Global importance of methane emissions from drainage ditches and canals

Inland waters including lakes, rivers, wetlands and ponds play an important role in the global carbon cycle and climate system because they emit large amounts of greenhouse gases (GHGs) including methane (CH₄), carbon dioxide (CO₂), and nitrous oxide (N₂O) to the atmosphere [1–5]. CH₄ is an important GHG emitted from inland waters; this is because many waterbodies have high fluxes and because CH₄ has a 100 year global warming potential (GWP) of 28 [6] resulting in a substantial radiative forcing effect [4, 7–9].

Total inland water CH₄ emissions have been estimated as 100–200 Tg CH₄ yr⁻¹ [10, 11], equivalent to ∼30% of global CH₄ emissions [11]. Globally and regionally important emissions of CH₄ have been documented from reservoirs [4], lakes [12], streams and rivers [13], small natural ponds [8], and artificial ponds [14]. It would therefore appear that CH₄ emissions from the major categories of inland waters have been quantified. However, emissions from another, potentially important category of inland waters have yet to be quantified at a global scale: ditches. Here, we present a synthesis of new and existing flux data from ditches showing that ditches release large amounts of CH₄ and that these emissions are significant on national and global scales.

The Ramsar Convention includes ditches as human-made wetlands under the category 'canals and drainage channels or ditches' [15]. Artificial irrigation channels are also classified as 'irrigated land'. However, there are no strict Ramsar definitions for these categories, which are a 'broad framework to aid rapid identification of the main wetland habitats' [15]. Ditches, canals and channels are created to: (a) improve the productivity of wet soils through drainage; (b) reclaim land from flooded areas; (c) move water through a landscape for agricultural or industrial use; (d) reduce agricultural soil erosion, or (e) remove stormwater in urban areas. For our purposes, we define ditches as constructed linear waterways, whilst acknowledging that their physical characteristics and function may vary widely (figure 1). Thus, irrigation channels and water supply canals in arid regions fall within our definition. We recognise that not all linear waterways included in our definition can strictly be described as drainage ditches in the traditional sense (i.e. they do not all drain land/lower the water table), however we use this term for simplicity, and because drainage ditches comprise the large majority of our dataset, as well as the majority of mapped linear water features globally (e.g. global ditch length is several orders of magnitude larger than global canal length; see SI text 9 (available online at stacks.iop.org/ERL/16/044010/mmedia)). Finally, it is worth noting that terminology is inconsistent and the words 'ditch' and 'canal' may even be used interchangeably; for example, larger drainage channels in SE Asian peatlands are navigable and often referred to locally as 'canals'.

Zoom In Zoom Out Reset image size

The spatial extent of ditches has been recognised on both small and large landscape scales. For instance, in a boreal catchment ditches doubled the length of the stream network [16], whilst in Great Britain the total length of ditches is estimated at ≈604 000 km which is more than twice that of streams and rivers combined (≈267 000 km) [17]. Data are scarce, but other estimates of drainage networks exist, and show that ditches can occupy a significant proportion of the waterscape in some countries and regions across the world [18–22].

Ditches possess particular characteristics which means that they cannot be assumed to function identically to most streams and rivers (although similarities may exist with streams in low-relief agricultural landscapes). Firstly, ditches do not in general follow natural topographic gradients and do not have natural catchments. They are often, but not always, situated in low-lying, low-relief landscapes which result in low flow rates. This creates a set of conditions favourable to CH₄ production and emission, namely: (a) the accumulation of sediment; (b) the development of anoxia; and (c) the growth of emergent plants. Ditches within agricultural and urban landscapes may also receive high inputs of labile organic matter and nutrients, providing a substrate for methanogenesis [21]. In such environments, CH₄ is produced autochthonously in anaerobic bottom waters or sediments by methanogenesis at rates dependent on an array of factors including temperature and labile carbon content [23]. In addition, ditches may act as conduits for the emission of CH₄ which is produced under anaerobic conditions in adjacent terrestrial environments [24]. This process could be particularly important because, unlike other inland waters, ditches are specifically created to drain (i.e. to receive lateral transfers of water from adjacent land), and at an extremely high drainage density [20, 25] when compared to natural streams. In summary, ditches have a particular set of hydrological, chemical and morphological characteristics that differentiate them from natural streams in ways that favour CH₄ production and emission. It has been suggested that ditches may have higher CH₄:CO₂ ratios than streams, albeit based on a small dataset (see SI text 1) [13].

Given that CH₄ emissions from streams have been recognized for over 100 years [26, 27], there have been surprisingly few quantitative assessments of CH₄ fluxes from ditches until the last two decades. Roulet and Moore [24] performed one of the first studies, in which they suggested that peatland ditches 'depending on the flow rate, depth, and morphology, could provide an ideal environment for the transport and in situ production of CH₄'. They found that drainage resulted in CH₄ uptake by the terrestrial peat surface, but that the ditches themselves emitted large amounts of CH₄. Thus, when scaled across the entire peatland they suggested that, depending on the spacing between ditches, drainage could result in a net increase in landscape CH₄ emissions. The potential of peatland ditches to emit CH₄ was confirmed by other studies [28, 29]. Ditches were subsequently incorporated into IPCC guidance for assessment of GHG fluxes from drained organic soils as part of the 2013 Wetlands Supplement [30]. This assessment identified a limited dataset of 19 publications reporting ditch CH₄ emissions from drained peat soils [31]. Because of the growing appreciation of the important role of CH₄ in regulating the global climate, the number of relevant studies has increased since 2013. However, to date no comprehensive, global-scale analysis of CH₄ emissions from ditches has been undertaken. Furthermore, whereas stream (and lake) emissions are a natural component of the global CH₄ budget (albeit subject to potential anthropogenic influences) ditches are constructed, therefore all associated CH₄ emissions must be considered anthropogenic. Our aim here is to draw attention to the potential importance of ditch fluxes in the global CH₄ cycle, and to highlight the lack of data from ditches which are a near ubiquitous landscape feature across climate zones and land use classes.

Here, we update and extend estimates of CH₄ emissions from ditches by including new literature and previously unpublished data. In total, we were able to calculate annual emissions for 64 unique ditch sites encompassing a wide range of ditch morphologies, latitudes and land uses. The majority of studies used the floating chamber method, and so we focused on these, and excluded a small number of flux estimates based on dissolved CH₄ concentrations. For all studies we extracted data on land-use, eutrophication status, and climate.

Reported CH₄ fluxes ranged from 0.1 to 386 g CH₄ m⁻² yr⁻¹ with a mean (±standard error) of 64.6 ± 11.1 g CH₄ m⁻² yr⁻¹. Emissions from ditches in tropical and sub-tropical zones are significantly higher than those from temperate and boreal zones (figure 2(A)) and higher fluxes are associated with higher temperatures, as shown by a significant correlation between CH₄ flux and temperature (figure 2(C)). Fluxes were significantly higher from continuously inundated ditches compared to intermittently dry ditches which can sometimes show negative fluxes (SI figure 1, SI text 8). There was no significant (p = 0.09) effect of nutrient status, but at sites where mean annual temperature (MAT) was above 15 °C eutrophic fluxes were noticeably higher (figure 2). Fluxes from eutrophic ditches also spanned the greatest range, with some eutrophic ditches exhibiting very low fluxes.

Zoom In Zoom Out Reset image size

Despite variation within each land-use category, mean fluxes tended to increase with increasing land use intensity in the order: natural land and forest ≈ peat extraction < low-intensity grassland < urban land < cropland ≈ high-intensity grassland (figure 3). Differences were significant between natural land and forest, and both cropland and high-intensity grassland.

Zoom In Zoom Out Reset image size

More flux estimates were obtained from ditches draining peat soils (n = 50) compared to mineral soils (n = 14). There was no significant difference between the two groups, with means of 58 ± 11 g CH₄ m⁻² yr⁻¹ and 87 ± 33 g CH₄ m⁻² yr⁻¹ for ditches draining peat and mineral soils, respectively (Mann–Whitney U test, p = 0.91). We compared these ditch emissions against the reduction in terrestrial emissions following drainage, assuming that ditch surface areas occupy, on average, 3% of the landscape (so called Frac_ditch = 0.03), and find that, on average, ditch emissions equate to 4% (error = 2%–8%) and 18% (15%–22%) of the terrestrial CH₄ reduction in mineral and peat soils respectively. The discrepancy in percentages between soil types arises primarily due to the large IPCC emission factors given for wetlands on mineral soils, which are substantially larger than fluxes from undrained peatlands.

A comparison shows that, on an areal basis, ditches emit similar amounts of CH₄ to reservoirs, streams and tropical wetlands (figure 4).

Zoom In Zoom Out Reset image size

Ditches have been mapped and total lengths calculated for some countries [18, 22], but for many regions of the world there is a lack of data on ditch densities. Additionally, although our synthesised CH₄ data covers a reasonable geographic spread, it is biased towards European and North American locations. Considering this, a detailed upscaling of global CH₄ emissions (i.e. aggregating by land use, soil type, climate zone, etc) from ditches would be flawed. Instead, and with the aim of calling attention to the possible global magnitude of these emissions, we present a rough 'back of the envelope' upscaling. For this, we took the average annual flux from our dataset and a global estimate of drained land area, and calculated total ditch surface areas using values of drainage ditch areas from the IPCC [30] and literature searches. This gives a mean flux of 3.46 Tg yr⁻¹ (0.61–10.5 Tg yr⁻¹) (table 1).

Our analysis shows that mean areal ditch CH₄ emissions are as large as those from other well characterised systems (inland waters and wetlands) that are considered to be important at national or global scales. Our global estimate is 3.5 Tg yr⁻¹ (0.6–10.5 Tg yr⁻¹) which is 0.1%–1.8% of global CH₄ emissions or 1.0% (0.2%–2.9%) of global anthropogenic CH₄ [11]. For comparison, our estimate is greater than emissions from permafrost soils (1 Tg yr⁻¹) [11] which receive considerably more attention compared to 'dull as ditchwater' drainage channels. On a 20 year and 100 year time frame using a sustained-flux GWP [32], global emissions respectively equate to 332 and 156 Tg CO₂ eq yr⁻¹. Thus, ditch emissions shift the balance and reduce the expected CH₄ sink from drained soils, which is estimated globally as 38 Tg CH₄ yr⁻¹ for all unsaturated oxic soils [11]. Ditch emissions could be even more important on a national scale. For example, in densely drained countries such as the Netherlands and Finland ditches could emit the equivalent of 7% and 9% of national anthropogenic CH₄ emissions (see SI text 10).

Multiple studies now show that ditch emissions can dominate the CH₄ budget of drained landscapes and, in some situations (e.g. if ditch density is high, or within-field water table is low) can outweigh any terrestrial CH₄ uptake [33–35]. Evidence of high emissions from ditches within drained landscapes counters the commonly held view, previously embedded in IPCC land-use emissions reporting guidance [36], that drained landscapes have zero CH₄ emissions (see SI text 11). Furthermore, a recent review of cropland drainage ignored the emissions from ditches and concluded that drainage could lead to a net reduction in CO₂ equivalents, and thus have a beneficial impact on climatic warming [37] (although wetland drainage can enhance terrestrial CO₂ and N₂O emissions [30]). We acknowledge that ditch emissions can be implicitly included in landscape budgets measured by eddy covariance towers, assuming that the ditch network within the tower footprint is representative of that in the wider landscape, but assume this is an uncommon occurrence (see SI text 11). Instead, we have shown that omission of ditch CH₄ emissions from landscape-scale GHG budgets will lead to underestimation of the overall global warming impact of drainage-based land management.

We found that ditch emissions increased with rising MAT, and that emissions were significantly higher in sub-tropical/tropical climate zones, reflecting the temperature-dependence of CH₄ production [38]. Higher fluxes were also associated with more intensive land uses, such as cropland. Annual fluxes were lower from ditches that periodically dried out, and this is likely due to sediments become aerobic and inhibiting methanogenesis [23]. For sites with MAT above 15 °C, eutrophic ditches had higher fluxes but, surprisingly, no significant difference (p = 0.09) was found between mean fluxes from different nutrient statuses (although this may be due to a relatively small sample size resulting in an underpowered test). Taken together, we infer that CH₄ emissions will tend to increase with temperature, in the absence of other biogeochemical constraints such as intermittent drying, low substrate supply, or a low nutrient status.

We consider our upscaled estimate to be conservative for the following reasons. Firstly, we did not quantify channel area in irrigated land or ditches in urban land. Secondly, our estimate is based on chamber measurements which typically do not include plant mediated fluxes, and may not capture ebullitive fluxes (see SI text 2) which are temporally and spatially highly variable [39]. We also note that ditches can emit CO₂ and N₂O [40] which further increase GHG forcing associated with drained landscapes [41]. N₂O emissions may be particularly large from ditches in drained wetlands that have been converted to croplands and subjected to extensive nitrogen fertilization. However, at this stage, there are insufficient data to enable these emissions of CO₂ and N₂O to be upscaled [42].

Our work also carries limitations. Firstly, our synthesised data is biased towards ditches draining peat soils (n = 50) compared to mineral soils (n = 14). Thus, if there is a difference in ditch fluxes between the two soil groups our upscaling could also be biased. There was no significant difference in mean fluxes between the two groups and we also found studies where, although annual fluxes could not be calculated, there was evidence for large CH₄ emissions from ditches in mineral soils. For example, an extremely high flux of 250 g CH₄ m⁻² d⁻¹ was reported from a ditch on a dairy farm in the Netherlands [43] and a UK survey detected a biogenic CH₄ hotspot near an urban canal [44]. There is therefore no reason to assume that emissions from mineral soils are smaller than those from peat soils, but future measurements of ditch fluxes on mineral soils should be a key priority to robustly test this assumption. It is feasible, however, that CH₄ production pathways may vary: in situ production may dominate in well-drained mineral soils, whilst lateral inputs of CH₄ into ditches are likely to also contribute in poorly-drained soils. A second limitation is the lack of information on ditch lengths and maps of drained land, which leads to a large uncertainty when upscaling. Note that the largest uncertainty in our calculations comes from ditch area: our upper estimate is ≈7.5 times larger than the lower, whilst for ditch CH₄ flux the upper estimate is ≈2.3 times larger than the lower. Drained agricultural land, the majority of which is on mineral soils, occupies the largest cumulative area of drained land, but few measurements of Frac_ditch exist for this land (see SI text 7). Mapping ditch networks within this land should therefore by another key priority for future work. Thirdly, there is an absence of nighttime ditch CH₄ measurements. It has been shown that CH₄ concentrations in channelized eutrophic streams [45] and ditches [46] can be approximately double those of daytime concentrations, due to diel fluctuations in oxygen concentrations. If this is a widespread phenomenon in ditches then our calculated annual fluxes may be low. Fourthly, ditch CH₄ measurements in boreal regions are often restricted to the growing season (SI table 1). Non-growing season emissions from shallow aquatic systems, marshes and peatlands in the boreal comprise, on average, 16% of annual emissions [47], and thus annual CH₄ emissions from ditches may be underestimated in our synthesis. Finally, the effect of aquatic plants is uncertain. Emergent vegetation in streams can act as a pathway for CH₄ emission and enhance sediment CH₄ concentrations, whilst floating plant species can act to increase or decrease net aquatic CH₄ emission depending on local conditions [48].

Increasing global populations and associated food and fibre demands are likely to lead to increased land drainage, ditch density and nutrient loadings [37]. Our results suggest that this will lead to significant increases in global ditch CH₄ emissions. Ditch emissions may also increase under a warmer climate. However, there are opportunities to mitigate these emissions. Nutrient enrichment increases CH₄ fluxes in other aquatic ecosystems such as lakes and reservoirs [4, 7], and there is no reason to assume that this effect does not occur in ditches. Reducing nutrient runoff into ditches may offer a pathway to mitigate CH₄ emissions to the atmosphere, whilst also improving water quality, thereby enhancing human and aquatic ecosystem health. Adaptive design and maintenance of ditch networks might also be an option to limit emissions. For example, clearing vegetation would prevent plant-mediated emissions and additionally remove a source of labile carbon for methanogens, although we acknowledge a conflict here with the biodiversity value provided by ditch vegetation. Maintaining deep, stable water levels may also limit emissions by minimising the warming of sediments and thereby reducing CH₄ production.

In some regions rewetting of drained wetlands is being pursued as a climate change mitigation measure. Rewetted sites often possess ditch networks that may be partially blocked or infilled, or that are still used for active hydrological management, and high CH₄ fluxes have been measured from such ditches [49, 50]. If CH₄ emissions from relict ditches are of a similar magnitude to emissions from the adjacent wetland area, as assumed by the IPCC [30], then this is not a cause for concern. However, for intermediately rewetted systems such as wet meadows and paludiculture, it is feasible that ditch emissions may make a significant contribution to the overall GHG balance of the restored landscape. At present, a lack of data means that no conclusions can be drawn on this topic, but it is a highly relevant area of scientific and policy uncertainty.

5.1. Experimental design

5.2. Comparison of terrestrial CH₄ reduction and ditch CH₄ emissions

5.3. Global and national ditch emissions

5.4. Statistical analysis

All data that support the findings of this study are included within the article (and any supplementary files).

No specific grant funded this work. MP acknowledges funding from Formas (Grant 2020-00950). DB contributions were supported by European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme (Grant Agreement No. 725546). JA acknowledges funding from Carl Tryggers Stiftelse. VG is grateful for support from the AXA Research Fund. We thank Marcus Klaus and Dongqi Wang for providing information on the sampling sites in their respective papers, and Frances Manning for providing data ahead of publication. We thank Jesper Rydén and Gareth Harvey for their advice on different approaches to statistically analysing the data. We thank Andy Baird and Sophie Green for providing ebullition data that were used in an earlier draft of this paper. We thank one anonymous reviewer and Tim Moore for their constructive review comments that helped to improve the manuscript.

The initial collation of ditch emission data was undertaken by CE as part of his contribution to the 2013 IPCC Wetland Supplement (Inland drained organic soils chapter), which was expanded by CM and MP for the Flooded Lands chapter of the Intergovernmental Panel on Climate Change 2019 Refinement to the 2006 Guidelines for National Greenhouse Gas Inventories (43), for which CL, CE, JH and DB were Lead Authors, and MP, SK and AG were Contributing Authors. MP, JA, AG, MK, SK, AJV contributed data. AG and MP created the figures. All authors searched for literature data and provided ideas. MP wrote the paper with input from all co-authors. The authors declare no competing interests.