CHARACTERIZATION OF THE BACTERIAL COMMUNITIES OF ROMAINE LETTUCE: INTERACTIONS WITH ENVIRONMENTAL CONDITIONS AND FOOD SAFETY

<p>The leafy green industry in the United States has positioned the country as the second world leader in lettuce production. Romaine lettuce has been associated with several outbreaks of <i>E. coli </i>O157:H7 during the last decade, producing economic losses, as well as negative impacts on human health and consumer confidence. This pathogen has been demonstrated to actively colonize plants and persist for weeks; therefore, dealing with this issue will require an understanding of the interactions happening between plant host, human pathogen, environment, and the resident microbial communities. This research aimed to provide insights for control strategies at the level of prevention in the field, as well as of detection. Based on this, our research goals were: 1. Describe how environmental factors affect the leaf properties and microbial ecology of romaine lettuce plants, as well as the fate of <i>E. coli </i>O157:H7 on their leaves; 2. Evaluate the application of a new light scattering technology (BARDOT) developed at Purdue, as an alternative tool to characterize culturable bacterial communities from plants through the recognition of scatter patterns produced by bacterial colonies. </p> <p>Lettuce plants were grown under three relative humidity (RH) levels: A. 83% (SD= 7.0); B. 62% (SD=9.0); C. 43% (SD= 7.4); significant changes in leaf properties such as responses of stomatal resistance to water loss were observed. RH was the main factor explaining the variation of resident bacterial communities, changes of leaf properties and the fate of <i>E. coli</i> O157:H7. Humid condition A produced the lowest bacterial diversity, which was mainly explained by the decreased transpiration rates of these plants, while at the same time this condition allowed the highest <i>E. coli</i> O157:H7 growth. Under RH condition C, differences in leaf properties and their distributions across the lettuce leaves were found to be correlated with the composition and localization of the resident microbial communities. <i>E. coli</i> O157:H7 growth on leaves was also negatively correlated with the inoculum dose, and it was enhanced on leaf areas with increased stomatal density and size, and decreased leaf wettability. We found that resident bacterial communities are disturbed after the introduction of the human pathogen, <i>Microbacterium, </i>and one unclassified genus from the <i>Rhizobiaceae </i>family were found as biomarkers of communities where <i>E. coli</i> O157:H7 reached higher and lower population counts, respectively. </p> <p>For the BARDOT technology, three libraries containing 8,418 images of scatter patterns from the nine most abundant bacterial genera from conventional and organic romaine lettuce were created. These libraries covered around 70-76% of the total isolated populations. The training parameters achieved classifiers at genus level with positive predictive values (PPVs) between 90.6-99.8%. The validation with blind samples resulted in sensitivity and average classification accuracy values above 90% for both pure and mixed cultures. The sensitivity and classification accuracy per genus when new lettuce samples were tested, showed values between 51.9-79.1% and 42.9%-100%, correspondingly. Some bacterial genera were identified as challenging for the BARDOT and improvements for the technology have been suggested. BARDOT technology represents a rapid and easy-to-use alternative to conventional microbiological and molecular methods for identification of culturable bacteria. </p>