Measured greenhouse gas budgets challenge emission savings from palm-oil biodiesel

The study investigates whether palm-oil biodiesel delivers greenhouse gas (GHG) emission savings compared with fossil diesel when real, measured ecosystem fluxes are considered and plantation age is accounted for. The context is the rapid expansion of oil palm in Indonesia and global reliance on palm oil for biodiesel, alongside EU policy requiring significant GHG savings for biofuels. Prior assessments assumed biogenic carbon neutrality and lacked ecosystem-scale, measurement-based GHG budgets. The authors aim to quantify CO2, CH4, and N2O fluxes across plantation development stages, incorporate them into a life cycle assessment (LCA), and evaluate how management options and land-use history (first vs. second rotation or degraded land) affect the GHG performance of palm-oil biodiesel.

Previous research highlighted that converting tropical forests to oil palm causes large carbon emissions, often exceeding the savings from substituting fossil fuels. Traditional LCA commonly assumes carbon neutrality of biofuels, disregarding biogenic CO2 turnover, though some studies examined biomass carbon stock changes. Field measurements in oil palm systems showed increases in leaf area index and transpiration with age, implying potential increases in carbon uptake, but comprehensive annual ecosystem CO2 budgets were lacking. Non-CO2 GHG measurements (CH4, N2O) in oil palm are limited, and plantations on organic (peat) soils are known hotspots for N2O and CO2 from decomposition. EU directives stipulate minimum GHG savings thresholds (60–80%) for biofuels relative to fossil fuels, yet results for palm-oil biodiesel have been mixed, particularly when land-use change (LUC) emissions are included. This gap motivates incorporation of measured fluxes and plantation age in LCA.

Study sites and measurements: Two oil palm plantations in Jambi Province, Sumatra, Indonesia, were studied: a 1-year-old smallholder plantation (non-productive; measurements for ~8 months from July 2013) and a 12-year-old state-owned plantation (productive; measurements for 24 months from May 2014). Both are on mineral Acrisols, with small peat-rich areas (5–10%) within the mature site’s eddy covariance footprint. CO2 fluxes: Net ecosystem exchange (NEE) of CO2 was measured using eddy covariance towers (7 m at the young site, 22 m at the mature site), with a sonic anemometer and open-path IRGA (10 Hz). Processing (EddyPro) included coordinate rotation, WPL correction, friction velocity filtering, footprint analysis, and gap-filling based on meteorological co-variation and temporal autocorrelation. The 8-month young-site data were linearly extrapolated to annual totals; two full annual budgets were reported for the mature site. Soil GHG fluxes: Soil CO2 (respiration), N2O, and CH4 were measured using static vented chambers. At the young site, nine chambers in three clusters at varying distances from trunks; four campaigns were conducted. At the mature site, 12 chambers on mineral soils and six on peat-influenced soils were sampled over 18 and 15 months, respectively (45 and 37 campaigns). Fluxes were derived from chamber concentration changes; annual estimates used trapezoidal interpolation. Second-rotation soil respiration: A decay function for soil carbon stocks after forest-to-oil-palm conversion (calibrated with regional forest C stocks and measured plantation C stocks) was used to infer equilibrium soil respiration (reached ~30–35 years post-conversion). Equilibrium soil respiration was estimated at 825 g C m−2 yr−1 for second-rotation analyses. Derived metrics: Net ecosystem productivity (NEP) was computed as NEE + harvest carbon export. For peat areas, NEEpeat = NEEmineral − SRmineral + SRpeat. For second-rotation cycles, NEE2nd = NEEmineral − SRmineral + SR2nd. Net GWP (100-year horizon) combined NEP (CO2) with CH4 and N2O contributions using GWPs of 25 and 298, respectively. Life Cycle Assessment: The functional unit was 1 MJ of biodiesel produced in Jambi. System boundaries included cultivation, milling, biodiesel production, product use (biogenic CO2 emissions from biodiesel, fibres, shells, POME), and LUC (including foregone forest sequestration). Traditional LCA assumed carbon neutrality; enhanced LCA incorporated measured ecosystem fluxes over plantation age and biogenic emissions during use. Allocation by energy was applied to co-products; background data came from ecoinvent v3; modeling used SimaPro 8. Monte Carlo simulations (1000 iterations) propagated uncertainties (ecoinvent pedigree matrix and measurement SE), reporting 25th and 75th percentiles. Scenarios: Business-as-usual (BAU) assumed a 25-year rotation on mineral soils with yield beginning in year 4 and plateau by year 8; fertilization increased to maturity (up to ~196 kg N ha−1 yr−1). Alternative scenarios: A) 30-year rotation; B) 40-year rotation; C) 30-year rotation with earlier-yielding variety (yield starts year 3; max by year 6). Second-rotation scenarios (or degraded-land establishment) excluded LUC emissions and used adjusted NEE reflecting lower soil respiration. LUC emissions: Annualized LUC emissions followed EU methodology using measured regional C stocks (forest 283.5 ± 12.2 vs. oil palm 109.9 ± 5.5 Mg C ha−1) and included foregone sequestration from nearby rainforest eddy-covariance data (−124 ± 13 g C m−2 yr−1).

Ecosystem fluxes on mineral soils: The 1-year-old plantation was a strong carbon source (NEE: 1012 ± 51 g C m−2 yr−1). The 12-year-old plantation was a substantial ecosystem CO2 sink (NEE: −799 ± 40 and −709 ± 35 g C m−2 yr−1 in two consecutive years), but when accounting for harvest export (NEP), it remained a net carbon source (52–240 g C m−2 yr−1). Diurnal patterns showed similar nighttime respiration between ages but much larger daytime CO2 uptake at the mature site. Soil GHGs on mineral soils: Soil respiration was similar between ages (young: 133.9 ± 56.9; mature: 91.7 ± 39.6 mg C m−2 h−1). Both plantations were small CH4 sinks (−17.0 ± 2.2 and −14.7 ± 4.6 mg C m−2 h−1). N2O emissions were negligible at the unfertilized young site; at the fertilized mature site, N2O showed peaks up to 296 ± 279 µg N2O-N m−2 h−1 with annual emissions of 0.3 ± 0.17 g N2O-N m−2 yr−1. Overall, plantation GWPnet was dominated by CO2; CH4 and N2O contributions were minor under mineral soils. The greenhouse gas intensity (GHGI) was 0.1–0.4 g CO2 per g yield. Peat-influenced soils: In the mature plantation, peat zones had 2.6× higher soil respiration than mineral soils; N2O fluxes were ~3× higher (0.95 ± 0.87 vs. 0.33 ± 0.17 g N2O-N m−2 yr−1), and CH4 was near neutral to a source (0.00 ± 0.18 g C m−2 yr−1) compared to mineral soil sinks (−0.13 ± 0.02 g C m−2 yr−1). Estimated ecosystem NEE for mature plantations on peat was a source (330 ± 288 g C m−2 yr−1). Ecosystem GWPnet was about seven times higher on peat than on mineral soils. LCA results (first rotation, mineral soil): Traditional LCA (carbon neutral assumption) yielded 186 g CO2-eq MJ−1 (173–199), 98% higher than fossil diesel reference (94 g CO2-eq MJ−1). Enhanced LCA (BAU) that included measured fluxes resulted in 216 (201–230) g CO2-eq MJ−1, 14% higher than traditional LCA, confirming the carbon neutrality assumption underestimates emissions. LUC-related emissions contributed 156 g CO2-eq MJ−1 (sum of stock change and foregone sequestration). Life-cycle process emissions (cultivation, milling, biodiesel production, POME handling, transport, inputs) totaled ~136 g CO2-eq MJ−1, partially offset by an ecosystem sink of 77 g CO2-eq MJ−1. Second rotation/degraded land: Excluding LUC and with lower soil emissions, enhanced LCA emissions dropped markedly. Net emissions (g CO2-eq MJ−1; mean [25th, 75th]) for 2nd rotation: BAU 33 (23, 44), Scenario A 28 (18, 38), Scenario B 23 (12, 36), Scenario C 21 (9, 34). These correspond to 65–77% savings versus fossil diesel (94 g CO2-eq MJ−1), meeting current EU thresholds and approaching the 80% threshold in Scenario C. Management effects: Extending rotation length (especially to 40 years) and adopting earlier-yielding varieties increase GHG savings by diluting early high emissions and advancing carbon uptake and yield. POME management substantially influences results: closed anaerobic treatment with methane flaring contributed ~1.8 g CO2-eq MJ−1; open lagoons would raise POME-related emissions to ~16.5 g CO2-eq MJ−1.

Measured ecosystem-scale GHG fluxes reveal strong age-dependent dynamics that invalidate a blanket carbon neutrality assumption in LCA for palm-oil biodiesel. First-rotation plantations after forest conversion accumulate substantial emissions due to large LUC carbon losses and elevated soil respiration during early years, which are not compensated by photosynthetic uptake, even at maturity after accounting for harvest exports. Consequently, biodiesel from such plantations does not achieve GHG savings relative to fossil diesel. In contrast, second-rotation plantations or those established on degraded land avoid major LUC emissions and exhibit reduced soil CO2 losses, enabling significant GHG savings that satisfy current policy thresholds. Management interventions—extending rotation cycles and deploying earlier-yielding varieties—further improve the GHG balance by increasing the productive, high-uptake phase and spreading fixed LUC or early emissions over longer, higher-yield periods. The study underscores the importance of incorporating measurement-based fluxes and temporal dynamics into LCA frameworks and databases. It also highlights critical roles of site conditions (mineral vs. peat soils) and processing choices (POME treatment) in determining net climate impacts, with peat cultivation likely precluding any GHG savings.

This work provides the first LCA of palm-oil biodiesel that integrates measured ecosystem GHG fluxes across plantation development stages. It shows that: (1) first-rotation-cycle plantations established after forest conversion on mineral soils do not deliver GHG emission savings relative to fossil diesel due to large LUC and early soil emissions; (2) second-rotation-cycle plantations or plantations established on degraded lands can meet or approach EU GHG-saving thresholds, especially with extended rotation cycles and earlier-yielding varieties; and (3) cultivation on peat soils likely eliminates any potential for GHG savings. The study demonstrates that accounting for biogenic CO2 dynamics and measured fluxes is essential for accurate LCA of biofuels. Future research should expand measurement-based datasets to second-rotation plantations and degraded lands, quantify full LCAs for peat systems, refine spatially explicit assessments, and evaluate additional mitigation options (e.g., optimized POME-to-energy recovery, soil carbon restoration practices).

Key limitations include: limited temporal coverage at the young site (8 months extrapolated to annual); lack of direct annual N2O totals for the young plantation (values adopted from nearby smallholder data); absence of direct ecosystem-scale NEE measurements for peat plantations (peat NEE inferred from differential soil respiration); no full LCA for peat soils due to incomplete data; reliance on assumed yield trajectories and management inputs from plantation reports rather than direct measurements over full rotations; scenario analyses depend on modeled NEE and soil respiration decay functions for later rotation stages; and regional specificity of LUC and background datasets. Uncertainty was addressed via Monte Carlo simulations but remains inherent, especially for LUC and CH4-related processes (e.g., POME management).