Peatland ditch blocking has no effect on dissolved organic matter (DOM) quality

The globally widespread drainage of peatlands has often been shown to lead to increased concentrations and fluxes of dissolved organic carbon (DOC) in streams and rivers. Elevated DOC concentrations have implications for carbon cycling, ecosystem functioning, and potable water treatment. Peatland rewetting, principally through ditch blocking, is often carried out with the expectation that this will reduce DOC concentrations. Uncertainty still remains as to whether drainage, or its reversal via ditch blocking, will also lead to changes in the molecular composition of DOC/dissolved organic matter (DOM), which have the potential to affect downstream processing and treatability of U.K. drinking water supplies. To investigate this question, we used a replicated experiment consisting of 12 parallel ditches on an upland bog and took samples of ditch water, pore water, and overland flow water for 4 years. After a brief preblocking baseline period, eight ditches were blocked using two methods. A complementary suite of optical metrics, chemical measurements, and analytical techniques revealed that ditch blocking had no consistent effect on DOM quality, up to 4 years after blocking. Where significant differences were found, effect size calculations demonstrated that these differences were small and would therefore have minimal impact upon water treatability. Furthermore, some differences between ditches were evident before blocking took place, highlighting the need for robust baseline monitoring before intervention. Based on our results from a hillslope‐scale experiment, we were unable to identify clear evidence that peatland ditch blocking will deliver benefits in terms of DOM treatability in potable water supplies, although we also did not find any evidence of short‐term deterioration in water quality during the restoration period. We conclude that, although peatland restoration can be expected to deliver other benefits such as reduced carbon loss and enhanced biodiversity, it is doubtful whether it will lead to improvements in drinking water treatability.

Increased exports of DOC are problematic for multiple reasons.
DOC in fluvial systems can be mineralized to CO 2 , thereby contributing to atmospheric CO 2 concentrations (Cole et al., 2007;Jones, Evans, Jones, Hill, & Freeman, 2016). Additionally, DOC affects light attenuation and can therefore affect the functioning of aquatic ecosystems (Karlsson, Byström, Ask, Persson, & Jansson, 2009), and DOC can bind with trace metals, some toxic (Lawlor & Tipping, 2003;Rothwell, Evans, Daniels, & Allott, 2007). Furthermore, DOC adds colour and odour to potable water which must be removed due to aesthetic concerns (Mitchell, 1991). Finally, when chlorinated during potable water treatment, high concentrations of DOC can lead to the formation of harmful disinfection by-products, including trihalomethanes (THMs; Chow, Kanji, & Gao, 2003). THM concentrations in potable water are strictly regulated; for example, the European Union limit is 100 μg L −1 for total THMs, whereas the World Health Organisation recommends concentration limits for individual THMs of between 60 and 300 μg L −1 (Werner, Valdivia-Garcia, Weir, & Haffey, 2016). Increased DOC concentrations therefore present a problem to water companies due to the cost associated with its removal and penalties for exceeding regulatory limits (Brooks, Freeman, Gough, & Holliman, 2015).
One potential method that has been proposed to reduce DOC concentrations in freshwaters is the rewetting of drained peatlands (Wilson et al., 2011). A recent synthesis by Evans, Renou-Wilson, and Strack (2016) suggests that drainage increases DOC in concentrations and  , and that individual storm events can increase DOC by~10 mg l −1 (Austnes, Evans, Eliot-Laize, Naden, & Old, 2009). It is worth noting that the majority of blanket bog rewetting studies compare drained and rewetted treatments, due to the fact that most U.K. blanket bog has been managed by drainage, grazing, or burning, leaving little undisturbed bog left (Ramchunder et al., 2009).
As well as uncertainty regarding the effectiveness of ditch blocking at reducing DOC concentrations, it is still largely unclear whether rewetting has the capacity to alter the chemical composition of dissolved organic matter (DOM), which might be expected due to hydrological changes (Thacker, Tipping, Gondar, & Baker, 2008). Numerous methods can be used to investigate DOM character in relation to drinking water quality. These include fluorescence and absorbance measurements, which are relatively fast and accessible techniques and provide information on DOM character; for example, the degree of aromaticity, humification, or autochthonous DOM. Alongside these, there are analytical approaches such as nuclear magnetic resonance spectroscopy, high-performance size-exclusion chromatography (HPSEC), and Fourier-transform infrared spectroscopy (Matilainen et al., 2011). These analytical methods provide increased detail on DOM composition, even down to the molecular level, but they require specialised and expensive equipment. Specific UV absorbance (SUVA), whereby DOC concentration is normalised to light absorbance, usually measured at 254 nm, is used as a proxy for DOM aromaticity (Weishaar et al., 2003) and is perhaps the commonest absorbance metric used within the water industry, having been in use for several decades (Edzwald, 1993). Ratios of absorbance at different wavelengths are also used, such as E2:E3, E2:E4, and E4:E6, which relate to DOM composition and molecular weight (Peuravuori & Pihlaja, 1997;Summers, Cornel, & Roberts, 1987). In addition to ratios, there are absorbance metrics that require measurements at multiple wavelengths, such as spectral slopes, whereby the slope of the absorbance spectrum is a function of DOM molecular weight (Helms et al., 2008). Although absorbance measurements are unable to provide fine-scale resolution on DOM structure, they can be used to reliably detect differences or changes in composition (Erlandsson, Futter, Kothawala, & Köhler, 2012).
Various studies have used some of the above techniques to investigate the effects of rewetting on DOM quality, although only a handful have been on blanket bogs (Table 1). Most studies use only a few metrics, however, and contrasting results are common. For example, Wallage et al. (2006) noted lower E4:E6 ratios in pore water of ditch-blocked peat when compared with drained peat. They also recorded higher specific absorbance at 400 nm in rewetted peat. Conflictingly, Wilson et al. (2011) suggested that ditch blocking decreased specific absorbance at 400 nm and increased E4:E6. By using a broader suite of DOM metrics, it might be possible to reduce uncertainty regarding the effects of rewetting.
The lack of more detailed knowledge from field studies is important, as DOM quality directly affects the treatability of potable water and the The aim of our study was therefore to test whether blanket bog rewetting would lead to alterations in DOM quality, which could result in associated changes in the treatability of drinking water.
To do this, we took pore water, ditch water, and overland flow water samples from an upland bog where a series of parallel ditches had either been blocked or left open as controls. Samples were collected on an approximate monthly basis for 4 years. We used optical metrics, chemical measurements, and analytical techniques to investigate the chemical composition of DOM. The main part of our analysis was a post-rewetting comparison of DOM from blocked and unblocked ditches, although we also had some limited pre-rewetting data which allowed us to test whether any differences in DOM between ditches existed before ditch blocking.

| Field site
The study was carried out on a hillslope on the Migneint blanket bog,

| Water sampling
Sampling of ditch water started in October 2010. Samples were collected from water flowing over v-notch weirs  or from pools behind weirs if there was no flow. Pore water sampling started in January 2011 (giving 1 month of baseline data) from piezometers placed 2-3 m west of each ditch. Piezometers were made of polyvinyl chloride with intakes at 10-15 cm depth, and pore water was collected using plastic tubing attached to a syringe. On each sampling visit, piezometers were emptied of water and allowed to recharge overnight, before samples were collected the next day.
Overland flow water sampling started after rewetting in January 2012. Overland flow water was collected using polyvinyl chloride crest-stage tubes (Holden & Burt, 2003). These comprised tubes that were sealed at both ends but with holes just above ground level to collect surface flow. For each ditch, two crest-stage tubes were sited 2 m west of the ditch, with another two 4 m west of the ditch. The water from these was bulked to give one sample, representing a Phenolic concentrations were measured for the first year using a method adopted from Box (1983). A total of 0.25 ml of sample was pipetted into a clear microplate well to which 12.5 μl of Folin-Ciocalteau reagent was added, followed by 37.5 μl of Na 2 CO 3 (1,340 Da). The mobile phase was milli-q water buffered with phosphate (2 mM KH 2 PO 4 + 2 mM K 2 PO 4 ·3H 2 O) to pH 6.8.

| Statistics
The amount of pre-and post-rewetting data for the various DOM metrics are summarised in

| Variation in DOM quality
When considering the three E ratios, the temporal variation was

| Effect of ditch blocking on DOM quality
There was no significant difference in DOC concentrations between open, dammed, or reprofiled ditches for ditch water, pore water, or overland flow water (Table 3). Additionally, there was no evidence of a consistent effect of ditch blocking on DOM quality as measured by all UV-vis metrics showed that temporal fluctuations were larger than any differences between treatments (Figures 2-8). Although a significant difference was present in ditch water phenolic:DOC before blocking, no significant difference was found for ditch water or pore water after blocking (Figure 8, Table 4).

UV-vis (summarised in
Two THMs were detected in ditch water samples from July 2012: CHCl 3 and CHBrCl 2 . Concentrations of CHCl 3 were two orders of magnitude larger than those of CHBrCl 2 . There was no significant difference in THM concentrations between open, dammed, and reprofiled ditches (Table 5). HPSEC showed that there was no difference in molecular weight of DOC in ditch water, pore water, or overland flow water after ditch blocking ( Figure 9). Chromatograms showed a minor high molecular weight peak at~4 min, followed by a dominant high molecular weight peak at~7.5 min, with a lesser peak at~9 min. In the pore water and overland flow water samples, there was a minor low molecular weight peak at~14 minutes. The height difference between chromatograms is due to differences in DOC quantity (i.e., concentration) rather than quality.

| Effect of ditch blocking on DOM quality
Our results indicate that peatland ditch blocking had no effect on DOC concentrations or on the composition of DOM in pore water, ditch water, or overland flow water, after nearly 4 years of rewetting, when measured by various metrics of organic matter quality. Although some significant differences were observed in UV-vis, the size of these effects was statistically shown to be small, very small, or <very small (Table 4) and would therefore have no substantive effect (detrimental or beneficial) on the treatability of potable water. Furthermore, some significant differences between ditches were observed both before and after ditch blocking took place. This finding emphasises the importance of collecting prerestoration baseline data, if only for a short period, due to the fact that small but significant differences in organic matter quality can occur over relatively small spatial scales, and in a visually homogeneous ecosystem. We argue that it is thus inadvisable to conclude that ditch blocking has resulted in reductions in DOC concentrations when no baseline data are available (e.g., Wallage et al., 2006) because differences in DOC quality and quantity could instead be driven by microscale variation in DOC processing. When considering all the UV-vis metrics measured here, temporal variation was larger than between-treatment variation, with seasonal variations in E4:E6 being particularly pronounced (Figure 2). Wilson et al. (2011) reported a change in mean E4:E6 from 2.7 to 5.8 in drains and streams after ditch blocking but did not report an associated E4:E6 for unblocked ditches. It is therefore impossible to confidently ascribe such a change to a rewetting effect, considering that temporal changes in our study for both blocked and open ditches ranged between 2 and 14 in ditch water.
We found no difference in the ratio of phenolics to DOC concentration in ditch water or pore water (Figure 8). If phenolic concentrations increased following hydrological changes (e.g., Freeman, Lock, & Reynolds, 1993) then this would have detrimental impacts on potable water, as phenolics are particularly difficult to remove by coagulation methods Tomaszewska, Mozia, & Morawski, 2004). Likewise, the lack of difference in apparent molecular weight distributions that we observed using HPSEC between blocked and unblocked ditches (Figure 9) is important, as changes to molecular weight can affect water treatment processes (Collins, Amy, & Steelink, 1986 Frimmel, 2012,). To our knowledge, ours is the first study to report field measurements of THMFP following ditch blocking (Table 5).
THMFP concentrations were in the same range as other field measurements from blanket peat (Delpla et al., 2015;Gough et al., 2012;Valdivia-Garcia, Weir, Frogbrook, Graham, & Werner, 2016) and did not differ between blocked and unblocked ditches. THM concentrations have been found to co-vary with molecular weight (Gang, Clevenger, & Banerji, 2003), and the lack of significant difference in THMs is expected due to the near identical chromatograms generated by HPSEC. Additionally, SUVA has been shown to relate to THM formation (Weishaar et al., 2003), and no strong effects of rewetting were detected for SUVA ( Figure 5).

| Reasons for lack of a rewetting effect on DOM quality
Detailed data concerning the dynamics of DOC quantity at this site are presented by Evans et al. (2018) but in summary (Table 3) show Note. Nonidentical letters mark significant differences for pairwise comparisons between treatments; comparisons were made only for each metric between treatments, that is, not between water types, metrics, or preblocking and postblocking periods. Dashed lines show either where no significant difference was detected or, for overland flow water, where no preblocking data were available. For most tests where significant differences occurred between all treatments both p and effect size were identical. The exception was S 350-400 postblocking pore water, hence two treatment statistics are reported there. a p ≤ 0.05, b p < 0.01, c p < 0.001. no effect of rewetting on concentrations or fluxes of DOC. Changes in water table following blocking were variable but very small (<2 cm; Holden et al., 2017), but rewetting did lead to increases in wetindicator testate amoeba suggesting the creation of wetter conditions across the site (Swindles et al., 2016). However, there was no difference in extracellular enzyme activity in the year following ditch blocking (Peacock et al., 2015), and Francez, Gogo, and Josselin (2000) noted a lag time in changes to microbial communities following restoration of a harvested raised bog. The lack of strong microbial or hydrological changes could be one reason for the associated lack of effect on DOM composition, as the water table was close to the bog surface despite the presence of open ditches .
Recent experimental work at this site, and in the wider peatland surrounding it, has led to the hypothesis of "self-rewetting" (Williamson et al., 2017). Briefly, the digging of a ditch leads to a lowering of the water table, which results in peat oxidation/compaction lowering the peat surface, and thus, the peat surface becomes wetter again (Williamson et al., 2017;Young, Baird, Morris, & Holden, 2017); in more actively drained and cultivated peatlands, this "self-rewetting" is avoided by repeated lowering of the drainage ditches (Kuntze, 1986). Such a process would explain the modest increases in water-  Worrall, Armstrong, and Holden (2007) recorded increases in ditch DOC and specific absorbance at 400 nm in the 10 months after ditch blocking. Conflictingly, a study of a Finnish peatland found little difference in DOM quality between a control and forest harvested area (Kiikkilä, Smolander, & Ukonmaanaho, 2014). Similarly, although Glatzel et al. (2003) found that ecosystem disturbance in the form of

| Assessment of methods
The majority of previous studies on blanket bog ditch blocking have reported only a few metrics of DOM quality alongside DOC concentrations and/or fluxes; for example, E4:E6 and/or specific absorbance at 400 nm (Wallage et al., 2006;Wilson et al., 2011;Worrall et al., 2007).
Although UV-vis measurements are undoubtedly useful, this technique has been described as a "black box," with little understanding of exactly how DOM composition affects light absorbance (Stedmon & Álvarez-Salgado, 2011). The expanded number of metrics that we used, which included additional optical and chemical measurements, has facilitated a more robust investigation of the effects of blocking on DOM quality.
It is perhaps noteworthy that the mesocosm study by Gough et al. (2016) that also measured optical and chemical metrics similarly found no evidence that ditch blocking improves water treatability. By complementing both basic (E ratios) and advanced (spectral slopes) UV-vis metrics with measurements of phenolics, THMFP, and molecular weight distributions (derived by HPSEC), a more complete picture of whether differences in water chemistry are significant and/or meaningful can be obtained.

| Wider implications
Analyses of peat chemistry from our site suggest that it is representative of other U.K. blanket bogs  and the type of ditching is also commonly found elsewhere . It can therefore be hypothesised that ditch blocking will not cause catchment-scale improvements or reductions in water quality at other upland sites, with no real-world effects for water treatment operations in the years immediately following rewetting especially when the local hydrological change (e.g., water table position) after rewetting is minimal. The caveat must be stated that such a lack of response will be noted at sites where ditches are relatively shallow or the blanket bog still relatively wet (due to the aforementioned self-rewetting effect). However, effects on DOM quality may be observed if ditch  blocking results in larger rises in water tables than those that we observed (e.g., 2.6 cm noted by Holden et al. (2011) for blanket peat).
Alternatively, results from studies on fens and raised bogs elsewhere in Europe have found changes in DOM composition after 10-20 years of rewetting (Frank et al., 2014;Höll et al., 2009), and it could be that such differences will eventually manifest themselves at our site. The difficulty then arises of untangling restoration effects on DOM from the effects of long-term environmental perturbations such as climate change and recovery from acidification that will also exert controls on DOM composition (Ekström et al., 2011;Ritson et al., 2014).

| CONCLUSIONS
We found no difference in the quality of DOM in the first 4 years following ditch blocking on an upland blanket bog, using a suite of both optical and chemical measurements. Ditch blocking is thus unlikely to lead to either positive or negative changes in the treatability of potable water at our site. Although the lack of improved treatability may prove disappointing to water utilities, the null result can also be perceived as a "no regrets" outcome if other benefits can be obtained from ditch-blocking, for example, reducing peak flows (Ballard, McIintyre, & Wheater, 2012), reducing sediment loss (Holden, Gascoign, & Bosanko, 2007), improving biodiversity (Carroll et al., 2011;Hannigan, Mangan, & Kelly-Quinn, 2011), restoring bog vegetation (Bellamy, Stephen, Maclean, & Grant, 2012), and improving landscape aesthetics (Bonn et al., 2014), without concern that these aims will interfere with potable water supplies.