The grasshopper genome reveals long-term gene content conservation of the X Chromosome and temporal variation in X Chromosome evolution

Abstract

We present the first chromosome-level genome assembly of the grasshopper, Locusta migratoria, one of the largest insect genomes. We use coverage differences between females (XX) and males (X0) to identify the X Chromosome gene content, and find that the X Chromosome shows both complete dosage compensation in somatic tissues and an underrepresentation of testis-expressed genes. X-linked gene content from L. migratoria is highly conserved across seven insect orders, namely Orthoptera, Odonata, Phasmatodea, Hemiptera, Neuroptera, Coleoptera, and Diptera, and the 800 Mb grasshopper X Chromosome is homologous to the fly ancestral X Chromosome despite 400 million years of divergence, suggesting either repeated origin of sex chromosomes with highly similar gene content, or long-term conservation of the X Chromosome. We use this broad conservation of the X Chromosome to test for temporal dynamics to Fast-X evolution, and find evidence of a recent burst evolution for new X-linked genes in contrast to slow evolution of X-conserved genes.

Grasshoppers (order Orthoptera, suborder Caelifera) represent an important phylogenetic and developmental comparison to many insect model systems. The first grasshoppers likely arose 250 million years ago during the Triassic period (Misof et al. 2014), and species within the group have since become some of the most prevalent herbivores on earth. The suborder, which contains more than 12,000 species, exhibits a worldwide distribution, with the greatest diversity in the tropics.

Grasshoppers normally possess XX/X0 sex chromosomes (Mao et al. 2020). X0 sex determination systems are thought to derive from XY systems with highly differentiated X and Y Chromosomes in species in which sex is determined by X Chromosome dose rather than Y Chromosome gene content (Furman et al. 2020). Because the Y Chromosome is completely lost in X0 systems, they represent the ultimate example of sex chromosome heterogamety. Extreme examples are often useful in revealing evolutionary patterns; however, despite their inherent utility for the study of sex chromosomes, X0 sex chromosomes are relatively rare compared with XY systems (Bachtrog et al. 2014; The Tree of Sex Consortium 2014), and therefore, their dynamics are not well understood. In XX/X0 systems, the entirely of the X Chromosome is haploid in males, and although theory predicts that extreme heterogamety will accelerate Fast-X evolution (Charlesworth et al. 1987) and the evolution of dosage compensation (Charlesworth 1996), empirical tests of this remain rare (Pal and Vicoso 2015).

Grasshoppers were an early genetic model, and Walter Sutton proposed the chromosome theory of heredity based in part on his work on grasshoppers at the start of the twentieth century (Crow and Crow 2002). Sutton's success was partly attributed to the large chromosomes in grasshoppers, which result from extreme genome size, which in turn has hampered effective genome assembly and subsequent molecular studies. As a result, the two currently available grasshopper genome assemblies, Locusta migratoria (Wang et al. 2014) and Schistocerca gregaria (Verlinden et al. 2020), remain fragmented, with contig N50s of 9.3 kb and 12.0 kb, respectively.

In this study, we combined the Pacific Biosciences (PacBio) HiFi reads (Wenger et al. 2019) and Hi-C technology (Belton et al. 2012) to assemble the first high-quality chromosome-level genome of a grasshopper, the migratory locust, L. migratoria. Our genome assembly allows unprecedented insight into the role of extreme heterogamety in sex chromosome evolution, and our results reveal widespread conservation of the X Chromosome gene content across broad swathes of the insect phylogeny, as well as temporal dynamics to the rate of X Chromosome evolution (Meisel et al. 2019; Chauhan et al. 2021).

Results

Genome features

We used PacBio HiFi sequencing to generate genome sequences for a female (XX) migratory locust and then used Hi-C reads to scaffold the contigs into a chromosome-level genome assembly comprising 12 chromosome-level scaffolds (Fig. 1A). The final assembled genome size is 6.3 Gb with a contig N50 value of 52.8 Mb, the largest to date among published chromosome-level insect genome assemblies. To assess the completeness of our assembly, we performed BUSCO analyses against the insect orthologous groups and recovered a score of 96%, a major improvement on the previous migratory locust and desert locust assemblies (Fig. 1B). Using our own and previously published RNA sequencing (RNA-seq) data sets, we identified 26,636 protein-coding genes with a total of 37,981 transcripts and 59,466 UTRs. Among the 26,636 genes, 19,481 were annotated by BLASTP against the RefSeq arthropod proteins, including top-hits to Zootermopsis nevadensis.

Figure 1.

Chromosome-scale genome assembly. (A) Hi-C contact map comprises 12 chromosome-level scaffolds, designated as linkage groups (LGs). (B) BUSCO assessment of our assembly, the previous migratory locust, and desert locust assemblies. (C) Male read depth along the genome in 250 kb windows.

The proliferation of repetitive elements is the main reason for the large size of the L. migratoria genome, and repetitive elements constituted 76.57% of the assembled genome, of which DNA transposons (19.87%) and LINE retrotransposons (28.13%) were the most abundant elements. The total repetitive content is much higher than previous reported (60%) (Wang et al. 2014), showing the advantage of the PacBio HiFi reads in assembly of highly repetitive genomes. To investigate the genome quality, we also quantified the satellite DNA distribution along each chromosome (Supplemental Fig. S1). The most dominant satellites are LmiSat02A-176 and LmiSat27A-57. In addition, we identified several centromere- and telomere-specific satellites (i.e., LmiSat01A-185 and LmiSat07A-5-tel), suggesting that centromere and telomere repetitive elements have successfully integrated into some chromosomes, further demonstrating the high quality of our genome assembly. Notably, after removing repetitive sequences, the L. migratoria nonrepetitive genome is ∼1.5 Gb, larger than many other insects.

X Chromosome identification and characteristics

To identify the X Chromosome, we sequenced a male (X0) to an average of 30× coverage, mapping the Illumina reads to our genome and calculating read depth in 100 kb windows. Chromosome 3 has a read depth nearly half of other chromosomes (Fig. 1C), consistent with an X0 male karyotype and previous cytogenetic work (Cabrero et al. 2009).

We next compared features between the X Chromosome and autosomes (Supplemental Table S1; Supplemental Fig. S2). Compared with the autosomes, the X Chromosome has lower gene density (Wilcoxon test, P < 0.001) (Supplemental Fig. S3) and larger intron length (Wilcoxon test, P < 0.001). The population recombination rate (ρ) is lower on average across the X Chromosome compared with all autosomes (t-test, P < 0.001), except for LG4 and LG12 (t-test, P > 0.05) (Supplemental Fig. S4). The X Chromosome also exhibits some differences in repetitive element distribution (Supplemental Fig. S2), with lower LINE transposon density (Wilcox test, P < 0.001) and higher DNA transposon density (Wilcox test, P < 0.001) compared with the autosomes. Particularly, the Maverick transposon is significantly enriched on the X Chromosome (Wilcox test, P < 0.001), in which it is nearly twice as dense compared with the autosomes.

Next, we calculated the Kimura two-parameter (K2P) (Kimura 1980) distance of all transposons (Supplemental Fig. S5). The profile of the X Chromosome is similar to the small chromosomes, with a wave of Helitron proliferation in both chromosome classes.

X-linked gene content conservation across insect orders

We identified conservation of L. migratoria X-linked gene content across eight insect orders: Odonata, Orthoptera, Phasmatodea, Hemiptera, Neuroptera, Coleoptera, Diptera, and Lepidoptera (Fig. 2A). In each comparison except with the Lepidoptera, the gene content shared on the X Chromosome was greater than expected by chance based on the relative proportion of protein coding sites (chi-squared test, 1 d.f., P < 0.05) (Supplemental Tables S4, S5), suggesting either repeated origin of sex chromosomes with highly similar gene content, or long-term conservation of the X Chromosome. Differentiating these alternatives phylogenetically would require X and Y Chromosome orthologs, which our species lacks as it is X0. Notably, the 800 Mb grasshopper X Chromosome shares significant gene content to Muller element F in Drosophila melanogaster (the ancestral fly X Chromosome; 55.6% of fly Muller F genes, chi-squared test, 1 d.f., P = 3.84 × 10⁻³¹) (Vicoso and Bachtrog 2013) despite 400 million years of divergence. Through functional enrichment analysis, we show that these conserved X-linked genes include GO terms such as learning and memory, neuron recognition, and growth hormone synthesis (Supplemental Fig. S6). The Lepidoptera Z Chromosome does not show significant conservation with the L. migratoria X and, instead, is more likely homologous to LG5, indicating the Z likely represents an independent evolution origin.

Figure 2.

X- and Z-linked gene content conservation across eight insect orders. (A) Bar plot showing the proportion of X/Z genes in other species represented in the L. migratoria genome. The segment in black is the one interpreted as X/Z gene content conservation. The second column shows the gene proportion of each chromosome in L. migratoria. Genome comparisons include the following (Iele) Ischnura elegans (Chauhan et al. 2021), (Spic) Schistocerca piceifrons (NCBI:GCA_021461385.2), (Sgre) Schistocerca gregaria (Verlinden et al. 2020), (Toce) Teleogryllus oceanicus (Pascoal et al. 2020), (Tcri) Timema cristinae (Parker et al. 2022), (Nlug) Nilaparvata lugens (Ye et al. 2021), (Lstr) Laodelphax striatellus (Zhu et al. 2017), (Sfur) Sogatella furcifera (Wang et al. 2017), (Pven) Pachypsylla venusta (Li et al. 2020), (Apis) Acyrthosiphon pisum (Li et al. 2020), (Rmai) Rhopalosiphum maidis (Chen et al. 2019), (Ccar) Chrysoperla carnea (NCBI:GCA_905475395.1), (Pcha) Pogonus chalceus (NCBI:GCA_002278615.1), (Ppyr) Photinus pyralis (Fallon et al. 2018), (Rful) Rhagonycha fulva (NCBI:GCA_905340355.1), (Tcas) Tribolium castaneum (Tribolium Genome Sequencing Consortium 2008), (Pser) Pyrochroa serraticornis (NCBI:GCA_905333025.1), (Csep) Coccinella septempunctata (NCBI:GCA_907165205.1), (Haxy) Harmonia axyridis (Chen et al. 2021a), (Hill) Hermetia illucens (Generalovic et al. 2021), (Dmel) Drosophila melanogaster (Celniker et al. 2002), (Cpom) Cydia pomonella (Wan et al. 2019), (Hmel) Heliconius melpomene (Davey et al. 2016), and (Bmor) Bombyx mori (Lu et al. 2020). For details about statistical analysis of conservation and phylogenetic reconstruction, see Supplemental Tables S3–S5 and Methods. (B) Circos plot showing synteny analysis of X-linked genes across five insect orders: (SPX) S. piceifrons X, (TCX) T. castaneum X, (APX) A. pisum X, (CCX) C. carnea X, and (HIX) H. illucens X.

Next, we asked if the X Chromosome gene order is also conserved across insect orders. Cross-order genome collinearity is usually weak owing to the long evolutionary distances. As expected, we detect strong collinearity between grasshoppers but few syntenic blocks across orders on the X Chromosome. However, with more gaps allowed (which is less meaningful), we could identify significant enrichment of gene synteny on the X Chromosome (Fig. 2B).

Variation in the tempo of Fast-X evolution

For the temporal variation study, we classified genes in L. migratoria into five, partially overlapping, categories. X-conserved genes (Supplemental Fig. S6) are X-linked in at least eight species from Figure 2 and represent genes with long-term X-linkage. X-Lmig genes are X-linked only in L. migratoria and are autosomal in all other species from Figure 2 and represent genes with the shortest history of X-linkage. X–X genes are X-linked in L. migratoria and only one other insect species and represent an intermediate period of X-linkage. A–A genes are autosomal in L. migratoria and all other species. A-X genes are autosomal in L. migratoria and are X-linked in other species and also represent an intermediate period of X-linkage. For each of these categories, we calculated average d_N/d_S (Fig. 3; Supplemental Table S2), comparing each category to X-Lmig via bootstrapping (1000 replicates). Tests between all categories and A–A were also significant, with P-values in all cases <0.005.

Figure 3.

Gene evolution rate between X Chromosome and autosomes. (A) Box plot of d_N/d_S values from different autosomal and X-linked categories. We classified genes in L. migratoria into five, partially overlapping, categories: X-conserved genes are X-linked in at least eight species from Figure 2, X-Lmig are X-linked only in L. migratoria and autosomal in all other species from Figure 2, X–X genes are X-linked in L. migratoria and only one other species, A–A genes are autosomal in L. migratoria and all other species, and A–X genes are autosomal in L. migratoria and X-linked in other species. Significant difference to the autosomal in the all-species (A–A) category shown. (****) P < 0.001. (B) The moving average values of d_N/d_S along the X Chromosome. (C) The moving average values of recombination rate (Rho) along the X Chromosome.

We observe elevated rates of evolution in X-Lmig genes compared with all the other categories of genes, consistent with Fast-X evolution (Fig. 3; Supplemental Fig. S9; Supplemental Table S2). Notably, we did not observe elevated d_N/d_S for X-conserved or X–X genes, suggesting that Fast-X primarily results from genes specific to the L. migratoria X Chromosome. Importantly, X-conserved genes showed significantly slower rates of average evolution compared with A–A genes (P = 0.002), suggesting both Fast-X and Slow-X in the same species depending on the age of X-linkage. Fast-X and Slow-X are mainly owing to differences in d_N values (Supplemental Table S2). To validate this, we performed the same analysis in true bugs (Hemiptera) and recovered similar results (Supplemental Fig. S7).

We next analyzed d_N/d_S patterns across the X Chromosome (Fig. 3B), recovering a region (360–670 Mb) of elevated d_N/d_S compared with both the autosomes (P < 0.0001 based on 1000 bootstraps) and the remainder of the X Chromosome (P < 0.0001 based on 1000 bootstrap replicates) level. This suggests that Fast-X might be explained by regional variation along the X Chromosome. We observe no clear difference in recombination rate for this region (Fig. 3C; Supplemental Fig. S8), suggesting the effect is independent of recombination. However, two points of very low recombination appeared at ∼450 Mb and 550 Mb, which might represent the legacy of an inactivated centromere. The higher d_N/d_S region is marginally enriched for genes that are X-linked only in L. migratoria (chi-squared test, P = 0.0275). This suggests that temporal dynamics in Fast-X coupled with a nonrandom distribution of genes X-linked only in L. migratoria likely drive this result.

X Chromosome dosage compensation

We next tested for the presence of complete dosage compensation in L. migratoria (Fig. 4A). In female (XX) grasshoppers, X-linked genes showed higher expression levels than autosomal genes in both somatic and gonad tissues. In male (X0) grasshoppers, X-linked gene expression is higher than or equal to autosomal genes in somatic tissues but is significantly lower in testis (P < 0.001). The overall male X-linked expression was equal to female X-linked expression in somatic tissues but was significantly lower in testis (P < 0.001) consistent with complete X Chromosome dosage compensation in somatic cells. The dosage compensation is overall homogeneous along the X Chromosome, although absent in gonads (Fig. 4B; Supplemental Fig. S10).

Figure 4.

Dosage compensation in L. migratoria. (A) Box plots of gene expression from different tissues, sexes, and chromosomes. P-values were calculated based on 1000 bootstrap replicates. (B) log₂ (female expression level:male expression level) along the X Chromosome.

Discussion

Hi-C and long-read sequencing resolves a large, complex insect genome into chromosomes

We used a combination of long-read DNA and Hi-C sequencing to successfully resolve and assemble an unusually large and highly repetitive insect genome. To date, this is the largest insect genome and one of the largest arthropod genomes, assembled to chromosome scale. The assembly of such relatively large and highly repetitive insect genomes into highly contiguous chromosomes was until very recently unattainable, largely owing to the difficulties presented by high amounts of repetitive content. Indeed, the unusually large size of the grasshopper genome is primarily caused by the high proportion of repetitive content, corresponding to 76.57% of the genome. Therefore, we show that using the highly accurate long-read PacBio HiFi sequencing is an effective assembly strategy for highly repetitive genomes. During the repeat annotation, we constructed a de novo repeat library based on our assembly using RepeatModeler (Flynn et al. 2020). Combining our custom repeat library, the Dfam_Consensus (Storer et al. 2021), and the Repbase libraries (Bao et al. 2015), we identified far more repeats than the previous short reads (Wang et al. 2014). Although we cannot say we annotated every repeat in the genome, our repeat annotation is the most complete to date and nearly complete. Using our new high-quality genome assembly and annotation of L. migratoria, we investigated X Chromosome dynamics.

Conservation of X Chromosome gene content

We observe high conservation of X Chromosome gene content across Insecta (Fig. 2). This is in contrast to previous observations of turnover in sex chromosomes within the Diptera from the ancestral Dipteran X Chromosome, the dot chromosome (Vicoso and Bachtrog 2015). Conservation of X gene content has been observed in a few well-chosen Insecta species (Meisel et al. 2019; Chauhan et al. 2021), and the X may in fact predate the origin of Insecta (Toups and Vicoso 2023). Our work illustrates a broader conservation across the class, with the exception of the Lepidoptera (Pease and Hahn 2012).

Previous work has documented convergent gene content on the mammalian Y, avian W, and snake W Chromosomes (Bellott and Page 2021) and convergent evolution of mammalian X and avian Z Chromosome gene content via gene acquisition (Bellott et al. 2017). It remains unclear whether the conservation in gene content in Insecta reflects conservation of the sex chromosome itself, repeated origin of X Chromosomes from the same underlying syntenic regions, or repeated movement of the same gene content to the sex chromosomes. These are difficult alternatives to differentiate without knowledge of the specific sex-determining locus throughout this group.

Fast-X and Slow-X evolution

The X Chromosome has several properties that distinguish it from the autosomes (Vicoso and Charlesworth 2006; Meisel and Connallon 2013) and that have the potential to influence the rate and pattern of evolution of X-linked genes (Charlesworth et al. 1987). Because males have only one copy of the X Chromosome and therefore only one copy of X-linked genes, recessive mutations on the X Chromosome are directly exposed to selection in males (Charlesworth et al. 1987). This can lead either to rapid fixation of recessive beneficial variation (Fast-X) or to more efficient purging of recessive deleterious mutations (Slow-X) (Xu et al. 2012). Additionally, the reduced effective population size of the X Chromosome compared with the autosomes can lead to elevated rates of genetic drift, a nonadaptive cause of Fast-X evolution (Charlesworth et al. 1987). This is a process distinct from the steady rate of divergence between X and Y orthologs observed in sex chromosome strata, regions where recombination between the sex chromosomes was halted at different times (for review, see Furman et al. 2020).

We observe Fast-X on genes that are X-linked only in L. migratoria and Slow-X for genes that are conserved on the X Chromosome across insects (Fig. 3A). This may suggest that the pool of adaptive recessive variation is quickly depleted following X-linkage, resulting in a limited burst of Fast-X. Over time, this dynamic appears to shift such that recessive deleterious variation is purged more effectively on the X, resulting in Slow-X over greater evolutionary distances. Slower-X has been predicted theoretically in the early stages of X Chromosome evolution (Mrnjavac et al. 2023). Interaction of Fast-X and Slow-X has previously been observed over far shorter timescales (Xu et al. 2012); however, the extraordinary conservation of X Chromosome coding content that we observe here across Insecta makes it possible to discern temporal dynamics in X Chromosome evolution across far larger timespans. Recombination is reduced on the X compared with the autosomes (Supplemental Fig. S3), and we would expect the reduction in recombination both to reduce the amount of standing variation on the X, all other forces being equal, and to reduce the adaptive potential of the X compared with the autosomes. However, we observed no clear relationship between recombination rate and rates of evolution along the X (Fig. 3C; Supplemental Fig. S7). The temporal dynamics of X evolution that we observe, in addition to the fact that Fast-X in L. migratoria is largely confined to a restricted region (360–670 Mb) (Fig. 3B), suggests that this Fast-X region may represent a recent addition to the X Chromosome.

Complete dosage compensation

Complete sex chromosome dosage compensation has evolved many times (Mank 2009, 2013) through a variety of mechanisms (Gu and Walters 2017) in both invertebrates and vertebrates (Nguyen and Disteche 2006; Rupp et al. 2016; Darolti et al. 2019; Huylmans et al. 2019; Hu et al. 2022; Metzger et al. 2023). Although dosage compensation is nearly always associated with high degrees of sex chromosome heterogamety, not all cases of heterogametic sex chromosomes evolve complete dosage compensation (Arnold et al. 2008; Chauhan et al. 2021), and it has been suggested that the evolution of complete sex chromosome dosage compensation may in fact accelerate the degeneration of the Y or W Chromosome and therefore exaggerate sex chromosome heterogamety (Lenormand and Roze 2022).

In addition to accelerating sex chromosome divergence, complete sex chromosome dosage compensation, which we observe for somatic tissues in L. migratoria (Fig. 4), is expected to exacerbate Fast-X and Slow-X evolution (Vicoso and Charlesworth 2009; Mank et al. 2010). The extreme heterogamety represented in X0 sex chromosome systems may also increase the rate of X evolution (Darolti et al. 2023). Fast-X, accelerated by dosage compensation and extreme heterogamety, may in turn increase the role of the X Chromosome in adaptation and speciation relative to its size and coding content, termed the Large-X effect (Lasne et al. 2017).

Approximately 60 years ago, Susumu Ohno predicted that the conservation of gene content on the X Chromosome across placental mammals evolved to maintain dosage relationships between X-linked genes and their autosomal counterparts (Ohno 2013). Linkage conservation is another notable feature of the mammalian X Chromosome, which may be a result of selection against rearrangements that might disrupt X Chromosome inactivation (Brashear et al. 2021). Our analysis shows that, compared with mammals, the insect X Chromosome follows Ohno's prediction but shows much weaker linkage conservation (Fig. 2B). Currently, Drosophila (Conrad and Akhtar 2012), aphids (Richard et al. 2017), and grasshoppers all exhibit X Chromosome upregulation, which is different from the X Chromosome inactivation in mammals. We hypothesize that, on the one hand, the dosage effect leads to the gene content conservation in the insect X Chromosome, as Ohno predicted. On the other hand, the insect dosage compensation mechanisms allowed some gene reshuffling between X Chromosome and autosomes. Further studies—especially dosage compensation mechanisms in more insect species—are needed to provide more evidence for this hypothesis.

Methods

Library construction and sequencing

For PacBio sequencing, genomic DNA of a female migratory locust was isolated and sheared to an average size of 20 kb using a g-TUBE device (Covaris). The sheared DNA was purified and end-repaired using polishing enzymes, followed by blunt end ligation and exonuclease treatment to create a SMRTbell template according to the PacBio 20 kb template preparation protocol. A BluePippin device (Sage Science) was used to size-select the SMRTbell template and enrich large (>10 kb) fragments. SMRTbell libraries were sequenced on a PacBio sequel II system, and consensus reads (HiFi reads) were generated using ccs software (https://github.com/pacificbiosciences/unanimity).

For Hi-C sequencing, Hi-C libraries were prepared from a male migratory locust at the BioMarker Technologies Company. Briefly, sample was collected and spun down, and the cell pellet was resuspended and fixed in formaldehyde solution. DNA was isolated, and the fixed chromatin was digested with the restriction enzyme DpnII overnight. The cohesive ends were labeled with biotin-14-DCTP using the Klenow enzyme and then religated with T4 DNA ligation enzyme. Subsequent DNA was sheared by sonication to a mean size of 350 bp. Hi-C libraries were generated using NEBNext Ultra enzymes and Illumina-compatible adaptors. Biotin-containing fragments were isolated using streptavidin beads. All libraries were quantified by Qubit2.0, and insert size was checked using an Agilent 2100 and then quantified by quantitative polymerase chain reaction (PCR). Hi-C sequencing was performed by a Illumina HiSeq 2500 platform, using the paired-end of 150 bp reads.

To assist gene prediction and dosage compensation analysis, 24 RNA-seq libraries were generated from brain, hindleg, and gonads with four biological replicates for each sex. Total RNA was extracted from each tissue using a TRIzol kit (Invitrogen). The mRNA fractions were isolated from the total RNA extracts with the MicroPoly(A)Purist kit (Ambion). cDNA libraries were prepared for each tissue with the RNA-seq library kit (Gnomegen) following the manufacturer's instructions. Each paired-end cDNA library was sequenced with a read length of 150 bp using the Illumina HiSeq 2500 sequencing platform. All sequencing was performed by the Biomarker Technologies Company.

Genome assembly

The PacBio long (∼12 kb) and highly accurate (>99%) HiFi reads were assembled to a contig-level assembly using Hifiasm (Cheng et al. 2021). The Hi-C data were mapped to Hifiasm contigs with BWA-MEM (Li and Durbin 2009). Uniquely mapped data were used for chromosome-level scaffolding. HiC-Pro version 2.8.1 (Servant et al. 2015) was used for duplicate removal and quality controls, and the filtered Hi-C data were then used to correct misjoins as well as to order and orient contigs. Preassembly was performed for contig correction by splitting contigs into segments with an average length of 300 kb, and then, the segments were preassembled with Hi-C data. Misassembled points were defined and broken when split segments could not be placed to the original position. Then, the corrected contigs were assembled using LACHESIS (Burton et al. 2013) with the parameters CLUSTER_MIN_RE_SITES = 225, CLUSTER_MAX_LINK_ DENSITY = 2; ORDER_MIN_N_RES_IN_TRUN = 105; ORDER_MIN_N_RES_IN_SHREDS = 105 with Hi-C valid pairs. Gaps between ordered contigs were filled with 100 N's.

To evaluate the quality of the genome assembly, we performed BUSCO (Simão et al. 2015) analyses using 1367 core conserved insect genes on the old assembly (Wang et al. 2014), the recent desert locust assembly (Verlinden et al. 2020), and our assembly.

Repeat annotation and gene prediction

De novo identification of repeats was performed by the RepeatModeler (Flynn et al. 2020) under default parameters. We also recovered 107 satellite DNA sequences belonging to 62 families in L. migratoria (Ruiz-Ruano et al. 2016). Using the ab initio repeat library and satellite DNA library and the Dfam_Consensus (Storer et al. 2021) and the Repbase libraries (Bao et al. 2015), we estimated the repeat content of the assembled genome using RepeatMasker (Smit et al. 2013–2015). Ab initio gene prediction was performed using AUGUSTUS (Stanke et al. 2008). GenomeThreader, implemented in BRAKER2 (Brůna et al. 2021), was run for homology-based prediction using protein sequences of D. melanogaster, Anopheles gambiae, Tribolium castaneum, Apis mellifera, Bombyx mori, Acyrthosiphon pisum, and Z. nevadensis. Publicly available NCBI transcriptome data (obtained from the NCBI Sequence Read Archive [SRA; https://www.ncbi.nlm.nih.gov/sra] under accession numbers SRR4041955, SRR6823282, SRR6823283, SRR6823285–SRR6823287, SRR1974300, SRR3315158, SRR3318255, SRR7221659, SRR2051024, SRR3372608–SRR3372610) and our own transcriptome data were aligned by HISAT2 (Kim et al. 2019) and assembled with StringTie (Pertea et al. 2015), and then coding regions were identified with TransDecoder (https://github.com/TransDecoder/TransDecoder). Finally, EVidenceModeler (Haas et al. 2008) was used to integrate the prediction results obtained with the above three methods. PASA v2.4.1 (Haas et al. 2008) was run to further improve the gene structure annotation.

X Chromosome identification via coverage in males

To identify the X Chromosome, a male migratory locust was sequenced to nearly 30× coverage. The Illumina reads were aligned to our genome assembly with BWA-MEM (Li and Durbin 2009), and SAMtools (Li et al. 2009) was used to extract uniquely mapped reads and remove PCR duplicates. mosdepth (Pedersen and Quinlan 2018) was used to calculate read coverage along the genome (parameters: -t 3 -n ‐‐fast-mode ‐‐by 250,000).

Gene density, GC content, nucleotide diversity

The gene density of each chromosome was calculated as the number of genes divided by chromosome length. When comparing gene or repeat density among chromosomes, we use 5 Mb sliding windows to calculate the number of genes or repeats. GC content along chromosomes was calculated within 50 kb sliding windows. VCFtools v0.1.13 (Danecek et al. 2011) was used to determine nucleotide diversity within 500 kb sliding windows.

Recombination rate estimation

To explore the recombination across the locust genome, we estimated the population recombination rate (ρ) using FastEPRR (Gao et al. 2016). First, the resequencing data of five female grasshoppers were downloaded from NCBI BioProject database (https://www.ncbi.nlm.nih.gov/bioproject/) under accession number PRJNA433455. BCFtools (Danecek et al. 2021) mpileup was used to call SNPs. Then Beagle 5.0 (Browning and Browning 2007) was used to phase the SNPs, and phased data were then input into the FastEPRR_VCF_step1 function in FastEPRR to scan each 10 and 50 kb window (with parameters inSNPThreshold = 30 and qualThreshold = 20). Next, FastEPRR_VCF_step2 was used to estimate the recombination rate for each window. Finally, we applied FastEPRR_VCF_step3 to merge the files generated by step 2 for each chromosome.

K2p analysis

RepeatMasker (Smit et al. 2013–2015) was used to construct the TE expansion history in the migratory locust genome by first recalculating the divergence of the identified TE copies in the genome with the corresponding consensus sequence in the TE library using the Kimura distance and then estimating the percentage of TEs in the genome at different divergence levels.

Gene content and synteny in insect orders

The species phylogenetic tree and divergence time were achieved from TimeTree (Kumar et al. 2022). Insect genomes with assembled sex chromosomes were retrieved from InSexBase (Chen et al. 2021b), InsectBase 2.0 (Mei et al. 2022), and NCBI GenBank database (https://www.ncbi.nlm.nih.gov/genbank/). The proportion of X-linked genes of each species with their L. migratoria homologs were identified by reciprocal best BLAST hit, including Ischnura elegans, Schistocerca piceifrons, S. gregaria, Teleogryllus oceanicus, Timema cristinae, Nilaparvata lugens, Laodelphax striatellus, Sogatella furcifera, Pachypsylla venusta, A. pisum, Rhopalosiphum maidis, Chrysoperla carnea, Pogonus chalceus, Photinus pyralis, Rhagonycha fulva, T. castaneum, Pyrochroa serraticornis, Coccinella septempunctata, Harmonia axyridis, Hermetia illucens, D. melanogaster, Cydia pomonella, Heliconius Melpomene, and B. mori. OrthoFinder v2.3.3 (Emms and Kelly 2019) together with DIAMOND v0.9.14 (Buchfink et al. 2015), MAFFT v7.305 (Katoh and Standley 2013), and FastTree v2.1.7 (Price et al. 2009) was used to cluster proteins into orthogroups, reconstruct gene trees, and estimate the species tree. Daphnia pulex (Colbourne et al. 2011) was used as an outgroup in phylogenetic tree construction.

To identify genome synteny over large evolutionary distances, we ran MCScanX (Wang et al. 2012) with the parameters “ -s 2 -k 100 -m 50.” MCScanX results were visualized with Circos (Krzywinski et al. 2009).

Functional enrichment of genes

Gene functions and Gene Ontology (GO) annotations were retrieved with eggNOG-mapper (Cantalapiedra et al. 2021). Because L. migratoria is not a model organism, a local OrgDb database was constructed based on eggNOG-mapper results. The functional enrichment was then determined using clusterProfiler (Yu et al. 2012).

Fast-X analysis

The evolution rate of genes (d_N/d_S) was calculated by comparing the L. migratoria protein sequence with grasshopper Oedaleus asiaticus, which belongs to the same subfamily, Oedipodinae, as L. migratoria. Because of a lack of reference genome, a gene set of O. asiaticus was constructed from transcriptomic data, which were downloaded from the NCBI SRA database (SRR IDs SRR2051024, SRR3372608, SRR3372609, and SRR3372610). Trinity (Grabherr et al. 2011) was used to assemble a transcriptome representing a nonredundant gene set of this species. Then, these Trinity-based genes and our genome-based genes were used for the d_N/d_S analysis. The reciprocal best BLAST hit pairs were used to identify orthogroups. KaKs_Calculator (v2.0; https://sourceforge.net/projects/kakscalculator2/) was used to calculate d_N/d_S values. Orthologous genes with d_N/d_S > 2 were removed.

To assess temporal variation in Fast-X evolution, we classified genes in L. migratoria into five, partially overlapping, categories. X-Conserved genes are X-linked in at least eight species and represent genes with the longest period of X-linkage. X-Lmig genes are X-linked only in L. migratoria and are autosomal in all other species and represent genes with the shortest history of X-linkage. X–X genes are X-linked in L. migratoria and only one other species and represent an intermediate period of X-linkage. A–A genes are autosomal in L. migratoria and all other species. A-X genes are autosomal in L. migratoria and are X-linked in other species and also represent an intermediate period of X-linkage. For each of these categories, we calculated average d_N/d_S, comparing each category to X-Lmig (1000 bootstrap replicates).

Dosage compensation analysis

RNA-seq reads from heads, hindlegs, and gonads of four adult females and four adult males were trimmed for adapter and low-quality bases (Q < 20) using fastp (Chen et al. 2018). Next, the RNA-seq reads were mapped to the genome using HISAT2 (Kim et al. 2019). An abundance estimation was performed with featureCounts (Liao et al. 2014). The raw counts were normalized by TPM methods. Genes with low expression support (sum of normalized read count of all samples less than one) were removed from downstream analysis. Dosage compensation was assessed by comparing the average expression between female autosomal and X genes, between male autosomal and male X genes, between female autosomal and male autosomal genes, and between female X and male X genes.

Statistical analysis

All statistical tests were performed using R Software (v4.1.1) (R Core Team 2021). A Student's t-test was used when the two groups of data follow a normal distribution and exhibit equal variances. If the variances were unequal, a wilcox.test() implemented in R was used. Bootstrapping was used as an alternative to statistical inference based on the assumption of a parametric model when that assumption is in doubt or when parametric inference is impossible.

A chi-squared test was used for the test of X-linked genes distribution against L. migratoria chromosomes. In short, we used a chi-squared test to determine whether any L. migratoria chromosome contained statistically more or fewer genes from a test organism X Chromosome, based on the total number of genes on that L. migratoria chromosome as a proportion of the total number of genes in the genome (chi-squared test, 1 d.f.). The expected value in the chi-squared test is calculated as

$$E_i={\displaystyle \frac{N_{xLmig}}{N_{Lmig}}{N}_x,}$$

where E_i denotes the expected value for chi-squared test, N_xLmig denotes the number of L. migrotoria X-linked genes, N_Lmig denotes the number of all genes in the L. migrotoria genome, and N_x denotes the number of X-linked genes in the test species. The formula of the chi-square statistic is shown as below:

$${\chi }^2={\displaystyle \frac{{\left(O_i-E_i\right)}^2}{E_i},}$$

where χ² denotes the chi-square statistic; O_i denotes the observed value, namely, the number of X-linked genes in the test species, which is also X-linked in L. migrotoria; and E_i denotes the expected value calculated by the previous formula. Bootstrapping was used for comparing d_N/d_S and gene expression between groups. Briefly, the mean value was calculated using the raw data. Then, we used the sample() function in R v4.1.1 to resampling the raw data with “replace=TURE.” Then the t.test() function was used to calculate statistical difference based on 1000 or 10,000 replicates.

Data access

All raw and processed sequencing data generated in this study have been submitted to the NCBI Sequence Read Archive (SRA; http://www.ncbi.nlm.nih.gov/sra/) under accession number SRP402730. The genome assembly CAU_Lmig_1.0 from this study has been submitted to the NCBI Assembly database (https://www.ncbi.nlm.nih.gov/assembly/) under accession number GCA_026315105.1. The genome annotations are openly available on Zenodo (https://doi.org/10.5281/zenodo.7538512) databases. The custom code for the data analysis can be openly accessed on the Zenodo database at https://doi.org/10.5281/zenodo.7538512 and in the Supplemental Scripts.

Competing interest statement

The authors declare no competing interests.