Independently evolved and gene flow‐accelerated pesticide resistance in two‐spotted spider mites

Abstract Pest species are often able to develop resistance to pesticides used to control them, depending on how rapidly resistance can emerge within a population or spread from another resistant population. We examined the evolution of bifenazate resistance in China in the two‐spotted spider mite (TSSM) Tetranychus uticae Koch (Acari: Tetranychidae), one of the most resistant arthropods, by using bioassays, detection of mutations in the target cytb gene, and population genetic structure analysis using microsatellite markers. Bioassays showed variable levels of resistance to bifenazate. The cytb mutation G126S, which confers medium resistance in TSSM to bifenazate, had previously been detected prior to the application of bifenazate and was now widespread, suggesting likely resistance evolution from standing genetic variation. G126S was detected in geographically distant populations across different genetic clusters, pointing to the independent origin of this mutation in different TSSM populations. A novel A269V mutation linked to a low‐level resistance was detected in two southern populations. Widespread resistance associated with a high frequency of the G126S allele was found in four populations from the Beijing area which were not genetically differentiated. In this case, a high level of gene flows likely accelerated the development of resistance within this local region, as well as into an outlying region distant from Beijing. These findings, therefore, suggest patterns consistent with both local evolution of pesticide resistance as well as an impact of migration, helping to inform resistance management strategies in TSSM.

. A fundamental question in the evolution of pesticide resistance is the origin of resistance mutations in natural populations (Ffrench-Constant, 2007), whether they emerge within a population from standing genetic variation or as new mutations, versus being introduced into it through migration. Answers to this question can contribute to insecticide resistance management (IRM) programs and theories of adaptive evolution more generally (Daborn & Le Goff, 2004;Hawkins et al., 2018;MacLean, Hall, Perron, & Buckling, 2010;Neve, Busi, Renton, & Vila-Aiub, 2014).
Resistance alleles can be maintained in standing genetic variation in populations before selection for resistance (Barrett & Schluter, 2008) or emerge from new mutations subsequent to a selective challenge (Woods, Schneider, Winkworth, Riley, & Lenski, 2006).
For contemporary evolution occurring on the time frame of less than a few hundred years (Hendry & Kinnison, 1999), standing genetic variation generally contributes more adaptive mutations than new mutations (Hendry et al., 2011). In pest insects, resistance mutations nevertheless can arise from new mutations as evidenced by cases of target-site resistance (Riveron et al., 2014;Weetman et al., 2015), as well as from standing genetic variation (Troczka et al., 2012) or a combination of processes (Hartley et al., 2006;Rose et al., 2011) as summarized by Hawkins et al. (2018).
Multiple origins of resistance mutations are evident from resistance being conferred by the same site mutation in different species/subspecies (Anthony, Brown, Markham, & Ffrenchconstant, 1995;Thompson, Steichen, & Ffrench-Constant, 1993;Weill et al., 2003), and when resistance to the same pesticide involves different mutations in different populations (Andreev, Kreitman, Phillips, & Beeman, 1999). However resistance allele originating from single sites can also be dispersed globally (Daborn et al., 2002;Raymond, Callaghan, Fort, & Pasteur, 1991), particularly in highly mobile pest species in agricultural ecosystems (Cao et al., 2017). The widespread dispersal of resistance alleles can increase the risk of resistance developing in distant populations (Osakabe, Uesugi, & Goka, 2009).
Although the development of resistance to pesticides is well established, less is known about the origin and dispersal of resistance mutations (Hawkins et al., 2018). The genetic basis of resistance needs to be understood to track the origin of resistance mutations (Daborn & Le Goff, 2004;Hawkins et al., 2018). For this reason, most work on the origin of pesticide resistance has been based on target-site resistance (Daborn & Le Goff, 2004;Raymond et al., 1991;Troczka et al., 2012;Weetman et al., 2015).
In this study, we used the two-spotted spider mite (TSSM) Tetranychus uticae Koch (Acari: Tetranychidae), one of the arthropods with very high levels of resistance (Ilias, Vassiliou, Vontas, & Tsagkarakou, 2016), as a model species to investigate the origin and dispersal of bifenazate resistance mutations. Bifenazate (Uniroyal Chemical Company, Inc., USA) is a new type of acaricide for controlling spider mites (James, 2002;Van Leeuwen, Tirry, & Nauen, 2006). It works as a cytochrome b (cytb) Qo-pocket inhibitor, targeting the mitochondrial cytochrome bc1 complex (complex III;Van Leeuwen et al., 2008). This mode of action was recently supported by a study revealing maternally inherited cross-resistance between bifenazate and acequinocyl, an HONQ acaricide (Van Nieuwenhuyse, Leeuwen, Khajehali, Vanholme, & Tirry, 2009). Previous studies showed the resistance of TSSM to bifenazate is mainly related to cytb nonsynonymous mutations including amino acid substitutions of G126S, I136T, and S141F (TSSM numbering) located in the cd1 helix or the P262T mutation near the ef helix aligning the cytochrome bc1 enzyme pocket (Van Leeuwen et al., 2008). A combination of two mutations (G126S and I136T, G126S and S141F) in cd1 helix or the P262T mutation seems to be necessary to confer extremely high resistance in TSSM in the laboratory. The G126S mutation alone only confers a lower level of resistance. However, this mutation may be coupled with a second mutation (I136T or S141F) to cause extremely high resistance.
Bifenazate was released in China in 2013 (Gong et al., 2013). It is a reliable pesticide for the control of TSSM (Xu et al., 2018), which is a serious pest on many crops, especially strawberry. After five years of usage, bifenazate has become less effective in several regions, suggesting that resistance development was in progress. Prior to the usage of bifenazate in the Beijing area, a resistance-related mutation G126S in one individual from 288 individuals was detected in natural populations (Gong et al., 2014), suggesting standing genetic variation for bifenazate resistance in TSSM in this area. Parallel evolution of bifenazate resistance mediated by mutation of cytochrome b was found the citrus red mite, Panonychus citri . The extensive transport of strawberry seedlings across China possibly enables high levels of gene flow among TSSM populations with the potential to spread the resistance mutation.
In this study, we examined the resistance status of TSSMs to bifenazate and detected mutations in the cytb gene in field populations across China. The population genetic structure of TSSMs was investigated based on microsatellites to trace the evolution and dispersal of resistance mutations among populations. We assumed that de novo mutations of TSSMs arising independently in China, based on the presence of resistance mutation prior to the usage of bifenazate (Gong et al., 2014), and parallel evolution of resistance mutation to bifenazate in relative spider mite species , and high genetic structure among populations of TSSM (Chen, Zhang, Du, Jin, & Hong, 2016;Navajas et al., 2002;Sun, Lian, Navajas, & Hong, 2012). The pattern of resistance evolution revealed in our study can help facilitate effective IRM and provides information on processes involved in resistance evolution against this pesticide.

| Sample collection and rearing
In total, ten populations of spider mites were collected from strawberry fields across seven provinces of China from February to May in 2017 (Table 1). When collecting the spider mites, we respectively chose about thirty scattered points from every field to avoid the collection of close relatives. Some of the spider mites collected were preserved in absolute ethanol for molecular analysis; the remaining mites were transferred to bean plants (Phaseolus vulgaris L.) to be cultivated for bioassays in the laboratory. TSSM does not readily move from strawberry leaves onto bean leaves, but once on bean leaves the mites are easily moved for bioassays as described below. A susceptible strain of TSSM maintained at the Institute of Vegetables and Flowers, Chinese Academy of Agricultural Sciences, was used as a control in the bioassays. This population had been reared in the laboratory for ten years without contacting any pesticides. Populations of TSSM were reared at 25 ± 0.5°C with 60% relative humidity and a 16:8 (light:dark) photoperiod.

| Bioassays
The bioassay was carried out using a slide-dip method with 43% bifenazate as described previously (Gong et al., 2013). In brief, TSSM adults were stuck onto one end of a slide with doublesided sticky tape. After 2 hr, inactive and dead individuals were removed with an insect needle, and only the active adult mites remained, leaving 20-30 individuals per slide. Based on preliminary tests, seven concentrations of 43% bifenazate from 15.625 to 1,000 mg/L were diluted using water (containing 0.1% Triton X-100). Water containing 0.1% Triton X-100 was used as a control.
Four replications were conducted for each treatment. After dipping in different treatments for 5 s, the slides with TSSM were dried naturally at room temperature and kept in the Animal Breeding System (LP-80CCFL-6AR/6ARS, NK system) at 25 ± 0.5°C, 60% relative humidity and a 16:8 (light:dark) photoperiod. Mortality was scored after 48 hr.

| Molecular analysis
Ten field populations of TSSM were used for sequencing of mitochondrial genes (both male and female) and genotyping of microsatellite loci (female). A rapid method was used to extract DNA from individual specimens (Mardulyn et al., 2013). Individuals stored in absolute ethanol were air-dried and picked into 0.2-ml tube containing 20 µl of lysis buffer (10 mM of Tris-HCl pH 8.2, 50 mM of KCl, 2.5 mM of MgCl 2 , 0.45% Tween-20, 0.01% gelatin, 60 μg/ml proteinase K) and homogenized using 20-μl pipette tips. After centrifugation, each tube of liquid was then incubated at −80°C for 30 min, at 65°C for 1 hr and finally at 95°C for 15 min. The DNA templates were stored at −20°C prior to usage.
The complete sequence of cytb gene (2,846 bp), which was found as the target gene for bifenazate resistance in TSSM, was amplified and sequenced from 207 adults (Table 1)   for 30 s, and extension at 60°C for 1 min (cox1) or 3 min (cytb), followed by a final extension at 60°C for 10 min. The amplified PCR products were sequenced using a primer walking strategy on an ABI 3730xl sequencer (Applied Biosystems) with the primer of LepF-TU for cox1 and bidirectional primers of cytbdiaR1 (5′-GAAACAAAAATTATTATTCCCCCAAC-3′) and cytbdiaR2 (5′-GGTACARATCGTAAAATTGC-3′) for cytb.
To reveal the population genetic structure of TSSM, we chose five microsatellite loci which had proved to be stably amplified and with a high level of polymorphism (Ge, Sun, Cui, & Hong, 2013). We added a PC-tail (Primer tail C; 5′ CAGGACCAGGCTACCGTG 3′) to the 5′ end of the candidate forward primers to improve amplification efficiency and reduce cost. A fluorescence-labeled PC-tail was added to the PCR volume to form a three-primer amplification system (Blacket, Robin, Good, Lee, & Miller, 2012;Schuelke, 2000).
In total, 180 female adults were genotyped from 10 populations (

| Species identification, mutation, and population genetic diversity analysis
For mitochondrial genes, we checked and revised sequencing using CHROMAS Pro v2.  (Rousset, 2008).

| Population genetic structure analysis
Population genetic structure was analyzed based on microsatellite loci.
First, phylogenetic relationships among the populations were inferred with POPTREE2 (Takezaki, Nei, & Tamura, 2010) using the neighborjoining (NJ) method (Saitou, 1987). Second, population differentiation was identified using the Bayesian analysis of population genetic structure (BAPS) analysis implemented BAPS version 6.0 (Cheng, Connor, Sirén, Aanensen, & Corander, 2013). We performed ten repeat runs of various K values (from 1 to 10). Additionally, the discriminant analysis of principal components (DAPC) was performed using R package adegenet version 2.0.1 (Jombart, 2008). This method does not rely on any biological hypothesis and provides complementary results to BAPS.

| Correlation analysis
Isolation by distance (IBD) analysis was performed to evaluate the correlation of pairwise genetic differentiation (F ST ) and geographic distance in TSSM populations using a Mantel test implemented in the R package ade4 (Elbrecht et al., 2014)

| Gene flow analysis
We used BAYESASS 3.0.4 (Wilson & Rannala, 2003) to calculate the migration rates between population pairs of TSSM. Preliminary runs (10,000,000 steps) were conducted to adjust mixing parameters for allele frequencies and inbreeding coefficients, after which ten longer runs of 100,000,000 steps with different start seeds were performed. The trace outputs of ten longer runs were combined using Tracer 1.6 (Rambaut, Drummond, Xie, Baele, & Suchard, 2018) to calculate mean migration with a burn-in of 50,000,000.

| Varied levels of resistance to bifenazate
Molecular identification based on the cox1 gene indicated that all specimens randomly selected from the tested populations were TSSM. Compared with susceptible population whose LC 50 is 13.12 mg/L, field populations of TSSM had variable levels of resistance to bifenazate (Table 3, Figure 1). Based on the resistance ratio,  Shandong showed compared to susceptible population.

| Resistance mutations in the cytb gene
In total, four haplotypes of cytb gene were identified from 10 populations of TSSM (Figures 2a and 3

| Population genetic structure
We obtain  show deviation on all loci (Table 6).
Phylogenetic analysis showed that five northern populations

| Correlation between genetic differentiation and geographical distance and resistance level
Mantel tests showed there was no correlation between genetic distance and geographic distance (r = 0.13, p = 0.25) or between these measures and resistance to bifenazate (r = −0.149, p = 0.759).
No correlation was found between resistance to bifenazate and the membership coefficient matrices (Q-matrices) when the optimal K was 2 (P1 = 0.089, P2 = 0.4).

| Gene flow
Relatively

| Bifenazate resistance and mutation of cytb gene
Compared with the laboratory-selected resistant strain (BR-VL strain, resistance ratio >164,000; Van Leeuwen et al., 2008), resistance is not extremely high in TSSM populations from China.
However, resistance has developed rapidly, especially in the Beijing and Sichuan regions where there are major areas of strawberry production and nursery activities that need frequent control of TSSM.
Resistance detected in our study is congruent with the poor control of TSSM by bifenazate in these regions.
We found four haplotypes of the target cytb gene with three mutations compared to the susceptible haplotype (Van Leeuwen et al., 2006 with low LC50 values (twofold to fourfold over the susceptible strain) may be caused by metabolic mechanisms or error of bioassay. The resistance ratio can be influenced by the susceptibility strain used for comparison. The susceptible population was reared in laboratory for many rears without any contact with pesticide.
The laboratory adaptation may in the population and lead to increase in susceptibility to pesticides (Hoffmann & Ross, 2018). A novel mutation of A269V was also found in the present study; it is located in the ef helix aligning in cytochrome bc1 complex, though not at the conserved site (Esser et al., 2004). Populations from Anhui and Hunan with the A269V mutation tend to have some resistance to bifenazate, compared to both susceptible strain and three field populations; and the more frequent the A269V mutation, the higher the resistance level. We therefore suspect that the A269V mutation may be associated with a low to medium level of resistance to bifenazate. However, metabolic mechanism may also lead to the resistance of TSSM to bifenazate, which need further validation.

| Standing genetic variation in resistance mutation
The major mutation G126S that confers resistance of TSSM to

| Multiple origins of resistance mutations
A phylogenetic approach has been used to identify the origin of resistance alleles in this species (Hawkins et al., 2018). Here, we used a population genetic analysis to examine the population structure of TSSM and infer the origin of resistance mutations similar to the method of using a haplotype network (Karasov, Messer, & Petrov, 2010 Note. S05, S19, S65, S158, and S167 represent five microsatellite loci, respectively, which are referred in previous work about microsatellite development in TSSM (Ge et al., 2013).
template DNA from each individual. This may bias the estimation of gene flow. However, five microsatellite loci were able to reveal the population genetic structure of a species as in previous study (Meng, Shi, & Chen, 2008). The genetic structure described here is consistent with a previous report that indicates spider mite populations represent metapopulations .
Two major clusters are located in north and south China, with relatively strong genetic differentiation (F ST ) and low gene flow between them. This is consistent with a previous study which found high genetic differentiation and strong evidence for limited gene flow among geographically separated populations (Sun et al., 2012;Xie et al., 2006)  resistance mutation and come from the same genetic cluster. The pairwise F ST values are very low while gene flow estimates are high, ranging from 0.035 to 0.090. Regional dispersal of spider mites can occur from wind-assisted movement and passive transport from human movement including agricultural practices Uesugi, Kunimoto, & Osakabe, 2009). Importantly, strawberry seedling transportation is an essential way of TSSM dispersal and is likely to spread resistance mutations. The location of the BJCW population covers the main production nursery of strawberries, and seedlings are transported to other planting areas around Beijing, promoting resistance dispersal.
Dispersal linked to cross-regional transportation also likely accounts for genetic similarity between Beijing and Sichuan popu- Nevertheless, a high level of gene flow does not necessarily mean an equivalent level of resistance. In particular, gene flow between Sichuan and Anhui was estimated as extremely high (from Sichuan to Anhui the average estimate was 0.069, while in the reciprocal direction it was estimated as 0.092). Yet, resistance to bifenazate was at a different level in these two locations. This may reflect differences in selection pressures in these regions, but this remains to be tested with additional sampling.

| CON CLUS ION
Resistance of TSSMs to the novel acaricide bifenazate is developing in China. Taking advantage of the well-studied genetic basis of bifenazate resistance, we monitored bifenazate resistance among TSSM populations and compared results to the population genetic structure of TSSM populations. The major resistance mutation in TSSM was probably present in populations before pesticide-related selection, although it appears to have evolved independently in populations of TSSM in China. Nevertheless, within a region and likely also across regions, gene flow among populations appears to have accelerated the development of resistance. Our study therefore suggests that the origin and development of pesticide resistance in fields can depend on local selection pressures as well as on movement. These results are relevant to IRM strategies in remaining regions where resistance has not yet developed.
Populations from these regions should be screened extensively for the major mutation affecting resistance. If present, attempts should be made to reduce local selection pressures. If the mutation is not found, a low level of resistance might still be expected, but it becomes important to reduce gene flow from resistant populations from other regions.

ACK N OWLED G M ENTS
We

CO N FLI C T O F I NTE R E S T
None declared.