To maximize yield throughout a crop's heterogeneous growing region, de
spite differences in cultivar rankings from place to place due to geno
type-environment interactions, frequently it is necessary to subdivide
a growing region into several relatively homogeneous mega-environment
s and to breed and target adapted genotypes for each mega-environment.
The objectives of this study are to identify relevant criteria for ev
aluating mega-environment analyses and to apply the Additive Main Effe
cts and Multiplicative Interaction (AMMI) model to mega-environment an
alysis. The proposed analysis is illustrated using a Louisiana corn (Z
ea mays L.) trial. Statistical strategies for identifying mega-environ
ments should meet four criteria: flexibility in handling yield trials
with various designs, focus on that fraction of the total variation th
at is relevant for identifying mega-environments, duality in giving in
tegrated information on both genotypes and environments, and relevance
for the primary objective of showing which genotypes win where. The A
MMI model meets these criteria effectively when the usual biplots are
supplemented with several new types of graphs designed to address ques
tions about mega-environments. Preliminary results indicate that a sma
ll and workable number of mega-environments often suffices to exploit
interactions and increase yields.