Ecosystem feedbacks from subarctic wetlands : vegetative and atmospheric CO 2 controls on greenhouse gas emissions

Wetland vegetation provide strong controls on greenhouse gas fluxes but impacts of elevated atmospheric carbon dioxide (CO2) levels on greenhouse gas emissions from wetlands are poorly understood. This study aims to investigate if elevated atmospheric CO2 enhance methane (CH4) 15 emissions from subarctic wetlands and to determine if responses are comparable or species specific within the Cyperaceae, an important group of artic wetland plants. To achieve this we carried out a combined field and laboratory investigation to measure of CO2 and CH4 fluxes. The wetland was a CH4 source with comparable fluxes from areas with and without vegetation and across the different sedge communities. In contrast, the net ecosystem exchange of CO2 differed with sedge species. Within the 20 laboratory experiment plants grown at double ambient (800 ppm) CO2, total biomass of Eriophorum vaginatum and Carex brunnescens increased, whereas the total biomass of E. angustifolium and C. acuta decreased, compared to the control (400 ppm CO2). These changes in biomass were associated with corresponding changes in CH4 flux. E. vaginatum and C. brunnescens mesocosms produced more CH4 when grown in 800 ppm atmospheric CO2 when compared to 400 ppm CO2 with E. angustifolium 25 and C. acuta producing less. Additionally, redox potential and carbon substrate availability in the pore water differed among the plant treatments and in response to the elevated CO2 treatment. Together, this suggests species specific controls of CH4 emissions in response to elevated CO2, which facilitate differential plant growth responses and modification of the rhizosphere environments. Our study highlights species composition as an important control of greenhouse gas feedbacks in a CO2 rich 30 future, which need to be considered in models aiming to predict how ecosystems respond to climate change.


Introduction 35
High latitude wetlands are important global carbon (C) stores with approximately half of global soil C in found in the northern circum-polar permafrost region (Tarnocai et al., 2009).These wetlands are under threat from climate warming (IPCC, 2013).Additionally, atmospheric CO 2 concentrations have increased from pre-industrial levels of 280 ppm to close to 400 ppm in 2013 with future atmospheric CO 2 concentrations predicted to increase to between 426 ppm (RCP 2.6) and 936 ppm (RCP 8.5) over 40 the next century (IPCC, 2013).These changes in climate and atmospheric CO 2 concentration have the potential to increase net primary productivity (NPP) and decomposition rate and hence greenhouse gas emissions (Curtis et al., 1989;Valentine et al., 1994).Wetlands contribute around 80 % of the powerful greenhouse gas methane (CH 4 ) production from natural sources and make up a third of overall global emissions (Kirschke et al., 2013).The largest CH 4 atmospheric mixing ratios are found north of 40 o N (Steele et al., 1987), with the distribution of wetlands in the northern hemisphere recognised as a significant contributor to the global CH 4 budget (Moore and Knowles, 1990).
Emissions of CH 4 from natural wetlands are closely related to the temperature and hydrology of the area (Updegraff et al., 2001;Bridgham et al., 2013).In subarctic and arctic regions, these factors are 50 strongly controlled by permafrost; hence future changes permafrost could impact greatly on regional CH 4 emissions (IPCC, 2013;Christensen et al., 2004).For example, waterlogging of previously aerobic soils may increase CH 4 emissions from arctic regions (ACIA, 2005).
Vegetation is a primary control of CH 4 emissions from wetlands (Heilman and Carlton, 2001;Strӧm et 55 al., 2005;Bhullar et al., 2013).This is in part because most of the organic matter stored in arctic peatlands is recalcitrant and unavailable for digestion by anaerobic bacteria (Bridgham et al., 2013).Therefore, input of recent photosynthates in the form of litter or root exudates are an important carbon source for methanogens (Torn and Chapin, 1993;Strӧm et al., 2005).The diffusion of oxygen through plant aerenchyma from the atmosphere into the roots and subsequent leakage into the rhizosphere leads 60 to 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 and root O 2 inputs differs among wetland plant species, potentially creating species specific impacts on CH 4 fluxes (Updegraff et al., 1995;Strӧm et al., 2005).

65
Elevated atmospheric CO 2 can influence CH 4 production through its impacts on plant C assimilation and allocation.For example, increased below ground biomass production in response to elevated CO 2 levels has been found to substantially increase CH 4 emissions from paddy rice fields (van Groenigen et al., 2012).Increases in plant biomass and productivity in a range of wetland species in response to elevated CO 2 have been found to increase CH 4 emissions (Megonigal and Schlesinger, 1997; Kao-70 Kniffin et al., 2011;Wang et al., 2013) while other species have been unresponsive (Angel et al., 2012).The contrasting responses found among wetlands may, in part, be controlled by the plant species composition as raised atmospheric CO 2 influences all aspects of plant activity including growth, photosynthetic rates and root exudate production, processes which vary strongly among species (Lawlor and Mitchell, 1991;Zak et al., 1993;Bellisario et al., 1999).These findings suggest that more detailed 75 understanding is required to tease apart the controls that govern plant mediated impacts on CH 4 emissions in response to elevated CO 2 .With regards to impacts of elevated atmospheric CO 2 concentrations on CH 4 fluxes, increased biomass may increase labile C inputs and hence production of CH 4, but also transport of O 2 to the rhizosphere and CH 4 to the atmosphere (Joabsson et al., 1999;Wolf et al., 2007;Laanbroek, 2010).However, our current understanding of impacts of elevated CO 2 on 80 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).This is an important knowledge gap as arctic wetlands are currently responding strongly to climate 85 warming resulting both in expansions of graminoid-dominated flooded areas (Prater et al., 2007;Åkerman and Johansson, 2008) and dramatic increases in CH 4 emissions (Christensen et al., 2004;Hodgkins et al., 2014).To explore how elevated atmospheric CO 2 impacts subarctic plant species and wetland CH 4 emissions we carried out in situ measurements of CO 2 and CH 4 emissions in a subarctic wetland in northern Sweden, comparing adjacent open water areas to areas vegetated by different Carex Eriophorum vaginatum and Eriophorum angustifolium respectively, to elevated CO 2 and quantified how this affected CO 2 and CH 4 fluxes, plant growth and peat physicochemical properties.This combined approach was used to test the hypothesis that: Elevated atmospheric CO 2 will increase 95 productivity of Carex and Eriophorum species and subsequently stimulate CH 4 emissions due to increased root inputs of dissolved organic carbon providing labile substrates for methanogens.

Site description
The study site is a subarctic wetland 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, over 40% of which occurs during summer, mean annual temperature is 0.7 o C with a summer average of 11 o C 105 (1913( -2000( average, Kohler et al., 2006)).The site is a palsa mire complex made up of two distinct communities of vegetation.The raised, mesic area is dominated by dwarf shrubs (Betula nana, Empetrum nigrum and Vaccinium uliginosum) and these hummocks have a summer active layer depth of 30 ± 0.9 cm.In the flooded areas there are three dominant Cyperaceae species: Carex acuta, Eriophorum angustifolium and Eriophorum vaginatum as well as the less common Carex brunnescens.On average, (n =5) C. acuta grew in locations with an active layer depth of 119 ± 21 cm, E. angustifolium at 122 ± 12 cm and E. vaginatum at 95 ± 21 cm.The water table 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.115 2.2.Experimental design and analysis 2.2.1.Field campaign Fieldwork was undertaken during July 2014.Summer precipitation in 2014 was 130 mm, slightly above 120 average and mean temperature over the sampling period was 13.3C.To establish the direction and extent of wetland-atmosphere carbon fluxes, in situ CH 4 and CO 2 fluxes were measured.Across the site, five plots were established for each of the three dominant Cyperaceae species.At each plot, gas fluxes were recorded using two 21.2 litre (15 cm diameter x 120 cm height) transparent fan-circulated headspace chambers over both individual plants and open water as experimental pairs, located within a 125 distance of one metre of each other.Air samples were taken at intervals of 5, 10, 20 and 30 minutes and stored in evacuated 12 ml exetainers (Labco, Lampeter, UK).Fluxes were measured twice for each of the 30 plots during the sampling week.CH 4 and CO 2 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.Thermal conductivity (TCD) 130 and H 2 flame ionization (FID) detectors were used to measure CO 2 and CH 4 , respectively.CH 4 and CO 2 evolution was examined for linearity.Gas 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.

Growth room experiments 135
Growth room experiments were established using two chambers which had fixed atmospheric CO 2 concentrations of 400 ppm and 800 ppm.Vegetated and non-vegetated peat treatments were split between the chambers.Mesocosms of control peat and peat planted with either C. acuta, C. Two types of head space chamber were used for the gas sampling.A taller chamber (15 cm diameter x 100 cm height, 17.7 litre volume) was used for the mesocosms with C. acuta and E. angustifolium and a 150 smaller chamber (15 cm diameter x 25 cm height, 4.4 litre volume) was used for the shorter E. vaginatum and C. brunnescens.
To define individual plant-mediated methane-controlling mechanisms over the experimental period, gas flux, redox, and plant extension growth measurements were measured fortnightly.These measurements 155 were taken at five time points over a 10 week period between January and April 2015.Gas fluxes were determined using static headspace chambers, taking air samples ca. 2 minutes after the chamber was closed and then again after 20 minutes.Air in the chambers was circulated using small computer fans.The air samples were stored in 12 ml exetainers and analysed for CH 4 and CO 2 using gas chromatography (as above).Redox potential of the soil was measured in three locations in each pot 160 using a redox probe connected to a millivolt pH meter.For plant samples, three leaves of each individual were labelled and extension growth recorded.At the conclusion of the experiment, porewater samples were extracted from each pot using rhizon samplers.From these samples, E4:E6 ratio, which is an indices of the humification capacity of dissolved organic carbon in the solution, was determined using a spectrophotometer (Cecil CE1011 1000 series) at wavelengths of 465 nm and 665 165 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.and the ratio of TOC:TN reflects the lability of carbon in the pore water (Kokfelt et al., 2009).Above and below ground biomass of plant samples as well as soil organic matter was separated, dried at 60 o C for 72 hours then weighed to calculate total above and below ground biomass.

Data analysis
All data analysis was carried out using GenStat (15 th Edition).Plant-related controls of in situ CH 4 and CO 2 fluxes were analysed using linear mixed models, using species and plant/open water factors as the fixed model and block as the random model.For the ex situ experiment, the fixed model included the CO 2 treatment, species treatment and time factors.Individual pots 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, denominator (i.e.residual) degrees of freedom, the P-value indicating significance when < 0.05.When required, data were transformed to meet the 180 normality assumption.Relationships between variables (e.g.CH 4 and CO 2 fluxes, biomass, pore water chemistry) were analysed using linear regression.195

Species
In the ex situ experiment, both above ground biomass (F 1,3 = 3.58, P = 0.064) and the shoot:root ratio (F 1,3 = 18.66,P < 0.001) were significantly or near significantly affected by the CO 2 treatment for the different species ( The elevated atmospheric CO 2 treatment resulted in contrasting effects on CH 4 fluxes among the four plant species treatments over time (CO 2 treatment × species × time; F 12, 12 = 2.65, P = 0.003; Fig. 2 a-d).CH 4 emissions were higher in the 400 ppm treatment for E. angustifolium and C. acuta although the effect was variable over time, with fluxes showing greatest CH 4 releases at the beginning of the study period for C. acuta while releases were greatest at the end of the period for E. angustifolium.In 210 contrast, the C. brunnescens treatment acted as a CH 4 sink at 400 ppm CO 2 , while fluxes of CH 4 were close to zero at 800 ppm CO 2 .Elevating CO 2 concentrations did not alter CH 4 fluxes in the unplanted control treatments.The changes in observed fluxes were similar irrespective of whether the data was expressed as a function of unit area or dry weight of peat in the mesocosms (Fig. 5, supplementary information).

215
Ex situ CO 2 fluxes (Fig. 2 b) were significantly effected by atmospheric CO 2 (F 1, 12 = 5.24, P = 0.026) and varied over time (F 4, 12 = 2.5, P = 0.044).In E. angustifolium, E. vaginatum and C. acuta treatments, CO 2 fluxes tended to be more positive under elevated atmospheric CO 2 when compared to ambient CO 2 conditions, showing increased CO 2 source potential.Fluxes from C. brunnescens 220 demonstrated an opposite response to elevated atmospheric CO 2 , switching from a CO 2 source to a CO 2 sink.Fluxes in the unplanted control treatment were smaller relative to the planted treatments and did not vary significantly between CO 2 treatments.CO 2 fluxes from the C. acuta mesocosms were highest at the beginning of the study period whereas releases from E. angustifolium mesocosms peaked at the end of the experimental period.

230
The elevated CO 2 treatment had a contrasting effect on total dissolved organic carbon (TOC) for the different species (species × treatment interaction (F 3, 3 = 2.82, P = 0.048), Fig. elevated CO 2 differed, 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 There was no significant difference between the CO 2 treatments for these ratios.

245
The in situ CH 4 fluxes (Fig. 1 a) were of similar magnitude to those measured in minerotrophic mires in the Torneträsk area (Ӧquist and Svensson, 2002;Christensen et al., 2004;Koelbener et al., 2010).The net ecosystem exchange of CO 2 (NEE) was most negative (representing net CO 2 uptake from the atmosphere) in the dense stands of the tussock forming E. vaginatum and the taller and more bulky C. acuta (Fig. 1 b).However, CH 4 fluxes did not vary substantially among areas dominated by different 250 plant species, irrespective of the area being vegetated or open water (Fig. 1 a).Since our paired measurements were done relatively close to each other spatially, the lack of difference between open and vegetated areas may be due to similar level of rhizosphere stimulation of CH 4 production (Dorodnikov et al., 2011) over small spatial scales.Additionally, as all sites were flooded, surface CH 4 oxidation is unlikely to cause differential net CH 4 emissions between vegetated areas, where CH 4 can be 255 transported to plant tissues, and unvegetated areas, potentially explaining why we saw no difference in net emissions from plots with and without out plants (Laanbroek, 2010).
The contrasting responses of above ground and total biomass of the four sedge species to elevated CO 2 (Table 1) indicate that wetland plant species differ in their capacity to respond to atmospheric CO 2 260 levels.Species specific biomass responses after two years exposure to elevated CO 2 (ambient + 340 ppm) were also found in temperature salt marsh plant species (Schoenoplectus americanus and Spartina patens) (Langley et al., 2013) as well as for three different 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).Typha species analysed by Sullivan et al., (2010) showed a uniform response with all species 265 responding to the increase in atmospheric CO 2 by increasing below ground biomass.This is in contrast to our study where above ground biomass of E. angustifolium, E. vaginatum and C. brunnescens increased in response to elevated CO 2 and below ground biomass only increased for two of the study species (E.vaginatum and C. brunnescens (Table 1)).The limited below ground biomass responses for two of our study species compares with the findings by Langley et al. (2013) who reported no 270 significant changes in below ground biomass of Schoenoplectus americanus and Spartina patens 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.

275
The switch in NEE found in three (E.angustifolium, E. vaginatum and C. acuta) of the plant treatments (Fig. 2 b) may be caused by reduced photosynthesis rates in the elevated CO 2 treatment, which have been found previously for beech tree saplings grown under elevated CO 2 (Urban et al., 2014).However, this does not match with the greater above ground biomass found for E. angustifolium and E. vaginatum under elevated CO 2 .Furthermore, increased root and/or soil respiration rates, possibly due to the greater root biomass as found in response to elevated CO 2 for E. vaginatum or greater levels of root exudation, (e.g.increased porewater TOC levels in the 800 ppm treatment for E. vaginuatum) may be the cause of the increased CO 2 emissions.The reduction in the CO 2 sink strength in response to elevated CO 2 in the E. angustifolium, E. vaginatum and C. acuta mesocosms contrast with studies suggesting that elevated atmospheric CO 2 will increase the CO 2 sink strength of wetland ecosystems by increasing NPP (King et 285 al., 1997;Megonigal and Schlesinger, 1997;Sullivan et al., 2010).
The pattern in the response of CH 4 production to atmospheric CO 2 mirror those found for biomass production (Fig. 2 a and Table 1).This suggests that the growth response of different species to elevated CO 2 concentration has an impact on how these species influence CH 4 emission.For example, CH 4 290 emissions from Taxodium distichum and Orontium aquaticum mesocosms increased by.65 and 28 % in response to an experimental increase in CO 2 levels from 350 to 700 ppm, reflecting changes in plant growth (Vann and Megonigal, 2003).Furthermore, CH 4 production in mesocosms planted with Typha angustifolia more than doubled, when CO 2 levels were increased from 380 to 700 ppm, due to increased root biomass (Kao-Kniffin et al., 2011).Similar large increases in CH 4 emissions (136 % increase) were 295 reported for Orontium aquaticum when exposed to double ambient CO 2 concentration.However, in this study only photosynthesis, and not plant biomass, increased significantly in response to the CO 2 treatment (Megonigal and Schlesinger, 1997).In contrast, 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).Limited responses to elevated CO 2 by some species may be linked to nutrient limitation, indicating that global 300 change responses of wetland CH 4 emissions may be strongly controlled by the nutrient demands of species and site nutrient status (Mozdzer and Megonigal, 2013).
In our study, the different plant species treatments controlled the amount and quality of substrate found in the pore water, with elevated atmospheric CO 2 influencing TOC concentrations in planted treatments

305
(Fig. 4a), largely reflecting trends in biomass (Table 1).This is in line with findings from temperate salt marshes exposed to elevated CO 2 (Marsh et al., 2005;Keller et al., 2009).The contrasting porewater chemistry with regards to the E4:E6 and TOC:TN ratios (Fig. 4 c and d) highlights the influence of species composition in these wetlands on rhizospheric carbon inputs, likely due to differences in the composition and quality of root exudates, with implications for CH 4 production (King et al., 2002; 310 Ström et al., 2005;Dorodnikov et al., 2011).In addition, it has also been shown that root exudates can stimulate decomposition of more recalcitrant soil organic matter (Basiliko et al., 2012).The length of our experimental period was too short to measure the effect of litter inputs.However, any observed increases in biomass as a result of raised atmospheric CO 2 (namely in C. brunnescens and E. vaginatum, Table 1) is expected to increase labile carbon inputs from litter production which may further stimulate 315 CH 4 production (Curtis et al., 1990).
The presence of vegetation was also found to lower redox potential (Fig. 3), a critical control of CH 4 production (Bridgham et al., 2013), which compares with findings from mesocosms with Phragmites australis grown under ambient elevated CO 2 (+330 ppm CO 2 ) (Mozdzer and Megonigal, 2013).We suggest that the redox-reducing potential of plant roots is due to increased provision of labile substrate for microbial respiration, depleting alternative electron donors in the micropores where CH 4 production takes places (Yavitt and Seidman-Zager, 2006;Laanbroek, 2010).Our findings and those of Mozdzer and Megonigal (2013)  oxidation rate (Fritz et al., 2011), potentially explaining a proportion of the variable responses observed in plant mediated changes in CH 4 emissions due to elevated atmospheric CO 2 .

330
In conclusion, we have demonstrated that elevated atmospheric CO 2 increased total biomass production in E. vaginatum and C. brunnescens but not in E. angustifolium and C. acuta.In parallel to this, elevated CO 2 only increased CH 4 emissions from the E. vaginatum and C. brunnescens treatments.These data suggest a link between increased productivity via CO 2 fertilisation that could drive changes in species composition, which may ultimately lead to an increase in wetland CH 4 emissions.Our results

335
highlight the need for improved mechanistic understanding, at the species level, of wetland plants to elevated CO 2 before assumptions can be made with regards to impacts on elevated CO 2 on greenhouse gas emissions from wetlands.

90and
Eriophorum species to determine variation in field CH 4 emissions.We then established a controlled environment experiment exposing peat mesocosms planted with Carex acuta, Carex brunnescens, Biogeosciences Discuss., doi:10.5194/bg-2016-105,2016 Manuscript under review for journal Biogeosciences Published: 6 April 2016 c Author(s) 2016.CC-BY 3.0 License. 110 Biogeosciences Discuss., doi:10.5194/bg-2016-105,2016 Manuscript under review for journal Biogeosciences Published: 6 April 2016 c Author(s) 2016.CC-BY 3.0 License.brunnescens, E. angustifolium or E. vaginatum were established with peat and plant material collected 140 from the field site.The degree of replication was; n =10 for C. acuta and E. vaginatum, n = 6 for E. augustifolium, and n =5 for C. brunnescens, and unplanted peat 'control' treatments (n=5).Peat samples were collected from submerged areas 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.The conditions used in the growth chambers simulated the subarctic growing season.Day length was 16 145 hours, day/night temperature was 21/15 o C, daytime light levels were 400 µmol m -2 s -1 and day/night humidity was 65/75%. 170
4 a).TOC was highest in the unplanted controls, followed by E. vaginatum and C. acuta.The response of these two species to Biogeosciences Discuss., doi:10.5194/bg-2016-105,2016 Manuscript under review for journal Biogeosciences Published: 6 April 2016 c Author(s) 2016.CC-BY 3.0 License.
235 contrast, total dissolved nitrogen (TN) (Fig. 4 b) was not significantly influenced by treatment or species effects.Pore water in the two species of Carex display the highest E4:E6 ratio (i.e.relatively more low molecular weight compounds) (F 3, 3 = 6.05,P = 0.001, Fig. 4 d) while the TOC:TN ratio was highest in 240 E. vaginatum (F 3, 3 = 7.91, P < 0.001, Fig. 4 c) out of the planted treatments.

Figure 1 .Figure 2 .
Figure 1.In situ paired-plot gas fluxes showing mean a) methane and; b) carbon dioxide fluxes at the field site in Abisko, Northern Sweden with standard error.

Figure 3 .
Figure 3. Mean redox potential with standard error for planted and unplanted treatments across 400 and 800 ppm atmospheric CO 2 treatments.530

Figure 4 .
Figure 4. Means with standard error for a) Total Organic Carbon; b) Nitrogen; c) TOC:TN ratio and; d) E4:E6 ratio in pore water samples from time point 5 in experimental growth period for all Cyperaceae species plus control pots under atmospheric CO 2 conditions of 400 and 800 ppm.
Table1).Specifically, elevated atmospheric CO 2 levels increased above ground biomass in E. angustifolium, E. vaginatum and C. brunnescens but decreased above ground biomass in The 800 ppm CO 2 treatment altered allocation of carbon in E. angustifolium increasing shoot:root ratio to 1.1 ± 0.14 compared to 0.39 ± 0.06 in the ambient CO 2 (400 ppm) treatment.Elevated atmospheric CO 2 increased total biomass production in E. vaginatum and C. brunnescens but did not effect total biomass production in E. angustifolium or C. acuta. 200C.acuta.