Cross-crop resistance of Spodoptera frugiperda selected on Bt maize to genetically-modified soybean expressing Cry1Ac and Cry1F proteins in Brazil

Spodoptera frugiperda is one of the main pests of maize and cotton in Brazil and has increased its occurrence on soybean. Field-evolved resistance of this species to Cry1 Bacillus thuringiensis (Bt) proteins expressed in maize has been characterized in Brazil, Argentina, Puerto Rico and southeastern U.S. Here, we conducted studies to evaluate the survival and development of S. frugiperda strains that are susceptible, selected for resistance to Bt-maize single (Cry1F) or pyramided (Cry1F/Cry1A.105/Cry2Ab2) events and F1 hybrids of the selected and susceptible strains (heterozygotes) on DAS-444Ø6-6 × DAS-81419-2 soybean with tolerance to 2,4-d, glyphosate and ammonium glufosinate herbicides (event DAS-444Ø6-6) and insect-resistant due to expression of Cry1Ac and Cry1F Bt proteins (event DAS-81419-2). Susceptible insects of S. frugiperda did not survive on Cry1Ac/Cry1F-soybean. However, homozygous-resistant and heterozygous insects were able to survive and emerge as fertile adults when fed on Cry1Ac/Cry1F-soybean, suggesting that the resistance is partially recessive. Life history studies revealed that homozygous-resistant insects had similar development, reproductive performance, net reproductive rate, intrinsic and finite rates of population increase on Cry1Ac/Cry1F-soybean and non-Bt soybean. In contrast, heterozygotes had their fertility life table parameters significantly reduced on Cry1Ac/Cry1F-soybean. Therefore, the selection of S. frugiperda for resistance to single and pyramided Bt maize can result in cross-crop resistance to DAS-444Ø6-6 × DAS-81419-2 soybean. The importance of these results to integrated pest management (IPM) and insect resistance management (IRM) programs is discussed.

. Percent mortality, stunting (larvae did not reach third instar) and mean weight (mg/larvae) of S. frugiperda strains after five days on leaves of Cry1Ac/Cry1F-soybean (event DAS-444Ø6-6 × DAS-81419-2) and non-Bt soybean (isoline) in laboratory trials. a P-R strain (selected for resistance to Cry1F/Cry1A.105/ Cry2Ab2-maize), H-R strain (selected for resistance to Cry1F-maize), and Sus strain (susceptible of reference). b Values represent means and respective 95% confidence intervals (CIs) or standard error (SE). For each pair of means, those followed by the same letter in a column and S. frugiperda strain are not significantly different due to nonoverlap of 95% CIs. c A separate t-test (P < 0.05) was conducted between Cry1Ac/Cry1F-soybean and the non-Bt soybean for each growth stage and S. frugiperda strain (for each pair of means, those followed by different letter in each column and S. frugiperda strain are significantly different). www.nature.com/scientificreports/ on Cry1Ac/Cry1F-soybean did not reach third instar at five days and larval weights were reduced by more than 50% compared to the same strains on non-Bt soybean. The susceptible strain (Sus) had higher mortality (97 and 82%), stunting and weight reduction on Cry1Ac/Cry1F-soybean than on non-Bt soybean (2 and 8% mortality and stunting, respectively) ( Table 1).
Life history traits of S. frugiperda strains on Cry1Ac/Cry1F-soybean. No significant differences in the duration and survival of egg and pupal stages of homozygous-resistant (P-R and H-R) and heterozygous insects on Cry1Ac/Cry1F-soybean and non-Bt soybean were detected (Fig. 1). However, larval stage duration of P-R and H-R strains were significantly shorter (~ 2 days) on Cry1Ac/Cry1F-soybean, and this also reduced the egg-to-adult period when compared to the same strains developing on non-Bt soybean. Larval and egg-toadult survival of P-R strain were significantly lower on Cry1Ac/Cry1F-soybean, while H-R strain showed higher survival on non-Bt (Fig. 1). For heterozygotes, the duration of larval (25 and 24 days) and egg-to-adult (43 and 38 days) periods was longer on Cry1Ac/Cry1F-soybean than non-Bt (18 and 33 days, respectively) ( Fig. 1). However, the survival of heterozygous larvae on Cry1Ac/Cry1F-soybean was lower than 35%, while on non-Bt was higher than 95%. There was also a reduction in the number of heterozygous insects that completed the life cycle on Cry1Ac/Cry1F-soybean (24% reaching the adult stage), when compared to non-Bt (more than 80% originated adults). In contrast, susceptible insects did not survive until adult stage on Cry1Ac/Cry1F-soybean, while on non-Bt more than 60% developed into adults in 35 days. When fed on the same plant ( Supplementary  Fig. S1), P-R insects had higher survival from neonate to adult on Cry1Ac/Cry1F-soybean (82%) than H-R (67%), heterozygous (28%) and Sus (no survival) insects. On non-Bt soybean, the survival of P-R and heterozygous (higher than 88%) insects were similar, while H-R and Sus insects had a lower survival (58% and 75%, respectively). Based on previous results, the dominance levels (D ML ) of resistance of S. frugiperda on Cry1Ac/ Cry1F-soybean were 0.32 (95% CI 0.29-0.36) and 0.40 (95% CI 0.37-0.43) for P-R and H-R strains, respectively, indicating that the resistance is partially recessive. No significant differences were detected in the larval weights of P-R and H-R insects on Cry1Ac/Cry1Fsoybean and non-Bt soybean ( Table 2). By contrast, pupal weight of resistant insects was significantly heavier on Cry1Ac/Cry1F-soybean than on non-Bt. Heterozygous larvae from both resistant strains presented lower weight on Cry1Ac/Cry1F-soybean, but the progeny from P-R♀ × Sus♂ had similar pupal weight on Bt and non-Bt soybean, while the insects from H-R♀ × Sus♂ had higher pupal weight on non-Bt. P-R females produced similar number of eggs when larvae developed on Cry1Ac/Cry1F-soybean or non-Bt, while H-R females produced more eggs when their development occurred on Bt-soybean. Females from P-R♀ × Sus♂ fed on Cry1Ac/ Cry1F-soybean oviposited less eggs than on non-Bt. On the other hand, females from H-R♀ × Sus♂ presented similar number of eggs on Bt and non-Bt soybean. On the same host plant, larval weight was higher for resistant insects on Cry1Ac/Cry1F-soybean than other strains ( Supplementary Fig. S2). In contrast, on non-Bt soybean heterozygotes had higher larval weight. Pupae were heavier for the H-R insects on Cry1Ac/Cry1F-soybean and H-R♀ × Sus♂ on non-Bt. Resistant and heterozygous females on Cry1Ac/Cry1F-soybean produced a similar number of eggs. However, on non-Bt heterozygous females produced more eggs.
Fertility life table parameters of P-R strain on the mean generation time, net reproductive rate, intrinsic and finite rate of population increase was similar on Cry1Ac/Cry1F-soybean and non-Bt (Table 3). Based on this, after ~ 40 days, 257 and 326 females from each P-R female are expected on Cry1Ac/Cry1F-soybean and non-Bt, respectively. However, H-R females presented higher fertility life table parameters on Bt soybean. For this strain, after ~ 40 days, 297 females/female are expected when feeding on Cry1Ac/Cry1F-soybean, while on non-Bt only 157 females/female in 45 days. By contrast, heterozygotes on Cry1Ac/Cry1F-soybean had their life history parameters negatively affected. For these insects, 94 and 105 females are expected from each female on Cry1Ac/ Cry1F-soybean in 44 to 49 days, while on non-Bt soybean more than 540 females/female would be produced in 39 days. This represents a reduction of 80% in the number of females produced per generation on Cry1Ac/Cry1Fsoybean. When fertility life table parameters were compared in a same host plant, the homozygous-resistant insects on Cry1Ac/Cry1F-soybean presented shortest generation time, better reproductive performance and rate of population increase (Supplementary Table S1).

Discussion
The colonies of S. frugiperda selected for resistance to single-and pyramided-Bt maize technologies showed high survival on Cry1Ac/Cry1F-soybean (event DAS-444Ø6-6 × DAS-81419-2). While quantitative measurements of Cry1Ac/Cry1F Bt protein expression were not collected in our study, observations from previous studies have shown that Cry1Ac/Cry1F protein expression in greenhouse-grown plants does fall within the range of expression observed across field environments (Corteva Agriscience, unpublished data). Measurements of Bt protein expression can vary across environments and be influenced by multiple factors, such as growth stage, position within the plant canopy, and type of plant tissue. The amount of Cry1Ac and Cry1F protein expressed in this soybean technology was reported in the USDA petition for nonregulated status 38 and by De Cerqueira et al. 39 . Therefore, the high survival of resistant strains on Cry1Ac/Cry1F-soybean can be explained by the crossresistance between Cry1 proteins expressed in Bt plants [34][35][36][37] and their low natural susceptibility to Cry1Ac protein as reported in studies with Cry1Ac-cotton 32,40-42 , Cry1Ac-soybean 43,44 , and diet bioassays containing Cry1Ac 43,45 . The cross-resistance among Cry1 proteins is attributed to their similar amino acid sequence 36 and also their same binding sites in the midgut of S. frugiperda 35 . Previous studies also showed that S. frugiperda resistant to Bt maize survived on single and pyramided Bt cotton 32,33 , indicating cross-crop resistance. Our results also revealed that homozygous-resistant insects had similar development and reproductive performance on Cry1Ac/Cry1Fsoybean and non-Bt soybean. These finding indicate that resistant insects have no adaptive disadvantage in the absence of the selection agent, maintaining the resistance frequency in the field 46  www.nature.com/scientificreports/ By contrast, heterozygous insects showed lower survival than homozygous-resistant insects on Cry1Ac/ Cry1F-soybean, but produced fertile adults. The survival of heterozygous insects on Cry1Ac/Cry1F-soybean indicated that the dominance levels of resistance is characterized as partially recessive for both resistant strains    www.nature.com/scientificreports/ In the current Brazilian crop production landscapes, with successive cultivation of maize, cotton and soybean, S. frugiperda populations are exposed to high selection pressure for resistance to Bt proteins. Resistance has been observed in S. frugiperda field populations to several Bt proteins expressed in maize (i.e. Cry1F, Cry1Ab and Cry1A.105) in Brazil [25][26][27] . Currently, field populations of S. frugiperda are composed predominantly of insects carrying Cry1Ab, Cry1F, and Cry1A.105 resistance alleles 26,28,[49][50][51] , reflected by the increasing use of insecticide applications in fields cultivated with crops expressing these proteins 52 . For example, on single or pyramided Bt maize technologies expressing Cry1 and Cry2 proteins up to four insecticidal sprays may be needed to manage S. frugiperda, under extreme infestations 52,53 . Based on this, it is expected that Cry1Ac/Cry1F-soybean may not provide stand-alone protection against S. frugiperda under Brazilian field conditions, making this species a non-target pest of this Bt technology. However, Cry1Ac/Cry1F-soybean was developed and does provide efficacy against the key soybean pests (A. gemmatalis, C. includens, and C. virescens) under field conditions in Brazil 13,14 , which are the driver pests for the development of Bt traits in soybean. Therefore, to maintain the effectiveness of Cry1Ac/Cry1F-soybean over time against the key soybean pests, the adoption of a structured refuge (20% of cultivated area should be planted with non-Bt soybean) will be important for delaying or preventing resistance evolution 33,36,47 .
According to our results, alternative IPM strategies will be necessary to control S. frugiperda on Cry1Ac/ Cry1F-soybean. Therefore, monitoring the presence of larvae and the damage to Cry1Ac/Cry1F-soybean are essential for supporting decision making regarding the use of other IPM tactics. The use of chemical insecticides probably will be the main tactic against S. frugiperda on Bt soybean. However, Cry1Ac/Cry1F-soybean could also be integrated with biological control agents as baculovirus-based insecticides (e.g. Spodoptera frugiperda multiple nucleopolyhedrovirus-SfMNPV) 54 and natural enemies 55 . In summary, IPM and IRM programs that integrate multiple control tactics with diverse mortality factors, rather than just relying on wide scale use of single control tactics like Bt crops, are needed to ensure the sustainability of Bt crops 56 in Brazil, where the resistance of S. frugiperda to Cry1 Bt proteins is already widespread.

Methods
Description of S. frugiperda strains. Two putative S. frugiperda resistant colonies were selected from a field population collected in maize in Paulínia, São Paulo, Brazil (22° 42′ 38′′ S and 47° 06′ 26′′ W) using the F 2 screen method developed by Andow and Alstad 57 . The selection and rearing of resistant colonies was described in detail by Muraro et al. 53 . The homozygous-resistant strains used in this study were H-R (selected for resistance to Cry1F-maize) and P-R (selected for resistance to Cry1F/Cry1A.105/Cry2Ab2-maize). We also used a strain of S. frugiperda that has been maintained in the laboratory since 2012 without exposure to Bt proteins. This population was collected in non-Bt maize during the 2011-2012 crop season in Mogi Mirim, São Paulo, Brazil (22° 28′ 31′′ S and 46° 54′ 21′′ W). We refer to this colony as a susceptible strain (Sus). To evaluate putative heterozygous insects, the crossing between resistant♀ × susceptible♂ were performed. We only used heterozygotes from this cross because inheritance of resistance is autosomal inherited, and heterozygous larvae have demonstrated similar mortality-response to Bt proteins in diet and leaf bioassays 25,27-29,33,53 . Soybean plants. Seeds from Cry1Ac/Cry1F-soybean and non-Bt soybean (isoline) (maturity group 5.0) were sown in 12-l plastic pots (four seeds/pot) in a greenhouse. Before the bioassays, Bt and non-Bt plants were tested for Bt protein expression using detection kits for Cry1Ac and Cry1F (Envirologix, QuickStix).

Plant tissue bioassays.
Bioassays were performed with soybean leaves from Cry1Ac/Cry1F-soybean and non-Bt soybean (isoline) in V 5-6 and R 4-5 growth stages. Leaves were removed from the upper third part of the plants and, in the laboratory, were placed on a gelled mixture of agar-water at 2.5% in 100 ml plastic cups. Subsequently, neonates from resistant or susceptible strains or their F 1 hybrid were placed on each cup. Cups were sealed and maintained in a room at 25 ± 2 °C, 60 ± 10% RH, and a photophase of 14 h. The experimental design was completely randomized with 10 replicates of 10 neonates/strain/growth stage. Mortality, stunting (larvae that did not reach the 3rd instar) and weight were assessed after 5 days.
Life history traits of S. frugiperda strains on Cry1Ac/Cry1F-soybean. Homozygous-resistant, heterozygous or susceptible neonates were reared on leaves of Cry1Ac/Cry1F-soybean and non-Bt soybean (isoline) excised from greenhouse-grown plants at the R 1 growth stage. In the laboratory, leaves were cut into pieces and placed on a gelled mixture of agar-water at 2.5% in 50 ml plastic cups. Then, a single neonate (6 replicates of 10 neonates/strain/treatment) was placed in each cup. Leaves were replaced every 48 h and cups were maintained in the same environmental conditions described above. The following life history traits were evaluated: duration and survival of egg, larva, pupa and total cycle periods (egg-to-adult); larval weight at 14 days; pupae weight 24 h after pupal formation; and number of eggs per female. The number of eggs were assessed daily from 18 couples kept in PVC cages (23-cm height × 10-cm diameter) internally coated with a paper towel and closed at the top with a voile-type fabric. To determine the embryonic period and survival, 100 eggs of the 2nd oviposition were obtained from each couple. The eggs were observed daily and the number of hatched larvae was counted. Data analysis. The number of insects tested, dead and those did not develop to 3rd instar on Cry1Ac/ Cry1F-soybean and non-Bt soybean (isoline) were used to estimate 95% confidence intervals (CIs) for the probability of mortality and stunting, according to a binomial distribution. For these analyses, the function binom. probit from the package binom in R 3.1.0 (R Development Core Team 2014) 58,59 was used. Percent mortality and stunting were considered significantly different when the 95% CIs on Bt-soybean did not overlap the 95% CIs on non-Bt soybean. The life history data of S. frugiperda strains on Bt and non-Bt soybean met the assumptions