Energy implications of the 21st century agrarian transition

The study investigates how the global agrarian transition—characterized by a shift from smallholder, low-input farming to large-scale, high-input commercial agriculture through large-scale land acquisitions (LSLAs)—alters fossil-fuel-based energy demand and associated greenhouse gas (GHG) emissions. Contextually, the green revolution overcame traditional limits of soil fertility, water, and labor via fossil-fuel-dependent inputs, tripling global crop production but increasing environmental impacts. While intensification and closing yield gaps are often proposed to meet future food demand sustainably, their social and environmental trade-offs are debated, especially with the rapid rise of LSLAs since 2000. Prior research has explored LSLAs’ political, socio-economic, land, water, and carbon impacts, but their direct energy implications remain understudied. This paper aims to quantify energy inputs before and after LSLAs under low- versus high-input scenarios, assess irrigation energy needs, and estimate resultant GHG emissions, thereby informing land governance with an energy-water-food nexus perspective.

The paper situates its inquiry within debates on agricultural intensification and LSLAs. While intensification can raise yields and offer macro-level benefits, evidence of local hunger alleviation and resource pressure reduction is mixed, with few clear win-win cases across SDGs. LSLAs have been linked to land dispossession, livelihood losses, and production for export markets potentially undermining local food security. Studies have covered LSLAs’ political dynamics, property system impacts, rural livelihoods, yield effects, water appropriation, food security outcomes, environmental impacts, and land-use-change emissions. Energy use in food systems is substantial (15–30% of global primary energy; 25–34% of GHG emissions), and prior estimates attribute 8 Gt CO₂e from deforestation and land-use change over 40 Mha of LSLAs (2000–2016). However, the direct fossil energy implications of the agrarian transition via LSLAs have largely been overlooked, with few exceptions, motivating this systematic assessment.

The authors compiled 197 agricultural land deals (>200 ha) from the Land Matrix database across 39 countries totaling 4.07 Mha. Deals were selected if their status was contracted, in start-up, or in production; intended for agricultural use; and had geographic coordinates. They characterized pre-LSLA conditions using spatial datasets on synthetic nitrogen application rates and yield gaps (Mueller et al. 2012) to infer baseline input levels and productivity. Two scenarios were defined: low-input (typical smallholder practices) and high-input (industrial fertilizers, pesticides, mechanization, mechanized irrigation, lower labor). Fossil-fuel-based energy inputs at farm level were compiled from literature (notably Pimentel & Pimentel) and included labor, machinery, fertilizers, chemicals, fuels, and seeds. For oil palm and jatropha, additional mill energy for biodiesel production was reported for context (not included in aggregated LSLA energy footprint in Fig. 4). Irrigation energy requirements were modeled using the WATNEEDS crop water model to estimate irrigation water demand and pumping energy under two systems: sprinkler and surface irrigation. Regional and crop-specific irrigation energy intensities were computed and aggregated. GHG emissions were estimated by converting fossil energy demand to CO₂e assuming oil-powered technologies with a combined combustion and upstream emission factor of 492.6 kg CO₂ per barrel of oil (EPA and Masnadi et al.). N₂O emissions from synthetic nitrogen fertilizer in high-input scenarios used FAO emissions factor (0.0132 kg N₂O per kg synthetic N) and a GWP of 265. Sensitivity analysis considered partial cultivation (75% of transacted area), 80% attainable yield, and an energy mix with 80% fossil fuels and 20% renewables, to assess impacts on total GHG emissions. Practical limitations of life cycle energy accounting were acknowledged.

• Pre-LSLA conditions: 80% of deals occurred on land previously under low-input agriculture; only 1% showed high-input fertilizer use. About 85% of deals targeted lands with high yield gaps (low current productivity), implying additional inputs (fertilizer, water) are generally needed.
• Energy intensity by crop: Among staple crops, high-input rice is the most energy-intensive; soybean and pulses are lower due to biological N fixation. Low-input sorghum is the least energy-intensive. Cotton is highly energy-intensive even under low inputs. Groundnut shows large variability between low- and high-input energy intensities.
• Aggregate energy demand: If all intended crops (4.07 Mha) are cultivated under low-input practices, annual energy input is ~3 million barrels of oil equivalent (boe). Under high-input practices, energy demand increases ~5-fold to ~15 million boe per year (~0.04% of global oil consumption).
• Associated GHG emissions: High-input farming over LSLAs would emit an additional ~6 million tons CO₂ per year relative to low-input conditions (using 492.6 kg CO₂ per boe). N₂O from synthetic fertilizers in high-input farming contributes ~1.3 Mt CO₂e yr⁻¹ (0.3 t CO₂e ha⁻¹ yr⁻¹).
• Irrigation energy: 20% of land deals are in regions where irrigation water demands cannot be sustainably met by local renewable water resources. Sprinkler irrigation requires more energy per hectare than surface irrigation due to higher operating pressure, though surface systems require more water. Africa shows the highest irrigation energy requirements due to drier climates and water-demanding crops (e.g., oil palm, sugarcane). Biofuel plantations are generally the most irrigation energy-intensive. If LSLAs were irrigated with sprinklers to meet full crop water demand, an additional ~4.3 million boe per year (~26 million GJ) would be required compared to rainfed conditions.
• Sensitivity scenario: Assuming 75% cultivated area, 80% attainable yield, and 80% fossil share in energy supply, GHG emissions from high-input farming would be lower by ~3 Mt CO₂ per year relative to the business-as-usual estimate.
• Comparative intensities (Table 1): Low-input fossil energy intensity ~4 GJ ha⁻¹ yr⁻¹ (GHG ~0.3 t CO₂e ha⁻¹ yr⁻¹); High-input ~19.1 GJ ha⁻¹ yr⁻¹ (GHG ~1.6 t CO₂e ha⁻¹ yr⁻¹). Irrigation adds ~3.2 GJ ha⁻¹ yr⁻¹ (~0.3 t CO₂e ha⁻¹ yr⁻¹). Land-use change emissions (from literature) can dominate at ~14.1 t CO₂e ha⁻¹ yr⁻¹ where deforestation occurs.
• Global extrapolation: If 360 Mha of small farms (<2 ha) globally transitioned from low- to high-input farming, additional emissions could reach ~0.6 Gt CO₂e yr⁻¹ (~4% of current global agricultural GHG emissions).

The findings demonstrate that agrarian transitions toward high-input, large-scale commercial agriculture markedly increase fossil energy use and GHG emissions relative to low-input smallholder systems in the lands targeted by LSLAs. This heightened energy intensity—driven by industrial fertilizers, mechanization, and irrigation—poses implications for climate mitigation, local energy security, and resource justice. Energy demands for irrigation are particularly salient in regions already facing water scarcity, potentially intensifying competition for limited energy resources and exacerbating energy poverty among local communities. While high-input practices can boost yields, their benefits are offset by increased embedded energy and emissions, and they may not directly alleviate local food insecurity if crops serve export markets. The study underscores the need to integrate energy accounting into land governance and LSLA assessments, applying a food-water-energy nexus approach to evaluate trade-offs and distributional impacts. Pathways to reduce energy and carbon intensity include improved fertilizer management, using recycled or biologically fixed nutrients, adopting agroecological practices (crop diversification, deficit irrigation, soil moisture conservation), and substituting fossil energy with local renewable systems (e.g., solar pumps, mini-grids, agrivoltaics). Such strategies can mitigate climate impacts and enhance equitable energy access for smallholders, but technical and economic barriers may limit their large-scale adoption.

This study provides the first systematic quantification of fossil energy requirements and associated GHG emissions for agrarian transitions induced by LSLAs, contrasting low- and high-input agricultural systems and explicitly accounting for irrigation energy. High-input transitions over 4.07 Mha raise energy demand roughly fivefold (~15 million boe yr⁻¹) and add ~6 Mt CO₂ yr⁻¹, with irrigation potentially adding a further ~4.3 million boe yr⁻¹. Given these footprints and the prevalence of water- and energy-scarce contexts, incorporating energy-intensity analysis into LSLA governance and rural development policy is essential. Future research should: expand datasets to include more deals and dynamic implementation statuses; refine spatially explicit life cycle energy accounting including milling/processing stages; integrate non-fossil energy scenarios and grid decarbonization; evaluate socio-economic outcomes of renewable energy interventions (mini-grids, solar irrigation, agrivoltaics); and develop decision-support tools embedding food-water-energy nexus and justice considerations to guide crop choice, irrigation strategies, and fertilizer management.

• Scope and representativeness: Analysis covers 197 LSLA deals (>200 ha) with available coordinates; results may not represent all LSLAs globally, nor evolving statuses or partial implementation.
• Scenario assumptions: Business-as-usual assumes full cultivation of transacted areas, adoption of high-input practices, and oil-powered technologies; this provides upper-bound estimates. Sensitivity scenarios partially relax these assumptions but uncertainty remains.
• System boundaries: Farm-level energy includes labor, machinery, fertilizers, chemicals, fuels, and seeds; aggregated energy footprints in Fig. 4 exclude oil palm and jatropha milling energy. Downstream processing, transport, and infrastructure construction are not comprehensively included.
• Irrigation modeling: Irrigation energy estimates assume meeting full crop water requirements via specified systems (sprinkler/surface) and generalized efficiencies; gravity-fed or localized practices could lower energy, while groundwater depth and pump performance variability introduce uncertainty.
• LCA limitations: Known ambiguities in energy ratio calculations and data variability affect life cycle assessments; emission factors (e.g., oil upstream/combustion, N₂O EF and GWP) are generalized and regionally variable.
• Land-use change: The study references but does not model deal-specific land-use-change emissions, which can dominate total GHGs where deforestation occurs.