Mixed model of additive main effects and multiplicative interaction for stability analysis of cassava

The research objective was to analyze genotype × environment interaction (gei) using ammi mixed model with restricted maximum likelihood (reml) method both with and without coefficient of coancestry matrix (a matrix) assuming residual error variance across environments were homogeneous and heterogeneous. multilocation trials were conducted at five districts of east java province,indonesia,from november 2010 to august 2011. the results showed that no pcs values that significantly different from ammi mixed model analysis,both without and with a matrix,assuming homogeneous error variance across environments. while the result of ammi mixed model analysis,both with and without a matrix,assuming heterogeneous residual error variance across environments had the same interpretation. the most stable genotype that located closest to the origin of biplot was genotype g13 (cmm 02033-1). the yield potential of g13 was not high (close to average). four genotypes namely g4 (adira 4),g6 (cmm 03036-7),g7 (cmm 03036-5),and g15 (cmm 02048-6) were the most unstable genotypes. environment s4 (malang) had the smallest interactions effect,while environments with the greatest interaction effect were s3 (probolinggo) and s1 (kediri),because these environments had a long vector. © pakistan academy of sciences.