The commercial production of genetically modified crops has led to a growing awareness of the difficulties of transgene confinement and the environmental risks associated with the escape of transgenes into natural plant populations. Because nearly all crop species mate with a wild relative in some part of their range, there is little question that transgenes released through crop systems will eventually be introduced to weedy or native plant species. Less certain is how transgenic plants will behave in the wild, or how populations of transgenic plants will affect natural plant and animal communities. Biological interactions, such as herbivory and competition, may promote the spread of transgenes when transgenic plants or hybrids are more fit than their progenitors. We evaluated the effects of herbivores, plant density and plant community composition on the fitness of canola (Brassica napus L.), canola engineered for insect resistance, field mustard (B. rapa L. (Brassicaceae), and their hybrids in environmentally controlled mesocosm chambers. Specifically we asked how plant fitness differs among genotypes, and how the frequency of the transgene evolves under weak and strong selection by herbivores and competitors.
Eight free-standing, open-topped chambers were fitted with three large pots filled with native soil. In 2006, each pot was planted with three replicates of each of five plant lines consisting of three parental lines: canola (B. napus), GM-canola (B. napus genetically modified for insect resistance (Bt cry1Ac)), and field mustard (B. rapa), and their hybrids: canola X field mustard, and GM-canola X field mustard. Competition was simulated in two pots by interplanting experimental plants with either a mixture of ruderal species or additional canola. Four mesocosms were inoculated with a specialist herbivore, Plutella xylostella. Selection by herbivores was manipulated in 2007 by increasing by approximately 100X the number of P. xylostella added to each chamber. ANOVA of the effects of competition, plant lineage and herbivore treatment on aboveground biomass and seed production revealed a significant three-way interaction of year*herbivory*genotype (P<0.001). Moreover, the main effect of competition was significant (P < 0.01). Post hoc analysis further demonstrated that seed mass was greater for transgenic than non-GM F1 plants in 2007 (P<0.05). Results of this work indicate that the fitness effects of a beneficial transgene are spatially and temporally variable and determined in part by community composition and level of selection.