Jung 2010
Jung 2010
Jung 2010
(Heteroptera: Anthocoridae)
Sunghoon Jung
a
, Hyojoong Kim
a,1
, Kazutaka Yamada
b
, Seunghwan Lee
a,
a
Laboratory of Insect Biosystematics, Division of Entomology, Research Institute for Agricultural and Life Sciences, School of Agricultural Biotechnology, Seoul National
University, San 56-1 Shilim-dong, Gwanak-gu, Seoul 151-742, South Korea
b
Tokushima Prefectural Museum, Bunka-no-Mori Park, Hachiman-ch, Tokushima 770-8070, Japan
a r t i c l e i n f o
Article history:
Received 16 April 2010
Revised 27 August 2010
Accepted 17 September 2010
Available online 1 October 2010
Keywords:
Anthocoridae
Flower bugs
Minute pirate bugs
Phylogeny
Ancestral character states
Bayes traits
Lyctocoridae
Lasiochilidae
a b s t r a c t
We performed a molecular phylogenetic study of the Anthocoridae, the ower bugs, based on maximum
likelihood, maximum parsimony, and Bayesian analyses of 3000 base pairs (bp) of DNA sequence from
the mitochondrial 16S rRNA and nuclear 18S rRNA and 28S rRNA genes for 44 taxa. Our phylogenetic
analyses indicates that (i) the tribe Cardiastethini (Dufouriellini) could be a paraphyletic group, as the
genera Amphiareus and Dysepicritus are not included in the tribe; (ii) the main subgroups, Oriini and
Anthocorini, are monophyletic within Anthocoridae; (iii) three tribes of Blaptostethini, Xylocorini, and
Scolopini are separated from the main anthocorid clade which is composed of Anthocorini, Cardiastethini,
and Oriini, suggesting that Anthocoridae could not be monophyletic. We compared our molecular
phylogeny to previous hypotheses of evolutionary relationships within Cimicoidea based on different
anthocorid classication systems using alternative hypothesis tests (KishinoHasegawa and Shimodaira
Hasegawa tests). BayesTraits were used to examine the ancestral character states inferring historical
habitat patterns of the Anthocoridae. Reconstruction of the ancestral habitat patterns of the Anthocoridae
suggests that deadplants may have served as animportant habitat for the commonancestor of anthocorids.
The biological events such as diversication of angiosperms and anthocorid prey might have provided
anthocorids with more habitat options, such as living plants; thereafter, Anthocorini and Amphiareus
appeared to have evolved increasingly specialized habitat relationships.
2010 Elsevier Inc. All rights reserved.
1. Introduction
Insects in the family Anthocoridae (Hemiptera: Heteroptera),
sometimes referred to as ower bugs or minute pirate bugs, are
small in size (1.44.5 mm). This family contains approximately
400600 species that are distributed worldwide (Pricart, 1972;
Schuh and tys, 1991; Pricart, 1996). Most species of Anthocoridae
are predaceous as nymphs and adults (Anderson, 1962c; Pricart,
1972; Lattin and Stanton, 1992), thus some species, such as Orius
laevigatus, O. strigicollis and O. insidiosus, are used as biological con-
trol agents and are commercially produced and traded (Yasunaga,
1997a; Kim et al., 2008). However, some species in this family con-
sume both plants and other insects and their eggs (Carayon and
Steffan, 1959; Chu, 1969; Pricart, 1972; Salas-Aguilar and Ehler,
1977; Armer et al., 1998). A few species appear to be entirely phy-
tophagous (Bacheler and Baranowski, 1975). Lattin (1999)
published a comprehensive review of the general biology and
bionomics of Anthocoridae, including a discussion of their use as
biological control agents.
Following the proposal of the infraorder Cimicomorpha by
Leston et al. in (1954), Carayon (1972) proposed a classication
of Anthocoridae that included three subfamilies: Anthocorinae,
Lyctocorinae, and Lasiochilinae (Fig. 1A; Table 1). Subsequently,
Ford (1979) and Schuh (1986) pointed out that Carayons Anthoco-
ridae may not be a monophyletic group. Schuh and tys (1991)
constructed a cladistic tree of relationships within Cimicomorpha,
and based on this tree, they elevated Carayon (1972)s three
subfamilies to family level (Lasiochilidae, Anthocoridae, and Lycto-
coridae) (Fig. 1B), and moved all the tribes in the subfamily Lycto-
corinae to the family Anthocoridae, except for the tribe Lyctocorini,
which was treated as a separate family (Table 1). After this revi-
sion, Cassis and Gross (1995) downgraded the rank of the three
families to the subfamily-level (Table 1). By elevating the three
subfamilies of Anthocorinae, Lyctocorinae, and Lasihchilinae to
family level (sensu Schuh and tys), the hypotheses of relation-
ships within Cimicoidea also changed (e.g., the position of Plokio-
philidae and Nabidae, Figs. 1A and B). Therefore, to perform a
phylogenetic study of the Anthocoridae, the relationships of higher
taxa within Cimicoidea should also be considered.
1055-7903/$ - see front matter 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.ympev.2010.09.013
(Exgene
1
1
8
3
1
1
7
5
The complete 18S rRNA gene was amplied using primer pairs
18S-1 (5-CTGGTTGATCCTGCCAGTAGT-3)/18S-4 (5-GATCCTTCT
GCAGGTTCACC-3) (Campbell et al., 1995), and sequenced using
the internal primers 18S-2 (5-AGATACCGCCCTAGTTCTAACC-3)
and 18S-3(5-GGTTAGAACTAGGGCGGTATCT-3) with 18S-1 and
18S-4, respectively. The annealing temperature was 4850 C.
The D3 region of 28S rRNA gene and a partial fragment of the mito-
chondrial 16S rRNA were amplied using primer pairs 28S-DD
(5-GGGACCCGTCTTGAAACAC-3)/28S-FF (5-TTACACACTCCTTAGC
GGAT-3) (Hillis and Dixon, 1991) and 16S-A (5-CGCCTGTTTAACA
AAAACAT-3) (Simon et al., 1994)/16S-B (5-CCGGTTGAACTCAGAT
CA-3) (Kambhampati and Smith, 1995). The annealing tempera-
tures were 45 C and 50 C, respectively.
2.3. Alignment and characterization of gene fragments
For the alignments, we retrieved GenBank reference sequences
for the three gene regions (Tian et al., 2008) with which to compare
our sequences. Raw sequences were examined and corrected using
SeqManII (version 5.01, 2001; DNA-star). All DNA sequences
were aligned using CLUSTAL X (version 1.83, 2003; Thompson
et al., 1997; default settings), MEGA 3.1 (Kumar et al., 2004), and
MAFFT (Katoh et al., 2002; Katoh et al., 2005) of the online sever
(v.6; <http://align.bmr.kyushu-u.ac.jp/mafft/online>). As the three
genes used in this study contain lots of gaps, to avoid bias in ren-
ing alignments, we compared the result using the three aligning
programs. Some ambiguous sites in the three gene fragments were
removed using GBLOCK 0.91b (Castresana, 2002; default settings
were used except for choosing the gap option with half).
2.4. Saturation tests
Given that nucleotide saturation can result in incorrect phylo-
genetic inferences (Swofford et al., 1996), uncorrected pairwise se-
quence distances were plotted against the numbers of transition
and transversion substitutions using MEGA 3.1 (Kumar et al.,
2004) to estimate the extent to which the DNA sequences were
saturated.
2.5. Phylogenetic analyses
Maximum parsimony (MP) analysis was implemented in
PAUP*4.0b10 (Swofford, 1998) using a heuristic search procedure,
tree bisection reconnection (TBR) branch swapping, and 1000 ran-
domsequence additions with 10 trees held at each pseudoreplicate.
All characters were treated as unordered and equally weighted for
MP analysis. One thousand MP bootstrap replicates were performed
using a heuristic search procedure, with a maximum tree setting of
200 trees. Decay index values (Bremer, 1988) were also calculated
to determine branch support using TreeRot V3 (Sorenson and
Farnzosa, 2007). The partition-homogeneity test (Farris et al.,
1994), as implemented in PAUP*, was performed to test for signi-
cant phylogenetic conict between the three gene fragments.
For maximum likelihood (ML) analysis, MODELTEST version
3.06 (Posada and Crandall, 1998) was used to select the best-tting
nucleotide substitution model, and then PAUP* settings were opti-
mized using data of the selected model before searching. ML anal-
yses were also performed with PAUP* using the heuristic search
procedure with TBR branch swapping, 100 random additions of se-
quences, and 10 trees held at each pseudoreplicate. One hundred
ML bootstraps were performed using a heuristic search procedure
and a maxtree setting of 200 trees.
Bayesian phylogenetic (BP) analysis was implemented in
MRBAYES (version 3.1.1; Ronquist and Huelsenbeck, 2003) for
the single and combined datasets. We analyzed the combined
dataset using two different methods (Supplementary data 4): (1)
partitioned Bayesian analysis (PBA), in which the data were parti-
tioned into 18S rRNA, 16S rRNA and 28S rRNA segments, with BP
performed using a partition scheme that maximized the likelihood
based on the GTR + I + G model with specic model scores esti-
mated by the MODELTEST for each gene region; (2) non-specic
model scores Bayesian analysis (NBA), in which, as for ML analysis,
we applied the GTR + I + G nucleotide substitution model with var-
iable model scores to the partitioned dataset. For the BP analyses of
the above two methods, four chains (three heated and one cold)
were run, starting from a random tree and proceeding for 10 mil-
lion Markov chain Monte Carlo (MCMC) generations, sampling
the chains every 100th cycle. To ensure that the distribution had
stabilized, Tracer version 1.4 (Rambaut and Drummond, 2008)
was used to view the graphical representation of MCMC chain mix-
ing. Burn-in was set at 15% of the sampled number of trees. Con-
vergence was conrmed by monitoring likelihood values
graphically. A 50% majority-rule consensus tree was constructed
from the remaining trees to estimate posterior probabilities (PP).
2.6. Evaluation of alternative hypotheses of relationships within
Cimicoidea
To assess whether there was signicant conict between previ-
ous phylogenetic hypotheses (Fig. 1) within Cimicoidea and our
molecular phylogeny, we employed the KishinoHasegawa (KH)
test (Kishino and Hasegawa, 1989) and ShimodairaHasegawa
(SH) test (Shimodaira, 2001, 2002). First, we reconstructed alterna-
tive, fully-resolved tree topologies consistent with the previous
taxonomic schemes or hypotheses of Kerzhner (1981) and Schuh
and tys (1991) (Fig. 1) using Mesquite version 2.6 (Maddison
and Maddison, 2007). To test Kerzhners hypothesis (Fig. 1A) based
on the classication of Carayon (1972), Table 1) for Anthocoridae,
we removed Velocipedidae and Polyctenidae, and combined
Medocostinae, Nabidae, and Prostemmatinae into a single group,
Nabidae, and constrained the anthocorid taxa based on Carayons
classication (1972) (Table 1). To test Schuh and tys hypothesis
(Fig. 1B), we also excluded Polyctenidae, and made Nabidae
(Naboidea) the sister group of all the taxa within Cimicoidea based
on the hypothesized family relationships within Cimicomorpha,
and constrained the anthocorid taxa based on the Anthocoridae
classications of Schuh and tys (1991) (Table 1). Second, we per-
formed ML heuristic searches under a GTR + I + G model for each
gene partition, and topological constraints were incorporated in
RAxML v.7.0 (Stamatakis, 2006) to evaluate the highest-likelihood
topology that satised the given hypothesis. In the third step, we
compared the ML tree obtained from this study with the two topol-
ogies from the rst and second steps using 2000 bootstraps and
RELL optimization of the concatenated dataset in PAUP*4.0b10.
2.7. Ancestral character states of anthocorids
To explore the relationships between divergence and ancestral
habitat patterns of Anthocoridae, we used the habitat types of
anthocorids, and applied the Bayesian estimation of ancestral
character states referencing phylogenetic result. Outgroup taxa,
Nabadae: Nabinae, damsel bugs including sub-outgroup taxa,
Cimicidae and Plokiophilidae, ectoparasites and cobweb dwellers,
respectively, were not included in the part of the study.
2.7.1. Habitat types
The ancestral habitats of Anthocoridae were inferred based on
the habitats occupied by extant taxa. Table 2 indicates the habitats
for each species used in this study. Although the host plants of spe-
cic taxa are usually reported in taxonomic papers, we collected
most anthocorid species in this study to conrm their habitats.
Although we collected Blaptostethus aurivillus from live owers of
1176 S. Jung et al. / Molecular Phylogenetics and Evolution 57 (2010) 11731183
Macaranga (Euphorbiaceae) from Malaysia, most taxa belonging to
Blaptostethus inhabit dead plant materials such as clusters of veg-
etable refuse and the nests of weaverbirds (Carayon, 1972). There-
fore, we specied that this species inhabits both living and dead
plants in our analysis.
Habitat types were coded according to the references (Table 2)
as follows: (A) living forbs (including owers of plants); (B) living
trees (including shrubs); (C) dead leaf clusters; (D) under the bark
of dead tree. Each species was assigned to one or two habitat types
based on reference papers and information collected in this study,
with the exceptions of Anthocoris montanus, A. thibetanus, and
A. zoui because of lack of habitat information.
2.7.2. Bayesian estimation of ancestral character states
A Bayesian approach as implemented in the BayesTraits (PC ver-
sion 1.0) software package (Pagel et al., 2004; Pagel and Meade,
2007) was used to reconstruct ancestral habitat character states
for selected nodes in the partitioned Bayesian phylogenetic analy-
sis (PBA). BayesTraits uses reversible-jump Markov chain Monte
Carlo (MCMC) methods to derive posterior probabilities and the
values of traits at ancestral nodes of phylogenies (Pagel et al.,
2004). BayesMultiState was selected as model of evolution and
MCMC as the method of analysis. Because some taxa occurred in
two habitats, we used the multiple character state option of Bayes-
Multistate. For example, the code AB signied that the habitat type
could be state A or B (with equal probability) but not state C or D
(Pagel and Meade, 2007). The rate deviation was set to 10. A hyper-
prior approach was employed with an exponential prior seeded
from a uniform prior in the interval 010. Thus, acceptance rates
in the preferred range of 2040% were achieved as recommended
(Pagel et al., 2004; Pagel and Meade, 2007). A total of 50 million
iterations were run for each analysis with the rst one million
samples discarded as burn-in, with sampling every 1000th gener-
ation. Because the posterior probabilities for ancestral habitat
patterns of the single runs differed slightly, we calculated the
arithmetic mean of all samples for reconstruction of ancestral
habitat types.
3. Results
3.1. Characteristics of the three gene fragments
The alignment results of the each gene region used in this study
using the three aligning programs are almost identical to each
other. The mitochondrial 16S rRNA dataset comprised 419 aligned
base pairs (bp), but some ambiguous sites were further excluded
using GBLOCK 0.91b with the half-option. Among the selected
399 bp, 253 bp were variable and 212 bp were parsimony informa-
tive. The average of the uncorrected sequence divergence among
taxa for 16S rRNA was 16.3%, and the average proportions of
T:C:A:G were 44:10:31:15. The AT richness observed is character-
istic of insect mitochondrial DNA sequences (Tauz et al., 1988;
Crozier and Crozier, 1993). The nuclear 18S rRNA dataset com-
prised 1949 bp of aligned sequences, but some ambiguous sites
were excluded using GBLOCK 0.91b with the half-option. Among
the selected 1565 bp, 408 bp were variable and 245 bp were parsi-
mony informative. The average of the uncorrected sequence diver-
gence among taxa for 18S rRNA was 4.8%, and the average
proportions of T:C:A:G were 24:23:25:28. For the nuclear 28S
rRNA dataset, 657 bp of sequences were aligned for all taxa, but
some ambiguous sites were excluded using GBLOCK 0.91b with
the half-option. Among the nal 552 bp, 136 bp were variable
and 85 bp were parsimony informative. The average uncorrected
sequence divergence of the 28S rRNA gene among taxa was
16.3%, while the average proportions of T:C:A:G were 27:21:26:26.
3.2. Substitution patterns
Uncorrected P-distances were plotted against the number of
transitions (Ts) and transversions (Tv). Tests were performed for
all three genes and the combined dataset. All three genes and com-
bined dataset were free of saturation; Ts and Tv values increased
linearly according to increasing uncorrected P-distances.
3.3. Phylogenetic analyses
3.3.1. Analyses of the single-gene datasets
Single datasets were analyzed by Bayesian analyses using the
GTR + I + G model selected as the best-t model for each of the
three genes by MODELTEST. For the individual-level phylogenetic
analyses of the three individual datasets (18S/16S/28S rRNA; Sup-
plementary Figs. 13), the branch support values were insufcient
to resolve relationships between lower taxa, but major nodes sub-
tending generic or tribal clusters were highly supported. The
monophyly of the Oriini was supported robustly in all three anal-
yses (PB = 100), and the monophyly of the Anthocorini and the Car-
diastethini except for the two genera Amphiareus and Dysepicritus
were clearly supported in all three analyses. The monophyly of
the genus Amphiareus was clearly supported by the 18S/16S rDNA
data (PB = 100).
3.3.2. Analyses of the combined dataset
For MP analyses, the partitionhomogeneity test found signi-
cant phylogenetic conict among all possible data partitions (gen-
erally P = 0.001). Nevertheless, there was no conict between
major nodes found in the analysis of individual datasets. Thus,
the concatenated data set was analyzed. MP analysis yielded three
equally parsimonious trees with a tree length of 2722. The MP (PP)
and decay index (DI) values for the strict consensus cladogram of
the three most parsimonious trees are shown in Table 3. For ML
analyses, the GTR + I + G model selected by MODELTEST analysis
as the most suitable for the combined dataset were used. ML anal-
ysis produced a single tree which was identical to the post-burn-in
consensus tree from NBA based on the GTR + I + G model for the
combined dataset. Bootstrap P-values (PL) for nodes in the ML tree
(ln L = 16765.98997) and PB of the NBA analysis are presented in
Table 3. Nodal supports for several nodes differed between the PBA
topology and the ML and NBA topologies. The topology and nodal
support values for major nodes based on the PBA analysis are
marked with capital letters in Fig. 3 and Table 3.
The monophyly of the superfamily Cimicoidea, all the remain-
ing taxa excluding Nabidae: Nabinae, were well supported by high
P-values (Fig. 2, node A). The family Lyctocoridae with tribes, Blap-
tostethini and Xylocorini, was poorly resolved in all analyses, but
Xylocorini clustered with Cimicidae + Plokiophilidae (Fig. 2, node
L), and Blaptostethini was positioned most basally except Lasiochi-
lidae within Cimicoidea (Fig. 2 node B), both of which separate
from the main Anthocoridae node (Fig. 2, node D) in the PBA
(ML, NBA) tree. The main Anthocoridae (Fig. 2, node D), including
the three tribes Oriini, Cardiastethini, and Anthocorini, was well-
supported (Table 3). The monophyly of the tribe Anthocorini
(Fig. 2, node E; Table 3) had strong support and included two
genera, Anthocoris and Tetraphleps. The genus Anthocoris was recov-
ered as paraphyletic group, clustering together with Tetraphleps
aterrimus. The monophyly of the genus Amphiareus (A. rucollaris,
A. obscuriceps, A. constrictus and A. morimotoi) (Fig. 2, node F; Table 3)
was well supported. In contrast, the two genera, Amphiareus and
Dyspicritus were separated from the main clade of tribe Cardiasteth-
ini (Fig. 2, node I), and clustered with Anthocorini (Fig. 2, node
N + R). The remaining members of Cardiastethini (Cardiastethus
exguus, Physopleurella armata, Buchananiella crassicornis and
B. leptocephala) (Fig. 2, node I; Table 3) formed a clade with strong
S. Jung et al. / Molecular Phylogenetics and Evolution 57 (2010) 11731183 1177
support. Therefore, Cardiastethini is not monophyletic due to the
separation of nodes F, node I, and Dysepicritus rufescens. Oriini was
composed of 10 species from three genera (Bilia sp., Montandonilola
moraguesi, Orius atratus, O. laevigatus, O. niger, O. agilis, O. insidiosus,
O. sauteri, O. minutus and O. strigicollis) and was monophyletic with
strong support (Fig. 2, node G; Table 3) while the subgenus Heterorius
was strongly supported by all analyses (Fig. 2, node P; Table 3).
In terms of the relationships between tribes and genera, Oriini
and Cardiastethini without the two genera Amphiareus and
Dysepicritus clustered together with moderate support on the ML,
BP (NBA, PBA), and MP trees (Fig. 2, node H; Table 3). Amphiare-
us + Dysepicritus and Anthocorini were sister clades on ML/BP
(NBA, PBA) and MP trees with relatively low support values
(Fig. 2, nodes N, R; Table 3). In the ML/BP (NBA, PBA) analyses,
Dysepicritus rufescens was basal to both node E (Anthocorini) + node
F (Amphiareus) with high support values (Table 3), and node F
(Amphiareus) clustered with node E (Anthocorini) with low support
values (Table 3).
3.4. Comparisons of alternative hypotheses within Cimicoidea
One of the two previous phylogenetic hypotheses within Cimi-
coidea was rejected on the basis of the KishinoHasegawa (KH)
and ShimodairaHasegawa (SH) tests (Table 4, P-value <0.05).
The hypothesis of Kerzhner (1981) differs from our molecular phy-
logeny in all the placements of families within Cimicoidea, regard-
less of the rank and the classication of the three subfamilies
Lashiochilinae, Lyctocorinae, and Anthocorinae (Carayon, 1972).
The hypothesis of Schuh and tys (1991) conicts with the
molecular phylogeny in several areas, particularly the relationship
Fig. 2. Phylogenetic relationships inferred from Bayesian analysis based on a GTR + I + G model with specic model scores for each gene partition (PBA) with posterior
probabilities (PP) under nodes. Thick branches indicate nodes also recovered in the other analyses. Capital letters on the nodes refer to nodes discussed in the text and support
values from other analyses are provided in Table 3. Higher-taxa names on the right side of the topology are based on Schuh and tys (1991)s classication with colored bar
regarding to the general habitat type for each higher taxon group. Gray boxes indicate taxa used as outgroup in this study. The habitat types of each species are color-coded
according to the four patterns of habitats (Table 2). BayesMultistate analysis results of ancestral habitat type reconstructions are indicated as pie charts under the nodes
showing the relative likelihoods of each habitat type at respective nodes.
1178 S. Jung et al. / Molecular Phylogenetics and Evolution 57 (2010) 11731183
between Cimicidae and Plokiphilidae, and the placement of several
tribes of the Anthocoridae. However, as the P-values of the KH and
SHtests were greater than 0.05, our molecular phylogeny partly
supports Schuh and tys (1991)s hypotheses of relationships
within Cimicoidea and the classication of Anthocoridae (with
Lyctocoridae and Lasiochilidae, Table 1), including the family level
rank of Lasiochilidae.
3.5. Ancestral character states analysis
Ancestral habitat types were estimated for the 10 nodes of high-
er taxa relationships in the Bayesian trees (Fig. 2; pie charts).
BayesMultiState analyses allowed free rates of habitat types
exchange between the four habitat types (Table 2). The analysis
suggests that the habitat type coded as C + D (dead leaf
clusters + bark of the dead tree: dead plant; light blue and dark blue
in Fig. 2, node A) at the root of Cimicoidea (Schuh and tys, 1991)
Fig. 3. Simplied view of higher taxa relationships for the hypothesized Anthocoridae (node D) found in this study showing the patterns of the change of habit type and
range. Capital alphabets and splits 13 for the nodes are corresponded with ones in Fig. 2 and discussed in the text, respectively. Species used as representatives of higher
groups are as follows: Physopleurella armata for Cardiastethini, Orius sauteri for Oriini, Amphiareus obscuriceps for Amphiareus, and Anthocoris confusus for Anthocorini.
Table 3
Clade recovered in the four analyses, maximum parsimony (MP), maximum likelihood (ML), and Non-specic model scores Partitioned Bayesian Analysis (NBA) and Partitioned
Bayesian Analysis with specic model scores for each partition (PBA) using PAUP and MrBayes. Values for MP with Decay Index (DI) and ML are boostrap values, those for the
MrBayes posterior probabilities. Indicates that a particular clade is not recovered in that analysis or values below 50.
Taxa clades Node on Fig. 2 MrBayes PAUP
NBA PBA MP DI ML
Cimicoidea A 1 1 98 29 95
Cimicoidea excluding Lasiochilidae B 1 0.85 100 30 100
Cimicoidea excluding Blaptostethini and Lasiochilidae C 0.78 1 65 1 57
Anthocoridae D 1 1 67 2 90
Anthocorini E 1 1 100 5 100
Amphiareus F 1 1 100 26 100
Oriini G 1 1 100 52 100
Cardiastethini + Oriini H 1 1 85 2 75
Cardiastethini I 1 1 100 21 100
Anthocoridae + Lyctocoridae J 0.57
Cimicidae + Plokophilidae K 1 1 68 1 95
Xylocorini + Cimicidae + Plokophilidae L 0.95 0.94 57
Lasiochilidae M 1 1 100 12 100
Anthocorini + Amphiareus + Dysepicritus N 0.99 0.97 90
Scolopini O 1 1 100 48 100
Subgenus Heterorius P 1 1 100 12 99
Cimicoidea excluding Lasiochilidae, Blaptostethini and Scolopini Q 0.78 0.98
Amphiareus + Anthocorini R 0.88 0.85 59
Table 4
Result from KishinoHasegawa (KH) and ShimodairaHasegawa (SH) tests among
tree topologies (within Cimicoidea). Previously proposed hypotheses (see Fig. 1) were
used to nd a ML tree under those constraints, and those trees were compared with
the ML tree (best). Maximum likelihood values (L) are reported for each tree and the
difference in log likelihoods (dif Ln L) between the proposed tree and the best tree is
used to assess statistical signicance of the test. Tests were run in PAUP. *P < 0.05.
ln L Dif ln L KH-test P SH-test P
Kerzhner (1981) 16861.76850 96.99891 0.007* 0.0045*
Schuh and tys (1991) 16779.24136 14.47177 0.3245 0.438
Best 16764.76959
S. Jung et al. / Molecular Phylogenetics and Evolution 57 (2010) 11731183 1179
had a reconstructed probability of greater than 90% compared to
alternative habitat types. On the next-most distal bifurcations after
the split of Lasiochilidae from the remaining taxa within Cimicoi-
dea, the ancestral habitat types are reconstructed as types C that
code for dead leaf clusters with the reconstructed probabilities
for the type was more than 90% (Fig 2, nodes B). The habitat ranges
for ancestral character states for the ancestor of the three tribes,
Anthocorini, Oriini, and Cardiastethini (Fig. 2, nodes H, R + N) sug-
gest that a mixture of live and dead plants were the ancestral hab-
itat type for the main lineages of Anthocorini, Oriini and
Cardiastethini. However, the analyses indicated that dead plants,
especially dead leaf clusters, were more likely to have been the
habitat type of the common ancestor of these main groups than liv-
ing plants (Fig. 2, node D) although the ancestral habitats of the
two tribes, Oriini and Anthocorini, were reconstructed as living
plants (Fig. 2, nodes E, G). The origin of habitat type for Anthocorini
(Fig. 2, node E) and for Oriini (Fig. 2, node G) is clear; it was recon-
structed as living trees with a probability of more than 99%. The
ancestral habitat types for the genus Amphiareus (Fig. 2, node F)
and for Cardiastethini except for the two genera, Amphiareus and
Dysepicritus(Fig. 2, node I) are also clear the habitat type of dead
leaves had a probability of greater than 99% for both these nodes,
and is almost the identical habitat seen for the ancestor of the
Anthocoridae (Fig. 2, node D; sensu Schuh and tys, 1991), exclud-
ing the three tribes, Blaptostethini, Scolopini and Xylocorini.
4. Discussion
4.1. Monophyly and paraphyly within Anthocoridae
The MP, ML, and BP (PBA, NBA) trees supported the monophyly
of Anthocorini (Anthocoris + Tetraphleps) and Oriini (Ori-
us + Bilia + Montandoniola) in the combined analyses with high
support values (Fig. 2and Table 3). In the single gene analyses,
the monophyly of Anthocorini and Oriini was also supported with
relatively strong PB nodal support values (18S rRNA: PB = 1, 1; 28S
rRNA: PB = 1, 1; 16S rRNA: PB = 0.98, 1, respectively; Supplemen-
tary Figs. 13). These results support Carayons (1972) morpholog-
ical classication of tribes. Anthocorini, containing 12 genera,
including the large genus Anthocoris with at least 50 described spe-
cies worldwide, is strongly represented in the Holarctic Region
(Schuh and Slater, 1995). Though most of the Palaearctic species
were sampled in this study, all species in the genus Anthocoris
and Tetraphleps aterrimus clustered together, thus the monophyly
of the tribe Anthocorini is conrmed (Fig. 2, node E). The tribe Ori-
ini contains 15 genera (Carayon, 1972). Carayon (1958) proposed
the new tribe Oriini based on reproductive morphology and other
structures (e.g., left paramere of male in the form of a spiral, endo-
soma of aedeagus short, copulatory tube single and short). Though
more sampling is required (e.g., Bilianella, Dokkiocoris, Kitocoris,
Wollastoniella) to conrm the classication proposed by Carayon
(1958), the clustering of Orius, Bilia, and Montandoniola in a clade
with high support values supports the monophyly of the tribe Ori-
ini (Fig. 2, node G).
However, our results do not support the monophyly of the Car-
diastethini in the single and combined analyses (Fig. 2, nodes F, I;
Table 3; Supplementary Figs. 13). The genus Amphiareus (four
species: A. rucollaris, A. obscuriceps, A. constrictus and A. morimo-
toi) and Dysepicritus rufescens clustered together with Anthocorini
outside of the main clade of Cardiastethini (with three genera: Car-
diastethus, Physopleurella, and Buchananiella). Additionally, the
genus Amphiareus (Fig 2, node F) clustered with Anthocorini sepa-
rate from Cardiastethini (Fig. 2, node I) with high support values;
this indicates that the current classication within the Anthocori-
dae is labile. As Schuh and Slater (1995) indicated, the tribe Dufou-
riellini (= Cardiastethini, restored by Carpintero and Dellap, 2009)
may not be monophyletic, and our molecular data indicate that
Amphiareus and Dysepicritus should not be treated as members of
the tribe Cardiastethini. Carayon (1972) provided some key charac-
ters for the tribe: a short rostrum, fossula spongiosa in males re-
duced or absent on foretibia, absent or vestigial on middle tibia,
and ovipositor greatly reduced. Although these morphological
characters are found in species in the genera Amphiareus and
Dysepicritus except for the short rostrum (Yamada and Hirowatari,
2003), the biological trait of living only in dead leaf clusters and
relatively longer rostrum among taxa in the tribe Cardiastethini
can be used as a reasonable character to exclude these genera from
the tribe Cardiastethini.
4.2. The systematic status of Lasiochilidae and related taxa
As indicated previously for the various classications of the
family Anthocoridae (table 1), the rank of Lasiochilidae (Lasiochili-
nae) and Lyctocoridae (Lyctocorinae) is controversial, and the
tribes Cardiastethini, Xylocorini, and Scolopini have been classied
as contained within Anthocoridae (sensu Schuh and tys, 1991) or
in Lyctocorinae (sensu Carayon, 1972) (Table 1). The family rank of
Lasiochilidae proposed by Schuh and tys (1991) is strongly sup-
ported by molecular data positioning most basally within Cimicoi-
dea (Fig. 2, node A). MP, ML, and BP (NBA, PBA) analyses support
the monophyly of Lasiochilidae separate from Anthocoridae within
Cimicoidea (Fig. 3, node M). Though Schuh and Slater (1995) indi-
cated that lasiochilids have the fewest novel features of all of the
cimicoid families, except for the presence of dorsal laterotergites
on abdominal segments 1 and 2 (Carayon, 1972), no traumatic
insemination is present in lasiochilids, in contrast to other cimic-
oids (Schuh and tys, 1991), which in light of our molecular data,
is an important biological characteristic that differentiates this
group from the cimicoids.
Three families, namely Nabidae, Cimicidae, and Plokiophilidae,
were used as outgroups in our study to determine the relationships
between Anthocoridae and higher related taxa (Fig. 1, Table 1). Like
the previous studies on family relationships of Cimicomorpha
(Schuh and tys, 1991; Tian et al., 2008; Schuh et al., 2009), Nabi-
dae was positioned more basally than Nabidae and Plokiophilidae
from the main clade of Anthocoridae (Fig. 3, node A) in this study.
Therefore, our results support the monophyly of the Cimicoidea
(Anthocoridae, Cimicidae, Lasiochilidae, Lyctocoridae, and Plokio-
philidae) which were also supported by morphological data (e.g.,
male genital structure) in the cimicomorphan relationships (Schuh
and tys, 1991). Additionally, by using several outgroup taxa in this
study, we found that the three tribes Xylocorini, Scolopini and
Blaptostethini were excluded from the main Anthocoridae clade
(Fig. 2, node D), and instead clustered with Cimicidae + Plokiophil-
idae and Lasiochilidae (Fig. 2, nodes C, Q, L). Either Cimicidae + Plo-
kiophilidae or Nabidae: Nabinae was the most basal node in this
study, while the three tribes, Xylocorini, Scolopini, and Blaptos-
tethini were always separated from the main Anthocoridae node
(Fig. 2D) and clustered with Cimicidae + Plokiophilidae or
Nabidae + Lasiochilidae + Lyctocoridae.
The rank of Lyctocoridae (Schuh and tys, 1991) and the phylo-
genetic relationships of the three tribes, Blaptostethini, Scoloponi,
and Xylocorini, could not be conrmed in this study due to rela-
tively low support values which must be caused from relatively
less number of sampling than the remaining higher taxa. However,
the three tribes were positioned between the groups Cimici-
dae + Plokiophilidae and Lasiochilidae with relatively high support
values (Fig. 2, nodes B, C, L, Q) on the PBA tree. Even though Lycto-
coridae proposed by Schuh and tys (1991) has unique morpholog-
ical features such as a modied left paramere (which does not
serve as a copulatory organ, but rather the vesica) and the posses-
sion of an apophysis on the female sternum VII, the molecular data
1180 S. Jung et al. / Molecular Phylogenetics and Evolution 57 (2010) 11731183
did not resolve the relationships between Lyctocoridae and these
tribes, and the position of Lyctocoridae itself was ambiguous due
to low nodal support (Fig. 2; Table 3, node J). Therefore, more sam-
pling of these and related taxa and/or additional molecular mark-
ers are required to infer their relationships.
4.3. The evolutionary implications of habitats and host plants
preferences within Anthocoridae
Based on our results, the family Anthocoridae is restricted to
node D (Fig. 2), which contains Oriini, Cardiastethini [node
H + node F (Amphiareus) + Dysepicritus refescens], and Anthocorini
(node E). Within the hypothesized Anthocoridae (Figs. 2 and 3,
node D), the relationships of tribes or genera and their habitat pat-
terns were mapped onto the simplied phylogram (Fig. 3); we
found that species in the hypothesized Anthocoridae evolved from
a broad-range of habitats to specic habitats (Fig. 3, large arrow).
We rst ruled out the exceptional habitat cases, such as Orius agilis,
which lives in the litter layer (Elov, 1976), and also excluded agro-
ecosystems, ornamental plants, articial habitats (e.g., green-
houses), and stored products facilities as habitat types, as
anthocorids have long been used as biological control agents. We
therefore only used habitat (host plants) information from natural
ecosystems.
Species within the tribe Cardiastethini used in this study, except
for the two genera Dysepicritus and Amphiareus, have been col-
lected from both living and dead plants (Lattin, 1999; Yamada
and Hirowatari, 2005b, 2007a, 2007b). Oriini species have been
collected from living plants (Yasunaga, 1997b), especially from for-
bs and owers of living plants including shrubs and trees (Pricart,
1972; Yasunaga, 1997b). Thus, habitat patterns suggest that spe-
cies in the tribe Cardiastethini (except for the two genera, Dysepicr-
itus and Amphiareus) have a relatively broader host range than
species in the tribe Oriini, as Oriini species inhabit only living
plants, especially various kinds of forbs (Fig. 3, nodes I, G). A few
taxa of Oriini have been reported to feed chiey on pollen (e.g.,
O. insidiosus, O. pallidicornis, O. tristicolor, and Paratriphleps laevius-
culus) (Barber, 1936; Carayon and Steffan, 1959; Bacheler and
Baranowski, 1975; Salas-Aguilar and Ehler, 1977), and in one
exceptional case, O. insidiosus has been reported to feed on blood
(Malloch, 1916). Furthermore, Elov (1976) reported O. agilis living
in the litter layer, an unusual habitat for species in this genus. Gen-
erally, species of Oriini feed on aphids, thrips, and mites or their
eggs, and usually inhabit living plants of forbs and owers of trees
(including agricultural crops) (Pricart, 1972; Yasunaga and
Miyamoto, 1993; Buchholz et al., 1994; Luettge and Sell, 1994;
Fiume, 1996; Frescata and Mexia, 1996). These species usually
aggregate with the host prey of specic plants, which may explain
why species of Oriini have been more effective as biological control
agents than species from other related anthocorid taxa.
Our results separate the two genera Dysepicritus and Amphiare-
us from the main clade of Cardiastethini. Species in these two gen-
era have been found in dead leaf clusters (Yamada and Hirowatari,
2002, 2003). We have yet to collect these species from living
plants, but have collected them from clusters of dead leaves from
various kinds of plants. Sometimes one or two individuals of this
group are collected from living plants, which may be accidental,
as they usually aggregate in the same habitat (Yamada and
Hirowatari, 2003). Pricart (1972) reported that Amphiareus inhab-
ited broadleafed trees, but it is unclear whether the trees they
were referring to were dead or alive. Thus, the habitats of species
in the genus are restricted to dead plants, which implies that
their prey would also be restricted to small arthropods living on
dead plants.
There are few biological studies of the Anthocorini. However,
their habitats are known to be more restricted to deciduous shrubs
and trees (Anderson, 1962a, 1962b; Hill, 1957, 1961; Pricart,
1972; Evans, 1976ad) than that of other anthocorids. Some
Anthocorini species have been reported to be specialist predators
against tree pests. For example, Anthocoris japonicus are usually
found on Zelkova serrata and Ulmus davidiana var. japonica (Ulma-
ceae) (even during winter hibernation) where Colopha moriokaensis
(Pemphigidae) occurs (eld observations by the rst author), and
A. whitei was reported as a specialist gall-forming aphid on Manza-
nita (Valenti et al., 1996). Anderson (1962a) reported A. bakeri in
aphid galls on Manzanita in southwestern Oregon, and Lattin
(1999) reported A. whitei on Cercocarpus ledifolius associated with
psyllids in Oregon. Most species in the tribe have been collected
from specic host plants at the genus or higher level, which may
indicate a preference for specic prey on the host plant (Sands,
1957; Anderson, 1962a; Madsen et al., 1963; Lussier, 1965; Cobben
and Arnoud, 1969; Pricart, 1972; Mendel et al., 1991; Lattin and
Stanton, 1992, 1993) (e.g., Elatophilus stigmatellus from Pinus syl-
vestris, Temnostethus ulmi from Ulmus pumila, and Anthocoris chibi
from Pinus densiora, etc.; see Kerzhner, 1988; Lattin, 1999). There-
fore, the strong preference of members of this tribe for specic host
plants implies that this group is a relatively more specialist group
than other related taxa within the hypothesized family Anthocori-
dae (Fig. 3).
The common Oriini ancestor appears to have evolved to live on
live plants (Fig. 3, node G, split 1; especially on forbs) from an
ancestral state of the mixed habitat (Fig. 3H, split 2). The common
ancestor of Cardiastethini and Amphiareus and Dysepicritus (Fig. 3,
nodes I, N + R, split 1) appears to have evolved to live on dead plant
(leaf clusters) from the mixed habitat (Fig. 3, node H, split 2). The
common ancestor for Anthocorini appears to also have evolved to
live on live plants (Fig. 3, node E, split 1) from an ancestral charac-
ter type of mixed habitat (Fig. 3, N + R, split 2). Those habitat tran-
sitions might be related with the emergence of deciduous or
herbaceous angiosperm plants providing various habitats which
could have allowed the common ancestors of anthocorids to
choose between two different habitat types (dead plants or living
plant; Fig. 3, nodes, H, N + R ?I, G, F, E). Putting these biological
data together with molecular phylogenetic analyses indicates that
the higher taxa within Anthocoridae evolved as a result of a nar-
rowing range of habitats that resulted in them selecting a different
type of habitat. The proliferation of angiosperms (Wikstrm et al.,
2001) is thought to have driven the diversication of major herbiv-
orous hemipteran groups (Grimaldi and Engel, 2005) such as an
radiation of aphids known as the prey of anthocorids (von Dohlen
and Moran, 2000), and it would appear that anthocorid diversica-
tion, too, closely tracks the rise of angiosperm-dominated forests
and their prey; anthocorids either chose new habitat, living plants
(Fig. 3, nodes D ?H ?G and D ?N + R ?E), or shared their
ancestors habitat of dead plant (Fig. 3, nodes D ?N + R ?F and
D ?H ?I). Morphologically, most extinct anthocorids are at-
bodied (Yao et al., 2006), which appears to be an adaptation to live
in dead plants whose habitat (under the bark and dead leaf
clusters; Fig. 2, node A) derived from the common ancestor. Addi-
tionally, extant anthocorids that inhabit living plants also have a
at-body as a symplesiomorphic character, and this differentiates
their feeding strategy from other predatory cimicomorphan spe-
cies (e.g., hunting behavior of Anthocoris spp., Anderson 1962a;
Valenti et al., 1996).
Acknowledgments
We would like to thank Dr. Tomohide Yasuanga, Dr. Ernst Heiss,
Dr. Tadashi Ishikawa, Mr. Sangwook Park, Mr. Gary Pixsely, Mr.
Jongok Lim, Ms. Ram Duwal, Dr. Eun Chan Yang, and the late Drs.
M. Josifov and I. M. Kerzhner for providing samples and their
encouragements. We are grateful to Dr. Randall T. Schuh, Dr.
S. Jung et al. / Molecular Phylogenetics and Evolution 57 (2010) 11731183 1181
Pavel tys, and Dr. Christiane Weirauch for the discussion about
this research during 4th IHS meeting. This research was supported
by a grant fromthe Agenda Project (Grant No. 102966290) and Bio-
green 21 Program (Grant No. 034032), by the Rural Development
Administration in the Republic of Korea, and the Eco-technopia
21 project Construction of a DNA barcode system for the conser-
vation and management of Korean insect fauna and partly by
Grant-in-Aid for Young Scientists (B) from the Japan Ministry of
Education, Culture, Sports, Science and Technology to the third
author (No. 20780043). This research was also supported by the
Research Institute for Agriculture and Life Sciences (RIALS).
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.ympev.2010.09.013.
References
Anderson, N.H., 1962a. Anthocoridae of the Pacic Northwest with notes on
distribution. Life histories and habitats (Heteroptera). Canadian Entomologist
94, 13251334.
Anderson, N.H., 1962b. Bionomics of six species of Anthocoris (Heteroptera:
Anthocoridae) in England. Transactions of the Royal Entomological Society of
London 114, 6795.
Anderson, N.H., 1962c. Growth and fecundity of Anthocoris sp. reared on various
prey (Hemiptera: Anthocoridae). Environmental Entomology 12, 990994.
Armer, C.A., Wiedenmann, R.N., Bush, D.R., 1998. Plant feeding site selection of
soybean by the facultatively phytophagous predator Orius insidious.
Entomological Experimental Application 86, 109118.
Bacheler, J.S., Baranowski, R.M., 1975. Paratriphleps laeviusculus, a phytophagous
anthocorid new to the United States (Hemiptera: Anthocoridae). Florida
Entomologist 58, 157163.
Barber, G.W., 1936. Orius insidiosus (Say): an important enemy of the corn earworm.
US Department of Agriculture Technical Bulletin 505, 24.
Bremer, K., 1988. The limits of amino acid sequences data in angiosperm
phylogenetic reconstruction. Evolution 42, 795803.
Bu, W., Zheng, L., 2001. Hemiptera Lasiochilidae, Lyctocoridae, Anthocoridae, Fauna
Sinica Insecta, vol. 24. Science Press, Beijing, China. p. 267.
Buchholz, U., Schmidt, S., Schruft, G., 1994. The use of immunological techniquein
order to evaluate the predationon Eupoecilis ambiguella (Hbn.) (Lepidoptera:
Cochylidae) in vineyards. Biochemical Systematics and Ecology 22, 671677.
Campbell, B.C., Ross, G.F., Woodward, T.E., 1995. Paraphyly of Homoptera and
Auchenorrhyncha inferred from 18S rDNA nucleotide sequence. Systematic
Entomology 20, 175194.
Carayon, J., Steffan, J.R., 1959. Observations sur le rgime alimentaire des Orius et
particulirement, dOrius pallidicornis (Reuter) (Heteroptera: Anthocoridae)
Cah.. National Bulletin Natural Parisiens 15, 5363.
Carayon, J., 1958. Etudes sur les Hmiptres Cimicoidea. 1. Position des genres Bilia,
Bilioa, Bilianella et Wollastoniella dans une tribe nouvelle (Oriini) des
Anthocoridae; differences entre derniers et les Miridae Isometopinae
(Heteoptera). Memories du Museum dHistoire Naturelle. Serial A, Zoology 16,
141172.
Carayon, J., 1972. Caractres systematiques et classication des Anthocoridae
(Hemipt.). Annals of the Society of Entomological France (NS) 7, 737770.
Carayon, J., Steffan, J.R., 1959. Observationssur le regime alimentaire des Oriuset
particuli