Kaare M. Nielsen and Jeffrey Townsend reply:

Davison's considerations may be relevant to citizens and policymakers as they decide upon policy toward the usage of plant marker genes; however, it is up to the scientific community to provide unbiased and accurate methods for assessing risk, which is what our article advocates. Our article did not discuss the specific effects of plant marker gene flow, but the general failure of current studies to detect HGT if it occurred, and the absence of population genetic considerations in these studies. Thus, none of the comments made by Davison is relevant to the scientific issues we point out. We wish to make our own clarifications of Davison's assertions.

First, he argues that HGT studies conducted with bacteria with inserted sequence-similarity to plant transgenes are not relevant to understanding HGT processes in natural environments, presumably because naturally occurring bacteria lack sequence-similarity to transgenes. In fact, a recent study shows that many commonly occurring bacteria have high sequence similarity with plant marker genes, suggesting homology-based recombination can occur1. Moreover, the plant marker gene can provide the anchor sequence necessary to initiate transfer of adjacent transgenes into bacteria. For instance, another recent study demonstrates that short stretches of DNA sequence similarity facilitate the incorporation of larger (>1 kb) heterologous DNA fragments, including entire plant genes, into naturally transformable bacteria2. These observations suggest that additive integration of transgenes can take place in bacteria after homology-initiated recombination; and additive integration is known to occur at high frequencies in bacteria3. More than ten peer reviewed studies now demonstrate that bacteria, including Acinetobacter baylyi and Pseudomonas stutzeri, take up either purified plant DNA or DNA naturally present in colonized whole plants when sequence similarity is present4. Natural transformation has also been demonstrated in Escherichia coli5.

Second, Davison suggests that nptII genes and other plant marker gene homologs are widespread in nature and that rare HGT from transgenic plants would add insignificantly to the dissemination of these. However, no reference to peer reviewed studies is given providing evidence for the uniform and geographically widespread occurrence of nptII genes in soils exposed to this transgene.

Third, he only considers the risk of HGT with regard to plant marker genes that represent only a subset of the transgenes in use. However, a main argument in our article is that novel transgenes that do not have natural counterparts are those that require particular attention in a HGT context. These are transgenes, including novel combinations of regulatory elements and toxin protein domains (e.g., for vaccines or biopharmaceuticals), that may differ substantially from those arising by natural evolutionary processes6.

We hope these considerations, along with our perspective and the methods suggested therein, focus interested parties on the issues important to resolving the long-term effect of transgenes exposed to complex microbial communities.