Systematic Entomology (2004) 29, 238–259
Phylogenetic analysis of mealybugs (Hemiptera:
Coccoidea: Pseudococcidae) based on DNA sequences
from three nuclear genes, and a review of the higher
classification
D . A . D O W N I E and P . J . G U L L A N
Department of Entomology, University of California, 1 Shields Avenue, Davis, CA 95616, U.S.A.
Abstract. Mealybugs (Hemiptera: Pseudococcidae) are small, plant-sucking
insects which comprise the second largest family of scale insects (Coccoidea).
Relationships among many pseudococcid genera are poorly known and there is
no stable higher level classification. Here we review previous hypotheses on
relationships and classification and present the first comprehensive phylogenetic
study of the Pseudococcidae based on analysis of nucleotide sequence data. We
used three nuclear genes, comprising two noncontiguous fragments of elongation
factor 1a (EF-1a 50 and EF-1a 30 ), fragments of the D2 and D10 expansion regions
of the large subunit ribosomal DNA gene (28S), and a region of the small subunit
ribosomal DNA gene (18S). We sampled sixty-four species of mealybug belonging
to thirty-five genera and representing each of the five subfamilies which had been
recognized previously, and included four species of Puto (Putoidae) and one
species each of Aclerda (Aclerdidae) and Icerya (Margarodidae), using Icerya as
the most distant outgroup. A combined analysis of all data found three major
clades of mealybugs which we equate to the subfamilies Pseudococcinae, Phenacoccinae and Rhizoecinae. Within Pseudococcinae, we recognize the tribes Pseudococcini (for Pseudococcus, Dysmicoccus, Trionymus and a few smaller genera),
Planococcini (for Planococcus and possibly Planococcoides) and Trabutinini
(represented by a diverse range of genera, including Amonostherium, Antonina,
Balanococcus, Nipaecoccus and non-African Paracoccus), as well as the Ferrisia
group (for Ferrisia and Anisococcus), some ungrouped African taxa (Grewiacoccus, Paracoccus, Paraputo and Vryburgia), Chaetococcus bambusae and Maconellicoccus. The ‘legless’ mealybugs Antonina and Chaetococcus were not closely
related and thus we confirmed that the Sphaerococcinae as presently constituted
is polyphyletic. In our analyses, the subfamily Phenacoccinae was represented by
just Phenacoccus and Heliococcus. The hypogeic mealybugs of the Rhizoecinae
usually formed a monophyletic group sister to all other taxa. Our molecular data
also suggest that the genera Pseudococcus, Dysmicoccus, Nipaecoccus and Paracoccus are not monophyletic (probably polyphyletic) and that Phenacoccus may be
paraphyletic, but further sampling of species and genes is required. We compare
our phylogenetic results with published information on the intracellular endosymbionts of mealybugs and hypothesize that the subfamily Pseudococcinae may be
characterized by the possession of b-Proteobacteria (primary endosymbionts)
capable of intracellular symbiosis with g-Proteobacteria (secondary endosymbionts).
Correspondence: D. A. Downie, Department of Zoology and
Entomology, Rhodes University, Grahamstown 6140, South
Africa. E-mail: [email protected]
238
#
2004 The Royal Entomological Society
Mealybug molecular systematics
239
Furthermore, our data suggest that the identities of the secondary endosymbionts may be useful in inferring mealybug relationships. Finally, cloning polymerase chain reaction products showed that paralogous copies of EF-1a were
present in at least three taxa. Unlike the situation in Apis and Drosophila, the
paralogues could not be distinguished by either the presence/absence or position of
an intron.
Introduction
Mealybugs are small, soft-bodied insects which constitute
the scale insect family Pseudococcidae (Hemiptera: Sternorrhyncha: Coccoidea). Their common name derives from the
white, powdery or mealy wax secretion which covers the
body of the nymphs and adult females of most species. Only
species which are adapted to living in concealed locations,
such as galls or grass sheaths, lack or have reduced amounts
of the mealy secretion (McKenzie, 1967; Miller, 1991). In
addition, the margin of the body frequently has a series of
white, lateral wax protrusions, arranged segmentally, and
often most prominent posteriorly. Adult females are wingless, resemble immature stages, and range from about 0.4 to
0.8 mm in body length, depending on species (McKenzie,
1967; Kosztarab & Kozár, 1988). Females either lay eggs
into a white, filamentous ovisac which they secrete from
glands in their cuticle, or are ovoviviparous and generally
lack the filamentous secretion (McKenzie, 1967; Cox &
Pearce, 1983; Williams, 1985). Reproduction is typically
sexual, although a few species are parthenogenetic (McKenzie,
1967; Nur, 1977; Williams, 1985; Kosztarab & Kozár,
1988). There are three immature instars in the female and
four in the male (McKenzie, 1967; Miller, 1991). Adult
males, if present, are short-lived, nonfeeding and rarely
collected. Nymphs and adult females feed by sucking plant
sap and some species cause considerable economic damage
to agricultural plants (McKenzie, 1967; Miller & Kosztarab,
1979; Williams, 1985; Miller et al., 2002). Damage results
from sap removal, the injection of toxins, honeydew contamination and associated sooty mould growth, and occasionally
from the effects of transmitted plant viruses (McKenzie,
1967; Williams, 1985; Rohrbach et al., 1988). Mealybugs
occur world-wide, but are most abundant in the tropics
and subtropics (Ben-Dov, 1994). Their favoured hosts are
herbaceous plants, especially grasses (Poaceae) and composites (Asteraceae) (Ben-Dov, 1994; Ben-Dov & German,
2003), although more than 20% of pest species may be
polyphagous (Miller et al., 2002).
The Pseudococcidae is the second largest family of scale
insects, with approximately 2000 described species in more
than 270 genera (Ben-Dov et al., 2003). It is a member of the
‘advanced scale insects’, an informal group usually called
the neococcoids or neococcids (Koteja, 1974b; Miller &
Kosztarab, 1979; Gullan & Kosztarab, 1997), and may
form the sister group of all other neococcoids (Koteja,
1974b; Danzig, 1980; Cook et al., 2002; Hodgson, 2002).
The current taxonomy and classification of mealybugs are
#
based on the morphology of adult females, although the
earliest study of relationships using numerical methods was
based on adult male morphology (Afifi, 1968) and a recent
cladistic morphological analysis of adult male scale insects
included representatives of ten mealybug species (Hodgson,
2002). There is no satisfactory or generally accepted suprageneric classification for mealybugs (reviewed by Ben-Dov,
1994). Although tribal and subfamilial classifications are
available for the other two large families of Coccoidea,
namely the Coccidae (Hodgson, 1994) and Diaspididae
(Danzig, 1980; Ben-Dov, 1990), for the Diaspididae at
least there is also disagreement on the content of the higher
groups. Furthermore, there has been controversy concerning the suprageneric classifications of the other sternorrhynchan superfamilies, although such classifications are
applied widely in all three groups, the Psylloidea (classification based on White & Hodkinson, 1985), Aphidoidea (e.g.
Remaudière & Remaudière, 1997; von Dohlen & Moran,
2000) and Aleyrodoidea (classification discussed by BinkMoenen & Mound, 1990).
Until recently the Holarctic and Neotropical genus Puto
(including Ceroputo and Macrocerococcus; see Williams &
Granara de Willink, 1992) was usually considered part of
the Pseudococcidae (e.g. Danzig, 1980; Miller & Miller,
1993) based on morphological similarities of the adult
females. However, cladistic analyses of molecular data
(Cook et al., 2002) and male morphology (Hodgson, 2002)
support placement of Puto outside both the pseudococcids
and the neococcoids, although the type species of Ceroputo
appears to be a mealybug (Williams & Granara de Willink,
1992; Hodgson, 2002). Puto is also plesiomorphic in its
cytology (Brown & Cleveland, 1968) and its endosymbionts
are very different from those of mealybugs (Buchner, 1965;
Tremblay, 1989). It appears that the similarities of adult
females of Puto to certain mealybugs are due to symplesiomorphy and we follow Beardsley (1969) in recognizing the
family Putoidae.
Between three and five subfamilies of mealybugs have
been recognized at various times by different workers (e.g.
Koteja, 1974a,b, 1988; Danzig, 1980; Williams, 1985;
Tang, 1992), but no subfamily classification has found
practical use, largely because only one subfamily, the
Rhizoecine (see Williams, 1998), is reasonably well defined
with stable generic content. Previously proposed subfamily
groups for mealybugs are Pseudococcinae, Phenacoccinae,
Trabutininae, Rhizoecinae and Sphaerococcinae, but it
must be stressed that these names are not in common use,
there is no agreed list of genera for the Pseudococcinae,
2004 The Royal Entomological Society, Systematic Entomology, 29, 238–259
240 D. A. Downie and P. J. Gullan
Phenacoccinae and Trabutininae, and it is not clear whether
any of the subfamilies are monophyletic or even whether the
Phenacoccinae and Trabutininae are mutually exclusive.
Occasionally authors have used informal groupings, such
as the eleven groups recognized and briefly defined by
McKenzie (1967) for North American mealybug genera
based on adult female morphology, and the six groups
recognized by Afifi (1968) based on male morphology. Tribal names have been used by some authors but, with a few
exceptions, these groups are not widely used, and often they
are equivalent to the subfamily groups of other authors (e.g.
Danzig, 1980 used tribal rank for taxa which other coccidologists called subfamilies). Ferris (1950, 1953) refrained
from recognizing higher taxa and emphasized that study of
the world mealybug fauna was prerequisite to establishing
an enduring classification.
teristic dark body pigments (Banks et al., 1976), often have
enlarged dorsal setae and comprise at least Trabutina,
Melanococcus, Nipaecoccus and Amonostherium (Ferris,
1950; Williams, 1985; Danzig & Miller, 1996), and the
Phenacoccincae would contain the large genus Phenacoccus
and an indeterminate number of other genera such as those
listed by Tang (1992) or some of the genera listed as
Phenacoccini by Danzig (1980). Koteja (1988) recognized
only Phenacoccinae and restricted this subfamily to a group
of genera sharing at least quinquelocular pores and a claw
denticle in the adult female. The genus Rastrococcus, which
is tropical and mostly Asian, is distinctive morphologically
(Williams, 1989) and especially in terms of its endosymbiosis
(Buchner, 1965; Tremblay, 1989) but has been put near
Phenacoccus (Williams, 1985, 1989) or in a more broadly
defined Phenacoccinae (Tang, 1992).
Pseudococcinae
Sphaerococcinae
Most mealybug species fall into the Pseudococcinae
(Williams, 1969, 1985; Koteja, 1974a,b; Tang, 1992),
which is equivalent to the Pseudococcini of Danzig (1980).
Three of the largest genera of mealybugs – Dysmicoccus,
Pseudococcus and Trionymus – belong here. Koteja (1988;
pers. comm. 2003) considered the Pseudococcinae to be a
relatively homogeneous group lacking lateral lines of simple
eyes in the adult male, and in the adult female lacking
quinquelocular pores, supernumerary cerarii and claw
denticles but with the dorsal setae hairlike, the labium
with fourteen pairs of setae, and the body colour uniformly
dark (brown to violet black). However, not all features in
this diagnosis apply to every taxon considered to belong in
the Pseudococcinae. There is no agreed tribal classification
for the subfamily Pseudococcinae, although some tribal
names such as Planococcini (Ezzat & McConnell, 1956)
have been used fairly consistently. Afifi’s (1968) study was
based mostly on mealybug species belonging to the Pseudococcinae. Similarly, most or perhaps all species in a recent
study of the relationships of the primary endosymbionts
(P-endosymbionts) of the Pseudococcidae (Thao et al.,
2002) would be placed in this subfamily. Although the
bacterial data were suggestive of relationships among the
pseudococcine hosts and did not contradict relationships
inferred by Afifi from adult male data, the taxa sampled
differed with only six genera and four species in common to
the two studies.
Sphaerococcinae contains the so-called ‘legless’ mealybugs, which are characterized by a reduction or loss of
legs, eyes and antennae and usually by sclerotization of at
least the posterior abdominal segments. Their specialized
morphology reflects adaptation to their feeding sites, which
are mostly under leaf sheaths or at the base of culms of
Poaceae (Hendricks & Kosztarab, 1999). However, the host
plants and habits of the two species of the type genus
Sphaerococcus are very different. The type species S. casuarinae
from Australia and S. durus from South Africa (Miller
et al., 1998; Hendricks & Kosztarab, 1999) feed, respectively, in association with galls or under the bark of host
trees. The only recent taxonomic treatment of Sphaerococcinae (Hendricks & Kosztarab, 1999) revised all species of
the redefined tribe Serrolecaniini and reviewed all other
genera, based on adult female morphology. In total, sixteen
genera (seven monotypic) have been assigned to the
Sphaerococcinae, its equivalent, or a subordinate group by
various authors. In addition to Sphaerococcus, these are
Chaetococcus, Idiococcus, Kermicus, Porisaccus, Serrolecanium
and Tangicoccus (these six placed in Serrolecaniini by
Hendricks & Kosztarab, 1999), Acinicoccus, Antonina,
Antoninoides, Cypericoccus, Nesticoccus, Paludicoccus,
Parapaludicoccus, Peridiococcus and Pseudantonina. Although
the six genera of Serrolecaniini were considered to form a
natural group, Hendricks & Kosztarab (1999) suggested that
all other genera were not closely related to each other or to
the Serrolecaniini. Thus, the Sphaerococcinae may be polyphyletic with true relationships being confounded by convergent simplification of structure.
Trabutininae and/or Phenacoccinae
One or the other (e.g. Koteja, 1974a,b; Danzig, 1980;
Williams, 1985) or both (e.g. Tang, 1992) of these groups
have been recognized at different times by different authors.
If only one subfamily is recognized, then Trabutininae has
priority over Phenacoccinae (Williams, 1969). If both are
recognized, then Trabutininae could be restricted to the
‘blue-black’ or ‘blue-green’ mealybugs which have charac-
#
Rhizoecinae
Mealybugs in this subfamily are small to minute and all live
underground (hypogeic) in soil, leaf litter or rotting logs and
are usually associated with plant roots, although the habits of
many species are unknown because they have only been
2004 The Royal Entomological Society, Systematic Entomology, 29, 238–259
Mealybug molecular systematics
collected from Berlese funnel samples (Williams, 1998). The
most recent review of this group (Williams, 1998) recognized
ten genera. With the exception of Rhizoecus, which has more
than 120 described species (Ben-Dov & German, 2003), the
genera are small with five of them monotypic (Williams,
1998; Ben-Dov & German, 2003). Several of the genera are
myrmecophilous and associated with ants of the genus Acropyga (Williams, 1998; Johnson et al., 2001). Hambleton
(1946) considered Rhizoecus and the other hypogeic genera
to be elements of a natural group, which was established as
the tribe Rhizoecini by Williams (1969) and later treated as a
subfamily (Koteja, 1974a,b). Adult females are characterized by their generally small size, their usually strongly geniculate antennae with six or fewer segments, the cerarii being
either absent or confined to the anal lobes, which are often
poorly or not developed and, if one or more circuli are present, they are often conical or raised (Hambleton, 1946, 1976;
Williams, 1969, 1998; Williams, 2001). In addition, most of
the members of this subfamily have a very narrow and comparatively small labium (Koteja, 1974a,b) with a shape
considered to be plesiomorphic (Koteja, 1974a). Most species
of Rhizoecus possess bitubular or tritubular cerores (peculiar
pores), which are unique to this group (Hambleton, 1946,
1976; McKenzie, 1967; Williams, 1998). The adult male of the
type species of Rhizoecus, R. falcifer, was studied by Beardsley
(1962), who considered it to be the most primitive of the
male mealybugs which he had examined. Koteja (1985) considered the hypogeic habit of Rhizoecus to be ancestral
and not secondarily acquired from an aerial habit on plant
foliage.
Efforts to generate phylogenetic hypotheses for the
family are confounded by a paucity of useful morphological
characters in adult females as a result of reduction (Ferris,
1950; Cox, 1987), substantial intraspecific polymorphism in
some species (Danzig, 1980; Cox, 1987), and environmental
influences on morphology affecting character states of the
wax ducts and pores (Cox, 1983; Charles et al., 2000), as
well as other features deemed useful in identification. Also,
significant electrophoretic differences have been detected
among collections of mealybugs which were considered to
be one species based on morphology (Nur, 1977). Thus, it is
likely that some species names are synonyms and that the
apparent close relationship of some others may be due to
convergences. Morphological data from adult males are
potentially informative phylogenetically (Afifi, 1968;
Hodgson, 2002), but some species lack males and, even if
present, adult males are short-lived and difficult to collect
for most species. First-instar nymphs, called crawlers, are
considered a good source of phylogenetic data in other
coccoid families (Howell & Tippins, 1990; Williams &
Kondo, 2002), but have not been used for phylogenetic
inference in Pseudococcidae. Molecular data, which have
been used in recent phylogenetic studies of coccoids and
related insects (e.g. Moran et al., 1999; Normark, 2000; von
Dohlen & Moran, 2000; Cook et al., 2002; von Dohlen et al.,
2002; Gullan et al., 2003), may have the greatest utility in
elucidating relationships at the species, genus and subfamily
levels in Pseudococcidae.
#
241
Here, we estimate the phylogeny of Pseudococcidae using
DNA sequences from three nuclear genes sampled from
mealybug taxa collected world-wide. This study is the first
such estimate for the Pseudococcidae, for any data type,
except for the very limited study by Afifi (1968). Specifically, the goals of the study were to: (1) assess the validity of
subfamilial concepts, and (2) assess the phylogenetic
relationships and monophyly of a sample of pseudococcid
genera.
Materials and methods
Material
Freshly collected specimens of sixty-four species in thirtyfive genera of Pseudococcidae (Appendix) were obtained
preserved in 95–100% ethanol. The sample included multiple
representatives from all putative subfamilies within Pseudococcidae. Fifteen genera were represented by two or more
species. As outgroups, four species of Putoidae and one
species of Margarodidae were collected, and one species of
Aclerdidae was included to represent a more derived neococcoid family. A recent molecular phylogeny of the Coccoidea
based on 18S nucleotide sequences did not resolve which
archaeococcoid taxon was the sister group of the Pseudococcidae plus other neococcoids (Cook et al., 2002).
The specimens were slide-mounted in Canada balsam
using the method described in Williams & Granara de
Willink (1992) except that xylene was used instead of clove
oil. Identifications were made mostly by reference to keys in
McKenzie (1967), Williams (1985), Cox (1987), Williams &
Granara de Willink (1992) and an unpublished manuscript
on the mealybugs of California by R. J. Gill. In some cases
identifications were confirmed by comparison with type
specimens deposited in the Coccoidea Collection of the
Bohart Museum of Entomology, University of California,
Davis. Slide-mounted voucher specimens were prepared for
all mealybug species sequenced and these vouchers have
been deposited in the Bohart Museum of Entomology.
DNA extraction, polymerase chain reaction (PCR) and
sequencing
Prior to DNA extraction, all specimens were examined
under the microscope for the presence of parasitoids. DNA
was extracted from single parasitoid-free adult females with
the DNeasy tissue kit (Qiagen, Inc., Valencia, California,
U.S.A.) with a final wash performed with sterile water
rather than the supplied buffer and at half volume.
PCR products were generated from three nuclear genes:
two noncontiguous fragments of a protein coding gene,
elongation factor 1a (EF-1a; hereafter termed EF-1a 50
and EF-1a 30 ), fragments of the D2 and D10 expansion
regions of the large subunit ribosomal DNA gene
(28S)(Michot et al., 1984), and 600 bp of the small subunit
ribosomal DNA gene (18S). These genes range from very
2004 The Royal Entomological Society, Systematic Entomology, 29, 238–259
242 D. A. Downie and P. J. Gullan
Table 1. Details of sequence size and location, amplification and sequencing primers for elongation factor 1a (EF-1a)), 28S and 18S genes.
All sequences read 50 to 30 .
Gene
Region
Amplicon
sizea
18S
24–585c
589–671 bp
575
28S
3686–4047c
D2 expansion region
D10 expansion region
310–356 bp
255
738–767 bp
721
2103–2342c
306–384 bp
238
2832–3149c
372–506 bp
316
EF-1a
Total
Characters
in matrixb
Primer sequence
CTGGTTGATCCTGCCAGTAG
CCGCGGCTGCTGGCACCAGA
AGAGAGAGTTCAAGAGTACGTG
TTGGTCCGTGTTTCAAGACGGG
GTAGCCAAATGCCTCGTCA
CACAATGATAGGAAGAGCC
CACATYAACATTGTCGTSATYGG
CTTGATGAAATCYCTGTGTCC
CARGACGTATACAAAATCGG
GCAATGTGRGCIGTGTGGCA
Primer
name
Primer
source
18S-2880
18S-B
None
von Dohlen
& Moran (1995)
Belshaw
& Quicke (1997)
Dietrich et al. (2001)
None
M3
rcM44.9
M51.9
rcM53-2
Cho et al. (1995)
Cho et al. (1995)
2105
a
Fragment length varied due to indels in 18S and 28S and a variable length intron in each fragment of EF-1a.
Size of data matrix after regions of ambiguous alignment in 18S and 28S and introns in EF-1a were deleted.
c
Nucleotide positions relative to the Drosophila melanogaster sequence (18S: Tautz et al., 1988; 28SD2: Hancock et al., 1988; EF-1a: Hovemann et al., 1988).
b
slowly evolving (18S) to rapidly evolving at some sites
(EF-1a) and were chosen to provide information at different
levels of the tree. Details on the sequences and primers
for both amplification and sequencing are given in Table 1.
The PCR reaction components and final concentrations
were 1.5–2.5 mM MgCl2, 0.2 mM dNTPs and 1 unit Taq
polymerase in a proprietary buffer (PCR Master Mix,
Promega, Madison, Wisconsin, U.S.A.), 0.2 mM each primer
and 4–5 ml of DNA template in a final volume of 20–25 ml.
The PCR cycling protocols were 94 C for 4 min followed
by thirty to fifty cycles of 94 C for 1 min, 48–56 C for 1 min
and 72 C for 1 min 30 s with a final extension at 72 C
for 4 min.
The PCR products were purified by exonuclease I and
shrimp acid phosphatase digestion of single-stranded
DNA (primers) and dNTPs (ExoSAP-IT, USB Corp.,
Cleveland, Ohio, U.S.A.). In Drosophila and Apis,
paralogues of EF-1a which differ in length due to the
presence or absence of introns have been reported
(Hovemann et al., 1988; Danforth & Ji, 1998). Paralogues may also differ in intron position (Danforth &
Ji, 1998). Co-amplification of paralogues which differ in
length should appear as two or more bands on agarose
gels. Generally this was not observed, but occasionally
two closely co-migrating bands appeared and some taxa
showing a single band produced a mixed sequence from
the EF-1a product when sequenced directly. To test for
the co-amplification of similar-sized fragments of
paralogous copies of the gene, the PCR products from
both EF-1a primer pairs from a small subsample of taxa
(n ¼ 7) were cloned using the pGEM1-T Easy Vector
System (Promega). Positive colonies (n ¼ 3–5 per PCR
product) were isolated from each library and plasmid
vectors purified using Wizard Plus SV Minipreps (Promega). Inserts were sequenced from each using the T7
promoter universal primer.
#
All PCR products were sequenced on both strands and
clones on a single strand, using the ABI Big Dye V3
terminator sequencing reaction kit (Perkin-Elmer/ABI,
Weiterstadt, Germany) on an ABI Prism 3100 automatic
sequencer (Perkin-Elmer) on 5% acryl/bisacryl Long
Ranger gels at the Division of Biological Sciences Automated Sequencing Facility at the University of California,
Davis.
Alignment
Sequences from both strands were assembled and edited
if necessary using SEQUENCHER ver. 4.0.5 (Gene Codes
Corp., Ann Arbor, Michigan, U.S.A.). Multiple alignments
were carried out on ingroup and outgroup sequences separately, which were then aligned together under the profile
alignment option with CLUSTAL X ver. 1.81 using the default
parameters (Thompson et al., 1997). Alignments were
edited in MACCLADE ver. 4.0 (Maddison & Maddison,
2000). Introns in the EF-1a sequences (delimited by
GT–AG splicing sites) and regions of ambiguous alignment
(due to indels) in the 28S and 18S sequences were deleted
prior to analyses.
Data analysis
Maximum parsimony as well as neighbour joining were
used to estimate trees with PAUP* vers. 4.0b10 (Swofford,
2003). For parsimony, heuristic searches with tree bisection–
reconnection branch swapping were run with 100 replications of random sequence addition. Search options for
the ribosomal genes included restricting the maximum
number of trees (MAXTREES) to between 50 000 and
100 000 with the maximum number of trees per replicate
2004 The Royal Entomological Society, Systematic Entomology, 29, 238–259
Mealybug molecular systematics
(NCHUCK) set from 500 to 1000 with constant characters
excluded. In addition to searches with all sites weighted
equally, different gene regions in combined analyses were
weighted by the proportion of potentially informative sites
each contributed to the combined dataset; and for both
EF-1a regions, alone and in combination, first and second
codon positions were weighted in inverse proportion to
their contribution of parsimony informative characters
(nine times third positions for both).
Sequences for each gene region were not obtained for
all taxa (Appendix). Therefore, in the combined analyses
two datasets were constructed: excluding all taxa with
missing data, or including all taxa. This procedure was not
followed for the two EF-1a fragments as inclusion of
all taxa sequenced for one or the other would result in a
substantial number of taxa missing 40–60% of the data.
Including taxa with missing data dramatically increased
the number of most parsimonious trees (MPTs) and
decreased the resolution of strict consensus trees for the
ribosomal fragments, so datasets excluding taxa with missing data are discussed in more detail. This effect tended to
be less pronounced as total sequence length increased (as
the proportion of missing data decreased; see Wiens, 1998),
so both types of dataset are discussed for datasets combining EF-1a with the rDNA fragments. In all analyses,
support for clades found by maximum parsimony and
neighbour joining was estimated by bootstrapping (200
and 500 replications for maximum parsimony and neighbour joining, respectively) using the same heuristic search
options as above, except with ten random sequence addition
replicates.
For distance-based analyses, the model of sequence
evolution which best fits the data was found by testing
fifty-six models using the iterative likelihood procedure
implemented in MODELTEST ver. 3.06 (Posada & Crandall,
1998). Maximum likelihood is a powerful means to estimate phylogenies and allows a statistical interpretation of
the strength of competing hypotheses. It is, however, time
prohibitive. An alternative is Bayesian inference, which
shares some of maximum likelihood’s satisfying statistical
properties but can be executed in a reasonable computing
time. Monte Carlo Markov chain methods allow the posterior probabilities (PP) of tree topologies and branch
lengths to be explored and estimated under different
models. The PP for nodes on the tree are calculated directly
from the frequencies of nodes found among trees after the
Markov chain has reached stationarity. Here, Bayesian
inference was implemented using MRBAYES ver. 3.0b3
(Huelsenbeck & Ronquist, 2001). Runs varied from
300 000 to 1 000 000 generations, sampling every 100
generations with base frequencies fixed at their empirical
values. Other parameters were set consistent with the
models supported by the hierarchical likelihood ratio
tests, namely, number of substitution types ¼ 6, rates ¼
invgamma (I þ G). The shape parameter (a) and the proportion of invariant (I) sites were estimated from the data.
The defaults were used for all other parameters. For the
fragments of EF-1a, runs were executed under the 4 4
#
243
model, and the codon model, which treats codons rather
than nucleotides as characters.
Heterogeneity among gene regions is a concern when
interpreting results and considering whether to combine
datasets. The incongruence length difference test (Farris
et al., 1994) has commonly been used to assess congruence globally (Mason-Gamer & Kellogg, 1996), but the
test is conservative (Cunningham, 1997), may be sensitive
to noise in the datasets (Dolphin et al., 2000), may fail to
assess the phylogenetic accuracy of a combined analysis
correctly (Yoder et al., 2001; Dowton & Austin, 2002),
or may lack power for some datasets (Darlu &
Lecointre, 2002). We use the incongruence length difference test here to assess heterogeneity among dataset
partitions, but not as a criterion for combining data.
Instead, the results are presented, or discussed, of
analyses on each gene fragment separately, and of all
possible combinations of data partitions.
The complete sequences for each fragment have been
submitted to GenBank (accession numbers in Appendix).
Results
Sequence details
Fifty-one sequences were obtained for EF-1a 50 , sixty for
EF-1a 30 , sixty-three for 18S and sixty-nine and sixty-seven
for the D2 and D10 regions of 28S, respectively. The length
of amplified fragments for each gene region is given in
Table 1. The 28S and 18S fragments had a large proportion
of regions with indels, consistent with inferred structure
(Michot et al., 1984; Rousset et al., 1991), although alignment of the D10 fragment was much more straightforward.
Weighting schemes using information from secondary
structure, and use of indels as characters, have been used
for these genes, but the lack of data on structure from a
close relative to the Pseudococcidae, and apparent random
distribution of indels, led us to take the conservative
approach of deleting regions of ambiguous alignment.
These regions were between bases 207–258 and 300–367 in
the initial 18S alignment, between bases 72–99, 123–152,
216–300 and 351–365 in the initial 28S D2 alignment and
between bases 315–547, 607–629 and 644–761 in the initial
28S D10 alignment.
Base frequencies were not heterogeneous among nucleotides for any gene and transitions were more frequent than
transversions for all datasets (Table 2). Plots of transitions
and transversions on pairwise divergence (not shown) suggest saturation at only the deepest divergences (20%).
All EF-1a sequences had an intron at the same location.
The intron in the EF-1a 50 fragment was found at base
103/104 in all taxa and varied in length from 64 to 145 bp,
and the EF-1a 30 intron was found at base 259/260 in all
taxa and varied in length from 63 to 191 bp, although the
Aclerda sp. and Icerya purchasi had introns of 276 and
295 bp, respectively (see Table 1 for reference positions in
Drosophila melanogaster). With few exceptions, uncorrected
2004 The Royal Entomological Society, Systematic Entomology, 29, 238–259
244 D. A. Downie and P. J. Gullan
Phylogenetic analyses
Table 2. Sequence details.
Nucleotide frequencies
2
Gene region
A
C
G
T
EF-1a 50
EF-1a 30
28S D2
28S D10
18S
0.335
0.274
0.179
0.220
0.270
0.199
0.229
0.270
0.260
0.239
0.219
0.230
0.346
0.305
0.265
0.247
0.267
0.205
0.215
0.225
39.34
47.67
112.37
63.27
11.56
P
ti:tv
1.00
1.00
0.99
1.00
1.00
5.22
2.03
2.96
1.46
2.31
ti:tv, transition/transversion ratio.
pairwise sequence divergences were less than 18%, and no
stop codons were found in any coding sequence.
The smaller sample of taxa in the EF-1a datasets arose
from amplification failures, or poor sequence quality in
some taxa. The presence of paralogous copies among the
PCR products is a potential cause of these difficulties. The
sequence details given above suggested that paralogy was
not a problem. However, three of seven taxa for which PCR
products were cloned revealed the presence of divergent
sequences within individuals. All three of these taxa had
previously given poor-quality sequences when sequenced
directly from PCR products. Unlike the situation in Apis
and Drosophila, these sequences did not differ in the presence/absence or position of introns. For EF-1a 30 , one of
five clones sequenced from Rhizoecus hibisci was nearly
25% divergent to the other sequences within this individual,
but was only 0.6% divergent from the sequence derived
from Phenacoccus colemani! Evidence that this sequence
was not a contaminant comes from the results from the
other genes, from its close similarity and identical intron
structure to that found in other pseudococcids, and from
the fact that PCR and cloning was not performed simultaneously for this taxon and any Phenacoccus group species.
Examination of the amino acid sequence revealed a serine
which was unique to this sequence, all others in the matrix
having a phenylalanine at this position. It was rejected as
the orthologous copy. Only a single clone was successfully
sequenced in Puto yuccae, which was identical to the
sequence from direct sequencing of PCR products from
two independent runs. This sequence was nearly identical
to that found in Ferrisia gilli, suggesting it to be a paralog,
or a contamination. For EF-1a 50 three distinct sequences
were found within an individual of Nipaecoccus nipae.
These sequences were not as divergent as those found in
EF-1a 30 for Rhizoecus hibisci. The largest difference was
10.9% with intron included, and 5.9% with intron excluded.
All three clones were placed as sister to the congeneric
Nipaecoccus viridis on a neighbour joining tree. The consensus of the two most closely related sequences was
included in the data matrix.
None of the cloned PCR products from taxa which produced quality sequences directly from PCR products
showed evidence of divergent sequences (ignoring small
differences suggestive of heterozygosity or PCR/sequencing
artefacts).
#
The total number of characters in each data matrix is
shown in Table 1. The number of parsimony informative
characters, tree statistics, and various measures of fit for
parsimony analyses of each gene fragment and different
combined datasets are shown in Table 3. For all data partitions the model of sequence evolution with the highest likelihood was that of Tamura & Nei (1993) with gamma
distributed rates. With the exception of the D2 region of
28S, all models specified a proportion of invariant sites (I),
generally greater than 0.5 (Table 4). In some cases, analyses
using these estimates led to a number of distances in the
matrices being undefined and the resulting trees were distorted due to the arbitrary assignment by PAUP of a distance
two times the largest in the matrix to these comparisons (but
in practice, usually much larger than this). This is apparently
a ‘bug’ in PAUP vers. 4.10b (Jim Wilgenbush, pers. comm.)
and the proportion of invariant sites in the model was therefore reduced until all distances were defined. As can be seen
from Table 4, Bayesian estimates of both alpha (a) and I were
lower than the likelihood estimates for all fragments.
18S. The strict consensus of 57 000 MPTs (see supplemental material; Fig. S1) is largely unresolved, although all
methods showed a monophyletic Puto and found support
for a clade consisting of a heterogeneous group of species,
including representatives from all subfamilies, as interpreted by various authors, except Rhizoecinae, and so
does not conform to any of the current views of mealybug
relationships. Seven base changes occur along the branch
leading to this clade, four of them synapomorphic. Within
this clade there was weak evidence from neighbour joining,
but a high PP (0.92), for monophyly of a group of species
from New Zealand (Paracoccus nothofagicola, Sarococcus
comis and Cyphonococcus alpinus). Other key features of the
tree are the paraphyletic relationship of the two Rhizoecus
species and moderate support for a group consisting
of Heliococcus and Neochavesia, and the species pair of
Planococcus citri and Pl. ficus. Bayesian analyses found
support for a clade within which the two Rhizoecus
species þ Geococcus coffeae were sister to the two Heliococcus
species þ Neochavesia caldasiae. Although this clade was
part of the unresolved polytomy in maximum parsimony
and neighbour joining trees, it was sister to the rest of the
Pseudococcidae in the Bayesian analysis.
28S D2. A large number of MPTs was found for this
fragment as well, but the strict consensus tree was more
resolved than the 18S tree (see supplemental material;
Fig. S2). Well-supported clades were a monophyletic Puto,
a Phenacoccus þ Heliococcus clade (except Phenacoccus
colemani, which was sister to all other Pseudococcidae), and
a clade consistent with that found in the 18S tree including a
heterogeneous assemblage of species, which in turn was split
into two clades with strong support from all methods
(one consisting of just four species: Paracoccus marginatus,
Paracoccus juniperi, Hypogeococcus pungens and Nipaecoccus
2004 The Royal Entomological Society, Systematic Entomology, 29, 238–259
Mealybug molecular systematics
245
Table 3. Tree statistics. For single gene fragments parsimony searches were conducted with all characters equally weighted, and with first and
second codon positions weighted inversely to their frequencies for both fragments of elongation factor 1a (EF-1a). Parsimony searches for
combined datasets were conducted with all characters weighted equally either including all taxa (equal/all), or only taxa sequenced for all
partitions (equal/delete), and with each partition weighted in inverse proportion to its contribution of parsimony informative characters
(prop.).
Partition
Treatment
Number of informative characters
MPT
Length
CI
RI
HI
n
18S
28S D2
28S D10
EF-1a 30
Equal
Equal
Equal
Equal
Codons – 9 : 1
Equal
Codons – 9 : 1
Equal
Codons 9 : 1
Equal/all
Equal/delete
Equal/all
Equal/delete
Equal/all
Equal/delete
Prop.
Equal/all
Equal/delete
Prop.
66
137
171
107a
57 000
68 014
33 000
12
328
30
45
26
2
2528
471
26 677
45
907
12
1
3321
5
5
263
652
712
779
798
533
576
1135
1693
1412
1354
1697
1494
2565
2017
2900
3175
2177
2179
0.574
0.443
0.475
0.288
0.281
0.253
0.234
0.287
0.344
0.444
0.453
0.458
0.489
0.391
0.448
0.468
0.358
0.447
0.446
0.780
0.751
0.803
0.615
0.602
0.569
0.522
0.510
0.522
0.766
0.756
0.762
0.754
0.705
0.706
0.701
0.673
0.638
0.637
0.426
0.557
0.525
0.713
0.719
0.761
0.766
0.713
0.656
0.556
0.587
0.542
0.563
0.653
0.552
0.596
0.642
0.595
0.554
63
69
67
60
EF-1a 50
EF-1a (both)
28S D2 þ D10
rDNA
EF-1a 30 þ rDNA
All fragments
76b
176
308
303
374
359
481
447
447
557
499
499
51
41
71
65
72
58
73
50
50
74
39
39
MPT, most parsimonious trees; CI, consistency index; RI, retention index; HI, homoplasy index; n, number of taxa; rDNA, 18S þ 28S D2 þ 28S D10.
a
Number of parsimony informative characters by codon position: first position ¼ 8, second position ¼ 6, third position ¼ 93.
b
Number of parsimony informative characters by codon position: first position ¼ 6, second position ¼ 0, third position ¼ 70.
nipae). A clade of Anisococcus þ Ferrisia was more weakly
supported, although the monophyly of these two genera
had high bootstrap support. Other relationships with
weaker support were a root feeding Rhizoecus group, with
Rhizoecus being paraphyletic within the clade, Planococcoides robustus sister to Planococcus, and a clade consisting
of Paradoxococcus mcdanieli þ a group of mostly Pseudococcus and Dysmicoccus species (PP ¼ 0.93). The congeneric
status of the species pairs of Heliococcus, Maconellicoccus
and Planococcus was supported, as was the congeneric
status of Paracoccus burneri and the Paracoccus sp. from
Namibia. It is notable that none of the genera with more
than three representatives was monophyletic (Pseudococcus,
Dysmicoccus, Paracoccus, Phenacoccus). The Bayesian analysis uncovered a sister group relationship of the Rhizoecus
group and the Phenacoccus group, although the PP was
Table 4. Estimates of the shape parameter (a) of the gamma distribution of rates of substitution and the proportion of invariant sites (pinvar)
from hierarchical likelihood ratio tests, and from Bayesian estimation under the general time reversible model. The model of sequence
evolution with the highest likelihood was Tamura & Nei’s (1993) (TrN þ G þ I), with equal base frequencies for 18S and no invariant sites for
28S D2. Standard deviations of the estimate are given in parentheses for Bayesian estimates.
Likelihood ratio
Bayesian
Partition
a
pinvar
a
18S
28S D2
28S D10
EF-1a 50
EF-1a 30
28S D2 þ D10
rDNA
EF-1a 30 þ rDNA
All fragments
0.468
0.456
0.512
1.387
1.490
0.583
0.498
0.523
0.492
0.563
0
0.565
0.616
0.586
0.487
0.525
0.527
0.523
0.050
0.193
0.078
1.063
1.080
0.262
0.397
0.439
0.509
a
Restricted dataset.
rDNA, 18S þ 28S D2 þ 28S D10.
#
2004 The Royal Entomological Society, Systematic Entomology, 29, 238–259
pinvar
(0.000)
(0.012)
(0.042)
(0.339)
(0.235)
(0.020)
(0.044)
(0.044)
(0.064)
0.432
n/e
0.400
0.545
0.510
0.330
0.449
0.473
0.513
(0.034)
(0.083)
(0.034)
(0.030)
(0.027)a
(0.031)a
(0.026)a
(0.030)a
246 D. A. Downie and P. J. Gullan
only 0.81 (not shown). Placement of the Aclerda sp. among
the Pseudococcus, Dysmicoccus and Trionymous species is
explicable only by invoking homoplasy and/or long-branch
attraction in this highly divergent sequence. The amplicon
for this taxon differed in length from all other Pseudococcidae (5–76 bp longer). The minimum uncorrected p-distance
of the Aclerda sequence in the distance matrix was 0.23 and
neighbour joining using uncorrected p-distances placed
Aclerda sister to all Pseudococcidae. Neighbour joining
using the TrN þ G þ I model resulted in distances three- to
five-fold larger and placement of Aclerda as in Fig. S2.
28S D10. The strict consensus of 33 000 MPTs is shown
in the supplementary material Fig. S3. Bootstrap support
was generally low; the greatest support was for the heterogeneous assemblage found in the D2 and 18S trees. Within
this clade the subclade consisting of two Nipaecoccus and
two Paracoccus species plus Hypogeococcus pungens was
strongly supported, consistent with the results from the
D2 sequence. The grouping of the Heliococcus and Maconellicoccus species pairs was supported here as well, as was
that of Anisococcus adenostomae and Anisococcus sp.,
although more weakly, whereas the two Planococcus species
formed a trichotomy with Saccharicoccus sacchari. The two
Trionymus species were grouped together, but with bootstrap support only from neighbour joining, and Ferrisia was
sometimes paraphyletic. High PP from Bayesian analyses
were found for some clades which had poor or only
moderate bootstrap support from maximum parsimony
and neighbour joining. Examples were the Anisococcus
þ Ferrisia clade (PP ¼ 1.00), the clade of two Paracoccus
species, Cyphonococcus alpinus and Sarococcus comis
(PP ¼ 0.85), all from New Zealand, and a larger clade
consisting mostly of Pseudococcus and Dysmicoccus species
(PP ¼ 0.99).
EF-1. The majority of potentially parsimony informative characters in both EF-1a sequences were at third codon
positions (pos1 ¼ 6, pos2 ¼ 0 and pos3 ¼ 70 for EF-1a 50 ;
and pos1 ¼ 8, pos2 ¼ 6 and pos3 ¼ 93 for EF-1a 30 ), as
might be expected in this highly conserved gene at the
amino acid level. Only three and two amino acid changes
occurred in EF-1a 50 and EF-1a 30 , respectively.
For EF-1a 30 , twelve MPTs were found when codon
positions were weighted equally, and 328 MPTs were
found when codon positions were weighted 9:1. The topology and measures of fit of the two strict consensus trees
differed little, although the near basal position for the
Anisococcus þ Ferrisia clade found in the equal weights tree
was not found in the weighted tree, which placed the Phenacoccus group as sister to all other taxa, and is more consistent
with the results from the ribosomal sequences. The strict
consensus tree discussed here (see Supplementary material
Fig. S4) is from the equally weighted analysis and is nearly
completely resolved (three trichotomies), although many
clades have poor bootstrap support, especially at deeper
nodes. The Maconellicoccus and Anisococcus þ Ferrisia
clades were placed deeper in the tree than the Phenacoc-
#
cus þ Heliococcus clade. The Planococcus species are sister
to two major clades which are consistent with the Pseudococcus þ Dysmicoccus group and the heterogeneous assemblage found in the previous analyses. None of these
relationships had bootstrap support however. Bayesian
analysis did, however, find strong support for a sister
group relationship of the large Pseudococcus þ Dysmicoccus
dominated clade (with Anisococcus þ Ferrisia as a subclade
within this clade), and the heterogeneous group found from
the rDNA sequences. With the exception of the placement
of Anisococcus þ Ferrisia, the Bayesian consensus tree was
quite similar to the strict consensus parsimony tree. The
four deepest nodes had PP ¼ 0.76–0.87. Setting the nucmodel to codon had little effect on either the topology or the
PP of nodes, indicating that no additional information was
provided by the larger number of character states. As with
the 28S D2 dataset, the placement of the Aclerda sp. with
Pseudococcus and Dysmicoccus shown in Fig. S4 is explicable only by invoking homoplasy and/or long-branch
attraction in a highly divergent sequence; as noted above,
its intron was considerably longer than those of all pseudococcids. Neighbour joining using uncorrected p-distances
placed Aclerda sister to all Pseudococcidae, except Rhizoecus
hibisci, Rhizoecus gracilis, and Geococcus coffeae. Using the
TrN þ G þ I model resulted in distances up to two times
greater and resulted in a neighbour joining tree with Aclerda
sister to the Phenacoccus þ Heliococcus clade (e.g. uncorrected p-distance from Pseudococcus odermatti ¼ 0.14,
TrN þ G þ I distance to Pseudococcus odermatti ¼ 0.29).
For EF-1a 50 , thirty MPTs were found when codon positions were weighted equally and forty-five MPTs were
found by weighting codon positions 9:1. The equally
weighted tree (see Supplementary material Fig. S5) is discussed here. As with the 30 fragment of this gene, the strict
consensus was fairly well resolved but lacked bootstrap
support in all but the most terminal nodes. In addition, a
number of cases of strongly incongruent relationships suggest that paralogy may have been a bigger problem for this
fragment than the 30 fragment. For example, all three Puto
species (removed from the tree shown) were placed in a
derived position as sister to Dysmicoccus boninsis and
D. neobrevipes, Geococcus coffeae was placed in a clade with
Planococcoides robustus and Planococcus citri embedded
within the Pseudococcus þ Dysmicoccus clade, Tridiscus distichlii was placed with the Maconellicoccus species, and
Chaetococcus bambusae was placed sister to all other mealybug taxa. The only relationships supported by bootstrapping or PP, however, were the species pairs which, for the
most part, were found from the other sequences as well.
Combined datasets
Incongruence length difference tests found significant
heterogeneity among all possible data partitions (generally
P 0.01). Significance was not sensitive to the inclusion or
exclusion of parsimony uninformative characters, taxa with
missing data, or outgroup taxa. As can be seen from the
2004 The Royal Entomological Society, Systematic Entomology, 29, 238–259
Mealybug molecular systematics
results from the individual datasets, the conflict among
partitions was often among poorly resolved nodes. As
only well-supported nodes which conflict can be considered
to pose a problem, the data were combined (although the
reader may make the choice to evaluate the results on the
basis of individual partitions).
Combined EF-1. Combining the two EF-1a fragments
produced twenty-six MPTs; the strict consensus was largely
unresolved (not shown). Interestingly, the Pseudococcus
þ Dysmicoccus clade was retrieved, whereas the large heterogeneous clade found in all other analyses was not, although the
species in this group were not sampled as well for EF-1a 50 .
Weighting codon positions, however, produced two MPTs,
which were consistent with the results from each fragment
separately, although bootstrap values were not improved.
Combined rDNA. Although only six taxa were not
sequenced for both fragments, combination of the two 28S
fragments with all taxa included resulted in 2528 MPTs,
whereas deleting taxa with missing data resulted in 417
MPTs and a reasonably resolved strict consensus (not
shown). As expected, the aclerdid species was placed sister to
the Pseudococcidae (separated from the Pseudococcidae in
Bayesian analysis, PP ¼ 1.0). The Rhizoecinae was placed
sister to all other mealybug taxa by parsimony but was sister
to the Phenacoccus clade in Bayesian analysis (PP ¼ 0.99). In
the full dataset, Rhizoecus hibisci and Rhizoecus gracilis were
paraphyletic. Support for other deep nodes was weak, but
three major clades had good support. These clades were an
Anisococcus þ Ferrisia clade (with good support for monophyly
of both genera), a clade consisting largely of Pseudococcus and
Dysmicoccus species (mostly unresolved except for the
grouping of Dysmicoccus boninsis and the two Trionymus
species), and a clade consistent with the heterogeneous
group found in the single gene analyses. This last clade
was further subdivided by splitting the Planococcus species
and Saccharicoccus sacchari from a clade of Planococcoides
robustus þ all others, within which two of the Paracoccus
species, Hypogeococcus pungens and Nipaecoccus nipae
formed a clade sister to a larger unresolved clade. A close
relationship for the New Zealand mealybugs Paracoccus
nothofagicola, Paracoccus sp., Sarocomis comis and Cyphonococcus alpinus was found within this group. The Bayesian
results dissected this heterogeneous group further, with ten
nodes having PP 0.95. All doublets supported in single
gene analyses retained support in this analysis.
The inclusion of taxa with missing data led to an even more
dramatic increase in the number of MPTs when all rDNA
fragments were combined (Table 3, Fig. 1). The strict consensus of forty-five MPTs (Fig. S6) of the reduced dataset shows
that the additional data from 18S improved resolution and
bootstrap values only slightly. In fact, support for some
nodes decreased. The best resolution and support were
found at intermediate nodes, with weaker support at deep
levels and poor resolution at many terminal nodes. In general,
the same clades were resolved as in the previous analysis.
Bayesian analysis retrieved much the same tree topology,
but the PP of nodes provided stronger confidence and the
#
247
resolution of deeper relationships within the Pseudococcus
þ Dysmicoccus group was supported by PP close to 1.00.
Combined EF-1 and rDNA. Seventy-three samples
(seventy taxa) were included in the combined dataset of
EF-1a 30 and all rDNA fragments (1867 bp), but this was
reduced to fifty when only taxa sequenced for all fragments
were considered (1847 bp). Nine hundred and seven MPTs
resulted from the analysis of the full dataset, twelve MPTs
were found from the reduced dataset under equal weighting,
but weighting partitions according to their contribution of
informative characters and excluding taxa with missing data
led to a single MPT (Fig. S7A). The strict consensus from the
full dataset (equal weights) is shown in Fig. S7B. The single
MPT from the reduced dataset featured strong support for a
number of important terminal and intermediate level nodes, but
most deep nodes received little support from parsimony. All
methods agreed in most details, including the monophyly of
Puto and the placement of the aclerdid as sister to the
pseudococcids, but interesting points of departure were the
relationships among the Rhizoecus group, the Phenacoccus
species, and the Heliococcus species. Heliococcus, Rhizoecus
and Geococcus formed a clade which was near the base of the
tree and split from Phenacoccus by parsimony in the reduced
dataset, whereas Rhizoecus, Geococcus, Heliococcus and
Phenacoccus formed a clade which was monophyletic
(PP ¼ 0.93) and sister to all other taxa (PP ¼ 1.00) by Bayesian
analysis. In the full dataset, Bayesian analysis retained the
monophyly of this clade, but Phenacoccus colemani was split
out. Parsimony had the Rhizoecus clade as monophyletic
and positioned sister to all other mealybug taxa with all
Phenacoccus and Heliococcus forming a more interior clade
(although not supported as monophyletic). All methods
supported Maconellicoccus as sister to the remaining Pseudococcidae (excluding the taxa already discussed). The position of
Planococcus was dependent on sampling: in the reduced dataset
they were next to Maconellicoccus and sister to the more derived
Pseudococcidae, but with additional sampling they were
ambiguously placed within a larger clade. All methods supported the monophyly of the Anisococcus þ Ferrisia clade
(and each genus) but its sister relationship to the rest of Pseudococcidae was only supported by Bayesian analysis (PP ¼ 0.99
reduced dataset, PP ¼ 0.97 full dataset). Two major clades
comprising the bulk of sampled species were retrieved by all
methods; additional species are ambiguously joined to these in
the full dataset. The sister group relationship of these two clades
received only modest support from Bayesian analysis for the
reduced dataset (PP ¼ 0.87), and no support in any of the
other analyses. Monophyly of the genera Heliococcus,
Maconellicoccus, Planococcus, Anisococcus, Ferrisia, Antonina
and Balanococcus (the latter as represented only by species from
New Zealand) was strongly supported by all methods, but all
other genera represented by more than two species were
paraphyletic or polyphyletic. The neighbour joining tree
differed from the MPT and the Bayesian tree in placing the
Anisococcus þ Ferrisia clade, not as sister to the larger clade
of Pseudococcidae, but within the more heterogeneous
clade (with 94% bootstrap support).
2004 The Royal Entomological Society, Systematic Entomology, 29, 238–259
248 D. A. Downie and P. J. Gullan
99/100/100
66/-/61/-/100
97/99/100
68/-/98/98/100
97/99/100
98/99/100
-/60/93
83/80/99 73/57/100
-/-/99
59/-/100
-/-/100
52/-/100
65/-/96
100/99/100
-/97/100
-/99/100
-/94/100
68/-/84/93/100
91/80/100
51/-/-
99/98/100
74/-/100 70/-/97
99/91/100
92/-/100
Pl.citri
Pl.ficus
Pla.robustus
Par. sp. NZ
Par.nothofagicola
Cy.alpinus
S.comis
Chry.longispinus
Chor. sp.
N.viridis
Au.grevilleae
B.diminutus
T.distichlii
Me.albizziae
Am.lichtensioides
An.pretiosa
An.graminis
B. sp.
Eu.europaea
Par.marginatus
Hy.pungens
Par.juniperi
N.nipae
N. nr gilli
Sa.sacchari
P.maritimus
D. sp.
P.viburni
D.brevipes HI
D.brevipes BOL
D.neobrevipes
Plo.eugeniae
D.ryani
P.longispinus
P.calceolariae
Tr.frontalis
D.boninsis
Tr.idahoensis
E.globosum
P.odermatti
Pa.mcdanieli
Paraputo sp.
V.amaryllidis
V.trionymoides
Par. sp. NAM
Par.burnerae
Ch.bambusae
A. sp.
A.adenostomae
F.malvastra
F.virgata
F.gilli
Grewiacoccus sp.
M.australiensis
M.hirsutus
Ph.solani CA
Ph.solani FL
Ph.madeirensis
H.clemente
H.adenostomae
Ph.colemani 2
R.hibisci
G.coffeae
R.gracilis
Ne.caldasiae
Aclerda sp.
Ph.colemani 1
Puto albicans
Puto yuccae
Puto barberi
Puto sp.
Icerya purchasi
Fig. 1. Strict consensus of 26 677 most parsimonious trees for the rDNA dataset including all taxa. The numbers to the left of nodes are
bootstrap values for maximum parsimony and neighbour joining, and posterior probabilities (PP) from Bayesian analysis. A hyphen indicates
that bootstrap values were less than 50% or PP values were less than 0.90 (here shown as a percentage). Neighbour joining was not conducted
for the unrestricted dataset due to the large amount of missing data, which resulted in a substantial number of undefined distances and led to
spurious branch lengths and topology. Support values for some terminal nodes are omitted for clarity.
#
2004 The Royal Entomological Society, Systematic Entomology, 29, 238–259
Mealybug molecular systematics
The combined dataset of both EF-1a fragments and all
rDNA fragments consisted of seventy taxa, but was reduced
to thirty-nine when taxa with missing data were excluded.
Five MPTs were found from both equally weighted and
weighted analyses. Figure S8 shows the strict consensus
tree from the equally weighted analysis of the reduced dataset and Fig. 2 is from the analysis of the full dataset. The
smaller number of taxa and larger sample of nucleotides led
to substantially greater bootstrap support for most nodes.
Again the aclerdid species was sister to the Pseudococcidae
and Puto was monophyletic, although without bootstrap
support in the neighbour joining analysis. The lone representative of the Rhizoecus group, Geococcus coffeae, was
placed next most basally. The monophyly of Phenacoccus
þ Heliococcus, Maconellicoccus, and Anisococcus þ Ferrisia,
was well supported, each of which was in turn sister to
the remaining Pseudococcidae. Both of the large clades
found in the previous analyses were monophyletic, although
their relationship as sister groups was still not supported.
Planococcus citri was placed as sister to the Pseudococcus
þ Dysmicoccus group, and further subdivision of these clades
received good bootstrap support.
The satisfying resolution seen in Fig. S8 was lost to some
extent when all taxa and missing data were included, but the
major features found in the smaller tree are evident. A
striking result is the placement of Planococcus in the heterogeneous clade, although Planococcus citri was placed with
the Pseudococcus þ Dysmicoccus group with 100% bootstrap support in the smaller tree. This suggests that the
results may at times be subject to sampling artefacts.
Discussion
Like other members of the Coccoidea, the Pseudococcidae
is typified by reduction and paucity of morphological characters of adult females useful for phylogenetic analysis. It is
likely that many characters of adult females used by coccoid
systematists are affected by homoplasy. This absence of useful characters and plasticity of others due to environmental
effects has led to difficulties in formulating phylogenetic
hypotheses and coherent classifications. Molecular characters
may have greater power to uncover relationships under these
circumstances, but the current study shows that molecular
data may be no panacea for unravelling pseudococcid relationships. Homoplasy is problematic in these molecular characters, as it is in morphological characters. None of the single
gene analyses led to a set of strongly supported relationships
which would allow confidence in inferring phylogeny across
the family. For the ribosomal genes this was due to a lack of
informative characters at appropriate levels as well as homoplasy, and for EF-1a, homoplasy was largely responsible for
a failure of any but the most recent divergences to be supported. Given the different rates of evolution of the genes
sampled here, we expected that resolution at different levels
of the tree would be realized when the sequences were combined. This expectation was not fully realized, but the results
#
249
are robust enough to allow us to make some inferences with
respect to our objectives.
Subfamily and tribal categories
Five subfamilies have been discussed in the literature:
Pseudococcinae, Trabutininae, Phenacoccinae, Sphaerococcinae and Rhizoecinae. These are, however, neither inclusive of all Pseudococcidae nor exclusive of each other, and
few workers use these categories. Our data are suggestive of
three major clades which might justifiably be given subfamily status (as indicated on Fig. 2). However, these do not
equate with the subfamilies recognized by previous workers
(Koteja, 1974a, b; Danzig, 1980; Williams, 1985; Tang,
1992), except that Koteja (1988) concluded that the mealybugs form just three natural groups, the Phenacoccinae,
Pseudococcinae and Rhizoecinae. He also suggested that
Antonina, Chaetococcus and possibly other genera usually
put in the Sphaerococcinae might be close to the Phenacoccinae (not supported by our data) and, from his diagnosis, it
is clear that he subsumed the Trabutininae under Pseudococcinae (as supported by our data).
Within the large clade which we tentatively consider to be
Pseudococcinae (Fig. 2), there are a number of subclades,
most of which are strongly supported in some or most
analyses, plus a few ungrouped species and genera. First, a
clade consisting of mostly Pseudococcus and Dysmicoccus,
but containing members of Erium, Paradoxococcus, Plotococcus and Trionymus, was consistently recovered and was
strongly supported in combined analyses. We refer to this
group as the tribe Pseudococcini. Representatives from
Vryburgia (probably native to Africa; Millar, 2002), the
African Paracoccus, and the undescribed African Paraputo
species were sometimes closely related to the Pseudococcini
(Fig. 1 and supplementary material Figs S2, S4, S6, S7B),
but bootstrap values and PP did not provide confidence in
that association. There was little differentiation among species within the Pseudococcini suggesting a recent divergence
of these species.
Second, a large clade consisting of a heterogeneous
assemblage of species assignable to at least three of the
subfamilies discussed in the literature was found in every
analysis, including of the highly conserved 18S (see Supplementary material Fig. S1). Even in this gene, additional
structure was found within this clade. This suggests that
divergence among species within this clade was ancient
relative to the other major clade and/or that evolutionary
rates have been faster. Relative rate tests (not shown)
showed that significant rate heterogeneity exists among
taxa, particularly between members of this clade and other
members of the family. An underlying cause for a higher
rate of evolution in this group is not clear, but it should be
noted that this group includes species native to the Australasian, Nearctic, Neotropical, Palaearctic and Oriental
regions. Some geographical structuring was recorded,
including a cluster formed of taxa native to New Zealand,
another cluster of species native to Australia, and a third
2004 The Royal Entomological Society, Systematic Entomology, 29, 238–259
250 D. A. Downie and P. J. Gullan
Fig. 2. Strict consensus of 3321 most parsimonious trees from the combined analysis of all fragments including all taxa. Neighbour joining
was not conducted for this dataset for the same reasons as in Fig. 1. The numbers to the left of nodes are bootstrap values for maximum
parsimony and posterior probabilities (PP) from Bayesian analysis, as in the previous figure. Support values for some terminal nodes are
omitted for clarity. Italicized numbers and letters beneath major nodes indicate support from individual datasets; 1 ¼ 18S; 2a ¼ 28S D2;
2b ¼ 28S D10; E3 ¼ EF-1a 30 ; E5 ¼ EF-1a 50 . Our proposed higher classification for the Pseudococcidae is indicated; subfamilies include taxa
marked with solid lines, whereas tribes and the Ferrisia group include taxa marked with dotted lines.
#
2004 The Royal Entomological Society, Systematic Entomology, 29, 238–259
Mealybug molecular systematics
cluster of taxa from the Americas. One or two of the
African genera Paracoccus (including the type species,
P. burnerae), Paraputo and Vryburgia grouped with the
heterogeneous clade in a few analyses (Figs S7A and S8),
but the positions of these taxa were labile (see above).
Amonostherium, Nipaecoccus and Hypogeococcus were all
found within this heterogeneous clade and have been
considered as part of Trabutininae (Williams, 1985; Tang,
1992). Unfortunately we were unable to sample from the
type genus Trabutina, but it is considered to be closely
related to Amonostherium and Nipaecoccus (Williams,
1985; Danzig & Miller, 1996). Thus, this heterogeneous
clade might be considered to be the tribe Trabutinini
within a broadly defined Pseudococcinae. However,
additional data, including nucleotide sequences from
Trabutina, might lead to a more narrow definition of
Pseudococcinae and recognition of the subfamily
Trabutininae for what we are calling Trabutinini in
Fig. 2.
The ‘legless’ mealybug Antonina falls in the Trabutinini
by our circumscription and appears to be closely related to a
group of species with fully formed legs. Chaetococcus bambusae, another ‘legless’ mealybug, appears to be completely
unrelated to Antonina, which supports the claim by
Hendricks & Kosztarab (1999) that the Sphaerococcinae is
not monophyletic. These were the only purported members
of the Sphaerococcinae sampled and so additional sampling,
including of the type genus Sphaerococcus, is required to
determine the relationships of the various ‘legless’ mealybugs. Should Sphaerococcus be found to belong in this
group, then the name Sphaerococcini has priority over
Trabutinini (Williams, 1969).
Third, the small clade consisting of Anisococcus and
Ferrisia was found to be monophyletic. The P-endosymbionts
of these two genera were also sisters in the molecular phylogeny of Thao et al. (2002). This clade probably represents
a new tribe but here we refer to it as the Ferrisia group until it
can be further studied and formally described. Its relationship to other pseudococcids was fluid, but the weight of
the evidence suggests that it is sister to the larger group of
Pseudococcini þ Trabutinini, as described. Ferrisia was considered part of Pseudococcinae by Koteja (1974b) and Tang
(1992), whereas Anisococcus has never been placed explicitly
in a subfamily. Ferrisia and Anisococcus are both New
World genera (Ben-Dov & German, 2003) and share auxiliary pores on the rim of the dorsal ducts in the adult female,
although their close relationship has not previously been
suggested because they are quite different morphologically.
Afifi (1968) recognized three groups among species which
we here consider to belong to Pseudococcinae; these he
called the Planococcus, Pseudococcus and Saccharicoccus
groups. The first two are equivalent to the tribes Planococcini and Pseudococcini (as indicated on Fig. 2). Planococcoides robustus, which is considered part of the Planococcini
(Ezzat & McConnell, 1956), was sister to Planococcus only
for the analysis of 28S D2 data, although no EF-1a 30 data
were obtained. Saccharicoccus did not consistently group
with any other pseudococcine genus in our sample but was
#
251
associated with both the Planococcini and Trabutinini in
combined analyses. Hodgson (2002) found one wellsupported mealybug clade based on male morphology and
it is equivalent to the Pseudococcinae, because it comprised
one species of each of Pseudococcus, Dysmicoccus, Ferrisia,
Nipaecoccus, Octococcus, Planococcus and Saccharicoccus.
Thao et al. (2002) also found that the P-endosymbionts of
the Pseudococcinae (sensu Fig. 2) formed a well-supported
clade with Maconellicoccus sister to all other species, but
they did not sample members of the Phenacoccinae or
Rhizoecinae. There was significant structure in their tree
with their clusters (A–F) equivalent to our tribes or groups
as follows: their cluster A equivalent to Pseudococcini, B to
the Ferrisia group, C þ D to Trabutinini, E to Planococcini,
and F is Maconellicoccus. Clusters C and D in Thao et al.
(2002) were sister groups with C comprising Antonina pretiosa, Amonostherium lichtensioides and three Australian
species, and D consisting of three New Zealand taxa.
Phenacoccus and Heliococcus were generally associated in
a clade situated as sister to the previous three clades plus
other ambiguously positioned taxa. The monophyly and
position of the Phenacoccus þ Heliococcus clade was only
weakly supported in some analyses. Adult females of Phenacoccus and Heliococccus share nine-segmented antennae,
quinquelocular pores, small lanceolate dorsal setae, and a
denticle on the plantar surface of the claw, although these
features are not unique to these two genera. For example,
Rastrococcus has all of these features and Euripersia, Heterococcus, Peliococcus and Spinococcus have most of them.
Our Phenacoccus þ Heliococcus clade corresponds to the
Phenacoccinae of previous workers but is more restricted
in taxon content. For example, Koteja (1988) also included
Puto, Euripersia and a number of genera which we did not
sample; Puto is clearly a separate family (Cook et al., 2002;
Hodgson, 2002) and Euripersia europaea always appeared
in a distant clade in our analyses. Tang (1992) put almost
forty genera into Phenacoccinae but we were able to sample
only six of them and four of these never grouped with the
type genus Phenacoccus in any of our analyses.
A clade of Rhizoecus, Geococcus and Neochavisia, which
are undisputed members of the Rhizoecinae, was most often
monophyletic and sister to all other mealybug taxa. This
relationship was not found in all analyses, however (e.g.
Fig. S7A); in one analysis, these three genera were placed
in a monophyletic clade with Phenacoccus and Heliococcus
(supplementary Fig. S3). The weight of the evidence summarized by the tree from the combined analysis of all fragments suggests that the Rhizoecinae forms a separate clade
from the Phenacoccinae, but this conclusion will need additional data to be robust.
Our hypothesis for the subfamily and tribal classification
of our sample of the Pseudococcidae is shown in Fig. 2. A
number of species and genera are left in ambiguous positions, which will only be resolved by the inclusion of further
taxa and, ideally, also additional molecular data. For
example, we have included Maconellicoccus and Grewiacoccus
tentatively in the Pseudococcinae as they are successively
sister to the remainder of the subfamily and the adult
2004 The Royal Entomological Society, Systematic Entomology, 29, 238–259
252 D. A. Downie and P. J. Gullan
females share some features with more typical members of
the subfamily. Ezzat (1958), Williams (1958, 1985) and Afifi
(1968) considered Maconellicoccus to be part of the Planococcini. Neither the endosymbiont phylogenetic data nor
the molecular data presented here support this traditional
view, although these studies also do not furnish clear evidence that Maconellicoccus lies outside of the Pseudococcinae. Adult females of the genus share nine-segmented
antennae with the Phenacoccinae, but this condition may
be a plesiomorphy of mealybugs, shared also with Putoidae.
Grewiacoccus, a gall-inducing mealybug from southern
Africa, was redescribed by De Lotto (1969), who mentioned
a resemblance to the North American grass-infesting species
of Pseudantonina. Unfortunately we did not sample any
species of Pseudantonina, although according to Hendricks
& Kosztarab (1999) none of the North American species
of Pseudantonina is congeneric with the type species,
Ps. bambusae from Sri Lanka.
our data of Pseudococcini appear to have related
S-endosymbionts. Mealybugs of Planococcus species have
genetically distinct S-endosymbionts, whereas mealybugs in
our Trabutinini had S-endosymbionts belonging to two
bacterial groups. However, in some of our analyses
(e.g. Fig. 1), the trabutinine mealybugs formed the same
two groups found with the S-endosymbiont data, with the
New Zealand species Paracoccus nothofagicola and Cyphonoccus alpinus forming a clade distinct from Amonostherium, Antonina, Australicoccus and Melanococcus.
S-endosymbionts were not found in the species of Ferrisia
and Maconellicoccus surveyed by Thao et al. (2002),
although their P-endosymbionts were related to
P-endosymbionts in other mealybug genera which did
have S-endosymbionts. Thus, the S-endosymbionts of
Ferrisia and Maconellicoccus may either have been lost
secondarily or never acquired.
Status of the larger genera
Mealybug relationships and endosymbionts
One significant observation is that the endosymbionts of
Phenacoccus, Heliococcus and four other genera which we
did not sample (Coccidohystrix [¼ Centrococcus], Ceroputo
[currently in Puto but a doubtful placement], Eumyrmococcus and Ripersia), are morphologically different from those
of Antonina, Dysmicoccus, Ferrisia, Nipaecoccus, Planococcus and Pseudococcus (Buchner, 1965; Tremblay, 1989). In
the latter genera (our Pseudococcinae), Buchner recorded
bacterial symbionts embedded in mucous spherules which
have no equivalent in other scale insects. It has since been
shown that these mucous spherules are the primary endosymbionts (P-endosymbionts), which all belong to the
b-subdivision of the Proteobacteria, and that they house the
secondary endosymbionts (S-endosymbionts), which belong
to the g-Proteobacteria (von Dohlen et al., 2001; Thao et al.,
2002). By contrast, the endosymbionts of Phenacoccus,
Heliococcus and some other genera are included directly in
the cytoplasm of the bacteriocyte and the mucous spherules
are lacking (Buchner, 1965; Tremblay, 1989). This observation may be highly significant to the phylogeny and classification of mealybugs because it may mean that a significant
synapomorphy of the subfamily Pseudococcinae is the
intimate association of two bacterial species, or at least the
acquisition of a particular kind of primary bacterium
capable of such an intracellular symbiosis with another
bacterium. Unfortunately, no bacterial endosymbionts of
the Phenacoccinae (or of the Rhizoecinae) have been
sequenced; all mealybugs studied by Thao et al. (2002)
belong to the Pseudococcinae sensu Fig. 2.
The identity of S-endosymbionts may provide additional
evidence for mealybug classification. The S-endosymbionts
of the Pseudococcinae appear to have colonized their
P-endosymbiont hosts on multiple occasions because different groups of mealybugs contain different S-endosymbionts
(Thao et al., 2002). Interestingly, related mealybugs from
#
A number of genera consistently received support as
monophyletic groups, but in each case these were only
represented by two taxa. With the exception of Ferrisia, all
genera sampled more than twice were not monophyletic.
This includes the important genera Pseudococcus, Dysmicoccus and Paracoccus, which clearly need to be studied
further. Phenacoccus and Nipaecoccus were also not monophyletic in most analyses. Thus, few large pseudococcid
genera as currently defined may be able to withstand further
scrutiny. Most of the speciose genera have broad distributions. For example, Paracoccus is represented in almost all
biogeographical regions (Ben-Dov & German, 2003), but
our data strongly suggest that species in Africa (including
the type species P. burnerae) are unrelated to species in New
Zealand, which in turn are unrelated to species from the
Americas. Historically, species of similar female morphology in different regions have been placed in the same
genus, largely due to an understandable reluctance to create
many new genera. Cox (1987) took the alternative approach
of creating a number of new genera for New Zealand mealybugs with the explicit intention of not confusing biogeographers who usually work solely from the literature. She
argued that mealybug species have a tendency to specialize
by reduction of features, with the females of species living in
similar habitats tending to converge in appearance.
Only the use of different datasets, such as from male
morphology and DNA sequences, combined with more
intensive sampling of species will resolve the status of
many genera.
EF-1a and paralogy
The downstream fragment of EF-1a 30 provided the best
resolved tree of the single gene analyses and is responsible for
much of the resolution found in the single MPT resulting
from combining it with three rDNA fragments. To the extent
2004 The Royal Entomological Society, Systematic Entomology, 29, 238–259
Mealybug molecular systematics
that EF-1a is prone to confounding from the presence of
paralogous copies, it is reasonable to ask whether this gene
has been misleading. The presence of paralogous copies
among PCR products was detected by cloning, and at least
four taxa were striking in their incongruent position for one
or the other fragment (Geococcus coffeae, Tridiscus distichlii
and Chaetococcus bambusae for EF-1a 30 , and Vryburgia
trionymoides for EF-1a 50 ). In those insect taxa where paralogues have been described for EF-1a, they have been highly
divergent (25%) and distinguished by intron presence/
absence or position (Hovemann et al., 1988; Danforth & Ji,
1998). Homology of the sequences in the matrices for both
EF-1a fragments here might be inferred from the presence of
an intron at the same location in all taxa. Furthermore,
excluding outgroups, only a few of the most divergent
sequences in the tree approached the 18% uncorrected
sequence divergence which Danforth & Ji (1998) suggested
as a threshold for caution. Although a paralogue need not be
nonfunctional, the absence of stop codons in any EF-1a
sequence makes it unlikely that a silenced copy was
sequenced. This does not rule out the possibility that a paralogous copy was sequenced. Cloning PCR products was
informative with regard to the nature of paralogous copies
of EF-1a in mealybugs, and indicates that the criteria for
recognizing paralogues may not be met (the paralogous
copy found in Rhizoecus hibisci, for example, was not distinguished by intron presence/absence or location). The best
evidence that our results are not affected, overall, by paralogy,
is the corroboration from other genes, especially the
consistent association of conspecifics and congenerics without significant displacement in tree topology. Inferences
drawn from sequences of EF-1a without corroboration
from other lines of evidence, or rigorous molecular
biology, should be viewed with caution.
Genome-wide sampling
The importance of sampling more than a single gene, or
gene fragment, has often been stressed (Pamilo & Nei, 1988;
Slade et al., 1994; Ruvolo, 1997). Concomitant on this
imperative is the desire that results from different regions
of the genome be congruent. Given the established variation
in rates of evolution among genes, among regions within
genes, and among codon positions, and the different properties of recombination and gene flow which affect sequence
variation within and among taxa, it should be no surprise
that results from different regions of the genome are often
incongruent. This being said, one might wonder what level of
confidence should be placed on any analysis based on a single
snippet of the genome, or what justification merits drawing
strong inferences from a collection of independent analyses
of different snippets of the genome, without an objective
basis for synthesis of these disparate lines of evidence. Thus,
we argue here that the greatest weight of evidence accrues
from the combined analysis of all sampled regions of the
genome.
#
253
Conclusions and future work
From our phylogenetic trees based on combined data,
there are three clades which can reasonably be equated to
previous ideas on classification. It is most parsimonious to
recognize these three major groups as the subfamilies Pseudococcinae, Phenacoccinae and Rhizoecinae. Within the large
clade which we call Pseudococcinae, there are a number
of subclades which we have either named as tribes according
to previous classifications or left as informal groups or
unplaced genera. The relationships among these various
suprageneric groups are largely poorly resolved, although
it is likely that the Rhizoecinae (which is well defined
morphologically and biologically) is sister to the other
higher taxa. Furthermore, Puto was mostly resolved as a
well-supported monophyletic group, except in trees from
the EF-1a dataset, in which is it possible that paralogous
copies were present. It is also possible that substitution
saturation or long-branch attraction may have affected the
results for Puto with EF-1a, as appears to be the case for the
Aclerda sp. It was sister to the Pseudococcidae plus Aclerda
(the other neococcoid) when trees were rooted on the
distant outgroup Icerya (Margarodidae), providing support
for the placement of Puto in its own family.
The relationships of Puto warrant further study with other
conserved genes and gene regions, and with additional
species of Puto, including species belonging to the two
genera, Ceroputo and Macrocerocccus, which are subjective
synonyms of Puto (Williams & Granara de Willink, 1992;
Ben-Dov, 1994; Ben-Dov & German, 2003).
Future work must assess morphological criteria to support
or diagnose the above groups. Koteja (1988) provided diagnoses of the Pseudococcinae, Phenacoccinae and Rhizoecinae, based largely on the Palaearctic taxa with which he is
familiar. For the Pseudococcinae at least, there do seem to be
a few exceptions to one or more of the features in Koteja’s
diagnosis (especially body colour and form of the dorsal
setae) among taxa which we here consider as Pseudococcinae. The Phenacoccinae cannot be defined adequately until
further taxa are sampled for DNA data. We sequenced only
two genera, but Koteja and others have suggested additional
genera which may belong in this group. Williams (1998)
diagnosed the genera of Rhizoecinae without explicitly
redescribing the subfamily, which is diagnosed briefly in
Williams (1969). One purported relationship which requires
further research is the placement in the Pseudococcinae of
the anomalous ant-attended mealybugs of the tribe Allomyrmococcini (Williams, 1978; Tang, 1992). Members of this
tribe might be confused with southern Asian genera of antattended Rhizoecinae because the adult females of most
species in both groups have unusual appendage and body
morphology with at least the dorsum densely covered in
minute setae (Williams, 1978, 2002). The Allomyrmococcini
contains eleven genera and thirty-seven species of mealybug
which feed aerially on their host plants (Williams, 2002) and
are attended by ants of the genus Dolichoderus which carry
the mealybugs around (Dill et al., 2002). The similarity in
structure of the Allomyrmococcini and certain Rhizoecinae
2004 The Royal Entomological Society, Systematic Entomology, 29, 238–259
254 D. A. Downie and P. J. Gullan
may be due to convergence associated with close association
with ants. Molecular data should resolve whether these
mealybug taxa are related. In particular, study of the DNA
of their endosymbionts might be informative because Buchner
(1965) recorded that the endosymbionts of Eumyrmococcus
(a member of the Rhizoecinae) is of the Phenacoccus type
rather than the Pseudococcus type.
Finally, DNA data should be acquired from many more
genera and species in order to provide a robust classification
of the Pseudococcidae. Particularly important or interesting
genera for future studies include Antoninoides, Atrococcus,
Brevennia, Cataenococcus, Ceroputo (currently in Puto),
Chorizococcus, Coccidohystrix, Crisococcus, Cryptoripersia,
Distichlicoccus, Ehrhornia, Eumyrmococcus, Formicococcus,
Heterococcus, Hopefoldia, Humococcus, Kiritshenkella,
Lenania, Leptococcus, Macrocerococcus (currently in
Puto), Nesopedronia, Nipaecoccus, Palmicultor, Paraputo,
Pseudantonina, Radicoccus (currently synonymized with
Rhizoecus), Rastrococcus, Serrolecanium, Sphaerococcus,
Spilococcus, Stemmatomerinx, Trabutina and Ventrispina.
As far as possible, it will be important to include the type
species of these genera in molecular analyses. In addition, it
is essential to sample many more species of the large and
economically important genera, such as Dysmicoccus,
Phenacoccus, Pseudococcus and Trionymus, to test for
generic monophyly and species synonymy. Nomenclatural
details plus taxonomic, distribution and host-plant data
can be obtained for any species by consulting ScaleNet
(Ben-Dov et al., 2003).
Supplementary material
The following material is available from http://www.
blackwellpublishing.com/products/journals/suppmat/SEN/
SEN241/SEN241sm.htm.
Fig. S1. Strict consensus of 57 000 most parsimonious
trees for 575 bp of 18S. Tree statistics can be found in
Table 3. The numbers to the left of nodes are bootstrap
values for maximum parsimony and neighbour joining,
and posterior probabilities (PP) from Bayesian analysis. A
hyphen indicates that bootstrap values were less than 50%
or PP values were less than 0.90 (here shown as a percentage). Support values for some terminal nodes are omitted
for clarity.
Fig. S2. Strict consensus of 68 014 most parsimonious
trees for 255 bp of the D2 expansion region of 28S. The
numbers to the left of nodes are bootstrap values for maximum parsimony and neighbour joining, and posterior
probabilities (PP) from Bayesian analysis. A hyphen indicates that bootstrap values were less than 50% or PP values
were less than 0.90. Support values for some terminal nodes
are omitted for clarity.
Fig. S3. Strict consensus of 33 000 most parsimonious
trees for 721 bp of the D10 expansion region of 28S. The
numbers to the left of nodes are bootstrap values for maximum parsimony and neighbour joining, and posterior
#
probabilities from Bayesian analysis, as in previous figures.
Support values for some terminal nodes are omitted for
clarity.
Fig. S4. Strict consensus of twelve most parsimonious
trees for 316 bp of elongation factor 1a (EF-1a 30 ). The
numbers to the left of nodes are bootstrap values for maximum parsimony and neighbour joining, and posterior
probabilities from Bayesian analysis, as in previous figures.
Support values for some terminal nodes are omitted for
clarity.
Fig. S5. Strict consensus of 810 most parsimonious trees
for 238 bp of elongation factor 1a (EF-1a 50 ). The numbers
to the left of nodes are bootstrap values for maximum
parsimony and neighbour joining, and posterior probabilities from Bayesian analysis, as in previous figures. Support
values for some terminal nodes are omitted for clarity.
Fig. S6. Strict consensus of forty-five most parsimonious trees from all rDNA fragments for the reduced dataset excluding taxa with missing data. The numbers to the
left of nodes are bootstrap values for maximum parsimony
and neighbour joining, and posterior probabilities from
Bayesian analysis. Support values for some terminal nodes
are omitted for clarity.
Fig. S7. (A) A single most parsimonious tree from the
combined analysis of elongation factor 1a (EF-1a 30 ), 28S
D2 and D10, and 18S from weighted analysis excluding taxa
with missing data. (B) Strict consensus of 907 most parsimonious trees from the combined analysis including all
taxa. The numbers to the left of nodes are bootstrap values
for maximum parsimony and neighbour joining, and posterior probabilities from Bayesian analysis. Neighbour joining was not conducted for the unrestricted dataset due to
the large amount of missing data which resulted in a substantial number of undefined distances and led to spurious
branch lengths and topology. Support values for some
terminal nodes are omitted for clarity.
Fig. S8. Strict consensus of five most parsimonious trees
from the combined analysis of all fragments excluding taxa
with missing data. The numbers to the left of nodes are
bootstrap values for maximum parsimony and neighbour
joining, and posterior probabilities from Bayesian analysis.
Support values for some terminal nodes are omitted for
clarity.
Acknowledgements
We acknowledge the government agencies and officers who
provided collecting permits, especially the Australian
National Botanic Gardens, Canberra; the Australian Capital Territory Parks & Conservation, Canberra; and the New
Zealand Department of Conservation (particularly Chris
Green). We are especially grateful to the following people
who either collected mealybug specimens for us or facilitated their collection: Greg Baker, Paul Baumann, Peter
Cranston, Jacques Delabie, Stuart Donaldson, Ray Gill,
Avas Hamon, Steve Heydon, Robert Hoare, My-My
Huynh, Marshall Johnson, Sandy Kelley, Demian Kondo,
2004 The Royal Entomological Society, Systematic Entomology, 29, 238–259
Mealybug molecular systematics
Heather Laflin, Rich Leschen, Dale Meyerdirk, Dug Miller,
Ross Miller, Nancy Moran, Tanya Price, Birgit SchlickSteiner and Florian Steiner, Maureen Stanton, Shawn Steffan
and Phil Ward. In particular, Dug Miller and Ray Gill
greatly improved the dataset by providing specimens of
many species. We also acknowledge Ray Gill, Jan Koteja,
Dug Miller and Doug Williams for ideas and helpful discussion on mealybug relationships, and Chris Hodgson, Jan
Koteja, Dug Miller, Douglas Williams and an anonymous
molecular phylogeneticist for reviewing the manuscript.
Lyn Cook kindly provided information aiding DNA extraction from voucher specimens. We again thank Sheryl
Bernauer, Shelly Williams and Kerry Cloud at the Division
of Biological Sciences DNA Sequencing Facility at University
of California, Davis. This research was supported in part by
Hatch funding from the California Agricultural Experiment
Station, by the Evert and Marion Schlinger Endowment
Fund of the Department of Entomology, University of
California, Davis, and by support from the U.S. National
Science Foundation (Partnerships for Enhancing Expertise in
Taxonomy program under Grant No. 0118718).
References
Afifi, S.A. (1968) Morphology and taxonomy of the adult males of
the families Pseudococcidae and Eriococcidae (Homoptera:
Coccoidea). Bulletin of the British Museum (Natural History),
Entomology Supplement, 13, 1–210.
Banks, H.J., Cameron, D.W. & Raverty, W.D. (1976) Chemistry of
the Coccoidea. II. Condensed polycyclic pigments from two
Australian pseudococcids (Hemiptera). Australian Journal of
Chemistry, 29, 1509–1521.
Beardsley, J.W. (1962) Descriptions and notes on male mealybugs
(Homoptera: Pseudococcidae). Proceedings of the Hawaiian
Entomological Society, 28(1961), 81–98.
Beardsley, J.W. (1969) A new fossil scale insect from Canadian
amber. Psyche, 76, 270–279.
Belshaw, R. & Quicke, D.L.J. (1997) A molecular phylogeny of the
Aphidiinae (Hymenoptera: Braconidae). Molecular Phylogenetics and Evolution, 7, 281–293.
Ben-Dov, Y. (1990) Classification of diaspidoid and related Coccoidea.
Armored Scale Insects, Their Biology, Natural Enemies and Control.
World Crop Pests, Vol. 4A (ed. by D. Rosen), pp. 97–128. Elsevier,
Amsterdam.
Ben-Dov, Y. (1994) A Systematic Catalogue of the Mealybugs of the
World (Insecta: Homoptera: Coccoidea: Pseudococcidae and
Putoidae) with Data on Geographical Distribution, Host Plants,
Biology and Economic Importance. Intercept, Andover.
Ben-Dov, Y. & German, V. (2003) Pseudococcidae. ScaleNet: a
Database of the Scale Insects of the World (ed. by Y. Ben-Dov,
D.R. Miller and G.A.P. Gibson). http://www.sel.barc.usda.gov/
scalenet/scalenet.htm.
Ben-Dov, Y., Miller, D.R. & Gibson, G.A.P. (2003) ScaleNet: a
Database of the Scale Insects of the World. http://www.sel.barc.
usda.gov/scalenet/scalenet.htm.
Bink-Moenen, R.M. & Mound, L.A. (1990) Whiteflies: diversity,
biosystematics and evolutionary patterns. Whiteflies: Their
Bionomics, Pest Status and Management (ed. by D. Gerling),
pp. 1–11. Intercept, Andover.
#
255
Brown, S.W. & Cleveland, C. (1968) Meiosis in the male of Puto
albicans (Coccoidea-Homoptera). Chromosoma, 24, 210–232.
Buchner, P. (1965) Endosymbiosis of Animals with Plant Microorganisms. Interscience, New York.
Charles, J.G., Froud, K.J. & Henderson, R.C. (2000) Morphological variation and mating compatibility within mealybugs
Pseudococcus calceolariae and P. similans (Hemiptera: Pseudococcidae), and a new synonymy. Systematic Entomology, 25,
285–294.
Cho, S., Mitchell, A., Regier, J.C., Mitter, C., Poole, R.W.,
Friedlander, T.P. & Zhao, S. (1995) A highly conserved nuclear
gene for low-level phylogenetics: elongation factor 1-alpha
recovers morphology-based tree for heliothine moths. Molecular
Biology and Evolution, 12, 650–656.
Cook, L.G., Gullan, P.J. & Trueman, H.E. (2002) A preliminary
phylogeny of the scale insects (Hemiptera: Sternorrhyncha:
Coccoidea) based on nuclear small-subunit ribosomal DNA.
Molecular Phylogenetics and Evolution, 25, 43–52.
Cox, J.M. (1983) An experimental study of morphological
variation in mealybugs (Homoptera: Coccoidea: Pseudococcidae). Systematic Entomology, 8, 361–382.
Cox, J.M. (1987) Pseudococcidae (Insecta: Hemiptera). Fauna of
New Zealand, 11, 1–230.
Cox, J.M. & Pearce, M.J. (1983) Wax produced by dermal pores in
three species of mealybug (Homoptera: Pseudococcidae). International Journal of Insect Morphology and Embryology, 12,
235–248.
Cunningham, C.W. (1997) Can three incongruence tests predict
when data should be combined? Molecular Biology and
Evolution, 14, 733–740.
Danforth, B.N. & Ji, S. (1998) Elongation factor 1-a occurs as two
copies in bees: implications for phylogenetic analysis of EF-1-a
in insects. Molecular Biology and Evolution, 15, 225–235.
Danzig, E.M. (1980) Coccids of the Far East USSR (Homoptera,
Coccinea) with Phylogenetic Analysis of Scale Insect Fauna of the
World. Nauka, Leningrad (in Russian).
Danzig, E.M. & Miller, D.R. (1996) A systematic revision of the
mealybug genus Trabutina (Homoptera: Coccoidea: Pseudococcidae). Israel Journal of Entomology, 30, 7–46.
Darlu, P. & Lecointre, G. (2002) When does the Incongruence
Length Difference test fail? Molecular Biology and Evolution, 19,
432–437.
De Lotto, G. (1969) The mealybugs of South Africa (Homoptera:
Pseudococcidae), II. Entomology Memoirs, Department of
Agricultural Technical Services, Republic of South Africa
Pretoria, 20, 1–30.
Dietrich, C.H., Rakitov, R.A., Holmes, J.L. & Black, W.C. (2001)
Phylogeny of the major lineages of Membracoidea based on 28S
rDNA sequences. Molecular Phylogenetics and Evolution, 18,
293–305.
Dill, M., Williams, D.J. & Maschwitz, U. (2002) Herdsmen ants
and their mealybug partners. Abhandlungen der Senckenbergischen Naturforschenden Gesellschaft, 557, 1–373.
Dolphin, K., Belshaw, R., Orme, C.D. & Quicke, D.L.J. (2000)
Noise and incongruence: interpreting results of the Incongruence
Length Difference test. Molecular Phylogenetics and Evolution,
17, 401–406.
Dowton, M. & Austin, A.D. (2002) Increased congruence does not
necessarily indicate increased phylogenetic accuracy – the
behavior of the incongruence length difference test in mixed
model analysis. Systematic Biology, 51, 19–31.
Ezzat, Y.M. (1958) Maconellicoccus hirsutus (Green), a new genus, with
description of the species (Homoptera: Pseudococcidae-Coccoidea).
Bulletin de la Socie´te´ Entomologique d’Égypte, 42, 377–383.
2004 The Royal Entomological Society, Systematic Entomology, 29, 238–259
256 D. A. Downie and P. J. Gullan
Ezzat, Y.M. & McConnell, H.S. (1956) A classification of the
mealybug tribe Planococcini (Pseudococcidae, Homoptera).
Bulletin of the University of Maryland Agricultural Experiment
Station, A-84, 1–108.
Farris, J.S., Kallerjo, M., Kluge, A.G. & Bult, C. (1994) Testing
significance of congruence. Cladistics, 10, 315–319.
Ferris, G.F. (1950) Atlas of the Scale Insects of North America, V, the
Pseudococcidae (Part I). Stanford University Press, California.
Ferris, G.F. (1953) Atlas of the Scale Insects of North America, VI, the
Pseudococcidae (Part II). Stanford University Press, California.
Gullan, P.J., Downie, D.A. & Steffan, S.A. (2003) A new pest
species of the mealybug genus Ferrisia Fullaway (Hemiptera:
Pseudococcidae) from the USA. Annals of the Entomological
Society of America, 96, 723–737.
Gullan, P.J. & Kosztarab, M. (1997) Adaptations in scale insects.
Annual Review of Entomology, 42, 23–50.
Hambleton, E.J. (1946) Studies of hypogeic mealybugs. Revista de
Entomologia, 17, 1–77.
Hambleton, E.J. (1976) A revision of the New World mealybugs of
the genus Rhizoecus (Homoptera: Pseudococcidae). United
States Department of Agriculture Technical Bulletin, 1522, 1–88.
Hancock, J.M., Tautz, D. & Dover, G.M. (1988) Evolution of the
secondary structure and compensatory mutations of the
ribosomal RNAs of Drosophila melanogaster. Molecular Biology
and Evolution, 5, 393–414.
Hendricks, H.J. & Kosztarab, M. (1999) Revision of the Tribe
Serrolecaniini (Homoptera: Pseudococcidae). Walter de Gruyter,
Berlin.
Hodgson, C.J. (1994) The Scale Insect Family Coccidae.
An Identification Manual to Genera CAB International,
Wallingford.
Hodgson, C.J. (2002) Preliminary phylogeny of some nonmargarodid Coccoidea (Hemiptera) based on adult male
characters. Bollettino di Zoologia Agraria e di Bachicoltura,
33(2001), 129–137.
Hovemann, B., Richter, S., Waldorf, U. & Cziepluch, C. (1988)
Two genes encode related cytoplasmic elongation factors 1a in
Drosophila melanogaster with continuous and stage specific
expression. Nucleic Acids Research, 16, 3175–3194.
Howell, J.O. & Tippins, H.H. (1990) The immature stages.
Armored Scale Insects, Their Biology, Natural Enemies and
Control. World Crop Pests, Vol. 4A (ed. by D. Rosen), pp.
29–42. Elsevier, Amsterdam.
Huelsenbeck, J.P. & Ronquist, F. (2001) MrBayes: Bayesian
inference of phylogeny. Bioinformatics, 17, 754–755.
Johnson, C., Agosti, D., Delabie, J.H., Dumpert, K., Williams,
D.J., von Tschirnhaus, M. & Maschwitz, U. (2001) Acropyga
and Azteca ants with scale insects: 20 million years of intimate
symbiosis. American Museum Novitates, 3335, 1–18.
Kosztarab, M. & Kozár, F. (1988) Scale Insects of Central Europe.
Akademiai Kiado, Budapest.
Koteja, J. (1974a) Comparative studies on the labium in the
Coccinea (Homoptera). Zeszyty Naukowe Akademii Rolniczej,
Kraków, Series Rozprawy, 27, 1–162.
Koteja, J. (1974b) On the phylogeny and classification of the
scale insects (Homoptera, Coccinea) (discussion based on the
morphology of the mouthparts). Acta Zoologica Cracoviensia, 19,
267–326.
Koteja, J. (1985) Essay on the prehistory of the scale insects
(Homoptera, Coccinea). Annales Zoologici (Warszawa), 38,
461–503.
Koteja, J. (1988) Reviewer’s remark. Scale Insects of Central
Europe (ed. by M. Kosztarab and F. Kozár), pp. 60–61.
Akademiai Kiado, Budapest.
#
Maddison, W.P. & Maddison, D.R. (2000) Macclade: Analysis of
Phylogeny and Character Evolution, Version 4.0. Sinauer
Associates, Sunderland, Massachusetts.
Mason-Gamer, R.J. & Kellogg, E.A. (1996) Testing for phylogenetic conflict among molecular data sets in the tribe Triticeae.
Systematic Biology, 45, 522–543.
McKenzie, H.L. (1967) Mealybugs of California with Taxonomy,
Biology and Control of North American Species (Homoptera:
Coccoidea: Pseudococcidae). University of California Press,
California.
Michot, B., Hassouna, N. & Bachellerie, J. (1984) Secondary structure
of mouse 28S rRNA and general model for the folding of the large
rRNA in eukaryotes. Nucleic Acids Research, 12, 4259–4279.
Millar, I.M. (2002) Mealybug genera (Hemiptera: Pseudococcidae)
of South Africa: identification and review. African Entomology,
10, 185–233.
Miller, D.R. (1991) The scales, scale insects or coccoids. Immature
Insects, Vol. 2 (ed. by F.W. Stehr), pp. 90–107. Kendall/Hunt,
Dubuque, Iowa.
Miller, D.R., Gullan, P.J. & Williams, D.J. (1998) Family
placement of species previously included in the scale insect
genus Sphaerococcus Maskell (Hemiptera: Coccoidea). Proceedings of the Entomological Society of Washington, 100, 286–305.
Miller, D.R. & Kosztarab, M. (1979) Recent advances in the study
of scale insects. Annual Review of Entomology, 24, 1–27.
Miller, D.R. & Miller, G.R. (1993) A new species of Puto and a
preliminary analysis of the phylogenetic position of the Puto
group within the Coccoidea (Homoptera: Pseudococcidae).
Jeffersoniana: Contributions from the Virginia Museum of
Natural History, 4, 1–35.
Miller, D.R., Miller, G.L. & Watson, G.W. (2002) Invasive species
of mealybugs (Hemiptera: Pseudococcidae) and their threat to
U.S. agriculture. Proceedings of the Entomological Society of
Washington, 104, 824–835.
Moran, N.A., Kaplan, M.E., Gelsey, M.J., Murphy, T.G. &
Scholes, E.A. (1999) Phylogenetics and evolution of the aphid
genus Uroleucon based on mitochondrial and nuclear DNA
sequences. Systematic Entomology, 24, 85–93.
Normark, B.B. (2000) Molecular systematics and evolution of the
aphid family Lachnidae. Molecular Phylogenetics and Evolution,
14, 131–140.
Nur, U. (1977) Electrophoretic comparison of enzymes of sexual
and parthenogenetic mealybugs (Homoptera: Coccoidea: Pseudococcidae). Virginia Polytechnic Institute and State University
Research Division Bulletin, 127, 69–84.
Pamilo, P. & Nei, M. (1988) Relationships between gene trees and
species trees. Molecular Biology and Evolution, 5, 568–583.
Posada, D. & Crandall, K.A. (1998) MODELTEST: testing the
model of DNA substitution. Bioinformatics, 14, 817–818.
Remaudière, G. & Remaudière, M. (1997) Catalogue of the World’s
Aphididae (Homoptera Aphidoidea). Institut National de la
Recherche Agronomique, Paris.
Rohrbach, K.G., Beardsley, J.W., German, T.L., Reimer, N.J. &
Sanford, W.G. (1988) Mealybug wilt, mealybugs, and ants on
pineapple. Plant Disease, 72, 558–565.
Rousset, F., Pelandakis, M. & Solignac, M. (1991) Evolution of
compensatory substitutions through G-U intermediate state in
Drosophila rRNA. Proceedings of the National Academy of
Sciences, 88, 10032–10036.
Ruvolo, M. (1997) Molecular phylogeny of the hominids:
inferences from multiple independent DNA sequence data sets.
Molecular Biology and Evolution, 14, 248–265.
Slade, R.W., Moritz, C. & Heideman, A. (1994) Multiple nuclear
gene phylogenies: application to pinnipeds and comparison with
2004 The Royal Entomological Society, Systematic Entomology, 29, 238–259
Mealybug molecular systematics
a mitochondrial DNA gene phylogeny. Molecular Biology and
Evolution, 11, 341–356.
Swofford, D.L. (2003) Paup* Phylogenetic Analysis Using Parsimony (*and Other Methods), Version 4. Sinauer Associates,
Sunderland, Massachusetts.
Tamura, K. & Nei, M. (1993) Estimation of the number of
nucleotide substitutions in the control region of mitochondrial
DNA in humans and chimpanzees. Molecular Biology and
Evolution, 10, 512–526.
Tang, F.T. (1992) The Pseudococcidae of China. Shanxi Agricultural University, Taigu, Shanxi (in Chinese; summary and key in
English).
Tautz, D., Hancock, J.M., Webb, D.A., Tautz, C. & Dover, G.A.
(1988) Complete sequences of the rRNA genes of Drosophila
melanogaster. Molecular Biology and Evolution, 5, 366–376.
Thao, M.L., Gullan, P.J. & Baumann, P. (2002) Secondary
(g-Proteobacteria) endosymbionts infect the primary
(b-Proteobacteria) endosymbionts of mealybugs multiple times
and coevolve with their hosts. Applied and Environmental
Microbiology, 68, 3190–3197.
Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F. &
Higgins, D.G. (1997) The CLUSTAL–Windows interface:
flexible strategies for multiple sequence alignment aided by
quality analysis tools. Nucleic Acids Research, 25, 4876–4882.
Tremblay, E. (1989) Coccoidea endocytobiosis. Insect Endocytobiosis: Morphology, Physiology, Genetics, Evolution (ed. by W.
Schwemmler and G. Gassner), pp. 145–173. CRC Press, Boca
Raton, Florida.
von Dohlen, C.D., Kohler, S., Alsop, S.T. & McManus, W.R.
(2001) Mealybug b-endosymbionts contain g-proteobacterial
symbionts. Nature, 412, 433–436.
von Dohlen, C.D., Kurosu, U. & Aoki, S. (2002) Phylogenetics and
evolution of the eastern Asian–eastern North American disjunct
aphid tribe, Hormaphidini. Molecular Phylogenetics and Evolution, 23, 257–267.
von Dohlen, C.D. & Moran, N.A. (1995) Molecular phylogeny of
the Homoptera: a paraphyletic taxon. Journal of Molecular
Evolution, 41, 211–223.
von Dohlen, C.D. & Moran, N.A. (2000) Molecular data support a
rapid radiation of aphids in the Cretaceous and multiple origins of
host alternation. Biology Journal of the Linnean Society, 71, 689–717.
White, I.M. & Hodkinson, I.D. (1985) Nymphal taxonomy and
systematics of the Psylloidea (Homoptera). British Museum of
Natural History (Entomology), 50, 153–301.
#
257
Wiens, J.J. (1998) Does adding characters with missing data
increase or decrease phylogenetic accuracy? Systematic Biology,
47, 625–640.
Williams, D.J. (1958) The mealy-bugs (Pseudococcidae: Homoptera) described by W.M. Maskell, R. Newstead, T.D.A.
Cockerell and E.E. Green from the Ethiopian Region. Bulletin
of the British Museum of Natural History (Entomology), 6,
203–236.
Williams, D.J. (1969) The family-group names of the scale insects
(Hemiptera: Coccoidea). Bulletin of the British Museum of
Natural History (Entomology), 23, 315–341.
Williams, D.J. (1978) The anomalous ant-attended mealybugs
(Homoptera: Pseudococcidae) of south-east Asia. Bulletin of the
British Museum of Natural History (Entomology), 37(1), 1–72.
Williams, D.J. (1985) Australian Mealybugs. British Museum of
Natural History, London.
Williams, D.J. (1989) The mealybug genus Rastrococcus Ferris
(Hemiptera: Pseudococciae). Systematic Entomology, 14,
433–486.
Williams, D.J. (1998) Mealybugs of the genera Eumyrmococcus
Silvestri and Xenococcus Silvestri associated with the ant genus
Acropyga Roger and a review of the subfamily Rhizoecinae
(Hemiptera, Coccoidea, Pseudococcidae). Bulletin of the Natural
History Museum, London (Entomology Series), 67, 1–64.
Williams, D.J. (2001) Appendix 1. Descriptions of a new genus and
three new species of Rhizoecinae (Hemiptera: Coccoidea)
associated with ants of the genus Acropyga Roger in Dominican
amber. American Museum Novitates, 3335, 16–18.
Williams, D.J. (2002) The mealybug tribe Allomyrmococcini and
its association with herdsmen ants of the genus Dolichoderus in
southern Asia. Abhandlungen der Senckenbergischen Naturforschenden Gesellschaft, 557, 115–181.
Williams, D.J. & Granara de Willink, M.C. (1992) Mealybugs of
Central and South America. CAB International, Wallingford.
Williams, M.L. & Kondo, T. (2002) Characteristics of first-instar
nymphs in the soft scale insects (Hemiptera: Coccidae):
surprising indicators of relationships. Bollettino di Zoologia
Agraria e di Bachicoltura, 33(2001), 35–42.
Yoder, A.D., Irwin, J.A. & Payseur, B.A. (2001) Failure of the ILD
to determine data combinability for slow loris phylogeny.
Systematic Biology, 50, 408–424.
Accepted 1 October 2003
2004 The Royal Entomological Society, Systematic Entomology, 29, 238–259
Host
Artemisia sp.
Adenostoma fasciculatum
Ephredra sp.
Bermuda grass roots
Bambusa multiplex
Grevillea sp.
Phormium tenax
Phylachne cushion
Bambusa dolichomerithalla
On grass crown
Poa roots, under rock
Hebe sp.
Saccharum officinalis
Laboratory culture
Under carton with ants
Laboratory culture
Ornamental juniper
Lechea sessiliflora roots
Acacia howittii
Ex roots in Lasius nest
Pistacio vera
Potato (in culture)
Mangifera indica
Ptychosperma elegans
Grewia sp.
Adenostoma fasciculatum
Gutierrezia sp.
Mandevilla sp.
Acacia dealbata
Unidentified
Acacia sp.
In Acropyga berwicki nest
Exocarpos cupressiformis
Palm foliage
Jatropha podagrica
Hibiscus sp.
Unidentified shrub
Juniperus sp.
Plumeria plumeria
Species
Amonostherium lichtensioides (Cockerell)
Anisococcus adenostomae (Ferris)
Anisococcus sp. (near A. ephedrae (Coq.))
Antonina graminis (Maskell)
Antonina pretiosa Ferris
Australicoccus grevilleae (Fuller)
Balanococcus diminutus (Leonardi)
Balanococcus sp.
Chaetococcus bambusae (Maskell)
Chorizococcus sp.
Chryseococcus longispinus (Beardsley)
Cyphonococcus alpinus (Maskell)
Dysmicoccus boninsis (Kuwana)
Dysmicoccus brevipes (Cockerell) HI
Dysmicoccus brevipes (Cockerell) BOL
Dysmicoccus neobrevipes Beardsley
Dysmicoccus ryani (Coquillett)
Dysmicoccus sp.
Erium globosum (Maskell)
Euripersia europaea (Newstead)
Ferrisia gilli Gullan
Ferrisia malvastra (McDaniel)
Ferrisia virgata (Cockerell)
Geococcus coffeae Green
Grewiacoccus sp.
Heliococcus adenostomae McKenzie
Heliococcus clemente Miller
Hypogeococcus pungens G. de Willink
Maconellicoccus australiensis (Gr. & Lid.)
Maconellicoccus hirsutus (Green)
Melanococcus albizziae (Maskell)
Neochavesia caldasiae (Balachowsky)
Nipaecoccus exocarpi Williams
Nipaecoccus nipae (Maskell)
Nipaecoccus viridis (Newstead)
Nipaecoccus sp. near N. gilli
Paracoccus burnerae (Brain)
Paracoccus juniperi (Ehrhorn)
Paracoccus marginatus Will. & G. de W.
Comins Lake, NV, U.S.A.
Mix Canyon, CA, U.S.A.
Moab, UT, U.S.A.
Mangilao, Guam
Davis, CA, U.S.A.
Aust. Nat. Uni. Canberra, Australia
Watsonville, CA, U.S.A.
Campbell Is., New Zealand
Fairchild Gardens, FL, U.S.A.
Ocala Nat. Forest, FL, U.S.A.
Campbell Is., New Zealand
Ohakune, New Zealand
Naples, FL, U.S.A.
Uni. Hawaii, HI, U.S.A.
Santa Cruz, Bolivia
Uni. Hawaii, HI, U.S.A.
Sacramento, CA, U.S.A.
Payne’s Prairie, FL, U.S.A.
Aust. Nat. Uni, Canberra, Australia
Feldberg bei Pulkau, Austria
Tulare, CA, U.S.A.
Tuczon, AZ, U.S.A.
Cali, Valle, Colombia
Ramona, CA, U.S.A.
Victoria Falls, Zimbabwe
Mix Canyon, CA, U.S.A.
Cuyama Valley, CA, U.S.A.
Largo, FL, U.S.A.
Tharwa, ACT, Australia
Chiang Mai, Thailand
Aust. Nat. Uni. Canberra, Australia
Ilheus, Bahia, Brazil
Canberra, Australia
Grove Beach, CA, U.S.A.
Lan Sang NP, Thailand
Calcutta, Belize
Ameib, Erongo Mts, Namibia
Del Puerto Canyon Road, CA, U.S.A.
Mangilao, Guam
Collection locality
#
AY426034
AY426068
AY426076
AY426028
AY426020
AY426075
AY426067
AY426019
AY426079
AY426066
AY426021
AY426071
AY426065
AY426072
AY426080
AY426033
AY426045
AY426031
AY426037
AY426046
AY426036
AY426035
AY429461
AY426030
AY426070
AY426017
AY426047
AY426027
AY426026
AY426069
AY426040
AY426064
AY426061
18S
AY427362
AY427317
AY427347
AY427358
AY427334
AY427315
AY179463
AY427318
AY427325
AY427314
AY427328
AY427360
AY427331
AY427353
AY427349
AY427329
AY427368
AY427355
AY427320
AY427321
AY427323
AY427332
AY427359
AY427333
AY427366
AY179455
AY179452
AY179458
AY427357
AY427313
AY427361
AY427356
AY427363
AY179433
AY427322
AY427326
AY427344
D2
28S
AY427421
AY427393
AY427422
AY427435
AY427387
AY427410
AY427431
AY427418
AY427399
AY427406
AY427411
AY427394
AY427415
AY427390
AY427428
AY427398
AY427383
AY427373
AY427420
AY427430
AY427425
AY427419
AY427426
AY427377
AY427403
AY427392
AY427381
AY427385
AY427424
AY427388
AY427378
AY427396
AY427391
AY427423
AY427402
AY427437
AY427413
D10
AY427294
AY427295
AY427270
AY427292
AY427287
AY427266
AY427289
AY427286
AY427290
AY427261
AY427275
AY427299
AY427297
AY427285
AY427279
AY427280
AY427277
AY427276
AY427284
AY427273
AY427304
AY427282
AY427268
AY427288
AY427307
AY427298
AY427267
AY427274
AY427293
50 a
EF-1a
AY427249
AY427224
AY427236
AY427257
AY427245
AY427211
AY427248
AY427243
AY427251
AY179472
AY427220
AY427208
AY427212
AY427240
AY427214
AY427253
AY179473
AY179475
AY179477
AY427244
AY427238
AY427215
AY427256
AY427246
AY427227
AY427226
AY179481
AY427218
AY427209
AY427213
AY427225
AY427250
AY427230
30 b
Collection details and gene regions sequenced for each taxon. With two exceptions (Dysmicoccus brevipes, 1993, and Rastrococcus iceryoides, 1995) all specimens were collected between
2000 and 2002.
Appendix
258 D. A. Downie and P. J. Gullan
2004 The Royal Entomological Society, Systematic Entomology, 29, 238–259
#
2004 The Royal Entomological Society, Systematic Entomology, 29, 238–259
Aristida sp.
Nandina domestica
Arctostaphylos viscida
Schefflera sp.
Arctostaphylos sp.
Ceanothus sp.
OUTGROUPS
Aclerda sp.
Icerya purchasi Maskell
Puto albicans McKenzie
Puto barberi (Cockerell)
Puto sp. (nymph)
Puto yuccae (Coquillett)
b
Including positions 2103–2342 in the Drosophila melanogaster sequence.
Including positions 2832–3149 in the Drosophila melanogaster sequence.
CA, California; FL, Florida; ID, Idaho; HI, Hawaii; NV, Nevada; UT, Utah.
a
Nothofagus solandri
Unidentified legume
Weinmannia racemosa
On grass roots
Acacia drepanolobium
Under stone with ants
Arctostaphylos viscida
Penstemon sp.
Hibiscus sp.
Bidens sp. crown
Mangifera indica
Citrus sinensis
Butternut squash (laboratory)
Eugenia sp. leaves
Citrus
Citrus
Vitis vinifera
Aglaonema stems
Vitis vinifera
Artemisia tridentata roots
Neodypsis decaryi
Saccharum sp.
Nothofagus menziesii
Leaf sheath of grass
Leymus arenarius
Grass roots
Narcissus sp.
Echeveria chihuahuaensis
Paracoccus nothofagicola Cox
Paracoccus sp. NAM
Paracoccus sp. NZ
Paradoxococcus mcdanieli McKenzie
Paraputo sp.
Phenacoccus colemani Ehrhorn 1
Phenacoccus colemani Ehrhorn 2
Phenacoccus madeirensis Green
Phenacoccus solani Ferris CA
Phenacoccus solani Ferris FL
Planococcoides robustus Ezzat & McC.
Planococcus citri (Risso)
Planococcus ficus (Signoret)
Plotococcus eugeniae Miller & Denno
Pseudococcus calceolariae (Maskell)
Pseudococcus longispinus (Targ. Tozz.)
Pseudococcus maritimus (Ehrhorn)
Pseudococcus odermatti Mill. & Will.
Pseudococcus viburni (Signoret)
Rhizoecus gracilis McKenzie
Rhizoecus hibisci Kawai & Takagi
Saccharicoccus sacchari (Cockerell)
Sarococcus comis Cox
Tridiscus distichlii (Ferris)
Trionymus frontalis McKenzie
Trionymus idahoensis Miller & McK.
Vryburgia amaryllidis (Bouché)
Vryburgia trionymoides (De Lotto)
Withlacoochee State For., FL, U.S.A.
Davis, CA, U.S.A.
11 km SE Placerville, CA, U.S.A.
Palmira, Colombia
McLaughlin Reserve, CA, U.S.A.
Saint Helena, CA, U.S.A.
Ohakune, New Zealand
Namib Naukluft Lodge, Namibia
Ohakune, New Zealand
Payne’s Prairie, FL, U.S.A.
Mpala Research Centre, Kenya
Mt George, CA, U.S.A.
Washington, CA, U.S.A.
Vacaville, CA, U.S.A.
Sacramento, CA, U.S.A.
Vero Beach, FL, U.S.A.
Hapur, Uttar Pradesh, India
Davis, CA, U.S.A.
Kearney, CA, U.S.A.
Tavernier, Key Largo, FL, U.S.A.
Swan Reach, Victoria, Australia
Wemen, Victoria, Australia
Witstrand, WA, U.S.A.
Naples, FL, U.S.A.
Winters, CA, U.S.A.
Holbrook Pass, ID, U.S.A.
Hilo, Hawaii, U.S.A.
Kendall Station, FL, U.S.A.
Ohakune, New Zealand
Bodega Bay, CA, U.S.A.
Santa Barbara, CA, U.S.A.
Holbrook Pass, ID, U.S.A.
Davis, CA, U.S.A.
Davis, CA, U.S.A.
AY426060
AY426078
AY426051
AY426048
AY426029
AY426052
AY426049
AY426056
AY426024
AY426074
AY426053
AY426063
AY426023
AY426054
AY426059
AY426039
AY426038
AY426043
AY426025
AY426050
AY426058
AY426032
AY426042
AY426055
AY426077
AY426044
AY426041
AY426062
AY426057
AY426073
AY427339
AY427348
AY427371
AY427338
AY427336
AY427364
AY427337
AY427330
AY427351
AY427324
AY179451
AY427341
AY427370
AY427335
AY179456
AY427312
AY427354
AY427309
AY427367
AY427346
AY427352
AY427319
AY427340
AY427345
AY427365
AY427311
AY427342
AY427369
AY427316
AY42327
AY427350
AY427343
AY427281
AY427308
AY427301
AY427296
AY427372
AY427408
AY427271
AY427303
AY427302
AY427278
AY427262
AY427265
AY427305
AY427260
AY427306
AY427264
AY427263
AY427291
AY427272
AY427269
AY427300
AY427283
AY427412
AY427432
AY427379
AY427416
AY427384
AY427436
AY427409
AY427427
AY427382
AY427407
AY427389
AY427395
AY427374
AY427434
AY427375
AY427405
AY427429
AY427401
AY427400
AY427386
AY427417
AY427376
AY427380
AY427433
AY427397
AY427404
AY427414
AY427438
AY427259
AY429462
AY427219
AY427232
AY427237
AY427252
AY427216
AY427234
AY179474
AY427233
AY427258
AY427229
AY179487
AY427217
AY427242
AY427222
AY427254
AY429463
AY427241
AY427255
AY427223
AY427228
AY427239
AY427235
AY427231
AY427247
AY427210
Mealybug molecular systematics
259