In the Mediterranean region, progress in selection for yield in harsh
environments is hampered by large environmental variation between seas
ons and locations. This study analyzes the genotype-by-environment (GE
) interaction of 36 two-rowed genotypes of barley (Hordeum vulgare L.)
, grown in 14 environments in Syria and North Africa. It assesses the
effect of growth type (winter or spring type) and heading date on the
GE interaction and determines whether or not high-yielding (HY) enviro
nments are representative of low-yielding (LY) ones. Average grain yie
ld per environment ranged from 7 to 513 g m-2. Genotypes and environme
nts were classified by a cluster analysis and the interaction was anal
yzed with an additive main effects and multiplicative interaction mode
l. Genotypes were classified into four clusters, related to their grow
th type and earliness of heading. Environments were clustered in a HY
and LY group; this classification was related to seasonal rainfall and
temperature. Medium-early heading winter types had a positive interac
tion with LY environments and a negative interaction with HY environme
nts, whereas late heading genotypes (spring and winter types) had the
opposite interaction pattern. Early beading spring types had above-ave
rage mean yields; the highest-yielding among them tended to have a low
interaction with environments. High-yielding environments did not dis
criminate well between genotypes with high or low yields in LY environ
ments, and may thus have limited value for yield selection for LY envi
ronments. For a breeding program aimed at improving yield in environme
nts where favorable conditions are rare, selection for yield should be
done in representative less-favorable environments.