Thrips are tiny insects from the order Thysanoptera (Hexapoda: Condylognatha), including many sap-sucking pests that are causing increasing damage to crops worldwide. In contrast to their closest relatives of Hemiptera (Hexapoda: Condylognatha) including numerous sap-sucking species, there are few genomic resources available for thrips. In this study, we assembled the first thrips genome at the chromosome level from the melon thrips, Thrips palmi, a notorious pest in agriculture, using PacBio long-read and Illumina short-read sequences. The assembled genome was 270.43 Mb in size with 4,120 contigs and a contig N50 of 426 kb. All contigs were assembled into 16 linkage groups assisted by the Hi-C technique. In total, 16,333 protein-coding genes were predicted, of which 88.13% were functionally annotated. Among sap-sucking insects, polyphagous species usually possess more detoxification genes than oligophagous species. The polyphagous thrips genomes characterized so far have relatively more detoxification genes in the GST and CCE families than polyphagous aphids, but they have fewer UGTs. HSP genes, especially from the Hsp70s group, have expanded in thrips compared to other hemipteran insects. These differences point to different genetic mechanisms associated with detoxification and stress responses in these two groups of sap-sucking insects. The expansion of these gene families may contribute to the rapid development of pesticide resistance in thrips, as supported by a transcriptome comparison of resistant and sensitive populations of T. palmi. The high-quality genome developed here provides an invaluable resource for understanding the ecology, genetics and evolution of thrips as well as their relatives more generally.

To facilitate both the curation of inaccurate annotations and the identification of previously undetected gene family copies directly from DNA sequences, we manually annotated seven gene families, including detoxification genes of cytochrome P450 monooxygenase (P450s), glutathione S-transferase (GSTs), carboxyl/cholinesterases (CCEs), UDP-glycosyltransferases (UGTs) and ATP-binding cassette (ABC) transporter, heat shock proteins (HSPs), and nicotinic acetylcholine receptors (nAChRs). Hidden Markov models (HMMs) and orthologs from related species of F. occidentalis and model species of D. melanogaster provided evidence for gene identification, run with HMMER v3.3 (Finn et al. 2011) and BLAST v2.2.31 (Altschul et al. 1990). The HMMs were downloaded from Pfam 32.0 (September 2018, 17929 entries) (El-Gebali et al. 2018). We used the bioinformatic pipeline BITACORA (full mode) to conduct HMMER and BLAST analyses (Vizueta et al. 2019). Hits were filtered with a default cut-off E-value of 10e-5. The annotated genes were further filtered manually based on gene length and the presence of conserved domains. We removed genes with a length shorter than 80% of the average gene length. Orthologs were aligned with the G-INS-I algorithm implemented in MAFFT v7.450 (Katoh& Standley 2013). A neighbor-joining tree was constructed for each gene family using MEGA7 (Kumar et al. 2016) with 500 bootstrap replicates.

Comparative and functional genomics.