculated with the Wetlands Supplement EF. As reported CO 2 -C on−site emissions are henceforth likely to be much higher for any 5 country that moves from the 2006 GPG to the 2013 Wetlands Supplement, some consideration of potential mitigation measures is required. Wetland Drainage and Rewetting is a new elective activity under Article 3.4 of the Kyoto Protocol (second commitment period) and applies to all lands that have been drained since 1990 and to all lands that have been rewetted since 1990. Countries that elect to report under this activity will also be able to claim C benefits from the rewetting of drained peatlands. In theory, this should provide an impetus for the rewetting of high emitting land use categories such as peatlands managed for extraction, particularly as these areas will remain persistent long term emission hotspots in the absence of rewetting actions (Waddington et al., 2002).

Information gaps
Greenhouse gas emissions from peatlands used for extraction are composed of (a) onsite emissions (i.e. from peat extraction areas, ditches and stockpiles) and (b) off-site emissions associated with water borne losses and the use of the peat for energy or horticulture. In this paper, we have focused solely on the on-site CO 2 -C emissions 20 from the peat extraction areas, and GHG emissions from fire. However, C losses from other pathways may also be substantial. Research has shown that GHG emissions from on-site peat stockpiles and ditches are considerable (Alm et al., 2007a and references therein). Currently, emissions data from stockpiles in the temperate zone are not available and the IPCC Wetlands Supplement does not provide a Tier 1 value, 25 and instead encourages countries to move to higher Tiers in terms of reporting (IPCC, 2014). However, countries such as Finland have developed a Tier 2 approach in which EFs (incl. CH 4 and N 2 O) depend on regional weather and in which emissions from 7515 Introduction ditches and stockpiles are taken into account (Alm et al., 2007a;Lapveteläinen et al., 2007). The IPCC Wetlands Supplement provides Tier 1 EFs for CH 4 emissions from both peat extraction areas and from ditches. The value for the latter is particularly high (542 kg CH 4 ha −1 yr −1 , expressed per unit area of ditch surface) and indicates the importance of this pathway in the full GHG balance (Evans et al., 2015). Similarly, 5 N 2 O emissions have been shown to be significant from drained peatlands (Regina et al., 1996), yet despite this, there are only a small number of published studies and more research is critical in order to provide regional specific EFs.
Other pathways may be of equal importance. For example, the loss of POC from bare peat surfaces may be considerable where the surface is exposed and subject to 10 wind or water erosion (Evans et al., 2006;Lindsay, 2010). While some of the windborne POC is likely to be deposited within the extraction field itself, a proportion undoubtedly leaves the peatland, although there are currently few data available to quantify losses from either wind or water erosion, or the extent to which POC is converted to CO 2 (IPCC, 2014). In addition, high losses of DOC from drained peatlands have been re-15 ported (Evans et al., 2015 and references therein). Although a Tier 1 EF value for DOC is provided in the IPCC Wetlands Supplement, disaggregated by climate zone, with the assumption that 90 % of the exported DOC is converted to CO 2 , there is an obvious need to quantify these losses on a regional basis given the high precipitation loads experienced by the ROI and the UK, and associated differences in peat type (Evans 20 et al., 2015). Emissions from burning are not currently reported in either the ROI or UK inventory reports. The EF provided in the IPCC Wetlands Supplement for CO 2 emissions associated with wildfire burning is similar to our value here (Table 3). Furthermore, given the high CH 4 emissions associated with the burning of the peat that we have reported here (Table 3), and taking cognisance of the strong GWP of CH 4 , more 25 research is urgently required to quantify this emission pathway, particularly under field conditions.
The provision of activity data for inventory reporting varies between the ROI and the UK, with the peat extraction industry the source of data in the former (Duffy et al.,

2014)
, and a multi-source approach (Directory of Mines and Quarries point locations with Google Earth imagery, scientific reports/papers) used in the latter (Webb et al., 2014). However, CO 2 emissions from domestic peat extraction in the ROI are not currently reported due to a lack of activity data and could potentially be very high (Wilson et al., 2013b). In the UK, areas under domestic extraction are included in the Grassland 5 category but may be moved as the UK considers changes post-Wetlands Supplement.

Conclusions
Peatlands managed for extraction are a substantial CO 2 -C emissions hotspot at the landscape scale and further contribute to climate change through significant GHG emissions when the peat is burned or utilised in horticulture. This study, which mea-10 sured and modelled emissions from a range of sites across the ROI and the UK, has highlighted the importance of generating robust Tier 2 values for different regions and land-use categories. Given that the IPCC Tier 1 EF was only based on 20 sites (all from Canada/Fenno-Scandia) we suggest that data from another 9 sites significantly expands the global dataset, as well as adding a new region. 15 Acknowledgements. Funding to D. Wilson  Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Glatzel, S., Basiliko, N., and Moore, T.: Carbon dioxide and methane production potentials of peats from natural, harvested and restored sites, eastern Québec, Canada, Wetlands, 24, 261-267, 2004. Griffith, D. W. T.: Synthetic calibration and quantitative analysis of gas-phase FT-IR spectra, Appl. Spectrosc., 50, 59-70, 1996. 5 Haddaway, N. R., Burden, A., Evans, C. D., Healey, J. R., Jones, D. L., Dalrymple, S. E., and Pullin, A. S.: Evaluating effects of land management on greenhouse gas fluxes and carbon balances in boreo-temperate lowland peatland systems, available at: http://www. environmentalevidencejournal.org/content/3/1/5 (last access: 18 May 2015), Environmental Evidence, 3, 30, doi:10.1186/2047-2382-3-5, 2014. Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Lafleur, P. M., Moore, T. R., Roulet, N. T., and Frolking, S.: Ecosystem respiration in a cool temperate bog depends on peat temperature but not water Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Paton-Walsh, C., Smith, T. E. L., Young, E. L., Griffith, D. W. T., and Guérette, É.-