The response of seagrass systems to a severe disturbance provides an opportunity to quantify the degree of resilience in different meadows, and subsequently to test whether there is a genetic basis to resilience. We used existing data on levels of long-standing disturbance from poor water quality, and the responses of seagrass (Zostera muelleri) after an extreme flood event in Moreton Bay, Queensland, Australia. Sites were grouped into high and low disturbance categories, in which seagrass showed high and low resilience, respectively, as determined by measuring rates of key feedback processes (nutrient removal, suppression of sediment resuspension, and algal grazing), and physiological and morphological traits. Theoretically, meadows with higher genotypic diversity would be expected to have greater resilience. However, because the more resilient meadows occur in areas historically exposed to high disturbance, the alternative is also possible, that selection will have resulted in a narrower, less diverse subset of genotypes than in less disturbed meadows. Levels of genotypic and genetic diversity (allelic richness) based on 11 microsatellite loci, were positively related (R2 = 0.58). Genotypic diversity was significantly lower at highly disturbed sites (R = 0.49) than at less disturbed sites (R = 0.61). Genotypic diversity also showed a negative trend with two morphological characteristics known to confer resilience on seagrass in Moreton Bay, leaf chlorophyll concentrations and seagrass biomass. Genetic diversity did not differ between disturbed and undisturbed sites. We postulate that the explanation for these results is historical selection for genotypes that confer protection against disturbance, reducing diversity in meadows that contemporarily show greater resilience.