A last important non-energy natural resource for biomass production is water. The potential for biomass production on suitable soils is strongly influenced by fresh­water availability (Tuskan 1998; Rabbinge and van Diepen 2000; Ryan et al. 2002; Kahle et al. 2002; Deckmyn et al. 2004), and in semi-arid areas, it may well be the main limiting factor (Ong and Leakey 1999). Of the yearly renewable stock of fresh­water, approximately 30-50% is currently used in the economy (Postel et al. 1996; Raskin et al. 1997; Rockstrom 2003). Worldwide, growing crops is the main con­sumer of freshwater (Pimentel et al. 1997a). Even if growing of rain-fed crops is not considered to be use of water, growing crops accounts for at least 66% of current wa­ter use (Gleick 2000; Wallace 2002; van Dijk and Keenan 2007; Gordon et al. 2008; Zimmerman et al. 2008). Water consumption is projected to go up considerably, grossly paralleling growth of the world population and economic growth, much of the increase being associated with the expected increase in food and feed production (Gleick 2000; Berndes 2002; Swedish Environmental Advisory Council 2007).

Geographically speaking, water stocks are unevenly distributed. Currently, struc­tural water shortages affect about 300-400 million people mainly in Africa and Asia in a band from China to North Africa (Gleick 2000; Wallace 2002). As much growth of the world population is expected in the same area, this does not bode well (Gleick 2000). But elsewhere there are problems, too. For instance, in the USA, roughly 20% of the irrigated area is supplied by groundwater pumped in excess of recharge (Tilman et al. 2002). Also, there are areas where climate change may lead to a struc­tural decrease in rainfall, such as the south and east of Africa (Funk et al. 2008).

Large additional water requirements follow from expected population growth and changes in dietary habits, especially the increased consumption of animal pro­duce (Rockstrom 2003; Falkenmark and Lannerstad 2005; Liu and Savenije 2008). The estimated size of the additional water requirement depends on the assumptions made. Assuming an intake of 3,000kcal person-1 day-1 and using available pre­dictions about world population growth, Rockstrom (2003) estimated an additional water requirement of 3,800km3year-1 in 2025 and of 5,800km3year-1 in 2050. Assuming business as usual, which does not include a substantial production of modern biomass-for-energy, it has been suggested that shortages of freshwater may well become a fact of life for up to 2.5-6.5 billion people by 2050 (World Water Council 2000; Wallace 2002). Expanding modern biomass-for-energy production may substantially exacerbate this trend.

The National Research Council of the USA has warned about more local water shortages due to the expanded production of corn for supplying ethanol (NRC 2007). This problem is especially pressing in the Western USA, where climate change threatens to exacerbate water shortages anyway (Barnett et al. 2008) and where there is already a substantial groundwater overdraft (Falkenmark and Lannerstad 2005). Excess water withdrawals may also be a problem in Brazilian sugar cane expansions (Dias de Oliveira et al. 2005), where hydrological flows are already much impacted by sugar cane cultivation (Gordon et al. 2008). India and China are expected to have increasing shortfalls in meeting national food demand due to water scarcity, necessitating increasing food imports (Falkenmark and Lannerstad 2005; Liu and Savenije 2008), and this leaves little scope for cropping feedstocks for transport biofuels.

In Hungary, there has been evidence that the use of cornstalks as biofuel wors­ened a drought by removing cover and reducing soil organic matter (Engelhaupt

2007) . Berndes (2002) has pointed out that in the case of large-scale bioenergy pro­duction, a European country such as Poland may face absolute water scarcity. Com­petition between fast-growing Eucalyptus and food for water resources in Ethiopia has led to government restrictions on the former (Jagger and Pender 2003). Else­where, competition between short rotation trees and food crops for scarce water re­sources has also been noted (Ong and Leakey 1999; Sanchez 1999; Berndes 2002; Ong et al. 2002; van Dijk and Keenan 2007). On the other hand, when growing feed­stocks are rain fed, there are some places subject to waterlogging and/or secondary salinization, where large uptakes of water by energy crops may be considered a ben­efit (Morris and Collopy 1999; Mahmood et al. 2001; Ryan et al. 2002; Palm et al.

2007) , though it should be noted that the use of Eucalyptus globulus for this pur­pose in Western Australia has turned out to be disappointing (Sudmeyer and Simons

2008) .

Data about seed-to-wheel water use are very patchy (Royal Society 2008). Data about water inputs into feedstock processing, especially, are largely lacking. Never­theless, it is clear that biomass processing can lead to substantial water consumption. A case in point is the factories that produce ethanol from sugar cane in the relatively dry winter period in Brazil (Smeets et al. 2008). Also, in generating electricity from biomass in power plants, there is often water cooling that consumes large amounts of water (King and Webber 2008). However, it would seem likely that most of the seed-to-wheel water inputs concern the growing of feedstocks, in line with agricul­ture being the dominant consumer of water in the economy.

In view of increasing water scarcity, the water efficiency of transport biofuels requires considerable attention. As to water use in biomass processing, there is, for instance, scope for improving the water efficiency of Brazilian factories for the con­version of sugar cane into ethanol by increasing the water recycling rate and by replacing cane washing with dry cleaning of cane (Smeets et al. 2008). Similarly, consumption of cooling water by power plants processing biomass can be much reduced by, for example, switching to closed loop and air cooling systems (King and Webber 2008). The water efficiency in the growth of feedstocks is especially important (Hanegraaf et al. 1998; Rytter 2005; Almeida et al. 2007; Rockstrom et al. 2007). On average, the water efficiency of C4 crop plants appears to be bet­ter than the water efficiency of C3 crop plants (Cernusak et al. 2007; Heaton et al.

2008) . Growing 1 kg of aboveground (dry) biomass of the C3 crop wheat on aver­age requires about 0.55 m3 of water (calculated from data in Zwart and Bastiaanssen

2004) , whereas 1 kg of dry biomass for the C4 crops sugar cane or Miscanthus prob­ably needs less than 0.2 m3 (Rockstrom et al. 1999; Beale et al. 1999).

The water efficiency of soybeans and other annual oil crops is much poorer than the water efficiency of wheat (Liu and Savenije 2008). The microalga Tetraselmis suecica, which is considered for biofuel production, has been estimated to require 0.31-0.57 m3 water kg-1 dry weight biomass (Dismukes et al. 2008). As transport biofuels are the product of biomass processing, such biofuels will need a larger amount of water per kilogram of biofuel than per kilogram of biomass. In the case of ethanol based on cornstarch, while assuming an average water productivity of corn (Zwart and Bastiaanssen 2004) and a 0.294 kg ethanol yield from 1 kg of corn kernels (Kheshgi et al. 2000), for each kilogram of bioethanol, 1.9 m3 of water is needed in cultivation, when all water use is allocated to ethanol production. In the case of wheat, the estimated water input in cultivation is about 3.1m3 water per kilogram of wheat-starch-based ethanol, when all water use is allocated to ethanol production. Water efficiency may change due to climate change linked to increas­ing greenhouse gas concentrations in the atmosphere. There is only very limited study thereof. A study on water use efficiency by a plantation producing willow as a feedstock for biofuels suggests that climate change may lead to a decrease in water efficiency (Tricker et al. 2009).

The difference in water efficiency between C3 and C4 crop plants is not the only noteworthy difference relevant to feedstock production. When the aim is to produce ethanol from carbohydrates, sugar crops tend, on average, to do so more efficiently than starch crops, whereas starchy root crops tend to be more water efficient than cereals (Rockstrom et al. 2007). In dry areas, the water efficiency of producing sugar by sugar beet may be better than that of sugar cane (Rytter 2005). Also, there may be ‘drought-resistant’ varieties of crops that would do better under such circumstances than other varieties (Rytter 2005). There is large variability in water efficiency be­tween woody plants that may be used for biofuel production (Cernusak et al. 2007). Growing 1 kg of aboveground dry oil palm biomass requires approximately 1.5 m3 of water (Rockstrom et al. 1999), and growing 1 kg of willow biomass (dry weight) requires 0.2-0.3 m3 of water (Linderson et al. 2007). The olive tree is well adapted to drought stress (Sofo et al. 2008), whereas current varieties of Eucalyptus are less so (Dale and Dieters 2007). Eucalyptus biomass on rain-fed soil would require about 0.46 m3 of water kg-1 dry biomass (calculated from data in Stape et al. 2008 and Kheshgi et al. 2000). So, choice of feedstock may have a large impact on water consumption.

Soil water conservation is also dependent on tillage (and soil organic carbon lev­els) with conservation tillage conducive to water conservation (Wallace 2002; Rock­strom 2003). Mulching and intercropping to maximize canopy cover also help in in­creasing water productivity or crop yield for a specified input of water (Rockstrom 2003). So, for instance, on the Loess Plateau in China, no tillage with mulching does better in soil water conservation than conventional tillage, leading to substantially improved water use efficiency of winter wheat (Su et al. 2007). It has furthermore been shown that there are irrigation techniques that lead to higher water efficiencies than traditional furrow irrigation. These include variable application (Al-Kufaishi et al. 2006), subsoil irrigation and drip irrigation. Improvements in water efficiency may be considerable. Subsoil irrigation, with controlled diffusion of irrigation wa­ter from a clay pipe, has been shown to increase grain yield of Iranian winter wheat per kilogram of water by at least a factor of 2.5 (Banedjschafie et al. 2008). And

in Uzbek cotton growing, which might supply cotton oil for biodiesel production, the improvement in yield by switching from furrow irrigation to drip irrigation was 35% to somewhat more than 100% (Ibragimov et al. 2007).

Apart from the quantitative impact on the availability of water, water quality may also be influenced by growing biofuels (NRC 2007). The expansion of corn production for the supply of ethanol is a case in point, as it leads to an increased nutrient and pesticide load of water resources (NRC 2007). And water discharges for algal biofuel production in open ponds may also be problematic in view of high pH, nutrient and/or salt levels present in pond water (Dismukes et al. 2008).