Response from Soleri and colleagues

摘要

Ortiz-García and colleagues (2005a) agreed with us that variance effective population size (Ne(v)), not census population size (n), should be used to estimate transgene frequency (Vencovsky and Crossa 1999), and reanalyzed their data using Ne(v) with two additional statistical tests. The first was based on the assumption that sampled maize populations had no significant structure, which is not valid (Cleveland et al. 2005). The second, Fisher's combined probability test, gave a minimum detection level for seeds in 2004 of approximately 1% (0.00775, P < 0.05) (Ortiz-García et al. 2005b), close to our estimate of approximately 1%–4% (0.00961–0.03586, P < 0.05) across individual locations, accounting for population structure (Cleveland et al. 2005), and contrasted with their original estimate of 0.01%, P = 0.00003 (Ortiz-García et al. 2005a). Therefore, there is no evidence to refute our conclusion that “we still do not have any data to support the proposition that transgenes are not present at other localities, or at frequencies below 1–4% in the localities in the Ortiz-García et al. study” (Cleveland et al. 2005, p. 205).