Most farmers prefer either short-duration cultivars alone (70-90 days), or a mixture of early,
medium and late cultivars (Haugerud, 1988). One rationale for the mixed strategy is that short-
cycle cultivars, by filling food and cash gaps, enable some farmers to grow long-cycle varieties
as well. Wealthier farmers with large land holdings can devote more land to late cultivars. Given
the nearly universal demand for some early cultivars, the Rwandan germplasm screening and seed
production programme, which had previously emphasised late cultivars, recently increased the
emphasis on short-duration cultivars. Previously, the programme had taken insufficient account
of the multi-crop and multiple-season framework in which farmers decide what cultivars to adopt.


Defining Appropriate Experimental Conditions

Efforts to match the conditions of resource-poor farmers in experimental fields are controversial.
Should varietal selection on the research station be conducted under husbandry conditions beyond
the reach of most African farmers? Identification of superior genotypes is more difficult under
low input conditions, where heterogeneity makes it difficult to apply equal selection pressure over
an entire plant population. More experimental replications are requried, since differences in
productivity may be small and statistical error large. Adjusting on-station research to farmers'
practices and priorities can complicate experimental design and analysis. Classic experimental
methodology, however, has its own shortcomings. Both conscious and unconscious decisions by
crop scientists produce more favourable crop environments on research stations than in farmers'
fields, and lead breeders to select genotypes that respond well to favourable environments
(Maurya et al, 1988; Simmonds, 1984).

One problem is to identify the changes to farmers' practices and priorities which it is reasonable
to expect them to adopt. The yield is in part due to circumstances beyond farmers' control (eg.
whether fertilizer or irrigation water arrives on time), as well as to farming practices that make
good biological, nutritional sense. Small farmers may use low inputs for a number of reasons: the
mix of production, consumption, and marketing priorities within the farming system; limited cash
resources; inadequate personal influence to obtain inputs; and limited capacity to risk high losses.
Small cultivators operate multiple enterprises as an integrated system, which requires compro-
mises in management, and therefore productivity, of any one constituent enterprise. Traditional
mixed cropping is a further dimension of this systems context with its own implications for
germplasm selection.

Another way in which germplasm screening can take greater account of the diversity of actual
farm conditions is to decentralise screening by the earlier release of promising material to farmers
for testing in on-farm trials, as in a successful rainfed rice breeding programme in India (Maurya
et al, 1988).

When scientists define treatment and non-experimental variables for cultivar selection, they
manipulate management practices such as time of planting, soil fertility, water availability,
chemical protection against diseases and pests, intercropping, relay cropping, cultivar mixtures,
crop rotation and plant spacing. The more explicitly they take such decisions from a knowledge
of farmers' practices, and the less tied they are to traditional textbook experimental design, the
more useful research results are likely to be. Some illustrations follow.


GATEKEEPER SERIES NO. SA30