The variability in phenotypic outcomes among biological replicates in engineered microbial factories presents a captivating mystery. Establishing the association between phenotypic variability and genetic drivers is important to solve this intricate puzzle. We applied a previously developed auxin-inducible depletion of hexokinase 2 as a metabolic engineering strategy for improved nerolidol production in Saccharomyces cerevisiae, and biological replicates exhibit a dichotomy in nerolidol production of either 3.5 or 2.5 g L −1 nerolidol. Harnessing Oxford Nanopore's long-read genomic sequencing, we reveal a potential genetic cause�the chromosome integration of a 2μ sequence-based yeast episomal plasmid, encoding the expression cassettes for nerolidol synthetic enzymes. This finding was reinforced through chromosome integration revalidation, engineering nerolidol and valencene production strains, and generating a diverse pool of yeast clones, each uniquely fingerprinted by gene copy numbers, plasmid integrations, other genomic rearrangements, protein expression levels, growth rate, and target product productivities. Τhe best clone in two strains produced 3.5 g L −1 nerolidol and ∼0.96 g L −1 valencene. Comparable genotypic and phenotypic variations were also generated through the integration of a yeast integrative plasmid lacking 2μ sequences. Our work shows that multiple factors, including plasmid integration status, subchromosomal location, gene copy number, sesquiterpene synthase expression level, and genome rearrangement, together play a complicated determinant role on the productivities of sesquiterpene product. Integration of yeast episomal/integrative plasmids may be used as a versatile method for increasing the diversity and optimizing the efficiency of yeast cell factories, thereby uncovering metabolic control mechanisms.