Evolutionary history of assassin bugs (insecta: hemiptera: Reduviidae): insights from divergence dating and ancestral state reconstruction

Abstract

Assassin bugs are one of the most successful clades of predatory animals based on their species numbers (∼6,800 spp.) and wide distribution in terrestrial ecosystems. Various novel prey capture strategies and remarkable prey specializations contribute to their appeal as a model to study evolutionary pathways involved in predation. Here, we reconstruct the most comprehensive reduviid phylogeny (178 taxa, 18 subfamilies) to date based on molecular data (5 markers). This phylogeny tests current hypotheses on reduviid relationships emphasizing the polyphyletic Reduviinae and the blood-feeding, disease-vectoring Triatominae, and allows us, for the first time in assassin bugs, to reconstruct ancestral states of prey associations and microhabitats. Using a fossil-calibrated molecular tree, we estimated divergence times for key events in the evolutionary history of Reduviidae. Our results indicate that the polyphyletic Reduviinae fall into 11-14 separate clades. Triatominae are paraphyletic with respect to the reduviine genus Opisthacidius in the maximum likelihood analyses; this result is in contrast to prior hypotheses that found Triatominae to be monophyletic or polyphyletic and may be due to the more comprehensive taxon and character sampling in this study. The evolution of blood-feeding may thus have occurred once or twice independently among predatory assassin bugs. All prey specialists evolved from generalist ancestors, with multiple evolutionary origins of termite and ant specializations. A bark-associated life style on tree trunks is ancestral for most of the lineages of Higher Reduviidae; living on foliage has evolved at least six times independently. Reduviidae originated in the Middle Jurassic (178 Ma), but significant lineage diversification only began in the Late Cretaceous (97 Ma). The integration of molecular phylogenetics with fossil and life history data as presented in this paper provides insights into the evolutionary history of reduviids and clears the way for in-depth evolutionary hypothesis testing in one of the most speciose clades of predators.

Figure 1. Maximum Likelihood phylogram with representative habitus images of reduviine clades.

Best tree (score = −83447.290932) based on RAxML analysis of 178 taxa using a partitioned molecular dataset of 5 gene regions (16S, 18S, 28S D2, 28S D3–D5, Wg) aligned with MAFFT G-INS-i. Bootstrap values are indicated on branches by colored triangles according to support strength (explained by inset). Reduviinae lineages are indicated as red branches and remaining reduviids as blue while outgroup taxa are black. Habitus images of Reduviinae species with RCW specimen ID numbers are grouped (A–L) according to the 11 separate reduviine clades. The shaded red box highlights members of the hematophagous Triatominae, here shown as paraphyletic. Red arrowheads refer to the polyphyletic Cetherinae; the asterisk refers to Physoderinae nested within a reduviine clade.

Figure 2. Strict consensus of 16 equally parsimonious trees with representative habitus images of reduviid subfamilies.

Shortest trees (tree length = 23413, C.I. = 0.21, R.I. = 0.57) generated by TNT using the same molecular dataset (178 taxa, G-INS-i aligned, 5 gene regions) with bootstrap values indicated by colored triangles on branches (explained by inset). Reduviinae lineages are indicated as red branches and other subfamilies as blue while outgroup taxa are black. Habitus images of reduviids with RCW specimen ID numbers are labeled 1–18 according to subfamily membership indicated beside the phylogeny. Reduviinae are separated into 14 clades here and Triatominae + Opisthacidius form an unresolved polytomy (red arrowhead).

Figure 3. Ancestral state reconstructions based on best maximum likelihood tree.

A. Microhabitats. Microhabitats of terminal taxa mapped onto ML best tree using Mesquite parsimony (P) model and maximum likelihood (ML) model in BayesTraits. Branches are color coded to represent different microhabitats (see color legends) based on parsimony and similarly-colored pie-charts represent probabilities generated from BayesTraits. Terminals without colored squares indicate unknown microhabitats and are coded as missing information in the matrix. Bark-associated lifestyle (brown arrowhead) is ancestral for all Higher Reduviidae except Peiratinae and Emesinae under both P and ML. Foliage-living (green arrowheads) has evolved at least six times independently within Reduviidae. Ancestral condition for all reduviids (red asterisk) remains ambiguous (bark associated/ground-dwelling/foliage-living) under P but ML favors bark-association (96.39%). Ancestral condition for Triatominae + Opisthacidius is mammal/“reptile” nest dwelling (red arrowhead). B. Prey Specialization. Prey specialization of terminal taxa mapped onto ML best tree using Mesquite parsimony (P) model and maximum likelihood (ML) model in BayesTraits. Branches and pie-charts (from ML) are color coded to represent different targeted prey (see color legends). Terminals without colored squares indicate unknown diets and are coded as missing information in the matrix. Ancestral condition for all reduviids is generalist predator (red asterisk). Hematophagy (red arrowheads) may have evolved once or twice independently under P while ML favors a single evolution (99.62%). Termite-specialization (cyan arrowheads) occurred at least three times independently while ant-specialization (black slanted arrowheads) evolved at least twice (Holoptilinae, Acanthaspis clade).

Figure 4. Divergence time estimates based on BEAST analysis using relaxed-clock model and 11 fossil calibration points.

Chronogram based on same G-INS-i aligned molecular dataset (178 taxa; 5 gene regions: 16S, 18S, 28S D2, 28S D3–D5, Wg), using unlinked substitution models (GTR+Γ+I), relaxed clock uncorrelated lognormal and 11 fossils as priors. Lineages are colored on the chronogram as follows: Outgroup taxa (black), Phymatine Complex (green), Ectrichodiinae (pink), Triatominae (red), all other reduviid subfamilies (blue). Posterior probabilities are indicated on branches by colored triangles (see inset). Shaded node bars indicate 95% highest posterior density (HPD) credibility intervals for clades of interest only. Placement of fossils as calibration points of clades indicated by red stars.