Phylogenomics and the evolution of
Kevin P. Johnson
, Christopher H. Dietrich
, Frank Friedrich
, Rolf G. Beutel
, Benjamin Wipfler
, Ralph S. Peters
Julie M. Allen
, Malte Petersen
, Alexander Donath
, Kimberly K. O. Walden
, Alexey M. Kozlov
, Lars Podsiadlowski
, Karen Meusemann
, Alexandros Vasilikopoulos
, Robert M. Waterhouse
, Stephen L. Cameron
, Daniel R. Swanson
, Diana M. Percy
, Nate B. Hardy
, Irene Terry
, Shanlin Liu
, Xin Zhou
, Hugh M. Robertson
, and Kazunori Yoshizawa
Illinois Natural History Survey, Prairie Research Institute, University of Illinois at Urbana–Champaign, Champaign, IL 61820;
Institut für Zoologie,
Universität Hamburg, 20146 Hamburg, Germany;
Institut für Zoologie und Evolutionsforschung, Friedrich-Schiller-Universität Jena, 07743 Jena, Germany;
Center of Taxonomy and Evolutionary Research, Arthropoda Department, Zoological Research Museum Alexander Koenig, 53113 Bonn, Germany;
Department of Biology, University of Nevada, Reno, NV 89557;
Center for Molecular Biodiversity Research, Zoological Research Museum Alexander
Koenig, 53113 Bonn, Germany;
Department of Entomology, University of Illinois at Urbana–Champaign, Urbana, IL 61801;
Scientific Computing Group,
Heidelberg Institute for Theoretical Studies, 69118 Heidelberg, Germany;
Institute of Evolutionary Biology and Ecology, University of Bonn, 53121 Bonn,
Evolutionary Biology and Ecology, Institute for Biology I (Zoology), University of Freiburg, 79104 Freiburg, Germany;
Australian National Insect
Collection, Commonwealth Scientific and Industrial Research Organisation National Research Collections Australia, Acton, ACT 2601 Canberra, Australia;
Department of Ecology and Evolution, University of Lausanne and Swiss Institute of Bioinformatics, 1015 Lausanne, Switzerland;
Entomology, Purdue University, West Lafayette, IN 47907;
Department of Entomology, University of California, Riverside, CA 92521;
Department of Life
Sciences, Natural History Museum, London, SW7 5BD United Kingdom;
Department of Botany, University of British Columbia, Vancouver V6T 1Z4, Canada;
Department of Entomology and Plant Pathology, Auburn University, Auburn, AL 36849;
School of Biological Sciences, University of Utah, Salt Lake City,
BGI-Shenzhen, Shenzhen, 518083 Guangdong Province, People’s Republic of China;
Department of Entomology, China Agricultural University,
100193 Beijing, People’s Republic of China; and
Systematic Entomology, Hokkaido University, Sapporo, 060-8589 Japan
Edited by David M. Hillis, The University of Texas at Austin, Austin, TX, and approved October 25, 2018 (received for review September 13, 2018)
Hemipteroid insects (Paraneoptera), with over 10% of all known
insect diversity, are a major component of terrestrial and aquatic
ecosystems. Previous phylogenetic analyses have not consistently
resolved the relationships among major hemipteroid lineages. We
provide maximum likelihood-based phylogenomic analyses of a
taxonomically comprehensive dataset comprising sequences of
2,395 single-copy, protein-coding genes for 193 samples of hemi-
pteroid insects and outgroups. These analyses yield a well-supported
phylogeny for hemipteroid insects. Monophyly of each of the three
hemipteroid orders (Psocodea, Thysanoptera, and Hemiptera) is
strongly supported, as are most relationships among suborders
and families. Thysanoptera (thrips) is strongly supported as sister
to Hemiptera. However, as in a recent large-scale analysis sam-
pling all insect orders, trees from our data matrices support
Psocodea (bark lice and parasitic lice) as the sister group to the
holometabolous insects (those with complete metamorphosis). In
contrast, four-cluster likelihood mapping of these data does not
support this result. A molecular dating analysis using 23 fossil
calibration points suggests hemipteroid insects began diversify-
ing before the Carboniferous, over 365 million years ago. We also
explore implications for understanding the timing of diversifica-
tion, the evolution of morphological traits, and the evolution of
mitochondrial genome organization. These results provide a phy-
logenetic framework for future studies of the group.
The hemipteroid insect orders, Psocodea (bark lice and para-
sitic lice), Thysanoptera (thrips), and Hemiptera (true bugs and
allies; i.e., hemipterans), with over 120,000 described species,
comprise well over 10% of known insect diversity. However, the
evolutionary relationships among the major lineages of these insects
are not yet resolved. Recent phylogenomic analyses questioned the
monophyly of this group (1) demanding a reconsideration of the
evolution of hemipteroid and holometabolous insects. We assess
these prior results, which placed Psocodea as the sister taxon to
Holometabola (insects with complete metamorphosis; e.g., wasps,
flies, beetles, butterflies), and uncover relationships within and
among hemipteroid insect orders by analyzing a large phylogenomic
dataset covering all major lineages of hemipteroid insects.
Knowledge of the phylogeny of these insects is important for
several reasons. First, major transitions between the mandibulate
(chewing) mouthpart insect groundplan and “piercing–sucking”
mouthparts occurred in this group. In particular, thrips and
hemipterans, and some ectoparasite lice in Psocodea, have highly
modified mouthparts adapted for feeding on fluids and, hence,
differ markedly from their mandibulate ancestors. Through a series
of remarkable modifications, hemipteroids acquired a piercing–
sucking mode of feeding in both immature and adult stages that
enabled them to feed not only on plant vascular fluids, but also
on blood and other liquid diets. Resolution of the evolutionary
tree of hemipteroid insects is needed to provide a framework for
Hemipteroid insects constitute a major fraction of insect diversity,
comprising three orders and over 120,000 described species. We
used a comprehensive sample of the diversity of this group
involving 193 genome-scale datasets and sequences from 2,395
genes to uncover the evolutionary tree for these insects and pro-
vide a timescale for their diversification. Our results indicated that
thrips (Thysanoptera) are the closest living relatives of true bugs
and allies (Hemiptera) and that these insects started diversifying
before the Carboniferous period, over 365 million years ago. The
evolutionary tree from this research provides a backbone frame-
work for future studies of this important group of insects.
Author contributions: K.P.J., C.H.D., F.F., R.G.B., B.W., R.S.P., K.M., X.Z., B.M., H.M.R., and
K.Y. designed research; K.P.J., C.H.D., R.G.B., B.W., R.S.P., J.M.A., M.P., A.D., K.K.O.W.,
A.M.K., L.P., C.M., K. M., A.V., R.M.W., S.L. , X.Z., and K.Y. performed re search; K.P.J.,
C.H.D., F.F., B.W., R.S.P., K.M., C.W., D.R.S., D.M.P., N.B.H., I.T., and K.Y. contributed new
reagents/analytic tools; K.P.J., C.H.D., R.G.B., B.W., R.S.P., J.M.A., M.P., A.D., K.K.O.W.,
A.M.K., L.P., C.M., K.M., A.V., R.M.W., and K.Y. analyzed data; and K.P.J., C.H.D., F.F.,
R.G.B., B.W., R.S.P., J.M.A., M.P., A.D., K.K.O.W., A.M.K., L.P., C.M., K.M., A.V., R.M.W.,
S.L.C., C.W., D.R.S., D.M.P., N.B.H., I.T., S.L., X.Z., B.M., H.M.R., and K.Y. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Published under the PNAS license.
Data deposition: Thedata reported inthis paper have beendeposited in NCBI(accession nos.
SRA SRR1821891–SRR1821980,SRR2051465–SRR2051515,andSRR921611–SRR921660). Gene
sets, alignments,trees, quartet likelihood mapping results, morphologicaldata matrices, and
dating analyses results were deposited in Dryad repository, 10.5061/dryad.t4f4g85.
To whom correspondence should be addressed. Email: email@example.com.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
www.pnas.org/cgi/doi/10.1073/pnas.1815820115 PNAS Latest Articles
understanding morphological transitions that occurred in this group, as
well as to provide a timeframe over which these changes occurred.
In addition, several lineages of hemipteroid insects (particularly
thrips and Psocodea) underwent major reorganizations of their
mitochondrial genomes, including the emergence of minicircles
(2). Understanding how these changes in mitochondrial genome
organization occurred requires knowledge of evolutionary rela-
tionships to document in which lineages these changes first arose.
Finally, hemipteroids are among the most abundant insects (3) and
are therefore key components of terrestrial and aquatic food webs
(4). Thus, a robust backbone phylogenetic framework is needed to
place ecological studies in their evolutionary context and for use in
comparative genomic and macroevolutionary analyses.
Despite their importance, relatively few studies have addressed
the relationships among the major groups of hemipteroid insects
[Paraneoptera, sensu stricto (excluding Zoraptera), also termed
Acercaria]. While a recent large transcriptome-based phyloge-
nomic analysis of insects (1) provided a well-resolved and strongly
supported phylogenetic framework for the insect orders in gen-
eral, it did not sample intensively within individual orders and
recovered some unexpected relationships. Among the most puz-
zling was the nonmonophyly of the hemipteroid insects, with
Psocodea as the sister taxon of holometabolous insects rather than
as sister to thrips plus hemipterans (Condylognatha). Although
this result was congruent with one earlier analysis based on three
nuclear protein-coding genes (5), it had not been proposed in
other molecular phylogenetic or morphological studies. Previous
morphological studies indicated monophyly of hemipteroid insects
with Psocodea sister to thrips plus hemipterans (6–9), or some-
times a group comprising thrips plus Psocodea (10, 11).
Another unexpected relationship recovered by Misof et al. (1)
was the placement of moss bugs (Coleorrhyncha) as sister to a
group comprising leafhoppers, cicadas, and relatives (Auchenor-
rhyncha) instead of sister to true bugs (Heteroptera). A recent
morphological study also found some support for moss bugs sister
to Auchenorrhyncha (12). In contrast, prior analyses based on
morphology (e.g., ref. 9) and DNA sequence data (e.g., ref. 13)
consistently placed moss bugs as sister to true bugs. An analysis of
a reduced gene set from transcriptome data (14) also recovered
moss bugs as sister to true bugs, while the full gene set placed moss
bugs as sister to Auchenorrhyncha. Analysis of mitochondrial
genomes (15) produced an even more unconventional result, with
moss bugs placed as the sister taxon of planthoppers (Fulgor-
oidea), making Auchenorrhyncha paraphyletic. Thus, it is impor-
tant to investigate the placement of moss bugs in more detail with
both expanded taxon and gene sampling.
We evaluated these possible conflicts among analyses by an-
alyzing a more comprehensive dataset comprising an increased
number of clusters of orthologous sequence groups (2,395
protein-coding, single-copy genes) as well as an increased taxon
sample within hemipteroid insects: 160 samples vs. 22 sampled
by Misof et al. (1). We included representatives of all major
hemipteroid lineages (sub- and infraorders). Outgroups com-
prised 33 species of holometabolous and nonholometabolous
insect orders. This dataset enabled us to test the hypothesis of
nonmonophyly of hemipteroid insects and also provides a more
detailed backbone framework for the hemipteroid phylogeny.
We evaluate the implications of this phylogeny for understanding
the evolution of feeding strategy, morphology, and mitochon-
drial genome organization of this major group of insects.
Phylogeny of Hemipteroid Insect Orders. Separate amino acid se-
quence alignments of the 2,395 single-copy genes across 193 ter-
minal taxa (SI Appendix,TablesS1–S4) yielded a concatenated
supermatrix of 859,518 aligned amino acid positions, which was
used in subsequent phylogenetic analyses. A concatenated nucleo-
tide sequence supermatrix of only first and second codon positions
resulted in ∼1.72 million aligned nucleotide sequence sites. Tree
reconstructions based on the nucleotide sequence data supported a
phylogenetic tree (Fig. 1 and SI Appendix,Figs.S1andS2)with172/
190 (∼90%) of all nodes supported in 100% of bootstrap replicates.
The tree based on amino acid sequence data (SI Appendix,Fig.S3)
was highly concordant with that based on nucleotide data. Analysis
of an optimized amino acid dataset (SI Appendix,Supplemental
Materials and Methods)producedatree(SI Appendix,Fig.S4)that
was identical to that based on all amino acids with respect to re-
lationships among orders, suborders, infraorders, and superfamilies,
but had some minor rearrangements within these groups.
Considering relationships within and among orders in more de-
tail, the thrips (Thysanoptera) were recovered with 100% bootstrap
support as the sister taxon of Hemiptera (i.e., monophyletic Con-
dylognatha), although only 68% of quartets supported this result in
four-cluster likelihood mapping (FcLM) (SI Appendix,TablesS5
and S6). As in the study of Misof et al. (1), Psocodea was placed as
the sister taxon of Holometabola in 100% of bootstrap replicates,
rendering hemipteroid insects paraphyletic. However, only 25% of
quartets supported Psocodea as sister to Holometabola, compared
with 67% of the quartets supporting hemipteroid insect mono-
phyly. Results from the FcLM imply that the placement of
Psocodea as sister to Holometabola is unstable and may be due
to confounding phylogenetic signal (e.g., from heterogeneous
composition of amino acid sequences, nonstationarity of sub-
stitution processes, or nonrandom distribution of missing data)
and is also dependent on the taxon sample. However, permutation
tests of these results suggested the impact of these potential
confounding signals on the topology was minor (SI Appendix,
Table S6). To evaluate whether the parasitic lice in particular
(Phthiraptera), which have elevated substitution rates compared
with other hemipteroids (16), were a possible source of conflicting
signal, we compared quartets with and without these ectoparasitic
insects as the representative of Psocodea. However, the support
from FcLM for monophyly of hemipteroid insects was highly
similar whether parasitic lice were included (66%) or not (67%).
Morphological character mapping over three possible alterna-
tive topologies (SI Appendix,Fig.S5) revealed no apomorphies
supporting Psocodea +Holometabola. In contrast, there are 14
potential apomorphies for the monophyly of Paraneoptera. These
results indicate that there is more agreement between morphology
and the FcLM results, compared with the supermatrix analyses
with all taxa. For Coleorrhyncha (moss bugs), three characters are
apomorphies for a sister relationship to Auchenorrhyncha (leaf-
hoppers and relatives) but two other characters appear to support
a sister relationship to Heteroptera (true bugs).
In general, the phylogenetic results from transcriptomes are
congruent with the generally accepted classification schemes
within these insect orders. Bark lice and parasitic lice (Psocodea)
together are monophyletic. As has been suggested based on both
morphological (17) and molecular (16, 18) analyses, the parasitic
lice are embedded within free-living bark lice, being the sister
taxon of book lice (Liposcelididae), which makes the bark lice
(“Psocoptera”) paraphyletic. In contrast to results based on 18S
rDNA sequences (18), parasitic lice (Phthiraptera) were sup-
ported as a monophyletic group in our analyses, which included
representatives of all four suborders of parasitic lice.
The thrips (Thysanoptera) were found to be monophyletic. The
thrips family Phlaeothripidae was recovered as the sister taxon to
the remaining thrips (Aeolothripidae +Thripidae), congruent
with previous molecular analyses and the current classification of
Thysanoptera into the suborders Tubulifera (i.e., Phlaeothripidae)
and Terebrantia (all other thrips) (19).
The order Hemiptera was also monophyletic. Within Hemi-
ptera, Sternorrhyncha (whiteflies, psyllids, scales, and aphids) was
recovered as the sister taxon of the remaining hemipterans. Re-
cent classification schemes (20) and prior molecular studies (13,
21) have placed the enigmatic moss bugs as the sister taxon of true
bugs. However, our results recovered moss bugs as the sister
taxon of Auchenorrhyncha (leafhoppers, planthoppers, and rel-
atives), which was also found by Misof et al. (1). In FcLM
analyses, 96% of quartets placed moss bugs with Auchenor-
rhyncha, suggesting little underlying conflict in the data for this
result (SI Appendix, Table S6).
www.pnas.org/cgi/doi/10.1073/pnas.1815820115 Johnson et al.
Within Sternorrhyncha, whiteflies (Aleyrodoidea) were sister
to the remainder of the suborder, and psyllids (Psylloidea) were
sister to a clade composed of aphids (Aphidoidea) +scale insects
(Coccoidea), also supported by 91% of quartets in FcLM analyses.
Previous phylogenetic analyses of Sternorrhyncha have tended to
focus within particular superfamilies or families (e.g., refs. 22–24)
Carboniferous Permian Jurassic CretaceousTriassic Paleog. Neo. Q.
bark lice and
Fig. 1. Dated phylogeny of hemipteroid insects (Hemiptera, Thysanoptera, and Psocodea) based on maximum likelihood analysis of a supermatrix of first and
second codon position nucleotides corresponding to 859,518 aligned amino acid positions from transcriptome or genome sequences of 193 samples. Colored
circles indicate bootstrap support. Timescale in millions of years (Bottom) estimated from MCMCTree Bayesian divergence time analyses using 23 fossil
calibration points and a reduced dataset. Number of species sampled from each group indicated in parentheses. Higher taxa are indicated as taxon labels and
below branches; most convenient generalized common names are above branches. Images represent five major groups: Heteroptera, Auchenorrhyncha,
Sternorrhyncha, Thysanoptera, and Psocodea.
Johnson et al. PNAS Latest Articles
rather than addressing relationships among major lineages
The earliest molecular phylogenetic analyses of Hemiptera (e.g.,
refs. 25 and 26) failed to recover Auchenorrhyncha as a mono-
phyletic group, as has a more recent analysis of mitochondrial
genomes (15). However, our analyses provided strong support for
monophyly of this group, corroborating results of other studies
based on multiple loci (13, 14). Within Auchenorrhyncha, our re-
sults strongly support the taxonomic status of the two recognized
infraorders Fulgoromorpha (i.e., Fulgoroidea, planthoppers) and
Cicadomorpha (leafhoppers/treehoppers, spittlebugs, and cicadas)
as monophyletic, as found previously (13). However, relationships
among the three superfamilies of Cicadomorpha were inconsis-
tently resolved. Cicadas (Cicadoidea) plus spittlebugs (Cercopoi-
dea) were sister to leafhoppers/treehoppers (Membracoidea) in the
analysis of nucleotide sequences (Fig. 1, FcLM 52% of quartets),
but cicadas were sister to spittlebugs plus leafhoppers/treehoppers
in the analysis of amino acid sequence data (SI Appendix,Fig.S1),
which was also found in 48% of quartets of nucleotide data in
Relationships among the earlier diverging lineages of true bugs
(Heteroptera) have not been resolved consistently across previous
analyses (14, 27–29), in which the deepest divergences received low
statistical branch support and recovered different relationships
among infraorders. In our analysis, which included representatives
of all seven currently recognized infraorders, the four infraorders
for which more than one species was included were found to be
monophyletic. Like two recent studies based on combined molec-
ular and morphological data (29) and transcriptome data (14),
we found 100% bootstrap support for (i) a clade comprising litter
bugs (Dipsocoromorpha), unique-headed bugs (Enicocephalo-
morpha), and semiaquatic bugs (Gerromorpha) (also found in
100% of quartets in FcLM analyses) and (ii) shore bugs (Lep-
topodomorpha) as the sister to Cimicomorpha +Pentatomo-
morpha (also found in 100% of quartets in FcLM analyses).
Divergence Time Analysis. The estimate of the root age for our
tree, the split between Paleoptera (dragonflies, damselflies, and
mayflies) and Neoptera (all other insects) at 437 million years
ago (mya) (95% CI 401–486) was only slightly older than that
estimated for this node by Misof et al. (1), at 406 mya. Di-
vergence dates for more interior nodes tended to be older than
those estimated by Misof et al. (1) and more similar to those of
Tong et al. (30), possibly due either to much denser sampling of
minimum age fossil calibration points throughout this part of the
insect tree or to different methodology (e.g., MCMCtree versus
BEAST or different prior distributions of expected ages for
Bayesian analyses). Analyses of divergence times postulated a
common ancestor of thrips and hemipterans as early as the Devo-
nian (∼407 mya, 95% CI 373–451). Radiation within Hemiptera is
also inferred to have begun in this period (∼386 mya, 95% CI 354–
427), with radiations within Sternorrhyncha, Auchenorrhyncha, and
Heteroptera having commenced by the late Carboniferous (all be-
fore 300 mya). Radiation within modern Psocodea dates to the
Carboniferous (328 mya, 95% CI 292–376), with divergence of this
lineage from other insects as early as 404 mya (95% CI 367–451).
Analysis of 2,395 protein-coding, single-copy genes derived from
transcriptomes of hemipteroid insects and outgroups provided
strong support for a backbone tree of hemipteroid insects largely
congruent with previous analyses and classification schemes. In
particular, we recovered with strong support monophyly of the
three orders of hemipteroid insects: Psocodea, Thysanoptera,
and Hemiptera. We also recovered monophyly of most currently
recognized suborders, infraorders, and superfamilies within these
groups as well as resolving relationships among these major
groups. Although the unconventional result of a sister relation-
ship between Psocodea and Holometabola of Misof et al. (1)
appeared to be robust to our substantially increased taxon
sampling based on maximum likelihood bootstrapping, it was not
supported by four-cluster likelihood mapping analyses. FcLM,
which can detect potentially confounding signal, suggests ex-
tensive underlying conflict for this result, with the majority of
quartets placing Psocodea with thrips and hemipterans, which
would imply monophyly of Paraneoptera in rooted trees. How-
ever, permutations appear to rule out several possible types of
confounding signal (e.g., among-lineage heterogeneity or non-
random distribution of missing data) in our dataset. Recent work
has suggested that bootstrap support from very large datasets
may provide an overestimate of confidence for phylogenetic re-
sults (31–33). Thus, the position of Psocodea in the insect tree is
still an open question. Monophyly of hemipteroid insects is
supported by several morphological autapomorphies (34); there-
fore, nonmonophyly of the group would imply homoplasy in these
traits. In addition, there is no known morphological apomorphy
supporting Psocodea +Holometabola (SI Appendix,Fig.S5). In
contrast, the other less conventional relationship, a clade com-
prising Coleorrhyncha and Auchenorrhyncha uncovered by Misof
et al. (1), was recovered by our trees with increased taxon sam-
pling and is supported by 96% of quartets in the FcLM analyses
and three morphological apomorphies, suggesting that this result
Divergence time estimates using a dense sampling of 23 fossil
calibration points suggest that the radiation of the hemipteroid
insect orders is relatively ancient, beginning before the early
Carboniferous, considerably older than initial expectations based
on available fossils. However, the insect fossil record of this
period is extremely fragmentary, and relatively old fossils of
modern lineages that are used as calibration points imply that
branches uniting these lineages must be older still, given that
fossil ages represent minimum ages.
Implications for Evolution of Feeding Strategy. Our phylogenetic
results generally agree with evidence from the fossil record that the
earliest hemipteroids fed on detritus, pollen, fungi, or spores (as in
most modern bark lice and thrips). Plant-fluid feeding probably
coincided with the origin of Hemiptera and was independently
derived in thrips. Today, Hemiptera is the fifth largest insect order,
surpassed only by the four major holometabolous orders (Hyme-
noptera, Coleoptera, Lepidoptera, and Diptera). It remains one of
the most abundant and diverse groups of plant-feeding insects.
Within Hemiptera, the origin of true bugs apparently coincided
with a shift from herbivory to predation, with subsequent shifts
back to herbivory (29, 35) in the more derived lineages (Pentato-
momorpha and Cimicomorpha). The two other large suborders of
Hemiptera (Auchenorrhyncha and Sternorrhyncha) feed almost
exclusively on vascular plant fluids.
Our results also suggest that the earliest hemipterans fed
preferentially on phloem. Phloem feeding remains predominant in
extant plant-feeding hemipterans, including nearly all Sternor-
rhyncha and most Auchenorrhyncha (36), while modern moss
bugs feed on phloem-like tissues in mosses (37). A shift to xylem
feeding appears to have coincided with the origin of Cicadomor-
pha (at least the crown group of this lineage), in which all cicadas
and spittlebugs retain this preference. This is also supported by the
fossil record in which the earliest leafhoppers had inflated faces
(38), indicating a preference for xylem feeding, despite the pre-
dominance of phloem feeding among modern leafhoppers and
treehoppers (Membracoidea). A shift to phloem feeding appar-
ently occurred early in the evolution of Membracoidea but at least
one reversal to xylem feeding [in Cicadellinae (sharpshooters)] has
been inferred previously (39), consistent with our results.
Implications for Morphological Evolution. Based on the conflicting
statistical support between the supermatrix analysis and four-
cluster likelihood mapping, the position of lice (Psocodea) ap-
pears to be unstable. Morphological evidence, in contrast, supports
the monophyly of hemipteroid insects (Paraneoptera). Our parsi-
mony mapping of 142 morphological characters (SI Appendix,Fig.
S5) found no apomorphies supporting Psocodea +Holometabola
but 14 apomorphies supporting hemipteroid insect monophyly.
www.pnas.org/cgi/doi/10.1073/pnas.1815820115 Johnson et al.
Some of these are reductions or losses, including the reduced
number of tarsomeres (three in modern hemipteroids), reduced
number of Malpighian tubules (four), and presence of only one
abdominal ganglionic complex. Nevertheless, these characters, to-
gether with characters of the forewing base, still appear to support
the sister group relationship between Psocodea and thrips plus
hemipterans (11, 34, 40). Thus, the phylogenetic position of Pso-
codea requires further study of morphological and molecular data.
In contrast to the equivocal support for Paraneoptera, Con-
dylognatha is strongly supported not only in the phylogenomic
analyses, but also with six morphological apomorphies. The or-
igin of this group apparently coincided with a distinct shift in
mouthpart morphology and feeding habits toward piercing and
sucking. These changes include anterior shifting of tentorial pits,
elongated and slender mandibles, stylet-like laciniae, and a
narrowed labium (SI Appendix, Fig. S5). Subsequent evolution-
ary transformations led to the very distinct and unique piercing–
sucking mouthparts of hemipterans that facilitate ingestion of
liquid from plant or animal tissues.
The sister-group relationship that we found between moss bugs
(Coleorrhyncha) and Auchenorrhyncha has not, to our knowledge,
been proposed previously in any explicit phylogenetic analysis other
than in recent phylogenomic analyses of transcriptomes (1, 14).
Traditionally, moss bugs were treated as one of three suborders of
“Homoptera”(along with Sternorrhyncha and Auchenorrhyncha),
largely based on the structure of the head. The mouthparts of moss
bugs arise posteroventrally (41), as in leafhoppers and relatives,
rather than anteriorly as in true bugs (42). Nevertheless, morpho-
logical evidence from fossil and living moss bugs, primarily from
wing structure and musculature, suggested a closer relationship to
true bugs (9, 41, 43). However, a recent comparative morphological
study (12) revealed that moss bugs share a unique derived feature of
the wing base with Auchenorrhyncha; a membranous proximal
median plate. The same study also showed that some previously
suggested morphological synapomorphies of moss bugs and true
bugs (SI Appendix,Fig.S5C) are either ambiguous or have been
misinterpreted (12). Prior molecular evidence supporting moss bugs
plus true bugs was also somewhat equivocal [ref. 13: maximum
likelihood (ML) bootstrap 83% and maximum parsimony (MP)
bootstrap 63%]. Our results support those of other transcriptome
studies (1, 14) in placing Coleorrhyncha sister to Auchenorrhyncha.
Implications for Evolution of Mitochondrial Genome Organization.
Several groups of hemipteroid insects have been shown to have
highly rearranged mitochondrial genomes (2). The sister re-
lationship between thrips and hemipterans indicates that the
heightened rates of mitochondrial (mt) genome rearrangements
observed in the lice (44) and thrips (45) are the result of con-
vergence between these two clades. Even if Psocodea is sister to
thrips plus hemipterans, and not to holometabolous insects, re-
cent analyses indicating that the ancestor of all Psocodea had a
generally standard insect mitochondrial gene order still result in
an interpretation involving convergence (46). This phylogenetic
evidence is also consistent with the absence of any shared, de-
rived gene arrangements between Psocodea and thrips, as both
have independently diverged from the inferred ancestral insect
mt genome arrangement (2, 45).
An interpretation involving convergence is also consistent with
the varying degrees of rearrangement observed within each order.
Within Psocodea, mt genomes vary wildly across different taxo-
nomic scales, from a single derived arrangement found in all
Psocomorpha (46), to wide variation within a single genus (Lip-
oscelis, ref. 47), and between closely related species of parasitic
lice. In contrast, for the thrips, mitochondrial genome arrange-
ments are relatively consistent at the family level (with only tRNA
rearrangements observed), albeit still highly rearranged relative to
the ancestral insect mt genome (48). Very few rearrangements of
any type are observed in the Hemiptera, with the vast majority of
families possessing the inferred ancestral arrangement (2).
In summary, although the exact phylogenetic position of
Psocodea remains to be resolved convincingly, our results based
on transcriptomes for hemipteroid insects provide a strong phyloge-
netic framework for future studies of genomic, morphological, eco-
logical, and behavioral characteristics of this important group of insects.
Materials and Methods
Our general approach closely followed methods described previously by
Misof et al. (1) and Peters et al. (49) for phylogenomic analyses of insect
transcriptomes (SI Appendix, Dryad repository, 10.5061/dryad.t4f4g85). Tran-
scriptomes of 140 samples of Paraneoptera were newly sequenced with
100 bp paired-end reads for this study using Illumina HiSeq2000 or HiSeq2500
machines to achieve at least 2.5 Gbp per taxon. The final taxon sample of
193 includes representatives of 97 hemipteroid families with several larger
families represented by multiple subfamilies.
All paired-end reads were assembled with SOAPdenovo-Trans (version
1.01; ref. 50) and the assembled transcripts were filtered for possible con-
taminants (SI Appendix, Table S2) as described in Peters et al. (49). The raw
reads and filtered assemblies were submitted to the NCBI SRA and TSA ar-
chives (SI Appendix, Table S1). We searched the assemblies for transcripts of
2,395 protein-coding genes that the OrthoDB v7 database (51) suggested to
be single copy across the genomes of six species (SI Appendix, Table S3) using
the software Orthograph (version beta4, ref. 52; for results of the orthology
search see SI Appendix, Table S4). Orthologous transcripts were aligned with
MAFFT (version 7.123; ref. 53) at the translational (amino acid) level. Cor-
responding nucleotide multiple sequence alignments were generated with a
modified version of the software Pal2Nal (54) (version 14).
Alignment sections that could not be discriminated from randomly aligned
regions at the amino acid level of each gene were identified with Aliscore version
1.2 (55, 56). To maximize the fit of our substitution models, we identified for
each gene the protein domains (clans, families) and unannotated regions using
the Pfam database (refs. 1 and 57 and SI Appendix,Supplemental Materials and
Methods). The phylogenetic information content of each data block was assessed
with MARE (version 0.1.2-rc) (58), and all uninformative data blocks (IC =0) were
removed. We subsequently used PartitionFinder (developer version 2.0.0-pre14,
ref. 59) to simultaneously infer the best partitioning scheme and amino acid or
nucleotide (removing third positions because of heterogeneity, SI Appendix,Fig.
S6) substitution models, using the rclusterf algorithm.
Phylogenetic treeswere inferred using a maximum likelihood approach with
ExaML version 3.0.17 (60)for both the nucleotide and amino acid datasets. We
performed 50 nonparametric bootstrap replicates mapping the support on the
best ML tree after checking for bootstrap convergence with the default
bootstopping criteria (61). An optimized dataset, which requires the presence
of at least one species from a given taxonomic group (SI Appendix,TableS5)in
each data block of the supermatrix (62), was used for testing the possible
impact of missing data at the partition level. Four-cluster likelihood mapping
(63) was used for assessing the phylogenetic signal for alternative phyloge-
netic relationships (SI Appendix,TablesS5andS6). Permutation tests in these
analyses assessed the impact of heterogeneous amino acid sequence compo-
sition among lineages, nonstationarity of substitution processes, and non-
random distribution of missing data on the inferred phylogenetic tree (1).
To understand the morphological transformations underlying the evolution
of the hemipteroid groups and to identify potential shared derived characters
(synapomorphies),we used the morphological data matrix of Friedemann et al.
(9) with 118 charactersof the entire body (with modifications from ref.12) and
additionally 25 characters associated with the wing base (8). By tracing char-
acters over the tree using maximum parsimony using Winclada (64), we
evaluated three possible phylogenetic alternatives: (i) paraphyletic Para-
neoptera and Coleorrhyncha sister to Auchenorrhyncha (result from ML
analysis of transcriptomes); (ii) monophyletic Paraneo ptera (as sugg ested by
FcLM analyses); and (iii) paraphyletic Paraneoptera, but with Coleorrhyncha
sister to Heteroptera (as suggested in previous literature).
To estimate divergence dates, we used the topology resulting from ML
analysis of first and second position nucleotides as the input tree and assigned
23 ingroup fossil calibration points (65) throughout the tree (SI Appendix,
Table S7). These calibrations were used as minimum ages in soft bound
uniform priors with a root age of 406 mya (1) as a soft bound maximum.
These priors were used in a Bayesian MCMCTree (66) molecular dating
analysis of a first and second position nucleotide dataset for which sites were
present in at least 95% of taxa.
ACKNOWLEDGMENTS. We thank E. Anton, M. Bowser, C. Bramer, T. Catanach,
D. H. Clayton, J. R. Cooley, G. Gibbs, A. Hansen, E. Hdez, A. Katz, K. Kjer, J. Light,
K. Schütte, W. Smith, K.-Q. Song, T. Sota, N. Szucsich, G. Taylor, S. Taylor,
S. Tiwari, and X. Tong for assistance with obtaining specimens; G. Meng and
Beijing Genomics Institute staff for their efforts in data curation; and O. Niehuis
Johnson et al. PNAS Latest Articles
for assistance preparing the ortholog gene set. R.M.W. was supported by Swiss
National Science Foundat ion Grant PP00P3_1706642. K.M. was supported by
David Yeates, the Schlinger Endowment, CSIRO National Research Council
Australia, J. Korb, and the University of Freiburg. This work was also sup -
ported by National Science Foundation DEB-1239788 (to K.P.J., C.H.D., and
1. Misof B, et al. (2014) Phylogenomics resolves the timing and pattern of insect evo-
lution. Science 346:763–767.
2. Cameron SL (2014) Insect mitochondrial genomics: Implications for evolution and
phylogeny. Annu Rev Entomol 59:95–117.
3. Adis J, Lubin YD, Montgomery GG (1984) Arthropods from the canopy of inundated
and terra firme forests near Manaus, Brazil, with critical considerations on the
pyrethrum-fogging technique. Stud Neotrop Fauna Environ 19:223–236.
4. Schaefer CW, Panizzi AR (2000) Heteroptera of Economic Importance (CRC Press, Boca
5. Ishiwata K, Sasaki G, Ogawa J, Miyata T, Su Z-H (2011) Phylogenetic relationships
among insect orders based on three nuclear protein-coding gene sequences. Mol
Phylogenet Evol 58:169–180.
6. Beutel RG, Gorb SN (2001) Ultrastructure of attachment specializations of hexapods
(Arthropoda): Evolutionary patterns inferred from a revised ordinal phylogeny. J Zool
Syst Evol Res 39:177–207.
7. Wheeler WC, Whiting M, Wheeler QD, Carpenter JM (2001) The phylogeny of the
extant hexapod orders. Cladistics 17:113–169.
8. Yoshizawa K, Saigusa T (2001) Phylogenetic analysis of paraneopteran orders (Insecta:
Neoptera) based on forewing base structure, with comments on monophyly of Au-
chenorrhyncha (Hemiptera). Syst Ent 26:1–13.
9. Friedemann K, Spangenberg R, Yoshizawa K, Beutel RG (2014) Evolution of attach-
ment structure in the highly diverse Acercaria (Hexapoda). Cladistics 30:170–201.
10. Whiting MF, Carpenter JC, Wheeler QD, Wheeler WC (1997) The Strepsiptera prob-
lem: Phylogeny of the holometabolous insect orders inferred from 18S and 28S ri-
bosomal DNA sequences and morphology. Syst Biol 46:1–68.
11. Kristensen NP (1991) Phylogeny of extant hexapods. The Insects of Australia, ed CSIRO
(Melbourne Univ Press, Melbourne), pp 125–140.
12. Yoshizawa K, Ogawa N, Dietrich CH (2017) Wing base structure supports Coleor-
rhyncha +Auchenorrhyncha (Insecta: Hemiptera). J Zool Syst Evol Res 55:199–207.
13. Cryan JR, Urban JM (2012) Higher‐level phylogeny of the insect order Hemiptera: Is
Auchenorrhyncha really paraphyletic? Syst Ent 37:7–21.
14. Wang Y-H, et al. (2017) When did the ancestor of true bugs become stinky? Disen-
tangling the phylogenomics of Hemiptera-Heteroptera. Cladistics, in press.
15. Li H, et al. (2017) Mitochondrial phylogenomics of Hemiptera reveals adaptive in-
novations driving the diversification of true bugs. Proc Biol Sci 284:20171223.
16. Yoshizawa K, Johnson KP (2013) Changes in base composition bias of nuclear and
mitochondrial genes in lice (Insecta: Psocodea). Genetica 141:491–499.
17. Lyal CHC (1985) Phylogeny and classification of the Psocodea, with particular refer-
ence to the lice (Psocodea: Phthiraptera). Syst Ent 10:145–165.
18. Johnson KP, Yoshizawa K, Smith VS (2004) Multiple origins of parasitism in lice. Proc
Biol Sci 271:1771–1776.
19. Buckman RS, Mound LA, Whiting MF (2013) Phylogeny of thrips (Insecta: Thysa-
noptera) based on five molecular loci. Syst Ent 38:123–133.
20. Bourgoin T, Campbell BC (2002) Inferring a phylogeny for Hemiptera: Falling into the
‘autapomorphic trap’.Denisia 4:67–82.
21. Ouvrard D, Campbell BC, Bourgoin T, Chan KL (2000) 18S rRNA secondary structure
and phylogenetic position of Peloridiidae (Insecta, Hemiptera). Mol Phylogenet Evol
22. Von Dohlen CD, Moran NA (2000) Molecular data support a rapid radiation of aphids
in the Cretaceous and multiple origins of host alternation. Biol J Linn Soc Lond 71:
23. Gullan PJ, Cook LG (2007) Phylogeny and higher classification of the scale insects
(Hemiptera: Sternorrhyncha: Coccoidea). Zootaxa 1668:413–425.
24. Percy DM, et al. (2018) Resolving the psyllid tree of life: Phylogenomic analyses of the
superfamily Psylloidea (Hemiptera). Syst Ent 43:762–776.
25. Campbell BC, Steffen-Campbell JD, Sorensen HT, Gill RJ (1995) Paraphyly of Homo-
ptera and Auchenorrhyncha inferred from 18S rDNA nucleotide sequences. Syst Ent
26. von Dohlen CD, Moran NA (1995) Molecular phylogeny of the Homoptera: A para-
phyletic taxon. J Mol Evol 41:211–223.
27. Li H, et al. (2012) The complete mitochondrial genome and novel gene arrangement
of the unique-headed bug Stenopirates sp. (Hemiptera: Enicocephalidae). PLoS One 7:
28. Weirauch C,
Stys P (2014) Litter bugs exposed: Phylogenetic relationships of Dipso-
coromorpha (Hemiptera: Heteroptera) based on molecular data. Insect Syst Evol 45:
29. Weirauch C, Schuh RT, Cassis G, Wheeler WC (2018) Revisiting habitat and lifestyle
transitions in Heteroptera (Insecta: Hemiptera): Insights from combined morpholog-
ical and molecular phylogeny. Cladistics, in press.
30. Tong KJ, Duchêne S, Ho SYW, Lo N (2015) INSECT PHYLOGENOMICS. Comment on
“Phylogenomics resolves the timing and pattern of insect evolution”.Science 349:
31. Salichos L, Rokas A (2013) Inferring ancient divergences requires genes with strong
phylogenetic signals. Nature 497:327–331.
32. Kobert K, Salichos L, Rokas A, Stamatakis A (2016) Computing the internode certainty
and related measures from partial gene trees. Mol Biol Evol 33:1606–1617.
33. Shen X-X, Hittinger CT, Rokas A (2017) Contentious relationships in phylogenomic
studies can be driven by a handful of genes. Nature Ecol Evol 1:126.
34. Yoshizawa K, Lienhard C (2016) Bridging the gap between chewing and sucking in
the hemipteroid insects: New insights from Cretaceous amber. Zootaxa 4079:229–245.
35. Cobben RH (1979) On the original feeding habits of the Hemiptera (Insecta): A reply
to Merrill sweet. Ann Entomol Soc Am 72:711–715.
36. Backus EA (1988) Sensory systems and behaviors which mediate hemipteran plant-
feeding: A taxonomic overview. J Insect Physiol 34:151–165.
37. Cronk QCB, Forest F (2017) The evolution of Angiosperm trees: From Palaeobotany to
genomics. Comparative and Evolutionary Genomics of Angiosperm Trees, eds
Groover A, Cronk Q (Springer, New York), pp 1–17.
38. Shcherbakov D (1996) Origin and evolution of the Auchenorrhyncha as shown by the
fossil record. Studies on Hemipteran Phylogeny, ed Schaeffer CW (Entomol Soc Am,
39. Dietrich CH, et al. (2017) Anchored hybrid enrichment-based phylogenomics of leaf-
hoppers and treehoppers (Hemiptera: Cicadomorpha: Membracoidea). Insect Syst
40. Beutel RG, Friedrich F, Ge S-Q, Yang X-K (2014) Insect Morphology and Phylogeny: A
Textbook for Students of Entomology (Walter de Gruyter, Berlin).
41. Spangenberg R, et al. (2013) The cephalic morphology of the Gondwanan key taxon
Hackeriella (Coleorrhyncha, Hemiptera). Arthropod Struct Dev 42:315–337.
42. Spangenberg R, Friedemann K, Weirauch C, Beutel RG (2013) The head morphology
of the potentially basal heteropteran lineages Enicocephalomorpha and Dipsocor-
omorpha (Insecta: Hemiptera: Heteroptera). Arthropod Syst Phyl 71:103–136.
43. Shcherbakov D, Popov YA (2002) Superorder Cimicidea Laicharting, 1781, Order
Hemiptera Linné, 1758. The bugs, cicadas, plantlice, scale insects, etc. History of
Insects, eds Rasnitsyn AP, Quicke DLJ (Kluwer, Dordrecht, The Netherlands), pp
44. Cameron SL, Yoshizawa K, Mizukoshi A, Whiting MF, Johnson KP (2011) Mitochon-
drial genome deletions and minicircles are common in lice (Insecta: Phthiraptera).
BMC Genomics 12:394.
45. Dickey AM, et al. (2015) A novel mitochondrial genome architecture in thrips (Insecta:
Thysanoptera): Extreme size asymmetry among chromosomes and possible recent
control region duplication. BMC Genomics 16:439.
46. Yoshizawa K, et al. (2018) Mitochondrial phylogenomics and genome rearrange-
ments in the barklice (Insecta: Psocodea). Mol Phylogenet Evol 119:118–127.
47. Shi Y, et al. (2016) The mitochondrial genome of booklouse, Liposcelis sculptilis
(Psocoptera: Liposcelididae) and the evolutionary timescale of Liposcelis.Sci Rep 6:
48. Yan D, et al. (2014) The mitochondrial genome of Frankliniella intonsa: Insights into
the evolution of mitochondrial genomes at lower taxonomic levels in Thysanoptera.
49. Peters RS, et al. (2017) Evolutionary history of the Hymenoptera. Curr Biol 27:
50. Xie Y, et al. (2014) SOAPdenovo-Trans: De novo transcriptome assembly with short
RNA-seq reads. Bioinformatics 30:1660–1666.
51. Kriventseva EV, Rahman N, Espinosa O, Zdobnov EM (2008) OrthoDB: The hierarchical
catalog of eukaryotic orthologs. Nucleic Acids Res 36:D271–D275.
52. Petersen M, et al. (2017) Orthograph: A versatile tool for mapping coding nucleotide
sequences to clusters of orthologous genes. BMC Bioinformatics 18:111.
53. Katoh K, Standley DM (2016) A simple method to control over-alignment in the
MAFFT multiple sequence alignment program. Bioinformatics 32:1933–1942.
54. Suyama M, Torrents D, Bork P (2006) PAL2NAL: Robust conversion of protein se-
quence alignments into the corresponding codon alignments. Nucleic Acids Res 34:
55. Misof B, Misof K (2009) A Monte Carlo approach successfully identifies randomness in
multiple sequence alignments: A more objective means of data exclusion. Syst Biol 58:
56. Kück P, et al. (2010) Parametric and non-parametric masking of randomness in se-
quence alignments can be improvedand leads to better resolvedtrees. Front Zool 7:10.
57. Finn RD, et al. (2014) Pfam: The protein families database. Nucleic Acids Res 42:
58. Misof B, et al. (2013) Selecting informative subsets of sparse supermatrices increases
the chance to find correct trees. BMC Bioinformatics 14:348.
59. Lanfear R, Calcott B, Kainer D, Mayer C, Stamatakis A (2014) Selecting optimal par-
titioning schemes for phylogenomic datasets. BMC Evol Biol 14:82.
60. Kozlov AM, Aberer AJ, Stamatakis A (2015) ExaML version 3: A tool for phylogenomic
analyses on supercomputers. Bioinformatics 31:2577–2579.
61. Pattengale ND, Alipour M, Bininda-Emonds ORP, Moret BME, Stamatakis A (2010)
How many bootstrap replicates are necessary? J Comput Biol 17:337–354.
62. Dell’Ampio E, et al. (2014) Decisive data sets in phylogenomics: Lessons from studies
on the phylogenetic relationships of primarily wingless insects. Mol Biol Evol 31:
63. Strimmer K, von Haeseler A (1997) Likelihood-mapping: A simple method to visualize
phylogenetic content of a sequence alignment. Proc Natl Acad Sci USA 94:6815–6819.
64. Nixon K (2002) Winclad a (Ithaca, NY), ve rsion 1.00. 08. Avai lable at http://w ww.
diversityofli fe.org/winclada/ . Accessed Januar y 24, 2018.
65. Parham JF, et al. (2012) Best practices for justifying fossil calibrations. Syst Biol 61:
66. Yang Z (2007) PAML 4: Phylogenetic analysis by maximum likelihood. Mol Biol Evol
www.pnas.org/cgi/doi/10.1073/pnas.1815820115 Johnson et al.