Introduction

151 countries, encompassing 92% of the world economy, have pledged “net zero” (NZ) targets1, but the terminology used to define these goals is often inconsistent, including terms such as “net zero”, “climate neutral”, “carbon neutral”, and “greenhouse gas (GHG) neutral”, often to denote what the IPCC calls GHG neutrality2. To limit warming to 1.5 °C, the IPCC2 identifies that it is necessary to achieve NZ carbon dioxide (CO2) by around 2050 and to reach NZ GHG by around 2070. Most national targets aim for NZ GHG rather than NZ CO2, but definitions of scope and implementation are often unclear3, and deviate from the 100-year global warming potential (GWP100) aggregation of GHGs applied in current national reporting to the UN Framework Convention on Climate Change4. The land sector has a critical role to play in meeting NZ targets, especially in providing CO2 removal, which will be required to reach the temperature goal of well-below 2 °C2,5. Deployment of land-based CO2 removal will compete with current land uses, especially agricultural land dominated by livestock grazing6. However, shifting land use has considerable implications for society and the environment7, including potential consequences for global food security and the risk of GHG “leakage” if food production (especially livestock-based) is displaced from countries where it is comparatively efficient8.

Livestock methane (CH4) emissions contribute greatly to global GHG emissions. Adopting alternative GHG accounting methods to GWP100 that reflect the short-lived nature of CH4, including the outcome that a gradual decline in CH4 emissions will be sufficient to halt global warming9,10,11, could result in different land use mixes compatible with national NZ. For example, the use of the GWP* metric may place less emphasis on reducing CH4 emissions12,13. Similarly, introducing separate CH4 targets for livestock may reduce demand for CO2 removals (to offset CH4 emissions on a GWP100 basis) elsewhere in the agriculture, forestry, and other land use (AFOLU) sector, whilst aligning with IPCC global emission trajectories for temperature stabilisation that reflect the distinct warming effect of CH42,10,14. However, despite scientific robustness at global level, downscaling these approaches to national scale has strong implications for perceived fairness in global CH4 emission burden-sharing among nations. NZ definitions with “grand-parenting” allocation principles, i.e., using historic emissions levels as a baseline, or definitions that require the same percentage CH4 reductions across all countries, may be perceived as unfair to countries with low baseline CH4 emissions, or countries with high CH4 emissions but that provide global food security through exports14,15,16,17. Such trade-offs make it challenging to conclude on the “best” method for addressing CH4 emissions, and may contribute to the lack of consensus to date on how to define national NZ targets.

Ireland is a significant global exporter of milk and beef products18,19, with the Irish AFOLU sector contributing over 40% of the country’s total GHG emissions20. As milk production continues to increase at a faster pace than the decline in suckler beef production18, and net forest increments decline owing to a continually low rate of afforestation21, national AFOLU emissions have been rising despite a national target to achieve NZ (defined in legislation as “climate neutrality”) by 205022. Using a unique series of detailed AFOLU GHG flux scenarios for Ireland, this study explores the land use combinations necessary to achieve NZ under various definitions and examines the potential effects on national milk and beef production, with and without additional ambitious GHG abatement measures. This research thus provides novel insight into the implications of, and sensitivity to, different definitions of NZ for national AFOLU sectors involving significant livestock production.

Results

Net zero definition implications for Ireland’s AFOLU sector

GOBLIN6,23, a national biophysical AFOLU model, was run to generate 3000 randomised scenarios of Irish agricultural activities (including varying production efficiencies) and land use combinations within biophysical constraints for the year 2050, calculating associated annual GHG emissions out to 2100 (Fig. 1a). Maximum animal numbers were constrained at 2021 levels, and land spared from livestock production allocated to widely accepted carbon-neutral or carbon-positive uses, specifically wetland restoration, afforestation, or ‘ungrazed’ grassland6,23,24 (Fig. S3, Table S1), in order to focus data points around NZ boundaries. An ambitious assumption of 30% reduction in agricultural CH4 and N2O emissions (representing optimistic but plausible technical abatement efficacy by 2050) was applied post hoc to the original 3000 scenarios, to generate a parallel suite of 3000 abated scenarios. Ten different NZ definitions were applied to filter each scenario (Fig. 1b), as described in the Methods Section. These definitions included achieving by 2050: NZ CO2 (only) emissions, NZ GHG emissions based on GWP100, no net warming based on GWP*, or separate (non-zero) CH4 targets14 alongside a GWP100 balance for N2O and CO2 fluxes. Variations of these definitions relating to international fairness and longer-term (LT) time horizons (up to 2100) were also explored. For each definition, scenarios were classified into whether they were successful in reaching NZ with and without abatement measures (S-NZ-A and S-NZ) or failed to reach NZ with and without abatement measures (F-NZ-A and F-NZ).

Fig. 1: Assessment of future agricultural production and land use in Ireland with various NZ definitions.
figure 1

a Workflow indicating generation, abatement modification, NZ filtering, and subsequent post hoc analysis of 3000 scenarios of future agricultural production and land use combinations in Ireland. b Summary of NZ definitions applied in this study, with ovals representing definitions. These definitions encompass CO2 only emissions (carbon neutrality), the balance of GHGs over 100 years (GWP100), warming potential (GWP*), and a national CH4 emission target based on equal percentage reduction across countries to achieve temperature stabilisation (CH4 Target Grand-parenting). Definitions on the right are derivatives of four fundamentally different definitions on the left. LT: Long-term, indicating warming or flux balance out to 2100 (as opposed to balance achieved in the year 2050 only). Fairness was explored within GWP* and CH4 Target definitions by differing Ireland’s future NZ CH4 emission targets and reference emissions levels based on allocation of global CH4 emissions compatible with temperature stabilisation equally per capita, globally, or by national protein production14. Further details can be found in the Methods Section.

Full size image

The definitions which saw most scenarios reach NZ were (Table 1): carbon neutrality, with 2969 of the 3000 scenarios attaining NZ; GWP*, with 2464 and 2547 of non-abated and abated scenarios accomplishing NZ; CH4 Target Grand-parenting, where 1744 and 2560 of non-abated and abated scenarios reached NZ; and eGWP* Protein, where 1816 and 2511 of non-abated and abated scenarios reached NZ. Conversely, the lowest counts of NZ were for CH4 Target Population (35 and 92 of the 3000 non-abated and abated scenarios achieved NZ), GWP100 LT (551 and 805 of non-abated and abated scenarios attained NZ), and eGWP* Population (770 and 1172 of non-abated and abated scenarios accomplished NZ).

Table 1 Percentage of the 3000 scenarios which achieved NZ according to each of the definitions explored
Full size table

Large differences in new forest areas, new wetland areas (rewetted organic soils), milk outputs, and suckler beef outputs for each NZ definition were observed between the scenarios which succeeded in reaching NZ (S-NZ and S-NZ-A) and the scenarios which failed to reach NZ (F-NZ and F-NZ-A) (Fig. 2). Overall, the scenarios that achieved NZ had considerably larger areas of new forestry and wetlands, but lower quantities of milk and beef output (Fig. 2). The median new forest areas for S-NZ were 1096–2267 kha across the ten definitions, whereas the median ranges for new forest area of F-NZ were between 313–1078 kha. The minimum new forestry area required to achieve S-NZ and S-NZ-A was lowest for the carbon neutrality definition, requiring a median average increase of 142% from 2021 forest cover, and highest for the CH4 Target Population and GWP100 LT definitions, requiring median increases of 294% and 255%, respectively, from 2021 forest cover in the non-abated scenarios (Fig. 2). Median new wetland areas (i.e., rewetted organic soils and peat bogs) equated to 209–339 kha and 39–207 kha for S-NZ and F-NZ scenarios, respectively, and 209–339 kha and 38–204 kha for S-NZ-A and F-NZ-A, respectively. Maximum possible rewetting (339 kha) corresponded to an inferred increase of 28%, but was also observed as the median average for GWP100 LT and eGWP* Population definitions for non-abated scenarios (S-NZ), and GWP100 LT and CH4 Target Population for abated scenarios (S-NZ-A) (Fig. 2). Milk and beef outputs in aggregate declined substantially relative to 2021 levels across all successful NZ scenarios, even with ambitious abatement. Although there was a decline in beef production in the majority of the S-NZ and S-NZ-A scenarios, maximum-beef outputs in S-NZ and S-NZ-A scenarios were close to 2021 levels for most definitions except CH4 Target Population, which only achieved a maximum total beef output of 22% and 38% of 2021 beef output for S-NZ and S-NZ-A (Fig. 2). However, maximum-milk output S-NZ and S-NZ-A scenarios involved significant reductions in milk output vis-à-vis 2021 for most definitions, especially in CH4 Target Population and Protein, eGWP* Population, and GWP100 LT definitions. Ambitious abatement measures allowed for scenarios to achieve NZ (S-NZ-A) with smaller areas of new forestry, and greater milk and beef outputs than S-NZ, but required similarly large areas of organic soil rewetting. Further statistics and data can be found in the Supplementary Data.

Fig. 2: Total forest and wetland area, and total milk output and suckler beef liveweight output variation generated by the 3000 scenarios that fit within the different definitions of NZ as percentage changes from 2021 values.
figure 2

Plots display local minimum, Q1 (25th percentile), median, Q3 (75th percentile), local maximum, and outliers. Outliers are considered when they lie 1.5 times the length of the interquartile range from either end of the box. X within bars: mean value, F-NZ Failed to reach NZ, S-NZ succeeded in reaching NZ, F-NZ-A Abated scenarios which failed to reach NZ, S-NZ-A Abated scenarios which were successful in reaching NZ, Raw data can be found in the supplementary materials.

Full size image

Even with the ambitious abatement assumption, it was not possible for any of the S-NZ-A scenarios to simultaneously achieve 2021 population levels for both adult suckler beef and dairy cow numbers for any definition except “carbon neutrality” across the 3000 scenarios (Fig. 3a). Few scenarios under few definitions were able to support 2021 dairy cow populations whilst complying with NZ; the scenarios that did so also involved a reduction of at least two thirds in suckler cow numbers, relative to 2021 as seen by the blank failure areas in the top right of each panel in Fig. 3 – except for the carbon neutrality definition panel where the full spectrum of scenario populations fall within the NZ frontier (Fig. 3a). To achieve NZ with abatement under CH4 Target Population and Protein, eGWP* Population, and GWP100 LT definitions, significant reductions in dairy cow populations were observed. Without optimistic abatement measures, only carbon neutrality and GWP* definitions allowed for 2021 levels of dairy cow populations to be maintained, though with large reductions in suckler-cow numbers for GWP*, leading to lower suckler-beef production (Fig. S1).

Fig. 3: Relationships of various parameters under different definitions of NZ.
figure 3

a The relationship between adult dairy cow and suckler beef cow populations across the 3000 abated scenarios which successfully achieved NZ (S-NZ-A) for each different definition. b The relationship between milk output and new forest planted for the 3000 abated scenarios which successfully achieved NZ (S-NZ-A) for each different definition of NZ. The dashed lines and orange points represent the animal numbers or milk output for the year 2021. (x1000): x and y axis values are multiplied by 1000, LT long-term. Non-abated relationships can be found in Fig. S1, S2.

Full size image

Milk output among S-NZ-A scenarios followed the dairy cow population patterns in Fig. 3a. Higher milk output was generally associated with smaller areas of new forest among S-NZ-A scenarios (Fig. 3b), reflecting smaller areas of land spared from cattle at higher milk outputs. However, the spread of new forest area narrowed in an upwards trend towards the highest milk yields for most definitions (especially GWP100), reflecting minimum new forest areas needed to offset emissions from milk production (Fig. 3b).

Maintaining high milk production under different net zero definitions

Given that milk production is far more profitable for farmers than beef or sheep production25, generating almost €7 billion of dairy exports from Ireland in 202226, we analysed the closest scenario to the 95th percentile of maximum milk output among the successful abated (S-NZ-A) scenarios for each definition of NZ. For each definition, 2050 (and beyond) land use in Ireland looks very different from 2021 land use (Fig. 4). To preserve Ireland’s profitable dairy sector, large changes in the national landscape are required, including a 26–90% reduction in grassland for grazing all animals, from 58% of national land cover in 2021 to 6–43% land coverage in 2050, alongside a 47–387% increase in forest land, from 11% land area in 2021 to 16–53% cover in 2050. Land and emissions constraints associated with maintaining high milk output also requires large (up to 98%) reductions in suckler beef live weight output, and up to 97% reductions in sheep populations (Fig. 5).

Fig. 4: National land cover configurations for the 95th percentile scenario of (maximum) milk output from the range of abated scenarios that successfully achieved NZ (S-NZ-A), from the 3000 tested scenarios, according to each definition of NZ.
figure 4

Also shown is land cover for Ireland as of 2021. Below each configuration is the 95th percentile milk output. Boxes represent total land use in Ireland. n: number of scenarios which achieve NZ for that definition.

Full size image
Fig. 5: Changes associated with scenarios that give the 95th percentile (maximum) level of milk output for each definition of NZ, expressed as percentage changes from 2021 values for key land uses, sheep populations, and production of milk and suckler-beef liveweight (NB: excludes dairy-beef outputs).
figure 5

Colours represent scale of transformation, with the intensity of red associated with levels of reduction, intensity of blue with levels of increase, and white with low levels of change from 2021 values. S-NZ scenarios which succeeded in achieving NZ, S-NZ-A abated scenarios which succeeded in achieving NZ.

Full size image

Technical abatement to aid net zero transition

We explored the sensitivity of NZ compliance to varying levels of agricultural emissions through technical abatement measures applied at source (see Abatement Section in Methods), considering a spectrum of technical abatement from 0–100% across the ten definitions (Fig. 6). 0% and 30% abatement levels corresponded with the S-NZ and S-NZ-A scenarios, respectively. At higher levels of abatement, the share of scenarios achieving NZ increased. For instance, increasing technical abatement from 20–80% abatement increased the share of scenarios complying with the GWP100 definition from 45–85%. For the CH4 Target Population definition, compliance rose dramatically from just 18% success at 70% abatement to 88% success at 90% abatement. Due to this differential effect across scenarios, the relative rankings of NZ definitions in terms of successful scenarios varied with abatement level. For example, at 20% abatement, GWP100 had 119 more successful NZ scenarios than CH4 Target Protein, but by 40% abatement, CH4 Target Protein had 231 more scenarios achieving NZ than GWP100. Notably, at 100% technical abatement, the two LT definitions only achieved 91% and 93% success rates, for GWP100 LT and GWP* LT, respectively – reflecting the challenge of maintaining a long-term balance in CO2 fluxes across soils and biomass, even before agricultural emissions are accounted for6.

Fig. 6: Impact of theoretical level of technical abatement on NZ success across different NZ definitions.
figure 6

Each line represents a distinct NZ definition. The vertical dashed lines represent the two default assumptions made within the study, these being scenarios achieving NZ with 0 abatement measures (S-NZ) and scenarios achieving NZ with 30% technical abatement (S-NZ-A). Level of technical abatement relates to reduction in agriculture emissions, assumed uniform for enteric and manure CH4, and manure, direct, and indirect N2O emissions.

Full size image

It is crucial to underscore the highly speculative nature of these abatement levels, which are extended to the impossible level of 100% simply to illustrate the theoretical bounds of technical abatement measures. Our default abatement assumption of 30% is already considered ambitious27,28.

Discussion

Identifying possible NZ AFOLU configurations is an important step in informing appropriate climate action. It can help stakeholders to visualise the scale of the challenge and to prioritise effective and strategic (future-proof) actions regarding sustainable food production and land use policy. Simply put, if policy makers do not know what national NZ commitments could look like in practice, they cannot effectively plan and implement policy to achieve those commitments. Randomised scenario modelling supports foresight analysis that does not prescribe a “solution” configuration, but rather informs stakeholders about the real constraints involved, and depicts boundaries around a NZ “space” that the AFOLU sector can occupy. Accordingly, this study has shown that future agricultural activities and land uses will need to change dramatically to achieve NZ, but the combination and level of changes will be strongly influenced by the specific NZ definition applied. The wide range of future visions associated with different NZ definitions could impede stakeholder action, leading to debates over appropriate measures and the necessity of action, reluctance to develop strong policies, and apathy towards adopting mitigation practices. More effort is needed to build international consensus on clear definitions for national NZ goals. However, there are important commonalities in inferred NZ AFOLU configurations across the wide range of definitions studied here, pointing to a clear set of actions consistent with reaching territorial NZ that could be prioritised in the absence of a clear definition and end goal. Dramatic expansion of forest cover and re-wetting of organic soils have been highlighted in previous studies24, and even the most moderate changes observed to achieve different NZ definitions here will require rapid ramping up of action beyond existing policy targets22. Our study further demonstrates that, even with optimistic abatement measures to reduce agricultural GHG emission intensities, substantial reductions in the cattle herd are required to reach NZ by 2050 for all definitions excluding carbon (only) neutrality (Figs. 2, 3). The quantification of these requisite actions necessary to achieve a NZ AFOLU sector in Ireland, and their sensitivity to a range of NZ definitions, constitute the main novel findings of this study. For instance, considering the average NZ forest area observed for each definition implies that Ireland will require 2.4–3.9 times more forest area by 2050 (relative to 2021). Although these values appear extreme, it is worth noting that Ireland has the third lowest forest cover area in EU-2729, standing at 11%29,30, far below the EU average of 39%29.

These results highlight the inescapable need for difficult decisions to be made on whether to prioritise milk or beef output (or reduce both similarly) if Ireland is to achieve its policy goal of NZ by 2050. Although some mitigation measures have low costs and can be applied without adverse production effects, most mitigation actions involve trade-offs that should be fully considered before they are implemented, including impacts on livelihoods, food security, biodiversity, and export income31. Reducing output could also risk international GHG “leakage” by displacing milk and/or beef production to regions with less efficient (higher GHG intensity) production systems8, as Ireland currently exports 90% of milk and beef output internationally32,33, and has amongst the lowest GHG intensity beef and milk production, globally34. That said, while the Paris Agreement presents NZ GHG as a global target for the second half of the century5, it has led to countries (and companies) adopting NZ targets on a territorial basis3, shifting discussion away from comparative efficiency of (milk and beef) production towards simultaneous zero-sum equations for emissions and land carbon uptake at national level. Such an approach is more in line with principles of absolute sustainability thresholds35, and with transformative changes needed to achieve food system sustainability – including diet shifts, waste reduction, and closing yield gaps in developing countries, as well as efficiency improvements36,37. In general, such an approach is also consistent with a shift in land use towards biomass production to support expansion of biomaterial and bioenergy value chains, and/or delivery of ecosystem services, as part of the transition towards a circular bioeconomy38. It has been shown that cascading use of wood from commercial forestry could support stronger and longer climate mitigation from afforestation than is considered in the national inventory accounting approach applied in this study39. There remains a need to better understand the potential economic value of these alternative uses, that could compensate farmers for foregone livestock production and stimulate a broader bioeconomy within a NZ future.

GWP100 is the internationally accepted metric to measure progress towards Paris Agreement commitments in Nationally Determined Contributions4, and so a GWP100 “net zero GHG” balance remains the default definition for national governments to work towards. However, alternative methods have been proposed to better represent global emissions profiles required to achieve temperature goals5,10,14,16,40. We explored ten definitions of NZ previously proposed in the scientific and policy literature. This list is not exhaustive, but does include distinctive features relating to, inter alia: alignment with long-term temperature stabilisation emissions profiles at global scale, international fairness, and flexibility across gases. No definition is ideal across all aspects, and trade-offs are involved in selecting one method over another. For example, GWP100 definitions are potentially better aligned with international fairness than GWP* definitions16, but the 2050 GWP100 definition is less well aligned with long-term temperature stabilisation. GWP100 and GWP* definitions enable flexibility across gases, whilst separate methane targets do not (though could be adapted to do so with a hybrid approach where variations from the CH4 target are compensated by a warming-equivalent balance for CO2 and N2O41). The population-based CH4 target and eGWP* definitions score well for international fairness and alignment with long-term temperature stabilisation, but the former would involve reducing milk and beef production more than three quarters for Ireland, even assuming ambitious emissions abatement via technical measures. Thus, distinguishing CH4 from CO2 and N2O based on its characteristics as a potent SLCP is a double-edged sword for countries such as Ireland with high CH4 emissions (owing to large ruminant, rice, or fossil fuel sectors) and could lead to increases in CH4 emissions globally in the short term if less efficient production fills the gap8. Calculating NZ solely based on CO2 (carbon neutrality) overlooks major climate forcing emissions for countries such as Ireland with large agricultural emission sources, thereby disregarding important actors directly contributing to, and capable of mitigating, climate impact. Consequently, although it is necessary to achieve global carbon neutrality by 20502, such a narrow metric is inadequate to design appropriate pathways towards achieving NZ within national AFOLU sectors. Definitions based on cumulative emissions between 2050 and 2100 are better aligned with long-term temperature stabilisation, but also depend on much less certain future activity-flux relationships (emission factors and CO2 removal opportunities) beyond 2050, and may be a step too far (ahead) for policy makers operating in the context of short voting cycles.

In summary, new evidence presented here highlights the striking diversity of AFOLU configurations that could successfully achieve NZ, depending on the definition applied. This is not necessarily a barrier to near-term action as several common activities emerged across NZ scenarios regardless of the different definitions, including high rates of afforestation, extensive re-wetting of organic soils, and modest cattle destocking. Achieving a high degree of technical abatement for agricultural CH4 and N2O emissions can moderate the required rates of the aforementioned activities but cannot avoid the need for substantial land use change and shifts in production. As nations work towards their NZ goals, governments will face difficult policy decisions as the zero-sum nature of land and GHG balances needed to achieve these goals become clearer, and consequences for agricultural activities, land use change, and livelihoods/rural communities emerge. Results for Ireland are stark, owing to the outsize contribution of AFOLU within the national emission profile, providing a powerful illustrative case study. Achieving NZ urgently requires evidence-based, yet sensitive, engagement with all stakeholders to drive the transformative action required. Greater clarity on the end-goal is crucial to ensure timely and progressive implementation of actions, particularly those associated with a significant delay in GHG-flux response (e.g., afforestation); and also ensure that bioeconomy opportunities are identified to support a just transition. This in turn requires national policy leadership to clearly define NZ, preferably based on international consensus, which remains lacking. Progress can still be made without an agreed definition given the commonalities we found across all NZ definitions, but the chances of a just transition based on consistent strategic policy making, and stakeholder buy-in around an end-point vision, will diminish with each passing year.

Methods

GOBLIN model

Detailed methodology describing the “GOBLIN” (General Overview for a Back-casting approach of Livestock INtensification) model can be found in Duffy et al.23 and (for harvested wood product accounting update) Duffy et al.6. To summarise, GOBLIN is a national biophysical AFOLU model which runs randomised scenarios of agricultural activities and land use combinations within biophysical constraints to calculate annual GHG emissions along trajectories to selected future target years, here 2050 and 2100. GOBLIN consists of eight modules (Figure S3), the first of which randomises key input parameters: national dairy and beef cattle and sheep numbers, animal level productivity, fertiliser application rates and distribution rules for grassland spared from livestock production. Deduced parameters include milk and beef production, areas of spared grassland, and new areas of rewetted organic soils, and broadleaf and commercial forestry. From these parameters, AFOLU GHG emissions and removals are calculated using IPCC Tier 1 and Tier 2 methodologies23,42,43, consistent with Ireland’s UNFCCC reporting20. Cropland emissions are included, but the area is assumed to be constant over time.

Scenarios

3000 randomised scenarios were generated using a Latin hypercube sampling method44 by randomly varying input parameters utilised in GOBLIN between set minimum and maximum values for each individual scenario (Table S1). This sampling approach ensures a comprehensive exploration of the future parameter space. Table S1 in the supplementary material provides a detailed overview of the key inputs into the GOBLIN model, presenting the value ranges of key variables from which each of the 3000 scenarios is randomly derived. Total animal numbers were set between one and values reported for 202145, with grassland utilisation rate calibrated at between 67% and 80% of grass produced being consumed by livestock (variable across scenarios) based on calculated grass uptake and total grassland area utilised by the updated national herd and flock numbers23. Further details on GOBLIN input parameters, including background data information, can also be found in Duffy et al.23 and Duffy et al.6. It is important to note that the target year for AFOLU configurations is 2050; no further changes to annual agricultural production or land use are considered after 2050. Emissions beyond 2050 thus represent this new “equilibrium” land use, incorporating forestry (re)growth and harvest cycles in pre-existing forests and “new” forests planted up until 2050. Although animal numbers in Ireland are increasing18, early runs of the model6 indicate current animal numbers are already exceeding the emissions levels necessary to achieve NZ across definitions. As such, animal numbers were capped at their recent levels to avoid returning excessive scenarios which failed to reach NZ, of less relevance to conclusions. It is important to note that by setting minimum animal populations at 1, and applying a randomized input algorithm, we are not pre-determining a future for Ireland – on the contrary, we are attempting to remove value judgments as far as possible in order to objectively explore what alternative futures could look like in terms of GHG emissions. However, to maximise data resolution around likely bounds of NZ under different definitions from the 3000 scenarios, parameter ranges forced static or reduced animal numbers, static or increased grass use efficiency, and allocation of land spared from livestock production to carbon-neutral or carbon-positive uses, such as organic soil rewetting and afforestation – reflecting insight from recent model runs6. The aim of this paper was not to make economic or feasibility assessments, but rather to explore the various AFOLU configurations that align with different definitions of NZ through randomised scenario simulations.

Net zero definitions

The following methods elaborate on the definitions pertaining to NZ for the year 2050.

GWP100

GWP100 is the most widely used climate metric. At COP24 it was decided to use GWP100 for reporting national emissions to the Paris Agreement4. GWP100 (Eq. 1) is defined as the ratio of the (100-year) time-integrated radiative forcing (RF) from the instantaneous release of 1 kg of a trace substance i relative to that of 1 kg of a reference gas (typically CO2)46:

$${GW}{P}_{i}\,\left(H\right)=\frac{{\int }_{0}^{H}R{F}_{i}\,\left(t\right)\,{dt}}{{\int }_{0}^{H}R{F}_{{{CO}}_{2}}\,\left(t\right)\,{dt}}=\frac{{AGW}{P}_{i}\,\left(H\right)}{{AGW}{P}_{{{CO}}_{2}}\,\left(H\right)}$$
(1)

where H is the time horizon over which the calculation is considered (here 100 years); RFi is the radiative forcing due to a pulse emission of gas i; (t) is the time-dependent decay in abundance of a pulse emission; and the corresponding quantities for the reference gas (typically CO2) are in the denominator. In other words, GWP is the ratio of Absolute GWP (AGWP) for i over AGWP for the reference gas CO246. The GWP100 in this study used AR5 emission metric values for CO2, CH4, and N2O emissions, with cumulative forcing over 100 years equalling 1, 28, and 265, relatively47. GWP100 was used to calculate GHG neutrality2, where to be classified as NZ (S-NZ) the net sum of GHG emissions had to balance with an equivalent amount of removals or be net-negative in the year 2050 (only – the trajectory after 2050 was not considered).

GWP*

GWP* defines equivalence based on the warming impact of a change in SLCP emissions to the warming impact of CO2 emissions, termed the CO2-warming-equivalent (CO2-we)40,48,49. GWP* is sensitive to the rate of change of SLCPs, with small absolute increases in emissions leading to a large warming impact and a larger value for CO2-we emissions than would be for GWP100 based CO2-e emissions. Conversely, any decline in SLCP emissions larger than about 0.3% per year leads to negative CO2-we emissions. CO2-warming-equivalent emissions of CH4 within this study were calculated from the latest GWP* equation (Eq. 2)40:

$${E}^{* }\left(t\right)=4.53\times {E}_{100}\,\left(t\right)-4.25\times {E}_{100}\,\left(t-20\right)$$
(2)

where E* (t) is the CO2-we emissions at time t; E100 (t) is the SLCP emission rate at t, calculated as CO2e using GWP100; and E100 (t-20) is the rate of SLCP emissions 20 years before t, calculated as CO2e using GWP100.

The CO2 and N2O emissions (E) within this definition were calculated based on GWP100 methodology as a CO2e emission (ECO2e) quantity by multiplying by the appropriate GWP conversion factor for the specified time-horizon (H), here 100 years (Eq. 3):

$${E}_{C{O}_{2}-e}=E\times {GW}{P}_{H}.$$
(3)

As noted in Allen et al.9, halting global warming requires NZ emissions of LLCFs such as CO2 and N2O, and declining (but not necessarily zero) net emissions of SLCFs such as CH4. Following the GWP* principles, if we have a scenario which maintains NZ CO2-we emissions for the year 2050 (S-NZ) then we have a scenario which should not drive temperatures upwards.

Methane targets

Prudhomme et al.14 developed a series of national biogenic CH4 quotas compatible with limiting global warming to 1.5 °C. International fairness was explored where global CH4 budgets for 1.5 °C scenarios were allocated to national quotas based on: grand-parenting (equal percentage reductions across countries) (Eq. 4); population (equality, or equal per capita emissions) (Eq. 5); and animal protein security (emissions proportionate to animal protein production in 2010) (Eq. 6). According to the three allocation methods above, CH4 quotas for Ireland for 2050 were calculated in Prudhomme et al.14:

$$C{H}_{{4}_{2050}}^{i}=\frac{C{H}_{{4}_{2010}}^{i}}{C{H}_{{4}_{2010}}^{{world}}}\times C{H}_{{4}_{2010}}^{{world}}\times \left({\alpha }_{E}^{{world}}-1\right)$$
(4)
$$C{H}_{{4}_{2050}}^{i}=\frac{{Po}{p}_{2010}^{i}}{{Po}{p}_{2010}^{{world}}}\times {\alpha }_{E}^{{world}}\times C{H}_{{4}_{2010}}^{{world}}$$
(5)
$$C{H}_{{4}_{2050}}^{i}=\frac{{Pro}{t}_{2010}^{i}}{{Pro}{t}_{2010}^{{world}}}\times {\alpha }_{E}^{{world}}\times C{H}_{{4}_{2010}}^{{world}}$$
(6)

where the national biogenic CH4 quotas in 2050 (C\({{{{{{\rm{H}}}}}}}_{{4}_{2050}}^{{{{{{\rm{i}}}}}}}\)) are dependent on specific allocation rules for a specific country, i, here Ireland. \(C{H}_{{4}_{2010}}\) is the national (i) or global (world) biogenic CH4 in 2010. Pop is the 2010 population of the country or globally, and Prot is the global or national animal-protein production. \({\alpha }_{E}^{{world}}\) refers to the reduction (between 0 and 1) of global biogenic CH4 emissions in 2050 compared to 2010 compatible with 1.5 °C scenarios (see Prudhomme et al.14 for further details). Deduced CH4 targets for Ireland for 2050 were 358, 60, and 229 kt of CH4 for Grand-parenting, Population, and Protein approaches, respectively. Here, scenarios that complied with the CH4 target and achieved at least NZ (or net negative) CO2 plus N2O using GWP100 by 2050 were assigned S-NZ.

eGWP*

Rogelj and Schleussner16 argue that when applied at national level, the CH4 emission grand-parenting implied via the reference emission level used in the GWP* calculation introduces a preferential treatment of countries that have large contemporary CH4 emissions. The eGWP* method is an attempt to introduce international fairness into the GWP* method by adjusting the national reference emission level for CH4 to a “fair share” of global emissions, rather than the contemporary emission level16. Here, we adapt eGWP* to incorporate the latest GWP* equation combined with reference level CH4 emissions derived from allocation rules introduced above for Ireland14,40. In its broken-down form for any SLCP, GWP* is given in Eq. 740:

$${E}^{* }\left(t\right)=\left[{gA}\times {E}_{H}\,\left(t\right)-{gB}\times {E}_{H}\,\left(t-\Delta T\right)\right]\times {GW}{P}_{H}$$
(7)

where g denotes a scaling factor of 1.13; A is a coefficient calculated as 75/ΔT + 0.25, where t is the GWP* time interval, the 75 is a rate-based component (0.75*100) (where 0.75 is the weighting given to the impacts of changing the rate of SLCP emissions49), 0.25 is the stock component (the weighting given to the impacts of the current emissions rate49); and B is the coefficient 75/ΔT. Equation 7 is identical to Eq. 2.

When the reference emissions level of CH4 is modified as per the eGWP* methodology, the output can be written as Eq. 8:

$$e{E}^{* }\left(t\right)=\left[{gA}\times {E}_{H}\,\left(t\right)-{gB}\times {{E}_{{Ref}}}_{H}\,\left(t-\Delta T\right)\right]\times {GW}{P}_{H}$$
(8)

where the eGWP* CH4 reference values \({{E}_{{Ref}}}_{H}\) in this study were extracted from Prudhomme et al.14. The calculations, performed in Prudhomme et al.14 for population and protein CH4 reference values and displayed in Eqs. 9, 10, respectively, result in reference values of 89.12 and 340.10 kt CH4 for equal-per-capita, and equal-per-protein in 2010, respectively.

$$C{H}_{{4}_{{ref}}}^{i}=\frac{{Po}{p}_{2010}^{i}}{{Po}{p}_{2010}^{{world}}}\,\times C{H}_{{4}_{2010}}^{{world}}$$
(9)
$$C{H}_{{4}_{{ref}}}^{i}=\frac{{Pro}{t}_{2010}^{i}}{{Pro}{t}_{2010}^{{world}}}\,\times C{H}_{{4}_{2010}}^{{world}}$$
(10)

It should be noted that although reference emissions are usually a value 20 years before the first term in the equation for CH4 (as per GWP*), reference values here were from the year 2010, based on availability of consistent data needed for international allocation calculations in the underlying study by Prudhomme et al.14, therefore introducing a larger ΔT of 40 years, between 2010 and 2050. Successful NZ scenarios achieve <0 kg CO2− we at the year 2050.

Carbon neutrality

The term “carbon neutrality” is often interchangeably characterised as an alternative to NZ (or GHG neutrality50). However, the IPCC50 defines “carbon neutrality” as the “condition in which anthropogenic CO2 emissions associated with a subject are balanced by anthropogenic CO2 removals”. As such, only the CO2 emissions and sequestered CO2 were calculated within this definition. Successfully achieving NZ (S-NZ) was assigned to scenarios where the net sum of CO2 emissions balances with (or were less than) an equivalent amount of removals in 2050.

Long-term net zero

The parties to the Paris Agreement have agreed to pursue mitigation measures to achieve a balance of anthropogenic GHG emissions by sources and removals by sinks in the second half of the 21st century5. To explore this definition of LT NZ, two further interpretations of NZ were analysed for both GWP100 and GWP*, termed “GWP100 LT” and “GWP* LT”, to assess which scenarios sustained NZ across the period 2051-2100, that is, balanced their cumulative GHG emissions during 2051–2100 whilst maintaining constant land use after 2050 (but with extended forest sequestration accounted for6,23), using the same methodology as described above, again assigning the scenarios to S-NZ or F-NZ.

Abatement

It is possible for NZ targets to be met not just through specific configurations of agricultural production and land use, but through the reduction of agricultural emissions at source following the implementation of abatement measures. Effective abatement measures for CH4 and N2O that can reduce emissions with minimal, neutral, or positive effect on productivity can include the displacement of mineral N fertiliser by biologically-fixed N (using grass clover or multi-species pastures) or mobilisation of soil N via pH manipulation, applying protected urea fertilisers, improving animal genetics, anti-methanogenic feed additives, low-emission storage and spreading techniques for manures, acidification of manures, and use of methane and nitrification inhibitors27,51,52. Within this study, abated scenarios assume an ambitious 30% reduction in agriculture emissions uniformly applied to enteric and manure CH4, as well as manure, direct, and indirect N2O emissions. This reduction represents the realistic upper end of cumulative abatement potentials possible with identified technologies27,28. Employing these abated scenarios reflects a conservative approach in drawing final conclusions with respect to the magnitude of impacts from efforts to meet NZ, and thus is our default assumption. Each of the 3000 abated scenarios was classified into whether they succeeded (S-NZ-A) or failed (F-NZ-A) to meet NZ with abatement, according to each NZ definition.

Although we refer to our abatement assumption as ambitious, some studies demonstrate significantly greater efficacy for certain technical abatement measures, far exceeding the 30% benchmark. For instance, red seaweed has exhibited an exceptional ability to reduce enteric CH4 emissions, with some early upper-end studies reporting over 80% CH4 mitigation53,54. However, these results may be overly optimistic for achieving a 2050 deadline for a national herd average, with several challenges still to be addressed on feasibility, long-term efficacy, and effects on animal production and health55. Nevertheless, a sensitivity analysis was undertaken on the level of technical abatement on the ability of scenarios to achieve NZ, ranging from 0% to 100%.

Limitations

While the GOBLIN model endeavours to simulate real-world consequences, it is important to acknowledge certain limitations inherent to its design, in addition to those outlined in the Methods Section. The model ran randomised scenarios within predefined parameter boundaries (Table S1); however, it is essential to note that the scope of land use changes was restricted exclusively to positive transformations on spared grassland. Specifically, spared grassland resulting from decreased animal numbers or increased productivity was constrained to either rewetting when on organic soil, afforestation on mineral soil, or preservation as ungrazed grassland. Notably, the model deliberately excluded any possibility of negative land use changes emanating from spared grassland. This design choice aligns with realistic pathways towards NZ based on identified necessary actions2,36, limiting the number of scenarios that would fail to reach NZ, but still constitutes a noteworthy limitation.

Additionally, the linear interpolation of scenario values from the baseline year to 2050 may marginally impact definitions influenced by trends, particularly the GWP* definitions. Further, the uncertainty inherent in long-term projections beyond 2050 is underscored by GOBLIN’s assumption of equilibrium AFOLU configurations and static flux factors from 2050 onwards (excluding forest growth-harvest cycles, which are modelled out to 2100). Consequently, many long-term effects, both positive and negative, arising from individual scenarios might be overlooked. There remains a need to consider interaction between land-based CO2 sinks and downstream bioeconomy innovations, especially cascading use of wood culminating in bioenergy with carbon capture and storage that could dramatically extend the longevity of critical carbon sinks. Further GOBLIN model limitations can be found at Duffy et al.6,23.