We constructed a mixed-species community designed to simulate roadside and field edge plant communities and exposed it to glyphosate drift in order to test three hypotheses: (1) higher fitness in transgenic Brassica carrying the CP4 EPSPS transgene that confers resistance to glyphosate will result in significant changes in the plant community relative to control communities; (2) given repeated years of glyphosate drift selective pressure, the increased fitness of the transgenic Brassica with CP4 EPSPS will contribute to an increase in the proportion of transgenic progeny produced in plant communities; and (3) the increased fitness of Brassica carrying the CP4 EPSPS transgene will contribute to decreased levels of mycorrhizal infection and biomass in a host species (Trifolium incarnatum). Due to regulatory constraints that prevented the use of outdoor plots for our studies, in 2005 we established multispecies communities in five large cylindrical outdoor sunlit mesocosms (plastic greenhouses) designed for pollen confinement. Three of the community members were sexually compatible Brassica spp.: transgenic glyphosate-resistant canola (B. napus) cultivar (cv.) RaideRR, glyphosate-sensitive non-transgenic B. napus cv. Sponsor, and a weedy B. rapa (GRIN Accession 21735). Additional plant community members were the broadly distributed annual weeds Digitaria sanguinalis, Panicum capillare, and Lapsana communis. Once annually in 2006 and 2007, two mesocosms were sprayed with glyphosate at 10% of the field application rate to simulate glyphosate drift as a selective pressure. After two years, changes were observed in community composition, plant density, and biomass in both control and treatment mesocosms. In control mesocosms, the weed D. sanguinalis (crabgrass) began to dominate. In glyphosate drift-treated mesocosms, Brassica remained the dominant genus and the incidence of the CP4 EPSPS transgene increased in the community. Shoot biomass and mycorrhizal infection in Trifolium incarnatum planted in 2008 were significantly lower in mesocosms that had received glyphosate drift treatments. Our results suggest that, over time, glyphosate drift can contribute to persistence of Brassica that express the CP4 EPSPS transgene and that increased representation of Brassica (a non-mycorrhizal host) within plant communities may indirectly negatively impact beneficial ecosystem services associated with arbuscular mycorrhiza.