Transport biofuels replace fossil fuels. But because, as pointed out in Chap. 2, much production in current societies is dependent on fossil fuels, it will come as no sur­prise that burning fossil fuels is often an important contributor to the greenhouse gas emissions associated with biofuels. N fertilizers are often made on the basis of natu­ral gas; tractors and transport are often powered by fossil fuels based on mineral oil. Factories, involved in converting biomass into fuels fit for powering vehicles, are more mixed in their fuel use. There are production facilities doing without burning fossil fuels. In Brazil, factories converting sugar into ethanol are often powered by harvest residues of sugar cane (Macedo et al. 2008). In Sweden, biofuel production in factories tends to use wood chips from forest logging residues (Borjesson and Mattiasson 2008). But, for example, in France and Germany, factories producing bioethanol or biodiesel are usually powered with fossil fuels (Reijnders and Huij — bregts 2007,2008b). Life cycle assessments of biofuels are characterized by a long­standing interest in the CO2 emissions linked to cumulative fossil energy demand (e. g. Pimentel et al. 1973; Weisz and Marshall 1979), and by now the standard is that they include the CO2 emission linked to burning fossil fuels to power the bio­fuel life cycle, though there are still exceptions to this rule (e. g. Chisti 2007, 2008; Dismukes et al. 2008).

There may also be non-CO2 carbonaceous emissions linked to fossil fuel use. For instance, there may be leakage of CH4 (methane) during transport and use of natural gas, and, if burning is not optimized, there may be emission of hydrocarbons and carbon monoxide (CO). The non-CO2 carbonaceous gases on a molecule-for — molecule basis tend to have a greater greenhouse effect than CO2. The non-CO2 carbonaceous greenhouse gases are often neglected in life cycle assessments, which will lead to an underestimate of the greenhouse effect of actual emissions. However, in advanced industrial economies, the error linked to this underestimation will be small, in the order of a few percent maximally.