Graduation date: 2007Stable expression of transgenes is required for commercial uses of genetically engineered trees. To better understand the stability of transgene expression under field conditions, we studied transgene expression and RNA interference (RNAi)-induced transgene suppression in 2,480 transgenic poplars (460 transgene insertion events) over three years. Stability of expression was assessed based on two reporter genes, green fluorescent protein (GFP) driven by a 35S promoter and the herbicide resistance gene BAR driven by a poplar rbcS promoter. No cases of gene silencing (complete breakdown of expression) were observed for either gene, although physical loss of the transgenes occurred in three events after 80 events had been subject to further organogenesis in tissue culture. Flanking MARs did not significantly elevate transgene expression or stability, but reduced variance in expression among the events. MARs increased the correlation of expression between events for GFP and BAR genes in the same T-DNA. A majority of transformants (85%) carried single copy inserts; transgene copy number was positively correlated with expression (r = 0.46 for GFP, and 0.35 for BAR). Formation of direct repeats was frequently observed in transgenic events containing multiple inserts of T-DNA, but did not adversely affect transgene expression. RNAi using inverted repeats (IR) directed at the coding sequence gave a high degree of gene suppression of a resident BAR transgene; 80% of transgenic events showed more than 90% suppression. IR directed at the promoter sequence, however, was very inefficient in inducing gene suppression; only 6% of transgenic events showed more than 90% suppression. RNAi efficiency was unaffected by the presence of MARs. The degree of RNAi suppression was stable over two years in the field, as well as during seasonal development. Copy number of integrated IR loci was also unassociated with frequency of gene suppression. DNA methylation was observed in the promoter region of the highly suppressed events containing an IR of the promoter sequence, and in the coding region of highly suppressed events containing IR directed at the coding sequence; however, there was no clear relationship of methylation to the level of gene suppression in coding region-directed RNAi.