A comparison of the climate and carbon cycle effects of carbon removal by afforestation and an equivalent reduction in fossil fuel emissions

. Afforestation and reduction of fossil fuel emissions are two major components of climate mitigation policies. However, their effects on the Earth’s climate are different because a reduction of fossil fuel emissions directly alters the biogeochemical cycle of the climate system and modi-ﬁes the physics of the atmosphere via its impact on radiation and the energy budget, while afforestation causes biophysical changes in addition to changes in the biogeochemical cycle. In this paper, we compare the climate and carbon cycle consequences of carbon removal by afforestation and an equivalent fossil fuel emission reduction using simulations from an intermediate complexity Earth system model. We performed two major sets of idealized simulations in which fossil fuel emissions follow extended Shared Socioeconomic Pathway (SSP) scenarios (SSP2-4.5, 3-7.0, and 5-8.5), and equal amounts of carbon are removed by afforestation in one set and by a reduction in fossil fuel emissions in another set. Our simulations show that the climate is cooler by 0.36, 0.47, and 0.42 ◦ C in the long term (2471–2500) in the case of reduced fossil fuel emissions compared to the case with af-forestation when the emissions


Introduction
Human activities in the industrial era have led to an increase in the concentration of greenhouse gases (GHGs) and an increase in global mean surface temperature (Masson-Delmotte et al., 2021). GHGs emitted by human activities include carbon dioxide, methane, nitrous oxide, etc., among which CO2 is the most important GHG because of its long 30 lifetime in the atmosphere (Archer et al., 2009;Montenegro et al., 2007;Archer, 2005;Archer and Brovkin, 2008;Moore and Braswell, 1994;Eby et al., 2009). The atmospheric CO2 concentration has increased from approximately 277 ppm to 415ppm during the period 1750-2021 (Joos and Spahni, 2008;Keeling et al., 1976). Most of the anthropogenic emissions of atmospheric CO2 result from either fossil fuel use or land use and land cover changes. In the recent decade (during the period 2010-19), the CO2 emissions from fossil fuel use and land use and land cover 35 changes are 9.6±0.5 PgC yr -1 and 1.6 ± 0.7 PgC yr -1 , respectively (Friedlingstein et al., 2020).
Approximately 50% of the emitted carbon stays in the atmosphere while the rest is taken up by the land and ocean on decadal timescales (Friedlingstein et al., 2020). As a result of the increasing atmospheric CO2, the global mean surface temperature has increased by 1.07°C from 1850-1900to 2010(Masson-Delmotte et al., 2021. Global warming has been directly linked to an increase in the frequency of floods, extreme rainfall events, and forest 40 fires in different parts of the world (Alfieri et al., 2015;Ali et al., 2019;Allan and Soden, 2008;Papalexiou and Montanari, 2019; Anderson et al., 2011;Canadell et al., 2021). Two major strategies are considered for mitigating climate change: i) reforestation/afforestation and ii) reduction of fossil fuel emissions. While both these methods reduce the carbon accumulation in the atmosphere, the net effect of these two actions on Earth's climate could be different. It may be noted that reforestation/afforestation is one of several carbon dioxide removal (CDR) options that 45 have been suggested to mitigate climate change (Pacala and Socolow, 2004;Psarras et al., 2017;van Kooten, 2020).
The nature of the source or sink of atmospheric CO2 could play a key role in determining its net effect on the earth's climate. For example, Jayakrishnan et al., 2022 investigated the contrasting response of the climate system to emissions from fossil fuel use and deforestation and showed that these two emissions are fundamentally different in how they affect the climate system. However, adequate emphasis is not given to the nature of the source or sink in 50 many contexts. An example for the implications of neglecting the non-radiative effects of the source of atmospheric CO2 is described by Simmons & Matthews, 2016, where they show the importance of accounting for the biophysical changes due to land cover changes for calculating the transient climate response to cumulative carbon emissions (TCRE; a metric that defines the response of the global surface temperature to cumulative carbon emissions). In the current study, we address another set of related questions where the nature of the source or sink is important: Are the 55 climate and carbon cycle effects of carbon removal by afforestation or an equivalent reduction of fossil fuel emissions the same? Which of these two actions is more beneficial from a climate change mitigation point of view?
Previous studies on the biophysical effects of land cover change are relevant in answering these questions (Anderson et al., 2011;Huang et al., 2018;Wang et al., 2014). The changes in land cover such as deforestation/afforestation have biophysical effects on the earth's climate, which results from changes in surface 60 albedo and moisture and heat fluxes at the surface. The land surface albedo depends on the vegetation type since each vegetation has different optical properties (Gao et al., 2005;Henderson-Sellers and Wilson, 1983;Houldcroft et al., https://doi.org/10.5194/bg-2022-227 Preprint. Discussion started: 6 December 2022 c Author(s) 2022. CC BY 4.0 License. 2009). Therefore, large-scale changes in the vegetation type can significantly affect the earth's climate by changing the surface albedo. The grasslands have a higher albedo than forests. Additionally, in mid-latitudes, the snow-albedo feedback increases the surface albedo further when forests are converted to grasslands (Bonan et al., 1992;Claussen 65 et al., 2001;Thackeray & Fletcher, 2016). Therefore, deforestation in mid-and high latitudes has a cooling effect because of an increase in the surface albedo (Bala et al., 2007;Bathiany et al., 2010;Govindasamy et al., 2001).
Deforestation in the tropical regions also results in a decrease in evaporation, causing a warming effect (Bathiany et al., 2010;Davin & de Noblet-Ducoudre, 2010;Lean & Rowntree, 1993). Therefore, the net effect of deforestation (afforestation) is determined by the balance of the biophysical effects and the biogeochemical warming (cooling) 70 effect from emission (removal) of carbon into (from) the atmosphere. The biophysical effects of afforestation are often neglected even though it could be comparable to the biogeochemical cooling effect of afforestation (Chen et al., 2012, Huang et al., 2018and Shen et al., 2022.
In this study, we compare the climate and carbon cycle effects of afforestation and reduction of fossil fuel emissions by considering two idealized simulations. In the first case, emissions follow three SSP scenarios (SSP2-4.5, 75 SSP3-7.0 and SSP5-8.5) (Meinshausen et al., 2020) , and some amount of carbon is removed by afforestation. In the second case, fossil fuel emissions are reduced by the same amount that is additionally stored on land by afforestation in each of the three SSP scenarios. Figure S1 gives a schematic representation of the two simulations. The final climate state of these two cases is compared to assess the difference in the climate and carbon cycle effects of afforestation and reduced fossil fuel emissions.  (Eby et al., 2009). The large-scale present-day climate is represented quite well in the UVic model (Weaver et al.,2001, Skvortsov et al., 2009, Eby et al., 2009and Cao and Jiang, 2017.

95
First, we spin up the model with the land use data corresponding to the year 1750 (Chini et al., 2014) for 7500 years to a steady state with an atmospheric CO2 concentration of 280.8 ppm ( Figure S2a, Table S1). The last 30 https://doi.org/10.5194/bg-2022-227 Preprint. Discussion started: 6 December 2022 c Author(s) 2022. CC BY 4.0 License. years of this preindustrial spin-up simulation (PI_1750) has a global mean surface air temperature (SAT) of 13.2°C ( Figure S2b, Table S1). Further details of the spin-up simulation are given in SI (Supplementary Information) TEXT S1. A historical simulation (HIST_1750_2005) is performed from 1750 to 2005 starting from the end of PI_1750 by 100 prescribing historical fossil fuel emissions (Hoesly et al., 2018), land cover change (Chini et al., 2014), and volcanic forcing (Crowley, 2000). In the UVic model, land cover change during the historical period is modeled by prescribing the fraction of agricultural land (cropland and pastureland) in each grid. The dynamic vegetation model has representation for five natural vegetation types (broad leaf tree, needle leaf tree, C3 grass, C4 grass, and shrub) and bare soil. The atmospheric CO2 concentration and SAT averaged over the last 30 years (1986)(1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005) of 105 HIST_1750_2005 are 349.1ppm and 13.5°C, respectively ( Figure S3, Table S1). A comparison of our historical simulation with observations shows that the model underestimates the amount of warming in the historical period (SI TEXT S2, Figure S3). The evolution of key climate variables during the historical simulation is shown in Figure S4 In the FIXED_AGR and REDUCED_FF cases, the fraction of the agricultural land is kept constant at values 115 corresponding to the year 2005. Note that the five natural vegetation types can compete outside the agricultural land, and thus, the land cover in the FIXED_AGR and REDUCED_FF cases can change dynamically depending on the climate conditions. In the AFFOREST experiment, vegetation is allowed to regrow over the agricultural land by abruptly setting the agricultural land fraction to zero everywhere, which leads to additional storage of carbon in the land and a reduction in the growth of atmospheric CO2. The fossil fuel emissions in FIXED_AGR and AFFOREST 120 cases follow three extended SSP scenarios (SSP2-4.5, SSP3-7.0 and SSP5-8.5; Meinshausen et al., 2020). The fossil fuel emissions peak in the year 2040, 2100 and 2100 in the SSP2-4.5, SSP3-7.0 and SSP5-8.5 scenarios, respectively, and reduces to zero by the year 2250 in the three scenarios. In the REDUCED_FF case, the fossil fuel emissions are reduced from the corresponding SSP scenarios by the same amount of carbon additionally stored over land in the AFFOREST case.

125
The AFFOREST (REDUCED_FF) simulations differ from the FIXED_AGR simulations only by afforestation (reduced fossil fuel emissions) in the AFFOREST (REDUCED_FF) simulations. Thus, the net effect of afforestation (reduced fossil fuel emissions) on the climate system is estimated by comparing the climate state of AFFOREST (REDUCED_FF) case with the FIXED_AGR case. Thus, in our analyses in the following sections, FIXED_AGR case is used as the reference case.

130
We recognize that the term "afforestation" in the real world refers to the intentional human activity of planting of trees to increase forest cover. However, the increase in forest in our AFFOREST simulations is due to dynamical https://doi.org/10.5194/bg-2022-227 Preprint. Discussion started: 6 December 2022 c Author(s) 2022. CC BY 4.0 License. natural evolution of tree type vegetation with no human intervention. Nevertheless, we use the term "afforestation" to refer to the increase in tree cover in these simulations.

Effects of afforestation on land carbon and land surface albedo
In this section, we analyze the effects of afforestation on land carbon and land surface albedo in our simulations. In the AFFOREST case, regrowth of forests in abandoned agricultural land results in an increase in tree fraction from approximately 0.2 to 0.4, while in the FIXED_AGR and REDUCED_FF cases, tree fraction remains nearly unchanged at around 0.2 ( Figure S5) in the three SSP scenarios. The larger tree fraction (averaged over 2471-140 2500) in the AFFOREST case compared to the FIXED_AGR case has similar spatial distribution in the three SSP scenarios, while there is virtually no difference in tree fraction (averaged over 2471-2500) between REDUCED_FF and FIXED_AGR cases everywhere in the three SSP scenarios ( Figure S6).
In our preindustrial spinup simulation, the land carbon stock is 1789 PgC (Averaged over the last 30 years of PI_1750) (Table S1). In the historical simulation, it stays nearly unchanged at the preindustrial value ( Figure S7) as 145 the land carbon averaged over the last 30 years (1986)(1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005) of HIST_1750_2005 is 1779 PgC (Table S1). In the UVic model, the atmosphere to land carbon flux is the difference between net primary productivity (NPP) and the sum of soil respiration and vegetation burning flux (VEGBURN). VEGBURN is estimated as the carbon that is released into the atmosphere either from the removal of natural vegetation for expansion of agricultural land or from the removal of trees and shrubs that regrow on the prescribed agricultural land fraction. A brief description of VEGBURN is 150 provided in SI TEXT S3. Because agricultural land fraction is zero everywhere in the AFFOREST case, VEGBURN is zero in the AFFOREST case ( Figure S8). In the FIXED_AGR, AFFOREST, and REDUCED_FF simulations, NPP increases initially until around the year when emissions peak (2040 in SSP2-4.5 and 2100 in SSP3-7.0 and SSP5-8.5) due to CO2 fertilization effect (Lobell and Field, 2008, Cernusak et al., 2019and Haverd et al., 2020 in which elevated atmospheric CO2 levels lead to increased plant productivity ( Figure S9). The increase in atmosphere to land carbon 155 flux due to this increase in NPP is partly offset by an increase in soil respiration ( Figure S10) due to an increase in SAT.
The land carbon stock initially increases in all nine simulations until near the end of the 21 st century ( Figure   S7) because the increase in NPP is larger than the increase in the sum of soil respiration and VEGBURN during this period. After the 21 st century, emissions decrease, causing NPP to become relatively constant ( Figure S9). Since soil 160 respiration is larger in the SSP3-7.0 and SSP5-8.5 scenarios due to larger warming (Sect. 3.2), land carbon stock  Table S2). In the SSP 5-8.5 and SSP3-7.0 scenarios, the additional carbon stored in land is larger than that of the SSP 2-4.5 scenario (Figure 1), because of the larger CO2 fertilization effect due to larger 175 atmospheric CO2 concentrations. However, added storage of land carbon is more in the SSP3-7.0 scenario than the SSP5-8.5 scenario which has a larger CO2 concentration. This is because the larger temperature in the SSP5-8.5 scenario causes a larger increase in soil respiration than the increase in net primary productivity (NPP) due to CO2 fertilization ( Figure S12). In the AFFOREST simulations, land carbon (averaged over 2471-2500) is larger in regions with forest regrowth ( Figure S13 and S6), while the spatial distribution of land carbon in the REDUCED_FF case is 180 similar to the FIXED_AGR case in the three SSP scenarios ( Figure S13). In the REDUCED_FF case fossil fuel emissions in corresponding SSP scenarios are reduced by the amount of carbon additionally stored over land in the AFFOREST simulations each year ( Figure S14).
In addition to the increased land carbon, afforestation can significantly change the land surface albedo. The land surface albedo in our preindustrial simulation (PI_1750) is 0.28 (Table S1), which remains nearly unchanged in 185 the historical simulation (HIST_1750_2005) ( Figure S15, Table S1). In the FIXED_AGR, AFFOREST, and REDUCED_FF simulations, land surface albedo decreases initially and becomes nearly constant after 2250 in the three SSP scenarios ( Figure S15). The land surface albedo is less in the AFFOREST case than in the FIXED_AGR case by 0.011 in the three SSP scenarios (Figure 2, Table S2), while the changes in land surface albedo in the REDUCED_FF case relative to the FIXED_AGR case is nearly zero in the three SSP scenarios (Figure 2, Table S2).

190
The land surface albedo (averaged over 2471-2500) is lower in the AFFOREST case compared to FIXED_AGR in regions with forest regrowth ( Figure S16 and S6), while in the REDUCED_FF case, the land surface albedo (averaged over the last 30 years) is similar to the FIXED_AGR case everywhere in the three SSP scenarios ( Figure S16).
In summary, we find that afforestation leads to additional carbon storage over land and lower land surface albedo in the AFFOREST case compared to the FIXED_AGR and REDUCED_FF cases where agricultural land 195 fraction is maintained at year 2005 values.

Evolution of Atmospheric CO2 and Surface Air Temperature
The atmospheric CO2 concentration and SAT in our preindustrial simulation (PI_1750) are 280.8ppm and 13.2 °C (averaged over the last 30 years of PI_1750) ( Figure S2, Table S1), respectively. In our historical simulation (HIST_1750_2005), atmospheric CO2 increases due to fossil fuel and land use change emissions. At the end of the 200 historical simulation, atmospheric CO2 concentration increases to 349.1ppm (averaged over 1976-2005) (Figure S3, Table S1), and consequently, SAT increases to 13.5°C ( Figure S3, Table S1).  Table S3). This is due to two reasons: i) the amount of carbon removed by land is larger in the SSP3-7.0 and SSP5-8.5 scenarios because of the larger CO2-fertilization effect as discussed in Sect. 3.1 ii)) larger ocean carbon uptake in the FIXED_AGR case relative to AFFOREST case in the SSP3-7.0 and SSP5-8.5 scenarios compared to SSP2-4.5 (Table S2).

215
The future projections of changes in SAT (averaged over 2471-2500) in our nine simulations relative to HIST_1750 (averaged over 1976HIST_1750 (averaged over -2005 vary from 2°C to 8°C ( Figure S18, Table S3). In the three SSP scenarios, the REDUCED_FF case simulates a smaller SAT increase compared to the AFFOREST and FIXED_AGR cases ( Figure   S18). The afforestation in the AFFOREST case results in a cooling of 0.31°C and 0.1°C and a warming of 0.05°C in the SSP2-4.5, SSP3-7.0, and SSP5-8.5 scenario, respectively, while the reduction of fossil fuel emissions in the 220 REDUCED_FF case results in a cooling of 0.66°C, 0.56°C and 0.36°C in the SSP2-4.5, SSP3-7.0, and SSP5-8.5 scenario, respectively when compared to the FIXED_AGR case ( Figure 4, Table S3).
In the AFFOREST case, the cooling effect of CO2 removal from the afforestation is partly offset by the warming effect of the changes in surface albedo because of the growth of forests. Hence, the AFFOREST case has a larger SAT than the REDUCED_FF case in the three SSP scenarios (Figure 4 and S18). In the SSP3-7.0 and SSP5-225 8.5 scenarios, this offsetting is almost full so that the AFFOREST and FIXED_AGR cases have similar SAT (Figure   4 and S18). However, in the SSP2-4.5 scenario, though the reduction in atmospheric CO2 is smaller (Figure 4 and S18), the cooling effect of CO2 removal is larger as temperature change scales with the logarithm of atmospheric CO2 levels. Therefore, in the SSP2-4.5 scenario, the warming effect of the regrowth of forests does not completely offset the cooling effect of removing atmospheric CO2.

230
The spatial patterns of SAT (averaged over 2471-2500) in the AFFOREST and REDUCED_FF cases are compared with the FIXED_AGR case in Figure 5. The REDUCED_FF case is cooler in all regions with respect to the FIXED_AGR case in the three SSP scenarios (Figure 5), while AFFOREST case shows regional warming in the SSP3-7.0 and SSP5-8.5 scenarios. This regional warming in the AFFOREST case is more prominent over land, where the afforestation results in a lower land surface albedo (Figure 5 and S16).

235
In summary, we find that a reduction in fossil fuel emissions is more effective than afforestation since the cooling benefits of storing atmospheric carbon in vegetation is partly offset by the decrease in the albedo of the surface https://doi.org/10.5194/bg-2022-227 Preprint. Discussion started: 6 December 2022 c Author(s) 2022. CC BY 4.0 License.
in the AFFOREST case. However, afforestation is beneficial for reducing ocean acidification, as shown in the next section.

240
The ocean carbon content in the PI_1750 simulation (averaged over 2471-2500) is 37287 PgC (Table S1). In our historical simulation (HIST_1750_2005), ocean carbon content increases as increasing CO2 levels in the atmosphere results in an increased carbon uptake by the ocean (Figure S19). The increase in ocean carbon content averaged over the period 1976-2005 of HIST_1750_2005 is 82 PgC (Table S1), The cumulative carbon uptake during the historical period is 113PgC, which falls in the observed range of 105±20PgC (Masson-Delmotte et al., 2021).

245
The ocean carbon content increases in the FIXED_AGR, AFFOREST and REDUCED_FF simulations in the three SSP scenarios. The FIXED_AGR case shows the largest amount of ocean carbon content in the three SSP scenarios ( Figure S19), because of larger atmospheric CO2 in the FIXED_AGR case compared to AFFOREST and REDUCED_FF cases. The spatial pattern of the ocean carbon content (averaged over 2471-2500) in AFFOREST and REDUCED_FF cases relative to the FIXED_AGR case shows that the ocean carbon content increase is less in the 250 AFFOREST and REDUCED_FF cases compared to FIXED_AGR case in all regions in the three SSP scenarios ( Figure S20). In the high emissions scenarios (SSP3-7.0 and SSP5-8.5), the reduction in the ocean carbon content in the AFFOREST and REDUCED_FF cases are less compared to SSP2-4.5 (Figure 6 and S19) because of the buffering effect (Middelburg et al., 2020). The reduction of ocean carbon content (averaged over 2471-2500) in the AFFOREST and REDUCED_FF cases compared to the FIXED_AGR case is more pronounced in the surface ocean as the surface 255 ocean adjusts more rapidly to the changes in atmospheric CO2 ( Figure S21). A longer simulation would be required for larger changes in carbon content in the deep ocean.
The surface ocean pH in our preindustrial state is 8.15 (averaged over the last 30 years of PI_1750). By year 2005, the surface ocean pH (averaged over 1976-2005) reduces to 8.09 as the ocean takes up more carbon as atmospheric CO2 increases during the historical period ( Figure S22). In the FIXED_AGR, AFFOREST and 260 REDUCED_FF simulations, surface ocean pH decreases until the fossil fuel emissions reduce to zero in the year 2250 and increases slightly after the emissions cease ( Figure 22). The AFFOREST and REDUCED_FF cases show larger and similar changes in surface ocean pH in comparison with the FIXED_AGR case in the three SSP scenarios ( Figure   7) because of smaller increase in ocean carbon content in the AFFOREST and REDUCED_FF cases (Figure 6 and S19, and Table S3).

265
The AFFOREST and REDUCED_FF cases show larger surface ocean pH (averaged over 2471-2500) in all regions in the three SSP scenarios relative to the corresponding FIXED_AGR cases, because of smaller ocean carbon content as a result of reduced atmospheric CO2 (Figure 8). In the high emissions scenarios (SSP3-7.0 and SSP5-8.5), the increase in surface ocean pH in the AFFOREST and REDUCED_FF cases are less compared to SSP2-4.5 ( Figure   7 and Figure 8) because the reduction in ocean carbon is smaller in higher emissions scenarios ( Figure 6). As discussed in the previous section, the cooling effect of afforestation is offset by the warming effect of surface albedo changes. However, as shown in this section, afforestation is useful to reduce the effects of increased ocean carbon content and thereby ocean acidification.

Conclusions
Afforestation and reduced fossil fuel emissions are two major components of climate change mitigation 275 currently adopted to slow climate change. Understanding the net effects of afforestation and reduced fossil fuel emissions is important for the development of climate mitigation strategies. In this paper, we have shown that the climate response to carbon removal by afforestation and an equivalent reduction in fossil fuel emissions is different because of the biophysical effects of afforestation, which is often neglected in the development of climate mitigation strategies.

280
We have analyzed the relative effectiveness of afforestation and reduction of fossil fuel emissions for mitigating climate change using climate model simulations. Our results show that allowing the forests to grow back by abandoning all the agricultural land in the year 2005 leads to an additional storage of carbon over land of 319.84 PgC, 418.93 PgC, and 379.21PgC by 2500(averaged over 2471-2500 in the SSP 2-4.5, SSP 3-7.0 and SSP5-8.5 scenarios, respectively. If fossil fuel emissions are reduced by the same amount of carbon that is additionally stored 285 over land, the climate is cooler in the reduced fossil fuel emission case compared to the afforestation case. The relative cooling is 0.36°C, 0.47°C and 0.42°C in the reduced fossil fuel emission case compared to the afforestation case in the year 2500 (averaged over 2471-2500) in the SSP 2-4.5, SSP 3-7.0 and SSP 5-8.5 scenario, respectively. In the case of afforestation, the change in vegetation cover from grasslands to forests has a warming effect which nearly offsets the cooling effect from carbon removed from the atmosphere. In our simulations, the cooling effect of afforestation is 290 completely offset by its warming effect in the higher emission scenarios (SSP 3-7.0 and SSP 5-8.5) and partially offset in lower emission scenario . This suggests that afforestation may have a larger climate benefit in the lower emission scenarios.
There are several limitations to our study. First, the afforestation in our model is highly idealized. In our afforestation simulations, we assume that the entire agricultural land in the year 2005 is abandoned and vegetation is 295 allowed to regrow abruptly, while in the real-world implementing afforestation at this scale would take a longer period.
Also, in our simulations, vegetation grows back naturally according to the climate conditions over the abandoned agricultural land, while in the real world, it might be possible to grow trees artificially in areas where the climate conditions do not support the growth of trees. Second, many processes in the model are highly simplified representations aimed at achieving a lower computational cost. For example, the dynamic vegetation model in our 300 simulation has only five plant functional types, while the real-world ecosystems are far more diverse and complex.
However, the simplified representation enables us to understand the role of climate-vegetation feedbacks in longer time scales with less computational cost. Even though the afforestation representation in our model is highly idealized and there are uncertainties in the processes that are represented in the model, we believe that the qualitative conclusions would not be affected by these limitations. Based on our results, we conclude that a reduction in fossil fuel emissions is more effective than afforestation in mitigating climate change. Though afforestation is relatively less effective in mitigating climate change, it has other benefits such as reducing ocean acidification: the removal of carbon from the atmosphere results in slightly reduced carbon in the ocean, which leads to higher surface ocean pH and less ocean acidification. Therefore, a better strategy to address climate change is to reduce fossil fuel emission as well as pursue afforestation efforts.

Data availability
All data that support the findings of the study will be made available at the Zenodo database. DOI: 10.5281/zenodo.7321684.

Author Contribution
Govindasamy Bala formulated the idea behind the study. Govindasamy Bala and K U Jayakrishnan designed the 315 experiments. K U Jayakrishnan performed the experiments. Govindasamy Bala and K U Jayakrishnan contributed to the writing and editing of the manuscript.   The REDUCED_FF case has lower SAT than the FIXED_AGR case in the three SSP scenarios case because of reduced fossil fuel emissions in the REDUCED_FF case. In the AFFOREST case, the cooling effect of removal of CO2 is nearly offset by the warming effect of regrowth of forests. Hence, the AFFOREST case has similar SAT as that of FIXED_AGR in the SSP3-7.0 and SSP5-8.5 scenarios. However, in the SSP2-4.5 scenario, though the reduction in atmospheric CO2 is smaller (Figures 4 and S17), the cooling effect of removal is larger because 450 global mean temperature change scales with the logarithm of atmospheric CO2 levels. Therefore, in the SSP2-4.5 scenario, the warming effect of the regrowth of forests does not completely offset the cooling effect of removing atmospheric CO2. https://doi.org/10.5194/bg-2022-227 Preprint. Discussion started: 6 December 2022 c Author(s) 2022. CC BY 4.0 License. Figure 5. The left (right) panel shows the spatial pattern of the difference in global mean surface air temperature (SAT) averaged over the last 30 years between the AFFOREST (REDUCED_FF) and FIXED_AGR cases. The top, middle and bottom panels correspond to the SSP2-4.5, SSP3-7.0 and SSP5-8.5 scenarios, respectively. The REDUCED_FF case shows lower SAT everywhere relative to the FIXED_AGR case in the three SSP scenarios, while the AFFOREST case shows regional warming in the SSP3-7.0 and SSP5-8.5 scenarios. Note that the regions of 460 warming in the AFFOREST case is more prominent over land where the forest regrowth results in a lower land surface albedo ( Figure S15). https://doi.org/10.5194/bg-2022-227 Preprint. Discussion started: 6 December 2022 c Author(s) 2022. CC BY 4.0 License. Figure 6. Changes in global total ocean carbon content in the AFFOREST (green; ΔAFFOREST) and REDUCED_FF (blue; Δ REDUCED_FF) cases relative to the FIXED_AGR case in the a) SSP2-4.5 b) SSP3-7.0 and c) SSP5-8.5 scenarios. The AFFOREST and REDUCED_FF cases have smaller ocean carbon than the FIXED_AGR case in the three SSP scenarios because of the reduction of atmospheric CO2 in the AFFOREST and REDUCED_FF cases by afforestation and reduced fossil fuel emissions, respectively.