I have toxin data from samples of corn genotypes (n=200) that were grown in two replicates in two environments. Data is from two random sub-samples of each sample. However, in one environment, samples of the two replicates were bulked (mixed) and four random sub-samples drawn for toxin analysis. What is the appropriate model for use to tell the effect of genotype and environment. No interaction needed because of the nature of the design. Any ideas please.