Dominant Inheritance of Field-Evolved Resistance to Bt Corn in Busseola fusca

Transgenic crops expressing Bacillus thuringiensis (Bt) toxins have been adopted worldwide, notably in developing countries. In spite of their success in controlling target pests while allowing a substantial reduction of insecticide use, the sustainable control of these pest populations is threatened by the evolution of resistance. The implementation of the “high dose/refuge” strategy for managing insect resistance in transgenic crops aims at delaying the evolution of resistance to Bt crops in pest populations by promoting survival of susceptible insects. However, a crucial condition for the “high dose/refuge” strategy to be efficient is that the inheritance of resistance should be functionally recessive. Busseola fusca developed high levels of resistance to the Bt toxin Cry 1Ab expressed in Bt corn in South Africa. To test whether the inheritance of B . fusca resistance to the Bt toxin could be considered recessive we performed controlled crosses with this pest and evaluated its survival on Bt and non-Bt corn. Results show that resistance of B . fusca to Bt corn is dominant, which refutes the hypothesis of recessive inheritance. Survival on Bt corn was not lower than on non-Bt corn for both resistant larvae and the F1 progeny from resistant × susceptible parents. Hence, resistance management strategies of B . fusca to Bt corn must address non-recessive resistance.


based on "mortality phenotypes"
A second estimation of dominance was based on the proportion of "mortality phenotypes" (mortality date) that were considered incompatible with the mortality pattern of susceptible individuals. Mortality in the S × S progenies reared on Bt was modeled as a function of time using a Weibull distribution characterized by both a shape parameter (k > 0) and a scale parameter (λ > 0). Maximum log-likelihood estimations of those parameters set the reference model (henceforth Φ ss ) which describes the mortality dynamics of susceptible individuals on Bt corn. Then, on the basis of this model, we computed the probability of individual mortality (in the R × R and R × S progenies) to be incompatible with the mortality phenotype of susceptible individuals (Φss). The density probability associated to the proportion of "nonsusceptible" individuals (phenceforth p R×S and p R×R relative to the crosses R × S and R × R, respectively) was calculated on the basis of a binomial distribution. The dominance (h ϕ ) was assessed using the following formula: h ϕ = (p S×Sp R×S ) / (p S×Sp R×R ), and its associated posterior probability distribution was computed accordingly.

Reference model
The mortality M(t) in the S × S progenies reared on Bt was modeled as a function of time using to a Weibull distribution: where k > 0 is a shape parameter and λ > 0 is a scale parameter of the distribution.
Under this model, we considered the probability for an individual i, originating from a S × S cross c, to die between two observation times T ci and T ci +1. The corresponding likelihood function used to fit the Weibull model was: The maximum likelihood estimations of the parameters k and λ defined the reference model (henceforth, Φ ss ) describing the mortality dynamics of susceptible individuals (i.e., originating from S × S) on Bt corn.

Proportion of "non-susceptible" phenotypes in the R × R and R × S progeny
Then, on the basis of this model, we discriminated individuals of the R × R and R × S progenies whose observed mortality date was not compatible with the phenotype of susceptible individuals (Φss). We classified as "non-susceptible" any individual mortality events that had, conditionally to Φ ss , a probability to be observed lower than 0.001. The following criterion was considered: where O ci is the time interval within which the death an individual R × R or R × S is observed.
Individuals surviving up to adult stage were de facto considered to meet this criterion.
The density probability of the proportion of individuals meeting this criterion in each cross (p henceforth p R×S and p R×R relative to the crosses R × S and R × R, respectively) was calculated on the basis of a binomial distribution: where n c denotes the observed number of individuals which were discriminated as noncompatible with the Φ ss model out of the N c individuals reared on Bt.
The dominance corresponding to those distributions was assessed using the same formula as before: h ϕ = (p S×Sp R×S ) / (p S×Sp R×R ), and the associated posterior probability distribution was computed accordingly.