The most abundant greenhouse gases in the troposphere, excluding water vapor, are carbon dioxide (CO2), methane (CH4), and nitrous oxide (N2O). Together they account for more than 80 % of the total radiative forcing (IPCC, 2013), and their current tropospheric mole fractions and rates of increase are unprecedented in recent Earth history (Ciais et al., 2013; Burke et al., 2018; Fig. 1a and b). While CO2 is the most abundant of the three greenhouse gases, CH4 and N2O both have a higher warming potential than CO2 (Montzka et al., 2011). Accurately constraining the contribution of CH4 and N2O to Earth's radiation budget and their representation in predictive models requires their sources and sinks to be quantified with high resolution at the global scale.

The oceans are a fundamental component of the global climate system and are a net source of tropospheric CH4 and N2O at the global scale, although local to regional budgets may include both source and sink components. There are far fewer measurements of dissolved CH4 and N2O than of dissolved CO2 and while there is substantial international coordination with regard to CO2 analysis, calibration, and data reporting, no such coordination yet exists for CH4 and N2O (Wilson et al., 2018). Given the increasing prominence of climate change on scientific and societal agendas, greater coordination among the marine CH4 and N2O scientific community to provide more targeted measurements and increase the quality and interoperability of CH4 and N2O observations is particularly timely.

Despite the lack of an international coordinating framework, there have been important advances in our understanding of marine CH4 and N2O in numerous research disciplines, ranging from cellular metabolism and model microbial systems to large-scale modeling. For example, recent work identified novel microorganisms and metabolic pathways in the production of N2O (Trimmer et al., 2016; Caranto and Lancaster, 2017) and CH4 (Repeta et al. 2016; Bižić et al., 2020). Earth system models now incorporate improved N2O parameterizations to better resolve the ocean's role in the global N2O cycle (Battaglia and Joos, 2018). New techniques enable the discrimination of ancient and modern dissolved CH4 (Sparrow et al., 2018) and the transfer of CH4-derived carbon to other carbon pools (Pohlman et al., 2011; Garcia-Tigreros and Kessler, 2018). Other technological and analytical advances include improved near-continuous spectroscopic analysis that yields greater sampling resolution in surface waters (e.g., Gülzow et al., 2011; Arévalo-Martínez et al., 2013; Erler et al., 2015) and the deployment of analytical devices on robotic vehicles (Nicholson et al., 2018).

These scientific advances and an improvement in the quantity and quality of CH4 and N2O observations are timely given that large areas of both the open and coastal ocean remain undersampled (Fig. 1c and d). Limited observations contribute to uncertainty in marine CH4 and N2O inventories, their rates of production and consumption, and their emissions. The uncertainty associated with CH4 and N2O inventories is particularly problematic given that the marine environment is susceptible to an accelerating rate of anthropogenic change that will continue to modify the global cycles of carbon and nitrogen into the future. Environmental impacts on marine CH4 and N2O distributions include increasing seawater temperatures, decreasing concentrations of dissolved oxygen (O2), acidification, retreat of ice and mobilization of carbon substrates from former permafrost altering coastal run-off, and eutrophication (IPCC, 2019). These impacts will undoubtedly alter future CH4 and N2O exchange with the atmosphere, but the directions and magnitudes of these modified fluxes remain insufficiently understood.

The need to resolve the marine CH4 and N2O inventories prompted an evaluation of the collective ability of the international scientific community to accurately determine the distribution and emissions of CH4 and N2O and the determining physical–biogeochemical factors. This became the focus of a marine CH4 and N2O workshop hosted by the Ocean Carbon and Biogeochemistry (OCB) program at Lake Arrowhead, California, in October 2018. The workshop considered CH4 and N2O equally on the same agenda, even though nearly all field, laboratory, and modeling studies examine these trace gases separately. The rationale for this dual approach is that CH4 and N2O share common considerations of the physical, chemical, and microbial processes that dictate their water column distributions (Bakker et al., 2014; Bodelier and Steenbergh, 2014). In addition, many of the analytical procedures for quantifying CH4 and N2O and the subsequent data quality assurances share many common requirements. The opportunity to bring a large research community together to increase dialogue and encourage the cross-fertilization of ideas was thus considered very valuable. This article articulates the workshop outcomes framed in the context of current marine CH4 and N2O research and explores future research opportunities and challenges.

Our understanding of the temporal and spatial distributions of oceanic CH4 and N2O derives from over 5 decades of open ocean and coastal observations, including targeted expeditions, repeat hydrographic surveys, and time-series monitoring, each of which has been crucial to the development of our current knowledge (Fig. 2). Targeted programs have enabled invaluable insights into the role of oxygen-deficient zones in N2O cycling (Babbin et al., 2015; Bourbonnais et al., 2017; Frey et al., 2020) and the exploration of CH4-rich seeps and vents (Foucher et al., 2009; Suess, 2010; Boetius and Wenzhöfer, 2013). Basin-scale repeat hydrographic surveys (e.g., the international GO-SHIP program) have facilitated extensive water column mapping to identify relevant water masses and evaluate ventilation rates (Fig. 2d) (de la Paz et al., 2017). Other oceanic surveys have focused exclusively on surface sampling, using continuous equilibrator systems connected to various gas analyzers to yield high-resolution surface concentration fields of CH4 and N2O (Gülzow et al., 2013; Erler et al., 2015; Kodovska et al., 2016; Thornton et al., 2016a; Pohlman et al., 2017). In contrast, sustained long-term time-series measurements of CH4 and N2O at fixed monitoring stations are relatively few, but they span a range of latitudes and biogeochemical provinces (Fig. 2a and b). The time-series observations provide the contextual background for seasonal and interannual variation that allow long-term temporal trends and episodic events to be identified and evaluated (Farías et al., 2015; Wilson et al., 2017; Ma et al., 2019). Overall, the majority of measurements enable the variability in marine CH4 and N2O to be quantified at the mesoscale or greater (i.e., from hundreds of kilometers to ocean basins), with monthly to annual resolution, but there are substantially fewer datasets at the sub-mesoscale level (i.e., < 10 km and hours to days) (Fig. 3). A major reason for the limited sampling at the sub-mesoscale level is that it necessitates high-resolution measurements to resolve the heterogeneous variability that exists at these time–space scales. Such analyses have only recently become technically feasible (discussed in more detail in Sect. 6).

Until recently there has been no formal coordination of observations across the CH4 and N2O scientific community. In response to this, a Scientific Committee on Oceanic Research (SCOR) Working Group was initiated in 2014 entitled “Dissolved N2O and CH4: Working towards a global network of ocean time series measurements”. A major goal of the SCOR Working Group was to unite the international community in joint activities conceived to improve and inform seagoing CH4 and N2O analyses. An important activity was the preparation and distribution of common, combined gaseous CH4 and N2O standards to 12 international laboratories, with the aim of improving and standardizing calibration (Bullister et al., 2017). A subsequent intercomparison of discrete seawater samples included the use of these standards and revealed the variability between laboratories. While there were some encouraging results from the intercomparison, such as the agreement between individual laboratories using contrasting techniques, overall a large range was observed in CH4 and N2O concentration data generated by the participating laboratories (Wilson et al., 2018). Such analytical discrepancies weaken our collective ability as a community to evaluate temporal–spatial variability in marine CH4 and N2O. The discrepancies also highlighted the need for standard operating procedures (SOPs) for CH4 and N2O analyses to facilitate standardization of sampling, measurement, and calibration, as well as the reporting of data and accompanying metadata in common repositories.

A data repository for oceanic CH4 and N2O data known as the MarinE MEthane and NiTrous Oxide database (MEMENTO) was established in 2009 (Bange et al., 2009; Kock and Bange, 2015). MEMENTO is now sufficiently mature to support descriptions of the broadscale surface distributions of CH4 and N2O (e.g., Suntharalingam et al., 2012; Zamora and Oschlies, 2014; Buitenhuis et al., 2018; Battaglia and Joos, 2018). Machine-learning mapping also recently identified the various contributions of physical and biogeochemical predictor variables for CH4 (e.g., depth, primary production; Weber et al., 2019; Fig. 4b) and N2O distributions (e.g., chlorophyll, sea surface temperature, apparent oxygen utilization, and mixed-layer depth; Yang et al., 2020; Fig. 4a). The application of gas transfer algorithms to the extrapolated oceanic CH4 and N2O distributions helped decrease the uncertainty in estimates of global air–sea exchange fluxes (Fig. 4c), thereby fulfilling one of the key goals of MEMENTO (Bange et al., 2009). Net global open ocean emissions of N2O are now similarly estimated at 3–5 Tg N yr-1 by both Yang et al. (2020) and the Global Nitrous Oxide Project (Tian et al., 2020). In comparison, net global ocean CH4 emissions from machine-learning mapping were estimated at 6–12 Tg CH4 yr-1 (Weber et al., 2019), compared to 9–22 Tg CH4 yr-1 in the most up-to-date CH4 synthesis (Saunois et al., 2020). However, the narrower range for machine-learning-derived CH4 emissions retains high uncertainty in regions such as the Arctic, where emissions are highly heterogeneous and compounded by seasonal ice cover. Identifying the causes for uncertainty in high-emission regions will greatly aid future sampling campaigns, as is discussed in the following sections.

In the surface waters of tropical and temperate oceans, a number of factors contribute to the low supersaturation of CH4 including direct aerobic production arising from the degradation of methylated sulfur compounds by phytoplankton (Klintzsch et al., 2019) and methyl phosphonate in phosphorus-depleted waters (Karl et al., 2008; Sosa et al., 2020), indirect production via grazing (Schmale et al., 2018), and abiotic photoproduction (Li et al., 2020). A recent study demonstrated that CH4 production by cyanobacteria is linked to general cell metabolism and does not rely on the presence of methylated precursor compounds (Bižić et al., 2020). Deep within the ocean's pelagic interior, CH4 is weakly undersaturated, reflecting depletion via microbial oxidation (Reeburgh, 2007; Weber et al., 2019). Towards the coastline, CH4 supersaturation increases by orders of magnitude (Fig. 5b), reflecting terrestrial inputs (e.g., river and groundwater), increased organic matter loading (Borges et al., 2018), and CH4 diffusion and ebullition from shallow anoxic methane-rich sediments (Zhang et al., 2008; Borges et al., 2016; Upstill-Goddard and Barnes, 2016). Supersaturation of CH4 occurs frequently in the Arctic Ocean and its relatively shallow marginal seas, with the most extreme values observed in the Eurasian Arctic (e.g., Shakhova et al., 2010; Damm et al., 2015; Kosmach et al., 2015; Thornton et al., 2016a; Lorenson et al., 2016; Fenwick et al., 2017; Lapham et al., 2017). Terrestrial and subsea permafrost are potential CH4 sources to shelf waters in addition to CH4 hydrates that are found in marginal shelves globally (Ruppel and Kessler, 2017). Large point source CH4 emissions, such as seafloor gas seeps, can be large sources to the atmosphere in small localized areas (e.g., Thornton et al., 2020), but these sites remain particularly difficult to parameterize in models. This reflects limited observations and a poor understanding of their spatial distributions, the driving mechanisms, and the wider context within the carbon cycle. For example, the upwelling of cold, nutrient-rich water that accompanies CH4 ascending the water column stimulates CO2 consumption by photosynthesizing phytoplankton, rendering such CH4 seeps an overall net sink for climate-forcing gases (Pohlman et al., 2017). Recent work using thermal infrared satellite retrievals indicates increased high-latitude oceanic CH4 release in late autumn, coincident with pycnocline breakdown and a deepening of the ocean mixed-layer depth, thereby bringing deep CH4 to the surface (Yurganov et al., 2019). This is especially notable in the Kara and Barents seas, but the remote observations have not yet been confirmed by surface ocean measurements which are difficult and therefore rare, except during the Arctic summer.

Seabed CH4 emissions are hypothesized to increase in a warming ocean through the decomposition of gas hydrates, the degradation of subsea permafrost under some high-latitude seas, and the increased biodegradation of sediment carbon (Romanovskii et al., 2005; Biastoch et al., 2011; Lapham et al., 2013; Ruppel and Kessler, 2017; Borges et al., 2019). Effort is thus focused on quantifying the fraction of CH4 generated in or released from marine sediments that ultimately enters the atmosphere, particularly on shallow continental shelves and in coastal ecosystems. Natural stable isotopes have been used to inform spatial and temporal changes in dissolved CH4 concentrations (e.g., Pack et al., 2011; Mau et al., 2012; Weinstein et al., 2016; Leonte et al., 2017; Chan et al., 2019), and incubation experiments with added stable isotopes and radiotracers have helped elucidate how oxidation (anaerobically in sediments and aerobically in the water column), ebullition (where CH4 pore water partial pressure exceeds sediment hydrostatic pressure), and subsequent bubble dissolution in the water column interact to mitigate CH4emissions to air (Steinle et al., 2015; Jordan et al., 2020). The information deriving from these various approaches is inherently different but complementary. Isotope tracer incubations provide snapshots of rates specific to the methanotrophic community and CH4 concentration at the time of sampling, whereas concentrations and isotopic gradients are used to infer in situ rates integrated over space and time. A recent study deployed a remotely operated vehicle to examine the isotopic fractionation of CH4 during bubble ascent and used this to constrain the extent of bubble dissolution (Leonte et al., 2018). This work demonstrated an experimental approach established for broadly constraining water column CH4 cycling directly from a surface research vessel.

Despite the range of analytical and experimental approaches available, determining whether the origin of the emitted CH4 is seafloor release or aerobic production in the upper water column remains problematic. To date there is no straightforward way to routinely distinguish between seafloor-derived and water-column-generated CH4 for all locations. Even so, stable carbon and hydrogen isotope measurements (i.e., δ13C-CH4 and δ2H-CH4) combined with ancillary data may provide valuable source information. For example, combining these measurements with the ratio of CH4 to higher-order hydrocarbons (e.g., ethene (C2H4) and ethane (C2H6)) can be used to infer, for example, whether the origin of the CH4 is thermogenic, sub-seafloor, or biogenic within the water column (Whiticar, 1999; Pohlman et al., 2009; Lan et al., 2019). Continuous shipboard measurement of CH4 isotopes in surface water (e.g., Pohlman et al., 2017) and in the atmospheric boundary layer (Pankratova et al., 2019; Berchet et al., 2020) are now possible and they have been used in combination with atmospheric inversion models to characterize and discriminate marine-emitted CH4 from other sources (Berchet et al., 2020). Application of this method to land-based monitoring stations appears promising for apportioning CH4 emissions from various marine regions and sources (Thonat et al., 2019). Additionally, in regions where aerobic CH4 oxidation is substantial, the resulting isotopic fractionation generates measurable vertical and/or horizontal seawater gradients that can also be used to identify contrasting biogenic CH4 sources (Leonte et al., 2020). However, the general overlap in isotope compositions of sediment CH4 (e.g., Thornton et al., 2016b; Sapart et al., 2017) can complicate purely isotope-based determinations of sources.

Measurements of the natural radiocarbon content of dissolved oceanic CH4, while being highly specialized and requiring substantial amounts of ship time and processing (Kessler and Reeburgh, 2005; Sparrow and Kessler, 2017), provide valuable source information because the 14C-CH4 measurements are normalized to the same δ13C value and are unaffected by the extent of oxidation. The bubbles sampled from hydrate and active seafloor seeps are largely devoid of radiocarbon (Pohlman et al., 2009; Kessler et al., 2008; Douglas et al., 2016). However, CH4 in sediments can also be derived from more modern or recently deposited organic material, and an exact determination of individual contributions is hard to achieve (Kessler et al., 2008; Sparrow et al., 2018). The powerful insights made by radiocarbon-CH4 investigations would be further strengthened by concurrent sampling of other analytes that offer CH4 source information, such as clumped isotopes. Isotope clumping, the co-occurrence of two or more of the less-abundant isotopes in a molecule (e.g., 13C and 2H or 1H and 2H), provides unique information on marine CH4 sources (Stolper et al., 2014; Wang et al., 2015; Douglas et al., 2017; Young et al., 2017; Labidi et al., 2020). In this approach, the isotopic deviations in samples from their random probability distributions can give insight into formation temperature and the extent of biochemical disequilibrium. However, the sample size required for a clumped isotope analysis in the oceanic environment away from areas of seafloor emission is large and exceeds the already demanding volume requirements for 14C analyses by 1–2 orders of magnitude (Douglas et al., 2017). While the requirement of large sample size and lengthy measurement time currently preclude their more widespread application, clumped isotope measurements offer future promise in refining our understanding of the processes of marine CH4 production and consumption.

The large-scale spatial distribution of N2O in the global ocean is reasonably well-established. The highest open ocean N2O values are in upwelling environments, where concentrations extend up to micromolar levels (Arévalo-Martínez et al., 2015) and production rates can be as high as 120 nM d-1 (Frey et al., 2020). The highly elevated N2O concentrations can be proximal to regions with some of the lowest recorded N2O concentrations, in the cores of O2-deficient zones. This coexistence of the highest and lowest observed N2O concentrations over vertical distances of tens of meters makes upwelling regions a focal point for N2O research, particularly since O2-deficient ocean zones are increasing in size (Stramma et al., 2011). In contrast, in the surface waters of the expansive oligotrophic ocean gyres, N2O is weakly supersaturated (103 %–105 %) with respect to atmospheric equilibrium (Weiss et al., 1992; Wilson et al., 2017; Charpentier et al., 2010). Nitrous oxide becomes more highly saturated in the surface waters of equatorial upwelling regions due to the upward advection of N2O-rich waters (Arévalo-Martínez et al., 2017). For the Arctic Ocean, the data indicate low net N2O emissions, with some areas acting as net N2O sources and others as N2O sinks (Fenwick et al., 2017; Zhang et al., 2015).

Several parameters control net N2O emissions from the ocean, including temperature, salinity, dissolved O2, apparent oxygen utilization (AOU), nutrients, and microbial community abundance and composition. A recent modeling study trained with just three of these variables (chlorophyll, O2, and AOU) accounted for 60 % of the observed variability in oceanic N2O concentrations (Yang et al., 2020; Fig. 5a), highlighting the importance of N2O in productive upwelling systems. Correlations between N2O and environmental variables provide some insight into the factors controlling its distribution, but they provide no information about the microorganisms or metabolic pathways involved. Microbial production of N2O occurs during the metabolic processes of nitrification and denitrification (Stein and Yung, 2003). To determine which process dominates N2O production at any given location requires the application of multiple methodological approaches, ideally in parallel.

One of the most commonly used approaches is the incubation of discrete water samples under in situ conditions with stable isotope (15N) addition such as 15N-enriched NH4+, NO2-, or NO3- to measure N2O production rates from nitrification and denitrification (e.g., Ji et al., 2017). These approaches also provide insight into the microorganisms involved. For example, N2O resulting from archaeal NH4+ oxidation is mostly formed from a combination of NH4+ and another N compound (e.g., NO2-) whereas bacteria produce N2O from NH4+ alone (Santoro et al., 2011; Stieglmeier et al., 2014; Carini et al., 2018; Lancaster et al., 2018; Frey et al., 2020). Unfortunately, as with all incubation-based approaches 15N techniques are subject to bottle artifacts, and the strong dependence of N2O production and consumption on ambient O2 increases the potential for contamination during the collection and manipulation of anoxic deep seawaters. Incubation-based rate measurements are also compromised by abiotic N2O production via chemodenitrification, specifically the reduction of NO2- coupled to Fe2+ oxidation, as observed in high-Fe environments (Ostrom et al., 2016; Buchwald et al., 2016; Wankel et al., 2017). These issues highlight the need for incubation techniques that mitigate the effect of experimental artifacts (Stewart et al., 2012).

In addition to isotope addition and incubation, natural abundance water column measurements of N2O concentrations, isotopes, and isotopomers yield valuable rate and process information. These measurements are free from experimental artifacts and can be used to integrate over appropriate temporal and spatial scales. For example, nitrification in sunlit waters has been inferred from N2O distributions (Dore and Karl, 1996), and N2O production close to the ocean surface is a large contributor to the uncertainty in oceanic N2O emissions (Ward et al., 1982; Zamora and Oeschlies, 2014). Isotopomers are isomers having the same number of each isotope of each element but differing in their structural positions. Nitrous oxide isotopomers are increasingly used, sometimes in combination with box models, to estimate the rates of different N2O production pathways, in the upwelling systems off southern Africa (Frame et al., 2014) and Peru (Bourbonnais et al., 2017). There is however some disagreement about whether isotopomer signatures are robust indicators of the formation pathway (Yoshida and Toyoda, 2000; Sutka et al., 2006) or whether there is fractionation during production (Schmidt et al., 2004; Casciotti et al., 2018). Greater clarity is therefore required in the use of N2O isotopes and isotopomers to infer metabolic pathways of N2O formation. Notwithstanding this issue, field measurements of N2O isotopes and/or isotopomers have the potential to greatly increase current experimental capabilities and robustness (Yu et al., 2020). However, the development of spectroscopic gas analysis systems that have been so advantageous to CH4 research has been slower for N2O. This is due to the higher costs and the increased complexity of the laser systems, although progress is being made to improve instrumental precision and to decrease matrix effects and spectral interferences (e.g., Harris et al., 2019).

A better understanding of the microorganisms responsible for N2O production and consumption is fundamental to deriving more accurate estimates of process rates. For example, the metabolic activity of ammonia-oxidizing archaea can exceed that of ammonia-oxidizing bacteria in the ocean (Santoro et al., 2010; Löscher et al., 2012; Fuchsman et al., 2017). The differing sensitivities of these archaea and bacteria to dissolved O2 (Stahl and de la Torre, 2012; Hink et al., 2017) are a critical factor in evaluating the microbial response to changing environmental conditions, as shown for the terrestrial environment (Prosser at al., 2020). Therefore, to understand the impact of deoxygenation on oceanic N2O emission requires a better understanding of both archaeal and bacterial metabolisms and their environmental niches. Field-based sequencing not only characterizes the community but can also highlight potential metabolic pathways when they might not otherwise be inferred. For example, transcripts encoding for N2O consumption (nosZ) have repeatedly been identified in the oxic water column, despite denitrification being an anaerobic metabolic process (Wyman et al., 2013; Sun et al., 2017). The transcription of nosZ has also been located in highly dynamic O2 permeable coastal sediments (Marchant et al., 2017). Denitrification under aerobic conditions is attributed to fluctuations in O2, NO3-, organic matter, and other parameters that affect the availability of electron donors and acceptors, which ultimately influences whether a coastal environment is a net source or sink of N2O, as discussed in the next section.

Coastal and other shallow (< 50 m) marine systems are globally relevant CH4 and N2O source regions. However, their emission rates to the atmosphere are weakly constrained in comparison with the open ocean. Several factors contribute to the uncertainty, including the high diversity of coastal and shallow marine ecosystems and lack of consistency in adequately defining them, locally heterogeneous conditions causing strong spatial and temporal concentration gradients, highly uncertain spatial distribution of CH4 seeps, a bias towards studies in the Northern Hemisphere, and incomplete or sometimes inappropriate sampling strategies (Al-Haj and Fulweiler, 2020). Until these issues are resolved it will remain difficult to adequately define the contribution from shallow marine systems to global CH4 and N2O budgets. An important illustration of this is reflected in the prevailing view that large geological sources (e.g., seeps, mud volcanoes, and hydrates) are the main contributors to marine CH4 emissions (Ciais et al., 2013). The most recent modeled estimate of global marine CH4 emissions (6–12 Tg CH4 yr-1) reported that nearshore environments (depths of 0–50 m) contribute a large and highly uncertain diffusive flux (Weber et al., 2019). A study of coastal ecosystems, in this case defined as shelf, estuarine, and tidally influenced rivers, estimated them to contribute 7 Tg CH4 yr-1 (Anderson et al., 2010) while another estimated 1–7 Tg CH4 yr-1 for estuaries alone (Borges and Abril, 2011). Similar uncertainties exist for N2O. Estimates of coastal N2O emissions (which include coastal, estuarine, and riverine sources) range from 0.1 to 2.9 Tg N yr-1 (Ciais et al., 2013), although a recent review of N2O production across a range of estuarine habitats placed N2O fluxes at the lower end of these estimates (0.17–0.95 Tg N yr-1) (Murray et al., 2015). Based on these data, coastal systems account for around one-third of total marine N2O emissions (Yang et al., 2020).

The direct quantification of CH4 and N2O emissions from shallow coastal ecosystems has historically involved using gas concentrations measured in discrete water and air samples combined with a gas transfer velocity (kw). For the coastal and open ocean, the dominant driver of gas exchange is wind speed (e.g., Nightingale et al., 2000; Wanninkhof, 2014) whereas in nearshore, shallow water environments the interaction of water, depth, and tidal current speeds may be a major contributor to near-surface turbulence. Several kw parameterizations are now in use for coastal waters (e.g., Raymond and Cole, 2001; Kremer et al., 2003; Zappa et al., 2003; Borges and Abril, 2011; Ho et al. 2011; Rosentreter et al., 2017; Jeffrey et al., 2018), which increases the uncertainties associated with CH4 and N2O emissions. For example, a 5-fold variation in CH4 emissions from a single system occurred when applying different parameterizations to the measured gradients in CH4 (Ferrón et al., 2007).

To constrain emissions over small areas, continuous air–sea fluxes can be measured using free-floating chambers (e.g., Bahlmann et al., 2015; Rosentreter et al., 2018; Yang et al., 2018; Murray et al., 2020), but issues related to turbulence modification may still generate flux artifacts (Upstill-Goddard, 2006). To overcome these problems in the future, a greater reliance on direct and robust continuous techniques for air–sea flux measurement, such as eddy covariance (e.g., Podgrajsek et al., 2016), that avoid any need for kw will be necessary. Eddy-covariance measurements also capture both diffusive and ebullitive flux components (Thornton et al., 2020). Combining this approach with new analytical techniques such as cavity enhanced absorption spectroscopy (CEAS) and non-dispersive infrared (NDIR) spectroscopy should continue to improve the quality of CH4 and N2O flux estimates (McDermitt et al., 2011; Nemitz et al., 2018; Maher et al., 2019). Indeed, eddy flux towers aboard ships (Thornton et al., 2020) and in coastal locations (Yang et al., 2016; Gutiérrez-Loza et al., 2019) are now being equipped with CH4 instrumentation that enables the integration of CH4 fluxes over large areas. There are fewer N2O flux estimates made with CEAS and NDIR, and the implementation of N2O sensors on eddy flux towers remains limited. Recently, N2O emissions from eastern-boundary upwelling systems were quantified using inversion modeling based on atmospheric measurements from coastal monitoring stations, highlighting the potential of this approach to constrain N2O emissions from remote oceanographic regions that have significant spatial and temporal heterogeneity (Ganesan et al., 2020; Babbin et al., 2020). Inverse modeling of atmospheric measurements was also recently used to constrain CH4 emissions from the East Siberian Arctic Shelf (Tohjima et al., 2020).

Coastal measurements of CH4 and N2O also require the collection of ancillary data such as water column depth, tidal motions (Rosentreter et al., 2018; Huang et al., 2019; Pfeiffer-Hebert et al., 2019), and other information relating to diel processes (Maher et al., 2016). Such data are important because for example, the magnitude of CH4 and N2O fluxes varies over a diel period depending on the redox environment as a result of tidal effects and changes in inorganic N and O2 availability (Seitzinger and Kroeze, 1998; Call et al., 2015; Vieillard and Fulweiler, 2014; Maher et al., 2015; Murray et al., 2015; Foster and Fulweiler, 2019). The magnitude of CH4 and N2O fluxes also varies over longer temporal scales (seasonally to yearly) due to additional factors such as groundwater inputs, adjacent land use, dissolved O2, organic matter content and quality, and macrofaunal distributions (Barnes and Upstill-Goddard, 2011; Upstill-Goddard and Barnes, 2016; Gelesh et al., 2016; Bonaglia et al., 2017; Borges et al., 2018; Wells et al., 2018; Ray et al., 2019; Al-Haj and Fulweiler, 2020; Reading et al., 2020). To determine the contributing factors and resolve the spatial distributions, mobile sampling platforms such as small vessels (Müller et al., 2016; Brase et al., 2017; Tait et al., 2017) and autonomous vehicles (Manning et al., 2019) are essential. Recent improvements in gas sensors and in technology such as sonar and ebullition sensors will further increase our ability to measure dynamic fluxes (Maher et al., 2019; Lohrberg et al., 2020). Improvements to the quality and quantity of CH4 and N2O measurements in coastal systems will enable the development of iterative forecast models, further improving estimates of global coastal CH4 and N2O fluxes.

A more complete understanding of marine CH4 and N2O necessitates closer integration between biogeochemistry, model requirements, and targeted microbiological studies involving both single microorganism isolates and enrichment cultures. Marine CH4 and N2O budgets deriving from both “bottom-up” (e.g., emissions inventories, ocean and terrestrial process models) and “top-down” (e.g., inverse analyses of atmospheric trace-gas measurements) approaches would greatly benefit from more highly constrained metabolic processes. Specifically, this includes rates of CH4 or N2O production and consumption for key model microorganisms and the kinetic parameters associated with these metabolic rates. Reliable inventories of key microbially mediated process rates will improve the robustness of Earth system models used for predicting climate-mediated changes to marine CH4 and N2O emissions.

For N2O, laboratory studies quantifying microbial process rates, such as for nitrification and denitrification, are relatively few (e.g., Frame and Casciotti, 2010; Santoro et al., 2011; Löscher et al., 2012; Ji et al., 2015; Qin et al., 2017). Consequently, models largely continue to use process rates optimized using water column concentrations of N2O, O2, and related nitrogen cycle quantities (e.g., Battaglia and Joos, 2018; Buitenhuis et al., 2018; Landolfi et al., 2017). Future model parameterizations for N2O will require information on the variability of microbial process yields derived from culture studies with controlled varying conditions of O2 (Goreau et al., 1980; Frame and Casciotti, 2010; Löscher et al., 2012; Ji et al., 2018), pH (Breider et al., 2019; Hopkins et al., 2020), temperature, and nutrients. Automated incubation systems have measured N2O production kinetics and yield as functions of the concentrations of O2 and total ammonia nitrogen (Molstad et al., 2007; Hink et al., 2017). Quantifying the physiology of relevant microorganisms and connecting them to environmental characteristics will provide insights into why, for example, some shallow marine habitats act as N2O sinks while others are N2O sources or how N2O is produced in well-oxygenated open ocean waters, compared to oxygen-deficient zones.

For CH4, a key requirement to relate in situ CH4 production with transport to atmospheric emissions is our ability to accurately determine rates of CH4 oxidation. Fundamental issues include the challenges of cultivating methanotrophs and of replicating environmental conditions such as pressure and the chemistry of CH4 gas bubbles. The increased emphasis on CH4 dynamics in shallow water environments highlighted in Sect. 5 must be supported by culture-based measurements of CH4 oxidation that control for temperature, O2, and other important variables. In comparison to CH4 oxidation, culture-based studies are used increasingly to identify organisms capable of aerobic CH4 production and their underlying metabolic pathways (Carini et al., 2014; Klintzsch et al; 2019; Bižić et al., 2020).

Specific cellular yields and consumption rates of CH4 and N2O are not the sole objective of culturing experiments. Cultivation of microorganisms involved in CH4 and N2O production and consumption provides vital information about the physiology, metabolism, and interactions of environmentally relevant clades. When combined with genomic approaches, insights can therefore be gained into the diversity and global distribution of organisms involved in CH4 and N2O cycling. For CH4 some unexpected physiologies have been revealed (Ettwig et al., 2010; Haroon et al., 2013; Ettwig et al., 2016), which has directed research into sources and sinks of CH4 in the natural environment. Similarly, our understanding of how and when ammonia oxidizers produce N2O has been facilitated by studies of cultured nitrifiers and detailed analysis of their biochemistry (Stahl and de la Torre, 2012; Caranto and Lancaster, 2017). Recent combinations of cultivation studies with environmental genomics, albeit largely for terrestrial systems, have revealed a variety of denitrifiers, many of which are only involved in specific denitrification steps (Ganesh et al., 2014; Lycus et al., 2017; Hallin et al., 2018; Marchant et al., 2018; Conthe et al., 2019).

This article has assessed the collective ability of the scientific community to determine the spatial variability of marine CH4 and N2O distributions, the underlying mechanisms that determine this variability, and the resulting sea-to-air emissions. Shallow marine environments and oxygen-deficient zones are widely recognized as deserving of greater attention because they have high CH4 and N2O concentrations with inherently high uncertainties that complicate any assessment of their emissions to air (Bange et al., 1994, 1996; Bakker et al., 2014; James et al., 2016; Borges et al., 2016; Tian et al., 2020). Fortunately, recent technological advances that have increased our ability to conduct high-resolution measurements allow an optimistic outlook for making substantial progress in quantifying the CH4 and N2O budgets of these ecosystems. Even so, the inherent complexity of shallow marine environments clearly warrants a strategically coordinated approach to optimize the value of future studies. Issues to consider include identifying the locations of complementary sampling sites, standardizing sampling strategies and techniques, and agreeing on the use of common ancillary measurements that set the broad biogeochemical context (Bange et al., 2019). In contrast to the open ocean, measurement campaigns in shallow water environments are amenable to the use of eddy covariance flux towers, and they have the potential to lever resources from existing observation networks, which in North America include the Long-Term Ecological Research (LTER) network and the National Estuarine Research Reserve (NERR) System (Novick et al., 2018). Indeed, such activities are already underway; an increasing number of flux towers are being equipped for CH4 measurements (Torn et al., 2019) and future efforts should focus on the inclusion of N2O (see Sect. 5).

We are encouraged that the Global Carbon Project with its objective of developing a complete picture of the global carbon cycle including interactions and feedbacks has expanded to include CH4 (Saunois et al., 2020) and is now incorporating N2O (Tian et al., 2020). These projects compile the most recent data from peer-reviewed analyses of the sources and sinks of atmospheric CH4 and N2O from both natural and human activities. For example, the aquatic components of the recent Global Carbon Project N2O budget reported emissions from the open ocean, inland waters, estuaries, and coastal zones. Low-oxygen oceanic regions associated with eastern-boundary upwelling zones and the coastal ocean were identified as key regions with significant N2O variability requiring more detailed assessment via measurement campaigns and model analyses (Tian et al., 2020). Contribution to the Global Carbon Project and similar initiatives will identify areas of synergistic CH4 and N2O research for oceanographers and other Earth observation scientists (Ganesan et al., 2019). Furthermore, as highlighted in Sect. 6, field observations alone are insufficient to improve the robustness of Earth system models, and leveraging laboratory-based microbial process studies is highly recommended.

The success of any coordinated CH4 and N2O research program relies heavily on having uniformly high confidence in the various resulting datasets and their interoperability, and we identify three key initiatives that are paramount to ensuring this. i.

The first is to develop and adopt standard operating procedures (SOPs) to help obtain intercomparable CH4 and N2O datasets of the highest possible accuracy and precision. Currently, there is no consensus definition of high-quality CH4 and N2O measurements. However, an analytical agreement of ≤ 1 % is considered achievable for the laboratories conducting repeat oceanographic surveys and time-series observations (Fig. 3). For context, an analytical agreement of ≤ 1 % would permit the ocean's response to the increasing tropospheric CH4 and N2O mole fractions to be resolved on timescales of 10 and 5 years, respectively. These values are based on the changes in surface ocean CH4 and N2O concentrations that are predicted to occur due to the ongoing increase in tropospheric CH4 and N2O mole fractions at a seawater temperature of 20 ∘C and a salinity of 35 g kg-1 and assuming all sources and sinks remaining constant. In our recent marine CH4 and N2O intercomparison exercise it was concluded that the diversity of analytical procedures employed by the participants was a major cause of high variability between the reported concentrations, highlighting an urgent requirement for CH4 and N2O SOPs (Wilson et al., 2018). Consequently, these SOPs are now being compiled by the scientific community.

The benefits of pursuing the three activities described above have already been clearly demonstrated for carbon system measurements in the ocean. The intercomparability and high accuracy and precision of carbon system measurements were achieved by streamlining methodological approaches, universally adopting agreed-upon SOPs, production of reference material, and following community-driven quality control procedures (Dickson et al., 2007; Dickson et al., 2010). It is encouraging to see the marine CH4 and N2O community beginning to move in a similar direction.