Impacts of Elevated Atmospheric CO2 and Plant Species Composition on Methane Emissions from Subarctic Wetlands

Elevated atmospheric CO2 may create greater methane (CH4) emissions from subarctic wetlands. To date such ecosystem feedbacks remain poorly understood, particularly in relation to how different wetland plant species will control such feedbacks. In this study we exposed plant-peat mesocosms planted with four Cyperaceae species to 400 and 800 ppm atmospheric CO2 concentrations and measured plant and peat properties as well as CH4 fluxes. Above ground biomass for plants grown at 800 ppm CO2 increased for E. angustifolium, Eriophorum vaginatum and Carex brunnescens, but the total biomass of C. acuta decreased relative to the ambient CO2 treatment. The plant species and elevated CO2 treatment affected both peat redox potential and pore water chemistry. There was no overall effect of the elevated CO2 on CH4 emissions, however, CH4 emissions were related to above ground biomass and redox potential, both of which were significantly altered by elevated CO2. Our study shows that species composition poses an important control on how wetland communities will respond to elevated CO2 and that plant mediated changes of peat biogeochemical processes, in response to elevated CO2 levels, may affect CH4 emissions from sub-arctic wetlands, but any such responses will differ among species.


Introduction
Northern peatlands store ca. half of global soil carbon (C), much of which is held in permafrost areas (Tarnocai et al. 2009). Arctic and sub-arctic peatlands are responding rapidly to climate warming, threatening their C storage capacity (IPCC 2013). In parallel with rising temperatures, atmospheric CO 2 levels have increased from pre-industrial levels of 280 ppm to 400 ppm with future atmospheric CO 2 concentrations predicted to increase to between 426 ppm (RCP 2.6) and 936 ppm (RCP 8.5) over the next century (IPCC 2013). These changes in climate and atmospheric CO 2 concentration may result in greater Net Primary Productivity (NPP) and decomposition rates which are both controls of greenhouse gas fluxes from ecosystems including wetlands (Curtis et al. 1989;Turetsky et al. 2014).
Wetlands release ca. 80% of methane (CH 4 ) emission from natural sources, equating to a third of overall global emissions (Kirschke et al. 2013) with the largest CH 4 atmospheric concentrations found north of 40 o N (Steele et al. 1987). Indeed, the large area of wetlands at northern high latitudes are recognised as an important component of the global CH 4 budget (Moore and Knowles 1990;Bridgham et al. 2013;Turetsky et al. 2014). Wetland CH 4 emissions are determined by temperature, substrate and hydrology (Updegraff et al. 2001;Bridgham et al. 2013). In subarctic and arctic regions, these factors are strongly controlled by permafrost. Therefore future changes to permafrost are predicted to impact on CH 4 emissions from high latitude regions (Christensen et al. 2004;IPCC 2013). For example, waterlogging of peatland soils as a result of permafrost thaw will likely increase CH 4 emissions from arctic regions both in response to the anoxic conditions per se but also due to vegetation shifts from moss-lichen woody shrub tundra to wet sedge communities (Christensen et al. 2004;ACIA 2005;Bridgham et al. 2013).
Vegetation directly impacts CH 4 emissions from peatlands (Joabsson et al. 1999;Heilman and Carlton 2001;Öquist and Svensson 2002;Ström et al. 2005;Bhullar et al. 2013a, b). The main controls of CH 4 emissions in relation to vegetation are; (i) release of labile organic compounds in to the root zone which increase CH 4 production; (ii) enhancing CH 4 oxidation due to root oxygen emissions in to the peat and; (iii) passive transport of CH 4 through plant tissues (Joabsson et al. 1999;Öquist and Svensson 2002). As most of the organic matter stored in arctic peatlands is recalcitrant and substrates for digestion by anaerobic bacteria are limiting (Bridgham et al. 2013;Sjogersten et al. 2016), input of labile photosynthates in the form of litter or root exudates are an important carbon source for methanogens (Torn and Chapin 1993;Ström et al. 2005). The diffusion of oxygen through aerenchyma from the atmosphere into the roots and leakage into the rhizosphere causes oxidation of CH 4 to CO 2 in the soil, substantially reducing net CH 4 emissions (Fritz et al. 2011). The quality and quantity of plant litter and root exudate as well as root O 2 inputs differs among wetland plant species, potentially creating species specific impacts on CH 4 fluxes (Updegraff et al. 1995;Ström et al. 2005). Plant mediated transport of CH 4 to the atmosphere can represent a major emission pathway but its contribution to net emissions varies among plant species as the rate of transport is strongly dependent on aerenchyma tissue (Kutzbach et al. 2004;Bhullar et al. 2013b). Indeed, while both Eriphorumn and Carex species can emit large amounts of CH 4 through their tissues the proportion of CH 4 emitted through plant tissues varies strongly among species (Bhullar et al. 2013b).
Elevated atmospheric CO 2 can influence wetland CH 4 production through its role in plant C assimilation and allocation. For example, greater root biomass in rice grown under elevated CO 2 concentration has substantially increased CH 4 emissions from paddy rice fields (Van Groenigen et al. 2011). Greater plant biomass and productivity in a range of wetland species in response to elevated CO 2 have resulted in increased CH 4 emissions from some wetland systems (Megonigal and Schlesinger 1997;Kao-Kniffin et al. 2011;Wang et al. 2013) while the growth of other species have not been affected resulting in no change in wetland CH 4 fluxes (Angel et al. 2012). Such contrasting responses may, in part, be controlled by the plant species composition of different wetlands as elevated atmospheric CO 2 concentrations influence plant activity including growth, photosynthetic rates and root exudate production, processes which also vary strongly among species (Lawlor and Mitchell 1991;Zak et al. 1993;Bellisario et al. 1999). These findings suggest that a more detailed understanding of how elevated CO 2 impact different plant species is required in order to tease apart the controls that govern plant mediated impacts on CH 4 emissions in response to elevated CO 2 . It is likely that if elevated atmospheric CO 2 concentrations increase biomass, providing growth is not nutrient limited in the first place (Gordon et al. 2001), this will increase labile C inputs into the peat and potentially production of CH 4. However, greater plant biomass may also increase transport of O 2 to the rhizosphere and CH 4 to the atmosphere (Joabsson et al. 1999;Wolf et al. 2007;Laanbroek 2010). Taken together, the understanding of impacts of elevated CO 2 on arctic wetland CH 4 emissions is limited at both the ecosystem and species level, creating large uncertainties in model predictions of the role of elevated CO 2 on CH 4 feedback mechanisms (Ringeval et al. 2011).
Arctic and sub-arctic peatlands are currently responding rapidly to climate warming, which has resulted in expansions of graminoid-dominated flooded areas (Prater et al. 2007;Åkerman and Johansson 2008). Such changes in the vegetation has lead to large increases in CH 4 emissions (Christensen et al. 2004;Hodgkins et al. 2014). In subarctic and arctic wetlands, graminoids (e.g. Eriophorum sp. and Carex sp.) are known to directly impact CH 4 emissions both by transport of CH 4 from the rhizosphere to the atmosphere (Bhullar et al. 2013b;Turetsky et al. 2014) and by impacting CH 4 production and oxidation in the rhizosphere (Ström et al. 2005;Koelbener et al. 2010;Fritz et al. 2011;Ström et al. 2012). However, these effects are species specific (Christensen et al. 2004;Ström et al. 2005;Prater et al. 2007;Hodgkins et al. 2014). To explore how variation in graminoid species composition and elevated atmospheric CO 2 impacts CH 4 emissions we established a controlled environment experiment exposing peat mesocosms planted with either C. acuta, C. brunnescens, E. vaginatum or E. angustifolium to elevated CO 2 and measured how this affected plant growth, peat physicochemical properties and CH 4 fluxes. This experiment was used to test the hypothesis that: Elevated atmospheric CO 2 will increase productivity of Carex and Eriophorum species and subsequently stimulate CH 4 emissions due to increased root inputs of labile substrates for methanogens, with the largest effect of elevated CO 2 on CH 4 emissions found in peats planted with C. acuta (the highest biomass species).

Site Description
The study site is a subarctic peatland located on the southern edge of Lake Torneträsk in Northern Sweden (68°21′ 30.96" N 18°46' 56.064" E). The mean annual precipitation is 310 mm, more than 40% of this occurs during summer. Mean annual temperature is 0.7°C, with a July average of 11°C (1913( -2000( average, Kohler et al. 2006. The site is a palsa mire complex, a common peatland type in the region. The area is made up of two distinct communities of vegetation (Sjogersten et al. 2016). The raised, mesic area is dominated by dwarf shrubs (Betula nana, Empetrum nigrum and Vaccinium uliginosum). The active layer depth in these hummocks is 30 ± 0.9 cm in summer. In the flooded areas there are three dominant Cyperaceae species: C. acuta, E. angustifolium and E. vaginatum as well as the less common C. brunnescens (Nilsson 1991;Mossberg and Stenberg 2008). On average, (n = 5) C. acuta grew in locations with an active layer depth of 119 ± 21 cm below the peat surface, E. angustifolium at 122 ± 12 cm and E. vaginatum and C. brunnescens at 95 ± 21 cm. The water level depth in the flooded areas varied, averaging +34 ± 7 cm where C. acuta was found, + 30 ± 3 cm for E. angustifolium and + 15 ± 2 cm for E. vaginatum and C. brunnescens, note that positive values means that the peat surface was submerged. The peat pH at 5 cm depth was 4.3 ± 0.04; with conductivity of 66 ± 30.8 μS; extractable PO 4 3− of 3.9 ± 1 μg g −1 and extractable NH 4 + of 0.12 ± 0.02 μg g −1 . Peat pH and conductivity was determined following mixing peat with DI water in a 1:2 ratio by volume and analysis on a dual pH and conductivity analyser. Extractable PO 4 3− and NH 4 + was determined using standard colourimetrically methods following a K 2 SO 4 extraction of 5 g of fresh weight soil.

Growth Room Experiments
Growth room experiments were established using two walk-in growth rooms (Unigrow, UK) which had fixed atmospheric CO 2 concentrations of 400 ppm and 800 ppm. Mesocosms planted with either C. acuta, C. brunnescens, E. angustifolium or E. vaginatum were established with peat and plant material collected from the field site. The degree of replication per treatment was; n = 10 for C. acuta and E. vaginatum, n = 6 for E. augustifolium, and n = 5 for C. brunnescens, this resulted in a total number of 42 mesocosms. Peat samples were collected as several 20 × 20 × 20 cm blocks taken ca. 30 cm below the peat surface from submerged areas free of vegetation with a water table depth of ca. 30 cm. The recovered plant and soil samples were transported, separated and transplanted into separate water-tight one litre pots ensuring good contact between plant roots and the peat. Care was taken that peat did not dry out prior to the experiment started, however, some oxygenation of the peat will have occurred. For C. acuta, C. brunnescens, E. angustifolium individual shoots were transplanted in to the peat while E. vaginatum was planted as small tussocks reflecting the plants growth form in the field. As the starting biomass differed among pots the mesocosms were grouped as pairs according to biomass and then randomly allocated to a CO 2 treatment. The volume of peat in the pots were ca. one litre, water levels were adjusted (using tap water) to 2-3 cm above the peat surface throughout the experiment. This water level is shallower than those in the field but where used to reduce ebullition and bubble formation on the inside edges of the mesocosm which were deemed to be a risk associated with using deeper water levels. The conditions used in the growth chambers was a day length of 16 h, day/night temperature was 21/15°C, reflecting the conditions found at the sites during warm summers to stimulate growth and microbial activity. The relatively high temperatures compared to the average air temperature during the field sampling period of 13.3°C. Daytime light levels were constant at 400 μmol m −2 s −1 and day/night humidity was 65/ 75%. The mesocosms were equilibrated for ca. 1 month before the measurements started to allow the system to settle from the disturbance and any remaining oxygen to be used up so that the experiment took place under anoxic conditions. Two types of head space chamber were used for the gas sampling. A taller chamber (15 cm diameter × 100 cm height, 17.7 l volume) was used for the mesocosms with C. acuta and E. angustifolium and a smaller chamber (15 cm diameter × 25 cm height, 4.4 l volume) was used for the shorter E. vaginatum and C. brunnescens.
To define individual plant-mediated methane-controlling mechanisms over the experimental period, CH 4 flux, redox, and plant extension growth measurements were measured fortnightly during daytime conditions. These measurements were taken at five time points over a 10 week period between January and April 2015. Methane fluxes were determined using static headspace chambers (Denmead 2008) with samples collected over 20 min. Air in the chambers was circulated using small computer fans. The air samples were stored in 12 ml exetainers (Labco, Lampeter, UK). CH 4 concentrations were determined by gas chromatography (GC-2014, Shimadzu UK LTD, Milton Keynes, UK) using a single injection system with a 1 mL sample loop that passed the gas sample using H 2 as carrier, the flow rate was 30 ml min −1 . The oven temperature was 40°C and the column was a molecular sieve, the injector temperature was 80°C. Flame ionization (detector temperature was 250°C) detectors were used to measure CH 4 . The analytical error was ca. 5%. Methane fluxes were calculated using the ideal gas law (e.g. Mangalassery et al. 2014) and were expressed as both per unit area and peat dry weight.
Peat redox potential was measured in three points in each pot using a redox probe (General Purpose ORP Electrode, Van London Phoenix, Randburg, South Africa) connected to a millivolt pH meter. To assess plant growth, three leaves of each individual plant were labelled and extension growth recorded. At the end of the experiment, pore-water samples were extracted from each mesocosm using rhizon samplers (Rhizosphere Research Products, Wageningen, Netherlands). In the soil solution we determined the E4:E6 ratio, which is an indices of the humification capacity of dissolved organic carbon in the solution, using a spectrophotometer (Cecil CE1011 1000 series) set at 465 nm and 665 nm (Worrall et al. 2002). TOC-TN analysis (Shimadzu TOC-V CPH; TNM-1) was used to measure the total dissolved organic carbon (TOC) and total dissolved nitrogen (TN) content of the water with the ratio of TOC:TN reflecting the lability of carbon in the pore water (Kokfelt et al. 2009). Above and below ground biomass of plant samples was separated, roots were picked from the peat using tweezers and then washed. The biomass samples and the remaining peat were dried at 60°C for 72 h then weighed to calculate total above and below ground biomass and peat dry weight.

Data Analysis
Data analysis was carried out using GenStat (15th Edition). Treatment effects on plant extension growth and biomass production, pore water chemistry and CH 4 fluxes was assessed using linear mixed models. In the model we used the CO 2 treatment, species treatment and time as fixed factors while individual mesocosms were used as the random factor. Statistics reported are the F-value, which is the ratio for between group variance and within group variance, numerator (i.e. fixed) degrees of freedom and denominator (i.e. residual) degrees of freedom, the P value indicating significance when <0.05. When required, data were transformed to meet the normality assumption. Linear regression was used to determine relationships between variables (e.g. CH 4 fluxes, biomass, pore water chemistry).

Results
Extension growth differed among species with the greatest extension growth found for C. acuta (species effect: F 3, 53 = 11.32, P < 0.001), the growth rates of the other three species was in the same range (e.g. 24 ± 2, 12 ± 2.8, 13 ± 2.6, 11 ± 2.0 mm week −1 , respectively for C. acuta, C. brunnescens, E. angustifolium, E. vaginatum at the start of the experiment). The elevated CO 2 treatment caused a significant overall reduction in extension growth (CO 2 treatment effect: F 1, 53 = 12.11; P < 0.001; Fig. 1) with a particularly strong negative impact for C. acuta. At the end of the experiment, above ground biomass differed significantly among species with the lowest biomass found for C. brunnescens (F 3,52 = 8.34, P < 0.001; Fig. 2a). The elevated atmospheric CO 2 levels tended to increase above ground biomass in E. angustifolium, E. vaginatum and C. brunnescens but decreased above ground biomass in C. acuta (near significant species × CO 2 treatment interaction: F 1,52 = 3.58, P = 0.064; Fig. 2a). Below ground biomass differed among species ( Fig. 2b; F 3,52 = 8.31, P < 0.001); furthermore the CO 2 treatment affected belowground biomass differently among the four species (species × CO 2 treatment interaction: F 3,52 = 2.58, P = 0.063) as below ground biomass declined in the elevated CO 2 treatment relative to ambient CO 2 for E. angustifolium and C. acuta. As a result of the contrasting above and below ground biomass responses among species to the CO 2 treatment, the response of the shoot:root ratios to elevated CO 2 differed among the species (Fig. 2c; species × CO 2 treatment interaction: F 3,52 = 6.45, P < 0.001). Specifically the shoot:root ratio increased for all species apart from C. brunnescens.
The CO 2 and species treatment affected soil properties. Specifically, redox potential was consistantly lower in the 800 ppm treatment (F 1, 53 = 3.39, P = 0.07, Fig. 4). Redox also differed among the plant species treatments (F 3, 53 = 27.01, P < 0.001) with the two Carex species having the highest redox potentials. The redox potential was negatively related to above ground plant biomass ( Fig. 5a; F 1, 7 = 4.27, P = 0.08). The elevated CO 2 treatment had a contrasting effect on total dissolved organic carbon (TOC) for the different plant species ((species × CO 2 treatment interaction, F 3, 52 = 2.82, P = 0.048), Fig. 6a). Specifically, E. vaginatum and C. acuta differed in their response to elevated CO 2 , with pore water in the 800 ppm treatment exhibiting 0.9 mg L −1 more organic carbon in E. vaginatum but 1.5 mg L −1 less in C. acuta when compared to ambient CO 2 conditions. In contrast, TOC levels for E. angustifolium and C. brunnescens were not affected by the CO 2 treatment.
Total dissolved nitrogen (TN) (Fig. 6b) was not significantly influenced by treatment or species effects. The TOC:TN ratio differed among species and was highest in E. vaginatum (F 3, 52 = 7.91, P < 0.001, Fig. 6c) out of the planted treatments but was not affected by the CO 2 treatment. Pore water in the two species of Carex display the highest E4:E6 ratio (i.e. relatively more low molecular weight compounds) (F 3, 52 = 6.05, P = 0.001, Fig. 6d) but again there was no significant difference between the CO 2 treatments.
Although the CH 4 fluxes were not directly related to plant biomass, redox potential, which was influenced by the above ground biomass, was negatively related to CH 4 fluxes ( Fig. 5b; F 1, 7 = 5.75, P = 0.05). In contrast, the pore water chemistry was not related to the CH 4 fluxes.

Discussion
The increase in above ground biomass for three of the four study species in response to the elevated CO 2 treatment (Fig. 2a) supported the hypothesis, which predicted greater biomass production under high CO 2 conditions. The different responses of above ground biomass of the study species to increased CO 2 concentrations (Fig. 2a) suggest that wetland plant species will respond in contrasting ways to rising CO 2 concentrations. Previous studies support the notion of species specific responses to elevated CO 2 . For example, following two years of experimental CO 2 treatments (ambient +340 ppm) different biomass responses was reported for Schoenoplectus americanus and Spartina patens in a salt marsh ecosystem (Langley et al. 2013). This was also the case for above ground biomass of Typha species (T. angustifolia, T. glauca and T. latifolia) exposed to 350-390 (control) to 550-600 ppm (treatment) CO 2 (Sullivan et al. 2010). The lower extension growth under elevated CO 2 indicates that greater above ground biomass is likely due to increased tillering and number of leaves. However, of the Typha species investigated by Sullivan et al. (2010), all increased their below ground biomass in response to the elevated CO 2 treatment. This differs to our study in which above ground biomass of E. angustifolium, E. vaginatum and C. brunnescens was higher in the elevated CO 2 treatment while below ground biomass only increased for two of the study species (E.vaginatum and C. brunnescens (Fig. 2b)). No below ground responses for Schoenoplectus americanus and Spartina patens was also reported by Langley et al. (2013) after two years exposure to elevated CO 2 . Species specific responses to atmospheric CO 2 are well known, with fundamental differences in stomatal numbers and size being observed (Woodward et al. 2002;Lomax et al. 2014), which can then influence physiology and ultimately impact on biomass.
This study has demonstrated that different plant species control the amount and quality of substrate found in the pore water, with elevated atmospheric CO 2 influencing TOC concentrations in planted treatments (Fig. 6a), largely reflecting trends in biomass (Fig. 2). This correlates with data from temperate salt marshes exposed to elevated CO 2 (Marsh et al. 2005;Keller et al. 2009). In addition, the trend of lower root biomass, lower TOC and TN concentrations and lower CH 4 emissions in the 800 ppm C. acuta treatment suggests a link among these parameters. Indeed, contrasting above/ below ground carbon allocation and quantity and quality of root exudates (e.g. Ström et al. 2005;Koelbener et al. 2010;Ström et al. 2012) may drive some of the species specific responses to elevated CO 2 and the subsequent impact on CH 4 emissions. The contrasting porewater chemistry with regard to the E4:E6 and TOC:TN ratios (Fig. 6 c and d) suggests that species composition alters rhizospheric organic matter inputs, likely due to differences in the concentration and lability of root exudates, with implications for CH 4 fluxes (King et al. 2002;Ström et al. 2005;Dorodnikov et al. 2011). Furthermore, root exudates have been found to enhance degradation of older recalcitrant soil organic matter which may further increase substrate availability for CH 4 production (Basiliko et al. 2012). The study was too short to measure how elevated CO 2 may alter litter chemistry, however, larger biomass as a result of elevated CO 2 (namely in C. brunnescens and E. vaginatum, Fig. 2) is likely to increase inputs from freshly produced litter which may also increase CH 4 production (Curtis et al. 1990).
The reduction in redox potential under the elevated CO 2 treatment (Fig. 4), together with the negative relationship between above ground biomass and soil redox potential (Fig. 5a), suggests that plant-mediated shifts in soil redox  potential in response to elevated CO 2 may impact soil processes. As soil redox potential is a strong driver of CH 4 production and oxidation (Fritz et al. 2011) such changes may impact CH 4 fluxes. Effects of elevated CO 2 on soil redox potential were also demonstrated in mesocosms with Phragmites australis grown under ambient and elevated CO 2 (+330 ppm CO 2 ) . We speculate that the lowering of the redox potential is caused by greater release of organic material into the rhizosphere contributing to microbial respiration and depletion of electron donors (Yavitt and Seidman-Zager 2006;Laanbroek 2010). Our findings and those of Mozdzer and Megonigal (2013) contrast with those of Wolf et al. (2007) who demonstrated higher soil redox potentials in mesocosms planted with Scirpus olneyi due to greater root O 2 inputs reflecting greater root biomass in the elevated CO 2 treatment. The differing impact of plant species on the soil redox conditions, possibly in response to different levels of root exudation among species and/or differences in gas exchange via aerenchyma, suggests that plant species composition is a key control of the redox environment and by extension CH 4 production and oxidation (Bridgham et al. 2013).
In contrast to our hypothesis, which predicted increased CH 4 production in the elevated CO 2 treatment, there was no significant effect of the elevated CO 2 treatment on CH 4 fluxes (Fig. 3b). However, the negative relationship between above ground biomass and redox potential, and redox potential and CH 4 fluxes, demonstrates the important role of the vegetation as a control of redox and CH 4 fluxes (Fig. 5). Our lack of direct responses of CH 4 fluxes to elevated CO 2 contrasts with studies on Taxodium distichum and Orontium aquaticum mesocosms exposed to an experimental increase in CO 2 levels from 350 to 700 ppm (Vann and Megonigal 2003) and mesocosms planted with Typha angustifolia when CO 2 levels were increased from 380 to 700 ppm (Kao-Kniffin et al. 2011), where increased root growth under elevated CO 2 translated in to greater CH 4 emissions. However, limited or no impact of elevated CO 2 on CH 4 emissions was found in two sedge dominated salt marsh communities (Marsh et al. 2005) illustrating that CH 4 responses to elevated CO 2 may vary strongly among species and ecosystems depending on the physiological responses of the plant species to elevated CO 2 . Furthermore, limited responses to elevated CO 2 by some species may be linked to nutrient limitation .
In conclusion, we have demonstrated that elevated atmospheric CO 2 increased above ground biomass production in E. vaginatum, E. angustifolium and C. brunnescens but not in C. acuta. Our study suggests links between plant biomass, soil redox potential and CH4 production but no direct impact of elevated CO 2 on CH 4 emissions. Our study demonstrates the importance for improved mechanistic understanding of how wetland plants species respond to elevated CO 2 before assumptions can be made with regard to impacts on elevated CO 2 on CH 4 emissions from wetlands.