Simultaneous radiation of bird and mammal lice following the K-Pg boundary
Abstract
The diversification of parasite groups often occurs at the same time as the diversification of their hosts. However, most studies demonstrating this concordance only examine single host–parasite groups. Multiple diverse lineages of ectoparasitic lice occur across both birds and mammals. Here, we describe the evolutionary history of lice based on analyses of 1107 single-copy orthologous genes from sequenced genomes of 46 species of lice. We identify three major diverse groups of lice: one exclusively on mammals, one almost exclusively on birds and one on both birds and mammals. Each of these groups radiated just after the Cretaceous–Paleogene (K-Pg) boundary, the time of the mass extinction event of the dinosaurs and rapid diversification of most of the modern lineages of birds and mammals.
1. Introduction
Parasites make up a very large fraction of all species on the Earth. The diversification of parasites, including herbivorous insects, is often suggested to be linked to the diversification of their hosts [1,2]. For example, several major lineages of herbivorous insects diversified around the same time as the major diversification of flowering plants [3–7]. However, the timing of diversification of the parasites of terrestrial vertebrates is less well understood. Birds and mammals are two groups of vertebrates that independently and rapidly diversified shortly after the Cretaceous–Paleogene (K-Pg) boundary, following the mass extinction of dinosaurs [8–11]. Ectoparasitic lice occur on both birds and mammals [12]. Thus, this group of insect parasites provides a unique opportunity to evaluate whether the timescale over which these ectoparasites diversified is similar to that of their two groups of hosts.
To reconstruct the evolutionary history of this group of parasites, we sequenced the genomes of 46 species of parasitic lice and two free-living bark lice outgroups, including representatives of all major louse groups that parasitize bird and mammal species throughout the world. From these genomes, we targeted and assembled 1107 single-copy orthologue genes and assembled these genes using an automated target restricted assembly method (aTRAM) [13]. A phylogenomic dataset was constructed from these genes and the resulting evolutionary tree was dated using several calibration points.
2. Material and methods
Total genomic DNA was extracted from single or pooled specimens (electronic supplementary material, table S1) of crushed parasitic lice (46 ingroup species) or bark lice (two outgroup species). Sonication was used to obtain fragment sizes ranging from 200 to 600 bp, which were prepared as paired-end libraries and sequenced for either 101 or 161 cycles on Illumina HiSeq2000 or 2500 machines to produce between 4 and 15 Gbp of raw genome sequence per species (NCBI SRA; see electronic supplementary material, table S1).
From aTRAM libraries, assemblies of 1107 single-copy 1 : 1 orthologue genes [14,15] were performed using Pediculus humanus [16] as the target for three iterations. Contigs in the resulting best file were passed through Exonerate [17] using the original target gene as the reference to identify intron–exon boundaries and concatenate exons into a single protein-coding gene (cDNA) annotated with respect to the reference sequence [14].
The resulting cDNA sequences were translated into amino acid sequences and aligned across taxa for each gene with UPP v. 2.0 [18]. The resulting aligned DNA sequences (deposited in Dryad upon acceptance) indicated substantial variation in base composition for the third codon positions, but not for the first and second codon positions, so third base positions were removed in further analyses. A k-means clustering approach [14] was used to develop a partitioning scheme and ML searches were conducted with RAxML v. 8.2.4 [19] using the GTR + GAMMA model and 100 bootstrap replicates.
A coalescence-based approach was also employed using an ASTRAL v. 4.9.9 [20,21] analysis of individual gene trees derived from RAxML analyses of individual genes to account for potential gene tree differences owing to stochastic lineage sorting. This analysis (electronic supplementary material, figure S1) produced a nearly identical tree topology to the concatenated RAxML analysis (figure 1), with the exception of only two changes in taxon position.
Figure 1. Phylogeny of parasitic lice (Phthiraptera) based on RAxML partitioned concatenated maximum-likelihood analysis of 1107 single-copy orthologue genes assembled using aTRAM from genome sequencing reads. Numbers indicate bootstrap support. Bird (red–light) and mammal (blue–dark) hosts indicated by colours. Pie charts are relative likelihoods for ancestral state (bird or mammal host) reconstruction under the equal rates model. Blue hashes on branches indicate the four inferred switches to mammals if ancestral host is a bird, and red hashes indicate four inferred switches to birds if ancestral host is a mammal. Images indicate generalized host associations for louse genera. (Attributions for images provided in electronic supplementary material, table S2.) Branch lengths are calculated from MCMCtree analysis with the ML topology and four calibration points (see §2). Scale unit is Mya. (Online version in colour.)
A molecular dating analysis was performed using relaxed clock models implemented in MCMCTree [22] of the concatenated dataset with only the first and second positions and a GTR + GAMMA model. Four internal calibration points were used: (i) oldest fossil Liposcelididae [23], 100 Myr minimum age, (ii) fossil Menoponidae [24], 44 Myr minimum age, (iii) cospeciation between Old World primates and Great Apes and their lice (20–25 Myr range [25]), (iv) cospeciation between humans and chimpanzees and their lice (5–7 Myr range [25]). The root of the tree was set to 200 Myr ago (Mya) based on fossil evidence [26–28] and was used to calibrate the substitution rate across the tree as well as to provide a maximum age calibration, and varying this from 150 to 250 Mya had almost no impact on the results. Ancestral states (bird or mammal) were reconstructed by maximum-likelihood using the ‘ace’ command in the R package APE [29] under the ‘equal rates’ model, which could not be rejected in favour of the more complex ‘all-rates-different’ model.
3. Results and discussion
The evolutionary tree estimated from our phylogenomic dataset identified three major groups of lice: one exclusively on mammals, one almost exclusively on birds and one found on both birds and mammals (figure 1). Each of these three groups began to diversify after 66 Myr, the time of the mass extinction of the dinosaurs and subsequent major radiations of modern bird and mammal lineages [9–11]. Four host-switching events are inferred to have occurred between birds and mammals, and, although the direction of these events is ambiguous, each of these happened during the early stages of louse diversification, between 50 and 66 Mya (figure 1).
Birds and mammals, which are distantly related, are the only hosts of parasitic lice. Although parasitism by lice is inferred to have arisen once between 90 and 100 Mya (figure 1), extant representatives of three major lineages of lice did not begin to diversify further until their bird and mammal hosts underwent major radiations following the K-Pg boundary. While it is possible that dinosaurs may have hosted these early lineages of lice [30], it is also plausible that lice persisted on early bird or mammal ancestors. In general, host switching between birds and mammals seems to have been relatively rare and occurred very early in the diversification of lice. Lice are faced with the prospects of needing to remain attached to the host and avoid host preening and grooming. Many lice have specific morphological and behavioural adaptations for attachment and avoiding host defence [12], which may make switching between mammals and birds difficult.
This phylogeny also sheds new light on how the lice of mammals and birds are related to each other. We found the sucking lice (Anoplura) to be embedded within the chewing lice (Amblycera, Ischnocera and Rhychophthirina), suggesting that lice with specialized piercing–sucking mouthparts for sucking blood are derived from mandibulate chewing lice [30]. Previous studies [31,32] have been limited both by incomplete taxon sampling, with the absence of major families, or by being based on very limited data (a few gene regions). We focused particularly on sampling distinctive lineages of mammalian lice (detailed below) that previously were poorly represented in prior work, including representatives of all four suborders. These included the elephant louse (Haematomyzus elephantis), representing a small distinct suborder (Rhynchophthirina) of chewing lice found only on elephants, warthogs and red river hogs [33]. These lice have chewing mandibles at the end of the long rostrum that they use to chew through the thick hides of their hosts to feed on blood. We found these lice to be the closest relatives of the blood-sucking lice (Anoplura), also found only on mammals. These chewing lice are thought to be the transition between chewing and sucking lice, which have piercing–sucking mouthparts.
We also sampled both of the major lineages of chewing lice within the major suborder Ischnocera that occur on mammals (the family Trichodectidae and the genus Trichophilopterus, which is placed in a family, Philopteridae, of otherwise avian lice). The family Trichodectidae is widespread on a variety of mammalian groups [33]. Surprisingly, we found this family to be the closest relative of the elephant louse and the blood-sucking lice, rather than related to the remaining Ischnocera. Interestingly, this relationship reduces the number of major host-switching events needed to explain the host distribution of these parasites. In particular, there is a larger group of lice parasitizing only mammals than would be predicted by classical taxonomy, suggesting that a reclassification of these groups is needed.
We found the Amblycera to be the earliest diverging group of lice. Importantly, we included samples of all six families within this group, three of which occur on mammals and three of which occur on birds. We found lice in the family Boopiidae (represented here by Heterodoxus), which typically occur on Australian marsupial mammals, to be the earliest diverging lineage in this group and not closely related to the other Amblycera (Gyropidae and Trimenoponidae, represented by Macrogyropus and Cummingsia, respectively) of mammals. Rather, these two families (Gyropidae and Trimenoponidae) comprised the nearest relatives of Menoponidae (represented by Osborniella and Myrsidea), the largest family of Amblycera widespread on birds.
These data are the first strong evidence for how the families of Amblycera are related to each other, which was largely unknown. Because members of the Trimenoponidae occur on marsupial mammals from South America and the Boopiidae on marsupial mammals in Australia, it might have been expected that these two families would be related, perhaps reflecting the break-up of Gondwana. However, we found that these mammal lice diverged from avian lice around 50 Mya, long after the break-up of Gondwana and long after the beginning of the diversification of marsupials. In summary, we found that the three major groups of lice all began to radiate shortly after the K-Pg boundary, the time when their hosts rapidly diversified.
Ethics
Research on animals was conducted according to University of Illinois IACUC protocols 10119, 13121 and 15212.
Data accessibility
Data reported in this paper are deposited in NCBI SRA (see electronic supplementary material, table S1 for Accession Numbers) and Dryad ([34]; http://dx.doi.org/10.5061/dryad.m453vh1).
Authors' contributions
K.P.J. and J.M.A. designed the research. J.M.A., N.-p.N., B.M.B. and A.D.S. performed the research and analyses. K.P.J., N.-p.N., J.M.A. and T.W. provided interpretation of the data. K.P.J. and T.W. provided resources for the study. K.P.J. wrote the paper and all authors contributed to editing the paper. All authors agree to be held accountable for the content herein and approve of the final version of the manuscript.
Competing interests
We declare we have no competing interests.
Funding
This work was supported by NSF DEB-0612938, DEB-1239788 and DEB-1342604 to K.P.J.; NSF XSEDE DEB-16002 to K.P.J., J.M.A. and B.M.B.; NSF DEB-1310824 to B.M.B.; NSF XSEDE ASC160042 to N.-p.N. and NSF DBI-1461364 and ABI-1458652 to T.W.
Acknowledgements
We thank R Wilson, DH Clayton, TD Galloway, E Osnas, T Spradling, L Mugisha, AK Saxena, T Chesser, E DiBlasi, F Daunt, V Smith, R Furness, E Mockford, J Jankowski, R Faucett, B O'Shea, K McCracken, RE Junge, J Weckstein, A Lawrence, KC Bell and MS Leonardi for assistance with obtaining specimens for this study. We thank A Paterson and N Longrich for comments on the manuscript.