The genome of the clonal raider ant Cerapachys biroi

Abstract

Social insects are important models for social evolution and behavior. However, in many species, experimental control over important factors that regulate division of labor, such as genotype and age, is limited. Furthermore, most species have fixed queen and worker castes, making it difficult to establish causality between the molecular mechanisms that underlie reproductive division of labor, the hallmark of insect societies. Here we present the genome of the queenless clonal raider ant Cerapachys biroi, a powerful new study system that does not suffer from these constraints. Using cytology and RAD-seq, we show that C. biroi reproduces via automixis with central fusion and that heterozygosity is lost extremely slowly. As a consequence, nestmates are almost clonally related (r = 0.996). Workers in C. biroi colonies synchronously alternate between reproduction and brood care, and young workers eclose in synchronized cohorts. We show that genes associated with division of labor in other social insects are conserved in C. biroi and dynamically regulated during the colony cycle. With unparalleled experimental control over an individual's genotype and age, and the ability to induce reproduction and brood care, C. biroi has great potential to illuminate the molecular regulation of division of labor.

Figure 1

A clonal raider ant (Cerapachys biroi) worker carrying a pupa. Ants of the genus Cerapachys are myrmecophagous and raid the nests of other ants [6]. The genus belongs to the dorylomorph clade of ants, which also includes the infamous army ants [6]. Since the early 1900s, introduced populations of C. biroi have become established on tropical and subtropical islands around the world, probably as a consequence of human traffic and trade [7, 8]. Like in many other dorylomorphs, colonies of C. biroi undergo stereotypical behavioral and reproductive cycles [5, 9]. Colonies of C. biroi lack queens and instead consist entirely of totipotent workers, all of which reproduce asexually [10, 11].

Figure 2

Three-dimensional projections of DAPI-stained chromosomes in < 2hr old eggs showing that C. biroi reproduces through automixis with central fusion. Embryos were prepared according to [14]. The diploid chromosome number in C. biroi is 2n = 28 [13]. A) Prophase I immediately post partum showing a single diploid nucleus close to the posterior pole of the egg. B) Meiosis I (reductional division) at approximately 30 min post partum. Two nascent haploid nuclei can be seen. C) Fusion of central products of meiosis II (indicated by arrow) within one hour post partum. D) Embryo after two rounds of mitotic division with four diploid nuclei, within two hours post partum. The polar bodies in panel C have fused (arrow) and migrated to the cell membrane, where they degenerate. Additional stages are shown in Figure S4.

Figure 3

Phylogenetic and genomic relationships between C. biroi individuals, colonies, and clonal lineages. The UPGMA tree shows the average number of substitutions per site for 100,608 informative sites, between 91 ants from 4 clonal lineages and 19 colonies. Colonies that are not recovered as monophyletic in the phylogeny are indicated by a “+” between colony names (there is no colony-level resolution for MLL6 and MLL13). Contiguous regions of homozygous loci ≥ 1Mb in size are shown in the map on the right: each colored square represents a putative single loss of heterozygosity (LOH) event. Columns indicate the 16 different scaffolds that contain LOH spanning ≥ 1Mb (scaffold number given above each column). Scaffold 113 (marked in red) contains the telomeric repeat sequence. Identical block colors within a column indicate homozygous fragments with identical beginning and end positions, which most likely arose from a single ancestral LOH event. Lighter colored blocks indicate individuals that had only un-scored loci in the focal positions and are therefore consistent with either sharing the LOH of the darker colored individuals or having the ancestral heterozygous genotype. Regions marked in grey contained un-scored loci that were heterozygous in the individuals lacking the LOH fragment, and may therefore actually represent two or more smaller homozygous fragments, each < 1 Mb. Figures beneath each lineage name are (from top to bottom): i. The percentage of 1,077 scaffolds that contain some heterozygosity in the ancestral state; ii. The number of loci that are heterozygous in the ancestral genotype; iii. The average heterozygosity per individual, calculated as the percentage of heterozygous loci among the loci that were inferred to be heterozygous in the lineage ancestor (for calculations, see Supplemental Methods: RAD-Seq Analysis); iv. The average within-colony relatedness (± SD); v. The average between-colony relatedness (± SD).

Figure 4

Schematic of the hypothesized evolutionary transitions from subsocial to eusocial to phasic eusocial, with the phase-specific expression of candidate genes throughout the C. biroi colony cycle. A) Schematic showing the compartmentalization of subsocial behaviors into eusocial queen and worker castes, and reintegration into the phasic colony cycle of C. biroi. The timing of behaviors and corresponding brood stages are indicated on the subsocial and phasic eusocial cycle. The C. biroi reproductive phase is subdivided into three stages based on the brood present: Grey – pupae (P) only; Blue – pupae and eggs (E); Orange – pupae and larvae (L). B) Whole-body gene expression for C. biroi Vgw, Vgq, Hmgcr, For and Mvl during the four stages described in panel A. Graphs show relative expression (mean ± SEM). Colors correspond to the different stages of the colony cycle in panel A. Brood stages present are also indicated on the × axis of each graph, and correspond to the colony cycle in A. Samples for the brood care phase (green) were collected at day 23, when foraging activity is highest. Sample size is indicated inside or above each column in bold. Letters above columns indicate significantly different groups (Bonferroni-corrected ANOVA (P < 0.05) with Tukey’s post-hoc tests (P < 0.05)). Numbers beneath gene names show average fold change in expression between significantly different groups. Maximum fold change for each gene is indicated in parentheses. C) Tissue-specific gene expression in some of the behavioral stages showing differences in panel B. Head and abdomen expression are indicated with light and dark colors, respectively. Graphs show mean ± SEM gene expression as described for panel B.

Figure 5

Vitellogenin sequence analysis. Previously annotated Vg genes were used to identify existing and novel Vg genes in all eight ant genomes (see Supplemental Experimental Procedures: Vitellogenin Annotation and Phylogeny). A) Phylogeny of all sequenced ant species with their corresponding Vg loci mapped. Maximum likelihood tree based on first and second codon positions constructed with RAxML (GTR+G model) [53] using 3,164 orthologous single-gene families present in all ants, A. mellifera, N. vitripennis (not shown) and the outgroup D. melanogaster (not shown) (see Supplemental Experimental Procedures: Phylogeny Reconstruction and Gene Expansions). Bootstrap support values (100 replicates) for all nodes are 100%. A. mel – Apis mellifera; H. sal – Harpegnathos saltator; C. bir – Cerapachys biroi; L. hum – Linepithema humile; C. flo – Camponotus floridanus; P. bar – Pogonomyrmex barbatus; S. inv – Solenopsis invicta; A. ech – Acromyrmex echinatior; A. cep – Atta cephalotes. Arrows indicate direction of transcription. Colors correspond to the reproduction-associated and brood care-associated Vg genes (Vgq (blue) and Vgw (green), respectively). Grey genes indicate an orthologous lipid transport protein immediately upstream of all hymenopteran Vgs. A tandem duplication occurred at the base of the Formicoid clade, followed by several independent duplications in different Formicoid lineages. B) Maximum likelihood phylogram of ant vitellogenin (Vg) genes. Colors and abbreviated species names correspond to those in A. Vgq and Vgw clades are indicated with solid bars. Bootstrap values based on 1,000 replicates are given for each node. Branch lengths indicate substitutions per site.