Replacing synthetic insecticides with transgenic crops for pest management has been economically and environmentally beneficial, but these benefits erode as pests evolve resistance. It has been proposed that novel genomic approaches could track molecular signals of emerging resistance to aid in resistance management. To test this, we quantified patterns of genomic change in Helicoverpa zea, a major lepidopteran pest and target of transgenic Bacillus thuringiensis (Bt) crops, between 2002 and 2017 as both Bt crop adoption and resistance increased in North America. Genomic scans of wild H. zea were paired with quantitative trait locus (QTL) analyses and showed the genomic architecture of field-evolved Cry1Ab resistance was polygenic, likely arising from standing genetic variation. Resistance to pyramided Cry1A.105 and Cry2Ab2 toxins was controlled by fewer loci. Of the 11 previously described Bt resistance genes, 9 showed no significant change over time or major effects on resistance. We were unable to rule out a contribution of aminopeptidases (apns), as a cluster of apn genes were found within a Cry-associated QTL. Molecular signals of emerging Bt resistance were detectable as early as 2012 in our samples, and we discuss the potential and pitfalls of whole-genome analysis for resistance monitoring based on our findings. This first study of Bt resistance evolution using whole-genome analysis of field-collected specimens demonstrates the need for a more holistic approach to examining rapid adaptation to novel selection pressures in agricultural ecosystems.

Adaptation to human-induced environmental change has the potential to profoundly influence the genomic architecture of affected species. This is particularly true in agricultural ecosystems, where anthropogenic selection pressure is strong. Heliothis virescens primarily feeds on cotton in its larval stages, and US populations have been declining since the widespread planting of transgenic cotton, which endogenously expresses proteins derived from Bacillus thuringiensis (Bt). No physiological adaptation to Bt toxin has been found in the field, so adaptation in this altered environment could involve (i) shifts in host plant selection mechanisms to avoid cotton, (ii) changes in detoxification mechanisms required for cotton-feeding vs.feeding on other hosts or (iii) loss of resistance to previously used management practices including insecticides. Here, we begin to address whether such changes occurred in H. virescens populations between 1997 and 2012, as Bt-cotton cultivation spread through the agricultural landscape. For our study, we produced an H. virescens genome assembly and used this in concert with a ddRAD-seq-enabled genome scan to identify loci with significant allele frequency changes over the 15-year period. Genetic changes at a previously described H. virescens insecticide target of selection were detectable in our genome scan and increased our confidence in this methodology. Additional loci were also detected as being under selection, and we quantified the selection strength required to elicit observed allele frequency changes at each locus. Potential contributions of genes near loci under selection to adaptive phenotypes in the H. virescens cotton system are discussed. K E Y W O R D SBacillus thuringiensis, cotton, Heliothis virescens, selective sweep, tobacco budworm

Adaptation to human-induced environmental change has the potential to profoundly influence the genomic architecture of affected species. This is particularly true in agricultural ecosystems, where anthropogenic selection pressure is strong. Heliothis virescens primarily feeds on cotton in its larval stages and US populations have been declining since the widespread planting of transgenic cotton, which endogenously expresses proteins derived from Bacillus thuringiensis (Bt). No physiological adaptation to Bt toxin has been found in the field, so adaptation in this altered environment could involve: 1) shifts in host plant selection mechanisms to avoid cotton, 2) changes in detoxification mechanisms required for cotton-feeding versus feeding on other hosts, or 3) loss of resistance to previously used management practices including insecticides. Here we begin to address whether such changes occurred in H. virescens populations between 1997-2012, as Bt cotton cultivation spread through the agricultural landscape. For our study, we produced an H. virescens genome assembly and used this in concert with a ddRAD-seq enabled genome scan to identify loci with significant allele frequency changes over the 15 year period. Genetic changes at a previously described H. virescens insecticide target of selection were detectable in our genome scan, and increased our confidence in this methodology. Additional loci were also detected as being under selection, and we quantified the selection strength required to elicit observed allele frequency changes at each locus. Potential contributions of genes near loci under selection to adaptive phenotypes in the H. virescens cotton system are discussed.

Replacement of synthetic insecticides with transgenic crops for pest management has been both economically and environmentally beneficial. These benefits have often eroded as pests evolved resistance to the transgenic crops, but a broad understanding of the timing and complexity of the adaptive changes which lead to field-evolved resistance in pest species is lacking. Wild populations of Helicoverpa zea, a major lepidopteran crop pest and the target of transgenic Cry toxin-expressing cotton and corn, have recently evolved widespread, damaging levels of resistance. Here, we quantified patterns of genomic change in wild H. zea collected between 2002 and 2017 when adoption rates of Cry-expressing crops expanded in North America. Using a combination of genomic and genotypic approaches, we identified significant temporal changes in allele frequency throughout the genomes of field-collected H. zea. Many of these changes occurred concurrently with increasingly damaging levels of resistance to Cry toxins between 2012 and 2016, in a pattern consistent with polygenic selection. Surprisingly, none of the eleven previously described Cry resistance genes showed signatures of selection in wild H. zea. Furthermore, we observed evidence of a very strong selective sweep in one region of the H. zea genome, yet this strongest change was not additively associated with Cry resistance. This first, whole genome analysis of field-collected specimens to study evolution of Cry resistance demonstrates the potential and need for a more holistic approach to examining pest adaptation to changing agricultural practices.