Offsetting unabated agricultural emissions with CO2 removal to achieve ambitious climate targets

total ERF on the RCP2.6 pathway even with additional methane and nitrous oxide agricultural emissions, fully accounting for interactions between the different species. To compare the climate impacts of the two offsetting approaches (GWP100- and ERF- based approach) and to verify whether they meet the RCP2.6 climate requirements, we sub- tract the CDR rates from the RCP2.6’s fossil fuel CO2 emissions. This has the effect to first reduce net positive CO2 emissions and then enhance net negative CO2. We then calculate, in FaIR, ERF and temperature anomaly. To account for uncertainty in the projections, we com- pute a 1000-member ensemble simulation using randomized parameters of transient climate response, equilibrium climate sensitivity, pre-industrial sensitivity of carbon sinks, sensitivity to cumulative CO2 emissions, sensitivity to temperature change, and ocean temperature response. While the first two are generated from a lognormal distribution informed by the CMIP5 ensemble [65], the other parameters are randomly perturbed by up to 10% from the best estimates reported in Smith et al. (2017) [65]. We then retain only the ensemble members that predict historical temperatures within observational uncertainty, using the method by Thompson et al. (2015) [67]. To allow direct comparison with the UNFCCC temperature goals, we express warming relative to pre-industrial (mean over 1850–1900). Lastly, we explore CDR costs by calculating the price ptax per ton agricultural emission (in $/tCO2-eq) necessary to finance the offsetting of agricultural methane emissions (see Eq 3):

ptax ¼ ECDR � cCDR P GWP100i � Escenario;iagri ð3Þ where ECDR is the policy-induced amount of CDR (in tCO2), cCDR is the cost of CDR (in $/tCO2) and Escenario;iagri denotes total agricultural emissions of species i (ti). To compute aggre- gate increase in prices of emissions for both agricultural methane and nitrous oxide emissions, we first convert each Ei,agri in tCO2-equivalent using GWP100 as in the IPCC AR5 [3]. This enables direct comparison with existing and planned carbon taxes that are expressed per ton CO2-equivalent emissions. The aggregate tax expressed per ton CO2 equivalent is not exact due to the imperfect equivalence of this conversion over a time frame different than 100 years. However, the use of the GWP100 conversion metric does not impact the tax itself which is lev- ied on the methane and nitrous oxide emissions and therefore does not rely on the conversion. We assume a constant, average cost of CDR by $150/tCO2 removed, while evaluating the cost uncertainty by considering a range of possible costs, in each year, between $35-235/tCO2 removed. These cost assumptions are based on 2050 CDR cost and potential ranges across the literature, meta-analysed by Fuss et al. (2018) in Table 2 [36]. An array of different negative emission technologies is needed to achieve high rates of carbon removal. Lower-cost CDR techniques (with cost estimates below $120/tCO2 by 2050), such as afforestation and reforesta- tion, biochar, and soil carbon sequestration, could contribute by 2050 an aggregate estimated

removal potential of 1.4–4 GtC/year. An additional 3–14 GtC/year by 2050 could be contrib- uted by higher-cost CDR techniques: enhanced weathering, bioenergy with carbon capture and storage, and direct air carbon capture and storage [36]. We combine the different ranges of estimates for costs and potential of each CDR technique reported by Fuss et al. (2018) [36] to estimate the average cost of the additional CDR envisaged in our analysis (cf. S1 Table). If all technologies contribute to additional CDR rates, their average cost is $100/tCO2eq (with a minimum-maximum scenario interval by around $35-170/tCO2eq). Since lower-cost options are likely to be exhausted due to the CDR deployed in the SSP1-2.6 alone, the average cost rises to $150/tCO2eq (with a range of around $65-235/tCO2eq) in a high-cost CDR only sce- nario. In our analysis, we use this latter value as an estimate of the long-term cost of additional CDR and explore the uncertainty across the whole range of cost scenarios (from lowest PLOS ONE Offsetting future agricultural emissions with CO2 removal PLOS ONE | https://doi.org/10.1371/journal.pone.0247887 March 17, 2021 6 / 19