Introduction

Northern peatlands contain about one third of global soil organic carbon (C) stocks, built up over many thousands of years under water-logged conditions that have limited decomposition (Gorham 1991). Future climate change might result in lower water tables, peat aeration and thus stimulated peat decomposition (Ise et al. 2008) as projections show an increase in the magnitude and frequency of drought events in the northern hemisphere (IPCC 2007). Observations of rising dissolved organic carbon (DOC) concentrations in rivers draining areas dominated by organic rich soils have lead to concerns that these carbon stores are beginning to destabilize (Freeman et al. 2001a). The mechanisms controlling DOC release from peats during droughts have been intensely debated over recent years, given the potential for increased DOC production and release in response to water table draw-down (Clark et al. 2010; Eimers et al. 2008b; Evans et al. 2002, 2006; Freeman et al. 2001a, 2004; Watts et al. 2001; Worrall et al. 2004). However, current dynamic peatland models are still inadequate in reflecting these processes, particularly the dynamic relationship between peatland hydrology and carbon (C) cycling (Heinemeyer et al. 2010). Further experimental work is needed to gain an improved understanding of the underlying decomposition processes that need to be incorporated into these models.

While it is generally accepted that water table draw-down increases peat decomposition and CO2 efflux, with a concurrent decrease in methane flux (CH4) (Blodau and Moore 2003; Freeman et al. 1993), the impact on DOC dynamics is less clear. Studies have shown conflicting results, with DOC concentrations increasing (Glatzel et al. 2006), decreasing (Clark et al. 2005; Fenner et al. 2005; Hughes et al. 1998; Pastor et al. 2003; Scott et al. 1998) and showing no change in both soil and stream waters following water table draw-down in peat (Blodau and Moore 2003; Eimers et al. 2008b). In some cases, stream water studies have not been able to report on dynamics during drought events as cessation of flow prevented sample collection (Eimers et al. 2008b; Jager et al. 2009), and therefore DOC dynamics during drought events are not recorded.

Explanations for DOC concentration responses during drought events are varied. Most simply, DOC concentrations might increase during droughts due to ‘evapoconcentration’ from decreasing water volumes (Waiser 2006). Other studies suggest that DOC concentrations increase as a result of increased biological activity and net DOC production, where more DOC is released during decomposition than is consumed and respired by soil biota (Pastor et al. 2003; Tipping et al. 1999). The DOC produced is either released directly to pore waters during drought events (Glatzel et al. 2006; Strack et al. 2008), or is adsorbed during the drought period and subsequently released as the peat rewets (Clark et al. 2009; Mitchell and McDonald 1992; Scott et al. 1998; Tipping et al. 1999; Watts et al. 2001; Worrall et al. 2003). Conversely, lower DOC concentrations during droughts could be due to decreased biological activity and DOC production (Scott et al. 1998); or because DOC consumption by soil biota exceeds DOC production (Fenner et al. 2005; Pastor et al. 2003; Scott et al. 1998). As DOC fluxes (i.e. mass of DOC transported per unit time per unit area) are controlled by water flow, stream water studies have consistently reported lower DOC fluxes during ‘dry’ years relative to ‘wet’ years, irrespective of changes in concentrations (Clark et al. 2007; Dillon and Molot 2005; Eimers et al. 2008b; Schindler et al. 1997; Worrall and Burt 2008).

In our previous work (Clark et al. 2005, 2006), we argued that observed soil water DOC concentrations decreased during drought events because sulphur (S) redox reactions caused an increase in acidity and ionic strength which decreased DOC solubility. This conclusion was based on evidence from one English blanket peat site, where both in-situ pore water monitoring data from the field and laboratory soil core experiments showed consistent results (Clark et al. 2006). A solubility control on DOC by drought-induced acidification could also explain observations of increasing DOC concentrations following drought periods; as both acidity and ionic strength decrease once the water table recovers and SO4 reduction takes place. Consequently, a store of compounds that are produced during the drought are only soluble at higher pH/lower ionic strength after water tables have recovered to the surface (Clark et al. 2005, 2006; Toberman et al. 2008). In this situation, DOC dynamics during drought events are indirectly hydrologically controlled by changes in soil water chemistry driven by water table fluctuations, rather than directly hydrologically controlled due to changes in water volume or flow.

However, increasing DOC following rewetting has also been attributed to enhanced enzyme activity increasing decomposition and DOC production (Fenner et al. 2005; Hughes et al. 1998; Toberman et al. 2008). Others have suggested that the post-drought recovery in DOC concentrations is due to hydrophobic recovery of peat, where structural changes have delayed complete rewetting of peat and the subsequent release of adsorbed DOC to pore waters (Watts et al. 2001; Worrall et al. 2003). In stream waters, others have argued that the main effect of drought on DOC is to limit its transport through cessation of flow (Eimers et al. 2008b). Therefore at the catchment scale, additional processes to those acting within peat pore waters at the plot/mesocosm scale contribute to overall observed DOC response to drought (Clark et al. 2010; Preston et al. 2011). As DOC in stream waters is controlled by supply from terrestrial sources as well as transport from soil to stream (Thurman 1985), understanding soil processes at the plot scale are important to understand the whole catchment response.

So far, no consensus has emerged on which of these processes controlling the production and release of DOC within pore waters (i.e. plot or mesocosm scale) under drought is ubiquitous with respect to variation in vegetation, peat type, climatic conditions and atmospheric inputs between locations. Studies have tended to focus on a single site or local region, each with a different scale of investigation (e.g. lab, plot or catchment) and/or sampling strategy, making comparisons between studies difficult (Clark et al. 2010). Unified studies across a range of sites at identical scales are required to fully examine the various hypothesised controls on DOC dynamics to ensure data interpretations between sites are based on comparable data. Therefore, we carried out controlled laboratory experiments to test whether evidence of DOC suppression by drought-induced acidification in pore waters (Clark et al. 2005, 2006) could be detected at other sites in Great Britain. We compared responses between peat collected from Moor House (our original study site) with five additional field sites with differing historic acid deposition and/or vegetation cover. We monitored changes in pore water chemistry, water volume and trace gas fluxes during simulated drying and rewetting to determine whether droughts resulted in: (i) decreased DOC concentrations in pore waters; (ii) increased soil water acidification and ionic strength; (iii) increased biological activity and potential consumption of DOC as a substrate for heterotrophic respiration; and (iv) whether decreased water volume affected DOC concentrations.

Materials and methods

Field sites

Cores from the top 10 cm of the peat were collected from six sites in May/June 2006. Sites in order of high to low S deposition were: River Etherow (ETH, South Pennines), Afon Gwy (GWY, Mid-Wales), Moor House (MH, North Pennines), Dargall Lane (DL, South-West Scotland), Loch Coire nan Arr (LCNA, North-West Scotland), Allt a’Mharcaidh (AM, Cairngorms). Mean total (marine and acid) S deposition for these sites over the period of 1985–2005 ranged from 10 to 57 kg S/ha/year; with non-marine S (xS) deposition ranging from 7 to 37 kg S/ha/year (Ron Smith, pers. comm.; Table 1). Five of the sites were located within catchments currently monitored by the Acid Waters Monitoring Network (AWMN) (see Patrick et al. 1991) and one site (Moor House) was monitored by the Environmental Change Network (ECN) (see Heal and Smith 1978). All sites were moorland areas used primarily for grazing. Four sites (GWY, MH, AM, ETH) were extensive areas of blanket peat and two sites were basin peat (DL, LCNA) that had developed in glaciated valleys on moraine. The main peatland vegetation species Calluna vulgaris and Sphagnum spp. were present at three sites (GWY, MH, AM) and absent at the others where Eriophorum spp. (ETH, DL) or Molina spp. (LCNA) were dominant. ETH is located in the Peak District (east of Manchester), where climate change, peat erosion and high levels of acid deposition following industrial development have been associated with the decline in Sphagnum spp. (Tallis 1994). Peats were acidic (pH 3.9–4.9 in H2O) with moderate to low base saturation (11.1–48.9%) (Table 2). Further details about the basic peat properties and site descriptions can be found in Clark et al. (2011).

Table 1 Site locations and description
Full size table
Table 2 Peat chemical and physical properties: soil pH (ratio 1:10 wet soil to solution), C:N ratio, base saturation and bulk density
Full size table

Experimental design

At each site, eight peat cores were collected at random locations within a 30 m radius of a chosen sample point (Table 1). To reduce the amount of live biomass (i.e. roots) within each core, soil samples were collected between tussocks and shrubs rather than directly underneath the plants. Other surface vegetation and litter were removed. Each intact core was extracted by carefully cutting a PVC tube (10 cm diameter, 11 cm length) into the peat using a serrated edge knife, taking care to leave a 0.5–1 cm gap between the surface of the peat and the top of the tube. Once excavated, cores were planced into straight-sided Nalgene poly-carbonate bottles (11 cm in diameter, 12 cm in length) and chilled during transport.

On return to the laboratory, a 10 cm Eijkelkamp Rhizon soil water sampler (SMS) was inserted into each core. Pore water is extracted by applying suction from a syringe attached to the Rhizon SMS (see Clark et al. 2006). Cores were then slowly rewetted with deionized water until the surface peat was covered with 0.5–1 cm water. Deionized water was used in preference to a synthetic soil water solution to ensure no additional solutes were added. Variation in precipitation and soil water chemistries between sites also precluded the use of a single standard synthetic solution. All cores were then placed in an incubator for 6 months at 10°C to allow soil water to stabilize, background SO4 and NO3 concentrations to reduce naturally to negligible concentrations, and any labile carbon produced as an artefact from decaying fresh roots to be consumed prior to the start of the experiment, as rates of carbon turnover are elevated in disturbed peat cores (Blodau and Moore 2003). Lids were placed loosely on the bottles to reduce evaporation whilst maintaining air exchange.

The experimental simulation was split into three stages: (1) initial pre-treatment (all cores saturated), (2) 12 weeks of simulated drought by water table draw-down (four cores each unsaturated and saturated for each site) and (3) 20 week rewetting period (all cores saturated) (Fig. 1). Water table draw-down was simulated by pouring off excess drainage water from soil cores during the first 4 weeks of the 12-week drought (i.e. weeks 0–4), and then by extracting an additional 20 mL every 2 weeks (weeks 4–12) using a syringe and the Rhizon SMS samplers. Cores were rewetted following soil water sample extraction in week 12 and monitored for a further 20 weeks until soil water concentrations stabilized.

For chemical analysis, 20 mL of soil water was collected every 4 weeks using Rhizon SMS samplers, starting with a pre-treatment sample taken just prior to Stage 2 (week 0). Core weights were recorded prior to sample collection. Water loss due to evaporation and sample extraction was replaced with deionised water after sampling in the saturated cores to maintain constant water levels throughout the experiment when saturated conditions were required.

Fig. 1
figure 1

Changes in water content over 32 week experimental period in a dry and b wet (i.e. control) treatments. Water context expressed as mass (g) of water per mass (g) of dried peat. Mean value shown (N = 4)

Full size image

At the end of the experiment, all peat cores were dried in an oven at 70°C for 48 h to determine their final water content and oven dry mass.

Soil water analysis

Soil water was filtered on extraction by the PTFE membrane in the Rhizon SMS samplers to <1 μm (Clark 2005). Water samples were analysed for pH; conductivity; SO4, chloride (Cl) and nitrate (NO3) by ion chromatography (Dionex); calcium (Ca), magnesium (Mg), sodium (Na) and potassium (K) by atomic absorption spectrophotometry (Hitachi); DOC by combustion infra-red analyser (Thermalox TOC-TN); and colour with scanning UV–VIS–NIR spectrophotometer (Shimadzu).

Gas analysis

Real-time trace gas flux measurements were taken before soil water sampling in weeks 0, 4, 8, 12, 16 and 32. A closed gas flow system was established using a Li-Cor 8100 carbon dioxide (CO2) infra-red gas analyser (IRGA) connected to a flux chamber (8100-102; Li-Cor Biosciences, USA). The chamber (diameter of 10 cm and a volume of 955 mL) was placed over each core for 10 min. CO2 flux data presented in this paper were calculated as the linear increase in humidity corrected CO2 concentrations (ppm) at a 1 s recording frequency during the first 2 min of chamber closure.

Further calculations

Changes in DOC characteristics were inferred from SUVA (specific ultra-violet absorbance) and E340 (extinction coefficient at 340 nm) values calculated from the absorbance and measured DOC concentration data. SUVA was calculated as the absorbance at 254 nm in units/m divided by the concentrations of DOC in mg/L. Analysis has shown a good relationship between SUVA and coloured aromatic humic acids (Weishaar et al. 2003). E340 was calculated as the absorbance at 340 nm in units/cm divided by the concentration of DOC in g/L; and has been shown to have a negative relationship with the hydrophilic fraction of DOC and positive relationship with the aluminium (Al) adsorption capacity (Thacker et al. 2008). ‘Evapoconcentration’ effects, that may cause a change in concentration due to a change in water volume, were examined by (1) taking the ratio of a solute (e.g. DOC) to Cl, assuming Cl is a conservative tracer (Waiser 2006) and by (2) calculating the total amount of a solute per mass of dry peat.

Acid neutralising capacity (ANC) is a conservative measure of acidity of pore waters, as unlike pH and alkalinity measurements, it is unaffected by CO2 degassing, aluminium reactions (Al) or organic species (Neal et al. 1999) and has been widely used a key chemical indicator of acidification status in Critical Load assessments (CLAG 1994). The ANC was determined here as the difference between the sum of strong base cations (SBC; calcium (Ca), potassium (K), magnesium (Mg), sodium (Na)) and the sum of strong acid anions (SAA; chloride (Cl), NO3, SO4) (Hemond 1990):

$$ {\text{ANC}} = \left( {\left[ {{\text{Ca}}^{2+} } \right] + \left[ {{\text{Mg}}^{2+} } \right] + \left[ {{\text{Na}}^{+} } \right] + \left[ {{\text{K}}^{+} } \right]} \right) - \left( {\left[ {{\text{Cl}}^{-} } \right] + \left[ {{{\text{NO}}_{3}}^{ - } } \right] + \left[ {{\text{SO}}_{4} } \right]} \right) $$
(1)

where concentrations of each ion was in μeq/L.

Ionic strength was estimated as:

$$ {\text{IS}} = 0.5\sum\limits_{i = 1}^{n} {C_{i} z_{i}^{2} } $$
(2)

where IS is in mol/L; C i is the concentration of a substance in mol/L and z is the valence of the substance (Sparks 2003). Ionic strength is typically calculated from all solutes in solution. However, in this case only base cation, acid anion and H+ data were available. Other studies on similar organic rich waters have included ammonia (NH4), aluminium (Al3+) and bicarbonate (HCO3 −) in their calculations (Hruska et al. 2009). As soil waters were acidic, HCO3 − concentrations will be negligible and inorganic carbon present as dissolved CO2. As blanket peat soils are typically low in N and Al, it is unclear whether these values were negligible. Al was present in semi-confined peats following analysis for exchangeable acidity (Clark et al. 2011), due to inputs of Al via throughflow from upslope sources (Mulder et al. 1991) and NH4 may also be present in sites impacted by N deposition. Therefore, values of IS calculated here are likely to have underestimated actual values.

Mass balance calculations were made to estimate the balance between (1) DOC consumption/production and subsequent respiration as CO2 and (2) changes in DOC due to chemical solubility. The rate of change in DOC between sample weeks was estimated (i.e. week 0–4, 4–8, 8–12 etc.). Total DOC amounts were corrected at each time point for removal of DOC by sample collection for analysis. For consistency, the average CO2 flux was estimated between the sampling periods (i.e. week 0–4, 4–8, 8–12 etc.). All changes were expressed in terms of unit mass of carbon (as either DOC or CO2) per unit mass of dry peat.

Further stochiometric calculations were carried out to determine possible changes in CO2 flux due to SO4 and NO3 reduction during peat rewetting. Since acidic conditions prevail in these soils, the reaction for SO4 reduction used was that for conditions of pH 4.5 and below:

$$ {{\text{SO}}_{4}}^{2-} + 2{\text{CH}}_{2} {\text{O}} + 2{\text{H}}^{+} \Rightarrow {\text{H}}_{2} {\text{S}} + 2{\text{CO}}_{2} + 2{\text{H}}_{2} {\text{O}} $$
(Reaction 1)

where CH2O is organic matter and H2S is hydrogen sulphide. For NO3 reduction, two possible reactions could occur depending on the presence of oxygen within the system. As the actual redox status was unknown, calculations were made using both equations. Under oxic conditions, the reaction is likely to be:

$$ 2{{\text{NO}}_{3}}^{-} + 2{\text{CH}}_{2} {\text{O}} + 2{\text{H}}_{2} {\text{O}} + 2{\text{H}}^{ + } \Rightarrow {\text{N}}_{2} {\text{O}} + 2{\text{CO}}_{2} + 5{\text{H}}_{2} {\text{O}} $$
(Reaction 2)

where N2O is nitrous oxide. Under anoxic conditions, dissimilatory NO3 reduction produces nitrogen gas (N2):

$$ 4{{\text{NO}}_{3}}^{ - } + 5{\text{CH}}_{2} {\text{O}} + 4{\text{H}}^{ + } \Rightarrow 2{\text{N}}_{2} + 5{\text{CO}}_{2} + 7{\text{H}}_{2} {\text{O}} $$
(Reaction 3)

From these reactions, the production of CO2-C per unit mass SO4-S reduction can be calculated by multiplying the mass of SO4-S reduced by 0.749; for NO3-N reduction, the unit mass production of CO2-C can be calculated by multiplying the mass of NO3-N reduced by 0.858 under oxic conditions and 1.07 under anoxic conditions.

Statistical analysis

The experiment design used here reflects a two-way ANOVA with repeated measures, where both peat site and dry–wet/wet–wet cycle are treatment factors and response variables (e.g. DOC) were measured at many points over time from the same replicate. However, data analysis was problematic as the variations between sites and over time in many response variables produced data sets with unequal variance that resulted in model fits with unequal error distributions. Transformations through log or power transformations did not improve this because of the nature of data distribution and in some cases was not possible due to negative values (e.g. ANC).

Instead, the data were analysed using standard two-way ANOVA for two critical points during the experiment where the change in response variable between the start and either week 12 (end of the dry period) or week 32 (end of experiment after rewetting) was calculated to remove the temporal pseudo-replication (Crawley 2007). Homogeneity of variance was tested using the Fligner–Killen test (Crawley 2007) and reported so analysis based on untransformed data where variances were unequal can be interpreted with caution. Post-hoc tests were only carried out when the equality of variance assumption was satisfied. Otherwise, graphical representation of standard error (i.e. standard deviation divided by the square root of the number of samples) in tables and figures were provided to allow comparisons. The Scheirer–Ray–Hare test, essentially a non-parametric version of two-way ANOVA were analysis is carried out on ranked data (Dytham 2011), was also preformed where variances were unequal. The 95% confidence interval of the mean was reported using the exact confidence interval from the Student’s T distribution.

Results

DOC response to simulated water table drawdown

Both pore water DOC concentrations and the total amount of DOC per core declined during the simulated dry period and increased during the wet period in all peat cores exposed to the ‘dry’ treatment (Fig. 2a, c). Results from both the parametric (score data) and non-parametric (rank) two-way ANOVA showed a significant difference between dry and wet treatments between 0 to 12 weeks (p < 0.001) but not between weeks 0 and 32 (p = 0.05–0.10) . Concentrations and amount of DOC released varied significantly between sites (p < 0.001). Overall change in DOC concentrations in the dry treatment from week 0 to 12 ranged from −36 to −87% with mean absolute concentrations in the dry cores ranging from 6.1 to 39.3 mg/L compared to 39.6 to 276.0 mg/L in the wet cores. Although the dry treatment period lasted from week 0 to 12, the greatest decline in DOC concentrations occurred during weeks 4–8.

Fig. 2
figure 2

Change in DOC amount and quality over the 32 week experimental period under the dry–wet and wet treatments: a, b DOC concentrations; c, d amount of DOC per unit mass dry peat; e, f molar ratio of DOC:Cl; g, h SUVA; i, j E340. Mean values shown (N = 4)

Full size image

The ratio of DOC:Cl also decreased during the simulated dry period for all peats except AM (Fig. 2e), where the ratio of DOC:Cl increased. If Cl is assumed to be a conservative tracer, ‘evapoconcentration’ would not change the DOC:Cl ratio. Decrease in the DOC:Cl show that DOC concentrations declined in both absolute and relative terms during the dry period.

All sites (except AM) showed a consistent decline in SUVA during the dry period indicating a change in DOC quality associated with a decline in the coloured humic acid fraction (Fig. 2g, h). Like DOC concentration and amount, both SUVA and E340 showed the greatest rate of decline between weeks 4 and 8, hence changes in DOC quality occurred at the same time as changes in DOC quantity. Following rewetting in week 12, DOC concentration and amount increased rapidly (Fig. 2a, c). SUVA and E340 in all peats between week 12 and 14 responded immediately to rewetting, showing that the DOC quality recovered faster than absolute concentration or amount.

In contrast, both concentrations and amount of DOC in the wet treatment cores showed either a steady decrease (LCNA, DL, ETH) or small increase (MH, AM, GWY) over time (Fig. 2). No significant pattern in SUVA or E340 was seen in the wet treatments, with values remaining constant or showing a slight decline as well (Fig. 2h, j).

Overall, observations support widespread evidence that a shift from saturated (i.e. ‘wet’) to unsaturated (i.e. ‘dry’) conditions resulted in a net decline in DOC, particularly the coloured hydrophobic humic fractions measured by SUVA and E340, and rewetting resulted in a rapid increase in both DOC and the proportion of humic material. These changes in DOC concentration and quality appeared to be independent of changes in the volume of water as both the amount of DOC in solution per g of peat and the ratio of DOC:Cl decreased.

Relationship between soil water acidification and DOC

Drying resulted in increased SO4 concentrations in all peats, which decreased when peats were rewetted, due to the oxidation and subsequent reduction of S (Fig. 3a) (Bottrell et al. 2004). The magnitude of peak SO4 concentrations varied considerably between peats (102–2,130 μeq/L), with greatest concentrations in peats that have received the greatest atmospheric S deposition (Table 1). Mean SO4 concentrations remained at <20.8 μeq/L (1 mg/L; Fig. 3b) in the wet treatment core throughout the experiment.

Fig. 3
figure 3

Change in soil water chemistry over the 32 week experimental period under the dry–wet and wet treatments, respectively: a, b SO4; c, d ANC; e, f pH; g, h conductivity; i, j ionic strength. Mean values shown (N = 4)

Full size image

Interestingly, the response of both conductivity and estimated ionic strength to drying and rewetting differed between those sites that had received the greatest levels of pollutant non-marine S deposition (DL, MH, GWY, ETH) and the least acid-impacted sites in north Scotland (AM, LCNA). Conductivity and ionic strength increased in the most acid-impacted peats, but decreased in the least acid-impacted peats during the dry treatment (Fig. 3g, i). The difference in responses between the peats can be attributed to differences in SBC concentrations, as drying resulted in a decline of −63 to −64% in the least impacted peats and increase in SBC concentration of 102–390% in the most acid impacted peats. However, the increase in the amount of SBC (μeq/g peat) in the most acid-impacted peats was proportionally smaller than increases in concentrations.

Drying and rewetting also resulted in the decrease and subsequent increase in ANC in all samples (Fig. 3c). The pH also showed a concurrent increase and decrease in peats from all sites except LCNA, where pH continued to rise (Fig. 3e). However, it should be noted that the pH meter seldom stabilised in LCNA solutions, so pH readings were taken after a fixed time period (1 min) and results should be viewed with caution. The decline in SBC at the least impacted sites contributed to the overall decline in ANC during dry periods where SAA increases were small. By contrast, at the more acid-impacted sites increased SBC partly offset increased SAA concentrations. Increased SAA was principally driven by increased SO4 due to the oxidation of reduced S, although significant increases in NO3 and were seen in ETH and DL peats (436 and 1,018 μeq NO3/L, respectively). The Cl concentrations showed a slow and steady decrease in both wet and dry treatments over the experiment, most likely due to continued removal of water during sample collection. As with other chemical variables, pH, conductivity and ionic strength remained reasonably constant or showed a slow decline over time in the wet treatment (Fig. 2e, g, i).

Inverse relationship between SO4 and DOC seen in MH peats were repeated in five other peats studied here (Figs. 2, 3). However, the magnitude of the SO4 increase (102–2,130 μeq SO4/L) and DOC decrease (−36 to −87%) varied between peats, such that no single relationship between SO4 concentration and the percentage change in DOC concentrations (ΔDOC%) emerged (Fig. 4). Furthermore, as noted above, two of the least acid impacted peats showed a decrease in conductivity with increasing SO4 concentrations, and one peat (LCNA) showed no pH response in spite of a change in ANC (Fig. 5).

Fig. 4
figure 4

Relative change in DOC (expressed as both a, b percentage and c, d concentration) with respect to change in SO4 concentrations, compared to the relationship between e, f observed values. Change relative to initial DOC concentration in week 0. Mean value shown (N = 4) for each treatment and soil type for each of the ten sampling points in time. Pearson’s correlation coefficient (r) and the significance value are shown

Full size image
Fig. 5
figure 5

Relationship between change in ANC and relative change in DOC concentrations (expressed as both a, b percentage and c, d change in concentration), compared to the relationship between the e, f absolute values. Change relative to initial concentration in week 0. Mean value shown (N = 4) for each treatment and soil type for each of the ten sampling points in time. Pearson’s correlation coefficient (r) and the significance value are shown

Full size image

However, ANC, like DOC, showed a consistent response between peats. Clear relationship between the change in ANC and change in DOC in terms of both percentage (r = 0.7) and absolute change (r = 0.94) were seen between sites (Fig. 5), with less variability between sites than the relationship between SO4 and change in DOC (Fig. 4). A linear relationship was observed between ANC and change in DOC concentrations in mg/L (Fig. 5c). The relationship between ANC and DOC observed between sites was more consistent when relative changes in ANC and DOC were considered rather than the absolute concentration (Fig. 5).

Changes in soil respiration: biological production and consumption

Observations showed CO2 fluxes in dry treatments were significantly greater than wet treatments in all peats (p < 0.001), with the lowest overall fluxes seen in ETH peats and greatest in LCNA (e.g. Fig. 6a). Comparison between the rate of change in DOC concentrations and CO2 fluxes were made to determine whether the increase in CO2 flux in response to drying could be explained by decreased DOC concentrations due to consumption and respiration. Only LCNA in weeks 0–4 showed a significantly greater change in DOC than elevated CO2 flux (p = 0.03), whereas GWY in week 8–12 showed a significantly lower change in DOC than elevated CO2 flux (p = 0.01). For all remaining times during the dry period, rate of change in DOC concentration and CO2 flux were similar.

Fig. 6
figure 6

Net C production of a CO2-C and b DOC between week 4 and 8 under both dry and wet experimental treatments. Mean value and standard error shown (N = 4). Two-way ANOVA for a CO2-C showed a significant difference between sites and dry/wet treatment (p < 0.001) but no significant interaction between these factors (p = 0.95). Two-way ANOVA for b DOC showed a weakly significant different between sites (p = 0.01) and significant difference between dry/wet treatment (p < 0.001) but no signification interaction between these factors (p = 0.12). As variances for DOC were not homogeneous (p < 0.01), the Scheirer–Ray–Hare test was also used for these data, producing similar findings: both site and treatment were significant (p < 0.001), with no significant interaction (p = 0.57)

Full size image

During the rewetting phase (week 12–14), the increases in DOC concentrations were generally greater than the CO2 fluxes (Table 3). CO2 fluxes declined during rewetting relative to elevated fluxes during the dry period, although were greater than under steady-state wet conditions. Stoichiometric calculations suggest that some of the CO2 fluxes during rewetting could be explained by CO2 released from SO4 and NO3 reduction, which would also consume DOC, however these fluxes were small in comparison to the total measured fluxes (Table 3).

Table 3 Average rates of change in the measured CO2-C flux and net DOC production compared to the calculated CO2-C flux from SO4 and NO3 reduction during the rewetting period (week 12–16) in dry treatment cores
Full size table

Discussion

Data from these experiments are consistent with our previous work at Moor House, where DOC concentrations in peat pore waters were shown to decrease during drought events and increase following rewetting (Clark et al. 2006). Similar observations have been reported elsewhere (Blodau and Moore 2003; Fenner et al. 2005; Hughes et al. 1998; Pastor et al. 2003). Other additional work reporting increased DOC concentrations in response to water table draw-down were only able to measure peat leachates or soil water during the rewetting phase (Mitchell and McDonald 1992; Tipping et al. 1999). It is unclear in these cases whether DOC concentrations increased or decreased during the actual dry period as no data were available from this time period. Likewise, no changes in DOC concentrations in streams have been shown when flow ceases (Eimers et al. 2008b), making it impossible to infer soil processes during these periods without stream water or additional pore water data.

Change in the water volume in stream and peat pore waters have also been suggested to affect DOC concentrations during drought periods (Eimers et al. 2008a; Waiser 2006). For instance, in the mineral horizon of an organo-mineral soil exposed to drought, increased DOC concentrations were attributed to decreased water volume (Sowerby et al. 2010). However, decreased water volume during drought periods and increased water volume following peat rewetting would have the opposite concentration/dilution effect to the actual observed decrease/increase in DOC concentrations seen in soil (Clark et al. 2006; Hughes et al. 1998) and stream waters (Scott et al. 1998; Watts et al. 2001). Instead of influencing DOC concentrations, the impact of ‘evapoconcentration’ appeared to be more apparent in terms of increasing base cation concentrations in the most acid-impacted soils, affecting the overall balance between acid anions and neutralising base cations (i.e. ANC) and hence the degree of acidification which in turn influences DOC solubility. However, H+ ions and Al can also displace cations from peat exchange sites, so increased base cation concentrations could be due to acidification induced cation exchange.

In this study and our previous work, clear relationships between SO4, acidity and DOC were only seen when changes in both variables were considered, rather than the absolute values. As many factors influence DOC dynamics, particularly in the field, variations due to seasonal changes in temperature needed to be removed before the effect of drought-induced acidification were seen (Clark et al. 2005). Integration of more processes with scale of measurement can make it difficult to disentangle the signal from processes operating in the soil (or laboratory experiment scale) to those operating at a catchment scale (Preston et al. 2011), unless care is taken with the analysis to factor out the larger scale catchment or site specific drivers (Clark et al. 2010). In peatland streams where DOC concentrations vary little with stream flow, seasonal changes in temperature are the main source of variation (Clark et al. 2007; Schiff et al. 1998). However, in catchments with organo-mineral soils and mixed soil types, changes in stream flow are the principle driver of variations in DOC concentrations (Clark et al. 2007; Hope et al. 1994); therefore, like temperature, the effect of stream flow needs to be removed before the effect of other factors, like drought-induced acidification, on DOC dynamics can be seen. For instance, in the analysis presented by Eimers et al. (2008b), no consistent relationship was found between absolute SO4 and DOC concentrations in stream waters. Rather than contrary to our findings, these observations are consistent with our earlier work, which also showed no clear overall relationship between SO4 and observed DOC concentrations when the effect of temperature was not removed from the data (Clark 2005; Clark et al. 2006).

Comparison between different peats showed that the relationship between drought-induced acidification and DOC mobility was more consistent when changes were measured in terms ANC rather than SO4. The ANC provided a more robust measure of acidification change, as SO4, NO3 and the base cation responses are all accounted for. Observations from minerotrophic fens have shown that increased SO4 concentrations in response to drying are not always associated with increased acidity (Knorr et al. 2009). An inverse relationship between the change in ANC and change in DOC during episodic droughts is consistent with seasonal inverse relationships between ANC and DOC seen in field observations in pore waters at both MH and GWY (Chapman et al. 2008) and also in long-term nitrogen addition experiments (Evans et al. 2008). It is possible that change or no change in ANC may also explain DOC dynamics in organic horizons of organo-mineral soils, where no relationship between SO4, pH and DOC has been found (Sowerby et al. 2010). Further work is required to determine whether this is likely.

Rapid and reversible changes in DOC quality measured in terms of changes in SUVA and E340 associated with drought-induced acidification in these experiments and other field observations (Scott et al. 1998) are also consistent with acidification driving changes in DOC. Both SUVA and E340 are associated with the larger coloured humic fractions, which carry greater charge and are more sensitive to changes in acidity in terms of their solubility. Acid addition experiments in humic rich lakes have observed changes in coloured humic DOC following long-term acidification linked to either Al-mediated coagulation and flocculation or chemical oxidation to CO2 (Donahue et al. 1998); and laboratory batch experiments have shown similar decreases in SUVA associated with declining DOC following increases in acidity and solute strength (Clark et al. 2011).

As noted above, some studies have suggested that decreased DOC concentrations in response to water table draw-down could be because consumption and respiration of DOC during heterotrophic respiration is greater than the net rate of DOC production (Fenner et al. 2005; Pastor et al. 2003; Scott et al. 1998). If correct, increased CO2 fluxes should be greater than decreased DOC. Mass balance calculations suggest this mechanism is plausible. However, without isotopically labelling the DOC pool, it is difficult to determine the source of DOC respired, making it difficult to quantify relative changes in DOC due to consumption and/or solubility. Studies have shown DOC released from peats to be resistant to degradation (Kalbitz et al. 2003). However, the quality of organic matter between sites is likely to differ due to differences in surface vegetation, particularly the presence or absence of Sphagnum spp. (Table 1); and studies have shown marked difference in the biodegradability between DOC leachates from Sphagnum spp. and Eriophorum spp. (Wickland et al. 2007).

Rapid increases in DOC concentrations following water table recovery have also been attributed to increased biological activity and net DOC production (Scott et al. 1998), especially in relation to increased activity of phenol-oxidase which is thought to play a key role in removing the inhibitors of the enzymes responsible for peat decomposition (Freeman et al. 2001b; Toberman et al. 2008). In unsaturated soils, rewetting has been associated with increased mineralisation and CO2 fluxes, possibly due to microbial priming by hydration and lysis of dead microbial cells as a substrate for respiration (Borken and Matzner 2009). However, rapid increases in DOC observed in peat pore waters here following rewetting were not accompanied by increased CO2 production rates, suggesting this is not a biologically mediated process (Blodau and Moore 2003). Whilst some of the CO2 released during rewetting could be attributed to organic carbon consumption during SO4 and NO3 reduction (Goldhammer and Blodau 2008), these fluxes were a small proportion of total values. Therefore, increased biological production of DOC during rewetting seems unlikely to fully account for the observed increases in DOC concentrations. However, it is worth noting that surface vegetation was absent during the experimental manipulations, therefore DOC production by ‘priming’ where heterotrophic decomposition is stimulated by inputs of labile plant exudates was excluded (Freeman et al. 2004).

Overall, these data suggest that conflicting results between studies conducted at a similar scale are not solely due to the methodological approach, but could reflect inherent differences in biogeochemical processes operating at each location due to different site characteristics. Therefore, further understanding of various processes operating between specific locations is needed to reconcile reported differences in observations between studies. There is a limit to how far these difference can be reconciled by focusing on process responses at one location, such as comparing blanket peatlands to minerotrophic fens (e.g. Clark et al. 2005; Eimers et al. 2008b; Preston et al. 2011), although these comparisons help to identify key research questions.

While it is clear that decreased stream flows during drought periods will undoubtedly limit the net transport of DOC from soils to surface waters (Eimers et al. 2008b), changes in biogeochemical cycling that occur within peat pore waters due to hydrologically driven changes in soil water chemistry will also affect the short and long-term exchange of carbon from peatlands with both the atmosphere and freshwater systems. As declining acid deposition will decrease the available S and N pool, the severity of drought-induced acidification driven by increased SAA concentrations is likely to decline over time (Tipping et al. 2003). However, our data suggest that where S and N deposition are low, drought-induced acidification can still occur in response to decreased SBC concentrations. Given the link between acid deposition and changes in Sphagnum distribution (Berendse et al. 2001), and the difference in the bioavailability of DOC leachates from different peatland vegetation substrates (Wickland et al. 2007), the interactions between acid deposition and vegetation change could result in substantial variations in the overall DOC responses to drought events between locations due to differences in DOC bioavailability and solubility.

Conclusions

Our re-examination of the relationship between drought-induced acidification and DOC dynamics during simulated droughts events in six peats broadly supported our earlier findings from one peat site—Moor House (Clark 2005; Clark et al. 2006). Peat pore waters consistently showed a decrease in DOC following water table draw-down and an increase in DOC after rewetting. There was no evidence that changes in water volume had a direct influence on altering DOC concentrations, although ‘evapoconcentration’ of base cations in the most acid impacted peats occurred.

Inverse relationships between increased SO4 and decreased DOC concentrations were found in all peat cores, although these varied in magnitude between sites. Instead, the relationships between ANC and DOC were more consistent between the different peats. In the most acid impacted peats, ANC decreased because of oxidation of large amounts of S to SO4 (combined with N to NO3 in two peats). By contrast, in the least acid impacted peats, ANC decreased because of the decline in base cations. As the ANC takes into consideration differences in both acidifying anions and neutralising base cations between sites, this parameter is therefore a more effective measure of drought-induced acidification between different locations.

Previous analysis of soil water chemistry and soil respiration suggest that decreased DOC could be attributed to both drought-induced acidification of pore water and microbial consumption and respiration. While biotic and abiotic controls could regulate DOC during periods of water table draw-down, the absence of a pulse in soil respiration following rewetting in our study suggest that abiotic chemical controls were the principle cause of rapid increases in DOC and DOC quality following rewetting.

While reduced stream flow ultimately limits DOC transport from soil to surface waters during drought periods (Eimers et al. 2008b), our study highlights the important balance between changes in soil water chemistry and net DOC production/consumption within peat soils that supply DOC to stream waters. Correctly understanding abiotic factors, like chemical and physical processes alongside biotic DOC production and consumption is necessary to enable their inclusion within dynamic peatland C models.