The production of transport biofuels can be accompanied by changes in carbon se­questration. Firstly, there can be changes in the carbon content of ecosystems. There may be both losses from, and increases of, C in the ecosystem, which in turn will change atmospheric CO2 concentrations. In early life cycle assessments, quantifica­tion of such changes was largely neglected (with some exceptions, such as Reijnders and Huijbregts 2003, Delucchi 2005; Kim and Dale 2005 and Cowie et al. 2006). However, by now, changes in C sequestration by ecosystems are increasingly rec­ognized as a major determinant of net seed-to-wheel greenhouse gas emissions (e. g. Fritsche 2007; Danielsen et al. 2008; Fargione et al. 2008; Gibbs et al. 2008; Ne — chodom et al. 2008; Searchinger et al. 2008; Wicke et al. 2009). In a number of cases, the link between change in C sequestration and expansion of biofuel produc­tion is direct. For instance, because forest is cleared or degraded land is planted for oil palm plantations serving the biodiesel market. In these cases, there is a large ef­fect on C sequestration (Danielsen et al. 2008; Fargione et al. 2008). There are also cases in which land use change has more limited direct effects on C stocks — for instance, cases where oil palms have replaced other plantation crops, such as rubber or coconut, that gave lower revenues (Tan et al. 2009).

There may also be indirect effects of expanding biofuel production on land use. These follow from the relative inelasticity of demand for food. When land used for food or feed (fodder) production is diverted to use for biofuel production, one may expect that food or feed production to a very large extent moves elsewhere (Searchinger et al. 2008). Some examples may illustrate this. Currently, in Brazil, there is substantial conversion of pasture to arable land to grow soybeans. Conver­sion of pasture to arable land tends to lower carbon in the ecosystem. But this is not the ‘whole story’. Farmers that used the pasture may move to new pastures or may import fodder from elsewhere, for which deforestation may be necessary (Nepstad et al. 2008). Also, the expansion of soybean cropping in Brazil is partly linked to the expansion of corn-for-biofuel production in the United States, which has dis­placed US soybean production (Nepstad et al. 2008; Scharlemann and Laurance 2008). Tilman et al. (2006) have proposed the use of native prairie species as a ba­sis for transport biofuel production from the US prairies, but to the extent that such production replaces current grazing by livestock, the latter has to be accommodated somewhere else. A last example refers to European wheat. In 2007, the price thereof was much elevated, in part due to the use of wheat starch for ethanol biofuel pro­duction. This led to a massive rise of corn imports with Brazil as a major supplier, which in turn was conducive to change of Brazilian land with natural vegetation into cropland.

There may also be other changes in carbon sequestration linked to biofuel life cycles. For instance, both in the case of electricity production of biomass and the production of synthesis gas from biomass, CO2 may be captured and stored in soils (e. g. aquifers or abandoned gas fields). In the case of electricity production, de­pending on the technology used, life cycle emissions of greenhouse gases linked to electricity production may be reduced by 75-84%, though other emissions such as those of eutrophying and acidifying substances may be increased (Odeh and Cock — erill 2008). There is currently only limited application of such capture and storage (abbreviated CCS), but this may change in the future (Odeh and Cockerill 2008; Gibbins and Chalmers 2008). Also, in generating transport biofuels from biomass by pyrolysis, significant amounts of charcoal, also called ‘biochar’, may be gener­ated (Demirba§ 2001), which in turn may be added to soils. This approach is cur­rently not practiced, but again, this may change in the future (Marris 2006; Renner

2007) .

Changes in carbon sequestration in ecosystems are currently common in trans­port biofuel production. Clearing forests to allow directly or indirectly for culti­vation of transport biofuel crops corresponds with substantial emissions of green­house gases (Fearnside and Laurance 2004; Righelato and Spracklen 2007; Far — gione et al. 2008; Reijnders and Huijbregts 2008a; Searchinger et al. 2008). This follows to a large extent from the large changes in aboveground and soil carbon stocks. For instance, estimates of aboveground C stocks in rainforests range from approximately 130-270Mg C ha-1 (Danielsen et al. 2008; Fargione et al. 2008; Reijnders and Huijbregts 2008a), and similar estimates for aboveground C stocks in temperate forests range from approximately 100-160 Mgha-1 (Searchinger et al.

2008) . Clearing other types of natural vegetation for transport biofuel production may lead to a lower but still substantial desequestration of C. For instance, clearing Cerrado savannah for soybean biodiesel production leads to an average deseques­tration of about 23 Mg C ha-1 (Reijnders and Huijbregts 2008b), whereas losses of C stocks in soil due to conversion of US grasslands into cropland for biofuels have been estimated at approximately 17-40 Mg C ha-1 (Fargione et al. 2008). On the other hand, Germer and Sauerborn (2007) have suggested that planting oil palms on degraded grassland may lead to a substantial accumulation of aboveground and soil organic carbon, estimated at 135 Mg C ha-1 over a period of 25 years. Ne — chodom et al. (2008) have suggested very favourable life cycle assessments regard­ing biomass from forest remediation in California, as it is assumed that without using such biomass, forest fire risk would be much higher.

If one takes account of changes in carbon sequestration due to ecosystem changes in life cycle environmental impact assessments, the distribution of changes in the carbon content linked to land use change over the subsequent period should be established, as pointed out in Sect. 4.2. The Intergovernmental Panel on Climate Change (IPCC 2006) has suggested the use of a 20-year period for this purpose, but calculations also have been made for periods of up to 100 years (Reijnders and Huijbregts 2008a; Wicke et al. 2009).

Another way to approach this matter is to balance the net reduction of C emis­sions from the use of biofuels against the ‘biofuel carbon debt’ due to C losses from ecosystems. This gives rise to a number of years to pay off the carbon debt. An ex­ample thereof for palm oil, reflecting a large effect on C stocks, is given in Fig. 4.1.

Direct changes in land use linked to expanded transport biofuel production are relatively easy to include in life cycle assessments. Including indirect changes is more difficult. A first possibility to take account of changes in C sequestration due to indirect changes in land use has been suggested by Fritsche (2007). Fritsche pro­posed that for such indirect effects, there should be a ‘risk adder’ or ‘iLUC’, re­flecting the risk that there may be clearance of natural vegetation or other forms of land use change that affect C sequestration and the net emission of CO2 (usually distributed over a 20-year period in line with recommendations of the Intergovern­mental Panel on Climate Change). The impact of emissions associated with differ­ent assumptions as to indirect effects on land use change (different risk adders or iLUCs) may be considerable. Illustrations are given by Table 4.1.

A more sophisticated approach uses modelling to estimate land use and land use change induced by expanding transport biofuel production. A proposal for such modelling has been published by Kkrverpris et al. (2008). Searchinger et al. (2008) have applied this type of modelling using a representation of outcomes in terms of carbon debt. Searchinger et al. (2008) evaluated expanding corn cropping to achieve the US government goals set for the transport biofuel supply in 2016. They used a model to estimate worldwide land use changes following from this development, which, they estimate, will divert 12.8 million ha of US cropland to transport bio­fuel production. Searchinger et al. (2008) estimate that this will bring 10.8 million

ha of additional land into cultivation. Searchinger et al. (2008) have calculated that over a 30-year period, including land use change linked to expanding US corn — based ethanol production will add about 93% to greenhouse gas emissions, if com­pared with fossil gasoline. The payback time for the carbon debt caused by land use change by corn ethanol is estimated by Searchinger et al. (2008) at 163 years. Mod­elling such as performed by Searchinger et al. (2008) is dependent on assumptions. Important among these are assumptions regarding the future productivity of land. Searchinger et al. (2008) assumed per hectare an average worldwide corn yield in­crease of 11.5% between 2007 and 2016. A lower increase would lead to a larger indirect effect on land use, a higher increase to a lower indirect effect. For instance, an additional 20% increase in yield per hectare would decrease the carbon debt from 163 to 133 years, whereas a yield increase of 1.5% between 2007 and 2016 would increase the carbon debt to 183 years (Searchinger et al. 2008).

Thus, land use change may have a very large impact on seed-to-wheel green­house gas emissions. Biofuel crops may, furthermore, differ in their consequences for carbon stocks in the soil on which they are grown after land use change. An­nual cropping of European arable land has been associated with losses of, on aver­age, 0.84Mgha~1year~1 soil carbon from the upper soil layer (Vleeshouwers and Verhagen 2002), and this has implications for the net greenhouse gas emissions as­sociated with biofuel crops such as rapeseed and wheat (Reijnders and Huijbregts 2007, 2008b). Net losses of soil carbon from arable soils under annual crops are not restricted to Europe. In Eastern Canada, arable land on average loses 0.07 Mg C ha-1year-1 to the atmosphere (Gregorich et al. 2005). In China, the estimated loss on average is 0.81 Mg C ha-1year-1 (Tang et al. 2006). The loss of soil or­ganic carbon in China parallels the removal of approximately 300 million tons of straw (Wright 2006). From Nepal, losses of soil carbon have been noted in a va­riety of cropping systems, with, for instance, losses in maize/millet cropping sys­

tems ranging between 0.11 and 0.23 Mg C ha-1year-1 (Matthews and Pilbeam 2005; Shrestha et al. 2006). In a set of 14 cropped Brazilian soils sampled annu­ally, losses of soil organic carbon were on average 0.15 Mgha-1year-1 (Zinn et al.

2005) , whereas for soybean-based crop rotation in the Brazilian Cerrado region, losses were reported to be between 0.5 and 1.5 Mg C year-1 ha-1 (Jantalia et al. 2007). In Africa, losses of carbon from soil are expected as little or no agricul­tural residues are returned to soils in many cropping systems (Syers et al. 1997). In the case of semi-arid Sudan, an annual loss from cropland during the twentieth century of, on average, 0.29 Mg C ha-1 year-1 has been found (Ardo and Olsson 2003). Long-term experiments in East Africa have suggested that losses of 0.69 Mg C ha-1 year-1 are common (Nandwa 2001). Over relatively short and recent time spans, measured C losses in the southern Ethiopian highlands amounted to approx­imately 0.85-1.75Mg C ha-1year-1 (Lemenih and Itanna 2004), and in Western Burkina Faso to approximately 0.31 Mg C ha-1 year-1 (Ouattara et al. 2006).

From peaty arable soils, losses may be much higher. They are especially high when there is deep drainage and intensive mechanical soil disturbance (Freibauer et al. 2004). Net carbon losses varying from 6 Mg in northern Norway up to 15 Mg C ha-1 year-1 in the tropics have been reported (Granlund et al. 2006; Reijnders and Huijbregts 2007, 2008a).

Mechanical tillage and deep ploughing tend to favour net losses of soil carbon stocks (Fontaine et al. 2007). No-till practices, ceteris paribus, lead to higher lev­els of soil carbon than tillage (Fontaine et al. 2007), so does the use of cover crops (Pretty et al. 2002). Also, returning harvest residues to soil is conducive to carbon sequestration in soils (Reijnders and Huijbregts 2007; Lal 2008). Intemperate cli­mates, soil tillage can be combined with a stable level of soil carbon, when fresh inputs of C (e. g. manure, crop residues) are large enough (Reijneveld et al. 2009).

Whereas losses of soil carbon are not uncommon under annual crops, net carbon sequestration can occur when perennials or multiannual crops are grown.

Growing switchgrass that may serve as a cellulosic feedstock for transport bio­fuel production is associated with the accumulation of soil carbon in the upper soil layer (Ma et al. 2000; Lal 2008). Carbon accumulation rates depend on cli­mate and soil and time period chosen. McLaughlin and Adams Kszos (2005) found in the mid-Atlantic region of the USA over a 6-year period an accumulation of 1.2-1.6Mg C ha-1 year-1. A simulation study considering a 30-year period of switchgrass cropping in the Eastern USA suggests a carbon accumulation rate of 0.53 Mgha-1 year-1 (McLaughlin et al. 2002). Long-term trials with ley grass in Sweden suggested an annual accumulation of 1.0-1.3Mg C ha-1 year-1 over a 30-year period (Boijesson 1999). Bjoresson (1999) has suggested that a switch from annual crops to perennial biofuel crops may be associated with a gain of about 0.5Mg C ha-1 year-1 in Swedish mineral soils. However, net sequestration does not always occur under perennial crops. In a grassland with Miscanthus sinensis in Japan, harvested yearly, reductions of carbon stocks were observed in the order of 0.56-1 Mgha-1year-1 (Yazaki et al. 2004). Finally, soil organic carbon levels may change when the climate changes. The effects of climatic change are complex and may be different dependent on geography and the (agro)ecosystem (Luo 2007)

For Europe, Vleeshouwers and Verhagen (2002) estimated that an increase in aver­age temperature of 1 °C may be associated with an average net loss of soil organic carbon in arable soils of about 0.04 Mgha^year-1.