SEARCH

SEARCH BY CITATION

Keywords:

  • CH4 flux;
  • CO2;
  • CO2 starvation;
  • glacial;
  • Last Glacial Maximum (LGM);
  • methane (CH4);
  • peatland;
  • wetland

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Wetlands were the largest source of atmospheric methane (CH4) during the Last Glacial Maximum (LGM), but the sensitivity of this source to exceptionally low atmospheric CO2 concentration ([CO2]) at the time has not been examined experimentally. We tested the hypothesis that LGM atmospheric [CO2] reduced CH4 emissions as a consequence of decreased photosynthate allocation to the rhizosphere.
  • We exposed minerotrophic fen and ombrotrophic bog peatland mesocosms to simulated LGM (c. 200 ppm) or ambient (c. 400 ppm) [CO2] over 21 months (= 8 per treatment) and measured gaseous CH4 flux, pore water dissolved CH4 and volatile fatty acid (VFA; an indicator of plant carbon supply to the rhizosphere) concentrations.
  • Cumulative CH4 flux from fen mesocosms was suppressed by 29% (< 0.05) and rhizosphere pore water [CH4] by c. 50% (< 0.01) in the LGM [CO2], variables that remained unaffected in bog mesocosms. VFA analysis indicated that changes in plant root exudates were not the driving mechanism behind these results.
  • Our data suggest that the LGM [CO2] suppression of wetland CH4 emissions is contingent on trophic status. The heterogeneous response may be attributable to differences in species assemblage that influence the dominant CH4 production pathway, rhizosphere supplemented photosynthesis and CH4 oxidation.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Ice core records indicate that atmospheric CH4 concentration ([CH4]) over the last 800 000 yr has varied from c. 350 ppb during glacial maxima to c. 800 ppb during interglacials (Loulergue et al., 2008). The reasons behind this natural variation are not fully understood, but changes in wetland CH4 emissions (Chappellaz et al., 1993, 1997) and the strength of the tropospheric sink (reaction with the OH radical) as a result of reductions in biogenic volatile organic compound (BVOC) fluxes from forests (Adams et al., 2001; Valdes et al., 2005; Kaplan et al., 2006) are possible major contributing factors.

During the Last Glacial Maximum (LGM), cool global temperatures (Guilderson et al., 1994; Jahn et al., 2005; Affek et al., 2008) and the presence of ice sheets across northern boreal latitudes (Abe-Ouchi et al., 2007) may have combined to limit the area of high-latitude wetlands and lower the rates of biogenic CH4 emissions (e.g. Chappellaz et al., 1993). Globally, however, wetland area at the LGM may have not differed substantially from that found at the present day, because of the expansion of wetlands onto exposed continental shelves (Kaplan et al., 2006) and a migration of boreal wetlands to regions south of the Northern Hemisphere ice sheet (Weber et al., 2010). A reduction in the atmospheric lifetime of CH4 as a result of an elevated atmospheric sink has been suggested as a possible alternative hypothesis to explain the low atmospheric [CH4] during the LGM (Valdes et al., 2005; Kaplan et al., 2006). However, the exact change in OH radical concentration in the atmosphere over glacial–interglacial cycles remains in doubt (Arneth et al., 2007). Proposed reductions in BVOC from terrestrial ecosystems during the LGM, caused by a contraction of global forests, may have been offset by increased BVOC at the leaf scale in response to the low atmospheric [CO2] present at this time (Possell et al., 2005; Wilkinson et al., 2009; Possell & Hewitt, 2010). In addition to this uncertainty, during isoprene oxidation in the atmosphere, OH radical recycling efficiency could potentially be as high as 40–80% (Lelieveld et al., 2008). Uncertainties in sink strength and wetland area during the LGM lead us to consider additional source-driven (wetland) processes to explain low atmospheric [CH4] during glacial periods.

Current approaches to explaining glacial–interglacial changes in CH4 recorded in ice cores are based on either ‘bottom-up’ or ‘top-down’ modelling. Bottom-up approaches simulate global-scale terrestrial carbon cycle processes that influence wetland CH4 emissions (e.g. Cao et al., 1996; Walter et al., 1996; Potter, 1997; Zhuang et al., 2004; Wania et al., 2009; Singarayer et al., 2011), and subsequent effects on atmospheric chemistry (e.g. Valdes et al., 2005; Singarayer et al., 2011). Top-down or inverse models infer the magnitude of wetland CH4 emissions by constraining atmospheric chemistry models with recorded ice core CH4 concentrations (Chappellaz et al., 1997; Brook et al., 2000; Dallenbach et al., 2000). These theoretical and experimental approaches provide insights into the effects of glacial conditions on CH4-relevant terrestrial biogeochemical processes at global and regional scales, such as reductions in net primary production (NPP) under glacial CO2 starvation.

Critically, we are unaware of experimental investigations into the direct effect of low LGM CO2 concentrations on wetland ecosystem carbon cycling processes or CH4 emissions complementing these earlier lines of enquiry. Consequently, the work presented here focuses on this unknown potential interaction. Previous work indicates that [CO2] is an important variable determining CH4 emissions from wetland ecosystems (Dacey et al., 1994; Megonigal & Schlesinger, 1997; Vann & Megonigal, 2003). Exposure of wetland ecosystems and mesocosms to elevated atmospheric [CO2] typically stimulates CH4 fluxes (Hutchin et al., 1995; Megonigal & Schlesinger, 1997; Saarnio & Silvola, 1999; Kang et al., 2001; Ellis et al., 2009). An increase in CH4 emissions results from an increase in NPP and increased allocation of plant photosynthates to the rhizosphere, with root exudates providing an important substrate for CH4 production (Chanton et al., 1995; Megonigal et al., 1999; Vann & Megonigal, 2003; Kim & Kang, 2008). However, during Pleistocene glacials, atmospheric [CO2] was approximately half the modern value (Luthi et al., 2008), and fossil evidence indicates that this imposes severe ‘CO2 starvation’ on the terrestrial biosphere. Trees in North America, for example, operated with a leaf-intercellular [CO2] approaching the CO2 compensation point for C3 photosynthesis (Van de Water et al., 1994; Ward et al., 2005). Under such conditions of carbon starvation, experimental studies (Polley et al., 1993; Kgope et al., 2010) and process-based modelling indicate substantially reduced terrestrial primary productivity and biomass (Beerling & Woodward, 2001; Harrison & Prentice, 2003; Pagani et al., 2009).

From these earlier studies, we hypothesize that CO2 starvation during glacial times exerted an important limiting effect on wetland CH4 fluxes to the atmosphere by decreasing the allocation of plant photosynthate to the rhizosphere, thereby limiting the substrate for CH4 production. To test this hypothesis, we established a replicated mesocosm-scale controlled environment experiment to examine the influence of LGM atmospheric [CO2] on wetland gaseous and dissolved CH4 dynamics, while also measuring pore water concentrations of volatile fatty acids (VFAs). VFAs are labile short-chain carbon molecules (e.g. acetate) that are used in methanogenesis. The study was conducted over two growing seasons on intact monoliths collected from two temperate British wetland ecosystems of contrasting nutrient status: a minerotrophic fen and an ombrotrophic bog. These two ecosystems represent contrasting ends of the peatland gradient (Charman, 2002). Fens receive water and nutrients from outside their boundaries and tend to be more nutrient-rich and alkaline. By contrast, bogs tend to receive all their water and nutrients from the atmosphere and are therefore acid and low in plant nutrients. In comparison to fens, bogs exhibit lower degradation rates (Aerts et al., 1999), lower methanogenic activity (Juottonen et al., 2005) and lower CH4 emissions (Nykänen et al., 1998) as a result of differences in biotic and abiotic factors (Belyea, 1996).

Materials and Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Field site and mesocosm description

Peat mesocosms (110 × 400 mm) were collected in autumn 2006 from a base-rich minerotrophic fen (Cors Goch) on the Isle of Anglesey, Wales, UK (UK grid reference SH 504 817), and from a base-poor ombrotrophic bog (Migneint) located in the Snowdonia National Park, Wales, UK (SH 816 440). The physicochemical characteristics of both sites are detailed in Kang & Freeman (1999), Kang et al. (2005) and Table 1. Cors Goch is a c. 25-ha base-rich (pH c. 5.5–6) wetland that occupies a former lake basin which has been mostly in-filled with calcium carbonate lacustrine sediments and peat deposits. The location (c. 2.5 km distance from the sea) and underlying geology were reflected in the anion and cation composition in the pore water concentrations of mesocosms (Table 1). Cors Goch has a fibrist peat type with a high (94%) organic matter content (Kang et al., 2005) and a dominant vegetation of both rushes and sedges (Fig. 1a). The most common plant species include Carex riparia curtis, Cladium mariscus, Phragmites australis, Juncus subnodulosus, Molinia caerulea and Myrica gale; however, because of the underlying Carboniferous limestone and calcareous groundwater, more alkaline species (e.g. Carex lepidocarpa and Schoenus nigricans) are also present. Sphagnum species (e.g. Sphagnum tenellum and Sphagnum recurvum) can be found in localized patches, usually colonizing sections maintained as firebreaks where vegetation height is restricted. Mesocosms collected from this site consisted of the smaller vascular plants indicative of the site (because of height restriction in the controlled environment laboratory) which included J. subnodulosus and C. lepidocarpa as the dominant vascular plant species and Campylium stellatum, S. tenellum and S. recurvum as the main bryophyte species.

Table 1.   Anion and cation characteristics of sites and mesocosm samples
IonMigneint (bog) ambient mesocosmsMigneint (bog) simulated LGM mesocosmsCors Goch (fen) ambient mesocosmsCors Goch (fen) simulated LGM mesocosms
  1. Mesocosm values (mg l−1) were determined from n = 6–8 in the autumn of year 1 of the experiment. Values in parentheses are ± 1 SE of the mean.

  2. *, < 0.05% uncertainty between ambient and simulated LGM mesocosms.

  3. LGM, Last Glacial Maximum.

F1.19 (0.37)1.15 (0.51)0.38 (0.05)0.57 (0.35)
HCOO1.30 (0.21)1.31 (0.39)1.99 (1.10)7.57 (6.78)
Cl75.3 (8.07)54.6 (3.94)*66.6 (2.97)62.5 (8.2)
NO31.23 (0.24)0.88 (0.14)0.72 (0.22)0.68 (0.24)
PO43−8.04 (5.09)6.40 (2.92)0.77 (0.22)1.18 (0.62)
SO42−2.45 (1.25)3.11 (1.82)1.95 (0.51)3.19 (1.24)
Na+56.9 (5.51)44.6 (4.18)40.2 (2.33)29.5 (6.05)
NH4+9.41 (1.97)5.29 (1.66)3.74 (1.07)3.13 (0.76)
K+38.5 (5.22)21.6 (4.70)*21.5 (5.32)17.1 (6.53)
Mg2+4.95 (1.04)3.14 (1.19)2.15 (0.25)1.81 (0.26)
Ca2+16.7 (2.45)9.67 (2.99)47.5 (4.87)30.8 (5.61)*
image

Figure 1. Photographs showing our sample sites (a) Cors Goch and (b) Migneint (UK), (c) a mesocosm before extraction and (d) multiple headspace chambers used to measure methane (CH4) flux. (a) A photograph taken at the boundary of Cors Goch. The tall common reed (Phragmites australis) can be seen in the background and the distinct red/brown of bog-myrtle (Myrica gale) in the foreground. (b) The contrasting darker heath vegetation (Calluna vulgaris) and the lighter, wetter areas of blanket bog vegetation that were sampled in this study at Migneint. (c) PVC pipe (110 × 400 mm) inserted into an area dominated by Sphagnum tenellum with four shoots of Juncus subnodulosus at Cors Goch before extraction. This photograph demonstrates that our sampling technique caused minimal damage to the surrounding area and no damage to the mesocosm plants. (d) Multiple static chambers capturing CH4 emissions from mesocosms in year 1 of the study for analysis by gas chromatography flame ionizing detector.

Download figure to PowerPoint

Migneint and the surrounding area of Arenig and Dduallt covers c. 20 000 ha, of which c. 50% can be categorized as bog/marsh/water fringed vegetation (Fig. 1b). The site supports a large area of hemist ombrogenous blanket bog (pH c. 3.5–4.5) with high organic matter content (99%) (Kang et al., 2005). It is particularly significant for the extent of Sphagnum-rich Calluna vulgarisEriophorum vaginatum blanket bog. Also present in the area is Erica tetralixSphagnum papillosum blanket bog, with localized patches of the bog-moss Sphagnum magellanicum. Other plant species include Carex magellanica and Carex pauciflora, which is towards the southern limit of its UK distribution. Mesocosms collected from Migneint were dominated by bryophyte species, such as Hypnaceous and Sphagnum mosses (e.g. Hypnum cupressiforme and S. papillosum), with Juncus effusus and E. vaginatum the most common vascular plants. The upland (c. 500 m above sea level) nature of this ombrotrophic bog is reflected in the low nutrient concentrations shown in pore water concentrations of mesocosms (Table 1).

Experimental design

A total of 16 bog and 16 fen mesocosms were collected with intact surface vegetation (Fig. 1c). Mesocosm containers were constructed from opaque PVC pipe segments and sealed base caps which maintained the anaerobic condition of the core throughout the duration of the experiment. Mesocosms were randomly assigned to one of two Snijders Microclima MC1750E (Snijders, Tilburg, the Netherlands) controlled environment units (CEUs) over two growing years (21 months). Light intensity was set to 250 μm m−2 s−1 and relative humidity to 70% during daylight hours (60% at night). This light intensity was typical of those used in other wetland CEU studies (e.g. Blodau & Moore, 2003; Blodau et al., 2004). A temperate climate using monthly temperatures based on national (England and Wales) 30-yr averages (1970–2000) and daylight hours calculated using the longitude and latitude of the local area was created within the CEUs. The water table was maintained to within 2–3 cm of the surface of the mesocosms by frequent (between one and three times per week depending on the temperature) applications of distilled water. Given the necessary limitations to our approach (i.e. a set-up designed to permit isolation of any atmospheric [CO2] effect on CH4 emissions and not the exact recreation of environmental variables encountered in the field), results from this experiment should be only used to provide an indication of the potential effect of CO2 starvation on CH4 emissions from natural wetlands and only cautiously extended to all wetland ecosystems pending further study, as there are different wetland types and assemblages of plants that are not covered in this study.

Modification of atmospheric [CO2]

An auto-regulating CO2 system was designed to maintain atmospheric [CO2] within the treatment CEU at LGM concentrations (i.e. c. 180 ppm) and modern-day [CO2] in the control CEU (i.e. c. 380 ppm). The CO2 regulating system included a purge gas generator (CMC Ltd, Eschborn, Germany) that created zero [CO2] air by using pressure swing adsorption technology to remove CO2 from compressed air. The purge gas generator created a near-continuous source of CO2-free (< 1 ppm) dry (< 0.01 ppm) air by switching between two adsorbent vessels (molecular sieves). The CO2 system created an atmospheric [CO2] in the CEUs that oscillated around the ambient and simulated LGM set points. When the scrubbed air purging the CEUs reduced the atmospheric [CO2] below their designated set points, a short automated injection of CO2-rich air was delivered into the units until the set point was achieved. During the experiment, the ambient [CO2] averaged 406 ± 23 (mean ± SE) and the simulated LGM 196 ± 28 ppm. Given the complexity and intensive nature of our approach, large-scale replication involving more CEUs was not feasible, and therefore we acknowledge this as a limitation. We were, however, able to mitigate against any undesirable block effect and pseudoreplication by rotating mesocosms (and their allocated CO2 exposure) both within chambers (i.e. altered position within a CEU) and between the two CEUs on a monthly basis.

CH4 measurements

CH4 emissions were measured from the mesocosms using headspace chambers (Fig. 1d) constructed from clear perspex pipe (110 × 500 mm). A small needle hole (0.8 mm) through a resealing membrane (Suba Seal, Sigma-Aldrich, St. Louis, MO, USA) enabled pressure change management. Gas samples were taken at the same point during the day cycle (midday) at each sample point throughout the experiment. Sampling frequency was kept to a minimum (bi-monthly on average) to allow characterization of seasonal CH4 flux responses to the treatment while minimizing the intrusion of ambient CO2 on the LGM simulation cabinet. Chamber [CH4] was determined using a gas chromatography flame ionizing detector (Ai Cambridge GC94, Ai Cambridge, Cambridge, UK; Molesieve 5A (Restek, Bellefonte, PA, USA) (40-60 mesh), and Porapak Q (Sigma-Aldrich, St. Louis, MO, USA) (50-80 mesh), columns and cavity ring down laser spectroscopy (Los Gatos Research RMA-200 Fast Methane Analyser, Los Gatos Research, Mountain View, CA, USA). CH4 fluxes were calculated from the linear increase in gas concentration in the chamber with time using a linear regression equation (e.g. Christensen et al., 1995). Total emitted CH4 was calculated using the mean gas flux of two sampling points multiplied by the number of days in the sampling interval (Melling et al., 2005).

One ml of unfiltered pore water was collected at bi-monthly intervals from the bog and fen mesocosms during the second year of the experiment for analysis of dissolved CH4 (DM). Samples were collected using pore water samplers constructed from 1-ml Plastipak syringes filled with glass wool (Plastipak, Becton Dickinson, Franklin Lakes, NJ, USA). These samplers were permanently fixed 10 cm below the surface vegetation in each mesocosm at the beginning of the experiment. The concentration of DM in pore waters was determined by using the headspace concentration determined by gas chromatography, the volume of headspace and water phase, and Henry’s Law. Calculating DM using this approach is common in belowground wetland/peatland experiments (e.g. Blodau et al., 2007).

Volatile fatty acids

A Supelco Visiprep 24 solid phase extraction SPE manifiold and solid-phase extraction tubes (Biotage Isolute Env+ 200 mg (3 ml)−1; Biotage, Uppsala, Sweden) were used to extract VFAs (acetic to octanoic) from pore water samples throughout the experiment. Isolute Env+ columns were conditioned using 4 ml of 0.01 M HCl. Pre-extraction, samples were acidified to c. pH 2 with HCl and 1.862 μg of 2-methylpentanoic acid was added as an internal standard. Two millilitres of sample was eluted through the SPE tubes, followed by 4 ml of 0.01 M HCl, with VFAs then eluted using 2 ml of methanol. After the extraction, anhydrous sodium sulfate was used to remove any residual water. A procedural blank of deionized water was processed with every batch of samples. Gas chromatography–mass spectroscopy (GC-MS) analysis was carried out using an Agilent Technologies 6890 gas chromatograph coupled to a 5973 mass spectrometer (Agilent Technologies, Santa Clara, CA, USA). Separation was performed on a Phenomenex FFAP column (Phenomenex, Torrance, CA, USA) (30 m length, 0.25 mm internal diameter and 0.25 mm film thickness) with a helium (He) carrier gas at a constant column flow rate of 1.1 ml min−1. The GC oven temperature was held for 1 min at 50°C and then ramped to 200°C at a rate of 10°C min−1 and then held for 2 min. The injection was at 190°C with a 10 : 1 split and 1 μl injected. The MS was run in full scan and for quantification in selective ion monitoring (m/z 43, 45, 60, 73, 74 and 87) with a dwell time of 50 ms for each ion. Response factors for the compounds of interest were calculated from six point calibration curves.

Statistical analyses

A general linear model (ANOVA repeated measures) was used to analyse within-subject (time) and between-treatment (CO2 effect) differences and the interaction of the two (treatment × time) during the experiment. Where the assumptions for this test were not met, transformations (common log and square root) and corrections (Greenhouse–Geisser) were applied accordingly. Data that continued to fail to meet the criteria were analysed using a nonparametric Kruskal–Wallis one-way analysis of variance and Friedman’s repeated measures test. Independent t-tests were used when analysing for statistical differences between total cumulative CH4 emitted from ambient and simulated LGM mesocosms. Differences in vascular plant density were accounted for by including this as a covariate in the repeated measures analysis and by expressing the total emitted flux per plant shoot when performing the t-test. Independent t-test and Mann–Whitney U tests were used to analyse between-treatment and between-ecosystem differences when data were segregated into calendar seasons. Analysis was performed using spss statistics version 18 (SPSS, Chicago, IL, USA).

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Mesocosm CH4 fluxes

In year 1, before the start of the CO2 treatment (0–95 d), all bog and fen mesocosms were maintained under winter conditions and in the same [CO2] (426 ± 4.5 ppm; mean ± SE). During this period, CH4 fluxes from ambient and simulated LGM mesocosms were indistinguishable (Table 2; > 0.05). The pretreatment period showed that average fen CH4 fluxes were 51% larger than bog fluxes, as expected from their different trophic status (Nykänen et al., 1998; Juottonen et al., 2005; Hornibrook & Bowes, 2007). Furthermore, in support of our experimental set-up, CH4 emissions at ambient atmospheric [CO2] were of a similar magnitude to those found in other mesocosm studies in nearby locations (Hutchin et al., 1995; Kang et al., 2001) and within the broad range of measured CH4 fluxes from northern latitude wetlands (Dise, 1993; Silvola et al., 2003).

Table 2.   Comparison of simulated Last Glacial Maximum (LGM) (200 ppm CO2) and ambient (400 ppm CO2) CH4 fluxes from bog and fen mesocosms during the experiment (n = 8 per CO2 treatment)
PeriodEcosystemCO2 treatmentMean CH4 flux (± 1 SE) (mg m−2 d−1)Difference between treatment means (%)aTotal CH4 emitted (± 1 SE (g m−2)Difference between total emitted CH4 (%)b
  1. aNo time × treatment interactions were observed.

  2. bStatistical analysis performed on plant-corrected data: mesocosm fluxes divided by the number of vascular plants present.

  3. *, P < 0.05% uncertainty; **, P < 0.01% uncertainty.

Year 1 (pre-CO2 treatment)BogAmbient12.0 (3.51)−210.78 (0.18)−33
Simulated LGM9.53 (3.12)0.52 (0.14)
FenAmbient24.0 (7.07)−181.39 (0.51)−25
Simulated LGM19.7 (6.53)1.04 (0.41)
Year 1 (post-CO2 treatment)BogAmbient23.2 (2.66)−204.64 (0.89)−23
Simulated LGM18.6 (2.21)3.55 (0.50)
FenAmbient30.5 (3.97)−177.14 (1.38)−26
Simulated LGM25.3 (3.10)5.28 (1.06)
Year 2BogAmbient17.6 (2.52)−135.30 (1.70)−15
Simulated LGM15.3 (2.84)4.51 (2.50)
FenAmbient26.3 (3.02)−31**7.61 (2.47)−32**
Simulated LGM18.1 (2.44)5.18 (2.04)
Years 1 + 2BogAmbient20.0 (1.84)−179.94 (2.04)−19
Simulated LGM16.7 (1.88)8.06 (2.26)
FenAmbient28.1 (2.42)−25**14.7 (2.93)−29*
Simulated LGM21.2 (1.94)10.5 (2.89)

After LGM atmospheric [CO2] treatment initiation (day 95), CH4 fluxes from simulated LGM bog and fen mesocosms were immediately reduced relative to ambient [CO2] controls (Fig. 2). This CO2 effect was most evident in bog mesocosms, where during the first month of CO2 manipulation, ambient mesocosms averaged 47 mg CH4 m−2 d−1 compared with 19 mg CH4 m−2 d−1 in the simulated LGM, a difference of c. 60%. Mesocosm CH4 fluxes in year 1 under LGM [CO2] peaked in August (Fig. 2) during the second-highest day temperature and length. In year 1 (during the CO2 treatment) there was no significant difference in CH4 flux between the simulated LGM and ambient [CO2] treated mesocosms (Table 2; > 0.05). There was also no statistically significant difference in total cumulative CH4 emitted between bog and fen simulated LGM treatments vs ambient controls (Table 2; > 0.05).

image

Figure 2. Average methane (CH4) fluxes measured over 2 yr (c. 650 d) from (a) bog and (b) fen mesocosms. Ambient, closed circles; simulated Last Glacial Maximum (LGM), open circles. Each point shows the mean flux of approximately eight peat mesocosms. Error bars represent ± 1 SE of the mean. The shaded area represents the temperature during the daylight simulation.

Download figure to PowerPoint

In year 2, bog ambient and simulated LGM CH4 flux values were similar until the late summer/autumn period, when successive simulated LGM fluxes were consistently lower (c. 68%) when compared with ambient controls (Fig. 2a). However, a repeated measures ANOVA showed that over the entire year there was no significant CO2 effect (Table 2; = 0.05). By the end of the season, bog ambient and simulated LGM mesocosms emitted similar (> 0.05) cumulative CH4 fluxes of 4.51 ± 2.50 and 5.30 ± 1.70 g CH4 m−2, respectively. By contrast, fen mesocosms subjected to LGM atmospheric [CO2] in year 2 emitted significantly less CH4 (total of 5.18 ± 2.04 g CH4 m−2) than their equivalent ambient control mesocosms (7.61 ± 2.47 g CH4 m−2) (Table 1; < 0.01). This statistically significant difference was predominantly driven by a notable period of divergence during the summer (Fig. 2b), where ambient fluxes were on average 38% higher than simulated LGM fluxes.

When considering the entire experiment (years 1 + 2), simulated LGM atmospheric [CO2] significantly suppressed fen CH4 flux by an average of 29% (Table 2; < 0.05). By contrast, CH4 emissions from the bog displayed no signs of CO2 treatment effect. The same pattern was observed in the mean CH4 flux values (Table 2). After 21 months, total CH4 emitted from the experimental groups followed the pattern: bog simulated LGM = bog ambient = fen simulated LGM < fen ambient.

Grouping the raw data (Fig. 2) into seasons (Fig. 3a) shows that CH4 fluxes from the simulated LGM bog mesocosms were only statistically different from ambient fluxes in the spring of year 1 (Fig. 3a, < 0.05). The simulated LGM fen CH4 fluxes were not significantly different from those of their equivalent ambient controls (because of large standard deviations and error) during this same period. Fen simulated LGM treatment CH4 fluxes were significantly lower (P < 0.05) than those of ambient controls in the summer of year 2 (Fig. 3a).

image

Figure 3. Seasonally grouped experimental data for (a) methane (CH4) flux, (b) dissolved [CH4] and (c) acetate concentration. Error bars represent ± 1 SE of the mean. Within each calendar season, bars with different letters are significantly different (t-test and Mann–Whitney U, < 0.05). LGM, Last Glacial Maximum.

Download figure to PowerPoint

Pore water CH4 (DM)

Average bog pore water concentrations of DM in the second year of the experiment ranged from 0.98 mg l−1 measured at the beginning of the year to 4.15 mg l−1 measured at the end. Both the ambient and simulated LGM bog concentrations peaked in autumn, lagging behind the peak in gaseous CH4 emissions measured in the summer (Fig. 3b). The average bog ambient DM concentration in year 2 was 5.23 mg l−1, which was 15% lower than the average bog simulated LGM figure of 6.02 mg l−1 (> 0.05). Grouped bog DM concentrations were not significantly different in any of the four seasons in year 2 (Fig. 3b; > 0.05). Fen mesocosms generally displayed lower concentrations of DM compared with the bog, where concentrations ranged from 0.30 to 3.03 mg l−1 during the year. The ambient and simulated LGM fen groups both showed a summer peak in DM that coincided with the gaseous flux peak. The fen ambient group averaged 3.96 mg l−1 over the year, which was 48% higher than the simulated LGM figure of 2.04 mg l−1 (P < 0.01). Grouped fen ambient (DM) values were significantly different from the fen simulated LGM in the spring, summer and autumn of the year (Fig. 3b; < 0.05). Grouping data also showed that fen simulated LGM mesocosms had significantly lower amounts of DM in their pore waters in every season compared with the other three experimental groups.

Volatile fatty acids

The concentration of VFAs in pore water samples was measured once during each calendar season in the experiment (Table 3). Acetate was the dominant molecule within the C2–C8 category, making up on average 66 and 57% of the total concentration in bog and fen samples, respectively. We will therefore only describe the pattern of acetate concentration during the experiment as it was broadly indicative of the behaviour and presence of all the VFAs measured. All values are blank-corrected from the procedural blanks.

Table 3.   Seasonal carboxylic acid concentrations (mg l−1) within pore water samples (n = 8)
BogYear 1Year 2
Volatile fatty acidSpringSummerAutumnWinterSpringSummerAutumnWinter
AmbientSimulated LGMAmbientSimulated LGMAmbientSimulated LGMAmbientSimulated LGMAmbientSimulated LGMAmbientSimulated LGMAmbientSimulated LGMAmbientSimulated LGM
  1. Carboxylic acid concentrations in mg l−1; values in parentheses, ± 1 SE.

  2. LGM, Last Glacial Maximum.

Acetic31.3 (7.20)18.5 (2.64)2.23 (0.64)3.17 (0.85)1.17 (0.17)2.42 (0.65)0.85 (0.09)1.05 (0.22)6.74 (4.23)3.47 (0.57)2.17 (0.58)1.43 (0.33)2.34 (0.79)1.41 (0.08)2.10 (0.59)1.59 (0.07)
Propanoic8.04 (2.64)4.09 (1.17)0.51 (0.29)0.44 (0.15)0.17 (0.02)0.55 (0.16)0.12 (0.01)0.12 (0.03)1.98 (1.73)0.44 (0.22)0.26 (0.14)0.12 (0.02)0.18 (0.02)0.15 (0.01)0.16 (0.02)0.15 (0.01)
Butanoic7.40 (4.86)5.08 (3.64)0.30 (0.12)0.57 (0.41)0.11 (0.02)0.56 (0.32)0.11 (0.01)0.15 (0.07)0.71 (0.51)0.61 (0.25)0.30 (0.08)0.17 (0.03)0.28 (0.004)0.25 (0.005)0.25 (0.01)0.27 (0.01)
Pentanoic7.06 (5.97)2.06 (1.17)0.13 (0.05)0.24 (0.17)0.07 (0.01)0.43 (0.14)0.04 (0.01)0.07 (0.01)0.24 (0.14)0.28 (0.20)0.16 (0.04)0.07 (0.02)0.16 (0.002)0.15 (0.002)0.14 (0.01)0.16 (0.005)
Hexanoic2.15 (1.42)1.27 (0.61)0.13 (0.02)0.19 (0.08)0.09 (0.02)0.39 (0.09)0.17 (0.02)0.16 (0.03)0.27 (0.05)0.20 (0.05)0.11 (0.01)0.09 (0.01)0.15 (0.01)0.14 (0.01)0.13 (0.01)0.15 (0.01)
Heptanoic0.21 (0.04)0.16 (0.03)0.04 (0.004)0.04 (0.008)0.03 (0.002)0.06 (0.005)0.05 (0.01)0.05 (0.01)0.07 (0.01)0.10 (0.03)0.04 (0.004)0.03 (0.01)0.05 (0.002)0.05 (0.001)0.05 (0.002)0.05 (0.001)
Octanoic0.2 (0.02)0.16 (0.04)0.05 (0.005)0.05 (0.002)0.04 (0.003)0.06 (0.002)0.05 (0.01)0.06 (0.01)0.09 (0.002)0.12 (0.03)0.05 (0.003)0.04 (0.004)0.06 (0.002)0.06 (0.002)0.06 (0.003)0.06 (0.003)
FenYear 1Year 2
Volatile fatty acidSpringSummerAutumnWinterSpringSummerAutumnWinter
AmbientSimulated LGMAmbientSimulated LGMAmbientSimulated LGMAmbientSimulated LGMAmbientSimulated LGMAmbientSimulated LGMAmbientSimulated LGMAmbientSimulated LGM
Acetic0.80 (0.22)18.0 (7.65)3.90 (2.36)2.33 (0.51)3.60 (1.79)1.30 (0.12)1.51 (0.06)1.60 (0.10)1.26 (0.15)7.10 (3.79)1.55 (0.06)1.22 (0.16)1.08 (0.03)1.41 (0.06)1.52 (0.11)1.29 (0.05)
Propanoic0.24 (0.09)6.04 (3.30)0.83 (0.66)0.58 (0.23)0.57 (0.29)0.40 (0.12)0.16 (0.02)0.24 (0.03)0.14 (0.01)0.42 (0.16)0.12 (0.003)0.11 (0.01)0.16 (0.004)0.15 (0.004)0.14 (0.01)0.14 (0.002)
Butanoic0.13 (0.05)20.8 (11.14)2.49 (2.30)0.63 (0.34)1.01 (0.67)0.19 (0.03)0.21 (0.07)0.20 (0.02)0.14 (0.02)2.35 (1.77)0.24 (0.01)0.14 (0.02)0.26 (0.01)0.26 (0.01)0.27 (0.01)0.25 (0.003)
Pentanoic0.06 (0.01)7.19 (4.13)1.21 (1.09)0.86 (0.49)0.61 (0.37)0.25 (0.07)0.12 (0.05)0.21 (0.04)0.08 (0.02)0.32 (0.20)0.13 (0.003)0.06 (0.01)0.15 (0.01)0.15 (0.01)0.14 (0.01)0.15 (0.003)
Hexanoic0.12 (0.05)7.50 (5.33)0.32 (0.20)0.35 (0.15)0.29 (0.14)0.24 (0.07)0.30 (0.01)0.31 (0.04)0.17 (0.02)0.94 (0.60)0.12 (0.01)0.10 (0.005)0.16 (0.01)0.15 (0.01)0.14 (0.01)0.14 (0.01)
Heptanoic0.04 (0.01)0.22 (0.04)0.04 (0.004)0.06 (0.01)0.08 (0.06)0.05 (0.004)0.09 (0.01)0.06 (0.01)0.05 (0.01)0.64 (0.39)0.04 (0.002)0.03 (0.005)0.05 (0.001)0.05 (0.001)0.05 (0.002)0.05 (0.001)
Octanoic0.04 (0.02)0.26 (0.03)0.04 (0.002)0.06 (0.01)0.10 (0.07)0.06 (0.004)0.11 (0.01)0.08 (0.01)0.07 (0.01)0.93 (0.44)0.05 (0.002)0.04 (0.005)0.06 (0.002)0.06 (0.002)0.06 (0.003)0.05 (0.002)

Only spring pore waters contained acetate at concentrations in both the bog and fen mesocosms that were notably above background concentrations (Fig. 3c). The bog acetate concentration was highest during the first spring period in the experiment in both the ambient (31.3 mg l−1) and simulated LGM (18.5 mg l−1) mesocosms (Fig. 3c). Bog ambient acetate concentration averaged 8.9 mg l−1 and simulated LGM 6.3 mg l−1 in year 1 (> 0.05). Values were lower in year 2, when the ambient group averaged 3.4 mg l−1 and the simulated LGM averaged 1.9 mg l−1 (> 0.05). Bog ambient and LGM [CO2] treatments during the autumn of year 1 were significantly different (< 0.05); however, the concentrations during this time were very low in both groups (ambient = 1.2 mg l−1 and simulated LGM = 2.4 mg l−1).

Fen acetate concentrations in ambient mesocosms remained consistently low throughout the experiment, with no notable peaks in any of the seasons. By contrast, the simulated LGM mesocosms exhibited a pattern of high concentrations in spring and much lower concentrations in the other seasons, a pattern that closely resembles that observed in the bog data. Over the first year, the fen ambient averaged 2.5 mg l−1 and the fen simulated LGM averaged 5.8 mg l−1, a statistically significant difference of 132% (< 0.05). A second peak in acetate concentration was measured in the second spring in the fen simulated LGM; however, like both the bog ambient and simulated LGM acetate concentration pattern, this is considerably lower than in the previous spring. In year 2, the ambient mesocosm average acetate concentration was 1.4 mg l−1, whereas the simulated LGM mesocosms averaged a higher 2.8 mg l−1 (> 0.05).

Influence of temperature and day length on CH4 flux

CH4 flux under laboratory controlled conditions responds exponentially to a linear increase in temperature (Daulat & Clymo, 1998). The influence of LGM atmospheric [CO2] on this relationship cannot be fully investigated in this experiment as temperature and day length are compounded. However, we plotted flux data from years 1 and 2 against the corresponding temperature when it was measured to cautiously suggest the potential impact of glacial [CO2] and ‘temperature’. In year 1, there was no clear relationship between CH4 flux and ‘temperature’ (data not shown), whereas in year 2, CH4 fluxes exhibited a linear response to ‘temperature’ (Fig. 4). Bog ambient and simulated LGM mesocosms shared a similar CH4 flux response across the experimental range, with statistically (ANCOVA) similar gradients and intercepts (> 0.05). By contrast, increasing temperature and day length caused a divergence in the response of CH4 between fen ambient and simulated LGM mesocoms. The simulated LGM effect caused a highly significant (< 0.001) shift in curve elevation (intercept) but did not change the gradient (> 0.05).

image

Figure 4. Temperature response of methane (CH4) flux measured in year 2 from (a) bog and (b) fen mesocosms. Error bars represent ± 1 SE of the mean. Ambient regression, solid line; simulated LGM regression, dashed line. (a) The ambient regression equation is = 1.91x − 8.80; the simulated Last Glacial Maximum (LGM) regression equation is = 1.87x − 10.6. (b) The ambient regression equation is = 2.67x– 10.6; the simulated LGM regression equation is = 2.00x − 9.59.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Applying an LGM atmospheric [CO2] starvation treatment to temperate wetland mesocosms resulted in a contrasting CH4 emission response. Reducing atmospheric [CO2] from present-day to glacial maxima concentrations suppressed both gaseous CH4 emissions and pore water DM concentration in minerotrophic fen mesocosms, whereas the ombrotrophic bog mesocosms showed no response (Figs 2, 3a). DM pore water concentrations showed a particularly clear differential response to CO2 starvation, with fen mesocosms exhibiting a c. 50% decrease in DM concentration compared with no change in their bog counterparts (Fig. 3b). Other studies where wetlands were exposed to elevated atmospheric [CO2] have shown increased CH4 emissions (Hutchin et al., 1995; Megonigal & Schlesinger, 1997; Saarnio & Silvola, 1999; Saarnio et al., 2000; Kang et al., 2001; Ellis et al., 2009) but the effect on DM concentration in pore waters is less straightforward. Marsh et al. (2005) showed no significant effects of elevated CO2 treatment on DM concentration compared with ambient controls on a brackish marsh after > 10 yr of exposure, as did Cheng et al. (2005) (> 0.05) when investigating the same treatment effect on DM concentration at 10 cm below the surface in rice (Oryza sativa) paddy soil. By contrast, Keller et al. (2009) measured an increased DM concentration in pore waters after a 5-yr period exposing a brackish marsh to elevated atmospheric [CO2]. Our results therefore provide further evidence that changing atmospheric [CO2] does not automatically change DM concentration in pore waters or gaseous CH4 flux from all wetland types.

The response of wetland ecosystems to changing atmospheric [CO2] may be contingent on trophic status (as highlighted in this study) as a result of differences in biotic and abiotic variables that control CH4 production and oxidation (e.g. pH, microbial diversity, plant functional groups and plant productivity). We therefore propose three possible reasons for the different responses of fen and bog mesocosms to CO2 starvation in terms of CH4 dynamics: contrasting nutrient limitation; opposing dissolved CO2 concentrations and dominant CH4 production pathways within the rhizosphere; and differences in soil-derived CO2 supplemented photosynthesis.

Low nutrient concentrations in ombrogenous bogs lead to low photosynthetic activity, plant growth, plant biomass, microbial biomass, ecosystem respiration and the overall decay rate of litter (Aerts et al., 1992, 2001; Basiliko et al., 2006). These factors may combine to limit CH4 emissions from bogs such that they are insensitive to changes in atmospheric [CO2] (Freeman et al., 2004). Hence, wetlands are unlikely to respond to CO2 starvation with a decrease in CH4 flux if nutrient availability is the dominant limitation. By contrast, if nutrient availability is less of a dominating factor for CH4 production, then reducing the atmospheric [CO2] may limit photosynthesis and plant production (Polley et al., 1993; Pagani et al., 2009) and ultimately the supply of fresh labile carbon substrate to methanogens (Whiting & Chanton, 1993; Chanton et al., 1995). The majority of root exudates tend to be lower molecular weight compounds (Walker et al., 2003), making them readily available for obligate proton-reducing acetogens and methanogens to utilize. However, under low atmospheric CO2 concentrations, the influence of nutrient availability on ecosystem productivity is reduced (Bazzaz, 1990) as atmospheric [CO2] becomes the predominant constraint on photosynthesis in C3 plants.

Our effort to infer the effect of CO2 starvation on plant related methanogenic substrate supply, from analysis of seasonal changes in pore water VFAs proved inconclusive (Fig. 3c). Bog mesocosms showed no difference between ambient and simulated LGM [CO2] during year 1 or 2 (> 0.05), whereas the fen simulated LGM mesocosms had a higher concentration in year 1 (< 0.05), and no difference in year 2 between the treatments (> 0.05). The highest acetate concentrations in both ecosystem types were measured during spring. These peaks represent an imbalance between the production and consumption of acetate in the rhizosphere, created by low temperatures allowing homoacetogenic bacteria to outcompete acetotophic methanogens and bacteria for H2 in the rhizosphere (Kotsyurbenko et al., 1996, 2001; Shannon & White, 1996; Duddleston et al., 2002; Hines et al., 2008; Hoj et al., 2008). As temperature increases, this shifts the balance in favour of acetoclasitc methanogenesis, resulting in acetate consumption in the rhizosphere and an increase in dissolved pore water [CH4] and gaseous CH4 emissions in the summer (Fig. 3) (Sugimoto & Wada, 1993; Shannon & White, 1996).

With no direct evidence of a reduction in root exudates in our experiment, we suggest the conflicting CH4 response may be attributable to different dissolved CO2 concentrations and dominant CH4 production pathways within the rhizospheres. Autotrophic and heterotrophic respiration within the rhizosphere elevates the concentration of dissolved CO2 within wetland pore waters (Smolders et al., 2001; Keller et al., 2009). The more neutral pore waters associated with fen ecosystems (Kang et al., 2001) do not favour the accumulation of dissolved CO2 as the equilibrium between the carbon species (CO2 : H2CO3 and HCO3 : CO32−) is shifted towards HCO3 because of the pH. By contrast, bogs have lower pH values and therefore an equilibrium that selects for the accumulation of dissolved CO2. As a result, it is possible that dissolved CO2 in simulated LGM mesocosms remained unchanged in the experiment. This would have significant implications for CH4 production as bog and fen mesocosms have contrasting dominant CH4 production pathways (Galand et al., 2010). Bogs that are Sphagnum-dominated show predominance for the hydrogentrophic CH4 production pathway (CO2 : H2) (Lansdown et al., 1992; Chanton et al., 1995; Duddleston et al., 2002; Horn et al., 2003; Hornibrook & Bowes, 2007), whereas nutrient-rich fens contain more methanogens that are obligate acetotrophs (Galand et al., 2005; Juottonen et al., 2005). Maintaining the aquatic source of CO2 in the simulated LGM bog mesocosms may have preserved CH4 emissions at levels associated with modern-day atmospheric [CO2] conditions.

Contrasting dominant CH4 production pathways explain why acetate concentrations measured in this study were higher in the bog than in the fen (Fig. 3c). Low pH values that are associated with bogs result in lower acetate turnover and a dominance of hydrogentrophic methanogenesis (Kotsyurbenko et al., 2007); consequently, acetate tends to accumulate in these ecosystems (Hines et al., 2001). By contrast, there is a higher demand for acetate in fens because of a greater presence of obligate acetotrophs (Galand et al., 2005), and hence CH4 emissions from minerotrophic wetlands could be susceptible to CO2 starvation without the drop in productivity being reflected in rhizosphere acetate concentration, as efficient removal would obliterate any potential signal.

Plant assemblage variations between the fen and bog mesocosms and a difference in utilization of soil-derived CO2 to supplement photosynthesis could also be part of a possible explanation for the contrasting CH4 dynamics we measured. Bog mesocosms in this study were mainly dominated by Sphagnum and Hypnaceous mosses and contained substantially fewer vascular plants compared with the fen mesocosms. The water table was maintained just below the surface in our experiment, therefore limiting the amount of CH4 oxidation in the mesocosms; however, Sphagnum mosses have been shown to form symbiotic relationships with methanotrophes in the hyline cells of the plant and on stem leaves (Raghoebarsing et al., 2005). The symbiosis results in the creation of CO2 as a result of O2 derived from photosynthesis driving methanotrophy of CH4 (Turetsky & Wieder, 1999), a CH4 recycling reaction that may also account for low CH4 emissions from Sphagnum areas. This CH4-derived CO2 provides an additional carbon source to submerged Sphagnum species in addition to that available from the atmosphere, and autotrophic and heterotrophic respiration. It is estimated that CH4-derived carbon accounts for between 5 and 35% of CO2 assimilated by Sphagnum (Raghoebarsing et al., 2005; Kip et al., 2010; Larmola et al., 2010). Consequently, Sphagnum-dominated bog mesocosms in our study may not have fully altered their physiological processes in response to atmospheric CO2 starvation as a result of the supplementing of photosynthesis with subsurface CO2. Therefore, it may be that plant assemblage is responsible for the contrasting bog and fen mesocosm response measured in this study, as this has a significant influence on dominant CH4 production pathways (Hines et al., 2008), the amount of rhizosphere supplemented photosynthesis (Raghoebarsing et al., 2005) and CH4 oxidation (Parmentier et al., 2011) and warrants further investigation.

Potential implications of findings

Our results suggest that CH4 emissions from minerotrophic temperate wetlands at the LGM may have been disproportionately affected by CO2 starvation when compared with ombrotrophic bogs occupying unglaciated boreal regions. Furthermore, mesocosm CH4 emissions exhibited a significant positive correlation with temperature and day length in year 2 (Fig. 4), with peaks in CH4 flux during the warmest periods (Figs 2, 3a), in agreement with laboratory (Thomas et al., 1996; Daulat & Clymo, 1998; Gauci et al., 2004) and field studies (Dise, 1993). In addition to this well-established relationship, our results indicate that the largest suppressions in fen CH4 flux induced by LGM atmospheric [CO2] occurred when temperature limitation on carbon mineralization was at its lowest during the warm summer months at temperatures > 15°C. Methanogen communities operate more rapidly, and increase diversity and relative abundance at temperatures > 10°C (Hoj et al., 2008); consequently, we hypothesize that, during the summer, CH4 emissions from the CO2-starved fen switched from being temperature-limited to being substrate-limited. Therefore, the response of wetland CH4 emissions to glacial atmospheric [CO2] may also be moderated by latitudinal temperature gradients. In which case, the largest suppression in wetland CH4 flux at the LGM would have been in the low latitudes, with a diminishing effect at higher latitudes. As a consequence, the warm-temperate and tropical wetland CH4 source may have responded sensitively to CO2 starvation during glacial episodes, a theory that is supported by modelling estimates in conjunction with carbon isotope ratios in CH4 found within ice cores (Fischer et al., 2008; Singarayer et al., 2011). Because lower latitude wetlands were probably the dominant source of CH4 during the LGM (Chappellaz et al., 1993; Dallenbach et al., 2000; Fischer et al., 2008; Weber et al., 2010; Singarayer et al., 2011), this challenges the assumption that global wetland CH4 emissions at the LGM would have been of similar magnitude to modern-day emissions (Kaplan et al., 2006).

In summary, we demonstrate that the low LGM atmospheric [CO2] significantly limited gaseous CH4 flux from, and DM concentration in, minerotrophic wetland ecosystems, while having no effect on ombrogenous bogs. The exact mechanism(s) that causes this heterogeneous response to glacial atmospheric [CO2] remains to be established. Recent studies suggest that glacial–interglacial variations in atmospheric [CH4] are caused by variations in global wetland CH4 source strength in response to changing orbital insolation patterns and greenhouse gas concentrations (Singarayer et al., 2011). Our experimental results support this hypothesis and provide direct evidence that wetland biogeochemical processes are sensitive to the effects of atmospheric glacial [CO2].

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The work was supported by research studentship funding provided by The Open University (OU) to C.P.B. (via V.G.), HEFCE Science Research Infrastructure Funding support for the OU Biogeochemistry Laboratory, seedcorn funding provided by the OU and Royal Society support to V.G. We thank Wesley Fraser for fieldwork assistance, Graham Howell and Corinne Rooney for laboratory assistance, and David Gowing and Mike Dodd for assistance with plant identification. We thank the Countryside Council for Wales for access to field sites. D.J.B. gratefully acknowledges a Royal Society-Wolfson Research Merit Award. We thank Pat Megonigal and anonymous reviewers for their thoughts and recommendations (particularly on potential mechanisms) on earlier versions of this manuscript.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References