Sawfly Genomes Reveal Evolutionary Acquisitions That Fostered the Mega-Radiation of Parasitoid and Eusocial Hymenoptera

Abstract

The tremendous diversity of Hymenoptera is commonly attributed to the evolution of parasitoidism in the last common ancestor of parasitoid sawflies (Orussidae) and wasp-waisted Hymenoptera (Apocrita). However, Apocrita and Orussidae differ dramatically in their species richness, indicating that the diversification of Apocrita was promoted by additional traits. These traits have remained elusive due to a paucity of sawfly genome sequences, in particular those of parasitoid sawflies. Here, we present comparative analyses of draft genomes of the primarily phytophagous sawfly Athalia rosae and the parasitoid sawfly Orussus abietinus. Our analyses revealed that the ancestral hymenopteran genome exhibited traits that were previously considered unique to eusocial Apocrita (e.g., low transposable element content and activity) and a wider gene repertoire than previously thought (e.g., genes for CO2 detection). Moreover, we discovered that Apocrita evolved a significantly larger array of odorant receptors than sawflies, which could be relevant to the remarkable diversification of Apocrita by enabling efficient detection and reliable identification of hosts.

Keywords: hexamerin; major royal jelly protein; microsynteny; odorant receptor; opsin; phytophagy.

Fig. 1.

Hymenoptera genome evolution. (A) Adult males of Athalia rosae^a and Orussus abietinus. Scale bar: 2.5 mm. (B) Number of described species (Apocrita: 144,593; Orussidae: 82; “Symphyta” excl. Orussidae: 7,983) of, relationships of, and ecological transitions in Hymenoptera (Aguiar et al. 2013; Peters et al. 2017). (C) Ratio of gain and loss of genes, domains, and domain arrangements, as well as ratio of gene families that experienced expansions or contractions. Gene and gene family evolution were analyzed by applying the maximum-likelihood optimality criterion, a single coupled birth and death rate, and using the divergence time estimates and phylogenetic relationships inferred by Peters et al. (2017). Domain and domain arrangement evolution were analyzed by applying the maximum parsimony optimality criterion. (D) Absolute number of nucleotides occupied by genomic components (left column), median length of various gene structure parameters (center column), and gene orthology in the genome of each species (right column; unit = number of genes). (E) Divergence distribution of transposable element (TE) copies in the genome of At. rosae and that of Apis mellifera, estimated from the Kimura distance of the nucleotide sequence of each TE copy to the TE family nucleotide consensus sequence. (F) Loss of synteny over time in the genomes of 12 Hymenoptera, inferred from the proportion of 3,983 shared single-copy orthologs (SCOs) retaining the same neighboring SCO, relative to the divergence time, in all possible pairwise comparisons. The curve represents the smoothed conditional mean. aa, amino acids; bp, base pairs; CDS, coding sequence; LINE, long interspersed nuclear element; LTR, long terminal repeats; Ma, million years ago; RC, rolling circle transposons; SINE, short interspersed nuclear element; TE, transposable elements; Aech, Acromyrmex echinatior; Amel, Apis mellifera; Aros, A. rosae; Bter, Bombus terrestris; Cflo, Camponotus floridanus; Dnov, Dufourea novaeangliae; Hsal, Harpegnathos saltator; Lalb, Lasioglossum albipes; Mrot, Megachile rotundata; Nvit, Nasonia vitripennis; Oabi, Orussus abietinus; Pdom, Polistes dominula; Tcas, Tribolium castaneum. All photographs by Oliver Niehuis, with assistance from Thomas Pauli and Ralph S. Peters. ^aNote that while the photograph shows a male of the nominate form, we sequenced and report the genome of the Eastern Palearctic subspecies At. rosae ruficornis.

Fig. 2.

Evolution of hymenoptera yellow, MRJP/-like, and immune response-related genes. (A) Relationships of hymenoptera yellow, major royal jelly protein (MRJP), and MRJP-like (MRJPl) amino acid sequences, inferred under the maximum-likelihood optimality criterion, modeling invariable sites, and approximating site-rate variation with a discrete gamma distribution. Branch support is estimated from 1,000 nonparametric bootstrap replicates. MRJP and MRJPl proteins of Athalia rosae and Orussus abietinus are highlighted in blue and red, respectively. (B) Gene structure comparison of mrjp and mrjpl genes and of two candidate sister group yellow genes, y-e3 and y-x2. Dashed lines indicate shared amino acid motifs conserved among species within each gene and between genes (supplementary section II.5.5, Supplementary Material online). Gene and motif lengths not to scale. (C) Heat map visualizing copy number variation in immune response-related genes between species. Modified Z-scores indicate the deviation from the median of each gene by SD units. Aaeg, Aedes aegypti; Aech, Acromyrmex echinatior; Amel, Apis mellifera; Apis, Acyrthosiphon pisum; Aros, Athalia rosae; Bter, Bombus terrestris; Cflo, Camponotus floridanus; Dnov, Dufourea novaeangliae; Dsim, Drosophila simulans; Gmor, Glossina morsitans; Hmel, Heliconius melpomene; Hsal, Harpegnathos saltator; Lalb, Lasioglossum albipes; Lhum, Linepithema humile; Mrot, Megachile rotundata; Oabi, Orussus abietinus; Pdom, Polistes dominula; Nvit, Nasonia vitripennis; Tcas, Tribolium castaneum; Znev, Zootermopsis nevadensis.

Fig. 3.

Hymenoptera vision gene, metabolic, hexamerin, and chemoreceptor repertoires. (A) Phylogenetic relationships of Hymenoptera, Nephotettix cincticeps (Hemiptera), and Drosophila opsin genes inferred under the maximum-likelihood optimality criterion. Branch support is estimated from 500 nonparametric bootstrap replicates. (B) Number of unique and shared enzymes (Enzyme Commission numbers) in the proteomes of Athalia rosae, Orussus abietinus, and Nasonia vitripennis. (C) Number of unique and shared metabolic pathways identified in the proteomes of At. rosae, O. abietinus, and N. vitripennis, inferred from enzyme and gene ontology annotations. (D) Phylogenetic relationships of Hymenoptera hexamerins inferred under the maximum-likelihood optimality criterion. Branch support is estimated from 1,000 nonparametric bootstrap replicates. Colors indicate deviation of the amino acid glutamine (Q) from the average amino acid content in percent (%). (E) Copy number variation of odorant and gustatory receptor gene repertoires among Hymenoptera. Data referring to At. rosae and O. abietinus are taken from the present study, those of all remaining species from literature (Robertson and Wanner 2006; Robertson et al. 2010; Zhou et al. 2012, 2015; Sadd et al. 2015; Robertson et al. 2018). Only full-length proteins comprising at least 350 amino acids were considered. Phylogenetic relationships taken from the study by Peters et al. (2017). Aech, Acromyrmex echinatior; Amel, Apis mellifera; Aros, Athalia rosae; Bter, Bombus terrestris; Ccin, Cephus cinctus; Cflo, Camponotus floridanus; Csol, Ceratosolen solmsi; Dmel, Drosophila melanogaster; Hsal, Harpegnathos saltator; Lalb, Lasioglossum albipes; Mdem, Microplitis demolitor; Ncin, Nephotettix cincticeps; Nvit, Nasonia vitripennis; Oabi, Orussus abietinus.