Careful selection of forest types in afforestation can increase carbon sequestration by 25% without compromising sustainability

Stringent climate mitigation pathways limiting warming to 2 °C or 1.5 °C rely on rapid emission reductions and substantial carbon dioxide removal (CDR) later this century. Land-based CDR options—afforestation/reforestation and bioenergy with carbon capture and storage (BECCS)—feature prominently in IPCC scenarios, but their large-scale deployment can create trade-offs for food security, biodiversity, and broader sustainability. Afforestation may conflict with ecosystems if natural open habitats are converted, and both afforestation and BECCS compete for land, potentially raising food prices and hunger risk. Prior work has explored CDR costs and potentials, but the implications of selecting different forest types for afforestation—and how such choices interact with food-related measures—remain underexplored. This study asks whether careful forest-type selection can increase carbon sequestration without exacerbating adverse impacts on food and land systems, and how complementary food-system measures modify these outcomes.

The study builds on IPCC assessments and integrated assessment modelling literature that positions afforestation/reforestation and BECCS as key CDR options for achieving 2 °C/1.5 °C pathways. Previous work estimates mid-century CDR volumes and mitigation costs for BECCS and afforestation and highlights feasibility concerns related to food security, biodiversity, and social acceptability. Research has also warned of biodiversity risks from afforesting naturally open ecosystems and emphasized the need for integrated strategies that align climate and biodiversity objectives. Studies indicate that food-system measures (agricultural intensification, trade liberalization, dietary changes) can mitigate land-pressure and enhance sustainable CDR deployment. However, existing IAMs generally do not differentiate forest types in afforestation or quantify how forest-type selection combined with food measures could influence sequestration potential and sustainability outcomes. This paper addresses that gap by explicitly modelling forest-type choices and their interaction with food-policy levers.

The authors employ an integrated framework coupling: (1) AIM/Hub (an economic IAM determining regional land demand across 17 global regions under a carbon budget and global uniform carbon price), (2) AIM/PLUM (a land-use allocation model downscaling regional land demands to 0.5° grids to allocate cropland, pasture, afforestation, and bioenergy crops to maximize landowner profit based on biophysical yields), and (3) VISIT (a gridded terrestrial vegetation model simulating vegetation and soil carbon dynamics and net primary production, NPP). A century-long global carbon budget of 600 GtCO2 consistent with a 2 °C target is imposed, with no negative emissions allowed after achieving net-zero CO2 emissions. Nine mitigation scenarios that meet the 2 °C target are analysed: varying the availability of land-based CDR (afforestation-only vs BECCS-only), forest-type selection schemes, and with/without food-related measures, plus a conventional scenario without mitigation or food measures. Forest-type selection schemes: Aff-Cur (native forest type), Aff-Div (select the most carbon-intensive forest type within the same agro-ecological zone, balancing carbon and ecological considerations), and Aff-Cmax (select the most carbon-intensive among 12 global forest types, maximizing carbon regardless of ecological fit). Forest growth is represented using a growth function (after Sohngen et al.) parameterized via VISIT-derived NPP: V = 8*exp[AB/age] with B=30 and 8=1; parameter A estimated per forest type so growth at age 20 matches VISIT NPP; maximum tree carbon set to 300 MgC/ha; tree age accumulates since planting. VISIT calculates NPP for 12 forest types by a hypothetical experiment where all land is afforested with a single type from 2010 onward; RCP4.5 climate is used for NPP, acknowledging inconsistency with the 2 °C mitigation scenario but deemed non-critical to conclusions. Natural disturbance via fire is included; other disturbances and afforestation operation emissions are excluded. BECCS yields are from the H08 hydrological model, with carbon sequestration from BECCS computed from bioenergy crop yields, the regional share of BECCS in total electrification, and the regional share of CCS in bioenergy electrification (from AIM/Hub). Land intensity of carbon sequestration (LIC) is defined as potential carbon sequestration per unit land area for afforestation and BECCS. Food-related measures include agricultural intensification and trade globalization (aligned with SSP1), dietary change to achieve the EAT-Lancet target by 2050 and maintain thereafter, and improved equity in food distribution (country CV of dietary energy consumption reaching 0.1 at per-capita GDP USD 50,000). Impacts on food security are assessed using FAO methodology: hunger risk equals the share of population with calorie intake below the mean minimum dietary energy requirement (M), assuming a lognormal/normal distribution parameterized by mean per-capita calorie intake and the coefficient of variation (CV). AIM/Hub computes calorie availability from commodity balances and conversion factors accounting for edible shares. Additional indicators include electricity price, GDP deviation from baseline, crop price index, water withdrawal for irrigation, and nitrogen fertilizer use (for food and bioenergy crops; not for afforestation due to data limitations). Regional results are reported for OECD and EU, Reforming Economies (REF), Asia, Middle East and Africa (MAF), and Latin America (LAM).

- Forest-type selection boosts afforestation carbon sequestration: selecting within-zone carbon-intensive types (Aff-Div) increases global sequestration by about 2% vs native (7.7 vs 7.6 GtCO2/yr in 2100). Selecting the most carbon-intensive type globally (Aff-Cmax) raises sequestration by about 25% vs native (9.5 vs 7.6 GtCO2/yr in 2100). - Regional gains: In 2100, Latin America sees +5.2% (Aff-Div) and +37% (Aff-Cmax) vs native, while Reforming Economies increase by +4.3% and +18%, respectively. Optimal types differ by region (e.g., tropical montane or dry forests in LAM; southern taiga in South Russia; semiarid wood in parts of East Asia). - Land-intensity (LIC) differences: Afforestation LIC rises with carbon-intensive types but remains below BECCS. In 2100, afforestation LIC (global mean) is 4.0 (Aff-Cur), 4.1 (Aff-Div), and 5.0 tCO2eq/ha/yr (Aff-Cmax). In Aff-Cmax, afforestation LIC is 7.7 and 5.0 tCO2eq/ha/yr in 2050 and 2100; BECCS LIC is ~15 and 19 tCO2eq/ha/yr in the same years. BECCS carbon yield per area exceeds afforestation by more than an order of magnitude. - BECCS vs afforestation sequestration: Even when maximizing afforestation (Aff-Cmax), global sequestration (8.5 GtCO2/yr in 2050; 9.5 in 2100) is below BECCS-only (7.3 and 14 GtCO2/yr in 2050 and 2100). - Land-use effects: Afforestation reduces cropland and pasture to expand forest; BECCS expands cropland for bioenergy while potentially allowing cropland for food to increase due to higher LIC and lower land competition per unit carbon. - Food-policy measures increase sequestration: Combining forest-type selection with food measures further increases afforestation sinks. Aff-Cmax with food measures (Aff-CmaxFodPol) reaches 11.3 GtCO2/yr in 2100, +19% vs Aff-Cmax without food measures and +49% vs Aff-Cur without food measures. - System impacts: Afforestation-only scenarios increase carbon prices, electricity prices, and GDP losses more than BECCS-only scenarios because afforestation’s lower LIC induces greater land competition and reliance on more expensive renewables. Both afforestation and BECCS raise land and food prices; afforestation generally poses a higher hunger risk than BECCS for a given carbon target due to larger land requirements. - Food measures mitigate some trade-offs: Food measures reduce carbon prices and GDP losses for both afforestation and BECCS. Their effect on hunger depends on distributional improvements: dietary shifts alone can increase hunger risk if food distribution inequality is not improved sufficiently, even as over-consumption falls. - Resource use: Afforestation scenarios entail lower nitrogen fertilizer use and irrigation water withdrawal than BECCS in the indicators considered (which cover food and energy crops, not afforestation inputs).

The study demonstrates that explicitly selecting high-growth, carbon-intensive forest types for afforestation can substantially raise global carbon sequestration at century’s end relative to planting native types, partially addressing the challenge of afforestation’s lower land productivity compared to BECCS. However, because afforestation still has a lower land intensity of carbon sequestration than BECCS, achieving a given climate target with afforestation alone induces stronger competition for land, elevating food and energy prices and causing larger GDP losses. BECCS is generally more effective at carbon removal per unit area, thereby reducing land competition, but faces social acceptability and feasibility concerns. Combining forest-type selection with supportive food-system measures (agricultural intensification, trade globalization, dietary change, improved food distribution) enables larger afforestation sinks while moderating adverse economic and food-system impacts. The ultimate effect on hunger hinges on improvements in food distribution equality; without adequate progress, dietary changes can raise hunger risk even as global over-consumption falls. Strategically mixing land-based CDR options, implementing complementary food policies, and undertaking early emissions reductions to avoid temperature overshoot can lessen reliance on land-based CDR and minimize trade-offs across economy, energy, food, land use, and biodiversity.

This work introduces forest-type selection into an integrated assessment framework and shows that choosing carbon-intensive forest types can increase afforestation carbon sequestration by up to 25% relative to native types and by up to 49% when combined with food measures, while highlighting the need for complementary policies to protect food and land sustainability. BECCS remains more land-efficient and can deliver larger carbon removal, but its deployment entails uncertainties regarding social acceptability and sustainability. The results support a best-mix strategy that combines forest-type-optimized afforestation with food-system measures and BECCS, tailored regionally to land availability and productivity, to meet a 2 °C target with fewer adverse impacts. Future research should extend modelling to additional land-based CDR options (e.g., agroforestry, biochar, soil carbon), incorporate biodiversity impact modules, account for the costs and feasibility of food measures, quantify albedo and climate feedbacks of large-scale afforestation, include water and nutrient use for afforestation and bioenergy crops, and undertake multi-model intercomparisons to assess uncertainty in forest-type selection effects.

- Scope of CDR options: Only afforestation/reforestation and BECCS are modelled; other land-based CDR options (agroforestry, biochar, soil carbon management) are excluded due to sectoral coverage and parameterization challenges. - Biodiversity modelling: Impacts on biodiversity are not dynamically modelled; more detailed biodiversity modules are needed to assess ecological trade-offs. - Food-measure costs: The economic costs and barriers to implementing food-related measures are not included, potentially biasing feasibility assessments. - Biogeophysical effects: Albedo reductions and associated radiative forcing from large-scale afforestation are not accounted for, nor are full climate–vegetation feedbacks; only fire disturbance is considered among natural disturbances. - Climate scenario inconsistency: VISIT NPP and afforestation growth use RCP4.5 climate, which is inconsistent with the 2 °C mitigation pathway used in AIM; this may affect absolute sequestration estimates. - Resource inputs: Water and nutrient inputs for afforestation and bioenergy crops are not fully represented (indicators cover only food and energy crops; afforestation inputs are omitted due to data limitations). - Single-model dependence: Results derive from a single integrated framework; multi-model analyses are needed to test robustness and quantify uncertainty, including the magnitude of forest-type selection benefits.