Journal of Cleaner Production 277 (2020) 124045

Contents lists available at ScienceDirect

Journal of Cleaner Production

journal homepage: www.elsevier.com/locate/jclepro

Certified palm oil reduces greenhouse gas emissions compared to

non-certified
Jannick Schmidt*, Michele De Rosa
Aalborg University, Rendsburggade, 14 9000, Aalborg, Denmark

a r t i c l e i n f o a b s t r a c t

Article history: Consumers are increasingly demanding products containing palm oil produced without harm to the
Received 25 November 2019 environment. The industry response to this demand has been the creation of the Roundtable on Sus-
Received in revised form tainable Palm Oil (RSPO) and the development of a certification system to ensure sustainable palm oil
28 July 2020
production. However, currently there is no scientific evidence of the benefit gained through the RSPO
Accepted 29 August 2020
certification schema. This paper quantifies the environmental impacts of RSPO certified and non-certified
Available online 3 September 2020
through a detailed Life Cycle Assessment (LCA) of 1 kg of RBD palm oil to factory gate, produced in
Handling editor. Yutao Wang Indonesia and Malaysia in 2016, to identify potential benefits and trade-offs of RSPO certification. The ISO
14040/14044 compliant LCA is carried out following both a consequential and an attributional LCA
Keywords: approach. The inventory model presents a high level of detail. Primary inventory data describing the
Palm oil certified production system are obtained from RSPO assessment reports, covering 73% (634 estates) of
Carbon footprint the certified estate, including 111 smallholders, and 58% (165 oil mills) of the certified mills. Data for the
Biodiversity total industrial production are drawn from national statistics and scientific literature. The non-certified
RSPO certified
flows are derived by subtracting the certified flows from the total industry flows. The consequential
GHG emissions
results show that RSPO certified oil reduces GHG emission by 35% compared to non-certified i.e. 3.41
(2.61e4.48) kg CO2 eq./kg for certified vs 5.34 (3.34e8.16) kg CO2 eq./kg for non-certified. Based on a
thorough data quality assessment and uncertainty analysis, this result is deemed sufficiently robust and
thus conclusive. Certified production achieves the largest GHG emissions reduction because of higher
yields, i.e. less land use per unit of product, less oil palm cultivated on peat soil and higher share of palm
oil mill effluents treated with biogas capture technologies. We also found that nature occupation is
reduced by 20% in certified production while respiratory inorganic is slightly higher (3%) in certified
production, due to the larger use of fertilisers. For other impact categories, results are associated with a
larger uncertainty and therefore shall be considered as indicative. Similar results are found in attribu-
tional modelling.
© 2020 Elsevier Ltd. All rights reserved.

1. Introduction Palm is not a land-use intensive product compared to the

alternative vegetable oils (Schmidt et al., 2015a). Nevertheless, oil
Palm oil is an agricultural commodity produced at large scale, palm plantations are spurring debates, especially in Europe, due to
consumed and traded globally. In the last decades, the oil palm the environmental effects caused by land-use changes (LUC) from
harvested area showed an outstanding growth, particularly in natural forest and peat swamps to plantations in tropical areas and
Indonesia and Malaysia, which in 2016 together accounted for more due to the related social implications (Wicke et al., 2011). The main
than 82% of the global production of fresh fruit bunches (FFB, environmental concerns are the emissions of greenhouse gasses
FAOSTAT, 2018). From 1990 to 2016, the oil palm area harvested (GHG) and loss of biodiversity caused by forest fires or logging,
increased from 0.7 to 9.3 million ha in Indonesia and from 1.7 to 5.0 changes in carbon stocks of soil and biomass triggered by drainage
million ha in Malaysia (FAOSTAT, 2018). of peatland. Tropical peatland stores a large share of the global
carbon stock (IPCC, 2014a) and the drainage of peatland for agri-
cultural purposes exposes the organic matter to oxygen, acceler-
ating its decomposition and the emission of CO2. Social concerns
* Corresponding author.
are human health issues, such as respiratory diseases caused by
E-mail address: [email protected] (J. Schmidt).

https://doi.org/10.1016/j.jclepro.2020.124045
0959-6526/© 2020 Elsevier Ltd. All rights reserved.
J. Schmidt and M. De Rosa Journal of Cleaner Production 277 (2020) 124045

forest fires, land tenure and human right conflicts caused by illegal certified versus non-certified production of palm oil. The certified
land occupation or forced displacement of local population (Wicke and non-certified palm oil are both assumed as produced in
et al., 2011). Indonesia and Malaysia, the two largest palm oil producing coun-
In response to these concerns, in 2004 a group of various or- tries in the world. The objective is to identify the potential envi-
ganizations including palm oil producers, traders, consumer goods ronmental benefits of the certification and the possible trade-offs
manufacturers, investors and NGOs, established the Roundtable on between the two production systems. This requires the quantifi-
Sustainable Palm Oil (RSPO). RSPO promotes the production of cation of the GHG emissions of the two production systems and
sustainable palm oil by setting credible global standards included other relevant impacts in palm oil production, such as nature
in the RSPO principles and criteria. In 2008, RSPO developed a occupation (biodiversity) and respiratory impacts. Other less rele-
voluntary certification schema and released the first certification. vant impacts for palm oil production are also assessed, e.g. eutro-
In 2017, certified production accounted for 12.1 million tons (RSPO, phication, toxicity etc.The comparison between these two palm oil
2018), 19% of the 2014 global palm oil and palm kernel oil pro- production systems allows the identification of the most promising
duction of 63.9 million tons (FAOSTAT, 2018). and the less obvious improvement options. The results also provide
The awareness of sustainable palm oil production rose further in useful insights to several stakeholders: palm oil consumers, to
2015 in conjunction with the European legislation on the provision highlight the benefits of choosing certified palm oil; palm oil sup-
of food information to consumers. From December 2014, the EU pliers, to define further requirements to producers, based on
(law No 1169/2011) requires that nutrition labels of products environmental performances; palm oil producers, to prioritise
specify information on the origin of refined vegetable oils and fats options to improve the palm oil product system. Therefore, this
(e.g. “palm oil” instead of “vegetable oil”), making palm oil visible to work addresses a pressing and currently unanswered question:
consumers (Gassler and Spiller, 2018). Because of the increased what are the environmental benefits and trade-offs of RSPO certi-
public attention, the need for a verifiable and efficient certification fied palm oil production compared to non-certified production?
schema has become even more urgent.
This paper compares the performances of RSPO certified and 2. Materials and methods
non-certified palm oil to quantify the difference between the two
production systems. Life Cycle Assessment (LCA) is a commonly 2.1. Functional unit and system boundaries
used framework particularly suitable for comparing the environ-
mental performances of alternative products or production sys- The functional unit (FU) of the LCA is 1 kg of refined, bleached
tems. Since 2007, a few LCAs of palm oil have been published in and deodorized (RBD) palm oil at refinery gate in 2016, produced in
literature (e.g. Yusoff and Hansen, 2007; Schmidt, 2007; Vijaya Indonesia and Malaysia. The formulation of the FU as refined oil
et al., 2008; Tan et al., 2010; Saswattecha et al., 2015; Morgans instead of crude palm oil (CPO) is because CPO contain impurities
et al., 2018) including simplified LCAs accounting only for GHG and free fatty acids typically removed at refinery stage, making CPO
emissions (Choo et al., 2011; Puah et al., 2013). The first ISO not substitutable in a one-to-one ratio.
compliant (i.e. review by an independent panel) LCA of palm oil was Palm kernels are a by-product of the palm oil mill sent to a
a comprehensive LCA of Palm Oil at United Plantations (Schmidt, palm-kernel crusher plant, where they are processed into crude
2008). Comparative LCAs allow the comparison of the potential palm kernel oil (CPKO) and palm kernel cake (PKC). After the CPKO
environmental impacts of palm oil with substitutable products in is refined, the output falls under the definition of the functional
the same economic sector. Examples of comparative LCAs of unit. We include the production stages of oil palm cultivation, palm
vegetable oils exist, e.g. Arvidsson et al., (2013), Schmidt (2015a). oil mill and palm oil refinery. The downstream life cycle stages, i.e.
LCAs have been performed to compare the performances of palm the use stage and the product’s end-of-life, are excluded from the
oil with rapeseed oil (Schmidt, 2010), or among palm oil, soybean assessment. Product packaging is also outside the model’s system
oil, rapeseed oil, sunflower oil and peanut oil (Schmidt, 2015a). boundaries, because most oils are traded as bulk oils in trucks or
Comparative LCA has also been performed in the biodiesel sector, ships.
e.g. comparing palm oil with jatropha oil (Lam et al., 2009), and
with rapeseed biodiesel (Yee et al., 2009). These studies focused on 2.2. Modelling approach
comparing palm oil with competing (i.e. substitutable) products,
but none of them assess the environmental effects of RSPO The LCA is performed according to attributional and conse-
certification. quential modelling approaches, described in Schmidt and Dalgaard
Previous works assessed the effectiveness of specific aspects of (2012), Weidema (2003) and Weidema et al. (2009). The conse-
the certification such as: the impact of palm oil certifications on quential approach follows a causal model, seeking to describe the
small holders in Indonesia (Brandi et al., 2013; Hutabarat et al., consequences of a change in demand of the functional unit and it
2018); the effect of RSPO certification in reducing forest fire in treats by-products by modelling their product’s substitution effect.
Sumatra and Kalimantan (Cattau et al., 2016); the impact of certi- In consequential modelling, the inputs to a system are assumed as
fication on deforestation and fire Carlson et al. (2017). Bessou et al. provided by the marginal suppliers of that input in the market. The
(2014) performed a simplified LCA of RSPO certified palm oil pro- attributional modelling instead follows a normative approach: it
duced by nine RSPO member companies, focusing exclusively on assumes the inputs to a system are the average of the products
global warming potential (i.e. a carbon footprinting). More recently, supplied in the market; by-products are allocated based on one of
Meijide et al. (2020) carried out a life-cycle carbon footprinting of their properties, e.g. the mass in the case of this study.
palm oil biodiesel, distinguishing between mature and immature The background database used for the consequential model is
Indonesian plantations. Yet, a comprehensive LCA comparing the the latest version of the global hybrid environmentally extended
environmental effects of RSPO certification against non-certified multi-regional input-output (EEMRIO) database EXIOBASE (version
palm oil production is currently missing in the scientific litera- 3.3.13), described in Merciai and Schmidt (2017a) and Stadler et al.
ture. Therefore, the potential environmental benefits of the RSPO’s (2018). The databases operates with no cut-off criteria, and thus it is
certification have yet to be demonstrated. significantly more complete in terms of product flows and in-
This work intends to fill this gap by performing a detailed and dustries. It also includes emissions from the production of more
comprehensive LCA, comparing the performance of average RSPO capital goods and services than any traditional processed-based
2
J. Schmidt and M. De Rosa Journal of Cleaner Production 277 (2020) 124045

LCA databases. EXIOBASE distinguishes among 164 activities/ ⁃ Ecotoxicity, aquatic

products (equivalent to LCA processes in a conventional LCA data- ⁃ Ecotoxicity, terrestrial
base), 43 countries and 5 Rest-Of-the-World (ROW) regions, rep- ⁃ Acidification
resenting the remaining countries for the reference year 2011. The ⁃ Eutrophication, aquatic
database also distinguishes between 34 emissions, 22 resources, ⁃ Eutrophication, terrestrial
including water, and 5 land use types. In this study, both direct land ⁃ Photochemical ozone, vegetat.
use change (dLUC) and indirect land use change (iLUC) are ⁃ Non-renewable energy
modelled. The land use change (LUC) model follows the framework ⁃ Mineral extraction
described Schmidt et al. (2015) and it is integrated in EXIOBASE.
This is further described by section 2.4.1.
The attributional approach follows, as close as possible, the 2.3. Data collection approach
modelling assumptions of ‘Palm GHG’, a GHG accounting tool
commissioned by RSPO, described in Chase et al. (2012). PalmGHG The inventory data for RSPO certified production and total in-
is chosen as a guiding principle for the attributional model, because dustrial production in Indonesia and Malaysia in 2016 are drawn
it is a tool and method provided by RSPO and because it is widely from different data sources. These are described in section 2.3.1 and
used by certified palm oil producers. However, the emission cal- section 2.3.2 below, respectively. Data for the non-certified pro-
culations and the background database in this study are different duction are obtained by subtracting the RSPO certified flows from
compared to PalmGHG. PalmGHG operates with a FU of 1 kg crude the total industry flows. Summary inventory tables for the total
palm oil (CPO). To compare results calculated with the PalmGHG industry, RSPO certified and non-certified production are available
calculator and this LCA, the contribution from the refining stage in section 3 of the SI. The summary inventory tables cover all the
shall be subtracted from the results of the current study. As in production stages of palm oil production: oil palm cultivation, palm
PalmGHG, the attributional model in the current LCA uses mass as a oil mil processing and kernel crusher plant, palm oil refining. The
property of the products to allocate the impacts (mass allocation) tables present inventory data for the consequential and attribu-
among by-products and it includes only direct land use changes. tional model separately (see SI, section 3).
This is consistent with the PAS 2050 (2011) specification. The direct The consequential model draws fuel combustion emissions for
LUC in the attributional model is calculated following a normative the whole product system from consistent emission factors found
approach also used by PAS2050. The GHG emissions from land in the EXIOBASE database. The marginal electricity mix is estimated
conversion are amortized (i.e. allocated) over an arbitrarily defined based on the expansion of the national electricity supply in the
period, in this case 25 years. Thus, only the LUC emissions occurred period 2006e2011. The attributional model draws fuel combustion
during the last 25 years are included. Although common in LCAs, emission factors and average electricity mixes from the ecoinvent
this approach does not reflect a cause-effect relationship. Allocating database (ecoinvent 2017).
historical land use change impacts to current oil palm cultivation
implies a causality that goes backwards in time (e.g. current de- 2.3.1. RSPO certified production
mand for oil palm caused deforestation 25 years ago). The changes The inventory data collection for certified production is based
in carbon stock associated with the land cover changes during the on a bottom-up approach, i.e. estate-by-estate and oil mill-by-oil
last 25 years are identified based on historical land cover data. The mill from the annual surveillance assessment reports, provided
calculated carbon stock changes are then annualized (i.e. divided by by RSPO member estates and mills for 2016 (RSPO, 2017). In order
25), as further explained in section 6 of the Supplemental Infor- to obtain data coverage for at least 50% of all certified FFB pro-
mation (SI) provided with this paper. duction and 50% of all certified Crude Palm Oil (CPO) production,
Similarly, for consistency with the PAS 2050 (2011) guidelines, data have been collected from 165 RSPO annual assessment reports.
the attributional model does not include capital goods in the The data describe 634 estates (of which 111 are smallholders),
foreground system. The background database applied to the attri- covering around 73% of the total certified oil palm planted area in
butional model is ecoinvent v3.4. (ecoinvent 2017) with system Indonesia (381 estates) and Malaysia (253 estates). Palm oil mills
model ‘Allocation at the Point Of Substitution’ (APOS). In accor- data have been collected for 165 mills in Indonesia (101 mills) and
dance to the ‘Palm GHG’ modelling approach, iLUC is not included Malaysia (64 mills), producing 58% of the certified palm oil in
in the attributional model. Indonesia and Malaysia. The thorough inventory data collected
The Life Cycle Impact Assessment (LCIA) method applied is describe the RSPO certified and the non-certified palm oil pro-
Stepwise 2006 version 1.7. The method is described and docu- duction systems with an unprecedented level of detail. For statistics
mented in the Annex II of Weidema et al. (2008). The weighting describing the data collected, see the SI, section 2.
module is documented in Weidema (2009). The most recent up- The RSPO’s annual assessment reports provide data on several
dates for the impact ‘nature occupation’ are described in Schmidt product flows and land use needed for the LCA. The assessment
and de Saxce
(2016). The characterization module of Stepwise is reports show data per oil mill, including the estates in its supply
based on a combination of the Impact 2002þ method (Jolliet et al., base. When the assessment report for an oil mill includes aggre-
2003) and the EDIP 2003 method (Hauschild and Potting, 2005). gated data for several estates (e.g. data on land use for nursery and
The indicators in the Stepwise method are explained in the SI, land set-aside for nature conservation aggregated for the mill’s
section 8. The full environmental impact of palm oil production is supply-base), data have been allocated to the estates proportionally
captured by considering a comprehensive set of environmental to the share of the total planted area occupied by each estate.
impact categories:
2.3.2. Total industrial production
⁃ Global warming While the data collection for the RSPO certified flows followed a
⁃ Nature occupation bottom-up approach, the inventory data collection for the fore-
⁃ Respiratory inorganics ground system of the total industry flows follows a top-down
⁃ Respiratory organics approach, when statistical data are available for this purpose.
⁃ Human toxicity, carcinogens Alternatively, representative coefficients drawn from literature are
⁃ Human toxicity, non-carc. used, e.g. diesel use per hectare oil palm estate and water use per
3
J. Schmidt and M. De Rosa Journal of Cleaner Production 277 (2020) 124045

tonne processed FFB in palm oil mills. The next sections describe emissions. This is not a problem when comparing palm oil with
how data have been obtained for three production stages: oil palm other bio-based alternative oils. However, if bio-based feed stocks
cultivation, palm oil milling and palm oil refining. are compared with fossil-based alternatives on a cradle-to-gate
assessment, the biogenic emissions shall be subtracted from the
2.4. Oil palm cultivation stage results, in order to avoid penalising the bio-based oil incorrectly.

Table 1 shows selected key performance indicators describing 2.4.1. Land use change and nature conservation modelling
the oil palm cultivation stage. The yields have large impact on the CO2 emissions from land use changes are responsible for 11% of
results. In order to make the results for 2016 significant and com- global GHG emissions (IPCC, 2014b). In the consequential model,
parable with other years, the FFB yields are adjusted to take into iLUC is modelled as described in Schmidt et al. (2015). This
account the particularly disadvantageous conditions that occurred approach is applicable to all crops (also forestland, rangeland, build
in 2016, due to a severe drought attributed to El Nin ~ oeSouthern areas etc.) in all regions in the world and avoids the arbitrary
Oscillation (ENSO) in the central and eastern tropical Pacific. In allocation/amortization of transformation impacts. The iLUC model
2016, weather changes in Southeast Asia caused by El Nin ~ o lowered reflects cause-effect relationships consistent with the LCA model-
the Malaysian yield of approximately 14%. (MPOB, 2018). The FFB ling in the consequential approach. The iLUC model has been used
yields reported in this study refer to the area occupied by mature extensively in LCA, carbon footprints and it ranks as the best per-
(producing) and immature (developing) oil palms to reflect the forming approach in comparison with six major LUC models (De
actual land occupation. Typically, yields are reported referring only Rosa et al., 2016) with respect to completeness, impact assess-
to mature area. In this study, they are converted to total planted ment relevance, scientific robustness, and transparency. We use
area assuming an average rotation time of 25 years: 2.5 years to version 4.3 of the iLUC model, which is integrated in the multi-
develop the crop and 22.5 years of harvesting (Woittiez et al., 2017). regional hybrid input-output model EXIOBASE v3.3.13 (Merciai
Further details on the data applied for the calculation of yields are and Schmidt, 2017b; Schmidt and De Rosa, 2018).
available in the SI, section 4. Certified oil palm plantations may reserve part of the land bank
The model includes emissions occurring in the estates, for nature conservation. We accounted for both the dLUC and the
regarding to fertiliser and crop residue inputs. This includes emis- iLUC impact of nature conservation according to Schmidt (2015b,
sions relating to the N-turnover, P and heavy metals from con- 2016; 2017).
taminants in fertilisers calculated through a detailed N and P The nature conservation dLUC (on-site) effects from avoided
balance, respectively, using the IPCC (2014a) approach. The bal- transformation of conserved land is modelled as one-year delay of
ances account for: 1) net inputs, 2) release from decomposition of the effects from transformation from non-productive land (i.e. the
crop residues, 3) uptake of nutrients from the soil in living biomass preserved land) to productive land (i.e. the land use cover, which is
and 4) harvested biomass. Emissions from peat oxidation (CO2 and avoided by the conservation). The nature conservation effect is
N2O), from anaerobic conditions in draining ditches (CH4) and from calculated on an annual basis (ha*year): the effect of preserving a
the use of fire for replanting (NOx and particulates) are also conservation area for one year is a one-year delay of the change in
modelled. carbon stock from nature conservation area to oil palm plantation.
The model accounts for carbon dioxide emissions from peat This is in line with the approach of the iLUC model (Schmidt et al.,
oxidation as fossil emissions and includes methane emissions from 2015).
draining ditches. We estimate that 18% of the industry-average The nature conservation iLUC (remote) effects are modelled as
planted area in Indonesia and Malaysia is on peatland. For RSO avoided transformation of conserved land in the iLUC model. The
certified planted area, we estimate that 11% of the planted area is on area of the nature conservation reserve is adjusted with a pro-
peatland. These percentages are based on data available in the ductivity factor to calculate the ha*year equivalents of land, which
scientific literature and on GIS elaborations performed for this will be occupied elsewhere to compensate for the land occupied by
study. The data and the calculations are described in the SI, section the nature conservation area. The effect of this further land occu-
5. pation is accounted by the iLUC model described by Schmidt et al.
Emissions of substances containing carbon, e.g. biogenic CO2, (2015).
are not based on a field balance. Instead, we conservatively assume The attributional model follows PalmGHG’s LUC accounting
that the CO2 uptake in biomass during growth is also released as approach, which means that iLUC is excluded. Note that the
CO2 when the biomass decomposes. The CO2 contained in the final PalmGHG approach only refers to LUC impact in terms of global
product, the so-called biogenic CO2, is included in the emissions warming (GHG emissions) and it does not provide guidance on how
account, because the functional unit is expressed at refinery gate, to model nature occupation impacts, i.e. impacts on biodiversity
where the carbon content of the oil is not yet converted to from land use changes. In the LCIA method Stepwise 2006 version

Table 1
Key performance indicators for oil palm cultivation.

Flows Unit Total industry (ID & MY) RSPO-certified Non-certified

Total planted area million ha 14.4 2.44 12.0

Share of oil palm on peat % 18% 11% 19%
Drainage depth (DD) of peat cm 73 57 75
FFB yield, mature t/ha 18.9 21.1 18.5
Fuel use MJ/ha 2940 2940 2940
Applied mineral N kg N/ha 82 170 64
Applied organic N kg N/ha 21 24 21
Applied mineral P2O5 kg P2O5/ha 41 103 28
Applied organic P2O5 kg P2O5/ha 28 31 27
Applied mineral K2O kg K2O 156 245 138
Applied organic K2O kg K2O 138 153 135

4
J. Schmidt and M. De Rosa Journal of Cleaner Production 277 (2020) 124045

1.7 (Weidema, 2009), used for this LCA, the methodology to account plants, because the certification schema does not demand specific
for the nature occupation impact is aligned with the concept of requirements concerning kernel crushing. The inventory data are
accelerated denaturalisation, i.e. avoiding amortization of land obtained from Schmidt (2015a). The consequential approach
transformation (Schmidt, 2015). The PalmGHG land use change models the kernels as material for treatment, jointly produced with
approach instead aims to account for the land transformation in the reference flow of the oil milling activity, CPO. CPKO and PKM are
absolute terms. Therefore, the PalmGHG approach to LUC is not both modelled as material for treatment. The attributional model
suitable for capturing the nature occupation impact with the Step- applied, instead, mass allocation between these two outputs. The
wise LCIA method. To overcome this issue in the attributional model, inventory data applied for the consequential and attributional
the nature occupation impact from the occupation of 1 ha*year by models are documented in the SI, section 3.
the oil palm estate is calculated by applying the same proportion Palm oil refineries remove palm fatty acid distillate (PFAD), palm
between iLUC GHG and iLUC nature occupation in the consequential kernel fatty acid distillate (PKFAD) and other impurities from CPO
model to the LUC GHG emissions to the attributional model. and CPKO, respectively. There is no distinction in inventory data
between certified and non-certified refining because the certifica-
2.5. Palm oil milling stage tion schema does not demand specific requirements concerning
refining. The inventory data are obtained from Schmidt (2015a). The
Table 2 shows the key performance indicators selected for palm refined CPO produces the reference product RBD palm oil, the
oil mill (POM) processing. Fig. 1 shows the main palm oil mill and reference flow of the LCA. The refined palm kernel oil (PKO) is traded
the involved activities in the palm oil mill life cycle stage included on the market for vegetable oil. In the consequential model, RPD PKO
in this study. substitute the marginal sources of edible oils, i.e. RBD palm oil.
Palm oil mills produces empty fruit bunches (EFB) and palm oil PKM from the palm kernel crusher plant, the PFAD and PKFAD
mill effluents (POME). The POME treatment activity is a major are by-product of the palm kernel crushing and refining processes.
contributor to GHG emissions, and the POME treatment method is a These by-products are used as animal feed. In consequential
crucial difference between RSPO certified and non-certified palm modelling, the by-products are modelled applying substitution.
oil. We implemented a state-of-the-art model to determine the Details about the modelling of the by-product and their substitu-
methane emissions from POME treatment with or without biogas tion effect on the global market are described in the SI, section 3.
capture facilities at the POM. The model calculates emissions based The attributional model applies mass allocation between RBD palm
on the fate of the captured methane, i.e. whether it is flared (dis- oil and PFAD/PKFAD.
tinguishing between open or enclosed flaring), used in POM boilers
or in biogas engines for electricity generation. The calculations of 3. Results and discussion
POME treatment emissions for these technologies are described in
the SI, section 7. Table 3 shows the characterised results of the comparative LCA
The amount of POME from a conventional palm oil mill, without of RSPO certified and non-certified palm oil according to the
a decanter and an Empty Fruit Bunches (EFB) press, has been found consequential and attributional model, respectively, for the FU of
to be 675 kg/t FFB (Ma et al., 2007); yet, estimations from UPRD 1 kg RBD palm oil produced in 2016. In Table 3, the lower score
(2007) shows that this amount might be low, and that 700 kg/t between certified and non-certified is marked in bold, for both
FFB may better reflect actual measurements. A decanter reduces consequential and attributional results. The Life Cycle Impact
the effluents by 35 kg/t FFB, while an EFB press increases them by Assessment (LCIA) method used is Stepwise 2006, version 1.7 (see
118 kg/t FFB. The EFB press also produces 120 kg EFB liquor. Based section 2.2). The impact categories are also included in Table 3.
on these data, the POME produced for each palm oil mill is calcu- Results from Ionizing radiation and Ozone layer depletion impact
lated depending on whether the mill has an EFB press and decanter. categories were only available for attributional model. For the
When no information is available, the default assumption is that consequential model, there are no contributions to these impacts,
mills do not have EFB an press and decanter. neither from the foreground system nor from the background
The EFB are either burned in oil mill boilers or used as mulch. database (EXIOBASE, does not include any elementary flows
POME is often applied to land. In the consequential model, the EFB contributing to these impact categories).
and POME used as fertiliser reduces the fertiliser applied to the field In the consequential results, an abnormal contribution from the
by a corresponding amount. The substituted fertiliser is assumed to use of Uranium to the non-renewable energy impact is observed.
be urea. The substitution efficiencies of nutrients in EFB and POME The cause may be poor data quality in the background database on
applied to land are 50% and 100%, respectively. Knowing the the use of uranium in Malaysian electricity, represented by ‘Rest of
nutrient content, the moisture content and the substitution effi- the World, Asia’ in Exiobase. Therefore, the consequential results
ciencies, the amount of substituted fertiliser is determined. for this impact category are considered non-reliable and are
excluded. The consequential results show that RSPO certified per-
2.6. Palm kernel crusher and palm oil refining stage forms better than non-certified in 8 out of 13 impact categories. The
attributional results show that RSPO certified performs better than
Kernel crusher plants process palm kernels into crude palm non-certified in 4 out of 16 impact categories. However, not all
kernel oil (CPKO) and kernel meal (PKM). There is no distinction in impacts have the same severity. Moreover, the differences between
inventory data between certified and non-certified kernel crusher results may not be sufficient for drawing conclusions.

Table 2
Key performance indicators for palm oil mills.

Flows Unit Total industry (ID & MY) RSPO certified Non-certified

OER % 20.2% 21.9% 19.8%

KER % 5.4% 5.6% 5.4%
Share of POME treated with biogas capture % 5.0% 16% 2.4%
Share of land bank in supply base set-aside as HCV % 0.6% 3.1% 0%

5
J. Schmidt and M. De Rosa Journal of Cleaner Production 277 (2020) 124045

Fig. 1. Overview of the processes and flows in the palm oil mill stage.

Table 3
Characterised results: RSPO certified and non-certified palm oil in Indonesia and Malaysia. The LCIA method applied is Stepwise 2006, version 1.7 (Weidema, 2009). The lower
score between certified and non-certified is marked in bold, for both the consequential and attributional results.

Impact category Unit Consequential results Attributional results

RSPO certified Non-certified RSPO certified Non-certified

Global warming kg CO2-eq 3.41 5.34 3.42 5.32

Nature occupation PDF*m2a 1.63 2.04 1.64 2.03
Respiratory inorganics g PM2.5-eq 2.58 2.50 0.98 0.67
Respiratory organics milli pers*ppm*h 2.46 2.73 1.58 1.78
Human toxicity, carcinogens g C2H3Cl-eq 13.3 13.5 4.92 2.73
Human toxicity, non-carc. g C2H3Cl-eq 35.5 21.8 30.0 17.2
Ecotoxicity, aquatic kg TEG-eq w 106 46.2 93.5 42.2
Ecotoxicity, terrestrial kg TEG-eq s 3.16 3.18 1.21 0.96
Acidification m2 UES 0.352 0.341 0.180 0.135
Eutrophication, aquatic kg NO3-eq 0.159 0.160 0.106 0.099
Eutrophication, terrestrial m2 UES 1.42 1.35 0.73 0.51
Photochemical ozone, vegetat. m2*ppm*hours 22.0 24.2 14.6 15.8
Non-renewable energy MJ primary n.a. n.a. 3.79 2.85
Mineral extraction KJ extra 1.47 1.56 12.2 9.80
Ionizing radiation Bq C-14-eq n.a. n.a. 1.65 1.33
Ozone layer depletion kg CFC-11-eq n.a. n.a. 0.044 0.038

6
J. Schmidt and M. De Rosa Journal of Cleaner Production 277 (2020) 124045

Weighting the results is an optional step in LCA according to ISO during cultivations are due to drainage of peatland and from LUCs.
14044 (ISO, 2006), because the value-choices of the weighting The remaining emissions are almost entirely caused by methane
method introduce a bias. However, weighting the results makes from POME treatment and from nitrous oxide emissions, occurring
possible to express the results for all the impact categories with a as field emissions during the application of nitrogen fertiliser.
common aggregated indicator (a single score) and, thus, to identify Sections 3.2 and 3.3 below provide a detailed analysis of the sources
the most severe impacts, i.e. the impacts with the largest contri- of GHG emissions contributing to global warming.
bution to the aggregated result. In order to focus on the most Nature occupation impact is measured in potentially dis-
relevant impact categories, the characterised results, presented in appeared fraction (PDF) per square meter of land occupied in a year,
Table 3, are ranked using the weighting method Stepwise (see per kg of product, i.e. PDF m2*year/kg. Nature occupation is 20%
section 2.2). The weighted results are presented as monetarised lower for RSPO certified palm oil than non-certified, i.e. 1.63
impacts (Weidema, 2009). The results were also weighted against (1.30e2.05) PDF m2*year/kg for certified versus 2.04 (1.12e3.34)
two other LCIA methods (ReCiPe Endpoint (H) V1.13 and IMPACT PDF m2*year/kg for non-certified. Nature occupation impacts are
2002þ V2.143) to ensure that the ranking is not biased by value- mainly caused by land occupation from oil palm plantations and, to
choices applicable only to one method. The three weighting a lesser extent, from roads, nurseries, housing in the estates, palm
methods show unanimously that global warming, nature occupa- oil mills, POME treatment facilities, workshops, and refineries. The
tion and respiratory inorganic are the main impacts, contributing to land occupation triggers the iLUC model accounting for the accel-
more than 90% of the total impact expressed as a single score, in erated deforestation effect. The consequential model also accounts
both certified and non-certified production. The results of the three for the avoided land occupation for wheat, corn and soybean in
weighting methods are shown in the SI, section 9. GHG emissions Brazil, US, Argentina, and Ukraine, as by-products of palm kernel
and nature occupation are also the impacts causing the highest meals and PFAD/PKFAD substitute feeds in the global market for
concern in the public debate on palm oil industry and its envi- animal feeds (SI section 3).
ronmental effects. Results for respiratory inorganics are similar for certified and
In order to compare the consequential results for the three main non-certified in both consequential and attributional modelling.
impact categories of RSPO certified and non-certified palm oil, an RSPO certified palm oil respiratory inorganics is slightly higher (3%)
uncertainty analysis was performed (Fig. 2). The uncertainty anal- than the non-certified production: 2.58 (1.93e4.17) gPM2.5 eq./kg
ysis concerns only the foreground system and is obtained through a RBD oil versus 2.50 (1.51e4.22) gPM2.5 eq./kg RBD oil for non-
Monte Carlo simulation, performed with the LCA software SimaPro certified. The main contributing substances are ammonia, particu-
version 8.4. The procedure to obtain data’s uncertainty values is lates and NOX, with a minor contribution from sulphur dioxide (SO2).
described in the SI, section 10.2. The uncertainty range in Fig. 2 Ammonia emissions originate largely from oil palm cultivation in
represents the 97.5% and 2.5% percentile. The largest uncertainty estates, while particulates are emitted from the oil mill boilers.
ranges are observed for global warming and respiratory inorganic The uncertainty ranges in Fig. 2 for certified and non-certified
results, because nature occupation uncertainties are almost entirely production overlap for all three impact categories. This means
related to the FFB yields. The uncertainty for non-certified is higher that it is not possible to formulate a conclusive comparative
because several data point are dependent on both certified and assertion based on these uncertainty ranges. Therefore, data quality
non-certified uncertainty values. This is because data for the non- and uncertainties were further examined to verify whether a robust
certified production are obtained by subtracting the RSPO certi- comparative assertion could be formulated for any of the impact
fied flows from the total industry flows (see section 2.3). categories.
Fig. 2 shows that RSPO certified RBD oil emits 35% less GHG
emissions than non-certified, i.e. 3.41 (2.61e4.48) kg CO2 eq./kg 3.1. Data quality and uncertainty analysis
RBD for certified versus 5.34 (3.34e8.16) kg CO2 eq./kg for non-
certified. Global warming is largely caused by CO2 emissions A qualitative assessment was performed for the data found for
occurring during oil palm cultivations. The majority of emissions the main elementary flows contributing to each impact category

Fig. 2. Result of Monte Carlo uncertainty analysis for the three most relevant impact categories. The error bars in the figure show the 97.5% and 2.5% percentile, respectively.
Consequential model.

7
J. Schmidt and M. De Rosa Journal of Cleaner Production 277 (2020) 124045

(see SI section 10.1). Impact categories based on not-sufficiently 0.92 kg CO2 eq., respectively (Table 5), corresponding to a reduction
reliable inventory data (SI Table 20) were excluded. The results of of 3.7% of the total non-certified emissions. However, the higher use
the qualitative test are reported in Table 4 (3rd column). Then, a of fertiliser in certified production also causes higher GHG emis-
discernibility analysis was performed to examine whether the re- sions associated with material use in the cultivation stage. Table 5
sults for certified and non-certified are confirmed in at least 95% of shows that the iLUC contribution is significant, accounting for
the Monte Carlo simulations (SI, Fig. 6). The results of the dis- approximately 11% of the GHG emissions in certified production
cernibility analysis are also reported in Table 4 (4th column). and 9% of the emissions in non-certified production. The iLUC GHG
Finally, a Null Hypothesis Significance Test (NHST) was performed emissions in certified palm oil are 20% lower than the non-certified
to test whether the means of the relative impacts of the two al- emissions (0.49 and 0.62 kg CO2 eq. respectively), due to the higher
ternatives are significantly different (see SI Table 21). The results of average yields found in RSPO certified oil palm cultivation (Table 1).
the discernibility analysis are summarised in the 1st and 2nd col- The remaining contributions to the oil crop cultivation stage are not
umn in Table 4. Two significance levels (alpha) were considered to significantly different in the two production systems (Table 5).
evaluate the p-value. The details of the data quality and uncertainty In Palm Oil Mills (POMs), methane emissions occur almost
analysis are reported in the SI, section 10. Table 4 summarizes the entirely during treatment of POME. In RSPO certified production,
results of the NHST test, the qualitative test and the discernibility GHG emissions from POME treatment are 21% lower than non-
analysis described in section 10 of the SI. From the results presented certified (1.19 and 1.51 kg CO2 eq., respectively). The certified pro-
in Table 4, it was possible to conclude that a robust comparative duction performs better than the non-certified because the share of
assertion can only be formulated for global warming. The com- POME treatment with biogas capture is higher in certified pro-
parison concerning the impact categories where only the NHST and duction (16% and 2.4% respectively, in Fig. 2). A large GHG emission
the qualitative test provide a positive outcome (indicated with ‘yes’ reduction can be associated to kernels, as the impacts from its
in Table 4) shall, instead, only be considered as indicative. Con- crushing process by-products, i.e. kernel oil and kernel meal, are
clusions cannot be drawn for the other impact categories. allocated as palm oil and animal feed, respectively. The outputs of
The following section presents a detailed contribution analysis kernel oil and meal substitute palm oil and animal feed. All these
for global warming in consequential and attributional models. effects are aggregated in the contribution related to the kernels in
Table 5. The table shows that the substitution for certified palm oil
mills is much smaller than for non-certified palm oil mills. This is
3.2. Contribution analysis: global warming (consequential) because the kernels in the certified palm oil mills substitute only
certified palm oil while this is not the case for non-certified,
Table 5 shows the contribution analysis for global warming in because the certified and non-certified product systems are
consequential modelling. The cultivation stage is the major source modelled as two separate systems. The smaller avoided impact
of GHG emissions, with 75.6% and 83.5% of the emissions for related to the kernels for the certified oil mills does not mean that
certified and non-certified, respectively. The palm oil milling stage the certified oil mills performs worse in that respect. Instead, it
is the second major source of GHG emissions, contributing to 24.9% means that part of the cultivation stage is substituted by a by-
and 16.7% of the emissions for certified and non-certified, respec- product in the oil mill stage. Hence, the difference in perfor-
tively. The refinery stage only contributes a small share of net mance must be assessed for the whole functional unit, rather than
negative GHG emissions. The negative values indicate that emis- for the oil mill stage alone.
sions are avoided in the refinery stage for both certified and non- The palm oil refinery shows a minor contribution to the global
certified production, respectively, due to the use of by-products. warming result. This is also valid for nature occupation and respi-
In palm oil cultivation, GHG emissions are mostly caused by peat ratory inorganic, in both consequential and attributional modelling.
drainage and LUC emissions. Peat drainage emissions are 1.59 kg Palm oil refining contributes 0.0 l kg CO2 eq./kg RBD palm to the
CO2 eq. lower in RSPO certified production compared to non- total 3.41 kg CO2 eq./kg RBD in both RSPO certified and non-
certified. This alone correspond to a 30% reduction of the non- certified production. This corresponds to a reduction of less
certified emissions. Field GHG emissions, mostly nitrous oxide than 0.3% and 0.2% of the total GHG emissions in certified and
emissions, are related to the nutrient cycle. They occur when ni- non-certified production, respectively. The negative value repre-
trogen fertiliser is applied to the crops. Field emissions are 22% sents a net GHG emission reduction, due to the by-product Palm
lower in certified production compared to non-certified (0.72 and

Table 4
Summary of the uncertainty analyses. “Yes” indicates when the uncertainty analysis supports reliable results and/or when a significant difference between certified and non-
certified is supported.

Impact category Null Hypothesis Significance Testing (NHST) Qualitative test Discernibility analysis

Sign. Diff compared to alpha? Sign. Diff compared to alpha-b? Reliable data? Reliable?

Global warming, fossil yes yes yes yes

Nature occupation yes yes yes no
Respiratory inorganics yes yes yes no
Respiratory organics yes yes yes no
Photochemical ozone, vegetat. yes yes yes no
Eutrophication, aquatic no no yes no
Eutrophication, terrestrial yes yes yes no
Acidification yes no yes no
Ecotoxicity, aquatic yes yes no yes
Ecotoxicity, terrestrial no no no no
Human toxicity, non-carc. yes yes no no
Human toxicity, carcinogens no no no no
Mineral extraction yes yes no no
Non-renewable energy yes yes no no

8
J. Schmidt and M. De Rosa Journal of Cleaner Production 277 (2020) 124045

Table 5
Contribution analysis: GHG emissions in kg CO2 eq. for 1 kg RBD palm oil in 2016 for the consequential model. The total results are shown with and without the iLUC
contribution at the bottom of the table. HCV: High Conservation Value.

Life Cycle Stage Contribution RSPO Certified Non-certified

Oil crop cultivation

Field emissions (related to nutrient cycle) 0.72 0.92
Field emissions (related to peat drainage) 0.77 2.36
Indirect Land Use Changes (iLUC) 0.49 0.62
Material inputs: fertiliser, pesticides, capital goods etc. 0.33 0.21
Energy 0.07 0.08
Other (transport, waste treatment, assets and services) 0.20 0.27
Total crop cultivation stage 2.58 4.46
Palm oil mill
POME treatment 1.19 1.51
Energy inputs 0.03 0.06
Other (transport, waste treatment, assets and services) 0.17 0.18
By-product: kernel 0.43 0.70
By-product: energy and EFB to field application 0.04 0.04
HCV nature conservation 0.01 0.00
Total palm oil mill stage 0.85 0.89
Refinery
Materials: chemicals, water etc. 0.02 0.02
Energy 0.03 0.03
Other (transport, waste treatment, assets and services) 0.02 0.02
By-products: PFAD/PKFAD 0.08 0.08
Total refinery stage ¡0.01 ¡0.01
All stages

Total Result with iLUC (default) 3.41 5.34

Total Result without iLUC 3.03 4.84

Fatty Acid Distillate (PFAD) and Palm Kernel Fatty Acid Distillate LUC emissions for RSPO certified. The cultivation stage is where the
(PKFAD). The major cause of GHG emissions in palm oil refining is largest share of emission reduction can be observed, also in the
energy use. attributional results. In particular, lower GHG emissions are ach-
ieved by certified production due to lower share of oil palm on peat
soil and lower LUC emissions. However, in the attributional model
3.3. Contribution analysis: global warming (attributional) the lower LUC emissions are not due to higher certified yields,
rather due to the historical uses of the land currently under oil palm
Table 6 shows the contribution analysis for global warming in cultivation, as described in section 2.2.
attributional modelling, including the LUC emissions. As for the The contribution to the palm oil milling stage is largely due to
consequential results, the attributional results show that the methane emissions occurring during the treatment of POME, for
cultivation stage is the major source of GHG emissions, followed by the consequential model. However, in the attributional results
the palm oil milling stage. In the cultivation stage, the largest there is no-negative contribution from the by-products, because
source of GHG emissions is peatland drainage for non-certified and

Table 6
Contribution analysis: GHG emissions in kg CO2 eq. for 1 kg RBD palm oil in 2016 for the attributional model. The total results are shown with and without the LUC contribution
at the bottom of the table.

Life Cycle Stage Contribution RSPO Certified Non-certified

Oil crop cultivation

Field emissions (related to nutrient cycle) 0.48 0.49
Field emissions (related to peat drainage) 0.65 1.98
Land use changes CO2 1.15 1.55
Materials: Fertilisers, chemicals and packaging 0.13 0.08
Energy 0.05 0.06
Other (transport, waste treatment, assets and services) 0.00 0.01
Total crop cultivation stage 2.47 4.16
Palm oil mill
POME treatment 0.91 1.13
Energy 0.03 0.05
Other (transport, waste treatment, assets and services) 0.05 0.05
By-product: energy and EFB to field application 0.01 0.01
Total palm oil mill stage 0.92 1.13
Refinery
Materials: chemicals and water 0.01 0.01
Energy 0.03 0.03
Other (transport, waste treatment, assets and services) 0.00 0.00
Total refinery stage 0.04 0.04
All stages

Total Result with iLUC (default) 3.42 5.32

Total Result without iLUC 2.27 3.77

9
J. Schmidt and M. De Rosa Journal of Cleaner Production 277 (2020) 124045

the effect of by-products is not modelled as an avoided production. Table 7 presents the result of the sensitivity analysis concerning
Instead, a share of the impact is allocated to the by-product, based the bullet points above. Overall, the table shows that the conclusion
on the product property used for allocation (mass allocation, in this of the study does not change by modifying the analysed parame-
case, as described in section 2.3). The palm oil refinery stage shows ters, i.e. RSPO certified palm oil still perform better than non-
a small negative contribution to global warming due to the by- certified palm oil in term of GHG emissions (kg CO2 eq.).
product PFAD/PKFAD. The largest contribution to GHG emission Doubling or halving the IPCC 2006, (chapter 2) values used to ac-
in the refinery stage is energy use. No-difference can be detected count for the carbon stock in the conserved nature, the change in
between certified and non-certified palm oil refining. This is also RSPO certified result is below 2% (Table 7, A). The conservative
valid for nature occupation and respiratory inorganic results in the assumption that no nature conservation activity occurs on water-
attributional models. logged peat soil was adapted. Assuming that nature conservation
occurring on peat soil would further lower the GHG contribution of
RSPO certified production, as shown in Table 7, B. These two pa-
3.4. Summary of the results
rameters do not affect the non-certified results because no-nature
conservation activity is assumed to take place in non-certified
Overall, the results indicate that the largest benefit of certified
production.
production is a significant reduction in terms of global warming
We tested the sensitivity of both the certified and non-certified
and nature occupation impacts. A slightly higher impact is found
GHG results to the peat soil emission factor (IPCC, 2014b). Assuming
for respiratory inorganics in RSPO certified, although in this case
a higher emission factor (Table 7, C), the GHG would increase of
the difference between the two production systems is much
approximately 23% for certified and 28% for non-certified produc-
smaller. The impact categories global warming, nature occupation
tion, respectively. Yet, the conclusion of the study would not be
and respiratory inorganic account for more than 90% of the single
affected. The peat drainage depth is based on data from 17 out of 63
score result (Fig. 3 in SI, section 9), in both the consequential and
estates, where peat soil was reported in the RSPO annual assess-
attributional LCA model. The single score expresses the results with
ment reports. Clearly, assuming a deeper peat drainage depth, as in
an aggregated indicator, a monetary value in Stepwise v1.7
Table 7, D - where the same depth is assumed for the compared
(Weidema, 2009).
production systems - would increase the GHG emission of certified
The environmental benefits of RSPO certified production are
production of approximately 6%. Yet, the certified production
mainly due to four factors:
would still show a considerable reduction in GHG emission.
1) Higher yields: lower production inputs (fertilisers, fossil fuel
etc.) and especially lower land use resources, and LUC effects,
4. Conclusion
per unit of product
2) Lower share of cultivated peatland: avoiding palm cultivation in
This paper presents a detailed LCA comparing average RSPO
peatlands prevents CO2 emission from peat drainage
certified and non-certified palm oil. The LCA accounts for a large
3) Nature conservation: setting-aside high conservation value
spectrum of impact categories, relying on a very detailed modelling
(HCV) land reduces GHG emissions and nature occupation
of both production systems. The comparison aims to shed light on
4) Higher share of biogas capture in POME treatment facilities: it
the benefits and trade-offs of RSPO certified palm oil. The objective
reduces methane emissions from POME
is to assess the potential effects of consumer and corporate choice
of palm oil and to help palm oil producers and suppliers to identify
Three out of four major benefits are related to the oil palm
potential areas of improvements in the production systems. The
cultivation. The palm oil refinery stage shows a minor contribution
model accounts for oil palm production, milling and refining, and
to the results and it does not show significant differences between
includes several detailed sub models such as: nutrient balances,
the two production systems under study. Since three impact cate-
global dLUC and iLUC modelling, POME treatment and modelling of
gories account for more than 90% of the single score, it can be
peat soil contribution to the impacts, among others. The conse-
concluded that the certified production performs significantly
quential model reflects the consequences of the market’s response
better than the non-certified, in particular with respect to global
to changes in production of certified or non-certified palm oil and
warming. The nature occupation of certified production is also
the effects of the by-products on markets. The attributional model
lower, but this result is associated with higher uncertainty.
follows the approach of PalmGHG, a GHG accounting tool
Although the global warming comparison is confirmed by thorough
commissioned by RSPO. The severity of the impact categories were
uncertainty analyses, such analyses do not test the sensitivity of the
assessed based on the results of three distinct LCA weighting
result to key modelling assumptions and value choices. Thus, a
methods, all of which indicate the following as the most relevant
sensitivity analysis was also performed, described in section 3.5
impact categories: global warming, nature occupation and respi-
below. The analysis confirmed the robustness of the comparative
ratory inorganic. Key findings of the consequential LCA are:
assertion formulated regarding global warming.
 1 kg of RSPO certified RBD palm oil causes 3.41 (2.61e4.48) kg
3.5. Sensitivity analysis CO2 eq. and non-certified 5.34 (3.34e8.16) kg CO2 eq. Therefore,
RSPO certified palm oil is associated with significantly lower
We evaluated the sensitivity of the LCA results to several GHG emissions (-36%) than non-certified production.
methodological and value choices. The following issues, related to  RSPO certified palm oil nature occupation is 1.63 (1.30e2.05)
key data applied to the model, have been identified throughout the PDF m2*year/kg RBD palm oil while non-certified is 2.04
study: (1.12e3.34) PDF m2*year/kg RBD palm oil. This means that RSPO
certified palm oil causes lower nature occupation (20%) than
⁃ Carbon stock in land set-aside for nature conservation non-certified production.
⁃ Share of nature conservation on waterlogged peat  RSPO certified palm oil respiratory inorganics is 2.58 (1.93e4.17)
⁃ Peat soil’s CO2 emission factor gPM2.5 eq./kg RBD palm oil, while for non-certified is 2.50
⁃ Peat drainage depth for certified estates (1.55e4.22) gPM2.5 eq./kg RBD palm oil. In other words, RSPO
10
J. Schmidt and M. De Rosa Journal of Cleaner Production 277 (2020) 124045

Table 7
Overview of the sensitivity analyses performed for GHG emission’s results (kg CO2 eq.). Consequential model.

Investigated parameter RSPO certified Non-certified

A) Carbon stock in nature conservation

Default: 165 t C/ha 3.41 5.34
Low carbon stock: 83 t C/ha 3.43 5.34
High carbon stock: 330 t C/ha 3.39 5.34
B) Nature conservation on waterlogged peat
Default: 0% nature conservation on peat 3.41 5.34
10% nature conservation on peat 3.38 5.34
50% nature conservation on peat 3.25 5.34
100% nature conservation on peat 3.09 5.34
C) Peat soil CO2 emission factor
Default: peat CO2 emission factor at 41.4 t CO2/ha 3.41 5.34
High peat CO2 emission factor at 84.5 t CO2/ha 4.22 7.76
D) Peat drainage depth for certified estates
Default: drainage depth at 57 cm for certified and 73 for non-certified 3.41 5.34
Drainage depth at 73 cm for both certified and non-certified 3.64 5.34

certified palm oil is associated with slightly higher respiratory References

inorganics (3%) than non-certified production.
Arvidsson, R., Perssonb, S., Fro €lingb, M., Svanstro
€m, M., 2013. Life cycle assessment
of hydrotreated vegetable oil from rape, oil palm and Jatropha. J. Clean. Prod. 19
Based on a thorough uncertainty analysis, only the comparison (2e3), 129e137.
between the GHG emissions (global warming results) of the certified Bessou, C., Chase, L.D.C., Henson, I.E., Abdul-Manan, A.F.N., Mila  i Canals, L., Agus, F.,
and non-certified RBD oil was deemed conclusive. The results for the Sharma, M., Chin, M., 2014. Pilot application of PalmGHG, the Roundtable on
Sustainable Palm Oil greenhouse gas calculator for oil palm products. J. Clean.
remaining impact categories shall be interpreted as indicative. Prod. 73, 136e145.
Uncertainties are mainly associated with toxicity impacts, non- Brandi, C., Cabani, T., Hosang, C., Schirmbeck, S., Westermann, L., Wiese, H., 2013.
renewable energies, mineral extraction and ionizing radiation and Sustainability Certification in the Indonesian Palm Oil Sector. Benefits and
Challenges for Smallholders. DIE Studies. German Development Institute, Bonn,
ozone layer depletion due to insufficient inventory data. Concern- ISBN 978-3-88985-581-7. ISSN 1860-0468.
ing the carbon footprint, data on the share of oil palm on peat are Carlson, K.M., Heilmayra, R., Gibbsd, H.K., Noojipadyg, P., Burnsg, D.N.,
rare, especially for certified production, and present a great range of Mortonh, D.C., Walkerg, N.F., Paolij, G.D., kremenk, C., 2017. Effect of oil palm
sustainability certification on deforestation and fire in Indonesia. Proc. Natl.
variation in literature. The share of oil palm on peat is a key aspect
Acad. Sci. U.S.A. 115 (1), 121e126. January 2, 2018.
in carbon footprinting of palm oil production. More accurate data Cattau, M.E., Marlier, M.E., DeFries, R., 2016. Effectiveness of Roundtable on Sus-
on peat soil occupation from oil palm estates would improve the tainable Palm Oil (RSPO) for reducing fires on oil palm concessions in Indonesia
accuracy of the LCA results. Moreover, a streamlined and quality- from 2012 to 2015. Environmentl Research Letters 11, 105007. https://doi.org/
10.1088/1748-9326/11/10/105007.
assured data collection in the RSPO annual assessment reports Chase, L., Henson, I., Faizal, A., Abdul-Manan, N., Agus, F., Bessou, C., Mila i Canals, L.,
would facilitate the quantification of the environmental effects of Sharma, M., 2012. Palm GHG e A Greeenhouse Gas Accounting Tool for Palm Oil
the certification. Products. The Roundtable for Sustainable Palm Oil e RSPO, Kuala Lumpur,
Malaysia. Accessed October 2017. https://www.rspo.org/file/PalmGHG%20Beta%
20version%201.pdf.
CRediT authorship contribution statement Choo, Y.M., Muhamad, H., Hashim, Z., Vijaya, S., Puah, C.W., Tan, Y., 2011. Determi-
nation of GHG contributions by subsystems in the oil palm supply chain using
the LCA approach. Int. J. Life Cycle Assess. 16 (7), 669e681.
Michele De Rosa: Data curation. De Rosa, M., Knudsen, M.T., Hermansen, J.E., 2016. A comparison of Land Use
Change models: challenges and future developments. J. Clean. Prod. 113,
Declaration of competing interest 183e193.
Faostat, 2018. FAOSTAT Agriculture Data, Food and Agriculture Organisation of the
United Nations (FAO). Rome. http://faostat.fao.org/. Accessed March 2017.
The authors declare the following financial interests/personal Gassler, B., Spiller, A., 2018. Is it all in the MIX? Consumer preferences for segre-
relationships which may be considered as potential competing gated and mass balance certified sustainable palm oil. J. Clean. Prod. 195, 21e31,
2018.
interests:This work is financed through a crowdfunded project by Hauschild, M., Potting, J., 2005. Spatial Differentiation in Life Cycle Impact Assess-
parties, which may be affected by this research.None of them ment - the EDIP2003 Methodology. Environmental News No. 80 2005. Danish
participated in the conceptualisation, design, data collection, Environmental Protection Agency, Copenhagen.
Hutabarat, S., Slingerland, M., Rietberg, P., Dries, L., 2018. Costs and benefits of
analysis or preparation of the manuscript.
certification of independent oil palm smallholders in Indonesia. Int. Food
Agribus. Manag. Rev. 21 (6), 681e700.
Acknowledgments IPCC, 2006. In: Eggleston, H.S., Buendia, L., Miwa, K., Ngara, T., Tanabe, K. (Eds.),
2006 IPCC Guidelines for National Greenhouse Gas Inventories, Prepared by the
National Greenhouse Gas Inventories Programme. IGES, Japan. Accessed
This study is the outcome of a crowdfunded project. The authors November 2017. http://www.ipcc-nggip.iges.or.jp/public/2006gl/vol4.html.
would like to thank the project partners for their financial support IPCC, 2014a. In: Hiraishi, T., Krug, T., Tanabe, K., Srivastava, N., Baasansuren, J.,
and for the valuable inputs provided throughout the development Fukuda, M., Troxler, T.G. (Eds.), 2013 Supplement to the 2006 IPCC Guidelines
for National Greenhouse Gas Inventories: Wetlands. IPCC, Switzerland.
of the project. The partners are: Beiersdorf, DuPont Nutrition and Accessed August 2014. http://www.ipcc-nggip.iges.or.jp/public/wetlands/pdf/
Biosciences, ERASM, Ferrero, Novozymes, Golden Agri-Resources Wetlands_Supplement_Entire_Report.pdf.
(PT Smart Tbk), and Unilever. The authors would also like to IPCC, 2014b. In: Pachauri, R.K., Meyer, L.A. (Eds.), Climate Change 2014: Synthesis
Report. Contribution of Working Groups I, II and III to the Fifth Assessment
thank the anonymous reviewers for the constructive feedbacks. Report of the Intergovernmental Panel on Climate Change. IPCC, Geneva,
Switzerland. Accessed August 2017. https://www.ipcc.ch/pdf/assessment-
Appendix A. Supplementary data report/ar5/syr/AR5_SYR_FINAL_All_Topics.pdf.
Jolliet, O., Margni, M., Charles, R., Humbert, S., Payet, J., Rebitzer, G., Rosenbaum, R.,
2003. Impact 2002þ: a new life cycle impact assessment methodology. Int. J.
Supplementary data to this article can be found online at Life Cycle Assess. 8 (6), 324e330.
https://doi.org/10.1016/j.jclepro.2020.124045. Lam, M.K., Lee, K.T., Rahmanmohamed, A., 2009. Life cycle assessment for the

11
J. Schmidt and M. De Rosa Journal of Cleaner Production 277 (2020) 124045

production of biodiesel: a case study in Malaysia for palm oil versus jatropha Schmidt, J., 2017. Life Cycle Assessment of Palm Oil at United Plantations Berhad
oil. Biofuels, Bioproducts and Biorefining 3 (6), 601e612. 2017, Results for 2004-2016. United Plantations Berhad, Teluk Intan, Malaysia.
Ma, A.N., Choo, Y.M., Toh, T.S., Chua, N.S., 2007. Renewable energy from palm oil Accessed March 2018. http://lca-net.com/p/2739.
industry. Not published. Updated version of chapter 17. In: Singh, G., Huan, L.K., Schmidt, J., Dalgaard, R., 2012. National and Farm Level Carbon Footprint of Milk e
Leng, T., Kow, D.L. (Eds.), Oil Palm and the Environment e A Malaysian Methodology and Results for Danish and Swedish Milk 2005 at Farm Gate. Arla
Perspective. Malaysian Oil Palm Growers Council, Kuala Lumpu, 1999. Foods, Aarhus, Denmark. http://lca-net.com/p/220.
Meijide, A., de la Rua, C., Guillaume, T., et al., 2020. Measured greenhouse gas Schmidt, J., De Rosa, M., 2018. Enhancing land Use change modelling with IO data.
budgets challenge emission savings from palm-oil biodiesel. Nat. Commun. 11. In: Presentation at the SETAC Europe 28th Annual Meeting, Rome 13-17 May
Merciai, S., Schmidt, J., 2017a. Methodology for the construction of global multi- 2018. Accessed June 2018. http://lca-net.com/p/3036.
regional hybrid supply and use tables for the EXIOBASE v3 database. J. Ind. Schmidt, J., de Saxce
, M., 2016. Arla Foods Environmental Profit and Loss Accounting
Ecol. 22 (3), 516e531. 2014. Environmental Project No. 1860, 2016. Danish Environmental Protection
Merciai, S., Schmidt, J., 2017b. Land use change and electricity models in a multi- Agency, Copenhagen. http://lca-net.com/p/2343.
regional hybrid input output framework. In: Paper 25th IIOA Conference in Schmidt, J., Weidema, B.P., Brand~ ao, M., 2015. A framework for modelling indirect
Atlantic City, 19-23 June 2017, USA. land use changes in life cycle assessment. J. Clean. Prod. 99, 230e238.
Morgans, C.L., Meijaard, E., Santika, T., Law, E., Budiharta, S., Ancrenaz, M., Stadler, K., Wood, R., Bulavskaya, T., So €dersten, C.J., Simas, M., Schmidt, S.,
Wilson, K.A., 2018. Evaluating the effectiveness of palm oil certification in Usubiaga, A., Acosta-Ferna
ndez, J., Kuenen, J., Bruckner, M., Giljum, S., Lutter, S.,
delivering multiple sustainability objectives. Environ. Res. Lett. 13, 064032, Merciai, S., Schmidt, J., Theurl, M.C., Plutzar, C., Kastner, T., Eisenmenger, N.,
2018. Erb, K.H., de Koning, A., Tukker, A., 2018. Exiobase 3: developing a time series of
PAS2050, 2011. Specification for the assessment of the life cycle greenhouse gas detailed environmentally extended multi-regional input-output tables. J. Ind.
emissions of goods and services. British Standard (BSI) (ISBN 978 0 580 71382 Ecol. 22 (3), 502e515.
8). Tan, Y.A., Muhammad, H., Hashim, Z., Vijaya, S., Puah, C.W., Let, C.C., Ngan, M.A.,
Puah, C.W., Choo, Y.M., Ong, S.H., 2013. Production of Palm Oil with Methane May, C.Y., 2010. Life cycle assessment of refined palm oil production and frac-
Avoidance at Palm Oil Mill: A Case Study of Cradle-To-Gate Life Cycle tionation (part 4). Journal of Oil Palm Research 22 (December), 913e926.
Assessment. UPRD, 2007. Data provided by the United Plantations Research Department, Di-
RSPO, 2017. Principles and criteria assessment progress reports. RSPO webpage. rector of Research Dr. Gurmit Singh. United Plantations Research Department,
Accessed December 2017. https://www.rspo.org/certification/principles-and- Teluk Intan, Malaysia.
criteria-assessment-progress. Vijaya, S., Ma, A.N., Choo, Y.M., Nik Meriam, N.S., 2008. Life cycle inventory of the
RSPO, 2018. Certification in numbers. Roundtable on sustainable palm oil (RSPO). production of crude palm oil - a gate to gate case study of 12 palm oil mills.
Accessed October 2017. http://www.rspo.org/about/impacts. Journal of Oil Palm Research (20), 484e494. June.
Saswattecha, K., Kroeze, C., Jawjit, W., Hein, L., 2015. Assessing the environmental Weidema, B., 2003. Market Information in Life Cycle Assessment, Environmental
impact of palm oil produced in Thailand. J. Clean. Prod. 100, 150e169. Project No. 863. Danish Environmental Protection Agency, Copenhagen.
Schmidt, J., 2007. Life Cycle Assessment of Rapeseed Oil and Palm Oil: Ph.D. Thesis, Weidema, B.P., 2009. Using the budget constraint to monetarise impact assessment
Part 3: Life Cycle Inventory of Rapeseed Oil and Palm Oil. Department of results. Ecol. Econ. 68 (6), 1591e1598.
Planning and Development, Aalborg University accessed March 2018. http://lca- Weidema, B.P., Wesnæs, M., Hermansen, J., Kristensen, T., Halberg, N., 2008. Envi-
net.com/p/2742. ronmental improvement potentials of meat and dairy products. In: Eder, P.,
Schmidt, J., 2008. Life Cycle Assessment of Palm Oil at United Plantations Berhad. Delgado, L. (Eds.), Sevilla: Institute for Prospective Technological Studies. (EUR
United Plantations Berhad, Teluk Intan, Malaysia. 23491 EN).
Schmidt, J., 2010. Comparative life cycle assessment of rapeseed oil and palm oil. Wicke, B., Sikkema, R., Dornburg, V., Faaij, A., 2011. Exploring land use changes and
Int. J. Life Cycle Assess. 15 (2), 183e197. the role of palm oil production in Indonesia and Malaysia. Land Use Pol. 28,
Schmidt, J., 2015a. Life cycle assessment of five vegetable oils. J. Clean. Prod. 87, 193e206. https://doi.org/10.1016/j.landusepol.2010.06.001, 2011.
130e138. Woittiez, L.S., van Wijk, M.T., Slingerland, M., van Noordwijk, M., Gillera, K.E., 2017.
Schmidt, J., 2015b. Nature conservation in life cycle assessment e new method and Yield gaps in oil palm: a quantitative review of contributing factors. Eur. J.
case study with the palm oil industry. In: Extended Abstract for Presentation at Agron. 83, 57e77, 2017.
the SETAC2015, Barcelona 3-7 May 2015. Accessed May 2018. http://lca-net. Yee, K.F., Tan, K.T., Abdullah, A.Z., Lee, K.T., 2009. Life cycle assessment of palm
com/p/1818. biodiesel: revealing facts and benefits for sustainability. Appl. Energy 86
Schmidt, J., 2016. Life cycle assessment of palm oil e investigating nature conser- (Suppl. 1), S189eS196.
vation and other GHG mitigation options. In: Paper Presented at the 5th In- Yusoff, S., Hansen, S.B., 2007. Feasibility study of performing an life cycle assess-
ternational Conference on Oil Palm and Environment (ICOPE), 2016. May 2018. ment on crude palm oil production in Malaysia. Int. J. Life Cycle Assess. 12 (1),
http://lca-net.com/p/2479. 50e58.

12