Human activities have led to substantial transformations in Earth’s natural vegetation, with forests particularly impacted due to extensive conversion to agricultural land. This large-scale land cover change (LCC) has significantly influenced the Earth system, affecting nearly half of the global land surface . As a central component of Earth’s biophysical and biogeochemical systems, the biosphere is markedly altered by LCC as it changes major biogeochemical cycles and modifies interactions with the atmosphere . Forests, which store approximately 45 % of the world’s terrestrial carbon, act as a critical carbon sink, capturing and sequestering carbon that would otherwise remain in the atmosphere and contribute to global warming . Additionally, forests play a key role in regulating the hydrological cycle through evapotranspiration, which influences atmospheric moisture levels, cloud formation, and precipitation patterns, ultimately impacting regional and global climate, including surface temperature . Land cover further modulates the planetary albedo; dense forest canopies can absorb up to 90 % of incoming solar radiation, thereby impacting surface reflectivity and playing a main role in the Earth’s surface energy balance .

The terrestrial biosphere is the primary source of biogenic volatile organic compounds (BVOCs), such as isoprene and various terpenes, which comprise nearly 90 % of total volatile organic compound (VOC) emissions into the atmosphere . Biogenic VOCs are highly reactive, readily interacting with tropospheric oxidants to produce lower volatility oxidation products that can partition into the aerosol phase, contributing to the formation of biogenic secondary organic aerosols (bSOA) and also participating in new particle formation in pristine regions .

The reactivity of BVOCs has been shown to influence global atmospheric aerosol loading, with implications for cloud properties and radiative balance . Numerous modelling studies have investigated how LCC influences aerosol loading, highlighting significant atmospheric and climate impacts. In particular, deforestation has been shown to reduce the global atmospheric aerosol burden, resulting in a net radiative warming effect .

Biogenic VOC oxidation not only contributes to atmospheric aerosol formation but also plays a crucial role in determining the concentrations of key atmospheric oxidants, such as hydroxyl (OH) radicals, ozone (O₃), and nitrate (NO₃) radicals, thereby modulating the atmosphere's oxidative capacity . The predominant reaction pathway for BVOCs is with OH radicals, which are primarily formed through the photodissociation of ozone in the presence of water vapour . While BVOCs initially deplete OH radicals through oxidation, they also contribute to OH recycling through subsequent reactions and, therefore, maintain the oxidation capacity of the atmosphere . This complex interaction, often termed OH reformation, is essential for sustaining the troposphere's self-cleaning ability in view of the high levels of reactive BVOCs emitted by vegetation. The recycling process involves the formation of peroxy radicals (RO₂), which react with nitrogen oxides (NO_x) to produce ozone (O₃) and other oxidised products. As a result, BVOCs modulate O₃ concentrations in the troposphere, which acts as both a pollutant and a greenhouse gas . In addition to daytime oxidation by OH, BVOCs can also be oxidised by O₃, and NO₃ radicals during the night, further influencing their atmospheric lifetimes and the concentrations of other trace gases . Changes in BVOC emissions and, therefore, OH reactivity also influence the atmospheric lifetimes of methane (CH₄) and carbon monoxide (CO), as OH radicals are a primary sink of these gases .

As forests are converted to agricultural or urban land, BVOC emissions decrease, changing the delicate balance of atmospheric chemical reactions and influencing global oxidant chemistry. Several studies (e.g. ) have suggested that lower BVOC emissions from deforestation lead to higher OH concentrations, thereby shortening the lifetimes of CH₄ and O₃ and influencing global radiative forcing. In contrast, increased BVOC emissions in future reforestation scenarios are found to have the opposite effect . At the same time, some studies suggest that the global atmospheric oxidising capacity, and thus methane lifetime, remains largely unaffected by LCC . In addition to changes in BVOC emissions, land cover practices involve direct emissions in the atmosphere, for example, black carbon (BC) and carbon monoxide (CO) from agricultural waste burning and NO_x and NH₃ emissions from manure management . LCC-driven NO_x emissions account for approximately 5 % of global emissions and play a key role in the HO_x cycle, significantly contributing to OH budgets alongside BVOCs .

In this work, we apply the chemistry–climate model EMAC coupled with the dynamic global vegetation model LPJ-GUESS to investigate the impacts on oxidant chemistry from changes in BVOC emissions and surface anthropogenic emissions following crop and grazing land expansion on potential natural vegetation (PNV). Potential natural vegetation refers to the type of vegetation that would naturally occur in a specific area under certain climate, soil, and environmental conditions without human influence. Land cover change is simulated through a deforestation routine in LPJ-GUESS (which simulates PNV), systematically clearing crop and grazing land areas based on 2015 land cover data. Additionally, we explore an extreme reforestation scenario where present-day grazing land is restored to natural vegetation. This work provides new insights into LCC-driven impacts on biogeochemical cycles by using state-of-the-art chemical representations in EMAC coupled with interactive vegetation calculations. This work provides comprehensive estimates of changes in atmospheric composition and non-aerosol radiative effects resulting from land cover change. For the first time, we present evidence that land cover change can affect the formation sensitivity of ozone, with important implications for both air quality and climate.

Changes in land cover can substantially impact atmospheric processes and, in turn, the climate system. Numerous studies have explored how land cover changes alter the climate by affecting carbon fluxes, the hydrological cycle, surface albedo, aerosols, and atmospheric chemistry. This study specifically examines the impact of land cover changes on atmospheric composition from perturbed BVOC and anthropogenic emissions associated with LCC. In , we showed that the model setup employed in this work can represent changes in vegetation cover and BVOC emissions in the scenarios considered (i.e. PNV, DCGL, and DCL). This study, which focuses on atmospheric chemistry, introduces two key modifications relative to the setup in : first, aerosol–cloud interactions are not included, thereby excluding aerosol-induced meteorological feedbacks; second, MEGAN-derived BVOC fluxes are tuned to better represent the present-day O₃ burden. All simulations are nudged to observed meteorology, ensuring consistent large scale transport and meteorological conditions across experiments. Additionally, dry deposition fluxes are computed interactively using LAI information from LPJ-GUESS and soil NO emissions were also modified to account for changes in land cover. We emphasise that while vegetation states are fully determined by EMAC prognostic variables, only BVOC emissions and dry deposition fluxes are interactively fed back from the vegetation calculations. This means that other feedbacks associated with land cover change (e.g. changes in evapotranspiration, albedo, roughness length, and aerosols) are suppressed, allowing us to focus solely on the impact of land cover change on oxidant chemistry via perturbed BVOC fluxes and surface emissions from LCC. Table 2 summarises the changes in vegetation and atmospheric states for the different scenarios considered in this work: (i) the difference between 2015 land cover and natural vegetation (DCGL - PNV) and (ii) the impact of restoring grazing land relative to 2015 land cover (DCL - DCGL). Changes are presented in both absolute and relative terms.

The oxidation capacity of the atmosphere is primarily controlled by hydroxyl (OH) radicals in the troposphere. Consistent with previous studies, our model calculations show the highest OH concentrations in the tropical troposphere, with a global mean (day and night) tropospheric OH of 11.0 × 10⁵ molecules cm⁻³ in the DCGL scenario, representative of present-day conditions. This value is close to the 11.3 × 10⁵ molecules cm⁻³ reported by and the multimodel mean of (11.1 ± 1.6) × 10⁵ molecules cm⁻³ derived for the year 2000 by .

Present-day land cover, relative to natural vegetation, increases the global mean surface OH concentration by 5.7 % while restoring current present-day grazing land would decrease it by 4.0 %. In both scenarios, surface OH changes correspond to regions where BVOC emissions were perturbed due to deforestation and reforestation. The EMAC results suggest that land-cover-driven BVOC emissions strongly perturb OH concentrations in the tropical troposphere, with relatively strong changes extending to the upper troposphere (Figs. 2c and S2c). While these changes are statistically significant, they remain small relative to the BVOC emissions changes (-25.9 % in the deforestation scenario and +21.3 % in the reforestation scenario). This is due to the efficient recycling of OH radicals, which buffers the global OH burden . Our simulations include OH recycling chemistry as detailed in . suggests a global annual tropospheric OH mean of 11.3 × 10⁵ molecules cm⁻³ based on the year 2000. Simulations that account for climate change, CO₂ trends, and land cover change project an OH increase of 1 % in 2050 and a decrease of 2 % in 2100 . One should note these changes reported in are primarily driven by the projected increase in BVOC emissions (isoprene: +10 % in 2050 and +25 % in 2100) due to climate warming. Our findings also align with , who suggest that significant local changes in the atmospheric oxidising capacity are confined to areas influenced by LCC, mostly in the tropics. estimates that complete global deforestation increases the annual tropospheric mean OH concentration from 13.6 × 10⁵ to 14.6 × 10⁵ molecules cm⁻³ (+7.4 %). This agrees with the 5.7 % increase simulated by EMAC under more realistic land cover changes.

Table S1 highlights the differences in prescribed global annual emission burdens across the simulations. The DCGL - PNV scenario shows increases of 37.1 Tg in CO, 35.25 Tg in NH₃, and 2.84 Tg in NO_x. In contrast, the DCL - DCGL scenario exhibits decreases of 37.1 Tg in CO, 26.58 Tg in NH₃, and 2.64 Tg in NO_x, reflecting the elimination of emissions associated with grazing land practices as these areas are restored.

We report a 7.5 ppbv decrease in the mean annual surface CO mixing ratio in the 2015 land cover scenario relative to pristine vegetation. As OH is the primary sink for CO, the increased availability of OH radicals enhances CO depletion. Notably, this 6.2 % reduction occurs despite an additional 37.1 Tg annual CO emissions burden from grazing land (mostly biomass burning), highlighting the dominant role of atmospheric oxidation in regulating CO levels . In DCL - DCGL, the 37.1 Tg decrease in surface emissions from grazing land is offset by increased BVOC emissions from reforestation, which lead to a reduction in OH radicals. This reduction in OH limits CO depletion, resulting in a net increase of 4.6 ppbv (4.0 %). We note that changes in CO, modulated by OH, may also feed back to influence OH concentrations, and vice versa. For example, the reduction in CO in the deforestation scenario may partly contribute to the observed increase in OH, while elevated OH levels in turn accelerate CO loss. Nevertheless, we argue that changes in BVOC emissions remain the primary driver of OH variability. OH enhancements resulting from reduced BVOC oxidation likely exert a stronger influence on CO concentrations than CO does on OH, reinforcing the observed decrease in CO.

In the DCGL - PNV scenario, a 2.84 Tg increase in prescribed NO_x emissions leads to a 7.8 % rise in the global mean NO_x mixing ratio. This increase is also partially driven by higher soil NO emissions (+3.1 Tg) due to deforestation, as reduced canopy deposition enhances their release, consistent with . Despite the increased availability of OH, the substantial rise in NO_x emissions outweighs the enhanced removal of NO_x via oxidation by OH. The increase in surface NO_x also promotes more efficient OH recycling, which can contribute to some of the observed increase in OH concentrations close to the source. Similarly, in the DCL - DCGL scenario, a 2.64 Tg reduction in NO_x emissions, along with a 2.3 Tg decrease in soil NO emissions, dominates surface NO_x levels, leading to a 6.4 % decline.

Land cover changes significantly shift both NO_x and VOC emissions, influencing the O₃ production and distribution. Table 2 confirms that, globally, surface ozone remains relatively unchanged; however, significant regional variations are evident (Fig. 6b). These changes are closely linked to ozone formation sensitivity. As shown in Fig. 6b, urban and polluted areas, which are typically VOC-limited, experience ozone production constrained by VOC availability. In these regions, reduced BVOC emissions lead to lower peroxy radical (RO₂) concentrations, slowing the conversion of NO to NO₂ and ultimately reducing O₃ formation rates. This results in the observed decrease in surface O₃ mixing ratios. Conversely, in rural areas of tropical South America, Africa, and SE Asia, ozone production is more sensitive to NO_x levels. Increased surface NO_x (from farming) enhances NO₂ formation, which photolyses to produce O₃, leading to higher regional O₃ levels. Our analysis using the sensitivity metric α(CH₃O₂) indicates that, under present-day conditions compared to pristine vegetation, ozone production shifts towards a VOC-sensitive regime, except over the Southern Ocean, which appears to shift towards a NO_x-sensitive regime (Fig. 7b). In remote marine environments, such as the Southern Ocean, BVOC concentrations are extremely low, making ozone formation primarily NO_x-sensitive. In these regions, the perturbation of BVOCs has minimal impact on ozone production, and even a slight increase in background NO_x levels can significantly enhance ozone production through NO_x-dependent photochemical processes.

Hotspots where ozone production sensitivity changes drastically were identified and plotted on an α(CH₃O₂) vs. NO curve. The selected locations are Central North America (100° W, 46° N), Southeastern South America (59° W, 28° S), West Africa (11° W, 20° N), and South Asia (91° E, 34° N). Figure 7c confirms that, in these regions, ozone production generally shifts toward a VOC-sensitive regime. While South America and Africa largely remain in the transition regime, a clearer shift is observed in North America, where ozone production moves from the transition regime to a VOC-sensitive regime. In DCL - DCGL, we observe opposite signals (Fig. S7), with ozone production sensitivity shifting toward a NO_x-sensitive regime. Similar trends were presented in , who showed that reforestation efforts in southern China increased BVOC emissions. This led to a reduction in surface O₃ by 1.6–2.5 ppbv in rural regions but caused an increase in O₃ peaks by up to 2.0–6.0 ppbv in VOC-limited urban areas. It is important to note that we limit changes in surface emissions related to LCC and do not account for past or future variations in anthropogenic emissions, which would also influence ozone production sensitivity .

Tropospheric O₃ and CH₄ are potent greenhouse gases with global warming potentials (GWPs) significantly higher than CO₂. As a result, even small changes in the abundance of these gases can impact the climate . We find that deforestation has had an indirect cooling effect on the climate, with RE_O3 and RE_CH4 of -10 mW m⁻² and -50.0 mW m⁻², respectively. In contrast, restoring grazing land would result in a positive radiative effect of 1.3 mW m⁻² for O₃ and 40.0 mW m⁻² for CH₄. Changes in RE_O3 are not statistically significant; however, this does not rule out the presence of meaningful effects that may be obscured by climate variability.

estimates a global annual mean tropospheric RE of CH₄ at -70 mW m⁻² and of O₃ at 170 mW m⁻² in a complete deforestation scenario. However, in , RE_CH4 is inferred from changes in steady-state CH₄ concentration, whereas in our calculations, RE_CH4 is derived from changes in CH₄ lifetime due to prescribed surface emissions. Additionally, RE_O3 in is calculated using a radiative kernel approach that accounts for stratospheric adjustments over the annual cycle, while we compute RE_O3 based on instantaneous radiative effects in the model. estimates an effective radiative effect (ERE_O3) between 32 and 57 mW m⁻², and ERE_CH4 between 7 and 20 mW m⁻² in a future reforestation scenario. For a fair comparison with the ERE_CH4 reported in , we have to scale our -50 to -43 mW m⁻² to account for additional chemical production of ozone and stratospheric water vapour, and accounting for the negative cloud adjustment including SW absorption by CH₄ . We also calculate the instantaneous RE_O3, excluding any adjustment (e.g. from stratospheric temperatures).

Given the complex response of O₃ to BVOC changes, RE_O3 exhibits high spatial and interannual variability compared to CH₄. As shown in Fig. 8, the top of atmosphere RE_O3 includes regional hotspots corresponding to increases in surface ozone mixing ratios in DCGL - PNV over North America, Europe, and Asia (Fig. 6b). Our findings are consistent with previous work, which shows a negative RE for O₃ and CH₄ linked to deforestation and a positive one linked to reforestation . We show that in our model, changes in RE_O3 is not as strong as previously suggested . Nevertheless, this discrepancy may partly arise from the slightly different model setups and methods used to estimate the radiative effects.

Based on the reforestation scenario, our results suggest that many biogeochemical and atmospheric responses scale approximately with the extent of tree cover restored. Biomass carbon storage, isoprene and monoterpene emissions, and changes in methane lifetime and RE_CH4 exhibit substantial, though partial, reversibility, indicating a relatively linear response to land cover change. However, the ozone response reveals a notable non-linearity. Despite reductions in NO_x levels, ozone mixing ratios show only modest recovery. This muted response likely reflects a shift in chemical regimes: as shown in Fig. 9f, reforested areas become increasingly NO_x-sensitive, reducing ozone production efficiency even in the presence of elevated biogenic precursor emissions.

Understanding the effects of land cover changes on atmospheric composition is essential for assessing the impact of human activities on the Earth system, both historically and in the future. This study provides a detailed analysis of two land cover change scenarios, examining how variations in BVOC emissions and emissions from crop and grazing land affect atmospheric oxidant chemistry. We focus on tropospheric changes in OH radicals, CO, NO_x, O₃, and CH₄, as well as their radiative effects. We find that present-day deforestation relative to pristine vegetation decreases global BVOC emissions. These reductions significantly alter atmospheric chemistry, particularly through their impact on oxidant levels. Global surface OH concentrations increase by 5.7 %, leading to a reduction in the oxidation of CO and a consequent 6.2 % decrease in its surface mixing ratio. In contrast, the global surface NO_x mixing ratio increases by 7.8 %, primarily due to enhanced emissions from cropland and grazing land, as well as increased soil NO emissions from reduced canopy deposition. Ozone changes are more complex, with a global surface mean mixing ratio decrease of less than 1 ppbv (about 1 %). However, significant regional variations occur, driven by shifts in ozone production regimes. Our assessment using the sensitivity metric α(CH₃O₂) indicates a consistent shift towards a VOC-sensitive ozone formation regime in the deforestation scenario. Shorter lifetimes of O₃ and CH₄ result in a small radiative effect of -10 mW m⁻² and -50 mW m⁻² (cooling), respectively. This cooling partially offsets the warming effect from reduced bSOA, which estimates to be 60.4 mW m⁻². However, a fully coupled Earth system model is necessary to capture the full interplay between these processes, as it enables all relevant feedbacks to be represented. Without such coupling, important secondary effects and compensating mechanisms may be overlooked, potentially leading to an incomplete understanding of the net climate impact. Ongoing land cover change may continue to reduce vegetation cover in pristine forested regions. In areas with low background concentrations of nitrogen oxides, such changes can promote further increases in surface ozone levels due to a shift toward VOC-sensitive ozone production. Increased BVOC emissions from large scale reforestation influence atmospheric oxidant chemistry in a manner largely opposite to deforestation, though with weaker intensity. However, the ozone response exhibits a marked non-linearity.