Modeling the evolution of resistance in cotton bollworm to concurrently planted 1 Bt cotton and Bt maize in China 2

30 Background: Transgenic maize expressing toxins derived from the bacterium Bacillus 31 thuringiensis (Bt) may be commercially planted in northern China where Bt cotton has been 32 planted for more than two decades. While Bt maize brings additional benefits for insect control, it 33 complicates the resistance management of cotton bollworm (CBW), Helicoverpa armigera 34 (Lepidoptera, Noctuidae), a common target of Bt cotton and Bt maize. 35 Results: We developed population genetic models to assess the risk of resistance in CBW when Bt 36 cotton and Bt maize are planted concurrently. Model simulations showed that when natural 37 refuges are absent, the time to resistance (TTR) is less than 10 generations in the case of one-toxin 38 Bt cotton and one-toxin Bt maize, but is more than 30 generations in the case of two-toxin Bt 39 cotton and two-toxin Bt maize. The differences in the TTR between the two cases become greater 40 as the proportion of natural refuge increases. Among the parameters we investigated, the fitness 41 cost has a relatively smaller effect on the TTR, while the dominance of resistance and the 42 proportion of natural refuge have a much greater effect. 43 Conclusions: We concluded that planting the first generation Bt cotton with Bt maize could 44 significantly increase the risk of CBW resistance to Bt toxins as compared to planting a pyramid 45 two-toxin Bt cotton. The strategies for reducing the risk of CBW resistance include replacing the 46 one-toxin Bt cotton with a pyramid two-toxin Bt cotton, adopting a pyramid two-toxin Bt maize, 47 and maintaining a sufficient proportion of natural refuges. 52 53 55 56 57

3 Background 58 Transgenic crops producing insecticidal proteins derived from Bacillus thuringiensis (Bt) have 59 become a major strategy to fight key insect pests in agriculture during the past two decades [1][2][3][4]. 60 By 2018, transgenic crops were planted on more than 190 million hectares worldwide [5]. Among 61 all transgenic crops, the majority is cotton, maize and soybean. 62 One of the main threats to the long-term use of Bt crops is that target pests may evolve 63 resistance to Bt toxin. To date, field resistance has been observed in the target pests of both Bt 64 cotton and Bt maize [6][7][8][9][10][11]. Resistance can be affected by various ecological and genetic factors 65 [12]. Among them, the most important factor is excessive planting of Bt plants [13], but other 66 factors such as insect susceptibility to the Bt toxin, dominance of resistance, and strategy of 67 resistance management are also very important [14][15][16][17]. 68 The refuge strategy is one of the general approaches to managing resistance. With this strategy, 69 a proportion of non-Bt host is planted as refuge to maintain susceptible insect populations. 70 Because abundant susceptible insects from refuge can compete for mating with rare resistant ones 71 and produce heterozygous offspring that cannot survive on Bt plants, evolution of resistance is 72 delayed [18]. The applications of the refuge strategies have been documented in literatures, in the 73 cases of structured refuges [14], natural refuges [19,20], and seed mixture [4]. 74 In China, Bt cotton is the only Bt crop that has been commercially planted so far [21]. One of 75 the main targets of Bt cotton is cotton bollworm (CBW), Helicoverpa armigera (Lepidoptera, 76 Noctuidae), a highly polyphagous insect pest that can feed on a number of different agricultural 77 crops. In northern China, CBW has 4 generations per year. The host crops of CBW include cotton, 78 maize, wheat, soybean, peanut, vegetables, and the availability of host crops vary among different 79 generations [22]. For the first generation, wheat is the primary host crop when other major host 80 crops like cotton and maize are absent. For the second through fourth generations, most of major 81 host crops are available. An earlier study has shown that abundant non-cotton host crops in 82 northern China served as natural refuges for CBW and contributed to delaying resistance of CBW 83 to Bt cotton [20]. However, a more recent study has found that resistance to Bt cotton in CBW is 84 accelerated by a dominant resistance allele [23]. 85 Because maize has been one of the main categories of natural refuges for CBW in northern 86 4 China so far, a practical question is what if the conventional maize is replaced with Bt maize in 87 the region? Planting Bt maize will increase the proportion of Bt plants while decrease the 88 proportion of natural refuge, so one can expect the risk of resistance to Bt to increase if the current 89 composition of host plants is not changed. However, it is unknown how such an increase in the 90 risk of resistance to Bt is affected by important factors associated with the Bt cotton and Bt maize 91 varieties that have been or will be planting. These factors include the number and types of toxins 92 contained in Bt cotton and Bt maize, the fitness parameters associated with Bt cotton and Bt maize, 93 the mode of action in resistance to Bt cotton and Bt maize. 94 Simulation models perhaps are the best approach to addessing the questions above. In this paper, a two-locus population genetic model is developed to analyze the resistance 103 evolution of CBW to Bt cotton and Bt maize when the two species of Bt crops are planted 104 concurrently. The model takes into account the actual Bt cotton variety and the existing resistance 105 to Bt cotton. The questions to be addressed include: (1) What are the differences in the risk of 106 resistance between an one-toxin Bt maize (Bt maize-1) and two-toxin Bt maize (Bt maize-2)? (2) 107 How does the preexisting resistance to Bt cotton affect the risk of resistance when Bt cotton and 108 Bt maize are planted together? (3) What are the key parameters that will impact on the risk of 109 resistance? 110

111
The temporal patterns in the frequency of resistance alleles 112 In the case of Bt cotton-1 & Bt maize-1 where resistance is governed by a single locus, the only 113 resistance allele R1 increases in frequency rapidly over time (Fig. 1A). Because of fitness cost, the 114 5 frequency of resistance allele goes to a level less than 100%. The rapid increase in the frequency 115 of resistance alleles is caused by a combination of dominant inheritance, high initial frequency and 116 lack of natural refuge. 117 In the case of Bt cotton-1 & Bt maize-2 where resistance is governed by two loci, the temporal 118 pattern in the frequency of R1 is very different from that in the frequency of R2 (Fig. 1B). The 119 frequency of R1 increases rapidly, similar to that in the single-locus case. Compared to that in the 120 frequency of R1, the frequency of R2 increases much slower. It is worthy to note that the frequency 121 of R1 increases slower than that in the single-locus case even if all parameters related to Locus 1 122 are the same. Namely, there is an interaction between R1 and R2, which results in a slower increase 123 in the frequency of R1. 124 In the case of Bt cotton-2 & Bt maize-1 and of Bt cotton-2 & Bt maize-2, where resistance is 125 governed by two loci, similar interactions between R1 and R2 are observed (Fig. 1C&D). Compared 126 to that in the single-locus case, the frequencies of R1 in all three two-locus cases increase slower. 127 128

The impact of fitness cost 129
To see how the fitness cost at Locus 1 affects resistance evolution, the time to resistance (TTR) is 130 derived when the fitness cost at Locus 1 varies from 0 to 0.5 (Fig. 2). In all four cases, i.e. Bt To see how the dominance at Locus 1 affects resistance evolution, the time to resistance (TTR) is 141 derived when the dominance at Locus 1 varies from 0 to 0.8 (Fig. 3). In all four cases, i.e. Bt Another reason that planting the first generation Bt cotton with Bt maize could increase the 177 risk of cotton bollworm resistance was that the Bt maize and Bt cotton contain a similar Bt toxin. 178 When the two Bt crops contain a similar Bt toxin, resistance is most likely governed by the same 179 locus or loci [32]. In this case, the resistance to Bt maize is superimposed on that to Bt cotton and 180 therefore evolves much faster than without the preexisting resistance to Bt cotton. Our simulation 181 results confirmed the above scenario. In particular, our results showed that the high initial 182 frequency of resistance could result in a rapid increase in the frequency of resistance when Bt 183 maize is planted and the proportion of natural refuge is reduced. 184 Our results showed that planting a pyramid product of two-toxin Bt maize with the first 185 generation Bt cotton could reduce the risk of resistance as compared with planting a single-toxin 186 Bt maize. This is consistent with our previous work in a more general setting [34]. It is also 187 generally consistent with the results in other literatures [35,36]. However, when the first 188 generation Bt cotton is planted, the differences between the two-toxin Bt maize and one-toxin Bt 189 maize are limited. This is because when the first generation Bt cotton is planted, the risk of 190 resistance in the landscape is mainly determined by that in cotton fields, even when a two-toxin Bt 191 maize reduces the risk of resistance in maize fields. 192 Our results showed that planting a two-toxin Bt maize with a two-toxin Bt cotton could 193 substantially reduce the risk of resistance as compared with planting the first generation Bt cotton. 194 When both two-toxin Bt cotton and Bt maize are planted, the risk of resistance in the landscape is 195 determined by two resistance alleles and the risk of resistance in the landscape is low as long as 196 one of the frequencies of the resistance alleles is low [41]. This is exactly the case when both 197 two-toxin Bt maize and Bt cotton are planted. Therefore, planting a two-toxin Bt maize with a 198 two-toxin Bt cotton could effectively counter the risk of resistance in CBW. We divided the host crops for CBW into three groups: cotton seed mixture, maize seed mixture 245 and natural refuge (i.e. non-Bt host crops other than cotton and maize). We assumed that the 246 effective proportions of the three groups are given, which are denoted by 1 , 2 , and = 1 − 247 1 − 2 , respectively. Here the effective proportion is the proportion of planting area weighted by 248 the relative effectiveness in producing susceptible insects [31]. Throughout this article, we 249 referred to the "effective proportion" simply as "proportion" unless mentioned otherwise.

The Bt cotton and Bt maize products 255
We considered two possible products for Bt cotton: a one-toxin product expressing cry1Ac or 256 similar protein and a two-toxin product expressing cry1Ac/cry2Ab or similar proteins, and denoted 257 them by Bt cotton-1 and Bt cotton-2, respectively. We also considered two possible products for 258 Bt maize: a one-toxin product expressing cry1Ab or similar protein and a two-toxin product 259 expressing cry1Ab/cry2Aj or similar proteins, and denoted them by Bt maize-1 and Bt maize-2,

The population genetic equations 265
We developed a general two-Bt-crop two-locus population genetic model to cover the four 266 combinations of Bt cotton and Bt maize mentioned above. In the case of Bt cotton-1 & Bt maize-1, 267 we assumed that the Bt-resistance is governed by a single-locus with two alleles because Bt 268 cotton-1 and Bt maize-2 express similar Bt proteins. In this case, a single-locus model is sufficient, 269 which can be achieved by setting no selection at the second locus in the two-locus model. In the 270 cases of "Bt cotton-1 & Bt maize-2", "Bt cotton-2 & Bt maize-1" and "Bt cotton-2 & Bt maize-2", 271 we assumed that the Bt-resistance is governed by two independently segregated loci with two 272 alleles at each locus because Bt cotton-2 and Bt maize-2 share similar Bt proteins. Because a 273 single-locus model is a special case of the two-locus model, we only described the two-locus 274 model as follows. 275 The two-locus model is a discrete-time, frequency-dependent one in which the frequencies of Where PBt1 is the proportion of Bt in the cotton seed mixture. 298 Similarly, we can obtain the corresponding probabilities for larval movement between maize 299 plants as follows. 300 2, Where PBt2 is the proportion of Bt in the maize seed mixture. 303 We assumed that moths emerged from different host crops mate randomly. This assumption is 304 reasonable because in the study area, host crops are planted by small-holder farmers and it is very 305 common that different host crops are planted side by side in small fields [1]. With the assumption 306 of random mating, the overall fitness of any two-locus genotype G across cotton plants, maize 307 plants and natural refuge is expressed by the following formula: 308 With the fitness function given above, the frequency of any genotype G in the next

Fitness parameters 330
We assumed the fitness of a two-locus genotype is multiplicative with respect to the two loci. 331 With this assumption, we only needed to specify the fitnesses of one-locus genotypes, that is those 332 of , , , where j=1,2 stands for loci 1 and 2, respectively. The fitnesses of one-locus 333 genotypes can be further computed based on fitness cost (c ), dominance of fitness cost (d ), 334 Bt-caused mortality to susceptible CBW (μ ), dominance of resistance (h ), and incomplete 335 resistance (σ ) ( Table 1). Because the Bt proteins in Bt cotton and Bt maize are similar, we 336 assumed that the Bt-caused mortality, dominance of resistance and incomplete resistance are the 337 same between Bt cotton and Bt maize. 338 The first category of fitness parameters that we need to specify is the fitness cost (c ) and 339 dominance of fitness cost (d ) for j=1,2. Among them, c 1 is the fitness cost of resistance to 340 cry1Ac or cry1Ab, while c 2 is the fitness cost of resistance to cry2Ab or cry2Aj. Experimental 341 results have showed that fitness cost of resistance to cry1Ac might be as large as 0.54 [23]. To be 342 conservative, here we used a smaller value of 0.36 as the default but also studied the cases when 343 this parameter varies between 0 and 0.5. Namely, we considered 0 ≤ 1 ≤ 0. of incomplete resistance to cry2Ab or cry2Aj, we adopted a conservative method ([32]) and 366 assumed that the resistance is complete, i.e. 12 = 22 = 0 . All fitness parameters were 367 summarized in Table 2. 368 The larval movement parameter 369 The probability of larval movement between plants depends on several factors, such as the 370 insect's tendency and ability to move, the distance between plants and the growing stage of the 371 plants. So far there is no evidence of significant CBW larval movement among plants. However, 372 because larval movement generally increases the risk of resistance evolution [39], we adopted a 373 conservative approach and considered a 10% larval movement between plants. Namely, we 374 assumed that the probability that a CBW larva moves from one plant to another during the entire 375 larvae stage is 0.1, i.e. m=0.1 (Table 2). 376

Initial frequencies of resistance alleles 377
A study showed that the frequency of resistance allele to Bt cotton in northern China was 0.1 in 378 2016 [23]. We used this value as the initial frequency of resistance allele associated with Locus 1.    Note: The parameters associated with Bt maize ( 2 , 2 and ℎ 2 ) are the same as those 598 with Bt cotton: 2 = 1 , 2 = 1 , ℎ 2 = ℎ 1 , for j=1,2. 599