Industrial strains have been traditionally improved by rational approaches and combinatorial methods involving mutagenesis and selection. Recently, other methods have emerged, such as the use of artificial transcription factors and engineering of the native ones. As methods for generating genetic diversity continue to proliferate, the need for quantifying phenotypic diversity and, hence, assessing the potential of various genetic libraries for strain improvement becomes more pronounced. Here, we present a metric based on the quantification of phenotypic diversity, using Lactobacillus plantarum as a model organism. We found that phenotypic diversity can be introduced by mutagenesis of the principal factor, that this diversity can be modulated by tuning the sequence diversity, and that this method compares favorably with commonly used protocols for chemical mutagenesis. The results of the diversity metric here developed also correlated well with the probability of finding improved mutants in the different libraries, as determined by recursive screening under stress. In addition, we subjected our libraries to lactic and inorganic acids and found strains with improved growth in both conditions, with a concomitant increase in lactate productivity.divergence ͉ phenotypic diversity ͉ strain improvement ͉ stress tolerance ͉ transcriptional engineering P roduction of chemicals and fuels from renewable sources has become a major focus of the biotechnology industry as petroleum sources become scarcer and riskier to secure (1). Metabolic engineering has emerged as the enabling technology for designing entire biochemical pathways, both heterologous and native to the engineered strain (2). By using mostly rational approaches for pathway design and construction, this field has met with remarkable success in manufacturing commercial products (3-5). A major remaining challenge is engineering environmental tolerance phenotypes to allow robust growth and production under adverse industrial conditions (6, 7). In general, these phenotypes are hard to engineer, because stress conditions elicit multigenic responses (8) coordinated at the transcriptomic and proteomic levels (9, 10).Robust industrial strains have been traditionally obtained through adaptation and selection (11,12). Because natural adaptation is time-and resource-intensive, mutagens are used to introduce diversity in the population and accelerate the process (13). Alternatively, and in recognition of the more direct mapping between transcriptome and phenotype, efforts in strain improvement have focused on direct manipulation of the transcript profile (14 -19). Global transcription machinery engineering (gTME) has been successfully applied to introduce transcriptional-level modifications that are transferable between strains. Examples include random mutagenesis of the TATA-binding protein of Saccharomyces cerevisiae (14) and of the principal -factor of Escherichia coli (15). In addition to artificial transcription factors and gTME, other methods for phenotypic alteration ...