The suggestion of mispriming in our i-PCR reaction is also warranted for sequences AF434756 and AF434759 (ref. 1). Significant homology with putative misamplifications is maintained across the length of these fragments, and the CaMV sequence was not recovered. However, this pattern is not found in our other i-PCR sequences. A revealing pattern of discontinuity is found at at least one end of five other sequences, indicating the integration junction between the transgenic DNA and the native host genome. Our critics choose not to recognize this feature in the majority of our i-PCR data. Partial homology with retrotransposon elements in maize is common in primers designed to amplify transposon-like sequences, and is not unique to our primers. Questions concerning the distortion of expected footprints at the DNA-integration junction certainly warrant future work.

The movement of transgenes into new populations and across generations is expected to result in diverse integration patterns^2-7. Our findings are compatible with recent studies^2-6 that characterize transgene/host DNA junctions where rearrangements include interspersion with host or unidentifiable DNA. As altered DNA species should also be an important focus of ecological research, we disagree with our critics who assume that only intact transgenes are worthy of attention in our study.

We agree that PCR-based methods are sensitive and therefore open to artefacts, but strongly disagree that the presence of these artefacts is unavoidable or uncontrollable. The consistent performance of our controls, as reported¹, discounts beyond reasonable doubt the possibility of false positives in our results. Nevertheless, the high sensitivity of the PCR reaction has incited some critics to request a non-PCR-based method to confirm our main statement. To address these challenges, we evaluated the same samples from our original publication¹ using DNA–DNA hybridization. The results of these experiments continue to support our primary statement.

Our analysis of Oaxacan maize is unique for several reasons. First, we wished to document changes that occur within diverse populations of landraces (rather than single varieties or lines), for which no markers, restriction-enzyme digestion maps or linkage analyses have been developed. Second, we could not have predicted which (or how many) specific transgenic constructs (or derivatives) were present in the samples that we analysed. Third, our samples of ground, pooled kernels from individual maize cobs do not represent individual genomes. All of these factors render the application of DNA-hybridization methods difficult. To minimize confusion in interpreting the multiplicity of bands that would have been created by Southern hybridization with our samples, we chose to use dot blotting for our experiments.

We extracted genomic DNA from dry maize kernels¹. Standards containing varying amounts of transgenic material were prepared by mixing flour from our positive control (Bt1) and our historical negative control¹. We blotted and immobilized 10–15 µg of DNA from each sample onto a nylon membrane using a Bio-Dot apparatus (Bio-Rad). We generated a horseradish peroxidase-labelled DNA probe from the same 220-base-pair fragment of the p-35S CaMV promoter that was amplified from our previously reported¹ positive control (Bt1). Hybridization conditions were as follows: 56 °C, 6 ng ml^-1 DNA probe, 1 hour. Washes were as follows: 3 5 min with 0.1 SSC/0.1% SDS at 56 °C, followed by 3 5 min with 2 SSC at room temperature. Loading homogeneity was confirmed by stripping and rehybridization of the experimental membrane with the 329-base-pair fragment from the maize-specific zein gene¹. Probe labelling, hybridizations and detection were carried out using a North2South kit (Pierce Endogen), according to the manufacturer's specifications.

DNA from four of our six criollo landrace samples, and from the Diconsa sample, hybridized with our CaMV probe (Fig. 1). By using standardized mixtures of transgenic and non-transgenic maize, dot-blot hybridization suggests a ratio of transgenic to non-transgenic kernels in criollo cobs of the order of 1:100, as we had previously suggested¹ and as was confirmed by Mexican government studies¹. This DNA-hybridization study confirms our original detection of transgenic DNA integrated into the genomes of local landraces in Oaxaca.