Summary 1.A high dose ⁄ refuge strategy has been adopted in the USA to manage the risk of Bacillus thuringiensis (Bt) resistance in target pests such as the cotton bollworm (CBW), Helicoverpa zea (Boddie) in transgenic Bt cotton Gossypium hirsutum L. Structured refuges, consisting of non-Bt cotton, have been a mandated part of this strategy to produce non-selected insects that are temporally and spatially synchronous with insects from the Bt crop, diluting Bt resistance alleles through mating. However, the bollworm is highly polyphagous and exploits a large number of crop and weedy hosts concurrently with Bt cotton. 2. A study was carried out in five major US cotton-producing states during 2002 and 2003 using the ratios of 13 C to 12 C in bollworm moths to estimate the proportions of the population originating from C 3 or C 4 plants. A separate study measured gossypol residues in moths from four states in 2005 and 2006, enabling the identification of moths whose natal hosts were cotton rather than other C 3 hosts. 3. C 4 hosts served as the principal source of bollworm moths from mid-to-late June to early September, depending on the state. Beginning in late August ⁄ early September and lasting 1-4 weeks, the majority of moths exhibited isotopic compositions characteristic of C 3 hosts. During this period, however, the minimum percentage of moths that developed as larvae on C 4 hosts was typically >25%. By mid-September and through October and November, the majority of the bollworm population exhibited C 4 isotopic compositions. 4. Between late June and early August, cotton-derived bollworm moths (moths with gossypol residues) comprised <1% of moths in all states, and remained below this level throughout the season in North Carolina. In other states, cotton-derived moths increased between early August and early September to peak at an average of 19AE1% of all moths. 5. Synthesis and applications. Data on 13 C ⁄ 12 C ratios and gossypol residues in CBW moths were used to assess the importance of structured non-Bt cotton refuges for the management of Bt resistance risk in H. zea. Weekly estimates of bollworm breeding on cotton, C 3 plants other than cotton and C 4 plants showed that, throughout the season, the majority of bollworm moths caught in pheromone traps adjacent to cotton fields did not develop as larvae on cotton. This result implies that management practices in cotton such as the use of structured cotton refuges will play a relatively minor role -particularly compared with maize Zea mays L. -in managing potential resistance to Bt cotton in populations of the CBW in the US Cotton Belt.

Laboratory studies were performed to characterize the lepidopteran toxicity of cotton plants expressing two different toxin proteins from Bacillus thuringiensis (Bt), in order to assess insect resistance management implications of a commercial, two‐toxin transgenic cotton. An independent and additive interactive effect of two Bt δ‐endotoxins expressed by the transgenic cotton variety 15985 was demonstrated by examining the responses of Heliothis virescens (F.), Helicoverpa zea (Boddie), and Spodoptera frugiperda (J.E. Smith) larvae to field‐ or greenhouse‐grown tissue from genetic near‐isolines, which expressed Cry1A only, Cry2Ab only, or both toxins. In all cases, the Cry2Ab component was the larger contributor to total toxicity in the two‐toxin isoline. Toxin‐specific, quantitative enzyme‐linked immunosorbent assay (ELISA) tests confirmed that the levels of each toxin in tissues of the two‐toxin isoline were not statistically different (P > 0.05) from the levels found in the corresponding tissues of the respective single‐toxin isoline. Resistance management considerations were discussed. Considering the additive interaction of toxins, a relatively simple insect resistance‐monitoring procedure was proposed for the monitoring of commercial cotton varieties expressing both toxins.

Selection pressure on bollworm, Helicoverpa zea (Boddie) (Lepidoptera: Noctuidae), by cotton, Gossypium hirsutum (L.) (Malvaceae), that produces one or more Bacillus thuringiensis Berliner (Bt) proteins is reduced by plantings of non‐Bt refuge cotton that produce non‐selected individuals. However, the contributions of non‐Bt, non‐cotton crop hosts to the overall effective refuge for H. zea on Bt cotton have not been estimated. A 2‐year, season‐long study was conducted in five US cotton‐producing states to assess the spatial and temporal population dynamics and host use of H. zea. Helicoverpa zea larval estimates in commercial crop fields demonstrated that non‐cotton crop hosts, such as maize, Zea mays L. (Poaceae), grain sorghum, Sorghum bicolor (L.) Moench (Poaceae), peanut, Arachis hypogaea L. (Fabaceae), and soybean, Glycine max (L.) Merrill (Fabaceae), collectively support much larger larval populations than cotton throughout the season. Larval populations were almost entirely restricted to maize in the middle part of the season (June and portions of July), and were observed in non‐cotton crop hosts more frequently and typically in larger numbers than in cotton during the period when production would be expected in cotton (July and August). Numbers of H. zea larvae produced in replicated strip trials containing various crop hosts paralleled production estimates from commercial fields. In contrast, the number of H. zea adults captured in pheromone traps at interfaces of fields of Bt cotton and various crop hosts rarely varied among interfaces, except in instances where maize was highly attractive. With the exception of this early season influence of maize, moth numbers were not related to local larval production. These data demonstrate that H. zea adults move extensively from their natal host origins. Therefore, non‐cotton crop hosts, and even relatively distant hosts, contribute significantly to effective refuge for H. zea on Bt cotton. The results presented here demonstrate that substantial natural refuge is present for Bt‐resistance management of H. zea throughout the mid‐South and Southeast portions of the US cotton belt.

Cry1Ac protoxin (the active insecticidal toxin in both Bollgard and Bollgard II cotton [Gossypium hirsutum L.]), and Cry2Ab2 toxin (the second insecticidal toxin in Bollgard II cotton) were bioassayed against five of the primary lepidopteran pests of cotton by using diet incorporation. Cry1Ac was the most toxic to Heliothis virescens (F.) and Pectinophora gossypiella (Saunders), demonstrated good activity against Helicoverpa zea (Boddie), and had negligible toxicity against Spodoptera exigua (Hübner) and Spodoptera frugiperda (J. E. Smith). Cry2Ab2 was the most toxic to P. gossypiella and least toxic to S. frugiperda. Cry2Ab2 was more toxic to S. exigua and S. frugiperda than Cry1Ac. Of the three insect species most sensitive to both Bacillus thuringiensis (Bt) proteins (including H. zea), P. gossypiella was only three-fold less sensitive to Cry2Ab2 than Cry1Ac, whereas H. virescens was 40-fold less sensitive to Cry2Ab2 compared with CrylAc. Cotton plants expressing Cry1Ac only and both Cry1Ac and Cry2Ab2 proteins were characterized for toxicity against H. zea and S.frugiperda larvae in the laboratory and H. zea larvae in an environmental chamber. In no-choice assays on excised squares from plants of different ages, second instar H. zea larvae were controlled by Cry1Ac/Cry2Ab2 cotton with mortality levels of 90% and greater at 5 d compared with 30-80% mortality for Cry1Ac-only cotton, depending on plant age. Similarly, feeding on leaf discs from Cry1Ac/Cry2Ab2 cotton resulted in mortality of second instars of S.frugiperda ranging from 69 to 93%, whereas exposure to Cry1Ac-only cotton yielded 20-69% mortality, depending on plant age. When cotton blooms were infested in situ in an environmental chamber with neonate H. zea larvae previously fed on synthetic diet for 0, 24, or 48 h, 7-d flower abortion levels for Cry1Ac-only cotton were 15, 41, and 63%, respectively, whereas for Cry1Ac/Cry2Ab2 cotton, flower abortion levels were 0, 0, and 5%, respectively. Cry1Ac and Cry2Ab2 concentrations were measured within various cotton tissues of Cry1Ac-only and Cry1Ac/Cry2Ab2 plants, respectively, by using enzyme-linked immunosorbent assay. Terminal leaves significantly expressed the highest, and large leaves, calyx, and bracts expressed significantly the lowest concentrations of Cry1Ac, respectively. Ovules expressed significantly the highest, and terminal leaves, large leaves, bracts, and calyx expressed significantly (P < 0.05) the lowest concentrations of Cry2Ab2. These results help explain the observed differences between Bollgard and Bollgard II mortality against the primary lepidopteran cotton pests, and they may lead to improved scouting and resistance management practices, and to more effective control of these pests with Bt transgenic crops in the future.