Changing precipitation patterns and shortages of surface and groundwater in important cropping regions pose a serious threat to China’s future food security. This paper presents a comprehensive analysis of water used for food production over the period 1998–2010 with a view to identifying pathways for achieving the national target of 580 million tons of grain output by 2020. The analysis was based on modelling of agricultural water use coupled with national and provincial statistics. The present situation was defined by (a) a slow declining trend in national precipitation and internally renewable water resources, (b) 12 out of the 13 so-called breadbasket provinces (which currently produce 74 % of national grain output) already facing water shortages and increasing competition for water from non-agricultural sectors, (c) national crop water productivity (CWP) increases of 19.5 % over the 13 years to 2010, and (d) a widening gap in CWP between breadbasket and non-breadbasket provinces. By 2020 an estimated 510 to 680 km³ of water will be required for food production depending upon future gains in CWP. A concern is that in many of the breadbasket provinces, recent CWP gains have already been substantial and additional large gains may prove difficult especially considering current environmental concerns related to agricultural intensification in China. That said, the historic efficiency gains give reason for optimism provided that there is continued investment in genetic improvement and innovation of farming systems.

The paper discusses the contribution of crop residues (CR) to feed resources in the context of the water productivity of CR in livestock feeding, using India as an example. It is argued that crop residues are already the single most important feed resource in many livestock production systems in developing countries and that increasing their contribution to livestock feeding needs to be linked to improving their fodder quality. Using examples from multi-dimensional crop improvement, it is shown that CR fodder quality of key crops such as sorghum, rice and groundnut can be improved by genetic enhancement without detriment to grain and pod yields. Improving crop residue quality through genetic enhancement, agronomic and management interventions and strategic supplementation could improve water productivity of farms and systems considerably. The draw-backs of CR based feeding regimes are also pointed out, namely that they result in only moderate levels of livestock productivity and produce higher greenhouse gas emissions than are observed under feeding regimes that are based on high quality forages and concentrates. It is argued that feed metabolisable energy (ME) content should be used as an important determinant of livestock productivity; water requirement for feed and fodder production should be related to a unit of feed ME rather than feed bulk. The paper also revisits data from the International Water Management Institute (IWMI) work on livestock-water productivity in the Indian state of Gujarat, showing that water input per unit ME can vary several-fold in the same feed depending on where the feed is produced. Thus, the production of one mega joule of ME from alfalfa required 12.9L of irrigation-derived water in south Gujarat but 50.7L of irrigation-derived water in north Gujarat. Wheat straw in south Gujarat required 20.9L of irrigation-derived water for 1MJME and was in this instance less water use efficient than alfalfa. We conclude that water use efficiency across feed and fodder classes (for example crop residue v. planted forages) and within a feed is highly variable. Feeding recommendations should be made according to specific water use requirement per unit ME in a defined production system.