restore biodiversity, improve soil permeability through root activity; return organic matter to the soil; protect against erosion by rain and wind, and provide protection from direct radiation and warming (Swift and Anderson, 1993; Swift et al., 1996). Natural fallows of this sort are no longer applicable in most places because population pressure is high; consequently shorter and more efficient fallows using leguminous shrubs and trees are being developed (Kwesiga et al., 1999). When soil fertility is severely depleted, some external mineral nutrients (phosphorus, calcium) or micronutrients may be needed to support plant growth and organic matter production.

In many intensive production systems, the efficiency of fertilizer nitrogen use is low and there is significant scope for improvement with better management.

The extent of soil degradation and loss of fertility is much greater in tropical than in temperate areas. Net nutrient balances (kg ha-1 per 30 years) of NPK are respectively: -700, -100, -450 for Africa; and +2000, +700, +1000 for Europe and North America. Low fertilizer recovery efficiency can reduce crop yields and net profits, increase energy consumption and greenhouse gas emissions, and contribute to the degradation of ground and surface waters (Cassman et al., 2003). Among intensive rice systems of South and Southeast Asia, crop nitrogen recovery per unit applied N averages less than 0.3 kg kg-1 with fewer than 20% of farmers achieving 0.5 kg kg-1 (Dobermann and Cassman, 2002). At a global scale, cereal yields and fertilizer N consumption have increased in a near-linear fashion during the past 40 years and are highly correlated. However, large differences exist in historical trends of N fertilizer usage and nitrogen use efficiency (NUE) among regions, countries, and crops. Interventions to increase NUE and reduce N losses to the environment require a combination of improved technologies and carefully crafted local policies that contribute to the adoption of improved N management practices. Examples from several countries show that increases in NUE at rates of 1% yr -1 or more can be achieved if adequate investments are made in research and extension (Dobermann, 2006). Worldwide NUE for cereal production is approximately 33% (Raun and Johnson, 1999). Many systems are grossly overfertilized. Irrigated rice production in China consumes around 7% of the global supply of fertilizer nitrogen. Recent on-farm studies in these systems suggest that maximum rice yields are achieved at N fertility rates of 60-120 kg N ha-1, whereas farmers are fertilizing at 180-240 kg N ha-1 (Peng et al., 2006).

Good soil management enhances soil productivity.

In the tropics, the return of crop residues at a rate of 10-12 tonnes dry matter ha-1 represents an input of 265 kg carbon ha-1 in the upper 10 cm soil layer (Sá et al., 2001ab; Lal, 2004). Given an appropriate C:N ratio, this represents an

increased water holding capacity of 65-90 mm, potentially a 5-12% increase in maize or soybean yield, and increased income of US$40-80 ha-1 (Sisti et al., 2004; Diekow et al., 2005). Soil carbon and yields can be increased on degraded soils through conservation agriculture (e.g., no-till), agroforestry, fallows with N-fixing plants and cover crops, manure and sludge application, and inoculation with specific mycorrhiza (Wilson et al., 1991; Franco et al., 1992). Organic matter can improve the fertility of soils by enhancing the cation exchange capacity and nutrient availability (Raij, 1981; Diekow et al., 2005).

Poor nutrient recovery is typically caused by inadequate correspondence between periods of maximum crop demand and the supply of labile soil nutrients

The disparity between periods of maximum crop demand and the supply of labile soil nutrients (Cassman et al., 2003) can be exacerbated by overfertilization (e.g., Peng et al., 2006; Russell et al., 2006). For elements like nitrogen which are subject to losses from multiple environmental pathways, 100% fertilizer recovery is not possible (Sheehy et al., 2005). Nevertheless, precision management tools like leaf chlorophyll measurements that enable real-time nitrogen management have been shown to reduce fertilizer N application by 20-30% while maintaining rice productivity (Peng et al., 1996; Balasubramanian et al., 1999, 2000; Hussain et al., 2000; Singh et al., 2002). From 1980 to 2000 in the US, maize grain produced per unit of applied N increased by more than 40%, with part of this increase attributed to practices such as split-fertilizer applications and preplant soil tests to establish site-specific fertilizer recommendations (Raun and Johnson, 1999; Dobermann and Cassman, 2002). Despite improved management practices, average N recovery in US maize remains below 0.4 kg N per kg fertilizer N (Cassman et al., 2002), indicating significant scope for continued improvement.

Precision application of low rates of fertilizer can boost productivity among resource poor farmers.

Resource constraints prevent many small-scale farmers from applying fertilizer at rates that maximize economic returns. ICRISAT has been working in SSA to encourage small-scale farmers to increase inorganic fertilizer use and to progressively increase their investments in agricultural production. This effort introduces farmers to fertilizer use thorough micro-dosing, a concept based on the insight that farmers are risk averse, but will gradually take larger risks as they learn and benefit from new technologies (Dimes et al., 2005; Rusike et al., 2006; Ncube et al., 2007). Micro-dosing involves the precision application of small quantities of fertilizer, typically phosphorus and nitrogen, close to the crop plant, enhancing fertilizer use efficiency and improving productivity (e.g., 30% increase in maize yield in Zimbabwe). Yield gains are larger when fertilizer is combined with the