Author's Abstract

Hybrid poplar plantations (Populus trichocarpa x Populus deltoides) are a relatively new feature on the landscape in the Pacific Northwest of the United States, and these plantations may soon include genetically engineered trees. Meanwhile, many wild poplar populations (Populus spp.) are highly degraded due in part to logging, dams, grazing, and agriculture. This raises the possibility that gene flow from plantation trees could have negative impacts on native tree populations. I analyzed gene flow from poplar plantations using a combination of large-scale field studies, genetic analyses, and simulation modeling. I describe development of a spatially explicit simulation model that depicts the processes of pollination, dispersal, establishment, competition, disturbance, and management activities on a landscape in western Oregon where hybrid poplar plantations are a prominent feature. Using sensitivity analyses, I demonstrate that competitiveness and fertility of transgenic trees were important factors determining the extent of modeled gene flow, and that these factors interacted such that the effects of enhanced competitiveness appeared to be obviated by cultivation of low-fertility transgenic trees. Disturbance regime, plantation silviculture, and the characteristics of the landscape surrounding plantations also had a strong influence on the rate of gene flow. Most modeled gene flow was due to long-distance transfer of pollen, and, to a lesser extent, seed. Modeled gene flow was insensitive to changes in vegetative dispersal and rates of vegetative establishment.

Field studies demonstrated low levels of gene flow from existing hybrid plantations in three settings. Gene flow was of a magnitude comparable to that achieved by the commonly cultivated horticultural variety Lombardy poplar (Populus nigra cv. Italica). I conclude that gene flow to wild populations is low under existing conditions. There was a close match between observed and modeled pollen flow. However, seed flow and establishment were overestimated by the model due to conservative assumptions in a number of functions and parameters.

The model proved useful for examining scenarios of cultivation of transgenic trees. Trees containing a neutral transgene were predicted to constitute less than 1% of the basal area of wild poplar populations after 50 years. In contrast, an insect resistance transgene that conferred a substantial growth benefit in the wild resulted in a continual increase in transgenic trees in wild populations. This trend was dramatically slowed or eliminated for trees with reduced fertility. Finally, an herbicide resistance gene had little effect on overall modeled gene flow, but establishment was enhanced in a narrow subset of agricultural fields where the herbicide was used as a primary means of weed control. All of these model estimates were likely inflated due to consistently conservative assumptions about processes and parameters for which there was little available information.