The large genotype x environment interactions of cassava clones constitute a challenge for
cassava breeders. The theory of site selection is not yet fully developed to incorporate as many
site variables as possible to form a more stable basis for choosing trial sites. We mapped the
agricultural environments of Nigeria to enable cassava breeders to select representative sites for
the agro-ecological evaluation of candidate genotypes. First, we divided Nigeria into 337 land
units or ecozones that measured 30' latitude x 30' longitude (i.e., almost 56 km2
). Each ecozone
was described by 100 variables, which covered aspects of relief, rainfall, geology, meteorology,
vegetation, soils, population, ground-water potential, and other related statistics. Data were
coded for computer analyses of principal components and for clustering ecozones. The number
of groups formed depended on level of resemblance among ecozones in each group. Our study is
the first to classify the ecozones of Nigeria to facilitate selection of trial sites for cassava
multilocational trials according to a broad-based list of variables and thus generate a practical
map for field use. Any ecozone in a selected group may be a trial site if it is near, secure, and has
other infrastructure.