Genetic structure and insecticide resistance characteristics of fall armyworm populations invading China

Abstract The rapid wide‐scale spread of fall armyworm (Spodoptera frugiperda) has caused serious crop losses globally. However, differences in the genetic background of subpopulations and the mechanisms of rapid adaptation behind the invasion are still not well understood. Here we report the assembly of a 390.38‐Mb chromosome‐level genome of fall armyworm derived from south‐central Africa using Pacific Bioscience (PacBio) and Hi‐C sequencing technologies, with scaffold N50 of 12.9 Mb and containing 22,260 annotated protein‐coding genes. Genome‐wide resequencing of 103 samples and strain identification were conducted to reveal the genetic background of fall armyworm populations in China. Analysis of genes related to pesticide‐ and Bacillus thuringiensis (Bt) resistance showed that the risk of fall armyworm developing resistance to conventional pesticides is very high. Laboratory bioassay results showed that insects invading China carry resistance to organophosphate and pyrethroid pesticides, but are sensitive to genetically modified maize expressing the Bt toxin Cry1Ab in field experiments. Additionally, two mitochondrial fragments were found to be inserted into the nuclear genome, with the insertion event occurring after the differentiation of the two strains. This study represents a valuable advance toward improving management strategies for fall armyworm.


| INTRODUC TI ON
The fall armyworm, Spodoptera frugiperda (J.E. Smith), is a polyphagous pest that is native to tropical and subtropical America, with a strong capacity for migration and reproduction (Johnson, 1987;Mitchell et al., 1991;Westbrook, Nagoshi, Meagher, Fleischer, & Jairam, 2016). It was first detected in Africa in 2016 (Goergen, Kumar, Sankung, Togola, & Tamò, 2016) and spread to 44 African countries within 2 years. It was detected in India in 2018, and has now spread to several southeastern Asian countries (Nagoshi et al., 2020). Such rapid spread poses a global threat to food production. The strong environmental adaptability of fall armyworm is not only reflected in its polyphagy for a wide range of host plants (Luginbill, 1928), but also in its evolution of resistance to chemical pesticides and genetically modified crops expressing Bacillus thuringiensis (Bt) toxins (Bernardi et al., 2015;Leibee & Capinera, 1995;Monnerat et al., 2015;Signorini et al., 2018;Storer et al., 2010). Studies have shown that gene families related to detoxification and metabolic processes in the fall armyworm have clearly expanded (Gouin et al., 2017;Liu et al., 2019). In addition, there are two morphologically identical, but genetically distinct, subpopulations or strains of fall armyworm, the rice-strain (R-strain) and the corn-strain (C-strain), which differ in their host plant selection and sex pheromone composition (Lima & McNeil, 2009;Pashley, 1986;Pashley, Hammond, & Hardy, 1992;Pashley & Martin, 1987). However, there is no absolute mating barrier between the two strains and productive hybridization has been confirmed in both laboratory and field studies (Dumas et al., 2015;Nagoshi, Meagher, Nuessly, & Hall, 2006).
To date, several field-evolved resistant populations of fall armyworm have been detected, including those displaying resistance to a variety of chemical pesticides and Bt crops (Chandrasena et al., 2018;Gutiérrez-Moreno et al., 2019;Zhu et al., 2015). The reported mechanisms of resistance to pesticides are mainly due to variation in receptor genes, such as amino acid changes in the ryanodine receptor (RyR) (diamide), acetylcholinesterase (AChE) (organophosphate) and voltage-gated sodium channel (VGSC) (pyrethroids) (Boaventura et al., 2020;Carvalho, Omoto, Field, Williamson, & Bass, 2013;Yu, Nguyen, & Abo-Elghar, 2003). In addition, the frameshift mutation resulting in early termination of the ATP-dependent Binding Cassette subfamily C2 gene (ABCC2) gene, caused by a 2-bp insertion, is linked to resistance to Bt toxin Cry1Fa . Field-evolved strains resistant to Bt toxin Vip3Aa20 were obtained by screening homozygous resistance loci in F 2 generations in the laboratory (Yang et al., 2018). Clarifying the development of pesticide-and Bt-resistance in fall armyworm would be helpful in providing scientific support for the commercialization of genetically modified crops and Bt biopesticides. resistance to conventional pesticides is very high. Laboratory bioassay results showed that insects invading China carry resistance to organophosphate and pyrethroid pesticides, but are sensitive to genetically modified maize expressing the Bt toxin Cry1Ab in field experiments. Additionally, two mitochondrial fragments were found to be inserted into the nuclear genome, with the insertion event occurring after the differentiation of the two strains. This study represents a valuable advance toward improving management strategies for fall armyworm.

K E Y W O R D S
gene insertion, population structure, resistance risk, Spodoptera frugiperda, subpopulations accurately assign the strain genetic background (Juárez et al., 2014;Nagoshi, 2019;Nagoshi, Goergen, Goergen, Du Plessis, van den Berg, & Meagher, 2019;Nagoshi et al., 2017). The dominant populations of fall armyworm invading Africa and Asia were speculated to be hybrid populations based on these two molecular markers (Zhang et al., 2019). In addition, an Africa-specific haplotype, different from those native to the Americas, was also reported in African and Chinese samples based on the Tpi gene Nagoshi et al., 2019), which makes strain identification and studies of population genetic structure more complicated. Therefore, a genome-wide analysis of the genetic characteristics of invasive fall armyworm is becoming imperative. Although several versions of the fall armyworm genome have now been published (Gouin et al., 2017;Kakumani, Malhotra, Mukherjee, & Bhatnagar, 2014;Liu et al., 2019;Nam et al., 2019;Nandakumar, Ma, & Khan, 2017), a high-quality genome assembly from a different geographical source is a valuable addition to the genomic resources for this species.
Moreover, the different biological properties of the C-and R-strains and the debate regarding strain identification will benefit from further genomic support and explanation. Here we report a chromosome-level genome sequence of a male moth from an inbred fall armyworm strain, which derived from field populations collected in Zambia in 2017 and would be classed as C-strain based on COI genotype but possessed an Africa-specific Tpi haplotype which differs from the Western Hemisphere (henceforth American) R-and C-strain. We also resequenced 103 fall armyworm samples from 16 Provinces in China, as well as four samples collected from two African countries (Zambia and Malawi). The genome-wide genetic backgrounds of the invading fall armyworm samples were compared, and insecticide-resistance risk was assessed based on analysis of potential resistance-related genes. Comparative genomic analyses of these data will help to reveal the resistance-related mechanisms and the population genetic characteristics of fall armyworm, which may facilitate its future management.

| Samples and sequencing for genome assembly
The fall armyworm samples were collected from maize fields in Lusaka, Zambia, in 2017 and reared to produce an inbred strain.
One male moth, derived from seven successive generations of single-pair sib mating, was selected for genomic sequencing for the primary assembly data set and all other individuals used in the Hi-C and RNAseq experiments were from the same inbred strain.
DNA was extracted using the Qiagen Genomic DNA kit (Cat. no. 13323, Qiagen) followed by purity assessment and quantification with a NanoDrop One UV-Vis spectrophotometer (Thermo Fisher Scientific) and Qubit 3.0 Fluorometer (Invitrogen), respectively. About 0.5 μg genomic DNA (gDNA) was used as input to generate a PCR-free Illumina genomic library using the Truseq Nano DNA HT Sample preparation Kit (Illumina), with 350-bp insert size and this library was sequenced in 2 × 150-bp format on the Illumina NovaSeq 6000 platform. Five micrograms of gDNA from the same individual was used as an input for ~20-kb insert libraries (SMRTbell Template Prep Kit 1.0, Cat. no. 100-259-100, PacBio) sequenced on the PacBio Sequel (Pacific Biosciences). Two third-instar larvae were selected for Hi-C library construction, and nuclear DNA was cross-linked in situ, extracted and then digested with the restriction enzyme DpnII.
Hi-C libraries were amplified by 12-14 cycles of PCR and sequenced on the Illumina NovaSeq 6000 platform with 2 × 150-bp reads. In addition, three fifth-instar larvae, three pupae, three female moths and three male moths were used for RNA sequencing. Total RNA was extracted using the RNeasy Mini extraction kit (Qiagen), and a NanoPhotometer spectrophotometer (Implen) and Qubit 2.0 Flurometer (Life Technologies) were used to check the purity and concentration of RNA, respectively. One microgram total RNA per sample was used to make indexed cDNA libraries using the NEBNext Ultra RNA Library Prep Kit for Illumina (NEB) following the manufacturer's recommendations. The libraries had insert sizes of 250-300 bp and were sequenced on the Illumina NovaSeq 6000 platform with 150-bp paired-end output.
The filtered Illumina reads were aligned to the assembled contigs by bwa mem version 0.7.17 , and single base errors in the contigs were corrected by pilon version 1.21 (Walker et al., 2014).

| Genome estimation and evaluation
A distribution analysis of 17 k-mer frequencies was performed to estimate the genome size of fall armyworm. The filtered Illumina reads were used as input to construct k-mer frequencies by jellyfish (https://github.com/gmarc ais/Jelly fish). Genome size was estimated using G = K_num/K_depth, where the K_num is the total number of K-mers, and K_depth is the frequency occurring more frequently than the others (Li et al., 2010). We used the arthropoda gene set

| Chromsome assembly based on Hi-C data
The Hi-C sequencing raw reads were filtered to remove reads containing <5 bases of adaptor sequence; >50% of bases with phred quality value of <19; and <5% of unknown bases (N). Filtered reads were then aligned to the assembled contigs using bowtie2 (version 2.2.3; http://bowti e-bio.sourc eforge.net/bowti e2/index.shtml) (Langmead & Salzberg, 2012). Invalid read pairs were filtered using default settings by hic-pro (version 2.7.8; https://github.com/nserv ant/HiC-Pro) (Servant et al., 2015). lachesis (https://github.com/ shend urela b/LACHESIS) (Burton et al., 2013) was applied to cluster, order and orient contigs based on the agglomerative hierarchical clustering algorithm. For each chromosome cluster, the ordered contigs were oriented by building a weighted, directed acyclic graph (WDAG). The orientation of each contig in each chromosomal group was predicted according to the maximum-likelihood path through WDAG. Finally, we cut the chromosomes predicted by lachesis into bins of equal length (100 kb) and constructed a heatmap based on the interaction signals revealed by valid mapped read pairs between bins using hic-pro.

| Gene prediction and annotation
A de novo repeat library of fall armyworm was constructed by repeatmodeler version 1.0.4 (http://www.repea tmask er.org/Repea tMode ler.html). Transposable elements (TEs) were identified by repeatmasker version 4.0.6 (http://www.repea tmask er.org/) using both the de novo library and Repbase library (Repbase-20150923), and tandem repeats were predicted using tandem repeats finder (Benson, 1999) version 4.07b. We used a combination of ab initio prediction, homology searches and RNA-seq annotation to predict genes in the Spodeptera frugiperda genome. We performed ab initio prediction using augustus 2.5.5 with default parameters (Stanke & Waack, 2003). For homology-based annotation, we queried the S. frugiperda genome sequences against a database containing nonoverlapping protein sequences from closely related species (Bombyx mori, Helicoverpa armigera, Spodoptera litura) by genblasta with default parameters (She, Chu, Wang, Pei, & Chen, 2009). genewise (Birney, Clamp, & Durbin, 2004) was used to refine the genblasta mappings to the genome. For the RNA-seq annotation, the RNA-seq data were mapped to the assembled genome of S. frugiperda using tophat version 2.0.12 and alignments were processed by cufflinks version 2.2.1 with default parameters to generate transcript predictions (Trapnell et al., 2012). evidence modeler (Haas et al., 2008) version 1.1.1 was used to combine ab initio predictions, homology-based searches and RNA-seq alignments. Predicted gene models supported by at least one of the annotations using the UniProt database, NR database and RNA-seq data were retained. Gene functional annotation was performed by aligning the predicted protein sequences to the NCBI NR, UniProt, eggNOG, and KEGG databases with blastp version 2.3.0+, apply- ing an E-value cut-off < 10 −5 .

| Phylogenetic tree construction and genomic comparison
Orthologous and paralogous gene families identified in a set of 10 species (Drosophila melanogaster, Plutella xylostella, Bombyx mori, Manduca sexta, Danaus plexippus, Heliconius melpomene, Operophtera brumata, Helicoverpa armigera, Spodoptera frugiperda, Spodoptera litura) with published genomes were analysed by orthofinder version 2.3.1 with default parameters. Orthologous groups that contain single-copy genes for each species were selected to construct the phylogenetic tree. The multisequence alignment of proteins was accomplished by muscle (Edgar, 2004) version 3.8.31. A neighbourjoining (NJ) phylogenetic tree was constructed using mega version 7.0.14. The current assembled genome was aligned with two published versions of fall armyworm genomes using the mummer3.23 (Kurtz et al., 2004) package with cutoff of identity >80% and coverage >80%. Alignments were filtered to generate a multi-alignment data set using the delta-filter utility with 85% minimum identity (-i 85) and minimum alignment length 10 (-l 10). A set of unique alignments was created using the same filter criteria but with the addition of the -r and -q flags.

| Sampling for resequencing and population genetic study
A total of 103 Chinese fall armyworm samples were used for resequencing. All samples were collected as larvae on maize or sugarcane from 50 cities of 16 provinces (autonomous regions or municipalities) of China. The larvae were fed with fresh maize leaves and brought back to the laboratory under ambient conditions during transportation. Larval bodies were cleaned and then stored in a freezer at −80°C. Detailed sample information is presented in Table S1 and the sample distribution in China is shown in Figure S1.
In addition, four fall armyworm samples from Africa were also used for resequencing, including two samples (AFR4 and 5) from the same inbred strain (AFR2017) as the genome sequencing in this study, and another two samples (AFR14 and 15) which were collected from maize fields in Bvumbwe, Malawi, in January 2019, which is also an inbred strain (AFR2019) reared in the laboratory. A total of 1.5 μg gDNA of each sample was used to construct a 350-bp insert library using the Truseq Nano DNA HT Sample preparation Kit (Illumina) sequenced in 150-bp paired-end mode as described in section 2.1.
Raw reads were aligned to the NCBI NT database using blastn, and reads with significant matches (identity > 95% and coverage > 80%) to microbes or host plants were removed. PCR was performed at 94°C for 5 min, 34 cycles of 94°C for 30 s, 60°C for 30 s and 72°C for 30 s, and finally 72°C for 5 min. A total of 10 µl of the PCR products containing the target fragment were sequenced by Life Technology. These samples were collected from the field as larvae or adult moths. Detailed sample information is presented in Table S2 and the sample distribution in China is shown in Figure S1. Mitochondrial COI and Tpi markers were used for strain identification. ABCC2 and AChE genes were detected based on primers designed according to published mutation sites Carvalho et al., 2013). Inserted mitochondrial fragments in the nuclear genome were detected using primers designed in this study. All primer sequence information in this study is shown in Table S3.

| Read mapping and SNP calling
The Illumina raw reads from resequenced samples were filtered using clean _ adapter and clean _ lowqual software as described in section 2.1, resulting in high-quality reads with an average error rate of <0.01. The high-quality reads were then aligned to the fall armyworm reference genome (American C-strain) and mitochondrial genome sequences using bwa mem software  version 0.7.5a with default parameters. Alignments for each sample were processed by removing duplicate reads using the samtools

| Bioassays of insecticides and Bt maize in the field
Bioassays were conducted by a topical application procedure (Armes, Jadhav, Bond, & King, 1992). Two inbred strains (cdcc and cdyc) collected from Yunnan Province and reared for multiple generations in the laboratory were tested using 14 types of pesticide commonly used in agricultural production (Table S4) Larvae were considered dead if they were unable to move in a coordinated manner when prodded with a small soft brush. We used median lethal doses (LC 50 ) to evaluate the resistance level of different fall armyworm populations. The LC 50 and 95% fiducial limit (FL) for each insecticide were estimated by probit analysis using the software package polo-pc (Russell, Robertson, & Savin, 1977)   Comparative analysis of orthogroups of nine Lepidoptera species and Drosophila melanogaster (Diptera) was performed (Table S7).

| High-quality genome assembly of fall armyworm
Among them, 17,180 genes in 10,755 orthogroups were found in the current genome of fall armyworm, and the remaining 5,080 lineage-specific genes were identified as unassigned genes. Compared with Spodoptera litura, S. frugiperda has more species-specific genes, and the number of unassigned genes is much greater than that of S. litura (Figure 2a). Phylogenomic analyses of the 10 species were conducted using 1,571 single-copy genes. As shown in Figure

| Genetic background of fall armyworm populations in China
A total of 103 fall armyworm samples from China were resequenced, as well as four samples from two countries in Africa (Zambia and Malawi). The generated Illumina data ranged from 8.6 to 18.9 Gb for each sample, with a median genome coverage of 32.5×. First, we analysed the whole mitochondrial genome sequences of all samples.
A total of 208 SNP loci were selected for analysis, based on comparison of the published mitochondrial sequences of both the American R-strain (AXE) and C-strain (ASW) (Gouin et al., 2017). Genotypes were obtained at these 208 sites for each individual after mapping the filtered sequence reads to the assembled mitochondrial genome.
We found that most of the samples were assigned to the R-strain, and all four samples from Africa were C-strain, while only four out of 103 samples in China were assigned to the C-strain based on the mitochondrial genome (Figure 3a). Note that most R-strain samples surprisingly contain heterozygous mitochondrial SNPs, which could be caused by inserted C-strain fragments or existing standing variation of low frequency. The proportion of the C-strain in this sample set was ~10% and was similar to that of the 173 Chinese fall armyworm samples identified by PCR based on the COI gene in this study (Table S2).
Next, we analysed the Tpi gene, which is commonly used in strain identification of fall armyworm (Nagoshi, 2012). By comparing the fulllength Tpi gene of the American R-strain (AXE) and C-strain (ASW), 22 SNP loci were found. The genotype of each individual was analysed based on these 22 sites. The results showed that all fall armyworm samples collected from China contained more C-strain SNP loci, as did the Malawi samples (AFR14, AFR15), but not those from Zambia samples (AFR4, AFR5) which represents the Africa-specific haplotype and which contained ~50% of R-strain SNP loci. Genotypes of seven Chinese samples were identical to the American C-strain (ASW) and the remaining samples contained a small proportion of R-strain genotypes or heterozygous SNPs (Figure 3b). However, none of the samples was found to be identical to the American R-strain genotype (AXE).
We further used PCR to analyse genotypes of 173 samples based on 10 strain-biased SNPs within the Tpi gene reported previously (Nagoshi, 2012). The results showed that almost all of the samples correspond to C-strain genotypes, although three samples (G-GXW11, G-GXW13, G-EP6) were identified as an Africa-specific haplotype, which was significantly different from known R-or C-strain genotypes ( Figure 4; Table S2). In summary, our genotyping results show that there are obvious contradictions between strain identification using mitochondrial and Tpi gene markers.

| Fall armyworm is developing a high risk of resistance to conventional pesticides
Insecticide resistance evolution is one of the most challenging problems in the control of fall armyworm. Identifying resistance-related genes is helpful for the monitoring and prevention of fall armyworm outbreaks.
We selected 14 previously reported resistance-related genes of lepidopteran pests and scanned the resequenced samples to analyse variation in F I G U R E 1 A genome-wide contact matrix from Hi-C data between each pair of the 31 chromosomes [Colour figure can be viewed at wileyonlinelibrary.com] target genes. The results showed that all the target genes had multiple variation sites with a high frequency of SNPs in the CDS region (Table S8).  be considered as the susceptible baseline (Bird, 2015) (Figure 6).
Resistance levels of the two populations to pyrethroids and organophosphate pesticides were very high; in particular, resistance ratios to chlorpyrifos of the two populations were more than 300-fold compared to a laboratory susceptible fall armyworm population that was sampled in 1975 (Yu, 1991) (Figure 5b). These results provide a susceptible baseline for fall armyworm populations invading China to different pesticides, which can provide guidance for resistance monitoring and pesticide management strategies.

Africa-specific
Intron2 I ntron3 Intron4  (Table S8), no reported resistant mutation was found in any target resistance genes.
Field tests showed that fall armyworm samples invading China were sensitive to GM maize expressing Cry1Ab compared with the control group. Damage assessment on larval density, the percentage of damaged plants and average damage ratings of GM maize were significantly lower than those of the control group (Figure 5c), indicating that the GM maize expressing Cry1Ab currently has good control effects on the invading population of fall armyworm in China.

| Insertion of mitochondrial fragments into the nuclear genome in a recent evolution event
We found that two mitochondrial fragments, with sequence lengths of 1.5 kb (partial COI gene and NADH2 gene) and 1.6 kb (partial NADH2 gene and 12S rRNA gene), were inserted into the nuclear genome,  (Table S2). At the same time, the resequencing data of 107 fall armyworm samples in this study also showed that there were varying numbers of reads covering the four junction points in 29 samples, and the percentage of samples with inserted reads was 27.1% (Table S9). Both the PCR and resequencing results showed that the insertion was not present in all samples, perhaps suggesting that it has a recent evolutionary origin.
Moreover, the genotype of the two inserted mitochondrial fragments was identical to that of the C-strain, indicating that the insertion occurred after differentiation of the R-and C-strains. Further analysis indicated that the two mitochondrial fragments were inserted into the intron region of the lysine-specific demethylase 3 B (Kdm3B) gene, which is not likely to affect expression of the gene.
The inserted partial COI and NADH2 gene fragments were also considered likely to be functionless.

| D ISCUSS I ON
The rapid spread of the fall armyworm has attracted popular attention worldwide. Accurate identification of its genetic characteristics (strain and pesticide resistance properties) has a direct and practical importance in terms of risk assessment and control strategies. A genome-wide analysis can reveal more in-depth genetic information than conventional gene-level analyses. org/fall-armyworm). The established R-strain fall armyworm in the Americas mainly feeds on turf grass, and there were few reports of damage to rice in 1970s (Bowling, 1978;Gallego, 1967). In addition, the established R-strain Tpi genotype has not been detected in any of the samples collected from Africa or Asia. We therefore speculate that the American R-strain fall armyworm did not invade Africa or According to our results, commonly used strain identification of fall armyworms by mitochondrial or Tpi markers is limited or even inaccurate. The nuclear insertion of two C-strain partial COI fragments in this study further underlines the need for caution in interpreting mitochondrial genotypes. We also found that the AT/GC SNP located at Tpi-intron3 (P173/174) was inadequate as a diagnostic marker. In addition, the TT/CC SNP located at Tpi-exon4 (P379/385) was associated with sequence variation in Tpi-intron4 (Figure 4; Figure S2), which could further be developed as a marker to subdivide C-strain samples.
It is noteworthy that a particular (Africa-specific) haplotype of the Tpi gene originally identified in Africa was tentatively designated as R-strain based on the E4 183 site (equal to P370 in Figure 4 in this study) in previous studies (Nagoshi, 2012). Our genome-wide SNP analysis revealed that this haplotype contained more C-strain SNPs than R-strain SNPs.
The sample used for the genome sequencing in this study represents a combination of the particular Tpi haplotype and C-strain COI. We also found combinations of the R-strain COI and particular Tpi (sample G-XW13), as well as heterozygous forms of the particular Tpi and Tpi-C with the R-strain COI in two samples (G-GXW11, G-EP6). These combinations of different genotypes show that the genetic boundaries between two established (American) R-and C-strains are obscure. The insertion of two mitochondrial fragments into the nuclear genome might be caused by random hybridization between different genotypes, which would suggest that fall armyworm invading China might be descendants of an interstrain hybrid population. This is the first report of DNA fragments transferred from mitochondria into the nuclear genome in a Spodoptera lineage, and two such fragments could be used to develop markers to identify specific populations and to follow further evolutionary events of fall armyworm.
The rapid evolution of insecticide resistance and the increasing levels of resistance observed in fall armyworm populations needs attention. In this study, reported mutations related to insecticide resistance were detected in the AChE gene. Although some mutation sites were detected as heterozygous in most samples, the frequency of resistant mutation sites will increase greatly under the selection pressure caused by application of related pesticides in the field. The bioassay results showed that armyworms invading China have evolved high levels of resistance to organophosphate pesticides, which was consistent with the results of molecular scanning of resistance-related genes, yet the resistance to pyrethroid pesticides cannot be explained by any reported mechanism. However, the fall armyworms invading China are currently sensitive to GM maize expressing Cry1Ab in field experiments, and are also sensitive to other Bt toxins in the laboratory, according to previous studies . At present, GM maize shows better application prospects in controlling fall armyworm in China, as larval density and damage rate of GM maize were significantly less than that of non-GM plants, although this crop is currently not registered for use in China.
This study provides a high-quality reference genome that demonstrates a genomic feature different from the established (American) C-or R-strain genotypes, as well as more comprehensive gene annotation. We also present resequencing data for 103 fall armyworm indi- important issues that remain for further exploitation using this whole genome approach, such as identifying the genes involved in polyphagy, migratory capability and olfaction, which could provide valuable tools for the future management of fall armyworms.

ACK N OWLED G EM ENTS
The following bodies provided funding that contributed to this work: Key Project for Breeding Genetic Modified Organisms Wei Fan https://orcid.org/0000-0001-5036-8733