3.1. Eh and metabolic processes
We observed a strong relation between Eh and porewater concentrations for Fe(II), phosphate, SO42−, S2− and CH4 across all five locations (Fig. 1, Fig. S2, Fig. S3). For Mn(II) and NO3− this relation was less persistent over study sites. Furthermore, we found that Eh variability for SSI plots is less than in the reference plots, with less observations (2% and 10% of the measurements for SSI and reference plots respectively) with higher Eh values (> 500 mV, Fig. 1, S2, S3, S4a,b). The lower variability in Eh matches the lower variability in (Fig. S5), and higher groundwater levels (on average + 11 cm) observed in the SSI plots.
Above an Eh of 400 mV, we find a higher variance and median for NO3− concentration in our dataset (Fig. S7a). However, combining data from all different measurement locations might obscure site-specific patterns and therefore we focus on detected breakpoints at individual plots. Measurements collected on reference plots, with high Eh values during summer, dominate the pattern of high NO3− concentrations above and low concentrations below the estimated breakpoints. This was especially true for the reference plots of Zegveld and Assendelft that had high NO3− concentrations (breakpoint at Eh = 665–685 mV and 366–433 mV respectively, Fig. S2). For the stable groundwater sites (SSI) no relevant breakpoint could be established, as Eh measurements were mostly below the expected threshold for NO3− reduction (Fig. S3). However, infrequently high NO3− concentrations were measured in the SSI plots indicating that NO3− can occur in strongly reducing environments. The interquartile range (25–75%) for mean breakpoints of all plots was 371–610 mV (n = 5, meaning that the model convergence was successful for 5 out of 10 locations) for plots that had Eh values spanning the theoretical breakpoint range in Table 1). This result corresponds well to the expected threshold for NO3− reduction (500 mV, Table 1) (Fig. 1).
For Mn(II), we found higher Mn(II) concentrations when Eh is below the Mn reduction range (Eh = 430–630 mV, Table 1) in the reference plots of Vlist, Zegveld and Aldeboarn (Fig. S2) and in the SSI plot in Vlist (Fig. S3). The interquartile range including the mean breakpoint estimated for each plot was 415–568 mV (n = 3), corresponding well to the theoretical breakpoint values in Table 1 (Eh = 430–630 mV). The absence of clear patterns in Mn(II) dependency on Eh for some research locations could be due to influences of carbonates and pH on Mn concentrations as described by Yu et al. (2007).
The evident patterns in Fe(II) concentration over the redox spectrum closely match the patterns we encounter for P concentration (Fig. 1, Fig. S7c, d). This finding matches previous studies that found P mobilization after reduction of Fe(III) and P immobilization after oxidation or Fe(II) (Smolders et al., 2006; Kjaergaard et al., 2012; Lamers et al., 2012; Van Der Grift et al., 2014). The lower and upper quantiles that were found for all detected Fe(II) and dissolved P mean breakpoint values of individual plots ranged from 90 to 356 mV (n = 14) and are in agreement with the Eh threshold for iron reduction of 170 mV in Table 1.
Peak S2− concentrations were found between − 50 mV and 0 mV (Fig. S2, S3, S7e, S8), which was within the expected range in Eh for SO42− reduction (Eh = -30 mV, Table 1). S2− was mostly detected between − 100mV and 150 mV. Concentrations of S2− above this Eh range were neglectable due to oxidation of S2− with other electron acceptors or precipitation with Fe(II) (i.e. pyrite formation, S6a, Smolders et al., 2006). Ratios of SO42−:Cl− between − 100 and 100 mV tended to drop below the standard marine ratio of 0.14 for mostly SSI plots (Fig. S2, S9), which indicates long term SO42− reduction leading to decreasing SO42− concentrations and decreasing SO42−:Cl− ratios. Lower SO42− concentrations in SSI plots compared to reference plots indicate a depletion of the SO42− stock due to more prevalent reducing conditions (Fig. S8). This depletion coincides with the low availability of Fe(II) in the SSI plot in Assendelft as compared to the reference plot (Fig S10), as it is likely that the Fe(II) has mostly precipitated with S2− (Smolders et al., 2006). Within the reference plot, Fe(II)S2 can become oxidized when aerobic conditions occur due to lower water tables as opposed to the SSI plots where oxidation is less likely due to higher groundwater levels (Knorr et al., 2009; Li et al., 2012). Due to precipitation with Fe(II), the Eh range with high S2− concentrations does not directly represent the range of SO42− reduction. The interquartile range of all mean breakpoint estimations for individual plots for SO42−:Cl− and SO42− is Eh = -49- 118 mV (n = 15).
In general, CH4 concentrations decreased with increasing Eh conditions. This could be explained by higher CH4 oxidation or inhibition of methanogenesis (Fig. 1, Fig. S7f). As a result of gas transport from deeper soil layers, the CH4 concentrations in porewater may not fully reflect methanogenesis at the measured depths. The minimum Eh was lower at the SSI plots compared to the reference plots and the CH4 concentrations at the SSI plot exceeded the concentrations at the reference plot at each monitoring location, except for Zegveld (Fig. S2, S3). Methanogenesis is generally limited during SO42− reduction (Fig. S6c) but both processes can co-occur (Sela-Adler et al., 2017; van Dijk et al., 2019). Similarly, methanogenesis can occur while Fe(III) reduction takes place (Fig. S6b, de Jong et al., 2020). This may be the case in the reference plot in Aldeboarn where SO42− reduction is absent and CH4 concentrations increase towards more reducing conditions from Eh = 200 mV (Fig. S2). The breakpoint estimation of CH4 varied widely, across all plots the interquartile range of the mean breakpoint was 65–381 mV (n = 7). This does not match the expected threshold for methanogenesis (Eh = -140 mV, Table 1), possibly due to gas transport from deeper soil layers and methanogenesis co-occurring with Fe(III) or SO42− reduction.
We aimed to assign Eh ranges to the dominant metabolic pathways in agricultural peat soils (Table 2, Fig. 2). Note that differing (overlapping or non-consecutive) minimum and maximum values for neighboring dominant metabolic process ranges were averaged although the estimated ranges were already highly consecutive. We found few samples with Eh values within the Eh range of breakpoints detected for NO3 and Mn(II) (366 and 685 mV, Fig. S4c), which can be explained by a high rate of NO3 consumption (Bougon et al., 2011; de Jong et al., 2020) and the low amount of available Mn(II) (< 100 µmol/L) in most soils. Therefore, and because of overlapping breakpoint estimations for NO3− and Mn(II), we decided to merge these two metabolic processes in our further analyses. Above this range we expect oxygen presence, which would be characterized by a low amount of dissolved Mn and higher NO3− concentrations as compared to anaerobic conditions (Bougon et al., 2011; van Beek et al., 2011). Our results showed that Mn(II) concentrations will approach zero between an Eh of 700–750 mV. This range approaches the mean estimated breakpoint for NO3− reduction for the reference plot in Zegveld, with high concentrations above 685 mV. According to Stępniewski et al. (2005), soils are oxygenated when the Eh exceeds 650 mV (normalized from pH 7 to pH 5.5). Here, we assume that oxygen becomes the dominant electron acceptor above an Eh of 700 mV. This Eh range for oxygen is characterized by a three times higher density of observations compared to the ranges for Mn(IV) and NO3− reduction (Fig. S4, 0.64 and 0.21 observations mV− 1 respectively) and thus we consider it better defined than the range for Mn(IV) and NO3−. Due to the variety of pore size classes and the heterogeneity of micro-niches in the soil, we expect that different metabolic processes can co-occur at similar depths as was found by Knorr et al. (2009) and suggested by Canfield and Thamdrup (2009). The estimated Eh ranges in Table 2 represent strong patterns of dissolved electron acceptors and/or reaction product concentrations along the Eh spectrum and indicate dominant (bulk soil) reduction processes.
Table 2
– Summary statistics of in Eh breakpoint analysis and Eh ranges for reduction that were used within further analysis for solutes that indicate reduction of electron acceptors (at pH 5.5). Thermodynamic thresholds in Table 1 are also included.
Dominant electron acceptor | Inorganic C | SO42− | Fe(III) | Mn(IV) | NO3− | O2 |
Breakpoint analysis first quantile | - | -49 | 90 | 415 | 371 | - |
Breakpoint analysis median | - | -20 | 282 | 548 | 412 | - |
Breakpoint analysis third quantile | - | 118 | 356 | 568 | 610 | - |
Breakpoint analysis n | - | 13 | 12 | 3 | 5 | - |
Lower Eh threshold (mV) | -250 | -49 | 104 | 364 | 364 | 700 |
Upper Eh threshold (mV) | -49 | 104 | 364 | 700 | 700 | 950 |
Thermodynamic threshold from Pourbaix diagrams (mV)* | -140 | -30 | 170 | 430** | 500 | - |
* Values are presented in Table 1. ** Note that the first thermodynamic reduction threshold of Mn(IV) is at Eh = 630 mV.
3.2. Continuous field Eh profiles
By applying the ranges that were found for reduction of electron acceptors (Table 2) to interpret Eh measurements over time and depth (Fig. 3, S11-20) we demonstrate and quantify the dynamics of metabolic processes in the soil, which can be compared for different research locations and plot treatments. Where groundwater levels are closer to the surface we find Eh values below 700 mV that indicate less oxygen intrusion due to a higher moisture content and consequentially low air intrusion. Low groundwater levels in summertime (> 50 cm below surface) coincide with Eh values above 700 mV and make the soil susceptible to aerobic peat respiration. The depth of Eh indicating oxygen presence (700 mV, Table 2) does not directly follow the groundwater level dynamics, which for example falls below 0.8 m in the reference plot Zegveld but is limited to a depth of 0.55 m in the dry summer of 2020 and 0.3 m in the wet summer of 2021 (Fig. 3a). This could be due to limited air diffusion related to a low amount of air-filled pores present near the groundwater level. Furthermore, we also find that deep oxygen intrusion can oxidize Fe(II) to Fe(III) that can be used during high groundwater levels in winter, as visualized in Fig. 3a where Fe(III) reduction plays an important role in the top 0.75 m of the soil profile in wintertime. The extent of iron reduction is lower in Fig. 3b, as oxygen intrusion was limited due to high groundwater levels. Probably, most Fe(III) has been reduced due to prevailing saturated conditions. After the oxidized conditions in the reference plot during the summer of 2020, we find that the lower extent of the iron reduction zone shifts upward stepwise during the wet summer of 2021 and the spring of 2022 (Fig. 3a). Due to prevailing reducing conditions in the Zegveld plot with higher groundwater levels (Fig. 3b) we expect that SO42− reduction and methanogenesis occur below a depth of 0.3–0.4 m. Low SO42− stocks at the SSI plot (Fig. S10) and the drain inlet of ditch water seems to cause the dominant metabolic system to switch from SO42− reduction to methanogenesis during warmer summer conditions. It is important to note that the Eh measurements have been interpolated linearly with depth for visualization purposes while Eh conditions are not likely to change linearly with depth.
After summing up the total number of days with a particular dominant metabolic pathway for peat mineralization within our complete timeseries, we can infer which processes are active at different measuring depths – independent of temporal dynamics (Fig. 4, S22). The observations in Fig. 3 and Fig. S11-20- are summarized in such a way that we can characterize important long-term metabolic states of all research locations. We find that above mentioned observations of deep iron reduction in the reference parcel, and enhanced SO42− and methanogenesis in the plot with high groundwater levels are immediately visible in Fig. 4. In addition, we also see that Eh suggests a higher amount of methanogenesis when measured closer the irrigation tube. This pattern is consistent over all measuring sites where SSI established higher groundwater levels (except Rouveen) and where higher concentrations of methane (> 100 µM) were measured (except Vlist, S22). Moreover, we find that the depth of oxygen intrusion seems limited by applying SSI for each site except the site where SSI does not establish a higher groundwater level (Rouveen, S22). As opposed to groundwater level or oxygen concentration measurements, soil Eh-pH measurements provide a more complete overview of all metabolic processes that are related to peat decomposition and related greenhouse gas emissions.
3.3. Research implications
We found that mean CH4 concentrations can be related to the relative amount of time that methanogenesis and SO42− reduction are expected to occur according to interpreted Eh measurements (Fig. 5). On the contrary, mean SO42− concentrations were higher when SO42− reduction was more dominant over time than methanogenesis (Fig. 5). However, the variation that is observed in CH4 concentrations cannot be explained by Eh only, but is probably a result of substrate quality, groundwater discharge, the microbial community, soil temperature, soil gas transport (from deeper layers), methanotroph activity and pH (e.g. Updegraff et al., 1998; Smolders et al., 2002; Freitag et al., 2010). When accounting for the availability of degradable organic matter and oxidation (with SO42− or O2) of CH4 by methanotrophs (Nazaries et al., 2013), CH4 emitting soils could be identified by using Eh measurements. Besides CH4 emissions, Eh conditions also function as a proxy for peatland CO2 emissions (Yu et al., 2007). Insights in the metabolic pathway dynamics in the different research sites arising from this paper can thus also be coupled to the aboveground GHG emission measurements in the future to evaluate effects of GHG mitigation measures.
Natural microbial competition generally establishes chemical zonation along the soil depth gradient in which each zone consists of dominant microbial communities that have a particular metabolism (Canfield et al., 2005; Canfield and Thamdrup, 2009; Lamers et al., 2012). Due to the relatively low hydraulic conductivity in peat soils and consequential low groundwater flow in natural peatlands (with high groundwater levels throughout the year) this chemical zonation would be rather stable (Boomer and Bedford, 2008; Canfield and Thamdrup, 2009). This situation is most comparable to the SSI sites where we found the least varying and highest groundwater levels that generally result in a more pronounced metabolic zonation and a lower amount of oxygen intrusion in the topsoil (at and above 0.5 m depth, Fig. S21). We find that such a situation is characterized by depletion of sulphates, and consequentially by a higher rate of methanogenesis (Fig. S10). In contrast, we found more variable topsoil redox conditions over space and time in drained peat soils without SSI that fit the more fluctuating groundwater levels (Fig. S21). The variability of redox conditions is important as it may impact peat decomposition by aerobic-carry-over-effects as following: (1) electron acceptors formed during aerobic conditions (such as iron-oxides and sulphate) can be reduced during anaerobic conditions, potentially increasing the rate of anaerobic peat decomposition during saturation as suggested by Li et al., (2012). Furthermore (2), it is likely that recalcitrant organic material is decomposed into more labile compounds during aerobic mineralization and these labile compounds can be consumed during anaerobic conditions (Stępniewski et al., 2005). Consequently, we hypothesize that fluctuating groundwater levels and dynamic topsoil redox conditions generally enhance the rate of peat decomposition, also during saturated conditions.
Here, we have shown that measurements of Eh and pH can be directly linked to dynamics in reduction-oxidation processes and solute composition in peat soils. By combining these relationships with continuous Eh-profile measurements we found that sites with seasonally variable groundwater levels also showed more variable oxidation processes, whereas sites with stable and higher groundwater levels showed less oxygen intrusion and dominance of methanogenesis and sulfate reduction. We hypothesize that fluctuating Eh conditions may enhance the rate of peat decomposition due to aerobic-carry-over-effects. Furthermore, we found that the time of Eh suggesting methanogenesis is related to porewater methane concentration and suggest that Eh measurements can be coupled to aboveground GHG measurements. Based on these mechanistic insights, Eh and pH measurements can now be used to locally optimize different GHG mitigation strategies for drained peatlands. Additionally, Eh and pH measurements can be valuable parameters to measure as part of a nation-wide GHG emission monitoring network.