Cassman et al., 2003). Nonetheless, further small gains are expected, through continued genetic gain and a better understanding and breeding for specific target environments (Reynolds and Borlaug, 2006). In developing countries and low yield potential environments the benefits of breeding for specific environments will be further enhanced with the adoption of more localized and/or participatory breeding, i.e., with the exploitation of G × E or local adaptation.

In several intensive production environments, cereal yields are not increasing.

In several of the most important regions for irrigated rice production (e.g., areas of China, Japan, Korea) there is strong evidence of persistent yield stagnation at approximately 80% of the theoretical productivity levels predicted by simulation models (Cassman et al., 2003). This type of stalled exploitation of potential production is primarily caused by economic factors since the rigorous management practices required for yield maximization are not cost effective (Pingali and Heisey, 1999; Cassman et al., 2003). Rice yield stagnation has also been observed in areas like Central Java and the Indian Punjab at levels significantly below 80% of the theoretical productivity. In long-term cropping system experiments (LTE) with the highest-yielding rice varieties under optimal pest and nutrient management, rice yield potential declined at several locations. Subsequent evidence from a larger set of LTEs suggested that this phenomenon was not widespread, but that rice yield potential was essentially stagnant in most regions despite putative innovations in management and plant genetic resources (Dawe et al., 2000). For irrigated production systems in the maize belt of the United States, yields achieved by the most productive farmers have not increased since the mid-1980s (Duvick and Cassman, 1999). For spring wheat producers in Mexico's Yaqui Valley, only nominal increases in yield have been observed since the late 1970s.

In many regions the production potential for the staple cereal crops has not been exhausted.

In contrast to concerns about limited future opportunities for yield improvement in cereals, there are some examples of yield increases. For example, coordinated efforts to improve management practices and profitability of Australian rice systems increased productivity from 6.8 tonnes ha-1 in the late 1980s to 8.4 tonnes ha-1 by the late 1990s (Ferrero and Nguyen, 2004). Farm-level maize yields in the United States are typically less than half of the climate-adjusted potential yield (Dobermann and Cassman, 2002). At the state level in India, an analysis (Bruinsma, 2003) suggests that rice productivity could be increased by 1.5 tonnes ha-1 (ca. 50%) without exceeding the 80% criteria commonly used to establish the economically-exploitable component of the biophysical yield potential (Bruinsma, 2003).

In developing countries the productivity of many smallscale farming systems is often constrained by limited access to inputs and modern varieties (MVs) and poor management practices.

In upland rice systems in Laos, the importance of the adoption of improved varieties and N fertilization has been demonstrated (Saito et al., 2006). By substituting MVs for traditional landraces, rice yields doubled to 3.1 tonnes ha-1 with a moderate dose of nitrogen fertilizer further improving yield by 1 tonne ha-1. Among farmers in Nepal, modern crop management practices (e.g., timely establishment, precision planting, two weedings) together with site-specific nutrient management boost rice productivity by 2 tonnes ha-1 over typical farmer practices (Regmi and Ladha, 2005). In West Africa, rural surveys show that most farmers have limited knowledge of soil fertility management and of optimal establishment practices for rice (Wopereis et al., 1999). In these areas, nitrogen deficiency, inadequate weeding, and late planting are commonly associated with low cereal productivity (Becker and Johnson, 1999). Poor knowledge of efficient practices for maintaining soil fertility has also been identified as an important component of the low yields achieved by Bangladeshi rice farmers (Gaunt et al., 2003).

Barriers to clonal forestry and agroforestry have been overcome by the development of robust vegetative propagation techniques, which are applicable to a wide range of tree species.

Techniques of vegetative propagation have existed for thousands of years (Hartmann et al., 1997), but the factors affecting rooting capacity seem to vary between species and even clones (Leakey, 1985; Mudge and Brennan, 1999). However, detailed studies of the many morphological and physiological factors affecting five stages of the rooting process in stem cuttings (Leakey, 2004) have resulted in some principles, which have wide applicability (Dick and Dewar, 1992) and explain some of the apparently contradictory published information (Leakey, 2004). Robust low-technology vegetative propagation techniques are now being implemented within participatory village-level development of cultivars of indigenous fruit/nut tree species to diversify cocoa farming systems in West Africa (Leakey et al., 2003).

Participatory domestication techniques are using low-tech approaches to cloning to develop cultivars of new tree crops for agroforestry.

Simple, inexpensive and low-tech methods for the rooting of stem cuttings have been developed for use by resource poor farmers in remote village nurseries (Leakey et al., 1990). These robust and appropriate techniques are based on a greatly increased understanding of the factors affecting