• Open Access

Opportunities for avoidance of land-use change through substitution of soya bean meal and cereals in European livestock diets with bioethanol coproducts

Authors


R. M. Weightman, tel. +44 1954 267 666, fax +44 1954 267 659, e-mail: Richard.weightman@adas.co.uk

Abstract

An analysis is presented which quantifies the potential for distillers dried grains with solubles (DDGS, a coproduct of wheat bioethanol production) to replace soya bean meal (SBM) and cereals in livestock rations. A major proportion of the SBM imported into Europe as a protein-rich feedstuff for livestock comes from South America, where land-use change (LUC) is associated with high carbon emissions. Production of DDGS can therefore reduce LUC in South America by substitution of SBM in animal feed. The analysis indicates that a single bioethanol distillery processing 1 million tonnes of wheat, and producing ca. 330 000 tonnes of DDGS per annum, would substitute at least 136 493 tonnes of whole soya beans grown on 47 725 ha of land, and save greenhouse gas emissions equivalent to 0.63 million tonnes CO2 per annum. By growing sugar beet and wheat in an average ratio of 0.06 : 0.94 on 1 ha of land in Europe, the net area of agricultural land required to produce feed ingredients equivalent to 6.08 t of sugar beet pulp (SBP) and 1.72 t of DDGS associated with 2363 L of bioethanol, is reduced to 0.40 ha. This accounts for 0.42 ha of soya that is not required when DDGS displaces SBM, and 0.18 ha of wheat that is not required when DDGS and SBP displace wheat in livestock rations.

Introduction

Greenhouse gas emissions (GHG) reduction is the main driver for biofuels development in the European Union (EU) with a target originally set in the Biofuels Directive (2003/30/EC) of a 5.75% inclusion of biofuel by volume in the road transport mix by 2010 (EU, 2003). In recognition of sourcing biofuels in a sustainable way, the Renewable Energy Directive (RED; 2009/28/EC) specifies the level of GHG savings to be expected. Guidelines are available in the RED (Annex V) which state default values for various biofuel supply chains to be used for reporting, in the absence of specific data from individual fuel suppliers (EU, 2009).

Following initial broad support for biofuels, Searchinger et al. (2008) raised important questions over the indirect land-use change (ILUC) associated with biofuel production, suggesting that using cereal grains such as maize in the United States, would lead to conversion of land elsewhere in the world in order to grow the food crops which had been displaced. If this conversion of land took place on high carbon stock land (for instance forest or permanent pasture) then their conversion to crop land would be associated with release of significant quantities of carbon into the atmosphere which would negate the potential GHG savings associated with any biofuel. Searchinger's analysis led to the so-called ‘Gallagher review’ in the United Kingdom, which led to a reduction in the 2010 target for the inclusion of biofuel in road transport fuel to 3.5% (Gallagher, 2008).

An economically important coproduct of bioethanol production from cereals, accounting for around one-third of the dry matter of the initial grain in the case of wheat, is distillers dried grains and solubles (DDGS), a protein-rich animal feed containing ca. 33 g crude protein 100 g−1 (Smith et al., 2006; Cottrill et al., 2007). In response to Searchinger's analysis, Croezen & Brouwer (2008) suggested that wheat (W-) DDGS produced in Europe would substitute locally produced wheat and maize, but also soya bean meal (SBM) originating in South America. Other protein-rich coproducts include vinasse (up to 35 g 100 g−1; Stemme et al., 2005) originating from spent yeast cells and unfermented sugars during production of bioethanol from beet sugar. This fraction is often added back to the sugar beet pulp (SBP) to give a feed which has a crude protein content in the range 10.6–13.5 g 100 g−1 (Ministry of Agriculture Fisheries & Food, 1975). For convenience in this paper, the SBP and vinasse have been considered together as a single product.

Soya beans are grown principally for the defatted SBM, with the oil as a secondary product (LMC International Ltd., 2006; Steinfeld et al., 2006). Dros (2004) forecasted the continued expansion of soya cropping in South America under a ‘business as usual scenario’ with soya cropping encroaching on both natural habitats and existing pastures, while pushing cattle ranchers into forest areas. Lapola et al. (2010) have described soya expansion replacing pasture in Brazil, with a resultant effect of pushing the boundaries of pasture into existing Amazonian forest. Therefore, given the environmental costs of soya bean production associated with extensive deforestation (Dros, 2004; Gasparri et al., 2008; Gasparri & Grau, 2009) as well as conversion of habitats such as cerrado and pasture (Sampaio et al., 2007; Morton et al., 2006; Zak et al., 2008) the displacement of SBM by W-DDGS and other protein-rich coproducts means that bioethanol production could be seen as one mechanism by which to avoid damaging land use. In this context, the potential benefits of bioethanol coproducts in Europe have not been thoroughly quantified, although Özdemir et al. (2009) have described the potential to avoid deforestation in Brazil based on the substitution of SBM with rapeseed meal. In addition to reducing the pressure on LUC in South America through substitution of soya, additional substitution of a proportion of the cereals in the diet by DDGS and by SBP, will at the same time remove some of the risk of ILUC from using cereals for biofuel production.

The value of coproduct credits, which can be used in the calculation of GHG savings from bioethanol production, are estimated using a proportional allocation method and have been discussed by Punter et al. (2004) and Edwards et al. (2006). In the European Joint Research Centre (JRC) (2007) report, the proportional allocation of SBM by wheat DDGS was based on a theoretical substitution ratio of 0.78 t SBM t DDGS−1, a value derived from the relative protein contents of 49 g 100 g−1 for SBM and 38.5 g 100 g−1 for W-DDGS. For SBP with vinasses added, it is assumed that 1 MJ dry SBP replaces 0.83 MJ of dry wheat grain based on their similar protein contents (JRC, 2007). In commercial practice; however, the achievement of full substitution of soya with coproducts will also be influenced by the inclusion levels of the coproduct in the diet for different species (Bremer et al., 2010), and limited by competition with other commodities available as cheaper ingredients in a least cost ration formulation (LCRF) system.

While proportional allocation is not proposed as a methodology for calculation of coproduct credits within the RED, determination of an appropriate substitution ratio is very relevant to quantifying the amount of SBM which will be displaced by bioethanol coproducts, and hence to what extent ILUC might be mitigated by biofuel production in Europe.

In order to test the hypothesis of Croezen & Brouwer (2008), it was therefore necessary to clarify and quantify certain elements of the SBM supply chain into the EU27 particularly (i) the proportion of substitution of DDGS for SBM in typical livestock diets, (ii) the proportion of EU27 SBM imports which are supplied by South American soya, (iii) the types of habitat which are being used for the production of SBM in the countries of origin within South America, (iv) the carbon stocks associated with those habitats, and (v) assumptions about the allocation of carbon stock loss to soya cropping following LUC.

The aim of the present study was to consider each of these elements, and to build an additional calculation of a coproduct credit based on avoidance of LUC, soya production and hence carbon emissions in South America, which could be allocated as a saving to W-DDGS produced in the EU, and to consider the implications for substitution of cereal grains and by-products by both DDGS and SBP.

Methodology

Quantification of soya imports into EU27 and countries of origin

Volumes of annual soya imports into the EU27 for two commodity groups relating to soya were collated for the period 1999 to 2008 from the Eurostat database (http://epp.eurostat.ec.europa.eu; accessed on September 29, 2009) and classified as follows:

  • 1Whole soya beans (excluding those for sowing),
  • 2SBM: Oilcake and other solid residues, whether or not ground or in the form of pellets, resulting from the extraction of soya bean oil.

The volumes of SBM were converted to whole soya bean equivalents (WSBe). Typical soya bean seed oil content is 20% (Weiss, 1983; Robinson, 1987) and therefore assuming a SBM yield of 80% from whole soya beans, the WSBe was estimated as 1/0.8 or 1.25 times the weight of SBM imported. Soya bean imports as total WSBe were therefore estimated as follows:

image

Crop yields

Wheat, sugar beet and soya bean yields, production and areas harvested were collated from the FAO statistical database, FAOStat (http://faostat.fao.org; accessed on March 4, 2010). Soya was identified using the FAO classification ‘Soybeans’.

Allocation of habitat undergoing deforestation and estimation of carbon stocks

Descriptions and areas of habitats undergoing deforestation in three South American Countries (Argentina, Brazil and Paraguay) to 2020 were taken from the scenarios described by Dros (2004). These habitats were linked to climate region classifications for soil organic carbon stocks to 30 cm depth as defined in the 2006 IPCC Guidelines for National Greenhouse Gas Inventories, Volume 4, Agriculture Forestry and Other Land Use (IPCC, 2006; table 2.3), except in the case of Atlantic Forest, Chaco and Yungas Forests in Argentina, where data for soils to 20 cm depth (Gasparri et al., 2008) were used and in the case of cerrado soils to 40 cm depth in Brazil (Corbeels et al., 2006) were used. Aboveground biomass estimates in the selected habitats were collated from values given by IPCC (2006; table 4.7) and the carbon in aboveground biomass for each habitat type was then estimated using a default value of 0.47 for the fraction of carbon in the biomass (IPCC, 2006; table 4.1). Root carbon stocks were estimated from the aboveground biomass, using ratios of belowground to aboveground biomass given by IPCC (2006; table 4.4). The total carbon stock per hectare of land was then estimated as the sum of the soil, root and aboveground carbon.

Estimation of carbon losses following deforestation

Following deforestation, it was assumed that all root and aboveground carbon was lost. The soil carbon was assumed to reduce to a constant after 20 years, based on a stock change factor for long term cultivated tropical soils, using values of 0.48 for moist/wet soils and 0.58 for dry soils (IPCC, 2006; table 5.5). The difference between the initial (i.e. total) and the final soil carbon values represent the soil carbon loss. The total carbon loss (soil+root+aboveground vegetation) was allocated to the subsequent cropping activities, and annualized over 20 years. The annualized losses for each habitat expressed as CO2e ha−1 were estimated from the carbon values, using a multiplier of 3.67.

Proportion of substitution of SBM by W-DDGS in livestock diets

In order to estimate the typical inclusion levels and substitution ratios of DDGS for SBM in commercial practice, a LCRF approach was used (data presented in Supporting Information Table S1) using ingredient prices in October 2009. For each livestock grouping studied, test diets including W-DDGS were compared with a control diet with no W-DDGS, and containing a typical amount of SBM. For pigs, the control diet contained 132 g SBM kg−1, and a number of runs were created, each representing a weighted average of 11 pig feeds representing both home-mix and compound diets, with maximum W-DDGS levels (in parentheses) set for different body weights and diet types as follows: 15–30 kg (50 g kg−1), 30–65 kg (100 g kg−1), 65 kg finishers (150 g kg−1), lactating sows (200 g kg−1), sows (250 g kg−1). For a typical diet containing W-DDGS introduced at 62 g kg−1, the substitution ratio was 0.35 t SBM t DDGS−1. For ruminants, the control diet contained 71 g SBM kg−1, and two runs were created, each representing a weighted average of 20 dairy, calf, beef and sheep feeds. The average ruminant test diet incorporated W-DDGS at an inclusion level of 45 g kg−1, determined by setting the W-DDGS price at the same level used in the pig test diets, and the substitution ratio was 0.26 t SBM t−1 DDGS. The LCRF takes into account the levels of digestible essential amino acids and energy concentrations in the case of nonruminants and metabolizable protein in the case of ruminants, as well as the digestibility of other key ingredients including phosphorus and micro-nutrients. For poultry diets incorporation rates of 65 g kg−1 were used, based on the example reported by Lywood et al. (2009) with a substitution ratio of 0.56 t SBM t−1 W-DDGS.

Potential inclusion rates of W-DDGS differ both between livestock categories (pigs, poultry and ruminants) and types of livestock within categories. In order to estimate the impact of W-DDGS on SBM use as a whole, the relative numbers of livestock units (LSU; reflecting the feed requirements of each individual animal category) in the EU27 in 2007 were collated from Eurostat. By combining the proportions of LSU for pigs, poultry and ruminants with the typical inclusion rates for W-DDGS for each livestock type, weighted average inclusion rates and substitution ratios for EU livestock diets were calculated.

The above scenario reflects the commercial position at the time of the study. In order to estimate the impact of W-DDGS on SBM in a future scenario, an assumption was made that W-DDGS would be included in more nonruminant than ruminant diets. It was also assumed that higher inclusion levels of W-DDGS, and higher substitution rates of SBM would be achieved in the future. The amounts of compound diets produced in Europe in 2007 were collated from the European Feed Manufacturers Association (FEFAC, 2009), and the potential substitution rates taken from Lywood et al. (2009). By combining the volumes of the different diet types produced in Europe with the typical inclusion rates for W-DDGS for each livestock type, weighted average inclusion rates and substitution ratios for future EU livestock diets were calculated.

Allocation of carbon losses following deforestation and soya cropping to DDGS and bioethanol production

Using the individual estimates of carbon losses associated with soya production collated above, the following approach was taken to allocate these as credits for W-DDGS production, where W-DDGS displaces SBM from EU livestock diets:

  • 1Carbon emissions (t C ha−1) in the key South American countries (Argentina, Brazil and Paraguay) supplying the EU27, following deforestation attributed to subsequent soya cropping were expressed as t CO2e ha−1,
  • 2The emissions in terms of CO2e/ha were expressed per tonne of soya, based on national yields of soya per hectare,
  • 3The emissions in terms of t CO2e t WSBe−1 were reduced pro rata, according to the proportions of soya imported into Europe from Argentina, Brazil and Paraguay,
  • 4The emissions of CO2e associated with soya production were expressed as a credit per tonne of W-DDGS, based on the average substitution ratio of DDGS for SBM estimated from LCRF across livestock types, and the proportion of livestock diets in Europe,
  • 5The credit calculated in (4) was reduced by a factor of 0.2 based on mass of products, to account for the fact that some soya beans would still need to be grown to supply oil for existing markets,
  • 6The credit in g CO2e kg DDGS−1 was expressed per kg of ethanol, based on a ratio of 1.14 kg DDGS (fresh weight basis) per kg ethanol from a UK wheat-bioethanol distillery (Renewable Fuels Agency, 2008) and also expressed per MJ of energy, based on the energy content of ethanol (26.72 MJ kg−1).

Results and Discussion

The EU27 imported 15.2 Mt of soya beans and 23.6 Mt of SBM in 2007 (Table 1), equivalent to an original production of ca. 45 Mt of whole soya beans per annum in the countries of origin. Three South American countries, Argentina (Ar), Brazil (Br) and Paraguay (Pr) account for the majority (0.89) of the soya imports into the EU27. Conversely, when the production of soya beans in these countries is examined, it is seen that 0.39, 0.25 and 0.18 of the total soya beans produced in Ar, Br and Pr, respectively, are exported to supply the EU market (Table 2).

Table 1.   Imports of whole soya beans and soya bean meal (SBM) into the EU27 by country of origin in 2007, converted to total WSBe
CountryWhole soya beans (t)SBM (t)WSBe from SBM* (t)Total WSBe (t)Proportion of
EU27 supply
[A]
  • Source: Eurostat.

  • *

    Whole soya bean equivalents (WSBe) estimated from SBM × 1.25.

Argentina312 89514 642 50118 303 12618 616 0200.420
Brazil9 492 8558 515 96310 644 95320 137 8080.450
Canada797 74460797598805 3420.018
China21 37825 74532 18153 5600.001
India10717989748983<0.001
Norway0155 168193 960193 9600.004
Paraguay1 046 466117414671 047 9330.023
Russia014 49418 11818 118<0.001
Ukraine143 67800143 6780.003
USA3 275 657150 447188 0593 463 7150.078
Uruguay79 2612100262581 8860.002
Others40 48230 37437 96878 4500.002
Total15 210 42423 551 22229 439 02844 649 4521.000
Table 2.   Production, areas harvested and yields of soya beans in 2007, and proportion of each countries total production supplying the EU27 market
CountryTotal production (t)Proportion of total
production supplying
export to EU27*
Area harvested (ha)Yield (t ha−1)
[B]
  • Source: FAOStat.

  • *

    For corresponding EU27 import figures see Table 1.

Argentina47 482 7840.39215 981 2642.97
Brazil57 857 2000.34820 565 3002.81
Canada2 695 7000.2991 171 5002.30
China13 800 1470.0048 900 0681.55
India10 968 0000.0018 880 0001.24
Paraguay5 856 0000.1792 429 0002.41
Russian Federation651 8400.028709 9000.92
Ukraine722 6000.199583 1001.24
USA72 860 4000.04825 960 0002.81
Uruguay779 9200.105366 5352.13

Dros (2004) predicted the continued expansion of soya in four key South American producing countries assuming a ‘business as usual’ scenario, including expansion of the global production for soya beans to 300 Mt, and a continual increase in soya bean yields based on historic trends. Excluding Bolivia (which was included in Dros' analysis but currently does not export to the EU27) the areas and types of habitat predicted to be deforested in Ar, Br and Pr are shown in Table 3.

Table 3.   Habitat types and areas subject to land use change associated with expansion of soya 2004–2020 forecasted by Dros (2004)
CountryHabitatExpansion of soya
2004–2020 (000 ha)
Average annual expansion
of soya (ha yr−1)
Proportion of habitat deforested
within each country
[C]
ArgentinaAtlantic Forest30018 7500.06
Chaco4850303 1250.91
Yungas20012 5000.04
BrazilCerrado9600600 0000.73
Transition and rainforest3600225 0000.27
ParaguayAtlantic Forest100062 5000.53
Chaco90056 2500.47
Total 20 4501 278 125 

The assumptions made in the present study rely heavily on figures from Dros (2004) forecasting an increase in soya production driven by external demand. These figures are broadly corroborated by other sources e.g. LMC (2006) who forecast 221 million tonnes supply of SBM equivalent to 278 Mt of WSBe by 2020, the Food and Agricultural Policy Research Institute (FAPRI) (2009) who forecast production of 295 Mt of soya in 2018 (applying an annual growth rate of 2% the estimate reaches 307 Mt by 2020), and the Organisation for Economic Co-operation and Development (OECD) (2009) who forecast 407 million tonnes of oilseeds by 2018, which assuming soybeans account for 74% of total oilseed production (based on soya bean, rapeseed and sunflower production in 2007; FAOStat) gives an expected world production of 301 million tonnes of soya beans in 2018, and ca. 312 Mt by 2020.

Given that world soya bean production is therefore expected to increase from 214 Mt in 2005 (FAOStat) to ca. 300 Mt, an average increase of 5.7 Mt yr−1 implies an increase in the world area sown to soya of ca. 2.4 Mha yr−1 (allowing for an annual yield increase of 0.02 t ha−1 yr−1 based on yield trends over the last decade; FAOStat). In this context the estimates from Dros (2004) for annual soya expansion in Ar, Br and Pr of 1.2 Mha yr−1 (Table 3) appear realistic. Similarly, based on historical production trends between 1998 and 2007 (FAOStat), the average annual increase in the area of soya harvested has been 0.96, 0.91 and 0.15 Mha yr−1 for Ar, Br and Pr, respectively (total 2.02 Mha yr−1). Therefore given that both historic trends in expansion of soya area and forecasted soya production increases broadly support the continued demand for land presented by Dros (2004), these figures were used to estimate the effects of LUC from deforestation and conversion of cerrado.

The carbon stocks associated with the various habitat types are shown in Tables 4–6 (soil, aboveground and root carbon, respectively). In each case, a high and low estimate is given from IPCC figures, and either a mid-point value chosen for further calculations, or where available, specific values from Gasparri et al. (2008) and Corbeels et al. (2006) have been used. Based on the losses of carbon from soil (using a stock change factor) and assuming all aboveground and root carbon lost, the annualized emissions are shown in Table 7. Based on the carbon loss for each habitat type and the proportion of each habitat within each country (Table 3), the 20-year annualized emissions could be allocated to each country studied (Table 8). Using the yields of soya (Table 2), annualized carbon losses were then converted to emissions per tonne of soya as WSBe produced by each of the three countries, and using the proportion of EU27 imports coming from each of the three countries (Table 2). Finally an average figure for total CO2 emissions associated with deforestation and expressed per tonne of soya entering the EU27 could be estimated. However, it is acknowledged that while SBM is the major product from soya beans, by reducing soya production, there would also be a parallel loss of the soya oil which currently supplies world markets, principally for food use and to a lesser extent biodiesel. This potential deficit in vegetable oil from South America might therefore have to be supplied from other crops e.g. oilseed rape, groundnut, sunflower and palm, each with differing yields of oil per unit area and different effects on LUC. Lapola et al. (2010) have suggested as an alternative to soya biodiesel, that palm oil production would be associated with the smallest LUC. It is not possible to state categorically which cropping systems would expand in response to a reduction soya oil, therefore for simplicity it was decided to reduce the coproduct credit by 0.2 (based on the oil content of soya beans by mass) reflecting the fact that a replacement vegetable oil would need to be supplied somewhere in the world and this could easily be in South America. This analysis gives a GHG emissions balance associated with soya imported into the EU27 of 4.62 t CO2e t WSBe−1 (Table 8). A further reduction in the coproduct credit could also arise from considering the oil content of soya bean on an energy or economic basis, rather than by mass. Modelling the demands for, and relative prices of commodities, was beyond the scope of this paper and may be worthy of further study.

Table 4.   Estimates of soil carbon (C) for soils in five different habitat types in Argentina, Brazil and Paraguay
CountryHabitatSoil C stock range from
IPCC 2006 (t ha−1)
Soil C stock used in
calculations (t ha−1)
Climate region or
reference used
UpperLower[D]
ArgentinaAtlantic Forest883435Gasparri et al. (2008)
Chaco701931Gasparri et al. (2008)
Yungas1096565Gasparri et al. (2008)
BrazilCerrado883487Corbeels et al. (2006)
Transition and rainforest1304487Tropical wet*
ParaguayAtlantic Forest883461Warm temp moist*
Chaco701945Warm temp dry*
Table 5.   Estimates of aboveground carbon in five different habitat types in Argentina, Brazil and Paraguay
CountryHabitatAboveground biomass range
from IPCC (2006) (DM t ha−1)
Aboveground biomass
used in calculations (DM t ha−1)
Biomass to
C factor
Aboveground C used in
calculations (t ha−1)
Ecological zone or
reference used
UpperLower[E]
ArgentinaAtlantic Forest2802103030.50152Gasparri et al. (2008)
Chaco280210960.5048Gasparri et al. (2008)
Yungas2802102330.50117Gasparri et al. (2008)
BrazilCerrado9040370.4717Corbeels et al. (2006)
Transition and rainforest4001203000.47141Tropical rainforest
ParaguayAtlantic Forest2802102200.47103Subtropical humid forest*
Chaco4102002100.4799Subtropical dry forest*
Table 6.   Estimates of root carbon in five different habitat types in Argentina, Brazil and Paraguay, estimated from aboveground carbon and the ratio of belowground biomass to aboveground biomass (R)
CountryHabitatR from IPCC, 2006 (t belowground
biomass : t aboveground)
R-value used in
calculations
Root carbon values used
in calculations* (t ha−1)
Ecological zone or
reference used
UpperLower[R][F]
  • *

    Root C estimated from R× aboveground biomass carbon values (E, Table 5), where F=E×R.

  • Classifications from IPCC (2006).

ArgentinaAtlantic Forest0.240.200.2030.0Gasparri et al. (2008)
Chaco0.240.200.2311.0Gasparri et al. (2008)
Yungas0.240.200.1922.1Gasparri et al. (2008)
BrazilCerrado0.400.400.407.0Corbeels et al. (2006)
Transition and rainforest0.370.370.3752.2Tropical rainforest
ParaguayAtlantic Forest0.330.220.2424.8Subtropical humid forest
Chaco0.280.270.2827.1Subtropical dry forest
Table 7.   Estimated total carbon losses following change in land use of five different habitat types in Argentina, Brazil and Paraguay
CountryRegionSoil C lost
over 20 years
(t ha−1)
Total C lost from soil+
root+aboveground
biomass (t ha−1)
Total emissions
(t CO2e ha−1)
Emissions annualized
over 20 years
(t CO2e ha−1)
[G]*[H][I][J]§
  1. Calculations :
    * Estimated from soil C stock (D, Table 4) and the stock change factor (SCF; 0.48 for moist wet soils and 0.58 for dry soils), where G=D× (1−SCF).
    †Estimated from sum of soil, root and aboveground C stock, where H=E+F+G.
    ‡Conversion of C stocks lost, to emissions as CO2e, where I=H× 3.67.
    §Annualized emissions estimated as J=I/20.

ArgentinaAtlantic Forest18.220073336.7
Chaco13.07226413.2
Yungas33.817263231.6
BrazilCerrado45.27025512.8
Transition and rainforest45.223887443.7
ParaguayAtlantic Forest31.716058629.3
Chaco18.714553026.5
Table 8.   Estimated total carbon losses (as CO2 equivalents) associated with land-use change expressed per tonne of soya, from Argentina, Brazil and Paraguay, for whole soya bean equivalents (WSBe) imported into the EU27
CountryHabitat(a) Emissions allocated
to habitat type within
country (t CO2e ha−1)
(b) Total emissions
allocated to country
of origin (t CO2e ha−1)
Emissions per tonne
of soya allocated to
country of origin
(t CO2e t WSBe−1)
Emissions per tonne of soya
imported into EU27 allocated
to individual countries
(t CO2e t WSBe−1)
Total emissions per tonne
of soya imported into
EU27 from S. America
(t CO2e t WSBe−1)
[K]*[L][M][N][O]
  1. Calculations :
    * Estimated from emissions annualized over 20 years (J, Table 7) and proportion of habitat deforested in each country (C, Table 3) where K=J×C.
    †Estimated from total emissions allocated to country (L) and soya yields (B, Table 2) where M=L/B.
    ‡Estimated from total emissions per tonne of soya (M) and proportion of supply by each country exporting into the EU27 (A, Table 1) reduced by a factor of 0.2, accounting for soya needed for supply of oil to existing markets, where N=[(M×A)Argentina+(M×A)Brazil+(M×A)Paraguay] × 0.8.

ArgentinaAtlantic Forest2.115.25.121.724.62
Chaco12.0    
Yungas1.2    
BrazilCerrado9.321.27.542.72 
Transition and rainforest11.9    
ParaguayAtlantic Forest15.428.011.610.19 
Chaco12.6    

Before these carbon emissions associated with soya can be attributed as a credit to DDGS, the proportion of SBM which will be substituted by W-DDGS across the European livestock industry needs to be estimated. In practice, the amount of soya which is displaced depends on the other protein-rich ingredients available for ration formulation, their relative prices, the composition of up to 20 nutrients (including the different fiber fractions and essential amino acids), and limits set on maximum inclusion for W-DDGS and other ingredients in diets for the different livestock types. Özdemir et al. (2009) stated that DDGS will principally replace cereals and not oilseed meals because of the low protein content of DDGS. However, Özdemir et al. appear to classify DDGS alongside low protein (60–150 g kg−1) feeding stuffs, whereas the protein content of DDGS is nearer 350 g kg−1, and Lywood et al. (2009) show that the digestible crude protein content of DDGS is slightly higher than that of rapeseed meal across the three major livestock categories. In reality DDGS is likely to replace a mixture of both cereals and oilseeds (Croezen & Brouwer, 2008). In order to model current scenarios, a pragmatic approach was therefore taken by examining the introduction of W-DDGS into different UK livestock diets containing SBM. The inclusion levels under the current scenario in Table 9 may appear conservative. Undoubtedly, the production of W-DDGS on a large scale from an EU bioethanol industry will increase the amount of W-DDGS in the market, but its use will be determined by the availability and price of other feeds against which it competes. In the case of the livestock types and diets above, it would be possible to force more W-DDGS into the diet to simulate this effect, but in reality, the effect of forcing more W-DDGS into the diet would be to lower the SBM : W-DDGS substitution ratio. This is because feed materials other than soya e.g. barley together with other high protein materials such as rapeseed meal, would be forced out of the diet in order to maintain the correct levels of essential nutrients. It is not possible to say precisely in which livestock diets the displaced ingredients will ultimately be found, because of the dynamic effects of the price availabilities of the other feed materials.

Table 9.   Calculation of coproduct credit allocation to DDGS and wheat-bioethanol based on current and future substitution ratios in EU livestock rations
Livestock typeRelative proportion
of each livestock type*
Typical inclusion
level (g kg−1)
Soya meal
substitution (t SBM t DDGS−1)
Soya bean equivalents
substitution
(t WSBe t DDGS−1)
Coproduct credit
Based on mass of
coproduct
(g CO2e kg W-DDGS−1)
Based on mass
of ethanol
(g CO2e kg EtOH−1)
Based on energy
content of ethanol
(g CO2e MJ EtOH−1)
[P][Q][S][T]§[U]
  1. Calculations :
    * For current scenario, proportions are estimated from livestock units in EU27 (Source: Eurostat), and for future scenarios, proportions are estimated from current manufactured compound feed volumes (Source: FEFAC). In both cases, proportions are expressed relative to 1.00 for poultry.
    †Conversion of substitution of SBM (P) to substitution of WSBe, where Q=P/0.8.
    ‡Based on substitution of WSBe (Q) and emissions per tonne of WSBe imported into EU27 from S. America (O, Table 8), where S=O×Q× 1000.
    §Based on ratio of DDGS : ethanol of 1.14 (JRC, 2008), where T=S× 1.14.
    ¶Based on energy content of ethanol of 26.72 MJ kg−1, where U=T/26.72.
    EU, European Union; DDGS, distillers dried grains with solubles; SBM, soya bean meal; WSBe, whole soya bean equivalents.

Current scenario
 Pig2.05610.350.442022230686
 Poultry1.00650.560.7032363689138
 Ruminant4.34450.260.331502171364
 Average  0.330.411912218082
Future high usage scenario
 Pig1.111500.590.7434093887145
 Poultry1.001000.580.7333523821143
 Ruminant0.814000.620.7835834084153
 Average  0.600.7634953984149

By combining these inclusion limits and substitution ratios with the production volumes of compound feeds per livestock type in the EU27, it was possible to calculate a weighted average substitution ratio across livestock types (0.33; Table 9). Combining this average substitution ratio with the carbon cost per tonne of WSBe from Table 8, it was possible to estimate a coproduct credit of 1912 g CO2e kg W-DDGS−1, or 82 g CO2e MJ bioethanol−1 (Table 9). This is a factor of 1.8 greater than the GHG savings from wheat bioethanol of 45 g CO2e MJ−1 stated in the RED (EU, 2009). It should be noted that the aim of the present analysis was not to revisit the estimation of GHG emissions associated with the whole supply chain (growing, transport, production and processing components) for bioethanol from wheat and sugar beet, as these have been adequately described elsewhere (JRC, 2008) and are now incorporated into sustainability and reporting guidelines (EU, 2009). Rather the aim was to quantify a coproduct credit based on substitution of soya with W-DDGS which is additional to the net savings used as current default values for wheat bioethanol.

The inclusion limits for all livestock categories are expected to increase in the future, as larger volumes of more consistent W-DDGS are produced by the biofuels industry, and confidence is gained from further nutritional research. However at the present time, inexperience and variability of DDGS from traditional (nonbiofuel) coproducts has tended to limit inclusion levels. In order to consider future potential, an additional scenario was studied using the same methodology as above, but using the higher substitution ratios (Lywood et al., 2009), an assumption of greater usage of DDGS in nonruminant livestock diets (based on the current proportions of compound feed manufactured in Europe) and acceptance of higher inclusion levels of W-DDGS in the diet. Such a scenario assumes that as the biofuels industry develops in Europe, and more nutritional research is carried out on bioethanol coproducts, this will give greater confidence to the animal feed industry, and more W-DDGS will be incorporated into rations. Using this future scenario gave a coproduct credit of 3495 g CO2e kg W-DDGS−1 (Table 9).

An important observation resulting from the use of LCRF is that other feed ingredients and in particular cereal grains and milling coproducts and some oilseeds, are also substituted from the diet by DDGS in addition to SBM (Table S1). Similarly, SBP can also be considered to displace wheat from certain animal diets (Lywood et al., 2009). Thus the use of bioethanol coproducts can release land which might otherwise be used to produce livestock feeds, thereby reducing the impact of ILUC. Table 10 shows different scenarios for net land area required to grow 1 ha of bioethanol feedstock crops (wheat and sugar beet) in Europe, taking account of land made available for other uses because of the use of the major bioethanol coproducts, W-DDGS and SBP. It can be seen that using the average relative proportions of wheat and sugar beet found in the rotation (0.94, 0.06; FAO Statistics 2006–2008 for EU27) the bioethanol coproducts from 1 ha of land will displace cereals from livestock diets releasing an additional 0.18 ha of land growing wheat, together with 0.42 ha of soya land displaced in South America giving a net land area requirement of 0.40 ha per ha of bioethanol feedstock crops. Taking into account the typical GHG savings from wheat and sugar beet bioethanol (JRC, 2008; EU, 2009) and adding to these the coproduct credit estimated in Table 9, conversion of the feedstock from 1 ha of land in the EU would be associated with GHG savings of around 115 g CO2e MJ−1 bioethanol produced.

Table 10.   Estimation of net land area requirement and ethanol output from 1 ha of land in Europe, growing variable proportions of wheat and sugar beet
Sugar beet
cultivation
in EU (ha)
Wheat
cultivation
in EU (ha)
Land area equivalent of
wheat in EU displaced
by SBP and DDGS* (ha)
Land area equivalent of
soya displaced by DDGS
in S. America (ha)
Net wheat
area (ha)
Net land
area (ha)
Total ethanol
output (GJ ha−1)
Total ethanol
output§ (L ha−1)
Total CO2 abated
(g CO2e MJ EtOH−1)
[V][W][X][Y]WX(V+W)−XY[Z] [AA]
  1. Calculations :
    * Estimated using land areas displaced by various coproducts (ha ha−1) as follows: Wheat/DDGS, 0.13 (Lywood et al., 2009) and wheat/SBP, 1.00 based on a coproduct yield of 6.08 t SBP ha−1, wheat yield of 5.20 t ha−1 (EU27 yields 2006–2008; FAOStat) and a substitution ratio of 0.85 t SBP t wheat−1 [average of 0.78 from Lywood et al. (2009) and 0.91 from JRC (2008)].
    †Estimated using land area for soya displaced by DDGS (ha ha−1) as follows: SBM/DDGS, 0.44 based on substitution ratio of 0.59 t SBM t DDGS−1, coproduct yield of DDGS of 1.72 t ha−1 [based on Lywood et al. (2009) adjusted for EU27 yields 2006–2008; FAOStat] and SBM yield of 2.29 t ha−1.
    ‡Estimated from ethanol productivity from sugar beet of 131.0 GJ ha−1 and wheat of 44.7 GJ ha−1 [Lywood et al. (2009) adjusted for EU27 yields 2006–2008; FAOStat].
    §Estimated using energy density of ethanol of 21.1 MJ L−1.
    ¶Estimated from proportion (p) of total ethanol output (Z) from either sugar beet (pSB) or wheat (pW) multiplied by the respective typical GHG saving of 52 and 45 g CO2e MJ−1 (EU, 2009) plus the average coproduct credit based on soya substitution (82 g CO2e MJ−1; Table 9) allocated to the wheat, where AA=(pSB × 52)+(pW × 45)+(pW × 82).
    EU, European Union; DDGS, distillers dried grains with solubles; SBP, sugar beet pulp; SBM, soya bean meal; GHG, greenhouse gas.

0.060.940.190.420.750.4049.92363115
0.100.900.220.400.680.3853.32527108
0.250.750.360.330.390.3166.3314190

If the higher yields representative of NW Europe are used (SBP 6.47 t ha−1 and wheat 7.75 t ha−1), the land area equivalent displaced by soya increases to 0.62 ha, and the net land area is reduced to 0.22 ha. With a higher proportion of sugar beet in the rotation, the net land area decreases as the equivalent land area of wheat diminishes (and at the same time the amount of ethanol produced increases). The intermediate scenario in Table 10 (relative areas; 0.90, 0.10) represents the rotational situation in Germany based on FAO Statistics. For a high sugar beet frequency scenario, the inclusion of 0.25 ha sugar beet in the rotation potentially enables a higher ethanol yield per hectare (3141 L ha−1) than growing 1 ha of wheat alone (2118 L ha−1) and results in a reduction in the net area of wheat to 0.40 ha and the net effective area of total land to 0.32 ha (Table 10). It is unlikely that sugar beet would be grown at levels higher than 0.25 in the rotation for agronomic reasons.

This analysis illustrates that when taking into account the substitution of other feed ingredients by biofuel coproducts, bioethanol produced from wheat and sugar beet in the EU27 has the ability to make more efficient use of existing arable land within Europe, while at the same time reducing the pressure on high carbon stock land outside Europe and delivering significant GHG savings from substitution of fossil fuels.

Acknowledgements

The authors would like to thank Mick Hazzledine of Premier Nutrition for running the least cost ration formulations, and Santiago Véron and José Volante of INTA for information on soya bean rotations in Argentina.

Ancillary