successful vegetative propagation (Leakey, 2004). The identification of selection criteria is being based on the quantitative characterization of many fruit and nut traits (Atangana et al., 2001, 2002; Anegbeh et al., 2003, 2004; Waruhiu et al., 2004; Leakey, 2005b; Leakey et al., 2005bc). Using participatory approaches (Leakey et al., 2003), the implementation of these techniques is being successfully achieved by small-scale farmers from 40 communities (Tchoundjeu et al., 2006).

Clonal approaches to the genetic improvement of timber tree species result in large improvements in yield and quality traits.

For example in timber species, clones of E. urophylla x E. grandis hybrid in Congo were planted in monoclonal blocks of 20-50 ha at a density of 800 stems ha-1 and resulted in mean annual increments averaging 35 m3 ha-l, compared with 20-25 m3 ha-l from selected provenances, and about 12 m3 ha-l from unselected seedlots (Delwaulle, 1983). In Brazil, mean annual increments between 45-75 m3 ha-l and up to 90 m3 ha-l have been recorded (Campinhos, 1999). Clonal approaches require (Leakey, 1987; Ahuja and Libby 1993ab) genetic diversity (Leakey, 1991), wise deployment (Foster and Bertolucci, 1994) and appropriate silviculture (Lawson, 1994; Evans and Turnbull, 2004) to maximize gains, minimize pest and pathogen risks and maintain species diversity in the soil microflora (Mason and Wilson, 1994), soil invertebrates (Bignell et al., 2005) and insect populations (Watt et al., 1997, 2002; Stork et al., 2003).

Increased private sector involvement in timber plantations has recently been more inclusive of social and environmental goals.

In the past, the cultivation of planted timber trees has mostly been implemented by national forestry agencies, often with inadequate attention to establishment techniques. In the last 20-30 years there has been increasing private sector investment, much of which has been multinational, and often in partnership with local companies or government agencies (Garforth and Mayers, 2005). These companies have focused on a few fast-growing species, especially for pulp and paper industries, often grown as exotic species outside their natural range. In these plantations genetic improvement has typically been achieved by provenance selection and clonal technologies. Increasingly, such plantations are being designed as "mosaic" estates with a view to greater synergies with both local agricultural conditions and areas protected for biodiversity (IIED, 1996) and as joint ventures with communities to provide non-fiber needs in addition to wood (Mayers and Vermeulen, 2002).

Livestock and fish

Domestication and the use of conventional livestock breeding techniques have had a major impact on the yield and composition of livestock products.

There has been widespread use of breed substitution in industrialized countries and some developing countries, often leading to the predominance of a few very specialized breeds, and often pursuing quite narrow selection goals. Organized within-breed selection has been practiced much less widely in many developing countries, partly because of the lack of infrastructure, such as national or regional performance recording and genetic evaluation schemes. Genetic improvement- breed substitution, crossbreeding and within-breed selection-has made an important contribution to meeting the growing global demand for livestock products. Selection among breeds or crosses is a one off, non-recurrent process: the best breed or breed cross can be chosen, but further improvement can be made only by selection within the populations (Simm et al., 2004). Crossbreeding is widespread in commercial production, exploiting complementarity of different breeds or strains, and heterosis or hybrid vigor (Simm, 1998). Trait selection within breeds of farm livestock typically produces annual genetic changes in the range 1-3% of the mean (Smith, 1984). Higher rates of change occur for traits with greater genetic variability, in traits that are not age- or sex-limited, and in species with a high reproductive rate, like pigs and poultry (McKay et al., 2000; Merks, 2000), fish and even dairy cattle (Simm, 1998). These rates of gain have been achieved in practice partly because of the existence of breeding companies in these sectors. Typically, rates of genetic change achieved in national beef cattle and sheep populations have been substantially lower than those theoretically possible, though they have been achieved in individual breeding schemes. The dispersed nature of ruminant breeding in most countries has made sector-wide improvement more challenging.

In most species, rates of change achieved in practice through breeding have increased over the last few decades in developed countries.

The greatest gains in productivity as a result of genetic improvement have been made in poultry, pigs and, to a lesser extent, dairy cattle. Greater success through breeding programs in developed countries has been the result of better statistical methods for estimating the genetic merit (breeding value) of animals, especially best linear unbiased prediction methods; the wider use of reproductive technologies, especially artificial insemination; improved techniques for measuring performance (e.g., ultrasonic scanning to assess carcass composition in vivo); and more focused selection on objective rather than subjective traits, such as milk yield rather than type. Developments in the statistical, reproductive and molecular genetic technologies available have the