Two experiments, each involving a set of 10 sweet potato clones, were conducted for three years at 4 sites (Ekona, Ebolowa, Nkolbisson, and Bambui Plain) in Cameroon . Data obtained were subjected to analysis of variance to determine the presence of genotype x environment (G x E) interactions, and to regression analysis to assess the performance of clones across environments . Environments were assessed in two ways : (i) the mean response of all clones (dependent assessment), and (ii) the average performance of a different set of clones (independent assessment) .
The first experiment (Expt 1) produced higher yields but had fewer stable clones than the second (Expt 2) .
The analysis of variance revealed that the clones interacted significantly with environments for all traits .
The study has identified high yielding and stable sweet potato clones for distribution to growers in the major areas of cultivation in the country. Despite slight differences in numbers of clones judged stable by the various regression indices in the two methods of environmental assessments, the rankings of clones on the basis of their linear regression coefficients were similar . In a developing country like Cameroon, with limited resources and where sophisticated equipment for obtaining physical or biological measures of the environment may be lacking, the mean performance of genotypes may still be the most reliable measure of environment in evaluating the stability of performance of crop cultivars . 'Environmental indices computed (A) as means of original set of clones (Expt 2) and (B) as means of second set of clones (Expt 1) . * b value significantly different from b = 1 .0; std value significantly different from 0 ; 62 and s2 values significantly greater than the pooled error, hence unstable.