The phylogeny of the superfamily Coccoidea (Hemiptera: Sternorrhyncha) based on the morphology of extant and extinct macropterous males

Authors


Correspondence: Dr. Chris J. Hodgson, Department of Biodiversity and Biological Systematics, The National Museum of Wales, Cardiff CF10 3NP, U.K. E-mail: hodgsoncj@cardiff.ac.uk

Abstract

Currently, 49 families of scale insects are recognised, 33 of which are extant. Despite more than a decade of DNA sequence-based phylogenetic studies of scales insects, little is known with confidence about relationships among scale insects families. Multiple lines of evidence support the monophyly of a group of 18 scale insect families informally referred to as the neococcoids. Among neococcoid families, published DNA sequence-based estimates have supported Eriococcidae paraphyly with respect to Beesoniidae, Dactylopiidae, and Stictococcidae. No other neococcoid interfamily relationship has been strongly supported in a published study that includes exemplars of more than ten families. Likewise, no well-supported relationships among the 15 extant scale insect families that are not neococcoids (usually referred to as ‘archaeococcoids’) have been published. We use a Bayesian approach to estimate the scale insect phylogeny from 162 adult male morphological characters, scored from 269 extant and 29 fossil species representing 43/49 families. The result is the most taxonomically comprehensive, most resolved and best supported estimate of phylogenetic relationships among scale insect families to date. Notable results include strong support for (i) Ortheziidae sister to Matsucoccidae, (ii) a clade comprising all scale insects except for Margarodidae s.s., Ortheziidae and Matsucoccidae, (iii) Coelostomidiidae paraphyletic with respect to Monophlebidae, (iv) Eriococcidae paraphyletic with respect to Stictococcidae and Beesoniidae, and (v) Aclerdidae sister to Coccidae. We recover strong support for a clade comprising Phenacoleachiidae, Pityococcidae, Putoidae, Steingeliidae and the neococcoids, along with a sister relationship between this clade and Coelostomidiidae + Monophlebidae. In addition, we recover strong support for Pityococcidae + Steingeliidae as sister to the neococcoids. Data from fossils were incomplete, and the inclusion of extinct taxa in the data matrix reduced support and phylogenetic structure. Nonetheless, these fossil data will be invaluable in DNA sequence-based and total evidence estimates of phylogenetic divergence times.

Introduction

The scale insects (Coccoidea) are small, sap-sucking true bugs (Hemiptera), sister to Aphidoidea in the suborder Sternorrhyncha. Their common name refers to the protective covering or ‘scale’ that is secreted by many species. More than 7700 species have been described (Ben-Dov et al., 2013). Among that number are many agricultural pests (Miller & Davidson, 1990) and invasive species (Miller et al., 2005), as well as species exhibiting varied endosymbioses (Buchner, 1965; Gruwell et al., 2005, 2007), diverse sexual systems (Nur, 1980; Normark, 2003; Ross et al., 2010) and sexual dimorphism (Gullan & Kosztarab, 1997).

The superfamily Coccoidea is divided routinely into two informal groups, the archaeococcoids and the neococcoids (reviewed by Gullan & Cook, 2007). The monophyly of neococcoids is supported by a synapomorphic genetic system (Paternal Genome Elimination: Nur, 1980; Cook et al., 2002; Normark, 2003) and DNA sequence data (Cook et al., 2002; Yokogawa & Yahara, 2009; Ross et al., 2013). Monophyly of the archaeococcoids has not been recovered (Cook et al., 2002; Gullan & Cook, 2007). Currently, 18 extant neococcoid families and 15 extant archaeococcoid families are recognised, along with 16 additional families known only from fossils (Ben-Dov, 2011; Ben-Dov et al., 2013). However, phylogeny estimates based on both morphological (e.g Gullan & Sjaarda, 2001; Hodgson, 2002; Hodgson & Foldi, 2005) and molecular data (Cook et al., 2002; Cook & Gullan, 2004; Gullan & Cook, 2007) have cast doubt on the monophyly of some scale insect families and, outside of neococcoid monophyly, very little is known about relationships among scale insect families.

Scale insect taxonomy is based almost entirely on adult females, which are paedomorphic and highly specialised for sedentary lives on host plants. They lack distinct tagmata, wings, compound eyes and sclerotic cuticle other than on the legs, mouthparts, antennae and certain anal structures. The female lifecycle involves two or three actively feeding instars prior to the adult stage, which is the longest-lasting stage in scale insect development (Gullan & Martin, 2009). In contrast, almost all adult male scale insects are dipterous, ‘normal’ insects, with clearly differentiated tagmata, numerous cephalic and thoracic sclerites, and obvious eyes, but without mouthparts (they are nonfeeding). Their lifecycle involves two feeding nymphal instars followed by two nonfeeding ‘pupal’ stages (Gullan & Kosztarab, 1997). They are small, ephemeral (living from a few hours to maybe 3 or 4 days) and capable of little more than delivering sperm to adult females.

Adult males are difficult to collect and rarely studied because they are so short-lived. However, many early scale insect workers (e.g. Morrison, 1928; Balachowsky, 1937, 1942; Ferris, 1942, 1950, 1957; Borchsenius, 1958, 1965) felt that adult male morphology could provide insights into the evolutionary relationships among major lineages of Coccoidea. Major groundwork in this area was performed by a cadre of Boratynski's students—Theron (1958), Ghauri (1962), Giliomee (1967) and Afifi (1968)—who produced the first detailed surveys of the morphology of adult male scale insects. Since then, the adult males of approximately 350 extant species have been described, along with several fossil taxa that have been described on the basis of adult male inclusions in amber (Beardsley, 1968; Koteja, 1984). Thus, the morphological and systematic framework is in place for us to attempt to improve our understanding of evolutionary relationships among scale insect families through phylogeny estimation based on adult male morphology. This approach also has the potential to shed light on the divergent evolutionary histories of extinct scale insect lineages. Therefore, our goals in this study are to (i) estimate phylogenetic relationships among family-level lineages of Coccoidea from adult male morphology and (ii) place adult male scale insect fossils in a phylogenetic framework.

Methods

The data for the analysis were taken from various sources (Table S1). An effort was made to include exemplars from all genera for which macropterous males were available, preferably with several examples. A few families are monotypic but, for other families, either all known descriptions were used (e.g. Beesoniidae and Diaspididae) or up to 84 species were included (e.g. Coccidae). The aphid species Eucallipterus tiliae L. was included as an outgroup. Thus, data for macropterous males of 269 extant species and a further 29 extinct species were analysed. Only two fully-winged extinct neococcoid taxa have been described and only one of these was included. Taxa were scored for 162 characters (Table S2)—although character state information was incomplete for fossil taxa. The data for the fossil taxa were taken entirely from published descriptions (Table S1) and the number of scored characters ranged from 40 to 80. Macropterous male scale insect specimens in amber tend to be distorted and shrivelled, and frequently are covered in what appears to be fungus. The amber can be discoloured or contain many air bubbles. Usually, only a single specimen is known for each extinct taxon and therefore features that are hidden in one specimen cannot be checked on others. Furthermore, specimens need to be illuminated from above (as they are uncleared) and therefore are only observable under low magnification (relative to the magnification possible with cleared specimens). Because of the incompleteness of the fossil data, two phylogenetic analyses were performed: the first included only the 269 extant species, and the second included fossil data. Our discussion of the phylogenetic relationships among scale insect families is based on the first (extant taxon) analysis (Nexus File S3). The second analysis (Nexus File S4) was used to place extinct taxa in a phylogenetic framework.

We used the Markov k-state 1-parameter (mk1) model of morphological character evolution (Lewis, 2001), and modelled the variation in the rate of evolution among characters with a gamma distribution. We employed a Bayesian approach to estimate phylogenetic relationships, using the Metropolis-coupled Markov chain Monte Carlo method (MC3) in MrBayes v3.2.1, with the default priors (Ronquist et al., 2012). Each tree-search consisted of two simultaneous runs of four chains for 15 million generations, sampling trees from the stationary distribution once every 1000 generations. The burnin value of trees to discard prior to stationarity was determined by examining likelihood and parameter traces. The set of high posterior probability trees was summarised with a majority rule consensus tree. Synapomorphic diagnoses for families and multi-family clades were determined by examining traces of parsimony ancestral state reconstructions on the majority rule consensus tree in Mesquite (Maddison & Maddison, 2013).

Results

Phylogenetic relationships among scale insect families

In the following discussion of phylogenetic relationships, we use the phrase ‘the rest of Coccoidea’ recursively. That is, we move from left to right on a left-ladderised tree, and refer to the species-poor group that is sister to the ‘rest of Coccoidea’ which consists of all of the family-level lineages that have not been referred to previously (Figs 1, 2; Nexus File S5).

Figure 1.

Majority rule consensus tree summarizing a Bayesian phylogeny estimate of relationships amongst extant non-neococcoid macropterous male scale insects. Nodal support indicated with posterior probabilities.

Figure 2.

Majority rule consensus tree summarising high posterior probability trees from a Bayesian phylogeny estimate of relationships amongst neococcoid macropterous adult male scale insects. Nodal support indicated with posterior probabilities.

Extant archaeococcoids

Monophyly was recovered for Matsucoccidae (PP 1), Ortheziidae (PP 1), Margarodidae s.s. (PP 1), Stigmacoccidae (PP 1), Xylococcidae (PP 1), the new Zealand members of Coelostomidiidae (although with low support PP 0.72), Monophlebidae (PP 1), Phenacoleachiidae (PP 0.97), and Putoidae (PP 1) (Fig. 1). Nonmonophyly was recovered for Kuwaniidae and Coelostomidiidae. The Kuwaniidae species Neosteingelia sp. was more closely related to what remains of Coccoidea after excluding Matsucoccidae, Ortheziidae and Margarodidae s.s., than it was to a second species of Neosteingelia (PP 0.99). The South American genera Nautococcus Vayssière (Coeolostomidiidae, not Monophlebidae as in Ben-Dov, 2011; see Williams & Gullan, 2008: 82) and Neocoelostoma Hempel (Coelostomidiidae) were sister to Monophlebidae (PP 1), rendering the Coelostomidiidae nonmonophyletic. The monophyly of the families Callipappidae, Steingeliidae and Pityococcidae was not tested because each group was represented by a single exemplar.

Strong support (PP 0.96) was recovered for a sister relationship between Matsucoccidae and Ortheziidae. Strong support was recovered for a clade comprised of all scale insect species other than Ortheziidae + Matsucoccidae and Margarodidae s.s. (PP 1). Each of the two Kuwaniidae species were recovered as successive sisters to the rest of Coccoidea. Callipappidae and Stigmacoccidae were sisters (low support PP 0.64) and this clade was sister to the rest of Coccoidea (PP 1). Xylococcidae was recovered as sister to the rest of Coccoidea (little support, PP 0.59). The Coelostomidiidae + Monophlebidae clade was sister to the rest of Coccoidea (PP 1). Phenacoleachiidae was sister to the rest of Coccoidea (PP 1). Putoidae was sister to the rest of Coccoidea (PP 0.85). Steingeliidae was recovered as sister to Pityococcidae with low support (pp 0.51). Strong support (PP 0.95) was recovered for Steingeliidae + Pityococcidae as sister to the neococcoids.

Neococcoids

Support was recovered for the monophyly of Rhizoecidae (PP 1), the Pseudococcidae subfamily Pseudococcinae (PP 0.97), Dactylopiidae (PP 1), Beesoniidae (PP 1), Stictococcidae (PP 1), Cerococcidae (PP 1), Asterolecaniidae (PP 1), Conchaspididae (PP 1), Diaspididae (PP 1), Kerriidae (PP 1) and Aclerdidae (PP 1) (Fig. 2). Monophyly was not recovered for the Pseudococcidae, the Pseudococcidae subfamily Phenacoccinae and Eriococcidae—although with the exception of Eriococcidae, no strong support was found for nonmonophyly. Eriococcidae was paraphyletic with respect to Stictococcidae and Beesoniidae (PP 1).

Strong support was recovered for the monophyly of the neococcoids (PP 0.98). Among neococcoids, strong support was recovered for Aclerdidae sister to Coccidae (PP 0.99) and marginal support was recovered for Conchaspididae sister to Diaspididae (PP 0.92). Otherwise, interfamily relationships were poorly supported.

Phylogenetic relationships of extinct taxa

Inclusion of the fossils resulted in decreased resolution and support for deeper nodes in the phylogeny. In addition, the inclusion of extinct taxa resulted in the nonmonophyly of the extant families Matsucoccidae, Ortheziidae and Xylococcidae—although in each case there was no strong support for nonmonophyly (Fig. 3; Nexus File S6). If the extinct taxa †Palaeotupo and †Marmyan are included within Putoidae, this would render the family paraphyletic with respect to Labiococcidae and Phenacoleachiidae (PP 0.85). Reciprocal monophyly was not recovered for †Electrococcidae and Pityococcidae; the two †Turonicoccus species were more closely related to Pityococcidae than to †Apticoccus minutus (†Electrococcidae) (PP 1). However, incorporating fossil data recovered the monophyly of the Kuwaniidae (PP 0.94). Weak support was recovered for a clade comprising all scale insects excluding Matsucoccidae and Ortheziidae (PP 0.76). The extinct taxa †Lebanococcidae and †Hammanococcidae were recovered in a polytomy with a clade made up of the rest of Coccoidea (PP 0.75). †Jerseycoccidae was recovered as sister to Margarodidae s.s. (PP 0.83). Weak support was recovered for †Jerseycoccidae + Margarodidae as sister to a clade comprising †Serafinidae, †Grohnidae, †Weitschatidae, Xylococcidae, Kuwaniidae, Callipappidae, Stigmacoccidae and Coelostomidiidae + Monophlebidae (PP 0.65). Within this clade, marginal support was found for †Serafinidae as sister to the remaining taxa (PP 0.93). Marginal support was recovered for a sister relationship between Coelostomidiidae + Monophlebidae and a clade made up of Stigmacoccidae, Callipappidae, and Kuwaniidae (PP 0.88). †Burmacoccidae was recovered as sister to the rest of Coccoidea (PP 0.84). †Albicoccidae was sister to the rest of Coccoidea (PP 0.76). †Grimaldiellidae was sister to the rest of Coccoidea (PP 0.51). The extinct taxa †Kukaspididae, †Pennygullaniidae and †Apticoccus (†Electrococcidae) were in a polytomy with the rest of Coccoidea (PP 0.63). The extinct steingeliid genus †Palaeosteingelia was recovered as sister to the extant steingeliid genus Steingelia Nasonov (PP 0.51). The extinct taxon †Kuenowicoccus (Eriococcidae) was recovered as sister to the rest of the neococcoids.

Figure 3.

Majority rule consensus tree summarising high posterior probability trees from a Bayesian phylogeny estimate of relationships among fossil and extant non-neococcoid macropterous male scale insects. Nodal support indicated with posterior probabilities. Extinct taxa in bold and preceded by †.

Synapomorphic diagnoses for well-supported clades are provided in Table S7.

Discussion

With a Bayesian analysis of adult male morphological data, we recovered a highly resolved estimate of phylogenetic relationships amongst scale insect families, with several highly supported interfamily relationships. This result is in stark contrast to the poor resolution and weak support in previously published estimates, with the exception of Yokogawa & Yahara (2009). Although taxonomic and genetic sampling in that study were limited (sequences of COI and COII sampled from 30 scale insect species representing nine families), support was recovered for (i) paraphyly of the archaeococcoids with respect to neococcoids, (ii) neococcoid monophyly, (iii) Cerococcidae sister to Asterolecaniidae, (iv) Cerococcidae + Asterolecaniidae + Kermesidae sister to Coccidae, (v) a sister relationship between the latter clade and Eriococcidae + Diaspididae, and (vi) Pseudococcidae sister to all other neococcoids. In the present study, we were able to test these results with independent data and a much expanded taxonomic sample. One of the most significant results is the high support we recovered for relationships among archaeococcoid taxa. The most comprehensive sampling of archaeococcoid families in a DNA sequence-based phylogeny estimate (Gullan & Cook, 2007) recovered all included archaeococcoid families in a single polytomy; i.e. a star phylogeny in which any of the possible interfamily relationships was equally probable. That result is an apt summary of the general lack of congruence or support among previous estimates of relationships among archaeococcoid families (e.g. Miller, 1984; Cook et al., 2002; Hodgson & Foldi, 2005). Here, we have statistical evidence that the archaeococcoids are paraphyletic with respect to the neococcoids (as in Yokogawa & Yahara, 2009), along with high support for several other interfamily relationships.

Coelostomidiidae + Monophlebidae

Coelostomidiidae, as defined currently (i.e. excluding Marchalina Vayssière), was found to be monophyletic and sister to Marchalinidae + Monophlebidae in a phylogenetic analysis of adult female and first-instar morphology (Gullan & Sjaarda, 2001). In the present study, the South American Neocoelostoma and Nautococcus (both Coelostomidiidae) formed a clade sister to Monophlebidae, rather than forming a clade with the New Zealand members of Coelostomidiidae. To preserve the monophyly of Coelostomidiidae and Monophlebidae, these South American taxa could be treated as a separate family-level group, or they could be transferred to Monophlebidae.

Phenacoleachiidae + Putoidae + Steingeliidae + Pityococcidae + neococcoids

With the exception of Phenacoleachia zealandicus (Maskell) in which the adult female's labium is four-segmented, all adult female scale insects (including Phenacoleachia australis Beardsley, the only other described phenacoleachiid) have a labium composed of three or fewer segments. This difference has been used to argue that the Phenacoleachiidae is sister to the rest of the extant scale insects (Koteja, 1974). Here, the Phenacoleachiidae were found to share a much more recent common ancestor with the families Putoidae, Steingeliidae, Pityococcidae and the neococcoids. The adult males of all species in this group have glandular pouches on the abdomen. The non-neococcoid species in this group have a ring of simple eyes, but a ring of simple eyes is found also in the neococcoid family Kermesidae (Sternlicht, 1969; Koteja & Zak-Ogaza, 1972) and in some Coccidae (e.g. Eulecanium, Philephedra; Giliomee, 1968; Miller & Williams, 2002). The adult males of some of the non-neococcoid species in this group have been recorded as having abdominal spiracles [Phenacoleachia zealandica (Theron, 1962); P. australis (Beardsley, 1964); Steingelia gorodetskia (Theron, 1958)]. However, Hodgson & Foldi (2006) were unable to see any actual spiracles in either Phenacoleachia or Steingelia even though tracheae were clearly visible near each segmental margin. Therefore, spiracles in these species are either extremely small or absent. In contrast, scale insects species outside of this group (i.e. those with compound eyes) have obvious and well-developed abdominal spiracles. Another feature that suggests a close relationship between the traditional ‘archaeococcoid’ lineages in this group and the neococcoids is the presence of ostioles on the males of Phenacoleachia australis (Maskell) and Pityococcus sp. (Beardsley, 1964; Hodgson & Foldi, 2006), a trait associated most often with the neococcid family Pseudococcidae.

Pseudococcidae

DNA sequence-based studies of pseudococcid phylogeny (Downie & Gullan, 2004; Hardy et al., 2008) have recovered two or three major clades that have been equated to the subfamilies Pseudococcinae, Phenacoccinae and Rhizoecinae (Hardy et al., 2008 recovered the Rhizoecinae within the Phenacoccinae). More recently, the Rhizoecidae have been elevated to family rank, on the basis of a comparative analysis of adult male morphology (Hodgson, 2012). In the present study, the Pseudococcinae was recovered as monophyletic, and the Phenacoccinae was not recovered as monophyletic; however, there was no strong support for nonmonophyly of Phenacoccinae. The Rhizoecidae were recovered as sister to all other neococcoids (as in Hodgson, 2012), and its family rank is weakly supported, although by the same type of data that have been used in past arguments for family rank. The male morphology estimate does not conflict with previous phylogeny estimates (Miller, 1984; Danzig, 1986; Cook et al., 2002; Hodgson, 2002; Cook & Gullan, 2004; Yokogawa & Yahara, 2009; Ross et al., 2013) that have indicated that the Pseudococcidae (or part of the Pseudococcidae) are sister to the rest of the neococcoids.

Eriococcidae

Eriococcidae paraphyly with respect to Dactylopiidae, Beesoniidae and Stictococcidae has been repeatedly demonstrated (Cook et al., 2002; Hodgson, 2002; Cook & Gullan, 2004; Gullan & Cook, 2007) in estimates based on both morphological and DNA sequence data. This study is congruent with that result. The Eriococcidae were recovered as paraphyletic with respect to Beesoniidae, Dactylopiidae and Stictococcidae, although the relationship with Dactylopiidae was poorly supported (PP 0.60). Monophyly was not recovered for an expanded concept of Eriococcidae that includes Dactylopiidae, Beesoniidae and Stictococcidae, although there was no strong support for the nonmonophyly of this group. In contrast to the close relationship between the Eriococcidae type species, Eriococcus buxi (Boyer de Fonscolombe), and the families Beesoniidae and Stictococcidae estimated by Cook & Gullan (2004), in the present study, E. buxi was recovered in a basal polytomy of neococcoid taxa excluding Rhizoecidae. Another conspicuous difference between DNA sequence-based estimates of eriococcid phylogeny and the estimate in this study is the failure of adult male data to support the ‘acanthococcid’ clade of felt scale species closely related to Dactylopiidae (as found by Cook & Gullan, 2004). Here, most putative acanthococcids were recovered in a basal eriococcid/pseudococcid polytomy.

Kermesidae

The kermesids have been included in several phylogenetic studies (Miller, 1984; Miller & Miller, 1993; Foldi, 1997; Miller & Hodgson, 1997; Cook et al., 2002; Hodgson, 2002; Gullan & Cook, 2007; Yokogawa & Yahara, 2009). Some of these studies have suggested that the Kermesidae is sister to all neococcoids except the Pseudococcidae and ‘eriococcids’. Here, they were recovered as sister to a clade composed of Aclerdidae, Asterolecaniidae, Cerococcidae, Coccidae and Lecanodiaspididae—i.e. all of the necoccoids except the Pseudococcidae, ‘eriococcids,’ Conchaspididae, and Diaspididae.

Lecanodiaspididae, Cerococcoidae, Asterolecaniidae, Conchaspididae, Diaspididae

The relationships amongst these families is particularly unclear. Adult females of Lecanodiaspididae, Cerococcidae and Asterolecaniidae all share large ‘8’-shaped pores and sclerotised anal plates. In the only published phylogeny estimate to include exemplars of all three of these families (Foldi, 1997) they were recovered as a clade, as was the case in this study. Marginal support was recovered for Diaspididae as sister to Conchaspididae. A close relationship between the latter two families has been suggested on the basis of similarities between adult females (Mamet, 1954; Boratyński & Davies, 1971). Others have felt that differences in labial segmentation would preclude such a close relationship (Koteja, 1974). A few species-poor families thought to be closely related to Diaspididae (Stickney, 1934)—Halimococcidae, Thysanococcidae, Phoenicococcidae—were excluded because the males are apterous. We recovered weak evidence for a sister relationship between Diaspididae + Conchaspididae and the rest of neococcoids excluding pseudococcids, and ‘eriococcids.’ Previously, a close relationship between Diaspididae and Asterolecaniidae or Lecanodiaspididae has been suggested (Stickney, 1934; Boratyński & Davies, 1971; Rosenblueth et al., 2012), along with several alternative arrangements. For example, from small ribosomal subunit DNA data, Gullan & Cook (2007) estimated a sister relationship between Diaspididae and the ‘Gondwanan clade’ of Eriococcidae (sensu Cook & Gullan, 2004). More recently, Yokogawa & Yahara (2009), using mitochondrial sequence data, recovered a sister relationship between Diaspididae and a different subgroup of Eriococcidae. None of these relationships have been estimated with strong support.

Coccidae, Aclerdidae, Kerriidae

The only previously published scale insect phylogeny to include exemplars of Coccidae and Aclerdidae (Hodgson, 2002) recovered Aclerdidae as sister to Coccidae, as was the case in this study. Several published phylogeny estimates that included exemplars of Coccidae and Kerriidae (Foldi, 1997; Cook & Gullan, 2004; Gullan & Cook, 2007) inferred a close relationship between the two. In the present analysis, Kerriidae was sister to a clade comprising Lecanodiaspididae, Cerococcidae, Asterolecaniidae, Aclerdidae and Coccidae. In an unpublished study of coccid phylogenetic relationships based on DNA sequence data (T. Kondo and L.G. Cook, unpublished data), Coccidae was found to be paraphyletic with respect to Aclerdidae and Kerriidae was sister to the Coccidae + Aclerdidae.

Fossil placement

One of the most striking features of the Cretaceous fossils is the abundance of species with lateral rows of simple eyes (Koteja, 2000a,b). The extinct genera considered in this study having lateral rows of simple eyes are †Albicoccus Koteja, †Apticoccus Koteja & Azar, †Grimaldiella Koteja, †Kukaspis Koteja, †Marmyan Koteja, †Palaeosteingelia Koteja & Azar, †Palaeotupo Koteja & Azar, †Solicoccus Koteja, †Turonicoccus Koteja and possibly †Pennygullania Koteja & Azar (Koteja, 2000a, 2001, 2004; Koteja & Azar, 2008). This group of extinct taxa was recovered in a clade with Putoidae, Steingeliidae, Pityococcidae and the neococcoids. However, †Burmacoccus Koteja, which also has compound eyes composed of a few ommatidia, was recovered in this clade. From our analysis, it appears that fossil Coccoidea with bands of simple eyes once included a large group of diverse families most of which are now extinct but from which the neococcoids and three extant smaller families survive today.

The exclusion of Pennygullania from the neococcoids is questionable. Recently, Gavrilov-Zimin & Danzig (2012) synonymised Pennygullaniidae with Pseudococcidae, arguing that the characters used by Koteja & Azar (2008) to define Pennygullaniidae were insufficient to distinguish it from Pseudococcidae. Furthermore, the adult male of Pennygullania has two pairs of glandular pouches on the abdomen, as is the case for adult males in the pseudococcid subfamily Phenacoccinae. However, careful study of Koteja & Azar's (2008) illustration suggests that there is a ring of at least three pairs of simple eyes on heavily sclerotised ocular sclerites that almost meet on the top of the head, a condition very similar to that in Putoidae. Thus, there may be cause to keep these families separate.

Conclusion

The intuition shared by many early scale insect systematists— that a comparative analysis of the morphology of adult male scale insects might shed light on deep phylogenetic relationships among scale insect lineages—appears to have been correct. We gained insights into scale insect phylogeny through analysis of macropterous adult male morphology data. In the future, as more comprehensive DNA sequence data become available, these insights will be tested and refined, and ultimately some may be discarded. Nevertheless, the evolutionary histories of the rich diversity of fossil scale insects—most of which are adult males in amber inclusions—always will only be accessible through their morphologies. Thus, the appearance of adult male scale insects will continue to provide indispensable information about the timing and pattern of scale insect diversification.

Acknowledgements

We would particularly like to thank Penny Gullan for her great help and advice, and Lyn Cook for her useful comments on an earlier version of the manuscript. We also wish to thank all those who kindly lent male specimens for study, particularly Dug Miller, the late Rosa Henderson, Imre Foldi, Penny Gullan, Ferenc Kozar, Jon Martin and Yair Ben-Dov. CJH is also extremely grateful to the National Museum of Wales for providing encouragement and facilities for his research. NBH thanks Gavin Svenson at the Cleveland Museum of Natural History (CMNH) for providing postdoctoral support through NSF grant DEB1216309. This project made use of CMNH Invertebrate Zoology computational resources. CJH produced the morphological dataset and NBH analysed it: both interpreted the results and wrote the manuscript.

Ancillary