Uncertainty in land-use adaptation persists despite crop model projections showing lower impacts under high warming

Climate change significantly impacts crop yields and resource availability. Understanding the potential and costs of agricultural land-use adaptation under varying warming scenarios is crucial. While incremental adaptation measures (e.g., improved irrigation, resilient cultivars, adjusted planting dates) can enhance resilience at the farm level, transformative changes at the land-use system level (e.g., cropland expansion, shifts in crop types, technological advancements, altered trade flows) are also necessary. Socioeconomic factors significantly influence adaptation decision-making, potentially slowing adoption rates. Most global land system models incorporate autonomous adaptation, but existing agroeconomic studies predominantly focus on specific measures, often neglecting the combined effects of multiple strategies and CO2 fertilization. This study addresses these gaps by using a comprehensive global land system model and incorporating the latest multi-model crop yield impact data, including CO2 fertilization effects, to assess the full potential and associated costs of land-use adaptation under different climate change scenarios.

Previous agroeconomic studies on climate change adaptation in agriculture have primarily focused on individual adaptation strategies, such as trade, cropland expansion, and investments in research and development (R&D). Studies often overlooked CO2 fertilization effects, leading to largely pessimistic projections of climate change impacts. Iizumi et al. (2020) estimated global adaptation costs at the farm level but did not consider transformative land-use changes or regional dynamics. This study builds upon previous work by incorporating a broader range of adaptation strategies, accounting for CO2 fertilization, and employing a more comprehensive land system model to analyze adaptation at both the farm and land-use levels. It leverages the latest climate change data from CMIP6 and crop model projections from AgMIP's GGCMI phase 3 ensemble.

This research utilizes the global land system Model of Agricultural Production and its Impact on the Environment (MAgPIE) version 4.4.0 to assess land-use adaptation. MAgPIE minimizes AFOLU costs based on spatial agricultural productivity, food and material demands, and international trade. The model considers various adaptation strategies: relocation of cropping areas (intra and internationally), irrigation, cropland expansion, technological intensification, and shifts in crop mixes. Two scenarios are analyzed: a low-emissions scenario (SSP1-RCP2.6) and a high-emissions scenario (SSP5-RCP8.5). Corresponding SSPx-NoCC scenarios (no climate change) serve as baselines. Climate impacts are incorporated through harmonized and calibrated crop yield data from nine Global Gridded Crop Models (GGCMs) (CYGMA1p74, EPIC-IIASA, LPJmL, CROVER, ISAM12, LandscapeDNDC, PEPIC, PDSSAT, PROMET) driven by five Climate Models (GCMs) (GFDL-ESM4, MRI-ESM2-0, UKESM1-0-LL, MPI-ESM1-2-HR, IPSL-CM6A-LR). Adaptation costs are calculated as the difference in costs between SSPx-RCPy and SSPx-NoCC scenarios, categorized into land conversion, intensification, irrigation, and trade & transport. The analysis includes global and regional assessments of adaptation responses and costs. The smoothing spline interpolation method is employed to extract long-term climatic trends and minimize the impacts of short-term weather variability, and data from the different GGCMs is further harmonized to the baseline simulation using a limited calibration approach. Post-processing calculations isolate the individual and combined effects of not adapting cropland patterns and technological change on crop production. The self-sufficiency ratio (SSR) is also calculated to assess regional production-demand balance. The shifts in crop mixes (SCM) are calculated as the percentage change in regional crop allocation between SSP5-RCP8.5 simulations and SSP5-NoCC scenarios.

Under the high-emissions scenario (SSP5-RCP8.5), the median relative change in global aggregated crop yields in 2100 compared to 2015 was approximately -3.8%, with maize and soybean showing higher sensitivity to climate change. MAgPIE simulations showed a median increase in rainfed cropland area (4.2%) and a decrease in irrigated area (-4.6%) by 2100 under SSP5-RCP8.5 compared to SSP5-NoCC. Technological change showed a slight median increase (0.23%). Not adapting would result in a modest overproduction (median +1%, range [-15%,+14%]) in 2100. In contrast, under SSP1-RCP2.6, adaptation entailed minor changes in cropland areas and technological change. Not adapting resulted in slightly higher overproduction (+2%) compared to SSP5-RCP8.5. The uncertainty in crop yield projections, particularly for high-emissions scenarios, significantly affected adaptation strategies. Regional analysis revealed diverse adaptation pathways depending on the specific GCM-GGCM combination. For instance, the LPJmL-MRI-ESM2-0 combination showed reduced irrigation and slightly increased rainfed cropland, along with moderately lower technological change. Regionally, reforming economies increased livestock and crop production, boosting their self-sufficiency ratio. Conversely, the CYGMA1p74-UKESM1-0-LL combination (most pessimistic yield projections) showed increased technological change and cropland expansion to meet demand, particularly in North America where crop production sharply declined. Global average adaptation costs for SSP5-RCP8.5 were positive (average +4.8 US$/ton of dry matter, range [+19, -1.5] US$/ton of dry matter), translating to absolute production costs ranging between -17 and +209 billion US$ per year. SSP1-RCP2.6 showed near-zero adaptation costs. The uncertainty in adaptation costs was particularly high in 2050 due to the peak in population. Adaptation costs in SSP5-RCP8.5 were influenced by intensification costs and land conversion costs.

The findings suggest that considering CO2 fertilization effects on crop yields can lead to lower cropland expansion and intensification needs compared to previous estimations. However, uncertainty in climate change impacts and adaptation responses remains significant, particularly at regional scales. Adaptation costs are not solely determined by global average yield impacts, but also regional variations, the rate of change, and previous land-use adjustments. The significant range of possible impacts under high-emissions scenarios highlights the necessity for a more flexible and adaptable food system. This might involve technological innovation, market liberalization, and broader economic transformations.

This study demonstrates that incorporating CO2 fertilization effects in crop yield projections can lead to more optimistic estimates of land-use adaptation needs and costs. However, substantial uncertainty remains, particularly at the regional level, underscoring the need for flexible food systems capable of adapting to diverse climate scenarios. Future research should explore additional adaptation strategies, including planned adaptation measures and policies, and consider a wider range of crop types and climate impacts.

This study focuses on autonomous adaptation, neglecting planned adaptation strategies. The impact data primarily considered four staple crops, potentially underrepresenting the full range of climate impacts. Farm-level management decisions are underrepresented, potentially leading to underestimation of adaptation potentials. Geopolitical instability, extreme weather events, and impacts on labor and animal productivity are not explicitly considered. The average values presented should not be interpreted as the most probable future outcomes.