Effects of dry-wet cycles on nitrous oxide emissions in freshwater sediments: a synthesis

Survey methodology

A search for relevant published studies focusing on dry-wet cycling and N₂O emissions from sediments was performed using the literature database search engines Web of Science and Google Scholar, with a combination of search terms: dry* wet* OR dry* rewet* OR dry* flood* OR drought AND “nitrous oxide” OR ”N₂O” AND sediment*. The asterisk (*) enables key word variations. The reference lists within the selected papers was also examined. The initial screening of titles and abstracts excluded papers considering N₂O emissions from soils exclusively, and from sediments other than from freshwater ecosystems. The following studies were excluded from this synthesis: soils (including wetland soils and peatlands), saline or brackish water environments, aquaculture ponds, and rice/paddy field studies. The studies include both field and laboratory observations. Rewetting of dry sediments include events caused by rainfall (natural or simulated), flooding and water table fluctuations. Drying events include water drainage from the sediment and sediment desiccation. N₂O emission peaks were considered as a direct response to sediment drying and rewetting events. Last database search: 28 of September 2020.

Effects of drying on N₂O emissions in freshwater sediments

Both physical mechanisms and microbial processes have been suggested to contribute to the initial N₂O pulse during sediment desiccation. The increase in gas diffusivity (i.e., the decrease in mass transfer resistance) and the lack of physical trapping, related to water content and oxygen fluctuations, determine the magnitude and temporal development of N₂O outgassing from the sediments (Reverey et al., 2016). Higher N₂O emissions are also attributed to periods of enhanced microbial activity due to coupled nitrification-denitrification occurring at the boundary of oxic-anoxic environments (Baldwin & Mitchell, 2000) (Fig. 2).

Drying provides suitable aerobic conditions for nitrification to occur and in turn, the NO${}_3^{-}$ and NO${}_2^{-}$ produced are reduced by denitrification in anoxic microsites, harbored within the oxic parts of the sediment. Substrates for N₂O production are also made available from the release of intracellular solutes from microbial cell lysis, and from mineralization of OM, including previously protected organic matter from aggregate disruption (shrinking), due to sediment desiccation (Baldwin & Mitchell, 2000; Borken & Matzner, 2009).

In general, studies in both lotic and lentic systems suggest that initial sediment desiccation stimulates N processing rates such as N mineralization and nitrification. A MC/WPFS threshold exists after water drainage and/or the onset of sediment drying when denitrification is also enhanced and coupled to nitrification (Fig. 2). This was observed in sediments from intermittent rivers and streams, where N mineralization and nitrification were found to be stimulated during initial sediment desiccation, while denitrification significantly decreased as %WFPS declined (Gómez et al., 2012). Between the onset of drying and the increase in net nitrification rates, nitrification and denitrification occurred simultaneously (coupled nitrification-denitrification), as evidenced by the NO${}_3^{-}$–N concentrations and enhanced denitrification rates detected during this time (Gómez et al., 2012). NO${}_3^{-}$–N tends to accumulate in the sediment with ongoing drought, while NH${}_4^{+}$–N tends to decrease (Fig. 2). During the first days or weeks of sediment desiccation, the increase in NO${}_3^{-}$–N content and loss of NH${}_4^{+}$–N most likely results from continuous nitrification throughout the dry period (Gómez et al., 2012; Arce et al., 2014; Merbt et al., 2016; Arce et al., 2018; Table 1). Furthermore, and similarly to highlighted observations for soils (Borken & Matzner, 2009), the duration of the desiccation period also controls the N concentration in the sediments (Gómez et al., 2012), with consequences for the N₂O/(N₂O + N₂) ratios, since denitrifiers prefer to transfer electrons to NO${}_3^{-}$ rather than N₂O.

Enhanced N₂O fluxes have been observed in different freshwater systems during initial sediment desiccation (Fig. 2). In an intermittent stream, Arce et al. (2018) found an N₂O peak average value after 11 days of sediment drying, after which N₂O fluxes decreased. This pulse in emissions was attributed to physical N₂O evasion favored by increased gas diffusivity (Fig. 2). The sources of N₂O fluxes were assumed to be denitrification under wet to saturated conditions, and nitrification as O₂ concentration increased. Dry sediments can be active spots for N₂O emissions even at low moisture contents (Pinto et al., 2020), underlining the importance of including dry phases in the N emission estimates.

In a seasonal floodplain lake, higher N₂O emissions were measured during the transition from flooded to dry sediment conditions (Koschorreck, 2005; Table 1). The majority of N₂O loss (94% of the total N₂O emission) occurred during an initial drying period of high N processing activity (59% of the total denitrification). After an initial lag phase, most likely limited by NO${}_3^{-}$ supply via nitrification, denitrification activity and N₂O flux peaked. During an intermediate state between flooding and sediment drying, a patchy distribution of oxic and anoxic microsites was observed. This high spatial heterogeneity in redox conditions increases the area of the oxic-anoxic boundary where coupled nitrification-denitrification can occur, explaining the high denitrification activity detected during this period. The denitrification rates during desiccation exceeded those during re-flooding. A previous study in the same system detected higher N₂O and denitrification rates in exposed sediments, and with greater distance to the waterline (Kern et al., 1996; Table 1). Namely, O₂ availability was the main factor controlling N₂/N₂O ratios. The coupling of nitrification and denitrification was also suggested, due to the lack of NO${}_3^{-}$ accumulation in the sediments.

Similar observations were made in a seasonal pond (Fromin et al., 2010; Table 1). There, potential denitrification was strongly stimulated during early desiccation of the sediment, while potential nitrification was stimulated during the late desiccation period. The potential for N₂O emissions was higher at sites with a longer dry period, pointing to different response magnitudes of the microbial processes depending on the length of the dry period. Both nitrification (at depth) and denitrification enzyme activities remained very high in sediment after 1 month of drought, and communities were able to maintain their enzymatic pool even under severe environmental conditions. Short-term variations in sediment water content strongly affected microbial functioning, but not the metabolic structure of the community.

During an extreme drought event in a hydroelectric reservoir, N₂O production potentials were measured along one transect, from flooded sediments to drier sediments at the water margin (Jin et al., 2016; Table 1). N₂O production was significantly higher in recently exposed sediments than in flooded or dry sediments, related to a temporary boost in OM processing in the drying sediment (Fig. 2). The increase in N₂O production as a consequence of enhanced O₂ availability during drying suggested that N₂O also derived from aerobic processes such as nitrification.

Effects of rainfall events on N₂O emissions in freshwater sediments

Recent studies recognize that as in soils, the Birch effect, a pulse in decomposition, mineralization and release of inorganic nitrogen and CO₂, is also observed in sediments (Birch, 1958; Wilson & Baldwin, 2008; Arce et al., 2019; Marcé et al., 2019). In addition, similar responses to a wetting pulse have been observed for nitrification and denitrification processes (Arce et al., 2014; Arce, Sánchez-Montoya & Gómez, 2015), with enhanced N and C mineralization increasing nitrogen turnover rates and N₂O emissions (Fromin et al., 2010), following a desiccation period. The main reasons leading to increased fluxes are enhanced microbial activities (including nitrification and denitrification), induced by a pulse of labile substrates (cell lyses, osmoregulation), and physical mechanisms such as gas displacement (Baldwin & Mitchell, 2000; Fierer & Schimel, 2003; Borken & Matzner, 2009; Congreves et al., 2018) (Fig. 3). The release of previously protected organic matter due to aggregate slaking may also contribute to the nutrient flush upon rewetting (Denef et al., 2001; Shrestha et al., 2012).

Nitrification and denitrification response to rewetting following dry periods of varying duration have yielded contrasting results. In an experimental stream, nitrification and denitrification rates recovered to levels equal or greater than those in the wet controls within 1 day of being rewetted in sediments dried less than 7 days, but failed to fully recover in sediments dried more than 7 days (Austin and Strauss, 2011; Table 1). On the other hand, the negative effect of desiccation duration was not detected upon rewetting of sediments from temporary streams (Arce et al., 2014; Arce, Sánchez-Montoya & Gómez, 2015; Table 1). A rapid recovery of N processing rates (within hours) to pre-drought levels or higher was observed, particularly for denitrification (Arce et al., 2014). After 1 h of sediment rewetting, there was a significant stimulation of denitrification rates and a peak after 24 h was detected (Arce, Sánchez-Montoya & Gómez, 2015). These results differed from those of temperate streams, which reported higher lag time between rewetting and recovery of denitrification rates to the pre-dry levels (Zaman & Chang, 2004). The lack of a clear microbial response to the rewetting stress suggests a lag phase between physiological adjustments or resilience of microbial communities to changing environmental conditions. A high export of NO${}_3^{-}$ (accumulated during sediment drying) was also observed during the first flushing events (Arce et al., 2014; Arce, Sánchez-Montoya & Gómez, 2015). Previous studies have highlighted the influence of the timing and duration of the drying-rewetting events on N availability and process dynamics (Gómez et al., 2012).

In ephemeral stream channels, N₂O emissions were significantly higher after a rainfall pulse following an extensive period of dry conditions (Gallo et al., 2014; Table 1). During an initial stage, the extremely high instantaneous fluxes were attributed to physical displacement of accumulated gases, nevertheless the potential contribution of immediate microbial responses warrants additional study. High fluxes were further sustained for some hours by rapid microbial activation as a response to moisture increase. The main driver of N₂O fluxes was water availability, followed by net mineralization and nitrification rates, and C and N stocks.

On the other hand, a rainfall pulse had a negative impact on N₂O emissions in sediments from a temperate intermittent river, as its magnitude increased (4 mm vs. 21 mm rainfall) (Arce et al., 2018). The effect of rewetting on N₂O fluxes was further modulated by the duration of the preceding drying period (6 and 9 weeks). Intense rainfall (21 mm) caused a decrease in N₂O fluxes, while following modest rainfall (4 mm) emissions increased immediately after the water pulse 24 h later. The decrease of the average fluxes was most likely related to physical trapping preventing gas exchange and displacement during the monitoring period (24 h). Comparing N₂O pulses immediately after drainage and rewetting, the former pulse was of smaller magnitude (≈ 5 vs. 17 µg N₂O-N m⁻² h⁻¹ increase in emissions, respectively) when considering the rewetting pulse after the longer preceding drying period under moderate rainfall (Arce et al., 2018; Table 1). These changes in N₂O emissions stress the importance of considering the previous hydrological history, as biogeochemical responses and emission sources are affected by the sediment’s preceding conditions (e.g., Gómez et al., 2012; Wilson & Baldwin, 2008; Reverey et al., 2018).

In seasonal pond sediments, rewetting of dry sediments after a rainfall event induced a peak of labile organic carbon and inorganic nitrogen and enhanced microbial activity (Fromin et al., 2010). The percentage of N₂O emission by denitrification remained high (60%) after rewetting and similar to the one measured during the dry period, suggesting a lack of recovery of the physiological ability of the denitrifying community to reduce N₂O, and an impact of drought on the community structure. Potential nitrification rates also slightly increased. Preceding dry conditions also affect sediment microbial activity. Sediments subject to frequent dry-wet cycles displayed a stronger stimulation of potential microbial activities (denitrification enzyme activity) and higher potential for N₂O emissions after rewetting compared to sediments undergoing infrequent drying phases. For instance, the potential denitrification rates after rewetting were highest at sites with longer dry periods, namely at the pond edges. The potential denitrification rates triggered by the rewetting pulse were comparable to the maximum rates measured during drying (Fromin et al., 2010).

Effects of flooding and water level fluctuations on N₂O emissions in freshwater sediments: aquatic-terrestrial transition zones

Specific regions present in both temporary and perennial systems may experience recurrent water level fluctuations (flooding, water drawdown). These include gravel bars and the parafluvial zone in lotic systems (rivers and streams) and the littoral zones of lentic systems (ponds, lakes and reservoirs), in which potentially large areas of sediments are exposed to the atmosphere during water level fluctuations, and are potential sources of N₂O. Similarly to the previous sections, the hydrological and O₂ dynamics will affect N₂O fluxes through physical mechanisms and microbial processes (Figs. 2 and 3). Rises in the water table and flooding create anoxic conditions, stimulating denitrification and N₂O production, while water drawdown increases aerobic processes and N₂O diffusion to the surface. For example, Morse & Bernhardt (2013) found that nitrification contributed significantly more to N₂O yields in sediments after drainage than denitrification. Water level drawdown and drying of sediments may temporarily increase NH${}_4^{+}$ and NO${}_3^{-}$ concentrations upon inundation. N losses might be high during initial rewetting and slow down with prolonged inundation (Arce, Sánchez-Montoya & Gómez, 2015; Reverey et al., 2016). Intensified water level fluctuations may also accelerate dissolved organic carbon mineralization, thereby increasing the availability of organic carbon to fuel denitrification (Shi et al., 2020).

High N turnover rates and locally high N₂O efflux were found in frequently flooded river gravel bars, including co-occurrence of high-rates of both, nitrification and denitrification (Shrestha et al., 2012; Table 1). During drying phases, accumulated or freshly produced NO${}_3^{-}$ was quickly denitrified, as suggested by higher potential denitrification rates compared to gross nitrification. The high spatial variability of N₂O emissions and a highly variable N₂:N₂O ratio, arising from the variety of processes involved in the production and consumption of N₂O, makes it difficult to correlate between fluxes and processes (e.g., potential denitrification). Furthermore, Shrestha et al. (2014) observed that flood pulses promoted a substantial temporary increase in N mineralization on frequently flooded gravel bars (Table 1). The cause of the transient increase in microbial activity was attributed to the input of fresh, highly available organic N (non-structured allochthonous material) and enhanced turnover of previously protected N from aggregate disruption during flood pulses. In turn, an increase in N mineralization stimulated coupled nitrification-denitrification during the drying phase, as oxic-anoxic conditions were favorable for the co-occurrence of both processes (spatially heterogeneous aggregate structure). Thus, areas exposed to short intensive floods with fast-flowing water can be considered as hot spots, and the dry phases after the flood as hot moments of N transformations, which may lead to temporarily high N₂O emissions. Similarly, McIntyre et al. (2009) observed that in sediments from an intermittent stream, large flood pulses resulting in full saturation cause a slower release of mineralization products, compared to small pulse events that stimulate a rapid cycle of C and N mineralization–immobilization. In sediments subject to 50% saturation, >90% of the mineralized N accumulated within the first 7 d of incubation, compared to only 48% when fully saturated (100% saturation).

In the parafluvial zone, N₂O emissions were measured along a hydrological gradient covering rarely-flooded and frequently flooded sediments, as well as elevated non-flooded soils in the floodplain (Pinto et al., 2020). Fluxes were higher during the period of higher water level variability, when sediments were subject to frequent drying and rewetting, compared to the desiccation period and post-flood period. Higher fluxes were attributed to a tight coupling between nitrification and denitrification during this period (Table 1). N₂O emissions were generally higher in sediments exposed to frequent dry-wet cycles, compared to non-flooded and rarely-flooded sediments. Emissions in the frequently flooded sediments were mainly C driven, and the variation in N₂O fluxes was explained by the quality of the dissolved organic matter pool, which can influence microbial processes such as denitrification (Barnes, Smith & Aiken, 2012).

The microbial use of available organic C is known to be affected by dry-wet cycles. The flow of allochthonous derived C through the microbial community has been tracked in sediments exposed to water level changes (Weise et al., 2016). Terrestrially derived allochthonous organic C was the main source of C under hydrological fluctuations, whereas in permanently dry sediments, the main source of OC originated from C stored in the sediments. Thus, allochthonous OC is more intensively used under dry-wet conditions showing that repeated dry-wet cycles promote OM degradation and increased substrate availability (Weise et al., 2016). Results from these studies indicate that even in perennial systems, certain regions are potential biogeochemical hotspots for GHG emissions during periods of water fluctuation, namely the parafluvial zone, which is a highly active and dynamic area.

The short-term effects of flood pulses on N₂O fluxes were quantified in artificial riparian marshes along a similar transverse gradient covering dry, alternate dry-wet and permanently flooded sediment conditions (Hernandez & Mitsch, 2006; Table 1). Permanently flooded plots showed the lowest N₂O fluxes, however, it remains unclear whether this was attributed to higher N₂:N₂O ratio due to complete denitrification or to restricted diffusion transport from the presence of standing water. Intermittently flooded sediments and sediments exposed to air showed significantly higher N₂O emissions during and after the flood pulse than before the pulse. Such pulsing conditions might have enhanced coupled nitrification-denitrification, but also caused suboptimal conditions for denitrification (inhibition of N₂O reductase by O₂), decreasing the N₂:N₂O ratio. At lower water table, N₂O emissions were most likely the by-product of nitrification, as seen by the high concentration of NO${}_3^{-}$ and low concentration of NH${}_4^{+}$. Under absence of standing water, N₂O diffusion to the atmosphere was unrestricted, increasing N₂O fluxes. Similarly, N₂O emissions were detected in exposed sediments during the low water period, but were never detected in flooded lake sediments, as nitrification and NO${}_3^{-}$ supply to denitrification were restricted (Kern et al., 1996; Table 1). As air enters the sediment during low water, O₂ conditions are favorable for OM mineralization, increasing NH${}_4^{+}$ and subsequent oxidation by nitrification. Since NO${}_3^{-}$ did not accumulate in the pore water of exposed sediments, N₂O was presumed to derive from coupled nitrification-denitrification.

Other studies in lakes highlight littoral zones as potential N₂O sources, where drying and wetting cycles are a significant factor controlling denitrification and N₂O emissions. Denitrification and N₂O emission rates were found to be higher in the exposed regions and increased with the degree of sediment dryness, whereas in the permanently flooded littoral zone rates were near zero (Huttunen et al., 2003; Akatsuka & Mitamura, 2011; Table 1). High emissions were attributed to enhanced N turnover and nitrification in the aerobic sediment surface, and/or denitrification in anaerobic microsites. Furthermore, potential denitrification rates and potential N₂O production rates are higher in sediments from littoral zones compared to land soils (Wang et al., 2012; Table 1). Nevertheless, a higher proportion of N₂O was transformed into N₂ in the sediments, and attributed to the short supply of NO${}_{\mathrm{x}}^{-}$. If drying events are expected to intensify in the future, these results suggest that sediments from littoral zones have higher potential than soils to emit N₂O, as NO${}_3^{-}$ and O₂ availability increase in exposed sediment with drying.

Hydroelectric reservoirs experience intensified water level fluctuations due to peak and off-peak electricity demand, creating frequent wetting–drying cycles over potentially large water drawdown areas. In these areas, potential denitrification rates were strongly related to the period of water level rising-falling cycle (Shi et al., 2020). Both denitrification rate and N₂O fluxes were higher closer to the water edge, where large amounts of N₂O were produced and emitted at a high rate (Table 1). Water level fluctuations further enhanced hyporheic exchanges, which accelerated organic carbon mineralization, increasing the carbon source supply for denitrification.

Kettle holes have likewise been identified as significant sources of N₂O (Merbach et al., 2002; Table 1). N₂O emissions in sediments negatively correlated with the water level along a depositional transect exposed to different drying frequencies, with higher water level slightly decreasing emissions. The legacy of dry-wet cycling is similarly important. Nitrogen loss from these ecosystems is potentially highest in hydrologically less stable sediments (frequently exposed pond margins, desiccated for longer periods), which were depleted in NH${}_4^{+}$ during sediment exposed conditions, likely due to nitrification (Reverey et al., 2018; Table 1). The hydrological history (predominantly inundated, intermediate/occasionally dry, and sediments frequently exposed to dry-wet cycles) in kettle holes created distinct microbial habitats, reflected by changes in the microbial community structure. During sediment exposure (higher oxygenation), DNRA potential was higher in sediments less exposed to drought, while potential denitrification was higher in frequently exposed sediments. Nevertheless, denitrification was always higher than DNRA. The nitrification potential increased towards the more frequently exposed sediments as a result of higher oxygenation due to longer exposure to the atmosphere (Reverey et al., 2018). These results highlight the importance of considering the biogeochemical legacy of dry-wet cycles when predicting ecosystem responses to changing conditions (inundation vs. sediment exposure), namely N₂O emissions.

Future research

In the reviewed studies, the experimental design and the different temporal scales between and within studies made it difficult to compare the dynamics of N₂O fluxes as each phase progresses (wet to dry or dry to wet phase), and hot moments may have been overlooked. Addressing this gap may be of significance when modeling ecosystem responses to dry-wet cycles, as the direction of the fluctuation may influence N₂O predictions. For instance (in soils), it has been observed that the intensities of N₂O emission peaks can be higher during the drying phase compared to the wetting (near-saturation) phase, and correlated with the amount of water drained, although with shorter peak duration (Rabot, Hénault & Cousin, 2014). In sediments, the driving role of water fluctuations on N₂O peak fluxes is evident, and future research should address the temporal development during both drying and rewetting phases in more detail, capturing the rapid changes at early stages. Therefore, efforts should be made to better assess the timing of these events, in order to estimate the relative contribution of hot moments and significance of these peaks to cumulative emissions. This will allow for a more accurate integration of emissions from inland waters. N₂O uptake in aquatic systems also merits further investigation, as the capacity of aquatic systems to act as sinks may change in the future with changes in net precipitation (Chapuis-Lardy et al., 2007; Kroeze, Bouwman & Slomp, 2007).

The importance of the dry-wet cycling legacy is also recognized, as it shapes the microbial community composition and function. Future research efforts tackling N₂O emissions during dry-wet cycles should likewise elucidate how environmental changes (i.e., drying and rewetting) shape the network interactions in microbial communities, as well as the functional potentials of the active microbial associations (RNA sequencing and ecological network analysis, e.g., see Fuhrman, 2009; Zhou et al., 2010; Cong et al., 2015). Studies have suggested that some microbial associations are found to alter predictably, and changes in activity under shifting conditions may be anticipated (Fuhrman, 2009). Focusing beyond the abundance of functional groups may prove valuable to develop and improve modeled N₂O emission estimates further, namely in temporary systems. A comprehensive study should include distinct inland waters exposed to different frequency, intensity and duration of drying and rewetting (perennial, intermittent, ephemeral, and episodic systems; according to definitions in Datry, Bonada & Boulton, 2017). Although it is challenging to predict the timing of events in these systems, this would allow for a broader view on the impacts of frequency and duration of dry-wet cycles on N₂O emissions. Laboratory incubations, mesocosms and microcosms studies are an asset to understand the evolution and determine the pathways for N₂O production and emissions during dry-wet cycles (e.g., ¹⁵N stable isotopic methods). Most studies assume the controlling N₂O production pathway depending on the predominant redox conditions, and more efforts should be made to determine the relative contribution of each production pathway to N₂O emissions during each phase (wet to dry or dry to wet). Nevertheless, these studies tend to oversimplify ecosystem interactions, namely between surface and subsurface sediments, potentially excluding significant N₂O input from deeper sediments (Lansdown et al., 2015; Welter & Fisher, 2016). Despite the inherent difficulties, experimental studies should be complemented by field studies covering natural hydrological gradients within and among ecosystems. Only connective knowledge can help to better understand the extremely dynamic behavior of N₂O in freshwater sediments under dry-wet stress.