The increased ambition of international climate policy, articulated in the Paris Agreement's goal of “holding the increase in the global average temperature to well below 2°C above preindustrial levels and pursuing efforts to limit the temperature increase to 1.5°C above preindustrial levels” (UNFCCC, 2015), has increased scrutiny on the role all sectors can play in climate change mitigation. This has included a particular focus on agriculture (for example, in IPCC, in press). In addition, a number of recent high profile publications have highlighted agricultural emissions (e.g., Poore and Nemecek, 2018) and how they may need to be reduced to meet environmental commitments (e.g., Springmann et al., 2018). Yet in many treatments of agriculture's role in climate change, some key principles appear to be increasingly overlooked or misunderstood: specifically, how the impacts of methane (CH₄) and nitrous oxide (N₂O), the major greenhouse gases emitted from agricultural production, are distinct from each other and, in particular, from carbon dioxide (CO₂). An appreciation of these differences is important not only to understand what the mitigation of different gases can achieve in the context of the Paris temperature goal, but can also inform policy decisions. In this paper we outline the roles of these different greenhouses gases, consider how their reporting might be improved, and explore some of the potential implications for overall climate change mitigation.

Anthropogenic climate change is caused by multiple climate pollutants, with CO₂, CH₄, and N₂O the three largest individual contributors to global warming (Myhre et al., 2013). Agriculture and food production is associated with all three of these gases, but direct agricultural emissions are unusual in being dominated by CH₄ and N₂O.

The global food system is responsible for ~21–37% of annual emissions (Mbow et al., in press), as commonly reported using the 100-year Global Warming Potential (more on this later). The composition of gases emitted by the food system does not reflect the overall global emissions balance, however, with agricultural activity generating around half of all anthropogenic methane emissions and around three-quarters of anthropogenic N₂O (Mbow et al., in press).

Food system CO₂ emissions are somewhat harder to quantify, due to the distinct processes through which they are generated and difficulty in applying uniform accounting methods or sectoral boundaries. A small amount of CO₂ emissions occur directly from agricultural production, following the application of urea and lime, but these sources constitute an extremely small portion of total CO₂ emissions. Energy-use CO₂ from either agricultural operations (e.g., tractor fuel) or embedded in inputs (e.g., fertilizer manufacture and transport) can also be included as food system emissions, but are highly uncertain (Vermeulen et al., 2012), and are considered as energy or transport emissions within the IPCC (Intergovernmental Panel on Climate Change) accounting framework. The routes to reducing most of these emission sources are likely to be in the overall decarbonization of energy generation, rather than specific agricultural mitigations.

In addition, the food system is the main cause of ongoing land-use change CO₂ emissions, primarily from clearing land for crop production or pasture. Net land-use related CO₂ emissions are estimated as being responsible for around 14% of annual anthropogenic CO₂ (Le Quéré et al., 2018), with 10% directly linked to agriculture (Mbow et al., in press).

A picture emerges of agriculture and the global food system as an important contributor to global greenhouse gas emissions: of CH₄ and N₂O in particular, but also significant amounts of CO₂ depending on whether energy or land-use related emissions are included. Understanding the climate impacts of agriculture, particularly with respect to other sectors, necessitates understanding the distinct impacts of these three greenhouse gases.

Carbon dioxide is by far the main contributor to anthropogenic global warming (Myhre et al., 2013). This is not surprising given the enormous, and as of 2019 still increasing (Jackson et al., 2019), amount of CO₂ that we emit every year. Yet it is not simply because emissions remain high that CO₂ is responsible for so much warming. For every ton of CO₂ we emit, a significant portion will remain in the atmosphere for millennia (Archer and Brovkin, 2008), and so the total amount of CO₂ ever emitted by human activities commits us to a significantly altered climate essentially indefinitely, from any normal human decision making perspective (Clark et al., 2016). The extremely long-term persistence of CO₂, and accumulating behavior that occurs as a result, is fundamental to our understanding of anthropogenic climate change, and is well-agreed upon by physical climate-carbon cycle models (Joos et al., 2013), but is not widely appreciated (Sterner et al., 2019).

This context reveals that achieving net-zero CO₂ emissions is not simply a slogan to encourage ambitious emission reductions—it is a necessary condition of stopping global warming, stemming directly from our geophysical understanding of how contemporary CO₂ emissions perturb the carbon cycle. This principle also suggests that in order to remain under a given temperature target there is a total, time-independent CO₂ budget we must keep within (Frame et al., 2014). Such “cumulative carbon budgets” have increasingly provided an overarching framework for climate policy and a valuable tool to understand climate change (Rogelj et al., 2019). However, it also appears that there has been some confusion in how non-CO₂ gases fit into this framework. As the cumulative carbon budget only applies to CO₂, it follows that in addition to not exceeding the carbon budget, we must globally also limit the level of warming from all other sources to achieve the Paris Agreement. The IPCC's Special Report on Global Warming of 1.5°C (IPCC, 2018) states that peak temperatures are dependent on cumulative CO₂ emissions and non-CO₂ radiative forcing, and suggests these non-CO₂ contributions decline from their peak, but not do not have to reach net-zero emissions. We discuss next how shorter lived gases relate to global warming.

The focus on reducing (to net zero) our CO₂ emissions is well justified not just because it is the major anthropogenic climate forcer but also because it acts cumulatively. Shorter-lived greenhouse gases than carbon dioxide will, by definition, automatically be removed from the atmosphere over a shorter timeframe, so emissions will not continue to act cumulatively over the very long term that CO₂ will. There follows two key implications for shorter-lived greenhouse gases in relation to CO₂.

First, it suggests that shorter-lived GHGs have the potential for a sustained equilibrium concentration to be reached where constant ongoing emissions can eventually be matched by natural atmospheric removals.¹ The timeframe at which this point is reached is determined by the atmospheric lifetime of the gas. For methane, such an equilibrium can be reached in decades, so we need to consider the gas as having a non-cumulative effect if we are to design a physically meaningful climate policy even in the near term, or simply to understand how past and present emissions affect the climate. The implication for a cumulative carbon budget is that pulse emissions of methane cannot be viewed as exhausting the budget in the same as way as pulse CO₂ emissions. Rather, an ongoing rate of methane emissions will contribute to the budget in an equivalent manner to a pulse release of CO₂ (Lauder et al., 2013; Pierrehumbert and Eshel, 2015; Allen et al., 2016; Cain et al., 2019; Collins et al., 2020).

For nitrous oxide it would take centuries to achieve this equilibrium between emissions and removals, so we would still need to treat emissions of the gas as acting approximately cumulatively in order to meet our climate policy targets over the next century. Over longer timeframes, such as the multi-centennial timescales associated with ice sheet loss, we might want to consider ongoing nitrous oxide emissions as part of a long-term cycle, also distinct from the impacts of fossil fuel CO₂.

In this context, “long-term” is relative. From the perspective of even the most far-sighted governments, multi-century climate policy seems fancifully long-termist, given that national infrastructure, economic, political, and emissions plans typically look not much further than 2050, by which point ambitions are already very vague. From a geological or Earth system perspective, however, a few centuries appears relatively brief compared to how long we anticipate it would take the Earth to recover from our CO₂ emissions (Pierrehumbert, 2014; Clark et al., 2016). This brings us to the second key difference between CO₂ and shorter-lived gases: the legacy of different emissions.

Even when net CO₂ emissions are finally brought down to zero, we (humanity, including our descendants) will either be stuck with the climate impacts of these emissions for millennia, or face the burden of actively removing the enormous quantities of carbon that we have added. For shorter-lived gases, if we can stop emissions, then much of their impact will automatically be reversed over the timescales of their natural atmospheric removals. Thermal inertia in the climate response and the risk of hysteresis after crossing “tipping points” beyond which the Earth cannot readily return to its unperturbed state mean we cannot fully anticipate a complete reversal of impacts even from very short-lived gases. This is still distinct from the impacts of CO₂, however, for which we not only have these long-term response elements, but also retain a portion of all past emissions in the atmosphere, continuing to exert a climate forcing.

The principles outlined above are well-recognized in the climate science literature and physically uncontested. Misunderstandings or oversimplifications are not because of debate over these dynamics, but arise from our communication of different emissions as “CO₂-equivalents.”

Non-CO₂ gases are conventionally reported as CO₂-equivalent emissions (“CO₂-e”) using the 100-year Global Warming Potential (GWP100). This metric is based on the total perturbation to the atmospheric energy balance (radiative forcing) by an idealized pulse-emission of different gases over the 100-years following this pulse, scaled relative to CO₂ (Myhre et al., 2013). The limitations of this metric have been discussed in detail elsewhere (for recent examples, see Pierrehumbert, 2014; Allen et al., 2016; Tanaka and O'Neill, 2018; Wigley, 2018). Here we simply emphasize some particularly fundamental points building on the observations above. First, by describing all emissions as direct equivalents using single, static weighting factors, conventional application of GWP100 (or any other pulse-based metric taking this approach), misses dynamics that are driven by changes in the rate of emissions, and in particular cannot distinguish the cumulative and non-cumulative nature of different gases. Second, even for what we can infer from the impacts of isolated pulse-emissions, GWP is blind to any impacts beyond its stated timeframe, and so does not reveal the differing legacies of emissions—including the contemporary legacy of past emissions.

Figure 1 illustrates some of these points but also draws attention to perhaps an even more important consideration: the extremely ambiguous warming impacts of emissions reported using the GWP100. This figure was generated using the FAIR simple climate model (Smith et al., 2018) in its default set-up, adding the stated CO₂-equivalent emissions as either nitrous oxide, methane, or CO₂ (or balances thereof) to RCP4.5 emissions, then deducting the modeled warming from the baseline RCP4.5 conditions to show the impacts of these emissions alone. GWP100 values of 265 and 32 were used for nitrous oxide (Myhre et al., 2013) and methane (Etminan et al., 2016), respectively.

It is immediately clear that emissions scenarios reported as CO₂-equivalents do not indicate an unambiguous warming path. Common statements such as “methane is an x times stronger greenhouse gas than CO₂” are inherently oversimplifying, as they cannot capture the contrasting dynamics of the two gases. Regardless of whether one might argue GWP100 CO₂-equivalent emissions still have a use in climate policy or as simplifying communication tool, it undeniably fails as a universal environmental indicator, shown by the very large spread of possible temperature responses to supposedly equivalent emissions. We should not use such an imprecise measure in scientific contexts, but this is more often than not how emissions are reported: researchers routinely discard essential climatic data by not reporting individual gases separately (Lynch, 2019).

The emissions pathway here—increasing over the second half of the twentieth century, stabilizing briefly and then rapidly falling to zero emissions by 2050—can be thought of as a providing an illustration of the warming that has resulted from anthropogenic emissions and their roles in ambitious mitigation (in terms of overall profile; it is not representative of the scale of different emissions). Exploring what the figure shows can therefore be informative as to the role of different gases, and highlight what we would get wrong by considering all emissions as directly analogous to CO₂.

The rate at which emissions initially increase results in methane having a much greater impact than the nominally equivalent amount of CO₂ would indicate. In an agricultural context, such rapid increases occurred for ruminant methane emissions over the past century, and their reported CO₂e likely underestimates their contribution to current warming (Reisinger and Clark, 2018). In general, the impact of increasing methane emissions at rates above ~1% per year will be understated by reporting using GWP100 CO₂e (Lynch et al., 2020a).

As emissions start to decline from 2020 to 2050, and then stay fixed at zero, an even starker difference between the gases becomes clear. As methane emissions are reduced, most of the warming they caused is reversed. The short lifetime of the gas means that the concentration of methane in the atmosphere falls when not maintained by ongoing emissions. Meanwhile, for CO₂, stopping emissions ends the ongoing temperature increases that result from any non-zero emissions, and we end with a relatively fixed level of long-term warming.

Reducing CO₂ emissions to zero is therefore necessary to prevent further warming, but for methane, completely eliminating emissions goes beyond what is required for temperature stabilization. A “net-zero” CO₂ emitter will continue to exert a significant climate impact long after their emissions cease, potentially much greater than a methane emitter who can only manage a partial emission reduction. So if we reach zero emissions of the two gases, a methane emitter has contributed a much greater role in climate change mitigation than a nominally “equivalent” CO₂ emitter, and this continues to be the case into the very long-term.

An alternative perspective on these dynamics can be gained by considering why they are not captured by the GWP100. As it covers a period of 100-years, the GWP100 is effectively open-ended for CO₂, but not for methane: for CO₂ there is relatively consistent warming contribution across the 100-year period after emission and well beyond, but for methane the impacts of an emission are largely experienced within the first few decades. As it is integrates total forcing over the 100-year period to a single value, the GWP100 undervalues the initial impact of a methane emission, but then also fails to clearly reflect that most of this initial impact is then reversed. To capture the difference between CO₂ and methane emissions with this dynamic detail, then, we could instead consider an individual methane emission as being equivalent to a large CO₂ release, but with a large CO₂ removal occurring shortly afterwards (Lynch et al., 2020a). To have a truly equivalent effect to a methane emitter reducing their emissions, a CO₂ emitter would therefore not only need to reduce their emission rates but also actively recapture most of their past emissions.

The overall temperature change contribution and eventual warming legacy of different actors (be it nations, sectors, or individuals emitting different combinations of GHGs) thus cannot be inferred from emissions in a given year or whether or not they have an eventual “[net]-zero” ambition, as climate is shaped (in a gas-specific manner) by all past emissions. Yet annual emissions and net-zero targets have become the common currency of climate change communications and policy discussions.

Clearly it is still climatically beneficial to reduce methane emissions as much as we can, provided this is not at the expense of stopping CO₂ emissions. However, the question of how much methane emissions must or should be expected to reduce by, especially in relation to what CO₂ emitters have now achieved by stopping emissions, is revealed as less physically straightforward than might be assumed if all gases really were directly equivalent.

For N₂O, the dynamics are approximately intermediate to those of CO₂ and methane. The initial impact of increasing emissions is undervalued if comparing to a nominally equivalent amount of CO₂, and in the longer-term the automatic reversibility of warming from N₂O is also not reflected. The reader can imagine a similar spread of possible warming between 100% N₂O and either CO₂ or CH₄ to again emphasize the ambiguity emerging from using GWP100 CO₂e to report emissions. Over this two-century example, the behavior of N₂O is closer to that of CO₂, however, and so, as noted above, N₂O can be treated as a cumulative pollutant in short/medium-term climate policy without giving a misleading indication of its impacts, unlike methane.

The significant limitations of reporting only GWP100 CO₂e lead us to suggest changes in how to communicate emissions and related concepts. The phrase “carbon emissions” is often used to refer either to carbon dioxide emissions or as shorthand for “all greenhouse gas emissions” (this second usage likely arising from either the dominance of CO₂ as a contributor to global warming, or the ubiquitous usages of “CO₂ equivalents”). This ambiguity in meaning has perhaps led to or cemented some misconceptions around the direct fungibility of different gases, but could easily be overcome by using “carbon emissions” to refer exclusively to carbon dioxide, while using the more precise “greenhouse gas emissions” (or often simply “emissions,” depending on the context) when discussing non-CO₂ emissions or combinations of multiple gases.

Clear and appropriate terminology is even more important in the context of “carbon budgets.” In the climate science literature, cumulative carbon budgets are CO₂-only, as they result from the cumulative nature of CO₂ emissions outlined above, and particularly the near-linear relationship observed between cumulative CO₂ emissions and their contribution to global warming (Matthews et al., 2018). Confusingly, in the policy context, “carbon budgets” are instead usually used to denote aggregated GWP100 CO₂-equivalent ambitions, as in the UK government's “carbon budgets,” which define reductions in all greenhouse gases over time. Increased clarity is required, particularly from researchers, to avoid these misinterpretable terms. In a scientific context, “carbon budgets” should be used exclusively for CO₂, or when using alternative equivalence approaches such as GWP^* CO₂-warming equivalents (Cain et al., 2019), CO₂-forcing equivalents (Jenkins et al., 2018), or CGWP/CGTP (Collins et al., 2020) that can report short-lived gases in a way that is compatible with cumulative carbon budgets.

These concerns are particularly notable in light of recent focus on “carbon neutral” and “[net-]zero carbon.” As explained above, the need for net-zero emissions in order to stabilize global temperatures is CO₂-specific and comes directly from our understanding of how cumulative CO₂ emissions affect the climate. It can become unclear what is inferred by “carbon neutrality” (or similar terms), as it has different implications for non-CO₂ gases depending on whether it refers to temperature stabilization (the objective and outcome of becoming “CO₂ neutral”), or net-zero emissions (the CO₂-specific requirement for temperature stabilization).

The non-CO₂ gases methane and nitrous oxide comprise a uniquely large share of agricultural emissions. We therefore need to appreciate how emissions of these gases contribute to temperature change in order to understand the role of agriculture in global warming, and what agricultural emission reductions can achieve. There is no satisfactory means by which a single pulse-emissions-based weighting can be used to describe a physical “equivalence” between gases, so our common reporting measure of GWP100 CO₂e, which is built on this approach, cannot provide clear climatic inference. These limitations are well-recognized: Fuglestvedt et al. (2000) noted “it is uncertain whether policy makers are aware of the significance of lifetime differences and the shortcomings associated with the GWP methodology.” We highlight these same concerns for environmental and food sustainability research, where in many cases emissions metrics are used in ways which are at best ambiguous and at worst positively erroneous. More attention should be paid to the uses and limitations of different metrics for different purposes. We call for more environmentally robust approaches in the future, including the use of multiple and alternative emission metric approaches, and modeling of the relevant impacts.

Revisiting the reporting of emissions, and appreciating that agricultural emissions are not direct analogs of fossil CO₂, might also encourage a more critical take on some of the approaches and assumptions that agricultural mitigation requirements are built upon. Climate science tells us what different mitigation options can achieve–it does not directly inform on what mitigations must be made, except for the principle, which emerges directly from geophysics, that CO₂ emissions must eventually reach net-zero to prevent further warming. There may be political discussions on how quickly net-zero CO₂ emissions can be reached, or how the limited cumulative emissions budget can be equitably shared out, but there is a clear ultimate requirement. For agricultural methane, and to some degree nitrous oxide, there is scope to negotiate what ongoing “sustainable” emission rates might be acceptable for different actors. Clarifying the impacts of different emitters can facilitate these negotiations and lead to workable mitigation policies. Other elements that need to be considered in balancing emission reductions from different sectors require broader political, ethical, and social considerations, and we encourage researchers in these areas to be open and transparent about these factors.

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.